 |
INTRODUCTION |
Transmembrane voltage-dependent calcium channels
control Ca2+ ion entry from the extracellular space and
thereby regulate various cellular processes such as muscle contraction,
neuronal development and plasticity, secretion, and gene expression.
Functional identification of voltage-dependent calcium
channel subtypes based on their biophysical properties have
distinguished low voltage-activated
(LVA)1 and high
voltage-activated (HVA) Ca2+ currents. LVA calcium
currents, mainly referred to as "T-type currents" due to their fast
inactivation and small conductance, have been described in a wide
variety of cell types (1). Although their physiological functions still
remain obscure, they are thought to play a role in neuronal burst
firing (2), pacemaker activity in heart (3, 4), aldosterone secretion
(5), or fertilization (6). T-type currents are mainly detected at early
stages of development as well as at a given transition
(G1/S) of the cell cycle (7-9). In addition, it has been
reported that enhanced expression of T-type current is associated with
a variety of diseases that include cardiac hypertrophy (10),
hypertension (11), and epilepsy (12). Unfortunately, the lack of
specific T-type channel blockers has considerably impaired their
functional characterization. The identification of the molecular
structure of T-type channels has therefore become an important
challenge in order to unravel their implication in cellular functions
(13).
The low level of knowledge about the molecular nature of T-type
channels contrasts with the extensive molecular and functional characterization of HVA channels (L-, N-, P/Q-, and R-types) related to
the discovery of specific drugs and toxins. HVA calcium channels exhibit an oligomeric structure composed of a pore-forming
1 subunit containing voltage sensors and drug/toxin
binding sites, associated with regulatory subunits (
,
2/
, and 
). The functional diversity of HVA calcium
channels is primarily related to the existence of several
1 subunits (1A-F and 1S)
encoded by distinct genes, many of which generate splice variants with specific properties, as described for the 1A isoforms
that generate P/Q-type channels (14).
A search in the genetic data bases for sequences homologous, but not
identical, to known Ca2+ channel 1 subunits
was an alternative to conventional molecular cloning strategies for the
identification of novel 1 subunits, and a major advance
in T-type Ca2+ channel studies has resulted from the
identification of several ESTs and genomic sequences corresponding to a
subset of distantly related 1 subunits (15). The
full-length cDNAs encoding three distinct 1 subunits
(1G, 1H, and 1I) have
recently been identified, which give rise, in expression systems, to
Ca2+ channel currents exhibiting hallmarks of native T-type
currents (15-17). The genes that encode mammalian 1G,
1H, and 1I proteins are related to the
Caenorhabditis elegans gene c54d2.5 described as
a putative Ca2+ channel. These recent data (15-18) provide
clear evidence that genes encoding T-type channels are members of a
third subfamily in addition to the gene family encoding neuronal
calcium channels (1A, 1B, and
1E) and the gene family encoding L-type calcium channels
(1C, 1D, 1F, and
1S). An important issue now is to elucidate the
molecular basis for the functional diversity of T-type Ca2+
channels (2).
To address this issue, we have cloned and expressed a human
1G subunit. The molecular data presented here describe
several new findings concerning the 1G subunit, such as
the existence of several splice variants, the partial gene structure,
and the mRNA expression pattern in adult and fetal human tissues.
Functional expression was achieved in order to determine the
electrophysiological and pharmacological profiles of the human
1G-related currents, and we show that the
1G subunit generates a sustained Ca2+
current with specific features. Characterization of the human 1G subunit is an important step that significantly
extends our knowledge of the molecular and functional properties of the
T-type Ca2+ channels.
| |
MATERIALS AND METHODS |
In Silico Strategies and Probe Design--
Genetic data bases
were searched for cDNA and genomic sequences homologous, but not
identical, to known 1 subunit cDNAs using (i)
sequence similarity search (BLAST; Ref. 19) and (ii) a search for
conserved motifs, such as a voltage sensor S4 = ((R/K)XX)4 (BioMotif).2 Among the
various human sequences that were identified, we focused on sequences
homologous to the C. elegans gene c54d2.5
(GenBankTM accession no. U37548). Two EST clones, H06096
and H19230, available from the IMAGE consortium (20), were sequenced
and compared with a large set of calcium channel 1 and
sodium channel cDNAs, using the ClustalW1.7 multialignment
software (21). These cDNA sequences were thus defined as candidates
for putative new calcium channel 1 subunits in humans
and subsequently used as probes for both Northern blot analysis and
screening of cDNA libraries. The genomic region of chromosome 17q22
that contains the CACNA1G gene was recently sequenced
(AC004590). The identification of the intron/exon structure was
performed using the GRAIL software (22) for the search of putative
exons. Alignment of the cDNA sequences characterized in this study,
together with the genomic sequence, led us to identify the exons of the
CACNA1G gene that form 1G transcripts.
Isolation and Characterization of Human cDNAs--
A 
gt10
human cerebellum cDNA library (CLONTECH) was
screened by conventional filter hybridization according to the
manufacturer's protocols, using a probe generated by PCR using H06096
cDNA as template. This probe covers nucleotides 3660-4590 of the
final cDNA. Hybridization was performed at 42 °C for 16-20 h in
a solution containing 50% formamide using a random primed
[-32P]dCTP-labeled probe that was added at a
concentration of 5 × 106 cpm/ml. The membranes were
then washed with a final stringency of 0.1× SSC, 0.1% SDS at
65 °C. Three gt10 clones named CC1/CC2/CC3 were identified. An
additional EST (H19230) corresponding to the 3' region of the messenger
was also identified. The completion of the 5' region of the sequence
was achieved using long reverse transcription-polymerase chain reaction
(RT-PCR) protocols (23) on mRNA from human total brain
(CLONTECH) using primers defined from the rat
sequence (Ref. 15; AF027984) and RACE-PCR (24). CC1 and CC3 are
overlapping clones that correspond to a splice variation of the human
1G. The resulting isoforms were designated 1G-a (AF126966) and 1G-b (AF126965). The
cDNA structure and sequence encoding these subunits was further
identified following the characterization of two overlapping cDNA
fragments of 4056 bp (nt 
120 to 3936) and 3432 bp (nt 2840-6272),
according to AF126965 (see Fig. 1A). Nucleotide sequences
were determined using automatic sequencing (Applied Biosystems) with
the dye terminator strategy. A full-length cDNA encoding the human
1G isoform designated 1G-b was
constructed using the unique restriction sites mentioned in Fig.
1A and subsequently subcloned in the mammalian expression vector pBK-CMV (Stratagene).
Long Range RT-PCR for Native Isoform Identification--
A
search for native isoforms of the 1G subunit was further
performed using long range RT-PCR (23) and sequence examination of the
cDNA fragment covering from nt 2840 (domain II) to nt 6272 (C-terminal region), according to AF126966. The size of this fragment
varied from ~3.4 to ~3.8 kb, as a function of the 1G isoform detected. This analysis was achieved using four different sources of human mRNA: adult whole brain, adult cerebellum, adult thalamus, and fetal kidney (CLONTECH). First strand
cDNA synthesis (reverse transcription) was performed using
Superscript 2 (Life Technologies) with a sequence-specific primer
(5'-GCTTTCTCCCAACAGCTTC-3') according to the manufacturer's protocol.
The enzyme was then heat-inactivated (15 min, 70 °C). After cooling
on ice, the samples were treated with ribonuclease H (Life
Technologies, 4 units) for 20 min at 37 °C. PCR amplification was
performed with 2 µl of the reverse transcription mixture in a final
volume of 50 µl using the Expand Long Template PCR system (Roche
Molecular Biochemicals) with 10× buffer 1 (10× buffer 1: 500 mM Tris-HCl, pH 9.2, 160 mM
(NH4)2SO4, 17.5 mM
MgCl2) supplemented with Me2SO to a final concentration of 5%, 200 nM dNTP, 50 pmol of forward
primer (5'-CTTCGGCAACTACGTGCTCTTC-3'), 50 pmol of reverse primer
(5'-TCCTGAAATCCAGCTCAGCTCC-3'), and 2.6 units of DNA polymerase mix. A
hot start PCR procedure was performed with the following parameters for
initial denaturation: 94 °C for 2 min, followed by 30 cycles (10 cycles with denaturation at 94 °C for 10 s, annealing at
65 °C for 30 s, extension at 68 °C for 6 min; and 20 cycles
with denaturation at 94 °C for 10 s, annealing at 65 °C for
30 s, extension at 68 °C for 6 min plus a 20-s increment/cycle)
and a final extension (at 68 °C for 7 min). The cDNA fragments
were subcloned in pUC18 using the Sureclone ligation kit (Amersham
Pharmacia Biotech). Sixty-eight positive clones were randomly chosen
and fully sequenced using automatic sequencing (Applied Biosystems)
with the dye terminator technology. A similar RT-PCR strategy was used
for the identification of cDNA fragments covering from nt 120 to
3936 with the forward and reverse primers 5'-TAGAGCCCACCAGATGTGCC-3'
and 5'-GAGATGAGCACCAACAGCCC-3', respectively. No evidence for sequence
variation has been identified to date in this region.
Northern and Dot Blot Analyses--
Commercial human Northern
blot membranes (CLONTECH) were hybridized using a
930-bp fragment (nt 3660-4590) generated by PCR amplification and
random-primed with [-32P]dCTP. The membranes were
treated according to the manufacturer's protocol. Relative amounts of
mRNA in each line were evaluated using two probes: a ubiquitin
probe and a -actin probe as internal controls. Two independent dot
blot membranes, normalized in their mRNA abundance by the
manufacturer according to the expression profile of eight housekeeping
genes (Master-blot mRNA membrane; CLONTECH)
were treated in similar conditions using a probe covering the
C-terminal region (nt 6440-6970). The exposure time for Northern and
dot blot membranes was 6 days. Densitometric analysis of the autoradiograms was performed using an AlphaImager system (Alpha Innotech Corp.) in order to provide semiquantitation of
1G mRNA in each condition. No signal was observed
for internal DNA controls, except for human genomic DNA.
Transient Transfection--
A plasmid encoding the reporter gene
CD8 (25) was used in transient transfection experiments,
together with the pBK-CMV construct that encodes for the
1G-b subunit, in a 1:10 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). For optimal transfection, cells were
plated at 30-40% confluence on Petri dishes coated with
poly-L-ornithine (Sigma). A standard calcium phosphate
transfection procedure was performed. Positively transfected cells were
identified with anti-CD8 antibody-coated beads (Dynal) and further
analyzed by voltage clamp experiments. cDNAs encoding the rat brain
1A-a, 2/1b, and
1b subunits were inserted in the vertebrate expression
vector pMT2 (14) and cotransfected in HEK 293 cells as a mix of
1A-a, 2/1b,
1b, and CD8 cDNAs at a molar ratio of 1:1:1:0.1.
Electrophysiology--
Macroscopic currents were recorded by the
whole-cell patch clamp technique at room temperature (~21 °C)
using an Axopatch 200B amplifier. Data were acquired on a PC computer
using the pClAMP6 software suite (Axon Instruments). Records were
filtered at 5 kHz. Leak and capacitive currents were subtracted using a P/5 procedure when needed (i.e. for tail
current recordings). Extracellular solution contained 2 mM
CaCl2 (or 2 mM BaCl2 or 2 mM SrCl2), 160 mM TEACl, 10 mM HEPES (pH to 7.4 with tetraethylammonium hydroxide).
Pipettes made of borosilicate glass with a typical resistance of 1-2
megaohms were filled with an internal solution containing 110 mM CsCl, 3 mM MgCl2, 10 mM EGTA, 10 mM HEPES, 3 mM Mg-ATP,
0.6 mM GTP (pH to 7.2 with CsOH). For pharmacological experiments, drugs were prepared daily in the external medium, freshly
(amiloride and mibefradil) or from a stock solution
(NiCl2). The various dilutions were applied to cells by a
gravity-driven homemade perfusion device, controlled by solenoid
valves. Single channel recordings were done in the cell-attached mode.
Sylgard coated pipettes (resistance of 7-15 mega-ohms) were filled
with a solution containing 110 mM BaCl2, 10 mM HEPES (pH to 7.2 with TEAOH). Membrane potential was
reduced toward 0 mV by bathing the cells in a high potassium solution
containing 140 mM potassium gluconate, 10 mM
EGTA, 10 mM glucose, 1 mM MgCl2, 10 mM 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 (Axon Instruments), Excel (Microsoft),
and GraphPad Prism (GraphPad Inc.) software programs. Whole cell
current-voltage curves were fitted using a modified Boltzmann equation
as described previously (26). Activation and inactivation curves were
fitted with a Boltzmann equation. Apparent dissociation constants
(Kd) were obtained from the fitting of the drug
dose-response curves using a sigmoidal function. Results are presented
as the mean ± S.E., where n is the number of cells used.
| |
RESULTS |
Cloning of a Human 1G Subunit--
Several cDNA
clones covering from domain II to the C-terminal region of the human
1G isoform were isolated from a cerebellum cDNA
library using a probe designed from the identified EST H06096 (Fig.
1A). Additional cDNA
clones that cover the entire coding region of the human
1G were identified using RT-PCR and RACE-PCR from human
total brain mRNA. Finally, two overlapping cDNA fragments of 4 and 3.4 kb covering from nt 120 to 6272 were obtained by RT-PCR from
human total brain mRNA to further confirm the cDNA structure of
human 1G isoforms. A full-length cDNA encoding the 1G-b isoform was then constructed using the overlapping
cDNAs (Fig. 1A).
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 1.
Cloning and molecular properties of
human 1G cDNAs.
A, the cloning strategy presented under "Materials and
Methods." Several partial cDNA fragments were isolated from
gt10 human libraries (CC1, CC2, and CC3). Two ESTs and two PCR
fragments covered the 5'- and 3'-ends. Two additional PCR fragments of
4 and 3.4 kb, according to 1G-b sequence (AF126965),
were subcloned and sequenced. This latter PCR fragment was used further
for 1G isoform determination as described under
"Materials and Methods" and in the legend to Fig. 2. A full-length
cDNA was constructed using overlapping PCR strategies as well as
the unique restriction sites. An additional restriction site
(BssHII) was introduced by PCR in the cDNA covering the
untranslated 5' region to allow final assembly and subcloning in the
pBK-CMV vector. B, the partial intron/exon structure of the
human gene CACNA1G was determined from sequence AC004590
using the GRAIL software for exon prediction. Exons are
boxed on the bottom line.
Black boxes correspond to the 34 exons that are
used in the human isoforms 1G-a and 1G-b.
The ATG and TGA codons are comprised in exon 1 and exon 38, respectively.  , exon 25 that is alternatively spliced to produce the
two isoforms a and b. *, exons 14, 26, 34, and 35 that were identified
in RT-PCR experiments.  , three predicted exons that are not
numbered, since they have not been detected in RT-PCR
experiments.
|
|
Alignment of our cDNA sequences to the corresponding genomic region
of chromosome 17q22 (AC004590), together with the use of GRAIL software
(22) for the identification of putative exons in the genomic sequence,
led us to identify the partial intron/exon structure of the
CACNA1G gene that encodes the 1G isoforms
(Fig. 1B). The CACNA1G gene covers <70 kb in the
17q22 region. A total of 41 putative exons was predicted using GRAIL
software (Fig. 1B). The 5'-end of CACNA1G (~500
bp), which has been determined recently (27), is comprised within a
single exon. Accordingly, this exon that also contains the first ATG
codon was designated exon 1. This ATG is comprised within a consensus
site for initiation of translation: AXXATGG (28). The coding
region of the two human 1G subunits described later in
this study is composed of 34 exons. One of these two variants which has
the highest homology with the original rat sequence (15), was named
1G-a. The isoform 1G-b resulted from the
use of an alternate 5'-splice donor site of exon 25 combined with the
acceptor site on exon 27 (Fig. 2, A and B). Four additional exons were identified
in RT-PCR experiments conducted on mRNAs from human total brain,
cerebellum, thalamus, and fetal kidney. These exons encode insertions
within the II-III loop (insertion e), within the III-IV loop (insertion
c), and within the COOH-terminal region (insertions f and d) (Fig.
2A). In our experiments, the insertion c, which corresponded
to the use of exon 26 (Fig. 2B), was found only in
combination with variation b. Three other predicted exons that are
included in Fig. 1B have not been retrieved experimentally
to date. Sequencing of cDNA fragments obtained using long range
RT-PCR experiments was performed in order to determine the potential
association of the a/b, c, d, e, and f regions in 1G
transcripts as well as their relative abundance in native human
tissues. A total of 68 cDNA fragments obtained from human total
brain, cerebellum, thalamus, and fetal kidney mRNAs was analyzed
(Fig. 2C). A majority of 1G-a isoform alone,
or in combination with insertion e, was found in neuronal tissues. By contrast, only the 1G-b isoform, mostly
associated with insertion c, was detected in fetal kidney.
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 2.
Identification of a variety of isoforms for
the human 1G subunit.
A, schematic structure of the 1G subunit that
illustrates the localization of the a/b variation generated by
alternate splicing of exon 25, as well as the positions of the
insertions e, c, f, and d encoded by exons 14, 26, 34, and 35, respectively. B, description of the alternative splicing
mechanisms that are expected to give rise to the 1G-a,
1G-b, and 1G-bc isoforms. Summary of the
analysis of the PCR-derived cDNA fragments encoding the 3.4-3.8-kb
fragment (from domain II to C terminus). A total of 68 clones deriving
from total brain (22), cerebellum (22), thalamus (7), and fetal kidney
(17) were analyzed by sequencing, and the occurrence of each
combination is reported. Note that a truncated form of
1G (+1*) was found three times. This
truncation (102 bp) removes most of the IIIS6 region of the
protein.
|
|
The construct 1G-a encodes a 2250-amino acid (aa)
protein with a calculated molecular mass of 249.333 Da. Primary
sequences of the two variants 1G-a and
1G-b were compared with the rat and mouse
1G proteins (Fig. 3). The
four transmembrane domains are highly conserved (98-100% identity).
The connecting loops between domains I and II and between domains II
and III, as well as the NH2 and COOH-terminal regions are
more divergent (85-95% identity). Within the II-III loop, overall
identity is 90% without taking into account insertion e (23 aa), which
occurs in human, rat, and mouse (Fig. 3A). The two variants
1G-a and 1G-b encode a distinct
intracellular III-IV loop (Fig. 3B). By contrast with the
mouse 1G sequence, insertion c (18 aa) was found
associated only with the splice variant b in our experiments (Fig.
2C). Nevertheless, this intracellular loop is shorter
(50-70 aa) compared with others (~300 aa). The III-IV loop in
isoform 1G-a is 100% identical to the rat
1G sequence. The other isoform, 1G-b,
exhibits a shorter III-IV linker resulting from a 21-bp deletion
described above. Sequence comparison of the III-IV loop of several
1G, 1H, and 1I subunits,
as well as the putative c54d2 gene product, revealed that
sequence variability within this intracellular loop is highly related
to the length of this region. The sequences surrounding this insertion
are more homologous, and several aa are conserved (Fig. 3B).
The aa sequences corresponding to insertions f (48 aa) and d (45 aa) in
the C-terminal region of the human 1G subunit that are
presented in Fig. 3C have not been described to date. The
search for putative regulatory sites, such as consensus phosphorylation
sites for protein kinase A and protein kinase C, led us to identify
eight protein kinase A and up to 20 protein kinase C sites contained
within the intracellular regions. Most of these consensus sites are
conserved among human and rat sequences. Interestingly, a protein
kinase A/protein kinase C consensus site identified in loop III-IV for
1G-a is removed in the isoform 1G-b, as a
consequence of splicing of exon 25 (Fig. 3B).
View larger version (50K):
[in this window]
[in a new window]
|
Fig. 3.
Multiple alignments and primary sequence
analysis of the human 1G
subunit. A, multiple alignments performed with
ClustalW1.7 (blosum matrix) of the deduced aa sequences that encode for
the intracellular loop between domain II and domain III of
1G. The sequences presented are human 1G-a (ha1Ga;
AF126966); human 1G-b (ha1Gb; AF126965); human partial
cDNA NBR13/1G-c (ha1G; AB012043), which carries out
insertion e (+e); rat 1G isoform 1 (ra1G1
(15); AF027984); rat 1G isoform 2 (ra1G2; AF125161);
mouse 1G (mA1G (30); AJ0112569). One should note that
the 23-aa insertion is retrieved in human, rat, and mouse. #,
insertion; *, nonconserved aa substitution; two dots:,
strongly similar aa substitution; one dot, weakly
similar aa substitution. B, multiple alignments performed
with ClustalW1.7 of the deduced aa sequences that encode the
intracellular loop between domain III and domain IV of
1G. In addition to the set of sequences defined above,
the sequences encoding for the C. elegans c54d2
(U37548); human 1H isoform 1 (ha1H1 (16); AF051946);
human 1H isoform 2 (ha1H2 (18); AF073931); and rat
1I (ha1I (17); AF086827) are added. Symbols
are as defined for A. The symbols in the
top line show the diversity of c54d2
with the mammalian sequences. The bottom line
illustrates the aa diversity among the various mammalian
1 subunits encoding T-type channels. Note that a
putative protein kinase A/protein kinase C phosphorylation site (#) is
removed in the human isoform 1G-b (AF126965; ha1Gb).
C, alignment of the C-terminal region of several human
1G isoforms that carry out no insertion
(df; AF126966), insertion f (+f), or
insertion d (+d), together with the mouse and rat
sequences.
|
|
Distribution of 1G mRNA in Human
Tissues--
Northern blot analysis of a large variety of adult human
tissues showed that 1G mRNA is predominantly present
in the brain (Fig. 4A).
Nonneuronal tissues that were positive were ovary and placenta. A very
weak band was also observed in testis, small intestine, colon, and
heart. Positive tissues exhibited a major band of approximately 8.5 kb
along with discrete larger bands. The reason for a smaller band in the
prostate mRNA lane is currently unknown. Considering neuronal
tissues, cerebellum as well as thalamus and substantia nigra displayed
a strong signal (Fig. 4B). Several other tissues from the
central nervous system also revealed a significant level of
1G expression, while other areas, like hippocampus, showed a weaker 1G mRNA signal. The pattern of
1G mRNA expression identified in Northern blots was
mostly confirmed using dot blot analysis with densitometric
quantification (Fig. 5). Two independent dot blot membranes were examined (see "Materials and Methods"). Brain tissues showed the strongest signal (thalamus > occipital lobe > cerebellum). The only difference between dot blot and
Northern blot analyses concerns the spinal cord, since expression of
1G mRNA was undetectable on Northern blot, although
the use of two independent internal controls, ubiquitin and -actin,
suggested that the quality this mRNA sample was correct on the
Northern blot membrane. Altogether, dot blot experiments revealed a
level of 1G expression in good agreement with Northern
blot analysis for other neuronal tissues.
View larger version (83K):
[in this window]
[in a new window]
|
Fig. 4.
Northern blot analysis of human
1G mRNAs from a variety of tissues
with commercial membranes (CLONTECH) using the same
probe as for the human cerebellum cDNA library screening.
Internal control was performed using a ubiquitin (Ubiq.) and
a -actin (not shown) probe provided by the manufacturer
(CLONTECH). A, the pattern shows strong
expression in brain but also in heart, placenta, and small intestine.
B, the expression profile for a variety of human central
nervous system regions is shown.
|
|
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 5.
Dot blot analysis of adult and fetal
tissues. Dot blot analysis of human 1G transcripts
with commercial membranes (CLONTECH) using a probe
that covers the 3'-end of the transcript (exon 38, nt 6440-6970) is
shown. A, relative intensities of the spots were determined
using the AlphaImager system (Alpha Innotech Corp.). B,
comparison and quantitation of 1G transcripts in fetal
(black bar) and adult (white
bar) tissues were determined by the densitometric analysis
on two independent master blot membranes (designated dot
1 and dot 2).
|
|
It is worth underlining the fact that analysis of two independent dot
blot membranes revealed that the 1G mRNA expression level was significantly higher in fetal peripheral tissues, heart, kidney, lung, spleen, and thymus, compared with adult tissues (Fig.
5B). This pattern of expression was not displayed in the brain, since the amount of 1G mRNA was equivalent in
adult and fetal brain. Since dot blot membranes are normalized for the
amount of mRNA present in each dot (see "Materials and
Methods"), we have performed a densitometric analysis that shows the
relative abundance of 1G mRNA in these tissues (Fig.
5B). Altogether, these data indicate that the expression of
1G is developmentally regulated.
Electrophysiological Description of the Human 1G
Currents--
Functional properties of human
1G-dependent channels were investigated
using the 1G-b isoform, as expressed in HEK-293 cells. This 1G-b isoform corresponds to the minimal structure
identified to date for the native human 1G subunit.
Robust T-type Ca2+ currents were recorded in the presence
of 2 mM Ca2+ (Fig.
6A). Activation kinetics was
voltage-dependent, with time constants ranging from 10 to
1.2 ms for potentials between 50 and +30 mV. Inactivation kinetics
was also strongly voltage-dependent. Current decay was best
described by a single exponential function, with time constants ranging
from 35 to 12 ms between 50 and +30 mV. This property implied
criss-crossing of current traces as observed for native T-type currents
(29).
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 6.
Electrophysiological properties of the
currents generated by the human 1G
subunit. A, currents evoked by increasing
depolarizations from 100 mV to +50 mV with 10-mV increments (from
holding potential of 100 mV). The arrow indicates the peak
current recorded for a test pulse to 30 mV. Note the criss-crossing
pattern of current traces. B, averaged current
density/voltage relationship obtained from 15 cells. C,
illustration of the permeation profile of 1G channel for
divalent cations. Equimolar substitution from Ca2+ to
Ba2+ ions induces a slight decrease (5%) in peak current
amplitude, while equimolar substitution with Sr2+ increases
current amplitude (35%). D, mean Ba/Ca and Sr/Ca
permeability ratios obtained from peak current amplitude values
measured in 2 mM Ca2+, Ba2+, and
Sr2+ (n = 6). E, single channel
measurements of 1G currents recorded using a tail
current protocol (see Refs. 16 and 17). The voltage protocol includes a
10-ms step to +30 mV followed by a test pulse to the indicated
potentials. F, unitary current amplitudes were obtained at
various potentials (120 to 0 mV) and plotted as a function of the
test potential value. Linear regression revealed a slope conductance of
7.3 picosiemens (n = 3).
|
|
This current activated around 60 mV and peaked at 30 mV, while
reversal potential was observed at +50 mV (Fig. 6B). When calcium (Ca2+) was replaced with equimolar barium
(Ba2+) or strontium (Sr2+) ions,
1G currents reflected permeation properties typical of that of T-type calcium channels (Fig. 6C). The amplitude of
Ba2+ currents was smaller (permeability ratio of
Ba2+/Ca2+ ~0.96), while Sr2+
currents were larger (permeability ratio of
Sr2+/Ca2+ ~1.35) than Ca2+
currents (n = 6; Fig. 6D). Interestingly,
Ba2+ current inactivation kinetics was faster, compared
with Ca2+ current. Single channel currents were measured in
the presence of 110 mM Ba2+ (Fig.
6E), using tail current protocols (16). Unitary current amplitude was recorded for membrane potentials ranging from 120 to 0 mV, indicating a slope conductance of 7.3 pS (n = 3;
Fig. 6F). Current amplitude at 0 mV was 0.39 ± 0.2 pA (n = 3).
Pharmacological Properties of Human 1G
Subunit--
We have studied the sensitivity of human
1G channels to three molecules, nickel
(Ni2+), amiloride, and mibefradil, that are considered to
be T-type channel blockers. The divalent ion Ni2+ was
described as a potent blocker of T-type currents; however, block of the
Ca2+ currents generated by the human 1G
subunit was modest, with an IC50 of 148 ± 10 µM (n = 8; Fig.
7A). This 1G
current was also poorly sensitive to amiloride, since less than 30% of
inhibition in the current amplitude (n = 5) was
obtained in the presence of 1 mM amiloride (not shown).
Finally, mibefradil, considered as the most potent T-type channel
blocker (30), affected 1G current amplitude with an
IC50 of 2 µM (n = 7; Fig.
7B) in good agreement with previous studies on native and
recombinant T-type channels.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 7.
Pharmacological profile of
1G channels. A, mean
dose-effect curve for Ni2+ (n = 8) shows an
IC50 of 148 ± 10 µM. The effect of
increasing Ni2+ ion concentrations on an 1G
current on peak Ca2+ current amplitude is shown in the
inset. The current was evoked by a 50-ms depolarization at
30 mV (stimulation rate of 0.1 Hz). Current traces from
bottom to top correspond (in µM
Ni2+) to control, 1, 10, 30, 100, 300, and 1000. B, the
mean dose-effect curve for mibefradil indicates an IC50 of
2 ± 0.2 µM (n = 7). The
inset illustrates the effect of increasing concentrations of
mibefradil on 1G current evoked using the same protocol
as in A. Current traces from bottom to
top correspond (in µM mibefradil) to control,
0.1, 0.3, 1, 3, and 10.
|
|
Gating Properties of Human 1G Channels--
The
voltage dependence of the channel activation was determined by
measuring tail current amplitudes at various depolarizing pulses (Fig.
8, A and C) that
fully activates 1G currents. For that purpose, duration
of each pulse was adjusted to the average time-to-peak current values
presented in the upper part of Fig. 8A. Normalized amplitudes of tail currents plotted as a
function of the membrane potential described a biphasic smooth curve
for steady state activation, which could be best fitted by the sum of
two Boltzmann functions (Fig. 8C) with half activation
values of V1 = 41.8 mV and
V2 = 14.7 V (n = 5). Steady
state inactivation properties were determined using a standard double
pulse protocol (Fig. 8B). A 5-s prepulse (100 to 35 mV)
preceded a test pulse at 30 mV. Data were fitted by a single
Boltzmann function with V0.5 = 71.6 mV (Fig.
8C). Superimposition of the steady state activation and
inactivation curves revealed the existence of a window current in the
range of 65 to 55 mV (Fig. 8C).
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 8.
Steady state activation and inactivation
of 1G calcium current.
A, determination of the steady state activation was based on
the measurement of tail current amplitudes according to the protocol
presented in the upper part of A. For
each depolarization, the tail current amplitude was measured for pulse
durations adjusted from the membrane potential/time-to-peak plot shown
in upper part of the figure.
B, steady state inactivation was estimated from the
measurement of the current amplitude at 30 mV after a 5-s
predepolarization pulse of increasing amplitude (100 to 35 mV, 5-mV
increment). C, normalized steady state activation obtained
from the protocol described in A (open
symbols, n = 5). The smooth curve
corresponds to the best fit obtained using two Boltzmann functions with
normalized activation (m ) being a function of the test
potential (V) according to the equation m = A1/1 + exp((V1 V)/K1) + A2/1 + exp((V2 V)/K2), where
A1 and A2 are the
respective contributions of each Boltzmann distribution;
V1 and V2 are the two
half-activation values; and K1 and
K2 are the slopes. Values from the fit are
A1 = 57%, A2 = 43%,
V1 = 41.8 mV, V2 = 14.7 mV, K1 = 5.1 mV, and
K2 = 11.9 mV. Mean steady state inactivation
curve (filled symbols, n = 10)
was best fitted with a single Boltzmann distribution with a
half-inactivation potential (V0.5) of 71.6 mV
and a slope of 4.33 mV. Superimposition of steady state activation
(m) and inactivation (h) curves revealed a
window current near 65 to 55 mV.
|
|
Deactivation Properties of 1G-related
Channels--
The deactivation properties of human 1G
channels were deduced from tail current analysis (Fig.
9). Fig. 9A shows that
kinetics of tail currents was slower when repolarizing membrane
potentials were more positive, reflecting a strong voltage dependence
of the deactivation process for 1G-related currents
(Fig. 9B). Additional experiments using action potential
waveforms (APWs) as a voltage clamp command were designed to evaluate
whether 1G-related channels allow brief or sustained
entry of Ca2+ in the range of physiological membrane
resting potentials (about 75 mV). When HEK cells overexpressing the
human 1G subunit were stimulated with neuronal APWs
(2-ms duration) applied from a holding potential of 75 mV, a slow
inward Ca2+ current following the APW was recorded (Fig.
9C). This current was totally suppressed by the application
of 1 µM mibefradil (not shown). By contrast, application
of this APW protocol to cells that expressed HVA current generated by
1A subunit resulted in a transient inward calcium
current that correlates with the duration of the APW (Fig.
9C). A similar result was obtained using native neurons
expressing only HVA currents (not shown).
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 9.
Deactivation properties of
1G currents and APW voltage clamp
experiments. A, 1G currents were
activated by a 7-ms test pulse at 30 mV from a holding potential of
100, and a family of tail currents was recorded for subsequent
repolarizations from 120 to 40 mV. B, plot of mean
deactivation time constants as a function of membrane repolarization
potential (n = 5). C, Ca2+
currents generated by the 1G and 1A
subunits (see labels) in response to a neuronal APW (2-ms
duration; holding potential of 75 mV). Note that, according to the
slow deactivation kinetics of the currents generated by the
1G subunit, activation and peak of the Ca2+
current occurred during repolarization. This inward current persisted
for more than 10 ms after repolarization. In contrast, HVA current
generated by the 1A-a subunit using the same APW
protocol evoked a transient Ca2+ current that matches the
APW duration.
|
|
| |
DISCUSSION |
This study is the first description of the molecular and the
functional properties of a cloned 1G T-type
Ca2+ channel subunit from humans. Specific findings include
(i) the description of several 1G isoforms generated by
alternative splicing, (ii) evidence for a developmental regulation of
1G transcripts, and (iii) the description of specific
functional properties related to slow deactivation and the existence of
a window current in the range of the resting membrane potential.
Our study provides the first evidence for alternatively spliced
isoforms of the 1G subunit, revealing therefore that the molecular diversity of T-type channels not only relies on the expression of three subunits encoded by distinct genes,
1G, 1H, and 1I (17). In
humans, the gene CACNA1G encoding 1G is
localized on chromosome 17q22 (15). Exon prediction using GRAIL (22) led us to identify 41 putative exons within the genomic sequence that
covers the coding region of that gene (AC004590). We have numbered 38 of them that were unambiguously identified in cDNA cloning and
RT-PCR experiments. Whether additional exons exist cannot be excluded.
Recently, Toyota et al. (27) have performed 5'-RACE analysis
of CACNA1G in order to identify the transcription start site
of this gene in humans (GenBankTM accession no. AF124351).
In good agreement with our results, no additional exon was found
upstream to the one identified here as exon 1. More importantly, these
authors have demonstrated that the CACNA1G gene is found
inactivated by aberrant methylation in several human tumors.
We have identified here that the 1G-b isoform
corresponds to the minimal structure of the native 1G
subunit in humans, and therefore it is important to consider that
analysis of this isoform provides a basis for the comparison of the
deduced aa sequences of 1G subunits among species, as
well as the identification of regions that are critical to explore for
further structure-function studies. Indeed, a mouse 1G
subunit recently described (31) exhibits two insertions within loops
II-III and III-IV, compared with rat 1G (15). We
demonstrate here that the insertion in loop II-III corresponds to the
use of an additional exon (exon 14 as referred to in the human
sequence). We have identified the expression of exon 14, i.e. insertion e, occurs in human brain mostly in
association with region a. In addition, we have identified two novel
insertions d and f in the C-terminal region of the human 1G subunit. The occurrence of these latter insertions is
low: 1 and 7 out of 68 for d and f, respectively. The functional role of these insertions in II-III loop and the C terminus remain to be
determined. Unexpectedly, while sequencing our batch of 3.4/3.8-kb fragments that covered from domain II to the C-terminal region, we have
found a truncated form of 1G cDNA (3 out of 68) that corresponds to a 102-bp deletion (nt 4498-4601; AF126965) of a region
that encodes the IIIS6 segment. The reason for the existence of such a
truncated form of 1G, which corresponds to an aberrant alternative splicing of exon 25, is unclear at the present time.
The differences among human, mouse, and rat 1G subunits
within the III-IV loop are likely to correspond to the combination of
two mechanisms. First, we have demonstrated here that the use of an
alternate splice site within exon 25 can generate the two human
1G isoforms, 1G-a and
1G-b. We referred here to these isoforms as
1G-a and 1G-b, by comparison to the
original rat 1G subunit (AF027984), which is identical
to the human 1G-a. A similar mechanism is likely to
occur in rodents, since a sequence similar to the III-IV loop of
1G-b can be retrieved in rat (32). Second, human, mouse,
and rat exhibit an additional insertion, designated c, in this region
(31, 32). In our study, insertion c was found only in combination with
b, resulting in 1G-bc isoforms that were mostly
expressed in fetal kidney. Again, the role of insertion c in the human
1G subunit has yet to be determined. Altogether, we
describe here two mechanisms that are able to generate 1G isoforms with a distinct III-IV loop. The isolation
and tissue distribution of cDNAs encoding the various human
1G isoforms as well the investigation of their
functional properties is now an important issue, as demonstrated for
the 1A isoforms that give rise to P/Q Ca2+
channel subtypes (14).
A common feature of the human and rodent 1G subunits, as
well as for the 1H and 1I subunits, is
the absence of the -subunit binding site, called AID (33), found in
every HVA 1 subtype. Such data are consistent with
previous experiments showing that -subunit knock-down using
antisense strategies does not affect T-type currents (34, 35).
Similarly, the G protein binding sites described in the non-L
1 subunits (36) are not retrieved. This is also the case
for the Ca2+ binding and calmodulin binding domains
involved in calcium dependent inactivation (37, 38). Several putative
phosphorylation sites for protein kinase A, protein kinase C, and
CaM-kinase II occur in the sequence, but their functional relevance
needs to be investigated. Interestingly, the shortest isoform,
1G-b, lacks a putative protein kinase A/protein kinase C
phosphorylation site in the III-IV loop.
Based on the 1G-a isoform, the aa sequence of the III-IV
loop is highly conserved among the three subunits 1G,
1H, and 1I, which would suggest a
conservation in function (75%; Ref. 17). It was proposed that this
connecting loop might be important for inactivation of T-type channels
(17, 39) on the basis of that described for Na+ channels
(40). However, although the various 1G isoforms exhibit important sequence variations in the III-IV loop, inactivation properties of currents related to the human 1G-b subunit
did not show significant differences from those of currents related to
rodents 1G isoforms (15, 31). Further work is needed to clearly examine whether this region of 1G influences the
voltage-dependent inactivation process and, subsequently,
if sequence variations within the III-IV loop have functional
consequences on T-type Ca2+ channel activity.
Northern blot and dot blot analyses have clearly indicated that the
highest level of 1G expression occurs in human brain tissues, particularly in thalamus, cerebellum, substantia nigra, and
frontal and occipital lobes. Overall, these data correlate well with
those of the recent study by Talley et al. (41). Expression of 1G in thalamus is expected to underlie the low
threshold spike activity described in thalamic neurons (42).
Furthermore, overexpression of T-channels in this structure has been
observed in GAERS rats, an animal model of absence epilepsy (12). More
importantly, our data have revealed that the 1G subunit
expression is developmentally regulated. While 1G is
rather similarly expressed in adult and fetal brain, it is much more
predominant in fetal peripheral tissues such as heart, kidney, or lung.
Enhanced expression of an 1 subunit encoding T-type
channels in fetal heart correlates well with electrophysiological data,
since it is generally accepted that T-type currents are preferentially
recorded in embryonic or newborn myocytes (43). It complicates,
however, the identification of the so-called cardiac T-type channel,
since another member of the T-type channel family, 1H,
has been cloned from heart tissue (16). The relative expression of both
1H- and 1G-related channels in
cardiovascular tissues is therefore an important issue that needs to be
further investigated.
The electrophysiological and pharmacological properties of human
1G currents were examined in a physiological external
calcium concentration (2 mM), except for the determination
of the single channel conductance (~7 picosiemens). Overall,
biophysical properties such as low activation threshold (<60 mV),
potential for peak current (30 mV), activation and inactivation
kinetics, cross-over of current traces, and permeation properties
(ISr > ICa 
IBa) are typical of that of T-type currents
recorded in native cells. Our data are also in good agreement with
those of other cloned 1G channels, although comparison
is impaired by the differences in experimental conditions. For
instance, threshold of activation at more positive potentials (50 mV)
and a shift of the I/V curve (peak at 10 mV) reported for
the mouse isoform (30) is likely to be due to the use of high
concentrations (20 mM) of divalent cations. It is worth
noting that steady state activation is best described by the sum of two
Boltzmann distributions. Similar results were obtained for human
1H channels (18) and rat 1G channels (44). A kinetic model that includes several closed state transitions prior to opening was proposed in Ref. 44, and this scheme can accurately account for most of the qualitative and quantitative features of human 1G channels described here. By
contrast, a single Boltzmann function fits the steady state
inactivation. This dual behavior for activation and inactivation is
reminiscent of that reported for several other channel types, including
N-type in dorsal root ganglion neurons (45), for which a bimodal
activation process (i.e. "willing"/"reluctant"
modes) was coupled to only one inactivation process.
The pharmacological profile of human 1G-related channels
was determined using the few molecules, nickel (Ni2+),
mibefradil, and amiloride, which have been used to discriminate LVA
T-type calcium channels from HVA calcium channels. The human 1G channel has a relatively high sensitivity to
mibefradil with an IC50 in the micromolar range. This value
is quite comparable with the mibefradil inhibition of human
1H currents (1 µM) recorded in the
presence of 15 mM Ba2+ (18). Amiloride seems to
affect differentially 1G- (IC50 > 1 mM) and 1H-related currents
(IC50 = 167 µM; Ref. 18), although the
differences in the recording conditions should again be considered. Similarly, we found a rather low sensitivity to Ni2+ ions
of human 1G currents (IC50 = 150 µM), as compared with 1H calcium channels
for which an IC50 of 6 µM was reported (18). One should note that the mouse 1G channel is blocked by
an even higher Ni2+ concentration (IC50 > 1.1 mM in 20 mM Ca2+; Ref. 31).
Nevertheless, discrepancies in the sensitivity to Ni2+ ions
are consistent with the data reported on native cells (2). It is
important to note that native T-type channels from brain regions
expressing a high level of 1G subunit exhibit low
sensitivity for Ni2+ ions in good agreement with that
described for expressed 1G channels (2, 41, 46). A
consequence of these data is that the block by micromolar
concentrations of Ni2+ should not longer be considered as
distinguishing LVA from HVA channels (47).
A hallmark of T-type channels is the slow kinetics of their current
deactivation, and our data clearly show that the deactivation kinetics
of the Ca2+ current related to human 1G
channels is voltage-dependent. We further demonstrate that
the consequence of a slow deactivation process can be visualized using
neuronal APWs as a voltage clamp command. Overexpression of T-type
calcium channel in HEK cells greatly improves such investigations, and
we show that APW stimulation (<2 ms) from cell's resting potential
values (75 mV) induces a sustained Ca2+ current that
occurs during repolarization and persists for more than 15 ms. The
kinetics of this Ca2+ current is consistent with the
kinetics of tail currents evoked by a square-pulse with membrane
repolarization at this potential. As expected, this Ca2+
influx is likely to be shaped by the pattern of APWs, such as duration,
resting membrane potential, and firing rate (48). Deactivation is a
feature that distinguishes LVA/T-type currents (
~ 5 ms; Ref.
2) from HVA currents ( ~ 0.3 ms; Ref. 49). McCobb and Beam
(50) have reported that distinct Ca2+ signals are generated
by LVA and HVA channels in voltage clamp experiments using neuronal
APWs. We show here that similar observations can be made using pure
populations of LVA and HVA channels, since APW stimulation of HEK cells
expressing P-type channels (1A-a isoform; Ref. 14)
induces a transient Ca2+ influx overlapping the APW
duration. This Ca2+ influx was ~7-fold smaller than the
one mediated by the 1G subunit. The Ca2+
influx gated by the 1G subunit is expected to play a
sp
1G Subunit That Forms T-type Calcium Channels*
§,
, and
TOP
ABSTRACT
INTRODUCTION
