Free Magnesium Concentration in Salamander Photoreceptor Outer Segments

Authors

  • Chunhe Chen,

    1. Department of Physiology and Biophysics, University of Colorado Health Sciences Center, Denver, CO 80262, USA
    Search for more papers by this author
  • Kei Nakatani,

    1. Institute of Biological Sciences, University of Tsukuba, Tsukuba, Ibaraki 305, Japan
    Search for more papers by this author
  • Yiannis Koutalos

    Corresponding author
    1. Department of Physiology and Biophysics, University of Colorado Health Sciences Center, Denver, CO 80262, USA
    • Corresponding author
      Y. Koutalos: University of Colorado Health Sciences Center, Department of Physiology and Biophysics, Box C-240, 4200 East Ninth Avenue, Denver, CO 80262, USA. Email: yiannis.koutalos@uchsc.edu

    Search for more papers by this author

Abstract

Magnesium ions (Mg2+) play an important role in biochemical functions. In vertebrate photoreceptor outer segments, numerous reactions utilize MgGTP and MgATP, and Mg2+ also regulates several of the phototransduction enzymes. Although Mg2+ can pass through light-sensitive channels under certain conditions, no clear extrusion mechanism has been identified and removing extracellular Mg2+ has no significant effect on the light sensitivity or the kinetics of the photoresponse. We have used the fluorescent Mg2+ dye Furaptra to directly measure and monitor the free Mg2+ concentration in photoreceptor outer segments and examine whether the free Mg2+ concentration changes under physiological conditions. Resting free Mg2+ concentrations in bleached salamander rod and cone photoreceptor cell outer segments were 0.86 ± 0.06 and 0.81 ± 0.09 mm, respectively. The outer segment free Mg2+ concentration was not significantly affected by changes in extracellular pH, Ca2+ and Na+, excluding a significant role for the respective exchangers in the regulation of Mg2+ homeostasis. The resting free Mg2+ concentration was also not significantly affected by exposure to 0 Mg2+, suggesting the lack of significant basal Mg2+ flux. Opening the cGMP-gated channels led to a significant increase in the Mg2+ concentration in the absence of Na+ and Ca2+, but not in their presence, indicating that depolarization can cause a significant Mg2+ influx only in the absence of other permeant ions, but not under physiological conditions. Finally, light stimulation did not change the Mg2+ concentration in the outer segments of dark-adapted photoreceptors. The results suggest that there are no influx and efflux pathways that can significantly affect the Mg2+ concentration in the outer segment under physiological conditions. Therefore, it is unlikely that Mg2+ plays a significant role in the dynamic modulation of phototransduction.

Magnesium ions (Mg2+) are indispensable for enzyme activity, as cofactors for ATP and GTP, and in the modification of channel function. The free Mg2+ concentration inside vertebrate cells is relatively stable and the turnover across the cellular plasma membrane appears to be extremely slow (for review see Romani & Scarpa, 2000). So, while the presence of Mg2+ is necessary for cellular function, it does not modulate cell function through changes in concentration in the cytoplasm. Consequently, the role of Mg2+ is very different from that of Ca2+. One reason for this difference is that the concentration of free Mg2+ in the cytoplasm is in the millimolar range (for review see London, 1991). This high resting concentration is unlikely to change in response to typical cation fluxes across the plasma membrane. Moreover, there is a large difference in the radius of hydrated and non-hydrated Mg2+, making coordination to proteins and a role as a transient regulator of protein function less likely.

Nevertheless, large Mg2+ fluxes across cell membranes have been observed in response to different hormonal stimuli in a variety of cell types, including liver and cardiac cells (for example, Vormann & Gunther, 1987; Romani & Scarpa, 1990; Romani et al. 1993). In the case of vertebrate photoreceptor cells, it is not yet clear whether there are significant Mg2+ fluxes across the cell membrane and whether Mg2+ plays any signalling role. Vertebrate photoreceptors are responsible for the conversion of incoming light to an electrical signal. Rod photoreceptors underlie vision at low light intensities, while cones underlie vision at high light intensities (for reviews see Fain et al. 2001; Ebrey & Koutalos, 2001). The phototransduction process takes place in the photoreceptor outer segment, where numerous reactions utilize MgGTP and MgATP, while Mg2+ itself regulates several of the phototransduction enzymes. As Mg2+ can enter the rod outer segment through light-sensitive channels in the dark (Nakatani & Yau, 1988), one might expect that closure of the channels by light would lead to a decrease in the Mg2+ concentration in the outer segment. Hence, it is possible that Mg2+ may also play a role in the dynamic modulation of phototransduction. However, no clear Mg2+ extrusion mechanism has been identified and removing extracellular Mg2+ for short periods of time has no significant effect on the light sensitivity or the kinetics of the photoresponse (Nakatani & Yau, 1988).

We have used the Mg2+-sensitive fluorescent dye Furaptra to directly measure and monitor the free Mg2+ concentration in photoreceptor outer segments and examine whether the free Mg2+ concentration is modulated in any way. A preliminary report of these results has appeared in abstract form (Koutalos et al. 2002).

Methods

Intact, isolated rod and cone photoreceptors were obtained from larval tiger salamander (Ambystoma tigrinum) retinas (Koutalos et al. 1995a). Animals were dark-adapted overnight and killed under dim red light by decapitation and pithing, which swiftly destroyed the brain and the spinal cord. The eyes were removed and subsequent manipulations were carried out in the dark with the aid of infrared image converters. The eyes were hemisected, and the retinas were isolated in amphibian Ringer solution (mm: 110 NaCl, 2.5 KCl, 1.6 MgCl2, 1 CaCl2, 5 Hepes, 5 glucose, pH = 7.55). For experiments with bleached cells, the retinas were bleached for 5 min with intense white light (60 W lamp at 10 cm) and subsequently incubated under room lights at room temperature with 1 % fatty-acid free bovine serum albumin (BSA) for 60 min, to remove the retinoids produced by bleaching. Control experiments with dark-adapted and BSA-treated retinas solubilized in 1 % Ammonyx-LO showed that no visual pigment was detectable in BSA-treated retinas by a spectrophotometric assay (> 99.7 % pigment bleached). Removal of the retinoids removes the long-lived rhodopsin photoproducts (Jager et al. 1996) and reduces the activity of the light-activated phosphodiesterase, allowing effective inhibition by 1 mm isobutyl-methyl-xanthine (IBMX).

Isolated, intact photoreceptor cells were obtained by chopping the retina with a razor blade under Ringer solution in a Petri dish coated with Sylgard elastomer (Dow Corning). Although the spectral sensitivity of the cones was not tested, they were most likely to be long-wavelength-sensitive (‘red’), on the basis of their morphology and since ‘red’ cones comprise ≈80 % of the cones obtained with the isolation procedures used (Perry & McNaughton, 1991). Isolated cells were transferred to 100 µl chambers that fitted on the microscope stage. The bottoms of these chambers were made of coverslips and were coated with concanavalin A. Cells were incubated in the dark and at room temperature for 30 min with 10 µm of the acetoxymethyl ester forms of the Mg2+-sensitive probe Furaptra or the Ca2+-sensitive probe Fluo-3 (TEF Labs, Austin, TX, USA; Molecular Probes, Eugene, OR, USA). Non-internalized dye was then washed away with Ringer solution, and the cells were ready for imaging. Cell autofluorescence (including the fluorescence of any outer segment all-trans retinol (Kaplan, 1984) or mitochondrial NADH) was not a concern, as it was much less than the fluorescence of internalized Furaptra.

Fluorescence imaging experiments were carried out on the stage of an inverted Zeiss Axiovert 100 microscope, using a xenon continuous arc light source from Sutter Instrument Company (Novato, CA, USA), a Zeiss × 40 Plan Neofluar objective lens, and a SensiCam CCD camera (Cooke Corporation, Auburn Hills, MI, USA). Fluo-3 fluorescence was excited at 490 nm and measured at 530 nm. The Fluo-3 fluorescence was used as an indicator of the free Ca2+ concentration. The ratio F340/F380 of Furaptra fluorescence was obtained by exciting sequentially at 340 and 380 nm and measuring emission at 540 nm with 50 nm bandwidth, i.e. 515-565 nm (although these filter settings are more than adequate for measuring Furaptra fluorescence emission, they are not, strictly speaking, optimal). The F340/F380 Furaptra fluorescence ratio was used as a measure of the free Mg2+ concentration. For each cell, regions of interests within the outer segment, the ellipsoid, or the nuclear areas were defined and the average fluorescence intensity was measured for each excitation wavelength. Average background fluorescence intensity for each wavelength was measured from an area near the cell that contained no other cell or debris. After subtracting background fluorescence, the F340/F380 ratio was calculated for each region of interest. As cone outer segments are small, cone outer segment data are typically quite noisy. Acquisition of the ratio data for each time point took less than a second. This total acquisition time per data point included the acquisition times at each wavelength (500 ms at 340 nm and 100 ms at 380 nm), the excitation wavelength switching time (≈100 ms), and acquisition software delays. It is unlikely that the free Mg2+ concentration can change significantly within seconds, so the 1 s resolution should be sufficient. Indeed, for experiments with BSA-washed cells, free Mg2+ changes occurred on a time scale of minutes if at all. For experiments with dark-adapted cells, we also carried out separate rapid measurements (100 ms resolution) of the relative fluorescence changes at 340 and 380 nm, demonstrating that the relative F340/F380 ratio did not change, even on a much shorter time scale.

Fluorescence ratios were converted to free Mg2+ concentrations after calibrating intracellular dye with different Mg2+ concentrations. For calibrations, Furaptra-loaded cells were exposed to a pseudointracellular solution (110 mm KCl, 5 mm Hepes, 5 mm glucose, pH = 7.55) containing different concentrations of Mg2+ and an ionophore cocktail (20 µm 4-bromo-A23187, 20 µm nigericin, 20 µm monensin) designed to collapse ionic gradients and equilibrate intracellular and extracellular Mg2+ concentrations (Tashiro & Konishi, 1997). The ionophore cocktail did not contribute any significant fluorescence. In the presence of the ionophore cocktail, the Furaptra F340/F380 ratio rapidly followed changes in the extracellular Mg2+ concentration, showing that it provides a direct measure of the free Mg2+ concentration. Figure 1 shows the calibration for internalized Furaptra in rod and cone outer segments. The data points were fitted with curves of the form [Mg2+]=K × (R - Rmin) / (Rmax - R), where [Mg2+] is the extracellular Mg2+ concentration, R=F340/F380, and K, Rmin and Rmax are the parameters to be determined from the fit (Grynkiewicz et al. 1985). These equations can then be used to convert ratios R to free intracellular Mg2+ concentrations. For rods, K= 3.88 mm, Rmin= 0.035 and Rmax= 0.069; for cones, K= 2.21 mm, Rmin= 0.034 and Rmax= 0.064. The Rmax values clearly underestimate the values that would be obtained if higher extracellular Mg2+ concentrations were used. However, the physiological free Mg2+ concentrations are expected to be well within the range of the concentrations employed, and, although Rmax is underestimated, the fits are appropriate for converting measured ratios to free Mg2+ concentrations. We carried out the calibration specifically for determining the value of the intracellular free Mg2+ concentration under physiological conditions, and all the relevant measurements were within the calibration range. For the rest of the experiments, the changes in Mg2+ concentration are reported as relative ratios, that is, ratios normalized to the initial fluorescence ratio. Image acquisition and analysis were carried out using the Intelligent Imaging Innovations (Denver, CO, USA) software.

Figure 1.

Intracellular Furaptra calibration curves for isolated salamander rods (○) and cones (▴)

Isolated cells were incubated with Furaptra-AM, and Furaptra-loaded cells were exposed to a pseudointracellular solution containing different concentrations of Mg2+ and an ionophore cocktail that allowed the equilibration of intracellular and extracellular Mg2+ concentrations. The curve is a least squares fit to the rod data (see text). The error bars represent standard deviations, and are not shown if smaller than the symbol size. Data were obtained from 26 rods and 12 cones.

One concern with the use of the AM forms to load cells with fluorescent probes is whether breakdown products interfere with the fluorescence measurements or with the cell's physiology. We have carried out experiments and characterizations with different kinds of AM dyes in outer segments (see Chen et al. 2002). Fura-2 and Fluo-3 produced from the AM forms give the same resting Ca2+ concentration. Calcein and Fluo-3 produced from the AM forms diffuse with the same diffusion coefficient as cGMP (Chen et al. 2002; Nakatani et al. 2002). Finally, Sampath et al. (1998) have shown that Fluo-3 introduced via a patch pipette gives the same results as Fluo-3 produced from the AM form. On the basis of these observations, we expect that the AM forms (including that of Furaptra, which, admittedly, we have not characterized separately) are de-esterified into the expected products in the outer segment cytoplasm, behave accordingly, and do not interfere with the physiology of the cells. Another, more general, concern is whether the F340/F380 ratio is a reliable indicator of free Mg2+ concentration changes. The calibration experiments suggest that the F340/F380 ratio reflects the free Mg2+ concentration. Moreover, in the presence of the ionophore cocktail, the F340/F380 changes rapidly in response to extracellular Mg2+ concentration changes (data not shown). The calibration experiments are also an indication that internalized Furaptra behaves as expected.

Finally, photoreceptor cells isolated from BSA-treated retinas appear to be similar to cells with their outer segment cGMP-gated channels closed, as judged by their Ca2+ concentration at rest. Cells isolated from BSA-treated retinas were loaded with the Ca2+-sensitive dye Fluo-3 and the dye fluorescence was calibrated as described previously (Chen et al. 2002). Rods and cones had resting Ca2+ concentrations of 18 ± 3 nm (n= 5) and 12 ± 4 nm (n= 7), respectively, values similar to those measured in outer segments under saturating light conditions (Gray-Keller & Detwiler, 1994; Sampath et al. 1998, 1999; Chen et al. 2002). Exposure to light did not change the Ca2+ concentration as monitored by the fluorescence of internalized Fluo-3, consistent with the lack of any significant amount of visual pigment left by the end of the BSA wash.

For experiments with 0 Mg2+, MgCl2 was removed from the Ringer solution; for 0 Na+, choline chloride was substituted for NaCl; for 0 Ca2+, CaCl2 was removed and 0.2 mm EGTA was added. Unless otherwise stated, all chemicals were from Sigma Chemical Company. All experiments were carried out at room temperature.

Results

In order to obtain the free Mg2+ concentration in rod and cone photoreceptor cell outer segments, we loaded cells with Furaptra and measured the ratio F340/F380 of dye fluorescence excited at 340 and 380 nm (emission: 515-565 nm). Fluorescence ratios were converted to free Mg2+ concentrations according to the calibrations shown in Fig. 1. These experiments were carried out with bleached photoreceptors, so that changes in the activity of the phototransduction cascade elicited by the fluorescence excitation light would not interfere with the measurements. The value of the free Mg2+ concentration in darkness is addressed below (Fig. 7). Figure 2 shows the histograms of the free Mg2+ concentrations for rods and cones. For rods, the average free Mg2+ concentration was 0.86 ± 0.06 mm (n= 114, range 0.23-3.66 mm), and for cones it was 0.81 ± 0.09 mm (n= 15, range 0.44-1.47 mm). Although only healthy looking cells were selected for the measurements, the higher free Mg2+ concentrations may indicate problems with metabolic regulation (see Discussion). The median values were 0.67 and 0.69 mm for rods and cones, respectively.

Figure 7.

The outer segment free Mg2+ concentration does not change after stimulation of dark-adapted photoreceptors by light

Dark-adapted salamander cones (A, n= 7, ▴) and rods (B, n= 6, ○) loaded with Furaptra. The light used for measuring the Furaptra fluorescence was also the light that stimulated the cell, leading to closure of the cGMP-gated channels. Error bars represent standard errors. C, the free Ca2+ concentration decreases in the outer segments of dark-adapted rod photoreceptors after stimulation by light. Dark-adapted isolated salamander rod photoreceptors were loaded with Fluo-3. The light used for measuring the Fluo-3 fluorescence was also the light that stimulated the cell, leading to closure of the cGMP-gated channels. Changes in free Ca2+ concentration were followed from the relative changes in Fluo-3 fluorescence, normalized to the initial value. Light leads to a rapid decrease in the free Ca2+ concentration in the rod outer segments. The excitation light was on continuously, and images were acquired every 530 ms. Error bars represent standard errors (n= 8).

Figure 2.

Distributions of resting free Mg2+ concentrations in bleached (A) rod and (B) cone cell outer segments

For rods, the free Mg2+ concentration was 0.86 ± 0.06 mm (mean ±s.e.m., n= 114) and for cones 0.81 ± 0.09 mm (mean ±s.e.m., n= 15). The micrographs show fluorescence images (excitation 380 nm, emission 515-565 nm) of an isolated salamander rod and a cone, loaded with Furaptra. The outer segments of the cells are the less brightly stained compartments in the lower parts of the micrographs. Scale bar is 10 µm.

The main question to address was whether the free Mg2+ concentration in the photoreceptor outer segment changes under physiological conditions. One possible route of Mg2+ entry is the cGMP-gated channels, which are known to be permeable to Mg2+ (Nakatani & Yau, 1988; Colamartino et al. 1991; Wells & Tanaka, 1997). Bleached cells have a significant phosphodiesterase activity that keeps the cGMP concentration low and the cGMP-gated channels closed. We inhibited the phosphodiesterase activity with 1 mm IBMX, thereby increasing the cGMP concentration and opening the cGMP-gated channels. Fig. 3 shows the results of such experiments. Opening the cGMP-gated channels in the presence of physiological concentrations of Na+ and Ca2+ had no discernible effect on the outer segment free Mg2+ concentration. In the case of cones, any increase in the Furaptra ratio was less than 5 %, corresponding to a free Mg2+ increase of less than ≈0.2 mm (calculated for a ratio at rest of ≈0.041). In the case of rods, any Furaptra ratio change was less than 1 %, corresponding to a free Mg2+ change of less than 0.05 mm. However, if the channels were opened in the absence of Na+ and Ca2+, the outer segment free Mg2+ concentration increased, indicating Mg2+ influx. In the case of cones, the average ratio increased by ≈20 %, corresponding to an increase in free Mg2+ by ≈1.6 mm (reaching on the average concentrations of ≈2.4 mm), while for rods, the average ratio increased by ≈8 %, corresponding to an increase in free Mg2+ concentration of ≈0.6 mm (reaching, on average, concentrations of ≈1.4 mm). Because rod cells were not as resilient as cones in terms of surviving the experimental manipulations, the IBMX exposures in the presence and absence of Na+ and Ca2+ were carried out with different cells (Fig. 3B). Figure 3 also shows that the free Mg2+ concentration increases over several minutes upon exposure to 1 mm IBMX, that is, much more slowly than the concentration of free Ca2+ (see Fig. 6). This is not unexpected, as the resting Mg2+ concentration is ≈0.8 mm, which is much higher than the resting concentration of Ca2+. As a result, the Mg2+ concentration will change relatively slowly, even in response to typical cation fluxes. In addition, the Mg2+ influx is probably significantly smaller than the typical cation influx.

Figure 3.

Mg2+ flows into photoreceptor outer segments in the absence of Na+ and Ca2+, but not in their presence

Isolated bleached salamander photoreceptors were loaded with Furaptra, and changes in free Mg2+ concentration were measured in response to 1 mm IBMX. IBMX inhibits the cGMP-phosphodiesterase, leading to an increase in the outer segment cGMP concentration and opening of the cGMP-gated channels. Free Mg2+ concentration changes were followed through the relative F340/F380 ratio. A, averaged data from 14 cone cells. The cells were initially in Ringer solution; after 5 min they were exposed for 15 min to Ringer solution containing 1 mm IBMX, but no significant change in the outer segment Mg2+ concentration was observed; the cells were returned to Ringer solution and, after another 5 min, were exposed for 15 min to a 0 Na+, 0 Ca2+ solution, containing 1 mm IBMX, which resulted in a significant increase in the outer segment free Mg2+ concentration; finally, the cells were returned to regular Ringer solution and the free Mg2+ concentration began to recover. B, data from rod cells. The cells were initially in Ringer solution; after 5 min the cells were exposed for 15 min to solutions containing 1 mm IBMX, and significant changes in the outer segment Mg2+ concentration were observed for the rods in the 0 Na+, 0 Ca2+ solution (n= 7, •), but not for the rods in Ringer solution (n= 9, Δ); finally, the cells were returned to regular Ringer solution and the free Mg2+ concentration recovered. Error bars represent standard errors.

Figure 6.

Ca2+ flows into rod and cone photoreceptor outer segments under conditions that Mg2+ does not

Isolated bleached salamander rod and cone photoreceptors were loaded with Fluo-3, and changes in free Ca2+ concentration were measured in response to 1 mm IBMX. Changes in free Ca2+ concentration were followed from the relative changes in Fluo-3 fluorescence. Fluorescence was normalized to the initial value. The cells were initially in Ringer solution; after 1 min the cells were exposed for 4 min to Ringer solution containing 1 mm IBMX, and large changes in the outer segment Ca2+ concentration were observed; finally, the cells were returned to regular Ringer solution and the free Ca2+ concentration recovered. A, averaged data from 9 isolated bleached cone cells. B, averaged data from 6 isolated bleached rod cells. Error bars represent standard errors.

One issue with the experiments of Fig. 3 is whether the increase in free Mg2+ in the absence of Na+ and Ca2+ is due to an increase in Mg2+ influx or due to a decrease in Mg2+ efflux. One possibility is that there is a Na+- or Ca2+-dependent Mg2+ extrusion mechanism that is inactivated in the absence of Na+ or Ca2+. However, the outer segment free Mg2+ concentration was not significantly affected by exposures to 0 Na+ or 0 Ca2+ solutions in the absence of IBMX, excluding a significant role for the respective exchangers (data not shown). Similarly, changes in extracellular pH between 6.50 and 8.00 did not significantly affect the outer segment free Mg2+ concentration (data not shown). Therefore, the change in free Mg2+ is due to an increase in Mg2+ influx. Indeed, Mg2+ has been shown to enter through the cGMP-gated channels in the absence of Na+ and Ca2+, as it sustains a light-sensitive current (Nakatani & Yau, 1988). Experiments with excised patches from rod outer segments have also shown that Mg2+ can enter through the cGMP-gated channels (Colamartino et al. 1991; Wells & Tanaka, 1997).

It is possible that the stability of the free Mg2+ concentration is due to the presence of a large influx of Mg2+ through the photoreceptor plasma membrane, balanced by a large efflux. If this were the case, opening the cGMP-gated channels with IBMX could have a negligible effect on the intracellular free Mg2+ concentration. If such fluxes were present, removal of extracellular Mg2+ should result in a decrease in the intracellular free Mg2+ concentration, as influx would cease while extrusion continued. Figure 4A shows that exposure to 0 Mg2+ in the absence of IBMX has no discernible effect on the free Mg2+ concentration, arguing against the presence of a large efflux. Experiments with rods, although they never showed a decrease in intracellular free Mg2+, were not as consistent: sometimes they showed an increase in intracellular free Mg2+, perhaps due to metabolic stress (see Discussion). Exposure to 0 Mg2+ in the presence of IBMX also did not lead to a decrease in the outer segment free Mg2+ concentration, suggesting that Mg2+ does not flow out through the cGMP-gated channels in the presence of Na+ and Ca2+ (Fig. 4B).

Figure 4.

There is no significant efflux of Mg2+ from cone photoreceptor outer segments

Isolated bleached salamander cone photoreceptors were loaded with Furaptra, and changes in free Mg2+ concentration were measured through the relative F340/F380 ratio, in response to exposures to 0 Mg2+ with or without 1 mm IBMX. A, averaged results from 6 cone cells. The cells were initially in Ringer solution; after 5 min they were exposed for 15 min to a 0 Mg2+ Ringer solution, but no significant change in the outer segment Mg2+ concentration was observed; the cells were returned to Ringer solution and, after another 5 min, they were exposed for 15 min to a 0 Na+, 0 Ca2+ solution, containing 1 mm IBMX, which resulted to a significant increase in the outer segment free Mg2+ concentration; finally, the cells were returned to regular Ringer solution and the free Mg2+ concentration began to recover. B, averaged results from 6 different cone cells. The cells were initially in Ringer solution; after 5 min they were exposed for 25 min to a 0 Mg2+ Ringer solution containing 1 mm IBMX. Again, no significant change in the outer segment Mg2+ concentration was observed. Error bars represent standard errors.

The experiments of Fig. 3 and Fig. 4 demonstrate that the outer segment free Mg2+ concentration does not change significantly upon opening of the cGMP-gated channels under physiological ionic conditions. The opening of the cGMP-gated channels also depolarizes the cell, which would activate and open cation channels in the cell body as well. The lack of change in the free Mg2+ concentration does not necessarily mean that there is no significant influx through all these different types of channels upon depolarization, as entering Mg2+ could be strongly buffered or taken up by internal stores. A large proportion of Mg2+ would be bound to ATP and GTP, and this buffering would dampen increases in the free Mg2+ concentration. Internal stores may also play a substantial part by sequestering the Mg2+ flowing through plasma membrane channels. We have examined the second possibility by comparing the changes in the free Mg2+ concentration in other photoreceptor compartments, specifically the mitochondria in the ellipsoid region, and the nucleus. Figure 5 shows the changes in free Mg2+ concentration in the mitochondrial compartment for the experiments of Fig. 3A and Fig. 4A. Similar changes were observed for the nuclear compartment. So the free Mg2+ changes in the mitochondria and nucleus parallel those in the outer segment. The lack of changes in the compartments when there is no change in the outer segment, along with the increases in the compartments when free Mg2+ increases in the outer segment, make it unlikely that Mg2+ is sequestered into a bound form, and is consistent with the lack of significant Mg2+ influx or efflux.

Figure 5.

The free Mg2+ concentration changes in the mitochondrial compartment parallel those in the outer segment

Isolated bleached salamander cone photoreceptors were loaded with Furaptra, and changes in free Mg2+ concentration were measured in response to 1 mm IBMX (A) and upon exposure to 0 Mg2+ (B). A, data from the same experiments as shown in Fig. 3A. B, data from the same experiments as shown in Fig. 4A. Error bars represent standard errors.

The experimental treatment of Fig. 3 and Fig. 4 (exposure to 1 mm IBMX) does result in significant opening of the cGMP-gated channels, as it gives rise to robust Ca2+ concentration changes (Fig. 6). Isolated cells were loaded with the Ca2+-sensitive dye Fluo-3, and Ca2+ concentration changes were followed by the relative fluorescence change, F/F0, that is, fluorescence normalized to its initial value. Experiments with other cells have shown that the IBMX-induced change in Ca2+ persists for at least 30 min (data not shown), suggesting that the cGMP-gated channels remain open for a prolonged period of time. We did not calibrate the internalized Fluo-3 for every cell in these particular experiments to obtain Ca2+ concentration values; however, calibration experiments with other cells (n= 5) indicate that 1 mm IBMX increases the Ca2+ concentration to 150-300 nm. This concentration represents a significant fraction of the dark-adapted level of 500-700 nm (Gray-Keller & Detwiler, 1994; Sampath et al. 1998; Matthews & Fain, 2002), indicating significant opening of the cGMP-gated channels. As the steady-state outer segment Ca2+ concentration is approximately proportional to the current through the cGMP-gated channels (Gray-Keller & Detwiler, 1994; Koutalos et al. 1995b), we estimate that exposure to 1 mm IBMX gives rise to ≈40 % of the dark current.

The results presented up to this point show that the opening of the cGMP-gated channels does not seem to have any significant, direct or indirect, effect on the free Mg2+ concentration. Therefore, one would expect that closure of the cGMP-gated channels upon light stimulation of dark-adapted photoreceptors should not affect the free Mg2+ concentration either. Figure 7A and B shows this to be the case for both cones (▴) and rods (○). Any ratio change was less than 1 %, corresponding to a free Mg2+ change of less than 0.05 mm for either rods or cones. Although the free Mg2+ concentration is unlikely to change rapidly, the slow time scale (every 5 s) of the measurements shown in Fig. 7A and B was of concern. In order to confirm that the outer segment free Mg2+ concentration did not change in a rapid time scale after light stimulation, we separately measured the changes in fluorescence excited with 380 or 340 nm light. By measuring the excitation from a single wavelength, we were able to carry out measurements on a shorter time scale (every 100 ms for 30 s). F380 and F340 changed in parallel in the light, so the relative F340/F380 ratio did not change (data not shown), thereby confirming that the free Mg2+ concentration does not change over a shorter time scale either. We conclude that the Mg2+ concentration does not change in the light. Figure 7C is a control showing the well known light-induced decrease in Ca2+ concentration in rod outer segments. Dark-adapted rod photoreceptors were loaded with Fluo-3, and the dye fluorescence was monitored after the fluorescence excitation light was switched on. The light used for the excitation of Fluo-3 fluorescence (490 nm) was also the stimulating light leading to closure of the cGMP-gated channels. Switching on the light led to a rapid decrease in Ca2+ concentration and Fluo-3 fluorescence with kinetics in general agreement with previously published reports (Gray-Keller & Detwiler, 1994; Sampath et al. 1998).

Discussion

The free intracellular Mg2+ concentration in photoreceptor outer segments is essentially the same in rods and cones, as the measured values, 0.86 ± 0.06 and 0.81 ± 0.09 mm, respectively, are not significantly different. A few photoreceptors had relatively large free Mg2+ concentrations, which may be the result of metabolic stress. Such stress would involve ATP and GTP depletion and release of ATP- and GTP-bound Mg2+. Since even the immediate byproducts of ATP breakdown, ADP and AMP, bind to Mg2+ with a significantly lower affinity than ATP (Martell & Smith, 1974), ATP depletion gives rise to the increased free Mg2+ values. The largest free Mg2+ concentrations we observed in rod outer segments are ≈3.5 mm (3 cells out of 114), which would correspond to an extra ≈2.5 mm Mg2+ above the average concentration. Two cells had free Mg2+ concentrations between 2.0 and 2.5 mm, which would correspond to an extra ≈1.5 mm Mg2+ above the average concentration. The extra 1.5-2.5 mm Mg2+ could come from release from nucleotide triphosphates (ATP and GTP), of which there is at least 5 mm in total in outer segments (Robinson & Hagins, 1979). ATP and GTP depletion may also be the cause of the observed increases in intracellular free Mg2+ after exposures to 0 extracellular Mg2+. The number of cells with high resting Mg2+ was small and did not significantly affect the estimated average free Mg2+ value. For comparison, the median values were 0.67 and 0.69 mm for rods and cones, respectively, only slightly less than the averages. As the free Mg2+ concentration did not change with light stimulation, these values are also the values of the free Mg2+ concentration in darkness. These values are comparable to those found in other cell types (London, 1991).

The free Mg2+ concentration in photoreceptor outer segments is significantly lower than the extracellular concentration of 1.6 mm (in Ringer solution). Taking into account the negative membrane potential (-30 mV in the dark and -65 mV at the peak of the photoresponse; see Schwartz, 1973; Baylor et al. 1974), there is a large driving force for Mg2+ flow into the outer segment, and the equilibrium intracellular concentration would be between 16 and 160 mm. This large electrochemical potential necessitates the presence of extrusion mechanisms for Mg2+. In addition, Mg2+-permeable components that would allow a large Mg2+ influx could play an important role in cellular Mg2+ homeostasis. The cGMP-gated channels are permeable to Mg2+ (Nakatani & Yau, 1988; Colamartino et al. 1991; Wells & Tanaka, 1997) and small light-sensitive currents carried by Mg2+ have been observed in the absence of the other permeant ions Na+ and Ca2+ (Nakatani & Yau, 1988). Apparently Mg2+ also enters through the cGMP-gated channels in the presence of Ca2+ but in the absence of Na+ (Nakatani & Yau, 1988). By extrapolating the results of those experiments to physiological ionic conditions, Nakatani & Yau (1988) estimated that Mg2+ carries ≈5 % of the light-sensitive current in the dark, in agreement with modelling based on data from experiments with excised patches (Wells & Tanaka, 1997). The results presented here show that opening or closing the cGMP-gated channels does not affect the free Mg2+ concentration in the photoreceptor outer segment under physiological ionic conditions. Although we have not measured them, the membrane potential values over which the Mg2+ concentration remains unaffected should be within the physiological range, as judged by the changes in the outer segment Ca2+ concentration. Since opening the cGMP-gated channels with 1 mm IBMX results in outer segment Ca2+ concentrations of 150-300 nm, the membrane potential should be between -65 and -30 mV. Therefore, the lack of change in free Mg2+ concentration upon exposure to 1 mm IBMX under physiological ionic conditions is not due to the lack of driving force for Mg2+ entry. The significantly larger influx of Mg2+ upon exposure to 1 mm IBMX in the absence of other permeant ions may be due to the anomalous mole fraction effect that has been described for other channels (Hess & Tsien, 1984).

The lack of change in the free Mg2+ concentration runs counter to some previous observations. Detached rod outer segments dialysed via a patch pipette with a 0 Mg2+ solution containing nucleotide triphosphates develop a dark current and can support phototransduction in the presence of physiological extracellular Mg2+ (Rispoli et al. 1993). As Mg2+ is required for the functioning of the phototransduction enzymes, this observation indicates that Mg2+ can enter the rod outer segment, though the influx may be through a non-specific leak pathway and below the resolution of our measurements. In the same experiments, removal of extracellular Mg2+ resulted in the slow decay of the dark current, indicating loss of the intracellular Mg2+ required for enzymatic function (Rispoli et al. 1993). In these experiments the Mg2+ loss may have occurred through the patch pipette containing the 0 Mg2+ solution. In agreement with the present results shown in Fig. 4 and Fig. 5B, experiments with sealed rod outer segments have shown that exposure to 0 Mg2+ solution does not result in Mg2+ loss, as judged by continued metabolic potency (Schnetkamp, 1981). Another observation that appears inconsistent with the present results is that the total Mg2+ concentration declines by 14 % in rod outer segments after 5 min of illumination (Somlyo & Walz, 1985). Such a decline in the total Mg2+ concentration would have involved an even smaller decline in the free Mg2+ concentration that may again have been below the resolution of our measurements. However, the reduction in total Mg2+ observed by Somlyo & Walz might not necessarily reflect a cessation of influx with illumination, since it could also reflect Mg2+ loss through extrusion associated with metabolic changes (as in Tessman & Romani, 1998).

The resolution of the experiments presented in this study is critical for evaluating the lack of any observed changes in the free Mg2+ concentration. By using Furaptra, which is a low affinity Mg2+ probe, the experiments are unlikely to interfere significantly with Mg2+ buffering and they should provide reliable measurements of the free Mg2+ concentration. The measurements show that light does not result in detectable changes in the free Mg2+ concentration and place an upper limit of ≈0.05 mm, corresponding to less than ≈6 %, for any change; and such a result is not inconsistent with Somlyo and Walz's (1985) measurements. Similarly, opening of the cGMP-gated channels in the presence of Na+ and Ca2+ does not lead to detectable changes in the free Mg2+ concentration. This would imply that the Mg2+ influx observed by Rispoli et al. (1993) is rather small. However, the measuring protocol employed in the present study is not especially sensitive for detecting Mg2+ influx. Measurements for detecting Mg2+ influx would be best carried out with high concentrations of a high affinity probe that ‘out-competes’ the cell's Mg2+ buffers and stores (see, for example, Neher & Augustine, 1992). Sensitivity is particularly relevant in the case of Mg2+, as the resting free Mg2+ concentration is in the millimolar range and is unlikely to change even with typical cation fluxes across the cell membrane. Mg2+ buffering and sequestration by stores would also reduce any change in the free Mg2+ concentration due to influx. The total Mg2+ concentration in rod photoreceptor outer segments has been reported to be ≈11 mm (Somlyo & Walz, 1985). As the free concentration is ≈0.8 mm, this suggests that there are ≈13 bound magnesium ions for every one free, so buffering could be responsible for masking changes in the concentration of free Mg2+, although it is unclear how freely exchangeable the bound magnesium is. On the other hand, sequestration by intracellular stores seems unlikely to play a significant role, as changes in free Mg2+ in the mitochondrial and nuclear areas parallel changes in the outer segment free Mg2+ concentration. Although in our experiments Mg2+ influx was undetectable under physiological conditions, we can estimate an upper limit for its value. In the case of rods, free Mg2+ increases no more than 0.05 mm after opening of the cGMP-gated channels with 1 mm IBMX. Assuming that efflux is small and that the increase in total outer segment magnesium concentration is proportional to the increase in free Mg2+, the total magnesium influx would provide no more than ▵[Mg]= 14 ×0.05 mm= 0.7 mm. The channels were open for a period (▵t) of 15 min, the volume of the outer segment (V) is ≈2 pl, the Faraday constant (F) is 96 500 Cb mol−1, and the Mg2+ current through the cGMP-gated channels should be less than IMg, where

display math(1)

so that IMg≈ 0.3 pA. From the changes in Ca2+ concentration (which reach levels of 150-300 nm), we estimate that 1 mm IBMX opens ≈40 % of the cGMP-gated channels that are normally open in the dark. Therefore, our results are consistent with a light-sensitive rod outer segment Mg2+ current of less than 0.8 pA under physiological conditions. This means that for a rod outer segment light-sensitive current of ≈60 pA, Mg2+ contributes less than ≈1.3 %, a value significantly lower than Nakatani and Yau's (1988) estimate of ≈5 %. An explanation for this discrepancy is that Nakatani and Yau extrapolated to physiological conditions from experiments in the absence of Na+. In the presence of an anomalous mole fraction effect, such extrapolation can lead to overestimations of the current carried by Mg2+ when Na+ is present as well. For a 0.05 × 60 pA = 3 pA light-sensitive Mg2+ current in the absence of Na+ and Ca2+, and using eqn (1) with ▵t= 15 min, we calculate a ▵[Mg]= 6.5 mm, corresponding to an increase in free Mg2+ of ≈0.5 mm. This is comparable to the increase of ≈0.6 mm observed after a 15 min exposure to 1 mm IBMX in the absence of Na+ and Ca2+ (Fig. 3B). Also, the number of cGMP-gated channels that are open in the dark is likely to be similar to the number opened by 1 mm IBMX, even in the absence of Na+ and Ca2+ (actually it should be larger because of the lack of Ca2+ inhibition of the guanylyl cyclase). Therefore, when ionic conditions are taken into account, the change in free Mg2+ concentration that would be expected by Nakatani and Yau's current measurements is similar to the change measured in Fig. 3B, indicating that the results are broadly consistent. Assuming that any influx in the dark is balanced by an efflux that continues after closure of the channels, then if indeed ≈5 % of the light-sensitive current were carried by Mg2+, we would expect a 3 pA light-sensitive Mg2+ current (see above), and a loss of ▵[Mg] of 2.2 mm over an illumination period of 5 min (as in Fig. 7A and B). This would correspond to a free Mg2+ drop of ≈0.16 mm (calculated from 1 free Mg2+ per 14 total), and a F340/F380 ratio decrease of ≈3 %, which would have been detectable in our experiments. Therefore, even with our limited resolution, we conclude that Nakatani & Yau (1988) have overestimated the magnitude of Mg2+ influx under physiological ionic conditions.

The results presented here also show no evidence for any significant extrusion mechanisms, though the Mg2+ concentration declines slowly after moderate loads (Figs 3-5). The slow recovery is indicative of slow extrusion mechanisms, but we have been unable to characterize these pathways because of the deleterious effects (the cells die) of large Mg2+ loads. There may be some differences in the kinetics of recovery between rods and cones, as the Mg2+ concentration appears to recover faster in rods (Fig. 3). Because of cell viability problems, we have not explored these differences in detail. They may reflect differences in extrusion, sequestration, or initial loading, as cones load with Mg2+ more than rods do (Fig. 3). In other cell types, Na+-Mg2+ exchange is involved in Mg2+ extrusion (Gunther & Vormann, 1985), but our experimental data argue against any significant contribution of Na+-, Ca2+-, or H+-dependent Mg2+ extrusion in photoreceptor outer segments. Mg2+-ATPases may also be involved in Mg2+ extrusion, but we have not been able to investigate their involvement in photoreceptors because experimental treatments that stress the cells may increase the free Mg2+ concentration through depletion of ATP and release of ATP-bound Mg2+ (see Tessman & Romani, 1998). As argued above, Mg2+ extrusion mechanisms have to be present to maintain the low intracellular Mg2+ concentration, but they are rather slow. At any rate, the free Mg2+ concentration remains relatively stable, in accordance with what is found in other cell types.

Mg2+ has been known to regulate the activity of the phototransduction enzymes. It stimulates the activity of the guanylyl cyclase (Koutalos et al. 1995a), and also modulates the interactions between the enzymes of the phototransduction cascade (for example through the association of transducin and phosphodiesterase with the disc membranes). The results presented here show that the outer segment free Mg2+ concentration does not change upon light stimulation and remains stable. Therefore, Mg2+ does not play a role in the dynamic modulation of phototransduction. The lack of a dynamic role for Mg2+ explains why exposure to 0 Mg2+ has no effect on the light sensitivity or the kinetics of the photoresponse of the cell (Nakatani & Yau, 1988). Of the two major intracellular divalent cations, Ca2+ and Mg2+, photoreceptors use Mg2+ as a cofactor for basic cellular processes, and maintain a stable Mg2+ concentration. By contrast, they use Ca2+ for modulating signal transduction, and have potent mechanisms to regulate its concentration.

Acknowledgements

The work was supported by National Institutes of Health Grant EY11351 and Human Frontier Science Program Organization Grant RG0204/2000-B (Y.K.).

Ancillary