Identification of common binding sites for calmodulin and inositol 1,4,5-trisphosphate receptors on the carboxyl termini of trp channels.

Homologues of Drosophila Trp (transient receptor potential) form plasma membrane channels that mediate Ca(2+) entry following the activation of phospholipase C by cell surface receptors. Among the seven Trp homologous found in mammals, Trp3 has been shown to interact with and respond to IP(3) receptors (IP(3)Rs) for activation. Here we show that Trp4 and other Trp proteins also interact with IP(3)Rs. The IP(3)R-binding domain also interacts with calmodulin (CaM) in a Ca(2+)-dependent manner with affinities ranging from 10 nm for Trp2 to 290 nm for Trp6. In addition, other binding sites for CaM and IP(3)Rs are present in the alpha but not the beta isoform of Trp4. In the presence of Ca(2+), the Trp-IP(3)R interaction is inhibited by CaM. However, a synthetic peptide representing a Trp-binding domain of IP(3)Rs inhibited the binding of CaM to Trp3, -6, and -7 more effectively than that to Trp1, -2, -4, and -5. In inside-out membrane patches, Trp4 is activated strongly by calmidazolium, an antagonist of CaM, and a high (50 microm) but not a low (5 microm) concentration of the Trp-binding peptide of the IP(3)R. Our data support the view that both CaM and IP(3)Rs play important roles in controlling the gating of Trp-based channels. However, the sensitivity and responses to CaM and IP(3)Rs differ for each Trp.

Binding of many cell surface receptors by hormones, neurotransmitters, and growth factors leads to the activation of phospholipase C, which in turn produces diacylglycerol and inositol 1,4,5-trisphosphate (IP 3 ). 1 Diacylglycerol activates protein kinase C, while IP 3 triggers Ca 2ϩ release from internal Ca 2ϩ stores by activating a set of intracellular Ca 2ϩ release channels, referred to as IP 3 receptors (IP 3 Rs). The release of Ca 2ϩ from internal stores in turn activates store-operated channels (SOCs) located on the plasma membrane, allowing Ca 2ϩ influx from the extracellular space. The store-operated Ca 2ϩ influx, also known as capacitative Ca 2ϩ entry (1,2), plays critical roles in controlling the duration and the frequency of cytosolic Ca 2ϩ changes (2)(3)(4).
In contrast to the well defined roles of IP 3 and IP 3 Rs in Ca 2ϩ release, the molecular makeup of channels that mediate Ca 2ϩ influx and their gating mechanism(s) remain to be elucidated. Drosophila transient receptor potential (Trp) protein and its mammalian homologues have been shown to form either Ca 2ϩ selective or nonselective cation channels that mediate Ca 2ϩ influx in response to phospholipase C activation (5)(6)(7). To date, seven trp genes have been cloned from mammalian species (6,8), probably reflecting the heterogeneity of Ca 2ϩ influx channels or pathways found in different cells (4,9). Expression of individual Trp proteins in heterologous systems revealed that Trp channels may be activated by a number of intermediaries involved in the phospholipase C-stimulated signaling cascade, including Ca 2ϩ (10), diacylglycerol (11), and activated IP 3 Rs (12)(13)(14). Although store depletion induced by an intracellular Ca 2ϩ -ATPase inhibitor, thapsigargin, appears to be sufficient to open some Trp channels (e.g. Trp1 (15), Trp2 (16), and Trp4 (17)), it remains controversial whether all Trp proteins participate in forming SOCs (18,19). Perhaps the answer lies within the structural organization of the channel, which could be composed of four different Trp subunits (5). A recent example showed that coexpression of two Drosophila store-insensitive Trp proteins, Trp␥ and Trp-like (TrpL), led to the formation of a SOC (20). Thus, the store sensitivity may be reconstituted with the proper combination of different Trp subunits. Consistent with this idea, Trp1, Trp3, and Trp4 have been shown to be part of SOCs in human submandibular gland cells, neurons, and adrenal cortex cells, respectively (15,21,22).
Recent studies showed that IP 3 Rs are involved in the activation of Trp3. Following the initial demonstration that human Trp3 (hTrp3) in inside-out membrane patches was activated by IP 3 Rs in the presence of IP 3 (12), Boulay et al. (14) identified the binding domains involved in the Trp-IP 3 R interaction, which were found to be located in the N terminus of type 3 IP 3 R (IP 3 R3) and the C terminus of Trp3. Overexpression of short peptide fragments containing these binding sites altered the activity of endogenous store-operated Ca 2ϩ influx in HEK293 cells (14). While the association with IP 3 Rs has also been shown for Trp1 and Trp6 by coimmunoprecipitation (14,23,24), it remains to be determined whether direct interaction with IP 3 Rs is common for all Trp proteins. In this study, we examined murine Trp4 (mTrp4) for interaction with the first and the stronger Trp3-binding domain of IP 3 R3 (F2q; Glu 669 -Asp 698 ) (14). In addition, we examined the interaction between Trp4 and calmodulin (CaM), which has been shown to bind to the C termini of Drosophila Trp (25) and TrpL (26,27) and has been implicated to cause inactivation of TrpL (28). We report here the presence of two CaM-binding and at least two IP 3 R3binding sites at the C terminus of Trp4. The first CaM-binding site overlaps closely with one of the IP 3 R3-binding site. Common binding sites for CaM and IP 3 Rs also exist in other Trp proteins. In functional studies, we show that currents are activated in inside-out membrane patches excised from Trp4expressing HEK293 cells by calmidazolium (CMZ), an antagonist of CaM, and by a peptide representing one of the Trpbinding domains of IP 3 R3.

MATERIALS AND METHODS
DNA Constructs-Fragments of IP 3 Rs and Trps were generated by polymerase chain reaction (PCR). All sense primers contain an NcoI recognition site at the 5Ј with ATG in frame with the codon for the first amino acid. The antisense primers start with an A nucleotide followed by the antisense codon for the last amino acid. PCR products were subcloned into pCRII (Invitrogen), and nucleotide compositions were confirmed by DNA sequencing. Glutathione S-transferase (GST) fusion constructs were made by subcloning NcoI/EcoRI fragments into a modified pGEX4T-1 vector (Amersham Pharmacia Biotech), in which an NcoI site was added after the BamHI site. By design, the insert in each fusion protein starts with a Met and ends with a "TAA" stop codon (T comes from the A in the antisense primer, while AA comes from the pCRII vector). Complementary DNA for enhanced blue fluorescence protein (EBFP) was excised from pEBFP-C1 (CLONTECH) and subcloned into pAGA (29) at NcoI/SmaI sites, while that for maltosebinding protein (MBP) was amplified from pMAL-C2 (New England Biolabs) by PCR, excised with MfeI/NcoI, and subcloned into pAGA at EcoRI/NcoI sites. Constructs for EBFP fusion proteins contained inserts downstream from the EBFP coding region and were made using existing restriction sites on mouse trp4 at the 5Ј-end and XbaI at the 3Ј-end. MBP fusion proteins contain the first 322 residues of MBP followed by inserts, of which the cDNAs were subcloned into NcoI/EcoRI sites.
In Vitro Binding Experiments-Preparation of GST fusion proteins; 35 S-labeled CaM, EBFP, and MBP fusion proteins; and procedures for pull-down experiments are as described (30). For interaction with CaM, CaM-Sepharose (Amersham Pharmacia Biotech) was used. The binding buffer used for IP 3 R3-Trp interaction contains 100 mM KCl, 2 mM MgCl 2 , 0.5% Lubrol, and 20 mM Tris-HCl, pH 7.5. The buffer used for CaM-Trp interaction generally contains 120 mM KCl, 0.5% Lubrol, 20 mM Tris-HCl, 10 mM EGTA, 9.96 mM CaCl 2 . pH was adjusted to 7.5. The estimated free Ca 2ϩ concentration for the buffer is 10 M. However, because EGTA is a poor buffer for Ca 2ϩ concentration at the micromolar range, the actual free Ca 2ϩ concentration of this buffer is about 70 M as determined by spectrofluorometric measurements using Fura2FF (TEF Laboratories, Austin, TX) as a low affinity Ca 2ϩ indicator and HEDTA-buffered solutions as standards.
For the determination of Ca 2ϩ -dependence of Trp binding to CaM, HEDTA or nitrilotriacetic acid instead of EGTA was used, and total CaCl 2 was added according to the MaxChelator program (C. Patton, Stanford University) to give rise to desired free Ca 2ϩ concentrations. 35 S-Labeled MBP fusion proteins containing the CaM-binding sites were incubated with CaM-Sepharose at room temperature for 30 min in varying free Ca 2ϩ concentrations. Each sample was washed twice with the same binding buffer that was used for the incubation; thus, the free Ca 2ϩ concentration was kept unchanged. Bound proteins were subjected to SDS-polyacrylamide gel electrophoresis. The radioactivity of [ 35 S]MBP fusion proteins retained was quantified by phosphorimaging analysis. The percentage of maximal increase over the value obtained in 10 mM EGTA with no added Ca 2ϩ was fitted with the Hill equation I/I max ϭ C n /(C n ϩ (K1 ⁄2 ) n ), where I/I max represents the relative binding, C is the Ca 2ϩ concentration, n is the Hill coefficient, and K1 ⁄2 is the Ca 2ϩ concentration that gives rise to half-maximal binding.
For competition studies, recombinant human CaM was prepared from bacterial lysate using phenyl-Sepharose (Sigma) as described (31). Peptide F2v (EYLSIEYSEEEVWLTWTD) was synthesized by Research Genetics. CaM or peptide F2v, in the desired final concentrations, was included in the binding and washing buffers containing 50 or 70 M free Ca 2ϩ .
Affinity Measurement for Trp-CaM Interaction-All fluorescence measurements were performed on a PerkinElmer Life Sciences LS5 Spectrofluorometer at 22°C. Peptides for CaM-binding sites of Trp1-7 were synthesized by Waterloo Peptide Synthesis (University of Waterloo, Ontario, Canada). Peptide for the second CaM-binding site of mTrp4␣ was synthesized by the Tufts University Core Facility (Tuft University, Boston, MA). Phosphodiesterase activity was assayed by monitoring the hydrolysis of fluorescent 2Ј-methylanthraniloyl cGMP (8 M) in 1 ml of solution containing 200 mM MOPS, pH 7.0, 90 mM KCl, 3 mM MgCl 2 , 2 mM EGTA, 100 M free Ca 2ϩ , 25 nM CaM, and the desired amount of peptide. Samples were excited at 330 nm, and emission at 450 nm was measured. The dissociation constant (K d ) with Ca 2ϩ /CaM was calculated from the activation curves of phosphodiesterase by CaM in the absence and the presence of the peptide as described (32).
Cell Lines and Electrophysiology-HEK293 cells stably expressing mTrp4 (T4-1 and T4-60 cells) and culture conditions were as described (33). Cells were seeded in 35-mm dishes 2 days prior to patch clamp recordings. Conditions for recording from inside-out patches were essentially as described (34). The pipette solution contained 140 mM Na-Hepes, 5 mM NaCl, and 2 mM CaCl 2 , pH 7.5. Inside-out patches were excised from the Trp4 cells into a Ca 2ϩ -free intracellular solution containing 140 mM potassium gluconate, 5 mM NaCl, 1 mM MgCl 2 , 5 mM EGTA, 10 mM Hepes, pH 7.5. CMZ, CaM, and peptide F2v were diluted to the final concentration in the intracellular solution containing no Ca 2ϩ (for F2v) or 18 M free Ca 2ϩ (for CMZ) and were applied to the cytoplasmic side of excised patches through perfusion. A step protocol containing continuous alternating steps to 40 and Ϫ40 mV, each for 1 s, from holding at 0 mV was applied throughout the experiments. Currents were recorded at sampling frequency of 5 kHz and filtered at 1 kHz. Currents recorded at Ϫ40 mV from 100 episodes under each condition were averaged using pCLAMP 8 (Axon Instruments) after digital filtering at 500 Hz and base-line adjustment.

RESULTS
The C Terminus of Trp4 Binds to IP 3 R3 and CaM-A previous study showed that a GST fusion protein containing the F2q fragment of human IP 3 R3 (Glu 669 -Asp 698 ) interacted with the C terminus of hTrp3 (14). The Trp3 peptides shown to interact with IP 3 R3 were T3C7 (Met 742 -Glu 795 ) and T3C8 (Gln 777 -Asp 797 ). T3C8 bound to IP 3 R3 more weakly than T3C7 and thus represents a partial IP 3 R-binding domain of Trp3. Because the sequence homology at the regions that align to T3C7 is relatively low (13,13, and 17% identical between Trp3 and Trp1, Trp2, and Trp4, respectively), it was difficult to predict whether the IP 3 R-binding domain is conserved among all Trp homologues. Therefore, we tested the binding of GST-IP 3 R3F2q to mTrp4. 35 S-Labeled MBP and EBFP fusion proteins containing the N and C terminus of mTrp4, respectively, were synthesized in vitro and incubated with GST or GST-IP 3 R3F2q bound to glutathione-Sepharose. Fig. 1A shows the Trp4 regions present in the fusion proteins, and Fig. 1B shows that while MBP (M) or EBFP (E) alone and MBP-Trp4 N terminus (NT) have no specific interaction with IP 3 R3F2q, the EBFP-Trp4 C-terminal fusion protein (CT) binds to GST-IP 3 R3F2q. To examine the interaction between Trp4 and CaM, we incubated the fusion proteins with CaM-Sepharose. Only EBFP-T4CT (CT) interacted with CaM. The binding of T4CT to CaM was very weak in the absence of Ca 2ϩ , but in the presence of 70 M Ca 2ϩ , it was greatly increased (Fig. 1B).
We divided T4CT into two portions. CT1 (residues 659 -750) includes the sequence homologous to T3C7 of Trp3, while CT2 (residues 733-974) contains the more C-terminal sequence unique to Trp4. Interestingly, both CT1 and CT2 bind strongly to GST-IP 3 R3F2q and to Ca 2ϩ /CaM. Another N-terminal construct (NT1) and the transmembrane region (TM) of Trp4 did not interact with IP 3 R3F2q or Ca 2ϩ /CaM (Fig. 1C). The overlapping sequence between CT1 and CT2 is relatively conserved among the Trp homologues (11 out 20 amino acids are identical between Trp3 and Trp4). However, an EBFP fusion protein containing this sequence (Cf) did not interact with either IP 3 R3F2q or Ca 2ϩ /CaM (Fig. 1D). On the other hand, a con-struct that contains mostly the sequence equivalent to T3C7 but not T4Cf (Cc) interacted with both IP 3 R3F2q and Ca 2ϩ / CaM. These results indicate that just as T3C7, a similar region at the C terminus of Trp4 also binds to IP 3 R3. In addition, there is at least one additional IP 3 R-binding site more C-terminal to the previously defined site. Moreover, both IP 3 Rbinding regions bind to Ca 2ϩ /CaM.
The First CaM-binding Site of Trp4 Closely Overlaps with an IP 3 R-binding Site-To determine how close the first CaM-binding site of Trp4 is to the site that binds to IP 3 R3F2q, we produced MBP-and EBFP-fusion proteins containing smaller fragments of T4Cc and tested their binding to GST-IP 3 R3F2q and Ca 2ϩ /CaM. Fig. 2 shows the composition of the fusion proteins and the binding results. In general, Trp4 fragments that bound to IP 3 R3F2q also bound to Ca 2ϩ /CaM. The smallest fragment that binds to IP 3 R3F2q and Ca 2ϩ /CaM with nearly the same strength as T4Cc is T4Cw (Arg 695 -Glu 724 ). Removing two residues from either the N or the C terminus of T4Cw reduced the interaction with both IP 3 R3F2q and Ca 2ϩ /CaM (Fig. 2C, T4Cv and T4Cx). Smaller fragments, such as Ca, Caa, and Cab, had much weaker binding to IP 3 R3F2q, and this was also accompanied by greatly reduced binding to Ca 2ϩ /CaM. Therefore, the binding sites for CaM and for the IP 3 R cannot be separated by deletion studies, indicating that a common sequence, referred to as the CaM/IP 3 R binding (CIRB) domain, is involved in the interaction with CaM and with the IP 3 R.
The Second CaM-binding Site Is Present in Trp4␣ but Not Trp4␤-The minimal domains that bind to CaM and IP 3 R3F2q at the CT2 region were sought in a similar manner (Fig. 3).
However, it appears that the CT2 IP 3 R-binding domain is not confined to a short sequence like the CIRB domain. When CT2 was divided into CT2a and CT2b, both retained weak interaction with IP 3 R3F2q (Fig. 3B). Further analyses suggested that the region from Gly 781 to Ser 864 retains significant binding to IP 3 R3F2q (Fig. 3C). Interestingly, molecular cloning and immunoblot analysis have revealed the presence of two major isoforms of Trp4, Trp4␣ and Trp4␤ (33). The ␤ form lacks the 84 amino acids corresponding to Gly 781 -Ser 864 of the ␣ form. Thus, CT2␤ binds very weakly to IP 3 R3F2q (Fig. 3B).
In contrast to IP 3 R3F2q, CaM binds to a short, confined sequence in CT2, CT2p (Arg 787 -Asn 812 , Fig. 3, C and D). Removing two residues from its N terminus (CT2r) nearly abolished the binding, whereas removing two residues from its C terminus (CT2q) reduced the binding by 45%, as determined by phosphorimaging analysis from two independent experiments. Although CT2p is contained within the CT2 IP 3 R3F2q-binding region, it does not bind to the IP 3 R fragment. Therefore, the binding sites for CaM and IP 3 R3F2q in the CT2 region do not overlap as closely as the CIRB sequence. Because the CaM binding site is not present in TRP4␤, CT2␤ does not bind to Ca 2ϩ /CaM (Fig. 3B).
CIRB Domain Is Conserved among Trp Proteins-As shown in Fig. 4, regions equivalent to T4Cw from hTrp1, mTrp2, hTrp3, mTrp5, mTrp6, mTrp7, and Drosophila Trp and TrpL also interacted with both GST-IP 3 R3F2q and Ca 2ϩ /CaM. For TrpL, the CIRB domain is in addition to the two CaM-binding sites (710 -725, 854 -875) identified previously (26,27). For Drosophila Trp, the CIRB domain is within a previously reported CaM-binding region (683-976) (25). For Trp3, the CIRB domain is the only CaM-binding domain, because deletion of this site eliminated the binding and the regulation of Trp3 activity by Ca 2ϩ /CaM (34). This may also be the case for Trp6 and Trp7, since they are very similar to Trp3. Trp1 terminates soon after the CIRB sequence and therefore is unlikely to have a second binding site at the C terminus. The C-terminal sequence of mTrp2 (residues 944 -1072) and of mTrp5 (residues 762-975) interacted with both Ca 2ϩ /CaM and GST-IP 3 R3F2q (not shown), suggesting that the second binding sites for the two modulators are conserved in Trp2, Trp4␣, and Trp5, despite the fact that the homology among the three Trps is very low at these regions.
Using synthetic peptides representing the CIRB domains of Trp1-7 and the second CaM-binding site of Trp4, we deter-mined the affinities to range from 10 nM for mTrp2 to 290 nM for mTrp6 (Table I). The second CaM-binding site of Trp4 has lower affinity (600 nM) than the CIRB domains of all Trp proteins. The apparent affinities for Ca 2ϩ , as measured by in vitro binding assays using [ 35 (Table I). The Hill coefficients of Ca 2ϩ dependence for the CaM-Trp interactions were about 1.5-3.5.
All IP 3 Rs Interact with Trp3-Just as the CIRB domain is conserved in all Trp homologues, the F2q Trp-binding site is also conserved in the three known mammalian IP 3 Rs. Fig. 5A shows that regions homologous to IP 3 R3F2q from IP 3 R1 and IP 3 R2 also interacted with the C terminus of Trp3. The alignment of the three sequences indicates that only the C-terminal halves are conserved. Further experimentation using smaller segments of IP 3 R3F2q fused to GST showed that the C-terminal half of F2q (represented by F2v) is the minimal domain of IP 3 R3 that interacts with Trp3 (Fig. 5B).
Competition between CaM and IP 3 R3 for Binding to Trps-In in vitro binding assays, we have shown that for T3C7, the interaction with IP 3 R3F2q was inhibited by purified recombinant human CaM and the binding to CaM was blocked by a synthetic peptide, composed of the sequence of F2v (34). Peptide F2v appears to be a potent inhibitor for CaM binding to Trp3, because the competition was observed in a binding buffer that contained Ca 2ϩ , a condition under which CaM has a high affinity for Trp3. In order to learn the relative abilities of CaM to compete with the IP 3 R for binding to different Trps, we tested the competition between CaM and IP 3 R3F2q for binding to the CIRB domains of other mammalian Trp homologues. The addition of CaM reduced the binding of 35 S-labeled MBP-T4Cw to GST-IP 3 R3F2q in a dose-dependent manner, with 50% inhibition (IC 50 ) occurring at 2.1 M (Fig. 6A), similar to the value obtained for Trp3 (1.2 M) (34). The inhibition of the Trp4-IP 3 R3 interaction by 20 M CaM was about 73%, slightly less than that of Trp3-IP 3 R3 interaction (88%) (34). In addition, 20 M CaM also inhibited the interaction between IP 3 R3F2q and the CIRB domains of Trp1, -2, -5, -6, and -7 to various degrees (Fig. 6B). The effect of CaM was Ca 2ϩ -dependent, since the inhibition by CaM was not observed in a Ca 2ϩ -free binding buffer (Fig. 6B). These results indicate that at high Ca 2ϩ concentrations, CaM competes with IP 3 Rs for binding to the CIRB domains of all Trp proteins. However, the extent of inhibition varies from Trp to Trp, which is probably related to the differences in their affinities to CaM. Although the CT2 IP 3 R-binding site does not overlap as closely to the CaM-binding site as the CIRB sequence, the binding between CT2c and IP 3 R3F2q was also blocked by CaM (Fig. 6C).
Peptide F2v blocked the interaction between CaM and the CIRB domain of Trp1-7 with IC 50 values ranging from 1.7 to 90 M (Fig. 6D). The peptide is more effective in inhibiting the interaction between CaM and Trp3, -6, and -7 than that between CaM and Trp1, -2, -4, and -5. The effect of the peptide is specific because the binding of CaM to Trp4CT2k (residues 781-814), which does not interact with IP 3 R3F2q (not shown), was not inhibited (Fig. 6D).
Peptide F2v and a CaM Antagonist Activate Trp4 in Excised Inside-out Patches-CaM bound to the CIRB domain has a general inhibitory function on Trp3, because in inside-out patches excised from HEK293 cells expressing Trp3, removing or inactivating CaM led to a large increase in Trp3 activity (34). In the absence of Ca 2ϩ and at 5 M, peptide F2v activated Trp3 by competing with CaM for binding to the CIRB domain. The activated channel was blocked by CaM (34). Consistent with the finding that F2v was 10 times less effective in competing with CaM for binding to Trp4 than to Trp3 (Fig. 6D), we found that in inside-out patches, F2v was also less potent in activating Trp4 than Trp3 (Fig. 7A). At 5 M, F2v only caused a small increase in Trp4 activity, which is significantly different (p Ͻ 0.05) from the basal activity in one Trp4 cell line (T4-1) but not in the other one (T4-60). When the concentration of F2v was increased to 50 M, both cell lines showed significant increase in activity (p Ͻ 0.01), which is more than 6 times higher than that stimulated by 5 M F2v. Additionally, patches excised from Trp4 cells were activated by 1 and 10 M CMZ (Fig. 7B). Under the same conditions, untransfected HEK293 cells did not show any significant increase in activity after treatment with peptide F2v or CMZ (Fig. 7, A and B). DISCUSSION The mechanism of activation for SOCs remains mysterious. Three major hypotheses have been proposed. The first assumes that a small diffusible soluble factor capable of stimulating SOCs appears in the cytosol upon store depletion. Although it has been shown that acid extracts from activated Jurkat cells and platelets or a Ca 2ϩ store-depleted mutant yeast strain contain such a factor, which when applied to naive cells could stimulate Ca 2ϩ influx (35,36) or cation conductance (37), the identity of the Ca 2ϩ influx factor has not been determined. The second hypothesis claims that a secretion-like process involving the insertion of channel-containing vesicles into the plasma membrane is required for activating SOCs (38). In agreement with this is the finding that actin redistribution affected capacitative Ca 2ϩ entry (39). The third hypothesis is called conformational coupling and is modeled after the well known mechanism of excitation-contraction coupling between the L-type Ca 2ϩ channel and the ryanodine receptor in skeletal muscle (40). In this case, IP 3 Rs are thought to serve not only as channels for Ca 2ϩ release but also as sensors for store depletion. The signal of store depletion is sent to the plasma membrane Ca 2ϩ entry channels via a direct protein-protein interaction (2,41). Consistent with this hypothesis, it has been demonstrated that IP 3 activates cation channels on the plasma membranes of endothelial cells, macrophages and A431 epithelial cells (42)(43)(44) and that the activity of IP 3 R1 purified from rat cerebellum is modulated by luminal Ca 2ϩ (45).
Despite the controversy about whether or not Trp proteins form SOCs, accumulating evidence has indicated that IP 3 and IP 3 Rs are involved in the activation of SOCs as well as Trp3 (12)(13)(14)46). Also, just like the native SOCs, Trp3 activation was prevented by actin filament condensation induced by calyculin A, a phosphatase inhibitor (46). Therefore, the conformational coupling mechanism, and perhaps the secretion-like coupling mechanism as well, are applicable to Trp-based channels. Using molecular and biochemical approaches, Boulay et al. (14) identified the interacting domains of Trp3 and IP 3 R3 and showed that they are involved in the regulation of native SOCs. In the current study, we further demonstrate that the interactions between Trp and IP 3 R are common for all Trp proteins and for all IP 3 Rs. Thus, conformational coupling involving direct physical interaction with activated IP 3 Rs may be a common mechanism for the activation of Trp-based channels. Moreover, we show that the IP 3 R-binding domains of the Trp proteins also bind to Ca 2ϩ /CaM and that the binding by CaM inhibits the association between Trp and IP 3 Rs. Therefore, the competition between CaM and IP 3 Rs for binding to a common site may play a key role in controlling the gating of Trp-based channels. Based on the functional study of Trp3 using insideout membrane patches and a non-Ca 2ϩ -binding CaM mutant, we concluded that at rest, CaM is tethered to the channel and prevents it from being spontaneously active (34). Maneuvers that displaced CaM from Trp3 strongly activated the channel (34). Here, we show that Trp4 is also strongly activated by inactivating CaM with CMZ, indicating that CaM probably plays the same inhibitory role in all Trp-based channels. Consistent with this, CMZ also activates current in patches excised from cells expressing Trp1 or Trp6 (not shown). Thus, binding to CaM is essential to prevent the spontaneous activity of Trp channels.
Analysis of genomic sequences revealed that for hTrp1, hTrp4, hTrp5, Drosophila Trp, and TrpL, the coding sequence for the N-terminal end of the CIRB domain coincides with the a K d for CaM was determined by the phosphodiesterase assay as described under "Materials and Methods" using the peptides shown. b Ca 2ϩ dependence was determined by in vitro binding assays using 35 S-labeled MBP fusion proteins shown in Fig. 4A and CaM-Sepharose. Free Ca 2ϩ concentrations in the binding solutions were buffered by HEDTA or nitrilotriacetic acid (for mTrp5). beginning of an exon, 2 indicating that conformational coupling involving binding by IP 3 Rs and its regulation by CaM were acquired early in evolution and that there may be a constraint to preserve these modes of regulation for Trps. However, the IP 3 R-deficient Drosophila photoreceptors showed normal response to light and unaltered function of Trp and TrpL (47,48), suggesting that the Drosophila IP 3 R is not involved in activating these channels. The fact that Drosophila Trp and TrpL interacted with human IP 3 R3F2q fragment in vitro does not implicate a similar interaction in vivo. Interestingly, a comparison between the Drosophila IP 3 R and the human IP 3 R3 showed a very low homology between the two at the F2q region. An extra 38-amino acid sequence in the Drosophila IP 3 R (725-762) essentially disrupts an otherwise intact F2q region. In addition, the Drosophila sequence contains only one of the two tryptophans critical for binding to Trp3 (34). Therefore, it is unlikely that the Drosophila IP 3 R would be involved in activating the Trp channels in the same manner as its mammalian counterparts. Since only a single gene for IP 3 R is present in the Drosophila genome (48), it is possible that in the insect's photoreceptors, a different protein would compete with CaM for binding to the CIRB domains of Trp and TrpL and hence activate the channels. Furthermore, the mechanism present here is only one of the several ways by which the Trp channels are activated. The involvement of IP 3 Rs in the activation of mammalian Trps does not rule out the participation of other molecules in Trp gating.
In addition to the CIRB domain, binding sites for CaM and IP 3 Rs are present downstream in Trp2, Trp4␣, and Trp5, suggesting that subtype-specific functions carried by CaM and IP 3 Rs also exist and contribute to the diversity in the regulation of Trp-based channels. Interestingly, the last 240 amino acids of human Trp4␣ including the 84-amino acid region that contains the binding sites for CaM and IP 3 Rs are encoded by a single exon. 2 In Trp4␤, the 84-amino acid region is deleted through the use of alternative splicing sites, and thus the second binding sites for CaM and IP 3 Rs are not present. Surprisingly, the same region was recently shown to bind to the C terminus of IP 3 Rs (49). It remains to be determined how the bindings of the N and C terminus of IP 3  region affect the function of Trp4␣ and how they differ from bindings to the more upstream CIRB domain.
Diversity also arises from the differences in the affinities of CIRB domains of different Trp homologues for CaM and for IP 3 Rs. The affinities for CaM differ by as much as 29-fold between the CIRB domains of Trp2 and Trp6 (Table I). Because peptide F2v is more effective in displacing CaM from Trp3 and the closely related Trp6 and Trp7 than from the CIRB domain of other Trps, the affinities of the CIRB domains for the Trpbinding site of IP 3 R3 may also be very different. In the context of conformational coupling, one important step accomplished by IP 3 Rs is to displace the inhibitory CaM from the CIRB domain. This may be accomplished mainly by the binding of the F2v domain of IP 3 R3 and homologous regions of IP 3 R1 and IP 3 R2. The effectiveness of the F2v may be dependent on local Ca 2ϩ concentrations, since Trp binding to CaM is greatly enhanced by Ca 2ϩ . In the binding assay shown in Fig. 6D, 70 M free Ca 2ϩ was used to facilitate the detection of CaM binding. Under this condition, peptide F2v blocked CaM binding to the CIRB domain of Trp3, -6, and -7 more effectively than to that of Trp1, -2, -4, and -5, suggesting that the closely related Trp3, -6, and -7 have higher affinity for the F2v domain and thus may be more sensitive to the activation by IP 3 Rs than the rest of the Trp homologues. However, it is important to note that under resting Ca 2ϩ concentrations of ϳ0.1 M, all CIRB domains would prefer binding to F2v in the in vitro binding assay. Because it is not clear how CaM is tethered to the channel under resting conditions when the cytosolic Ca 2ϩ concentration is low, it is difficult to predict whether the F2v domain alone is sufficient to displace CaM from all Trp channels in excised patches. The fact that a higher concentration of peptide F2v is needed to activate Trp4 than to activate Trp3 agrees well with the binding data, suggesting that when acting alone, F2v is a weak competitor for the CaM-Trp4 interaction. This supports our conclusion that peptide F2v activates Trp3 by displacing CaM from the common binding site.
CaM regulation through competition with another protein appears to be a common mechanism shared with other Ca 2ϩpermeable channels. Similar to Trp-based channels, a C-terminal site of the N-methyl-D-aspartate receptor binds to Ca 2ϩ / CaM and actin-associated ␣-actinin. CaM inhibits whereas ␣-actinin promotes channel opening (50,51). A common binding site for Ca 2ϩ /CaM and the C terminus of the olfactory cyclic nucleotide-gated channel has been found at the N terminus of the same channel. CaM binding disrupts the interdomain coupling between the N and C terminus and inactivates the channel (52). For these channels, it is not known whether CaM is tethered to the channels at rest and whether it plays any functional role without the presence of any physiological stimulus. To date, only Trp channels have been shown to be activated by removing or inhibiting CaM. The relatively harsh treatment may be effective in releasing the channel from inhibition by CaM, but this does not result in a complete activation of Trp channels (34). A full stimulation of Trp channels should therefore include multiple events, such as the activation of IP 3 Rs, Ca 2ϩ store-depletion, and perhaps the increase in the level of diacylglycerol.
Native SOCs may be regulated by IP 3 Rs and CaM in a similar manner. However, because the single channel conductance of the endogenous channel is very low (12) and the channel density is much lower than the expressed Trp channels, it has been difficult for us to determine confidently whether or not similar mechanisms that stimulate the Trp channels also stimulate the native SOCs in control HEK cells. In A431 epithelial cells, IP 3 has been shown to activate SOCs via IP 3 Rs (53). In whole-cell studies, intracellular infusion of CaM inhib-ited the SOC in bovine endothelial cells by slowing down the activation and speeding up the inactivation (54). Ca 2ϩ -dependent inactivation has been documented for SOCs found in a number of different cells (55,56). Thus, conformational coupling involving IP 3 Rs and its negative control via CaM may be a primary mechanism of gating for SOCs.