Timing of Ca²⁺ Release from Intracellular Stores and the Electrical Response of Limulus Ventral Photoreceptors to Dim Flashes

Microvillar photoreceptors of invertebrates use the phophoinositide cascade to couple the absorption of light by rhodopsin to the activation of ion channels in the microvillar plasma membrane (Ranganathan et al. 1995; Dorlöchter and Stieve 1997). The phosphoinositide cascade produces several diffusible messengers, including inositol (1,4,5) trisphosphate (InsP₃), diacylglycerol, and Ca²⁺ ions, the latter being released from intracellular stores by the action of InsP₃ (Berridge and Irvine 1989). In this paper, we describe a rapid light-induced release of Ca²⁺ from intracellular stores in the giant ventral photoreceptor of Limulus polyphemus, which is necessary for a highly amplified electrical response to dim flashes and which limits temporal acuity.

The role of light-induced Ca²⁺ release in excitation of invertebrate photoreceptors is uncertain and may depend on the species studied. In Drosophila photoreceptors, light-induced release of Ca²⁺ has been observed (Arnon et al. 1997), but the light-induced elevation of intracellular free calcium ion concentration (Ca_i) results mostly from an influx through the light-sensitive channels, which may therefore be activated by a messenger other than Ca²⁺ ions or InsP₃ (Ranganathan et al. 1994; Peretz et al. 1994; Acharya et al. 1997; Chyb et al. 1999). For Limulus ventral photoreceptors, however, light-induced elevation of Ca_i is dominated by the release of Ca²⁺ from intracellular stores (Brown and Blinks 1974), which may be necessary to maintain the sensitivity of visual excitation (Bolsover and Brown 1985). There is also evidence that release of Ca²⁺ excites Limulus ventral photoreceptors. Intracellular pressure injection of InsP₃ or Ca²⁺ ions in darkness activates a conductance with a reversal potential and relative permeability to sodium ions similar to that of the light-sensitive conductance (Fein et al. 1984; Brown et al. 1984; Payne et al. 1986). Rapid release of Ca²⁺ by flash photolysis of intracellular caged InsP₃ or caged Ca²⁺ activates this conductance within a few milliseconds (Ukhanov and Payne 1997). Pharmacological blockade of light-induced Ca²⁺ release greatly diminishes the electrical response to dim or moderately intense light flashes and slows the onset of the electrical response at all intensities (Payne and Fein 1986; Frank and Fein 1991; Shin et al. 1993; Nagy and Contzen 1997; Johnson et al. 1998). Cyclopiazonic acid (CPA), an agent that depletes intracellular Ca²⁺ stores (Mason et al. 1991), eliminates the highly amplified response to single quanta (Dorlöchter et al. 1999).

Measurement of Ca²⁺ at spots within a few microns of the photosensitive membrane by confocal fluorescence microscopy has permitted the detection of elevations of Ca_i that precede the electrical response to very intense focal stimulation (Ukhanov and Payne 1995). We now investigate similarly rapid elevations of Ca_i to moderate and dim flashes, for which the pharmacological evidence of a role for Ca²⁺ release in excitation is most convincing.

Ventral optic nerves were dissected as described by Clark et al. 1969 and placed in artificial sea water (ASW) containing (mM): 435 NaCl, 10 KCl, 20 MgCl₂, 25 MgSO₄, 10 CaCl₂, 10 HEPES, pH 7.0. 0 Ca-ASW was prepared as for ASW, but with no CaCl₂ and 1 mM EGTA added. For some experiments, cells were chemically bleached to reduce their quantum efficiency (Faddis and Brown 1992). A bleaching solution containing (mM) 200 hydroxylamine chloride (Sigma-Aldrich), 235 NaCl, 10 KCl, 20 MgCl₂, 25 MgSO₄, 10 CaCl₂, 10 HEPES, adjusted to pH 6.5 with 10 N NaOH and cooled to 4°C. Nerves were rinsed with 25 ml of bleaching solution, and then immediately exposed to 10 min of intense white light at 4°C. Nerves were then washed five times at 10-min intervals with 50 ml of ASW at 4°C. Nerves were bleached in this manner four times with at least 8 h between bleaches, with the intervals spent at 4°C in a modified culture medium (Kass et al. 1988, modified from Bayer and Barlow 1978). This treatment reduced the quantum efficiency of the bleached photoreceptors by ∼3 log₁₀ U. Cyclopiazonic acid (Calbiochem) was stored as a 25-mM stock solution in DMSO, which was diluted to 100 μM (0.4% DMSO) in ASW immediately before use.

Membrane potential was recorded through the micropipette used for pressure-injecting dye solutions. For some experiments, cells were impaled with a second electrode filled with 3 M KCl, through which current was supplied from a voltage-clamp amplifier (Axoclamp2A; Axon Instruments, Inc.).

Ventral nerves were viewed with an LSM 410 laser-scanning, inverted, confocal microscope (Carl Zeiss, Inc.) equipped with a 488-nm Ar-Kr laser (Uniphase Inc.) attenuated 300-fold and focused through a Neofluar ×40/0.75 objective lens (for details, see Ukhanov and Payne 1995). A high-speed shutter (Uniblitz No. 26L; Vincent Associates) was placed in the path of the laser beam to control the timing and duration of flashes. To generate a dim flash, followed by a bright one, Kodak Wratten neutral density filters were placed in the path of the laser, attached to the mechanism of a second high-speed shutter that was capable of removing the filters from the light path within 10 ms. Illumination not subject to this filtering is referred to in the text as “unattenuated.”

Photoreceptors were impaled with micropipettes containing 500 μM of the fluorescent Ca²⁺ indicator dyes, fluo-4 or Oregon green-5N (Molecular Probes, Inc.; dissociation constants determined to be 2 and 18 μM, respectively, in 400 mM KCl, 10 mM MOPS, pH 7.0) in a carrier solution containing 100 mM potassium aspartate, 10 mM HEPES, pH 7.0. 5–15 pressure injections of 1–10 pl were delivered into the cells. After loading the cells with dye, the site of origin of the Ca²⁺ release was determined. Photoreceptors were first viewed in the transmission mode of the microscope so as to determine the likely position of the light-sensitive rhabdomeral lobe (R-lobe). The position of the R-lobe was then confirmed by performing a line scan of fluorescence along the longitudinal axis of the cell. If the R-lobe was correctly identified, then, upon opening the shutter covering the laser beam, a delayed wave of increased fluorescence spread into the cell from the point at which the scanning line intersected the edge of the R-lobe (Ukhanov and Payne 1995). To reduce noise, scans were spatially filtered, limiting to 4 μm the resolution of the site at which light-induced Ca²⁺ release originated. Subsequent measurements of the time course of Ca²⁺ release were made at this point of intersection at the edge of the R-lobe in the microscope's spot mode, with the scanning movement of the laser beam halted.

Light intensities are expressed as “effective photons.” One effective photon is the light energy required to photo-isomerize one rhodopsin molecule and so to elicit one quantal electrical response, a “quantum bump,” from a dark-adapted photoreceptor (Yeandle and Spiegler 1973). The laser beam could therefore be calibrated by counting the number of quantal events per second generated by a photoreceptor during continuous illumination when the beam was attenuated by a known amount. Using this calibration method, the unattenuated laser was estimated to deliver ∼10⁸ effective photons/s to ventral photoreceptor cells that were not chemically bleached and ∼10⁵ effective photons/s to chemically bleached cells.

Autofluorescence from cells not injected with indicator dye was observed to be negligible. Photomultiplier records of dye fluorescence, F, from injected cells were used to determine the mean fluorescence, F_rest, during the latent period of the light response and the change in fluorescence, ΔF = F − F_rest, during the light response. For each cell, the saturated peak fluorescence, F_max, during illumination by the unattenuated laser was also recorded. On the assumption that the dye is saturated by the peak elevation of Ca_i that follows flashes delivered by the unattenuated laser (see results), Ca_i can be estimated using the standard equation (Grynkiewicz et al. 1985):

where K_d is the dissociation constant of the dye and F_min is the fluorescence of the indicator with no Ca²⁺ bound. Two assumptions were adopted to obtain F_min. For the low affinity indicator, Oregon green-5N, the dye was assumed to bind negligible Ca²⁺ at resting Ca_i, but to have a comparatively significant calcium-independent fluorescence. F_min is then equal to F_rest. In terms of the fluorescence changes recorded following illumination, becomes :

or :

where ΔF_max = F_max − F_rest.

For the high affinity indicator, fluo-4, calcium-independent fluorescence was assumed to be negligible compared with the fluorescence due to resting Ca_i. F_min is then equal to zero. In terms of the fluorescence recorded after illumination, becomes :

The protocol used for reconstructing elevations of Ca_i after dim flashes is similar to that adopted by Gray-Keller and Detwiler 1996 in their study of vertebrate rod photoreceptors, with the exception that all of the measurements for a given dim flash intensity could be made on a single cell, since photobleaching of visual pigment does not occur in Limulus photoreceptors (Lisman and Sheline 1976). The method takes advantage of the latent period that precedes the elevation of Ca_i and electrical response. Dim flashes do not elicit sufficient dye fluorescence to enable measurement of Ca_i. However, a dim flash can be paired with a subsequent bright flash delivered by the unattenuated laser. Dye fluorescence sampled during the 20–30-ms latent period of the response to the bright flash is assumed to reflect the [Ca²⁺] that would have prevailed during the response to the dim flash alone. The latent period of the response to the bright flashes can be defined as the time at which receptor potentials following the bright flashes (Fig. 1 A, 1–5) depart significantly from the response to the dim flash alone. This deviation from the response to the dim flash occurred after the bright flash was over, so that fluorescence was only collected during the latent period.

By varying the time between dim and bright flashes, the time course of the elevation of Ca_i that followed the dim flash could be reconstructed (Fig. 1), provided adequate time for dark adaptation was allowed between the paired flashes. To extend the time-window during which measurements of Ca_i could be made, the latent period of the response to the bright flash could be extended to 70–150 ms by lowering the quantum efficiency of the cell through chemical bleaching of rhodopsin. In comparing the waveforms of the reconstructed calcium signals and the receptor potential in Fig. 2, Fig. 4, Fig. 6, and Fig. 10, it is important to note that the receptor potential shown is a representative response to the dim flash alone, recorded between presentations of the dual flashes. As Fig. 1 illustrates, there was some variation in the latency of the receptor potential from flash to flash, which becomes significant for dim flashes delivering <1,000 effective photons. This variation reduces the accuracy with which the latency of the estimated elevations of Ca_i and that of the receptor potential can be compared. Stable recordings of dye fluorescence and membrane potential were generally obtained for up to 90 min after injection of the dye. With ∼10 min between dual flash presentations required for dark adaptation in unbleached photoreceptors, reconstructions of elevations of Ca_i were usually limited to a maximum of 9 or 10 dual flash presentations.

Membrane potential or current traces were digitally sampled at a rate of 1 kHz and stored on a computer, simultaneously with samples of the command pulses sent to the amplifier controlling the shutter placed in front of the laser beam. The shutter command pulses stored in these records were used to synchronize the current or voltage traces with the onset of fluorescence in the digitized records of photomultiplier output stored in the confocal microscope, after allowing 1.5 ms for shutter opening, as described in Ukhanov and Payne 1995.

Photoreceptors were filled with the low affinity indicator, Oregon green-5N. A dual flash protocol (see materials and methods) was used to reconstruct changes in dye fluorescence after dim flashes. Photoreceptors were dark adapted between presentations of dual flashes so that bumps of 1–5 mV could be recorded. For each cell, the saturated peak fluorescence, F_max, was also recorded during illumination by a step of light from the unattenuated laser. After dim flashes, transient increases in dye fluorescence could be reconstructed that increased in peak amplitude with increasing flash intensity (Fig. 2). The peak amplitude of the fluorescence saturated at F_max for flash energies delivering >5 × 10⁴ effective photons (Fig. 3 A). This saturation is assumed to reflect that of the Ca²⁺-indicator dye rather than the value of Ca_i, since peak Ca²⁺ concentrations exceeding 150 μM have been recorded in response to a step of illumination by the unattenuated laser beam using a lower affinity dye, Calcium-Green 5N (K_d = 67 μM; Ukhanov et al. 1995). On this assumption and also assuming that the dye binds negligible Ca²⁺ in dark-adapted cells at rest (see materials and methods), the peak amplitudes of the light-induced elevations of Ca_i were calibrated (Fig. 3 B). Estimated peak Ca_i rose approximately as light intensity raised to the power 0.6, a much more steep relationship than that observed for Ca_i during prolonged background illumination, which increases as light intensity raised to the power 0.1–0.2 (Levy and Fein 1985; Ukhanov et al. 1995), but comparable with that observed by Oberwinkler and Stavenga 1998 during sustained illumination of blowfly photoreceptors. The dimmest flashes for which elevations of Ca_i could be reliably detected at the peak of the receptor potential delivered ∼50 effective photons and generated estimated peak increases in Ca_i of 5.0 ± 1.8 μM (SEM, n = 7), or ∼300 free Ca²⁺ ions per effective photon in an estimated confocal measurement volume of ∼5 μm³.

For flashes delivering <200 effective photons, the elevation of Ca_i recorded using Oregon green-5N was too small to reliably reconstruct its time course, and only the elevation of Ca_i at the peak of the receptor potential was sampled (but see below for fluo-4 signals). For more intense flashes, the time course of the initial elevation of Ca_i could be reconstructed (Fig. 4). The peak receptor potentials after these flash intensities were all close to saturation, but their latencies greatly differed. Despite this difference in response latency, the rising edge of the reconstructed elevation of Ca_i was always approximately coincident with that of a representative receptor potential. For the dimmest flashes, the latencies of the response varied considerably from cell to cell (Fig. 5 A), but for any given cell, the electrical signal and elevation of Ca_i were approximately coincident. As a measure of the latency of the elevation of Ca_i, we chose the time, T_{c, Ca2+}, taken for Ca_i to exceed a criterion of 2 μM, the smallest increase that could be reliably distinguished in individual reconstructions when using Oregon green-5N as the Ca²⁺ indicator dye. T_{c, Ca2+} fell from an average of 141 ms for four cells illuminated by the dimmest flash to 21 ms as flash intensity increased (Fig. 5 A). For this range of response latencies and light intensities, T_{c, Ca2+} was approximately equal to the time, T_{c, mV}, taken to exceed an 8-mV criterion receptor potential (Fig. 5 B). The relationship could be well described (r ² = 0.987; n = 13) by the equation:

with Δt equal to 2.2 ± 6.4 ms. This time difference is not significant (P > 0.05).

For flashes delivering 50 or fewer effective photons, the change in Oregon green-5N fluorescence was insufficient to allow a detailed analysis of the time course of the response. A higher affinity indicator dye, fluo-4, was therefore used. The rising edge of the elevation of Ca_i induced by these dim flashes was also approximately coincident with the receptor potential (Fig. 6A and Fig. B; representative of recordings from three cells). T_{c, Ca2+} and T_{c, mV} were both equal to 146 ms for criteria of 2 μM and 8 mV, respectively. In the same cell, the response to a step of illumination from the unattenuated laser beam also demonstrated coincidence of the receptor potential and the elevation of Ca_i (Fig. 6 B). T_{c, Ca2+} was 30 ms, and T_{c, mV} was 29 ms. The receptor potential and elevation of Ca_i were, therefore, approximately coincident in this cell for responses to illumination that differed in intensity by five orders of magnitude.

The elevation of Ca_i and receptor potential arise from the same stochastic process. Chemical bleaching of rhodopsin by hydroxylamine (see materials and methods) reduced the quantum efficiency of photoreceptors by ∼3 log₁₀ U, while maintaining quantum bumps that were up to 10 mV in amplitude and of normal time course. Neither the latency of the responses delivering approximately the same number of effective photons nor the resting membrane potential was qualitatively altered by bleaching. This treatment reduced the effective intensity of the unattenuated laser from ∼10⁸ to ∼10⁵ effective photons/s, prolonging the latency of the response to continuous illumination from 20–30 to 70–150 ms. Because the laser illumination was unattenuated and continuous, the entire time course of an elevation of Ca_i accompanying a receptor potential could be determined and its latency correlated with that of the receptor potential. Thus, the problem of comparing elevations of Ca_i reconstructed from many flash responses with a representative receptor potential was eliminated.

After chemical bleaching, successive receptor potentials to unattenuated step illumination by the laser exhibited variation in their latent period (Fig. 7). Variation in the latency of the receptor potential after dim flashes is thought to result from the stochastic nature of the underlying responses to single effective photons (Yeandle and Srebro 1970; Kraemer et al. 1989). Elevations of Ca_i, measured using fluo-4, that accompanied each receptor potential showed latency variations similar to those of the receptor potential (Fig. 7). To quantitatively compare the fluctuations in latency, a criterion of either 1 or 2 μM Ca_i was selected for each of five cells so as to best match the mean value of T_{c, Ca2+} to that of T_{c, mV}. The values of T_{c, Ca2+} and T_{c, mV} were then compared for individual flashes and were found to be highly correlated and approximately equal. For the cell of Fig. 7, their relationship could be well described (r ² = 0.951; n = 5; Fig. 8 A) by , with Δt = 1.4 ± 1.14 ms. This time difference is not significant (P > 0.05).

To compare data from all five cells, the mean values of T_{c, mV} and T_{c, Ca2+} recorded from a given cell were subtracted from those of individual signals (Fig. 8 B). The resulting deviation scores of T_{c, mV} were equal to those of T_{c, Ca2+} (r ² = 0.829). Thus, if an elevation of Ca_i rose, for example, to exceed its criterion 10 ms earlier than its mean for that cell, it was highly probable that the electrical response would also exceed its criterion 10 ms earlier than its mean. This correlation indicates that the process generating the stochastic latency is shared by both signals.

While the results so far demonstrate that the timing of the electrical and Ca²⁺ signals are highly correlated, the absolute value of Δt does not indicate whether the Ca²⁺ signal “leads” or “lags” the electrical response of a cell. Δt rather indicates the relative time for the two signals to reach criterion amplitudes and its value is therefore dependent upon the choice of those criteria. In further considering the significance of Δt, we noted that it varied from +1.8 to −13 ms for the five cells studied, for criterion responses of 8 mV and 2 μM. We wished to determine whether this variation of Δt between cells was due to systematic differences between recordings from different cells or due to noise inherent in estimating Ca_i. Fig. 9 illustrates five responses each from two cells. To eliminate the variation in response latency from flash to flash, the time base of each step response has been normalized by subtracting T_{c, mV}. The two sets of responses show that the time course of the elevation of Ca_i relative to that of the receptor potential varies between the two cells in a systematic manner. The elevation of Ca_i was first detected 17.3 ± 4.5 ms before T_{c, mV} in Fig. 9 A (arrow) and 7.4 ± 2.1 ms before T_{c, mV} in B (arrow) (detection was based on a criterion of two successive samples of Ca_i greater than 2 SD above the initial noise level). This variation in the relative timing of the signals from cell to cell, which may arise from small displacements of the confocal spot relative to Ca²⁺ release sites (see discussion), prevents us from determining an accurate relationship between Ca_i and the receptor potential amplitude. However, it is apparent from Fig. 9 that the light-induced elevation of Ca_i is detectable at very small depolarizations. For the five cells, the mean depolarization was not significant (1.1 ± 1.2 mV; P > 0.05) at the time the light-induced elevation of Ca_i was first detected.

After chemical bleaching, the increased latency of the response to the unattenuated laser allowed a greater time “window” in which to measure the elevation of Ca_i. Also, because the flashes used to elicit dye fluorescence were less effective, the time for dark adaptation between pairs of flashes was reduced from ∼10 to ∼2 min, allowing more samples to be taken while dye fluorescence levels were stable in the cell. These improvements allowed the measurement of elevations of Ca_i in cells filled with fluo-4 that accompanied extremely dim flashes, delivering ∼10 effective photons, which demonstrated considerable stochastic variation in the amplitude and latency of individual receptor potentials (Fig. 10 A). Although the accuracy of the comparison was limited to ∼20 ms by variation in individual responses (Fig. 10 B), the first appearance of a calcium signal was still approximately coincident with the appearance of the electrical response. In the same cell, the response to a step of illumination from the unattenuated laser beam also demonstrated coincidence of the receptor potential and the elevation of Ca_i (Fig. 10 B). Similar elevations of Ca_i were obtained from two other chemically bleached cells.

For bright steps of light, delivered by the unattenuated laser, the timing of the elevation of Ca_i, recorded from unbleached cells, relative to that of the electrical response, is unaffected by voltage clamping cells to their resting potential or by removal of extracellular Ca²⁺ (Ukhanov and Payne 1995). We wished to confirm this result using bleached photoreceptors and for dim flashes. In hydroxylamine-treated cells that were voltage clamped to their membrane potential in darkness (typically between −50 and −65 mV), the elevation of Ca_i measured by fluo-4 was approximately coincident with the appearance of the photocurrent initiated by a step of illumination from the unattenuated laser beam (Fig. 11 A). Similar results were obtained in a total of five cells. Bathing the cell of Fig. 11 A in 0 Ca-ASW for 15 min reduced the initial dye fluorescence, reflecting a decrease in resting Ca_i (Bolsover and Brown 1985; Levy and Fein 1985), and increased the latency of both the photocurrent (Martinez and Srebro 1976) and the elevation of Ca_i. However, the onset of the photocurrent and the elevation of Ca_i were still approximately coincident (Fig. 11 B).

Reconstruction of the complete time course of elevations of Ca_i after dim flashes delivered to unbleached cells bathed in 0 Ca-ASW was not possible because of the progressive depletion of Ca²⁺ stores by the repeated bright flashes required to measure dye fluorescence (see below). However, up to five dual flashes could be delivered without greatly affecting the magnitude or latency of successive responses, so that a portion of the rising phase of elevations of Ca_i could be reconstructed (Fig. 11 C). The magnitude of reconstructed elevations of Ca_i measured using Oregon green-5N after dim flashes and their timing relative to that of the receptor potential were similar to those recorded from cells bathed in ASW. Similar results were obtained from a total of four cells. The correlation between the latencies of the receptor potential and the elevation of Ca_i cannot, therefore, be ascribed to the entrance of Ca²⁺ ions through light- or voltage-activated channels in the plasma membrane.

Bolsover and Brown 1985 demonstrated that repeated stimulation with bright flashes, while bathing cells in 0 Ca-ASW, prolongs the latency of the electrical response, slows, and diminishes it. Light-induced elevations of Ca_i estimated from recordings using aequorin are diminished under these conditions, indicating that the calcium stores become depleted by the repetitive illumination. We wished to determine whether the prolongation of the latency of the electrical response under these conditions was accompanied by a similar prolongation of the latency of the elevation of Ca_i. Bleached photoreceptors were bathed in 0 Ca-ASW and subjected to repetitive x-y scans by the unattenuated laser beam so as to uniformly deplete Ca²⁺ stores within the cell. At intervals between the scans, the elevation of Ca_i and receptor potential elicited by spot illumination at the edge of the R-lobe were measured. As the length of time in 0 Ca-ASW and the number of scans delivered increased, the latency of the elevation of Ca_i increased and its magnitude and rate of rise decreased (Fig. 12). Despite the large differences in the absolute latency of the responses, the rise of the receptor potential was still approximately coincident with that of the elevation of Ca_i. Results similar to those of Fig. 12 were obtained in three other cells.

Bathing photoreceptors filled with fluo-4 in 0 Ca-ASW containing 100 μM CPA, an inhibitor of endoplasmic reticulum (ER) Ca²⁺ ATPase (Mason et al. 1991; Walz et al. 1995), dramatically reduced the light-induced elevation of Ca_i and slowed and diminished the receptor potential (Fig. 13 A). As reported by Dorlöchter et al. 1999, single photon signals were also reduced in amplitude to below the noise level (Fig. 13B and Fig. C). However, despite the large reduction in the sensitivity of the cell to dim flashes, the receptor potential was never completely abolished by treatment with CPA in 0 Ca-ASW. If the response remaining after CPA treatment is considered to be a separate component of the transduction process, not dependent on Ca²⁺ release (Nagy 1993), it is relatively insignificant during the rising edge of the normal response to a dim flash. For the cell of Fig. 13 A, the response after CPA treatment was 1.2 mV at 123 ms after the flash, compared with 25 mV at the same time before CPA treatment. However, as previously reported (Ukhanov and Payne 1995), intense, saturating illumination by the unattenuated laser resulted in the generation of large receptor potentials in the absence of a significant elevation of Ca_i (Fig. 13 D). Thus, for bright flashes, the current generated by this mechanism may be significant. To check that the loss of the light-induced elevation of Ca_i was not caused by the loss of the Ca²⁺ indicator dye from the cytosol, the cell was bathed for 2 min in 10 mM Ca-ASW, without added CPA. Upon illumination of the chosen spot by the unattenuated laser beam, fluorescence rose after a latency of 30 ms, with a half time of 140 ms, to a peak that was 4.0× its value in 0 Ca-ASW. This slower, but substantial, rise in fluorescence presumably reflects Ca²⁺ influx, as well as any renewed release from stores, and it demonstrates the presence of calcium indicator after the prolonged incubation with CPA. Results similar to those of Fig. 13 were obtained in a total of four cells.

The above result, our failure to ascribe the elevation of Ca_i to influx of Ca²⁺, and the conclusions of previous work (see discussion) are consistent with the idea that the correlation between the timing of the light-induced elevation of Ca_i and inward current might result from the activation of an inward current by released Ca²⁺. As noted above, variation in the timing of the Ca²⁺ signal from cell to cell makes it impossible to define an accurate relationship between the estimated value of Ca_i and the receptor potential or the light-induced current. However, we thought it valuable to determine the approximate form of a process that might couple elevation of Ca_i to the receptor potential.

Fig. 14 shows the mean elevation of Ca_i and inward current recorded from five bleached photoreceptors filled with fluo-4 that were voltage clamped to their resting membrane potentials and illuminated with a step of light from the unattenuated laser. For each cell, the time taken for the light-induced current to reach a criterion of −2 nA has been subtracted from the time base so as to compare the relative timing of Ca²⁺ and electrical signals, despite the variation in response latency. As with the voltage responses of Fig. 9, the elevation of Ca_i was detectable at the earliest times of the electrical response to light. For three of the five cells individually, as well as for the mean data in Fig. 9, elevation of Ca_i is detectable before significant inward current flow (detection based on a criterion of two successive samples >2 SD above the initial noise level). To relate the elevation of Ca_i to inward current flow, a simple model was assumed in which we ignore the small component that is not dependent on Ca²⁺ release during the rising edge of the response and assume that Ca²⁺ reversibly binds to the light-activated channel or an associated protein. The channel was assumed to be open for as long as Ca²⁺ is bound.

For low [Ca²⁺], before saturation of the dye or inward current occurs, the rate of change of the mean inward current, i_L(t), can be described by the differential equation:

Fig. 14 shows a fit of to i_L(t) obtained through numerical integration using the observed light-induced elevation of Ca_i, k′₁ = 0.35 nA · μM⁻¹ · ms⁻¹ and k₋₁ = 0.2 ms⁻¹. For constant [Ca²⁺], then describes a first-order process with a sensitivity of inward current to Ca_i of 1.75 nA · μM⁻¹ and a time constant of 5 ms. Aside from parsimony, our choice of a linear relationship between [Ca²⁺] and i_L is dictated by the following consideration. After the latent period, the rise of [Ca²⁺] with time during the light step of Fig. 14 approximates to a linear ramp. then predicts that the current will also rise as a linear ramp after a brief delay, as is observed. Models that invoke a cooperative action of several calcium ions to activate the channel would predict a sigmoidal rise in current with time.

Pharmacological experiments have indicated that light-induced release of Ca²⁺ is necessary, in Limulus photoreceptors, for a rapid, highly amplified initial response to dim flashes of light (Bolsover and Brown 1985; Payne and Fein 1986; Frank and Fein 1991; Shin et al. 1993; Nagy and Contzen 1997; Dorlöchter et al. 1999). Ours is the first measurement of an elevation of Ca_i that is sufficiently rapid to mediate this effect. Our reconstructed elevations of Ca_i are similar in peak amplitude (10–100 μM) and duration (1–3 s) to those recorded previously using other methods (Brown and Blinks 1974; Brown et al. 1977; Levy and Fein 1985; O'Day and Gray-Keller 1989). However, the high temporal resolution of fluorescent indicators and the focal origin of the fluorescence within a few microns of the photoreceptive membrane (Ukhanov and Payne 1995) allow more accurate measurement of elevations of Ca_i that might modulate the activity of the light-sensitive conductance in the plasma membrane.

For a wide range of flash intensities and conditions, the detection of the electrical response of dark-adapted Limulus ventral photoreceptors is, within a few milliseconds, coincident with that of the elevation of Ca_i. For responses of five chemically bleached photoreceptors to steps of light, where an uninterrupted Ca²⁺signal could be recorded, the mean receptor potential was insignificant at the time that the elevation of Ca_i was detected. Elevation of Ca_i was detected in three of another five cells before significant inward current flowed under voltage clamp. These elevations of Ca_i appeared to result from the release of Ca²⁺ ions from intracellular stores, since they were undiminished by removal of extracellular Ca²⁺ ions, but were abolished by exposure to CPA. Thus, light-induced Ca²⁺ release precedes the generation of most, if not all, of the electrical response.

Our observations necessitate a revision of a previous model of a role for Ca²⁺ release in phototransduction in Limulus photoreceptors. Payne and Fein 1986 observed that the rising edge of the photocurrent rose too abruptly to be plausibly modeled by a linear cascade of first-order reactions, as was first proposed by Borsellino et al. 1965. To explain the abrupt rise, they proposed that a gradual light-induced elevation of Ca_i accelerated the rate of three sequential steps in the proposed cascade of reactions that link photoisomerized rhodopsin to the activation of ion channels in the plasma membrane. Our measurements of Ca_i are not consistent with a gradual elevation. It is apparent from inspection of Fig. 4, Fig. 6, and Fig. 7 that the elevation of Ca_i shares the abrupt rising edge of the electrical response to light. InsP₃-mediated calcium release in other cells also displays abrupt release after a variable latent period, a behavior that has been attributed to an apparent threshold concentration of InsP₃ required to activate the InsP₃ receptor in the ER. (Marchant and Taylor 1997).

The simplest explanation for the correlation between the elevation of Ca_i and the receptor potential is that light-induced elevation of Ca_i activates a component of the photocurrent (Shin et al. 1993). Our simple analysis shows that a first-order process with a steady state sensitivity of inward current to Ca_i of 1.75 nA · μM⁻¹ and a time constant of 5 ms would model the relationship between the initial mean elevation of Ca_i and mean inward current recorded from five cells. However, variation in the timing of the Ca²⁺ signal relative to that of the inward current from cell to cell, possibly due to experimental error, makes it impossible to say definitively that such a deterministic linkage exists. Variation in timing of the Ca²⁺ signal may be due to error in placing the confocal measurement spot over the microvillar membrane. Assuming a diffusion coefficient for Ca²⁺ of 227 μm² · s⁻¹ (Allbritton et al. 1992), an error of only 2 μm could result in a delay of several milliseconds in the detection of Ca²⁺ released from the ER, which lies within 100 nm of the microvillar membrane (Calman and Chamberlain 1982; Payne et al. 1988).

In support of the proposal above, rapid Ca²⁺ release by flash photolysis of caged InsP₃ can activate an inward current in Limulus ventral photoreceptors within 2.5 ± 3.3 ms of the detection of the InsP₃-induced elevation of Ca_i (Ukhanov and Payne 1997). The molecular mechanism of this activation is unknown, as is the identity of the channels carrying the current. In addition, pharmacological blockade of light-induced Ca²⁺ release greatly slows and diminishes the initial, transient component of the response of dark-adapted Limulus ventral photoreceptors to dim and moderate flash intensities (Fig. 13; Bolsover and Brown 1985; Payne and Fein 1986; Frank and Fein 1991; Shin et al. 1993; Nagy and Contzen 1997; Dorlöchter et al. 1999). However, even though treatment with CPA to eliminate measured Ca²⁺ release has a dramatic effect on the amplitude and rate of rise of the response to dim light flashes, it does not eliminate the electrical response to bright light. Thus, a slower, less sensitive, pathway of transduction might also exist that activates the light-sensitive conductance in the absence of calcium release (Nagy 1993).

Since treatment with CPA abolishes the large quantum bumps that comprise the response of the cell to very dim illumination (Dorlöchter et al. 1999), Ca²⁺ release appears to be a necessary step in the bump-generating mechanism. The large peak amplitude of the quantum bump, which averages 5 mV (0.5 nA in voltage clamp recordings; Yeandle and Spiegler 1973), and requires the activation of hundreds of ion channels (Bacigalupo and Lisman 1983), may then be the consequence of the amplification inherent in InsP₃-induced Ca²⁺ release. We calculate that, for dim flashes, at least 300 Ca²⁺ ions are released per effective photon at the peak of the response.

Previous attempts to demonstrate an elevation of Ca_i that precedes or accompanies the rising edge of the electrical response using aequorin or Arsenazo III as an indicator were not successful (Brown and Blinks 1974; Payne and Flores 1992; Stieve and Benner 1992) and it has been argued that light-induced calcium release is not rapid enough to contribute to the initial electrical response of invertebrate photoreceptors (Clapham 1996; Sakakibara et al. 1998). While this may be true for the photoreceptors of Drosophila, which are capable of an electrical response within 6 ms of a light stimulus (Hardie 1995), it is clearly not true of Limulus ventral photoreceptors. Our findings might also be relevant for photoreceptors of other slow-moving species, such as the file clam (del Pilar Gomez and Nasi 1998) and the marine mollusk, Hermissenda (Sakakibara et al. 1998), where intracellular injection of InsP₃ has been also shown to induce rapid bursts of inward current or depolarization. However, the price of the amplification inherent in the Ca²⁺ release mechanism may be a longer, more variable, latent period of the response in Limulus photoreceptors. We have demonstrated that the elevation of Ca_i displays variation from flash to flash, similar to that of the receptor potential. Variation in the latency of the photocurrent in response to dim flashes results in an inaccuracy in the coding of temporal information, a component of “transducer noise” (Lillywhite and Laughlin 1979).

We thank Dr. Alain Dabdoub and Ms. Monika Deshpande for critical reading of the manuscript and Dr. Ian Mather of the Animal and Poultry Science Department, University of Maryland, for the use of the confocal microscope.

This work was supported by National Institutes of Health EY07743.