Functional Differences between TRPC4 Splice Variants*

Functional characterizations of heterologously expressed TRPC4 have revealed diverse regulatory mechanisms and permeation properties. We aimed to clarify whether these differences result from different species and splice variants used for heterologous expression. Like the murine (cid:1) splice variant, rat and human TRPC4 (cid:1) both formed receptor-regulated cation channels when expressed in HEK293 cells. In contrast, human TRPC4 (cid:2) was poorly activated by stimulation of an H 1 histamine receptor. This was not due to reduced expression or plasma membrane targeting, because fluorescent TRPC4 (cid:2) fusion proteins were correctly in-serted in the plasma membrane. Furthermore, currents through both human TRPC4 (cid:2) and TRPC4 (cid:1) had similar current-voltage relationships and single channel conductances. To analyze the assembly of transient receptor potential channel subunits in functional pore complexes in living cells, a fluorescence resonance energy transfer (FRET) approach was used. TRPC4 (cid:2) and TRPC4 (cid:1) homomultimers exhibited robust FRET signals. Furthermore, coexpressed TRPC4 (cid:2) and TRPC4 (cid:1) subunits formed heteromultimers exhibiting comparable FRET signals. To promote variable heteromultimer as-semblies, TRPC4 (cid:2) /TRPC4 (cid:1) were coexpressed at different molar ratios. TRPC4 (cid:1) was inhibited in the presence

Functional characterizations of heterologously expressed TRPC4 have revealed diverse regulatory mechanisms and permeation properties. We aimed to clarify whether these differences result from different species and splice variants used for heterologous expression. Like the murine ␤ splice variant, rat and human TRPC4␤ both formed receptor-regulated cation channels when expressed in HEK293 cells. In contrast, human TRPC4␣ was poorly activated by stimulation of an H 1 histamine receptor. This was not due to reduced expression or plasma membrane targeting, because fluorescent TRPC4␣ fusion proteins were correctly inserted in the plasma membrane. Furthermore, currents through both human TRPC4␣ and TRPC4␤ had similar current-voltage relationships and single channel conductances. To analyze the assembly of transient receptor potential channel subunits in functional pore complexes in living cells, a fluorescence resonance energy transfer (FRET) approach was used. TRPC4␣ and TRPC4␤ homomultimers exhibited robust FRET signals. Furthermore, coexpressed TRPC4␣ and TRPC4␤ subunits formed heteromultimers exhibiting comparable FRET signals. To promote variable heteromultimer assemblies, TRPC4␣/TRPC4␤ were coexpressed at different molar ratios. TRPC4␤ was inhibited in the presence of TRPC4␣ with a cooperativity higher than 2, indicating a dominant negative effect of TRPC4␣ subunits in heteromultimeric TRPC4 channel complexes. Finally, Cterminal truncation of human TRPC4␣ fully restored the channel activity. Thus, TRPC4␤ subunits form a receptor-dependently regulated homomultimeric channel across various species, whereas TRPC4␣ contains a Cterminal autoinhibitory domain that may require additional regulatory mechanisms.
In addition to their well defined ability to release Ca 2ϩ from internal stores, phospholipase C-coupled receptors activate at least two different classes of cation entry. Capacitative calcium entry (CCE) 1 channels are activated as a consequence of the depletion of intracellular Ca 2ϩ stores and receptor-operated channels are gated independently of the filling-state of the endoplasmic reticulum. Although both Ca 2ϩ -selective and nonselective CCE channels have been described, their common feature is a receptor-independent activation by thapsigargin, an inhibitor of sarcoendoplasmic reticulum ATPases. By contrast, receptor-operated channels are activated by poorly defined mechanisms that include the formation of diacylglycerols or other membrane-resident second messengers such as arachidonic acid. Several members of the recently cloned transient receptor potential channel (TRPC) superfamily are candidate molecular substrates for these mechanisms (1,2). Despite a number of studies that have been conducted to define regulatory and biophysical properties of these channels, no clear picture has emerged. Most prominently, members of the first characterized classical, canonical, or short TRPC family (TRPC1-7) have been implicated in both CCE and store-independent cation entry (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15).
For almost each of the TRPCs both modes of activation have been proposed. In the case of TRPC4, the initial characterization reported an increased CCE activity in human embryonic kidney (HEK) 293 cells expressing the bovine TRPC4␣ splice variant (6). Murine TRPC4 and TRPC5 were shown to be nonselective cation channels that share the characteristics of receptor-operated channels (12). In contrast, human TRPC4 (hTRPC4) displayed only basal activity without acute regulation (16). A rat TRPC4 splice variant, also referred to as TRP-R (17), has been reported to enhance store-operated Ca 2ϩ entry when expressed in Xenopus laevis oocytes (18,19). Furthermore, Strü bing et al. (20) described the formation of heteromultimeric channel complexes of TRPC1 with either murine TRPC4␣ or TRPC5. These channel complexes behave like a receptor-operated channel and display biophysical properties distinct from those of homomultimeric TRPC4␣ or TRPC5. Thus, the assembly of TRPC4 splice variants in heteromultimeric TRPC complexes might add an additional level of complexity.
In this study, we have attempted to clarify the contradictory findings with heterologously expressed TRPC4 by directly comparing the regulation and biophysical properties of different species and splice variants. In addition, a fluorescence resonance energy transfer (FRET)-based approach was established to investigate the multimerization of TRPC subunits. Our data provide evidence that TRPC4␣ and TRPC4␤ subunits form receptor-dependently but store-independently regulated homoor heteromultimeric cation channels. TRPC4␤ shows cation fluxes of similar amplitudes across various species. In contrast, human TRPC4␣ displays a markedly lower efficiency of receptor-induced activation and exerts a dominant negative effect in heteromultimeric complexes with TRPC4␤.

EXPERIMENTAL PROCEDURES
Materials-Culture media and trypsin were purchased from Invitrogen. Fetal calf serum and phosphate-buffered saline were obtained from Biochrom (Berlin, Germany). Thapsigargin was from Calbiochem (Bad Soden, Germany). The rabbit anti-TRPC4 antibody was from Alomone Labs (Jerusalem, Israel). Unless otherwise stated, all other reagents were purchased from Sigma.
Cell Culture, Transient Transfections, and Fluorometric Techniques-HEK 293 cells were grown in minimal essential medium supplemented with 10% fetal calf serum, 50 units/ml penicillin, and 50 units/ml streptomycin. For transient transfections, the cells were seeded on 6-cm dishes and transfected the following day at 30 -70% confluency as described (12). HEK 293 cells were transfected with a Fugene6 reagent (Roche Molecular Biochemicals) according to the manufacturer's recommendations. The transfected cDNA plasmid concentration was kept constant by cotransfecting 0.2 g/dish of the rat histamine H 1 receptor in pcDNA3 and 3.8 g/dish of the indicated TRPC subunit in pcDNA3 or the corresponding amount of the empty expression vector (vector-transfected controls). After 18 -24 h, the cells were trypsinized and seeded on glass coverslips. All fluorometric experiments were conducted 48 -72 h after transfection, as described previously (12). Statistical analysis was done with the Mann-Whitney test, and significance was accepted at p Ͻ 0.05.
Cloning of TRPC4 Splice Variants and Generation of Fluorescent Fusion Proteins-Total RNA was prepared from Wistar rat brain, embryonic rat vascular smooth muscle cells (21), pooled human testis, and bovine brain using the Trizol reagent (Invitrogen). The first strand synthesis (Superscript II, Invitrogen) was primed with gene-specific oligonucleotides (rat TRPC4, 5Ј-GGT GCG TTT ATT CAA CA; human TRPC4, 5Ј-GCA GGC TAC AAA ACA AA; and bovine TRPC4, 5Ј-GCG GGT AAC AAA ACA A) and allowed to proceed for 80 min at 42°C. TRPC4 was amplified by 36 cycles of PCR (Expand-HF; Roche Molecular Biochemicals). Annealing temperature was 58°C (30 s), and the extension time (3.5 min) was further prolonged by 10 s/cycle in cycles 11-36. The forward primers were 5Ј-GCC AGC ATG GCT CAG TTC TAT TAC AAA (rat) or 5Ј-GCC ACC ATG GCT CAG TTC TAT TAC AAA AG (human). Reverse primers were 5Ј-GCA GGC TAC AAA AGA GGA GGG TT (rat) or 5Ј-CCC ACC CAG AGC ACT ACG GAA A (human). PCR products were subcloned in a pcDNA3.1-V5-His vector (eucaryotic TA-TOPO cloning system, Invitrogen) and confirmed by sequencing (ABI-Prism 377, Perkin Elmer). Human TRPC4␣ was C-terminally truncated (hTRPC4␣ ⌬842-977 ) by PCR with a reverse primer 5Ј-TTA ATC GGT CAC AAA ATT CAC TTT. C-terminal truncation at the splicing donor site Glu 784 of human TRPC4␣ (hTRPC4 ⌬785-977 ) was generated with a 5Ј-TTA TTC GCT ATC ACT CTT TTC AT reverse primer. The human TRPC4␣ and hTRPC4␤ characterized in this study correspond to the GenBank TM accession numbers AF421358 and AF421359.
For C-terminal fusion with cyan (CFP) or yellow (YFP) fluorescent protein, the stop codons of human and rat TRPC4 variants were removed by PCR and primers that replaced the stop codons by NotI or XhoI restriction sites, respectively. Subsequent in-frame subcloning into pcDNA3-CFP or pcDNA3-YFP fusion vectors (22) resulted in the TRPC4-CFP and TRPC4-YFP fusion plasmids. The absence of mutations was confirmed by DNA sequencing. The YFP-or CFP-tagged TRPC4 fusion constructs were used only for confocal microscopy and for FRET imaging, whereas all functional data were acquired with the wild-type TRPC4 variants.
Electrophysiological Techniques-For patch clamp experiments, the standard extracellular solution contained 140 mM NaCl, 5 mM CsCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM glucose, 10 mM HEPES (pH 7.4 with NaOH). For Na ϩ -and Ca 2ϩ -free extracellular solutions, Na ϩ was replaced by N-methyl-D-glucamine, and Ca 2ϩ was omitted. The standard intracellular solution contained 110 mM cesium methane sulfonate, 25 mM CsCl, 2 mM MgCl 2 , 0.362 mM CaCl 2 , 1 mM EGTA, 30 mM HEPES (pH 7.2 with CsOH) with a calculated [Ca 2ϩ ] of 100 nM. In some experiments Ca 2ϩ -free (10 EGTA) solutions or solutions with a higher Ca 2ϩ buffer capacity (3.62 CaCl 2 , 10 EGTA) were used. Whole cell recordings were made with an EPC-7 amplifier using Pulse software (HEKA, Lambrecht, Germany). Patch pipettes made of borosilicate glass had resistances of 3-5 M⍀ when filled with the standard intracellular solutions. The cells were held at a potential of Ϫ60 mV, and current-voltage relations were obtained from voltage ramps from Ϫ100 to ϩ100 mV with a duration of 400 ms applied at a frequency of 0.2 or 0.4 Hz. Ramp data were acquired at a frequency of 4 kHz after filtering at 1 kHz. The holding current was acquired at 30 Hz. Single channel data were stored on a digital tape recorder (Biologic, Claix, France) after filtering at 10 kHz. The currents were subsequently digitized at 10 kHz after filtering at 1 kHz, and analysis was performed with the pCLAMP6 software (Axon Instruments, Foster City, CA). Channel activity is expressed as NP o , the product of the number (N) of channels in the patch and the open probability (P o ). NP o values were calculated for consecutive 2-s periods.
Fluorescence Resonance Energy Transfer Measurement-The FRET efficiency (E) was measured by recording the fluorescence intensities of the donor with acceptor (F da ), and without acceptor (F d ). F d was quantified by detecting the increase in donor (CFP) fluorescence during acceptor-bleach (YFP). The fluorescence imaging system (TILL-Photonics, Martinsried, Germany) was equipped with a Lambda 10/2 motorized filter wheel (Sutter Instruments, Novato, CA) that was synchronized using the digital output of the Polychrome II monochromator control unit (TILL-Photonics). A dual reflectivity dichroic mirror (Ͻ460 and 505-525 nm; Chroma, Brattleboro, VT) was used in combination with the monochromator to excite CFP at 440 nm or YFP at 515 nm. The applied emission band pass filters (Chroma) were 460 -500 nm for CFP and 540 -580 nm for YFP. All FRET experiments were performed using a Plan-Apochomat 63ϫ/1.4 objective (Carl Zeiss). The acceptorbleach protocol consisted of 30 -40 cycles with 10 -20 ms of exposure to detect the CFP and YFP fluorescences without YFP-bleach and 80 -120 additional cycles with an additional 2 s of illumination at 510 nm to bleach YFP. The data of three to five independent experiments (each representing four to eight transfected single cells) were averaged and normalized to the initial intensity at the beginning of the experiment (CFP init ). The donor fluorescence in the absence of the acceptor F d was assessed by measuring the half-maximal increase in the donor fluorescence ⌬CFP 50 after half-maximal bleach of YFP. The FRET efficiency E was calculated using the following formula.
The data of each FRET experiment were averaged over five to eight single cells. The means and S.E. were computed from n ϭ 3 independent experiments for each donor-acceptor pair.

Ca 2ϩ and Mn 2ϩ Entry through TRPC4 Species and Splice
Variants-By cloning TRPC4 from different species and tissues, we and others could identify a number of TRPC4 splice variants. In rat and human, the most abundant transcripts, TRPC4␣ and TRPC4␤, differ in an 84-amino acid domain in the cytosolic C terminus of TRPC4␣ that is absent in TRPC4␤. A TRPC4␤-like rat TRPC4␤ 2 (AF421365) shares the same splicing acceptor site but is shortened by an additional amino acid. Our data indicate that the functional properties of rat TRPC4␤ 2 are similar to rat TRPC4␤ (data not shown). A number of rat and human splice variants (hTRPC4, AF421361; rTRPC4, AF421366; rTRPC4, AF421367; and rTRPC4, AF421368) are shortened at a position that corresponds to the predicted second transmembrane domain. Because these alternative splice events result in a shifted reading frame and premature stop codons, we did not characterize these truncated proteins.
When coexpressed with the rat H 1 histamine receptor in HEK 293 cells, rTRPC4␣ and, more efficiently, rTRPC4␤ could be activated by histamine (100 M; Fig. 1). Only a weak activation was seen for the human TRPC4␣ (Fig. 2). In control cells that were cotransfected with the H 1 receptor and the expression vector without insert, the addition of histamine induced an immediate [Ca 2ϩ ] i elevation from 94 Ϯ 3 to 455 Ϯ 31 nM (means Ϯ S.E. of five independent transfection experiments). The basal [Ca 2ϩ ] i was not significantly higher in HEK 293 cells that expressed any of the investigated TRPC4 variants. How-ever, markedly higher peak values for the histamine-induced [Ca 2ϩ ] i transients were detected in cells that expressed rTRPC4␤ (1512 Ϯ 49 nM, n ϭ 5, p Ͻ 0.05; Fig. 1) or hTRPC4␤ (1133 Ϯ 81 nM, n ϭ 5, p Ͻ 0.05; Fig. 2). A weaker increase in histamine-induced peak [Ca 2ϩ ] i was observed after expression of rTRPC4␣ (594 Ϯ 122 nM, n ϭ 3, not significant) or hTRPC4␣ (575 Ϯ 24 nM, n ϭ 6, not significant).
Because the peak [Ca 2ϩ ] i results from both Ca 2ϩ mobilization and Ca 2ϩ influx, we used Mn 2ϩ as a Ca 2ϩ surrogate that permeates through TRPC4 and allows a differentiation between Ca 2ϩ release and cation entry (12). In control cells that were cotransfected with the H 1 receptor and the empty expression vector instead of TRPC4, the quench rate of the fura 2 fluorescence in the presence of extracellular Mn 2ϩ (1 mM) was 1.2 Ϯ 0.7%/30 s before and 2.4 Ϯ 1.0%/30 s after the addition of histamine (100 M), respectively. In rat or human TRPC4␤expressing cells, only a marginal and not statistically significant acceleration of Mn 2ϩ entry was evident in the absence of the agonist. Stimulation with histamine, however, resulted in 15-and 17-fold increases in Mn 2ϩ entry for rat and human TRPC4␤, respectively ( Figs. 1 and 2). Regarding the ␣ splice variants, we observed a difference between rat and human clones; histamine application accelerated the Mn 2ϩ entry rate through rTRPC4␣ by more than 10-fold ( Fig. 1), but hTRPC4␣ was only weakly (3.8-fold; Fig. 2) yet significantly (p Ͻ 0.05 versus vector-transfected controls) activated by histamine.
Because a store-dependent mode of activation has been described for bovine TRPC4␣ (6), we tested the effects of thapsigargin, an inhibitor of sarcoendoplasmic Ca 2ϩ -ATPases, on human TRPC4 splice variants. Thapsigargin (2.5 M) did not induce an acceleration of Mn 2ϩ entry through hTRPC4␣ (Fig.  3A), hTRPC4␤ (Fig. 3B), or the corresponding rat orthologues (data not shown). The subsequent addition of histamine (100 M) activated only hTRPC4␤ but not hTRPC4␣ (Fig. 3). The absence of a histamine-induced Ca 2ϩ mobilization in hTRPC4␣-expressing cells (Fig. 3A) indicates that the thapsigargin treatment effectively depleted InsP 3 -sensitive intracellular Ca 2ϩ stores. In hTRPC4␤-expressing cells, however, a histamine-induced increase in [Ca 2ϩ ] i indicative of cation entry paralleled the Mn 2ϩ quench (Fig. 3B). Thus, store depletion is not sufficient to activate any of the investigated TRPC4 variants. Likewise, the addition of the membrane-permeable diacylglycerols 1,2-dioctanoyl-sn-glycerol or 1-oleoyl-2-acetyl-snglycerol at concentrations up to 200 M did not induce Ca 2ϩ or Mn 2ϩ entry through human or rat TRPC4␣ and TRPC4␤ variants.
Electrophysiological Whole Cell Recordings-We studied the properties of human and rat TRPC4␣ and TRPC4␤ expressed in HEK 293 cells in whole cell patch clamp recordings. Following patch rupture, no constitutive activity was observed at a holding potential of Ϫ60 mV for any of the clones (Fig. 4, A and  B). Currents at the start of whole cell recording were in the range of Ϫ0.72 to Ϫ1.45 pA/pF for each of the investigated TRPC4 variants (n ϭ 5-34). These values are not significantly different from those in vector-transfected control cells (Ϫ0.80 Ϯ 0.22 pA/pF, n ϭ 7), indicating the low basal activity of the TRPC4 variants studied. Again, we also tested whether hTRPC4␣ could be stimulated by a store depletion protocol. Four hTRPC4␣-expressing cells showed no response to infusion with 100 M InsP 3 in a solution containing 10 mM EGTA with 100 nM free Ca 2ϩ but responded to subsequent application of histamine. At a holding potential of Ϫ60 mV, histamine evoked transient inward currents in cells expressing any of the splice variants (Fig. 4C). Despite the continuous presence of the agonist, currents rapidly reached a maximum and then decayed with an initial rapid phase followed by a slow phase. Not all cells displayed the rapid phase of current decay, particularly when the current amplitude was very small. Histamine-induced inward currents were completely abolished by the removal of Na ϩ and Ca 2ϩ from the extracellular solution (Fig. 4,  A and B). Upon removal of histamine, the currents returned to the prestimulation level.
As reported previously for murine TRPC4␤ (12), the currentvoltage relations of all of the splice variants had a typical doubly rectifying form (Fig. 4, A and B) and reversal potentials close to 0 mV (7.8 Ϯ 0.5 for hTRPC4␣, n ϭ 31; 6.3 Ϯ 0.6 for hTRPC4␤, n ϭ 22; 2.8 Ϯ 0.4 for rTRPC4␣, n ϭ 8; and 4.6 Ϯ 1.2 for rTRPC4␤, n ϭ 5), indicative of poor cation selectivity. In a minority of cells expressing hTRPC4␣ or hTRPC4␤, we observed current-voltage relations that displayed a clear minimum in the inward direction and no plateau in the outward direction. These current-voltage relations closely resembled those of heteromultimers of TRPC1 and TRPC4 (20). In vectortransfected control cells, this type of cation current was not observed upon histamine application. Consistent with the low rates of Mn 2ϩ influx in fluorometric measurements (Fig. 2), the histamine-induced increases in current density were much smaller in cells that expressed hTRPC4␣ than in cells expressing other TRPC4 variants (Fig. 4C).
A characteristic feature of murine TRPC4␤ and TRPC5 is their stimulation by micromolar concentrations of La 3ϩ (12,20). Likewise, extracellular La 3ϩ rapidly and reversibly potentiated inward currents through hTRPC4␣ (Fig. 4D) and the other TRPC4 variants investigated (data not shown). The mean increases in current at Ϫ60 mV with 100 M La 3ϩ were typically 3-4-fold (n ϭ 3-5 for each TRPC4 variant). At a concentration of 10 M, La 3ϩ had weaker stimulatory effects (about 50% augmentation of inward currents). The low concentrations at which it is effective and the absence of a shift in reversal potential indicate that La 3ϩ potentiates Na ϩ currents rather than carrying the additional inward currents. Furthermore, some potentiation, although less than for inward currents, was also observed at positive potentials where Cs ϩ is the charge carrier.
Single Channel Recordings-The single channel properties of hTRPC4␣ and hTRPC4␤ were characterized in outside-out patches. Under the same conditions used for whole cell recordings, single channel events with an amplitude of around Ϫ2 pA at Ϫ60 mV were observed in outside-out patches from cells expressing either of the hTRPC4 splice variants. Levels of single channel activity were higher in patches from cells to which histamine had been applied during the whole cell recording than in patches from cells that had not previously been exposed to the agonist. Application of histamine to outside-out patches resulted in a transient, reversible stimulation of single channel activity (Fig. 5A). The potential dependence of the current amplitude is shown in Fig. 5B. Because patches were unstable at positive membrane potentials under the conditions used, we limited the analysis to potentials below 0 mV. Current amplitudes were similar at all negative potentials for both hTRPC4␣ and hTRPC4␤ (Fig. 5B). The chord conductances at Ϫ60 mV were 30.3 Ϯ 0.6 pS (n ϭ 7) and 29.7 Ϯ 1.0 pS (n ϭ 6) for hTRPC4␣ and hTRPC4␤, respectively. Under similar conditions, rTRPC4␣ displayed similar properties to the human clones (data not shown). The chord conductance was 27.5 Ϯ 0.7 pS (n ϭ 4) at Ϫ60 mV.
Membrane Targeting and Multimerization of TRPC4 Splice Variants-The reason for the weak receptor-induced cation entry through hTRPC4␣ was obscure. Because expression plasmids for both human splice variants contain the same 5Јuntranslated regions and ribosomal docking sequences preceding the start codon (GCC ACC ATG), the low receptor-induced Ca 2ϩ and Mn 2ϩ entry through hTRPC4␣ is unlikely to be caused by different expression levels of these clones. The Western blot analysis of heterologously expressed human TRPC4 variants demonstrated comparable expression levels for both TRPC4␣ and TRPC4␤ (Fig. 6A). Moreover, the expression of the C-terminally YFP-fused subunits resulted in similar fluorescence intensities of the TRPC4 splice variants in fluorescence imaging experiments (data not shown).
A retention in intracellular compartments such as the endoplasmic reticulum or the Golgi apparatus has recently been demonstrated for murine TRPC2 (23). The targeting of the splice variants was therefore studied with TRPC4 fusion proteins C-terminally tagged with the YFP. Both human and rat TRPC4 variants were integrated into the plasma membrane with similar efficiency (Fig. 6B). The clustered plasma mem-brane distribution of rat TRPC4␣ resembles that of human TRPC4␣ (15) or murine TRPC4␤ (12). The lateral mobility of these clusters was highly restricted, indicating a tight connection to rigid structures such as elements of the cytoskeleton. When coexpressed, CFP-fused TRPC4␣ and YFP-fused TRPC4␤ strictly colocalize within the same clusters (data not shown).
Despite correct membrane targeting, hTRPC4␣ might still fail to assemble into functional channel complexes presumably requiring assembly of four channel subunits. Thus, the formation of multimeric channel complexes was analyzed. Because of the tight coupling to large immobile clusters harboring a large number of other proteins, coimmunoprecipitation was considered as inappropriate to prove direct interactions between channel subunits. Channel multimerization was therefore assessed with a FRET approach. Static FRET efficiencies between fluorescent TRPC fusion proteins were determined by measuring the donor recovery during acceptor bleach. The maximal FRET efficiency in an intramolecularly coupled CFP-YFP tandem protein was 53.6 Ϯ 0.5%. Because the Förster radius (R 0 ) at which the CFP/YFP combination exhibits a FRET efficiency of 50% was assumed to be 5-6.5 nm (24), an additional 5-nm distance of the tags results in a drop of the FRET signals below 1%. Correspondingly, no significant FRET (FRET efficiency, 0.07 Ϯ 0.2%) could be observed when CFP and YFP were cotransfected on separate plasmids (data not shown). The applicability of the FRET technique was verified using the rat vanilloid receptor (VR1), which has been demonstrated to form homomultimers (25). Coexpressed C-terminal fusion proteins of VR1 (VR1-CFP and VR1-YFP) yielded FRET efficiencies of 18.5 Ϯ 0.5% (Fig. 7A). For coexpressed hTRPC4␣-CFP/VR1-YFP or hTRPC4␤-CFP/VR1-YFP, no FRET signals were detectable (Fig. 7A) confirming the sensitivity and specificity of the FRET-based analysis of TRPC multimerization. For coexpressed CFP-and YFP-tagged hTRPC4␣, a FRET efficiency of 9.4 Ϯ 0.04% was observed (Fig. 7B). Similarly, a FRET efficiency of 7.7 Ϯ 0.3% was detected for hTRPC4␤. Thus, both hTRPC4 subunits appear to form homomultimers with similar efficiencies. Interestingly, a heteromultimeric assembly of hTRPC4␣-CFP and hTRPC4␤-YFP (FRET efficiency, 8.3 Ϯ 0.2%) or hTRPC4␤-CFP and hTRPC4␣-YFP (FRET efficiency, 7.9 Ϯ 0.3%) resulted in FRET signals similar to their respective homomultimers (Fig. 7, B and C).
TRPC4␣ Is a Dominant Negative Regulator in Heteromultimeric Channel Complexes-The weak activation of hTRPC4␣ may be compensated by neighboring TRPC4␤ subunits in the same channel complex. The receptor-induced activation of heteromultimeric TRPC4 complexes was assessed by cotransfecting different cDNA plasmid amounts of hTRPC4␣ and hTRPC4␤. Unexpectedly, when HEK 293 cells were cotransfected with hTRPC4␣ and hTRPC4␤ at a 1:1 plasmid ratio, the histamine-induced Mn 2ϩ entry was almost indistiguishable from cells that only expressed hTRPC4␣ (Fig. 8). A further reduction of the TRPC4␣ plasmid concentration in favor of hTRPC4␤ (1:4 plasmid ratio) revealed a reduction in histamine-stimulated Mn 2ϩ entry of about 40% when compared with hTRPC4␤ expressed alone (Fig. 8). Thus, hTRPC4␣ subunits inhibit the conductance of heteromultimeric TRPC4 channel complexes with a high cooperativity. These data are in agreement with a model in which a single hTRPC4␣ channel subunit is sufficient to block a heterotetrameric TRPC4 channel complex. Because the pore region is not affected by differential splicing of TRPC4␣ or TRPC4␤, either binding of the channel-activating agent or the subsequent allosteric movement may be prevented by the hTRPC4␣ subunit. To clarify whether the 84-amino acid domain that is unique for hTRPC4␣ contains an autoinhibitory domain, we constructed expression plasmids encoding C-terminally truncated hTRPC4␣ subunits. A C-terminal truncation that contains only a part of the hTRPC4␣-specific domain (hTRPC4␣ ⌬842-977 ) partially recovered the activation by histamine (Fig. 9A). A deletion at the splicing donor site (Glu 785 in hTRPC4␣) resulted in hTRPC4 ⌬785-977 , which is shortened by 192 C-terminal amino acids. This gross deletion fully restored the histamine-induced Mn 2ϩ entry (Fig. 9B) and whole cell currents (Fig. 9C) to the same levels as observed for hTRPC4␤. DISCUSSION In this study we show that splice and species variants of TRPC4 exhibit major differences in their regulation despite similarities in their mechanism of activation and biophysical properties. In particular, human TRPC4␤ showed much stronger responses to phospholipase C-coupled receptor activation than the ␣ splice variant. Neither reduced expression levels, nor a defect in the plasma membrane targeting account for the weak receptor-induced activation of hTRPC4␣. Furthermore, we have successfully applied the FRET technique to study TRPC homo-and heteromultimerization. The similar FRET signals observed for coexpressed fluorescent TRPC4␣ and TRPC4␤ subunits indicate that homo-and heteromultimers form with comparable efficiencies. The functional properties of heteromultimeric hTRPC4 channel complexes revealed a dominant negative effect of the hTRPC4␣ subunit. C-terminal truncation experiments confirmed the presence of an autoinhibitory domain in hTRPC4␣.
Our functional data on heterologously expressed TRPC4␣ or TRPC4␤ species variants are consistent with a receptor-operated but store-independent mode of activation. By contrast, either store-dependent activation or basal activity without acute stimulation have been reported for bovine and human TRPC4␣ (6,16). In our hands, hTRPC4␣ is only poorly activated by histamine and not responsive to store depletion protocols. Earlier reports that a frameshifted splice variant that ends at the second transmembrane domain and corresponds to our rat TRPC4 might act as a store-operated CCE channel (18,19) are in contrast with our findings for the full-length rTRPC4 variants. Interestingly the N-terminal 765 amino acids of human, bovine, mouse, or rat TRPC4 display a similarity of more than 99% whereas the C-terminal 212 amino acids are significantly less conserved (76% similarity). The intact receptorinduced gating of rTRPC4␣ is in agreement with recent data obtained with the phylogenetically more closely related murine TRPC4␣ (20). Although this does not exclude the presence of functionally relevant domains in the C terminus, its structural constraints appear to be less conserved during evolution. In a number of human tissues, TRPC4␣ transcripts appear to be more abundant than the TRPC4␤ splice variant (26). Another study, however, reported the abundant expression of the hTRPC4␤ in gastrointestinal myocytes (27). Interestingly, the biophysical and regulatory properties of endogenous receptoractivated currents in ileal smooth muscle cells (28) display a striking similarity to those of heterologously expressed TRPC4. Furthermore, the requirement for intracellular Ca 2ϩ and a potentiation of inward currents by external La 3ϩ has been also reported for the endogenous acetylcholine-induced currents in ileal myocytes (29,30).
Our data favor a model in which human TRPC4␣ subunits contain an autoinhibitory domain that inhibits the receptorinduced activation of homo-or heteromultimeric TRPC4 channel complexes. To date, a dominant negative effect between TRPC subunits has only been described for artificial N-terminal TRPC3 or TRPC6 fragments, which might interfere with the multimerization process (31,32). However, the mechanism as well as the selectivity of those dominant negative effects are not substantiated by experimental data. In contrast to the dominant negative effects of a MaxiK channel splice variant (33) or Kv1.1 channels (34), the surface targeting and multimerization of hTRPC4␣ subunits are not affected. Alternative concepts like the inactivation by a ball-and-chain domain in voltage-gated "Shaker" potassium channels (35) or a phosphorylation-sensitive autoinhibitory domain as described in the CFTR (36) will have to be tested.
The store-independent gating and the biophysical properties of TRPC4 variants argue against previous findings that TRPC4 by itself may be a part of Ca 2ϩ -selective, store-operated channels. Nonetheless, our data are not in conflict with other concepts how TRPC4 variants may be linked to CCE. Because the autoinhibitory domain of TRPC4␣ overlaps with recently described binding motifs for calmodulin (37,38) and/or InsP 3 receptors (26), binding of these proteins may serve additional regulatory mechanisms. There is evidence for the existence of two independent calmodulin-binding sites on TRPC4␣. The first is present in both TRPC4␣ and TRPC4␤ and alternatively binds to a N-terminal F2q segment of InsP 3 receptors (37,38). A second calmodulin-binding site is present only in TRPC4␣ and does not bind the F2q segment (38) but a C-terminal portion of InsP 3 receptors (26). Although infusion of InsP 3 was not sufficient to activate any of the TRPC4 variants investigated in this study, we do not want to exclude the possibility that the binding site for InsP 3 receptors may play a role in TRPC4 activation at low expression levels. Alternatively, TRPC4␣-containing complexes may anchor the endoplasmic reticulum via InsP 3 receptors. Thus, it is plausible that TRPC4 channels may act as a scaffold to attach the endoplasmic reticulum in the vicinity of store-operated channels and thereby optimize CCE. Although InsP 3 receptors, calmodulin, and the Na ϩ /H ϩ exchanger regulatory factor (39) are proteins currently known to interact with TRPC4, the identification of other proteins that reside in the same clusters as TRPC4 may reveal the composition of TRPC4 signaling complexes in vivo.
The investigation of TRPC4 subunit multimerization is complicated by its tight docking to rigid supramolecular structures as inferred from the clustered appearance of TRPC4 and the very low lateral mobility of these patches within the plasma membrane. Attachment of TRPC4 to supramolecular structures precludes the identification of directly interacting proteins with standard coimmunoprecipitation approaches. Therefore, the TRPC4 subunit assembly was studied in living cells using the FRET technique. The vanilloid receptor 1, which is an essential component of the pain pathway (40) and which is known to form homotetramers (25), was chosen as a positive control. The robust FRET signals for coexpressed CFP-and YFP-tagged VR1 support these findings. The inability of CFPfused hTRPC4 splice variants to generate a FRET signal with VR1-YFP further confirmed the sensitivity and specificity of the FRET-based multimerization assay. This first application of the FRET technique to detect the assembly of ion channel subunits in living cells will be extended to clarify the specificity and promiscuity of TRPC heteromultimerization in future studies.
Considering that hTRPC4␣ is a dominant negative modulator of hTRPC4␤, it may predominantly function in TRPC complexes that contain other TRPC subunits such as the ubiquitously expressed TRPC1 (20). Our preliminary data, however, failed to demonstrate a receptor-induced activation of channel complexes formed by TRPC1 and the ␣ splice variant of hTRPC4. Although coexpressed in selected structures in the central nervous system, the restricted expression pattern of TRPC5, the closest relative of TRPC4, makes it unlikely that these TRPC subunits combine in other tissues.
Taken together, the regulatory properties of TRPC4␣ cloned from different species are inconsistent, presumably because of the variability of the C-terminal cytosolic domain. We provide evidence that some of the apparently conflicting findings for TRPC4 regulation rely on the different function of TRPC4 species and splice variants. In addition, we show that FRET between fluorescent subunits is a valuable tool for studying TRPC homo-and heteromultimerization in living cells. Our data support a concept in which TRPC4␤ is a functional cation channel across various species, whereas hTRPC4␣ subunits prevent full channel activation but offer additional sites for protein-protein interactions.