Presence of a light-independent phospholipase A2 in bovine retina but not in rod outer segments.

Rod outer segments (ROS) are responsible for the visual transduction process. Rhodopsin, which constitutes 85-90% of ROS proteins, absorbs light photons, changes its conformation, and then binds to a heterotrimeric G-protein called transducin. As a consequence, transducin dissociates into Tα and Tβγ subunits. The presence in ROS of a phospholipase A2 (PLA2) stimulated by light and guanosine 5′-O-(3-thio)triphosphate was first demonstrated in 1987 (Jelsema, C. L. (1987) J. Biol. Chem. 262, 163-168). This led that author to conclude that ROS PLA2 could be involved in the phototransduction process, and raised the possibility of receptor-mediated activation of PLA2 via G-proteins in cell types other than rods. However, the biochemical characteristics and the role of this PLA2 have not been fully elucidated. We have tried to reproduce some of the results previously reported in order to further characterize this enzyme. We have found that, in our hands, there is neither light-dependent nor GTP-dependent PLA2 activity in intact purified ROS. We also failed to detect PLA1 activity in those ROS preparations. Nevertheless, we detected significant amounts of PLA2 activity in two subretinal fractions adjacent to ROS: RPE (enriched with retinal pigment epithelial cells) and P200 (presumably containing neuronal cells, Müller cells, and rod inner segments). The enzyme present both in RPE and P200 is light- and GTP-independent, Ca2+- and Mg2+-independent, and seems to be optimally active in the alkaline pH range. Our results suggest that there is, if any, vanishingly little PLA2 or PLA1 activity in intact purified ROS and that the activity levels previously reported in the literature could have been due to a contamination by either RPE or P200. This is supported by our observation that some contaminated ROS preparations were “PLA2 active.”

Rod outer segments (ROS) are responsible for the visual transduction process. Rhodopsin, which constitutes 85-90% of ROS proteins, absorbs light photons, changes its conformation, and then binds to a heterotrimeric G-protein called transducin. As a consequence, transducin dissociates into T␣ and T␤␥ subunits. The presence in ROS of a phospholipase A 2 (PLA 2 ) stimulated by light and guanosine 5-O-(3-thio)triphosphate was first demonstrated in 1987 (Jelsema, C. L. (1987) J. Biol. Chem. 262, 163-168). This led that author to conclude that ROS PLA 2 could be involved in the phototransduction process, and raised the possibility of receptor-mediated activation of PLA 2 via G-proteins in cell types other than rods. However, the biochemical characteristics and the role of this PLA 2 have not been fully elucidated. We have tried to reproduce some of the results previously reported in order to further characterize this enzyme. We have found that, in our hands, there is neither light-dependent nor GTP-dependent PLA 2 activity in intact purified ROS. We also failed to detect PLA 1 activity in those ROS preparations. Nevertheless, we detected significant amounts of PLA 2 activity in two subretinal fractions adjacent to ROS: RPE (enriched with retinal pigment epithelial cells) and P200 (presumably containing neuronal cells, Mü ller cells, and rod inner segments). The enzyme present both in RPE and P200 is light-and GTP-independent, Ca 2؉ -and Mg 2؉ -independent, and seems to be optimally active in the alkaline pH range. Our results suggest that there is, if any, vanishingly little PLA 2 or PLA 1 activity in intact purified ROS and that the activity levels previously reported in the literature could have been due to a contamination by either RPE or P200. This is supported by our observation that some contaminated ROS preparations were "PLA 2 active." In the mammalian eye, rod outer segments (ROS) 1 consist of a stack of 1000 -2000 disks which contain the visual pigment, rhodopsin. ROS are thus responsible for the phototransduction process. It has been clearly shown that following absorption of light, photoexcited rhodopsin binds to and activates transducin (T␣␤␥), which is a member of the heterotrimeric GTP-binding protein family. During the activation of transducin, the GDP molecule (normally associated with the inactive state of the protein) is exchanged for GTP. As a consequence, transducin dissociates into T␣ and T␤␥ subunits. T␣ is well known to activate a cGMP-phosphodiesterase whose activity eventually leads to hyperpolarization of the rod through closure of Na ϩ / Ca 2ϩ cGMP-dependent channels (see Refs. 1-4 for reviews).
Jelsema reported in 1987 (5) that phospholipase A 2 (PLA 2 ) activity was present in "crude ROS" and "partially purified ROS," and that this activity was stimulated by light and a non-hydrolyzable analog of GTP, GTP␥S. Moreover, Jelsema and Axelrod (6) demonstrated that T␤␥ was responsible for the activation of this ROS PLA 2 . These results led them to conclude that ROS PLA 2 could be involved in the phototransduction process. However, conflicting results have been reported since that time. In fact, although Zimmerman and Keys (8) detected phospholipase A activity in their ROS preparations, the activity that they measured, either PLA 2 or PLA 1 , was neither light-dependent nor GTP-dependent. It was rather stimulated by ATP and coenzyme A (CoA). Moreover, the maximum activity that they observed was approximately 1 order of magnitude lower than that reported by Jelsema (5). In addition, Castagnet and Giusto (9) published data on ROS PLA 2 activity but the maximum activity that they obtained was almost 3 orders of magnitude lower than that reported by Jelsema (5).
The mechanism of stimulus-response coupling between Gproteins and phospholipase A 2 could be important in many cells other than the retinal rods, given the wide distribution of PLA 2 in tissues (10 -20) and its potential role in controlling the biosynthesis of prostaglandins, leukotrienes, and other inflammatory mediators (21)(22)(23)(24)(25), as well as the turnover of phospholipids. Moreover, the biochemical characteristics of this retinal PLA 2 have not been fully studied yet and its role is still unknown. We present here results showing that two subretinal fractions, namely RPE (enriched with retinal pigment epithelial cells) and P200 (presumably containing neuronal cells, Mü ller cells, and rod inner segments) are rich in PLA 2 activity having an alkaline pH optimum. Our results also point out that purified ROS isolated by three different methods, vortexing, homogenizing or hand shaking, are devoid of significant PLA 2 activity. The varying levels of activity that have been reported in ROS preparations by the aforementioned authors (5,8,9) could thus, in most cases, be accounted for by a contamination by adjacent retinal cell types.  (26). ROS were broken off the retinas by vortexing (10 ϫ 1 s) and then purified as described (26). Purified v-ROS were assayed for purity and intactness as described below. They were either used directly for PLA 2 activity measurement, processed for electron microscopy, or aliquoted and stored at Ϫ80°C.
Isolation of ROS by Homogenizing (h-ROS)-Dark-adapted h-ROS were prepared from fresh bovine eyes kept either on ice or at 15°C. Dissection was made under dim red light according to a modification of the procedure described by Feeney-Burns and Berman (27). Briefly, 64 eyes were cut along the ora serrata. The anterior segment and the vitreous were then eliminated by tilting the eyes. Eyecups were gently filled with buffer A (20 mM Tris-HCl, 11% sucrose, 0.5 mM DTT, pH 7.4) and incubated at room temperature for 15 min. After removal of buffer A, retinas remained attached only at the optic disk and were cut with scissors. Sixteen retinas were collected in 40 ml of buffer B (20 mM Tris-HCl, 20% sucrose, 0.5 mM DTT, pH 7.4) and were either homogenized immediately or kept on ice in total darkness for about 2 h (to allow for preparation of RPE). Homogenization was done as described by Zimmerman and Godchaux (28). The loose-fitting Potter-Elvehjeim homogenizer (clearance of 2 mm) was operated at 300 rpm. Six strokes (20 s/each) were used. The homogenate was then sedimented at 200 ϫ g (4°C, 5 min; Sorvall HB4 rotor). The supernatant was collected and kept on ice while the pellet was gently resuspended by inversion in buffer B and centrifuged in the same conditions. Supernatants were pooled and further centrifuged at 7,000 ϫ g (4°C, 6 min; Sorvall SS34 rotor). The 7,000 ϫ g supernatant was discarded and the pellet resuspended in buffer B with a wide mouth plastic pipette. This suspension was layered on the top of six 27-50% continuous sucrose gradients and centrifuged (140,000 ϫ g, 4°C, 16 h; Beckman SW28 rotor). ROS bands were collected, diluted to 20% sucrose with buffer B, and pelleted (17,500 ϫ g, 4°C, 5 min; Sorvall SS34 rotor). Purified h-ROS were resuspended in a minimal volume of buffer B and assayed for purity and intactness (see below). They were then either used directly for PLA 2 activity measurement, processed for electron microscopy, or aliquoted and stored at Ϫ80°C.
Isolation of ROS by Hand Shaking (hs-ROS)-Dark-adapted hs-ROS were purified from fresh bovine eyes kept either on ice or at 15°C according to the procedure of McDowell and Kü hn (29). Briefly, 60 retinas were gently peeled off the eyecup and rinsed with buffer C containing 45% sucrose (buffer C: 50 mM Na 2 HPO 4 , 50 mM NaH 2 PO 4 , 1 mM MgCl 2 , 0.5 mM DTT, 0.1 mM EDTA, pH 7.0). They were then cut at the optic disk, collected in a 250-ml Erlenmeyer containing 54 ml of buffer C containing 45% sucrose, and gently hand shaken for 3 min. The suspension was centrifuged (3,000 ϫ g, 4°C, 5 min; Sorvall SS34 rotor) and the supernatant collected (SN 3,000 ϫ g ϭ crude ROS). It was used as the starting material for further purification of ROS and for determination of PLA 2 activity associated with particulate and soluble fractions at each purification step. The crude ROS-supernatant (SN 3,000 ϫ g) was diluted 1:1 with buffer C and centrifuged (4,400 ϫ g, 4°C, 7 min; Sorvall SS34 rotor). The 4,400 ϫ g pellet (P 4,400 ϫ g ϭ semi-purified ROS) was resuspended in buffer C containing 25% sucrose with a wide-mouth plastic pipette and layered on the top of discontinuous sucrose gradients (27-32%). Gradients were centrifuged at 140,000 ϫ g (4°C, 2 h; Beckman SW28 rotor). Purified diluted ROS (SN 140,000 ϫ g) were collected at the 27-32% interface, mixed with 1 volume of buffer C and centrifuged (17,500 ϫ g, 4°C, 5 min; Sorvall SS34 rotor). The 17,500 ϫ g pellet (P 17,500 ϫ g ϭ purified concentrated hs-ROS) was resuspended in a minimum of buffer C containing 45% sucrose. hs-ROS were assayed for purity and intactness as described below. They were then either used directly for PLA 2 activity measurement, processed for electron microscopy, or aliquoted and stored at Ϫ80°C. Aliquots of the 4,400 and 17,500 ϫ g supernatants (SN 4,400 and SN 17,500 ϫ g, respectively) were kept for measurements of rhodopsin and total protein concentrations, as well as PLA 2 activity.
Preparation of RPE-RPE cells were purified according to a modification of the method described by Feeney-Burns and Berman (27). After the retinas were cut from the optic disc, eyecups were gently filled with buffer A containing 2 mM EGTA and incubated at room temperature for 15 min. In the next steps, eyes are treated one at a time. Buffer A-EGTA was discarded and replaced by 1 ml of buffer A. RPE cells were immediately brushed off from the choroid with a soft camel hair brush (number 12) and the suspension was aspirated with a wide-mouth plastic pipette. RPE suspension was pelleted (400 ϫ g, 4°C, 10 min; Sorvall HB4 rotor), resuspended in a minimal volume of buffer B and either used directly or aliquoted and stored at Ϫ80°C.
Preparation of P200 -The pellet obtained after centrifugation of the homogenized retinas at 200 ϫ g (see "Isolation of ROS by Homogenizing") was used as the "P200" fraction. For some experiments, it was rehomogenized with a tight-fitting (0.5 mm clearance) Potter-Elvehjem homogenizer before use. It was either assayed directly for PLA 2 activity or aliquoted and stored at Ϫ80°C.
Phospholipase A 2 Assay-In vitro assays of PLA 2 activity were performed on each type of purified ROS, using [ 14 (30). When [ 14 C]PAPE was used as the substrate, unlabeled DPPC (80 g/ml) was added to [ 14 C]PAPE before the evaporation of solvent. The sonication and incubation steps were done at 42°C. Reactions were initiated by addition of 20 l of the sonicated radiolabeled substrate to ROS aliquots containing 20 -75 g of proteins. The buffer used was either 30 mM Tris-HCl, pH 8.8, 5 mM CaCl 2 , 30 mM MgCl 2 , 0.6 mM NaCl, 4 mM glutathione as described by Jelsema (5) or 20 mM Tris-HCl, pH 7.5, 5 mM CaCl 2 , 30 mM MgCl 2 , 50 mM KCl, 0.5 mM DTT, 1 mM ATP, 0.25 mM CoA as described by Zimmerman and Keys (8). The total volume was 250 l. Incubations were done at 37°C under either dim red light or white light (1330 lx; we used a 250 watt tungsten lamp which practically does not emit in the UV range (31)). Reactions were stopped at specified times with 1.25 ml of Dole's reagent (isopropyl alcohol, n-heptane, 1 N H 2 SO 4 , 40:10:1 v/v/v). After addition of 0.75 ml of n-heptane and 0.5 ml of water, samples were vortexed and allowed to stand for 5 min for phase separation. The enzymatically released [ 14 C] oleic or [ 14 C]arachidonic acids were separated from unreacted substrate by elution of the heptane phase on columns containing 200 mg of dessicated Bio-Sil A silicic acid with 1 ml of diethyl ether. After addition of 7.5 ml of Econofluor-2 to the 1-ml eluates, radioactivity was measured using a Beckman scintillation counter programmed to correct for quenching and counting efficiency. PLA 2 activity was expressed as nanomoles of [ 14 C]arachidonic or [ 14 C]oleic acid released/mg of proteins and was corrected for the dilution of the radiolabeled substrate by the endogenous unlabeled phospholipids. This was done by using a phospholipid to rhodopsin weight ratio of 1:1 (32). Zero-time control values were subtracted. This method was also used to assay PLA 2 activity in the particulate and soluble fractions collected during the ROS purification procedure, as well as in the RPE and P200 fractions. In these cases, we used phospholipid to protein weight ratios of 0.09:1 for RPE, which is in agreement with the data published by Berman et al. (33), and 0.35:1 for P200 (as determined in our laboratory by measuring the phosphorus content of a P200-phospholipid extract (see below)).
Measurement of ROS PLA 2 or PLA 1 Activity Toward Endogenous Phospholipids-In vitro investigations of ROS PLA 2 or PLA 1 activity were done by HPLC. Purified ROS were incubated at 37°C for different periods of time (0, 0.5, 10, 30, and 60 min). The endogeneous unlabeled phospholipids were the only source of substrate. Reaction was started by transferring samples from ice to 37°C. Samples were either exposed to white light (1330 lx) or a dim red light during the incubation. Reaction was stopped by cooling the tubes to 0°C and adding 0.5 ml of argon-saturated methanol/mg of ROS proteins. Phospholipids were extracted and quantitated by measuring its phosphorus content according to a modification of the procedure described by Miljanich (32). Briefly, aliquots of the phospholipid extracts were evaporated to dryness at 110°C. Then, 750 l of H 2 SO 4 , 1.8 M, HClO 4 , 7% (40:25, v/v) were added and samples were heated for 1 h at 210-220°C on an aluminum heating block with occasional shaking. After cooling to room tempera-ture, 1 ml of ascorbate, 2%, and 1 ml of molybdate, 0.5%, were added to each tube. Tubes were vigorously vortexed and incubated for 1 h at 37°C. Absorbance was read at 825 nm. Typically, 800 g of ROS phospholipids were injected on two HPLC columns placed in tandem: 1) a 4.6 ϫ 250-mm analytical Ultrasphere Si and 2) a 10 ϫ 250-mm semi-preparative Ultrasphere Si (Beckman). The solvent system consisted of hexane:isopropyl alcohol (3:4, v/v) with a final concentration of water increasing from 1 to 9.1%. The elution was monitored at 206 nm using a Waters M490 multi-wavelength detector. Appearance of lysophospholipid peak(s) and/or decrease in phospholipid peak height was considered to be suggestive of endogenous ROS PLA 2 or PLA 1 activity. The retention time of lysophospholipids that could be expected from ROS PLA 2 or PLA 1 activity was determined as described above with commercially available lysophospholipids.
Electron Microscopy of ROS-Samples of freshly purified ROS (v-ROS, h-ROS, and hs-ROS) were fixed for 1 h at 20°C in 3% glutaraldehyde in a cacodylate buffer (100 mM, pH 7.4, 20% sucrose or 140 mM NaCl, 12 mM CaCl 2 , 2 mM MgCl 2 ) and then washed 3 times with the buffer alone. Post-fixation was done for 30 min at 4°C in 1% OsO 4 in the same buffer. After they were washed 3 times, the fixed samples were progressively dehydrated in 30, 50, 70, 85, 95, and 100% ethanol, and finally in 100% propylene oxide. Fixed and dehydrated ROS samples were then embedded in Epon 812 and stained with uranyl acetate.
Other Methods-Rhodopsin concentration was determined according to the procedure of Raubach et al. (35). Total protein concentration was determined using the Lowry protein assay kit from Sigma. ROS purity was evaluated by measuring the ratio of A 280 /A 500 nm (26). ROS intactness was estimated by measuring the production of NADPH as described by Schnetkamp and Daemen (36).

RESULTS
Effect of Light on ROS PLA 2 Activity-Considering the discrepancy between ROS PLA 2 activity levels reported in the literature and that its role is still unknown, we have attempted to reproduce some of the reported results in order to further characterize this enzyme. We first tried to measure PLA 2 activity in the conditions described by Jelsema (5) because she reported the highest level of activity. Surprisingly, we could not detect any significant light-dependent or light-independent PLA 2 activity in our ROS preparations (Fig. 1). In an attempt to detect PLA 2 activity, we have tried: 1) to preincubate the eyes on ice, in total darkness, for different periods of time (1, 2, or 4 h) prior to dissection, 2) to vary the composition of the ROS purification buffer, 3) to vary the method used to either collect the retinas or detach ROS from the retina (see the three procedures for ROS isolation and purification), and 4) to use mixed substrate vesicles ([ 14 C]PAPC ϩ DPPC) in the activity assay, but none of these conditions allowed us to detect significant PLA 2 activity in ROS. Fig. 1 shows typical results that were obtained in all those diverse conditions.
Effect of Different Substrates on ROS PLA 2 Activity-We have tested ROS PLA 2 activity toward other exogenous radiolabeled substrates differing from [ 14 (5), we did not observe any PLA 2 or PLA 1 activity that could be considered significantly different from the background. In fact, as can be seen in Fig. 2, none of the substrates we used was significantly hydrolyzed by our ROS preparations.
Intactness of ROS-As we could not reproduce the results of Jelsema (5) and Zimmerman and Keys (8), we have investigated the intactness of our ROS preparations. Using electron microscopy of freshly purified v-ROS, we found that our preparations were in fact completely burst (Fig. 3A). Since our negative results could be due to the loss of the PLA 2 (or PLA 1 ) enzyme itself or some soluble activating factors during the purification procedure, we changed our method of ROS preparation in order to get intact plasma membranes. We used homogenization (as described by Zimmerman and Godchaux (28)) and hand shaking (as described by McDowell and Kü hn (29)) to isolate h-ROS and hs-ROS, respectively. Fig. 3, B and C, clearly show that hand shaking gives better results; only ROS isolated by hand shaking have densely packed discs as well as a sealed plasma membrane (Fig. 3C). Moreover, we measured the production of NADPH as a quantitative criteria for evaluating the intactness of v-ROS, h-ROS, and hs-ROS. Given that the NADPH-recycling enzyme required for the reduction of rhodopsin's chromophore is cytosolic (36), we measured the production of NADPH, in the presence of exogenously added substrates (D-glucose 6-phosphate ϩ ␤NADP), prior to FIG. 1. Effect of light on ROS PLA 2 activity. PLA 2 activity was measured as described by Jelsema (5)  and after solubilization of ROS membranes with Triton X-100 (36). The results were compared to negative controls where addition of both substrates was omitted. An increase in the absorbance at 340 nm after solubilization of the plasma membrane indicates that cytosolic NADPH-recycling enzymes were present in ROS and thus provides an indication that ROS were initially intact. Fig. 4 shows results obtained for the three types of ROS preparations. The three curves show a slow production of NADPH prior to membrane solubilization with Triton X-100. The addition of the detergent clearly leads to a large increase of NADPH production, mainly for h-ROS and hs-ROS. Obviously, v-ROS contain much less intact ROS as observed in Fig. 3A. From Figs. 3 and 4, the intactness of these ROS preparations can be assessed as hs-ROS ͘ ͘ h-ROS ͘ ͘ ͘ ͘ ͘ v-ROS. Nonetheless, no matter whether we used v-ROS, h-ROS, or hs-ROS preparations, we have failed to detect significant PLA 2 activity. Fig. 1 shows typical results that we observed with each type of preparation.
Distribution of PLA 2 Activity Among Particulate and Soluble Fractions Generated during the Purification of hs-ROS-To further investigate the possibility that PLA 2 enzymes were lost during the purification procedure, we measured PLA 2 activity in the particulate and soluble fractions generated at each step of ROS purification. As can be seen in Table I, there is no significant PLA 2 activity in any of these fractions. Moreover, coincubation of the corresponding pellet and supernatant obtained after the centrifugation run at either 4,400 or 17,500 ϫ g was not sufficient to restore PLA 2 activity. The only PLA 2 active fraction was "P 3,000 ϫ g" which is equivalent to the P200 fraction (presumed to contain neuronal cells, Mü ller cells, and rod inner segments). These results argue strongly against the loss of PLA 2 enzymes during ROS purification and supports our results showing the absence of endogenous PLA 2 activity in intact ROS (see Figs. 1 and 2, and Table I).
Effect of Guanine Nucleotides on ROS PLA 2 Activity-It has been well established that guanine nucleotides such as GTP and GDP influence the active-inactive state transition of Gproteins. Moreover, Jelsema and Axelrod (6) have shown that light-activation of ROS PLA 2 occurred through the ␤␥ complex of transducin (T␤␥). That is, photoexcited rhodopsin undergoes conformational changes which allow its binding to the inactive undissociated transducin ␣-subunit containing a GDP molecule in its catalytic site (T␣-GDP). After photoexcited rhodopsin has bound to T␣-GDP, the GDP molecule is exchanged for GTP which confers the active state to transducin and allows it to dissociate into T␣-GTP and T␤␥ subunits. T␤␥ would then, according to the results of Jelsema and Axelrod (6), be able to activate ROS PLA 2 until GTP is hydrolyzed to GDP by T␣ and T␣-GDP reassociates with T␤␥. We thus have attempted to measure PLA 2 activity in the presence of a non-hydrolyzable GTP analog (GTP␥S) or a GDP analog that cannot be phosphorylated (GDP␤S) to promote, respectively, the permanent dissociation or association of transducin T␣ and T␤␥ subunits. As a consequence, light-stimulated ROS PLA 2 activity was expected to be enhanced while dark-adapted basal activity was expected to be lowered. We have also tested the effect of unmodified hydrolyzable GTP. Unfortunately, as shown in Fig. 5, we did not observe any significant light-dependent or lightindependent PLA 2 activity in any of these conditions.
Effect of Mg 2ϩ and Ca 2ϩ Concentration, EGTA, and Detergent on ROS PLA 2 Activity-Since Mg 2ϩ is an essential cofactor of G-proteins and since there are several types of PLA 2 differing in their Ca 2ϩ requirement (see Refs. 37-39 for reviews), we have tested the effect of two Mg 2ϩ concentrations on ROS PLA 2 activity as a function of Ca 2ϩ concentration. Moreover, it has been shown by Marshall and McCarte-Roshak (40) that the addition of EGTA to the assay buffer can reduce the Ca 2ϩ requirement of some PLA 2 . So we have used two types of PLA 2 assay buffers in our experiments: one containing increasing concentrations of Ca 2ϩ and the other one containing an increasing excess of Ca 2ϩ over a fixed concentration of EGTA. We have also tested the effect of emulphogene on ROS PLA 2 activity since this detergent is efficient at extracting ROS proteins (41) and could thus favor the interactions between the substrate vesicles and ROS PLA 2 . However, as shown in Fig. 6, we did not detect any significant PLA 2 activity in any of these conditions. ROS PLA 2 or PLA 1 Activity Toward Endogenous Phospholipids-As can be seen in Figs. 1, 2, 5, and 6, we did not observe significant PLA 2 activity in ROS incubated with exogenous radiolabeled substrate, whatever the conditions we used, and whatever the type of ROS preparations we used. Moreover, the results obtained with [ 14 C]DOPC also suggest the absence of PLA 1 activity (see Fig. 2). We have thus tested ROS PLA 2 or PLA 1 activity toward endogenous unlabeled ROS phospholipids. Purified ROS were incubated for different periods of time under either white light or dim red light. A decrease in any phospholipid peak height and/or the appearance of lysophospholipid peak(s) on HPLC chromatograms, as compared to zero time controls, was considered to be indicative of ROS PLA 2 or PLA 1 activity. Fig. 7 represents a typical HPLC elution profile of ROS phospholipids. We did not find any significant decrease in any phospholipid peak height from samples incubated under light or in the dark. This result is in agreement with those shown in Figs. 1, 2, 5, and 6, and suggests that there is neither PLA 2 nor PLA 1 activity in intact ROS. PLA 2 Activity in Subretinal Fractions Other Than ROS-In order to explain the discrepancy between the activity levels reported by Jelsema (5), Zimmerman and Keys (8), Castagnet and Giusto (9), and in the present study, we have measured PLA 2 activity in two subretinal fractions containing cell types which are found adjacent to ROS in vivo, RPE (enriched with retinal pigment epithelial cells) and P200 (presumably containing neuronal cells, Mü ller cells and rod inner segments). Significant levels of light-independent PLA 2 activity were detected in both fractions (Fig. 8). Considering the possibility that ROS PLA 2 activity detected by Jelsema (5), Zimmerman and Keys (8), and Castagnet and Giusto (9) is due to a contamination by adjacent cell types, our results on a light-independent PLA 2 activity would be in agreement with the characteristics deter-FIG. 5. Effect of guanine nucleotides on ROS PLA 2 activity. A, PLA 2 activity of ROS exposed to dim red light (dark-adapted). B, PLA 2 activity of ROS exposed to white light. The activity was measured as described in Fig. 1   a The assay was performed over 15 min as described in the legend to Fig. 1. P 3,000 g (not shown) is equivalent to the P200 fraction (see "Preparation of P200"). These results are representative of mean Ϯ S.D. of triplicates from two separate experiments. mined by Zimmerman and Keys (8) but would contradict the results of Jelsema (5) and Castagnet and Giusto (9) who observed a 3.3-fold and a 35% increase upon light-stimulation, respectively.
Influence of pH and Ca 2ϩ on RPE and P200 PLA 2 Activity-In order to better identify the type of PLA 2 present in RPE and P200, and because it is known that there are at least three types of PLA 2 differing by their Ca 2ϩ requirement and their pH optimum (37)(38)(39), we have tested the effect of pH and Ca 2ϩ concentration on its activity. As shown in Fig. 9, the enzyme present in both fractions was found to be Ca 2ϩ -independent and optimally active at alkaline pH. Moreover, Fig. 9, A and C, suggest that Tris-HCl is not a good buffer to measure PLA 2 activity, at least in our systems, when used in conjunction with calcium. In fact, we always observed an inhibitory effect of this buffer on both RPE and P200 PLA 2 activity. DISCUSSION In 1987, Jelsema (5) reported high levels of PLA 2 activity in "crude" and "partially purified" ROS preparations. Moreover, the enzyme was shown to be stimulated 4.5-and 3.8-fold by light and GTP␥S, respectively, as compared to dark-adapted controls. Values of 133.6 Ϯ 24.0 and 110.5 Ϯ 12.7 nmol of [ 14 C]arachidonic acid released/mg of proteins/10 min were reported for light-and GTP␥S-stimulated samples, respectively, as compared to a value of 29.4 Ϯ 2.6 for dark-adapted controls. Thus, it was suggested that ROS PLA 2 could be regulated by light through a GTP-binding protein, herein transducin (5). In 1993, Castagnet and Giusto (9) also published data on the presence of a light-stimulated PLA 2 in ROS. However, the maximum activity level that they observed (see Table II) was more than 2 orders of magnitude lower than that reported by Jelsema (5), although using the same assay conditions. That is, Castagnet and Giusto (9) obtained 350 times less activity than Jelsema (5) for a 6-fold longer incubation period (see Table II). The reported effect of light and GTP␥S was also much less important with only a 35 and 63% increase, respectively, over dark-adapted controls (9).
Zimmerman and Keys (8) also reported results on a ROS phospholipase A activity but, in their hands, it was found to be light-independent but CoA-and ATP-dependent. The substrate ([ 14 C]DOPC) they used did not allow them to discriminate between PLA 2 or PLA 1 activity. Moreover, the maximum activity that they measured (see Table II) was approximately 1 order of magnitude lower than that reported by Jelsema (5). In fact, Zimmerman and Keys (8) observed seven times less activ- Intact ROS were incubated in 30 mM Tris-HCl, pH 8.8, 5 mM CaCl 2 , 30 mM MgCl 2 , 0.6 mM NaCl, 4 mM glutathione without any exogenous substrate added. During the assay, samples were kept at 37°C and were either exposed to white light or dim red light for 0, 0.5, 15, 30, and 60 min. Phospholipids were then extracted, quantitated, and separated by HPLC as described (see "Experimental Procedures"). FFA, free fatty acids; PE, phosphatidylethanolamine; PS, phosphatidylserine; PC, phosphatidylcholine; lysoPE, lyso-phosphatidylethanolamine; lysoPS, lyso-phosphatidylserine; lysoPC, lyso-phosphatidylcholine. The elution profile is typical of three separate injections for each condition.

FIG. 8. PLA 2 activity in subretinal fractions other than ROS.
A, PLA 2 activity of RPE and P200 exposed to dim red light. B, PLA 2 activity of RPE and P200 exposed to white light. The activity was measured as described in the legend to Fig. 1 using [ 14 C]PAPC as the substrate. Samples were incubated for 10 min at 37°C. The results are mean Ϯ S.D. of triplicates from three separate experiments. ity than Jelsema (5) for a 6-fold longer incubation period (see Table II). They reported activity values of 11.22 Ϯ 2.11 mmol of [ 14 C]oleic acid released/mol of phospholipid/h which, once expressed in the same units as Jelsema (5) and Castagnet and Giusto (9), represents 19.16 Ϯ 3.60 nmol of [ 14 C]oleic acid released/mg of proteins/h.
Recently, Jung and Remé (43) reported that light could stimulate the release of [ 3 H]arachidonic acid in intact retinas by approximately 2-fold as compared to dark-adapted controls. It is, however, difficult for us to compare their activity values with those cited above as they were not expressed in terms of specific activity but rather as a ratio of [ 3 H]arachidonic acid released/[ 3 H]arachidonic acid preincorporated in retinal phospholipids (43). Moreover, they worked with whole retinas on a retinal light-damage model using lithium-treated albino rats. These conditions are very different from those used by Jelsema (5, 6), Zimmerman and Keys (8), Castagnet and Giusto (9), and in the present paper.
Considering the high levels of ROS PLA 2 activity reported by Jelsema (5), its likely regulation by a G-protein (transducin) (6), the discrepancy between ROS PLA 2 activity in the literature (see Table II), and that its biochemical characteristics and its role have not been fully elucidated, we have attempted to reproduce the aforementioned results. We were first interested in characterizing this enzyme because the mechanism of stimulus-response coupling between G-proteins and PLA 2 could be important in many cells other than retinal rods. Moreover, the activation of PLA 2 through G-protein-coupled receptors in cells is still a matter of controversy (7). Unfortunately, we did not find any significant PLA 2 or PLA 1 activity in ROS, whatever the conditions we used and whatever the type of ROS preparations we used (Figs. 1-7). We did not detect either light-and GTP-dependent (Fig. 5, Table I), or light-independent but CoAand ATP-dependent PLA 2 or PLA 1 activity (Fig. 2). In our hands, there was no significant increase in ROS PLA 2 or PLA 1 activity over time (Figs. 1, 2, 5, and 7). This means that the amount of 14 C-fatty acids collected after the enzymatic reaction had been quenched was, in all cases, equivalent to the level of  Considering that our negative results could be due to the loss of PLA 2 enzymes, or some soluble-activating factors, from ROS during the purification procedure, we investigated the intactness and PLA 2 activity of ROS isolated using three fundamentally different methods. Indeed, there are essentially three basic methods described in the literature for ROS purification, the major difference between them being the way ROS are separated from the retina: 1) vortexing (v-ROS), 2) homogenizing (h-ROS), and 3) hand shaking (hs-ROS). We thus prepared v-ROS, h-ROS, and hs-ROS as described (see "Experimental Procedures") and assayed each preparation for PLA 2 activity, after we looked at their intactness both qualitatively and quantitatively (Figs. 3 and 4). The intactness was quantitatively estimated as described by Schnetkamp and Daemen (36), by measuring the production of NADPH for each type of ROS preparations. This assay is based on the fact that sealed ROS plasma membranes prevent access of cytosolic NADPH-recycling enzymes to exogenously added substrates. Thus, a larger production of NADPH following solubilization of the membrane with Triton X-100 is expected to reflect a higher degree of intactness prior to solubilization.
Since we did not observe any contaminating mitochondria, which would give erroneously high results (36), in the electron micrographs of our v-ROS, h-ROS, and hs-ROS preparations ( Fig. 3 as well as other micrographs not shown), the extent of NADPH production effectively correlates with the extent of intactness. However, the highest production of NADPH (Fig. 4) by the most intact hs-ROS (Fig. 3C), did not correlate with a higher level of PLA 2 activity. None of the ROS preparations show significant levels of PLA 2 activity (Figs. 1, 2, 5, 6, and 7). We also failed to detect PLA 2 activity in the different particulate and soluble fractions that were generated during the purification of hs-ROS (see Table I). Incubating for longer (30 min) or shorter (1, 5, or 10 min) periods of time did not appear to be beneficial since no significant PLA 2 activity was detected (data not shown). Another important point is that we could not recover any significant PLA 2 activity by coincubating the corresponding pellet and supernatant generated after centrifugation at either 4,400 or 17,500 ϫ g (Table I). These results thus strongly suggest that the absence of PLA 2 or PLA 1 activity in our ROS preparations is not due to the loss of PLA 2 or PLA 1 enzymes or some soluble-activating factors during ROS purification.
In order to understand why we did not find PLA 2 activity in ROS whereas others did (5,8,9), we have assayed PLA 2 activity in two other subretinal fractions and found that RPE and P200 contained reproducibly high levels of PLA 2 activity (Fig.  8). This result suggests that the varying levels of activity that have been reported for ROS preparations could be due, in most cases, to contamination to varying extent by adjacent retinal cell types. As a matter of fact, we found that there was a relationship between the level of PLA 2 activity in our h-ROS preparations and their purity coefficient (A 280 /A 500 ) (results not shown). That is, rhodopsin constitutes 85-90% of total ROS proteins (44) and has two principal absorption bands: 1) aromatic amino acids of the protein moiety at 280 nm and 2) the 11-cis-retinal Schiff base at 500 nm. Thus, measurement of this A 280 /A 500 ratio for ROS preparations is indicative of their contamination by proteins from adjacent cell types other than ROS. Moreover, Salesse et al. (26) have demonstrated that this purity criterion is useful to evaluate the contamination of ROS by non-ROS proteins. In our hands, PLA 2 activity was detected only in contaminated h-ROS. Samples with the highest A 280 / A 500 ratios being the most PLA 2 active ROS preparations. For example, we detected 12.5 Ϯ 0.8 and 4.4 Ϯ 0.7 nmol of hydrolyzed [ 14 C]arachidonic acid/mg of proteins/10 min for ROS preparations having a A 280 /A 500 ratio of 6.9 and 3.4, respectively (data not shown). As a comparison, results presented in Figs. 1-7 were all obtained with ROS preparations having A 280 /A 500 ratios which were always lower than 3.
This observation supports the idea that the different levels of ROS PLA 2 activity reported in the literature could be due to a contamination by proteins coming from adjacent retinal cell types. However, it still does not explain why Jelsema (5) observed such a strong light effect on PLA 2 activity whereas Castagnet and Giusto (9) only noted a slight effect, and Zimmerman and Keys (8) and our laboratory did not (see Figs. 1 and 5). We have tried to restore this light-stimulation by incubating RPE and P200 in the presence of ROS. Unfortunately, ROS were unable to induce light-activation of PLA 2 in either fraction. The activity detected in light-exposed samples containing ROS were equivalent to dark-adapted controls without ROS (not shown).
It could be suggested that the light-sensitive PLA 2 detected by Jelsema (5) is normally associated with the outside of the plasma membrane and that, depending on the treatment during the purification, we lose it either partially or totally. This could explain why Castagnet and Giusto (9) saw only very low levels of PLA 2 activity whereas we saw none (Figs. 1, 2, and 5-7). We cannot exclude this possibility. However, given that PLA 2 hydrolyzes phospholipids, which account for up to 80% of total ROS lipids (45), it seems improbable to us that the enzyme is present on the outside of the plasma membrane since it certainly requires to be highly regulated. Moreover, it was shown (5, 6) to be regulated by G-protein and guanine nucleotide components of intracellular signal transduction. One alternative hypothesis is that the PLA 2 activity measured by Jelsema (5) was in fact present in rod inner segments still present in her crude and partially purified ROS preparations and was activated secondary to light-activation of the ROS visual cascade. The extent of light-sensitive PLA 2 activity detected in ROS would thus be dependent on the extent of contamination by rod inner segments still attached to ROS (via the connective cilium). Since electron microscopy of our ROS preparations did not reveal the presence of mitochondria, this would mean that our preparations were essentially free of rod inner segments and this could thus explain why we did not observe any light-sensitive PLA 2 activity. This latter hypothesis suggests that PLA 2 activity found in our P200 fractions could come, at least partly, from rod inner segments. The absence of a light-effect (Fig. 8) could then be due to the disruption of the connection between rod inner and outer segments. PLA 2 activity of rod inner segments would only be light-sensitive, although probably indirectly, when rod inner and outer segments are joined to each other by the connective cilium. Moreover, this light-stimulation could effectively be mediated by a G-protein as described by Jelsema (6) but, since cross-reactivity between different G-proteins has been demonstrated in different systems (46,47), the exogenously added transducin could have mimicked the role that should normally be held by a different but similar G-protein.
Considering the high PLA 2 activity levels that we detected in RPE and P200, we have tested the influence of pH and Ca 2ϩ concentration on this activity to identify the type of enzyme present in both fractions. As shown in Fig. 9, PLA 2 present in both RPE and P200 is optimally active in the alkaline pH range and is Ca 2ϩ -independent. However, care should be taken when interpreting the effect of Ca 2ϩ because Reynolds et al. (48) have recently shown that, in the absence of Ca 2ϩ , high salt concentrations can overcome the requirement of PLA 2 for divalent metals. They showed that 10 mM Mg 2ϩ could stimulate human cytosolic PLA 2 to almost the same extent as did 2 mM Ca 2ϩ . Since those pH-and Ca 2ϩ -dependence measurements were made in the assay conditions described by Jelsema (5) and thus contained a high concentration of Mg 2ϩ , additional experiments using different Mg 2ϩ concentrations will be required to characterize this enzyme. As can be pointed out in Fig. 9, the maximum activity detected in RPE and P200 are, respectively, lower than those shown in Fig. 8. This might be explained by the fact that, for these experiments (see Fig. 9), we used buffers at 133 mM final concentration, instead of 30 mM (Fig. 8), to make sure that substrate and sample buffers would not significantly affect the final pH needed for the assay. It may be possible that such high buffer concentration partially inhibits PLA 2 activity by affecting either the PLA 2 -substrate interactions or the enzyme itself. Otherwise, it is possible that part of the hydrolyzed [ 14 C]arachidonic acids were further metabolized during the 1-h incubation period, thus rendering them unavailable for quantitation as free fatty acids. Another point is that Tris-HCl seems to inhibit the enzyme activity at pH 7, 8, and 9 when used in conjunction with calcium. This is evident both in RPE and P200 (see Fig. 9, A and C). We also had similar difficulties with cacodylate and imidazole buffers having a profound inhibitory effect on PLA 2 activity at pH 5, 6, and 7 and 6, 7, and 8, respectively (results not shown). Although we were able to successfully replace cacodylate and imidazole buffers by Na ϩ acetate, pH 5 and 6, and bis-Tris, pH 6 and 7, buffers, we did not find an adequate combination of buffers which could be used at pH 7, 8, and 9 instead of Tris-HCl. Glycine seems to be a more "permissive" buffer since we repeatedly obtained higher levels of PLA 2 activity with glycine, pH 9, compared to Tris-HCl, pH 9 (see Fig. 9). We also tested CHES and AMP buffers at pH 9 and we observed results which were similar to those obtained with glycine, pH 9. In fact, activity was greater with CHES, AMP, and glycine buffers compared to Tris-HCl (not shown). This suggests that Tris-HCl is not a good buffer to assay PLA 2 activity at pH 7, 8, or 9 in RPE and P200.
Taken together, our results suggest that both RPE and P200 contain PLA 2 activity which is light-independent, Ca 2ϩ -independent, and optimally active at alkaline pH (Figs. 8 and 9). Considering that the maximum activity observed in RPE is less than that observed in P200, the possibility that RPE preparations are contaminated by P200 cells cannot be completely excluded. However, it seems unlikely that neuronal cells, Mü ller cells, and/or rod inner segments could have contaminated our RPE preparations since retinas were removed from the eyecup after incubation with buffer which allows ROS to retract from RPE microvilli, thus facilitating detachment of retinas (49). Moreover, it is well known that RPE cells and ROS are in intimate contact (50). Most of the contamination of RPE should thus come from ROS which, according to our results, does not contain PLA 2 . Berman et al. (33) have demonstrated that most RPE preparations purified by brushing out the cells are contaminated with varying amounts of ROS and red blood cells. Again, this should not affect our results since, as shown in Figs. 1-7, ROS do not contain PLA 2 activity. Moreover, PLA 2 activity present in RPE is Ca 2ϩ -independent (Fig. 9, A and B) whereas that in red blood cells is Ca 2ϩ -dependent (51). This suggests that RPE-PLA 2 is different from red blood cell-PLA 2 and probably not due to a contamination by P200. More experiments will be necessary to verify this hypothesis.
In summary, we have shown that there is no significant PLA 2 or PLA 1 activity in ROS and that the activity levels previously reported by Jelsema (5), Castagnet and Giusto (9), and Zimmerman and Keys (8) could be accounted for by a contamination by adjacent retinal cell types. We identified two potential sources of such contaminating activity: RPE and P200. The enzyme present in both fractions is light-and Ca 2ϩindependent and is optimally active at alkaline pH. Other experiments are needed to further characterize the PLA 2 present in RPE and P200. It will be particularly important to study PLA 2 activity in P200 because of its cell composition. Indeed, we hypothesize that PLA 2 activity found in our P200 fractions could come, at least partly, from rod inner segments. The absence of a light effect in our hands could then be due to the disruption of the connection between rod inner and outer segments. PLA 2 activity of rod inner segments could be lightsensitive only when these two segments are joined by the connective cilium. It will thus be very interesting to isolate and purify rod cells where the inner and outer segments are still attached to test this attractive hypothesis.