The effect of light on outer segment calcium in salamander rods

Abstract

Calcium acts as a second messenger in vertebrate rods, regulating the recovery phase of the light response and modulating sensitivity during light-adaptation. Since light not only decreases the outer segment calcium concentration ([Ca²⁺]_i) by closing cyclic nucleotide-gated channels but can also increase [Ca²⁺]_i by releasing Ca²⁺ from buffer sites or intracellular stores, we examined in detail the effect of light and circulating current on [Ca²⁺]_i by making simultaneous measurements of suction pipette current and [Ca²⁺]_i from isolated rods of the salamander Ambystoma tigrinum after incorporation of the fluorescent dye fluo-5F. When the release of Ca²⁺ is measured in 0 Ca²⁺−0 Na⁺ solution, minimising fluxes of Ca²⁺ across the plasma membrane, it is substantial only for light bright enough to bleach a significant fraction of the photopigment and is restricted to the part of the outer segment in which the bleach occurred. It is unlikely, therefore, to make a large contribution to [Ca²⁺]_i for most of the physiological operating range of the rod. Nevertheless, since release is half-maximal for a bleach of less than 10 %, it cannot be produced by a simple mechanism such as a change in the affinity of a binding site on rhodopsin itself but must instead require some more complex interaction. In Ringer solution, the Ca²⁺ in the light-releasable pool can be discharged merely by the decrease in [Ca²⁺]_i that occurs as the outer segment channels close. In steady background light or after exposure to saturating illumination, the fraction of Ca²⁺ in the pool decreases essentially in proportion to [Ca²⁺]_i as if Ca²⁺ were being removed from a buffer site within the cytoplasm. Furthermore, [Ca²⁺]_i itself changes in proportion to the circulating current, with little evidence for a contribution from Ca²⁺ release or other mechanisms of Ca²⁺ homeostasis. This indicates that flux of Ca²⁺ across the plasma membrane is the major determinant of outer segment Ca²⁺ concentration within the rod's normal operating light intensity range. Once Ca²⁺ has been discharged from the releasable pool, it is restored following dim illumination apparently as the simple result of the subsequent restoration of dark [Ca²⁺]_i and the rebinding of Ca²⁺ to its release site, but after brighter light perhaps also as a consequence of regeneration of the photopigment.

Considerable evidence indicates that calcium acts as a second messenger in vertebrate photoreceptors controlling the rate of several steps in the transduction cascade (Fain et al. 2001). This modulation is essential for the regulation of sensitivity during adaptation to backgrounds and after bright bleaching light (reviewed by Pugh et al. 1999; Fain et al. 2001). Several attempts have been made to account for the role of Ca²⁺ by modelling its effect on the transduction cascade (Tamura et al. 1991; Miller & Korenbrot, 1993; Koutalos et al. 1995; Nikonov et al. 1998), but these calculations all assume that the light-induced closure of the cyclic nucleotide-gated channels produces a linearly proportional decrease in the influx of Ca²⁺ into the outer segment (Yau & Nakatani, 1984; Hodgkin et al. 1985). It is then also assumed that the closure of the channels produces a linearly proportional decrease in outer segment calcium concentration ([Ca²⁺]_i), since the K_m of Ca²⁺ for the Na⁺-Ca²⁺-K⁺ exchanger is considerably higher than physiological levels of [Ca²⁺]_i (Cervetto et al. 1989).

There are several reasons for supposing that these assumptions might not be valid, and that the relationship between channel opening and [Ca²⁺]_i may be more complicated than simple linearity. There is evidence, for example, that in addition to decreasing outer segment [Ca²⁺]_i (McNaughton et al. 1986; Ratto et al. 1988; Gray-Keller & Detwiler, 1994; McCarthy et al. 1994, 1996; Sampath et al. 1998), bright light can increase[Ca²⁺]_i by releasing as much as 10–50 μm Ca²⁺ per litre tissue volume from some cytosolic store or buffer site. This effect can most clearly be seen during exposure of the outer segment to a 0 Ca²⁺−0 Na⁺ solution designed to minimise both the influx and efflux of Ca²⁺ across the plasma membrane (Matthews & Fain, 2001, 2002; Brockerhoff et al. 2003), but the light-induced release of Ca²⁺ has also been shown to alter [Ca²⁺]_i even in Ringer solution (Matthews & Fain, 2001; Brockerhoff et al. 2003). There is also evidence suggesting that [Ca²⁺]_i may be lower during presentation of background light or after recovery from a bright flash than would be predicted from the value of the photocurrent (Gray-Keller & Detwiler, 1994). Furthermore, the disks of rods are known to contain a high concentration of Ca²⁺ (Schnetkamp, 1979; Fain & Schroder, 1985; Chen et al. 2002), which may communicate in some way with the cytoplasm.

We therefore thought it important to take advantage of the previous techniques we have devised for measuring [Ca²⁺]_i and light-induced Ca²⁺ release (Matthews & Fain, 2001, 2002) to examine more carefully the relationship between light exposure, photocurrent and [Ca²⁺]_i. We approached this problem in two ways. First, we measured the intensity dependence of light-induced Ca²⁺ release and showed that it is half-maximal at an intensity bleaching about 8–9 % of the pigment. Second, we studied the relationship between [Ca²⁺]_i and photocurrent, both for steady light of moderate intensity and after a saturating light flash. We found that photocurrent and [Ca²⁺]_i are very nearly proportional to one another. Consequently, for most of the range of intensities over which rod light responses occur, the value of [Ca²⁺]_i is linearly proportional to the number of open channels, and the release of Ca²⁺ seems not to produce a significant effect on [Ca²⁺]_i. A release of Ca²⁺ large enough to affect [Ca²⁺]_i is produced only by light sufficiently intense to bleach a significant proportion of the photopigment, and this release is localised to the illuminated region of the outer segment. Preliminary results of this study have been reported both to the Physiological Society (Matthews & Fain, 2003) and to the Association for Research in Vision and Ophthalmology (Fain & Matthews, 2003).

METHODS

Preparation

Aquatic tiger salamander (Ambystoma tigrinum) were purchased from Charles Sullivan (Nashville, TN, USA) and dark-adapted overnight. Animals were killed according to Schedule 1 of the Animals (Scientific Procedures) Act by stunning followed by decapitation and pithing, their eyes were removed, and their photoreceptors dissociated mechanically from the isolated retina under infrared illumination. Dissociated cells were incubated for 30 min in darkness with 10 μm fluo-5F acetoxymethyl ester (fluo-5F AM; Molecular Probes Europe) as described previously (Matthews & Fain, 2001, 2002), in amphibian Ringer solution consisting of 111 mm NaCl, 2.5 mm KCl, 1.6 mm MgCl₂, 1.0 mm CaCl₂, 10 μm EDTA, and 3 mm Hepes, adjusted to pH 7.7–7.8 with NaOH. They were then transferred to the recording chamber, and excess unhydrolysed dye was removed by bath perfusion. With this method of loading, the dye appears to remain within the outer segment cytosol and very little seems to become compartmentalised within the disks (see Chen et al. 2002; Nakatani et al. 2002).

An isolated rod photoreceptor was drawn inner segment first into a suction pipette so that the outer segment was exposed to the bathing solution. Rapid solution changes from Ringer solution to 0 Ca²⁺−0 Na⁺solution were made as described previously using a computer-controlled stepper motor coupled to the microscope stage (Matthews & Fain, 2001, 2002). The 0 Ca²⁺−0 Na⁺ solution was identical in composition to the ‘0 Ca²⁺, 0 Mg²⁺, 0 Na⁺ solution’ in earlier experiments (Matthews, 1995,1996) and consisted of 111 mm choline chloride, 2.5 mm KCl, 2 mm EGTA, and 3.0 mm Hepes, adjusted to pH 7.7–7.8 with tetramethylammonium hydroxide. All experiments were performed at 22°C.

Ca²⁺ measurement and recording and light stimulation

Ca²⁺ was measured as in our previous experiments (Matthews & Fain, 2001, 2002). In brief, dye fluorescence was excited by an air-cooled argon ion laser (Model 60, American Laser Corporation, Salt Lake City, UT, USA), tuned to 488 nm. Fluorescence was collected with a 505 nm dichroic mirror and a 510 nm emission filter (Types 505DRLP and 510ALP, Omega Optical, Brattleboro, VT, USA). Laser intensity at the plane of the preparation was adjusted to around 2 × 10¹¹ photons μm⁻² s⁻¹ with reflective neutral density filters (8 × 10¹⁰ photons μm⁻² s⁻¹ for the experiment of Fig. 4). A beam stabiliser was used to decrease noise from laser intensity fluctuations (Noise Eater, Model CR-200A, Thorlabs, Newton, NJ, USA). Fluorescence intensity was measured with a photomultiplier tube (PMT, Model 9124A, Electron Tubes Ltd, Ruislip, UK). The signal from the PMT was amplified, filtered at 1 kHz, and digitally sampled at 4 kHz as in earlier studies (Matthews & Fain, 2001, 2002). Curve fitting was carried out with the plotting program Origin (Microcal Software, Inc., Northampton, MA, USA).

Figure 4. Spatial localisation of Ca²⁺ release.

The outer segment was stepped into 0 Ca²⁺−0 Na⁺ solution and 2 s later exposed for 10 s to an intense slit of 520 nm light calculated to bleach over 99 % of the photopigment within the area of the stimulus, which was positioned over the base (A) or tip (B) of the outer segment. One second after the extinction of the slit stimulus, the tip of the outer segment was exposed to two laser flashes separated by a 5 s interval. Arrows denote the magnitudes of fluorescence evoked by the first (P1) and second (P2) laser flashes and of the pedestal of fluorescence after return to Ringer solution (Ped). Insets show examples of rods in which the bleaching slit was separate (A) or coincident with the laser spot (B). C, ratio of the fluorescence evoked by the second laser flash to that evoked by the first (P2/P1). Separate, data from 7 cells as in A; Coincident, data from 7 cells as in B; data represent means ±s.e.m. Intensity of the slit bleaching light stimulus 2.8 × 10⁷ photons μm⁻² s⁻¹ at 520 nm.

In some experiments, the Ca²⁺ signal was normalised to the small ‘pedestal’ of fluorescence remaining after bright light stimulation in Ringer solution. To justify this procedure, the absolute levels of Ca²⁺ were estimated with a calibration method modified from one used previously (Sampath et al. 1998, 1999). The rods in Ringer solution were exposed to 50 ms laser flashes to estimate fluorescence in darkness and after bright illumination. The minimum fluorescence (F_min) was then determined by moving the outer segment with the solution stepper into a solution containing 10 μm ionomycin (No. I-0634, Sigma-Aldrich Ltd) and 0 Ca²⁺ of the following composition: 111 mm NaCl, 2.5 mm KCl, 2.05 mm MgCl₂, 3 mm Hepes, no added CaCl_2, and 2 mm EGTA, pH. 7.7. To determine the maximum fluorescence (F_max), the 0 Ca²⁺ solution bathing the outer segment was replaced with an isotonic CaCl₂ solution containing 76.6 mm CaCl₂, 2.5 mm KCl, and 3.0 mm Hepes, pH 7.7. Once the fluorescence amplitude reached steady state in this solution, the outer segment was exposed to an identical isotonic CaCl₂ solution which also contained saponin (1 mg per 100 ml, Sigma S-4521). Fluorescence intensities were then translated into values for [Ca²⁺]_i using the K_d for fluo-5F determined previously at 22°C in the presence of 1 mm Mg²⁺ (1.28 μm; see Woodruff et al. 2002). Further details of the calibration procedure are given in the Results section.

Light stimuli were delivered in the majority of experiments with a dual beam optical stimulator from a stabilised high-intensity mercury discharge lamp (Lumatec SUV-DC, Ultrafine Technology, Brentford, UK); stimuli were of wavelength 436 nm or 546 nm, selected with a narrow band interference filter (Comar Instruments, Cambridge, UK). The brightest intensity which could be delivered by the optical stimulator at 546 nm was of the order of 1.6 × 10⁸ photons μm² s⁻¹, which was typically attenuated with calibrated, neutral density filters. In the experiment of Fig. 4 light stimuli of wavelength 520 nm were delivered from a tungsten-halogen light source. Rhodopsin bleaching was calculated from the photosensitivity for a vitamin-A₂-based pigment in solution (Dartnall, 1972), corrected for the difference in dichroism in free solution and in disk membranes (Jones et al. 1993).

RESULTS

Intensity dependence of Ca²⁺ release

To characterise the intensity dependence of Ca²⁺ release, rods were pre-exposed to conditioning light that bleached a variable fraction of the photopigment, and then to a series of laser flashes to measure the amount of Ca²⁺ still remaining in the releasable pool. The protocol for these experiments is shown in Fig. 1. A rod held in a suction pipette was first moved into 0 Ca²⁺−0 Na⁺ solution and then exposed to intense conditioning light from the optical bench (Bleach). The conditioning exposure was delivered in 0 Ca²⁺−0 Na⁺ solution to minimise the fall in [Ca²⁺]_i during the light response, since our previous experiments have shown that if the light is delivered in Ringer solution, Ca²⁺ can be lost from the releasable pool, even in relatively dim light, apparently as a result of the closure of the cyclic nucleotide-gated channels and ensuing decrease in outer segment [Ca²⁺]_i (Matthews & Fain, 2001). The conditioning exposure began 1 s after the solution change and was of constant intensity but variable duration: the shortest was 11 ms and bleached approximately 1 % of the rhodopsin, whereas the longest was 5 s and bleached 99 % of the pigment. Following the light exposure, the rod was given a series of eight logarithmically spaced, 20 ms laser flashes to measure fluorescence and determine the amplitude and time course of the light-induced Ca²⁺ release.

Figure 1. Effect of prior bleaching in 0 Ca²⁺−0 Na⁺ solution on fluo-5F fluorescence recorded from salamander rods in response to brief laser flashes.

A dark-adapted rod was stepped to 0 Ca²⁺−0 Na⁺ solution 1 s before the presentation of a period of intense illumination of variable duration designed to bleach a proportion of the photopigment. Six seconds after the solution change, a series of 8 laser flashes was delivered, each 20 ms in duration, spaced at approximately logarithmically equal intervals, beginning at the following times (in ms): 0, 75, 150, 300, 600, 1250, 2500 and 5000 (i.e. 5 s). Uppermost traces give timings of solution changes, bleaching light and laser exposures. Prior light exposure was calculated to bleach 0 % (darkness; A), 5 % (B), and 99 % (C) of the photopigment; each panel represents an experiment on a different rod. Markers denote magnitude of fluorescence evoked by the first and last laser flashes (P1 and P8) and the low level of fluorescence obtained when the laser exposure was repeated following return to Ringer solution (Ped). Bleaching light intensity 1.6 × 10⁸ photons μm⁻² s⁻¹ at 546 nm.

Figure 1A shows the raw photomultiplier current elicited by the dye fluorescence evoked by the laser flashes delivered according to this protocol but with no conditioning illumination. P1 denotes the fluorescence evoked by the first laser flash and is a measure of the dark-adapted [Ca²⁺]_i in the outer segment. The dotted line shows the response to P8, the last of the laser flashes, and the difference between P8 and P1 (normalised to P1) represents the fractional increase in free Ca²⁺ evoked by the laser flashes (Matthews & Fain, 2001, 2002). This increase is about half as large in Fig. 1B, for which the conditioning light bleached about 5 % of the rhodopsin, and it is eliminated in Fig. 1C, for which 99 % of the pigment had been bleached.

Figure 2 presents a summary of all the experiments with conditioning light exposures. The traces on the left show a rapid component of fluorescence increase in response to the first laser flash, which we have shown previously not to reflect a change in [Ca²⁺]_i but instead some interaction between rhodopsin and the indicator dye (Matthews & Fain, 2002). The amplitude of this early component is reduced by approximately half by a conditioning light bleaching 50 % of the rhodopsin. When 99 % of the rhodopsin was bleached, this initial component could no longer be observed (Matthews & Fain, 2002).

Figure 2. The effect of bleaching light on rapid and slow components of fluorescence increase.

Experimental protocol as for Fig. 1: before presentation of first laser flash, rods were exposed in 0 Ca²⁺−0 Na⁺ solution to a period of intense illumination designed to bleach the proportion of photopigment indicated beside each trace. Traces to the left compare the mean waveform during the initial 12.5 ms of the first laser flash for prior bleaches of 0, 50 and 99 % (10 cells for each bleach). Traces to the right represent means and standard errors of the fluorescence evoked by each of the 8 laser flashes as in Fig. 1 after prior bleaches between 0 and 99 % (10 cells for each bleach). Fluorescence amplitude has been measured as mean photomultiplier current during the interval 15–19 ms after the beginning of the laser flash and has been normalised for each cell to the amplitude of the first laser flash. Continuous curves in the right hand panels represent a two-exponential asymptotic rise with the time constants following prior bleaching constrained to the values of 0.12 s and 1.5 s obtained for the cells in darkness (0 % bleach).

The symbols plotted to the right give the photomultiplier current evoked by the eight laser flashes, for a conditioning light bleaching the percentage of pigment indicated to the right of each curve. The amplitude of the photomultiplier current was normalised cell by cell to the current evoked by the first laser flash, effectively normalising dye fluorescence to the value corresponding to the level of [Ca²⁺]_i after release by the conditioning light. Means and standard errors have been calculated from 10 rods for each bleaching light exposure. Fluorescence was measured over the interval 15–19 ms from the beginning of each 20 ms laser flash, to avoid a contribution from the rapid increase in fluorescence shown to the left. This procedure may somewhat overestimate the amplitude of fluorescence in darkness, due to the very earliest stages of Ca²⁺ release, but the size of this error is likely to be small (Matthews & Fain, 2002).

The time courses of Ca²⁺ increase have been fitted with a two-exponential asymptotic rise, which we have shown previously provides a good description of the kinetics of light-induced Ca²⁺ release (Matthews & Fain, 2002). The time constants of the curves fitted to the data obtained after prior bleaching light have been constrained to their values in the dark-adapted rods. We have shown previously that the fluorescence intensity 5 s after the first laser flash gives a good estimate of the maximum amplitude of the Ca²⁺ release (Matthews & Fain, 2001). The data in Fig. 2 indicate that, in contrast to the rapid initial component shown to the left, the maximum amplitude of fluorescence increase is halved by a conditioning light bleaching only about 5–10 % of the pigment.

This difference is investigated further in Fig. 3. To estimate the amplitude of the rapid initial component, a single exponential was fitted individually for each cell to the rise of fluorescence evoked by the first laser flash and this fit was then used to calculate the fractional contribution of this initial component to the total fluorescence signal (Matthews & Fain, 2001). The plot of Fig. 3A shows that the mean amplitude of the rapid component of fluorescence declined nearly linearly with the amount of rhodopsin bleached by the conditioning exposure. This is consistent with the notion that this component reflects some change in the properties of fluo-5F in the outer segment produced by activation of rhodopsin, as we have argued previously (Matthews & Fain, 2001).

Figure 3. Summary of the effect of bleaching light on the rapid and slow components of fluorescence increase.

Experimental protocol as in Figs 1 and 2. A, fractional magnitude of the rapid rise in fluorescence evoked by the first laser flash plotted as a function of the percentage of the photopigment bleached by the prior light exposure in 0 Ca²⁺−0 Na⁺ solution. A single exponential fitted individually to data from each cell was extrapolated back to time zero in order to calculate the fractional contribution of this initial component to the total fluorescence. Data represent means and standard errors, and are normalised individually for each cell. Fitted curve is a linear regression line. The number of cells was 9–10 for each bleach, except for the 99 % bleach, for which the rapid component was only large enough for the exponential fit to converge for 4 cells; the fractional contribution for the remaining cells at 99 % bleach (for which the rapid component was too small to be fitted) was set to zero. B, fraction of releasable Ca²⁺ pool discharged by the preceding light exposure in 0 Ca²⁺−0 Na⁺ solution. The rise in fluorescence between the first and last laser flashes has been normalised for each cell to the fluorescence evoked by the last laser flash ((P8 − P1)/P8). The means and standard errors of the data from cells at each bleaching level were scaled to set the mean values for the 0 % bleach to zero and for the 99 % bleach to unity to obtain the fraction of the pool remaining at the time of the first laser flash. This was then subtrated from unity to give the fraction of the pool released by the conditioning light; this quantity is plotted as a function of the prior bleach (10 cells for each bleach). Fitted curve is an asymptotic exponential, with a decay constant of 12.3 % bleached pigment.

The slower increase in fluorescence indicating an increase in outer segment Ca²⁺ has quite a different intensity dependence (Fig. 3B). The data have not been normalised to fluorescence of the first laser flash (P1), as in Fig. 2 (right), since the conditioning light progressively discharged the releasable Ca²⁺ pool, and consequently P1 would not reflect a constant level of [Ca²⁺]_i but instead the sum of [Ca²⁺]_i in darkness and the increase in [Ca²⁺]_i resulting from release by the conditioning stimulus. Instead, the rise in fluorescence between the first and last laser flashes has been normalised for each cell to the fluorescence evoked by the last laser flash, by calculating for each cell (P8 − P1)/P8. These data can be converted into the fraction of the pool remaining after conditioning illumination by assuming that the value after a 0 % bleach reflects release from a fully charged pool, whereas that after a 99 % bleach reflects the absence of release with the pool completely empty. The modest decrease in fluorescence following the 99 % bleach may represent residual exchanger activity, re-uptake of Ca²⁺, or diffusion from the outer segment into the inner segment (Matthews & Fain, 2002).

With these assumptions, the fraction of the pool remaining after each conditioning light exposure was normalised to the maximum value of the pool, estimated from the difference in the release for the 0 % and 99 % bleaches. That is, the means and standard errors of (P8 − P1)/P8 were offset for each conditioning intensity by the mean value for the 99 % bleach, and the result was then divided by the difference between the mean values for the 0 % and 99 % bleaches. These numbers were then subtracted from unity to give the fraction of the pool released by the conditioning light; this quantity is plotted in Fig. 3B as a function of the percentage of rhodopsin bleached. The data have been fitted with an asymptotic single exponential function with a decay constant of 12.3 % pigment bleached. This curve predicts that the pool should be half-discharged by a bleach of 8.5 %.

Spatial localisation of Ca²⁺ release

The ability of bleaching light to empty the pool of releasable Ca²⁺ raises the question of whether the signal that triggers the Ca²⁺ release is restricted to the area bleached by the laser spot, or whether this signal might spread throughout the outer segment. To investigate this question, spatially restricted bleaches were delivered from the optical bench in 0 Ca²⁺−0 Na⁺ solution, prior to the measurement of Ca²⁺ release induced by the laser spot. The protocol for this experiment is illustrated in Fig. 4. First, the outer segment was stepped into 0 Ca²⁺−0 Na⁺ solution and then exposed for 10 s to a spatially restricted slit of intense light from the optical bench, calculated to bleach over 99 % of the pigment in either the base (Fig. 4A, inset) or the tip (Fig. 4B, inset) of the outer segment. Two laser flashes were delivered at a 5 s interval to measure Ca²⁺ release. When the bleaching slit and the laser spot were spatially separated, a clear rise in fluorescence took place between the first and second laser flashes (Fig. 4A). In contrast, when, in another rod, the bleaching slit and the laser spot were coincident, no rise in fluorescence was observed between the two laser flashes (Fig. 4B).

Data from a number of such experiments are summarised in Fig. 4C, which compares the ratio of the fluorescence evoked by the first and second laser flashes in the two conditions. When the slit and spot were separated, a rise in fluorescence of 23 ± 3 % was observed, indicating that Ca²⁺ release could still be evoked by the laser from the tip of the outer segment even though its base had previously been bleached. In contrast, when the slit and spot were coincident, no significant rise in fluorescence was observed between the two laser flashes, consistent with the notion that the pool had already been voided by the bleaching slit at the outer segment tip prior to the first laser flash. These results indicate that the process that generates light-induced Ca²⁺ release is localised to the bleached region of the outer segment. The Ca²⁺ increase produced by the release is also likely to be spatially restricted, given the low diffusion constant of Ca²⁺ within the outer segment (Nakatani et al. 2002).

Measurement of changes in free Ca²⁺ concentration

In the analysis of Figs 1–4, the fluorescence signals were normalised by dividing the photomultiplier current by the current obtained from the first or last laser flashes. This effectively normalises the light-induced rise in Ca²⁺ either to the value of [Ca²⁺]_i before the first laser exposure, or to the value after the light-releasable Ca²⁺ pool has been completely discharged. Either procedure corrects for varying levels of dye incorporation in different rods.

The experiments that follow were designed to estimate changes in [Ca²⁺]_i in the presence of steady background light or during recovery after bright flashes delivered in Ringer solution. Dye fluorescence cannot be normalised as in Figs 1–4, since light exposure in Ringer solution not only alters the dark-adapted [Ca²⁺]_i (and therefore P1), it also partially discharges the releasable Ca²⁺ pool. Exposure to light in Ringer solution closes the cyclic nucleotide-gated channels, and the consequent decrease in [Ca²⁺]_i removes some of the Ca²⁺ from releasable buffer sites or internal stores (Matthews & Fain, 2001), thereby altering P8 (or P2 in Fig. 4). Consequently, some other method must be adopted to compensate for variations in dye loading among different cells.

In the following experiments, the fluorescence signals have been normalised to the level measured in Ringer solution 30–60 s after exposure to intense laser light. We have called this value the pedestal (see Fig. 4A and Sampath et al. 1998, 1999; Matthews & Fain, 2001, 2002). Since even a brief laser flash closes all of the cyclic nucleotide-gated channels in the outer segment for an extended period (Sampath et al. 1998), the fluorescence intensity decreases to a small value. The remaining dye fluorescence is due in part to [Ca²⁺]_i, which is greatly diminished after bright light exposure but is not zero (Gray-Keller & Detwiler, 1994; Sampath et al. 1998), and also to a Ca²⁺-insensitive, non-specific component of dye fluorescence (Sampath et al. 1998). We demonstrate below that the Ca²⁺-insensitive component contributes the majority of the pedestal fluorescence, indicating that it may be possible to use the pedestal as a measure of dye loading.

To determine the relative contribution of [Ca²⁺]_i and non-specific fluorescence to the pedestal value, we modified a procedure previously used to estimate the absolute value of [Ca²⁺]_i in the outer segment (Sampath et al. 1998). Figure 5A shows a continuous fluorescence record from such an experiment. A dark-adapted rod in Ringer solution was first exposed to a single 50 ms laser flash to estimate the level of fluorescence corresponding to the resting [Ca²⁺]_i. Approximately 50 s later, four 50 ms laser flashes were delivered to obtain the pedestal fluorescence value (Ped). The outer segment was next rapidly stepped into a 0 Ca²⁺ solution containing the ionophore ionomycin (see Methods), to reduce [Ca²⁺]_i to a very low level. Laser flashes were delivered every 10 s; once the fluorescence reached a steady-state value, it was averaged over a series of five flashes to determine the minimum fluorescence in the absence of Ca²⁺ (F_min). Then the solution flowing in that stream of the rapid perfusion system was changed to an isotonic CaCl₂ solution, resulting in an increase in dye fluorescence, presumably because the ionomycin introduced with the 0 Ca²⁺ solution remained within the membrane of the outer segment and facilitated a large flux of Ca²⁺ into the cell. After fluorescence reached steady state, the solution was changed to an isotonic CaCl₂ solution containing saponin. This resulted in a further increase in fluorescence followed by an abrupt decline, which was often accompanied by gross changes in the appearance of the outer segment and most probably reflected the loss of dye from the cell. The largest of the fluorescence values recorded during perfusion with saponin was taken to represent the maximum fluorescence in the presence of saturating Ca²⁺ (F_max).

Figure 5. Calibration of free Ca²⁺ concentration in salamander rods with fluo-5F.

Method of calibration described in text (see Methods). A, rod in Ringer solution in darkness pre-incubated in 10 μm fluo-5F AM was exposed to a series of 50 ms laser flashes to measure fluorescence in darkness (Dark) and after exposure to bright light (Ped). The rod was then stepped into a 0 Ca²⁺ solution containing 10 μm ionomycin to estimate F_min. The solution was then changed to isotonic Ca²⁺, and finally to isotonic Ca²⁺ containing 0.001 % saponin (1 mg (100 ml)⁻¹). B–D, waveform of photomultiplier current during calibration at higher temporal resolution. Same cell as for A. Illustrated recordings are as follows: B, first laser flash in Ringer solution (Dark) and average of 5 responses in Ringer solution beginning 50 s after first laser flash, at steady state after exposure to bright illumination (Ped); C, pedestal response from B at higher amplification compared to average of 5 responses in 0 Ca²⁺ Ringer solution containing 10 μm ionomycin; D, response in 0 Ca²⁺ Ringer solution compared to responses in isotonic Ca²⁺ and isotonic Ca²⁺ containing saponin. Note large value of the ratio F_max/F_min.

Figure 5B compares the response to the first laser flash with the pedestal. The ratio of these two values for 38 cells in darkness was 12.4 ± 0.67 (mean ±s.e.m.). Comparison of the pedestal to the fluorescence in 0 Ca²⁺-ionomycin (Fig. 5C) shows that, for the 15 cells for which the calibration procedure was carried to completion, 72 ± 3 % of the pedestal signal in the outer segment represents fluorescence which is insensitive to [Ca²⁺]_i. Figure 5D compares F_max and F_min, whose ratio for the 15 rods averaged 74.5 ± 4.3.

Using a K_d for fluo-5F of 1280 nm (Woodruff et al. 2002), these data give a value for [Ca²⁺]_i in the outer segment when the channels have all been closed of only 8 nm, a level considerably smaller than previous estimates from isolated photoreceptors (Gray-Keller & Detwiler, 1994; Sampath et al. 1998). It is sufficiently small that, if it is assumed to be zero (McCarthy et al. 1994), the value calculated for the dark resting Ca²⁺ concentration changes by only 2 %, from 545 ± 66 nm to 534 ± 65 nm. Since most of the pedestal fluorescence is produced by a signal insensitive to Ca²⁺, variability in its value is likely to reflect in large part the relative loading of outer segments with dye, with only a small contribution from differences in [Ca²⁺]_i in different rods after exposure to bright light. Fluorescence data from different cells can therefore be combined by normalising cell-by-cell to the pedestal value, with the likelihood that this will not introduce additional error due to variation in [Ca²⁺]_i between different photoreceptors.

The effect of steady background light on current and [Ca²⁺]_i

The effect of a steady background light on sensitivity and photocurrent was investigated as shown in Fig. 6A. For each rod, a bright flash was first given in Ringer solution, to determine the saturating photocurrent in darkness (i_D). Once the rod had recovered from this flash, steady background light was delivered in Ringer solution, upon which were superimposed a series of dim flashes to measure the sensitivity of the rod. Sensitivity is plotted as a function of background intensity in Fig. 6B for nine rods at each background intensity: each rod was exposed to a single background (see also Matthews et al. 1988). A bright flash was then delivered to determine the saturating photocurrent in the presence of the background (i_B)_. The ratio i_B/i_D is a measure of the fraction of channels still open in the presence of the background, so 1 −i_B/i_D corresponds to the fraction closed. Figure 6C plots 1 −i_B/i_D as a function of background intensity for the same cells.

Figure 6. Effect of background light on sensitivity and circulating current.

A, protocol for measurement of suppression of circulating current and reduction of dim flash sensitivity by background light in Ringer solution. Dim flashes were delivered in darkness (not shown) followed by a saturating flash, both at 546 nm. Steady background light was then presented, accompanied by a series of dim flashes and a single saturating flash. Background intensity 35 photons μm⁻² s⁻¹ at 436 nm, a wavelength which is blocked by the fluorescence emission filter and so which does not interfere with the measurement of dye fluorescence. B, dim flash sensitivity during background illumination, and C, response to background light, both normalised for each cell to their values in darkness and plotted as functions of steady background intensity. Each data point plots mean ±s.e.m. (n = 9 cells).

The background light was then extinguished, and the rod left in darkness for several minutes to permit the photocurrent and sensitivity to return to their dark-adapted values. A second background light exposure was then delivered to each rod, identical in intensity to the one used to measure sensitivity and photocurrent. At a time corresponding to the presentation of the saturating flash in Fig. 6A, the outer segment was stepped into 0 Ca²⁺−0 Na⁺ solution and exposed to a series of eight laser flashes to measure dye fluorescence (as in Fig. 1), in order to determine [Ca²⁺]_i and light-stimulated Ca²⁺ release in the presence of the background. The outer segment was then returned to Ringer solution, and 60 s later a series of laser flashes were given to measure pedestal fluorescence.

Collected results from these experiments are presented in Fig. 7. Figure 7A shows the relative values of photocurrent, calcium and calcium release as a function of background intensity. The photocurrent (•) has been plotted as the means and standard errors of i_B/i_D, for the same nine rods at each background intensity (as in Fig. 6). Calcium concentration (▵) was determined for each rod by subtracting the pedestal (Ped) from the response to the first laser flash (P1) in the presence of the background, and then normalising this to the pedestal value ((P1 − Ped)/Ped). The means and standard errors were then normalised relative to the mean value of this quantity for an additional nine rods in darkness (i.e. without background), measured in parallel from the same retinas. The amplitude of Ca²⁺ release (▿) was calculated as the difference between the responses to the first (P1) and last (P8) laser flashes, again normalised to the pedestal ((P8 − P1)/Ped). This is plotted relative to the mean value of this quantity for the nine rods in darkness.

Figure 7. Effect of background light on current, [Ca²⁺]_i and Ca²⁺ release.

A, photocurrent (•), [Ca²⁺]_i (▵; (P1 − Ped)/Ped) and releasable Ca²⁺ store (▿; (P8 − P1)/Ped) following steady illumination in Ringer solution normalised to matched controls in darkness and plotted as functions of steady background intensity (see text for details). Same cells as Fig. 6 (mean ±s.e.m., n = 9 cells for each data point). Continuous lines join data points for normalised photocurrent. Data have been replotted below by treating background intensity as an implicit variable. B, normalised [Ca²⁺]_i as a function of normalised photocurrent. C, normalised Ca²⁺ release as a function of normalised [Ca²⁺]_i. Continuous curves are lines of unit slope.

The measurements in Fig. 7A, have been replotted in Fig. 7B and C with background intensity as an implicit variable. Figure 7B shows normalised [Ca²⁺]_i as a function of photocurrent, and Fig. 7C normalised Ca²⁺ release as a function of normalised [Ca²⁺]_i. The data in both Fig. 7B and C fall close to a straight line of unity slope passing through the origin. These measurements are therefore consistent with the hypothesis that [Ca²⁺]_i during steady illumination is effectively determined simply by the fraction of open channels in the outer segment, at least for the range of light intensities we have investigated. Furthermore, the size of the pool of Ca²⁺ available to be released seems to be primarily governed by the [Ca²⁺]_i in the cytoplasm.

Recovery of photocurrent and [Ca²⁺]_i after bright light

As a further test of the relationship between photocurrent and [Ca²⁺]_i, we investigated the recovery of the response after a bright, saturating light. The protocol for these experiments is illustrated in Fig. 8. A bright, saturating light was presented for 30 s in Ringer solution. A saturating flash was delivered at a fixed time after the extinction of the background, to measure the time course of current recovery. The rod was then allowed to recover in darkness for several minutes and subsequently presented with the same 30 s light exposure (Fig. 8B). At the same interval after extinguishing the background, the outer segment was stepped into 0 Ca²⁺−0 Na⁺ solution and a series of eight laser flashes delivered to determine [Ca²⁺]_i and the amplitude of light-evoked Ca²⁺ release.

Figure 8. Recovery of photocurrent and [Ca²⁺]_i after bright steady light.

A, rod exposed for 30 s in Ringer solution to saturating light delivering 3.3 × 10³ photons μm⁻² s⁻¹ at 546 nm, followed by a bright flash at a fixed time after the background was extinguished. B, same rod exposed again to the same background intensity in Ringer solution and then stepped to 0 Ca²⁺−0 Na⁺ solution at the time at which the bright flash was delivered in A. Sequence of laser flashes presented 1 s after the solution change to measure light-induced Ca²⁺ release. Arrows denote magnitude of fluorescence evoked by the first and last laser flashes (P1 and P8) and the low level of fluorescence obtained when the laser exposure was repeated following return to Ringer solution (Ped).

Figure 9A shows mean data for the relative photocurrent, calcium and calcium release, calculated as for Fig. 7A; each time interval after extinction of the background corresponds to measurements from 10 rods. The recovery of the circulating current (•) has been fitted with a single exponential with a time constant of 15.4 s. At time zero just after extinction of the saturating light, both [Ca²⁺]_i (▵) and the fraction of Ca²⁺ in the light-releasable pool (▿) were nearly zero, since closure of the channels for a prolonged period produced a substantial decrease in [Ca²⁺]_i and a depletion of Ca²⁺ from the releasable pool (see also Matthews & Fain, 2001). Within 30–60 s, however, [Ca²⁺]_i had returned to virtually its dark-adapted level and the pool had completely refilled; the time course of these processes appeared to correspond closely to the time course of photocurrent recovery. The larger standard errors for Ca²⁺ measurements at longer times probably result from the difficulty of keeping the rod exactly positioned in the laser measuring beam for such an extended period before giving the laser flashes. When time is treated as an implicit variable, [Ca²⁺]_i can be seen to vary nearly linearly with the normalised photocurrent (Fig. 9B), and the size of the releasable pool can again be seen to vary nearly linearly with the level of outer segment Ca²⁺ (Fig. 9C).

Figure 9. Recovery of photocurrent, [Ca²⁺]_i and Ca²⁺ release after bright light.

A, photocurrent (•), [Ca²⁺]_i (▵; (P1 − Ped)/Ped) and releasable Ca²⁺ store (▿; (P8 − P1)/Ped) following the cessation of steady saturating light in Ringer solution normalised to matched controls in darkness and plotted as functions of time after extinction of the background (see text for details). Data were plotted as mean ±s.e.m., with n = 10 cells for each data point. Fitted curve is an asymptotic single exponential with decay constant 15.4 ± 0.8 s. Data have been replotted below by treating time as an implicit variable. B, normalised [Ca²⁺]_i as a function of normalised photocurrent. C, normalised Ca²⁺ release as a function of normalised [Ca²⁺]_i. Continuous curves are lines of unit slope.

Recovery of [Ca²⁺]_i and Ca²⁺ release after a bleach

The data of Fig. 8 and Fig. 9 show that the light-releasable Ca²⁺ pool can be depleted in Ringer solution merely by exposing the rod to saturating illumination. This presumably occurs because the decrease in [Ca²⁺]_i in the outer segment removes Ca²⁺ from cytoplasmic buffer sites or from some store (Matthews & Fain, 2001). Once the pool has been voided, it is rapidly restored when the rod is returned to Ringer solution (Fig. 9A). It is also possible to empty the light releasable pool in 0 Ca²⁺−0 Na⁺ solution when little change in [Ca²⁺]_i occurs, provided the outer segment is exposed to a light that bleaches a significant fraction of the photopigment (Figs 1–4).

To investigate refilling of the pool after intense light exposure, light-induced Ca²⁺ release was measured both from dark-adapted rods, and from rods which had been bleached and then allowed to recover a proportion of their circulating current and sensitivity during a prolonged period in darkness. The results of these experiments are shown in Fig. 10. First, the response-intensity relation (Fig. 10A, •), [Ca²⁺]_i (Fig. 10B, Dark) and light-adapted Ca²⁺ release (Fig. 10C, Dark) were measured from a number of dark-adapted rods. Then a further preparation of rods from the same animal was placed in the recording chamber and exposed to light from the optical bench at an intensity calculated to bleach 99 % of the photopigment. The isolated photoreceptors were then left in darkness for a period of 45–60 min, to permit sensitivity and photocurrent to reach steady state (Cornwall et al. 1990). Response-intensity relations were again recorded (Fig. 10A, half-filled circle), and individual rods were stepped into 0 Ca²⁺−0 Na⁺ solution to measure [Ca²⁺]_i and light-activated Ca²⁺ release.

Figure 10. Refilling of the light-releasable Ca²⁺ pool after pigment bleaching.

A, response–intensity relations for flashes of increasing intensity obtained from dark-adapted rods (•; n = 9 cells), rods exposed to intense light calculated to bleach 99 % of the photopigment and then allowed to stabilise (half-filled circles; n = 9 cells), and rods presented with a second bleaching light exposure (○, n = 9 cells). Relative [Ca²⁺]_i (B) and relative Ca²⁺ release (C) in dark-adapted (Dark), singly bleached (Bleach) and doubly-bleached (Double) rods from A. Data represent means ±s.e.m. Bleaching light exposure of intensity 1.6 × 10⁸ photons μm⁻² s⁻¹ at 546 nm and duration 5 s.

Bleaching produced a pronounced decrease in sensitivity and photocurrent, which previous experiments have shown to be partly the result of a decrease in quantum catch, and partly of stimulation of the visual cascade by opsin (Cornwall & Fain, 1994). The resting fluorescence (Fig. 10B, Bleach) also decreased, falling by about the same amount as the photocurrent in Fig. 10A (see also Sampath et al. 1999). The pool of Ca²⁺ available for release (Fig. 10C, Bleach) was also smaller, again by about the same fraction as the decrease in [Ca²⁺]_i. Since the pool will initially have been completely emptied by exposure to light of this intensity (see Fig. 1C and Fig. 3B), at least a part of the Ca²⁺ must have been restored during the recovery period in darkness.

We were surprised to see such a large recovery of Ca²⁺ release after bleaching so much of the visual pigment. It seemed to us possible that at least part of the recovery might reflect rhodopsin regeneration, since even isolated rods retain some capacity for regenerating photopigment (Azuma et al. 1977), perhaps from a small outer segment reserve of 11-cis-retinal. In order to investigate this possibility, a further group of rods was exposed to two bleaches in sequence, both of the same intensity; the second bleach was delivered 60 min after the first. Since any outer segment reserve of chromophore should have been exhausted during recovery from the first bleach, the second bleach should have left the rods with very little remaining rhodopsin. A further 45–60 min after this second bleach, the rods showed an additional decrease in photocurrent and sensitivity (Fig. 10A, ○). Furthermore, both the resting [Ca²⁺]_i (Fig. 10B, Double) and the pool of releasable Ca²⁺ (Fig. 10C, Double) were substantially depressed. The decrease in the size of the releasable pool was relatively larger than the decrease in the resting [Ca²⁺]_i, suggesting that pigment bleaching may affect the ability of the store to refill, independent of the value of [Ca²⁺]_i.

DISCUSSION

Since Ca²⁺ is an essential second messenger in vertebrate photoreceptors that modulates several steps in the transduction cascade, making an important contribution to the recovery phase of the light response and to light adaptation, it is important to know how [Ca²⁺]_i changes during and after illumination We have therefore taken advantage of techniques we have developed for measuring circulating current, [Ca²⁺]_i and light-induced Ca²⁺ release to study changes in [Ca²⁺]_i as a function of light intensity and the open probability of cyclic nucleotide-gated channels. Our experiments show that light changes the [Ca²⁺]_i in a rod outer segment predominantly by decreasing influx through the cyclic-nucleotide-gated channels across the plasma membrane. In background light or after a bright flash, [Ca²⁺]_i decreases essentially in proportion to the decrease in the circulating current (Fig. 7B and Fig. 9B). As the circulating current recovers, so does [Ca²⁺]_i. These results are in accord with previous observations (reviewed in Fain et al. 2001) that the flux of Ca²⁺ across the plasma membrane is the primary determinant of the [Ca²⁺]_i in rod outer segments.

We were unable to substantiate a prolonged decrease in [Ca²⁺]_i during background lights or after bright flashes disproportionate to the reduction in circulating current, as previous experiments on isolated Gekko rod outer segment seemed to have indicated (Gray-Keller & Detwiler, 1994). There are, however, a number of differences in method between our experiments and this earlier study. In these previous experiments the dye Indo-dextran was introduced into isolated rod outer segments from a patch pipette which remained in the whole-cell configuration throughout the experiment to enable simultaneous measurements of [Ca²⁺]_i and voltage-clamped current. The free Ca²⁺ concentration of the patch pipette solution (4 nm) was considerably lower than the dark resting Ca²⁺ concentration of the outer segment. Even though the region near the pipette was masked during the Ca²⁺ measurement, it seems possible that diffusion between pipette and outer segment may have influenced the time course of [Ca²⁺]_i during and after illumination. Alternatively, there may be some difference in the physiology between the isolated outer segments used in their experiments and the intact rods of the present investigation.

Although changes in Ca²⁺ flux across the plasma membrane are primarily responsible for changes in outer segment [Ca²⁺]_i, there is now ample evidence that an increase in [Ca²⁺]_i can also occur from the release of Ca²⁺ from a buffer site or sequestered pool within the outer segment (Matthews & Fain, 2001, 2002; Brockerhoff et al. 2003). Our measurements show that [Ca²⁺]_i can be released from this pool in two ways: by exposure to saturating illumination in Ringer solution, presumably as a result of the decrease in outer segment [Ca²⁺]_i; and by bright bleaching light, in 0 Ca²⁺−0 Na⁺ solution, when little change in [Ca²⁺]_i occurs (see also Matthews & Fain, 2001). In Ringer solution, the amount of Ca²⁺ in the releasable pool declines in proportion to [Ca²⁺]_i (Fig. 7C and Fig. 9C), and the pool recovers in proportion to the recovery of [Ca²⁺]_i (Fig. 9A). The releasable Ca²⁺ pool seems therefore to behave as if it were freely exchangeable with cytosolic Ca²⁺, possibly suggesting that it may represent Ca²⁺ bound to disk membrane or cytosolic protein rather than a sequestered store. It seems unlikely that the Ca²⁺ is coming from a sequestered pool within the disks, since a similar light-dependent release has now been detected in cones (Leung et al. 2002; Brockerhoff et al. 2003).

Light-dependent release can also occur even when there is little change in outer segment Ca²⁺, as in 0 Ca²⁺−0 Na⁺ solution, provided the light is made sufficiently bright (Fig. 3B). This release is occurring from the same pool of Ca²⁺ as the one depleted by low [Ca²⁺]_i (see Matthews & Fain, 2001), and the signal that triggers the release is localised to the region of outer segment that is illuminated (Fig. 4). A simple model consistent with these observations is that Ca²⁺ might be released from a buffer site whose affinity is light dependent. Although this release requires strong illumination, its magnitude in 0 Ca²⁺−0 Na⁺ is not simply proportional to the fraction of rhodopsin bleached but is half-maximal when less than 10 % of the photopigment has been activated. This observation seems to exclude a simple mechanism for the release, such as a change in the conformation of a binding site directly on the rhodopsin molecule. The intensity dependence of the release is reasonably well fitted by a single exponential, as if it were occurring from a finite pool at a rate proportional to light intensity. It is possible that the interaction of activated rhodopsin with a protein like recoverin that binds Ca²⁺ may change the conformation of this protein to induce the release of previously bound Ca²⁺. Rod outer segments contain a number of Ca²⁺-binding proteins (see Polans et al. 1996), whose affinity for Ca²⁺ might be affected by interaction with bleached rhodopsin. Our previous experiments showing that Ca²⁺ release takes place with bi-exponential kinetics suggest that more than one site may be involved (Matthews & Fain, 2002). At present, too little is known about Ca²⁺ buffering in outer segments to form a more definite hypothesis. Once the Ca²⁺ has been released by bright light, it can be restored to the releasable pool, in part as the result of the recovery of outer segment [Ca²⁺]_i, but perhaps in part also as the result of regeneration of the photopigment (see Fig. 10).

Our experiments show that the release plays only a minor role in determining [Ca²⁺]_i in the rod outer segment within this receptor's normal operating range of light intensities from threshold to increment saturation. This may not be true in cones, however, which are capable of responding to much brighter light. In a cone, light-dependent Ca²⁺ release has been shown to produce a decrease in circulating current, even in the absence of transducin, evoking a response independent of the normal transduction cascade (Brockerhoff et al. 2003). The functional role of this release in cones may become clearer through study of the light and [Ca²⁺]_i dependence of this process.