Drosophila TRP and TRPL are assembled as homomultimeric channels in vivo

The transient receptor potential (TRP) family of cation channels serves as sensors and transducers of environmental stimuli and also as regulators of ion homeostasis in neuronal and epithelial cells. The founding members of the TRP family are the Drosophila TRP (Hardie and Minke, 1992; Minke et al., 1975; Montell and Rubin, 1989) channel and its closest relative TRP-like (TRPL) (Phillips et al., 1992). To date more than 80 family members have been isolated from C. elegans, Drosophila, mice and humans (for reviews see Clapham, 2003; Hardie, 2007; Minke and Cook, 2002; Minke and Parnas, 2006; Montell et al., 2002; Montell, 2005), which have been grouped into seven subfamilies (TRPC, TRPV, TRPM, TRPA, TRPN, TRPP and TRPML) on the basis of amino acid sequence identity. By analogy to other channels with a similar transmembrane structure that have been more extensively studied [e.g. voltage-gated K⁺ channels and cyclic nucleotide-gated channels (Kaupp and Seifert, 2002; MacKinnon, 1991)], TRP channels are most likely formed by tetramers of the pore-forming subunits. Given the seven mammalian members of the TRPC subfamily and the total of 28 TRP channel isoforms of mammals, an important question arises as to the ability of TRP channels to form heteromultimers and what structural features are required for channel assembly. Indeed, heteromultimerization of TRP channels within single subfamilies has been described for vertebrate members of the TRPC, TRPM and TRPV subfamilies (for reviews see Cheng et al., 2010; Schaefer, 2005). However, even within the TRPC subfamily not all members co-assemble with each other. The current view, though still debatable, is that TRPC1,4,5 and TRPC3,6,7 are two district assembly groups that do not inter-assemble (Goel et al., 2002; Hofmann et al., 2002; Parnas et al., 2012; Schaefer, 2005).

In Drosophila photoreceptors two TRPC channels, TRP and TRPL are expressed together where they generate the light-induced current (LIC). In dark-raised flies, TRP and TRPL are localized to the highly packed microvilli, which form the signaling compartment of fly photoreceptor cells called the rhabdomere. In the rhabdomere the channels are activated in response to light, downstream of a G_q protein and phospholipase C (PLCβ) mediated cascade, generating the LIC (Hardie and Raghu, 2001; Huber, 2004; Minke and Parnas, 2006). Although in wild-type flies both TRP and TRPL are expressed together in each photoreceptor cell, they can form functional light-activated ion channels in photoreceptor cells of Drosophila mutants in isolation, clearly showing that each channel can function independently of the other channel (Niemeyer et al., 1996; Reuss et al., 1997). Electrophysiological studies of trp mutants (such as trp^P301, trp^CM and trp^P343), lacking the TRP channel, and of the trpl³⁰² null mutant, lacking the TRPL channel, revealed different biophysical properties of TRP and TRPL in vivo (Delgado and Bacigalupo, 2009; Hardie and Minke, 1994; Hardie and Mojet, 1995; Liu et al., 2007; Raghu et al., 2000; Reuss et al., 1997). Furthermore, the light response is completely abolished in the trpl³⁰²;trp^P343 double null mutant, indicating that both channels are necessary for the response to light (Niemeyer et al., 1996).

In addition to their different biophysical properties, TRP and TRPL also display differences in the dynamics of their subcellular localization (Bähner et al., 2002; Cronin et al., 2006; Meyer et al., 2006). In dark-raised flies both ion channels are present in the rhabdomere. However, upon continuous illumination, TRPL translocates from the rhabdomere to an intracellular storage compartment, while TRP remains in the rhabdomeral compartment. An additional difference between the TRP and TRPL channels is their ability to bind to the INAD scaffold protein. Some of the key elements of the phototransduction cascade are incorporated into supramolecular signaling complexes via the scaffold protein INAD (Shieh and Niemeyer, 1995), which binds the TRP channel but also its activator PLCβ and its regulator protein kinase C (Chevesich et al., 1997; Huber et al., 1996; Tsunoda et al., 1997). A specific interaction of INAD with TRP is required for rhabdomeric localization of the entire INAD signaling complex. When this interaction is disrupted the INAD and the entire scaffold proteins including the TRP channel moves from the rhabdomere to the cell body (Tsunoda et al., 1997). In contrast, TRPL appears to be separated from the INAD signaling complex (Tsunoda et al., 1997; but see Xu et al., 1998).

The first suggestion that TRP channels can assemble as heteromultimers came from studies on the Drosophila channels TRP and TRPL (Xu et al., 1997). This report provided electrophysiological and biochemical evidences for the formation of TRP–TRPL heteromultimers obtained from cell culture experiments, in vitro studies and co-immunoprecipitation (co-IP) from fly heads and cell culture. However, the functionality of heterologously expressed TRP channels was questioned (Minke and Parnas, 2006) and the existence of TRP–TRPL heteromultimers in vivo was challenged by showing that the electrophysiological properties of WT flies could be attained by a weighted sum of two independent TRP and TRPL components (Reuss et al., 1997). Moreover, the natively expressed TRP channel protein in the photoreceptor cells outnumbers the TRPL channel protein by approximately tenfold (Xu et al., 2000). This observation together with massive TRPL (but not TRP) translocation makes functionally significant formation of TRP–TRPL heteromultimers unlikely, casting doubt on the reported functional TRP–TRPL heteromultimers. Therefore, the issue whether or not TRP and TRPL form heteromultimers in fly photoreceptor cells is still unresolved.

In the present study we investigated whether TRP and TRPL are able to form heteromultimers, by in vivo assembly studies of tagged TRP and TRPL channels and of chimeric TRP and TRPL channels. Interactions of tagged and native channels as well as native and chimeric TRP–TRPL channels using co-IP, immunocytochemistry and functional electrophysiology critically tested the ability of TRP and TRPL to interact. We found that TRP and TRPL assemble exclusively as homomultimeric channels in their native environment. The above analyses revealed that the transmembrane regions of TRP and TRPL did not determine assemble specificity of these channels. However, the C-terminal regions of both TRP and TRPL, predominantly specify the assembly of homomultimeric TRP and TRPL channels.

In order to address the issue of TRP and TRPL channel subunit assembly in vivo, we generated Drosophila transgenes expressing TRP and TRPL channel subunits fused to enhanced green fluorescent protein (eGFP) at their C-termini (Fig. 1A) and expressed the tagged channels under the rhodopsin 1 (Rh1) promoter, driving the expression in R1–6 photoreceptor cells. The transgenes were introduced into a wild-type (WT) background, which expresses the TRP and TRPL subunits endogenously. The expression of eGFP-tagged TRP and TRPL could be readily observed by inspecting the flies under a fluorescence microscope with low magnification (Fig. 1A). In western blot analyses the eGFP-tagged TRP and TRPL channels were identified using TRP and TRPL antibodies generated against the C-terminal region. The western blots showed an apparent molecular mass of about 30 kDa higher than that of native TRP and TRPL proteins consistent with the molecular weight addition of the eGFP-tag (Fig. 1B, Input). This difference in migration on SDS-gels allowed distinguishing between native and tagged channels. Co-immunoprecipitation (co-IP) studies were carried out using antibodies against the eGFP-tag and the immunoprecipitates were blotted using TRP and TRPL antibodies. This analysis enabled determining which of the native subunits interact with the tagged subunit (Fig. 1B). These experiments revealed that TRPL–eGFP interacted with native TRPL, but not with the native TRP, while TRP–eGFP interacted with the native TRP but not with the native TRPL. The specificity of the tag was shown in control flies (WT), which did not express eGFP-tagged proteins, showing no signal on the western blot. The obtained results clearly show that TRP and TRPL channels assemble exclusively as homomultimers.

In order to exclude the possibility that the eGFP-tag hinders specifically heteromeric interactions between TRP and TRPL, we performed the co-IP studies with WT flies using TRP and TRPL antibodies raised against the C-termini (Fig. 1C). The results of these experiments showed that TRP and TRPL were specifically and exclusively immunoprecipitated with TRP and TRPL antibodies, respectively (Fig. 1C). Furthermore, in order to exclude the possibility that the TRP and TRPL antibodies raised against the C-termini hinder specifically heteromultimeric interactions, we used a TRPL antibody raised against the N-terminal region (α-TRPL-NT) of the channel. The results of these experiments showed that TRPL was specifically and exclusively immunoprecipitated with TRPL antibodies (Fig. 1C). Thus, in line with the anti–eGFP studies, these experiments reveal that TRP and TRPL channels assemble exclusively as homomultimers.

Additional evidence against the formation of TRP–TRPL heteromultimers came from the previously described translocation of TRPL. Accordingly, TRPL–eGFP showed translocation between the rhabdomere and the cell body in dark- and light-adapted flies, respectively (Meyer et al., 2006) while, TRP remained in the rhabdomere irrespective of the light rearing condition indicating no TRP–TRPL interactions (Fig. 1Da–h). Further evidence against the formation of TRP–TRPL heteromultimers came from analysis of transgenic flies expressing the TRP–eGFP channels. In flies expressing TRP–eGFP in the WT background, the GFP signal was detected in the rhabdomeres in both light- and dark-adapted flies and did not affect the light-triggered translocation of TRPL to the cell body (Fig. 1Di–p). The localization of TRP–eGFP was different in dark-adapted transgenic flies expressing TRP–eGFP in the trp^P343 background. In these flies the GFP signal was predominantly detected in the cell body, while the TRPL signal was detected in the rhabdomere (Fig. 1Dq–t). These results further support the notion that the TRP and TRPL channels do not interact.

A possible explanation for the observed localization of TRP–eGFP in the trp^P343 background is that in the absence of TRP, TRP–eGFP cannot associate with INAD promoting TRP–eGFP localization to the cell body (Tsunoda et al., 1997). To directly examine this possibility, we performed co-IP experiments; supplementary material Fig. S1 shows that in the WT background, in which native TRP and TRP–eGFP are co-expressed, INAD was precipitated by the anti-eGFP antibody. In contrast, in the absence of native TRP (TRP–eGFP in the trp^P343 null background) INAD was not precipitated, indicating that the eGFP-tag disrupted TRP-INAD interaction, while the TRP–TRP–eGFP heteromultimers can interact with INAD. Similar results were obtained when using anti-TRP antibody, thus demonstrating that the anti-eGFP antibody did not disrupt the TRP–INAD interaction.

The co-IP experiments, together with the observation that TRP and TRPL can be localized in separated non overlapping cellular compartments supports the notion that TRP and TRPL do not interact with each other and assemble only as homomultimers.

In order to examine the functional consequence of eGFP-tag attachment to the channels, we performed whole-cell patch-clamp recordings. Attachment of the eGFP tag had different effects on the TRP and TRPL channels. Attachment of the eGFP-tag to the TRPL channel (TRPL–eGFP) had virtually no effect on the light response amplitude as shown by the intensity–response curve (Fig. 2A–C) (Meyer et al., 2006) demonstrating that the eGFP tag had no effect on the function of the TRPL channel. However, for the TRP channel (TRP–eGFP) the situation was different depending on the genetic background. Accordingly, flies expressing TRP–eGFP in WT and trpl³⁰² backgrounds showed response amplitudes similar to WT and trpl³⁰². However, TRP–eGFP in the trp^P343 background showed reduced sensitivity to light compared to WT and trpl³⁰² mutant flies. This reduction in sensitivity to light is readily explained by the observed disruption of the TRP-INAD interaction required for retention of the entire INAD signaling complex to the rhabdomere (Fig. 1Di–l; Fig. 1Dq–t; Fig. 1Du–x; supplementary material Fig. S1) (Tsunoda et al., 1997). This complex includes the TRP channel and its activator, phospholipase C, which are critical components for the normal sensitivity to light. However, TRP–eGFP in the trp^P343 background revealed also a small reduction in sensitivity compared to trp^P343. This result cannot be explained solely by the disruption of TRP–INAD interaction as the same situation is also observed in the trp^P343 mutant fly (Tsunoda et al., 2001). Nevertheless, this result can be explained by at least two mechanisms. One possible mechanism is that non-functional TRP–eGFP induces a dominant-negative effect by the formation of TRPL–TRP–eGFP heteromultimers. In order to test this possibility, we examined whether TRP–eGFP is non-functional. To this end we measured the reversal potential (E_rev) of WT, trpl³⁰², trp^P343 and TRP–eGFP in the trp^P343 background. Importantly, the LIC of TRP–eGFP in the trp^P343 background revealed a biphasic E_rev typical of two functional channels with different E_rev similar to WT (supplementary material Fig. S2). Moreover, this E_rev was significantly more positive than the E_rev of trp^P343, also indicating that TRP–eGFP is functional (Reuss et al., 1997), thus making the above mechanism unlikely. A second possible mechanism is that the reduced sensitivity to light of the TRP–eGFP in the trp^P343 background arises from functional interaction of independent TRP–eGFP and TRPL channels (Reuss et al., 1997). Accordingly, when functional TRP and TRPL channels are activated together, the Ca²⁺ influx via the TRP channels suppresses the activity of the TRPL channels (Reuss et al., 1997) so that mainly the TRP current is maintained (e.g. in Fig. 2, notice the similarity of the LIC peak amplitude of WT, expressing both TRP and TRPL channels, and trpl³⁰², expressing only the TRP channel). However, in the trp^P343 background fly, the amount of functional TRP–eGFP in the rhabdomere is small (Fig. 1Dq; supplementary material Fig. S2), contributing more to the suppression of the TRPL channel than to the overall current. Together, elucidating the functional consequence of eGFP-tag attachment to the TRP channel indicated that the electrophysiological data are consistent with the co-IP and immunocytochemistry data.

To further examine TRP and TRPL subunit assembly and to identify regions in the channel proteins that determine the formation of specific multimers, we analyzed four different TRP–TRPL chimeric channels. We have previously generated chimeric Drosophila transgenes expressing constructs, in which either the C-terminus, the N-terminus, both regions, or the transmembrane domains of TRP were replaced by the corresponding regions of TRPL (Richter et al., 2011). They were referred to as chimera 1 (N^TRP-TM^TRP-C^TRPL-eGFP), chimera 2 (N^TRPL-TM^TRP-C^TRP-eGFP), chimera 3 (N^TRPL-TM^TRP-C^TRPL-eGFP) and chimera 4 (N^TRP-TM^TRPL-C^TRP-eGFP) (Richter et al., 2011). Expression of those chimeras in the WT background in Drosophila eyes was verified by in vivo fluorescent measurements (Fig. 3A) and by western blot analysis with antibodies against TRP or TRPL (Fig. 3B) (Richter et al., 2011). As the TRP and TRPL antibodies used here are directed against the C-terminal regions of the channels, anti-TRPL detected chimera 1 and 3, while anti-TRP detected chimera 2 and 4.

All the chimeric constructs formed functional light-activated ion channels in photoreceptor cells when expressed in the trpl³⁰²;trp^P343 double null mutant background (Fig. 4; Figs 6–f07,8) (Richter et al., 2011). However, the amplitude of the LIC was markedly reduced in chimeras 1–3 compared with trpl³⁰² mutant fly, while for chimera 4 small reduction in response amplitude was observed compared to the trp^P343 mutant fly, when the chimeric channels were expressed in the trpl³⁰²;trp^P343 double null mutant background.

We first performed co-IP experiments on chimera 3 in the WT background, applying antibodies against the eGFP-tag and the immunoprecipitates were blotted using TRP and TRPL antibodies (Fig. 3B, lane 4). Since the tagged subunit is distinguishable from the endogenous subunit on western blots by its higher molecular weight, this assay allowed identification of the in vivo interaction partners of chimera 3. Fig. 3B shows that chimera 3 interacted with native TRPL, but not with the native TRP channel (compare Fig. 3B upper and lower panels, lane 4). This result indicates that chimera-3–TRPL heteromultimers are formed, while chimera-3–TRP heteromultimers do not form.

To further test the above conclusion, we examined the cellular localization of chimera 3, TRP and TRPL channels using immunocytochemistry in dark- and light-raised flies. The immunocytochemical studies were used to examine whether or not the two channel subunits found to interact in co-IP experiments also reside in the same cellular region. Fig. 4A shows that both native TRPL and chimera 3 translocated normally from the rhabdomere to the cell body upon illumination as previously reported (Fig. 4Aa–h) (Richter et al., 2011). This finding is compatible with an interaction between the native TRPL and chimera 3. However, it should be noted that the TRPL antibody detected both chimera 3 and native TRPL. Hence, while translocation of chimera 3 was clearly revealed by observing the eGFP fluorescence (Fig. 4Aa,e), translocation of native TRPL from the rhabdomere to the cell body was revealed only indirectly by the absence of labeling of the rhabdomeres in the light (Fig. 4Ab,f). The situation was different when analyzing chimera 3 and TRP localization. In dark-raised flies colocalization was observed in the rhabdomere for both TRP and chimera 3 (Fig. 4Ai–p). However, in light-raised flies, chimera 3 translocated to the cell body and did not colocalize with the native TRP channel, which remained localized to the rhabdomere (Fig. 4Ai–p). These results are consistent with the co-IP experiments.

To determine the physiological effects of the expression of chimera 3 in photoreceptor cells we carried out whole-cell recordings from isolated ommatidia. Chimera 3 expressed in the trpl³⁰²; trp^P343 double null background revealed highly reduced sensitivity to light when compared to both trpl³⁰² and trp^P343 mutant flies (Fig. 4B–D). Strikingly, chimera 3 expressed in the trpl³⁰² background showed sensitivity to light similar to WT and trpl³⁰² indicating that chimera 3 does not affect the LIC through TRP channels. However, chimera 3 in the trp^P343 background (expressing only the native TRPL channel) showed highly reduced sensitivity to light when compared to WT, trpl³⁰² or trp^P343 mutant flies. Because the current and hence Ca²⁺ influx produced by light activation of chimera 3 in isolation was very small the above finding is best explained by a dominant-negative effect of chimera 3 on the native TRPL channel with which it forms multimers (Fig. 4B–D).

These results also show no chimera-3–TRP interactions as no dominant-negative effect was observed in chimera 3 in the trpl³⁰² background.

The macroscopic response to light of the fly is composed of single photon responses (quantum bumps), which sum to produce the LIC (for review see Katz and Minke, 2009). Reduced sensitivity to light can arise from two different phenomena: reduction in mean bump amplitude or from reduction in bump frequency. Therefore, dominant-negative effect at the quantum bump level is manifested by a reduction in bump amplitude or rate of occurrence. To further examine the dominant-negative effect of chimera 3 on the native TRPL channel, we measured responses to dim light, which elicit quantum bumps (Fig. 5). The figure shows that quantum bump formation is highly attenuated in chimera 3 in the trp^P343 background, while no dominant-negative effect was observed in chimera 3 in the trpl³⁰² background (Fig. 5A–D). These results are compatible with the observation of the macroscopic response to light.

Together, the analysis of chimera 3 demonstrated that chimera 3 with N- and C-termini of TRPL exclusively formed multimers with TRPL and had a dominant-negative effect on the TRPL- but not on the TRP-mediated current. We conclude that the requirement for TRPL multimeric formation is specified by either the N- or C-terminal region or both regions but is not specified by the TRPL transmembrane domains.

As with chimera 3 we first used co-IP studies with antibodies against the eGFP-tag of chimera 4 (Fig. 3B, lane 5) and asked which of the native TRP and/or TRPL subunits interact with the tagged subunit. Fig. 3B shows co-IP experiments using flies expressing chimera 4 in the WT background. These experiments revealed that chimera 4 interacted with native TRP, but not with the native TRPL channel (compare Fig. 3B upper and lower panels, lane 5).

Immunocytochemical studies on chimera 4 expressed in the WT background showed that this chimera was localized to both rhabdomere and cell body, without considerable changes in its distribution between light and dark rearing condition (Fig. 6Aa,e,i,m). The expression of chimera 4 in WT photoreceptors did not affect the localization of native TRPL, which displayed its typical translocation between the rhabdomere and cell body upon illumination (Fig. 6Ab,f). Labeling with TRP antibodies was indistinguishable from eGFP-fluorescence detection (Fig. 6Ai–p). However, it should be noted that the TRP antibody detected both chimera 4 and native TRP. Nevertheless, the anti-TRP signal showed relatively low intensity in the R1–6 rhabdomeres compared with the R7 rhabdomere signal, indicating a reduction in the TRP expressed in the rhabdomere. Hence, these findings are in line with the co-IP results showing interaction of chimera 4 with TRP but not with TRPL.

Single cell recordings revealed that chimera 4 in the trpl³⁰²; trp^P343 double null background showed the largest peak amplitude and sensitivity to light relative to all the other 3 chimeras, reaching ∼4 nA peak amplitude of the LIC in response to a light intensity of 1.3×10⁵ effective photons per second, similar to that of the trp^P343 mutant fly (Fig. 6B–D). This can be explained by the fact that chimera 4 has the transmembrane regions of TRPL, forming the large conduction pore of TRPL (i.e. at least approximately fivefold larger single channel conductance than that of the TRP channel; Raghu et al., 2000) and it is also present at the rhabdomeres (Fig. 6Aa,e,i,m), while chimeras 1–3 harbor the TRP transmembrane domains forming the TRP pore with a small estimated single channel conductance (Richter et al., 2011). Similarly, the quantum bumps of chimera 4 in the trpl³⁰²; trp^P343 background revealed quantum bumps of smaller amplitude similar to the quantum bumps of the trp^P343 mutant, indicating small reduction in the sensitivity to light of chimera 4 in isolation compared with trp^P343 mutant flies (Fig. 5D–E). Analysis of flies expressing chimera 4 in the trpl³⁰² background showed that the peak amplitude of the LIC in photoreceptors expressing chimera 4 in the trpl³⁰² background was similar to WT and trpl³⁰² mutant fly (Fig. 6B–D). Thus, although chimera 4 and TRP form heteromultimers as revealed by co-IP and immunocytochemistry studies, chimera 4 had no dominant-negative effect on the TRP-mediated current. This may be explained by the minor reduction in sensitivity to light of chimera 4 in the trpl³⁰²; trp^P343 background together with the still observed localization of both chimera 4 and TRP in the rhabdomere, as revealed by immunocytochemistry.

A puzzling observation is that chimera 4 in the trp^P343 background (expressing native TRPL) showed reduced sensitivity to light compared with trp^P343 flies (Fig. 6B–D). This phenomenon was also observed and explained in detail for TRP–eGFP in the trp^P343 background (Fig. 2D–F). However, in this case Ca²⁺ influx was mediated by the large TRPL single channel conductance and the relative high chimera 4 expression in the rhabdomere (see above).

Together, the analysis of chimera 4 by two independent methods, namely co-IP and immunocytochemistry, demonstrated that chimera 4 channels are unable to interact with the native TRPL channel, while they are able to interact with the native TRP channel regardless of their transmembrane domain. Electrophysiology could not be used to demonstrate chimera-4–TRP interactions, because chimera 4 did not show severely reduced sensitivity in isolation.

The results obtained from chimera 3 and 4 expressing flies revealed that chimera 3 with its N- and C-terminal regions of TRPL is able to interact only with TRPL, while chimera 4 with its N- and C-terminal regions of TRP is able to interact only with TRP. However, these chimeras could not resolve whether the N- or C-termini (or both) specify channel assembly. To answer this question we studied flies expressing chimera 1 and chimera 2.

Co-IP experiments using flies expressing chimera 1 (N^TRP-TM^TRP-C^TRPL-eGFP; Fig. 3A) in the WT background revealed that it interacted with native TRPL, but not with native TRP (Fig. 3B, lane 2). A previous study on chimera 1 revealed that this chimera does not translocate upon illumination and resides predominantly in the cell body (Fig. 7Aa,e,i,m) (Richter et al., 2011). A small fraction of chimera 1 resides in the rhabdomere as evidenced by the very small LIC when expressed in the trpl³⁰²; trp^P343 double null background (Fig. 7B–D). To further examine the ability of this chimera, expressed in the WT background, to interact with the native TRPL and TRP, we performed immunocytochemical experiments. Chimera 1 affected the localization of native TRPL that was no longer detected in rhabdomeres of photoreceptor cells R1–6 in dark reared flies. Instead, TRPL antibodies (detecting both, chimera 1 and native TRPL) labeled only the cell body and the rhabdomere of R7 cells in dark-reared flies (Fig. 7Ab,f). Labeling of the R7 cell rhabdomere was expected as chimera 1 was not expressed in this cell and served as a positive control. Fig. 7Ai–p further shows that chimera 1 did not colocalize with the TRP channels, which are mainly located to the rhabdomere. These data revealed that chimera 1 hindered TRPL translocation to the rhabdomere, presumably due to formation of heteromultimers between chimera 1 and TRPL. These results are consistent with the co-IP experiments and together show that while chimera 1 interacts with the native TRPL it does not interact with the native TRP channel.

We next applied whole-cell patch-clamp recordings from isolated Drosophila ommatidia. Chimera 1 expressed in the trpl³⁰²;trp^P343 double null background revealed highly reduced sensitivity to light when compared to both trpl³⁰² and trp^P343 mutant flies (Fig. 7B–D). Chimera 1 in the trpl³⁰² and WT backgrounds showed similar sensitivity to light compared to both WT and trpl³⁰² indicating that no interaction exists between chimera 1 and TRP channel subunits (Fig. 7B–D). Analysis of the microscopic response to dim light stimulation confirmed the results of the macroscopic response by showing that the mean amplitude quantum bump and bump frequency of chimera 1 in the trpl³⁰² background was similar to that of the trpl³⁰² mutant fly (Fig. 5A–C). Strikingly, chimera 1 in the trp^P343 background showed a highly reduced sensitivity to light compared to WT, trpl³⁰² and trp^P343 mutant flies indicating a strong dominant-negative effect of chimera 1 on TRPL (Fig. 7B–D). The observation of the dominant-negative effect on the macroscopic response to light was also obtained in chimera 1 in the trp^P343 background, at the quantum bump level, by showing virtually no bump production (Fig. 5D). Thus the electrophysiological data strongly support the conclusions obtained from the co-IP and immunocytochemical experiments that chimera 1 interacts with the native TRPL but not with the native TRP channel.

The co-IP studies of flies expressing chimera 2 (N^TRPL-TM^TRP-C^TRP-eGFP; Fig. 3A) in the WT background revealed that it interacted with both native TRPL and native TRP (Fig. 3B, lane 3). As is the case for chimera 1, chimera 2 was located almost exclusively in the cell body irrespective of the light rearing condition (Fig. 8Aa,e,i,m) (Richter et al., 2011). To further examine the ability of chimera 2 in the WT background to interact with the native TRPL and TRP channels we performed immunocytochemical experiments. As revealed by staining with TRP antibodies (detecting both chimera 2 and native TRP), expression of chimera 2 resulted in mislocalization of TRP, which was no longer observed in the rhabdomeres of R1–6 cells (Fig. 8Aj,n). A superficial observation of Fig. 8 shows that chimera 2 had no effect on the localization of TRPL, which was correctly located in the rhabdomeres or in the cell body of dark- or light-reared flies, respectively (Fig. 8Ab,f). However, a close examination of the merged images showed colocalization of chimera 2 and TRPL, indicating possible interaction. Further examination of immunocytochemical analysis of chimera 2 in the trp^P343 background, in which the TRP and TRPL channel subunits do not compete for chimera 2 interaction, showed enhanced colocalization between TRPL and chimera 2 (supplementary material Fig. S3).

In whole-cell patch-clamp recordings, chimera 2 in the trpl³⁰²; trp^P343 background showed strongly reduced sensitivity to light compared with WT, trpl³⁰² and trp^P343 mutant flies (Fig. 8B–D). Chimera 2 in the trp^P343 background as well as in the trpl³⁰² background showed reduced sensitivity to light compared with WT, trpl³⁰² and trp^P343 mutant flies indicative of a dominant-negative effect of chimera 2 on both, TRP- and TRPL-mediated currents. Chimera 2 in the WT background showed partial rescue of the dominant-negative effect of chimera 2 observed in the trpl³⁰² background. This can be readily explained by competing binding of chimera 2 to both TRP and TRPL channels, resulting (in the case of the WT background) in additional homomultimeric TRP channels (Fig. 5A–C). The observation of the dominant-negative effect on the macroscopic response to light was also obtained at the quantum bump level showing reduced mean bump amplitude of chimera 2 in the trpl³⁰² background with little change in bump frequency (Fig. 5A–C), while chimera 2 in the trp^P343 background showed virtually no bumps (Fig. 5D).

Together, the analysis of chimera 1 and 2 by three independent methods, demonstrated that the C-terminal region of TRPL determines assembly specificity with the native TRPL channel, while the C-terminal region of TRP determines assembly specificity with TRP. The N-terminal region of TRPL may also contribute to TRPL assembly specificity, as evidenced by the interaction of chimera 2 with TRPL in co-IP and electrophysiological studies. In contrast, we did not find evidence indicating that the N-terminal region of TRP can specify the formation of TRP multimers (but see Discussion).

In the present research we studied molecular aspects, which determine the specificity of the Drosophila TRPC channels assembly. Many of the disparate results regarding TRPC function and regulation were reconciled by showing that several TRPC channel subunits are assembled into heteromultimeric channels with diverse properties. These studies have also shown that heteromultimeric assembly of various members of the TRPC channel members are characterized by heteromultimeric assembly into two distinct groups: the TRPC1,4,5 and TRPC3,6,7 groups (Goel et al., 2002; Hofmann et al., 2002; Parnas et al., 2012; Schaefer, 2005). Clapham and colleagues have reported that TRPC1 co-assembled with TRPC4 and TRPC5 in rat brain (Strübing et al., 2001). The biophysical properties of TRPC(1,4) and TRPC(1,5) heteromultimers are distinct from those of the channel homomultimers. Considering that many TRPCs are ubiquitously expressed and often co-expressed in a given cell (Abramowitz and Birnbaumer, 2009), heteromultimerization in addition to homomultimerization among members of this protein family represents an efficient way to increase the functional diversity of TRPC channels. A structural understanding of how TRPC subunits combine to form functional ion channel complexes is an essential prerequisite to evaluate the contribution of a given TRPC channel to endogenous PLC-dependent cation currents.

Since the two Drosophila TRPC channels TRP and TRPL are expressed in each one of the photoreceptor cells of the Drosophila compound eye, the previous report on the existence of both homomultimeric and heteromultimeric assembly of TRP and TRPL (Xu et al., 1997) was in agreements with later studies on TRPC channels. Unlike the study by Xu et al. the present study shows that TRP and TRPL channels do not form heteromultimers (Table 1) (Xu et al., 1997). Our results are consistent with previous studies showing that the WT light-activated conductance is composed of two independent TRP and TRPL components (Reuss et al., 1997), that only TRPL translocates from the rhabdomere to the cell body upon illumination (Bähner et al., 2002), and that only TRP interacts with INAD (Tsunoda et al., 1997) (see also Xu et al., 1998).

Co-IP studies with native TRP and TRPL subunits extracted from Drosophila eyes suffer from the drawback that per se only heteromultimers but not homomultimers can be detected. It is therefore difficult to adjust the experimental conditions in a way that excludes precipitation artifacts. We have chosen here to use eGFP-tagged subunits allowing detection of multimers composed of the tagged and the native subunit, which can be distinguished on western blots by their different molecular mass. These co-IP experiments demonstrated that the TRPL and TRP channels do not interact with each other (Fig. 1B,C; Table 1).

Previous studies have demonstrated that coiled-coil (cc) domains and ankyrin repeats participate in TRP channel assembly (Engelke et al., 2002; Lepage et al., 2006; Lepage et al., 2009; Xu et al., 2000). In order to better understand previous (Xu et al., 1997) and current results and to explain the possibilities of association between TRP and TRPL, we analyzed the ankyrin repeats and cc-domains of TRP and TRPL channels (supplementary material Fig. S4). Two ankyrin repeats (aa 78–107 and 152–181 of TRPL; supplementary material Fig. S4) and a cc-domain (aa 233–266 of TRPL; supplementary material Fig. S4) at the N- termini of both the TRP and TRPL channel were identified with high confidence. Sequence alignments of these domains showed sequence homology of the first ankyrin repeat (63% similarity) and very high sequence homology (93% similarity) of the second ankyrin repeat (supplementary material Fig. S4). The alignment of the N-terminal cc-domain showed high sequence homology both globally (62% similarity) and in its a–d pattern which is a repetitive pattern in cc-domains responsible for multimerization [80% similarity (Fujiwara and Minor, 2008)] (supplementary material Fig. S4). The situation for the C-termini is different. Accordingly, depending on the prediction program used, no or one cc-domain was predicted for the C-terminal region of the TRP channel (aa 760–781), while one to three cc-domains were predicted for the TRPL channel (aa 722–747; 768–789; 899–922). Sequence alignment of TRP and TRPL in these regions showed low sequence homology in the a–d pattern of the first, and high sequence homology in the second cc-domain, while the third predicted cc-domain of TRPL has no TRP counterpart and the a–d pattern is not conserved (supplementary material Fig. S4). The analysis indicates high homology between the N-termini cc-domain and ankyrin repeats of TRP and TRPL, while low homology between the C-termini cc-domain of these channels.

A model in which neighboring subunits of specific channel interact through a C-C-terminal assembly may explain our results. This model also explains why TRP and TRPL do not interact, as the C-terminal pattern of their cc-domains differs. However, this model is inconsistent with previous data showing that the N-terminal region determines subunit assembly, while the C-terminal region does not participate in subunit assembly (Engelke et al., 2002; Lepage et al., 2006; Lepage et al., 2009; Liu et al., 2005; Xu et al., 2000; Xu et al., 1997). In order to reconcile this apparent contradiction, we partially adapted a previously proposed model in which the N-terminal interaction assembles the channels. We modified this model by including participation of the C-terminal in subunit interactions via repulsion. Accordingly, the N-terminal fragment of TRP and TRPL can homo- and heterooligomerize. However, the TRP and TRPL do not heteromultimerize because of the difference in their C-termini, resulting in repulsion and dissociation of the subunits. This model fits well with previous in vitro results from Drosophila, as well as mammalian TRPC channels (Engelke et al., 2002; Lepage et al., 2006; Lepage et al., 2009; Liu et al., 2005; Xu et al., 2000; Xu et al., 1997), showing that N-terminal but not the C-terminal fragments of these channel subunits interact.

Together, the data of this study indicate that the C-termini of TRP and TRPL have a critical role in subunit assembly and that TRP and TRPL do not form heteromultimers. The proposed model enables a new framework to evaluate the participation of the C-termini in determination of the assembly partners of the TRPC channels.

The following strains of Drosophila melanogaster were used: w¹¹¹⁸ (here referred to as wild type), yw;trpl³⁰² (Niemeyer et al., 1996), yw;trpl³⁰²;trp^P343 (Yang et al., 1998), yw;;trp^P343 (Pak, 1979), yw;;P[Rh1 >TRPL-eGFP,y⁺] (Meyer et al., 2006), yw,P[Rh1 >TRP-eGFP,y⁺];; and yw;P[Rh1 >Chimera 1-4-eGFP,y⁺] (Richter et al., 2011). Transgenic flies were crossed with trp^P343, trpl³⁰² or a trpl³⁰²; trp^P343 double mutant to obtain the genotypes indicated in the figure legends using standard Drosophila genetics. Flies were raised at 25°C on standard corn meal food. For all biochemical and immunocytochemical experiments adult flies were used at an age of 1–2 days after eclosion. For whole cell experiments newly eclosed flies were used. At least 12 hours before eclosion the vials containing the flies were wrapped in aluminum foil and transferred into a light-sealed box (dark adapted).

For generating the DNA construct used to express a TRP–eGFP fusion protein, the stop codon and the 3′ untranslated region of a trp cDNA clone were removed by substituting the sequence 3′ of a SacI restriction site with a PCR fragment containing SacI and ApaI cloning sites. The modified trp cDNA was subcloned after partial digestion with EcoRI and ApaI into a p-Bluescript vector between a Drosophila Rh1-promoter fragment [base pairs −833 to +67 (Mismer and Rubin, 1987)] and the coding sequence for eGFP (obtained from the vector pEGFP-1, BD Biosciences, Germany). Rh1-promoter, trp and eGFP coding sequences were then cloned into the XhoI restriction site of the P-element transformation vector YC4 (Patton et al., 1992). Chimeric constructs were generated as described (Richter et al., 2011) and have the following structure: Chimera 1, amino acids (aa) 1–675 of TRP + aa 681–1124 of TRPL + eGFP; chimera 2, aa 1–336 of TRPL + aa 328–1275 of TRP + eGFP; chimera 3, aa 1–336 of TRPL + aa 328–675 of TRP + aa 681–1124 of TRPL + eGFP; chimera 4, aa 1–328 of TRP + aa 336–675 of TRPL + aa 675–1275 of TRP + eGFP.

100 to 300 fly heads were obtained by mass isolation from flies frozen in liquid nitrogen using sieves as described (Voolstra et al., 2010). Fly heads were homogenized in buffer A [Triton X-100 buffer: 1% (v/v) Triton X-100, 150 mM NaCl, 4 mM phenylmethylsulfonyl fluoride in 50 mM Tris-HCl, pH 8.0; 2 µl/head] using a micro pestle (Roth, Germany). Thereafter, the homogenates were subjected to sonification for 5 minutes and extracted on ice for 30 minutes. Head extracts were incubated with 6 µg anti-GFP antibodies (Roche, Germany), anti-TRP antibodies (Mab83F6-c; Developmental Studies Hybridoma Bank, University of Iowa) or anti-TRPL antibodies directed against the C-terminal (Meyer et al., 2008) or N-terminal region (amino acids 5–19) coupled to 35 µl of protein-G–agarose beads (Roche, Germany) or protein-A–agarose beads (Thermo Fisher Scientific, Germany) overnight at 4°C. The wash and elution steps were performed as previously described (Voolstra et al., 2010) except that 0.1% instead of 1% Triton X-100 was used in the wash buffer. The eluate was subjected to western blot analysis as described (Meyer et al., 2006). The antibodies used were α-TRP antibody (Mab83F6-c; Developmental Studies Hybridoma Bank, University of Iowa), α-TRPL antibody (Richter et al., 2011) or anti-INAD antibody (Bähner et al., 2002).

Immunocytochemistry was carried out as described before (Chorna-Ornan et al., 2005; Meyer et al., 2008). The eGFP-tagged ion channels were visualized by their GFP fluorescence while AF546-coupled phalloidin (Invitrogen, Germany) was used for labeling of the rhabdomeres. For labeling of endogenous subunits anti-TRPL (Meyer et al., 2008) or anti-TRP antibodies (Mab83F6-c; Developmental Studies Hybridoma Bank, University of Iowa) were used. Cyosections were observed with the AxioImager.Z1m microscope using an ApoTome module (Zeiss, Germany; objective: EC Plan-Neofluar 40×/1.3 NA oil immersion). All images were captured with a AxioCam MrM (Zeiss) camera and AxioVision 4.6./4.8. (Zeiss) software.

For whole-cell patch-clamp measurements, a xenon high-pressure lamp (PTI, LPS 220, operating at 75 W) was used, and the light stimuli were delivered to the ommatidia by means of epi-illumination via an objective lens (in situ). Absolute calibration of the effective number of photons in the stimuli was achieved by counting quantum bumps in dark-adapted WT photoreceptors under control conditions with dim light.

Dissociated Drosophila ommatidia were prepared from newly eclosed dark-adapted adult flies (<1 hour posteclosion) and transferred to a recording chamber on an inverted Olympus microscope. Whole-cell voltage-clamp recordings and bump detection were performed as described previously (Katz and Minke, 2012). The bath solution contained (in mM): 120 NaCl, 5 KCl, 4 MgSO₄, 1.5 CaCl₂, 10 N-Tris-(hydroxymethyl)-methyl-2-amino-ethanesulphonic acid (TES), 25 L-proline, 5 L-alanine. The recording pipette solution contained (in mM) 140 potassium gluconate, 2 MgSO₄, 10 TES, 4 MgATP, 0.4 Na₂GTP, and 1 nicotinamide adenine dinucleotide (NAD). For reversal potential measurements, the pipette solution contained (in mM) 140 CsCl, 15 tetraethylammonium chloride, 2 MgSO₄, 10 TES buffer, 4 MgATP, 0.4 NaGTP, and 1 NAD. All solutions were adjusted to pH 7.15.

We thank the Developmental Studies Hybridoma Bank, University of Iowa for providing the anti-TRP antibody and Andreas Heinhold for help in generating the TRP–eGFP fly.