Rescue of light responses in the Drosophila "null" phospholipase C mutant, norpAP24, by the diacylglycerol kinase mutant, rdgA, and by metabolic inhibition.

Light responses in Drosophila are reportedly abolished in severe mutants of the phospholipase C (PLC) gene, norpA. However, on establishing the whole-cell recording configuration in photoreceptors of the supposedly null allele, norpAP24, we detected a small ( approximately 15 pA) inward current that represented spontaneous light channel activity. The current decayed during approximately 20 min, after which tiny residual responses (<2 pA) were elicited by intense flashes. Both spontaneous currents and light responses appeared to be mediated by residual PLC activity, because they were enhanced by impairing diacylglycerol (DAG) kinase function by mutation (rdgA) or by restricting ATP but were reduced or abolished by a mutation of the PLC-specific Gq alpha subunit. It was reported recently that metabolic inhibition activated the light-sensitive transient receptor potential and transient receptor potential-like channels, even in norpAP24, leading to the conclusion that this action was independent of PLC (Agam, K., von Campenhausen, M., Levy, S., Ben-Ami, H. C., Cook, B., Kirschfeld, K., and Minke, B. (2000) J. Neurosci. 20, 5748-5755). However, we found that channel activation by metabolic inhibitors in norpAP24 was strictly dependent on the residual PLC activity underlying the spontaneous current, because the inhibitors failed to activate any channels after the spontaneous current had decayed. By contrast, polyunsaturated fatty acids invariably activated the channels independently of PLC. The results strongly support the obligatory requirement for PLC and DAG in Drosophila phototransduction, suggest that activation by metabolic inhibition is primarily because of the failure of diacylglycerol kinase, and are consistent with the proposal that polyunsaturated fatty acids, which are potential DAG metabolites, act directly on the channels.

Light responses in Drosophila are reportedly abolished in severe mutants of the phospholipase C (PLC) gene, norpA. However, on establishing the whole-cell recording configuration in photoreceptors of the supposedly null allele, norpA P24 , we detected a small (ϳ15 pA) inward current that represented spontaneous light channel activity. The current decayed during ϳ20 min, after which tiny residual responses (<2 pA) were elicited by intense flashes. Both spontaneous currents and light responses appeared to be mediated by residual PLC activity, because they were enhanced by impairing diacylglycerol (DAG) kinase function by mutation (rdgA) or by restricting ATP but were reduced or abolished by a mutation of the PLC-specific G q ␣ subunit. It was reported recently that metabolic inhibition activated the light-sensitive transient receptor potential and transient receptor potential-like channels, even in norpA P24 , leading to the conclusion that this action was independent of PLC (Agam, K., von Campenhausen, M., Levy, S., Ben-Ami, H. C., Cook, B., Kirschfeld, K., and Minke, B. (2000) J. Neurosci. 20, 5748 -5755). However, we found that channel activation by metabolic inhibitors in norpA P24 was strictly dependent on the residual PLC activity underlying the spontaneous current, because the inhibitors failed to activate any channels after the spontaneous current had decayed. By contrast, polyunsaturated fatty acids invariably activated the channels independently of PLC. The results strongly support the obligatory requirement for PLC and DAG in Drosophila phototransduction, suggest that activation by metabolic inhibition is primarily because of the failure of diacylglycerol kinase, and are consistent with the proposal that polyunsaturated fatty acids, which are potential DAG metabolites, act directly on the channels.
Phototransduction in the microvillar photoreceptors of invertebrates is mediated by the inositol lipid or phosphoinositide signaling cascade. Largely because of its molecular genetic potential, Drosophila has become an important model system not only for phototransduction but for inositol lipid signaling in general (1)(2)(3). The generally accepted mechanism of excitation conforms to the canonical phosphoinositide cascade (see Fig. 8). After photoisomerization in the microvillar membrane, rhodopsin activates a G q protein which, in turn, binds to and activates phospholipase C (PLC␤4 1 isoform), resulting in the hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) to generate inositol 1,4,5-trisphosphate and DAG. By a still unknown mechanism, this process activates at least two classes of light-sensitive channels in the microvilli, encoded by the transient receptor potential (trp) and trp-like (trpl) genes, which are the prototypical members of the large TRP ion channel superfamily (4,5). Recent evidence suggests that lipid products of PIP 2 hydrolysis are the messengers of excitation (6,7), but this is still a subject of controversy (3, 8 -10).
The key argument for the conclusion that phototransduction in Drosophila is an obligatory PLC-signaling cascade is the finding that the norpA (no receptor potential A) gene encoded an eye-enriched PLC␤ and that severe mutations in norpA abolished the response to light (11)(12)(13). Less severe hypomorphic norpA mutants show reduced sensitivity with dramatically slowed kinetics (14 -16) and reduced amplification, as indicated by the small amplitude of responses to single photons known as quantum bumps (17). We recently discovered that sensitivity in norpA hypomorphs could be greatly facilitated by mutations in the retinal degeneration A (rdgA) gene and also by depletion of cytosolic ATP (17). This supports the hypothesis that DAG or its metabolites are excitatory messengers, because the rdgA gene encodes diacylglycerol kinase (DGK) (18), which is the major enzyme controlling DAG in most cells (19). In the present study we show that even the supposedly null mutant norpA P24 has a residual sensitivity to light, which can be massively enhanced by either ATP depletion or the rdgA mutation but still appears to be mediated by PLC. In light of these findings we repeated some key experiments, which assumed that norpA P24 was functionally null. In particular we found that contrary to a recent report (8), activation of the light-sensitive channels by metabolic inhibitors is strictly dependent on PLC activity, and we present evidence here that this represents the failure of DGK. By contrast, excitation of the channels by PUFAs (6), suggested previously to be caused indirectly by metabolic inhibition (8) or by the activation of endogenous PLC (20), appeared to be independent of PLC activity and seems likely to represent a direct effect on the channels or their lipid environment.

EXPERIMENTAL PROCEDURES
Flies-Flies (Drosophila melanogaster) were raised on standard medium in the dark at 25°C. The wild type (WT) strain was white-eyed Oregon R. Mutant alleles included: G␣q 1 , a severe hypomorphic mutation (approximately 1% protein) of the eye-enriched G q protein ␣ subunit (21); norpA P24 , a reportedly protein-null mutant of PLC with a 28-bp deletion (14); Df(1)HC244/FM7C, a deficiency covering 3E8-4F11, which includes the entire norpA genomic region; rdgA 1 (also termed rdgA BS12 ), the most severe reported allele of an eye-specific DAG kinase with Ͻ5% DGK activity and complete degeneration on the day of eclosion (18); and sl 2 and sl 1 , severe mutants of PLC␥ (22).
Whole-cell Recordings-Dissociated photoreceptor clusters (ommatidia) were prepared as described previously (23,24) from recently eclosed adult flies and transferred to a recording chamber on an inverted Nikon Diaphot microscope. The bath solution contained (in mM): 120 NaCl, 5 KCl, 10 TES, 4 MgCl 2 , 1.5 CaCl 2 , 25 proline, and 5 alanine. Unless otherwise stated, intracellular solution was (in mM): 140 potassium gluconate, 10 TES, 4 MgATP, 2 MgCl 2 , 1 NAD, and 0.4 NaGTP. The pH of all solutions was adjusted to 7.15. To determine currentvoltage relationships (Fig. 7), potassium gluconate was replaced with 125 mM cesium gluconate and 15 mM tetraethylammonium chloride to block K ϩ channels. Carbonyl cyanide m-chlorophenylhydrazone (CCCP), 2,4-dinitrophenol (DNP), and linolenic acid (LNA) were obtained from Sigma and were applied from a puffer pipette (tip diameter, 2-4 m) positioned ϳ10 m from the recorded cell. Whole-cell voltage clamp recordings were made at a holding potential of Ϫ70 mV unless otherwise stated, using electrodes of approximately 10 -15 megaohms resistance. Series resistance values were generally below 30 megaohms and were compensated to approximately 80% when recording macroscopic responses but not for collecting bumps. Data were sampled at 0.5-1 kHz and filtered at 100 Hz using Axopatch 1-D or 200B amplifiers and pCLAMP 6 or 8 software (Axon Instruments, Foster City, CA). Cells were stimulated via a green light-emitting diode. All intensities are expressed with respect to effectively absorbed photons in WT flies, which were estimated by counting quantum bumps at the lowest intensities and then calibrating relative intensities by using a linear photodiode. Quantum bumps were detected and analyzed using Mini-analysis (Synaptosoft, Decatur, GA). The quantum efficiency (QE) (equal to the percentage of absorbed photons eliciting a quantum bump) was estimated relative to WT by counting the number of quantum bumps in response to calibrated flashes (25), or when they could not always be clearly separated, by integrating the current under the entire response and dividing by the average bump integral current recorded in the same cell. Amplitudes of the spontaneous currents were measured with respect to the final level reached after complete decay of the current to a quiet base line. The data are presented here as mean Ϯ S.D.

Spontaneous Currents and Light Responses in norpA P24 -It
is well known that light responses in severe norpA hypomorphs have extremely slow kinetics, with responses to brief flashes continuing for several minutes (14) because of the generation of quantum bumps with exceptionally long latencies (15,16). This behavior is believed to reflect a general principle of G-proteincoupled signaling, namely that the effector enzyme (in this case PLC) also functions as an obligatory GTPase-activating protein required for inactivation of the activated GTP-bound G q ␣ subunit. In WT flies the density of G-protein and PLC (approximately 100 copies/microvillus) is such that diffusional encounters of G-proteins with PLC occur within milliseconds, resulting in rapid activation of PLC swiftly followed by GTPase activity and inactivation. However, when PLC levels are very low (Ͻ Ͻ1 copy/microvillus in severe norpA hypomorphs), activated G-protein ␣ subunits can remain in the active GTPbound form and diffuse for many seconds or even minutes before finally encountering a PLC molecule, resulting in a greatly delayed cycle of activation and inactivation (15,16). Consequently, on establishment of the whole-cell recording configuration there is an ongoing barrage of noisy inward current because of the summation of bumps with extremely long latencies generated by previous illumination. Depending on the severity of the allele, this activity decays in the dark during several seconds or minutes after which responses can be elicited from a quiet base line (15)(16)(17).
Although it is considered as a null mutation, we wondered whether a similar behavior might also be observed in norpA P24 . Indeed, in almost all whole-cell recordings from dissociated norpA P24 photoreceptors we detected a small ϳ15-pA noisy inward current on establishment of the whole-cell configuration. This current appeared to be mediated by the light-sensitive channels, because it was blocked by La 3ϩ (Fig. 1), which is known to specifically block TRP channels in these cells (26,27). As in other norpA hypomorphs, the noisy current gradually subsided and in favorable recordings eventually revealed isolated ϳ1-2-pA events (1.4 Ϯ 0.1 pA, n ϭ 5 cells; mean Ϯ S.D.) and finally, after ϳ20 min, a quiet base line. At this point bright flashes delivered to the cells induced tiny (Ͻ2 pA) light responses consisting of little more than a train of miniature "quantum bumps," again ϳ1-2 pA in amplitude and lasting for several minutes (Figs. 1 and 5). Because they would be buried in the noise, such responses could be detected only after the complete decay of the spontaneous current and even then only in recordings with excellent signal-noise ratios. Under standard recording conditions, we never detected any response to even saturating flashes (corresponding to Ͼ10 6 rhodopsin photoisomerizations) in norpA P24 when they were delivered before the decay of the initial spontaneous activity.
Enhancement of Sensitivity by ATP Depletion-We recently discovered that sensitivity to light in less severe norpA alleles FIG. 1. Spontaneous PLC activity and light response in norpA P24 . A, spontaneous inward current recorded in a norpA P24 photoreceptor as a function of time after establishment of the whole-cell configuration (minutes after w-c) with 5-s samples of recordings shown at the time points indicated. A noisy inward current of ϳ5 pA was evident on establishing the whole-cell configuration, which decayed slowly during 20 min until a quiet base line with occasional 1-2-pA "bumps" was reached. The current was recorded at Ϫ70 mV with a standard electrode solution containing nucleotide additives. B, the initial noisy current was blocked by application of 40 M La 3ϩ (bar) from a puffer pipette (n ϭ 4). C, response to a bright 100-ms flash containing ϳ1 ϫ 10 6 WT effective photons (arrow) recorded in norpA p24 after decay of the spontaneous current. The additional segments were recorded at ϳ70 and 380 s after the flash, showing that isolated 1-2-pA bumps were still occurring at least 6 min after photon absorption. such as norpA P12 and norpA P16 could be greatly enhanced by the omission of ATP from the recording pipette. This facilitation was effectively mimicked by the rdgA mutation encoding DAG kinase, suggesting that DGK normally suppresses bump amplitudes in norpA by metabolizing DAG (17). We performed similar experiments to see whether ATP depletion could also enhance sensitivity in norpA P24 . In recordings made with no ATP in the electrode, the spontaneous activity seen on establishment of the whole-cell configuration in norpA P24 photoreceptors initially increased in magnitude during a period of several minutes, consistent with an increase in the amplitude of the underlying events (quantum bumps). However, the current again eventually decayed during 15-20 min, revealing single quantum bump-like events with amplitudes that were now restored to near WT levels (ϳ7 pA) but terminated more slowly (Figs. 2 and 5). Once a quiet base line had been reached, it was maintained indefinitely in the dark (Ͼ30Ј) with no sign of spontaneous activation of the light-sensitive channels as happens in WT flies recorded without nucleotide additives (8,28). Strikingly, sensitivity to light was greatly enhanced with maximum responses up to ϳ75 pA, and large quantum bumps, similar to the spontaneously occurring bumps during the latter stages of the decay, could be elicited by dimmer flashes (Fig. 2).
Facilitation in norpA,rdgA Double Mutants-To test whether the rescue of light responses in norpA P24 by ATP depletion could be explained by the failure of DGK as reported in other norpA hypomorphs (17), we generated a norpA P24 ,rdgA double mutant using the most severe allele, rdgA 1 (18). As will be described in detail elsewhere, 2 retinal degeneration was effectively rescued in such double mutants. In norpA P24 ,rdgA 1 , even with standard electrode solution containing nucleotide additives, the constitutive inward currents recorded immediately on establishing the whole-cell configuration were greatly enhanced (mean 145 Ϯ 31 pA; n ϭ 19) compared with norpA P24 . Once again, however, the spontaneous currents slowly decayed, reaching a quiet base line after ϳ20 min. Toward the end of this decay, large (ϳ8 pA) slowly terminating quantum bumps could be resolved, very similar in both amplitude and kinetics to those seen in recordings from norpA P24 without ATP (Figs. 3 and 5). Sensitivity to light was also massively enhanced on the rdgA background, and even before the decay of the spontaneous current, bright flashes elicited responses up to ϳ200 pA in amplitude superimposed on the constitutive current. After decay to base line, large bumps could be elicited by dimmer flashes, whereas bright flashes elicited responses of up to ϳ400 pA that decayed with the slow kinetics that are characteristic of norpA P24 (Fig. 3). Overall, the rdgA mutation therefore closely mimics the effect of depletion of ATP, and we conclude that impairment of DGK function is sufficient to account for the effect of ATP depletion in enhancing the spontaneous currents, increasing bump amplitude, and rescuing the macroscopic response to light in norpA P24 . In fact, the overall rescue of sensitivity was even more pronounced in norpA P24 ,rdgA 1 , probably indicating that residual endogenous ATP can support limited DGK function in recordings made without ATP.
Compared with recordings from norpA P24 made with ATP, the overall facilitation in norpA P24 ,rdgA 1 (estimated by comparing the integrals of light-induced currents) was in excess of 2000-fold (2215 Ϯ 884-fold, n ϭ 3). An ϳ50-fold increase can be attributed to the increase in the size of the bump integral current (Fig. 5), the rest presumably represents an increase in QE. In fact QE in norpA P24 ,rdgA 1 was now estimated to be reduced only ϳ30-fold with respect to that seen in WT flies. This comparison highlights both the critical role of DGK, presumably in regulating the supply of DAG required for excitation, and also the considerable potential sensitivity to light in the supposedly null PLC mutant norpA P24 .
Light Responses in norpA P24 Are because of Residual PLC Activity-The spontaneous currents and light responses revealed in norpA P24 could represent residual PLC activity or could be evidence of an alternate phototransduction pathway. To test whether the residual responses in norpA P24 and 2 R. C. Hardie and P. Raghu, unpublished data.

FIG. 2. Effect of omitting ATP in norpA P24
. A, spontaneous currents recorded in norpA P24 with electrode solution containing no ATP or other nucleotide additives. Initially the spontaneous current and associated noise increased severalfold, but it then decayed during ϳ20 min. As a quiet base line is approached, large spontaneous quantum bumps are clearly resolved with variable deactivation defects. B, light responses are recorded after Ͼ20 min of recording in norpA P24 without ATP. Top trace was recorded in darkness. Middle and lower traces are responses to flashes (arrows) containing ϳ2000 and 10 6 WT effective photons. C, response to a saturating flash (10 6 photons) recorded in another norpA P24 cell with ATP in the electrode. norpA P24 ,rdgA 1 were mediated by the same PLC-specific G q protein, we generated norpA p24 ;G␣q 1 double mutants and norpA P24 ,rdgA 1 ;G␣q 1 triple mutants using the severe hypomorphic allele G␣q 1 in which levels of the G-protein ␣ subunit are reduced to ϳ1% of WT levels (21). The spontaneous currents in these double and triple mutants were completely abolished, indicating that they were strictly dependent on activation of G q . Responses to light in norpA P24 ;G␣q 1 were also greatly reduced compared with norpA P24 controls, with at most a few sporadic 1-2-pA bumps to saturating illumination (Fig. 4) and in 5 of 11 cells no detectable response at all. In norpA P24 ,rdgA 1 ;G␣q 1 , however, the rdgA background again resulted in a massive enhancement of bump amplitude and QE and large slowly terminating quantum bumps very similar to those seen in norpA P24 ,rdgA 1 could be elicited by bright flashes from the start of the recording in all cells (Figs. 4 and 5, n ϭ 20). Nevertheless, compared with norpA P24 ,rdgA 1 , sensitivity in norpA P24 ,rdgA 1 ;G␣q 1 was reduced ϳ100-fold. These findings show that both the spontaneous current and the light-induced current in norpA P24 and norpA P24 ,rdgA 1 are dependent on G q and hence likely to be mediated by its only known target in the photoreceptors, namely PLC. Although in principle one cannot exclude the possibility that G q might have alternative targets, the striking rescue of the light response in norpA P24 ,rdgA 1 and norpA P24 ,rdgA 1 ;G␣q 1 (compared with norpA P24 and norpA P24 ; G␣q 1 , respectively) provides strong independent evidence that the residual responses in norpA P24 and norpA P24 ;G␣q 1 are mediated by PLC, because the rdgA mutation is expected to result in a greater net production of one of the two products of PLC activity, namely DAG.
The question remains whether the PLC responsible for the spontaneous currents and responses in norpA P24 represents residual norpA P24 protein or an alternative PLC isoform. The Drosophila genome contains only two further identifiable PLC sequences: a PLC␥ encoded by the sl (small wing) gene (22) and PLC-21C, which is a second PLC␤ isoform (29). PLC␥ can be excluded, because spontaneous currents and residual light responses similar to those in norpA P24 could still be recorded in norpA P24 ,sl double mutants (data not shown). Because there are no mutants of PLC-21C, we attempted to test whether the residual norpA P24 protein might be responsible by generating flies with only one copy of norpA P24 by crossing norpA P24 to a lethal deficiency strain, Df(1)HC244, in which the entire norpA gene and the surrounding genomic region were deleted. The results, however, were inconclusive; the constitutive currents and quantum bump amplitudes in photoreceptors from norpA P24 /Df(1)HC244 (n ϭ 11) were not significantly different from homozygous norpA P24 controls, but QE did in fact appear to be reduced ϳ2-fold. Although the latter suggests that residual mutant norpA P24 protein is responsible, because the responses were minimal and near the limit of detectability, we treat this result with caution.
Activation of TRP Channels by Metabolic Inhibitors Is PLCdependent-Because any residual light sensitivity in norpA P24 still appears to be mediated by PLC, our findings only serve to strengthen the conclusion that Drosophila phototransduction is an obligatory PLC-based pathway. Nevertheless, numerous studies have assumed that norpA P24 is a null mutant, and we wondered whether the finding of significant PLC activity in norpA P24 had any implications for the conclusions of such studies. For example, we recently demonstrated that PUFAs such as arachidonic acid and LNA could activate the light-sensitive channels. Because LNA was still effective in norpA P24 , we concluded that activation was downstream of PLC and therefore likely to reflect a direct effect on the channels (6). Subsequently, Minke and colleagues (8) reported that the light-sensitive channels could be activated by mitochondrial inhibitors; since they too found these agents to be effective in norpA P24 , they also concluded that the agents acted downstream of PLC and that metabolic inhibition might affect the channels directly. These authors also noted that PUFAs can act as mitochondrial uncouplers and proposed that this was responsible for their ability to activate the light-sensitive channels. In view of the importance of these arguments for the mechanism of excitation and because of the significant PLC activity remaining in norpA P24 , we repeated and extended these experiments. We first confirmed that mitochondrial inhibitors could activate light-sensitive channels as reported previously (8). Indeed, large currents were reversibly induced within only seconds of application of either 10 M CCCP or 0.1 mM DNP in WT and norpA P24 in recordings made soon after establishing the wholecell configuration (Fig. 6A). Note that both CCCP and DNP induced a noise-free inward current that is probably because of their mode of action as electrogenic proton ionophores, but this current was clearly distinguishable from the larger and noisy currents representing activation of the light-sensitive channels. Because the DNP-induced proton currents were smaller (ϳ5-10 pA) than those induced by CCCP (ϳ50 pA), most experiments were performed using DNP. The light-sensitive channels were invariably activated by DNP or CCCP either with or without ATP in the electrode, although activation was quicker and not indefinitely reversible when no ATP was included. However, when DNP or CCCP was applied after the spontaneous current had first decayed in norpA P24 , we never detected any activation despite repeated and prolonged (Ͼ2 min) application (Fig. 6A). In norpA cells recorded with ATP and in cases where isolated spontaneous bumps were still occurring, DNP caused an increase in the size of these bumps. Similarly, light-induced bumps and macroscopic responses were greatly enhanced by DNP, as in recordings made in norpA,rdgA or in norpA without ATP in the electrode (Fig. 6B,  cf. Figs. 3 and 4).
In case this behavior was specific to norpA P24 , similar exper-iments were also performed in other severe norpA alleles, including norpA P12 , norpA P16 , and norpA EE5 . Again DNP and CCCP induced large inward currents superimposed on ongoing spontaneous or light-induced currents, but in no case (n ϭ 8) could any channels be activated after the spontaneous currents had decayed to base line, although again responses to light were greatly facilitated (data not shown). In case the failure to activate channels was related to the long recording time required for the spontaneous current to decay (e.g. because of washout of some essential factor), we also applied metabolic inhibitors to norpA P24 ;G␣q 1 (n ϭ 4) and norpA P24 ,rdgA 1 ;G␣q 1 (n ϭ 3), in which the lack of constitutive current allowed the inhibitors to be tested immediately after establishment of the whole-cell configuration. However, we were never able to detect any activation of the light-sensitive channels by CCCP or DNP at any stage of recording with or without ATP in the electrode, although again the size of the quantum bumps in norpA P24 ; G␣q 1 was greatly increased (Fig. 6C). These results indicate that the activation of light-sensitive channels by metabolic inhibition is strictly dependent on PLC activity and strongly suggest that it reflects enhancement of the spontaneous miniature quantum bumps induced by residual PLC activity, which underlie the spontaneous currents in norpA.
The most obvious interpretation of these results is that metabolic inhibition activates channels via the same mechanism by which light responses and bump amplitudes are enhanced by ATP depletion in norpA, namely the failure of DGK (see Fig. 8). If this is the case, then the mitochondrial inhibitors should be ineffective on rdgA backgrounds where DGK activity is already minimized. Indeed, in recordings from norpA P24 ,rdgA 1 double mutants, DNP failed to activate any inward current, whether applied during or after the decay of the spontaneous current (n ϭ 4, Fig. 6D). In fact, there was typically a slight inhibition of any ongoing spontaneous or light-induced currents. The cause for this was not investigated further, although possible factors might include pH changes or depletion of the GTP required for G-protein activation.
PUFAs Activate Channels Independently of PLC-We performed similar experiments using PUFAs, reasoning that if the ability of PUFAs to activate light-sensitive channels was because of the effect of these acids as mitochondrial uncouplers, then they should also be ineffective in norpA mutants after decay of the spontaneous current. To quantify the data, we applied a subsaturating dose (20 M) of LNA from a puffer pipette and determined the time taken to elicit a criterion response (10 pA) and also the peak current reached. In marked contrast to the effects of metabolic inhibitors, the potency of PUFAs appeared to be unaffected in norpA mutants. Currents with characteristic high frequency channel noise were elicited by the application of LNA in every photoreceptor tested, both in norpA P24 after the spontaneous currents had decayed and in norpA P24 ;G␣q 1 , which lacks any spontaneous currents (Fig. 7). To confirm that these currents represented the light-sensitive channels, in several cases we tested their current-voltage relationship using voltage ramps, which showed the characteristic outward rectification (Fig. 7F). Compared with WT, the potency of LNA in norpA P24 backgrounds was if anything slightly enhanced in terms of the time taken to elicit the criterion response, although the maximum current reached was significantly reduced, and the overall waveform of the currents showed a more gradual development without the accelerating overshooting phase typically observed in WT photoreceptors. This overshooting phase could represent positive feedback by Ca 2ϩ -dependent activation of PLC or possibly an additional effect of LNA as a mitochondrial inhibitor or inhibitor of DGK, but it was not explored further. For the present argument, the FIG. 6. Activation of channels by metabolic inhibitors is dependent on PLC activity. A, 0.1 mM DNP applied by puffer pipette (solid lines) rapidly activated an inward current of 200 -300 pA in both WT (left) and norpA P24 photoreceptors recorded soon after establishing the whole-cell configuration (i); in another norpA P24 cell after decay of the spontaneous current, DNP only amplified the residual quantum bumps (electrode solution containing nucleotide additives) (ii). The small (ϳ5-10 pA) noise-free inward currents are proton currents because of the H ϩ ionophore action of DNP. B, massive facilitation of light response in norpA P24 by DNP. Left trace, before application of DNP, a flash (ϳ10 6 photons at arrow) elicited a barely detectable response consisting of sporadic ϳ1-pA bumps. Middle trace, DNP perfusion some minutes later in the dark resulted only in an increase in the size of remaining bumps still being elicited by the earlier flash. Immediately afterward, however, the same light flash elicited a greatly facilitated response (right trace). C, DNP applied to norpA P24 ;G␣q 1 failed to activate any current apart from an ϳ10-pA ionophore current (left). Right, quantum bumps in norpA P24 ;G␣q 1 generated by an ϳ10 6 -photon flash were greatly enhanced by DNP (above, before DNP application; below, after DNP application). D, DNP applied during an ϳ150-pA spontaneous current in norpA P24 ,rdgA 1 and recorded shortly after establishment of the whole-cell configuration caused a slight suppression of the ongoing current (three superimposed traces). Similar results obtained in four cells (mean suppression, 46 Ϯ 18 pA, n ϭ 4). crucial finding is that the light-sensitive channels were activated by LNA in every norpA cell tested after the decay to base line, whereas DNP or CCCP invariably failed to activate any channels under the same conditions. This result indicates that PUFAs can activate the light-sensitive channels independently of any possible role as mitochondrial uncouplers.
The proposal that PUFAs directly activate the light-sensitive channels (6) was also questioned recently by Schilling and colleagues (20), who concluded that the activation of recombinant TRPL channels by PUFAs when expressed in Sf9 cells was at least in part because of activation of endogenous PLC in the membrane. Despite the residual PLC activity demonstrated in norpA P24 , the near elimination of NORPA protein in this mutant would make it unreasonable to suggest that activation of the PLC encoded by norpA could underlie activation by PUFAs in norpA P24 . However, if an alternative PLC isoform were responsible, in principle this could still be a valid explanation. Previous reports of PLC activation by PUFAs have specifically concerned PLC␥ isoforms (30,31). We therefore investigated two severe mutants (sl 1 and sl 2 ) of the only PLC␥ gene known in Drosophila. Despite a mild eye developmental phenotype (22), light responses in both sl 1 and sl 2 photoreceptors had essentially WT properties (data not shown), whereas LNA activated the light-sensitive channels with a potency indistinguishable from that in WT (Fig. 7). Similarly, channels in norpA P24 ,sl double mutants were also reliably activated by LNA after decay of the constitutive currents in all cells tested (n ϭ 4) with undiminished potency (Fig. 7). These results indicate that activation of the light-sensitive channels by LNA is not mediated indirectly by activation of PLC␥.
Finally, if activation of the PLC-21C (or any other undiscovered PLC isoform) were responsible for the effects of PUFAs, then we would predict that, like the spontaneous activity and residual response to light in norpA P24 , the effects of PUFAs should be greatly enhanced on an rdgA background or in recordings made without ATP. However, we found no significant difference in either latency or the magnitude of the currents activated by LNA in norpA P24 ,rdgA 1 or norpA P24 ,rdgA 1 ;G␣q 1 compared with norpA P24 or norpA P24 ;G␣q 1 controls or in norpA P24 recorded with or without ATP in the electrode (Fig. 7).
In conclusion, activation of the light-sensitive channels by PUFAs cannot readily be explained either by their action as mitochondrial uncouplers or by their reported ability to activate PLC in some systems. This suggests that, as originally proposed (6), PUFAs may act directly on the light-sensitive channels or their lipid environment.

DISCUSSION
In this study we found that the supposedly null PLC mutant, norpA P24 , has detectable spontaneous and light-activated currents. Although this might have indicated an alternative parallel transduction pathway, further experiments strongly supported the essential role of PLC. Thus, the residual responses still depend on the same PLC-specific G q protein and were massively enhanced in norpA,rdgA double mutants in which metabolism of one of the products of PLC activity, i.e. DAG, is blocked. This massive facilitation underscores other recent evidence indicating that DAG or a downstream metabolite is the essential messenger of excitation in Drosophila phototransduction (6,7,17).
There is still uncertainty, however, as to the identity of the PLC responsible for the responses in norpA P24 . The molecular lesion in norpA P24 , which has been identified as a 28-bp deletion resulting in a frameshift and a premature stop codon, eliminates the domains believed to be essential for G-protein interaction and renders protein levels undetectable on Western blots (14). Although one would thus expect any residual protein to be non-functional, the mutation leaves the catalytic site still intact so that the possibility of a low efficiency of activation cannot be completely excluded. For technical reasons, we were unable to convincingly demonstrate a further loss of sensitivity in norpA P24 /Df(1) flies with only one copy of the norpA P24 gene, and so the possibility also remains that an alternative PLC isoform mediates the residual responses. Of the two remaining PLC isoforms known in the Drosophila genome, we could rule out the PLC␥ encoded by the sl gene (22), leaving PLC-21C (29) as the only other obvious candidate.
Mechanism of Activation by Metabolic Stress-When nucleotide additives are omitted from the electrode solution, the light-sensitive channels in Drosophila photoreceptors open spontaneously, generating a so-called rundown current after several minutes of whole-cell recording (8,28). The underlying cause and its relevance to the physiological mechanism of activation had remained obscure, but the present results now provide an explanation. Although the ability of metabolic inhibitors to activate TRP and TRPL channels in norpA P24 was taken as evidence that the mechanism must be downstream of PLC (8), we found that metabolic inhibitors failed to activate FIG. 7. Activation of channels by linolenic acid. 20 M LNA (bars) applied to a WT photoreceptor (A) and a norpA P24 photoreceptor after complete decay of the spontaneous current (B) in both cases elicited a current with a characteristic high frequency channel noise. Although the final current in WT was larger, the response in norpA developed more quickly (see G and H for statistics). C-E, application of LNA elicited similar currents in norpA P24 ;G␣q 1 norpA P24 ,rdgA 1 and norpA P24 ,rdgA 1 ;G␣q 1 . F, current-voltage relationship of the LNA-induced current (in this case, from norpA P24 ,rdgA 1 ;G␣q 1 ) showed reversal potential and outward rectification characteristic of the light-sensitive channels (39). G and H, summary of peak amplitudes and delays (to a 10-pA criterion) of LNA-induced currents in WT, norpA P24 the double mutants norpA P24 ;G␣q 1 and norpA P24 ,rdgA 1 , the triple mutant norpA P24 ,rdgA 1 ;G␣q 1 (NR;G␣q), the phospholipase C␥ mutant (sl, data from sl 1 and sl 2 , pooled), and the norpA P24 ,sl 2 double mutant. LNA also elicited currents with similar potency after decay of the constitutive current in other norpA alleles including norpA P12 , norpA P16 , and norpA EE5 (data not shown). channels in the absence of spontaneous currents caused by ongoing residual PLC activity. Both here and elsewhere (17), we found that quantum bump amplitude was reduced in norpA mutants but could be greatly increased by depleting ATP. Hence, metabolic inhibitors can be expected to increase the amplitude of the spontaneous bumps responsible for the constitutive currents in norpA, giving the appearance of the activation of a large inward current. Because the rdgA mutation and ATP depletion have similar effects on bump amplitude, it seems likely that the mechanism of channel activation by metabolic inhibition in these cases is by the impairment of DGK (Fig. 8). Failure of DGK could have at least two consequences: 1) build-up of excess DAG and 2) impaired ability to resynthesize PIP 2 (Fig. 8). As argued elsewhere, excess DAG seems to be responsible for enhancing the bump amplitude in norpA (17). We therefore attribute the ability of metabolic inhibitors to activate an inward current in norpA mutants to the impairment of DGK and consequent amplification of the small bumps underlying the spontaneous current by increased levels of DAG. In WT flies, basal PLC activity is presumably sufficient to generate enough DAG to cause activation of the channels if DGK function is compromised, and indeed in the single rdgA mutant, we have found previously that the light-sensitive channels are always constitutively active (7).
Although impairment of DGK seems a sufficient explanation for activation of the channels under the conditions of these experiments, metabolic inhibition will have many other consequences that might also affect the phototransduction cascade. For example, in the long term PIP 2 is likely to be depleted, because apart from DGK there are at least three high-energy phosphate-dependent steps required for PIP 2 resynthesis (Fig.  8). Furthermore, rhodopsin, arrestin, the PDZ domain-scaffolding molecule INAD, and the TRP channels are all phosphoryl-ated during the light response (32)(33)(34)(35); although with the exception of arrestin (36), the functions of these phosphorylations remain obscure. We have in fact found one situation in which metabolic inhibitors appear to activate the light-sensitive channels independently of DAG. Thus, in mutants of the PIP 2 recycling pathway such as cds, microvillar PIP 2 can be completely and irreversibly depleted by prolonged illumination (37). Surprisingly, under these conditions of PIP 2 depletion metabolic inhibitors still reliably activated both TRP and TRPL channels, although there should no longer have been any substrate for DAG generation. 3 Recent evidence has suggested that under some conditions PIP 2 depletion may contribute to TRP or TRPL channel activation (20,37), and this result might be explained for example if the phosphorylation state of the channels or related proteins determines whether or not PIP 2 depletion can activate the channels.
Activation by PUFAs-In marked contrast to metabolic inhibitors, PUFAs such as LNA invariably activated the lightsensitive channels in all genetic backgrounds tested, regardless of the presence or absence of PLC and whether or not there was an ongoing spontaneous current. The effects of PUFAs are hence distinct from those of metabolic inhibitors and cannot be explained by their action as mitochondrial uncouplers as suggested previously (3,8). Neither can their action be readily accounted for by stimulation of PLC, as has been proposed also (20). Thus, the potency of LNA was undiminished in mutants of the only PLC␥ isoform in Drosophila genome and was also unaffected in norpA P24 ,rdgA 1 , although the rdgA mutation massively facilitates the action of the residual PLC activity in norpA P24 . The proposal that PUFAs act directly on the channels (6) or their lipid environment thus still seems the most likely explanation for their effect. Whether PUFAs are the endogenous excitatory messengers, presumably released from DAG by DAG lipase (Fig. 8), or whether they mimic the effect of DAG, for example, remains to be determined. Key questions that need to be answered in this respect include: whether a DAG lipase (which has yet to be cloned in any eukaryote) is expressed in Drosophila photoreceptors; if so, whether this lipase is required for phototransduction; and whether there are lipid binding domains on the TRP and TRPL channels or associated proteins.
Concluding Remarks-In summary, although we have shown that even a supposedly null norpA mutant can respond to light, our results only strengthen the conclusion that activation of PLC is absolutely required for phototransduction in Drosophila. As also discussed elsewhere (17), the striking rescue of the light response in norpA by the rdgA mutation highlights the essential role of DGK in regulating the supply of the excitatory messenger and provides compelling support for the proposal that DAG rather than inositol 1,4,5-trisphosphate is the critical product of PIP 2 hydrolysis by PLC in Drosophila phototransduction. Finally, we propose that this critical role of DGK means that inhibition of this enzyme is the primary although not necessarily exclusive mechanism by which metabolic inhibition can activate the light-sensitive TRP and TRPL channels. Given the widespread distribution of TRP channels in mammalian tissues and the emerging view that many of these may also be regulated by lipid messengers such as DAG and arachidonic acid (10,38), it will be interesting to see whether this proves to be a more general mechanism of hypoxia-induced Ca 2ϩ influx and resultant cytotoxicity.