Photoreceptor GTP binding protein mediates fluoride activation of phosphodiesterase.

In this report, we show that fluoride activates dark-adapted rod outer segment phosphodiesterase, and that this activation is mediated, in analogy with adenylate cyclase, through a GTP binding protein. The GTP binding protein is released from dark-adapted rod outer segment membranes by exposure to fluoride and subsequent centrifugation. The 39-kilodalton subunit of the GTP binding protein, released from the membrane by this procedure, exhibits altered susceptibility to limited trypsin proteolysis, identical to that seen when hydrolysis-resistant GTP analogs are bound to that subunit. Repeated exposure of dark-adapted rod outer segment membranes to fluoride and subsequent centrifugation results in maximal activation of the membrane-bound phosphodiesterase. Thus, activation of phosphodiesterase by fluoride in the dark appears similar to fluoride activation of adenylate cyclase.

In this report, we show that fluoride activates darkadapted rod outer segment phosphodiesterase, and that this activation is mediated, in analogy with adenylate cyclase, through a GTP binding protein. The GTP binding protein is released from dark-adapted rod outer segment membranes by exposure to fluoride and subsequent centrifugation. The 39-kilodalton subunit of the GTP binding protein, released from the membrane by this procedure, exhibits altered susceptibility to limited trypsin proteolysis, identical to that seen when hydrolysis-resistant GTP analogs are bound to that subunit. Repeated exposure of dark-adapted rod outer segment membranes to fluoride and subsequent centrifugation results in maximal activation of the membrane-bound phosphodiesterase. Thus, activation of phosphodiesterase by fluoride in the dark appears similar to fluoride activation of adenylate cyclase.
Adenylate cyclase (EC 4.6.1.1.) and photoreceptor phosphodiesterase I (EC 3.1.4.1.) have been shown to be remarkably similar enzyme complexes (1-3). In both systems, activation of the catalytic moiety through light (for phosphodiesterase) and hormone or neurotransmitter (for adenylate cyclase) requires receptor aytivation, a GTP binding protein' (also called G protein, N protein, transducin, etc.), and GTP (or analogs). Activation of the enzyme by fluoride ion has been demonstrated in both the adenylate cyclase and phosphodiesterase systems as well (4, 5). However, other than demonstrations that GTP binding protein mediates adenylate cyclase activation by fluoride, the molecular events responsible for this process remain unclear. The single report of fluoride activation of photoreceptor phosphodiesterase shows only that the enzyme can be activated in dark-adapted rod outer segment 9081 (ROS') membranes (5).
Guanine nucleotide (GTP or hydrolysis-resistant GTP analog) is required for light activation of ROS phosphodiesterase (6). After illumination and the addition of GTP, the GTP binding protein complex is released from ROS membranes after centrifuguation (7,8). In order to understand the effects of fluoride on the photoreceptor phosphodiesterase cascade, we examined the release of the GTP binding protein from ROS membranes after incubation with fluoride and centrifugation. We also measured the effects of fluoride on the activation of phosphodiesterase. The data show that, in the dark, KF mimics the effects of guanyl nucleotide and light in the activation of phosphodiesterase and the release of the GTP binding protein from ROS membranes. Furthermore, limited trypsin proteolysis of the KF-released GTP binding protein indicates that KF, in the absence of light, causes a conformational change in the GTP binding protein similar to that induced by light and hydrolysis-resistant GTP analogs (%lo).

MATERIALS AND METHODS
Preparation of ROS-Retinas were dissected from dark-adapted (12-18 h) Bufo marinus and ROS were prepared as previously described (10). The harvested ROS were resuspended in 100 mM Tris (pH 7.3), 1 mM MgClz, and 5 mM dithiothreitol (Buffer A). An aliquot of the ROS suspension was solubilized in Emulphogene BC-720 and the absorption spectrum of rhodopsin was measured from 650-250 nm using a Shimadzu UV-3000 spectrophotometer. The rhodopsin concentration was calculated using 40.6 X lo3 cm2/mol as the molar extinction coefficient.

Release of GTP Bindiw Protein from
ROS-We assessed the association of the GTP binlding protein with both bleached and dark adapted ROS by monitoring the appearance of the GTP binding protein into solution after incubation of the ROS membranes (25-30 p~; 100-pl reaction volume) with KF or guanyl nucleotide followed by centrifugation. At the end of their incubation time, the membranes were centrifuged at 12,000 X g for 30 min. The supernatant solutions were recentrifuged for an additional 30 min under the same conditions to remove any traces of ROS membrane. Aliquots were removed, solubilized in 3% SDS Laemmli sample buffer (11) with 0.1 mM dithiothreitol. Samples were heated to 90 "C for 5 min and subjected to electrophoresis on 8-20% gradient gels or 12.5% gels (as previously described (10)). Gels or gel photographs were scanned with a Shimadzu Instruments scanning densitometer and peak areas were determined by a Shimadzu C1-A Integrator.
Trypsin Proteolysis of the GTP Binding Protein-ROS membranes were prepared as above and split into three aliquots. One aliquot was washed (in the dark) four times with Buffer A and then three times with Buffer A containing 2.5 mM KF. The KF extracts containing the GTP binding protein were pooled. The two remaining aliquots were bleached and washed four times with Buffer A and three times with Buffer A containing either GTP or GTPyS. The respective extracts were pooled. Aliquots of each GTP binding protein preparation containing approximately 10 pg of GTP binding protein were incubated with trypsin (14 pg/ml, final concentration). At 1, 5, 10, and 30 min, the proteolysis was stopped by the addition of soybean trypsin inhibitor (56 pg/ml, final concentration). After the addition of 3% SDS Laemmli sample buffer, the samples were analyzed on 12.5% polyacrylamide gels.
KF Effects on Phosphodiesterase Activity-To assess the effects of KF washing on membrane-bound phosphodiesterase activity, ROS membranes (prepared as above) were washed four times with Buffer reaction volume) were preincubated in the dark or bleached and preincubated for 4 min with KF or GTPyS in 125 mM KCI, 1 mM MgC12, 5 mM dithiothreitol, and 2.5 mM Tris, pH 8.0. A t the end of the preincubation period, 5 mM cyclic GMP was added and the rate of proton evolution (a direct indication of cyclic GMP hydrolysis (13)) was monitored using a pH electrode (MI 410, Microelectrodes Inc., Londonderry, NH) whose output was fed into a voltage follower (WP Instruments, New Haven, CT) and amplifier (WP Instruments, New Haven, CT) and recorded on a Model 220 Brush Recorder (Gould Instruments, Cleveland, OH). Trypsin Proteolysis of the GTP Binding Protein-Several previous studies (9,10) showed that the active (capable of activating phosphodiesterase) and inactive conformations of GTP binding protein may be distinguished by their digestion patterns during limited trypsin proteolysis. We employed this technique to analyze the conformation of GTP binding protein released with KF. Fig. 3 (Panels A and C) shows that trypsin digestion of the KF-released protein is similar to that generated for the protein with GTPrS bound. After release from the membrane with either KF or GTP$$ trypsin digestion eventually generates a 32-kilodalton fragment stable to further digestion. By contrast, when GTP binding protein is extracted with GTP and then subjected to trypsin proteolysis the digestion proceeds past the 32-kilodalton stage to generate fragments of 23 and 12 kilodaltons (Fig. 3, Panel B). Thus, limited trypsin proteolysis indicates that the conformation of GTP binding protein, when released by KF in the dark, is C. Fluoride Actiuation of Phosphodiesterase-To assess the influence of fluoride on phosphodiesterase activity, we measured cGMP hydrolysis in dark-adapted ROS membranes exposed to buffers containing different KF concentrations. Fig. 4 shows that phosphodiesterase activity is optimally stimulated, in the dark, at 5.0 mM KF. Half-maximal stimulation appears to occur at about 1 mM KF. It is interesting to note that, at 10.0 mM KF, enzyme activity is reduced. Inhibition of the catalytic moiety by fluoride3 may be responsible for this effect. Similarly, adenylate cyclase from rat cerebral cortex synaptic membranes is maximally stimulated by 18 mM NaF while concentrations above 25 mM are inhibitory (15).

Fluoride Releases GTP
To analyze the mechanism by which phosphodiesterase is activated by fluoride, we washed dark-adapted ROS membranes with Buffer A containing either 2.5 mM KC1 or KF. The activity of phosphodiesterase which (unlike the GTP binding protein) remains membrane-bound was then assayed in the light. Table I shows

DISCUSSION
Incubation of dark-adapted ROS membranes with KF containing buffer results in a change in the conformation of the GTP binding protein such that it is capable of activating phosphodiesterase. This change is also indicated both by the KF concentration-dependent release of the protein from the ROS membrane and by the fact that the trypsin digestion pattern resembles that of the GTP binding protein with hydrolysis-resistant guanine nucleotide analog bound. Since, in this state, the protein is in a conformation capable of activating phosphodiesterase, it appears that fluoride activation of phosphodiesterase is mediated by the GTP binding protein. This idea is supported by reconstitution experiments which show that the fluoride-solubilized GTP binding protein can be added back to dark-adapted ROS membranes and activate the enzyme? Both fluoride and guanine nucleotides release the GTP binding protein from ROS membranes. However, fluoride releases the protein most efficiently in dark ROS membranes while guanine nucleotides are most efficient in bleached ROS membranes. Washing dark-adapted ROS membranes after exposure to fluoride activates phosphodiesterase. The mechanism of activation appears to involve release of an inhibitory protein from the catalytic moiety, since a similar mechanism was recently demonstrated to account for light-dependent guanyl nucleotide activation of phosphodiesterase (12). However, fluoride activation occurs in dark-adapted membranes while guanyl nucleotide activation requires bleaching. Thus, the light-dependent rhodopsin-GTP binding protein interaction required for guanyl nucleotide activation of phosphodiesterase is apparently not required for fluoride activation of this enyzme. This parallels observations made for adenylate cyclase where hormone (and hormone receptor) is not necessary for fluoride activation of the enzyme (14, 15). Furthermore, fluoride activation of the adenylate cyclase stimulatory GTP binding protein (N.) has been shown to resemble activation by GTP analogs (17). The trypsin digestion experiments reported above demonstrate that the GTP binding protein released by fluoride or GTPyS generate identical proteolytic fragments and therefore have similar conformations. Since photoreceptor GTP binding protein can substitute for adenylate cyclase GTP binding protein in the activation of the catalytic moiety (16), it appears likely that a similar conformation change must occur in the adenylate cyclase GTP binding protein during the activation process.