in each of these examples, the mutant (or knockout) responses deviate from the control responses at roughly 100 msec after the flash, during the rising phase of the flash response. This suggests that, normally, phosphorylation begins to reduce rhodopsin's activity very soon after the R* is formed.

Experiments on mouse rods that do not express arrestin (Arr-/—) have likewise provided important clues regarding the in vivo time course of rhodopsin inactivation. First, Arr—/— responses do not differ greatly from wild-type responses until fairly late in the falling phase of the dim flash response (Xu et al., 1997; see also Fig. 16.6). The partial recovery (approximately halfway back to baseline) of the Arr—/— response suggests that phosphorylation alone is sufficient to reduce rhodopsin's catalytic activity by at least 75% (keeping in mind that R* activity in the absence of all rapid inactivation, such as in the RK—/— rods, generates a response with a peak amplitude roughly twofold larger than normal). On this basis, one might view arrestin as contributing relatively little to inactivation. However, in the absence of arrestin, the final time constant of recovery is nearly 200-fold longer than normal, and roughly 10-fold longer on average than responses of rods that lack rhodopsin phosphorylation. Thus, although the magnitude of the effect of arrestin during the single-photon response is rather small, the implications for the overall function of the rod during prolonged light exposure are astounding.

Inactivation of G*/PDE*. Knockout and transgenic approaches have also proven useful for elucidating the mechanisms of G*/PDE* inactivation. Experiments on mouse rods expressing a mutant form of PDEg that binds poorly to G* (W70A; Slepak et al., 1995) have shown that G*/PDE

interactions are required for normal rates of GTP hydrolysis measured in in vitro assays, as well as for the normal amplification and recovery of the flash response (Tsang et al., 1998). An even greater effect on the rate of GTP hydrolysis has been observed in mouse rods expressing neither RGS9-1 nor G05 L (RGS9—/—). Retinal homogenates of these mice have shown an impaired rate of GTP hydrolysis that could not be accelerated by the addition of exogenous PDEg (Chen et al., 2000), thereby confirming the notion that RGS9/GP5 L acts on the G*/PDE* complex.

Calcium-dependent Feedback Finally, knockout mice have proven to be very effective at isolating the calcium-dependent negative feedback mechanisms. Although such mechanisms have been studied by more conventional physiological and biochemical approaches for decades, the abolition of one such mechanism (e.g., calcium regulation of guanylate cyclase by knocking out GCAPs) in intact cells allows the other calcium-dependent mechanisms to be studied unperturbed. The results and interpretations of these experiments will be described in the section that follows.

Calcium feedback regulation of phototransduction

It has been clear from many investigations conducted since the mid-1980s that the photoreceptor's response to light is accompanied by a decline in cytoplasmic calcium concentration (e.g., as reviewed by Koutalos and Yau, 1996). Here we begin by summarizing those findings.

In darkness, the flow of current into the photoreceptor outer segment is carried primarily by Na+ ions, but a substantial proportion (~15% in rods) is carried by Ca2+ ions.

This steady influx of Ca2+ in darkness is balanced by an equal and opposite efflux, driven by the Na+/Ca2+, K+ exchanger that resides in the plasma membrane along with the cyclic nucleotide gated channels. Different estimates have placed the dark resting concentration of free cytoplasmic Ca2+ in the range of 200 to 700 nM. During illumination, the outer segment ion channels close (see the section entitled "Channel Opening/Closing by cGMP"), reducing the influx of Ca2+ ions. As a result, the exchanger drives down the cytoplasmic Ca2+ concentration to a level that has been estimated to be in the range of 10 to 50 nM in saturating light. In amphibian rods, the decline is not especially fast, with an effective time constant of around 1 s. In cones, where the surface-to-volume ratio of the outer segment is much greater, the time course of decline is substantially faster.

It has been known for several decades that a lowered calcium concentration increases the rod's circulating dark current (Hagins and Yoshikami, 1974) and activates guany-late cyclase (Lolley and Racz, 1982), and that elevated calcium concentration enhances light-activated PDE activity (Torre et al., 1986). Furthermore, minimizing the light-induced decline in intracellular calcium by incorporation of calcium buffers or by "calcium clamp" protocols dramatically increases the flash sensitivity of dark-adapted photore-ceptors and prevents the onset of normal light adaptation (Torre et al., 1986; Fain et al., 1989; Matthews et al., 1990). However, it has taken some years to unravel the relative importance of the individual molecular mechanisms that participate.

We now know that the decline in calcium concentration orchestrates at least three separate changes in the outer segment. These are often referred to as negative feedback mechanisms because any perturbation of the state of the system tends to be opposed. It is now clear that the most important of these feedback mechanisms is the calcium-dependent regulation of guanylate cyclase activity, mediated by GCAPs. In addition, there are also effects of calcium on the disc membrane cascade of reactions leading to PDE activation, probably mediated by the calcium-sensitive protein recoverin, and on the sensitivity of the plasma membrane ion channels to cGMP, mediated in rods by calmodulin. We will discuss each of these three processes in turn.

Calcium Regulation of Guanylate Cyclase Evidence for calcium regulation of guanylate cyclase originated in the 1980s. Biochemical experiments showed that cyclase activity was regulated in a cooperative manner by a decline in calcium concentration (Pepe et al., 1986; Koch and Stryer, 1988), whereas electrophysiological experiments suggested that cyclase activity was elevated during recovery from saturating flashes (Hodgkin and Nunn, 1988). Today we know that the rapid restoration of circulating current after a flash is, to a major extent, attributable to activation of guanylate cyclase by the lowered intracellular free calcium concentration.

Ca2+ regulation of guanylate cyclase activity is mediated by GCAPs, of which there are three retinal isoforms: GCAP1, GCAP2, and GCAP3. GCAPs belong to the calmodulin superfamily of EF-hand proteins, but contain only three functional Ca2+ binding sites (reviewed in Dizhoor and Hurley, 1999). Biochemical studies have shown that at low calcium concentrations, GCAPs activate guanylate cyclases, whereas at high calcium concentrations, the calcium-bound form of GCAPs inhibits guanylate cyclase activity (Dizhoor and Hurley, 1996; Rudnicka-Nawrot et al.,

1998). The calcium concentration that has been reported to activate guanylate cyclases half-maximally (K1/2) has varied widely depending on experimental conditions, with most reports clustering near ~200nM (Dizhoor and Hurley, 1996). The activation by calcium is cooperative, and although an early in vitro experiment gave an estimate for the Hill coefficient of about 4 (Koch and Stryer, 1988), subsequent experiments have given values of about 2 (reviewed in Pugh et al., 1997).

Photoreceptors express not only multiple GCAPs, but also multiple isoforms of guanylate cyclase (GC1 and GC2; reviewed in Pugh et al., 1997). The reason for having multiple guanylate cyclases and GCAPs is not known, but presumably they serve distinct functions. GCAP1 activates primarily GC1, whereas GCAP2 and GCAP3 activate both GC1 and GC2 with similar potencies (Haeseleer et al.,

1999). It has generally been thought that the calcium dependence of cyclase activation by the GCAPs is similar. However, expression of a GCAP2 transgene in mouse rods lacking endogenous GCAPs (only GCAP1 and GCAP2 occur in mouse rods) has been found to restore maximal calcium-dependent cyclase activity but not to restore normal flash response kinetics (Mendez et al., 2001). In contrast, expression of a GCAP1 transgene in GCAPs-/- rods did restore normal kinetics (Howes et al., 2002). This suggests that GCAP1 and GCAP2 may serve different temporal roles in rods.

Comparison of the steady-state responses of GCAPs-/-rods and wild-type rods suggests that, in a wild-type rod, a given relative change in the cGMP concentration produces an 11-fold larger relative change in the guanylate cyclase activity (Burns et al., 2002). Assuming that the Hill coefficient for channel activation is about 3, this suggests that the Hill coefficient for guanylate cyclase activation by calcium is about 4 in normal rods. However, most biochemical measurements of calcium dependence of guany-late cyclase activity have given a Hill coefficient of about 2 (reviewed in Dizhoor and Hurley, 1999). A satisfactory explanation for this large discrepency remains to be elucidated.

The power with which calcium regulates guanylate cyclase activity perhaps helps to achieve the surprising speed with which feedback can occur. Experiments on intact GCAPs-/- rods have shown that feedback to guanylate cyclases normally begins very quickly, within roughly 40 ms of the flash (Burns et al., 2002). The rapid feedback to guanylate cyclases causes the rising phase, as well as the amplitude, of the response to be severely attenuated, such that calcium feedback to guanylate cyclases decreases the dark-adapted flash sensitivity (Mendez et al., 2001). In addition, feedback speeds the restoration of the dark current, causing a more rapid recovery of the dim flash response.

Calcium feedback to guanylate cyclase not only reduces the amplitude of the signal (single-photon response) of rods, but also of the noise. In darkness, rods are noisy; spontaneous fluctuations in dark current arise primarily from spontaneous fluctuations in cGMP concentration stemming from thermal activation of rhodopsin and PDE (Baylor et al., 1980; Rieke and Baylor, 1996). Calcium feedback via GCAPs reduces these spontaneous fluctuations, reducing the standard deviation of the dark noise by roughly a factor of 6.3, while at the same time reducing the amplitude of the response by a factor of 4.5. Thus, calcium feedback in the dark-adapted rod serves at least two distinct purposes: (1) it speeds the recovery of the flash response, improving temporal responsiveness at the cost of sensitivity; and (2) it improves the signal-to-noise ratio of the cell by roughly 40%.

Perhaps not surprisingly, GCAPs-/- rods do not adapt normally to steady lights (Mendez et al., 2001). The incremental flash sensitivity is markedly reduced at the brightest backgrounds, reflecting the importance of the steady-state feedback regulation of guanylate cyclase for range extension (see the section entitled "Calcium-dependent Mechanisms of Adaptation"). Thus, both the dynamic and steady-state changes in intracellular calcium have profound effects on guanylate cyclase activity and significant implications for rod function.

Calcium Regulation of the Disc Membrane Cascade of Reactions

Electrophysiology. Electrophysiological evidence for the calcium dependence of the disc membrane reactions that lead to activation of PDE came from experiments in which cGMP was infused into rods. The duration of the bright flash response was shown to be greatly extended under conditions in which Ca2+ was expected to have been elevated, even though guanylate cyclase activity had been overridden by the supply of exogenous cGMP (Torre et al., 1986). Prolongation of the flash response has also been seen in experiments in which recombinant recoverin was dialyzed into rods (Gray-Keller et al., 1993; Erickson et al., 1998).

Experiments lowering the Ca2+ concentration around the time of flash delivery (Hodgkin et al., 1986; Matthews, 1997) have revealed a shortened response duration and have suggested that the extent of light-dependent PDE activation is determined by the intracellular calcium concentration only near the time that the flash is delivered, presumably while R* remains in existence. Finally, experiments on truncated salamander rods have suggested that lowered intracellular calcium can speed the quenching of rhodopsin's activity by phosphorylation, as the rate of rise of the flash response is sensitive to both calcium and ATP, but not GTPgS (Sagoo and Lagnado, 1997).

Biochemical experiments. Biochemical evidence supporting the calcium-dependent regulation of light-dependent PDE activity was first derived from experiments on frog rods (Kawamura, 1992). These experiments demonstrated that rhodopsin phosphorylation was dependent on free calcium concentration and was mediated by the calcium-binding protein recoverin (S-modulin; Kawamura, 1993). The molecular target of recoverin was identified by cross-linking as RK (Sato and Kawamura, 1997). These observations have been extended by several research groups, leading to the hypothesis that recoverin calcium binds to RK in the dark, thereby inhibiting rhodopsin phos-phorylation. During light exposure, when the calcium concentration deceases, recoverin unbinds its calcium and its affinity for RK is reduced, so that it no longer inhibits rhodopsin phosphorylation. One as yet unresolved caveat to this hypothesis is that some in vitro measurements have claimed that the extent of rhodopsin phosphorylation is unaffected by light adaptation and changes in intra-cellular calcium (Ohguro et al., 1995; Otto-Bruc et al., 1998).

Calcium Regulation of cGMP-gated Ion Channels: Calmodulin Light-evoked changes in intracellular calcium concentration cause a shift in the K/ for activation of the cGMP-gated channels by cGMP (Hsu and Molday, 1993). In rod channels, this shift is thought to be mediated by the binding of calmodulin (Weitz et al., 1998) or a calmodulin-like protein (Gordon et al., 1995) to the b subunit of the rod channel (Chen et al., 1994) at high calcium concentrations. The decline in calcium that accompanies the steady-state response to light causes the modulator to dissociate from the channel, increasing the affinity of the channel for cGMP. The contribution of this mechanism to range extension in rods is thought to be rather modest (2- to 10-fold; Koutalos et al., 1995b; Sagoo and Lagnado, 1996; Nikonov et al., 2000;). A similar but more powerful mechanism also appears to operate for cone channels (Rebrik and Korenbrot, 1998; Rebrik et al., 2000; Muller et al., 2001).

In the next section, we will try to evaluate the relative roles of the molecular mechanisms that we have described above in the context of adaptation.

Mechanisms of photoreceptor light adaptation

Photoreceptors have the daunting task of reporting the time course of illumination across an intensity range of at least 10 orders of magnitude. The rod (scotopic) system operates primarily over the dimmest four orders of magnitude, whereas the cone (scotopic) system covers the brightest seven, with mesopic vision (both rods and cones functioning) covering a couple of orders of magnitude in the midrange. In human vision, the lowest useful intensity corresponds to ~10-2 photoisomerizations s-1, whereas the rods have saturated at an intensity of ~103 photoisomerizations s-1, and yet the cones never saturate under steady illumination (see discussion that follows, and Burkhardt, 1994).

Light adaptation causes two characteristic changes in the incremental light response of photoreceptors: it reduces the cell's sensitivity (measured as the peak amplitude per unit of flash intensity) and it speeds the kinetics of response recovery. The cellular events known to contribute to light adaptation encompass at least five categories: (1) response compression; (2) calcium-dependent mechanisms that act on a relatively short time scale; (3) elevated steady-state activity of the PDE, which acts by reducing cGMP's turnover time; (4) recently discovered large-scale movements of protein out of the outer segment; and (5) pigment bleaching. In the following sections we briefly describe the contributions of these phenomena to the overall light adaptation of photoreceptors; for more extensive coverage we refer the reader to several recent reviews (Pugh et al., 1999; Govardovskii et al., 2000; Fain et al., 2001).

An important clue to possible molecular mechanisms of light adaptation comes from the observation that, over a wide range of background intensities, the gain of the trans-duction process is unaltered by light adaptation (Nikonov et al., 2000). Accordingly, over this range of intensities, any modulation of the cascade of reactions leading up to PDE activation must operate only via effects on the lifetimes of active substances, rather than on the gain of any activation step, a concept that will be discussed in more detail later in this chapter.

Response Compression During steady illumination, the steady level of photoreceptor circulating current decreases, leaving less current available for suppression by additional photon absorptions. Hence, if sensitivity is measured (as it usually is) in terms of absolute, rather than relative, changes in circulating current, then during illumination the sensitivity is bound to decrease because of this effect. Although this effect clearly contributes to measurements of light adapta tion, it is difficult to view the phenomenon as an adaptation to background light; rather, it represents an unavoidable consequence of the way that phototransduction operates. In this context, it is worth noting that, in cones, such a drawback is not apparent in measurements of intracellular voltage in the eyecup preparation, where the steady-state hyperpolarization does not exceed half the maximum available range, even at exceedingly high steady intensities (Normann and Perlman, 1979; Burkhardt, 1994). Thus, for the intact cone system, the compressive effect never exceeds a factor of 2, and the cone is able to undergo voltage excursions of roughly equal magnitude for intensity increments and decrements on a background.

Calcium-dependent Mechanisms of Adaptation Calcium was clearly established as a messenger of light adaptation by the experiments of Nakatani and Yau (1988a) and Matthews et al. (1988). These and subsequent experiments showed that, in the absence of the normal light-induced change in cytoplasmic calcium concentration, the photoreceptor did not display the normal characteristic features of light adaptation. Instead, the photoreceptor simply exhibited response saturation with a much more abrupt decline of sensitivity as a function of increasing background intensity (reviewed in Fain et al., 2001).

These results have sometimes been misinterpreted to indicate that the decline in calcium concentration actually causes the decline in flash sensitivity during light adaptation, but nothing could be further from the truth. Instead, the decline in calcium concentration must be viewed as rescuing the photoreceptor from the far greater desensitization that would occur in the absence of its intervention, so that it thereby provides an extension of the range of operating intensities (Pugh et al., 1999). In considering the role of calcium, a distinction needs to be made between dynamic and steady-state effects (Pugh et al., 1999; Torre et al., 1986). The rapid transient drop in calcium concentration has been proposed to underlie the rapid recovery of the flash response to its baseline level (Torre et al., 1986), whereas the steady-state change is thought to account for light-adaptation behavior (Koutalos and Yau, 1996).

The role of calcium-dependent modulation of the discbased reactions in photoreceptor light adaptation is not fully understood, despite our knowledge of the molecular mechanisms listed in a previous section ("Calcium Regulation of the Disc Membrane Cascade of Reactions"). It is clear that the calcium concentration at the time of flash delivery has an effect on the duration of the flash response (see section just cited). In addition, the so-called step-flash effect (whereby the recovery of a bright test flash is accelerated by the prior application of a background) is abolished when changes in calcium concentration are prevented (Fain et al., 1989). Furthermore, this effect also exists in mammalian rods and is eliminated in rods that do not express recoverin (Dodd, 1998). Hence, the electrophysiological results are consistent with the notion that recoverin-mediated steady-state effects of calcium modulate the disc-based reactions and contribute significantly to overall light-adaptation behavior. Moreover, because of the invariance of the amplification constant of transduction with background intensity (Nikonov et al., 2000; see also the section entitled "Gain Reduction: A Matter of Intensities"), the steady-state effect of calcium must be on the lifetime of intermediate(s), rather than on gain. Finally, a small part of the rescue of sensitivity comes from calcium regulation of the ion channels, through the reduction in Ky for channel opening by cGMP

Contribution of Elevated Steady-state PDE Activity Continuous illumination directly causes a steady-state increase in PDE activity, thus increasing the rate constant of cGMP hydrolysis (Hodgkin and Nunn, 1988; Kawamura and Murakami, 1986); this serves to reduce the effective lifetime of the cGMP molecule. Somewhat paradoxically, though, it was not realized until quite recently that this mechanism makes a major contribution to photoreceptor light-adaptation behavior (Nikonov et al., 2000; Pugh et al., 1999). Indeed, this phenomenon appears to represent the rod's most powerful mechanism for steady-state adjustment of sensitivity, and furthermore, it is calcium independent.

A useful way of thinking about this mechanism is to apply the so-called "bath tub" analogy (Pugh et al., 1999), whereby water flows into a bath from a tap and out via a plug hole. The size of the plug hole represents steady PDE activity, whereas the steady inflow through the tap represents steady cyclase activity; the depth of the water represents the cGMP level. In darkness, with a small plug hole and a slow trickle of water from the tap, any perturbation in the water level dies away only slowly. However, in moderate illumination, with a gigantic plug hole and the tap running rapidly, any perturbation comes to equilibrium much more rapidly. From such an analysis it can be shown that, for a given driving function (opening-up of a given additional plug hole), the response measured by the change in water level is both faster and smaller in the adapted state. For a further discussion of this concept, see Govardovskii et al. (2000) and Pugh and Lamb (2000).

Gain Reduction: A Matter of Intensities A simple mechanism that one could imagine for reducing flash sensitivity during light adaptation would be a reduction in amplification of the cascade, for example, by reducing the amplifying step between rhodopsin and transducin. Although some studies have suggested that there can be a reduction in rhodopsin-transducin gain during light adapta tion (Lagnado and Baylor, 1994; Jones, 1995; Gray-Keller and Detwiler, 1996), other studies have found no evidence for a change in amplification (Torre et al., 1986; Hood and Birch, 1993; Thomas and Lamb, 1999; Nikonov et al., 2000). In attempting to resolve this issue, it is important to restrict consideration to the very earliest times in the light response, before inactivation reactions come into play. Because both the reduction of rhodopsin's catalytic activity by phosphorylation (Chen et al., 1995; Chen et al., 1999; Sagoo and Lagnado, 1997) and calcium activation of guany-late cyclase (Burns et al., 2002) begin during the rising phase of the flash response, the window of relevant times is very short. Bearing in mind this temporal restriction, the most recent analyses have indicated that, for intensities eliciting up to 70% suppression of the circulating current in rods, the amplification of transduction is invariant with intensity (Nikonov et al., 2000, suction pipette; Sokolov et al., 2002, electroretinogram).

By contrast, brighter illumination, sufficient to hold the rods in saturation for a prolonged time, has been found to reduce the amplification (Kennedy et al., 2001; Sokolov et al., 2002). Furthermore, this reduction in gain is associated quantitatively with a large-scale redistribution of trans-ducin within the rod, with a substantial fraction of both the a and bg subunits of transducin translocating from the outer segment to the inner segment (Sokolov et al., 2002). The gain reduction that results from the massive movement of transducin from the outer segment is a novel form of light adaptation. However, it is not yet clear whether the purpose of this mechanism is to rescue rods from saturation at higher light intensities than would otherwise be possible, or whether it might perhaps act as a means for minimizing unnecessary metabolism while the rod's photocurrent is saturated.

Contribution of Pigment Bleaching to Light Adaptation The bleaching of photopigment makes a negligible contribution to light adaptation, except at the very highest intensities experienced by cones. In human rods, complete saturation occurs at around 103 photoisomerizations s-1, which, with a pigment regeneration time constant on the order of 400 s, corresponds to a steady-state bleach level of 4 x 105 out of ~108 rhodopsin molecules, or less than 0.5%. Hence, rods are completely inoperative at light intensities that cause any appreciable level of bleaching.

Cones, on the other hand, are capable of functioning even when almost all of their pigment has been bleached (Burkhardt, 1994), and it has long been known psychophys-ically that the cone system does not show saturation under steady illumination of any intensity (Barlow, 1972). This seems to indicate that cones are much less susceptible than rods to the presence of bleaching products, although an additional contributory factor no doubt stems from the fact that cone pigments exhibit much faster removal of all-trans retinoid and faster return to the native 11-cis form (Shichida et al., 1994). The physiological relevance of the cones' ability to function at very large bleaches is not clear, because large steady-state bleaches are so unpleasant to the observer that they are actively avoided wherever possible. In practice, it seems likely that the steady-state cone pigment bleaching level experienced by human observers seldom exceeds about 90%, contributing a factor of, at most, ~ 10-fold to reduced quantal catch, even though the cones are fully capable of functioning at much higher bleach levels.

Composite Effect of All the Adaptational Mechanisms Koutalos et al. (1995a, 1995b) undertook experiments to separate the steady-state effects of calcium feedback on guanylate cyclase, on the cascade leading to PDE activation, and on ion channel activation. They provided a compelling theoretical model of the contributions to light adaptation of these different calcium mechanisms. They reported that the cyclase-mediated effect was most important at relatively low intensities, whereas the effect on the PDE cascade became more important at higher intensities; the effect on the channels was modest at all intensities.

That description would appear to remain valid, with one modification. In view of the analysis of Nikonov et al. (2000) showing the role of steady PDE activation in desensitization, it seems likely that much of the effect that Koutalos et al. (1995b) attributed to calcium-dependent modulation of the cascade might instead arise from light-dependent activation of the cascade. Indeed, Nikonov et al. (2000) presented calculations of the extent to which light adaptation would be perturbed by "knocking out" different components of the calcium feedback system (their Fig. 14). That analysis confirmed the important role of the GCAP/cyclase effect at relatively low intensities, but also attributed a substantial component of desensitization to the light-induced cascade activation of PDE, over a range of intermediate intensities. Their analysis suggested relatively small effects for both the recoverin-mediated effect on the cascade and the modulation of channel sensitivity.

Summary and conclusions

Photoreceptors have a seemingly simple job description: to accurately relay to the rest of the visual system information regarding the timing and number of photon absorptions, and to do this across more than 10 log units of light intensity. Although nature has evolved two separate neural receptors and pathways to mediate vision across this enormous range of intensities, this has clearly not been enough. Rods and cones possess a multitude of specialized mechanisms: (1) mechanisms that serve both to amplify and to deactivate the response in a timely manner; (2) mechanisms that serve to adapt the cell to light in both the short- and long-term; and finally, (3) mechanisms that serve to maintain these functional abilities, via pigment regeneration and other homeo-static mechanisms that lie beyond the scope of this chapter (e.g., disc shedding and renewal). Our current level of understanding is beginning to reveal our true ignorance regarding the engineering principles with which nature has achieved the generation of signals essential for the visual perceptions of the world around us.

0 0

Post a comment