In Vivo Light-induced and Basal Phospholipase C Activity in Drosophila Photoreceptors Measured with Genetically Targeted Phosphatidylinositol 4,5-Bisphosphate-sensitive Ion Channels (Kir2.1)* □ S

The phosphatidylinositol 4,5-bisphosphate (PIP 2 )-sen- sitive inward rectifier channel Kir2.1 was expressed in Drosophila photoreceptors and used to monitor in vivo PIP 2 levels. Since the wild-type (WT) Kir2.1 channel ap- peared to be saturated by the prevailing PIP 2 concen-tration, we made a single amino acid substitution (R228Q), which reduced the effective affinity for PIP 2 and yielded channels generating currents proportional to the PIP 2 levels relevant for phototransduction. To isolate Kir2.1 currents, recordings were made from mutants lacking both classes of light-sensitive transient receptor potential channels (TRP and TRPL). Light resulted in the effective depletion of PIP 2 by phospho- lipase C (PLC) in approximately three or four microvilli per absorbed photon at rates exceeding (cid:1) 150% of total microvillar phosphoinositides per second. PIP 2 was re- synthesized with a half-time of (cid:1) 50 s. When PIP 2 resynthesis was prevented by depriving the cell of ATP, the Kir current spontaneously decayed at maximal rates representing a loss of (cid:1) 40% loss of total PIP 2 per minute. at a maximum rate of (cid:1) 5%/min. In a trpl cell recorded with no nucleotide additives ( (cid:3) ATP ) the Kir2.1 R228Q current rapidly decayed to baseline within (cid:1) 6–7 min (the large initial run-up indicates that this cell was exposed to more red light than usual during electrode approach and is presumably mediated by endogenous ATP reserves). D , in the PLC mutant, norpA , the spontaneous decay is delayed and slowed (cid:1) 4-fold; in G (cid:2) q 1 and rdgA 1 mutants decay was as rapid as WT or trpl controls. E , without added nucleotides ( (cid:3) ATP ) a brief flash delivered after (cid:1) 4 min irreversibly depleted all remaining PIP 2 -sensitive current ( trpl ); but with nucleotide additives PIP 2 was still rapidly resynthesized following a similar flash delivered after more than 10 min (Kir2.1 WT channel). F , summarizes the maximum rate of PIP 2 depletion (steepest slope of decay normalized to maximum Kir current) in a variety of mutant and experimental backgrounds ( n (cid:9) 6 or more cells) recorded using Kir2.1 R228Q channels in the absence of nucleotide additives ( (cid:4) ATP is WT control with nucleotide additives; DNP , continuous application of 100 (cid:3) M dinitrophenol).

tion of two classes of light-sensitive channels TRP and TRPL. These are the prototypical members of the large and diverse family of "transient receptor potential" (TRP) channels many of which, including all of the most closely related "canonical" TRPC subfamily, are also regulated by PLC (reviewed in Refs. [1][2][3]. Although in most cases the exact mechanism of TRP channel gating remains unresolved, mounting evidence, both in Drosophila and in at least a subset of mammalian TRP homologues, now suggests that diacylglycerol (DAG) rather than inositol 1,4,5-trisphophate (InsP 3 ) is the critical product of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) hydrolysis (4,5), whereas some evidence suggests that the reduction in PIP 2 itself may also be important (3,6,7). Because of its central role in phototransduction and PLC signaling generally, it is important to understand the dynamics of PIP 2 turnover. In recent years a number of attempts have been made to monitor PIP 2 levels in vivo, many making use of the GFP-tagged PIP 2 binding PH domain from PLC␦ (8). While this has provided valuable insight into dynamic and spatial aspects of PIP 2 mobilization, the PLC␦ PH domain also binds InsP 3 with high affinity, and it is not always clear whether PIP 2 or InsP 3 levels are being monitored (Ref. 9, but see also Ref. 10). Fluorescence-based technologies are also of restricted value in photoreceptors, because the excitation light represents a saturating and usually damaging stimulus for the cell. Therefore we adopted an alternative approach by using an electrophysiological biosensor in the guise of the well characterized PIP 2 -sensitive ion channel, Kir2.1 (7). Like all members of the inwardly rectifying Kir family, these channels require phosphoinositide binding for their activity, but Kir2.1 has the highest specificity for PIP 2 showing, for example, no detectable activation by PI or phosphatidylinositol 3,4-bisphosphate and Ͻ5% activation by phosphatidylinositol 4-phosphate (PIP) or phosphatidylinositol 3,4,5-trisphosphate (11,12). In an initial study we expressed eGFP-tagged Kir2.1 channels in Drosophila photoreceptors under the control of the rhodopsin promoter and found them to be specifically targeted to the light-transducing microvillar membrane. In whole cell recordings from dissociated cells, the channels generated large constitutive currents, which could be rapidly and reversibly suppressed by light, representing PIP 2 hydrolysis by PLC and its subsequent resynthesis (7).
In the present study we have developed this technology further by expressing a mutant version of the Kir2.1 channel with reduced affinity for PIP 2 with a dynamic range that more effectively covers the range of physiologically relevant PIP 2 levels. We expressed the channels in several mutant backgrounds, including flies lacking the light-sensitive TRP and TRPL channels so that Kir2.1 currents could be recorded in isolation. Our results indicate that activated PLC can deplete PIP 2 at rates well in excess of 100% s Ϫ1 . In addition our data provide in vivo measurements of basal rates of PLC activity that, although they are orders of magnitude less than that of activated PLC, can still deplete all detectable PIP 2 within minutes if PIP 2 resynthesis is blocked. As well as providing unique quantitative information on the in vivo activity of PLC, the results also shed light on some recent studies concerning the mechanism of phototransduction.

EXPERIMENTAL PROCEDURES
Flies-Drosophila melanogaster were raised in the dark at 25°C. The eGFP-KiR2.1 R228Q construct was generated from the wild-type (WT) eGFP-Kir2.1 construct previously described (7,13) using the QuikChange mutagenesis system (Stratagene, La Jolla, CA) according to the manufacturer's instructions. Following mutagenesis, the presence of the R228Q mutation and no other was confirmed by sequencing of the full gene. The eGFP-KiR2.1 R228Q gene was then subcloned into pUAST and injected into yw embryos to obtain transformants as previously described (7,13). Expression was driven by the Gal4-UAS system (14) (15); trpl 302 , null mutant of the second class of light-sensitive channel (15); rdgA 1 , the most severe allele of the photoreceptor DAG kinase (16); norpA P24 , a null or near-null allele of the photoreceptor phospholipase C (17); and G␣q 1 , a severe allele of the photoreceptor G q protein with Ͻ1% protein (18).
Whole Cell Recordings-Dissociated ommatidia were prepared as previously described from recently eclosed adult flies (7,19) and transferred to the bottom of a recording chamber on an inverted Nikon Diaphot microscope. Unless otherwise stated, the bath was composed of (in mM): 110 NaCl, 10 KCl, 4 CsCl, 10 TES, 4 MgCl 2 , 1.5 CaCl 2 , 25 proline, and 5 alanine. The standard intracellular solution was (in mM): 140 potassium gluconate, 10 TES, 4 Mg-ATP, 2 MgCl 2 , 1 NAD, and 0.4 Na-GTP. In some experiments nucleotide additives (ATP, GTP, and NAD) were omitted and replaced with sucrose to maintain osmolarity. The pH of all solutions was 7.15. Whole cell voltage clamp recordings were made using electrodes of ϳ10 -15 M⍀ resistance, and series resistance values were generally below 25 M⍀ and were routinely compensated to Ͼ80%. Data were collected and analyzed using Axopatch 1-D or 200 amplifiers and pCLAMP 8 or 9 software (Axon Instruments, Foster City, CA). Cells were stimulated via one of two green lightemitting diodes, with maximum effective intensity of ϳ2 ϫ 10 5 and 8 ϫ 10 7 photons s Ϫ1 per photoreceptor, respectively. Relative intensities were calibrated using a linear photodiode and converted to absolute intensities in terms of effectively absorbed photons by counting quantum bumps at low intensities in WT flies (e.g. Ref. 20).
With normal bath solutions, the constitutive currents generated by Kir2.1 channels were often very large (Ͼ5 nA), potentially leading to unacceptable series resistance errors. We therefore reduced the current amplitudes by recording in the presence of 4 mM Cs ϩ , which partially blocks Kir2.1 channels. Furthermore, the block is voltage-dependent such that the current voltage (I-V) relationship reaches a maximum inward current at voltage of ϳϪ84 mV (Fig. 1). The current at this unique voltage can always be found by applying voltage ramps (typically from Ϫ100 to Ϫ40 mV) and determining the maximum current level, which will always occur at the same absolute voltage irrespective of any series resistance error. More usually, cells were unambiguously clamped at Ϫ84 mV using a manual offset voltage adjustment to find the maximum current, which was regularly checked throughout the recording.
Estimating Resting PIP 2 Levels-To compare PIP 2 levels in different genetic backgrounds, dark-adapted Kir current levels were normalized to the background-subtracted GFP fluorescence measured from at least 10 randomly selected ommatidia from the same preparations using a microfluorimetry system (Photon Technology International) incorporat-ing a photomultiplier, which measured the fluorescence above 510 nm induced by 485-nm excitation from a 75-watt xenon arc lamp.

RESULTS
The Dynamic Range of Kir2.1 R228Q Matches Endogenous PIP 2 Levels-To track PIP 2 levels in vivo we previously expressed the wild-type eGFP-tagged Kir2.1 channel (Kir2.1 WT ) in Drosophila photoreceptors under control of the rhodopsin (Rh1) promoter. The channels localize almost exclusively to the light-transducing microvilli, where they generate large constitutive inwardly rectifying currents, which can be suppressed by light as PIP 2 is hydrolyzed by PLC, and recover in the dark as PIP 2 is resynthesized. Most notably, when Ca 2ϩ influx via TRP channels was prevented by blocking them with La 3ϩ , modest stimulation by light resulted in near complete suppression of Kir currents, indicating essentially total depletion of PIP 2 as well as PI and PIP. This suggested that Ca 2ϩ influx via TRP channels is normally required to inhibit PLC and/or facilitate PIP 2 resynthesis (7). We concluded that the resulting lightinduced depletion of PIP 2 in the trp mutant was the underlying cause of the long-debated trp ("transient receptor potential") phenotype (21), whereby the response to maintained light in trp mutants decays to baseline, and thereafter the photorecep- . I-V curves determined using voltage ramps from Ϫ110 to Ϫ50 mV (duration, 2 s; variable interval between traces) in the presence of 10 mM K ϩ and 4 mM Cs ϩ in the bath, and 140 mM K ϩ in the pipette. Repeated ramps were delivered after varying stages of depletion; in the last trace (4) of the Kir2.1 R228Q family, all detectable PIP 2 has been depleted. Because of the (effectively instantaneous) voltage-dependent block by Cs ϩ , a maximum current was reached at Ϫ84 mV. Trace B3 is replotted on an enlarged scale (dotted line) to show that the apparent voltage required to reach the maximum current is already shifted by 3 or 4 mV in the first and largest current trace (1) due to a small series resistance error. By repeating ramps or by manually adjusting the voltage to find the maximum inward current, an accurate holding potential of Ϫ84 mV can always be found and maintained irrespective of series resistance error.
tors remain insensitive to further stimulation, recovering their sensitivity slowly in the dark as the PIP 2 is resynthesized.
The loss of PIP 2 inferred from the suppression of the Kir current correlated in general with the loss of sensitivity of the light-induced current (LIC); however, there were significant quantitative discrepancies. In particular, suppression of the Kir current (representing PIP 2 loss) required higher light intensities than the suppression of the LIC, whereas recovery of the Kir current was faster than the recovery of the light response, although both presumably reflect resynthesis of PIP 2 as recovery of both was blocked in mutants of the PIP 2 recycling pathway (7). We attributed these differences to the known very high affinity of the Kir2.1 channel for PIP 2 (22). To confirm this and in an attempt to generate a more accurate probe for PIP 2 , we generated a point mutation (R228Q) in Kir2.1, which had previously been reported to substantially reduce the effective affinity of the channel for PIP 2 (22). We first confirmed a reduced effective affinity of ϳ4-fold by expressing the channels in Drosophila S2 cells and measuring the dose response function to exogenously applied PIP 2 in inside-out patches (Supplemental Fig. S1). We then expressed the channels in the photoreceptors and found that, like Kir2.1 WT , the channels were also almost exclusively targeted to the microvillar membrane and mediated large inwardly rectifying K ϩ currents. After normalization for expression levels, the Kir currents in cells expressing Kir2.1 R228Q channels were only 36% (Ϯ4% S.E., n ϭ 27) of those expressing Kir2.1 WT , suggesting that the endogenous PIP 2 levels in the dark were only sufficient to partially activate Kir2.1 R228Q . As will be described in more detail elsewhere, 2 this was confirmed by showing that Kir2.1 R228Q channel activity in the photoreceptors could be further increased at least 2-fold by application of exogenous PIP 2 while Kir2.1 WT channels seemed to be saturated (i.e. additional PIP 2 did not significantly increase currents). Furthermore, like the WT channel (23), Kir2.1 R228Q channels have a PIP 2 dose response function with a Hill coefficient close to 1.0 (Supplemental Fig. S1); together with the low (Ͻ50%) level of activation, this means that the current should be approximately directly proportional to physiological PIP 2 levels.
We quantitatively compared the ability of calibrated light stimuli to suppress Kir currents on the one hand and to inactivate the LIC on the other. With Kir2.1 WT channels, increasing intensities of light, delivered to cells in the presence of La 3ϩ to block TRP channels, progressively suppressed both Kir current and the LIC; however, ϳ9ϫ higher intensities were required to suppress the Kir current (Fig. 2). By contrast, in flies expressing Kir2.1 R228Q , suppression of sensitivity and Kir current were much more closely matched with only a 2.5-fold shift. A very similar behavior (2-fold shift) was also seen in trp mutants expressing Kir2.1 R228Q and recorded in the absence of La 3ϩ .
We also compared the time courses of the recovery of Kir current (representing PIP 2 resynthesis) and the LIC following stimuli inducing substantial PIP 2 depletion (Fig. 3). Currents mediated by Kir2.1 WT channels recovered relatively rapidly (t1 ⁄2 ϳ 30 s), but significant sensitivity to light only began to be restored after the Kir current had almost fully returned to pre-stimulation levels. By contrast, using Kir2.1 R228Q the LIC and Kir current recovered over a similar time course (t1 ⁄2 ϳ 50 s). We also discovered that PIP 2 synthesis could be routinely monitored at the start of any recording made in the presence of La 3ϩ or the absence of Ca 2ϩ , because without Ca 2ϩ influx via the TRP channels even the red light used for observation during electrode approach (equivalent to ϳ200 -400 effectively absorbed photons per second) was sufficient to significantly deplete PIP 2 . Consequently sensitivity was often very low im-mediately after establishing the whole cell configuration (break-in) but then increased over a period of 2-3 min in the dark. Similarly, the Kir current was typically relatively small at first but then increased over a similar time course, starting as soon as (but not before) the red light was turned off. In cells expressing Kir2.1 WT channels there was only a modest (28 Ϯ 7%, n ϭ 7; mean Ϯ S.E.) increase in Kir current, which was not well correlated with the increase in sensitivity to light. However, over the same experimental time period in cells expressing Kir2.1 R228Q the current typically increased over 3-fold (369 Ϯ 38%, n ϭ 34) over the first ϳ3 min in the dark and, when measured, closely paralleled a similar increase in the sensitivity to light (Fig. 3C). A similar "run-up" behavior was first described over ten years ago in recordings made in Ca 2ϩ free Ringer (24,25), but at that time it was speculated that it might represent refilling of intracellular Ca 2ϩ stores.
In summary, the activity of Kir2.1 R228Q channels shows a close quantitative correlation with the sensitivity to light over a range of PIP 2 levels, and the channels' operating range thus appears to effectively cover the PIP 2 levels relevant to phototransduction. Although Kir2.1 WT channels are more sensitive to low levels of PIP 2 , e.g. during the initial period of PIP 2 resynthesis following depletion, they seem to be largely saturated by the dark-adapted resting level of PIP 2 . We also note that the large differences in sensitivity to suppression by light and the time course of the subsequent recovery, following a point mutation in Kir2.1 known to reduce its effective affinity for PIP 2 , represent strong additional confirmation that Kir2.1 channels are directly monitoring PIP 2 levels under these conditions.
Light-induced PLC Activity-We previously found that, under control "physiological" conditions with both TRP and TRPL function intact, even very bright stimuli only suppressed the Kir2.1 WT current by ϳ20%. It could also not be excluded that even this modest suppression might have represented modulation by the large light-induced ion fluxes (which include Na ϩ , K ϩ , Ca 2ϩ , and Mg 2ϩ ) rather than PIP 2 depletion. We therefore repeated these experiments using flies expressing Kir2.1 R228Q , reasoning that if PIP 2 levels were in fact reduced under these conditions, there should now be substantially more suppression of the Kir current. Indeed, the Kir2.1 R228Q channels were significantly more sensitive to suppression by light; nevertheless, even the brightest stimuli tested (ϳ10 6 photons, roughly equivalent to full daylight intensities) failed to suppress more than ϳ40% of the current (Fig. 4). The currents then recovered with the typical time course seen in the presence of La 3ϩ (t1 ⁄2 ϳ 50 s: Figs. 3D and 4B). These results indicate that there is indeed a significant, though not debilitating loss of PIP 2 during light adaptation under physiological conditions.
In experiments described thus far, the Kir current was simultaneously recorded with the LIC. Although this approximates physiological conditions, it compromises accurate measurement of the Kir current, because it cannot be cleanly separated from the LIC and because the Kir current may be indirectly influenced by ion fluxes associated with the LIC. To isolate the Kir current we expressed and recorded from Kir2.1 channels in trp;trpl double mutants or in trpl mutants in the presence of La 3ϩ . In either case this results in total elimination of all native light-sensitive currents (15,26), and the electrophysiological response to light now consisted solely of a suppression of the constitutive Kir current uncontaminated by any other conductance. Suppression and recovery of the Kir current should thus now provide basic quantitative information on rates of PIP 2 hydrolysis and synthesis in the absence of feedback by Ca 2ϩ . Both the overall suppression and the rate of suppression increased with brief flashes of increasing intensity (Fig. 5). In terms of the overall suppression reached, the data could be reasonably well fitted by a simple equation that as- where p is the number of effectively absorbed photons, m is the total number of microvilli (estimated at 45,000), and I/I max is the normalized residual current after suppression. Using Kir2.1 R228Q channels, recorded under control conditions, the data were fitted assuming ϳ3-4 microvilli were depleted of PIP 2 (and PI and PIP) per effectively absorbed photon (Fig.  5D). Using the higher affinity Kir2.1 WT channel, a value of only ϳ0.6 microvilli per photon was estimated. The suppression of both Kir2.1 WT and Kir2.1 R228Q currents in response to brief flashes showed complex kinetics. At relatively low intensities, suppression was rather slow (several seconds) and approximately monoexponential, but with brighter flashes a biphasic time course became increasingly apparent with an initial rapid suppression of maximally ϳ70% of total current, lasting only ϳ500 ms at the brightest intensities, and a slower suppression that continued for ϳ20 s (Fig.  5C). The maximum slope of the overall response reached rates well in excess of 100% total current s Ϫ1 (150 Ϯ 7% s Ϫ1 , n ϭ 9) before saturating at intensities roughly equivalent to one absorbed photon per microvillus (Fig. 5E). These kinetics are probably limited by the Kir2.1 channel response to PIP 2 and its gating kinetics, which are relatively slow (e.g. mean open time ϳ200 -300 ms (27,28)). Consequently, even the rapid phase can only be taken as a minimum estimate of the rate of PLC activity. In particular, it should be emphasized that it is very unlikely that the slow response with dimmer flashes, or the slow component of fast flashes, are reflected in similarly slow kinetics of PLC activity and PIP 2 depletion, which probably occur on a sub-second time scale. A direct indication of this can be seen in traces such as Fig. 3B where the LIC typically starts to recover (presumably representing PIP 2 resynthesis) while the Kir current is still decreasing.
Basal PLC Activity-In vitro basal activity of many PLC isoforms, including the Drosophila NORPA PLC␤4 (29), has often been reported in biochemical experiments; however, to our knowledge basal PLC activity has not previously been monitored in real-time in living cells. We reasoned that by tracking PIP 2 levels in the dark under conditions where PIP 2 synthesis was blocked, it would be possible to detect and measure the unstimulated basal rate of PLC activity in vivo. We first established that PIP 2 levels, as monitored by Kir currents, remained stable under control conditions in the dark. Indeed, currents mediated by both Kir2.1 WT and Kir2.1 R228Q channels remained stable for considerable periods in photoreceptors recorded with the standard nucleotide enriched pipette solution. Kir2.1 WT currents remained stable indefinitely (Ͼ15 min), with at most a very gradual loss of current amounting to less than 1% maximum current/min. After the initial run-up (see above), currents mediated by Kir2.1 R228Q channels did eventually decay, but only very slowly (maximum rate of ϳ5%/min; Fig. 6). This apparent slow loss of PIP 2 is presumably non-physiological and may reflect "wash-out" of some factor required for resynthesis. This is suggested by the observation that PIP 2 resynthesis rates following PIP 2 depleting flashes almost invariably slowed down with repeated flashes or prolonged recording times (e.g. Fig. 5A).
Under control conditions, therefore, it appears that any loss of PIP 2 by basal PLC activity must be largely replaced by resynthesis of PIP 2 and to detect basal PLC activity, it is first necessary to prevent this resynthesis. We found that this could terms of effectively absorbed photons/microvillus) elicited rapid suppression followed by slow recovery; sensitivity of the high affinity Kir2.1 WT channel was less than for Kir2.1 R228Q , but recovery was faster. Maximum currents normalized to Ϫ1.0 (i.e. 0.25 on the scale bar represents 25% suppression), dotted lines represent zero current (total suppression). C, on a faster timescale two kinetic components of suppression can be recognized (intensity 1.3 photons/microvillus). D, intensity (photons/microvillus) plotted against normalized suppression; the data were fitted with Equation 1 assuming that 3.4 microvilli are completely depleted of PIP 2 (and PIP and PI) per absorbed photon for Kir2.1 R228Q but only 0.56 for Kir2.1 WT . E, the maximum slope of suppression expressed as percent total current/s shows that PI is depleted at rates in excess of 100%/s with stimuli representing one or more effective photons absorbed per microvillus (same symbols as in D).
be simply achieved without significantly affecting PLC activity by omitting ATP and other nucleotide additives from the electrode. Because the TRP channels activate spontaneously after several minutes of whole cell recordings under these conditions, generating a so-called rundown current (30,31), measurements were made in trp mutants or in the presence of La 3ϩ to block TRP channels. We also made measurements in trpl mutants in the presence of La 3ϩ where no light-sensitive channel activity is possible at all. Under these conditions, the Kir current still initially increased for 2-3 min after turning off the red light, which we interpret as PIP 2 resynthesis mediated by the cells endogenous ATP reserves. However, soon afterward, the current decayed spontaneously, usually reaching baseline after a further 3-4 min (Fig. 6). As independent confirmation that this decay represented loss of PIP 2 , in several cases in WT photoreceptors exposed to La 3ϩ (n ϭ 6) we simultaneously monitored the response to brief test flashes and found that the decay was invariably paralleled by a similar loss in sensitivity responses are shown in B. C, in trpl controls recorded with nucleotide additives (ϩATP) the Kir2.1 WT channels show a gradual ϳ40% run-up but no sign of spontaneous loss over 15 min. Kir2.1 R228Q channels also run up, but then decay slowly at a maximum rate of ϳ5%/min. In a trpl cell recorded with no nucleotide additives (ϪATP) the Kir2.1 R228Q current rapidly decayed to baseline within ϳ6 -7 min (the large initial run-up indicates that this cell was exposed to more red light than usual during electrode approach and is presumably mediated by endogenous ATP reserves). D, in the PLC mutant, norpA, the spontaneous decay is delayed and slowed ϳ4-fold; in G␣q 1 and rdgA 1 mutants decay was as rapid as WT or trpl controls. E, without added nucleotides (ϪATP) a brief flash delivered after ϳ4 min irreversibly depleted all remaining PIP 2 -sensitive current (trpl); but with nucleotide additives PIP 2 was still rapidly resynthesized following a similar flash delivered after more than 10 min (Kir2.1 WT channel). F, summarizes the maximum rate of PIP 2 depletion (steepest slope of decay normalized to maximum Kir current) in a variety of mutant and experimental backgrounds (n ϭ 6 or more cells) recorded using Kir2.1 R228Q channels in the absence of nucleotide additives (ϩATP is WT control with nucleotide additives; DNP, continuous application of 100 M dinitrophenol).
to light (Fig. 6). Because we presume that the initial phase of the decay represents a period where basal PIP 2 loss begins to outweigh the capacity to resynthesize (as the ATP reserves of the cell run out), the maximum slope of these curves, normalized to the maximal current, was taken as a conservative estimate of the rate of basal PIP 2 depletion. In WT flies expressing Kir2.1 R228Q , maximum rates represented ϳ40 -50% of the maximum PIP 2 pool per minute with 50% loss occurring on average ϳ4 -5 min (270 Ϯ 34 s n ϭ 8) after establishing the whole cell configuration. The maximum rates were similar when measured using Kir 2.1 WT ; however, the time taken before the suppression began to develop was considerably longer, with 50% loss occurring only after 8 -10 min (510 Ϯ 76 s, n ϭ 5). This very significant additional delay of ϳ4 min is consistent with our conclusion that the Kir2.1 WT channels are saturated with respect to the resting PIP 2 levels and only begin to report the falling PIP 2 levels once they have already been substantially reduced.
To confirm that omitting ATP from the electrode did indeed prevent the cell from resynthesizing PIP 2 , in several cases we delivered a stimulus sufficient to deplete substantial PIP 2 . With ATP in the electrode, PIP 2 was regenerated as usual; however, without ATP in the electrode, in most cells little or no recovery was observed over a time course of several minutes (Fig. 6E). By contrast the light-induced activity of PLC in cells recorded without ATP (as measured by the rate and extent of Kir current suppression by light) was similar to controls with ATP in the electrode. We also attempted to accelerate the loss of ATP by applying the mitochondrial inhibitor dinitrophenol, which should rapidly reduce any residual ATP levels; however, this appeared to make negligible difference to the maximum rates indicating that omitting ATP from the electrode alone was sufficient to effectively deplete the cell of ATP. Initially we also tried to prevent PIP 2 resynthesis using wortmannin, reported to block PI 4-kinase at concentrations above 10 M (32, 33). However, PIP 2 resynthesis following depleting stimuli appeared unaffected even by 50 M wortmannin applied over several minutes (recovery t1 ⁄2 ϭ 51 Ϯ 5 s, n ϭ 6) suggesting that the PI kinases in the microvilli are insensitive to wortmannin.
Spontaneous Rundown Is PLC-dependent-In principle the spontaneous loss of PIP 2 could be mediated by basal PLC activity or, e.g. basal activity of lipid phosphatases. To distinguish these we expressed Kir2.1 channels in norpA P24 , a severe PLC mutant with no detectable biochemical activity (17). In norpA flies expressing Kir2.1 WT , and recorded without ATP in the electrode, Kir currents were stable for at least 15 min of recording (n ϭ 6); in norpA photoreceptors expressing Kir2.1 R228Q a small residual decay (9.5 Ϯ 2.2% min Ϫ1 n ϭ 6) could be detected after a delay of ϳ5 min (Fig. 6D). This 4-fold reduction compared with controls suggests that most of the PIP 2 loss in WT or trpl backgrounds can be attributed to PLC activity. The significant difference in behavior between Kir2.1 WT and Kir2.1 R228Q emphasizes that even this residual Kir current rundown represents PIP 2 depletion, in this case probably mediated by lipid phosphatases and/or alternative PLC isoforms.
We also tried to confirm that the spontaneous PIP 2 loss was due to PLC by applying the PLC inhibitor U73122. Surprisingly, however, rather than preventing the spontaneous decay, U73122 (6 M), but not the control compound U73443, caused a rapid suppression of Kir currents even in the presence of nucleotide additives. This action was independent of PLC, because it was equally pronounced in norpA mutants and appears to represent a direct and novel action of U73122 as a Kir2.1 inhibitor (Supplemental Fig. 2). From dose response data based on the rate of suppression the IC 50 was estimated at ϳ1.5 M for Kir2.1 R228Q and ϳ10 -15 M for Kir2.1 WT . The increased sensitivity in Kir2.1 R228Q may indicate that U73122 acts by inhibiting the PIP 2 channel interaction as suggested for a similar inhibitory action of U73122 on another member of the Kir family (the acetylcholine-activated K channel or GIRK (34,35).
Basal activity of PLC could be due to spontaneous activity of PLC itself or spontaneous activation of G protein. However, we found that the rate of spontaneous PIP 2 loss in a mutant of the G q protein ␣ subunit, G␣q 1 , was indistinguishable from that measured in WT or trpl backgrounds (Fig. 6D) suggesting that the basal PLC activity is a property of the PLC molecule itself. Not surprisingly, the ability of light to suppress Kir currents was essentially eliminated (tested with flashes containing up to ϳ2 ϫ 10 5 effective photons) in both G␣q and norpA mutants, representing a genetic demonstration that suppression of Kir2.1 currents by light is G protein-and PLC-dependent.
Basal PLC Activity in rdgA Mutants-The primary genetic evidence for a role of DAG as a messenger of excitation in Drosophila photoreceptors comes from studies of mutants of the retinal degeneration A (rdgA) gene, which encodes DAG kinase. TRP channels are constitutively active in rdgA mutants, which in principle would be consistent with the suggestion that DAG is an excitatory messenger, because DAG might be expected to accumulate in the absence of metabolism by DAG kinase (36). This presupposes, however, that there is sufficient PIP 2 to act as substrate for PLC and that there is still significant PLC with basal activity. Neither can be taken for granted, because conversion of DAG to phosphatidic acid by DAG kinase is also the first step in PIP 2 resynthesis, whereas the severe degeneration seen in rdgA results in virtual elimination of the transduction compartment, i.e. the microvilli. We therefore expressed both WT and R228Q Kir2.1 channels in the rdgA 1 mutant. Despite the severe retinal degeneration, the channels were still successfully expressed and targeted to the residual rhabdomeres as judged by GFP fluorescence. After blocking the constitutive TRP channel activity with La 3ϩ , substantial Kir currents remained indicating that PIP 2 was still present, but absolute PIP 2 levels appeared to be reduced as GFP-normalized Kir currents in rdgA cells expressing Kir2.1 R228Q were ϳ5-10 ϫ lower than in WT controls (7.8 Ϯ 2.4-fold, n ϭ 7 cells from three flies). Importantly, in cells recorded without nucleotide additives, the Kir currents, whether Kir2.1 WT or Kir2.1 R228Q , decayed spontaneously with a time course at least as fast as that seen in WT flies (Fig. 6D). To confirm that this decay was PLC-dependent, we also expressed Kir2.1 WT channels in a norpA,rdgA double mutant and observed no decay of the Kir current even after Ͼ10 min recording time without nucleotide additives (n ϭ 4, data not shown).

DISCUSSION
The present results build on a previous study, which introduced PIP 2 -sensitive Kir2.1 channels as electrophysiological biosensors for PIP 2 (7). We first demonstrated that a point mutation previously shown to reduce the effective affinity of Kir2.1 for PIP 2 generates channels with a dynamic range matched to the PIP 2 levels relevant for phototransduction and then used these to derive in vivo estimates of both light induced and basal PLC activity as well as PIP 2 resynthesis In the following we discuss the suitability of Kir2.1 WT and Kir2.1 R228Q as PIP 2 biosensors and explore the significance of the results for our current understanding of phototransduction.
Kir2.1 R228Q Is a More Appropriate PIP 2 Sensor-The Kir2.1 WT channel proved very informative as a PIP 2 biosensor (7); however, several lines of evidence indicated that the Kir2.1 channel is more or less saturated by the prevailing darkadapted PIP 2 levels in Drosophila photoreceptors, prompting us to investigate a mutant Kir2.1 channel with reduced affinity for PIP 2 . Extensive studies from several laboratories have described a number of point mutations that reduce the effective affinity of Kir2.1 for PIP 2 . From these we chose the R228Q mutation (22) and confirmed that the effective affinity was reduced ϳ4-fold (Supplemental Fig. S1). Because the Kir2.1 R228Q channel was less than 50% saturated by the prevailing dark-adapted PIP 2 levels and was found to have a Hill coefficient of close to 1.0 (Supplemental Fig. S1), this means that in the steady state, the Kir2.1 R228Q current should be approximately proportional to PIP 2 over the entire dynamic range and in principle can be further corrected by transforming the data via the Hill equation. Importantly, the close quantitative correlation between sensitivity to light and currents mediated by Kir2.1 R228Q (e.g. Figs. 3 and 6) also implies that, under certain conditions, such as in the trp mutant or in recordings made in Ca 2ϩ -free solutions, the sensitivity to light itself can also be interpreted as an approximately linear indicator of PIP 2 levels.
Rate of PIP 2 Hydrolysis by Light-activated PLC-In the present study, not only did we use a probe with an approximately linear response to physiological levels of PIP 2 , but we also expressed the channels on genetic/pharmacological backgrounds that eliminated TRP and TRPL channel activity allowing us to record the Kir current in isolation. In principle this should provide a measure of the dynamic changes in PIP 2 during illumination; however, the kinetics of rapid changes in PIP 2 following stimulation by light are almost certainly compromised by the kinetics of the Kir2.1 channel. This has a mean open time reported to be between 200 and 300 ms (27) and multiple closed states, with one component of ϳ10 s (28). In addition, although the kinetics of Kir2.1-PIP 2 interactions are far from fully understood, the response time of Kir2.1 to PIP 2 sequestration by PIP 2 antibodies takes tens of seconds (22,37). For relatively slow changes in PIP 2 , such as resynthesis and basal depletion, these kinetics appeared to have little impact, as witnessed by the close correlation with measurements based on the sensitivity to light (e.g. Figs. 3 and 6); however, the kinetics of the light-induced PLC activity, particularly at low stimulus strength, are likely to be significantly underestimated. Even the maximum slope of the suppression of Kir current (Fig. 5E), which reached saturating values of ϳ150% maximum current s Ϫ1 , should be considered only as a minimum estimate for the maximal rate of PIP 2 depletion. Thus, these rates saturated with stimuli representing only ϳ1 absorbed photon per microvillus but would be expected to increase further with multiple hits per microvillus, which should recruit additional PLC molecules. The slow channel kinetics probably also result in a slight underestimation of the total light-induced suppression, because there may be significant resynthesis before the channels have completely adjusted to the new level (e.g. Fig. 3B). This seems likely to account for the slight (ϳ2-fold) residual mismatch between sensitivity to suppression of Kir2.1 R228Q and the response to light (Fig. 2D).
As discussed previously (7) the prolonged recovery time (t1 ⁄2 , ϳ50 s) following depletion, and the lack of any substantial recovery in rdgB mutant defective in PI transfer protein, strongly suggest that the estimated rates of PLC activity reflect not only the immediately available PIP 2 but also all PI and PIP that we propose is rapidly converted to PIP 2 "on demand" on a sub-second timescale. As will be discussed in more detail elsewhere, together with quantitative estimates of PI and PIP 2 levels in the microvilli, the present results allow a conservative estimate of the absolute rate of PLC hydrolysis in Drosophila phototransduction of PIP 2 as greater than 10 4 molecules s Ϫ1 photon Ϫ1 . These rates of PLC were measured in the absence of Ca 2ϩ influx by blocking or genetically eliminating both TRP and TRPL channels; as previously reported the Ca 2ϩ influx associated with the LIC rapidly inhibits PLC and under physiological conditions prevents this precipitous depletion of PIP 2 (7). Presumably, however, these rates of hydrolysis are also approached during the 20-to 100-ms latent period of the response, resulting in large transient and localized increases in DAG and InsP 3 .
Basal PLC Activity-Previous measurements of basal, unstimulated PLC activity have relied on biochemical experiments (e.g. Ref. 33). In the present study we were able to estimate the rate of spontaneous PIP 2 depletion in vivo in real-time by depriving the cells of ATP required for PIP 2 resynthesis. The rundown of Kir2.1 and other Kir channels in the absence of ATP has been widely reported, particularly in excised patches (37), but has generally been attributed to depletion of PIP 2 by lipid phosphatase activity. Because the spontaneous decay of Kir2.1 channels was either blocked or greatly slowed in the near null PLC mutant, norpA P24 , it seems likely that the spontaneous PIP 2 loss in the photoreceptors is largely due to basal PLC activity. Formally, however, we cannot exclude the possibility that PLC activity under these conditions also indirectly promotes some other mode of PIP 2 degradation (e.g. DAG might stimulate a lipid phosphatase). The maximum rates of PIP 2 loss amounted to ϳ40% of the total PIP 2 per minute with loss of all detectable PIP 2 even by the higher affinity Kir2.1 WT occurring typically within 10 min of break-in.
Although this apparent high rate of basal turnover may at first seem surprising, it is in keeping with the few available measurements in other cells. For example, a recent comprehensive biochemical study of PI turnover (33) estimated that the total cellular pool of PIP 2 in human neuroblastoma cells was turned over every 5 min under basal conditions. In fact, compared with the maximum light-activated rate (150% s Ϫ1 , Fig.  5), basal turnover in Drosophila photoreceptors appears exceptionally low. Thus, although the light-activated rate is believed to represent not only PIP 2 but also the microvillar PI and PIP pool, the estimate of basal activity presumably represents just PIP 2 . If, as in most cells, PIP 2 represented only 5-10% of the total PI pool, this would imply that the basal rate is ϳ2000 -4000 times less than the maximal activated rate. This compares, for example with a basal rate in neuroblastoma cells that is only ϳ20-fold less than the maximum activated rate (33). The fact that the low basal rate in Drosophila has the potential to deplete all detectable PIP 2 within 10 min is presumably a reflection of the extremely high density of PLC in the microvillar membrane (ϳ100 copies per microvillus (38)).
Under normal conditions, the basal activity of PLC may be of little consequence for the cell, because the PIP 2 is continually replaced. However, the demonstration of such a significant overall rate of basal activity is important for the interpretation of some recent studies on the mechanism of phototransduction. First, it contributes to the debate over the long-standing finding that the light-sensitive TRP channels activate spontaneously in the absence of ATP or when cells are challenged with metabolic inhibitors (30,31). On the one hand, it has been proposed that this is due to protein dephosphorylation, possibly of the light-sensitive channels themselves (31,39). However, in a recent study we found that activation of TRP channels by metabolic inhibition was dependent upon PLC, and we proposed that the spontaneous activation under these conditions was most likely due to failure of DAG kinase to metabolize DAG (40). Because substantial other genetic evidence is consistent with an excitatory role for DAG (reviewed in Refs. 3, 36, 41), this seems a plausible explanation, but it assumes the existence of a substantial basal PLC activity for which there was no evidence. This conclusion has since been challenged (39). The present study strongly indicates that there is indeed a high level of basal PLC activity, consistent with this interpretation. Second, the finding that the TRP channels in the rdgA mutant were constitutively active represented the first genetic evidence for the excitatory role of DAG in Drosophila phototransduction (36). It presupposed, however, both that there were significant amounts of PIP 2 remaining in rdgA mutants and that there was basal PLC activity, which could convert this to DAG. Again our results provide direct evidence supporting both these assumptions, thus consolidating our previous conclusions and adding strength to the proposal that DAG is an excitatory messenger in Drosophila phototransduction.
Concluding Remarks-The present study has demonstrated the potential of the Kir2.1 PIP 2 -sensitive ion channel, and in particular the mutant Kir2.1 R228Q as a biosensor for tracking PIP 2 in vivo providing unique quantitative insight into the dynamics of PIP 2 turnover. Their usefulness in other systems remains essentially unexploited; however, Kobrinsky et al. (35) demonstrated that a similarly engineered Kir2.1 channel (R218Q) also responded to agonist stimulation of PLC when expressed in Xenopus oocytes. Drosophila phototransduction has long been an influential model for PLC signaling. The ability to directly monitor both the activity of PLC and the PIP 2 resynthesis machinery in vivo, coupled with the rich repertoire of transduction mutants should provide the opportunity to dissect many features of the regulation of the cascade that have previously been intractable to analysis. We also note that the present study has established conditions whereby PIP 2 levels can be rapidly and quantitatively manipulated in vivo. PIP 2 is increasingly implicated in the regulation of an ever-growing list of membrane proteins, including a variety of ion channels and transporters. Targeted expression of proteins such as Kir2.1in Drosophila photoreceptors may prove a useful approach for exploring their PIP 2 dependence under physiological conditions.