Article

Top Summary Article References

SOCE Is a Widespread Phenomenon

Ca²⁺ entry is critical for the development and/or physiology of virtually all cells. A diversity of voltage- and ligand-gated ion channels are responsible for Ca²⁺ influx in different types of cells; however, there are similarities in the mechanisms of Ca²⁺ influx among highly divergent cell types, such as excitable versus nonexcitable cells, that are greater than recognized previously. This common Ca²⁺ entry process, SOCE (formerly referred to as capacitative Ca²⁺ entry) (1, 2), has been observed in a plethora of cell types ranging from T lymphocytes, pancreatic acinar cells, hepatocytes, and vascular endothelial cells to Drosophila photoreceptor cells (reviewed in Ref. 3). This mode of Ca²⁺ entry may be ubiquitous because it has been detected in most cells that have been specifically examined for SOCE. Moreover, SOCE has been implicated in a variety of processes ranging from mitogenesis in fibroblasts, osteoclast function, and Drosophila phototransduction to T cell activation (reviewed in Ref. 3). This latter possibility is strengthened by the intriguing correlation between a form of severe immunodeficiency and a defect in SOCE in T cells (4).

SOCE is activated by depletion of Ca²⁺ from internal Ca²⁺ storage sites. Although the exact location of such pools remains elusive, it seems that portions of the endoplasmic reticulum compose a major component of these stores (reviewed in Ref. 5). Release of Ca²⁺ from the stores results from activation of tyrosine kinase receptors or G protein-coupled receptors, which in turn stimulates PLC-. Production of IP₃ by PLC induces a biphasic rise in Ca²⁺. The initial phase is transient and caused by activation of the IP₃R, a Ca²⁺-release channel situated in the Ca²⁺ stores. Then, depletion of Ca²⁺ from the internal stores induces a more sustained plasma membrane Ca²⁺ conductance. It may be that not all IP₃-sensitive Ca²⁺ stores are involved in SOCE. Some IP₃-sensitive stores may be involved only in Ca²⁺ release, whereas depletion from other stores may function in SOCE (6). Despite the apparent prevalence and importance of SOCE, the mechanism by which depletion of Ca²⁺ from the internal stores activates the plasma membrane Ca²⁺ conductance is not understood. Moreover, the identity of SOCs expressed in vertebrate cells has, until recently, been elusive.

Inositol Phospholipid Signaling and Ca²⁺ Stores in Drosophila Vision

The founding member of a family of related SOCs has emerged through a genetic approach to the study of vision in the fruitfly Drosophila melanogaster. As is the case in vertebrates, Drosophila vision is initiated by photoactivation of rhodopsin and interaction with a heterotrimeric G protein. However, in contrast to vertebrate vision, phototransduction in the fruitfly utilizes the inositol phospholipid signaling system. Definitive evidence that this is the case is the observation that null mutations in norpA, which encodes the eye-enriched PLC, render the flies completely unresponsive to light (7). Upon activation of phosphoinositide signaling, there is sustained plasma membrane Ca²⁺ and Na⁺ conductance.

Release of Ca²⁺ from internal stores seems to be essential for Drosophila vision, although this issue is controversial. Evidence has been provided that Drosophila photoreceptor cells contain Ca²⁺ stores (8) and receptors for both IP₃ (9, 10) and ryanodine (10). The ryanodine receptor is another Ca²⁺-release channel which is activated through Ca²⁺-induced Ca²⁺ release. Inhibition of the IP₃R or exhaustion of the ryanodine pools with ryanodine blocks the photoresponse, indicating that Ca²⁺ depletion via both the IP₃- and ryanodine-sensitive stores is required in Drosophila vision (11, 12). It has been proposed that release of Ca²⁺ from the IP₃-sensitive stores triggers Ca²⁺-induced Ca²⁺ release from the ryanodine-sensitive stores, resulting in amplification of the light response (11).

Despite the pharmacological evidence for a role of the IP₃-sensitive stores in Drosophila phototransduction, a mutation in a Drosophila IP₃R has been reported to be lethal (13, 14) but to have no effect on vision (14). Due to the requirement of the IP₃R for viability, the eye phenotype was analyzed in mosaic flies containing mutant patches in the compound eye of the fly. An additional observation suggesting that Ca²⁺ stores do not function in activation of the light-induced conductance is that thapsigargin, an agent which interferes with the Ca²⁺-ATPase in the endoplasmic reticulum and causes Ca²⁺ depletion via leak currents (15), does not activate Ca²⁺ influx in photoreceptor cells (8, 16). However, the lack of effect of thapsigargin may be due to insufficient leak current from the stores in Drosophila photoreceptor cells. Furthermore, several lines of evidence indicate that the absence of phenotype in the IP₃R mosaics eyes is due to the presence of one or more additional IP₃Rs that remain to be identified: (i) norpA flies are blind, demonstrating a crucial role for IP₃ and/or diacylglycerol in activation of the light-induced conductance (7); (ii) application of heparin, an IP₃R blocker, interferes with the photoresponse (12); and (iii) thapsigargin activates the light-sensitive channels in vitro (see below). Based on a recent study in a lymphocyte cell line suggesting that there are separate IP₃-sensitive pools which function in Ca²⁺ depletion and influx (6), it is plausible that a similar phenomenon exists in Drosophila photoreceptors and that separate IP₃Rs function in each process. Alternatively, there may indeed exist just one Drosophila IP₃R, and the lack of phenotype in the genetic mosaics may be the consequence of residual IP₃R protein in the mutant patches.

An alternative possibility is that there are dual mechanisms for activating the light-dependent current, both of which require the NORPA PLC and production of IP₃. The initial light-dependent conductance may be induced by a very rapid mechanism independent of the stores, and the sustained response may involve the internal stores. This latter mechanism involving Ca²⁺ depletion would require diffusion of IP₃ from the rhabdomeres to the cell bodies, Ca²⁺ release, and production of a signal that is sent back to the rhabdomeres, leading to production of the cation conductance. The rhabdomeres, which are the microvillar structures containing most of the components required in Drosophila phototransduction, do not seem to contain Ca²⁺ stores. Thus, an involvement of Ca²⁺ depletion in the light response would necessitate signal transmission to and from the cell bodies. However, the light-dependent conductance is activated within a few milliseconds, and this time scale may be too rapid for information to flow between the rhabdomeres and cell bodies. Therefore, it is appealing to propose that production of IP₃ may activate the initial phase of Ca²⁺ and Na⁺ influx by directly binding the light-sensitive channels or binding to a channel-associated protein in the rhabdomeres. The IP₃-sensitive Ca²⁺ stores would be proposed to function in sustaining the cation influx for an extended duration.

Identification of TRP and TRPL

Many mutants affecting Drosophila vision have been identified in genetic screens using either ERGs or behavioral assays, such as phototaxis (reviewed in Ref. 17). A locus identified in one such screen, referred to as trp, seems to encode a SOC subunit. trp mutant flies display normal phototaxis but behave as though blind in the optomotor assay, a test in which wild-type flies orient to visual cues (18). The impaired optomotor response is associated with an ERG phenotype (Fig. 1). The initial response to light is indistinguishable between trp and wild-type flies. However, in response to bright light, the sustained component of the ERG quickly decays. No difference between the trp and wild-type ERG is detected if dim or moderate light intensity is used as a stimulus.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   The trp ERG phenotype. Wild-type (A and B) and trpCM (C and D) flies were exposed to a white light. The flies were dark adapted for either 2 min (A and C) or 5 sec (B and D). The initiation (on) and cessation (off) of the light stimulus, a signal amplitude, and time scale are indicated. [Figure reprinted with permission from Montell, C., K. Jones, E. Hafen, and G. Rubin. Rescue of the Drosophila phototransduction mutation trp by germline transformation. Science (Washington D. C.) 230:1040-1043 (1985). © 1997 American Association for the Advancement of Science.]

The basis of this ERG phenotype was greatly clarified upon the advent of technology to perform whole-cell recordings on Drosophila photoreceptor cells (19, 20). In wild-type, the channels are ~25-40-fold more permeable to Ca²⁺ than to Na⁺ (20, 21); however, in trp photoreceptors, the permeability to Ca²⁺ is reduced ~10-fold, whereas the permeability to Na⁺ is decreased by only ~30% (21). A defect in Ca²⁺ entry is further supported by experiments using Ca²⁺-sensitive electrodes and Ca²⁺-indicator dyes in photoreceptor cells (22, 23). These data indicate that the trp locus either encodes a Ca²⁺ channel or is required for activation of such a channel. In addition, the observation that the cation conductance is reduced but not eliminated in trp flies demonstrates that at least one additional class of channel is still functional in the mutant flies.

A suggestion that trp might encode an ion channel was obtained through molecular identification of the gene (24). trp encodes a photoreceptor cell-specific 1275-amino acid protein with multiple transmembrane domains, most likely six (Fig. 2) (25, 26); this predicted structure is reminiscent of the superfamily of voltage- and second messenger-gated ion channels (27). In addition, analysis of the deduced amino acid sequence suggested that if TRP is a channel, it is not voltage gated because the fourth putative transmembrane segment lacks the positively charged residues believed to form the voltage sensor in voltage-gated ion channels (28).

View larger version (27K): [in this window] [in a new window]   Fig. 2.   Domain structures of TRP, TRPL, cTRP, and TRPC1. The putative ankyrin repeats, putative transmembrane domains (TMD), calmodulin binding sites (CaM), putative PEST degradation signal (PEST), region contain multiple repeats of the dipeptide lysine and proline (KP), domain containing nine tandem copies of an eight amino acid repeat, D-K-D-K-K-P-G/A-D (8 × 9), and a putative nucleotide binding site (N) are indicated. Also indicated in TRPC1 is a region with a high probability of forming a coiled-coiled (cc) structure and a segment near the carboxyl terminus with weak homology to dystrophin (D). The third putative ankyrin repeat is missing in one of the two isoforms of TRPC1.

In a screen for calmodulin binding proteins expressed in the Drosophila head, an 1124-residue protein was identified, referred to as TRPL, that is ~40% identical to TRP over the amino-terminal ~700 amino acids but differs significantly from TRP over the carboxyl-terminal ~400 residues (Fig. 2) (29). The identification of TRPL raised the possibility that it is a second light-dependent ion channel, perhaps responsible for the remaining conductance in a trp mutant background. This explanation seems to be an oversimplification because trpl mutant flies have become available that are reported to elicit a light response indistinguishable from that of wild-type, yet trpl; trp double mutants are nearly completely unresponsive to light (30). A possible explanation for this conundrum is discussed below.

TRP and TRPL Form Channels but Only TRP Is Store Operated

Consistent with the analysis of the deduced amino acid sequence, expression of TRP in vitro, such as in Sf9 or 293T cells, results in a Ca²⁺ conductance that is not voltage dependent. However, a cation conductance is induced upon depletion of Ca²⁺ from the internal stores with thapsigargin (Fig. 3) (31, 32). The observation that TRP is activated in vitro by depletion of Ca²⁺ from the internal stores strongly indicates that it is a bona fide SOC. TRP seems to function as a multimer, similar to other members of the superfamily of voltage- and second messenger-gated ion channels (reviewed in Ref. 27), because the TRP-dependent conductance is almost completely inhibited by a dominant negative form of TRP (32). Such derivatives consist of a substitution of the putative "pore-loop" domain, between transmembrane segments 5 and 6, with the corresponding region from the Shaker B voltage-gated K⁺ channel.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   TRP, TRPL, and TRP/TRPL currents in 293T cells before and after thapsigargin treatment. The cDNA constructs pTRP and pTRPL were transfected either singly or in combination as indicated. Na+ currents were recorded either before or after depletion of the intracellular Ca2+ stores by perfusing the cells for 7-10 min with a Ca2+-free solution containing 1 µM thapsigargin. Traces, currents produced using a set of step voltages from 100 to +80 mV in 15-mV increments. The holding potential was 10 mV. Dotted lines, 0 pA. [Adapted from Xu, X.-Z. S., H.-S. Li, W. B. Guggino, and C. Montell. Coassembly of TRP and TRPL produces a distinct store-operated conductance. Cell 89:1155-1164 (1997). © 1997 Cell Press.]

The ion selectivity and pharmacology of TRP expressed in vitro have parallels with in vivo currents in Drosophila photoreceptor cells. The relative permeability of Ca²⁺ is higher than that of Na⁺ both in vivo (~25-40:1) (20, 21) and in vitro (~10:1) (32). The TRP channel is also less permeable to Mg²⁺ than to Ca²⁺ in both photoreceptor and tissue culture cells (21, 31). Thus, the TRP conductance has selectivity for Ca²⁺ but is not nearly as selective as one of the best characterized store-operated conductances in vertebrate cells, I_CRAC (33-35). However, a wide variety of vertebrate store-operated conductances have been described. Some are Ca²⁺ selective but display higher conductances than I_CRAC (36); others are similar to TRP in their relative Ca²⁺-to-Na⁺ permeability (37); and others are nonselective (38). Thus, Ca²⁺ selectivity is not a prerequisite for a store-operated conductance.

Despite the higher permeability to Ca²⁺ than Na⁺, the TRP-dependent currents measured in vitro are larger when Na⁺ rather than Ca²⁺ is used as the permeant ion. This phenomenon seems to be due to Ca²⁺-mediated inactivation of TRP, another feature of TRP observed in vitro (32) and in photoreceptor cells (39, 40). Furthermore, inhibition by Ca²⁺ may be a common phenomenon among SOCs because I_CRAC is also inactivated by Ca²⁺ (35, 41). In accordance with in vivo studies (42, 43), TRP expressed in 293T cells is blocked by Mg²⁺ and La³⁺ (32). The Mg²⁺ block has been proposed to play an important role in determining the gain, kinetics, and signal-to-noise of phototransduction (42).

Despite the similarities in the pharmacology and permeabilities of TRP in vivo and in vitro, TRP alone does not account for all the characteristics of the light-activated conductance. In contrast to the TRP conductance, the light-activated current is strongly outwardly rectifying (21). Moreover, at least one other channel seems to be present in Drosophila photoreceptor cells because the initial response to bright light is similar for TRP and wild-type flies. TRPL would seem to be a plausible candidate for this other channel; however, TRPL mutant flies fail to exhibit a discernible phenotype (30). Furthermore, TRPL expressed in vitro is constitutively active (Fig. 3) (32, 44-46), although it seems to be further activated by IP₃ (47). Release of Ca²⁺ from the internal stores does seem to have some effect on TRPL activity because the addition of thapsigargin increases the outwardly rectifying current (Fig. 3) (32). There are reports that Ca²⁺/calmodulin and/or direct interaction with G₁₁ functions in TRPL activation (46, 48); however, direct activation of TRPL by G proteins is unlikely to be the main mode of TRPL activation in vivo because PLC is required for the response to light (7).

Despite the observation that TRPL is not store operated, conclusive evidence that TRPL is an ion channel has been obtained using single-channel recordings (48, 49). In addition to the constitutive activity, TRPL differs from TRP in its lack of ion selectivity, with respect to Na⁺, Ba²⁺, and Ca²⁺, and relative insensitivity to inhibition by La³⁺; however, as is the case with TRP, TRPL is blocked by Mg²⁺ (32, 49).

Interestingly, TRPL can be converted into a SOC by fusing the TRPL amino-terminal and transmembrane domains with the region of TRP carboxyl-terminal to the transmembrane domains (50). As is the case with the TRPL conductance, the TRPL/TRP current is nonselective. Conversely, a constitutively active conductance with an ion selectivity typical of TRP is produced by joining the amino-terminal and transmembrane regions of TRP with the carboxyl-terminal domain of TRPL (50). Thus, the portion of TRP critical for conferring store-operated activity is the region carboxyl-terminal to the transmembrane domains.

TRP and TRPL Heteromultimerize to Produce a Functional SOC

The observation that TRP and TRPL are expressed in the same cells, Drosophila photoreceptor cells, raises the possibility that the two proteins interact directly to form heteromultimeric channels. Consistent with this proposal, TRP and TRPL coimmunoprecipitate from Drosophila photoreceptor cells (32). This interaction is most likely direct, rather than through another protein, because TRP and TRPL associate directly in vitro. Both the amino-terminal and transmembrane segments contribute to the interaction (32).

Coexpression of TRP and TRPL in vitro leads to store-operated currents distinct from currents produced by expression of the individual proteins. In Xenopus laevis oocytes, coexpression of both proteins leads to an inward store-operated Mg²⁺ current that was not detected when the individual channels were expressed separately (51). The Mg²⁺ current is different from the light-dependent conductance in that it does not seem to be permeable to Na⁺ and is detected mainly at hyperpolarizations more extreme than those that occur in vivo. However, introduction of TRP and TRPL into 293T cells produces a store-operated conductance, which exhibits many features of the light-activated current (32). In particular, the TRP/TRPL store-operated current is outwardly rectifying (Fig. 3), more selective for Ca²⁺ than Na⁺, and inhibited by either Ca²⁺ or Mg²⁺ but relatively insensitive to La³⁺. However, the extent of inward rectification produced by expression of TRP/TRPL is more subtle than that of the light-activated conductance. The store-operated activity, ion selectivity, and inhibition by Mg²⁺ are features reminiscent of the TRP-dependent current, whereas the outward rectification and relative insensitivity to La³⁺ are characteristics most similar to those of current generated from TRPL alone. It is possible that TRP/TRPL in vivo has some sensitivity to La³⁺ because TRP expressed in vitro (32) is inhibited to a lesser extent than in vivo (43). Further electrophysiological evidence that TRP and TRPL coassemble is that dominant negative forms of TRP suppress the TRPL conductance and vice versa.

It has been reported that TRP and TRPL are capable of responding independently to light activation (30), but several observations suggest that this is not the case and that TRPL exists solely as a heteromultimer in Drosophila photoreceptor cells (32). First, TRPL expressed in vitro is not a SOC but requires TRP for store-operated activity. Second, although TRPL homomultimers form in vitro, TRPL binds preferentially to TRP. Third, TRP seems to be 10-fold more abundant than TRPL. Thus, two store-operated channels that function in Drosophila photoreceptor cells seem to be TRP homomultimers and TRP/TRPL heteromultimers.

In addition to TRP and TRPL, a third TRP-related channel (TRPR) may function in phototransduction. The existence of a third light-activated channel in photoreceptor cells could account for the seemingly conflicting observations that (i) elimination of TRPL has been reported to have no discernible effect and (ii) the photoresponse is dramatically reduced, although not completely eliminated, in trpl; trp photoreceptors (30). TRPR may heteromultimerize with either TRP or TRPL; if so, the TRP/TRPL and TRP/TRPR complexes may be functionally redundant. Hence, there is no phenotype in trpl photoreceptors. In contrast, TRP/TRPL may not be redundant, with TRPL/TRPR resulting in a significant decrease in the photoresponse in trp flies. The small remaining conductance in trpl; trp flies may be due to TRPR, which displays little activity as a homomultimer.

Possible Mechanisms Coupling TRP and TRP/TRPL Activation to Ca²⁺ Depletion

The mechanism by which depletion of Ca²⁺ from the internal stores is coupled to the opening of SOCs is not known; however, the signal does not seem to be Ca²⁺ because SOCE can be activated in the presence of Ca²⁺ chelators (33). Two proposals have received considerable attention (reviewed in Ref. 3), one of which suggests that depletion of Ca²⁺ from the stores activates SOCs through a small soluble factor, referred to as Ca²⁺ influx factor. Such a factor of a molecular weight of <500 was originally isolated from a lymphocyte cell line, Jurkat cells (52), and has been reported to stimulate Ca²⁺ influx in a variety of cell lines (53, 54). However, other studies have cast doubt on the role of such a factor in Ca²⁺ influx (55, 56). Currently, there are no reports addressing the potential role of the Ca²⁺ influx factor in TRP activation, and further studies will be required to either confirm or dismiss a critical role for the Ca²⁺ influx factor in mediating Ca²⁺ influx via other SOCs.

A second proposal is that there is direct coupling between the IP₃R and SOCs (reviewed in Ref. 3). On depletion of Ca²⁺ from the stores, there is a putative conformational change in the IP₃R, which in turns activates the SOC through direct interaction. In support of this proposal, referred to as the conformational coupling hypothesis, it has been reported that TRP is spatially restricted to the base of the rhabdomeral microvilli adjacent to the presumed position of the IP₃-sensitive Ca²⁺ stores (57). However, subsequent studies indicate that TRP is uniformly distributed throughout the rhabdomeral microvilli (58-60), a spatial distribution which would preclude a direct interaction between the vast majority of TRP and the IP₃R. It remains possible, however, that SOCs other than TRP are activated via conformational coupling with the IP₃R.

An appealing modification of the conformational coupling hypothesis is that the putative conformational change in the IP₃R activates TRP indirectly through the actin-based cytoskeleton. The IP₃R could be linked to the cytoskeleton via ankyrin because there are reports indicating that some mammalian IP₃Rs associate with this adapter protein (61, 62). TRP could also be linked to the actin cytoskeleton because the rhabdomeres contain actin filaments (63) and squid TRP seems to associate with the detergent-insoluble cytoskeletal fraction (64). The putative interaction of TRP with the cytoskeleton could be mediated through one or more of the four ~33-amino acid ankyrin repeats located in the amino terminus of TRP. Furthermore, it is noteworthy that other ion channels, such as the N-methyl-D-aspartate receptor (65) and the amiloride-sensitive Na⁺ channels (66), bind directly to and may be regulated by components of the actin cytoskeleton such as ankyrin and -actinin. Disruption or stabilization of actin filaments, with specific drugs, has been shown to regulate a variety of ion channels in whole-cell and patch-clamp studies (67-69).

INAD Links TRP to Upstream Signaling Proteins

Drosophila TRP forms a large signaling complex with at least five other proteins that are essential for phototransduction. In additional to binding TRPL, TRP has been shown to associate with the major rhodopsin (RH1) (60), the PLC encoded by norpA (59, 60), an eye-specific protein kinase C encoded by the inaC locus (59), and calmodulin (60). Thus, most if not all of the phototransduction cascade may be linked into one supramolecular complex (Fig. 4). The interaction between TRP and calmodulin is direct, although the association with NORPA and RH1 is mediated through INAD (60), a protein with five ~90-amino acid PDZ domains (70-72). PDZ domains are protein interaction motifs which bind a variety of ion channels and other signaling proteins (reviewed in Refs. 70 and 73).

View larger version (39K): [in this window] [in a new window]   Fig. 4.   Model of the signaling complex in Drosophila rhabdomeres. TRP binds directly to calmodulin and INAD and indirectly to NORPA, RH1. and INAC via INAD. The DGq trimeric G protein (Gq, G, and G) is proposed to transiently interact with the complex. [Adapted from Chevesich, J., A. J. Kreuz, and C. Montell. Requirement for the PDZ domain protein, INAD, for localization of the TRP store-operated channel to a signaling complex. Neuron 18:95-105 (1997). ©1997 Cell Press.]

TRP binds directly to INAD, and this interaction is disrupted in an allele of InaD, InaD^P215, which contains a point mutation in the third PDZ domain (72). NORPA (60) also binds directly to INAD, although this has not been demonstrated with RH1, INAC, and calmodulin. In InaD^P215 mutant photoreceptors, TRP no longer associates with either NORPA or RH1, and the spatial distribution of TRP, but not NORPA or RH1, is severely disrupted (60). These data indicate that INAD is a scaffolding protein which serves to link TRP to upstream signaling proteins.

There may be two functions of the phototransduction signaling complex, one of which may be to facilitate feedback regulation. Compartmentalization of TRP with upstream signaling molecules may be to facilitate feedback regulation by highly localized increases in Ca²⁺ due to the TRP-dependent Ca²⁺ influx. The Ca²⁺ fluxes are likely to be spatially restricted as a consequence of extensive buffering by calmodulin and other Ca²⁺ binding proteins. Release of TRP from the signaling complex, as occurs in InaD^P215 photoreceptors, would therefore preclude the feedback regulation by the TRP-dependent Ca²⁺ influx. Consistent with the model that one function of TRP may be for Ca²⁺-dependent feedback regulation are the findings that the activities of proteins associated with INAD, such as NORPA and RH1, may be down-regulated by Ca²⁺ (74, 75). Furthermore, InaD^P215 has been reported to display a slow deactivation of the light-induced current and increased sensitivity to dim light (71). In the absence of extracellular Ca²⁺, the InaD^P215 response is indistinguishable from wild-type (71). Ca²⁺ is also likely to mediate feedback regulation of TRP and TRP/TRPL directly in photoreceptor cells because high Ca²⁺ levels inhibit the light-induced current (20, 39, 76) and both TRP and TRP/TRPL expressed in vitro are inactivated by Ca²⁺ (32). The inactivation of these SOCs may be through association with calmodulin, although this has not been demonstrated; however, the issue of whether the INAD signaling complex has a role in feedback regulation is controversial. It has been reported that InaD^P215 flies seem to display an increase in the mean latency between light stimulation and bump activation rather than a defect in inactivation (77), although it seems plausible that the Ca²⁺-dependent feedback regulation requires greater light intensities than the single photons that give rise to unitary bumps.

A second potential function of the TRP/INAD supramolecular complex may be in activation. Production of IP₃ in close proximity to the light-sensitive channels could potentially result in rapid activation of the cation influx independent of the Ca²⁺ stores. As discussed above, such a mode of activation may operate during the initial light response, and SOCE may be required for a sustained response to light. Consistent with the proposal that the signaling complex also functions in activation, an InaD allele, InaD¹, with a stop codon in the second PDZ domain, displays a severely reduced response to light (77). Thus, the signaling complex may be critical for both activation and feedback regulation of Drosophila vision.

Proteins Related to TRP Are Conserved Throughout Animal Phylogeny

Proteins related to Drosophila TRP and TRPL have been identified in a variety of invertebrates, including a homolog of TRP in the related fly, Calliphora, which is ~85% identical to TRP over a amino-terminal segment of ~900 amino acids and ~40% identical to TRPL over the same region (Fig. 2) (59). However, in the carboxyl-terminal end, the homology between Drosophila and Calliphora TRP falls to <50%. Significant heterogeneity in the carboxyl-terminal region is a recurring theme because TRP-related proteins in the squid (sTRP) (64) and C. elegans (cTRP1) (78) contain only ~800 residues, a length ~450 residues shorter at the carboxyl terminus than Drosophila TRP. These latter proteins are ~40-45% identical to TRP; however, a much more distantly related protein in C. elegans emerged from the Genome Project (cTRP2; accession No. Z66514) which is only ~25% identical to TRP but has an overall predicted topology of six transmembrane domains, similar to other members of the TRP family. The low level of homology in cTRP2 is intriguing because this putative channel may display quite different functional characteristics than Drosophila TRP.

The conservation of TRP in a variety of invertebrates raised the possibility that the TRP family is also conserved in vertebrates and may account for some of the store-operated conductances characterized in vertebrate cells. Consistent with this proposal, a related human protein, TRPC1, has been identified which is widely expressed in many tissues but at the highest levels in the brain, heart, testis, and ovaries (78, 79). TRPC1 is ~40% identical to TRP or TRPL over the ~625 amino acids spanning most of the amino terminus and the putative transmembrane segments. Furthermore, TRPC1 is distinct from the Drosophila proteins because it contains a relatively short segment carboxyl-terminal to the transmembrane domains (148 residues) compared with TRP and TRPL (614 and 464 amino acids, respectively; Fig. 2).

In view of the requirement for the extended carboxyl terminus of TRP for store-operated regulation (50), the relatively short carboxyl-terminal segment in TRPC1 suggested that it might not be a SOC. However, functional expression of TRPC1 in COS or Chinese hamster ovary cells results in a store-operated nonselective cation conductance (80, 81). In contrast to these studies, TRPC1 expressed in Sf9 cells produces a constitutively active cation conductance (82), a feature reminiscent of Drosophila TRPL. Thus, as is the case with TRPL, TRPC1 may require interaction with another protein to confer regulation by Ca²⁺ depletion. In Sf9 cells, TRPC1 may not be store operated due to the requirement for another TRPC protein that is absent in insect cells but present in mammalian cells. In support of this proposal, TRPC1 forms heteromultimers in vitro with TRPC3 (32), a brain-enriched member of the TRP family (80). TRPC3 expressed in COS cells seems to be a SOC (80); however, it has not been analyzed in other cell types, such as Sf9 cells.

The full extent of the vertebrate TRP family remains to be determined, although the minimum size is six because the partial or complete sequences of six types of TRPC genes have been identified in various animals; these include the full-length sequence for the mouse homologs of human TRPC1 (83) as well as portions of mouse homologs for human TRPC3 and TRPC2, the latter of which may be a pseudogene (80). The partial sequence of a fourth mouse TRPC gene, mtrp4, has been reported (80) and is a homolog of the full-length bovine clone, bCCE (84), and a TRP-related rat protein, TRP-R (85). These genes, which are collectively referred to here as TRPC4, are widely expressed in many tissues, including the brain and retina. Although rat and bovine TRPC4 are highly related, the conservation falls off considerable in certain portions of the carboxyl-terminal tail. Finally, segments of two other mouse sequences, Mtrp5 and Mtrp6, have also been reported (80). Evidence that some of these additional vertebrate TRPC proteins may be SOCs is that expression of antisense constructs inhibit the endogenous SOCE in mouse L cells (80). Furthermore, bovine TRPC4 has been introduced into human embryonic kidney cells and seems to be an inwardly rectifying SOC with somewhat higher permeability for Ca²⁺ than for Na⁺ (P_Ca/P_Na ~8:1) (84).

Overall, the regions most highly related among the TRPC family members are the ~300 residues amino-terminal to the transmembrane domains, which typically contain three or four ankyrin repeats, followed by the ~50 residues immediately carboxyl-terminal to the sixth transmembrane segment and then the transmembrane domains. The carboxyl-terminal regions of each TRPC protein are quite heterogeneous in sequence. It is noteworthy that rat and bovine TRPC4s have a longer carboxyl-terminal extension (~260-300 amino acids) than either of the other TRPC proteins for which a complete sequence is available (148 and 178 residues for TRPC1 and TRPC3, respectively). This raises the possibility that the TRPC proteins with longer carboxyl-terminal extensions, such as TRPC4, can form homomeric SOCs, whereas TRPC proteins with shorter extensions are constitutively active unless they heteromultimerize with the appropriate TRPC partner.

Remaining Questions and Perspective

A key question on the minds of many researchers studying I_CRAC is whether any of the TRPC channels are responsible for the highly Ca²⁺-selective, low-conductance, store-operated current typical of I_CRAC. Although it is clear that there are many types of store-operated cation conductances, none of the invertebrate or vertebrate TRP proteins induce a conductance with the properties of I_CRAC. One possibility is that I_CRAC is mediated by channels unrelated to TRP. However, an alternative possibility is that a diversity of currents are generated by heteromultimeric interactions between different TRPC proteins and that I_CRAC is generated through formation of specific sets of heteromultimers. Regardless of whether I_CRAC can be accounted for by any homomultimeric or heteromultimeric TRPC channel, it is of great interest to characterize the biophysical properties of the various conductances produced by these channels. A second important question concerns the mechanism by which depletion of Ca²⁺ from the internal stores activates members of the TRP family. Are there biologically relevant alternative modes for activating TRPC proteins independent of Ca²⁺ depletion from the internal stores? Furthermore, do the vertebrate TRPC proteins form supramolecular signaling complexes, and what is the nature of the proteins in the complex? Equally interesting are questions concerning the biological roles of the TRPC proteins. Are there human diseases, such as severe immunodeficiencies, that are due to perturbations in TRPC proteins? Although there has been considerable recent progress in characterizing the TRP family, many critically important questions clearly remain.