L-Type Calcium Channels Mediate Calcium Oscillations in Early Postnatal Purkinje Neurons

Abstract

Ca²⁺ signaling is important in many fundamental neuronal processes including neurotransmission, synaptic plasticity, neuronal development, and gene expression. In cerebellar Purkinje neurons, Ca²⁺ signaling has been studied primarily in the dendritic region where increases in local Ca²⁺ have been shown to occur with both synaptic events and spontaneous electrical activity involving P-type voltage-gated Ca²⁺ channels (VGCCs), the predominant VGCC expressed by Purkinje neurons. Here we show that Ca²⁺ signaling is also a prominent feature of immature Purkinje neurons at developmental stages that precede expression of dendritic structure and involves L-type rather than P-type VGCCs. Immature Purkinje neurons acutely dissociated from postnatal day 4–7 rat pups exhibit spontaneous cytoplasmic Ca²⁺ oscillations. The Ca²⁺oscillations require entry of extracellular Ca²⁺, are blocked by tetrodotoxin, are communicated to the nucleus, and correlate closely with patterns of endogenously generated spontaneous and evoked electrical activity recorded in the neurons. Immunocytochemistry showed that L-, N-, and P/Q-types of VGCCs are present on the somata of the Purkinje neurons at this age. However, only the L-type VGCC antagonist nimodipine effectively antagonized the Ca²⁺ oscillations; inhibitors of P/Q and N-type VGCCs were relatively ineffective. Release of Ca²⁺from intracellular Ca²⁺ stores significantly amplified the Ca²⁺ signals of external origin. These results show that a somatic signaling pathway that generates intracellular Ca²⁺ oscillations and involves L-type VGCCs and intracellular Ca²⁺ stores plays a prominent role in the Ca²⁺ dynamics of early developing Purkinje neurons and may play an important role in communicating developmental cues to the nucleus.

Cerebellar Purkinje neurons are a favored model for studying Ca²⁺ signaling because of the prominent role played by Ca²⁺ in dendritic excitability and synaptic transmission in this neuronal type (Tank et al., 1988;Hockberger et al., 1989; Gruol et al., 1992; Miyakawa et al., 1992;Eilers et al., 1996). In mature Purkinje neurons, dendritic voltage-gated Ca²⁺ channels (VGCCs), predominantly the P-type (Hillman et al., 1991), have been shown to mediate changes in dendritic Ca²⁺ levels associated with spontaneous or evoked electrical activity (Tank et al., 1988; Sugimori and Llinas, 1990). In addition, stimulation of either climbing fibers or parallel fibers, the excitatory afferents to Purkinje neurons, produces local increases in dendritic Ca²⁺ via activation of glutamate receptors and VGCCs (Ross and Werman, 1987; Sugimori and Llinas, 1990; Konnerth et al., 1992; Miyakawa et al., 1992; Eilers et al., 1995). Release of Ca²⁺ from intracellular Ca²⁺ stores also contributes to Ca²⁺ signaling in Purkinje neuron dendrites and is mediated both by ryanodine receptors (RyR) and inositol-1,4,5-trisphosphate receptors (IP3R) (Kano et al., 1995a;Gruol et al., 1996; Finch and Augustine, 1998; Narasimhan et al., 1998).

Ca²⁺ signaling is also an important physiological property of developing neurons, but relatively little is known about this process in immature Purkinje neurons. Components essential for Ca²⁺ signaling in Purkinje neuron dendrites are expressed early in Purkinje neuron development, when Purkinje neurons are structurally relatively simple and consist primarily of a somata and axon with little or no dendritic structure. These Ca²⁺-signaling components included VGCCs, RyRs, IP3Rs, and the Ca²⁺-binding proteins calbindin and parvalbumin (Gruol and Franklin, 1987; Regan, 1991; Gruol et al., 1992; Mintz et al., 1992; Llano et al., 1994;Milosevic and Zecevic, 1998). Large Ca²⁺transients can be experimentally evoked in early postnatal Purkinje neurons (Kano et al., 1995b), indicating the functional capacity for Ca²⁺ signaling. However, the relationship between this capacity and the physiology of the immature Purkinje neurons is unknown. Physiological events associated with dendritic Ca²⁺ signaling in mature Purkinje neurons are expressed at the somatic level in early developing Purkinje neurons including endogenously generated electrical activity and excitatory synaptic input (Woodward et al., 1969a,b; Gruol and Franklin, 1987;Mason et al., 1990; Gruol et al., 1991; Altman and Bayer, 1997), but the potential role of this activity in Ca²⁺ signaling has not been investigated.

In the current study we investigated Ca²⁺dynamics in early developing Purkinje neurons acutely isolated from the cerebella of postnatal rats. Results show that immature Purkinje neurons exhibit spontaneous somatic Ca²⁺oscillations that are generated by endogenous electrical activity, amplified via release of Ca²⁺ from intracellular stores, and transmitted to the nucleus. The Ca²⁺ oscillations occur via a signaling pathway involving L-type VGCCs rather than P-type VGCCs, which mediate the electrically evoked Ca²⁺ signals of Purkinje neuron dendrites (Usowicz et al., 1992). In addition to playing an important role in regulating Ca²⁺ dynamics of immature Purkinje neurons, this signaling pathway may contribute to the early developmental program by influencing downstream events such as gene expression.

MATERIALS AND METHODS

Preparation of acutely dissociated cerebellar cells.Purkinje neurons were acutely isolated from postnatal rats according to the methods of Surmeier et al. (1991). Rat pups between the ages of postnatal day 4 (P4) and P7 were rapidly decapitated, and the cerebellum was immediately removed and kept moistened with ice-cold, oxygenated artificial CSF (in mm: 130 NaCl, 3.5 KCl, 1.25 NaH₂PO₄, 24 NaHCO₃, 10 glucose, 0.2 CaCl₂, and 5.0 MgSO₄). The meninges were dissected away, 350–450 μm sagittal slices were cut manually from the vermis, and the white matter was removed under a dissecting microscope. Slices were minced in isethionate buffer (in mm: 140 sodium isethionate, 2 KCl, 4 MgCl₂, 23 glucose, and 15 HEPES, pH adjusted to 7.4 with NaOH) supplemented with 5 μm glutathione, 1 mmN-ω-nitro-l-arginine, and 10 mm kynurenic acid at room temperature (21–23°C). The tissue was incubated 45 min in Earl's Balanced Salt Solution (supplemented as above) in a spinner flask in the presence of 5% CO₂ and 95% O₂ at room temperature. The tissue was then incubated for 30 min with 0.1% papain (Sigma, St. Louis, MO) in HBSS (supplemented as above) in an oxygenated spinner flask maintained at 37°C. The tissue was washed three times in isethionate buffer (see composition above) and gently triturated through a size series of three fire-polished Pasteur pipettes to dissociate the tissue and isolate the cells. Cells were plated on Lab-Tek chamber slides (Nunc, Naperville, IL) for immunological studies or glass coverslips (Fisher Scientific, Houston, TX) for physiological recordings. Experiments were performed in physiological saline (in mm: 140 NaCl, 3.5 KCl, 0.4 KH₂PO₄, 1.25 Na₂HPO₄, 2.2 CaCl₂, 2 MgSO₄, 10 glucose, and 10 HEPES, pH-adjusted to 7.3 with NaOH) or Mg²⁺-free saline supplemented with 5 μm glycine depending on the experiment.

Measurement of Ca²⁺levels. Dissociated cells were centrifuged to pellet them (5 min at 6000 rpm) and loaded with the fluorescent Ca²⁺ indicator fura-2 AM (Molecular Probes, Eugene, OR) by resuspending the cells in physiological saline (see composition above) containing 3 μm fura-2 and 0.02% pluronic F-127 (Molecular Probes) for 30 min in the dark at room temperature (21–23°C). Cells were then centrifuged and resuspended in physiological saline. The cell suspension was plated on glass coverslips that had been precoated overnight at 4°C with Thy1 antibody (100 μg/ml; Sigma), washed once with PBS, and incubated in physiological saline during the attachment period. Poly-l-lysine (Sigma) and Matrigel (Collaborative Biomedical Products, Bedford, MA) were also tested as substrates, but Thy1 antibody-coated glass coverslips were found to most effectively immobilize the Purkinje neurons (Baptista et al., 1994). Results were not dependent on the relative plating density or substrate used. Coverslips with fura-2-loaded cells were used for experiments 1 hr after plating.

Standard microscopic fura-2 digital imaging techniques were used as described previously (Gruol et al., 1996). Neurons were viewed with a Nikon Eclipse TE300 inverted fluorescent microscope and light intensity monitored with a silicon-intensifier target camera (SITCAM) (MTI VE1000SIT; Dage, Michigan City, IN) for capturing live video images. Data were collected at 2–5 sec intervals (four frames per wavelength were averaged for each time point) over a total time of 90 sec to 10 min depending on the experiment. Ratio images of individual neurons in each field were formed by pixel-by-pixel division of the averaged images collected at 340 and 380 nm (340/380 nm). Real-time digitized display, image acquisition, and calculation of intracellular Ca²⁺ levels were made with Microcomputer Imaging Device imaging software (Imaging Research, St. Catharines, Ontario). Intracellular Ca²⁺levels were estimated using the following formula: [Ca²⁺]_i =K_d(R −R_min)/(R_max−R)*F_o/F_s, where R is the ratio value,R_min is the ratio for a Ca²⁺ free solution,R_max is the ratio for a saturated Ca²⁺ solution,K_d is 135 (the dissociation constant for fura-2), F_o is the intensity of a Ca²⁺-free solution at 380 nm, and F_s is the intensity of a saturated Ca²⁺ solution at 380 nm.R_min andR_max values were determined from standardized solutions of known Ca²⁺concentration (Molecular Probes calibration kit C-3009) containing fura salt (100 μm). Calculated levels of intracellular Ca²⁺ obtained by this method are considered approximations and are used primarily to provide information about the relative level of intracellular Ca²⁺ under a variety of experimental conditions. Typical R_max,R_min, andF_o/F_svalues were 0.61, 2.85, and 2.5, respectively. The low level of background fluorescence and adjustment of the black level of the SITCAM eliminated the need for background subtraction methods. Measurements were made of baseline Ca²⁺ levels, the maximum amplitude of the Ca²⁺oscillations relative to baseline, and the average Ca²⁺ level above baseline (see Fig.2B). Baseline for a neuron was defined as the lowest Ca²⁺ level observed in the Ca²⁺ recording.

Fig. 2.

Early postnatal Purkinje neurons exhibit spontaneous oscillations of intracellular Ca²⁺levels. A, Ca²⁺ recordings from three representative Purkinje neurons; spontaneous Ca²⁺oscillations persisted throughout the entire recording period. A granule neuron is included in each example for comparison; Ca²⁺ oscillations were small or absent in granule neurons (see also Figs. 4, 6, and 7). Collections were performed at 5 sec intervals for 10 min. The animal age is indicated in thebottom right-hand corner. B, Quantification of Ca²⁺ oscillations. Measurements were made using Axograph software (Axon Instruments). The baseline Ca²⁺ level was defined as the lowest Ca²⁺ level measured in a neuron. The highest Ca²⁺ level relative to baseline was defined as the maximum amplitude. Average amplitude was measured as the average Ca²⁺ level above baseline calculated over the recording period.

Drugs. Experiments were performed in the presence of 5 μm2,3-dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide disodium (Tocris Cookson, Langford, Bristol, UK), 100 μmpicrotoxin (Sigma), and 1 μm CGP55845A (Novartis, Basel, Switzerland) in the bath saline to block AMPA, GABA_A, and GABA_B receptors, respectively. Test agents were diluted in bath saline and introduced by bath addition or bath exchange to achieve the desired concentrations. A variety of drugs were tested including the P-type VGCC antagonist ω-agatoxin IVA (0.1 μm; Calbiochem, La Jolla, CA), the N-type VGCC antagonist ω-conotoxin GVIA (1 μm; Calbiochem), the L-type VGCC antagonist nimodipine (1 μm; Research Biochemicals International, Natick, MA), the competitive NMDA receptor antagonist APV (50 μm; Tocris), the NMDA receptor open channel blocker 5R,10S)-(+)-5-methyl-10,11-dihydro-5H-dibenzo[a,d]-cyclohepten-5,10-imine/dizocilpine (MK801; 10 μm; Tocris), the sodium channel inhibitor tetrodotoxin (TTX; 0.5 μm; Calbiochem), and the dihydropyridine receptor agonistS(−)-BayK-8644 (1 μm; Research Biochemicals International). Stock solutions of nimodipine and BayK-8644 were prepared in DMSO. DMSO by itself did not affect spontaneous Ca²⁺ oscillations (data not shown). In most studies at least three different cell preparations were tested to determine the effect of a given drug. n is the number of cells studied. Results are presented as mean ± SEM.

Electrophysiological techniques. Electrophysiological recordings were performed in combination with recordings of intracellular Ca²⁺ or in parallel with Ca²⁺-imaging studies (see above). Current-clamp recordings were made from the somatic region of acutely isolated Purkinje neurons using the nystatin, perforated-patch method of whole-cell recording and the Axopatch-1C amplifier (Axon Instruments, Foster City, CA) following previously described methods (Netzeband et al., 1997). Briefly, patch pipettes (2–4 MΩ) were filled with a pipette solution containing 6 mm NaCl, 154 mm K⁺-gluconate, 10 mm HEPES, and 200 μg/ml nystatin (Sigma), pH adjusted to 7.3 with KOH. A stock solution of nystatin (50 mg/ml DMSO) was prepared fresh for each experiment and diluted in the pipette solution. Recordings were made at the resting membrane potential of the cell and monitored on a polygraph (Gould, West Warwick, RI) and oscilloscope. For better resolution of fast events, selected data were recorded on FM tape (Store 4DS; Racal Recorders, Irvine, CA) for playback at reduced tape speed onto the polygraph recorder. Voltage responses elicited by a standardized series of hyperpolarizing and depolarizing current pulses (500 msec duration) were used to assess membrane properties. pClamp software and the Labmaster interface (Axon Instruments) were used for acquisition and analysis of the current-evoked membrane responses. All recordings were made at room temperature.

Immunohistochemistry. The acutely dissociated cerebellar cells were immunostained according to standard methods (Netzeband et al., 1997). Cells were fixed for 15 min in 4% paraformaldehyde (or 4% paraformaldehyde and 0.1% glutaraldehyde for GABA immunostaining) and washed three times (10 min) with 5% sucrose in PBS and once with PBS (10 min). The cells were then incubated 30 min in PBS containing 0.05% Triton X-100, followed by three washes (10 min) in PBS. Primary antibodies were diluted in PBS with 0.5 mg BSA/ml and 1% serum and incubated overnight at 4°C. The following antibodies were used: mouse monoclonal anti-calbindin (1:3000; Sigma); mouse monoclonal anti-glial fibrillary acidic protein (GFAP) (1:2000; Boehringer Mannheim, Indianapolis, IN); rabbit polyclonal anti-GABA (1:500; Incstar, Stillwater, MN); rabbit polyclonal anti-α_1Asubunit of P/Q-type VGCC (1:1000; Alomone Labs, Jerusalem, Israel); rabbit polyclonal anti-α_1B subunit of N-type VGCC (1:500; Alomone Labs); and rabbit polyclonal anti-rat α_1C subunit of L-type VGCC (1:1000; Alomone Labs). Vectastain kits (Vector Laboratories, Burlingame, CA) were used for the secondary antibodies and the peroxidase reaction.

RESULTS

Developing Purkinje neurons show spontaneous Ca²⁺ oscillations

The experiments focused on the postnatal age interval between P4 and P7 (Fig. 1A–C), known to represent a small but significant time window of Purkinje neuron development that precedes the main period of dendritic elaboration and synapse formation (Altman, 1972). Purkinje neurons in cerebellar cell preparations obtained by acute dissociation of P4 to P7 rat cerebella were identified for study by morphological criteria established by immunostaining cerebellar cell preparations with an antibody to the Ca²⁺-binding protein calbindin (Fig. 1E–G), a marker for Purkinje neurons (Enderlin et al., 1987). Morphologically identified Purkinje neurons also immunostained with an antibody to GABA (Fig.1H), the neurotransmitter used by Purkinje neurons, but did not immunostain with an antibody to GFAP (Fig.1D), a marker for glial cells. Granule neurons, cerebellar interneurons that use glutamate as a transmitter, were identified by their small, uniform size and lack of immunoreactivity for antibodies to calbindin, GABA, and GFAP (Fig. 1).

Fig. 1.

Morphological and immunological characteristics of acutely isolated immature Purkinje neurons. A–C, Live Purkinje neurons photographed under high power (Nomarski optics) to illustrate features of early developing Purkinje neurons at P4 (A) to P7 (B, C).A, Perisomatic processes (arrow) were characteristic of Purkinje neurons early in development (P4). Theasterisk indicates granule neurons that were present in all microscopic fields. B, A short axon (arrow) and prominent lateralized nucleus (arrowhead) were generally evident at P7.C, The initial outgrowth of the apical dendritic tree (arrow) was occasionally seen at P7. It should be noted, however, that a variety of morphological forms could be found within any given preparation because the cerebellar folia mature at different rates. D, Glial cell (arrowhead) immunostained with an antibody to GFAP. Arrow points to an unstained Purkinje neuron. As noted for acutely prepared P7 to P14 cerebellum (Hockberger et al., 1994), we found that some glial cells resembled Purkinje neurons, especially at the earliest stage studied. However, preliminary removal of the white matter during the dissection significantly reduced the number of glial cells. E–G, Purkinje neurons immunostained with an antibody to calbindin, a specific marker for Purkinje neurons (Enderlin et al., 1987).Arrows in E–G correspond to the structures labeled in A–C, respectively. Granule neurons (E, asterisk) did not immunostain for calbindin. H, Purkinje neuron (arrow) immunostained with an antibody to GABA, the neurotransmitter used by Purkinje neurons. Granule neurons, which use glutamate as a neurotransmitter, were unstained. All immunostaining was photographed using Hoffman optics. Scale bars: C,H,15 μm (refer to A–C and D–H, respectively).

Measurement of intracellular Ca²⁺ levels in the immature Purkinje neurons revealed that they exhibit sustained spontaneous oscillations of intracellular Ca²⁺ (Fig.2) and that such activity is present as early as 4 d after birth, the earliest age examined. To obtain an overview of the patterns of spontaneous Ca²⁺ activity, collections of 5 or 10 min at 5 sec intervals were performed on oscillating Purkinje neurons isolated from P4 to P7 rat pups. Patterns of Ca²⁺ oscillation were fairly consistent for a given Purkinje neuron but were highly variable among Purkinje neurons at all ages studied. Figure 2 illustrates three typical patterns of Ca²⁺ oscillations observed in the Purkinje neurons. The average frequency of Ca²⁺ oscillations was measured as the number of peaks occurring during the recording period. The average frequency for a population of P4 to P7 neurons studied was 2.6 ± 0.1 oscillations/min (n = 24), a rate significantly higher than, for example, that measured for spontaneous Ca²⁺ spikes (2.4 ± 0.3 spikes/hr) and waves (2.1 ± 0.2 waves/hr) in embryonic Xenopusspinal neurons (Gu et al., 1994) or for spontaneous Ca²⁺ fluctuations in the leading process of migrating cerebellar granule cells (13 ± 3 fluctuations/hr;Komuro and Rakic, 1996).

Statistics on various measures of the Ca²⁺oscillations at different ages were obtained from collections of 1–5 min at 3–5 sec intervals. Measurements were made of baseline Ca²⁺ levels, the maximum amplitude of the Ca²⁺ oscillations relative to baseline, and the average Ca²⁺ level above baseline (Fig. 2B). The minimum amplitude (relative to baseline) for defining an oscillation was 20 nm. Approximately two-thirds (219 of 351 cells) of the P4 to P7 Purkinje neurons studied displayed oscillations in Ca²⁺ levels. In oscillating cells, age-related differences were observed in all measurements with the most prominent changes occurring between P4 and P5, when a significant increase in baseline Ca²⁺ and decrease in maximum amplitude and average Ca²⁺ were observed (Table 1). In nonoscillating cells, measurements were made of baseline Ca²⁺ levels, which showed no significant change between P4 and P7 (Table 1).

Table 1.

Descriptive characteristics of spontaneous Ca²⁺transients observed in Purkinje neurons

Age	Non-oscillating cells		Oscillating cells
	n	Baseline Ca2+ (nm)	n	Baseline Ca2+ (nm)	Maximum amplitude (nm)	Average Ca2+ (nm)
P4	21	54  ± 6	82	48  ± 3	85  ± 8	29  ± 3
P5	33	61  ± 2	79	62  ± 2*	47  ± 5*	16  ± 2*
P6	50	54  ± 3	95	61  ± 3*	49  ± 6*	20  ± 2*
P7	28	61  ± 3	94	68  ± 2*,1-160	56  ± 5*	22  ± 2*,1-160

Values are mean ± SEM. Oscillating cells were defined as having Ca²⁺ transients with an amplitude >20 nm. Baseline Ca²⁺ was defined as the lowest Ca²⁺ level during the recording period. Maximum amplitude indicates the peak amplitude of the Ca²⁺ transients relative to baseline. Average Ca²⁺ refers to the mean Ca²⁺ level above baseline during the recording period. Data were based on measurement of Ca²⁺ levels during collections of 1–5 min at 3–5 sec intervals.

^*Significant difference compared with the corresponding value at P4 (p < 0.05; ANOVA; post hocanalysis by Fisher's protected least significant difference test).

F1-160: Significant difference compared with the corresponding value at P5 (p < 0.05; ANOVA; post hoc analysis by Fisher's protected least significant difference test).

Nuclear Ca²⁺ changes

In the above studies, the intracellular Ca²⁺ levels represent an average of levels measured over the entire somata of the neurons. However, in a subpopulation of the neurons studied (n = 20) the nuclear and cytoplasmic regions were clearly distinguishable, enabling measurements of these two subcellular regions. This analysis indicated that a pattern of Ca²⁺ oscillations similar to that observed in the cytoplasmic region occurred in the nuclear region (Fig. 3). Baseline Ca²⁺ levels were typically lower (24 ± 4%) in the nucleus compared with the cytoplasm. However, the maximum amplitude of the Ca²⁺ oscillations in the nucleus and cytoplasm were comparable. Mean values for the 20 neurons studied were 103 ± 20 nm and 119 ± 28 nm for the maximum amplitude of the nuclear and cytoplasmic Ca²⁺oscillations, respectively.

Fig. 3.

Cytoplasmic Ca²⁺ oscillations are communicated to the nucleus. A, Ca²⁺ dynamics in the Purkinje neuron soma. Thetoprow shows digital ratio (380/340 nm) images at selected time points during the Ca²⁺recording shown in B; a paler blueindicates a higher ratio value. The bottomrow shows the corresponding digital 380 nm wavelength;warmer colors correspond to lower Ca²⁺ levels, and cooler colors to higher Ca²⁺ levels. Numbers below the images inA correlate with numbers above collection time points in the Ca²⁺ trace in B. B,Ca²⁺ oscillations occurring simultaneously in the cytoplasm and nucleus of a P7 Purkinje neuron. Images were collected at 3 sec intervals for 90 sec. The imaging software was used to outline regions of similar size in the nucleus (red; seeinset) and cytoplasm (blue) of the digital Ca²⁺ images.

Generation of Ca²⁺ oscillations requires spontaneous electrical activity

Mature Purkinje neurons in vivo exhibit spontaneous electrical activity generated both by synaptic input and endogenous voltage-sensitive ionic mechanisms. This activity is expressed early in the developmental program (Woodward et al., 1969a,b; Gruol and Franklin, 1987). The acutely isolated Purkinje neurons are dissociated from their synaptic connections, making it unlikely that synaptic components contribute to the generation of the Ca²⁺ oscillations observed in the current study. However, several studies have shown that the endogenously generated component of the electrical activity is retained in acutely isolated Purkinje neurons (Nam and Hockberger, 1997; Raman and Bean, 1997, 1999). To determine whether the endogenously generated electrical activity plays a role in the generation of the Ca²⁺ oscillations, we tested the effect of TTX, which blocks the endogenously generated electrical activity of Purkinje neurons (Gruol and Franklin, 1987; Nam and Hockberger, 1997), on the Ca²⁺ oscillations. TTX (1 μm) completely abolished the Ca²⁺ oscillations, indicating an underlying involvement of Na⁺-dependent electrical activity in the generation of the Ca²⁺ transients (Fig.4). TTX also blocked all spike activity recorded under current clamp (data not shown).

Fig. 4.

Spontaneous Ca²⁺oscillations require Na⁺ channel activity. Representative example of the effect of the Na⁺channel blocker TTX (1 μm) on the spontaneous Ca²⁺ oscillations of an acutely isolated Purkinje neuron. As shown in this cell, TTX completely blocks the Ca²⁺ oscillations.

To examine the relationship between the spontaneous electrical activity and the Ca²⁺ oscillations, whole-cell (perforated patch), current-clamp recordings combined with Ca²⁺ imaging were performed on acutely isolated Purkinje neurons. In most cases, Purkinje neurons showing spontaneous Ca²⁺ oscillations were selected for study. The majority of Purkinje neurons studied (n = 11 of 13) exhibited sustained spontaneous electrical activity composed of single and doublet spikes with intervening quiescent periods of varying length, patterns similar to that reported by others for acutely isolated Purkinje neurons of similar postnatal age (Nam and Hockberger, 1997) and immature Purkinje neurons in vivo (Woodward et al., 1969a,b) and in culture (Gruol and Franklin, 1987). Spontaneous Ca²⁺ oscillations correlated closely with the spontaneous electrical firing pattern in the combined recordings (Fig. 5A; see also Fig. 7). Moreover, manipulation of the endogenous electrical activity by applying hyperpolarizing or depolarizing current (Fig. 5B) produced a corresponding alteration in the Ca²⁺ oscillations, whereas in nonoscillating neurons depolarization of the membrane potential to elicit electrical activity produced Ca²⁺signals similar to the spontaneous Ca²⁺oscillations. Sustained electrical activity elevated resting Ca²⁺ levels because of summation of Ca²⁺ signals (Fig. 5B).

Fig. 5.

Ca²⁺ dynamics correlate closely with patterns of electrical activity in immature Purkinje neurons.A, Combined recording of the spontaneous Ca²⁺ oscillations (top trace) and the electrical activity (middle trace) in a single Purkinje neuron recorded at rest (i.e., without applied current; bottom trace). Note that the Ca²⁺ oscillations correlate with the appearance of spontaneous action potentials. Thearrows indicate a burst event in the electrical activity and the corresponding increase of intracellular Ca²⁺. Electrical activity was recorded under current clamp. B, Combined Ca²⁺imaging/current-clamp recording in another Purkinje neuron. This cell exhibited little change in electrical activity or Ca²⁺ levels at rest. However, action potentials and corresponding Ca²⁺ signals could be induced by injecting depolarizing current (5 pA), first in two 2 sec pulses and then in a 50 sec pulse. The elevated resting Ca²⁺level during the 50 msec current injection reflects the summation of high-frequency Ca²⁺ peaks. For both cells, thedashed lines indicate intracellular Ca²⁺ levels of 60 nm and a membrane potential of −60 mV.

Fig. 7.

L-type VGCCs mediate Ca²⁺influx that generates intracellular Ca²⁺oscillations in immature Purkinje neurons. A, Representative example of the effect of nimodipine (1 μm; added to the bath saline) to nearly block Ca²⁺oscillations in an acutely isolated Purkinje neuron. B, Incubation of the cells with BayK-8644 (1 μm) dramatically enhanced the Ca²⁺ oscillations of oscillating Purkinje neurons. C, Combined Ca²⁺-imaging, current-clamp experiment showing simultaneous inhibition of spontaneous Ca²⁺oscillations and spontaneous electrical activity by nimodipine (1 μm). Nimodipine was applied to the bath saline at thearrow and was present throughout the remainder of the recording period. D, Combined Ca²⁺imaging, current-clamp experiment showing the simultaneous effect of BayK-8644 (1 μm) on spontaneous Ca²⁺oscillations and spontaneous electrical activity. BayK-8644 was applied to the bath saline at the arrow and was present throughout the remainder of the recording period. Recordings inC and D are at the resting membrane potential of the cells; dashedlinesdenote Ca²⁺ levels and membrane potential as indicated to the left of the lines.Insets in C and D show records of the spike activity at an expanded time scale.

Ca²⁺ oscillations require Ca²⁺ entry mediated by L-type VGCCs

The correlation of electrical activity with the Ca²⁺ oscillations suggests that Ca²⁺ influx plays a primary role in the generation of the Ca²⁺ signals. To test for a dependence on Ca²⁺ influx, extracellular Ca²⁺ was reduced from 1.5 mm to 20 μm with EGTA (1 mm) in the bath saline. The Ca²⁺ oscillations were completely blocked by the reduction of extracellular Ca²⁺ (n = 3 from two preparations; Fig. 6A), indicating that influx of Ca²⁺ was required for the oscillations.

Fig. 6.

Spontaneous Ca²⁺ oscillations in developing Purkinje neurons require Ca²⁺ influx through VGCCs. A, The Ca²⁺ traces show a typical example of the effect of EGTA (1 mm) to block the Ca²⁺ oscillations in an acutely isolated Purkinje neuron. For these studies, EGTA was applied by bath exchange in physiological saline containing low Ca²⁺ (20 μm). B, Representative example of the Ca²⁺ oscillations in a Purkinje neuron before and after bath application of the P/Q-type VGCC antagonist ω-agatoxin IVA (0.1 μm). ω-Agatoxin IVA had little effect on the amplitude or frequency of the Ca²⁺ oscillations.C, ω-Conotoxin GVIA (1 μm) was also ineffective at blocking the Ca²⁺ oscillations as shown in this cell.

The two major avenues for Ca²⁺ entry into the neuron are through ligand-gated channels permeable to Ca²⁺ or through VGCCs. We first determined whether Ca²⁺ was entering through NMDA receptors because at this stage of development Purkinje neurons express NMDA receptors (Garthwaite et al., 1987; Krupa and Crepel, 1989;Rosenmund et al., 1992), which may have been activated by glutamate released from granule neurons that are present in the cerebellar cell preparation (Fig. 1). However, the oscillations were not affected by blocking NMDA receptors either with the competitive inhibitor APV (50 μm; n = 2) or with the open channel blocker MK801 (10 μm; n = 3; data not shown). AMPA, GABA_A and GABA_B receptors were also not involved in Ca²⁺ entry because antagonists for these receptors were routinely included in the bath saline (see Materials and Methods).

We then tested inhibitors specific for the three principal classes of VGCCs (P/Q-, N-, and L-types) to determine whether the Ca²⁺ oscillations were dependent on Ca²⁺ influx through VGCCs. For these studies, measurements were made of the average Ca²⁺ above baseline and the maximum amplitude of the Ca²⁺ oscillations under control conditions and after addition of the antagonist to the bath saline. The baseline Ca²⁺ level for a cell was defined as the lowest Ca²⁺ level occurring during either the control or antagonist treatment period. Antagonists were considered to be effective if they reduced the average Ca²⁺ of a cell by >10%.

Other laboratories have shown that P-type VGCCs are the primary Ca²⁺ channels mediating Ca²⁺ influx in 2- to 3-week-old Purkinje neurons (Regan, 1991; Mintz et al., 1992). However, in our studies on the P4 to P7 Purkinje neurons, the P/Q type VGCC blocker ω-agatoxin IVA (0.1 μm) was relatively ineffective (Fig.6B), decreasing Ca²⁺levels in only 3 of 13 Purkinje neurons. Mean values (±SEM) for average Ca²⁺ in the three neurons sensitive to ω-agatoxin IVA were 69 ± 19 nm under control conditions and 38 ± 17 nm after addition of ω-agatoxin IVA to the recording saline, an ∼45% decrease. Mean values for the maximum amplitude of the Ca²⁺ oscillations were similar under the two conditions, 169 ± 32 nm under control conditions and 161 ± 36 nm in the presence of ω-agatoxin IVA (n = 3). Similarly, the N-type Ca²⁺ channel blocker ω-conotoxin GVIA (1 μm) decreased Ca²⁺levels in only 3 of 8 Purkinje neurons (Fig. 6C). Mean values for average Ca²⁺ in the three neurons were 27 ± 6 nm under control conditions and 18 ± 3 nm in the presence of ω-conotoxin GVIA, an ∼33% decrease. Mean values for the maximum amplitude of the Ca²⁺ oscillations in the three neurons were 102 ± 16 nm under control conditions and 62 ± 17 nm in the presence of ω-conotoxin GVIA (n = 3), a 39% decrease.

Surprisingly, the most consistent effects were observed with the L-type VGCC antagonist nimodipine, which was an effective antagonist of the Ca²⁺ oscillations in 27 of 29 neurons (Fig.7A,C). Nimodipine (1 μm) reduced both the average Ca²⁺ and the maximum amplitude of the Ca²⁺ oscillations in all 27 neurons tested (from 17 preparations between P4 and P7). In those 27 neurons, mean values for average Ca²⁺ were 40 ± 6 nm under control conditions and 16 ± 4 nm in the presence of nimodipine, a 60% decrease. Mean values for the maximum amplitude of the Ca²⁺ oscillations were 101 ± 14 nm under control conditions and 28 ± 4 nm in the presence of nimodipine (n = 27), a 72% decrease.

When the above data were separated by age, the sensitivity of the Ca²⁺ oscillations to nimodipine (as measured by the effects of nimodipine on average Ca²⁺) decreased between P4 and P6, suggesting a developmental regulation of either the activity or expression of the L-type Ca²⁺ channel. Mean values for average Ca²⁺ at P4 were 36 ± 7 nm (n = 9) under control conditions and 12 ± 3 nm in the presence of nimodipine, a 67% decrease, whereas at P6 mean values for average Ca²⁺ were 42 ± 13 nm (n = 8) under control conditions and 24 ± 12 nm in the presence of nimodipine, a 43% decrease. In combined electrophysiological and Ca²⁺-imaging experiments, electrical spiking continued in the presence of nimodipine, although Ca²⁺ oscillations were abolished within seconds of introducing the drug (Fig. 7C), confirming that although the Ca²⁺ patterns were directly determined by the electrical activity, the specific influx of Ca²⁺ mediated by the L-type VGCCs was fundamental and essential to generating the Ca²⁺ oscillations. Thus, L-type VGCCs appear to be the primary mediators of Ca²⁺influx generating the Ca²⁺ oscillations in the P4 to P7 Purkinje neurons.

Consistent with a dominant role for L-type VGCCs in the generation of the Ca²⁺ oscillations, the L-type channel-specific agonist BayK-8644 (1 μm) enhanced Ca²⁺ levels in six of eight cells tested (Fig. 7B,D). For the six cells, mean values for average Ca²⁺ were 46 ± 19 nm under control conditions and 135 ± 16 nm in the presence of BayK-8644, a 193% increase. Mean values for the maximum amplitude of the Ca²⁺ oscillations were 107 ± 32 under control conditions and 246 ± 86 nm in the presence of BayK-8644 (n = 6), a 130% increase. These results are consistent with the known action of BayK-8644 in prolonging the open time of L-type VGCCs (Hess et al., 1984). In combined electrophysiology/Ca²⁺imaging experiments, addition of BayK-8644 (1 μm) to the recording chamber changed the pattern of electrical activity from single-spike to burst events, which correlated with a large sustained increase in the Ca²⁺ resting level resulting from summation of the Ca²⁺ oscillations (Fig.7D).

Expression of VGCCs in developing Purkinje neurons

Because our pharmacological results showing a prominent role for L-type VGCCs in the early developing Purkinje neurons were somewhat unexpected, it was important to determine the expression of the L-, P/Q-, and N-type VGCCs at these early developmental stages. Immunostaining of acutely dissociated P5 cerebellar preparations with antibodies against specific subunits of the P-, N-, and L-type VGCCs showed that all three classes of Ca²⁺channels were expressed by P5 Purkinje neurons (Fig.8A). Somatic immunostaining for the P/Q-type channel α_1Asubunit was strong and uniform for a given neuron, but showed variation from one Purkinje neuron to another. P/Q-type staining extended into the developing apical dendrite when dendritic structure was present (Fig. 8B) (n = 10). Immunostaining for L-type Ca²⁺ channels was uniform over the soma, almost as intense as for the P/Q-type, and heterogeneous in intensity within a given neuronal population. In contrast with the P/Q-type antibody, the L-type antibody did not consistently stain the developing apical dendrite (Fig. 8B;n = 12), suggesting a predominately somatic role for L-type VGCCs in developing Purkinje neurons. Somatic immunostaining for the N-type VGCCs was much fainter but also uniform for a given neuron. Staining with the N-type antibody was too faint to clearly resolve the presence of dendritic staining.

Fig. 8.

P/Q-, N-, and L-type VGCCs are expressed by immature Purkinje neurons. A, Photomicrographs showing immunoreactivity of early developing Purkinje neurons (arrows) for P/Q- (top, left panel), L- (middle, left panel), and N-type (bottom, left panel) VGCCs. P/Q-, L-, and N-type VGCCs were immunostained with antibodies to the α_1A, α_1C, and α_1B subunits of Ca²⁺ channels, respectively. Immunoreactivity is indicated by the brown color and was blocked in the presence of the corresponding blocking peptide (1:1 ratio of the antibody and peptide by weight) as shown in the panels to the right. Allpanels show P5 neurons. B, Photomicrographs demonstrating the relative expression of P/Q-type (left panel) and L-type (right panel) VSCCs in the developing apical dendrites (arrows) compared with the somata of immature Purkinje neurons. The dendrites typically expressed P-type VGCCs with relatively low levels of immunoreactivity for L-type VGCCs. Bothpanels show P7 neurons. Scale bars, 15 μm.

Amplification of the Ca²⁺ signal by release of Ca²⁺ from intracellular Ca²⁺stores

Ca²⁺ influx through VGCCs is known to be amplified by release of Ca²⁺ from intracellular stores in Purkinje neurons (Llano et al., 1994; Kano et al., 1995a; Gruol et al., 1996). To test for an involvement of intracellular Ca²⁺ stores in the Ca²⁺ oscillations of the immature Purkinje neurons, the neurons were incubated for 20 min in 10 μmdantrolene, an agent known to block release of Ca²⁺ from RyR-gated intracellular Ca²⁺ stores. Dantrolene decreased the amplitude of the Ca²⁺ oscillations in 8 of 10 cells (from three preparations; Fig.9). In those eight neurons, mean values for average Ca²⁺ were 52 ± 12 nm under control conditions and 26 ± 10 nm in the presence of dantrolene, a 50% decrease. Mean values for the maximum amplitude of the Ca²⁺ oscillations were 91 ± 18 nm under control conditions and 39 ± 13 nm in the presence of dantrolene (n = 8), a 57% decrease. These results suggested that a large component of the Ca²⁺ transients originated from intracellular Ca²⁺ stores. To verify that point, we tested ryanodine, which at relatively high concentrations blocks Ca²⁺ release from RyR-gated intracellular Ca²⁺ stores (Ehrlich et al., 1994). Ryanodine (100 μm; 20 min incubation) depressed the Ca²⁺ oscillations in 20 of 30 cells (from two preparations). In the 20 cells, mean values for average Ca²⁺ were 28 ± 4 nm under control conditions and 10 ± 2 nm in the presence of ryanodine, a 64% decrease. Mean values for the maximum amplitude of the Ca²⁺ oscillations were 68 ± 9 nm under control conditions and 30 ± 9 nm in the presence of ryanodine (n = 20), a 56% decrease.

Fig. 9.

Ca²⁺ signals of extracellular origin are amplified by release of Ca²⁺ from intracellular Ca²⁺ stores. A, Representative examples of the Ca²⁺ oscillations observed in an immature Purkinje neuron at time 0 (0′) and 20 min (20′) later. Under control conditions there was little change in the amplitude or frequency of the Ca²⁺ oscillations over time. B, Typical Ca²⁺ traces showing the effect of the RyR antagonist dantrolene (10 μm; 20 min incubation) to depress Ca²⁺ oscillations in an another Purkinje neuron. C, Current-clamp recording of the spontaneous electrical activity in another Purkinje neuron showing the persistence of electrical spiking after the 20 min incubation with dantrolene (10 μm). Resting membrane potential (−64 mV) is indicated by the dashed line. Vm, Membrane voltage.

DISCUSSION

Results from the current study show that spontaneous oscillations in intracellular Ca²⁺ are a prominent feature of early developing Purkinje neurons and that the Ca²⁺ oscillations are elicited by endogenously generated electrical activity that is expressed by this neuronal type (Woodward et al., 1969a,b; Crepel, 1972; Gruol and Franklin, 1987; Gruol et al., 1991; Nam and Hockberger, 1997; Raman and Bean, 1997, 1999). The Ca²⁺ oscillations are initiated by influx of Ca²⁺ through L-type VGCCs, which is then amplified by release of Ca²⁺ from RyR-gated intracellular Ca²⁺ stores and transmitted to the nucleus, a result that may have implications for gene regulation during neuronal development. These observations made with early developing Purkinje neurons contrast with the known predominant role of P-type VGCCs in Ca²⁺ signaling in the dendritic region of mature Purkinje neurons (Hillman et al., 1991; Usowicz et al., 1992).

Ca²⁺ oscillations have been reported to occur spontaneously in several neuronal types (Sorimachi et al., 1990;Yuste et al., 1992; Gu et al., 1994; Gomez et al., 1995; Gu and Spitzer, 1995; Wong et al., 1995; Komuro and Rakic, 1996; Owens and Kriegstein, 1998; Gomez and Spitzer, 1999), and in some cases have been shown to play a role in the developmental process. For example, developmental signals for neurotransmitter expression and growth cone migration in embryonic Xenopus spinal neurons are encoded in patterns of spontaneous intracellular Ca²⁺transients (Gu et al., 1994; Gu and Spitzer, 1995; Gomez and Spitzer, 1999). Spontaneous Ca²⁺ oscillations also play a critical role in regulating the advancement of the leading process of migrating cerebellar granule neurons (Komuro and Rakic, 1996), as well as, of chick retinal ganglion cell growth cones (Gomez et al., 1995). In these neuronal types, the trigger for spontaneous Ca²⁺ oscillations is still under investigation. However, our combined Ca²⁺imaging and electrophysiological recordings show that the Ca²⁺ oscillations observed in the early developing Purkinje neurons are generated by spontaneous electrical activity that is endogenously generated.

The ability of immature Purkinje neurons to endogenously generate electrical activity both in the immature and mature states has been documented for Purkinje neurons in acutely isolated cerebellar cell preparations (Nam and Hockberger 1997; Raman and Bean, 1997, 1999), in cerebellar culture preparations (Gruol and Franklin, 1987; Gruol et al., 1991), and in vivo as early as 2 d postnatal (Woodward et al., 1969a,b; Crepel, 1972). Ionic models involving “resurgent” sodium currents (Raman and Bean, 1997) or sodium “window currents” (Nam and Hockberger, 1997) have been proposed to account for this component of the electrical activity of Purkinje neurons. Synaptic input also contributes to the spontaneous activity of mature and developing Purkinje neurons in vivo and in culture and may play a role in the Ca²⁺dynamics of early developing Purkinje neurons. However, this component could not be examined in the current study because the synaptic connections were removed during the isolation procedure.

Our results show that the patterns of spontaneous electrical activity and Ca²⁺ oscillations are tightly linked in the early developing Purkinje neurons and that clusters of action potentials produce peaks of Ca²⁺. Moreover, a higher frequency of action potentials corresponds to a Ca²⁺ response of greater amplitude. Thus, the level of electrical activity of early developing Purkinje plays a critical role in defining the level of intracellular Ca²⁺. Different patterns of spontaneous electrical activity were present within a population of the acutely isolated Purkinje neurons at a single age, presumably resulting from ongoing developmental processes. The pattern of spontaneous electrical activity in Purkinje neurons is known to change with the developmental program (Woodward et al., 1969a,b; Gruol and Franklin, 1987). Therefore, it is likely that the patterns of Ca²⁺ oscillations produced by the spontaneous electrical activity will also change during development.

Pharmacological analysis showed that the Ca²⁺ oscillations in the developing Purkinje neurons are generated by the entry of extracellular Ca²⁺ through L-type VGCCs. This was unexpected because the P-type channel blockers ω-agatoxin IVA and funnel web spider toxin almost completely inhibit Ca²⁺ currents in acutely isolated Purkinje neurons from 2- to 3-week-old rats (Mintz et al., 1992) and in the dendrites of mature Purkinje neurons (Usowicz et al., 1992), respectively, indicating the functional importance of P/Q-type VGCCs at these stages. The dihydropyridine agonist BayK-8644 did not enhance Ca²⁺ influx in mature Purkinje neurons (Usowicz et al., 1992), implying that L-type Ca²⁺ channels do not play a prominent functional role in the mature dendritic arbor. However, the N-type Ca²⁺ channel blocker ω-conotoxin GVIA inhibited 6%, and the L-type Ca²⁺ channel blocker nitrendipine inhibited 8% of the high-threshold Ca²⁺ currents in acutely isolated Purkinje neurons from 1- to 3-week-old rats (Regan, 1991), indicating that both of these channel types contributed to Ca²⁺influx in developing Purkinje neurons. In our experiments, spontaneous Ca²⁺ oscillations in P4 to P7 Purkinje neurons were strongly inhibited by nimodipine as well as enhanced by the dihydropyridine agonist BayK-8644. Together, these results indicate that functional L-type VGCCs are present on Purkinje cells at an early age, and that these Ca²⁺ channels play an important role in Ca²⁺ signaling in early development.

Intracellular Ca²⁺ stores in the endoplasmic reticulum represent a critical intermediary step in many neuronal Ca²⁺ signaling pathways. Our results show that early in development, a somatic Ca²⁺ signaling pathway links Ca²⁺ entering through L-type VGCCs to Ca²⁺-mediated release of Ca²⁺ from the RyR-gated intracellular Ca²⁺ stores. A similar Ca²⁺ signaling pathway has been shown to generate spontaneous Ca²⁺ oscillations in the rat frontal cortex (Hayashi et al., 1997) and agonist-evoked Ca²⁺ signals in other neuronal types (Chavis et al., 1996; Ronde and Nichols, 1997). IP3R-gated intracellular Ca²⁺ stores could also contribute to the spontaneous Ca²⁺ signals observed in the current study, if ambient IP3 and Ca²⁺ levels were sufficient to activate this pathway. IP3-mediated release of Ca²⁺from the IP3R-gated intracellular Ca²⁺stores has been shown to be important in dendritic Ca²⁺ signaling in Purkinje neurons (Gruol et al., 1996; Netzeband et al., 1997; Finch and Augustine, 1998) in addition to Ca²⁺ release linked to RyRs (Llano et al., 1994; Gruol et al., 1996). Future studies will address this issue.

Immunostaining by other laboratories has shown that all three types of high-threshold VGCCs (P/Q-, L-, N-) are present on mature Purkinje neurons (Ahlijanian et al., 1990; Hillman et al., 1991; Westenbroek et al., 1992). Our results show that all three VGCC types are already expressed in the soma of immature Purkinje neurons by the first postnatal week. However, pharmacological analysis showed that P- and N-type Ca²⁺ channels do not play a prominent role in generating the Ca²⁺oscillations and that L-type VGCCs were the predominant Ca²⁺ channel involved. This selective involvement of L-type VGCCs in the Ca²⁺oscillations could reflect functional coupling between L-type VGCCs and RyR-gated intracellular Ca²⁺ stores, as has been shown to exist in other neuronal systems (Chavis et al., 1996). Such an association produces selective amplification of Ca²⁺ signals through L-type VGCCs. Alternatively, other features of the Ca²⁺signaling system such as the topographic distribution of VGCCs or intracellular Ca²⁺ stores may underlie the prominent role of L-type VGCCs. Further studies will be necessary to resolve this issue.

Activation of L-type Ca²⁺ channels has been directly linked to second messenger pathways that lead to increased expression of immediate early genes in several neuronal cell types (Murphy et al., 1991; Bading et al., 1993; Deisseroth et al., 1998; Rajadhyaksha et al., 1999). Our studies showing that electrically induced Ca²⁺ oscillations are transmitted to the nucleus suggest that such a pathway may be important for signaling in developing Purkinje neurons. Studies in other cell types have shown that cytosolic Ca²⁺signals can be translated into changes in nuclear Ca²⁺ and that nuclear Ca²⁺ signals can also be initiated in the nucleus (Kocsis et al., 1994; Gerasimenko et al., 1996). Both RyRs and IP3Rs are present on the nuclear membrane (Kocsis et al., 1994;Gerasimenko et al., 1996; Humbert et al., 1996), a membrane that is formed by an extension of the endoplasmic reticulum. Nuclear Ca²⁺ has been shown to be capable of modulating transcription (Hardingham et al., 1997) and dynamic changes in the nuclear Ca²⁺ pool, including Ca²⁺ oscillations and bursts of Ca²⁺ transients have been shown to act as specific switches for increasing gene expression (Gu and Spitzer, 1995;Dolmetsch et al., 1998; Li et al., 1998), potentially by activating Ca²⁺-sensitive intermediary components such as the enzyme CaM kinase II (DeKoninck and Schulman, 1998) or the transcription factor cAMP response element-binding protein (Hardingham et al., 1997). Thus, Ca²⁺ oscillations generated by the coordinated action of spontaneous electrical activity, L-type voltage-sensitive Ca²⁺ channels, and RyR-gated intracellular Ca²⁺ stores may represent a developmental signal linking electrical activity at the neuronal cell membrane with changes in gene expression in developing Purkinje neurons.