ABSTRACT Cyclic guanosine monophosphate (CGMP) gated channels serve as the downstream target in vertebrate visual transduction. However, it is not known whether cGMP-gated channels play a similar role in photoreceptors from invertebrates. Inside-out patch clamping of solitary photoreceptor cells isolated from the retina of the squid (Loligo opalescens) revealed that the probability of channel opening increases from a baseline of Po = 0.20 in the absence of cGMP to near saturation (Po = 0.91) at 100 uM cGMP, suggesting the existence of cGMP-gated channels. The probability of channel openings was independent of membrane potential over the range -60 to +60 mV. Data from these macroscopic patches revealed multiple channel types with unitary conductances ranging from 10 to 125 pS in symmetrical Nat solutions. Conductance of a 100 pS channel was blocked by the addition of 2 mM Ca2 to the inner face of the patch. However, channel activity persists in the presence of 60 mM Mg2-, a divalent cation concentration sufficient to block cGMP-gated channels in vertebrate photoreceptors. This suggests a significant difference in vertebrate and invertebrate cGMP-gated channel structure. The results of this study indicate that cGMP-gated channels are the effectors in invertebrate phototransduction. INTRODUCTION Phototransduction in vertebrate retinal rod cells involves an enzyme cascade in which the second messenger cyclic guanosine monophosphate (cGMP) directly controls the gating of cation channels in the cell membrane (Fensenko et al., 1985) to regulate the photoreceptor potential. The transduction cascade begins when light isomerizes the chromophore of rhodopsin (11-cis retinal) and activates the GTP-binding protein, transducin. Activated transducin activates CGMP phosphodiesterase, which lowers the concentration of cGMP in the cell, leading to the closing of cation channels and hyperpolarization of the photoreceptor (Stryer, 1986). Significant homology exists between the squid and vertebrate phototransduction cascades, although it has been shown that light causes depolarization, not hyperpolarization, of squid (Loligo pealei) photoreceptor cells (Pinto and Brown, 1977; Nasi and Gomez, 1992) and it is generally accepted that invertebrate photoreceptors depolarize in response to light. Studies have shown that the rhodopsin and GTP-binding protein from the vertebrate and squid systems cross-react (Saibil and Michel-Villaz, 1984). However, details of the events between rhodopsin activation of GTP- binding protein and depolarization of the invertebrate photoreceptor cell are not known. Cyclic GMP has been implicated as a second messenger in the squid phototransduction cascade by studies demonstrating that exposure to a flash of light increases the cGMP concentrations in squid photoreceptor homogenates (Saibil, 1984) and pieces of squid retina (Johnson et al., 1986). More recently, this effect was called into question when the results could not be reproduced in replicate experiments (Brown et al., 1992) and other research suggested that cGMP levels do not change in octopus retinal cells following a flash of light (Seidou et al., 1993). The presence of cGMP-gated ion channels in the squid photoreceptor cell would support the idea that cGMP is a second messenger in this system; by analogy with the vertebrate system, such a channel might serve as the downstream target responsible for the depolarization in response to light. This idea is supported by the discovery of a cGMP-activated channel in the light-sensitive lobe of Limulus ventral photoreceptors (Lisman et al., 1992; Bacigalupo et al.. 1992; Johnson and Bacigalupo, 1992). Until now however, the search for such a channel in squid has been hampered by the difficulty of obtaining solitary photoreceptor cells, although one bilayer reconstitution experiment has indicated that squid photoreceptors contain a cGMP-gated channel (Nasi and Gomez, 1990). A recently developed protocol for dissociation of squid retina was used to obtain healthy solitary photoreceptor cells (Nasi and Gomez, 1992) for patch clamp experiments. This report describes the properties of cGMP-gated channels in inside-out patches. MATERIALS AND METHODS Isolation of photoreceptors Photoreceptors were isolated according to a procedure adapted from Nasi and Gomez (1992). Reagents were obtained from Sigma unless otherwise specified. Squid (Loligo opalescens) obtained in Monterey Bay, CA, were dark adapted for at least 30 min before use. All manipulations of photoreceptors up to the mechanical trituration step were performed in darkness or dim red light (Kodak Safelight Filter no. 1A). Squid were decapitated and the head quickly immersed in ice cold normal artificial seawater (ASW): (mM) 460 NaCl, 10 KCl, 10 CaCl,, 50 MgClz, 10 HEPES (pH 7.8). Eyes were dissected out and transferred to ice cold oxygenated calcium-free artificial seawater (Ca2--free ASW): (mM) 460 NaCl, 10 KCl, 60 MgCl,, 10 HEPES (pH 7.8). The sclera was peeled from the back of the eye and the retina cut out. The retina was transferred to clean ice-cold Ca2--free ASW and cut into strips of approximately 1 mm x 5 mm, which were transferred to a test tube containing 800-1000 units/ml Pronase (Calbiochem, San Diego, CA) in Ca2--free ASW and incubated for 1 hour in a refrigerator at 3-4°C. The Pronase solution was freshly made 10-15 min before use. Äfter incubation, the pieces of retina were washed twice with 2 ml aliquots of 3% bovine albumin in normal ASW. Dissociation of photoreceptor cells was then performed by gentle mechanical trituration with a pulled and fire- polished Pasteur pipette in ice cold 100uM leupeptin and 50 uM pepstatin in Ca2--free ASW. An aliquot of cells was transferred to a recording chamber pre-treated with poly-(D)-lysine to facilitate cell adhesion. Inside-out patch recording Macroscopic currents were recorded in the inside-out patch configuration from patches obtained from the microvillar region and cell body of isolated photoreceptor cells. The recording chamber was maintained at 15°C. Patch pipettes were pulled from fiber-filled borosilicate capillary tubing (Sutter Instruments Co.) just before experiments and coated near to the tip with Silgard elastomer (Dow Corning). The standard extracellular medium and pipette solution was Ca2+free ASW. Na-cGMP solutions applied to patches were prepared by serial dilution from a 10 mM stock solution in distilled water. Cyclic GMP solutions were added to the recording chamber via a gravity perfusion system or by pipetting the solutions directly into the chamber. In calcium block experiments, both pipette solution and extracellular medium were divalent free Nat saline containing (mM) 520 NaCl, 10 EGTA, 10 HEPES (pH 7.8). Recordings from macroscopic patches were made using an Axopatch 200A amplifier. Data analysis was done using PCLAMP (Axon Instr.) and DEmpsTER software. Records were filtered before digitization using a 8-pole bessel filter (Frequency Devices). RESULTS Morphological characteristics of solitary cells Intact, solitary photoreceptor cells and cell clusters were obtained in high yield. Immediately after dissociation, the cells were long (up to 400 um) and rod-shaped, with the cell body, axon, and distal segment clearly visible as described in Nasi and Gomez (1992). Microvilli were visible at the end of the distal segment with DIC optics. Within 15-30 min after dissociation, the cells began to shorten while retaining a rod-shaped morphology (Fig. 1) or rounded up (Fig. 2). The change in morphology was not prevented if the cells were kept on ice. Commonly, the cells were observed to shorten from both ends, resulting in a dumbbell-shaped cell with spheres forming from the axon/cell body region and the distal segment (Fig. 2). Good recordings were obtained from the cells and cell clusters (Fig. 3) up to three hours after dissociation. The addition of divalent free Nat saline solution to the extracellular medium was observed to speed cell shortening. Cyclic GMP increases open probability of inside-out patches The existence of a cGMP-gated channel was detected in macroscopic inside-out patches obtained from the cell body region of photoreceptor cells. Brief examination of each patch while still cell-attached showed that all patches contained some voltage stimulated channel activity Recordings were performed with Ca2-free ASW on both sides of the membrane. It was observed that channel openings in some patches were greatly increased upon the addition of cGMP to the inner membrane face. In the example in Figure 4, the voltage across the patch was 20 mV, pipette negative. The inner membrane face was continuously perfused with Ca2--free ASW containing various concentrations of cGMP. In the control, the perfusion solution contained no cGMP and few channel openings were observed (Fig. 4A). The addition of 10 uM cGMP activated multiple channels in the patch (Fig. 4B). Dose response data was obtained by perfusing the cytosolic face of the patch with Ca2--free ASW containing 1, 10, and 100 uM cGMP, switching back to a control solution of Ca2--free ASW without cGMP between different cGMP concentrations. Membrane voltage was held constant at either 20 or 40 mV, pipette negative. The open probability (Po) of the patch was calculated by dividing mean open time by mean closed time, using 50% of the smallest unitary conductance measured directly from the current recording as the criterion for channel openings. Open probability of the patch reversibly increased from a basal Po of 0.20 in the absence of cGMP to 0.91 in the presence of 100 uM cGMP on the inside face of the patch. A plot of open probability versus cGMP concentration shows that the increase in open probability tends toward a saturating value, with open probability increasing rapidly at low cGMP concentrations and approaching a probability of 1 as cGMP concentrations increase (Fig. 5). Voltage independence of patch open probability Experiments were conducted with Ca2--free ASW on both sides of patches obtained from the microvillar region and cell body. Saturating concentrations of cGMP were applied to the inner face of the patch. As the patch was depolarized, patch open probability showed no significant voltage dependence. Figure 6 shows data from several different patches plotted as patch open probability versus membrane voltage. A least squares linear regression fit of the data points (Microsoft Excel) is shown in the figure. Unitary conductance analysis of macroscopic patches Analysis of the unitary conductance of channels measured directly from current amplitude recordings revealed that patches contained multiple channel types. Scatter plots of conductances were made such that clusters of data points indicated the approximate conductance of specific channels within the patch. In the example shown in Figure 7, data was recorded with Ca2--free ASW on both sides of the patch from the cell body and a pipette potential of-40 mV. When no cGMP was present, data points were clustered around a conductance of 19 pS (Fig. 7A). Upon addition of 10 uM cGMP to the inside face of the patch, a second set of data points appeared clustered around a conductance of approximately 14 pS (Fig. 7B). This indicates the existence of at least two channels in the patch, a 14 pS cGMP-channel and a 19 pS non-cGMP-gated channel. Multiple channels of the same type were also observed within patches. During an experiment performed in divalent free Nat saline on both sides of the membrane with 200 uM cGMP applied to the inside membrane face and a pipette potential of -46 mV, up to four channels of the same conductance (approximately 65 pS) were seen opening at the same time (Fig. 8). Channel openings longer than 100 mS were common at this high cGMP concentration. Ca't block of channels Calcium ions were shown to block at least one channel type with a conductance of approximately 100 pS. The experiment was performed with divalent free Nat saline on both sides of a patch from the microvillar region and a voltage of 40 mV, pipette positive. Saturating cGMP was applied to the inside face of the patch. The channel exhibited long openings and brief closures (Fig. 9A). In part B, a 2 mM CaCl, solution in Nat saline was added directly to the recording chamber, resulting in complete channel block. Part C also shows channel block by Ca2-. While voltage was increased in a ramp from -50 to +50 mV, the patch consistently displayed higher conductance in the absence of Ca2- than in the presence of Ca2-. Without Ca2, the trace shows that channels remain closed for the duration of the sweep whereas a short closure in the Ca2 free trace at about -50 mV shows that the channel remains open most of the time in the absence of Ca2- The closed state conductance of Ca2 free patch is lower than the closed state conductance with Ca2t, possibly because calcium improves the seal resistance. These current recordings were not adjusted for leak through the patch and thus would not account for variations in the seal resistance. Channel openings persist in the presence of high Mg2- Both dose response and voltage independent data were recorded in the presence of 60 mM Mg2t, a concentration that would completely block conductance in bovine rod cGMP-gated channels (Haynes, 1995; Sesti, 1994). The persistence of channel conductance under these conditions indicates that squid ion channels are not sensitive to Mg2- block. DISCUSSION Thave shown that the open probability of macroscopic patches of squid photoreceptor cell membrane increases with cGMP concentration. The data, along with the fact that the patch dose- response curve saturates as cGMP is raised, indicate that there is at least one cGMP-gated channel in the squid photoreceptor membrane. There are three possible explanations for the observation that open probability is not zero the absence of cGMP: 1) multiple channel types exist in the patch, including non-cGMP-gated channels which open independent of cGMP concentration, 2) the preparation and patch clamping process irreversibly damaged the cells and allowed cGMP to accumulate externally, and 3) the cGMP-gated channels in the patch are modulated by cGMP, but have a non-zero open probability in the absence of cGMP. Multiple channels within the patches made data difficult to analyze. Conductance of each different channel type could not be definitively determined from the few patches available for analysis. However, as shown in Figure 6A, there is clear indication of the presence of non- CGMP-gated channels in the patch. Non-zero open probability in the absence of cGMP may have also been due to spontaneous openings caused by exposure of the inside-out patches to Ca-free ASW. Experiments on inside- out patches from Limulus ventral photoreceptors showed that brief exposure of excised patches to ASW altered cGMP-gated channel behavior such that the patches exhibited spontaneous openings when examined in a Nat free solution (Bacigalupo et al., 1991). These observations are consistent with the increased rate of cell shortening and rounding seen in squid photoreceptors when divalent free Nat saline is added to the extracellular medium. The shortening and rounding of the photoreceptors also suggests that some damage may have occurred to the cells during the dissociation procedure. Nasi and Gomez (1992) reported that the cells were highly susceptible to deterioration upon exposure to temperatures above 15°C. In these experiments, the cells were maintained at a temperature just at or below 15°C, so it is possible that cell shortening and rounding was caused by high temperatures. Slices of squid retina are also known to be very sensitive to hypoxia (Pinto and Brown, 1977). The in vivo sensitivity to hypoxia is easily understood because of the high density of cells in the retina, but it is not known whether dissociated cells still have an unusually high oxygen requirement. Since freshly oxygenated extracellular solution was not used after cell dissociation due to the complexity of the perfusion system, membranes may have been altered at the relatively low atmospheric partial pressure of oxygen. The non-zero open probability of the patch in the absence of cGMP may also be an inherent characteristic of the cGMP-gated channels themselves. The cGMP-gated channels observed in these experiments might be members of a subclass of cyclic nucleotide-modulated channels. In such channels, cGMP binding is not obligatory for opening, but instead increases open probability (Yau and Chen, 1995). A cAMP-modulated Kt channel has been observed in larval Drosophila muscle (Delgado et al., 1991) that has a low, voltage independent, basal probability of opening that is increased by cAMP. The probability of opening for the squid photoreceptor cGMP-gated channel could be similarly modulated by CGMP. No significant voltage dependence of probability of opening was observed in the patches. However, it is possible that voltage dependence was masked by the changing activity of multiple types of channels within one patch. Every patch contained voltage-stimulated channel activity while cell-attached and ripped-off. This is in contrast to previous experiments performed in the dark in which most patches did not contain voltage-stimulated activity (Nasi and Gomez, 1992). Nasi and Gomez observed a few patches which required concurrent light and voltage stimulation before channel openings were observed, which could explain the differences in channel activity observed in the light and dark. The calcium blockage of the 100 pS channel in Figure 8 suggests some similarity between this channel and vertebrate rod cGMP-gated channels, which are also blocked by calcium (Haynes 1995, Sesti 1994). Ca2t block may provide a possible mechanism for recovery from the light response and light adaptation as it does in vertebrate rods. However, the 100 pS channel cannot be definitively identified as a cGMP-gated channel because no control recordings in the absence of cGMP were obtained from this patch. In contrast, other channels, in particular the 14 pS CGMP gated channel, do not exhibit Mg2t block. This is similar to other invertebrate cyclic nucleotide- gated channels (Bacigalupo et al., 1991; Baumann et al., 1994; Christy, 1996). Lack of Mg2- block suggests that the structure of squid cGMP-gated channels differs significantly from the vertebrate rod cGMP channels, which would be completely blocked at 60 mM Mg2¬ concentrations. In the context of the controversy surrounding whether cGMP is an internal messenger in the squid phototransduction, the data from this experiment supports the hypothesis that increased levels of cGMP are responsible for the depolarization of a photoreceptor in response to light and that cGMP-gated channels are the effectors. To obtain more compelling evidence for this model, it is necessary to definitively determine the cellular localization of these cGMP-gated channels. The shortening and rounding of dissociated photoreceptors raises the question of whether patches were truly taken from the microvillar and cell body regions of the photoreceptor since the cell shape changed. The possibility of lateral movement of ion channels within the lipid bilayer during these morphological changes also exists. Localization of these channels to different areas of the cell membrane could have important implications on their cellular function. Further experimentation to obtain single-channel patches is needed to better characterize these cGMP-gated channels, determine their similarity to the light-dependent channels, and conclusively determine how many types of cGMP-gated channels exist in the squid photoreceptor. ACKNOWLEDGMENTS thank Taylor Liu, Chris Mathes, Thomas Preuss, and Matthew McFarlane for their advice and help in obtaining squid, and Stuart Thompson for his inexhaustible energy and phenomenal support. REFERENCES Bacigalupo, J., Johnson, E.C., Vergara, C., Lisman, J.E. (1991). Light-dependent channels from excised patches of Limulus ventral photoreceptors are opened by cGMP. Proc. Natl. Acad. Sci. USA 88, 7938-7942. Baumann, A., Frings, S., Godde, M., Seifert, R., Kaupp, U.B. (1994). Primary structure and functional expression of a Drosophila cyclic nucleotide-gated channel present in the eyes and antennae. EMBO J. 13(21), 5040-5050. Christy, I. (1996). Cyclic nucleotide and IPz gated ion channels in hermit crab olfactory receptor neurons. Stanford University Hopkins Marine Station Final Papers Biology 175H 1996. Delgado R., Hidalgo, P., Diaz, F., Latorre, R., Labarca, P. (1991). A cyclic AMP-activated K¬ channel in Drosophila larval muscle that is persistently activated in dunce. Proc. Natl. Acad. Sci. USA 88, 557-560. Fensenko, E.E., Kolesnikov, S.S., Lyubarsky, A.L. (1985). Induction by cyclic GMP of cationic conductance in plasma membrane of retinal rod outer segment. Nature 313,310-313. Haynes, L.W. (1995). Permeation and block by internal and external divalent cations of the catfish cone photoreceptor cGMP-gated channel. J.Gen. Physiol. 106, 507-523. Johnson, E.C., Robinson, P.R., Lisman, J.E. (1986). Cyclic GMP is involved in the excitation of invertebrate receptors. Nature 324, 468-470. Nasi, E. and Gomez M. (1992). Electrophysiological recordings in solitary photoreceptors from the retina of squid, Loligo pealei. Visual Neurosci. 8, 348-359. Nasi, E. and Gomez M. (1992). Recording from solitary photoreceptors and reconstituted rhadbomeric membranes of the squid. Biophys. J. 57, 368a. Pinto, L.H., and Brown, J.E. (1977). Intracellular recordings from photoreceptors of the squid (Loligo pealii). J. Comp. Physiol. 122, 241-250. Saibil, H.R. (1984). A light-stimulated increase of cyclic GMP in squid photoreceptors. FEBS Lett. 168(2), 213-216. Saibil, H.R., Michel-Villaz, M. (1984). Squid rhodopsin and GTP-binding protein crossreact with vertebrate photoreceptor enzymes. Proc. Natl. Acad. Sci., USA 81, 5111-5115. Seidou, M., Ohtsu, K., Yamasita, Z., Narita, K., Kito, Y. (1993). The nucleotide content of the octopus photoreceptor cells: no changes in the octopus retina immediately following an intense light flash. Zool. Sci. (Tokyo) 10, 275-279. Sesti F., Straforini M., Lamb T.D., Torre, V.(1994). Gating, selectivity and blockage of single channels activated by cyclic GMP in retinal rods of the tiger salamander. J. Physiol. 474, 203- 222. Stryer, L. (1986). Cyclic GMP cascade of vision. Annu. Rev. Neurosci. 9, 87-119. Yau, K.W., Chen, T.Y. (1995). Cyclic nucleotide-gated channels. In Handbook of Receptors and Channels: Ligand-and voltage-gated ion channels (ed. R.A. North), pp. 307-335. Boca Raton: CRC Press. FIGURE LEGENDS Figure 1. An enzymatically isolated solitary squid photoreceptor which has shortened to a length of approximately 100 um. The cell body can be seen as a rounded region with a short section of axon below it and a long, rod-shaped distal segment above it. The microvillar region, located at the apical end of the distal segment, is seen as a less clearly defined region of the cell membrane. Figure 2. Rounded up photoreceptor cells. The length of the longest cell shown is approximately 40 um. One dumbbell-shaped cell is visible (upper right corner) with two spherical regions at either end of the cell connected by a short region of distal segment. Figure 3. A cluster of partially shortened photoreceptor cells. Cells are attached below the cell body region. Figure 4. Cyclic GMP increases channel openings in inside-out membrane patches. A. Current record extracted from a continuous recording immediately after patch has been ripped off into an extracellular solution of Ca2--free ASW: (mM) 460 NaCl, 10 KCl, 60 MgCl,, 10 HEPES (pH 7.8). Pipette contained Ca2+free ASW also. Pipette voltage clamped to-20 mV. None or few channel openings. B. Current record extracted from a continuous recording from the same patch as in A after inside membrane face exposed to 10 uM cGMP. Pipette voltage remained clamped to -20 mV. Multichannel openings shown as downward deflections. Filtered with an 8-pole bessel filter with 1 kHz cutoff frequency. Figure 5. Increase in patch open probability with increasing cGMP concentrations applied to inside face of the same inside-out patch. Both sides of patch were exposed to the same Ca2+-Free ASW: (mM) 460 Nacl, 10 KCI, 60 MgCh, 10 HEPES (pH 7.8). Patch open probability calculated using 50% of the smallest unitary conductance measured directly from current recordings as open criterion. Figure 6. Patch open probability shows no significant voltage dependence. Both sides of patch were exposed to the same Ca2+-Free ASW: (mM) 460 NaCl, 10 KCl, 60 MgCl,, 10 HEPES (pH 7.8). Saturating cGMP concentrations applied to the inner membrane face. Data compiled from several different patches. Least squares linear regression fit performed (Microsoft Excel). Figure 7. Unitary conductance analysis indicating 19 pS non-cGMP-gated channel and 14 pS cGMP-gated channel. Scatter plots of conductance against conductance reveal clusters of points around channel conductances. Both sides of membrane exposed to same Ca2-Free ASW: (mM) 460 NaCl, 10 KCl, 60 MgCl, 10 HEPES (pH 7.8). Voltage across the patch was 40 mV, pipette negative. A. Points clustered around 19 pS in the absence of cGMP. B. Same patch with 10 uM CGMP added to the inner membrane face has a second cluster of points appear at approximately 14 pS. Figure 8. Up to four 65 pS channels opening at the same time in a patch with divalent free Nat saline on both sides: (mM) 520 NaCl, 10 EGTA, 10 HEPES (pH 7.8) and 200 uM cGMP on the inner face. Pipette clamped to -46 mV. Multiple sweeps are superimposed. Each additional channel opening increases current by an integer amount of approximately 3 pA. Number of channels open is indicated (arrows). Filter cut-off frequency was 1 kHz. Figure 9. Ca2t block of a 100 pS channel. Divalent free Nat saline on both sides of patch: (mM) 520 NaCl, 10 EGTA, 10 HEPES (pH 7.8), saturating cGMP applied to inner membrane face. Membrane voltage was 40 mV, pipette positive. A. Long openings and short closures occurred in the absence of Ca2t. Dotted line is zero current level. Filter cut-off frequency was 1 kHz. B. All openings blocked when 2 mM CaCl, solution in Nat saline was added to inner membrane face. C. Patch conductance decreased with the addition of 2 mM CaCl, to inside face of patch as evidenced by flatter slope in the presence of calcium. Voltage is varied in a ramp from -50 mV to +50 mV. A control mpjvimngmnge B 10 uM CGMP M 250 msec wp 0.9 0.8 0.7 0.6 g0.5 0.4 0.3 0.1 Po vs. CGMP concentration 60 80 20 ICGMPI (UM) 100 Po vs. voltage at saturating CGMP 0.9 y = 0.0014x + 0.4254 0.8 R2 = 0.088 0.7 0.6 0.5 0.3 0.2 0.1 -60 -40 -20 40 20 Voltage (mV) 60 7 . . A 2 10 10 5 —— 10 Unitary conductance at control, -40mV conductance (pS) Unitary Conductance at 1OuM cGMP, —40mV conductance (p3) u2 zu — 20. 0 V 2 1.* A B CGMF svn Mr V ------------------ -----——----- CGMP + Ca gnebeen poblgng p 25 ms 40 CGMP PA CGMP+Ca 50 50 mV -40