|Home | About | Journals | Submit | Contact Us | Français|
The rate of synaptic transmission between photoreceptors and bipolar cells has been long known to depend on conditions of ambient illumination. However, the molecular mechanisms that mediate and regulate transmission at this ribbon synapse are poorly understood. We conducted electroretinographic recordings from dark- and light-adapted mice lacking the abundant photoreceptor-specific protein, phosducin, and found that the ON-bipolar cell responses in these animals have a reduced light-sensitivity in the dark-adapted state. Further desensitization of their responses, normally caused by steady background illumination was also diminished compared to wild type animals. This effect was observed in both rod- and cone-driven pathways, with the latter affected to a larger degree. The underlying mechanism is likely to be photoreceptor-specific because phosducin is not expressed in other retina neurons and transgenic expression of phosducin in rods of phosducin knockout mice rescued the rod-specific phenotype. The underlying mechanism functions downstream from the phototransduction cascade, as evident from the sensitivity of phototransduction in phosducin knockout rods being affected to a much lesser degree than b-wave responses. These data indicate that a major regulatory component responsible for setting the sensitivity of signal transmission between photoreceptors and ON-bipolar cells is confined to photoreceptors and that phosducin participates in the underlying molecular mechanism.
The vertebrate retina has a remarkable ability to adapt its sensitivity to dramatic changes in the ambient illumination which occur during the normal diurnal cycle (Shapley, 1984). Dark and light adaptation take place both in the biochemical phototransduction cascade converting the information coded in incoming photons into electrical activity of rods and cones (Pugh et al., 1999; Fain et al., 2001; Burns and Arshavsky, 2005; Luo et al., 2008) and at essentially every stage of the retinal circuitry (Demb, 2008; Dunn and Rieke, 2008). One intriguing aspect of light adaptation is the regulation of light-sensitivity of ON-bipolar cells, the neurons relaying photoreceptor inputs to the inner retina. ON-bipolar cells are very sensitive to light under dark-adapted conditions and can be saturated by light at least 10-fold dimmer than the light required to saturate rods (Dowling, 1967; Green and Powers, 1982; Dunn et al., 2006; Cameron et al., 2008). Sustained illumination causes desensitization of bipolar cells, which results in extension of their operating range. This desensitization is due in part to the adaptation mechanisms operating in the phototransduction cascade in photoreceptor outer segments, which ultimately reduce photoreceptor synaptic output. However, additional adaptation mechanisms have been considered to operate within bipolar cells themselves and/or on the level of the photoreceptor-to-ON-bipolar cell synapses (Thoreson, 2007; Snellman et al., 2008). Yet, the molecular aspects of these mechanisms remain far from understood.
Here we report an unexpected observation that an abundant photoreceptor-specific protein, phosducin, previously thought to regulate phototransduction, participates in regulating the light-sensitivity of synaptic transmission between photoreceptors and ON-bipolar cells. This was established by analyzing flash sensitivity and light adaptation in the retinas of phosducin knockout mice by measuring the electroretinogram (ERG) which monitors the light-evoked activities of retinal cells, including both photoreceptors and bipolar cells in living animals. Phosducin knockout affected two distinct features of rod- and cone-driven ERG b-waves, which reflect the activation of ON-bipolar cells by photoreceptors (Pugh, 1998; Robson and Frishman, 1998). First, the knockout reduced the dark-adapted sensitivity of flash responses. Second, it diminished the degree of b-wave desensitization observed under sustained background illumination. Both effects were more pronounced for cone- than for rod-driven ERG b-waves. In contrast, a-waves, which reflect the activation of phototransduction in photoreceptor outer segments, were essentially unchanged by phosducin knockout. Combined with our recent study indicating that phosducin knockout does not affect any feature of light adaptation of rod outer segments (Krispel et al., 2007), these data argue that phosducin has minimal impact on phototransduction and instead is involved, directly or indirectly, in regulating the light sensitivity of synaptic transmission between photoreceptors and ON-bipolar cells. The rescue of the rod-specific ERG b-wave phenotype by transgenic restoration of phosducin in rods of phosducin knockout mice indicates that the underlying mechanism is confined to photoreceptors. Therefore phosducin is likely to be a part of a novel presynaptic mechanism responsible for regulating the information flow from rods and cones to ON-bipolar cells.
Mice were handled following an approved protocol from the Institutional Animal Care and Use Committees of Duke University and reared under the normal diurnal cycle. C57/Bl6 WT mice were purchased from Charles River. Phosducin knockout (Pd−/−) mice were described in (Sokolov et al., 2004). Transducin γ-subunit heterozygous (Gγ1+/−) mice were described in (Lobanova et al., 2008) . Transducin α-subunit knockout (Gαt−/−) mice were provided by J. Lem (Tufts University, Medford, MA). The Gαt−/−Pd−/− double knockout mice were produced by crossing these two lines.
We generated two transgenic mouse lines: one expressing mouse phosducin (tgPd+ mice) and another expressing the mutant phosducin lacking all 5 previously detected protein kinase A and calmodulin-dependent kinase II (PKA/CaMKII) phosphorylation sites (Thulin et al., 2001) (noPoP+ mice). The pCMV-SPORT 6 plasmid carrying the WT mouse phosducin sequence was purchased from ATCC (Manassas, VA). To generate the noPoP+ mutant, we introduced five alanine substitutions into this cDNA (S6A, S36A, S54A, S71A and S104A) using the PCR-based site directed mutagenesis (Quickchange Kit, Stratagene). Both constructs were amplified by PCR and extended by C-terminal c-myc epitope coding sequences. The PCR products were cloned at the SalI/BamHI restriction sites into a rod-specific expression vector carrying the mouse opsin promoter and a polyadenylation signal of the mouse protamine gene (Lem et al., 1991). The constructs were confirmed by sequencing. Linear constructs were excised by the KpnI/NotI digest, purified using the Quiagen Gel Extraction Kit and injected into pronuclei of mouse oocytes of superovulated B6SJLF1/J females. Transgene insertion was determined by the PCR analysis of tail DNA. To establish the final mouse lines, the founders were crossed with Pd−/− mice for three generations. For experiments, we used mice of the F3 generation on the Pd−/− background (tgPd+Pd−/− and noPoP+Pd−/−), as well as non-expressing littermate controls also on the Pd−/− background (called tgPd–Pd−/− for both lines).
To confirm that the noPoP phosducin mutant folds as a functionally active protein, we cloned the DNA above into the pET101/D-TOPO expression vector and produced His-tagged recombinant noPoP phosducin in E. coli based on published methods (Bluml et al., 1997). This protein was able to efficiently bind transducin Gβγ-subunits in pull-down binding assays, which indicated that mutations did not affect its overall integrity.
Determination of rhodopsin concentration in retina lysates and the expression level of phosducin in transgenic mice were performed as described earlier (Lobanova et al., 2008). Samples of retina lysates from tgPd+Pd−/− or noPoP+Pd−/− mice containing known amounts of rhodopsin were subjected to SDS-PAGE alongside with the samples from Pd−/− mice containing the same amount of rhodopsin and mixed with various amounts of the recombinant phosducin standard. Western blots were probed using the sheep anti-phosducin antibody described in (Sokolov et al., 2004) and the bands were visualized using fluorescent secondary anti-sheep antibodies (Invitrogen). The fluorescence of the individual bands on the blot was determined using the Odyssey Infrared Imaging System (LI-COR Biosciences). Phosducin content in retina lysates was determined based on the calibration curves obtained with phosducin standards and expressed as a molar ratio with rhodopsin.
To produce the phosducin standard, we obtained the pET15b plasmid (Invitrogen) carrying the coding sequence of N-terminally His6-tagged rat phosducin and the C-terminal c-myc epitope from B. M. Willardson (Brigham Young University, Provo, UT). To precisely match the electrophoretic mobility of the standard with that of the myc-tagged phosducin expressed in the transgenic mice, we subcloned this construct without His6-tag into the pTYB1 vector (NEB) at the NdeI/SapI restriction sites, which allowed us to use the intein-mediated purification system (IMPACT-CN, New England Biolabs) (Xu and Evans, 2001). The resulting construct encoded the C-terminally myc-tagged phosducin, N-terminally fused to intein. This construct was expressed in E. coli, bound to chitin beads, washed and cleaved by β-mercaptoethanol to remove the intein tag (see (Herrmann et al., 2004) for experimental details). The resulting recombinant phosducin was at least >95% pure, as estimated by SDS-PAGE.
For detection of the cone transducin alpha subunit (cGαt), we generated rabbit antiserum against the cGαt peptide CGIDYAEPSCADAGRQLNNL. The peptide was coupled to the maleinimide-activated keyhole limpet hemocyanin (Pierce Biotechnology, Rockford, IL) and the conjugate was used to immunize rabbits. Antibodies were purified on a Sulfolink Plus column (Pierce) containing covalently attached peptide. Antibodies against mouse cone arrestin, Arr4, were a gift from C. M. Craft (USC, Los Angeles, CA). For comparison of protein expression levels, retinas were dissolved in 2% SDS, Tris·HCl (pH 6.8) upon sonication. The total protein content of samples was determined using the Dc Protein Assay (Biorad). Samples containing equal amounts of total protein were subjected to Western blotting as described (Lobanova et al., 2008), and probed with either of the above antibodies or mouse anti-actin antibody (sc-47778, Santa Cruz). Fluorescent secondary antibodies against mouse and rabbit were from (Invitrogen). The fluorescence of the individual bands on the blot was determined using the Odyssey Infrared Imaging System (LI-COR Biosciences) and their densities were quantified using the Odyssey software.
Transmission electron microscopy in 0.15 μm-thick retina cross-sections was performed as described by (Petters et al., 1997). Immunostaining of the myc-tagged phosducin and its mutant lacking phosphorylation sites was performed as described in (Lobanova et al., 2007) using the mouse anti-c-myc primary antibody (1:500, Santa Cruz) and goat anti-mouse Alexa Fluor 488 secondary antibody (1:500, Invitrogen). Hoechst dye was used for nuclear staining (1:1000, Invitrogen). For immunofluorescence localization of cone transducin, eyes were enucleated and fixed for 1 hour with 4% formaldehyde in PBS buffer, pH 7.5 (Fisher Scientific, Hampton, NH). The anterior portions of the fixed eyes were removed, and the eyecups were further fixed overnight at 4°C. Eyes were rinsed three times in PBS and embedded in 4 % low melt agarose (Gibco). Retina cross-sections (100 μm) from the central retina were collected with a vibratome (Microm HM650 V). Floating sections were further incubated in 24-well plates placed on an orbital shaker (Stovall, Greensboro, NC) to ensure complete penetration of antibodies. Sections were incubated during 4 hours with blocking solution containing 3 % donkey serum and 0.1% Triton X-100 in PBS buffer. For detection of cone transducin, floating sections were incubated overnight with a mixture of the rabbit anti-cGαt antibody and anti-peanut agglutinin lectin (PNA) antibody tagged with Alexa-488 fluorophore (Molecular Probes). Sections were washed three times with PBS, and incubated for 2 hours with goat anti-rabbit Alexa Fluor 594 secondary antibody (1:500, Invitrogen). Sections were rinsed three times in PBS, mounted with Fluoromount G (Electron Microscopy Sciences, Hatfield, PA) under glass coverslips, and visualized using a Nikon (Tokyo, Japan) Eclipse 90i confocal microscope.
Suction electrode recordings of rod outer segment currents in Gγ1+/− mice were performed and analyzed as described in (Lobanova et al., 2008). Briefly, mice that had been dark-adapted overnight were killed, and the retinas dissected under infrared light. Small pieces of retina were placed in a recording chamber and perfused with oxygenated, bicarbonate-buffered Locke’s solution at 34-37°C (pH 7.4). Brief flashes (10 ms, 500 nm) of calibrated stimulus strength were delivered to a single rod whose outer segment was drawn into a suction electrode filled with HEPES-based buffer and connected to a current-to-voltage amplifier (Axon Instruments). Signals were filtered (8-pole Bessel with 30 Hz corner frequency; Frequency Devices) and digitized at 200 Hz acquisition frequency using IgorChart (Wavemetrics).
ERGs were recorded using the Espion E2 system (Diagnosys LLC, Littleton, MA) according to published protocols (Saszik et al., 2002). 8-12 week old mice were dark-adapted overnight and prepared for recordings in darkness using infrared goggles. Mice were anesthetized by intraperitoneal injection of a ketamine/xylazine mixture (75/10 mg/kg). Pupils were dilated with 1% cyclopentolate-HCl, 2.5% phenylephrine (~2.8 mm in diameter). A drop of the Gonak solution (Akorn, Buffalo Grove, IL) was placed on the cornea. Recordings were performed simultaneously from both eyes with silver loop electrodes supplemented with contact lenses to keep the eyes immersed in Gonak solution. This prevented eyes from drying and cataract development during recording periods of up to 60 min. The reference electrode was a toothless alligator clip wetted with Gonak and attached to the mouse cheek. Mouse body temperature was maintained at 37°C using a Deltaphase Isothermal Pad (Braintree Scientific, Braintree, MA). ERG signals were sampled at 1 kHz and recorded with 0.15 Hz low frequency and 500 Hz high frequency cut-offs.
The light stimuli consisted of brief full field white flashes delivered by a set of LEDs (duration <4 ms) for flash strengths ≤ 1 cd·s/m2 or a Xenon light source for flashes ≥ 1 cd·s/m2 (flash duration 5 μs to 2 ms). All flashes were produced by a Ganzfeld stimulator (ColorDome, Diagnosys LLC), either in darkness or on steady blue background light (λmax = 465 nm). Internal calibrations of LED and Xenon light sources were verified using a light meter (International Light IL 1700, USA) by applying a train of flashes and measuring the mean luminance. The rate of rhodopsin isomerization evoked by steady background light was determined as reported before (Lobanova et al., 2007). Briefly, mice were treated as they were for ERG recordings and exposed to background light (10 min) of various intensities. The amount of rhodopsin bleached over the illumination period was determined by removing the retinas and measuring their rhodopsin contents before and after rhodopsin regeneration with 11-cis-retinal (Sokolov et al., 2002). The rate of rhodopsin isomerization was calculated by multiplying the fraction of photoactivated rhodopsin by the total number of rhodopsin molecules per mouse rod (7·107 from (Lyubarsky et al., 2004)) and dividing this product by the duration of background illumination. For our experimental setup, we found that blue background light of 1 cd/m2 activated ~800 rhodopsins/rod/sec.
Intravitreal injections of 2-amino-4-phosphonobutyric acid (APB, Sigma) were performed as described before (Shirato et al., 2008) to block ON-bipolar cell light responses (Slaughter and Miller, 1981). A solution of APB in sterile saline was prepared freshly and adjusted to neutral pH using NaOH. A volume of ~1.5 μl was injected with a 30-gauge needle into the vitreous just behind the limbus 1 h before recordings, yielding a final vitreal APB concentration of ~2 mM based on a vitreal volume of 20 μl.
ERG responses in mice, either dark-adapted or adapted to blue background light (0.005, 0.1, 1 cd/m2), were evoked by a series of flashes ranging from 0.0001 cd·s/m2 to 1000 cd·s/m2. For dim flashes up to 0.1 cd·s/m2, responses of 10 trials were averaged. For 1 cd·s/m2 flash responses, 3 trials were averaged. For brighter stimuli, responses to single flashes were recorded without averaging. Intervals between individual flashes were chosen to ensure that the retina recovered completely from each flash, i.e. no indications of flash-induced reduction of response amplitudes, enlargement of oscillatory potentials or shortening of implicit times were observed. Based on these criteria, the inter-flash intervals were 5 sec for the 0.0001 – 0.1 cd·s/m2 flashes and 15 sec for the 0.5 cd·s/m2 and 1 cd·s/m2 flashes. Recovery times were 30 sec after the 1 cd·s/m2 flash, 60 sec after the 5 cd·s/m2 and 10 cd·s/m2 flashes, and 120 sec after 100 cd·s/m2 and 180 sec after the 1000 cd·s/m2 flash. For each mouse, the dark-adapted series were recorded first, followed by the series of increasing intensity of background illumination. Complete adaptation to background light was achieved by a 2 min exposure to steady background before starting the flash series (longer adaptation times did not change the results). To account for any potential changes in response amplitudes which may occur during a recording session due to electrode shifts or condition of the subject, mice were dark adapted after the brightest background and a subset of test flashes was applied to verify that the dark-adapted response amplitudes were not significantly changed during the course of experiment. In rare cases when more than 15% changes were observed, the entire dataset was dismissed. To isolate cone-driven ERG responses, we modified this protocol as follows. Flash strengths for the dark-adapted series were 0.001, 0.005 and 0.01 cd·s/m2 (5 trials averaged with 15 s trial interval time), 0.1, 1 cd·s/m2 (3 trials averaged with 15 s trial interval time), and 5, 50, 100, 500 and 1000 cd·s/m2 (single flash responses). For the light-adapted series, we used background light of 10, 50, 100 and 500 cd/m2 and flashes of 0.1, 1, 5, 50, 100, 500 and 1000 cd·s/m2 (up to 10 trials averaged with interval times of 10 s). The interval time of 10 s was sufficient to ensure complete recovery between the flashes due to the fast cone recovery rate.
To determine a-waves and b-wave amplitudes, ERG data were processed and analyzed using MATLAB software (Version R2007a, Mathworks Inc.). Rod-driven a-waves were measured from the baseline recorded prior to the flash to the response at 8 ms. For b-waves, the oscillatory potentials were removed from the signals by 55 Hz FFT low-pass frequency filtering since oscillatory potentials obscure the amplitude and position of the b-wave peak, especially under light-adapted condition (Lei et al., 2006). The b-wave amplitude was calculated from the bottom of the a-wave response to the peak of the b-wave peak.
The data points from the a-wave and b-wave stimulus-response curves were fitted by Equation 1 using the least-square fitting procedure (Scientist Software, MicroMath, Salt Lake City, UT):
The first term of this equation describes rod-mediated responses (index 1), the second term accounts primarily for responses that were cone-mediated (generally observed at the flash intensities ≥ 1 cd·s/m2 for dark-adapted mice; index 2). Rmax,1 and Rmax,2 are maximal response amplitudes, , and I0.5,1 and I0.5,2 are the half-saturating flash intensities. The use of a simple hyperbolic for each curve was adopted with the assumption that stimulus response functions of retina cells increase in proportion to stimulus strength and then saturate in a characteristic way that is adequately described by this function (Robson and Frishman, 1995).
where R/Rdark is the response amplitude normalized to the dark-adapted value, Ib is the background light intensity, Ihalf is the background luminance causing half-maximal reduction of normalized response amplitudes, and n the Hill coefficient.
To characterize the role of phosducin in dark- and light-adaptation of photoreceptors and ON-bipolar cells we recorded ERGs from Pd−/− and WT mice. ERGs are massed field potentials which reflect the combined light-evoked activity of many retinal cells. The negative ERG deflections immediately following the flash stimulus, called a-waves, originate primarily from the suppression of the circulating dark current in rod and cone outer segments. The subsequent positive deflections, called b-waves, originate mainly from the light-induced depolarizing currents in ON-bipolar cells (Pugh, 1998; Robson and Frishman, 1998).
Representative ERG recordings from dark- and light-adapted Pd−/− and WT mice are shown in Fig. 1A. Mice were subjected to a series of flashes, with the dimmest photoexciting only ~0.02 rhodopsin molecules (R*) in each rod and the brightest exciting ~2·105 R*. Recordings were conducted in the dark and in the presence of three background light intensities activating rhodopsin at the rates of ~4, 80 and 800 R*/rod/sec. These measurements revealed that phosducin knockout resulted in reduced amplitudes of dark adapted b-waves evoked by dim flashes. However, the presence of background light essentially reversed this effect, so that the b-wave amplitudes in Pd−/− mice were larger than in WT mice under most conditions tested. To confirm that b-waves indeed served as adequate readouts of ON-bipolar cell activity in our experiments, we blocked ON-bipolar cell responses with DL-2-amino-4-phosphonobutyric acid (APB), a potent agonist of the mGluR6 receptor localized specifically at the dendritic tips of both rod and cone ON-bipolar cells (Nomura et al., 1994; Masu et al., 1995; Vardi and Morigiwa, 1997). The data shown in Fig. 1B confirmed that APB completely eliminated b-waves in both mouse types under all illumination conditions that we used.
ERG data averaged from 17 eyes of Pd−/− and WT animals are presented in Fig. 2A and 2B. Phosducin knockout did not affect the absolute values of maximal dark-adapted b-wave amplitudes, Rmax, dark (809±56 μV and 785±32 μV in WT and Pd−/− mice, respectively; mean ± SEM). To characterize the effects of phosducin knockout quantitatively, we normalized b-wave amplitudes (R) to the corresponding Rmax, dark values, plotted the resulting ratios as functions of flash strength and fitted the data to Equation 1 (see Table 1 for the complete set of parameters; note that for Pd−/− mice, an additional set of recordings was performed in the presence of 0.02 cd/m2 background light). The resulting plots were biphasic, with the first phase reflecting the responses of rod ON-bipolar cells (‘the rod-component’) and the second phase reflecting mainly the responses of cone ON-bipolar cells (‘the cone-component’) (Toda et al., 1999; Weymouth and Vingrys, 2008). In some plots (Fig. 2A and 2C), the amplitude of the response was elevated slightly for the weakest stimulus, probably due to a small contribution from the positive scotopic threshold response (pSTR) (Saszik et al., 2002). The plots also do not take into account the small contribution of rod signals that enter into cone pathways via the electrical synapse between rods and cones when rods signals are nearing saturation (Abd-El-Barr et al., 2009).
To better visualize the effects of phosducin knockout on rod and cone components of b-waves, we plotted average responses for each component separately in Fig. 2C and 2D. The analysis of the rod-specific b-wave component in Fig. 2C revealed two effects of the knockout. First, the light sensitivity of dark-adapted b-waves (defined as the flash strength evoking half-saturating response amplitude) was reduced by ~2.7-fold (p=0.0003 based on the unpaired, two-sample t-test) and the b-wave desensitization caused by background light was less pronounced (compare solid lines to dashed lines in the lower panel of Fig. 2C).
Normally, b-wave desensitization has two distinct features: an increase in half-saturating light (“stimulus scaling”), and a reduction in the maximal response amplitude (“response compression”) (Dowling, 1967; Green, 1986). To correctly analyze stimulus scaling in Pd−/− mice, we had to account for the fact that the flash sensitivity of their b-waves was already reduced in the dark. Therefore, we normalized the values of half-saturating flash intensities in the presence of background illumination to the corresponding values obtained in the dark (I0.5/I0.5, dark) and compared the resulting ratios to the corresponding parameters obtained from WT mice. Assuming that the mechanisms which cause sensitivity reduction of flash responses in the dark would similarly affect the desensitizing ability of the background light, we also corrected the nominal value of background luminance applied to Pd−/− mice for the same 2.7-fold sensitivity reduction factor. For example, light of 0.005 cd/m2 was calculated to have the effective equivalent of 0.0019 cd/m2. The resulting plots for WT and Pd−/− mice followed similar trends (Fig. 3A), which indicates that stimulus scaling was not significantly affected by phosducin knockout. On the contrary, b-wave response compression in Pd−/− animals was reduced. This is evident from plotting the normalized values of saturating b-wave amplitudes (Rmax/Rmax rod, dark) as a function of corrected background luminance (Fig. 3B).
To formally describe the degree of b-wave desensitization caused by phosducin knockout, we determined the Rmax/I0.5 ratio, according to (Saszik et al., 2002). We calculated this ratio at each condition of background illumination and normalized it to the dark-adapted value in order to account for the difference in the dark sensitivities of WT and Pd−/− mice. The resulting values were plotted as a function of background luminance in Fig. 3C, indicating that the 50% reduction of b-wave Rmax/I0.5 ratio in Pd−/− mice required ~4-fold higher background luminance than in WT animals.
The analysis of cone b-wave components (Fig. 2D) revealed that their sensitivity was also reduced in Pd−/− mice. However, a quantitative analysis of this effect required applying rod suppressing background light or employing animal models in which ERG responses lack any rod contributions. These data are presented below in Figs. Figs.9,9, ,1010 and and1111.
One potential explanation for the decreased sensitivity of rod ERG b-waves in Pd−/− mice is that their rods contain ~20-30% less transducin than WT rods (Krispel et al., 2007). Single cell recordings from rods of Pd−/− mice conducted in (Krispel et al., 2007) indicated that this reduction in transducin was reflected by ~1.5-fold reduction in dark-adapted response sensitivity. However, our current ERG data indicate that rod-driven b-waves in Pd−/− mice are ~2.7-fold less sensitive than in WT mice under dark-adapted conditions. Therefore, the lack of phosducin is likely to affect the synaptic transmission between rods and rod bipolar cells in addition to any effects related to a reduced transducin expression.
To obtain further evidence that the reduction in transducin content alone cannot explain most phenomena observed in Pd−/− mice, we analyzed the parameters of ERG b-waves in another mouse model, the transducin γ-subunit heterozygous mouse (Gγ1+/−), in which rods are characterized by a similar ~30-40% reduction in their transducin content (Lobanova et al., 2008). Suction electrode recordings obtained with rods of Gγ1+/− mice (Fig. 4; see Table 2 for complete set of parameters) indicated that the flash sensitivity of their dark-adapted responses was reduced by ~1.4-fold, very similar to the corresponding reduction in Pd−/− rods (Krispel et al., 2007). ERG recordings from dark-adapted Gγ1+/− mice (Fig. 5) revealed ~1.7-fold reduction in b-wave flash sensitivity (p=0.044), which is more in line with the reduction in transducin content than the effect caused by phosducin knockout. The unpaired, two-sample t-test revealed that the remaining difference between the reductions in b-wave sensitivities in Pd−/− and Gγ1+/− mice was statistically significant (p=0.017). Furthermore, the reduction in transducin content in Gγ1+/− mice did not cause any notable abnormality in response desensitization caused by background light after correcting the data for the reduction in dark-adapted b-wave sensitivity (Fig. 3, violet symbols; Table 3). This is in striking contrast to the 4-fold Rmax/I0.5 ratio shift observed in Pd−/− mice (Fig. 3C, red symbols). These data demonstrate that a part of the reduction in dark-adapted b-wave flash sensitivity and the entire reduction of light adaptation caused by the phosducin knockout do not originate from impaired phototransduction.
Unlike b-waves, the absolute dark-adapted sensitivity and adaptation of a-waves were not significantly affected by phosducin knockout (Fig. 2B). Notably, the ~1.5-fold light-sensitivity reduction in Pd−/− mice documented by single cell recordings (see above; (Krispel et al., 2007)) was not paralleled by any reliable reduction in the sensitivity of a-waves. The same was true for Gγ1+/− mice (Table 3). Perhaps this reflects the fact that b-waves are better suited for measuring rod-driven responses in dim light at which the suction electrode recordings are the most sensitive to reflect small changes in phototransduction.
An alternative ERG protocol for studying light adaptation in the retina is based on applying a single saturating flash presented in the dark and at several background light intensities (Cameron et al., 2006). We used this protocol to analyze light adaptation in WT and Pd−/− mice at the flashes of 1 cd·s/m2 (Fig. 6A). While the amplitudes of the dark-adapted b-waves were identical, the presence of background light caused much smaller b-wave suppression in Pd−/− mice than in WT controls. The analysis of normalized b-wave amplitudes (R/Rdark) plotted as a function of background luminance and fitted by Equation 2 demonstrated that the background luminance causing a half-maximal reduction of response amplitude in Pd−/− was ~11-fold higher than in WT animals (0.086 vs. 0.008 cd/m2; Fig. 6B; Table 4). This large shift can be explained by a combination of the 2.7-fold reduction in the effective background luminance, given the dark-adapted sensitivity (Fig. 2C) and the ~4-fold difference in Rmax/I0.5 ratio reduction (Fig. 3C). Another feature of Pd−/− mice was that the plot in Fig. 6B had a steeper slope than the plot for WT mice, as reflected by a larger Hill coefficient (n = 0.61 vs. n = 0.38; Table 4). As a consequence, the range of background luminance causing b-wave desensitization in Pd−/− mice was both shifted and reduced. Overall, the single saturating flash protocol is a rapid and efficient approach to describe the major features of the phosducin knockout phenotype, which we used in several control experiments outlined in this and the next section.
To rule out any possible contributions to ERG recordings arising from the strain differences between WT (C57Bl/6J) and Pd−/− mice (mixed C57Bl/6J and 129sv backgrounds), we bred heterozygous mice to obtain Pd−/− and Pd+/+ littermates and repeated ERG experiments using the single saturating flash protocol. The results, plotted as open symbols in Fig. 6B, C and fitted by Equation 2 (Table 4), were very similar to those obtained from non-littermate animals, indicating that the strain difference is not a significant factor in this study. Finally, we should note that the analysis of a-wave amplitudes using this protocol did not reveal a reliable difference between Pd−/− and WT mice (Fig. 6C). This is another indication that phosducin knockout does not significantly affect phototransduction, at least within the sensitivity limits which a-waves may provide (see above).
All published immunolocalization studies concluded that phosducin’s localization in the retina is restricted to rods and cones (Lee et al., 1988; von Schantz et al., 1994; Sokolov et al., 2004; Chen et al., 2005), which argues that the ERG b-wave phenotype of the Pd−/− mice has a pre-synaptic origin. However, the spatial resolution of published immunogold data is not sufficient to completely exclude that a portion of phosducin might also be expressed in dendritic tips of bipolar or horizontal cells (Chen et al., 2005), which calls for additional experiments eliminating a possibility that phosducin may act post-synaptically.
We addressed this concern by transgenically expressing a myc-tagged phosducin in Pd−/− mice under the control of the rod-specific rhodopsin promoter. The localization pattern of this construct in retina cross-sections visualized by anti-myc antibodies (Fig. 7A) was similar to that of endogenous phosducin previously described in wild type retinas (Sokolov et al., 2004), with most immunostaining observed in inner segments and synaptic terminals. Quantification of myc-tagged phosducin expressed in this mouse, conducted by quantitative Western blotting using recombinant myc-tagged phosducin standards (Fig. 7B), revealed that it was expressed at ~1:50±12 molar ratio to rhodopsin (n=6), which corresponds to ~25% of its normal amount in mammalian rods (Sokolov et al., 2004). Testing these mice with the single flash ERG protocol indicated that this amount was sufficient to provide a nearly complete rescue of the rod b-wave phenotype observed in Pd−/− mice (Fig. 7C, D; Table 4). These data provide further confirmation for the pre-synaptic origin of the phosducin knockout phenotype.
Phosducin is a phosphoprotein, and it has been suggested that its function in rods is regulated by phosphorylation with PKA and/or CaMKII kinases (Lee et al., 1984; Bauer et al., 1992; Yoshida et al., 1994; Thulin et al., 2001; Lee et al., 2004). Therefore, we repeated the transgenic experiment using a construct encoding the myc-tagged phosducin mutant lacking all known phosphorylation sites (we called it noPoP for “no phosphorylation of phosducin”). Immunostaining revealed a pattern similar to that of WT myc-tagged phosducin, although the staining intensity was higher (Fig. 7A). Accordingly, quantitative Western blotting indicated that the expression level of this mutant was higher and identical to that of normal phosducin in WT rods (~1:13±3 molar ratio to rhodopsin, n=6; Fig. 7B). The ERG analysis of these animals revealed that noPoP phosducin rescued the phosducin knockout phenotype as efficiently as the wild type transgene (Fig. 7C, D; Table 4), indicating that phosducin phosphorylation is unlikely to play a role in the underlying mechanism.
We next addressed whether the b-wave phenotype of phosducin knockout could be associated with any morphological abnormalities of the rod-to-rod bipolar cell synapse. Transmission electron microscopic analysis of three WT and five Pd−/− retinas (at least 15 synapses per retina) did not reveal any notable difference. This was judged from the overall dimension of the synaptic terminal, the appearance and length of synaptic ribbons, the invaginations of horizontal cells and dendritic tips of rod bipolar cells, and the presence of numerous synaptic vesicles. Three randomly chosen images of synapses from each retina type are shown in Fig. 8.
We next studied the effects of phosducin knockout on cone pathway function. A common approach to isolating cone-driven ERG components is to subject mice to constant bright light sufficient to completely suppress, any rod inputs (Lyubarsky et al., 1999; Weymouth and Vingrys, 2008). Representative responses from Pd−/− and WT mice in the presence of background luminance of 50, 100 and 500 cd/m2 (each completely suppressing rod ERGs) are shown in Fig. 9A and the data averaged from 5 eyes of each type are presented in Fig. 9B,C. The latter panels also contain data for dark-adapted cone b-wave responses recorded from the same animals prior to their light-adaptation and calculated by subtracting the rod b-wave component from the biphasic stimulus-response curve (see description of Fig. 2 for details of this analysis). Overall, the responses of Pd−/− mice appeared as if they were already significantly desensitized in the dark. Both the amplitudes and the light sensitivities of dark-adapted b-waves were reduced compared to WT animals and background light caused only a small additional desensitization. This was very different from desensitization of WT responses, in which background light reduced the maximal b-wave amplitude and increased the half-saturating flash strength (see Table 5 for a complete set of parameters).
To investigate this phenomenon more quantitatively, we switched to using transducin α-subunit knockout mice (Gαt−/−), in which rods are completely unresponsive to light and ERGs are driven exclusively by cones (Calvert et al., 2000). This enabled us to measure dark-adapted cone b-wave responses directly and to include measurements at background light intensities insufficient to completely suppress rods in WT animals. The Gαt−/−Pd−/− double knockout mice have been previously shown to have normal retina morphology (Lobanova et al., 2008), just as the single Gαt−/− knockout mice (Calvert et al., 2000) which served as controls. Both mouse types were tested with the complete ERG protocol similar to that utilized in Fig. 2, but using brighter background lights (10 cd/m2, 50 cd/m2, 100 cd/m2 and 500 cd/m2).
A representative set of ERG responses (Fig. 10A) illustrates that b-waves recorded from Gαt−/−Pd−/− mice had reduced dark-adapted flash sensitivity and resisted further desensitization by background light which was well-displayed in control Gαt−/− animals. As in the case of rod b-waves (Fig. 1B), we confirmed that b-waves recorded in this experiment originated primarily from cone ON-bipolar cells by repeating ERG recordings after intravitreal injections of APB, which efficiently abolished b-wave responses under all conditions of illumination in both animal types (Fig. 10B). B-wave amplitudes averaged from 11 Gαt−/−Pd−/− and Gαt−/− eyes were plotted as functions of flash strength (Fig. 11A, B; see Table 5 for a complete set of individual parameters), revealing that the most striking effect of the phosducin knockout is a ~20-fold reduction in the sensitivity of dark-adapted cone b-wave responses. Background light caused further b-wave desensitization in Gαt−/−Pd−/− mice, reaching approximately the same light sensitivity level as Gαt−/− controls for the brightest background applied. Therefore, cone-driven b-waves of Gαt−/−Pd−/− mice behaved approximately normally in the strongly light-adapted state, but lacked the ability to regain sensitivity in the dark or at lower levels of background illumination.
To determine whether the desensitization observed in each animal type followed Weber’s law, we plotted both b-wave half saturation and the inverse of the Rmax/I0.5 ratio as a function of background luminance (Figs. 11C,D). The data for the Gαt−/− mice followed Weber’s law reasonably well (apart perhaps from the Gαt−/− responses recorded at the brightest background). However, Gαt−/−Pd−/− mice displayed not only a major reduction in dark-adapted sensitivity but also a large change in the background luminance that desensitized the b-wave, indicative of a major impairment of light adaptation in cone-driven b-wave responses.
Overall, this phenotype resembles the behavior of rod-driven b-waves, although the extent of the dark-adaptation impairment was significantly larger. We also noted that cone-driven b-waves recorded from Pd−/− mice were characterized by accelerated recovery (particularly in the case of and Gαt−/−Pd−/− mice), although we can not offer a mechanistic explanation to this phenomenon based on available information.
Overall, the analyses of cone-driven b-waves by the two approaches illustrated in Figs. Figs.99 and and1111 yielded similar results. Interestingly, the maximal dark-adapted cone-driven b-wave amplitudes were nearly identical in Pd−/−, Gαt−/−Pd−/− and Gαt−/− mice and ~100 μV smaller than the calculated value for WT animals from Fig. 9B. Given that the values of dark-adapted b-wave amplitudes in Fig. 9 were calculated whereas in Fig. 11 directly measured, we consider the latter more reliable. Therefore, we conclude that the maximal amplitude of dark-adapted cone b-waves is not affected by the phosducin knockout, just as seen for the corresponding parameter in rods.
In the last set of experiments we investigated whether the large rod/cone difference in the loss of dark-adapted b-wave sensitivity may be explained by a larger reduction in transducin expression level in cones vs. rods of Pd−/− mice. This was accomplished by analyzing both intracellular localization and the expression level of cone transducin in retinas of Pd−/− mice. Immunostaining of cone transducin α-subunit in Pd−/− mice indicated that it is normally localized to cone outer segments (Fig. 12A). Quantitative Western blotting revealed that it its expression level is actually reduced by ~43% compared to WT retinas (Fig. 12B; 44±1% reduction in Pd−/− compared to WT mice, n = 2; 42±1% reduction in Gαt−/−Pd−/− compared to Gαt−/− mice, n = 2). This reduction is somewhat more significant than the ~20% reduction in the content of rod transducin in these animals (Krispel et al., 2007). However, the hyperbolic relation between the outer segment content of transducin and the rate of its activation (Heck and Hofmann, 2001; reviewed in Arshavsky et al., 2002), suggests that this 43% reduction could not cause more than a similar reduction in dark-adapted photoreceptor sensitivity, which is insufficient to explain the 20-fold effect on b-wave sensitivity that we observed. Similarly, the large reduction of b-wave desensitization in Gαt−/−Pd−/− mice as reflected in the Weber plots in Figs. 11C,D is more significant than the reduction in the transducin concentration, suggesting that the underlying mechanism resides downstream from cone outer segments. Finally, the content of the cone-specific isoform of arrestin, Arr4, was normal in Pd−/− mice (Fig. 12B), indicating that the overall health of cones was not affected by the phosducin knockout.
The modulation of bipolar cell sensitivity by light-adaptation was discovered from the analysis of ERG b-waves over 4 decades ago (Dowling, 1967), yet it remains very poorly understood. Light adaptation at the level of the bipolar cells occurs at background luminance far dimmer than in photoreceptors and are reflected by complex biphasic changes in the gain of bipolar cell response (Dunn et al., 2006). These changes likely occur both at the level of the bipolar cells themselves and/or at the level of the photoreceptor-to-bipolar synapse (Thoreson, 2007; Snellman et al., 2008). At higher intensities, light-induced desensitization follows adaptive changes in the phototransduction cascade in photoreceptor outer segments (Dunn et al., 2006). Our current study argues that at least one major underlying mechanism has a pre-synaptic origin. We report that knockout of the photoreceptor-specific protein, phosducin, reduces bipolar cell response sensitivity in the dark and diminishes the degree of response desensitization, which normally takes place upon persistent illumination. These effects were observed in both rod- and cone-driven pathways, with the latter affected most significantly.
Phosducin is perhaps the most mysterious photoreceptor-specific protein. Discovered over two decades ago as a major phosphoprotein in rods (Lee et al., 1984; Lee et al., 1987; Lee et al., 1988; von Schantz et al., 1994) it was subsequently shown to interact with multiple βγ-subunits of heterotrimeric G proteins (Lee et al., 1987; Lee et al., 1992; Muller et al., 1996). However, both the scope and specific role(s) of such interactions remains poorly understood.
Early studies suggested that phosducin contributes to light adaptation in the phototransduction cascade by sequestering the βγ-subunits of transducin and preventing efficient transducin activation by photoexcited rhodopsin (Lee et al., 1984; Lee et al., 1992; Lee et al., 2004). However, this hypothesis was rejected in the study of light adaptation in phosducin knockout rods (Krispel et al., 2007). On the other hand, the same study established that phosducin is involved in the regulation of transducin expression level since rods of phosducin knockout mice contain 20%-30% less transducin than normally. Here we demonstrate that a comparable ~40% reduction in transducin takes place in cones of these mice. Another function of phosducin, also revealed in the studies of Pd−/− mice, is to facilitate the light-driven translocation of transducin βγ-subunit from rod outer segments (Sokolov et al., 2004), presumably by promoting the βγ-subunits’ dissociation from the disc membranes of rod outer segments (Lee et al., 1992; Yoshida et al., 1994; Kosloff et al., 2008). However, both of these phosducin knockout phenotypes are relatively modest, which may suggest that the major function of phosducin remains to be uncovered.
The role of phosducin in photoreceptor synaptic transmission was first proposed in (Nakano et al., 2001), based on the observation that a fraction of phosducin resides in rod synaptic terminals. However, this idea had not been tested experimentally and in fact phosducin localizes not just to synapses but to all subcellular compartments of the photoreceptor cell (Thulin et al., 1999; Sokolov et al., 2004).
Here we provide the first experimental evidence that phosducin indeed contributes to the regulation of synaptic transmission in photoreceptors. Both rods and cones of phosducin knockout mice behaved as if they were persistently desensitized. The dark-sensitivity of ERG b-waves was reduced by ~20-fold for the cone-driven and by ~3-fold for the rod-driven responses and the extend of response desensitization caused by background light was also diminished for both cell types. We conclude that this phenomenon originates primarily from mechanism(s) in photoreceptors acting downstream from phototransduction, based on the following arguments.
Although phosducin knockout causes a reduction in transducin content in rods and cones, the degree of this reduction is grossly insufficient to explain the loss of b-wave sensitivity in dark-adapted Pd−/− mice. This is well-illustrated by a direct comparison of the light responses in two mouse lines, phosducin knockout and Gγ1 hemizygote. Both display a similar reduction in transducin content in their rods, causing a nearly identical ~1.5-fold reduction in the sensitivity of dim flash rod photoresponses recorded by suction electrodes. However, the sensitivity reduction for dark-adapted rod-driven b-waves in Pd−/− mice is 2.7-fold while in Gγ1+/− mice it is only 1.7-fold. Thus the sensitivity reduction in the phototransduction cascade of Gγ1+/− mice translates nearly linearly into the sensitivity reduction of their rod-driven b-waves. In contrast, the lack of phosducin strongly amplifies the sensitivity reduction in phototransduction on the level of synaptic transmission. Furthermore, b-wave desensitization by background light is normal in Gγ1+/− mice (after applying appropriate corrections to account for the dark sensitivity reduction; Fig. 3), whereas in Pd−/− mice this desensitization is diminished. This is despite a complete lack of alteration in light adaptation in the phototransduction of Pd−/− mice documented in exhaustive studies of their rods by suction electrode recordings (Krispel et al., 2007).
Another argument that phosducin regulates photoreceptor output independently of phototransduction is that dark-adapted Pd−/− mice have normal maximal amplitudes of rod- and cone-driven ERG b-waves despite their reduced light sensitivity. However, when a comparable reduction in b-wave sensitivity is caused by persistent illumination in wild type mice, the maximal b-wave amplitudes are reduced as well. Therefore, the loss of b-wave sensitivity in dark-adapted Pd−/− mice must depend upon alternative mechanisms.
The mechanisms by which phosducin regulates synaptic transmission are completely unknown. Because the only well-documented molecular partners of phosducin are the βγ-subunits of heterotrimeric G proteins, it is tempting to speculate that synaptic transmission in photoreceptors is regulated by a yet-to-be-identified mechanism involving G protein action. Indeed, there are multiple indications that G protein signaling mechanisms distinct from phototransduction may affect photoreceptor synaptic output. For example, Koulen et al. (1999), Higgs and Lukasiewicz (2002), Hosoi et al. (2005) and Koulen et al. (2005) suggested that presynaptic metabotropic glutamate receptors from groups II and III inhibit the glutamate release from photoreceptor synapses thus imposing a negative feedback loop which restricts the strength of synaptic output in the dark. Another G protein-coupled receptor found in photoreceptors is the dopamine D4 receptor, which regulates the intracellular cAMP level (Cohen et al., 1992; Jackson et al., 2009). The mechanistic details of how these pathways are organized remain to be explored. One possibility is that Gβγ-subunits could directly act on the synaptic release machinery, a regulatory mechanism documented in other neurons (Blackmer et al., 2001; Yoon et al., 2007; Yoon et al., 2008). However the relevance of this Gβγ function to ribbon synapses, which are present in photoreceptors and bipolar cells in the retina and several other cell types, has not been yet explored. Phosducin binding to Gβγ in synaptic terminals could potentially restrict Gβγ availability for the downstream targets. Alternatively, phosducin could sequester Gβγ from their Gα subunit partners, which could impair their ability to be subsequently activated by the corresponding receptors. Alternatively, phosducin may be engaged in the regulation of Ca2+ dynamics at the synapse. In this case it would modulate the synaptic output indirectly by affecting the light-induced suppression of vesicle release caused by flash or background light.
Finally, we should note that neither of the above hypothetical mechanisms is likely to rely on phosducin phosphorylation since our data with transgenically expressed noPoP phosducin mutant indicate that phosducin phosphorylation is not required for its role in b-wave modulation. The same likely applies to the previously documented interaction of phosducin with the 14-3-3 protein (Nakano et al., 2001; Thulin et al., 2001) because 14-3-3 interacts specifically with phosphoproteins.
In summary, we have established that the light-sensitivity of synaptic output from rods and cones is regulated by a novel phosducin-dependent mechanism. Elucidating the molecular details of this regulation is a challenge for future studies.
We thank P.S. Ferry-Leeper for help in maintaining mouse colonies and Y. Ling for assistance with EM analysis. This work was supported by the NIH grants EY10336 (VYA), EY 14047 (MEB), EY06671 (LJF), the NIH Core Grant for Vision Research EY5722 to Duke University and FFS fellowship to RH.