Role of recoverin in rod photoreceptor light adaptation

Abstract

Key points

• Recoverin is a small molecular‐weight, calcium‐binding protein in rod outer segments that can modulate the rate of rhodopsin phosphorylation. We describe two additional and perhaps more important functions during photoreceptor light adaptation.

• Recoverin influences the rate of change of adaptation. In wild‐type rods, sensitivity and response integration time adapt with similar time constants of 150–200 ms. In Rv–/– rods lacking recoverin, sensitivity declines faster and integration time is already shorter and not significantly altered.

• During steady light exposure, rod circulating current slowly increases during a time course of tens of seconds, gradually extending the operating range of the rod. In Rv–/– rods, this mechanism is deleted, steady‐state currents are already larger and rods saturate at brighter intensities.

• We propose that recoverin modulates spontaneous and light‐activated phophodiesterase‐6, the phototransduction effector enzyme, to increase sensitivity in dim light but improve responsiveness to change in brighter illumination.

Abstract

Recoverin is a small molecular‐weight, calcium‐binding protein in rod outer segments that binds to G‐protein receptor kinase 1 and can alter the rate of rhodopsin phosphorylation. A change in phosphorylation should change the lifetime of light‐activated rhodopsin and the gain of phototransduction, but deletion of recoverin has little effect on the sensitivity of rods either in the dark or in dim‐to‐moderate background light. We describe two additional functions perhaps of greater physiological significance. (i) When the ambient intensity increases, sensitivity and integration time decrease in wild‐type (WT) rods with similar time constants of 150–200 ms. Recoverin is part of the mechanism controlling this process because, in Rv–/– rods lacking recoverin, sensitivity declines more rapidly and integration time is already shorter and not further altered. (ii) During steady light exposure, WT rod circulating current slowly increases during a time course of tens of seconds, gradually extending the operating range of the rod. In Rv–/– rods, this mechanism is also deleted, steady‐state currents are already larger and rods saturate at brighter intensities. We argue that neither (i) nor (ii) can be caused by modulation of rhodopsin phosphorylation but may instead be produced by direct modulation of phophodiesterase‐6 (PDE6), the phototransduction effector enzyme. We propose that recoverin in dark‐adapted rods keeps the integration time long and the spontaneous PDE6 rate relatively high to improve sensitivity. In background light, the integration time is decreased to facilitate detection of change and motion and the spontaneous PDE6 rate decreases to augment the rod working range.

Key points

• Recoverin is a small molecular‐weight, calcium‐binding protein in rod outer segments that can modulate the rate of rhodopsin phosphorylation. We describe two additional and perhaps more important functions during photoreceptor light adaptation.

• Recoverin influences the rate of change of adaptation. In wild‐type rods, sensitivity and response integration time adapt with similar time constants of 150–200 ms. In Rv–/– rods lacking recoverin, sensitivity declines faster and integration time is already shorter and not significantly altered.

• During steady light exposure, rod circulating current slowly increases during a time course of tens of seconds, gradually extending the operating range of the rod. In Rv–/– rods, this mechanism is deleted, steady‐state currents are already larger and rods saturate at brighter intensities.

• We propose that recoverin modulates spontaneous and light‐activated phophodiesterase‐6, the phototransduction effector enzyme, to increase sensitivity in dim light but improve responsiveness to change in brighter illumination.

Introduction

Recoverin (or S‐modulin) was first identified as a small molecular‐weight, Ca²⁺‐binding protein that binds to and regulates G‐protein receptor kinase 1 (GRK1), the kinase that phosphorylates rhodopsin (Kawamura, 1993; Chen et al. 1995). In its Ca²⁺‐bound form, recoverin binds to and inhibits the kinase. Decreases in Ca²⁺ cause recoverin to be released and kinase activity to be enhanced. These observations immediately suggested that recoverin could play a role in photoreceptor light adaptation (Kawamura, 1993). In darkness, when the Ca²⁺ concentration of the rod outer segment is relatively high (Woodruff et al. 2002), kinase activity would be down‐regulated and relatively low. In the light, when the Ca²⁺ concentration decreases, recoverin would become unbound from the kinase and the kinase would phosphorylate rhodopsin more rapidly. This effect could, at least in theory, reduce the lifetime of light‐activated rhodopsin (Rh^*), decrease the gain of production of transducin‐GTP and lower the sensitivity of the rod.

The first recordings from rods in mice lacking the gene for recoverin (Rv–/–) initially appeared to be consistent with a role of recoverin in regulation of Rh^* lifetime (Makino et al. 2004) because deletion of recoverin accelerated the decay of the rod light response. However, deletion had little effect on the initial time course or peak amplitude of the response to flashes (Sampath et al. 2005), indicating that the gain of transduction was almost unaltered. Peak amplitude could be modestly decreased in Rv–/– rods also lacking the guanylyl cyclase‐activating proteins (the GCAPs), indicating a small effect of recoverin on Rh^* lifetime (Makino et al. 2004), as confirmed in subsequent experiments (Chen et al. 2010a). There was, however, little effect of recoverin deletion on adaptation to background light (Makino et al. 2004; Chen et al. 2010d), an observation apparently inconsistent with a major role of recoverin in regulating photoreceptor sensitivity.

Because of the uncertainty of the role of recoverin in rod function, we have reinvestigated the physiological effects of recoverin deletion. We show that deletion alters the rate of change of sensitivity during light adaptation. In wild‐type (WT) rods, background light produces a decrease in both sensitivity and response integration time with similar time constants of 150–200 ms. Recoverin is an essential part of this process because, in Rv–/– rods, sensitivity adjusts much more rapidly when the ambient illumination is increased and integration time remains unaltered. Recoverin is also essential for a mechanism that produces a slow increase in circulating current during prolonged illumination, which increases the rod response amplitude and helps to modulate the background light intensity at which rods saturate. The results of the present study and those of previous experiments (Chen et al. 2010d, 2012, 2015) indicate that neither of these effects of recoverin is likely to be produced by regulation of Rh^* phosphorylation or by channel modulation, and that each is more probably the consequence of an effect on spontaneous and light‐activated PDE6.

Methods

Ethical approval and animals

Experiments were performed in accordance with the policy of The Journal of Physiology, as well as with the rules and regulations of the NIH guidelines for research animals, as approved by the institutional animal care and use committees of the University of California, Los Angeles, USA. Animals were kept under a 12:12 h light/dark cycle in approved cages and supplied with ample food and water. Animals in all experiments were dark‐adapted for 3–5 h and killed before tissue extraction by cervical dislocation.

Mice were aged 3–6 months and were either WT C57BL/6 from Jackson Laboratory (Bar Harbor, ME, USA), Rv–/– mice lacking the gene for recoverin, or GCAPs–/–;Rv–/– lacking the gene for recoverin and both genes for the GCAPs. Both Rv–/– and GCAPs–/– mouse lines were generously supplied to us by Professor Jeannie Chen of the University of Southern California (Los Angeles, CA, USA). Makino et al. (2004) showed, for the Rv–/– rods, that levels of rhodopsin and rhodopsin kinase, arrestin, transducin α‐subunit, the α‐, β‐, γ‐subunits of PDE, RGS9‐1, and guanylate cyclase E were similar in WT and knockout animals; there was no sign of degeneration of Rv–/– rods even up to 1 year, which is older than any of the animals used in our experiments. We have also seen no sign of degeneration in these animals. We have additionally used the N‐terminal antibody of Sarfare et al. (2014) to demonstrate that there was no detectable change in GARP2 expression in Rv–/– animals by comparison to WT. For GCAPs–/– mice, Mendez et al. (2001) showed that there were no detectable changes in the expression levels of GC1, GC2 and PDE subunits, or of rhodopsin, transducin, rhodopsin kinase, arrestin, recoverin or RGS9; moreover, there was no sign of photoreceptor degeneration up to 8 months of age.

Tissue dissection and preparation

Eyes were enucleated under dim red light. The anterior portion of the eye was cut and the lens and cornea were removed in darkness by means of infrared image converters. The retina was isolated from the eyecup; the retinal pigment epithelium was removed with fine tweezers, and the retina was chopped into small pieces with a razor blade. The pieces were then transferred to the recording chamber in complete darkness by means of infrared goggles (American Technologies Network Corporation, San Francisco, CA, USA).

Solutions and pipettes

During recording, the photoreceptors were continuously perfused with Dulbecco's modified Eagle's medium (D‐2902; Sigma Chemical, St Louis, MO, USA) supplemented with 15 mm NaHCO₃, 2 mm Na succinate, 0.5 mm Na glutamate, 2 mm Na gluconate and 5 mm NaCl, bubbled with 5% CO₂ in O₂ (pH 7.4). The total concentrations of the major salts were (in mm): 135 Na⁺, 5.4 K⁺, 1.9 Ca²⁺, 0.8 Mg²⁺, 123.5 Cl^– and 15 ${\mathrm{HCO}}_3^{-}$. The medium used for dissection and storage of the tissue during the experiment was also supplemented Dulbecco's but, in addition, contained 10 mm glucose, 5 mm sodium ascorbate and 10 mg 10 mL⁻¹ (0.1%) BSA. Temperature was maintained at 35–38°C. The recording electrodes were filled with Locke's solution, which contained (in mm) 93 NaCl, 2.1 KCl, 2.6 CaCl₂, 1.8 MgCl₂, 2.0 NaHCO₃ and 10.8 Hepes at pH 7.4. Fire‐polished borosilicate glass was pulled with a micropipette puller (P‐97; Sutter Instruments, Novato, CA, USA) to produce pipettes with rapidly tapering shanks used for our recordings. The tip size was further adjusted under a compound microscope by moving the pipette close to a platinum heating wire until the tip had melted to an inner diameter that would fit the outer segment of the photoreceptor and provide a good seal. We used the pipettes uncoated.

Suction‐electrode recording and light stimulation

Responses of single photoreceptor outer segments were recorded at 35–38°C with the suction‐electrode technique (Baylor et al. 1979). The change in outer‐segment membrane current produced by a stimulus was recorded with a current‐to‐voltage converter (Axopatch 200A; Axon Instruments, Foster City, CA, USA), low‐pass filtered at 30 Hz with an eight‐pole Bessel filter (Kemo Limited Electronic Filters, Dartford, UK) and sampled at 100 Hz. Digitized data were recorded with Clampex, version 8.0 (Axon Instruments) and were analysed with Origin Pro (OriginLab Inc., Northampton, MA, USA). Curve fitting and plotting of data were also performed in Origin Pro. Statistical calculations were conducted either in Origin Pro or Excel (Microsoft Corp., Redmond, WA, USA); values are given as the mean ± SEM and the t tests were two‐tailed and assumed unequal variances.

Cells were stimulated with a dual‐beam optical bench; the light of halogen lamp bulbs was passed through electronic shutters (Uniblitz; Vincent Associates, Rochester, NY, USA) and interference filters at 500 nm, near the peak of spectral sensitivity of mouse rods (Nymark et al. 2012). The intensity of the light was attenuated with absorptive neutral‐density filters (Fish‐Schurman Corp., New Rochelle NY, USA). The light intensity was calibrated with a photodiode (UDT Instruments, San Diego, CA, USA).

Results

Effect of recoverin deletion on the time course of light adaptation

Earlier experiments have shown that recoverin can regulate the rate of photoreceptor response decay (Chen et al. 2012; Chen et al. 2015). To investigate the time course of this effect and its relationship with the change in sensitivity in background light, we gave a 4 s step of light at one of three intensities: 38, 438 or 1350 photons μm⁻² s⁻¹ (Fig. 1). We stimulated the rod with a flash of 70 photons μm⁻² either in darkness before the light step or at a series of delays after presentation of the light step; delays ranged from 100 to 2000 ms. For each presentation of the step, only one flash was given. After each step presentation, the rod was left in darkness for 30 s to 1 min to permit sensitivity to return to its dark‐adapted value. The next step was then given with a different step intensity or flash delay. For all of the rods shown in Fig. 1, steps were given at each of the three step intensities and at all five time delays. The mean responses at the different time points are superimposed on each of the three step‐intensity values to illustrate the progression of the change in response amplitude as a function of time.

Figure 1. The effect of background light on the onset of adaptation.

Data traces are means of responses of nine WT rods (A–C) and 12 Rv–/– rods (D–F), exposed to steps of light 4 s in duration and stimulated with a 20 ms flash having the fixed value of 70 photons μm⁻² at the following times after presentation of the light step (in ms): 100 (red traces), 250 (blue traces), 500 (pink trances), 1000 (green traces) and 2000 ms (navy traces). Background light intensities were 38 photons μm⁻² s⁻¹ (A and D), 438 photons μm⁻² s⁻¹ (B and E) and 1350 photons μm^‐2 s^‐1 (C and F). Mean maximum values of dark‐adapted currents $(r_{max}^D)$ were 12.6 ± 0.3 pA for (A) to (C) and 12.6 ± 0.4 pA for (E) to (F). [Color figure can be viewed at http://wileyonlinelibrary.com]

These results are given in Fig. 1 A–C for WT rods and Fig. 1 D–F for Rv–/– rods. The recordings from WT and Rv–/– rods show two differences. First, the response to the step itself in WT rods had a different waveform from that of the Rv–/– rods. This difference is particularly evident in Fig. 1 C and F. We explore this feature of the response in more detail below (Figs 3 and 4). In the second place, there were differences between WT and Rv–/– rods in the time course of the decrease in amplitude of the flash response. This effect is more easily seen in the records of Fig. 2 A–D, which show the responses of Fig. 1 at the various time points only for the two brighter step illuminations, where the differences are most evident. To obtain these records, we recorded for each rod the response to the step alone; then, for each test flash, we subtracted the response to the step alone from the response to the step with test flash. The result of this subtraction was averaged among the rods at each of the time points and step intensities, and these averages are shown in Fig. 2 A and B for WT rods and Fig. 2 C and D for Rv–/– rods. These results suggest that the time course of decrease of sensitivity in the presence of the background light was slower for WT rods than for Rv–/– rods.

Figure 3. Responses of WT and Rv–/– rods to prolonged light steps.

WT (A–C) and Rv–/– (D–F) rods were stimulated with 60 s steps of light at intensities of 72 (A and D), 210 (B and E) and 950 (C and F) photons μm⁻² s⁻¹. Black traces give the mean response of (A–C) eight rods; (D) seven rods; (E) eight rods; and (F) nine rods. Grey traces give ± SEs calculated point by point. Maximum values of dark‐adapted response $(r_{max}^D)$ were 12.1 ± 0.4 pA for (A and B), 11.9 ± 0.4 for (C) and 11.5 – 11.8 pA for (D–F).

Figure 4. Responses of GCAPs–/– and GCAPs–/–;Rv–/– rods to 60 s steps of light at an intensity of 950 photons μm⁻² s⁻¹ .

A, mean of 16 GCAPs–/– rods. B, mean of 19 GCAPs–/–;Rv–/– rods. Deletion of recoverin eliminated both the slow decline of current during illumination and the undershoot at stimulus cessation. Grey traces in both (A) and (B) give ± SEs calculated point by point.

Figure 2. Time course of change in response amplitude after presentation of light step.

The same nine WT (A and B) and 12 Rv–/– rods (C and D) as in Fig. 1 but showing only mean responses to the 70 photons μm⁻² flashes, obtained by mathematically subtracting rod by rod the response to the step only from the response to both step and test flash (see text). Same colour coding as in Fig. 1; black traces are responses to 70 photons μm⁻² flash from same rods, dark‐adapted and in the absence of a light step. Background light intensities were 438 photons μm⁻² s⁻¹ (A and C) and 1350 photons μm⁻² s⁻¹ (B and D). E, mean amplitude of responses for rods of (A) to (D) to 70 photons μm⁻² flash as a function of time after presentation of light steps of 438 photons μm⁻² s⁻¹ (circles) and 1350 photons μm⁻² s⁻¹ (squares). WT rods, black symbols; Rv–/– rods, red symbols. SEs are plotted but, in some cases, were smaller than the diameter of the data points. F, mean ± SE integration time of responses of Fig. 2 B and D to 70 photons μm⁻² flash as a function of time after presentation of light steps of 1350 photons μm⁻² s⁻¹ for WT rods (black squares) and Rv–/– rods (red squares). Curve for WT data is a best‐fitting single‐exponential decay function with a time constant of 172 ms. [Color figure can be viewed at http://wileyonlinelibrary.com]

This difference can be more easily seen in Fig. 2 E, where we have plotted the mean peak amplitude of the flash response at the two brighter backgrounds, for WT rods (black) taken from Fig. 2 A and B, and for Rv–/– rods (red) taken from Fig. 2 C and D. To facilitate comparison, responses have been normalized to their peak amplitudes in darkness. The means have been fitted with single‐exponential decay functions, with best‐fitting time constants of 163 ms (black circles), 127 ms (black squares), 58 (red circles) and 49 ms (red squares). Cell‐by‐cell fits gave best‐fitting time constants of 163 ± 5 ms (black circles) and 153 ± 4 ms (black squares) for nine WT rods, and 77 ± 3 ms (red circles) and 47 ± 1 ms (red squares) for 12 Rv–/– rods. The differences in decay time constants between WT and Rv–/– rods were highly significant at both light intensities (P < 10⁻⁸). These results show that WT rods adapt to the presence of semi‐saturating backgrounds more slowly than Rv–/– rods. In experiments not shown, we extended the time of presentation of the test flashes out to 2 min, and no further change in flash response amplitude was detected.

Our previous experiments showed that background light produces a quickening of the decay of the WT rod flash response, which is no longer observed after recoverin is deleted (Chen et al. 2012, 2015). To substantiate and quantitate a similar effect in the experiments of Figs 1 and 2, we measured the integration time of the rod response as a function of time after the presentation of the brightest of the three backgrounds. We used integration times instead of exponential fits because, as first shown by Makino et al. (2004), the responses of Rv–/– rods have a discontinuous time course of decay characterized by a pronounced ‘kink’ and cannot be fit by a single exponential (Fig. 2 C and D). For WT rods (Fig. 2 F, black symbols), the integration time declined exponentially from 242 ± 11 ms in a dark‐adapted rod to 204 ± 7 ms after 100 ms of background exposure, and then gradually thereafter to 166 ± 7 ms after 2000 ms (n = 9). These differences were significant between dark‐adapted and 100 ms (P < 0.02), dark‐adapted and 2000 ms (P < 10⁻⁴) and 100 ms and 2000 ms (P < 0.002). The mean integration times in Fig. 2 F were fitted with a single exponential decay function with a best‐fitting time constant of 172 ms. Cell‐by‐cell fits to exponential decay functions gave a mean time constant of 190 ± 30 ms, which was not significantly different from the time constant of decay of response amplitude at this same background intensity (P > 0.24) (Fig. 2 E, black squares). Our measurements are therefore consistent with the supposition that, in WT rods, both sensitivity and integration time decline with the same time course after presentation of a background light.

For Rv–/– rods (n = 12), the dark‐adapted integration time was 186 ± 13 ms, which is significantly less than the integration time of dark‐adapted WT rods (P < 0.001); it was then 195 ± 13 ms after 2000 ms of presentation of the background and was not significantly different at these two time points (P > 0.1) or at any other two time points where measurements were made (e.g. 100 ms vs. 2000 ms, P > 0.6). Mean integration times for the Rv–/– rods are given as the red squares in Fig. 2 F and confirm our earlier observations that responses of Rv–/– rods are not significantly accelerated in background light (Chen et al. 2012, 2015).

Effect of recoverin deletion on step response decay

In addition to the relatively rapid effects on sensitivity and response waveform illustrated in Fig. 2, the data in Fig. 1 show differences in the waveform of the response to the step itself. To explore this phenomenon in greater detail, we presented 60 s light steps to WT and Rv–/– rods at three intensities and compared the waveforms of their responses (Fig. 3). As Calvert et al. (2002) first demonstrated for bullfrog, and as we subsequently showed for mouse rods (Chen et al. 2010d), responses to prolonged light steps decay in two phases. In the mouse, the faster component is well fit by a single‐exponential decay time constant with a value of a few hundred ms; it is eliminated by deletion of the GCAPs (Chen et al. 2010d). The slower time phase of decay lasts many tens of seconds and is unaffected by GCAPs deletion.

The responses in Fig. 3 D–F demonstrate that this slower decay is eliminated by recoverin deletion. To analyse this effect in more detail, we calculated the mean values of the currents cell by cell between 4 and 5 s near the beginning of the step after decay of the fast component, as well as between 58 and 59 s near the end of the step. For each time interval, the 100 separate values of current were averaged. The difference was then taken cell by cell between the mean value of current at 4–5 s and at 58–59 s, and these differences were averaged across cells. The mean differences for WT rods at the three step intensities were (mean ± SE): 0.44 ± 0.19 pA (n = 9) (Fig. 3 A); 0.67 ± 0.11 pA (n = 9) (Fig. 3 B); and 0.93 pA ± 0.14 (n = 8) (Fig. 3 C). For the Rv–/– rods, the corresponding values were: −0.08 pA ± 0.11 (n = 7) (Fig. 3 D); 0.01 ± 0.20 pA (n = 8) (Fig. 3 E); and 0.17 ± 0.24 pA (n = 9) (Fig. 3 F). These values were then evaluated for significance by comparing WT and Rv–/– rods cell by cell for steps at the same light intensity. The following comparisons were significant at the indicated P values: P < 0.037 (Fig. 3 A and D); P < 0.02 (Fig. 3 B and E); and P < 0.02 (Fig. 3 C and F).

If, instead of giving these differences as current values, we calculated the percentage drop in current from the peak amplitude to the light step, the values for WT at the three step intensities were remarkably consistent: 8.7 ± 3.8%, 8.0 ± 1.4% and 9.7 ± 1.5%, with a global mean of 8.8 ± 1.4% (n = 26). For the Rv–/– rods, the global mean was 0.15 ± 2.0% (n = 24). Because all of these measurements were independent, we evaluated the significance of the percent difference across all three light intensities and P < 0.001. The deletion of recoverin therefore removes a slow component of current decay that is present in WT rods.

If the faster component of response decay is produced primarily by cyclase activation by means of the GCAPs, and the slower component primarily by recoverin modulation, deletion of both the GCAPs and recoverin should produce a step response with no decay at all. To examine this possibility, we first repeated the experiment of Burns et al. (2002) and recorded step responses from rods lacking the GCAPs (Fig. 4 A). Responses decayed without an initial fast time course and with a slow time course similar to that of WT rods (Chen et al. 2010d). At the cessation of illumination, there was a pronounced undershoot, during which the current was transiently larger in darkness than before illumination. This undershoot then decayed back to the baseline with a time constant for the rods of Fig. 4 A of ∼4 s.

In Fig. 4 B, we show the result of deleting both the GCAPs and recoverin. Both the fast and slow phases of current decay were eliminated, leaving only a small initial component of current increase produced presumably by electrogenic Na⁺/Ca²⁺‐K⁺ exchange (Hodgkin et al. 1987). To compare the waveforms of the responses, we again calculated cell by cell the mean current between 4 and 5 s and between 58 and 59 s. We took the difference of these values across all of the recordings, which gave 1.2 ± 0.3 pA for the GCAPs–/– rods (n = 16) and 0.2 ± 0.1 for GCAPs–/–;Rv–/– rods (n = 18). These values were significantly different (P < 0.004). The mean value for the GCAPs–/–;Rv–/– rods was somewhat above zero, and comparison with the T distribution indicated that the probability that the difference was in fact zero (such that the current was completely flat between 4 and 59 s) was 0.052. We were therefore unable to reject the possibility that some decay of current occurred in GCAPs–/–;Rv–/– rods during the 60 s light step, although the decay was significantly less than in the GCAPs–/– rods.

The undershoot of the GCAPs–/– rods was also eliminated by deletion of recoverin. To compare undershoots from our sample of rods, we integrated the current response between 60 and 80 s, the time interval that immediately followed the light step during which the undershoot occurred. For the GCAPs–/– rods, the integral was –34.0 ± 5.5 pC (n = 16) and was negative, reflecting the swing of the current below the dark resting level (which we had set to zero). For GCAPs–/–;Rv–/– rods, the integral was 21.4 ± 3.9 pC (n = 18) and was positive, reflecting the slow decay of the response back to the baseline. This difference was highly significant (P < 10⁻⁸).

Effect of recoverin deletion on sensitivity in background light

The slow decay of the response to steps of light in WT rods in Fig. 3 A–C would have the effect of slowly increasing the circulating current, potentially increasing the amplitude of responses to flashes superimposed upon steady illumination. We therefore considered whether deletion of recoverin, which alters this slow decay, might also alter sensitivity during light adaptation in a way previous experiments had not revealed. To explore this question, we re‐examined background adaptation for WT and Rv–/– rods by recording responses over their entire range of illumination. Accordingly, we first recorded a family of responses without a background, and we then turned on the dimmest background. After waiting 2 min for the circulating current to reach steady‐state, we recorded a further flash family. The background light was extinguished and the rod left in darkness for 2–3 min before turning on the next brighter background. This procedure was repeated for each of the backgrounds up to saturation of the rod response.

Typical results are illustrated in Fig. 5. For both WT and Rv–/– rods, background light produced an decrease in circulating current, indicated by the decrease in maximum amplitude of the response. Sensitivity was also decreased, as can be seen by comparing responses in each of the backgrounds to the dimmest values of ϕ, the number of photons μm⁻² delivered by the flash. For ease of comparison, we show, in red, the responses of WT and Rv–/– rods to the same stimulus delivering 221 photons μm⁻². For both WT and Rv–/– rods, there was a gradual decrease in the amplitude of these responses as the background intensity was increased.

Figure 5. Light adaptation of WT and Rv–/– rods. Responses to 20 ms flashes from rods dark‐adapted or exposed to steady background light.

A–D, means of four WT rods in dark (A) and in the following backgrounds (photons μm⁻² s⁻¹): 201 (B), 764 (C) and 2060 (D). (E–H), means of six Rv–/– rods in dark (E) and in the following backgrounds (photons μm⁻² s⁻¹): 201 (F), 2060 (G) and 7460 (H). Values of ϕ were as follows (photons μm⁻²): (A) 2.4, 8, 21, 70, 122, 221, 403, 733, 1430 and 2600; (B) 8, 21, 70, 122, 221, 403, 733, 1430, 2600 and 4730; (C) 21, 70, 122, 221, 403, 733, 1430, 2600 and 4730; (D) 70, 221, 733, 1430, 2600, 4730 and 8620; (E) 2.4, 8, 21, 70, 221, 733 and 2600; (F) 8, 21, 70, 221, 733, 2600 and 8620; (G) 21, 70, 221, 733, 2600, 8620 and 22700; and (H) 70, 221, 733, 2600, 8620 and 22700. Red traces in all of (A) to (H) are responses to 221 photons μm⁻². [Color figure can be viewed at http://wileyonlinelibrary.com]

These effects can be assessed more quantitatively in Fig. 6, where we have plotted the peak amplitude of the flash response for the rods of Fig. 5, normalized to the maximum response amplitude of the dark‐adapted photoreceptor. As we and others have shown for vertebrate photoreceptors in a variety of species (Fain et al. 2001), background light causes the response–intensity functions to shift to the right and downward, reflecting both a decrease in circulating current and in photoreceptor sensitivity. The curves have been fitted with the equation (Lamb et al. 1981):

$$r(\phi )=r_{\max }\left[1-exp\left(-a\phi \right)\right]$$ (1)

where r is the peak response amplitude, r _max is the maximum value of r either in darkness or in the presence of a background illumination, ϕ is the product of the intensity of the flash (in photons μm⁻² s⁻¹) and the flash duration (20 ms), and a is a constant with units of photons⁻¹ μm², whose value depends upon the intensity of the background light. We will let $r_{\max }^{\mathrm{D}}$ and a ^D denote the values of r _max and a for the rods in darkness (that is, without a background light).

Figure 6. Response–intensity curves and flash–response decay waveform.

A, WT. The same cells as in Fig. 1 A–D. Data points give mean response amplitude (with SE) as a function of ϕ, normalized to the maximum response amplitude in dark‐adapted WT rods (15.5 pA). Data have been fitted with Eqn (1) at the following backgrounds (in photons μm⁻² s⁻¹) with the following values of a and of r _max normalized to its value in darkness: dark‐adapted (■), 1.0, 0.0139; 75 (●), 0.82, 0.0106; 201 (▲), 0.73, 0.0075; 764 (▼), 0.41, 0.0043; 2060 (♦), 0.21, 0.0030; and 7460 (◂), 0.02, 8.3 × 10⁻⁴. B, Rv–/–. The same cells as in Fig. 1 E–H. Mean ± SE amplitude normalized to mean maximum amplitude of dark‐adapted Rv–/– rods (12.9 pA) fitted with Eqn (1) for the following backgrounds with the following values of a and of r _max normalized to its value in darkness: dark‐adapted (□), 1.0, 0.011; 75 (◯), 0.85, 0.014; 201 (△), 0.71, 0.0045; 764 (▽), 0.56, 0.0031; 2060 (◇), 0.37, 0.0019; and 7460 (◃), 0.15, 0.0015.

Although the effects of backgrounds on WT and Rv–/– rods were similar, there were some important differences. WT rods saturated at dimmer intensities. This difference is most easily seen for the three brightest backgrounds, where response amplitudes at the same values of ϕ were uniformly larger for Rv–/– rods than for WT rods. To evaluate the effect of this difference, we used the fits to the response–intensity curves to calculate the sensitivities of the rods. We define the flash sensitivity (S _F) as the peak response amplitude r(ϕ) of a small‐amplitude response, divided by ϕ, the number of photons per μm² delivered by the flash. Thus, S _F = r(ϕ)/ϕ for small r(ϕ). Because we define sensitivity for small r(ϕ), we can expand the exponential in Eqn (1) to its first two terms to give:

$$\begin{array}{ccc}\hfill r(\phi )& =& r_{\max }[1-exp(-a\phi )]\hfill \\ & \cong & r_{\max }[1-(1-a\phi )]\hfill \\ & =& r_{\max }a\phi \mathrm{for}\phi <<1/a\hfill \end{array}$$ (2)

The sensitivity is therefore equal to r _max a, where both r _max and a are functions of the intensity of the background light.

We begin with r _max. In Fig. 7 A, we give $r_{\max }/r_{\max }^{\mathrm{D}}$with means and SEs calculated rod by rod for both kinds of rods at each of the background intensities at which measurements were made. The black circles and red squares give values for the WT and Rv–/– rods of Figs 5 and 6. The red triangles show results of a second series of six Rv–/– rods for which measurements were made only at a few brighter background intensities.

Figure 7. Effect of background light on WT and Rv–/– rods.

A, maximum response amplitude divided by value in darkness $(r_{\max }/r_{\max }^{\mathrm{D}})$ from fits to Eqn (1), plotted as a function of I _B. Data points give the mean ± SE for four WT rods (●) and six Rv–/– rods () at dim‐to‐moderate background intensities, plus a further six Rv–/– rods () at bright background intensities. Black curve for WT rods is Eqn (3) with I _½ of 527 photons μm⁻² s⁻¹; red curve for Rv–/– rods is Eqn (4) with I _½ of 1110 photons μm⁻² s⁻¹ and n of 0.76. B, value of constant a from fits to Eqn (1) relative to its value in darkness (a/a ^D), plotted as a function of I _B. Same rods as in (A) with the same symbols. Data have been fitted with Eqn (6) with, for WT rods, values of I ₀ of 105 photons μm⁻² s⁻¹, I _½ of 527 photons μm⁻² s⁻¹ and n = 1 (black curve); and for Rv–/– rods I ₀ of 188 photons μm⁻² s⁻¹, I _½ of 1110 photons μm⁻² s⁻¹ and n = 0.76 (red curve). C, sensitivity relative to its value in darkness (S _F/$S_{\mathrm{F}}^{\mathrm{D}}$) for same rods as in (A) and (B) with the same symbols, plotted as a function of I _B. Data have been fitted with Eqn (5) with values of I ₀ of 105 photons μm⁻² s⁻¹ for WT rods (black curve) and I ₀ of 188 photons μm⁻² s⁻¹ for Rv–/– rods (red curve). Black dashed curve is a straight line between the last two WT data points and was drawn to show the departure of WT rods from the Weber–Fechner curve and their approach to increment saturation. [Color figure can be viewed at http://wileyonlinelibrary.com]

To fit these data, we first observed that $r_{\max }^{\mathrm{D}}=r_{\mathrm{SS}}+r_{\max }$where r _SS is the response of the rod at steady‐state to background light. This is because $r_{\max }^{\mathrm{D}}$ is the value of the circulating current in darkness, and the sum of the steady‐state response to the background (r _SS) and the current remaining in the presence of the background (r _max) must sum to the total circulating current of the rod ($r_{\max }^{\mathrm{D}}$). It therefore follows that $r_{\max }/r_{\max }^{\mathrm{D}}=1-r_{\mathrm{SS}}/r_{\max }^{\mathrm{D}}$. This result is useful because, in data that we do not show, we could fit the normalized response to background light $r_{\mathrm{SS}}/r_{\max }^{\mathrm{D}}$ as a function of the intensity of the background by a Michaelis–Menten curve,

$$\frac{r_{\mathrm{SS}}}{r_{\max }^{\mathrm{D}}}=\frac{I_{\mathrm{B}}}{I_{\mathrm{B}}+I_{1/2}}$$

were I _B is the intensity of steady illumination in units of photons μm⁻² s⁻¹ and I _½ is the intensity required to produce a steady response that is one‐half its maximum value. We could therefore fit values of $r_{\max }/r_{\max }^{\mathrm{D}}$ with:

$$\frac{r_{\max }}{r_{\max }^{\mathrm{D}}}=\frac{I_{1/2}}{I_{\mathrm{B}}+I_{1/2}}$$ (3)

This is the black curve in Fig. 7 A with a best fitting value of I _½ of 527 photons μm⁻² s⁻¹.

The data for the dependence of r _SS on I _B for Rv–/– rods had a different shape and were not well fit by a Michaelis–Menten curve but could be fit with the Hill equation. We could therefore fit the values of $r_{\max }/r_{\max }^{\mathrm{D}}$ for the Rv–/– rods with:

$$\frac{r_{\max }}{r_{\max }^{\mathrm{D}}}=\frac{I_{1/2}^{\mathrm{n}}}{I_{\mathrm{B}}^{\mathrm{n}}+I_{1/2}^{\mathrm{n}}}$$ (4)

which reduces to Eqn (3) when n = 1. We have fitted Eqn (4) to the Rv–/– data in Fig. 7 A and plotted the result as the red curve, with best‐fitting values of I _½ of 1110 photons μm⁻² s⁻¹ and n of 0.76. At a background intensity of ∼80,000 photons μm⁻² s⁻¹, the mean value of $r_{\max }/r_{\max }^{\mathrm{D}}$ was 0.013, and the Rv–/– rods were effectively saturated. These results show that the deletion of recoverin alters the dependence of $r_{\max }/r_{\max }^{\mathrm{D}}$ on background light.

In Fig. 7 B, we show the effect of background intensity on a/a ^D, which is a normalized to its value in darkness. To fit these data, we assumed that the normalized sensitivity of the rod follows the Weber–Fechner equation,

$$\frac{S_{\mathrm{F}}}{S_{\mathrm{F}}^{\mathrm{D}}}=\frac{I_0}{I_0+I_{\mathrm{B}}}$$ (5)

where I _B is the intensity of the background (in photons μm⁻² s⁻¹) and I ₀ is a constant called the ‘dark light’, equal to the background intensity required to reduce sensitivity by one‐half. We felt justified in making this assumption because earlier experiments have shown good agreement with the Weber–Fechner equation for sensitivity in background light for both WT and Rv–/– rods (Mendez et al. 2001; Makino et al. 2004; Chen et al. 2010d). Because, in Fig. 7 A, we have determined the effects of I _B on $r_{\max }/r_{\max }^{\mathrm{D}}$, and because S _F / $S_{\mathrm{F}}^{\mathrm{D}}$= r _max a/$r_{\max }^{\mathrm{D}}{a}^{\mathrm{D}}$ (where $S_{\mathrm{F}}^{\mathrm{D}}=r_{\max }^{\mathrm{D}}{a}^{\mathrm{D}}$ is the sensitivity of the dark‐adapted photoreceptor), we can derive the following equation for a/a ^D:

$$\frac{a}{a^{\mathrm{D}}}=\frac{I_{0(}{I}_{1/2}^{\mathrm{n}}+I_{\mathrm{B}}^{\mathrm{n}})}{I_{1/2}^{\mathrm{n}}\left(I_0+I_{\mathrm{B}}\right)}$$ (6)

This equation was then used to fit the data in Fig. 7 B with I ₀ as a free parameter. For WT rods, with I _½ = 527 photons μm⁻² s⁻¹ and n = 1 from Fig. 7 A, we obtained a best‐fitting value for I ₀ of 105 photons μm⁻² s⁻¹. This curve is shown in Fig. 7 B in black. For Rv–/– rods, with I _½ = 1110 photons μm⁻² s⁻¹, and n = 0.76 from Fig. 7 A, we obtained a best‐fitting value for I ₀ of 188 photons μm⁻² s⁻¹. This is the red curve in Fig. 7 B.

In Fig. 7 C, we have plotted the relative sensitivities $S_{\mathrm{F}}/S_{\mathrm{F}}^{\mathrm{D}}$. Values of sensitivities were calculated rod by rod from $S_{\mathrm{F}}=r_{\max }a$ and then averaged at each of the background intensities. The curves in Fig. 7 C are Eqn (5) with values for I ₀ obtained from the fits to Eqn (6) in Fig. 7 B. That is, for WT rods, we used a value of I ₀ of 105 photons μm⁻² s⁻¹ or ∼50 Rh^* s⁻¹ (assuming a collection area of 0.5) (Field & Rieke, 2002) similar to values obtained by other investigators for WT rods (Mendez et al. 2001; Makino et al. 2004). The value that we used for the Rv–/– rods was 188 photons μm⁻² s⁻¹ (or 95 Rh^* s⁻¹), which is somewhat larger than for the WT rods, although this difference was small and was not statistically significant when WT and Rv–/– rods were fitted cell by cell with Eqn (5).

The results in Fig. 7 C reveal a striking difference between WT and Rv–/– rods. At a background of ∼8000 photons μm⁻² s⁻¹ (or ∼4000 Rh^* s⁻¹), WT rods were very nearly saturated: only two of the four rods gave any response at all. The Rv–/– rods, on the other hand, were not saturated but still gave robust responses (Fig. 5 H). At this background intensity, the mean value of $S_{\mathrm{F}}/S_{\mathrm{F}}^{\mathrm{D}}$ was 0.018 for Rv–/– rods, almost ten times greater than the mean value of 0.0025 for WT rods; this difference was quite significant (t test, P < 0.0028) even though the dark‐adapted sensitivity of WT and Rv–/– rods was almost the same (Figs 5 and 6) (Makino et al. 2004). Brighter backgrounds elicited no further responses from WT rods, but Rv–/– rods continued to respond as the background light was increased and approached saturation only at the brightest background we used, which was ∼80,000 photons μm⁻² s⁻¹ or ten times brighter than the background saturating WT rods. Although $S_{\mathrm{F}}/S_{\mathrm{F}}^{\mathrm{D}}$ at this background intensity still fell on the curve for Eqn (3), only two of the six rods gave any response, and responses were very small.

Discussion

Recoverin was first described as a small molecular‐weight, Ca²⁺‐binding protein that complexes with GRK1 and modulates the rate of phosphorylation and life‐time of Rh^* (Kawamura, 1993). Although recoverin does indeed appear to do this (Makino et al. 2004; Chen et al. 2010a), we have discovered two further effects of recoverin that may be of greater physiological significance. First, recoverin regulates the rate of change of sensitivity after presentation of a background light. When the ambient intensity increases, sensitivity and integration time decrease in WT rods with similar time constants of 150–200 ms (Fig. 2 E and F). This process preserves a longer integration time in dim light to maximize absolute sensitivity but a shorter integration time in brighter light to improve detection of change and motion. Recoverin is part of this mechanism because, in Rv–/– rods, sensitivity declines much more rapidly and integration time is already shorter and not further modulated. Second, recoverin is an essential part of a mechanism controlling the operating range of the rod. During steady light exposure, rod circulating current slowly increases over tens of seconds, gradually augmenting the mean number of channels available to be closed by illumination. This process may keep the spontaneous PDE6 rate relatively high in darkness to improve sensitivity to single photons but reduce the spontaneous rate in background light to extend the operating range of the rod. In Rv–/– rods, this mechanism is deleted, steady‐state currents are already larger and rods saturate at brighter intensities.

We believe that neither of these effects of recoverin are produced by modulation of the rate of Rh^* phosphorylation or the lifetime of Rh^* but, instead, are the result of regulation of phophodiesterase‐6, the phototransduction effector enzyme. The difference in the rate of change of the flash response amplitude between WT and Rv–/– rods shown in Figs 1 and 2 was accompanied by little change in the steady‐state sensitivity S _F at the background intensities used in these experiments (Figs 5, 6, 7), which is a result that precludes any significant effect on the life‐time of Rh^*. Moreover, the time‐dependent changes in sensitivity in WT rods are accompanied by a decrease in response integration time, produced largely by acceleration of response decay and ablated when recoverin is deleted (Fig. 2 F) (Chen et al. 2012). Previous experiments have shown that acceleration of response decay is accompanied by acceleration of the limiting time constant τ_D (Woodruff et al. 2008), which is no longer modulated after recoverin is deleted (Chen et al. 2015). Because, under the conditions of our experiments, the limiting time constant reflects the rate of decay of light‐activated PDE6 (Krispel et al. 2006; Chen et al. 2010a), the change in integration time and response decay after exposure to background light is also probably determined primarily by an alteration in the rate of light‐activated PDE6 decay.

Recoverin is a Ca²⁺‐binding protein at a relatively high concentration in the rods (Kawamura, 2000); deletion of recoverin could alter Ca²⁺ buffering and the time course of changes in Ca²⁺ in the outer segment (Makino et al. 2004). However, the decrease in integration time and acceleration of response decay in Figs 1 and 2 are probably not the result of a simple change in Ca²⁺ buffering because they are accompanied by acceleration of the limiting time constant (Chen et al. 2015) measured so as to minimize any effect of changes in the Ca²⁺ concentration (Krispel et al. 2006). Moreover, the response decay waveform is accelerated by background light even in rods lacking the GCAPs (Chen et al. 2010d).

The second effect that we have discovered is a slow decline of photocurrent in WT rods, which gradually increases the mean number of channels available to be closed by illumination, and which is ablated upon deletion of recoverin (Fig. 3). This effect is also probably not produced by modulation of Rh^* phosphorylation. In Fig. 4 A, we show that the light response to prolonged illumination of rods lacking the GCAPs has a slow decay similar to that in WT rods, accompanied by a pronounced undershoot when the light is turned off (Burns et al. 2002; Chen et al. 2010d). During the undershoot, the circulating current becomes larger than in darkness before illumination. This undershoot cannot result from modulation of Rh^* because a change in the life‐time or activity of Rh^* cannot produce a current at the offset of illumination that is greater than before the light was presented. Moreover, the lifetime of Rh^* is of the order of 40 ms (Burns & Pugh, 2011) and is much less than the time course of decay of the undershoot. The undershoot is also not produced by calmodulin‐dependent channel modulation (Hsu & Molday, 1993) because deletion of the calmodulin‐binding site of the rod cyclic nucleotide‐gated channel does not diminish undershoot amplitude (Chen et al. 2010d).

The results summarized in Fig. 4 B show that both the slow decline of photocurrent and the undershoot are ablated by deletion of recoverin. This result suggests that the two are largely the result of the same phenomenon and, if the undershoot is not produced by modulation of Rh^* phosphorylation or calmodulin‐dependent channel modulation, then neither is the slow photocurrent decline. Both the undershoot and slow decrease in photocurrent are too slow to be caused by a change in Ca²⁺ buffering from deletion of recoverin, whose Ca²⁺ dissociation constants are in the micromolar (Ames et al. 1995). We have proposed that these effects are rather the result of modulation of spontaneous and light‐activated PDE6 (Chen et al. 2010d). Although we cannot exclude other possible mechanisms such as modulation of the rod outer‐segment channels by some mechanism not involving calmodulin binding (Savchenko et al. 2001; McKeown & Kraft, 2014), an effect on PDE6 could explain both the slow decay and undershoot in Fig. 4 A (Chen et al. 2010d). A decrease in the rate of spontaneous activation, perhaps together with a slowly developing increase in the rate of light‐activated PDE6 decay, would cause a decrease in total PDE6 activity in steady background light, slowly increasing the concentration of cGMP and the mean number of channels available to be closed by illumination. When the light is turned off, the slow return of spontaneous PDE6 activity to its normal dark level could cause the current to become transiently greater in darkness than before the light was presented.

Recoverin in the outer segment is known to modulate GRK1, and it is possible that the effects on response decay and circulating current are also mediated by this kinase. Overexpression of GRK1 can accelerate response decay much like deletion of recoverin (Chen et al. 2012; but see also Gross et al. 2012; Chen et al. 2015). There is, however, no biochemical or molecular information that would indicate how recoverin and GRK1 could directly interact with and modulate PDE6.

The effects of recoverin that we have described have important ramifications for the physiology of rod vision. In the dark, a relatively slow decay of light‐activated PDE6 keeps the response integration time relatively long. This long integration time will probably not affect visual sensitivity near absolute threshold at the level of the single‐photon response (Umino et al. 2012) because only the rising phase of the rod response is communicated across the rod/bipolar synapse (Sampath et al. 2005). It would, however, augment sensitivity in somewhat brighter light, when single rods were receiving multiple photons. This is because a longer integration time would increase the time interval during which two or more single‐photon responses could add together. As the ambient light intensity becomes even brighter, the response decay time accelerates and the integration time decreases, we think via recoverin‐dependent modulation of light‐activated PDE6 decay (Chen et al. 2012, 2015). This effect may produce a small decrease in sensitivity, although its more important result is probably to increase temporal resolution in brighter light (Umino et al. 2012).

The slow decay of photocurrent and increase in the mean number of open channels illustrated in Fig. 3 A–C would have quite a different function. We propose that this effect is produced at least in part by a change in the rate of spontaneous activation of PDE6. In darkness, the spontaneous PDE6 activation rate may be kept relatively high by recoverin bound to Ca²⁺ to reduce photoreceptor dark noise. Theoretical calculations indicate that the spontaneous PDE6 rate must be at least of the order of one active PDE6 molecule per second per disk, or ∼1000 per second per mouse rod outer segment (Reingruber et al. 2013). The rate could be higher but, if the rate were much lower, the dark noise would be greater because rare spontaneous PDE6 activations would cause large deflections of current in the presence of a lower steady‐state activity of guanylyl cyclase. The cGMP would slowly accumulate in darkness and, when a PDE6 molecule suddenly became spontaneously active and hydrolysed cGMP, there would be a large excursion in current more slowly returning to baseline. We propose that, in the presence of a background, the rate of spontaneous activation normally decreases to increase the operating range of the rod. This decrease could occur without affecting background noise because the overall rates of PDE6 and guanylyl cyclase increase in the light, and photon noise becomes much more important than the noise produced by spontaneous PDE6 activation. If the dark spontaneous rate in a WT mouse rod were as low as one PDE6 per disk, it is possible that Rv–/– rods would have less dark noise than WT rods. Experiments to test this possibility are best conducted with voltage clamp recording from rods in slices and will be part of a future study.

The results shown in Fig. 7 C and previous results both from Makino et al. (2004) as well as our own laboratory (Chen et al. 2010d) show that, in dim to moderate background light, the sensitivity of the rod is almost the same in WT and Rv–/– rods. Thus, the deletion of recoverin has little effect on the basic mechanism controlling the modulation of sensitivity. Sensitivity regulation is known to occur in part from GCAPs‐dependent modulation of guanylyl cyclase and in part from some other process that is not yet identified (Mendez et al. 2001). We previously proposed that this second process is by direct modulation of spontaneous and light‐activated PDE6 (Chen et al. 2010d). The results in this present study, however, indicate that deletion of recoverin eliminates regulation of response waveform, as well as the slow increase of circulating current in steady illumination, which we attribute to PDE6 modulation; but recoverin deletion has little effect on steady‐state sensitivity except near saturation. These observations leave us with no remaining theory for how sensitivity is regulated over most of the operating range of the rod. Additional biochemical and physiological mechanisms of transduction modulation must be present in the outer segment, but these mechanisms remain to be discovered.

Additional information

Competing interests

The authors declare that they have no competing interests.

Author contributions

AM helped conceive and design the experiments, carried out most of the experimental work, and analysed most of the data. GLF helped conceive and design the experiments, helped analyse the data, and wrote the manuscript. MLW obtained the WT data used in Figs 6–8. Both AM and GLF approve of the final manuscript submitted for publication, are responsible for the integrity of the results, and qualify for authorship. Unfortunately, MLW could not read the final draft of the paper but qualified for authorship.

Funding

This work was supported by National Institutes of Health Grant R01‐EY001844.

Biography

Ala Morshedian has recently acquired his doctorate in the Molecular, Cellular and Integrative Physiology Program at UCLA and is currently working as a Postdoctoral Scholar in Ophthalmology at the Jules Stein Eye Institute in David Geffen School of Medicine.