Kinetic modification of the α_1I subunit-mediated T-type Ca²⁺ channel by a human neuronal Ca²⁺ channel γ subunit

Abstract

1. Voltage-sensitive Ca²⁺ channels (VSCCs) are often heteromultimeric complexes. The VSCC subtype specifically expressed by skeletal muscle has long been known to contain a γ subunit, γ₁, that is only expressed in this tissue. Recent work, initiated by the identification of the mutation present in the stargazer mouse, has led to the identification of a series of novel potential Ca²⁺ channel γ subunits expressed in the CNS.

2. Based on bioinformatic techniques we identified and cloned the human γ₂, γ₃ and γ₄ subunits.

3. TaqMan analysis was used to quantitatively characterise the mRNA expression patterns of all the γ subunits. All three subunits were extensively expressed in adult brain with overlapping but subunit-specific distributions. γ₂ and γ₃ were almost entirely restricted to the brain, but γ₄ expression was seen in a broad range of peripheral tissues.

4. Using a myc epitope the γ₂ subunit was tagged both intracellularly at the C-terminus and on a predicted extracellular site between the first and second transmembrane domains. The cellular distribution was then examined immunocytochemically, which indicated that a substantial proportion of the cellular pool of the γ₂ subunit was present on the plasma membrane and provided initial evidence for the predicted transmembrane topology of the γ subunits.

5. Using co-transfection techniques we investigated the functional effects of each of the γ subunits on the biophysics of the T-type VSCC encoded by the α_1I subunit. This revealed a substantially slowed rate of deactivation in the presence of γ₂. In contrast, there was no significant corresponding effect of either γ₃ or γ₄ on α_1I subunit-mediated currents.

VSCCs play a critical role in a wide variety of biological functions, including pre-synaptic transmitter release, muscle contraction and gene expression (Hille, 1992). On the basis of their voltage dependence of activation, VSCCs are subdivided into two major classes known as high voltage-activated (HVA) channels and low voltage-activated (LVA) channels. HVA channels are heteromeric complexes that are believed in all cases to consist of at least an α₁, β and α₂δ subunit. Of these, the α₁ subunit is the major determinant of the channel phenotype, and alone encodes the Ca²⁺-selective pore, the voltage-sensing apparatus and major drug-binding sites. To date, seven distinct HVA channel-encoding α₁ subunit genes are known, which are named α_1A through to α_1F, plus the skeletal muscle-specific α_1S. LVA Ca²⁺ channels are also based around α₁ subunits, of which three are currently known, α_1G, α_1H and α_1I (for review see Perez-Reyes, 1999; Randall & Benham, 2000). In contrast to HVA channels, less is known about the subunit composition of LVA VSCCs, and indeed it remains possible that some or even all LVA channels exist as monomers of α₁ subunits alone. Contrary to this, there are reports of a significant functional association between the α_1G LVA channel and α₂δ subunits (Dolphin et al. 1999; Hobom et al. 2000); although others (Lacinova et al. 1999) noted some small effects of α₂δ in similar experiments, they failed to reach statistical significance.

It has long been known that the α_1S-based VSCC contains an additional subunit known as γ. Like the α_1S subunit with which it associates, expression of this γ subunit is entirely restricted to skeletal muscle (Powers et al. 1993). As no other γ subunits were identified by almost a decade of homology screening, it was widely believed that only the one γ subunit existed, which was exclusively associated with α_1S-containing VSCCs, and therefore in some way reflected the unique functional role of these channels in the excitation-contraction coupling of skeletal muscle. This dogma was recently challenged by data from a genetic investigation of the spontaneously epileptic mouse line stargazer. Determination of the mutation present in these animals uncovered a novel neuronal specific gene that encoded a protein originally dubbed stargazin. Sequence analysis and structural prediction of stargazin pointed to considerable homology with the skeletal muscle VSCC γ subunit (Letts et al. 1998).

Subsequent work with stargazin suggested that its expression in BHK cells (baby hamster kidney cell line) could, albeit subtly, modulate the properties of a co-expressed HVA VSCC, α_1A (Letts et al. 1998). This observation led to stargazin being renamed as the γ₂ VSCC subunit (with the original skeletal muscle subunit being termed γ₁). The identification of murine γ₂ rapidly led to the isolation of its human orthologue plus the cloning of an additional human paralogue, γ₃ (Black & Lennon, 1999). Subsequent to this, two additional subunits known as γ₄ and γ₅ were isolated from mice (Klugbauer et al. 2000). Of these subunits γ₂ and γ₄ have been recently reported to alter the inactivation of α_1A-mediated VSCC, whereas γ₅ seemingly interacts with the LVA subunit α_1G (Letts et al. 1998; Klugbauer et al. 2000).

Here we confirm the cloning of the human γ subunit γ₄, as predicted by Burgess et al. (1999), and report the mRNA distribution in human tissues for each of the subunits γ₂, γ₃ and γ₄. We additionally describe the effects of co-expression of each of these human γ subunits on the properties of the LVA Ca²⁺ currents produced by stable heterologous expression of the α_1I subunit.

METHODS

Identification of human γ subunit genomic sequences and prediction of cDNA sequences

Using the mouse stargazin (i.e. γ₂) protein as a template, sequence comparison searching of human EST (expressed sequence tag) and high-throughput genomic sequence data was performed with the WU-BLAST 2.0 (W. Gish, 1997, unpublished, see http://blast.wustl.edu/blast/README.html) version of BLAST software (basic local alignment search tool; Altschul et al. 1990). An assumption that the regions of high similarity (high scoring pairs) were roughly equivalent to actual exons, led to the identification of individual exons for three possible γ subunit sequences. Accurate positions for exon-intron boundaries were determined by visual analysis of the three-frame translations generated by the MAP program (Wisconsin package version 9.0, Genetics Computer Group) on the appropriate regions of genomic sequence, and applying the consensus splice site intron rule (gt … ag) as a guide. Sequence identities were calculated using the BESTFIT program (Wisconsin package version 9.0), sequence alignments of the predicted protein sequences were performed using CLUSTALW (Thompson et al. 1994) and shading was applied using the Belvu program (E. Sonnhammer, unpublished, see http://www.sanger.ac.ukpfam/help/belvu-setup.shtml).

Isolation of γ subunit cDNAs

To complete the 5′ end of the CACNG4 cDNA sequence (i.e. the γ₄ subunit gene), primers derived from the predicted partial sequence were used in 5′ RACE (rapid amplification of cDNA ends; Frohman, 1993) from Marathon-Ready human brain and brain sub-regional cDNAs (Clontech, Basingstoke, UK). The first of nested RACE-PCR reactions used a gene-specific primer (5′-AGAGGATGCCGCACTGAGGACGATG-3′) in combination with the AP1 primer (Clontech), which is complementary to the adaptor ligated to both ends of Marathon-Ready cDNA. The secondary reaction used 1/25 of the product of the first, with a second gene-specific primer (5′-GCAGGAGCAGGATGGTGCTGAGGATG-3′) and AP2 (Clontech) as nested primers. The RACE reaction mix contained 0.5 μm primers, 0.2 mm dNTPs, 1.5 mm MgCl₂ and 20 U ml⁻¹ of Platinum Taq (Life Technologies, Paisley, UK). The thermal cycle for the primary reaction was: 94 °C for 2 min, 5 cycles of 94 °C 30 s, 72 °C 3 min, then 25 cycles of 94 °C 30 s, 70 °C 3 min, and ended with 7 min at 72 °C. The same cycle was used for the second reaction, except that the number of cycles at 94 °C and 70 °C was increased to 30. A second 5′ RACE experiment was done similarly, using further nested gene-specific primers (5′-GTTGCAGATGTGCGCGCTGGAGTAC-3′, 5′-GGCCGTGGTCAGCAGCATCTGCAG-3′) derived from the sequence determined in the first RACE experiment.

Full-length clones of CACNG2, CACNG3 and CACNG4 were obtained by RT-PCR as follows. Oligo dT-primed first-strand cDNA was made from 1 μg human brain poly-A⁺ mRNA (Clontech), using the Superscript II pre-amplification system (Life Technologies). High-fidelity PCR amplification of the coding region of each γ subunit cDNA from 1/500 of the RT reaction product was performed with primers that incorporated the trinucleotide ACC immediately upstream of the start codon to optimise expression (forward (F) 5′-ACCATGGGGCTGTTTGATCGAG-3′ and reverse (R) 5′-CCCGCGGTCTTTTATACG-3′ for CACNG2; F 5′-ACCATGAGGATGTGTGACAGAAGG-3′ and R 5′-CCATCACAAGGACCATGC-3′ for CACNG3; F 5′-ACCATGGTGCGATGCGAC-3′ and R 5′-ACATGCCATCCTGTGTGAC-3′ for CACNG4). The PCR components were 0.5 μm primers, 0.2 mm dNTPs, 1.5 mm MgCl₂, 40 U ml⁻¹, PfuTurbo DNA polymerase (Stratagene, Amsterdam, The Netherlands). The PCR thermal cycle consisted of: 94 °C for 2 min, 8 cycles of 94 °C 10 s, 60 °C 30 s, 72 °C 1 min, then 23 cycles of 94 °C 10 s, 56 °C 30 s, 72 °C 1 min, and ended with 4 min at 72 °C.

cDNA cloning

PCR products were agarose-gel separated and purified by a silica-membrane-based method (Qiaquick gel extraction kit, Qiagen, Crawley, UK). RACE-PCR products were cloned into the topoisomerase-activated pCR2.1-topo vector (Invitrogen, Groningen, The Netherlands). Full-length RT-PCR products were cloned into the topoisomerase-activated pcDNA3.1-topo vector (Invitrogen). The resulting plasmids were transformed into Escherichia coli K12 TOP10 cells (Invitrogen). Insert-containing pCR2.1-topo clones were identified by colour selection with isopropyl-β-d-thiogalactopyranoside and clones were tested for inserts of the expected size by PCR with vector-specific primers. Plasmid DNA was prepared from suspension cultures of selected clones by alkaline lysis followed by anion-exchange chromatography (Qiaprep miniprep kit, Qiagen).

Sequence analysis

Clones were sequenced on both strands from standard vector primers and, where necessary, internal gene-specific primers using an ABI 377 automated sequencer (PE Applied Biosystems, Warrington, UK). Sequences were assembled with the Seqman software package (DNASTAR, Madison, WI, USA), and the alignments optimised manually to give an overall consensus. For each of the three γ subunit full-length PCR products, two independently derived clones were sequenced to rule out ambiguities or PCR-introduced mutations.

Distribution analysis of γ subunits by TaqMan RT-PCR

TaqMan RT-PCR analysis was carried out as previously described (Medhurs Talley et al. 2000). Human poly-A⁺ mRNA samples extracted from the CNS and peripheral tissues were obtained from Clontech. cDNA synthesis was performed in triplicate. For each 20 μl reverse transcription reaction, 200 ng human poly-A⁺ mRNA in 9 μl water was mixed with 1 μl oligo dT primer (0.5 μg; Life Technologies) and incubated for 5 min at 65 °C. After cooling on ice the solution was mixed with 4 μl 5 × first strand buffer, 2 μl of 0.1 m DTT, 0.5 μl each of dATP, dTTP, dCTP and dGTP (each 10 mm), 1 μl RNAseOUT (40 U; Life Technologies) and 1 μl SuperScript II reverse transcriptase (200 U; Life Technologies). Reactions were performed for 60 min at 42 °C and terminated by incubating for 15 min at 70 °C. Parallel reactions for each RNA sample were run in the absence of SuperScript II to assess the degree of any contaminating genomic DNA.

TaqMan PCR assays for CACNG2, CACNG3, CACNG4 and GAPDH were performed in triplicate on cDNA samples or genomic DNA standards in 96-well optical plates on an ABI Prism 7700 sequence detection system (PE Applied Biosystems). For each 25 μl TaqMan reaction, 1 μl cDNA (or genomic DNA standard) was mixed with 11.25 μl PCR-grade water, 11.25 μl 2 × TaqMan universal PCR master mix (PE Applied Biosystems), 0.5 μl sense primer (10 μm), 0.5 μl antisense primer (10 μm) and 0.5 μl TaqMan probe (5 μm). Primer sequences were as follows. CACNG2 F 5′-ATCATCGCCGAGATGGTCG-3′, CACNG2 R 5′-GTGGCCCGCAGCTGTTT-3′ and CACNG2 probe 5′-TGGCGGTGCACATGTTTATCGACC-3′; CACNG3 F 5′-CCACCGCAGCAGACACAAC-3′, CACNG3 R 5′-GTTGCTTAACCCTGCAGAGACAA-3′ and CACNG3 probe 5′-TCATTCTCAGCGCGGGCATCTTT-3′; CACNG4 F 5′-GGCACTGCTTCCGGATCA-3′, CACNG4 R 5′-AGGAGGTACTCCGAGCTGTCG-3′ and CACNG4 probe 5′-TCACTTCCCAGAGGACAATGACTACGACC-3′.

PCR parameters were 50 °C for 2 min, 95 °C for 10 min, 40 cycles of 95 °C for 15 s and 60 °C for 1 min. Data were captured using an ABI Prism 7700 sequence detector system and analysed using relative standard curve method (Livak, 1999). Each sample was normalised to GAPDH to correct for differences in RNA quality and quantity (Medhurs Talley et al. 2000).

Myc tagging of the γ₂ subunit

Addition of a myc tag at the predicted intracellular C-terminus of γ₂ was carried out by performing high fidelity PCR amplification of the coding region cDNA with primers which incorporated a 5′Hin dIII site upstream of the start codon and an in-frame 3′Xba I site downstream of the coding sequence minus the stop codon (F 5′-TATAAGCTTATGGGGCTGTTTGATCGAGG-3′ and R 5′-TATTCTAGATACGGGGGTGGTCCGGCG-3′). The PCR components were 10 ng plasmid DNA, 1 μm primers, 1 mm dNTPs, 1 ×PfuTurbo reaction buffer (incorporating 2 mm MgCl₂) and 2.5 U PfuTurbo DNA polymerase. The following PCR cycles were then carried out: 94 °C for 1 min, then 30 cycles of 94 °C 1 min, 57 °C 1 min, 72 °C 1 min, and ended with 10 min at 72 °C. The PCR products were agarose gel separated and purified using a Qiaquick gel extraction kit (Qiagen), and then digested simultaneously with Hin dIII and Xba I and again were agarose gel separated and purified using a Qiaquick gel extraction kit. A modified pcDNA3.1A myc/His vector (Invitrogen) was digested with Hin dIII and Xba I and was agarose gel separated and purified using a Qiaquick gel extraction kit. The PCR product was cloned into the digested vector and the resulting plasmid was transformed into E. coli XL1 blue cells (Stratagene). Plasmid DNA was prepared from suspension cultures of selected clones using Qiaprep miniprep kits (Qiagen) and insert-containing clones were identified by enzyme digestion of this plasmid DNA. Plasmid DNA from clones containing inserts of the expected size were scaled up using Qiagen maxiprep kits.

Incorporation of a myc tag into the predicted extracellular loop of γ₂, between the first and second transmembrane domains at amino acid position 85 was carried out by performing ‘all the way round’ PCR with primers incorporating a myc tag. Forward and reverse primers each contained half of a myc tag at the 5′ end of the primer (F 5′-TCAGAAGAGGATCTGGATGCAGATTACGAAGCTGACACAGCAG-3′ and R 5′-GATGAGTTTTTGTTCCTCTGGGAAGTGATCAATTTGCTTGCAC-3′), the myc-tag in bold, with the forward primer being phosphorylated at the 5′ end. The PCR components were 50 ng plasmid DNA, 100 ng each primer, 0.2 mm dNTPs, 1 × Herculase reaction buffer and 2.5 U Herculase DNA polymerase (Stratagene). The following PCR cycles were then carried out: 92 °C for 2 min, then 35 cycles of 92 °C 30 s, 58 °C 30 s, 72 °C 6 min 40 s, and ended with 10 min at 72 °C. The PCR product was agarose gel purified using a Qiaquick gel extraction kit, and a series of ligation reactions using increasing amounts of the eluted PCR product were set up overnight at 14 °C. Parallel reactions containing no ligase enzyme were set up as controls. The ligation reactions were transformed into E. coli XL1 blue cells, and plasmid DNA was prepared from suspension cultures of selected clones using Qiaprep miniprep kits and sequenced to confirm the presence of the myc tag.

Transient transfections of γ subunits into HEK 293 α_1I cells

HEK 293 cells (human embryonic kidney cell line) stably expressing the rat T-type α_1I Ca²⁺ channel were obtained from E. Perez-Reyes (University of Virginia, USA). This version of the rat α_1I Ca²⁺ channel (GenBank accession no. AF086827) used to generate this stable cell line is a C-terminally truncated version of the predicted full length rat α_1I Ca²⁺ channel, and appears to be missing a short section at the 3′ end of the coding sequence. Functional data generated from this C-terminally truncated rat α_1I cell line (Lee et al. 1999), when compared to the full length human α_1I Ca²⁺ channel, are very similar, with the human subunit retaining most of the properties described initially for the rat α_1I subunit (see Monteil et al. 2000). Therefore, this C-terminal truncation appears to have little functional effect on the electrophysiological properties of α_1I.

These cells were transiently co-transfected with plasmid DNA for each of the γ subunits (tagged and non-tagged versions) and green fluorescent protein (GFP) by the Lipofectamine Plus method (Life Technologies, according to the manufacturer's instructions). After approximately 24 h, GFP-positive cells were selected for electrophysiological recordings.

Anti-myc staining of transiently transfected HEK 293 cells

Wild-type HEK 293 or HEK 293 α_1I cells were transiently transfected in poly-lysine-coated 24-well dishes with either the extracellular or intracellular myc-tagged version of the γ₂ subunit and left for approximately 24 h before staining. Live staining to detect the extracellular myc-tagged γ₂ was carried out by addition of 7 μg ml⁻¹ mouse anti-myc monoclonal antibody (Sigma, Gillingham, UK) in 1 % bovine serum albumin-phosphate-buffered saline (PBS) to the cells for 1 h at room temperature. Cells were then carefully washed three times in PBS, then fixed in 4 % paraformaldehyde/PBS for 10 min at room temperature, washed three times in PBS, then incubated with 90 μg ml⁻¹ fluorescein isothiocyanate (FITC)-conjugated rabbit anti-mouse immunoglobulins (Dako Ltd, High Wycombe, UK), for 1 h at room temperature.

To detect the C-terminal myc-tagged γ₂, cells were fixed in 4 % paraformaldehyde-PBS for 10 min at room temperature, washed three times in PBS, permeabilised in 0.1 % Triton X-100 for 20 min at room temperature and washed three times in PBS. Cells were then incubated with the anti-myc monoclonal and FITC-conjugated rabbit anti-mouse immunoglobulins as for the live staining. Cells were visualised on an inverted microscope.

Electrophysiology

Electrophysiological experiments were carried out using the whole-cell variation of the patch-clamp technique. Recordings were only made from cells that were strongly GFP positive. Current-voltage relationship data were gathered from cells from a minimum of four and a maximum of six different transfections. Steady-state inactivation data were gathered from between one and three different transfections, depending on the subunit under investigation. Data on inactivation kinetics were gathered from one transfection and deactivation data were gathered from four transfections. All experiments were performed at room temperature with pipettes of ≈2-5 MΩ resistance. Series resistances (R_ser) were always < 20 MΩ (mean R_ser 8.45 ± 0.46 MΩ, n = 110) and were additionally compensated by 80-95 % using the appropriate controls on the patch clamp amplifier (Axopatch 200B, Axon Instruments Inc., Foster City, CA, USA). For the deactivation experiments shown in Fig. 6. R_ser was always below 11 MΩ (mean 6.8 MΩ) and there was no significant difference between the four data sets shown.

Figure 6. Altered deactivation kinetics of α_1I-mediated Ca²⁺ currents in the presence of the γ₂ subunit.

A, the applied voltage protocol (top) and resultant current traces (bottom) from typical experiments used to characterise α_1I subunit-mediated current deactivation. Examples are shown for cells transiently expressing GFP alone (left) or GFP plus γ₂ (right). The voltage protocol consisted of a 40 ms depolarising pulse to -15 mV from a holding potential of -80 mV followed by repolarisation to potentials between -120 and -60 mV; only the portion of the current traces around the repolarising step is shown. Similar experiments were also performed on cells expressed GFP plus either the γ₃ or the γ₄ subunit (not shown). B, a graph compiled from populations of recordings like those in A. The graph plots the time constant of deactivation (derived by fitting an exponential function to the decay of the tail current) against the repolarisation potential (V_tail). In cells expressing the γ₂ subunit, deactivation was significantly slower than control cells (GFP transfected) or cells expressing γ₃ or γ₄. This was the case at all repolarisation potentials between -120 and -60 mV. Data are shown for cells expressing either GFP alone or GFP plus γ₂, γ₃ or γ₄.

Gigaseals were formed in an extracellular solution consisting of (mm): NaCl, 138; KCl, 5; CaCl₂, 2; MgCl₂, 1; glucose, 30; Hepes-NaOH, 15; pH 7.3, 310-315 mosmol l⁻¹. To record well-isolated Ca²⁺ currents this solution was exchanged for one of the following composition (mm): tetraethylammonium chloride, 165; CsCl, 5; CaCl₂, 2; MgCl₂, 1; Hepes, 10; glucose, 10; pH 7.3, 310-315 mosmol l⁻¹. The pipette contained a solution of the following composition (mm): CsCl, 140; MgCl₂, 4; Hepes, 10; EGTA, 10; pH 7.3, 295 mosmol l⁻¹.

All experiments were performed under control of the pCLAMP8 software suite (Axon Instruments). This program was also used extensively in data analysis. The current output of the amplifier was filtered at f_c= 2 kHz and sampled at 10 or 20 times f_c. Leak subtraction was performed in software using a standard P/4 method; both leak-subtracted and raw data were stored on computer disk. The voltage clamp protocols employed are described within the body of the text. Throughout data are presented as means ± 1 s.e.m. Statistical analysis was carried out using Student's t test with significance set at P < 0.05

RESULTS

Prediction of human γ subunit cDNA sequences from genomic sequence

To identify human genes encoding neuronal voltage-gated Ca²⁺ channel γ subunits, the mouse stargazer gene CACNG2 was used to search for similar sequences in publicly available human sequence databases. We found human genomic sequences and ESTs that potentially encoded three γ subunits, which we termed CACNG2, CACNG3 and CACNG4. The four exon structure of CACNG2 was conserved in all three human subunits. Exon 1 of CACNG2 was present in a genomic clone from human chromosome 22q12-13 (GenBank accession number Z83733) and a neighbouring clone (AL022313) contained exons 3 and 4. A genomic clone (AC004125) from 16p12.1 contained all four exons of CACNG3, which was also represented by ESTs (GenBank accession numbers H04905, H11833 and W29095) from an infant and adult brain. Exons 2, 3 and 4 of CACNG4 were present in a genomic clone (AC005544) from chromosome 17q24.3. CACNG4 was also represented by an EST (AA970202) from the kidney.

Isolation of γ subunit cDNAs

The first exon of the human CACNG4 gene was not identified in the available public genomic sequence, and the predicted cDNA sequence was instead completed by two sequential 5′ RACE experiments. The first RACE reaction yielded a clone from cerebral cortex cDNA, and the second a clone from hippocampus cDNA, which in composite extended the known sequence by 1 kb. Comparison with the mouse CACNG2 cDNA sequence, and the presence of stop codons in all three reading frames in the 815 nucleotides upstream of the putative initiation codon, which complied with Kozak's rules, indicated that the start of the coding region had been found.

The full-length coding regions of all three human γ subunit cDNAs were cloned by RT-PCR from brain poly-A⁺ mRNA and primers derived from the predicted cDNA sequences as described above. Sequence results were identical to those deposited in GenBank (see accession nos AF096322, AF100346 and AF142625 for γ₂, γ₃ and γ₄, respectively). In Fig. 1 the putative translations of the human exons are shown in comparison with the mouse stargazin protein. The three human γ subunits have identities of 97 % for CACNG2, 75 % for CACNG3 and 62 % for CACNG4, when compared to the mouse stargazin protein. However, all of these neuronal γ subunits share only approximately 20 % amino acid identity to the original γ₁ subunit, but the predicted membrane topologies of all the subunits are highly conserved suggesting strong selective pressure on these regions of the structure (not shown). The predicted structure for all of the γ subunits is for four transmembrane domains with both N- and C-termini located intracellularly (see Burgess et al. 1999 for more details).

Figure 1.

Amino acid sequence alignments of human CACNG2 (γ₂), CACNG3 (γ₃) and CACNG4 (γ₄) against mouse stargazin (MMCACNG2).

TaqMan analysis of γ subunit distribution

γ₂ was found to be highly expressed in the cerebellum, with some expression in other brain regions such as the cortex, hippocampus and nucleus accumbens (Fig. 2a). In contrast to γ₂, γ₃ exhibited almost no expression in the cerebellum, but was more highly expressed in other brain regions such as the nucleus accumbens, amygdala, hippocampus and cortex (Fig. 2B). There was no or very little expression of either γ₂ or γ₃ in the periphery, apart from a low level of expression in the testes. γ₄ like γ₃ had very low levels of expression in the cerebellum and seemed to be expressed at low levels in most brain regions tested, but was very highly expressed in the fetal brain (Fig. 2C). γ₄, in contrast to the other two subunits, had much wider peripheral expression particularly in the lungs and prostate.

Figure 2. TaqMan RT-PCR analysis of γ subunits.

TaqMan distribution profiles of γ₂ (A), γ₃ (B) and γ₄ (C) are shown in CNS and peripheral tissues. Units are presented as arbitrary fluorescence units and are normalised to the housekeeping gene GAPDH. Measurements are from three independent samples and are means ±s.e.m. Samples of whole brain, heart, liver and lung are present on each plate tested to control for plate to plate variation (for detailed explanation, see Medhurs Talley et al. 2000).

Localisation of tagged γ subunits

The live staining pattern of the extracellularly myc-tagged γ₂ subunit in HEK 293 cells stably expressing rat α_1I T-type Ca²⁺ channel, showed that γ₂ was expressed on the cell surface (Fig. 3a), with an increase in intracellular staining evident when the cells were permeabilised (Fig. 3B). When the intracellularly myc-tagged γ₂ was used, there was no staining present on the cell surface until the cells were permeabilised, when staining was evident throughout the cell (Fig. 3C and D). Similar staining patterns were seen for both constructs when transfected into wild-type HEK 293 cells (not shown), thus indicating that the presence of an α₁ subunit is not required for γ₂ to reach the cell surface.

Figure 3. Extracellular and intracellular staining of myc-tagged γ₂ subunits expressed in α_1I HEK 293 cells.

Extracellularly myc-tagged γ₂ subunit staining: A, live, and B, fixed/permeabilised cells stained with anti-myc monoclonal antibodies. Intracellularly myc-tagged γ₂ subunit staining: C, live, and D, fixed/ permeabilised cells stained with anti-myc monoclonal antibodies.

Electrophysiological analysis of functional effects of VSCC γ subunit expression

In previous studies it has been reported that co-expression of the murine γ₂ and γ₄ subunits altered the inactivation curve of α_1A HVA VSCCs whereas γ₅ accelerated both the activation and inactivation of the LVA channel encoded by α_1G (Letts et al. 1998; Klugbauer et al. 2000). In this study we chose instead to investigate if the human γ subunits we had isolated would interact with LVA VSCCs, which are highly prevalent in brain, encoded by the α_1I subunit, a channel which had not previously been analysed for potential interactions with γ subunits.

Our experiments used transient transfection of human γ₂, γ₃ or γ₄ into HEK 293 cells stably expressing rat α_1I. A plasmid bearing a GFP gene was used as a transfection marker. To generate an appropriate control population, some cells were transfected solely with the GFP-encoding plasmid. LVA Ca²⁺ currents in the presence of each γ subunit paralogue were assayed with a panel of standard voltage-clamp protocols. From these we derived measurements of current amplitude as well as activation, inactivation and deactivation behaviour.

Figure 4a illustrates typical sweeps extracted from current-voltage series recorded from cells expressing either GFP alone or GFP plus one of the three γ subunits. Analysis of such data produced the mean current-voltage relationships shown in Fig. 4B. This illustrates that none of the γ subunits examined produced large changes in the current amplitude, although γ₂ stood out as eliciting a small but non-significant apparent increase in current amplitude. Normalisation of these curves to the maximum inward current observed in each recording produced the current-voltage relationships shown in Fig. 4C. These indicate that none of the γ subunits tested here produced a detectable change in the voltage dependence of activation of α_1I, which remained classically LVA in both the absence and presence of γ subunits.

Figure 4. Electrophysiological analysis of γ subunit interaction with a LVA Ca²⁺ channel.

A, typical traces extracted from current-voltage series recorded from HEK 293 cells stably expressing the rat α_1I Ca²⁺ channel subunit. The cells were additionally transiently transfected with GFP (top left), γ₂ and GFP (top right), γ₃ and GFP (bottom left), and γ₄ and GFP (bottom right). The test pulses were applied from a holding potential of -80 mV and were of 400 ms duration. The traces shown were elicited by test potentials of -60 to +20 mV in 10 mV increments. B, pooled peak current-voltage relationships from large numbers of recordings like those in A. C, a peak-normalised version of the data in B, illustrating the lack of effect of γ subunits on the voltage dependence of current activation. D, a graph plotting the 10-90 % rise time of the currents elicited at test potentials between -50 and +20 mV. For panels B-D data are from cells transiently expressing GFP alone or GFP plus γ₂, γ₃ or γ₄.

The LVA Ca²⁺ currents produced by expression of the α_1I subunit exhibit rather slow activation kinetics when compared with most other voltage-gated Ca²⁺ channels, including other recombinant T-type channels (Klockner et al. 1999; Lee et al. 1999). We examined if these slow kinetics were altered in any way by co-expression of γ subunits. Similar to the lack of effect on the voltage dependence of α_1I current activation (Fig. 4C), we failed to detect any significant γ subunit-induced changes in α_1I-mediated current activation kinetics. This was the case at all test potentials between -50 and +20 mV (Fig. 4D).

Both the γ₂ and γ₄ subunits have been reported to alter the steady-state inactivation behaviour of the HVA α_1A VSCC subunit. Figure 5a illustrates sample sweeps extracted from two typical steady-state inactivation experiments on α_1I-expressing cells. In this experiment cells co-expressing either GFP alone or GFP plus one of the γ subunits were studied and traces from a typical control and γ₂ subunit co-expression experiment are shown. Analysis of such experiments including the mean mid-points of inactivation for each experimental population are shown in Fig. 5B. This indicates that the steady-state inactivation of the α_1I-mediated T-type channel is another property of the channel not markedly affected by γ subunit expression. Similarly the rate of α_1I-mediated current inactivation at +15 mV was also clearly unaltered by γ subunit expression (Fig. 5C and D).

Figure 5. γ subunit expression fails to modulate inactivation of α_1I-mediated Ca²⁺ currents.

A, typical traces extracted from steady-state inactivation protocol where cells were subjected to a 50 ms depolarising pulse to 0 mV applied from a range of different holding potentials between -120 and -50 mV (in 5 mV increments). Prior to the test depolarisation the cell was clamped at a different holding potential for at least 5 s. Data are shown for α_1I-expressing HEK 293 cells transiently transfected with GFP alone or GFP plus the γ₂ subunit. Identical experiments were also carried out for cells expressing GFP plus either γ₃ or γ₄ (not shown). B, pooled steady-state inactivation curves recorded from cells expressing GFP alone or GFP plus a γ subunit from experiments similar to those shown in A. The inactivation curve from each individual experiment included in the pooled data sets was fitted with a standard Boltzmann function and a bar graph plotting the mean mid-points of steady-state inactivation and the corresponding standard errors is shown as an inset. The data shown are for cells transiently expressing GFP alone, and GFP plus γ₂, γ₃ and γ₄. C, average peak-normalised current responses to maintained depolarisations to -15 mV applied from a holding potential of -80 mV and traces for γ₂ are shown. D, a graph showing the mean time constants of inactivation at a test potential of -15 mV; analysis is compiled from recordings such as those shown in C. Data for control (GFP only) and γ₂-, γ₃- and γ₄-transfected cells are shown, but appear indistinguishable.

In contrast to the absence of substantial changes to current activation and inactivation, the deactivation of the α_1I T-type Ca²⁺ channel was significantly slowed in the presence of the γ₂ subunit. In Fig. 6a our standard deactivation voltage protocol and the currents it elicited in typical control and γ₂ co-expressing cells can be seen. Figure 6B plots the relationship between repolarisation potential and deactivation time constant for a number of such experiments, plus those using γ₃ and γ₄ subunits. It is clear that γ₂ significantly slowed deactivation at all voltages between -120 and -60 mV. In contrast, no such effects on deactivation of α_1I were observed with co-expression of either γ₃ and γ₄ (Fig. 6B), suggesting that γ₂ may be a specific partner for the α_1I channel.

DISCUSSION

The main features of the data presented here are the cloning of three human VSCC γ subunits, the determination of their respective expression patterns in human tissues and the demonstration of an interaction of one of their number, γ₂, with a recombinant T-type Ca²⁺ channel that is expressed mainly in the brain.

The tissue distribution of the mRNA of the γ subunits in humans clearly shows that both γ₂ and γ₃ are predominantly localised to the brain, with very little peripheral expression, whereas γ₄, although also highly expressed in the brain, exhibits significant peripheral expression. This mRNA distribution pattern in humans appears to be very similar to that recently reported in the mouse (Klugbauer et al. 2000). Although all three γ subunits are expressed in the brain they have quite distinct distributions within different brain regions. Notably, there is a very high level of expression of γ₂ in the cerebellum, whereas γ₃ and γ₄ exhibit relatively low levels of expression in this area.

As a result of its high levels of expression in the cerebellum, mutations in the γ₂ subunit may have a more significant effect in this tissue, and indeed may relate to certain phenotypic facets of the stargazer mutation, for which loss of γ₂ is responsible (Letts et al. 1998). Cerebellar changes associated with the stargazer mutation, and therefore presumably reflecting the lack of γ₂, include large alterations in brain-derived neurotrophic factor (BDNF) levels and a substantially altered expression of the α₆ subunit of the GABA_A receptor (Qiao et al. 1996; Thompson et al. 1998), a protein with a highly cerebellar-specific expression pattern.

A second striking feature of the γ subunit distribution pattern is the high level of γ₄ expressed in fetal brain. It has been shown that both the α_1A HVA subunit (Vigues et al. 2000) and other auxiliary subunits of HVA VSCCs, the β subunits, are regulated during development (Vance et al. 1998). It is possible that the high level of γ₄ subunit expression in fetal brain seen in this study reflects a potential role for this subunit during development.

Comparison of the mRNA expression patterns of the T-type Ca²⁺ channels with the γ subunits described here shows that α_1G and α_1I, like γ₂ and γ₃, are predominantly expressed in the brain, and have very low levels of peripheral expression. In contrast α_1H, like γ₄, has significant peripheral expression (Talley et al. 1999; M. Pangalos & A. D. Medhurst, unpublished results), but to date, γ₄ has only been examined against predominantly CNS-located VSCCs, such as α_1A, α_1G (Klugbauer et al. 2000) and α_1I channels (this study). A modulatory effect of γ₅ was shown recently on α_1G T-type VSCCs (Klugbauer et al. 2000). However, given the minimal overlap in expression patterns of these two subunits (i.e. α_1G is predominantly expressed in the brain whereas γ₅ appears to be in the liver, heart, lungs, kidneys and skeletal muscles), it is interesting to speculate as to whether this fifth γ subunit may instead be an accessory subunit for the α_1H T-type VSCC, which has broadly similar biophysical properties to α_1G and also exhibits a widespread peripheral expression pattern.

Using both intracellularly and extracellularly tagged γ₂ subunits, we demonstrated that a significant proportion of the γ₂ subunit pool seems capable of reaching the plasma membrane of HEK 293 cells. This was the case irrespective of the presence or absence of the α_1I VSCC subunit. Such trafficking to the extracellular surface of the cell is a clear requirement for a protein that produces direct functional alterations in VSCC activity. Furthermore, by confirming that the C-terminus is intracellular and the area between the predicted first and second transmembrane domains is extracellular, these tagging experiments provide initial evidence for the correctness of the sequence-based predictions of γ subunit topology.

Our data clearly demonstrate that the γ₂ VSCC subunit has the ability to modify the gating properties of the T-type channel encoded by the α_1I subunit. Specifically, we observed a considerable slowing of channel deactivation (i.e. closing) across a wide range of membrane potentials in the presence of γ₂ but not γ₃ or γ₄ subunits. This is the first demonstration of any of the auxiliary Ca²⁺ channel subunits having an effect on the α_1I T-type Ca²⁺ channel. It also shows that the γ subunit interactions are not restricted to changing activation and steady-state inactivation properties of Ca²⁺ channels (Letts et al. 1998; Klugbauer et al. 2000), but clearly can also alter the rate of deactivation.

A slowed deactivation of α_1I might be expected to produce some enhancement in current amplitude during a test depolarisation. Although there was a trend in this direction it failed to quite reach statistical significance in our study (Fig. 4B), and this may in part reflect the variation in current amplitude from preparation to preparation. In some individual transfections that contributed to the total data set presented, γ₂ co-expression resulted in current amplitudes that were significantly enhanced with respect to control GFP-transfected cells. This parallels data from the original skeletal muscle γ subunit (i.e. γ₁) which was also reported to slightly increase the amplitude of α_1S-mediated currents (Wei et al. 1991; Singer et al. 1991).

Even in the absence of γ subunits, T-type channel deactivation is slow compared to that of HVA VSCCs. Along with their distinctive activation properties, this facet of the channel's biophysical profile is thought to be crucial in determining the form of Ca²⁺ transients that T-type channels deliver to cells (McCobb & Beam, 1991; Randall & Tsien, 1997). An additional slowing of deactivation by γ₂ could further exaggerate these marked differences between the Ca²⁺ delivery patterns elicited by HVA and LVA channel activity.

Work from our laboratory (Warre & Randall, 2000) has recently demonstrated an additional mechanism by which α_1I-mediated channel deactivation can be modified. This recent study shows that the induction of inactivation considerably slows T-type channel deactivation. Thus, interactions that enhance the rate or extent of α_1I inactivation are also predicted to slow deactivation. This novel mechanism, however, does not seem to be responsible for the effects of the γ₂ subunit reported here since the observed slowing of deactivation (Fig. 6) was not accompanied by changes to channel inactivation behaviour (Fig. 5).

A number of the newly identified γ subunits have been linked to the pathophysiology of epilepsy (Letts et al. 1998; Black & Lennon, 1999; Burgess et al. 1999), particularly that typified by absence seizures. T-type VSCCs have also long been implicated in this condition (Talley et al. 2000). It is interesting, therefore, to speculate what the significance of the interactions between γ₂ and α_1I shown here may mean in vivo. α_1I channels are strongly expressed by neurones of the reticular thalamic nucleus (RTN; Talley et al. 1999), cells which are believed to be involved in a circuit that sustains the thalamocortical oscillatory burst firing in absence seizures (see review by Futatsugi & Riviello, 1998). γ₂ mRNA is also expressed in the thalamus (Klugbauer et al. 2000, this study), although its localisation within individual nuclei is currently not described.

All other things being equal, the slowed deactivation (and possibly slightly larger currents) of the α_1I T-type VSCC associated with the γ₂ subunit would be predicted to result in a net increase in neuronal excitation if compared with cells lacking γ₂. Hence, co-expression of α_1I and γ₂ on the GABAergic neurones of the RTN may result in an increased activity of these inhibitory neurones, which in turn would enhance dampening of the thalamocortical circuit, thus producing an anti-seizure outcome. From this it follows that in the absence of the γ₂ subunit decreased excitability of these inhibitory neurones may result, thus leading to an epileptic phenotype, such as is observed in the stargazer mouse.

We feel that the demonstration of an interaction between a T-type VSCC expressed by the RTN and the γ₂ subunit is an exciting step forward into understanding the role these channels may have in such pathologies. It will be interesting to see if the levels of either the other γ subunits or any of the other T-type Ca²⁺ channels are modulated in this and other models of epilepsy, thus paralleling the changes in expression of the β_1b, β₄ and the HVA α_1B VSCC reported in the epileptic lethargic mouse (McEnery et al. 1998).