Ca_V3 T-type calcium channel isoforms differentially distribute to somatic and dendritic compartments in rat central neurons

Animal care

Male Sprague–Dawley rats (n = 38; Charles River, Canada) were maintained and prepared for histology according to guidelines approved by the Canadian Council for Animal Care.

Immunocytochemistry

All chemicals were obtained from Sigma (St. Louis, MO) unless otherwise indicated. General procedures for Ca_v3.x immunocytochemistry followed the procedures detailed in Molineux et al. (2006). Briefly, male rats (P18–P40) were anaesthetized with sodium pentobarbital (65 mg/kg; MTC Pharmaceuticals, Cambridge, ON, Canada) and perfused intracardially with 250 mL of 0.1 m phosphate-buffer (PB, pH 7.4), followed by 100 mL of 4% paraformaldehyde (Para, pH 7.4) at room temperature. Brains were postfixed in 4% Para at room temperature for 1 h and overnight at 4 °C. Brains used for glutamic acid decarboxylase (GAD) 67 staining were fixed with 4% paraformaldehyde and 0.3% gluteraldehyde, postfixed 1 h on ice, and overnight at 4 °C. All tissue was cut as free-floating 30–40 µm parasagittal or coronal sections by vibratome in ice-cold PB, and reacted in a working solution of 3% normal donkey, horse or goat serum (Jackson Immuno-Research, West Grove, PA), 0.1% Tween, and 1% dimethylsulphoxide (DMSO) in PB, with gentle agitation throughout all reactions. Primary antibodies were added to the working solution and reacted at 4 °C for 24–48 h and washed 3 × 15 min in working solution at room temperature. Secondary antibodies were incubated for at least 4 h at room temperature and washed in PB. Sections were then mounted on gel-coated slides and cover-slipped with antifade medium (90% glycerol/PBS/0.1%p-phenylenediamine; pH 10), sealed with nail polish and stored at −20 °C. Controls consisted of omitting the primary antibodies and preabsorbing primary antibodies with an excess of purified protein.

Antibodies directed to the Ca_v3.1 calcium channel were produced using the Ca_v3.1 I–II linker sequence (ELRKSLLPPLIIHTAATPMS), which corresponds to amino acids #1010–1027 (accession number # AF290212). The channel peptide was cross-linked to KLH, injected into rabbits, and the serum purified for use. The epitopes used for the Ca_v3.2 and 3.3 antibodies were produced using the Caulobactor expression system (Invitrogen) and corresponded to amino acids 1195–1273 (accession number # AF290213) and 1013–1115 (accession # AF290214) for the Ca_v3.2 and Ca_v3.3 calcium channel II–III linkers, respectively. Purified fusion proteins were injected into rabbits and the polyclonal antibodies were IgG purified before use. The specificity of each of the Ca_v3 antibodies has been thoroughly established (Molineux et al., 2006). All reactions were carried out using flourophore-conjugated secondary antibodies to avoid spurious labelling of endogenous biotin (McKay et al., 2004). Ca_v3.x antibodies were used at concentrations of Ca_v3.1 (1 : 250–1000), Ca_v3.2 (1 : 1000–10 000), Ca_v3.3 (1 : 2000–10 000) and revealed using a secondary Cy3-donkey anti-rabbit IgG (1 : 1000; Molecular Probes, Eugene, OR). Monoclonal or polyclonal antibodies used to counterlabel cell specific proteins were localized using secondary (1 : 1000) Alexa Fluor 488-goat anti-mouse IgG or -donkey anti-mouse IgG (Molecular Probes), or Alexa Fluor 555-goat anti-rabbit IgG or -goat anti-chicken IgG. Primary antibodies used for colabelling studies were monoclonal anti-microtubule-associated protein (MAP-2, 1 : 500), calbindin-28DK (1 : 1000, Swant, Bellinzona, CH), parvalbumin (1 : 25 000), or GAD67 (1 : 1000), Chemicon, Temecula CA), and polyclonal chicken anti-lamin A (1 : 1000, Cedarlane, Hornby ON Canada) and rabbit anti-annexin A3 (1 : 500, Abcam, Cambridge UK). Immunolabel was imaged using an Olympus FV300 BX50 confocal microscope (Carsen Group, Markham ON) equipped for isolating Cy3 and FITC-like fluorolabels. All image adjustments were confined to brightness/contrast and intensity levels using Photoshop 7 and Illustrator 9.

Western blotting

Male Sprague–Dawley rats were anaesthetized with sodium pentobarbital and the hippocampus removed and placed on ice. Cell compartment components from the hippocampus were isolated as to the protocol detailed in the Qiagen Qproteome Cell Compartment Kit (Qiagen, Mississauga ON). For Western blots 5 µg of whole protein from each cell compartment sample was prepared in 2× SDS loading buffer enriched with 0.1 m DTT, separated by a 10% SDS-PAGE gel, and transferred to PVDF membrane (Millipore, Billerica, MA). Membranes were blocked in PBS, 0.1% Tween 20 and 5% (w/v) powdered milk prior to overnight incubation with polyclonal antibodies against annexin A3 (1 : 1000), lamin (1 : 1000), Ca_v3.1 (1 : 5000), Ca_v3.2 (1 : 15 000) or Ca_v3.3 (1 : 15 000). Secondary antibodies used were horseradish peroxidase (HRP)-conjugated rabbit anti-chicken (1 : 5000, Abcam, Cambridge UK) or goat anti-rabbit IgG (1 : 5000) and detected using the Amersham's HCL detection kit as to the manufacturer's instructions.

Subcellular distribution of Ca_v3.1, Ca_v3.2 and Ca_v3.3 T-type calcium channels

The current study examined Ca_v3 T-type calcium channel isoform distribution in key regions of rat brain in which T-type currents have been suggested to have important roles in modulating cell output. We were particularly interested in Ca_v3 channel distribution between the soma and dendritic structures of individual cell types in neurons of the cortex, hippocampus, thalamus, and cerebellar input and output cells.

The polyclonal antibodies used to localize Ca_v3 channel isoforms have previously been shown to be specific to each of the rat Ca_v3 isoforms, and brain sections are negative if preabsorbed with purified Ca_v3 fusion protein or if the primary antibody is omitted (Molineux et al., 2006). We expect the antibodies to label all known alternatively spliced isoforms of Ca_v3.1, 3.2 and 3.3 given the conserved regions used to generate the antibodies. We observed consistent penetration of Ca_v3 antibodies to ∼10–15-µm depth under our conditions, and always compared the pattern of Ca_v3 label to a counterlabel against internal or structural proteins to improve identification of positive or negative cell structures. The use of cell-specific immunolabels (i.e. parvalbumin, calbindin-28K, GAD) allowed both analysis of Ca_v3 expression in identified cell classes and the detailed examination of dendritic structures (MAP-2).

Interpretation of immunolabel distributions

The localization of immunofluorescent signals in relation to functional ion channels in situ requires careful consideration of the limitations inherent to resolving low-level signals in light microscopy. For instance, identifying plasma membrane labelling of channels in neurons is more demanding than that encountered in expression systems, where overexpression of an ion channel can readily support visual identification of membrane-associated label. As often found in localization studies of native channels, a distinct membrane-associated Ca_v3 label that can be interpreted as channel insertion in the plasma membrane was detected in only specific cell types. It is important to recognize that a lack of apparent membrane labelling does not rule out insertion of these channels, as LVA T-type calcium currents are recorded in many cell types that did not present with distinct Ca_v3 immunofluorescence at the membrane level. Rather, in most cells Ca_v3 immunolabel was seen as a diffuse label in the cell interior, a pattern expected for channels localized to at least the Golgi complex or in the process of being transported to the membrane (referred to here as ‘cytoplasmic’, which includes organelles). In some cell types Ca_v3 immunolabel could also be detected in the region of the nucleus, a pattern not typically expected but readily visualized in particular cell types (i.e. Ca_v3.3 in dentate gyrus granule cells, Fig. 3J). We previously performed exhaustive control studies to establish that these antibodies are specific to Ca_v3 proteins in rat CNS, ruling out non-specific labelling (Molineux et al., 2006). There is ample precedence for the expression of ion channels in the nuclear envelope, including voltage-gated potassium, chloride and cation channels, and calcium-activated potassium channels, with additional expression of Na-K ATPase (Maruyama et al., 1995; Draguhn et al., 1997; Valenzuela et al., 1997; Franco-Obregon et al., 2000; Mazzanti et al., 2001; Garner, 2002; Marchenko et al., 2005). Moreover, a resting transmembrane potential can be recorded across the nuclear envelope at a level more negative than the plasma membrane potential, and extensive systems exist to regulate calcium concentration in the nucleoplasm (Draguhn et al., 1997; Mazzanti et al., 2001; Marchenko et al., 2005; Marchenko & Thomas, 2006). Therefore, to understand better the source of Ca_v3 channel immunolabel we first carried out a series of protein separation/Western blot and colocalization studies.

To test for the subcellular distribution of T-type calcium channel isoforms we separated plasma membrane, cytoplasmic, nuclear and cytoskeletal proteins from hippocampus using a commercially available protein separation kit. We first verified the specificity of the extraction by establishing that an antibody against annexin A3, a member of an extensive family of membrane and cytoplasmic proteins, labelled the appropriate fractions on a Western blot (Fig. 1A; Gerke et al., 2005). Conversely, the nuclear-specific protein lamin was detected on Western blots only in the nuclear protein fraction (Fig. 1A; Gruenbaum et al., 2005). A nuclear localization of lamin but exclusion of annexin A3 from the nucleus was verified further through double-label immunocytochemistry of CA3 hippocampal pyramidal cells (Fig. 1B). Finally, Western blot analysis of isolated protein fractions from hippocampus established that antibodies to the three Ca_v3 channel isoforms detected bands of the correct molecular weight in all three fractions of plasma membrane, cytoplasmic and nuclear proteins, but not cytoskeletal proteins (Fig. 1C). These data indicate that each of the Ca_v3 calcium channel isoforms are expressed at the plasma membrane at levels at least as high as that found in the cytoplasm. We cannot fully distinguish whether labelling in the nuclear region corresponds to channel expression in the nuclear envelope, internal nuclear membranes or even potential inclusion of membranes of the smooth ER, which is continuous with the outer membrane of the nuclear envelope. However, these data confirm that labelling in the nuclear region, when observed, is specific to Ca_v3 calcium channel isoforms.

It also became clear that Ca_v3 immunolabel could differ in terms of relative fluorescent intensity between nuclei or in its distribution over the soma-dendritic axis of individual cells. Differences in the intensity of immunolabel were also apparent between Ca_v3 isoform labels processed in parallel from the same animal. The extent to which this could reflect actual differences in membrane channel expression or density of membrane-inserted calcium channels is unknown. We did not attempt to quantify these differences. Ultimately the level of channel expression in the membrane needs to be verified with direct electrophysiological measurement. Yet, relative differences in the perceived fluorescent intensity that were apparent between Ca_v3 isoforms when compared at the same confocal laser intensity were prominent enough to warrant a descriptive comparison (Table 1). Comparisons between Ca_v3 intensity levels made here are thus necessarily subjective and expressed in terms of relative intensity levels, with no implied relation to protein level or density of membrane-associated and functional channel isoforms.

Ca_v3 neuronal expression

Ca_v3 channel isoforms were widely expressed, with most cell types labelling for one or more isoforms. As a general rule, Ca_v3.1 subtype immunolabel was primarily restricted to the somatic and/or proximal dendritic region and exhibited the lightest labelling of all three isoforms. Ca_v3.2 channel immunolabel was often more intense than Ca_v3.1, with a distribution at the soma and over at least the proximal to mid portions of dendrites of a number of cell classes. Ca_v3.3 channels typically exhibited the most intense immunolabel, and could be expressed at the soma as well as extended regions of dendrites. It is important to note that this represents only a general pattern to which several exceptions could be found. The extent of Ca_v3 isoform distribution between somata and dendrites was also highly cell-specific and is detailed for each of the regions indicated below. In each case we first summarize the available physiological data implicating T-type channels in cell output to help place the immunolabel results in context with known electrophysiological patterns.

Cortex

Pyramidal neurons within layer V of the neocortex express robust LVA calcium currents (Sayer et al., 1990; de la Pena & Geijo-Barrientos, 1996). Synaptic or intracellular stimulation triggers a low threshold calcium spike (LTS) that drives a burst of somatic Na⁺ spikes (Markram & Sakmann, 1994; Larkum & Zhu, 2002). Patch-clamp recordings have identified a dendritic origin for LTS that is greatest in the region of the distal dendritic tuft (Larkum & Zhu, 2002). LTS behaviour has been identified further in layer V interneurons where it enables dendritic-initiated signalling to activate the soma (Goldberg et al., 2004). Ca_v3.1 immunolabel has been reported in cortical pyramidal cell bodies and proximal apical dendrites (Craig et al., 1999).

We observed immunolabel for Ca_v3 channel isoforms in all cortical layers and regions, with the most prominent cell type corresponding to large diameter pyramidal cells. Colabelling with MAP-2 revealed labelling of most cells for Ca_v3 channels, although cells negative for a given Ca_v3 isoform could be detected in all cases. A distinct layer-specific banding pattern previously reported for mRNA expression (Kase et al., 1999; Talley et al., 1999) was not apparent in immunostaining, with a fairly homogeneous intensity across layers for each of the Ca_v3 isoforms. The greatest distinction between isoform labelling was in the extent of Ca_v3 distribution over the soma-dendritic axis of individual cells. Ca_v3.1 T-type channel immunolabel was largely restricted to the soma and proximal regions of pyramidal cells as a cytoplasmic label that was sufficient to delineate the boundaries of the plasma membrane (Fig. 2A). Ca_v3.2 T-type immunolabel was slightly less intense than Ca_v3.1, but included the soma as well as more extended (<200 µm) regions of the apical dendrites (Fig. 2B). Ca_v3.3 T-type channel immunolabel stood out in labelling somatic regions, and in particular, extended lengths of pyramidal cell apical dendrites that emanated primarily from cells in layer V (Fig. 2C). In many cases, Ca_v3.3 immunolabel highlighted elongated regions of individual apical dendrites that could be tracked visually over 600 µm from cells in layer V (Fig. 2C). The distal tuft of neocortical pyramidal cell apical dendrites also expressed immunolabel but often with less apparent intensity than the primary shaft, with the thin branching dendrites readily distinguished only for Ca_v3.3 immunolabel (Fig. 2C). By comparison, virtually no label for any of the T-type channel isoforms was found in basilar dendrites. An additional aspect of Ca_v3.1 distribution was a distinctive label associated with some, but not all, local GAD-positive interneurons. A higher power image in Fig. 2D reveals that many neurons colabelled for GAD and Ca_v3.1 at the soma were positioned adjacent to GAD-negative and presumed pyramidal cell bodies. However, Ca_v3.1 label did not extend noticeably to interneuron dendritic membranes or the synaptic terminal boutons that surround pyramidal cell somata, indicating a specific subcellular distribution for this T-type calcium channel isoform in local interneurons. As found in many brain regions, Ca_v3 labelling in the nuclear region was primarily evident for Ca_v3.3 immunolabel in either pyramidal cells or interneurons (Fig. 2C).

A comparison between the cortical expression pattern of Ca_v3 channels to previous physiological studies indicate that at least one or more of the Ca_v3 isoforms are expressed in cell types exhibiting LVA calcium responses. From this pattern we can infer that LVA spikes in the somatic region of pyramidal neurons may involve all three isoforms, while dendritic T-type activity in layer V pyramidal cells likely incorporates Ca_v3.2 and particularly Ca_v3.3 in mid-distal dendritic regions.

Hippocampus

Patch-clamp recordings have shown that single T-type channels are present in the somata and apical dendrites of hippocampal pyramidal cells, with a potential increase in channel density in dendritic regions (Fisher et al., 1990; Magee & Johnston, 1995; Kavalali et al., 1997). LVA calcium currents have also been reported to contribute to burst discharge in both hippocampal and subicular pyramidal cells (Jung et al., 2001; Metz et al., 2005). Dentate gyrus granule cells express LVA calcium channels and generate LVA spike responses (Fisher et al., 1990; Zhang et al., 1993). Ca_v3.1 protein has further been reported in the cell body region of dentate granule cells (Craig et al., 1999). The hippocampal and dentate regions contain a multitude of interneuron subtypes (Freund & Buszaki, 1996). Although interneurons typically exhibit a fast spiking phenotype compared to the burst discharge normally associated with calcium channel expression, recordings have shown calcium conductances in dendrites (Goldberg et al., 2004).

We observed immunolabel for all three T-type calcium channel isoforms in pyramidal cells, with a gradient of relative intensity for all isoforms that could be seen in single tissue slices to increase from a low level in the CA3 field to a higher intensity in CA1, and highest in the subiculum. The general subcellular distribution for the three Ca_v3 isoforms was apparent in pyramidal cells, with a restricted expression of Ca_v3.1 in the soma/proximal dendritic region, more extended labelling by Ca_v3.2 in soma and apical dendrites, and finally the soma and extended lengths of apical dendrites for Ca_v3.3 (Fig. 3A–G). The most prominent distribution of Ca_v3 immunolabel was found for Ca_v3.3 in the soma and over the apical dendritic axis of CA1 pyramidal cells, and in subicular pyramidal cells reached an intensity and degree of dendritic extension similar to that in neocortex (Fig. 3C and D). The pattern of expression in single cells was generally a diffuse internal label, with no clearly definable puncta at the light microscopic level. Apical dendritic labelling of pyramidal cells in CA and subicular regions could only be discerned for the main dendritic shafts and not the oblique dendrites in stratum radiatum, while branches in the distal tuft of apical dendrites in stratum lacunosum-moleculare were less evident than in cortical pyramidal cells. Similarly, we could not detect a noticeable change in Ca_v3.3 immunolabel between mid- and distal-apical dendritic regions, and no distinct label was detected in pyramidal cell basilar dendrites in hippocampal or subicular regions. In the dentate gyrus we observed a relatively weak expression of Ca_v3.1 but stronger staining for Ca_v3.2 and Ca_v3.3 at the cell body level of granule cells (Fig. 3H–J). We could detect no label for any of the Ca_v3 channel isoforms in granule cell dendrites despite resolving these structures with a MAP-2 counterlabel (Fig. 3H–J). When present, the more prominent labelling in the nuclear region was for Ca_v3.3, particularly in CA1 and dentate regions (Fig. 3C and J).

We identified GAD-labelled interneurons or cells with a morphology and location consistent with interneurons that were either positive or negative for different Ca_v3 channel immunolabels (Fig. 4). Most interneurons were positive for Ca_v3.3 or in some cases Ca_v3.2 or Ca_v3.1. Figure 4A indicates a positive label for Ca_v3.3 at the soma and at least proximal regions of horizontally projecting processes of a parvalbumin-positive stratum oriens-alvear interneuron. GAD-positive interneurons colabelled for Ca_v3.3 were located within or adjacent to stratum pyramidale of the CA1 region (Fig. 4B), while many presumed multipolar and fusiform interneurons in the hilar region were positive for Ca_v3.3 (Fig. 4C) or Ca_v3.2 (Fig. 4D). Potential MOPP or VIP-containing basket cell interneurons positioned in the inner molecular layer of the dentate gyrus were positive for all three Ca_v3 isoforms at the soma (Fig. 3H–J; Freund & Buszaki, 1996). An additional clear extension of Ca_v3.1 into the dendrites of these cells that project into the molecular layer was apparent (Fig. 3H).

Ca_v3 isoforms can thus be predicted to contribute to LVA calcium-dependent responses in somatic membrane of pyramidal cells, and for at least Ca_v3.3 in apical dendritic regions of CA1 and particularly subicular pyramidal cells. The dendritic LVA responses reported in dentate granule cells may derive from other channel subtypes, given an apparent restriction of Ca_v3 channel distribution to the somatic and nuclear region. The expression of Ca_v3 channel isoforms in interneurons in both the hippocampal and dentate gyrus regions was also more prominent than for interneurons in neocortical regions.

Thalamus

The thalamic nuclei represent one of the most extensively studied regions for the expression and functional output of T-type calcium channels. Previous recordings provide a detailed account of the electrical properties of LVA calcium currents and their role in generating rebound discharge and membrane potential oscillations (Pape et al., 2004). Most studies have focused on three primary cell types: (i) relay cells; (ii) local GABAergic circuit neurons within principal thalamic nuclei, and (iii) neurons of the nucleus reticularis (nRT). Similar work has been carried out in the midline habenular nuclei and dorsal paraventricular (PV) thalamic nuclei (Huguenard et al., 1993; Kim & Chang, 2005; Richter et al., 2005). Together these studies provide a wealth of evidence for LVA-mediated responses associated with T-type calcium channel activity, and the importance of distributing these channels to dendritic regions in specific cell types (Huguenard, 1996). However, the extent to which the activity of thalamic subnuclei reflects a differential expression or distribution of Ca_v3 channel isoforms has not been fully determined. We focused attention on the regions most examined in previous electrophysiological studies, including the medial and lateral habenular nuclei (MHb, LHb), PV, dorsal thalamic nuclei [dorsal lateral geniculate (dLGN), lateral posterior (LP), and lateral dorsal (LD)], and ventrobasal (VB) nuclei [ventroposterolateral (VPl), ventroposteromedial (VPm)], and nRT.

Midline thalamic nuclei

A prominent T-type current that underlies a LVA calcium spike and rebound discharge has been distinguished in LHb neurons (Kim & Chang, 2005). T-type currents in dissociated LHb cells can be further distinguished from relay or nRT cells on the basis of kinetic properties (Huguenard et al., 1993). In contrast, cells in the adjacent MHb exhibit only tonic firing, with no evidence for rebound discharge (Kim & Chang, 2005).

We observed a difference in the intensity and number of cells labelled for Ca_v3 isoforms that was sufficient to clearly define habenular nuclear boundaries. Ca_v3.1 T-type channel label was detected in MHb and LHb nuclei as both a cytosolic and putative membrane-associated punctate label (Fig. 5A and D). Although Ca_v3.1 label could be detected in the MHb the relative intensity of label in individual cells was far greater in LHb neurons. In the LHb distinct puncta of Ca_v3.1 label outlined the circumference of individual cell somata (Fig. 5D), with little or no evidence for Ca_v3.1 dendritic label in either nucleus despite resolving these structures with a MAP-2 counter-label. Ca_v3.2 channel label was present in both MHb and LHb cells as a diffuse or a punctate membrane-associated label in individual cells (Fig. 5B and E). We further detected some extension of Ca_v3.2 label into proximal dendritic stumps of cells in LHb but not MHb cells, and an additional dense Ca_v3.2 cytosolic label in a less frequently encountered and smaller diameter cell type in the LHb (Fig. 5E). Unlike other structures examined, Ca_v3.3 label in the MHb and LHb was expressed in a lower percentage of cells and with a lower relative intensity (Fig. 5C and F). Similarly, we could detect no evidence for a dendritic extension of Ca_v3.3 T-type channels in either the MHb or LHb, with all label restricted to a punctate somatic label or in some cases an additional nuclear label.

Neurons of the PV were recently shown to generate LVA calcium spikes that support rebound discharge through T-type calcium channels (Richter et al., 2005). The boundaries between habenular nuclei and the PV were readily apparent in Ca_v3.1 immunolabelled tissue, with a decrease in the number of cells in the PV compared to MHb, but with fluorescent intensity as high as Ca_v3.1 in the LHb. Ca_v3.1 immunolabel was again apparent as an intense punctate and membrane-associated signal that surrounded PV cell somata (Fig. 5G). Ca_v3.2 was detected as a more sparsely distributed but intense punctate label surrounding PV cell somata (Fig. 5H). Ca_v3.3 labelling in the PV was light compared to Ca_v3.1 and 3.2, in a much lower percentage of cells and included both somatic and nuclear labelling (Fig. 5I). As found for the MHb and LHb, we could detect no Ca_v3 label in PV cell dendrites even with high power resolution of surrounding dendrites with MAP-2.

Ca_v3 labelling in the midline thalamic nuclei is consistent with previous reports of rebound discharge in LHb and PV neurons as compared to a relatively lower intensity label for Ca_v3 protein and only tonic spike activity in the adjacent MHb (Huguenard et al., 1993; Richter et al., 2005). These structures were also distinct in exhibiting one of the clearest examples of a punctate pattern of label that can be associated with membrane-associated channels. As compared to cortical and hippocampal neurons, labelling for Ca_v3.1 in the LHb and PV was among the highest intensity of all structures examined but without detectable dendritic label.

Thalamic relay and nRT cells

Several experimental approaches have provided evidence for a distribution of Ca_v3 channels over extended regions of the soma-dendritic axis of thalamic relay, local circuit cells, and nRT cells. Patch recordings, calcium imaging and modelling of relay cells in the dLGN and ventrobasal (VB) nuclei indicate that T-type calcium channels are located at the soma and dendrites, with a higher peak density and activation in dendrites 18–60 µm from the soma (Munsch et al., 1997; Zhou et al., 1997; Destexhe et al., 1998; Williams & Stuart, 2000). Local GABAergic interneurons in thalamic nuclei exhibit a greater activation of LVA current in dendrites at 60 µm over that at the soma (Munsch et al., 1997) and less calcium current in dissociated cells when dendrites have been removed (Pape et al., 1994; Munsch et al., 1997; Tarasenko et al., 1997; Zhuravleva et al., 1999, 2001). Similar evidence exists for nRT cells, where models and recordings show that multiple aspects of LVA calcium-mediated responses depend on the dendritic distribution of T-type channels (Destexhe et al., 1996). Moreover, differences in the rate of inactivation of somatic and dendritic T-type currents have been measured with on-cell patch recordings in nRT cells (Joksovic et al., 2005a), directly supporting a selective distribution or modulation of Ca_v3 isoforms in somatic vs. dendritic regions.

Relay cells

We observed an extensive distribution of immunolabel for Ca_v3 isoforms in all thalamic nuclei, with Ca_v3 immunolabel localized to large diameter presumed relay cells, in GABAergic local circuit cells, and nRT cells (Fig. 6). There were few differences in the relative intensity of fluorescent label for Ca_v3 isoforms between main thalamic nuclei and the nRT. Rather, differences were apparent in the relative intensity of label for different isoforms between cells and over the soma-dendritic axis of individual cell classes. In general, Ca_v3.1 and Ca_v3.2 immunolabels were present in the soma of relay cells in both dorsal and ventral divisions, and Ca_v3.3 as a lighter somatic label. All Ca_v3 isoforms could be distinguished in individual cells to some degree in the somatic region and dendrites ≤30 µm from the soma, although no attempt was made to quantify the percentage of cells labelled in different nuclei. The most prominent dendritic distribution was apparent for Ca_v3.3 channels, with label detected in relay cell dendritic segments up to 50 µm in length more frequently than Ca_v3.1 or Ca_v3.2 (Fig. 6F and I). For each isoform the relative intensity and number of cells exhibiting a Ca_v3 label in soma or dendrites detected by MAP-2 colabelling varied, with some identified cells being negative for a given Ca_v3 isoform. In fact, neighbouring relay cells with dendrites of equivalent length could be negative or positive for dendritic Ca_v3.3 (data not shown). Labelling for any of the Ca_v3 isoforms could be distinguished in the nuclear region for a number of cells, but most frequently for Ca_v3.3 (Fig. 6A–F).

GABAergic local circuit neurons

Immunolabel for GAD revealed extensive axonal arborizations throughout all thalamic nuclei that are expected to arise from both nRT afferents and local circuit neurons (Harris & Hendrickson, 1987). A light and relatively diffuse label for Ca_v3 channels often found in thalamic nuclei could well correspond to labelling of this dense axonal network, but Ca_v3 immunolabel was not conclusively identified in axons or terminals in GAD colabelling studies. As previously reported, GABAergic cells were infrequently encountered in ventral thalamic nuclei, but were easily distinguished in the LP, lateral dorsal (LD), dLGN and medial geniculate nuclei (MGN) using a GAD counterlabel. Local circuit cells displayed a relatively light level of somatic labelling for Ca_v3.1, but label could be identified in some cases in dendritic regions (Fig. 7A). However, other GAD-positive processes of the same cell or nearby cells were unlabelled by Ca_v3.1 in the same section (Fig. 7A). Light or variable labelling for Ca_v3.2 could be found on somata of GAD-positive cells (Fig. 7B). Ca_v3.3 label was most intense in this cell class but was restricted to the nuclear and somatic region (Fig. 7C). The labelling reported here indicates that thalamic cells exhibit a distribution of Ca_v3 channels consistent with an important role for dendritic LVA calcium channels in determining cell output. A greater variability between MAP-2 labelled dendrites and Ca_v3 immunolabel compared to cortical or hippocampal regions also suggests a greater heterogeneity of Ca_v3 channel distribution over the soma-dendritic axis. Thus, relay cells exhibit a distribution of any of the three isoforms in at least the proximal stump of dendrites, but the extent of this distribution that could be detected at the light microscopic level varies between cells. Local circuit interneurons are somewhat unique in distributing Ca_v3.1 channels to extended regions of dendrites and prominent labelling for Ca_v3.3 in the nuclear region.

Cerebellum and inferior olive

Electrophysiology and calcium imaging studies have established that Purkinje cells and neurons of the deep cerebellar nuclei (DCN) generate LVA-mediated responses attributed to T-type calcium current. In Purkinje cells T-type single channels can be recorded in dendritic regions (Mouginot et al., 1997) and contribute to both the burst and interburst depolarizations (Swensen & Bean, 2003; Isope & Murphy, 2005). In DCN cells Ca_v3 channels contribute to a rebound burst discharge following Purkinje cell inhibition (Gauck et al., 2001; Molineux et al., 2006). Inferior olivary neurons provide climbing fibre input to both DCN and Purkinje cells and were the first cell type identified as generating LVA calcium spikes (Llinas & Yarom, 1981). Unlike other major cells with prominent I_T, the LTS in inferior olivary neurons was proposed to be initiated at the soma rather than dendrites (Llinas & Yarom, 1981).

The distribution of Ca_v3 channel proteins in cerebellar neurons was recently described, revealing a widespread expression of at least one or more Ca_v3 isoforms in the main cerebellar cell types, and a striking distribution of Ca_v3.3 to at least Purkinje cell dendrites (Molineux et al., 2006). Here we focus on the extent of dendritic Ca_v3 channel distribution in the three major input and output cells of cerebellum: Purkinje, DCN and inferior olivary neurons.

Purkinje cells

Ca_v3 channels were distributed in Purkinje cells with the same general pattern we detected in other brain regions: primarily Ca_v3.1 and Ca_v3.2 at the soma or proximal dendrites, and Ca_v3.3 at the soma and over extensive regions of the dendritic tree of most Purkinje cells (Molineux et al., 2006). Craig et al. (1999) found that Ca_v3.1 label could also extend over the dendrites of at least some Purkinje cells of mouse cerebellum. In our hands, Ca_v3.1 and Ca_v3.2 could be detected in some rat Purkinje cell dendrites, but only in a very small fraction of cells.

Figure 8A and B provides magnified images of the distribution of Ca_v3.3 channels in Purkinje cell dendrites. We found that Ca_v3.3 immunolabel was distributed as either a diffuse cytosolic label (Fig. 8A) or as a distinct punctate or patchy distribution over the primary dendritic shafts across the entire extent of the molecular layer (Fig. 8B). A lower intensity of Ca_v3.3 immunolabel extended into the smaller diameter secondary and tertiary dendritic branches (Fig. 8B), but no label of dendritic spines could be detected even at 100× magnification. In thin confocal sections the patchy distribution of Ca_v3.3 immunolabel could be visualized further as being positioned on the circumference of the dendritic shaft, as highlighted by an internal cytoplasmic core of MAP-2 counterlabel (Fig. 8B; arrows). Labelling in the region of the nucleus was also distinguished for stellate cells in the molecular layer for Ca_v3.3 (Fig. 8A and B).

Inferior olivary neurons

Inferior olivary cells identified through a calbindin counterlabel expressed detectable levels of only Ca_v3.1 and Ca_v3.3 at the soma (Fig. 8E and F) and were negative for Ca_v3.2 (not shown). The most intense label was found for Ca_v3.1 at the soma, with no apparent dendritic label for Ca_v3.1 (Fig. 8E). Ca_v3.3 label was comparatively light and associated with the somatic and nuclear regions (Fig. 8F).

Determination of Ca_v3 channel distribution

The function of T-type calcium channels has been investigated in depth through physiological recordings, while more recently the pattern of Ca_v3 mRNA expression has been detailed (Craig et al., 1999; Kase et al., 1999; Talley et al., 1999). The distribution of Ca_v3 protein described here agrees well with the cell types previously demonstrated to be capable of generating LVA calcium-dependent responses. The Ca_v3 immunostaining also largely agrees with that reported for Ca_v3 mRNA. However, it has become recognized that the correspondence between detectable mRNA and protein levels obtained through in situ hybridization and immunocytochemistry can differ considerably, with neither serving as a faithful predictor of the other (Gygi et al., 1999; Zhang et al., 2002). One example is the finding that ∼80% of the nRT cell T-type current is blocked by N₂O (a specific blocker of Ca_v3.2 channels in expression systems) despite much higher apparent levels of Ca_v3.3 mRNA (Talley et al., 1999; Joksovic et al., 2005b). Thus, the location and relative intensity of immunolabel was similar to that reported for mRNA in some cell types (i.e. hippocampal pyramidal cells, inferior olive), while others showed less correspondence (i.e. thalamic relay nuclei vs. nRT). We previously found that the distribution of Ca_v3 immunolabel in cerebellar cell types was also different from the reported mRNA expression (Molineux et al., 2006). However, Ca_v3 protein distribution proved to entirely match the ability to generate LTS in six different cerebellar cell types when competing outward potassium currents were blocked. Thus, even when calcium channels are transported to the membrane, their activation can be masked by the coexpression of potassium channels (Pape et al., 1994; Molineux et al., 2005, 2006). Work to date then indicates that Ca_v3 immunostaining with these antibodies can be an accurate reflection of the cells that express T-type channels. Nevertheless, the relation between mRNA expression, immunolabel distribution, and cell activity ultimately needs to be verified through physiological recordings. For this reason we will use electrophysiological data from the literature to provide a physiological context to our results and assess the potential functional significance of our labelling patterns.

Ca_v3 distribution and physiological activity

Distributing select Ca_v3 channel isoforms with distinguishable kinetic properties to different parts of a cell would be expected to influence cell output. Indeed, one consequence of expressing multiple Ca_v3 isoforms may be to increase the potential range for post-translational modifications of channel function. A comparison of our immunolabel results with previous electrophysiological studies identifies some correlations between Ca_v3 channel expression and different T-type channel-mediated responses.

Calcium spike discharge

A prominent electrophysiological response associated with T-type calcium channel activation is a LVA calcium spike. Many of the cell types labelled for Ca_v3 isoform(s) generate somatic or dendritic calcium spike discharge. In particular, dendrites were most often associated with an extended distribution of Ca_v3.3 immunolabel, suggesting that at least this isoform can contribute to dendritic LVA calcium responses. LVA calcium influx in cortical pyramidal cell dendrites (Yuste et al., 1994; Jung et al., 2001; Larkum & Zhu, 2002) thus correlates with a widespread distribution of Ca_v3.3 channels over the apical dendritic axis. The prominent calcium spikes of cerebellar Purkinje cells can be expected to include the activation of Ca_v3.3 channels that are distributed over primary, secondary and tertiary dendritic branches (Swensen & Bean, 2003; Isope & Murphy, 2005). Note that although a dendritic localization of Ca_v3.3 immunolabel was prominent in many cell types, we could also detect a proximal to mid-dendritic distribution of Ca_v3.2 channels or even Ca_v3.1 channels, suggesting the additional involvement of these isoforms. In contrast, LVA calcium currents of inferior olivary neurons appear to derive from Ca_v3.1 and Ca_v3.3 isoforms localized to the soma.

Rebound discharge

A second output that results from the activation of T-type calcium channels is a rebound depolarization following a preceding hyperpolarization. This activity can initiate a high frequency spike output in response to inhibition or contribute to oscillatory swings in membrane potential. Recent work has indicated that rebound discharge in thalamic relay cells is blocked in a Ca_v3.1 knockout animal (Kim et al., 2001). An extensive analysis of mRNA through RT-PCR also reported a correlation between the expression of Ca_v3.1 mRNA and the ability to generate rebound bursts (Toledo-Rodriguez et al., 2004). We recently established a correlation between Ca_v3.1 channel expression and rebound discharge in large diameter DCN cells (Molineux et al., 2006). Cerebellar Golgi cells and Purkinje cells also label for Ca_v3.1 and generate rebound depolarizations, while non-bursting DCN cells and stellate cells that lack this isoform do not. A similar correlation between Ca_v3.1 expression and rebound burst discharge appears to exist between the habenular nuclei, where intense somatic Ca_v3.1 immunolabel in LHb cells correlates to rebound burst capability, whereas MHb cells that exhibit only light Ca_v3.1 label do not exhibit rebound bursts (Kim & Chang, 2005).

Collectively, the available data suggests that expression of Ca_v3.1 channels may be sufficient and potentially necessary to generate rebound discharge in a wide variety of neurons. Nevertheless, the extent to which these correlations may also reflect the relative density of Ca_v3.1 channel expression, varying kinetic properties of Ca_v3 isoforms through post-translational modulation, or the degree of interplay with other ion channels remains to be determined.

Kinetic properties

Kinetic properties of individual Ca_v3 isoforms have been characterized primarily in expression systems at room temperature (McRory et al., 2001; Chemin et al., 2002; Perez-Reyes, 2003). The degree of correspondence between the properties of expressed channels and those in native neurons is not fully known. As most of the cell types examined here express more than one Ca_v3 channel isoform, it may be difficult to predict the properties of whole-cell Ca_v3 current. Differences in the properties of T-type currents have been reported between thalamic regions (Huguenard et al., 1993; Joksovic et al., 2006). Our immunolabel did not reveal striking differences in the expression pattern of the Ca_v3 isoform protein between thalamic nuclei, but this does not rule out the selective translocation of a given isoform to the plasma membrane or post-translational modifications as a source for physiological differences. It is also difficult to relate differences in burst properties to soma-dendritic Ca_v3 channel distributions (i.e. relay vs. nRT cells). In general, relay cells are positive for all three Ca_v3 isoforms at the soma and proximal dendrites and shift to a distribution of Ca_v3.3 label in distal dendrites. Ca_v3.1 and Ca_v3.2 label is readily detected at the soma of nRT cells, but Ca_v3.1 and/or Ca_v3.3 immunolabel can be distributed over extensive regions of dendritic membrane in a given cell. Recent patch recordings reveal that T-type calcium channels at the soma of nRT cells exhibit a faster rate of inactivation (τ = 28 ms) than those recorded in proximal dendrites (τ = 53 ms; Joksovic et al., 2005a). These findings are at least consistent with a higher density of Ca_v3.3 channels in dendritic regions that may influence spike output in these cells (Destexhe et al., 1996; Destexhe et al., 1998; Williams & Stuart, 2000).

Synaptic responses

LVA calcium currents have been shown to contribute to synaptically activated calcium influx over extensive regions of the dendritic arbor (Yuste et al., 1994; Magee et al., 1995). Our data suggest that all three T-type calcium channel isoforms may contribute to this activity in proximal dendritic regions of many cell types, but that more distal involvement of T-type channels to synaptic input is likely to be mediated primarily by Ca_v3.3 channels. Another important potential location of T-type channel distribution is dendritic spines, where low voltage synaptic responses could be amplified by Ca_v3 activation. Putative T-type mediated calcium influx has been identified in the spines of Purkinje cell dendrites from younger animals (Isope & Murphy, 2005). We were unable to obtain any evidence here for Ca_v3 channel distribution to dendritic spines, including that of Purkinje cells, where these structures can be readily visualized using a calbindin counterlabel. A distribution to dendritic spines may then be restricted to early developmental stages or simply fall below resolution at the light microscopic level. Alternatively, calcium entry into dendritic spines may be secondary to calcium influx in the adjacent dendritic shaft or involve other LVA-like channels (Sabatini & Svoboda, 2000).

Nuclear regulation

The most unexpected labelling pattern for Ca_v3 protein was in the nuclear region of some cells. To our knowledge this is the first report and test for voltage-dependent calcium channels localized to the nuclear protein fraction, even though several other ion channel types have been directly recorded (Mazzanti et al., 2001). The exact location at which these channels are expressed remains to be determined. Membranes of the smooth ER are continuous with the outer membrane of the nuclear envelope, where a high density of large diameter nuclear pore complexes (NPCs) are expressed. The classical view of the NPC is of a fixed and open structure that allows free transfer of compounds between the cytoplasm and nucleoplasm. This understanding has been substantially revised with findings that the NPC is actually semipermeable, with an additional set of eight smaller diameter pores in the outer ring of the NPC that allows passage of smaller inorganic ions (Pante & Aebi, 1995). Patch-clamp recordings have measured conductances in nuclear membranes at 800–1000 nS for the unblocked NPC core, as well as smaller conductances of 2–200 nS. The source of these smaller conductances has not been determined, but could potentially correspond to the small diameter outer pores of the NPC. Altogether, the nuclear envelope acts as a high resistance and semipermeable membrane that maintains transmembrane gradients for ion species (Mazzanti et al., 2001). Nucleocytoplasmic transmembrane potentials of up to −33 mV below that of the plasma membrane potential have been measured, although this waits to be determined for neuronal nuclei. The nuclear envelope is recognized further for acting as a calcium store for the release of calcium following IP3 or ryanodine receptor activation that can react to events at the level of the plasma membrane or cytoplasm (Gerasimenko et al., 1995; Hardingham et al., 2001; Marchenko & Thomas, 2006; Zhang et al., 2006). Indeed, postsynaptic activity can regulate the extent and time-course of nuclear calcium increases, providing a means of integrating neuronal firing patterns in terms of the level of gene transcription (Hardingham et al., 2001; Power & Sah, 2002). The nucleus thus exhibits the necessary calcium concentration gradient and potentially voltage gradient to employ the LVA properties of Ca_v3 channels. As such, our results suggest the potential for T-type calcium channels to have previously unrecognized roles in regulating nuclear membrane activity, and add to a growing list of voltage-activated ion channels that can be localized to the nuclear region.