The Physiology of Dark Adaptation: Progress and Future Directions

Abstract

Exposure of the eye to bright bleaching light produces a large decrease in photoreceptor sensitivity, followed by a slow return during adaptation to darkness. Although much progress has been made understanding the nature of this phenomenon, particularly its biochemistry, less is known about the physiology of dark adaptation. In this review, we summarize the evidence for desensitization produced by photoproducts of bleaching, especially apo-opsin, that is opsin without bound chromophore. We describe the relationship between these studies and diseases such as vitamin A deprivation and congenital stationary night blindness; the effects of analogs of chromophore on photoreceptor sensitivity; and the roles of transducin, rhodopsin kinase, and arrestin. We review many specialized features of dark adaptation in cones, including the role of retinal G protein-coupled receptor (RGR) opsin. For both rod and cone dark adaptation, we summarize some of the principal uncertainties in our understanding. We hope our review will provide a guide to past work as well as an indicator of many possible areas of future research.

1. Introduction

When the two of us were beginning our careers in Boston in the early seventies, little was known about the molecular and cellular mechanism of dark adaptation. Gordon was a student of John Dowling, who did his PhD with George Wald, who in turn studied with Selig Hecht. Carter was in the lab of Tony Gorman, a good friend of Dowling and a prominent biophysicist and photoreceptor physiologist. Both Carter and Gordon knew of Hecht’s famous paper (Hecht et al., 1937), showing the two-branched recovery of rod and cone vision after exposure to bright bleaching light (Fig. 1A). We also knew of the key discovery of the Wald laboratory that the chromophore of the photopigment is 11-cis retinal, which is photoisomerized by light to an all-trans isomer and then regenerated in darkness.

Fig. 1.

(A). Dark adaptation of a human observer to increasing intensities of bleaching light. The logarithm of the psychophysical threshold is plotted versus time in darkness after the bleach. Recovery of sensitivity had two distinct lobes, especially evident for the brightest bleach. The first was recovery of cone vision, which could reached a plateau if the bleach was sufficiently strong. The second slower recovery was of the rods, which could continue for over 30 minutes. Reprinted with permission from Hecht et al (1937). (B). The relationship between rhodopsin content of the rat retina and visual threshold measured with the b-wave, in animals recovering from bright-light exposure (○) and after vitamin-A deprivation (●). Figure reprinted with permission from Dowling (1960).

As physiologists, we were curious about the effect of bleaching illumination on retinal light responses, a field pioneered by Dowling when he was a student in the Wald lab. Dowling was the first to study the relationship between the sensitivity of recovery of the b-wave of the electroretinogram (ERG) during dark adaptation and the concentration of pigment. He made the surprising discovery that, during recovery from a bright bleach in the rat retina, the logarithm of the sensitivity of rod vision seemed to be nearly proportional to the concentration of rhodopsin in the eye. A summary of his results is given in Fig. 1B, which plots the log of the b-wave threshold against pigment concentration. The open circles are for bleaches, and the closed circles are for vitamin A deprivation. Both seemed to follow the same relationship. Similar observations were made a little later by William Rushton and his collaborators for human retina (Rushton, 1961). Although various explanations were offered for why the photoreceptors would be calculating logarithms, none of them seemed very convincing. Several groups in the early seventies were recording from rods with micropipettes, which we were soon to do ourselves. This early work showed that rods are initially insensitive to light after a bright flash but slowly recover response amplitude and sensitivity with time (Grabowski and Pak, 1975; Kleinschmidt and Dowling, 1975). Nothing was then known about the mechanisms that produced these changes.

The cellular biochemistry of pigment regeneration was also beginning to be clarified. Kühne in the 1870’s showed that the retinal pigment epithelium (RPE) could facilitate photopigment regeneration in the photoreceptors, but Dowling made the first quantitative measurements of the distribution of the chromophore after bright light. He showed that most of the bleached retinal transits to the RPE and then comes back slowly to the photoreceptors as sensitivity recovers (Dowling, 1960). This scheme is now known as the visual cycle (Fig. 2) and is still thought to be primarily responsible for the regeneration of visual pigment in rods.

Fig. 2.

The visual cycle. The 11-cis retinal chromophore of rhodopsin is isomerized to all-trans retinal by light. The all-trans retinal falls off the opsin binding site and is converted to all-trans retinol by a retinol dehydrogenase, probably RDH8. The all-trans retinol leaves the rod bound to interphotoreceptor retinoid-binding protein (IRBP). Once delivered to the retinal pigment epithelium (RPE), it is esterified into all-trans-retinyl ester by the enzyme lecithin retinol acyltransferase (LRAT), and then isomerized to 11-cis retinol by the retinoid isomerase (RPE65). The 11-cis retinol is converted to 11-cis retinal by another retinol dehydrogenase, probably primarily RDH5. The 11-cis retinal is transported to the RPE membrane by cellular retinaldehyde-binding protein (CRALBP), leaves the RPE and is transported by IRBP back to the rod, where it recombines with apo-opsin to regenerate the visual pigment. For an excellent, in-depth review of the biochemistry, see Kiser and Palczewski (2016). Figure reprinted with permission from Fain (2019).

There was at that time some question whether the visual cycle was the only mechanism of pigment reformation. We both remember presentations at ARVO meetings of the experiments of Bruce Goldstein (1970) and Don Hood and Peggy Hock (1973), showing that cone vision can recover after exposure to bright bleaching light even in the absence of the RPE. At the time, these results evoked bewilderment and disbelief, but we can now appreciate that they were the first indications that pigment regeneration can be different for rods and cones and can occur within the neural retina (Sato and Kefalov, 2024; Wang and Kefalov, 2011).

During the succeeding 50 years, we have gained a greatly enhanced understanding of what happens to rods and cones during and after exposure to light bleaching a significant fraction of the visual pigment. It has become clear that desensitization and recovery primarily depend upon events within the photoreceptors. After bright light exposure, the photoreceptors are desensitized by what Stiles and Crawford called an “equivalent background light” (Stiles and Crawford, 1932), produced by the photoproducts of bleaching including metarhodopsin II (meta II), metarhodopsin III (meta III) (Kolesnikov et al., 2003), and apo-opsin, that is opsin without chromophore. These photoproducts excite the transduction cascade like a steady background of real light, by a mechanism dependent on the change in.free-Ca²⁺ concentration in the photoreceptor outer segment (Burns and Pugh, 2014; Fain et al., 2001; Yau and Hardie, 2009). Once the pigment is entirely regenerated, the photoreceptor can return to its sensitivity before the bleach (Cornwall et al., 1990; Cornwall et al., 2000; Pepperberg et al., 1976; Yoshikami and Noll, 1982).

The mechanism of photopigment regeneration in the RPE was also clarified (Fig. 2). The discovery of lecithin retinol acyltransferase (LRAT) which catalyzes the esterification of all-trans-retinol into all-trans-retinyl ester, together with the retinoid isomerase (RPE65) which converts all-trans retinyl ester to 11-cis retinol, provided a detailed understanding of the enzymatic regeneration of the chromophore. These two proteins along with binding and transport proteins (CRBP1, CRALBP, and IRBP) and retinol dehydrogenases interconverting retinal with retinol are now thought to provide most of the supply of regenerated chromophore to rods (Kiser and Palczewski, 2016). The classic visual cycle of Fig. 2 can also contribute to cone pigment regeneration, but cones and even rods can in addition rely upon other pathways in the RPE as well as in the Müller glial cells of the retina (Sato and Kefalov, 2024), as we describe in more detail in Section 7.

In this review, we will not attempt to delineate all of the experiments that have led to our present understanding of the initial desensitization and then recovery of sensitivity and photopigment concentration after exposure of the eye to bright light. Instead, our aim is to highlight certain findings and to pinpoint unresolved questions and unexpected observations still in need of explanation. We will look to the past but hope also to point to the future.

2. Dark Adaptation in Rods

The early experiments of Dowling (1960) and Rushton (1961) created a puzzle. The logarithm of rod sensitivity seemed to be related to the concentration of rhodopsin during recovery after a bleach, but no satisfactory explanation could be given for the basis of this relationship despite many attempts (see for example Pepperberg, 1984; Wald, 1961). More measurements were required and from the photoreceptors themselves, but there was little hope of reproducing Dowling’s sensitivity measurements on single rods. Regeneration of rod photopigment requires the RPE, as Kühne first showed. It seemed therefore necessary to record from rods in the intact eye or eye-cup in the presence of the RPE, and for a prolonged period. The recordings would have to be made with sharp electrodes and would be more difficult in the presence of the RPE; and the rod (and retina) would have to remain stable for at least an hour in a rodent (Dowling, 1960). No one could do the experiment.

Early efforts to study the effects of bleaching on single rods did however suggest an alternative approach. If the rods in the isolated retina in the absence of the RPE are subjected to bright bleaching light and then monitored for a prolonged period, the sensitivity of the rod stabilizes and reaches a steady state (Cornwall et al., 1990; Grabowski and Pak, 1975). Since there is no appreciable regeneration of rod pigment in the absence of the RPE, the rods under these conditions have only a fraction of their normal rhodopsin concentration and behave much as if they are light adapted (Jones et al., 1996). We show typical results in Fig. 3 from stably bleach-adapted mouse rods. Responses are less sensitive and smaller in peak amplitude than dark-adapted rods, and the decay time of the response is accelerated (Nymark et al., 2012) much as in light-adapted mouse rods (Woodruff et al., 2008). If the rods are subsequently exposed to 11-cis retinal, as Pepperberg and colleagues first showed (Pepperberg et al., 1976), the bleached rods can recover completely to their dark adapted condition both in their sensitivity and waveform of response (Fig. 4).

Fig. 3.

Responses of isolated mouse rods before and after exposure to bright bleaching light. (A). Dark-adapted rod, flashes from 4 to 1120 photons μm⁻². (B). 5% bleach, flashes from 4 to 1120 photons μm⁻². (C). 10% bleach, flashes from 17 to 1120 photons μm⁻². (D). 20% bleach, flashes from 17 to 1870 photons μm⁻². (E). 50% bleach, flashes from 43 to 10500 photons μm⁻². (F). 90% bleach, flashes from 159 to 10500 photons μm⁻². Flashes were 20 ms. Reprinted with permission from Nymark et al. (2012).

Fig. 4.

Recovery of mouse rods following 50% bleach after addition of exogenous 11-cis retinal. (A). Averaged responses to 453 photons μm⁻² in darkness (black), at steady state after 50% bleach (green), and after addition of vesicles containing 11-cis retinal at 60 min (blue) and at 150 min (red). Flashes were 20 ms. (B). Same responses as in A but normalized cell by cell to the peak amplitude of the response. (C). Mean response–intensity curves for rods of (A) and (B) at each condition. Reprinted with permission from Nymark et al. (2012).

There is every indication that stably bleached rods are indeed excited by a steady equivalent background light, just as Stiles and Crawford (1932) first postulated. The question then becomes, what produces this equivalent background?

2.1. Opsin stimulates the transduction cascade

An initial approach to the identification of the equivalent background was based on experiments of Hodgkin and Nunn (1988). They showed that rapid solution jumps could be used to estimate the activity of the phosphodiesterase (PDE) and guanylyl cyclase in a salamander rod. The PDE is the effector enzyme of phototransduction, which hydrolyzes cyclic guanosine monophosphate (cGMP) and gates channel closing in the outer segment. Guanylyl cyclase is the enzyme that synthesizes cGMP. Since these two enzymes have antagonistic effects synthesizing and degrading cGMP, their rates can be estimated by inhibiting one and observing the change in outer-segment current produced by the activity of the other. If for example the rod is rapidly perfused with 3-isobutyl-1-methylxanthine (IBMX) to inhibit the PDE, the rod photocurrent rapidly increases as the guanylyl cyclase synthesizes cGMP and opens cGMP-gated channels. From the initial rate of increase of current, it is possible to estimate the rate of the cyclase. Hodgkin and Nunn (1988) showed that the rates of both enzymes are increased by background illumination (Hodgkin and Nunn, 1988).

Cornwall and Fain (1994) adapted this method for stably bleached rods in salamander and showed that bleached pigment produces an acceleration of both PDE and cyclase much like steady light. The PDE and cyclase activities in an isolated, bleached rod continue to be accelerated indefinitely and are only reversed when the pigment is regenerated by extrinsically applied chromophore. Although the identity of the bleaching intermediate producing the steady excitation could not be definitively determined, Cornwall and Fain postulated that excitation was produced by apo-opsin, that is opsin without bound chromophore, since all of the other bleaching intermediates would have decayed by the time the measurements were made. They were careful to say that “the opsin may or may not be phosphorylated, capped (by arrestin), or otherwise modified biochemically.” As we shall see later (Section 5), subsequent experiments have indicated that the apo-opsin is likely to be still phosphorylated and capped by arrestin in the stably bleached photoreceptor.

By comparing the effects of bleaches and background light on PDE and cyclase activity, Cornwall and Fain calculated that a single opsin molecule would activate transduction in salamander with an efficacy of about 5 × 10⁻⁷ Rh* s⁻¹ per opsin molecule. That is to say, about 2 × 10⁶ opsin molecules need to be present (or about 0.1% of the total rhodopsin in a salamander rod) to produce a background light equivalent to 1 Rh* s⁻¹. Although this efficacy is small, so much pigment is bleached in bright light that the activity of the opsin is sufficient to mimic a steady background light of appreciable intensity.

The exact mechanism of opsin excitation is in dispute. Cornwall and Fain showed that activation of the cascade is proportional to the opsin concentration and suggested that excitation could be produced by single opsin molecules rarely entering a transient configuration capable of binding a single transducin and stimulating the cascade. The recent experiments of Yue and colleagues (Yue et al., 2019) seem largely to support this hypothesis. They showed that bleached pigment molecules produce unitary responses of the same amplitude as a weakly transducing form of rhodopsin called REY, with an altered G protein-interaction motif. Since REY rhodopsin would be expected to produce events generated by only single transducin molecules, Yue and colleagues concluded that apo-opsin molecules would also be having a similar effect. Other results contradict this conclusion and argue that the gain of a single apo-opsin molecule may be greater and similar to that of Rh* (Sato et al., 2019). These experiments were however done only with weak exposures bleaching less than 2% of the visual pigment (see also Leibrock et al., 1994), and it is unclear how they would be extended to brighter light exposures.

2.2. Opsin activation in mouse rods

The measurement of opsin efficacy in salamander poses an interesting conundrum. Mouse rod outer segments have a volume about 70 times smaller than salamander outer segments and have about 70 times fewer rhodopsin molecules. A 50% bleach of rhodopsin in a mouse rod produces only about 3 × 10⁷ opsin molecules. If mouse opsin had an efficacy the same as salamander opsin, a 50% bleach would produce an equivalent background light of only about 10 Rh* s⁻¹, which would have almost no effect on the sensitivity of the rod (see for example Makino et al., 2004; Mendez et al., 2001; Morshedian et al., 2018). If mouse rods had a rhodopsin and transduction cascade like the one in salamander, opsin activation would be nearly irrelevant, and recovery from bright background light would depend only on the recovery of meta II and meta III. It would be much faster. As the experiments of Figs. 3 and 4 show, however, mouse rods are stably desensitized after exposure to bleaching light and recover only when perfused with 11-cis retinal. Moreover, the extent of the desensitization is quite comparable to that in salamander rods for equivalent percentage bleaches (Cornwall et al., 1990). The efficacy of mouse opsin in stimulating transduction in a mouse rod must therefore be much greater than for salamander.

To measure the efficacy in mouse, two approaches were used. The first exploited RPE65^−/− rods, that is rods lacking the retinoid isomerase. These rods cannot produce 11-cis retinal (see Fig. 2), are markedly desensitized, and will eventually degenerate (Redmond et al., 1998; Seeliger et al., 2001). If however these mice are kept for several months in darkness, they can incorporate 9-cis retinal apparently produced in the liver to form functional pigment (called isorhodopsin) at a concentration up to a third of that of normal rhodopsin (Fan et al., 2003). The rods, even so, are less sensitive than normal rods and appear to be light-adapted, presumably from the fraction of the pigment remaining in the rods as apo-opsin. From biochemical measurements in one eye and physiology in the other, the extent of desensitization could be related to pigment concentration and compared to desensitization produced by steady real light (Fan et al., 2005).

In a second series of experiments, bleaches were given to WT rods whose strength was determined by measuring the decrease in rhodopsin absorption directly with microspectrophotometry (Nymark et al., 2012). Although in mouse it is not then possible to measure pigment in single rods, Nymark and collaborators made measurements from groups of 6 to 8 photoreceptors, which provided enough signal to detect pigment absorption reliably. Both these experiments and those with RPE65^−/− mice gave the same value of equivalent light (2.5 × 10⁻⁵ Rh* s⁻¹ per opsin molecule). This value is about 100 times greater than for salamander, approximately the same as the change in the total number of rhodopsin molecules in the two rods. As a consequence, nearly the same percentage bleach produces an equivalent desensitization in the rods of the two species (Fan et al., 2005).

It is remarkable that the equivalent background produced by bleaching in mouse is so much greater per opsin molecule than in salamander. Although some of this difference can be attributed to the higher body temperature of the mammal, most of it must be the result of intrinsic differences in the structure of the photopigment and/or differences in transducin and the rest of the visual cascade. One might have thought that mammals would have exploited the quieter salamander equivalent light to speed recovery from bright bleaches, but that is not so. It seems possible that the prolonged desensitization produced by strong illumination is somehow adaptive, perhaps to keep the rod outer-segment channels closed to reduce energy consumption (Okawa et al., 2008). It may be interesting to examine equivalent light after bleaching in other species with outer-segment dimensions intermediate between those of salamander and mouse.

2.3. Opsin activation and desensitization

If desensitization in a stably bleached rod is caused by activation of the cascade by opsin, and if this desensitization acts like an equivalent background light, it should be possible to calculate the relationship between opsin concentration and the change in sensitivity. Since activation of transducin by opsin is a rare event (Yue et al., 2019), it is reasonable to suppose that activation is linear with opsin concentration. Direct measurement supports this assumption (Cornwall and Fain, 1994; Yue et al., 2019). We can then postulate that opsin stimulation produces an equivalent background that desensitizes the rod like real background light, according to the Weber-Fechner relation

$$\frac{S_F}{S_F^D}=\frac{I_0}{I_0+I_B}$$ (1)

where S_F is the sensitivity in the presence of a background light, $S_F^D$ is the sensitivity without a background, I_B is the intensity of the background, and I₀ is a constant sometimes called the dark light. We assume that the value of the equivalent background light for a stably bleached rod is proportional to the concentration of opsin, given by the fraction F of pigment bleached. To this change in sensitivity, we must add that produced by the simple decrease in rhodopsin concentration (the decrease in quantum catch), which is proportional to the amount of pigment still remaining (1 – F). The result (Jones et al., 1996) is

$$\frac{S_F}{S_F^D}=\frac{1-F}{1+kF}$$ (2)

where k is a constant. This relation is plotted for single mouse rods in Fig. 5A, with a value for the constant k of 35 (Nymark et al., 2012).

Fig. 5.

Sensitivity as a function of per cent pigment bleached. (A). Mouse rods. Same cells as in Fig. 3. Means with SEM are plotted as a function of percent bleached for rods that were dark-adapted or at steady state after bleaches of 5%, 10%, 20%, 50%, and 90%. Continuous curve is Eqn. (2) with best-fitting value of k of 35. Dashed curve is change in sensitivity predicted by loss of quantum catch. Reprinted with permission from Nymark et al. (2012). (B). Rat rods during dark adaptation. Data taken from open circles in Fig. 1B, plotted on same axis and with same curve as in part (A). See text.

The satisfactory fit of sensitivity and fraction bleached to Eqn. (2) for stably bleached rods suggests that the apparent log-linear relation of Dowling and Rushton may have a simple explanation. In Fig. 5B, we have taken the open circles from Dowling’s paper in Fig. 1B and placed them on the same plot with the same curve from mouse rods in Fig. 5A, with no further adjustment. The fit is fairly good. The points for the greatest fraction bleach lie above the curve, perhaps because these measurements were made early during dark adaptation in Dowling’s experiments, when the contributions of other intermediates of bleaching such as meta II and meta III are likely to have been greatest (Frank and Dowling, 1968; Kennedy et al., 2001; Kolesnikov et al., 2003).

The measurements in parts A and B of Fig. 5 were done on different animals with different methods, and they were done on stably bleached rods in Fig. 5A but on rods during the course of recovery from a bleach in Fig. 5B. Because there is evidence that the value of the constant k can vary with the method of measurement even in the same species (Pahlberg et al., 2017), the fit in Fig. 5B should not be given too much weight. It is nevertheless consistent with the conclusion that the log-linear relationship of Dowling and Rushton is not strictly log-linear but reflects the effect of an equivalent background produced by activation of the cascade together with the simple loss of rhodopsin concentration, as Jones and colleagues first postulated (Jones et al., 1996).

Activation of the transduction cascade by opsin does not, however, fully explain what happens to rods during desensitization and recovery of sensitivity after bleaching exposure. In both rods and cones, absorption of light by the photopigment produces an intermediate state called metarhodopsin II (meta II or R*), which activates the G-protein transducin and initiates a cascade of reactions that ultimately results in changes in membrane voltage communicated to the rest of the nervous system. Meta II can form meta III, which ultimately decays to apo-opsin when the chromophore leaves its binding pocket. We will not describe these reactions in detail because they have been extensively reviewed elsewhere (Fain et al., 1996; Hofmann and Lamb, 2023; Lamb and Pugh, 1992). Both meta II and meta III can persist in photoreceptors after bright light exposure (Frank and Dowling, 1968; Kolesnikov et al., 2003; Sommer and Farrens, 2006; Zimmermann et al., 2004), presumably in a form that is multiply phosphorylated. Meta II is likely to act much like opsin by directly triggering the cascade with low gain rather than by reversion back to Rh* (Leibrock et al., 1994). Meta III, though mostly inactive itself (but see Kolesnikov et al., 2003), can revert to meta II and in this way continue to stimulate the cascade (Heck et al., 2003; Kolesnikov et al., 2003; Zimmermann et al., 2004). Stimulation by meta II and (by reversion) meta III is likely to be especially important for weak bleaches and at early times after strong bleaches, and alteration of the rate especially of meta III can accelerate or delay recovery from bright bleaching light (Frederiksen et al., 2016). The relative roles of these intermediates and of opsin has, however, still not been clarified in detail. It is also possible that all-trans retinal can form a noncovalent complex with opsin that can activate transducin (Jager et al., 1996). Nothing is known about the persistence of all-trans retinal or its efficacy in stimulating the cascade in vivo. We return to these interesting questions when we summarize rod dark adaptation in Section 6.

3. Experiments with chromophore analogs

One approach to understanding pigment activation and recovery during dark adaptation has been to create analogs of 11-cis retinal in which focal changes have been engineered in the structure of the chromophore. Once it became possible to regenerate pigment from exogenous chromophore (see Fig. 4 and Pepperberg et al., 1976), the chromophore structure could be modified to test changes to its shape and chemical composition. Such modifications can include truncation of the polyene chain to form analogs too short to form the Schiff base linkage (Crouch et al., 1982; Jin et al., 1993; Matsumoto and Yoshizawa, 1975), addition of a hydroxyl group to the ionone ring (Corson et al., 1990b), deletion of methyl groups from the polyene chain attached to the ring (Corson et al., 2000), and formation of a hydrocarbon ring around the 11–12 bond to inhibit photoisomerization (Corson et al., 1990a).

3.1. 9-cis C17 aldehyde and beta-ionone

As we explained in section 2.3 and Fig. 5, desensitization in stably bleached rods is composed of two separate components. The first is due to loss of quantum catch from the decrease in the concentration of visual pigment, and the second results from decreased gain produced by an equivalent background light generated by bleached pigment (i.e. apo-opsin). It is possible to distinguish these two components with truncated analogs, which compete for opsin binding with 11-cis retinal but are not themselves able to form a covalent bond with lysine or initiate transduction. One such analog is 9-cis C17 aldehyde, in which the polyene chain is truncated (see Fig. 6B and Jin et al., 1993). When this analog was added to a bleached cone, sensitivity increased (Fig. 7A) and response decay was slowed (Fig. 7B), much as occurs after pigment regeneration (see Fig. 4). Recovery of sensitivity was however incomplete because the 9-cis C17 aldehyde was unable to form a covalent bond with opsin and initiate transduction. Thus the analog removed the equivalent light but had no effect on pigment concentration or quantum catch.

Fig. 6.

Structures with carbons numbered of (A) 11-cis retinal, (B) 9-cis Cl7 aldehyde, and two ring-locked analogs: (C) 11-cis-locked retinal and (D) 11-cis-locked-13-cis retinal.

Fig. 7.

Effects of all-trans C17 aldehyde and 9-cis Cl7 aldehyde on bleached cones. (A). Time course of sensitivity changes before and following 90% pigment bleach and subsequent addition (from left to right) of 1 μM all-trans Cl7 aldehyde, 1 μM 9-cis Cl7 aldehyde, and 3 μM 11-cis retinal. Retinoids in lipid vesicles were added to the superfusate as indicated by arrows. (B) Comparison of time course of flash responses before bleach (l), after bleach (2), and after treatment with 9-cis Cl7 aldehyde (3) at times indicated in (A). The intensity of the flash in (2) was more than 100 times brighter than in (1). (C) Comparison of the relative spectral sensitivity in darkness before bleaching (open circles) and following bleaching and treatment with 9-cis Cl7 aldehyde (closed triangles). The curves are absorbance spectra of salamander cone outer segments made with a microspectrophotometer. The large open triangles in the upper curve were derived by normalizing the data in the lower curve to the peak sensitivity at 600 nm. Reprinted with permission from Jin et al. (1993).

Similar results for cones were found for beta-ionone (Crouch et al., 1982; Matsumoto and Yoshizawa, 1975), in which the polyene chain is truncated so as to retain only C1 – C9. When however rods were treated with beta-ionone, they underwent a further decrease in sensitivity, and flash responses remained accelerated. When rapid solution jumps were used to estimate rates of PDE and cyclase as in section 2.1, beta-ionone was found to activate these enzymes in bleached rods and inactivate them in bleached cones (Kefalov et al., 1999). Similar results were also found with 11-cis 13 demethyl retinal, which lacks the methyl group at position 13 (Corson et al., 2000).

A related phenomenon can be detected in normal rods and cones, when the photoreceptors are rapidly jumped into solutions containing 11-cis retinal. Rapid exposure to chromophore causes rods to be further desensitized, followed soon thereafter by recovery of sensitivity to dark-adapted levels. In contrast, isolated bleached cones treated in the same way experience only a rapid monotonic increase in sensitivity (Kefalov et al., 2001). This difference may be related to the large difference in the rates of spontaneous activation of rod and cone pigments (Fu et al., 2008; Kefalov et al., 2005). It is curious that no one has ever attempted to discover what changes in the structure of these pigments are responsible for these differences.

3.2. 11-cis ring-locked retinal

For the retinoid analogs of Figs. 6C and 6D, the 11–12 bond of the chromophore is bound in a ring structure, which impedes photoisomerization. These analogs can bind to the opsin chromophore binding site but cannot be isomerized and cannot therefore initiate visual transduction. If these analogues are added to salamander rods after bleaching, they produce a recovery of sensitivity and a slowing of the waveform of the light response by relieving the equivalent background; but as for 9-cis C17 aldehyde in Fig. 7, recovery is only partial because the analogs cannot compensate for the loss in quantum catch (Corson et al., 1990a).

3.3. Analogs: the summary

The most important result to emerge from treatment of bleached rods with retinal analogs, at least for our purposes, is that substances such as 9-cis C17 aldehyde, beta-ionone, and ring-locked compounds can in many cases greatly reduce or eliminate the equivalent background light produced by bleached pigment. They cannot however remove the decrease in sensitivity caused by the loss of quantum catch because they do not form a pigment capable of phototransduction. These experiments provide further evidence that the equivalent background in stably bleached rods and cones is produced by activity of apo-opsin—that is opsin without chromophore, and not by some other intermediate of bleaching (Leibrock and Lamb, 1997). Unless the chromophore binding pocket were empty, the analogs would not be able to bind to opsin to relieve the desensitization.

4. Transduction activation and disease

One further way of probing the importance of the structure of the photopigment and activation of the cascade by bleaching light has been to study disease, in particular vitamin A deprivation and some naturally occurring mutations of rhodopsin which cause congenital stationary night blindness. These conditions—like bleaches—produce an equivalent background light and can help us understand the nature of this desensitization.

4.1. Vitamin-A deprivation

When animals (or humans) are deprived of vitamin A, they become night-blind from a reduction in rod sensitivity (Dewett et al., 2021). After prolonged deprivation, the rods begin to degenerate. These effects are all the direct result of opsin activation of the cascade (Lem and Fain, 2004).

Our first understanding of the etiology of deprivation came again from experiments John Dowling did as a part of his Ph.D. thesis with George Wald. He showed that a decrease in chromophore in the rods produces a decrease in sensitivity with the same relation as for bleaches—these are the filled circles in Fig. 1B (Dowling, 1960). Deprivation produces some structural disorganization of outer-segment membranes, but rod morphology remains otherwise intact with normal concentrations of opsin and other proteins. This outcome explains why humans deprived of vitamin A for many weeks can regain their vision in a matter of days when given supplemental vitamin A.

The evidence from Dowling’s work and from many more recent investigations demonstrates that the decrease in rod sensitivity in vitamin A deprivation is produced by the same mechanism producing desensitization in a stably bleached rod, that is by stimulation of the transduction cascade by the opsin remaining in the rods and unbound to chromophore (Dewett et al., 2021). Moreover, degeneration of the rods is also produced by opsin activation. The best evidence for this conclusion comes from experiments on Rpe65^−/− mice lacking the RPE65 retinoid isomerase (see Fig. 2 and Redmond et al., 1998). Because these animals cannot isomerize all-trans to 11-cis retinal, they become progressively deprived of chromophore. Rods are desensitized and behave as if in the presence of a constant background illumination (Fan et al., 2005; Woodruff et al., 2003). Moreover, rods slowly degenerate over a time course of several months. If, however, the Rpe65^−/− rods are also Gnat1^−/− lacking the G-protein transducin so that the opsin in their outer segments cannot activate the cascade, the rods do not degenerate (Woodruff et al., 2003). It is therefore activation of the cascade by opsin and not the presence of opsin per se that is responsible for rod cell death.

The effect of continuous activation of the cascade by opsin in an Rpe65^−/− mouse is similar to the effect of constant stimulation of the cascade by light. For reasons we still do not fully understand, constant illumination even by relatively weak room lighting can cause rodent rods to degenerate though only in albinos (Noell et al., 1966). Pigmented animals are spared (LaVail, 1980). Even in pigmented animals, degeneration proceeds if rods lack either rhodopsin kinase (Chen et al., 1999a) or arrestin (Chen et al., 1999b) and, as a result, have prolonged light responses. If however animals lacking rhodopsin kinase or arrestin are also Gnat1^−/− and cannot produce light responses, the mice are no longer sensitive to continuous illumination and do not degenerate, much as in RPE65^−/−/Gnat1^−/− mice (Hao et al., 2002). What these experiments show is that both equivalent light produced by opsin and continuous real light can kill the photoreceptors and can do so by continuous activation of the transduction cascade (Fain, 2006; Fain and Lisman, 1993). This mechanism contributes at least in part to several forms of retinal degeneration, including some kinds of Leber’s amaurosis and mutations to guanylyl cyclase and cyclic-nucleotide-gated (cGMP-gated) channels. When the channels are maintained mostly or even entirely closed for a prolonged period, some change occurs in the physiology of the rod that leads apoptosis. The cGMP-gated channels are permeable to Ca²⁺ and open in darkness (Fain, 2019). Light closes the channels and produces a decrease in the Ca²⁺ concentration in the outer segment (Gray-Keller and Detwiler, 1994; Woodruff et al., 2002). It is possible that a prolonged decrease in Ca²⁺ produces cell death (see Fain, 2006), but as yet there is no direct evidence that this is so.

The death of the rods from prolonged closing of the channels may also explain why transducin translocates from the outer to the inner segment in bright light (Lobanova et al., 2007). Translocation of transducin reduces the gain of transduction, and by reducing transduction gain it prevents the rods from completely saturating to allow at least some of the cyclic-nucleotide-gated channels to remain open even in bright light (Frederiksen et al., 2021). An explanation of why rods die during constant light or continuous activation of the cascade by equivalent light remains one of the important unsolved problems of retinal cell biology.

4.2. Congenital Stationary Night blindness

Mutations in many photoreceptor genes can produce a condition called stationary night blindness, in which the rods are stably desensitized and either do not degenerate or degenerate very slowly (Zeitz et al., 2015). The most interesting of these conditions for our purposes are mutations of rhodopsin, which appear to act by a common mechanism: they disrupt the salt-bridge between Lys296 and the counterion Glu113, which ordinarily helps to constrain dark-adapted rhodopsin in an inactive conformation (Rao et al., 1994; Rao and Oprian, 1996). The most thoroughly studied of these mutations is G90D, which was first described by Paul Sieving and colleagues (1995). Rao and coworkers (1994) showed that G90D rhodopsin is constitutively active: that is, the G90D pigment in darkness can stimulate the transduction cascade at a much more elevated rate than WT rhodopsin. It therefore acts as an equivalent background light, desensitizing the rods and producing night blindness in patients.

Rao et al. (1994) speculated that the desensitizing signal is produced by a small number of pigment molecules lacking chromophore, that is by G90D opsin. These few G90D opsins were thought to be sufficiently active to produce an equivalent background, desensitizing the visual system. The Oprian group tested this hypothesis by making transgenic Xenopus with G90D rhodopsin and with two other rhodopsin mutations which also cause stationary night blindness in humans (Jin et al., 2003). They showed that rods from these animals were desensitized, but that the desensitization could be relieved by adding extrinsic 11-cis retinal. These observations seemed therefore to support their theory that the desensitization was produced by opsin rather than by G90D rhodopsin.

This explanation was then disproved by Dizhoor and colleagues, at least for mammals (2008). They recorded from single G90D mouse rods and showed that the photoreceptors were effectively light adapted, just as Sieving et al (2001) had earlier claimed. Dark-adapted G90D rods had decreased sensitivity and an accelerated time course of response decay, much as in background light. In contrast to the study of Jin et al on Xenopus (2003), however, the addition of excess 11-cis retinal to mouse G90D rods was without effect, even though excess retinal under the same conditions could relieve desensitization after pigment had been bleached. In an important further experiment, Dizhoor et al showed that 11-cis retinal could also relieve desensitization when the rods were deprived of retinal by deletion of the gene for the RPE65 retinoid isomerase.

Dizhoor and colleagues concluded that desensitization in a G90D rod was not produced by G90D opsin but rather by noisy G90D rhodopsin, that is by the holoprotein, which could trigger the transduction cascade. By measuring the variance of the photocurrent, they showed that the events produced by G90D rhodopsin were much smaller than those produced by actual light and speculated that only one or a few transducin molecules were stimulated by activated G90D rhodopsin, considerably fewer than the number stimulated by WT rhodopsin in actual light (Krispel et al., 2006; Reingruber et al., 2013; Yue et al., 2019). They were however unable to characterize these events in greater detail, because the concentration of G90D rhodopsin was so high in their heterozygous G90D mice that individual noise events could not be visualized, and because the activity of so much G90D protein caused the rod to be light-adapted and decreased the amplitude of the noise events.

Their experiments were extended by the observations of Chai and colleagues (2024), who confirmed that the desensitization of G90D mouse rods could not be relieved by extrinsic addition of 11-cis retinal. Chai and colleagues (2024) were able to resolve single events in G90D rods with three, carefully chosen genetic alterations. In the first, they expressed the G90D rhodopsin at a very low rate of only about 0.1% of normal rhodopsin. In this way, they reduced the number of noise events and prevented the G90D protein from light-adapting the rod. They then put this G90D rhodopsin on a background of rhodopsin with the REY mutation (Yue et al., 2019), which greatly reduced the affinity of activated rhodopsin for transducin and effectively silenced the rest of the visual pigment. As a result, events produced by the G90D protein could be more easily resolved. Finally, these mutations were further placed on a background of Gcaps^−/−, which removed the guanylyl-cyclase activating proteins from the rods and increased the amplitude of the light response (Mendez et al., 2001).

As a consequence of these alterations, Chai et al. (2024) were able to record noise events produced by the G90D pigment and discovered that they consisted of the transient activation of only single transducin molecules. A similar conclusion was reached by this same group for WT apo-opsin molecules (Yue et al., 2019) produced after bleaching of the photopigment. Though both WT opsin and G90D rhodopsin are noisy and can activate single transducin molecules, G90D rhodopsin does so at a much higher rate. In patients heterozygous for the G90D gene, half of the rhodopsin will be mutant rhodopsin, of the order of 10⁷ molecules. Even though each G90D rhodopsin stimulates the cascade only occasionally, the large number of these molecules in the rod can cumulatively activate transduction at a rate high enough to light-adapt the visual system and render the patients night-blind. Future research may help us understand in greater detail how this mutation changes the structure of the rhodopsin molecule to facilitate this aberrant activity.

5. Transducin, rhodopsin kinase, and arrestin

Recovery of the rod from bleaching exposure requires more than just the regeneration of the photopigment and the extinction of the equivalent background light produced by apo-opsin and other intermediates of bleaching. Bright light exposure produces a large translocation of transducin from the rod outer segment to the rod inner segment (Lobanova et al., 2007; Sokolov et al., 2002), which reduces the gain of the initial step in transduction. This reduction in gain allows the photoreceptor to recover from saturation and continue to produce small responses even in very bright illumination (Frederiksen et al., 2021; Tikidji-Hamburyan et al., 2017). In this way, the rods can avoid the complete closure of the outer segment channels and cell death.

For the rod to recover fully from bright light exposure, this transducin must migrate back into the outer segment, which occurs slowly in rodents over a time course of many minutes (Belcastro et al., 2012; Elias et al., 2004; Sokolov et al., 2002). The preponderance of evidence indicates that both the movement from outer to inner segment and the reverse migration back into the outer segment occur by simple diffusion (Frederick et al., 2020; Slepak and Hurley, 2008). Translocation of transducin continues to be normal in mouse retina depleted of ATP (Nair et al., 2005), indicating that molecular motors are unlikely to be required. Both the alpha and beta-gamma subunits of transducin move to the inner segment, where they appear to recombine before migrating back to the outer segment. The return of transducin is delayed in animals depleted of the protein Uncoordinated 119 (UNC119), which is known to bind to the transducin alpha subunit (Zhang et al., 2011); and delayed in mice with mutations in the protein phosducin (Belcastro et al., 2012), which is known to bind to the transducin beta-gamma subunits (Gaudet et al., 1999). The precise roles of UNC119 and phosducin in transducin return are however still unclear.

Light also causes inactivation of rhodopsin by phosphorylation by the enzyme rhodopsin kinase, also known as GRK1; and by the subsequent binding of arrestin to the phosphorylated pigment (see Fain et al., 2010). Arrestin sterically hinders the further binding of transducin to rhodopsin and quenches the transduction cascade. Arrestin binding also speeds the decay of meta III (Frederiksen et al., 2016). Recovery of the photoreceptor back to its dark-adapted state requires arrestin unbinding and dephosphorylation, in addition to regeneration of the 11-cis retinal chromophore. The enzyme responsible for dephosphorylation is likely to be the phosphatase PP2A (Kolesnikov et al., 2017).

Our understanding of this process was assisted by a key observation made by Berry and coworkers (2016), who showed that phosphorylated rhodopsin is capable of initiating transduction. Rods containing mostly phosphorylated rhodopsin have a sensitivity within a factor of between 1.5 and 2 of that of rods containing mostly unphosphorylated rhodopsin. If arrestin were bound to phosphorylated rhodopsin, the photopigment would not be able to bind transducin. We can, we think, safely conclude that by the time the chromophore has been regenerated and the 11-cis retinal rebound to the pigment, arrestin can no longer be bound even if the rhodopsin is still phosphorylated.

It is sometimes thought that in darkness after a bleach, arrestin comes slowly unbound from phosphorylated opsin to leave phosphorylated opsin free of arrestin (Hofmann and Lamb, 2023; Hofmann et al., 1992). We think it more likely that arrestin is bound to all the phosphorylated intermediates of bleaching including apo-opsin, in order to reduce their activity (Sommer et al., 2005); and that it only becomes unbound when the 11-cis chromophore reattaches to opsin and regenerates the pigment (Zhuang et al., 2013). The rebinding of 11-cis retinal to phosphorylated opsin may produce a change in the structure of the protein, such that arrestin binding is no longer favorable. Phosphorylated rhodopsin with arrestin no longer bound would then still be able to bind transducin and activate transduction with only a small diminution of sensitivity. The phosphate groups are eventually removed with a slow time course (in mouse) of many minutes or even hours (Berry et al., 2016; Kennedy et al., 2001).

6. Rod dark adaptation: a summary and some important unanswered questions

When the eye is exposed to bright bleaching light, the rods are at first nearly saturated with all of the cGMP-gated channels closed, and they cannot respond to further illumination. Bleached pigment produces an equivalent background light, which decreases rod sensitivity by a process similar to adaptation in real background light. The pigment intermediates producing the equivalent background are likely to be primarily meta II/meta III and apo-opsin, with the relative importance of each form depending upon the brightness of the bleaching light and the amount of bleached pigment. For small bleaches, meta II and meta III probably predominate, both probably multiply phosphorylated and capped by arrestin. For large bleaches, a significant fraction of the equivalent background is produced by apo-opsin, probably again multiply phosphorylated and capped by arrestin.

With time, this equivalent background subsides as the concentrations of meta II and meta III decrease, and as apo-opsin recombines with chromophore to produce dark-adapted rhodopsin. The rate at which the equivalent background decreases and sensitivity recovers depends again on the extent of the bleach and can be only a few minutes if little pigment has been bleached. Rods appear to have a small store of free 11-cis retinal, since isolated rods without the RPE can fully recover from bleaches of only a few percent (see for example Azuma et al., 1977; Krispel et al., 2003; Lamb, 1980; Leibrock et al., 1994). Regeneration of rhodopsin is rapid, with the time course of recovery depending mostly on the rates of decay of meta II and meta III (Frederiksen et al., 2016). For bright bleaches, on the other hand, recovery can require 30 minutes to several hours in different species (Fig. 1A). The time course is rate-limited by decay of apo-opsin, which will depend upon the rate of synthesis of 11-cis retinal in the RPE as well as the rate of diffusion of the chromophore within the aqueous medium of the rod cytosol (Frederiksen et al., 2012). Once rhodopsin is reformed, arrestin falls off its binding site, and phosphate groups are slowly removed to produce the unphosphorylated pigment present in a fully dark-adapted retina.

This summary leaves a number of important, unanswered questions.

(1). What are the relative contributions of the different intermediates of bleaching to the equivalent background, and how do these contributions depend upon the intensity of the bleach? Although several attempts have been made to characterize different stages in dark adaptation and to attribute them to different pigment intermediates (see especially Kennedy et al., 2001; Lamb and Pugh, 2004), these models rely upon many assumptions. What we lack are actual measurements of the concentrations and relative effectiveness of meta II and meta III in producing activation of the transduction cascade during dark adaptation (as in Kolesnikov et al., 2003), together with the time course of the changes in sensitivity and pigment regeneration in an intact eye. Although the bold outlines of recovery after bleaching are clear, many details are still lacking.

(2). What is the role of RGR opsin in rod pigment regeneration? RGR opsin can regenerate 11-cis retinal from all-trans retinal by the absorption of light rather than through an enzymatic process. We will encounter this interesting protein during our discussion of cone pigment regeneration in Section 7.3. RGR opsin is known to be highly concentrated in the RPE and can affect rhodopsin concentration (Chen et al., 2001) and rod dark adaptation (Tworak et al., 2023), but there have been no actual measurements of the rate or extent of rod pigment regeneration by this pathway, or of the relative proportions of 11-cis chromophore produced by RGR opsin on the one hand and the classical visual cycle (Fig. 2) on the other.

(3). The large translocation of transducin from the rod outer segment to the rod inner segment reduces the gain of transduction (Sokolov et al., 2002). This reduction has two effects: it allows the rods to recover from saturation to permit them to continue to respond even in bright illumination (Frederiksen et al., 2021), and it decreases the sensitivity of the transduction cascade (Majumder et al., 2013). Translocation of transducin back into the outer segment during dark adaptation is slow, requiring many minutes (Belcastro et al., 2012; Elias et al., 2004; Sokolov et al., 2002). Although measurements have been made of the dependence of transducin translocation from the outer to the inner segment on background light intensity (for example Lobanova et al., 2007; Sokolov et al., 2002), no systematic study has been made of the effects of bleaches on translocation back from the inner segment to the outer segment, which is likely to be slower when the bleach is greater (Belcastro et al., 2012). Such measurements are important, because the changes in gain produced by transducin movement are likely to contribute to the decrease and recovery of sensitivity during dark adaptation.

(4). If apo-opsin produced by bleaching is phosphorylated and bound to arrestin, what about opsin during vitamin A deprivation? As Dowling’s measurements showed, the relationship between sensitivity and apo-opsin concentration is nearly the same for opsin produced by bleaching and opsin produced by deprivation (Fig. 1B). So either both forms of opsin are phosphorylated and capped by arrestin, or the efficiency of opsin in stimulating the transduction cascade is the same regardless of whether the opsin is capped by arrestin. A similar argument would apply to opsin accumulating in an RPE65^−/− mouse, since the efficiency of this opsin in stimulating the cascade (Fan et al., 2005) is the same as that of opsin produced by bleaching (Nymark et al., 2012).

(5). Although some studies have explored the role of the inner retina during adaptation to background light (for example Dunn et al., 2007; Griffis et al., 2023), virtually nothing is known about network adaptation during desensitization and recovery produced by bright light exposure.

7. Dark Adaptation in Cones

The first indication that recovery from pigment bleaching might occur differently in rods and cones came from the observations of Goldstein (1967, 1970) and Hood and Hock (1973). They showed that following bleaching illumination in the isolated frog retina in the absence of the RPE, cone but not rod pigments could regenerate, and cone but not rod sensitivity could recover. These experiments indicated that cones can use a mechanism of pigment regeneration different from the one used by rods. Although this idea was initially greeted with skepticism, it was given renewed attention by the work of Mata and coworkers in the Travis laboratory (2002). They persuasively argued that cones must utilize some different mechanism, because the rate of regeneration by the RPE is insufficient to supply chromophore to cones in bright light. They further demonstrated an enzymatic activity in isolated retinas of cone-dominated chicken and ground-squirrel that, in the absence of the RPE, could convert all-trans retinol to 11-cis retinol. Because cones but not rods had been shown to take up and utilize 11-cis retinol to form 11-cis retinal and regenerate visual pigment (Jones et al., 1989), Mata and colleagues proposed that this production of 11-cis retinol by the isolated retina would provide a privileged pathway for chromophore to cones to speed their recovery after bleaches and in continuous bright light. Mata et al. further proposed that the cells mediating this activity were the Müller glial cells, first because cultured Müller cells from chicken retina had been shown to take up all-trans-retinol and release 11-cis-retinol into the medium (Das et al., 1992); and because Müller cells are known to contain cellular retinaldehyde-binding protein (CRALBP), which is also present in the RPE and binds to and helps transport both retinol and retinal (Bunt-Milam and Saari, 1983).

These speculations were verified and their physiological importance demonstrated by microspectrophotometry and physiological recordings from photoreceptors and whole retina. Wang and collaborators (2009) made pigment absorption measurements from single salamander photoreceptors and showed that single cones can regenerate pigment and recover sensitivity after a bleach if the cones are still attached to the rest of the retina (Fig. 8). Rods could not do this, nor could single cones isolated from the rest of the retinal tissue. Recovery of cone sensitivity in the isolated retina was abolished if the Müller cells were poisoned with L-ɑ-aminoadipic acid (see Fig. 9A and Wang et al., 2009), a glial cell toxin which kills Müller cells (Jablonski and Iannaccone, 2000). This same group showed that cones could also recover sensitivity in the isolated retina of mice and even primates (Wang and Kefalov, 2009).

Fig. 8.

Effect of 80% bleach on pigment absorbance spectrum in cones from dissociated retina (a), cones from intact retina (b), and rods from intact retina (c), in dark-adapted state (left), 2 h after a bleach (middle), and following treatment with exogenous 11-cis retinal (right). Note recovery of pigment content after bleach of cones from intact retina (b), but not of cones from dissociated retina (a) or rods from intact retina (c). Reprinted with permission from Wang et al. (2009).

Fig. 9.

Kinetics of salamander cone dark adaptation in isolated retina and eye-cup with RPE, following 40 s exposure to bright white light. (a) Recovery of sensitivity in isolated retina in Ringer solution (black) or following 48-h treatment with L-ɑ-aminoadipic acid (red). (b) Recovery of sensitivity in eye-cup preparation in Ringer solution (black) or following 48-h treatment with L-ɑ-aminoadipic (red). Reprinted with permission from Wang et al. (2009).

The question then becomes, what is the mechanism of chromophore isomerization and pigment regeneration for cone pigment? Present evidence indicates that there are at least three pathways: (1), the classical visual cycle (Fig. 2); (2), photoconversion of N-retinylidene-PE (N-ret-PE); and (3), direct light-dependent isomerization by retinal G-protein coupled receptor, also known as RGR opsin. In the following we will summarize the evidence for these pathways and then attempt to give a coherent view of how cone pigment is restored during and after bright light exposure, pointing out all of the uncertainties still remaining in our understanding.

7.1. Cone pigment regeneration by the RPE

One clear indication of an important role of RPE in cone pigment regeneration comes from the experiments of Dwight Burkhardt on turtle cones (1994). Burkhardt was able to measure pigment concentration in a turtle eye with intact pigment epithelium. He did this by detecting small changes in light transmission optically across a patch of retina approximately 2 mm in diameter in an eye-cup preparation, after scraping off the sclera from the back of the eye to provide a clearer optical path. The measuring beam was a 633 nm laser light, which detected light absorbed predominantly by the 617-nm, long-wavelength-sensitive cones (Loew and Govardovskii, 2001). Burkhardt showed that, after exposure to an intense 630 nm light, pigment concentration recovered with a time constant of between 5 and 10 minutes. He then measured the pigment concentration in steady 630-nm background illumination and found that it fit the relation

$$1-F=\frac{I_h}{I_h+I_B}$$ (3)

where F is the fraction of pigment bleached, I_B is the background light intensity, and I_h is a constant. The value of I_h can be shown to be equal to 1/(Pτ), where τ is the time constant of regeneration and P is the photosensitivity of bleaching, which is nearly the same for rod and cone pigments (see for example Makino et al., 1991; Yoshizawa and Kuwata, 1991). From I_h it is therefore possible to calculate the time constant τ for pigment regeneration in the presence of the background illumination, and it is about 8 minutes and in rough agreement with the time constant derived from recovery of pigment in darkness.

These experiments show that for red-sensitive turtle cones illuminated with red light, the rate of pigment regeneration is the approximately the same in darkness and in constant light. That conclusion is important because, as we shall see, the other two further mechanisms of cone pigment regeneration that we will describe in Sections 7.1 and 7.2 both depend upon illumination. It is probably relevant that both of these light-dependent mechanisms are maximally activated at wavelengths much shorter than the 630-nm light Burkhardt used in his experiments. Burkhardt’s experiments nevertheless demonstrate that, at least under some conditions, and perhaps in many circumstances apart from very bright illumination (Burkhardt, 1994; Mata et al., 2002), pigment regeneration in the light is mediated by the same process that regenerates pigment after bleaches in darkness. This process in Burkhardt’s experiments was very likely to have been the classical visual cycle in the RPE.

A similar conclusion emerges from the experiments of Wang and collaborators (Wang et al., 2009). The results in Fig. 9B show that cones in an eye-cup preparation (in the presence of the RPE) can recover sensitivity after a bleach even after a 48-hour treatment of the retina with L-ɑ-aminoadipic acid to kill the Müller cells. Cones can therefore recover from bleaches without the Müller cells presumably using the RPE alone. The rate of recovery may be somewhat slower after the destruction of the Müller cells, but it is possible that L-ɑ-aminoadipic acid has additional effects on retinal tissue which slow regeneration, for example by disrupting transport of 11-cis retinal from the RPE to the photoreceptors.

7.2. Cone pigment regeneration by N-ret-PE

In the disk bilayer, retinaldehydes can rapidly and reversibly condense with phosphatidylethanolamine (PE) to form the retinyl-lipid called N-retinylidene-PE (N-ret-PE). All-trans-N-ret-PE can undergo photoisomerization to 11-cis-N-ret-PE in visible light (Shichi and Somers, 1974), and 11-cis-N-ret-PE can transfer 11-cis retinal to opsin (Kimbel et al., 1970). It would therefore be possible for the formation of N-ret-PE to provide a mechanism for the photoconversion of all-trans retinal back to 11-cis retinal within the photoreceptor itself, without requiring transport to the RPE.

To see if this mechanism actually functions in the retina, Kaylor and colleagues (2017) showed that chromophore conversion from trans to cis could take place in a cell-free system from the mouse retina. They further demonstrated that rhodopsin could be regenerated by exposure to 450 nm light near the peak of absorption of N-ret-PE. Cone photoreceptors were shown to be less desensitized in 450 nm light (which maximally activates N-ret-PE isomerization) than in 560 nm light. A difficulty with these observations, however, is that much of the recovery of sensitivity could have been mediated by RGR opsin, whose role had not yet been discovered. Later experiments seem to indicated that N-ret-PE can contribute to pigment regeneration even in retinas lacking RGR opsin (see Fig. 7 of Morshedian et al., 2019), but the extent of this contribution remains unclear.

7.3. Cone pigment regeneration by RGR opsin

Retinal G protein-coupled receptor (RGR) opsin, a rhodopsin-related protein, was first identified in bovine retina and localized to intracellular membranes in the RPE and Müller cells (Pandey et al., 1994). Hao and Fong (1999) showed that RGR opsin isolated from bovine RPE could bind to all-trans retinal in darkness and convert it to 11-cis retinal upon illumination with blue light. This same group made a mouse in which the gene for RGR opsin had been deleted (Chen et al., 2001). Mutant mice freely behaving in an illuminated environment had less rhodopsin than WT mice. The amount of rhodopsin decreased with increasing steady illumination, and this decrease was greater for animals lacking RGR opsin than for WT mice.

These experiments indicated that RGR opsin may have a role in regenerating rod photopigment, which later experiments have confirmed (Tworak et al., 2023). The activity of such a light-dependent mechanism of pigment regeneration would, however, seem more useful to cones than to rods, to support regeneration in bright illumination. To pursue this possibility, Morshedian and colleagues (2019) made a mouse that was lacking RGR opsin but also lacked Gnat1, the G protein of rod transduction. With such a mouse they were able to record cone responses from whole isolated retina without interference from rods, to investigate the contribution of RGR opsin to cone physiology.

Some of the results of their experiments are given in Fig. 10. The traces in this figure are extracellular recordings from isolated retina treated with synaptic blockers to isolate photoreceptor responses. Each group of traces gives responses to a series of increasing light intensities in the presence of a continuous bright background light, with the red trace in each group always for the same flash intensity to facilitate comparison. In a mouse lacking Gnat1 but with RGR opsin, cone responses are maintained for a prolonged period, only gradually decreasing in sensitivity and maximum amplitude (top traces). When RGR opsin was deleted (middle traces), or if the Müller cells were poisoned with L-ɑ-aminoadipic acid (lower-most traces), responses decayed much more rapidly. Deletion of RGR also nearly eliminated recovery after exposure to bright light but only for prolonged bleaches. For a brief (350 ms) bleaching exposure, recovery of sensitivity in the isolated retina was much smaller. Since RGR opsin needs light to regenerate chromophore, recovery after bleaching could be facilitated by this mechanism only if the light were sufficiently prolonged to generate a significant pool of 11-cis retinal.

Fig. 10.

Whole retina extracellular recordings of light responses of mouse cone photoreceptors with and without deletion of RGR opsin. All mice were Gnat1^−/− to prevent any contribution of rods. Isolated retinas (without RPE) were placed into recording chambers (Vinberg et al., 2014) and perfused with synaptic blockers to suppress responses from retinal interneurons and glial cells. Perfusate also contained 10 mM all-trans retinol to maintain constant substrate concentration. Responses were recorded to 565-nm flashes stimulating mouse M cones, in dark-adapted retina (DA) and at 15, 30, and 60 minutes of continuous exposure to 505-nm background light estimated to bleach 10⁶ M-opsin pigment molecules per second. (A) Retinas were RGR^+/+ and Gnat1^−/−. (B) Retinas were RGR^−/− and Gnat1^−/−. (C) Retinas were RGR^+/+ and Gnat1^−/− but perfused with 10 mM L-ɑ-aminoadipic. Each panel shows responses to a series of increasing flash intensities with the red trace in each group always for the same intensity of 3.7 × 10⁵ photons μm⁻² to facilitate comparison. Flashes were 10 ms. Reprinted with permission from Morshedian et al. (2019).

Two recent studies have further advanced our understanding of the “retinal” visual cycle. Tworak and colleagues (2023) were able to delete the mouse gene for RGR opsin selectively in either the Müller cells or RPE. From recordings of the b-wave in intact animals, they concluded that deletion of RGR opsin affects sensitivity in the presence of continuous background light (as in Fig. 10) as well as sensitivity during recovery from illumination (as in Fig. 9). Both the Müller cells and RPE seemed to contribute about equally. In another series of experiments, Kaylor and collaborators (2024) identified the enzyme (called a retinol dehydrogenase) in zebrafish which converts 11-cis retinol to 11-cis retinal. Recall that cones can use 11-cis retinol to regenerate visual pigment, but rods cannot (Jones et al., 1989). The 11-cis retinol must then be enzymatically converted in the cones from retinol to retinal for the chromophore to bind to the cone photopigment. The zebrafish enzyme that does this conversion was called ZCRDH and was abundantly expressed in cone inner segments. When the gene for ZCRDH was deleted, cones could no longer use 11-cis retinol to regenerate pigment. The human retinol dehydrogenase most similar to ZCRDH is RDH12, which is the only mammalian candidate protein to exhibit 11cROL-oxidase activity. RDH12 may very well be the form of retinal dehydrogenase used in mammalian cones.

8. Cone dark adaptation: a summary and some important unanswered questions

These experiments taken together indicate that visual pigment regeneration is not as simple as once thought. In addition to the classical visual cycle (Fig. 2), all-trans chromophore can be regenerated to 11-cis within Müller cells in the retina. The Müller cells appear to use RGR opsin, and this RGR-driven regeneration is light-dependent. This mechanism is especially suitable for pigment regeneration in the presence of bright background light. It seems to produce only 11-cis retinol, which cones but not rods can convert to 11-cis retinal with a retinol dehydrogenase, which is ZCRDH in zebrafish and probably RDH12 in mammals. Moreover, cones but not rods can take up and utilize chromophore from their inner segments (Jin et al., 1994). RGR opsin is even more heavily expressed in the RPE than in the Müller cells (Morshedian et al., 2019; Tworak et al., 2023; Zhang et al., 2019) and may contribute to regeneration of rod as well as cone pigment (Chen et al., 2001; Tworak et al., 2023).

These studies leave a number of important, unanswered questions.

(1). What is the rate of pigment regeneration in isolated retina? Many experiments have investigated the time course of recovery of cone sensitivity after a bleach with and without RGR opsin and with and without the Müller cells, in isolated retina or in eye-cup (Fig. 9). These results approach the problem from the wrong direction, because RGR opsin works only in the light and not in the dark after bleaching. RGR opsin can assist recovery after exposure to bright illumination only as a result of accumulation of 11-cis chromophore during the bleach. The bleach must be prolonged, because brief bleaches are much less effective even when they are quite bright (Morshedian et al., 2019). No measurements have been made of the extent of pigment bleaching at steady state as a function of background light intensity. These measurements—and these measurements alone—make possible a determination of the actual rate at which RGR regenerates pigment.

(2). What is the contribution of the Müller-cell pathway to regeneration in the isolated retina? The only information we presently have derives from poisoning the Müller cells with L-ɑ-aminoadipic acid. It is however possible that L-ɑ-aminoadipic acid has effects on retinal cells in addition to those on Müller cells, for example by disrupting transport of 11-cis retinal to and from the RPE and the photoreceptors. Now that we have the gene for the retinol dehydrogenase (at least in zebrafish), it should be possible to eliminate production of 11-cis retinal by the Müller-cell pathway without the use of a toxin.

(3). Is light-dependent regeneration of chromophore via RGR opsin the only mechanism of pigment regeneration in the isolated retina? And if so, why does the isolated retina show some partial recovery even after brief bleaching illumination, which wouldn’t be expected to lead to much accumulation of 11-cis chromophore via the Müller-cell pathway?

(4). What is the contribution of the Müller-cell pathway in the intact eye? Measuring pigment in the intact eye is challenging but not impossible (Burkhardt, 1994). Mutant animals lacking RGR opsin are available in both mouse and zebrafish, and RGR opsin is minimally affected by red light so that its effects can be gauged by comparing stimulation with light of different wavelengths. From measurements of pigment concentration and sensitivity at a variety of background illuminations, either simultaneously or in succession, it should be possible to calculate and compare the rates of pigment renewal with and without RGR opsin activity.

(5). Why is there so much RGR opsin in the RPE, and what is it doing there? Experiments like those in Figs. 9 and 10 and in Kaylor et al. (2024) indicate that if the Müller-cell conversion of all-trans to 11-cis retinol is blocked, the isolated retina cannot regenerate cone photopigment. If it becomes possible to measure the rate of pigment regeneration in the intact eye, it would be possible to compare the rates at wavelengths of bleaching light that favor or disfavor RGR opsin activity, at (say) 500 nm and 635 nm. Similar experiments could be done on animals lacking RGR opsin on the one hand, or lacking the “retinal” cycle by deleting the retinol dehydrogenase. It may also be possible to compare results in animals with RGR opsin selectively knocked out in the RPE and Müller cells (Tworak et al., 2023), though perhaps not in mouse because direct measurements of pigment concentration are next to impossible for mouse cones with present techniques.

(6). How is 11-cis retinol produced by the Müller cells transported to cones? If no mechanism were available to target the chromophore to cones, it is possible that much of it would be taken up by the much more numerous population of rods (in mouse), which could not use it. An accumulation of 11-cis retinol may even be toxic to the rods (Jones et al., 1989). It is possible that chromophore produced in the retina cannot reach rod outer segments but only their inner segments, which do not seem to be able to take up chromophore (Jin et al., 1994). At present, we know nothing about the properties of the mechanism that allows cones specifically to take up chromophore produced by the Müller cells, if indeed such a mechanism exists.

(7). Is the 11-cis chromophore produced by RGR opsin in the RPE exported to the retina as 11-cis retinol or as 11-cis retinal? It seems possible that 11-cis retinol produced in the RPE by RGR opsin is bound immediately to a carrier protein and exported as such. If so, the chromophore produced by RGR opsin in the RPE may be utilized only by cones. The 11-cis retinol may, however, be converted by a retinol dehydrogenase in the RPE to 11-cis retinal, to be used by both rods and cones. Experiments in the intact retina with RPE from animals with RGR opsin knocked out in the Müller cells and with and without deletion of the retinol dehydrogenase could help to answer the mechanism of pigment chromophore production in the RPE.

(8). Does the relative contribution of the “retinal” visual cycle depend upon the position of the cones within the retina? Is regeneration by RGR opsin and Müller cells in primates of greater (or lesser) importance in the fovea than in the periphery? Both fovea and periphery have Müller cells, but it is possible that these cells function somewhat differently in different parts of the eye.

(9). Finally, what is the role (if any) of N-ret-PE in regeneration of cone pigment? Although previous experiments indicated that N-ret-PE may make some contribution in animals lacking RGR opsin (Morshedian et al., 2019; Tworak et al., 2023), it should be possible to specify its role more clearly at least for the isolated retina. Measurements of pigment concentration can be made directly with microspectrophotometry from some species (Kaylor et al., 2024). If it is possible to show with this technique that pigment can be regenerated in single cones still attached to the retina (Wang et al., 2009), and if pigment regeneration is nulled either by deletion of RGR opsin or retinol dehydrogenase (or both), a role for N-ret-PE would seem to be rather small.

Data availability

No new data were used in preparing this article.