Specific Properties of T-type Calcium Channels Generated by the Human a 1I Subunit*

, We have cloned and expressed a human a 1I subunit that encodes a subtype of T-type calcium channels. The predicted protein is 95% homologous to its rat counter-part but has a distinct COOH-terminal region. Its mRNA is detected almost exclusively in the human brain, as well as in adrenal and thyroid glands. Calcium currents generated by the functional expression of human a 1I and a 1G subunits in HEK-293 cells were compared. The a 1I current activated and inactivated ; 10 mV more pos- itively. Activation and inactivation kinetics were up to six times slower, while deactivation kinetics was faster and showed little voltage dependence. A slower recovery from inactivation, a lower sensitivity to Ni 2 1 ions (IC 50 ; 180 m M ), and a larger channel conductance ( ; 11 picosiemens) were the other discriminative features of the a 1I current. These data demonstrate that the a 1I subunit encodes T-type Ca 2 1 channels functionally distinct from those generated by the human a 1G or a 1H subunits and point out that human and rat a 1I subunits have species-specific properties not only in their primary sequence, but also in their expression profile and electrophysiological behavior.

Voltage-dependent calcium channels control the rapid entry of Ca 2ϩ ions into a wide variety of cell types and are therefore involved in both electrical and cellular signaling. Electrophysiological studies have identified two major Ca 2ϩ channel types as high voltage-activated and low voltage-activated channels (1,2) with this latter class being also identified as T-type Ca 2ϩ channels (3). T-type Ca 2ϩ channels were originally defined by their activation at low membrane potential, their fast time course, and their small single channel conductance (4,5). These channels have been identified on a large variety of neurons, and it has become obvious that significant functional diversity exists in the gating behavior of T-type channels, particularly in inactivation kinetics, voltage dependence of steady-state inactivation, and pharmacology (6). The recent identification of several novel genes encoding a subset of homologous Ca 2ϩ channel ␣ 1 subunits, e.g. the ␣ 1G subunit (7)(8)(9), the ␣ 1H subunit (10,11), and the rat ␣ 1I subunit (12), has revealed that diversity of T-type voltage-dependent calcium channels is primarily related to the expression of distinct ␣ 1 subunits. Indeed, the expression of the various ␣ 1G and ␣ 1H subunits (7)(8)(9)(10)(11)(12) produces Ca 2ϩ currents with the typical signature of T-type channels but with specific features, such as the block by Ni 2ϩ , which discriminates between ␣ 1G and ␣ 1H currents (13). By contrast, the biophysical properties of T-type channels generated by the ␣ 1I subunit markedly differ from those made of ␣ 1G and ␣ 1H subunits (12,14,15), and it was postulated that the ␣ 1I subunit is responsible for native "slow" T-type currents observed in rat thalamic neurons (16). To date the ␣ 1I subunit has only been cloned from rat (12) and whether the ␣ 1I subunit encodes an atypical T-type Ca 2ϩ channel certainly needs further investigation. We have addressed this issue by characterizing a novel ␣ 1I subunit. We have cloned and functionally expressed a fulllength cDNA encoding the human ␣ 1I subunit. The properties of the Ca 2ϩ currents generated by the human ␣ 1I subunit were compared with a neuronal isoform of the human ␣ 1G subunit cloned recently, ␣ 1G-a (9). The data also described several species-specific properties of the human ␣ 1I channels.

MATERIALS AND METHODS
Cloning of a Complete cDNA for Human ␣ 1I Subunit-An initial search of genetic data bases for cDNAs weakly homologous to ␣ 1 subunit cDNAs encoding high voltage-activated channels (9) identified several sequences homologous to the Caenorhabditis elegans gene c54d2.5 (GenBank TM accession number U37548), including three expressed sequence-tagged clones, H55225, H55223, and H55617 from human chromosome 22 (17). In the mean time, a genomic sequence covering human chromosome 22q12.3-q13.2 (GenBank TM accession number AL008716) was sequenced and released by the Sanger Center (Cambridge, United Kingdom), and more recently, a cDNA encoding the rat ␣ 1I subunit was described by the group of Perez-Reyes (Ref. 12, GenBank TM accession number AF086827). The partial intron/exon structure of the CACNA1I gene encoding the human ␣ 1I subunit was determined using the GRAIL software (18). Alignment of the rat cDNA with GenBank TM accession number AF086827 was then performed to identify more precisely the putative human ␣ 1I cDNA coding region and to design PCR 1 primers for cDNA cloning. Pairs of primers were designed to amplify the entire coding region of the ␣ 1I subunit cDNA in five partial fragments designed PCR-I1 to PCR-I5 (PCR-I1: sense, 5Ј-TCAGTGTGGACATGGCTGAG-3Ј; antisense, 5Ј-CCATTATCCCATT-GTCGCCC-3Ј; PCR-I2: sense, 5Ј-ATGGAGCTGATCCTCATGTCCC-3Ј;  antisense, 5Ј-AGTACTTGCTGTCCACGATGCC-3Ј; PCR-I3: sense, 5Ј-CCTCCCCTGGAAATGATCAC-3Ј; antisense, 5Ј-TGAGCAGGAAG-GAGATGAAG-3Ј; PCR-I4: sense, 5Ј-CACCCGCAACATCACCAAC-3Ј;  antisense, 5Ј-TGGGCAGGCGAGTAGCAG-3Ј; and PCR-I5: sense, 5Ј-CCCATCAATCCCACCATCATCC-3Ј; antisens, 5Ј-TTAGATCCTGC-CCCTTGCCC-3Ј, see Fig. 1A). Reverse transcription (RT)-PCR protocols were performed using mRNA from human total brain * This work was supported in part by the Program Génome du CNRS, Association pour la Recherche contre le Cancer (number ARC9011), Association Française contre les Myopathies (AFM). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF211189.
(CLONTECH). Nucleotide sequences were determined using automatic sequencing (Applied Biosystems) using dye terminator. A full-length cDNA encoding the human ␣ 1I subunit was constructed using the restriction sites mentioned in Fig. 1A and subsequently subcloned into the mammalian expression vector pBK-CMV (Stratagene). Sequence comparisons of this ␣ 1I subunit with a subset of T-type ␣ 1 subunits was performed with BESTFIT (Genetics Computer Group) multialignment software.
Northern and Dot Blot Analyses-Commercial human Northern and dot blot membranes (CLONTECH) were hybridized using a 321-bp fragment (nt 256 to nt 577) generated by RT-PCR amplification (sense, 5Ј-ATGCTGGTGATCCTGCTGAAC-3Ј; antisense, 5Ј-GCACGCGGTT-GATGGCTTTGAG-3Ј) from human brain mRNA (CLONTECH) and random-primed with [␣ 32 P]dCTP. The membranes were treated according to the manufacturer's protocol, as reported previously (9). The exposure time for Northern and dot blot membranes was 6 days at Ϫ80°C. Densitometric analysis of the autoradiograms was performed using an alphaimager system (Alpha Innotech Corp.) to provide semiquantitation of ␣ 1I mRNA in each condition and comparison with the ␣ 1G mRNA signal (9). No signal was observed for internal DNA controls.
Transient Expression Experiments and Electrophysiology-Transfection experiments were performed with the pBK-CMV construct that encodes for the ␣ 1I subunit, cotransfected with a plasmid encoding the reporter gene CD8 or GFP, in a 10:1 ratio. Human embryonic kidney cells (HEK-293) were grown in Dulbecco's modified Eagle's medium (Eurobio) supplemented with 10% fetal bovine serum and 1% penicillin/ streptomycin (v/v). Cells were transfected using a calcium phosphate transfection procedure, and the positives were identified with anti-CD8 antibody-coated beads (Dynal). Whole cell currents were recorded using the patch clamp technique at room temperature (ϳ21°C) with an Axopatch 200B amplifier. Data were acquired on a PC computer using the pCLAMP6 software (Axon Instruments). Records were filtered at 5 kHz. Leak and capacitive currents were subtracted using a P/Ϫ5 procedure when needed. Extracellular solution contained (in mM): 2 CaCl 2 , 160 TEACl, 10 HEPES (pH to 7.4 with TEAOH). Pipettes made of borosilicate glass with a typical resistance of 1-2 M⍀ were filled with a solution containing (in mM): 110 CsCl, 10 EGTA, 10 HEPES, 3 Mg-ATP, 0.6 GTP (pH to 7.2 with CsOH). Mibefradil and Ni 2ϩ solutions were prepared daily in the external medium, freshly for mibefradil (a gift from Roche, Basel, Switzerland) and from a stock solution for Ni 2ϩ (NiCl 2 ), and the various dilutions were applied to cells by a gravity driven perfusion system. Single channel recordings were done in the cell-attached mode using sylgard-coated pipettes (resistance of 7-15 M⍀) filled with a solution containing (in mM): 100 BaCl 2 , 10 HEPES (pH to 7.2 with TEAOH). Membrane potential was reduced toward 0 mV by bathing the cells in a high potassium solution containing (in mM): 140 potassium gluconate, 10 EGTA, 10 glucose, 1 MgCl 2 , 10 HEPES (pH to 7.3 with KOH). The sampling frequency for acquisition was 10 kHz, and data were filtered at 1 kHz. Data were analyzed using pCLAMP6, Excel (Microsoft), and GraphPad Prism (GraphPad Inc.) software. Activation and inactivation curves were fitted with a Boltzmann equation. Apparent dissociation constants (K d ), obtained from the fitting of the drug dose-response curves using a sigmoidal function, were presented as the concentration responsible for 50% of block of the currents (IC 50 ). Results are presented as the mean Ϯ S.E., where n is the number of cells used.

RESULTS AND DISCUSSION
A putative cDNA encoding the human ␣ 1I subunit was predicted from the genomic sequence of chromosome 22q12.3-q13.2 (GenBank TM accession number AL008716), and the partial intron/exon structure of the human gene encoding the ␣ 1I subunit, CACNA1I, was identified (19,20). Five pairs of primers were designed to clone overlapping cDNA fragments that covered its complete coding sequence. The five partial cDNA clones were obtained by using RT-PCR with human total brain mRNA and subsequently sequenced and assembled (Fig. 1A). Finally, the full-length cDNA encoding the ␣ 1I subunit was sequenced (GenBank TM accession number AF211189) and subcloned into the pBK-CMV expression vector. The human ␣ 1I subunit has an open reading frame of 5943 bp and encodes a protein of 1981 aa (calculated molecular mass of 220,747 Da). It is highly homologous to the rat ␣ 1I subunit (GenBank TM accession number AF086827 (12)), as well as to the human ␣ 1G (GenBank TM accession number AF126965 (9)) and ␣ 1H (Gen-Bank TM accession number AF051946 (10); GenBank TM accession number AF073931 (11)) subunits. Considering the complete aa sequence, the human ␣ 1I subunit is 70 -75% homologous to ␣ 1G and ␣ 1H subunits, with a 85-90% homology for the transmembrane domains (Fig. 1B). The three human T-type ␣ 1 subunits share a similar percentage of homology (ϳ63%) with their putative ortholog in C. elegans, C54D2.5 (GenBank TM accession number U37548). Expectedly, the human and rat ␣ 1I subunits are highly homologous with a 95.4% of homology calculated for the complete sequence. This percentage, however, does not consider the main sequence variations that occur in the COOH-terminal region (Fig. 1C). First, a proline/cysteine-rich region identified in the rat ␣ 1I protein (aa The construction of the full-length cDNA was realized using the restriction sites BamHI, EagI, BsrGI, and HindIII. B, percentage of homology of the human ␣ 1I subunit (GenBank TM accession number AF211189 (Hum. ␣ 1I )) with the human ␣ 1G subunit (GenBank TM accession number AF126965 (Hum. ␣ 1G )), the human ␣ 1H subunit (GenBank TM accession number AF073931 (Hum. ␣ 1H )), the rat ␣ 1I subunit (GenBank TM accession number AF086827 (Rat ␣ 1I )), and the C. elegans C54D2.5 putative protein (GenBank TM accession number U37548) calculated for the complete proteins (black numbers, white boxes) and for the transmembrane domains (white numbers, black boxes). C, alignment of the predicted COOH-terminal regions for the human (aa 1705-1981) and rat (aa 1699 -1835) ␣ 1I proteins. The symbols are: #, insertion; *, conserved aa substitution; :, strongly similar aa substitution; ., weakly similar aa substitution. The stretch of cysteine (C), proline (P), glycine (G) described for the rat sequence (aa 1720 -1737) is not found in the human protein. A corrected rat sequence (Rat ␣ 1I (*)) obtained following the introduction of two frameshifts at nt 5437 and 5615 (छ) is proposed. In this corrected rat sequence a stop codon (X) is found 115 bp upstream to the predicted end. our cDNA cloning experiments and from the analysis of the human genomic sequence. Second, the rat ␣ 1I subunit was described with a rather short COOH-terminal region, as compared with other T-type ␣ 1 subunits (12). Here we describe that COOH-terminal region of the human ␣ 1I subunit is 158 aa longer. The stop codon identified by Lee et al. (12) is likely to shorten prematurely the rat ␣ 1I protein, since pairwise alignment of the human and rat cDNAs can predict a novel 3Ј-end for the rat ␣ 1I cDNA sequence, which is highly homologous to the human cDNA. The corrected rat sequence was obtained by introduction of two frameshifts at nt 5437 and 5615 (Fig. 1C). However, the presence of a stop codon ϳ106 bp upstream to the putative end of this corrected rat sequence most likely indicates that cDNA cloning and/or sequencing of the 3Ј-end sequence of the rat ␣ 1I cDNA should be performed again before further investigations.
Northern blot and dot blot analyses of a large variety of human tissues showed that ␣ 1I mRNA is almost exclusively present in the brain (Fig. 2). The ␣ 1I mRNA was found as a single band of approximately 10.5 kb (Fig. 2, A and B). By contrast, the rat ␣ 1I mRNA was detected as at least two bands (10.5 and 8 kb) in several tissues, including brain, kidney, and liver (12). A complementary analysis of the ␣ 1I mRNA expression profile was performed using a dot blot membrane, which confirmed the lack of ␣ 1I expression in human peripheral tissues with the exception of adrenal and thyroid glands (Fig. 2C). In addition, it is worth noting that in brain the intensity for ␣ 1I signal in fetal tissue is ϳ25% higher than for adult brain (compare A1 and G1; Fig. 2, C and D). Such a difference was not retrieved for the human ␣ 1G mRNA signal in identical experimental conditions (9).
Expression of human ␣ 1I channels was performed in HEK-293 cells, and the properties of the resulting current (␣ 1I current) were compared with those of the current generated by the human ␣ 1G-a subunit, in the presence of 2 mM Ca 2ϩ (Fig. 3). Isoform "a" is the major isoform of the ␣ 1G subunit expressed in the human brain and differs from the ␣ 1G-b isoform described in Monteil et al. (9) in its cytoplasmic III-IV loop. Superimposition of ␣ 1I current traces obtained for depolarizing pulses from Ϫ90 to ϩ50 mV (holding potential Ϫ110 mV) revealed an activation at low voltages (Ϫ60 mV) and a crossing over of the current traces (Fig. 3A) that are typical features of T-type currents (21). Peak of the normalized current-voltage curve for ␣ 1I current (n ϭ 12) was obtained near Ϫ24 mV, while for ␣ 1G current, the peak of the normalized current-voltage relationships was 12 mV more negative ( Fig. 3B; Table I). The conductance-voltage curves (Fig. 3C) were then deduced from each current-voltage relationships to estimate the activation parameters. For ␣ 1I and ␣ 1G currents, the potentials for half-maximal activation (V 0.5 ) were ϳϪ40 mV and ϳϪ51 mV, respectively (Table I). Steady-state inactivation properties for ␣ 1I and ␣ 1G currents (Fig. 3C) were determined by potentials for half-inactivation (V 0.5 ) of ϳϪ69 mV for ␣ 1I currents (n ϭ 8) and ϳϪ75 mV for ␣ 1G currents (n ϭ 5) that were significantly different (Table I). For ␣ 1I current, similar steady-state inactivation properties were obtained (V 0.5 ϳϪ70 mV, n ϭ 5) when using a 15-s prepulse duration. These data demonstrate that the human ␣ 1I subunit generate channels with significant differences in steady-state activation and inactivation properties as compared with the channels made of ␣ 1G subunit, with activation and inactivation V 0.5 values more positive by at least 11 and 5 mV, respectively. As a consequence of its steady-state activation and inactivation properties, the ␣ 1I subunit can generate a window current at membrane potentials in the range of Ϫ60/ Ϫ50 mV, i.e. ϳ10 mV more positive than the one related to the ␣ 1G-a subunit (Ϫ70/Ϫ55 mV). Considering the expression profile of the ␣ 1I subunit, this channel could therefore regulate intracellular Ca 2ϩ concentrations at resting membrane potential in endocrine tissues where it could play a role in hormone secretion (22), as well as in neuronal tissues for the induction of burst firing (5).
voltage-dependent and was described by a single exponential function with time constants ranging from 273 to 95 ms for potentials between Ϫ50 and ϩ30 mV (Fig. 3F). Inactivation kinetics of ␣ 1I current was ϳ6-fold slower than for ␣ 1G current at membrane potentials up to Ϫ30 mV. The difference was even more pronounced at lower voltages (inactivation kinetics of ␣ 1I current ϳ13-fold slower at Ϫ50 mV). The recovery from fast inactivation was determined using a double pulse protocol that comprises an inactivating pulse followed by a test pulse at variable interpulse durations. Plot of the relative peak current amplitude as the function of the interpulse duration (Fig. 3G) was described by a single exponential for ␣ 1I current ( ϭ 297 Ϯ 43 ms, n ϭ 11) but not for ␣ 1G current, which was best fitted by a two exponential function (1 ϭ 56 Ϯ 5 ms (64%) and 2 ϭ 238 Ϯ 32 ms (36%)). A distinctive feature between T-type currents is their slow deactivation kinetics (21). For ␣ 1I current, however, deactivation was fast (ϳ3-fold) as compared with ␣ 1G current and showed little voltage dependence (Fig.  3H). In addition, we have also found that the sensitivity to Ni 2ϩ and mibefradil was significantly different between human ␣ 1I and ␣ 1G channels (Table I). Dose-responses curves for mibefradil indicated IC 50 values of 2.3 and 1 M for ␣ 1I and ␣ 1G currents, respectively. Similarly, ␣ 1I current was blocked by higher concentration of Ni 2ϩ , as compared with ␣ 1G current (IC 50 of 184 and 133 M, respectively; Fig. 3I). Finally, single channel currents measurements performed as indicated under "Material and Methods" (Fig. 3J) revealed a slope conductance of 11.2 pS (Fig. 3K) with a current amplitude at 0 mV of Ϫ0.24 Ϯ 0.04 pA (n ϭ 3) typical of that obtained for recombinant and native T-type Ca 2ϩ channels.
Altogether, the detailed functional analysis performed in this study by comparing the channels generated by the human ␣ 1I and ␣ 1G subunits as expressed in HEK-293 cells provide compelling evidence that the diversity of T-type channel activity in native neurons (6) is primarily related to the expression of distinct subunits. Although these two channel proteins are highly homologous and do not display striking differences in their primary sequence, we have demonstrated here that they produce distinct patterns of Ca 2ϩ currents with specific signatures. Molecular identification of the human ␣ 1I and ␣ 1G subunits that generate currents with significant difference in their activation and inactivation kinetics and steady-state properties, as well as in their deactivation kinetics and recovery from inactivation, should offer the opportunity to identify the molecular determinant(s) responsible for the hallmark properties of T-type Ca 2ϩ channels.
Our study also suggests that there are functional differences between the currents generated by the human and rat ␣ 1I FIG. 3. Electrophysiological properties of the ␣ 1I currents, as compared with the ␣ 1G currents. A, current traces evoked by increasing depolarizations from Ϫ90 mV to ϩ50 mV (holding potential Ϫ110 mV) on a HEK-293 cells transfected with the human ␣ 1I -pBKCMV construct. The peak current was obtained for a test pulse to Ϫ25 mV (arrow). Note the criss-crossing pattern of current traces. B, averaged current/voltage relationships obtained for ␣ 1I (squares) and ␣ 1G-a currents (circles). C, average steady-state inactivation curves (hϱ) for ␣ 1I (squares) and ␣ 1G currents (circles) were obtained from the measurements of the current amplitude at Ϫ30 mV, which allows measurement of the channel availability, after a 5-s predepolarization pulse of increasing amplitude (Ϫ110 to Ϫ35 mV, 5-mV increment). The mean steady-state inactivation curves (filled symbols) were best-fitted with a single Boltzmann distribution with a half-inactivation potential values (V 0.5 ) reported in Table I. The conductance-voltage relationship for activation was calculated from each current-voltage curve, and the mean steady-state activation curves (mϱ) were fitted with a Boltzmann distribution (smooth curves) with parameters reported in Table I. Superimposition of steady-state activation (mϱ) and inactivation (hϱ) curves revealed a window current in the range of Ϫ60 to Ϫ50 mV for ␣ 1I current and Ϫ70 to Ϫ55 mV for ␣ 1G current. D, superimposed ␣ 1I and ␣ 1G current traces (test pulses: Ϫ25 and Ϫ35 mV, respectively), recorded on a 500-ms time scale, showed their differences in activation and inactivation kinetics. Vertical scale bars (250 pA) refer to ␣ 1G current (regular) and ␣ 1I current (bold). E, kinetics of activation is presented as a plot of the 10 -90% rise time (rise 10 -90) as a function of the test potential. F, kinetics of inactivation is illustrated by plotting the time constant () of the current decay, as a function of the test potential. G, recovery from fast inactivation is estimated from the fit (exponential function) of the normalized percentage of availability from inactivation as a function of the interpulse duration. The values can be retrieved in the Table I. H, for the analysis of the deactivation properties, current traces were elicited by repolarization at Ϫ80 mV after a short test pulse at Ϫ25 mV (35 ms) and Ϫ35 mV (6 ms) for ␣ 1I and ␣ 1G currents, respectively. Vertical scale bars (1 nA) refer to ␣ 1G current (regular) and ␣ 1I current (bold). The deactivation time constant () plotted as a function of the repolarization potential is shown as an inset. I, mean dose-effect curve for Ni 2ϩ of ␣ 1I current (n ϭ 7; IC 50 of 184 Ϯ 23 M) and ␣ 1G current (n ϭ 6; IC 50 of 133 Ϯ 4 M). J, single channel traces obtained for depolarization from Ϫ100 to Ϫ30 mV (upper part) and to Ϫ10 mV (lower part) were leak-substracted using averaged blank sweeps. Scale bars are 20 ms (horizontal) and 1 pA (vertical). K, plot of the unitary current amplitudes obtained at various potentials (Ϫ70 to 10 mV) as a function of the test potential value. Linear regression reveals a slope conductance of 11.2 pS (n ϭ 3).
subunits, although the human subunit retained most of the properties described initially for the rat ␣ 1I subunit (12). It was recently reported that the rat ␣ 1I and ␣ 1G currents have identical steady-state inactivation properties with a potential for half-maximal inactivation (V 0.5 ) of Ϫ72 mV (14). Here we report that the human ␣ 1I and ␣ 1G currents have significantly different V 0.5 values (Ϫ69 and Ϫ75 mV, respectively). It is of course important to consider that the rat and human ␣ 1G subunits used in the present study and in Ref. 14 are highly homologous (␣ 1G-a isoforms; 95% of homology on the complete sequence), while the ␣ 1I subunits are potentially different in their carboxyl terminus (Fig. 1C). Also, while rat ␣ 1I current was more sensitive to the block by Ni 2ϩ than rat ␣ 1G current (IC 50 of 216 and 250 M, respectively; Ref. 13), we found a reverse sequence of sensitivity for the currents generated by the human subunits (␣ 1G Ͼ ␣ 1I ). It will be important to determine precisely whether species specific differences in the proteins can account for functional differences described here.
The specific features described here for the currents generated by the recombinant ␣ 1I and ␣ 1G subunits should be useful for the identification and the discrimination of the corresponding currents in native cells. According to the expression profile of ␣ 1I and ␣ 1G mRNA in rat brain (23), it was suggested that ␣ 1I and ␣ 1G currents could coexist in thalamic structures. Indeed, it was reported that the LVA current in rat thalamic reticular and laterodorsal neurons could be separated into fast and slow components (16,24,25). By contrast, our Northern and dot blot data have not confirmed that human thalamic structures expressed the ␣ 1I mRNA at high level. By comparing ␣ 1I and ␣ 1G mRNA expression profiles in human brain (this study and Ref. 9), we found that the occipital lobe, putamen, and nucleus accumbens are the regions that exhibit rather high level of both ␣ 1I and ␣ 1G transcripts (not shown). Unfortunately, these human brain structures remains poorly explored, and no slow T-type current has been reported in any human neurons to date. Altogether, our data provide important new insights into the molecular properties and the function of T-type Ca 2ϩ channels in humans, suggesting that there are species specificity in the structure, the expression profile, and the function of the Ca 2ϩ channels generated by the ␣ 1I subunit. Functional expression of this human channel also offers a bioassay to investigate in detail the pharmacological profile of this peculiar T-type calcium channel and should help further in the identification and interpretation of the physiological role(s) of ␣ 1I T-type Ca 2ϩ channels in humans.
Interest for the characterization of human ␣ 1I subunit Ca 2ϩ channels is also raised by the identification of several neuronal disorders mapped to human chromosome 22q13. The CACNA1I gene is located in 22q12.3-q13.2, a susceptibility locus for familial schizophrenia (26) and a form of spinocerebellar ataxia, SCA10 (27). Another Ca 2ϩ channel gene, CACNG2, encoding the ␥2 subunit and localized on chromosome 22q13 has been hypothesized as a candidate for SCA10 (28). Similarly, it is tempting to suggest that CACNA1I is a candidate gene for these human diseases and linkage analysis of this gene with human neuronal disorders should be investigated.