Subcellular translocation of the eGFP-tagged TRPL channel in Drosophila photoreceptors requires activation of the phototransduction cascade

TRP channels constitute a large and diverse protein family of cation channels, which is conserved through evolution (for reviews, see Minke and Cook, 2002; Clapham, 2003; Clapham et al., 2003; Montell, 2003; Moran et al., 2004; Montell, 2005). TRP channels are employed in a number of sensory systems, including invertebrate vision, mechanoreception, taste perception, nociception, thermoreception, pheromone detection and in non-neuronal cells, where they function in vasorelaxation or in cell cycle control. Drosophila TRP, which is the founder of this ion channel family, and its homologue TRP-like (TRPL) are required for visual transduction. The physiological site of action of TRP and TRPL is the rhabdomeral photoreceptor membrane formed by a densely packed stack of microvilli along the side of the photoreceptor cells in the fly compound eye. The cation channels are activated in response to light absorption by the visual pigment rhodopsin through a Gq protein-mediated signaling pathway (Devary et al., 1987; Bloomquist et al., 1988; Ranganathan et al., 1995; Montell, 1999; Hardie and Raghu, 2001). Although the exact gating mechanism of TRP and TRPL is not yet known, phospholipase Cβ that hydrolyzes phosphatidylinositol 4,5-bisphosphate to form the second messengers diacylglycerol and 1,4,5-inositol trisphosphate is mandatory for the activation of the ion channels. Both second messengers have been implicated in TRP and TRPL activation [see Hardie (Hardie, 2003) and Minke and Agam (Minke and Agam, 2003) for a discussion of possible gating mechanisms].

We reported previously that TRPL undergoes a light-regulated subcellular translocation. High level of rhabdomeral TRPL characteristic of dark-raised flies was reduced to a low level upon continuous illumination, whereas the content of rhabdomeral TRP is not altered by exposing Drosophila to different light conditions (Bähner et al., 2002). The change of the TRP/TRPL ratio has physiological consequences. Flies with high TRPL level in the rhabdomere respond to a wider range of light intensities than flies with a reduced TRPL content, and they are more sensitive to adaptation by dim background lights (Bähner et al., 2002). Translocation of the Drosophila TRPL channel thus constitutes a powerful in vivo model system for studying the still unclear mechanisms underlying translocation of mammalian TRPC and TRPV channels that have been recently reported (Kanzaki et al., 1999; Bezzerides et al., 2004).

To understand further the cellular mechanism underlying TRPL translocation we generated transgenic Drosophila, which express an eGFP-tagged TRPL channel in photoreceptors R1-6. We found that TRPL-eGFP forms functional ion channels with native properties. The analysis of TRPL-eGFP translocation in various mutants with defects in phototransduction proteins shows that activation of rhodopsin and of downstream signaling proteins of the phototransduction cascade is mandatory for inducing TRPL internalization.

For the expression of an eGFP-tagged TRPL ion channel in photoreceptor cells R1-6 of the compound eye, a DNA construct containing the promoter of Drosophila rhodopsin 1 (Rh1) and the fused coding sequences for TRPL and the enhanced green fluorescent protein (eGFP) was cloned into a P-element vector and used to generate transgenic Drosophila. Western blot analysis with anti-TRPL antibodies revealed that flies that contain this transgene in a wild-type background express native TRPL (molecular mass 128 kDa) and TRPL-eGFP (Fig. 1A). Two protein bands with an apparent molecular mass of ∼145 kDa and 135 kDa, representing TRPL-eGFP, were detected in flies carrying the TRPL-eGFP transgene (Fig. 1A, left panel). The protein bands corresponding to TRPL-eGFP were also recognized by an anti-GFP antibody (Fig. 1A, middle panel). The second, lower molecular mass band is likely to represent a degradation product of the full-length protein. The amount of the second cation channel subunit, TRP, present in the photoreceptor cells is not significantly affected by the expression of TRPL-eGFP (Fig. 1A, right panel).

When flies expressing the TRPL-eGFP reporter gene were kept in the dark for 16 hours and were then inspected at low magnification under a fluorescence microscope, a very bright green fluorescing deep pseudopupil (dpp) was observed (Fig. 1B). The dpp is a virtual image consisting of seven spots that reflect the trapezoidal pattern of the rhabdomeres in a single ommatidium. It is generated by the superimposed images of several rhabdomeres of photoreceptor cells from different ommatidia which have the same optical axis (Franceschini and Kirschfeld, 1971b). Since the Rh1 promoter drives TRPL-eGFP expression only in R1-6 cells, a central spot corresponding to the rhabdomere of R7 and R8 photoreceptor cells was missing in the fluorescing dpp of the transgenic flies and served as an internal control. The eGFP-fluorescence of the dpp in flies kept in the dark suggests that TRPL-eGFP is located in the rhabdomeres under this condition. A different picture was obtained when the flies were kept in the light. In these flies no fluorescing dpp was observed, suggesting that TRPL-eGFP was not present in the rhabdomeres of the photoreceptor cells (Fig. 1B). These findings are fully compatible with the notion that TRPL-eGFP undergoes a light-dependent translocation between the rhabdomere and an intracellular compartment.

A high resolution image of the cellular TRPL-eGFP localization can be obtained when the cornea of the eye is optically neutralized by water immersion as has been described by Franceschini et al. (Franceschini and Kirschfeld, 1971a; Franceschini et al., 1981). This method allows visualization of single rhabdomeres and cell bodies of single ommatidia in a living fly. We used a water immersion objective to study the distribution of TRPL-eGFP in transgenic flies which were kept in the dark or in the light for 16 hours (Fig. 1C). In flies kept in the dark, eGFP-fluorescence was restricted to the rhabdomeres of photoreceptor cells R1-6, whereas flies kept in the light showed fluorescence in the cell bodies, but not in the rhabdomeres (which appear dark in Fig. 1C right). This light-dependent redistribution of TRPL-eGFP was further confirmed when compared to the redistribution of native TRPL by analyzing eGFP fluorescence and immunofluorescence of antibodies against TRPL in cross sections through Drosophila eyes (supplementary material Fig. S1). A similar distribution in the rhabdomeres or in the cell bodies was observed for native TRPL and TRPL-eGFP in dark- or light-raised flies, respectively.

In order to determine the kinetics of TRPL-eGFP relocation, we quantified the intensity of eGFP fluorescence inside and outside the rhabdomere as a function of time, using fluorescent images obtained from intact flies with the water immersion technique. The obtained time courses (Fig. 2A,B) revealed that the internalization of TRPL-eGFP in the light and its movement into the rhabdomeres in the dark occurred within several hours with half-times of 3.25 hours and 1 hour, respectively. The light-dependent redistribution of Calliphora TRPL reported previously was significantly shorter (Bähner et al., 2002). This quantitative difference may have resulted from species differences. Qualitative evaluation of the time-dependent translocation of native Drosophila TRPL by immunocytochemistry suggests a similar time course for the translocation of native TRPL and TRPL-eGFP in Drosophila (supplementary material Fig. S2).

In another stimulation paradigm we subjected either dark- or light-adapted flies to alternating light and dark intervals (30 minutes each). Under these conditions an equilibrium was established in which about 75% of TRPL-eGFP was located in the rhabdomeres whether the flies were light- or dark-adapted before the experiment (Fig. 2C). This is in line with our findings showing that the internalization of TRPL-eGFP is slower than its translocation to the rhabdomeres. Finally, we investigated TRPL-eGFP internalization in different light intensities (supplementary material Fig. S3). Even under the lowest light intensity tested (2 Lux) complete internalization of TRPL-eGFP was observed in most photoreceptor cells. However, under this condition few photoreceptor cells were observed in which TRPL-eGFP was located completely in the rhabdomeres.

In order to determine whether or not the eGFP-tag of TRPL interferes with the formation of functional ion channels, we generated flies that expressed TRPL-eGFP in a trp^P343; trpl³⁰² double mutant background. The trp^P343; trpl³⁰² double mutant lacks functional photoreceptor channels and does not respond to light at all (Niemeyer et al., 1996; Scott et al., 1997). Thus, the presence of light responses in flies that express TRPL-eGFP in this genetic background would indicate the formation of functional TRPL-eGFP channels. We examined the properties of TRPL-eGFP channels electrophysiologically at two levels: (i) in vivo, using the electroretinogram (ERG) and (ii) in situ, using patch-clamp whole cell recordings from isolated ommatidia. ERG-recordings from TRPL-eGFP-expressing mutants revealed responses to medium intensity lights, which declined to that of the dark baseline during illumination, within 5 seconds. The response to light recovered in the dark as revealed by a significant transient response to a second light stimulus applied 10 seconds after the first light. (Fig. 3A, middle trace). The transient light-responses from TRPL-eGFP-expressing flies are reminiscent of light-responses obtained from trp mutants that express only native TRPL channels (Fig. 3A bottom trace). The light responses of both TRPL-eGFP-expressing flies and trp mutants are very different from that of wild-type (WT) flies, which have a sustained response during illumination and the responses to the first and second light pulses are virtually identical (Fig. 3A, upper trace). To further compare the physiological properties of TRPL-eGFP-expressing flies and trp mutants, intensity-response relationships (V-log I curves) were plotted from the peak amplitude of ERG recordings in response to increased intensities of orange light. V-log I curves of mutants expressing either native TRPL or TRPL-eGFP were almost indistinguishable (Fig. 3B). In both mutants the V-log I curve was shifted to higher levels of light intensity by about 1 log unit as compared to wild-type flies. A quantitative difference between TRPL- and TRPL-eGFP-expressing flies was found in the amplitude of the response to a second light pulse that was applied 10 seconds after the first pulse (Fig. 3A,C). Responses of TRPL-eGFP-expressing flies to the second light pulse were significantly larger than the corresponding responses of trp mutants, indicating that the photoresponse of the latter recovered slower from the light-induced inactivation (Fig. 3A,C). The kinetics of the dark recovery from response inactivation is one of the hallmarks of the trp mutant phenotype, where slower recovery corresponds to stronger trp phenotype (Minke, 1982). It is possible that the observed difference in the responses to a second light pulse resulted from differences in the amounts of TRPL or TRPL-eGFP present in the corresponding photoreceptor cells.

To compare more directly the characteristics of the native TRPL to TRPL-eGFP channels, light-induced currents (LIC) were measured in TRPL-eGFP-expressing flies and trp mutants. The waveforms of the LIC in response to increasing light intensities were very similar in both fly strains (data not shown). Furthermore, both strains revealed virtually identical current-voltage relationship and the reversal potential (which is the membrane voltage in which the light-induced inward current changes to outward current in response to the constant light pulse) was ∼0 mV in both fly strains (Fig. 3D) (Hardie and Minke, 1992). Altogether, the physiological properties of the TRPL-eGFP and the native TRPL channels were very similar, indicating that the attachment of the eGFP-tag to the TRPL channel did not modify the functional properties of the TRPL channel both in vivo and in situ.

In an attempt to determine how the light signal leads to the translocation of the TRPL channel we examined whether rhodopsin is required for this process and whether the efficiency of TRPL relocation is proportional to the amount of activated rhodopsin. Fig. 4 shows that TRPL-eGFP failed to move following illumination with orange light in the ninaE mutant, which lacks Rh1 (O'Tousa et al., 1985; Zuker et al., 1985) (Fig. 4C,D). This experiment showed that rhodopsin, a major protein of the microvilli, is required for light-dependent TRPL-eGFP internalization. This result did not rule out the possibility that the defect in TRPL-eGFP internalization is due to a requirement of Rh1 for proper rhabdomere formation rather than its function as a photon capturing protein. To distinguish between these possibilities we made use of a Drosophila mutant that ectopically expresses the UV-sensitive rhodopsin Rh3 in the major photoreceptor cells of a ninaE mutant (Feiler et al., 1992). The presence of Rh3 in R1-6 photoreceptor cells rescues the morphological defect of the rhabdomeres, but Rh3 cannot be activated by long wavelength light (Feiler et al., 1992). Illumination with orange light resulted in internalization of TRPL-eGFP in wild-type flies (Fig. 4A,B) but not in flies carrying the UV-sensitive Rh3 instead of the native Rh1 (Fig. 4E,F). However, partial internalization of TRPL-eGFP was observed when the Rh3-expressing flies were illuminated with UV (Fig. 4G). These results show that activation of rhodopsin is required for TRPL internalization and identifies rhodopsin as the light-sensor that initiates the relocation of the ion channel. The mutant ninaB^P315 cannot form the retinal chromophore from carotenoids and shows a dramatically reduced amount of opsin protein on western blots (von Lintig et al., 2001). In this mutant we observed variable degrees of TRPL-eGFP translocation (Fig. 4H-J). Whereas in some individual flies orange-light illumination resulted in almost complete translocation of TRPL-eGFP to the cell body, in other individuals TRPL-eGFP was confined to the rhabdomeres after illumination. Dark-raised flies always displayed rhabdomeral localization of TRPL-eGFP. A plausible explanation for the above variability is that ninaB^P315 flies differ in the individual content of rhodopsin generated from carotenoid-independent sources of retinal, and hence show variable TRPL-eGFP translocation phenotypes.

Activated rhodopsin (i.e. in its metarhodopsin state) is removed from the rhabdomeral membrane in a process that requires the formation of Arrestin 2-metarhodopsin complexes and binding to clathrin-coated vesicles (Dolph et al., 1993; Alloway et al., 2000; Kiselev et al., 2000). A question arises as to whether TRPL-eGFP is internalized with activated rhodopsin, in the same endocytotic vesicles. If this were the case, the efficiency of TRPL internalization should be proportional to the amount of activated rhodopsin generated by illumination. The difference in the absorption spectra of rhodopsin and metarhodopsin lead to production of a very small amount (<2%) of metarhodopsin during orange illumination, whereas green and blue light convert about 50% or 70% of rhodopsin into metarhodopsin, respectively. We, therefore, exposed TRPL-eGFP-expressing flies to orange, green and blue light for 16 hours, and found, by evaluation of TRPL-eGFP-localization with the water immersion method, that each light condition resulted in significant TRPL-eGFP internalization. However, orange light was the most efficient in removing TRPL-eGFP from the rhabdomere, thus indicating that there is no correlation between TRPL translocation and internalization of the photopigment (Fig. 4K). For this reason orange light was used for inducing TRPL internalization in all the experiments described in this paper, except when noted otherwise. Taken together, the results indicate that rhodopsin is necessary for triggering TRPL internalization, which is completely inhibited in the rhodopsin null mutant ninaE¹⁷. Furthermore, the data suggest that continuous activation of a relatively small percentage of wild-type rhodopsin by orange light or activation of residual amounts of rhodopsin present in some individual flies of the ninaB^P315 mutant are sufficient for triggering TRPL internalization.

Two arrestins, Arrestin 1 (Arr1) and Arrestin 2 (Arr2), are expressed in Drosophila photoreceptor cells. According to a recent report Arr1 is required for light-dependent endocytosis of rhodopsin (Satoh and Ready, 2005), whereas the major function of Arr2 is inactivation of activated rhodopsin (Dolph et al., 1993). In addition to visual G protein-coupled receptors, arrestins have also been implicated in the internalization of Frizzled and TGF-β receptors (Chen et al., 2003a; Chen et al., 2003b). Investigation of TRPL-eGFP translocation in arr1 null mutant revealed no significant difference from wild-type photoreceptors (Fig. 5A,B,G) and only partial impairment of TRPL-eGFP translocation was observed in the arr2 null mutant (Fig. 5C,D,G). However, translocation of TRPL-eGFP was severely inhibited in the arr1; arr2 double mutant (Fig. 5E-G), in which we observed TRPL-eGFP in the rhabdomeres of both light- and dark-raised flies. Hence, the presence of either Arr1 or Arr2 is sufficient at least for partial TRPL-eGFP internalization whereas the lack of both arrestins interferes with proper translocation of TRPL-eGFP.

Each one of the three signaling proteins, Gq, PLC and INAD is essential for the generation of the light-induced current and these three proteins together with TRP are essential for the ensuing Ca²⁺ influx. It is therefore important to examine whether these proteins are required for TRPL translocation.

In order to test a possible requirement of the visual G protein for TRPL translocation we used a hypomorphic mutant in the Gα subunit, Gαq¹, which produces ∼1% of the wild-type levels and shows a dramatic (~1000-fold) loss in light sensitivity (Scott et al., 1995). In most of the photoreceptor cells (typically five out of the six R1-6 cells) TRPL-eGFP was translocated to the cell body as in wild type when the flies were raised in the light (Fig. 6A,B). On average, ommatidia contained 0.95±0.01 photoreceptor cells in which TRPL-eGFP remained completely in the rhabdomere. The amount of Gαq present in individual photoreceptor cells of the Gαq¹ mutant may vary and may fall below a value required for TRPL-eGFP internalization in some cells, whereas most cells still contain enough G protein to efficiently trigger TRPL internalization. A similar all-or-none response as in the Gαq mutant was observed when internalization of TRPL-eGFP was triggered with very dim orange light (2 Lux) illumination (see supplementary material Fig. S2). In order to further examine this possibility we generated flies carrying the mutated Gαq allele over a deletion (Gαq¹/Df(2R)vg135). This mutant displays a significantly larger reduction in light sensitivity (Scott et al., 1995) than the homozygous Gαq¹ mutant, suggesting that the photoreceptors of this fly express negligible amounts of Gαq. In Gαq¹/Df(2R)vg135 TRPL-eGFP internalization was strongly inhibited. On average, in 3.65±0.43 photoreceptor cells per ommatidium TRPL-eGFP remained in the rhabdomeres, indicating that further reduction of the Gαq content increases the number of cells that fail to show translocation of TRPL-eGFP (Fig. 6C,D). Altogether, Fig. 6A-D strongly suggest that even a dramatic reduction of the level of the G protein to ∼1% did not block TRPL-eGFP translocation and, most likely, almost complete elimination of Gαq was required to block TRPL translocation.

Gαq-GTP activates the norpA-encoded phospholipase Cβ. In norpA^P24, an almost complete null mutant of phospholipase Cβ, no light-dependent translocation of the eGFP-tagged TRPL to the cell body was observed (Fig. 6E,F). In another allele of norpA (norpA^P57), which produces ∼20% of the amount of PLC of wild-type photoreceptors and is characterized by a photoresponse of normal amplitude but with slow response termination (Pearn et al., 1996; Cook et al., 2000), translocation of TRPL-eGFP from the rhabdomere to the cell body was observed upon illumination (Fig. 6G,H) indicating that a large reduction of PLC is required to block TRPL translocation.

Absence of the major ion channel, TRP, in the trp^P343 mutant resulted in the complete loss of the translocation of TRPL-eGFP from the rhabdomere to the cell body (Fig. 6I,J). Inhibition of TRPL-eGFP internalization was also observed in the null mutant of the INAD scaffold protein inaD¹. INAD is essential for maintained localization of PLC and TRP to the rhabdomere (Chevesich et al., 1997; Tsunoda et al., 1997), (Fig. 6K,L). Accordingly, the absence of INAD affects two proteins that are essential for TRPL movement and thus its absence was expected to block TRPL translocation.

The inaF mutant shows a severe reduction in the amount of TRP and the function of the remaining TRP molecules is impaired (Li et al., 1999). Fig. 6M,N shows that TRPL translocation is blocked in this mutant. It thus confirmed that TRP function is required for TRPL translocation. Additional support for this conclusion came from studies of a mutant with impaired light control of TRP gating. The mutant trp^P365 expresses constitutively active TRP channels (Yoon et al., 2000). Photoreceptors of trp^P365 show fast and severe degeneration in flies homozygous for this allele, owing to uncontrolled Ca²⁺-influx into the cell (Yoon et al., 2000; Wang et al., 2005). Heterozygous mutants kept in the dark maintain intact photoreceptors for up to 4 days post-eclosion. We investigated the localization of TRPL-eGFP in a trp^P365/+ background and found that it localized outside the rhabdomeres in light- and dark-raised flies (Fig. 6O,P). This result strongly suggests that light-independent activation of TRP can lead to localization of TRPL-eGFP in the cell body.

Previously, a requirement for activation of phospholipase Cβ and of the visual G protein for TRPL translocation was ruled out because it was shown that native TRPL is translocated in norpA^P24 (Bähner et al., 2002) and in Gαq¹ (Cronin et al., 2004) mutants. In this study we show that a reduction of Gαq to 1% or of PLCβ to 20% of the wild-type level did not inhibit TRPL-eGFP translocation in most photoreceptor cells. Much larger reduction in the levels of these signaling proteins in Gαq¹/Df(2R)vg¹³⁵ and norpA^P24, respectively, was required to interfere with TRPL-eGFP translocation. The complete inhibition of light-induced TRPL-eGFP translocation by the norpA^P24 mutation (Fig. 6E,F) and the reported translocation of the native TRPL in the same mutant (Bähner et al., 2002) can be explained by a reduced sensitivity to activation of phototransduction of the TRPL-eGFP internalization process as compared to that of the native TRPL. The norpA^P24 mutant and also the inaD¹ mutant show very small but significant activation of the phototransduction cascade (Hardie et al., 2003; Tsunoda et al., 1997), which seems to be sufficient to trigger translocation of native TRPL. Alternatively, the discrepancy between native and eGFP-tagged TRPL might imply the existence of a distinct norpA-independent mechanism that is specifically compromised for TRPL-eGFP.

To examine directly whether light-induced Ca²⁺ influx is necessary for TRPL-eGFP translocation, we compared light-induced TRPL-eGFP translocation in the presence and absence of Ca²⁺ in the extracellular medium. To this end we incubated sliced heads of trpl³⁰²; trpl-eGFP flies in oxygenated solutions containing either 1 mM Ca²⁺ or 1 mM EGTA (no Ca²⁺ added). The incubated heads were illuminated with orange light (OG 590 edge filter) for 4 hours. After incubation and illumination, the retinas were isolated and examined under a confocal microscope. Fig. 7B,E shows a very significant translocation of TRPL-eGFP to the base of the rhabdomeres that was observed in illuminated retinas incubated with extracellular Ca²⁺. Removal of extracellular Ca²⁺ by EGTA completely blocked the light-induced translocation of TRPL-eGFP (Fig. 7A,D), and the localization of TRPL-eGFP remained very similar to its localization in retinas of dark-raised flies (Fig. 7C,D). Fig. 7 thus demonstrates that removal of extracellular Ca²⁺ inhibits light-induced translocation of TRPL-eGFP.

Towards gaining some insight into the cell biological mechanism which may underlie TRPL internalization we studied the dependence of TRPL-translocation in the mutant shibire (shi). shi encodes a dynamin required for budding-off of vesicles from the membrane. A temperature-sensitive mutant (shi^ts1) fails to recycle synaptic vesicles at the restrictive temperature and renders the flies paralyzed. TRPL-eGFP translocation was unaffected in shi^ts1 indicating that the internalization mechanism of TRPL-eGFP is dynamin-independent (Fig. 8A,B). In ninaC⁵ that is a null-mutant of an unconventional myosin III, TRPL-eGFP was properly localized in the rhabdomeres in dark-raised flies. Illumination of the ninaC⁵ mutant with orange light resulted in only partial internalization of the ion channel (Fig. 8C,D). Thus, this protein, which affects organization of the cytoskeleton (Matsumoto et al., 1987) is not required for the incorporation of TRPL-eGFP into the rhabdomere, but its mutation affects translocation of TRPL-eGFP to the cell body.

The light-dependent translocation of TRPL in Drosophila photoreceptor cells alters the composition of ion channels in the rhabdomeral membrane. Thus, the rhabdomeral membrane contains functional TRP and TRPL channels in dark-raised

flies but harbors only functional TRP in light-raised flies because of the reduced number of TRPL channels (Bähner et al., 2002) and because the Ca²⁺-influx via the TRP channels inactivates the remaining TRPL channels (Reuss et al., 1997).

Identification of the triggering signal, which initiates the specific internalization of TRPL is important for understanding TRPL translocation. One possible triggering mechanism is the light-dependent activation of the TRPL channels themselves, which could form an ion channel conformation susceptible for internalization. Other possibilities include activation of TRPL internalization by increase in intracellular Ca²⁺ concentration, which results from Ca²⁺ influx through activated TRP, or internalization of TRPL together with metarhodopsin in an arrestin-dependent way. Our findings that functional rhodopsin is required for TRPL-eGFP internalization are in agreement with each one of these three possibilities. However, since activation of a small fraction of the rhodopsin molecules by orange light in wild-type flies was sufficient for maximal TRPL-eGFP internalization and since TRPL-eGFP internalization could be triggered by the residual rhodopsin present in the ninaB^P315 mutant, it seems unlikely that removal of TRPL-eGFP from the rhabdomeral membrane is achieved by co-internalization with metarhodopsin.

A recent paper by Satoh and Ready (Satoh and Ready, 2005) reveals that Arr1 and Arr2 have separate functions. Arr1 is required for rhodopsin internalization whereas Arr2 is required for rhodopsin inactivation. Thus, if TRPL-eGFP was internalized together with rhodopsin one would expect that mutation of arr1 inhibits internalization of TRPL-eGFP. This is not the case. On the other hand, the defect of rhodopsin inactivation in the arr2 mutant is amplified in the arr1/arr2 double mutant (Dolph et al., 1993) as is the case with the inhibition of TRPL-eGFP internalization. This correlation may suggest that proper inactivation of rhodopsin is required for complete internalization of TRPL-eGFP. In addition the arr2 mutant shows light-dependent degeneration of the photoreceptor cells. Although in our experiments we used young flies which did not show any sign of degeneration we can not exclude that this may affect the internalization of TRPL-eGFP.

Our data indicate that activation of signaling proteins such as Gαq, PLCβ and TRP, which operate downstream of rhodopsin in the phototransduction cascade, are essential for TRPL-eGFP translocation. However, activation of a relatively small percentage of these signaling molecules during many minutes triggers complete TRPL-eGFP translocation. Because of the huge amounts of signaling proteins in the photoreceptor cell (Hardie and Right, 2001) and the high gain of phototransduction, activation of a relatively small fraction of the signaling molecules is sufficient to induce nearly saturated responses via opening of the TRP channels. Activation of TRPL alone is not sufficient and possibly not even required for its translocation to the cell body, rather the second ion channel, TRP, must be activated. This is evident from the lack of TRPL-eGFP translocation in the trp null mutant and in mutants with altered TRP function, such as inaD¹ and inaF^P106x. The observed localization of TRPL-eGFP to the cell body irrespective of the light condition in a mutant expressing constitutively active TRP (trp^P365) suggests that activity of TRP, independent of activation of the phototransduction cascade and independent of the activation of the TRPL channel, causes translocation of TRPL-eGFP to the cell body. Taken together, an increase of the intracellular concentration of Ca²⁺ through TRP channels in the light is an attractive mechanism for triggering TRPL translocation. This hypothesis is strongly supported by our finding that large reduction of extracellular Ca²⁺ in the eye inhibits light induced TRPL-eGFP internalization.

To further elucidate the mechanism by which TRPL-eGFP is transported from the rhabdomeral membrane to the cell body, we analyzed the possible requirement of dynamin and of the myosin III NINAC for TRPL-eGFP translocation. Dynamin is crucial for the budding of vesicles from the plasma membrane in several endocytotic pathways (Conner and Schmid, 2003). However, vesicular internalization pathways may also be independent of dynamin. For example, for the endocytosis of G protein-coupled receptors dynamin-dependent and dynamin-independent pathways have been reported (Zhang et al., 1996; Pals-Rylaarsdam et al., 1997; Vickery and von Zastrow, 1999). The internalization of TRPL-eGFP is not affected in a temperature-sensitive shibire mutant at the restrictive temperature, indicating that the translocation of TRPL-eGFP is dynamin independent. Furthermore, we observed no defect in the translocation of TRPL-eGFP from the cell body to the rhabdomere in a mutant lacking the myosin III NINAC, but TRPL-eGFP translocation in the opposite direction was compromised in this mutant. Because myosin III is a plus ended myosin and the plus ends of actin filaments are oriented towards the tips of the microvilli (Lee and Montell, 2004), it is not feasible that this myosin mediates the transport of proteins out of the rhabdomere. However, the ninaC⁵ mutant exhibits secondary defects such as disruption of the actin cytoskeleton and retinal degeneration (Matsumoto et al., 1987; Hicks and Williams, 1992). We suggest that the observed partial inhibition of TRPL-eGFP transport from the rhabdomere to the cell body is due to these secondary defects.

Besides TRPL, at least two other proteins mediating Drosophila phototransduction, Arrestin 2 and the visual Gαq, undergo light-dependent translocation between the rhabdomere and the cell body (Byk et al., 1993; Kiselev et al., 2000; Kosloff et al., 2003; Lee et al., 2003; Lee and Montell, 2004). Likewise, in vertebrate photoreceptors arrestin and the visual G protein transducin translocate between the inner and outer segment in a light-dependent way (Arshavsky, 2003). In both visual systems, arrestin and G protein movements occur in opposite directions, that is, in the light arrestin accumulates whereas the G protein is depleted in the photoreceptive membrane and vice versa in the dark. Accordingly, these light-dependent relocations of visual signaling proteins make the photoreceptor more sensitive in the dark and less sensitive in the light and mediate long-term adaptation of the Drosophila and vertebrate visual systems (Sokolov et al., 2002; Lee et al., 2003). A third protein that translocates in vertebrate photoreceptors is the Ca²⁺ binding protein recoverin (Strissel et al., 2005).

The mechanisms underlying these protein translocations have been elucidated in part for Drosophila arrestin (Lee and Montell, 2004), the Gαq subunit (Kosloff et al., 2003; Cronin et al., 2004; Elia et al., 2005) and for vertebrate transducin (Sokolov et al., 2004). Translocation of Drosophila arrestin from the cell body to the rhabdomere has been reported to require the ninaC-encoded myosin III which may actively transport arrestin along the actin cytoskeleton of the photoreceptor microvilli through PIP₃-enriched vesicles, to which arrestin binds (Lee and Montell, 2004). However, the requirement of the myosin III NINAC for arrestin translocation has been challenged in a more recent publication (Satoh and Ready, 2005). Removal of arrestin from the rhabdomeral membranes in the dark does not require cytoskeletal elements and may thus occur passively (Lee and Montell, 2004). Likewise, Gαq translocation into the rhabdomere, but not its removal, is facilitated by the myosin III NINAC (Cronin et al., 2004). Translocation of vertebrate transducin is aided by phosducin, an abundant photoreceptor-specific protein that binds to the βγ subunits of transducin (Sokolov et al., 2004). Phosducin increases the solubility of the G protein subunits and may thereby facilitate transducin translocation.

These mechanisms are markedly different from the mechanism underlying TRPL translocation because TRPL is a transmembrane protein that cannot enter the soluble fraction and needs to be removed from the rhabdomere by an endocytotic pathway, whereas arrestin and the visual G protein change from a membrane attached state to a soluble state. Therefore, elucidating the triggering mechanism of TRPL translocation reported in the present study is the first step for unraveling the mechanism underlying an important cellular process.

The following strains and mutants of Drosophila melanogaster were used: w Oregon R, yw; trpl³⁰² (Niemeyer et al., 1996), yw; trpl³⁰²; trp^P343 (Yang et al., 1998), yw;;trp^P343 (Pak, 1979), w;;trp^P365 (Yoon et al., 2000), yw; ninaE¹⁷ (O'Tousa et al., 1985), yw; ninaE¹⁷/P[Rh1+3] (Feiler et al., 1992), w;;ninaB^P315 (von Lintig et al., 2001), arr1¹cn bw (Dolph et al., 1993), w;; arr2³ st (Dolph et al., 1993), w; arr1¹ cn bw; arr2³ (Dolph et al., 1993), w; Gαq¹ (Scott et al., 1995), Df(2R)vg135, nompA^vg135/CyO, S^* bw¹ (Lasko and Pardue, 1988), w norpA^P24 (Bloomquist et al., 1988), norpA^P57; bw; st (Pearn et al., 1996), yw; inaD¹ cn bw (Tsunoda et al., 1997), w inaF^P106x (Li et al., 1999), w shi ^ts1 (Grigliatti et al., 1973), and w; ninaC⁵ (Pak, 1979). The transgenic flies expressing TRPL-eGFP, which were generated in this study, were crossed with the mutants to obtain the genotypes indicated in the figure legends. If the mutations were located on the second or third chromosome, the flies were made homozygous for both the mutation and the trpl-eGFP transgene, inserted into the other autosome. If the mutation mapped to the X-chromosome, male flies of the F1 generation that were hemizygous for the corresponding mutation and heterozygous for the trpl-eGFP transgene were used for analysis.

Flies were raised at 24°C in a 12 hours light/12 hours dark cycle. 12-18 hours before the experiment, flies were kept in the dark, or were illuminated with white light (18 W fluorescent lamp, ∼700 Lux), orange light (acrylic glass cut-off filter transmitting light >560 nm, 450 Lux), green light (acrylic glass wide-band filter transmitting light between 460 nm and 610 nm, 30 Lux), blue light (acrylic glass wide-band filter transmitting light between 310 nm and 490 nm, 3 Lux) or with UV (344 nm monochromatic light, xenon high-pressure lamp 150 W). Dark-raised flies were dissected under dim red light (Schott RG 630, cold light source KL1500, Schott, Germany), whereas light-raised flies were dissected under white light.

To generate the DNA construct used to express a TRPL-eGFP fusion protein, the stop codon and the 3′ untranslated region of a trpl cDNA clone (Phillips et al., 1992) in p BluescriptII SK (Stratagene, Germany) were removed by substituting the sequence 3′ of a AflII restriction site with a PCR fragment containing AflII and ApaI cloning sites. The modified trpl cDNA was subcloned after partial digestion with EcoRI and ApaI into a p-Bluescript vector containing 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). In the resulting construct the trpl gene was placed in between the Rh1 promoter and the eGFP gene to produce a fusion protein in which the eGFP tag was located at the C-terminus of TRPL. Rh1 promoter, trpl and eGFP coding sequences were then cloned into the XhoI restriction site of the P-element transformation vector YC4 [a gift from S. Britt, University of Colorado; the YC4 vector is derived from the vector Y.E.S. (Patton et al., 1992)]. P-element-mediated transformation of Drosophila was carried out as described previously (O'Tousa, 1992). Host strains used were Drosophila yellow white (yw) and the trpl³⁰² mutant in a yw background. The transformants were made homozygous for the P-element inserts.

Western blot analysis was carried out with proteins obtained from homogenates of Drosophila heads. Proteins were extracted with 1× SDS-PAGE extraction buffer (4% SDS in 65 mM Tris-HCl, pH 6.8) for 15 minutes at room temperature. Proteins from three heads were subjected to SDS-PAGE according to Laemmli (Laemmli, 1970), using 7.5% polyacrylamide gels (Midget System, Amersham Pharmacia Biotech, Germany). For immunoblotting, proteins were electrophoretically transferred to PVDF membranes (BioRad Laboratories, Germany) and processed as described previously (Huber et al., 2000), except that the ECL Western Blotting Analysis system (GE Healthcare, Germany) was used for signal detection instead of ¹²⁵I-labeled secondary antibodies. The antibodies used for western blot analysis were α-DmTRP (Bähner et al., 2000), α-DmTRPL (Bähner et al., 2002) and α-GFP (Roche, Germany).

Electroretinograms (ERGs) were performed as described previously (Peretz et al., 1994a). ERG traces were recorded using Axon analog to digital converter operated by pClamp 8 software on a PC computer.

To measure light-induced currents, orange light (OG 590 Schott edge filter) from a Xenon high-pressure lamp (75 W) was delivered to isolated ommatidia via the objective lens (40×, Zeiss, Germany) and attenuated up to seven orders of magnitude by neutral density filters. The maximal luminous intensity of the orange light at the level of the ommatidia was about 3.0 log units above the intensity required for a half-maximal response of the R1-6 photoreceptors.

Dissociated ommatidia were prepared from newly eclosed adult flies (<1 hour post-eclosion). Whole-cell patch-clamp recordings were performed as described previously (Hardie and Minke, 1992; Peretz et al., 1994b). Recordings were made at 21°C using patch pipettes of 5-10 MΩ pulled from fiber-filled borosilicate glass capillaries. Series resistance of 7-14 MΩ was carefully compensated (>80%) during all experiments. Signals were amplified with an Axopatch-1D (Axon Instruments) patch-clamp amplifier, sampled at 2 kHz, and filtered below 1 kHz. The bath solution contained (in mM): 120 NaCl, 5 KCl, 10 TES buffer (N-Tris-(hydroxymethyl)-methyl-2-amino-ethanesulphonic acid, pH 7.15), 4 MgSO₄, 1.5 CaCl₂. For all the experiments an internal solution that blocked K⁺ channels was used. The whole-cell recording pipette contained (in mM): 120 CsCl, 15 tetraethylammonium (TEA) chloride, 2 MgSO₄, 10 TES buffer (pH 7.15), 4 MgATP, 0.4 Na₂GTP, 1 NAD. The external solution was perfused via a perfusion system at a rate of 25 chambers/minute.

Flies at the age of 2-4 days expressing TRPL-eGFP were anaesthetized with diethylether for 2 minutes and kept on ice. The flies were then spiked on an insect needle and mounted with plasticine on an object slide. Fluorescence microscopy (microscope: DM LFS, Leica, Germany; light source: ebq 100 dc-1 [100 W], Jena GmbH, Germany) was carried out with a water immersion objective (HCX APO, L 20×/0.5 W U-V-I, Leica, Germany) using the Leica I3 filter set (illumination path: BP 450-490 nm, dichroic mirror/reflector: 510 nm, observation path: LP 515 nm) for excitation and detection of eGFP-fluorescence. The fluorescence was documented with a digital camera (DC200, Leica, Germany).

For observing the fluorescence in the deep pseudopupil the flies were anaesthetized with CO₂ and were then examined with a 5× objective (HC PL Fluotar, 5×/0.15, Leica, Germany) using the same filter set as described above or with a Zeiss Stemi SV 11, equipped with an epifluorescence device.

For quantitative analyses of the amount of TRPL-eGFP located in the rhabdomeres, fluorescence images obtained with the water immersion technique were analyzed with ImageJ 1.32j software (National Institute of Health, USA). The relative amount of TRPL-eGFP present in the rhabdomeres (R) was calculated using the formula R=(Ir-Ib)/[(Ir-Ib)+(Ic-Ib)], where Ir, Ib, and Ic are the fluorescence intensities in the rhabdomeres, in the background, and in the cell body, respectively. The data were normalized to the values obtained for dark raised flies that were set to 100%. Background intensities were determined in the center of the ommatidium where the rhabdomere of the R7/R8 cells is located.

The number of photoreceptor cells of the Gαq mutant, in which TRPL-eGFP translocation was inhibited, was determined by counting fluorescing rhabdomeres of 16 to 31 ommatidia per individual fly from three to five light-raised flies.

For the Ca²⁺-dependent TRPL-eGFP translocation, sliced heads of TRPL-eGFP-expressing flies were placed in oxygenated modified extracellular solution used for electrophysiology and illuminated for 4 hours with orange light. The live retinas were then isolated and examined with LSM (Olympus Fluoview 200 confocal microscope with 60×/0.9 w LUMPlan objective).

The authors thank Bill Pak, Charles Zuker and Steve Britt for providing Drosophila mutants. We are grateful to Reinhard Paulsen for discussion of results and to Gabi Gerdon and Renu Walia for their help in the generation of DNA constructs and transgenic Drosophila, respectively. Tanja Landmann provided technical assistance with immunocytochemistry and Elkana Kohn helped with the ERGmeasurements. This work was supported by grants from the Deutsche Forschungsgemeinschaft (Hu 839/2-4) to A.H., the German-Israel Foundation (GIF) to A.H. and B.M., the National Eye Institute to B.M. (R01-EY 03529), and the Israel Science Foundation (ISF) to B.M.