A G Protein-Coupled Receptor Phosphatase Required for Rhodopsin Function

See allHide authors and affiliations

Science  01 Aug 1997:
Vol. 277, Issue 5326, pp. 687-690
DOI: 10.1126/science.277.5326.687


Heterotrimeric guanine nucleotide-binding protein (G protein)–coupled receptors are phosphorylated by kinases that mediate agonist-dependent receptor deactivation. Although many receptor kinases have been isolated, the corresponding phosphatases, necessary for restoring the ground state of the receptor, have not been identified. Drosophila RDGC (retinal degeneration C) is a phosphatase required for rhodopsin dephosphorylation in vivo. Loss of RDGC caused severe defects in the termination of the light response as well as extensive light-dependent retinal degeneration. These phenotypes resulted from the hyperphosphorylation of rhodopsin because expression of a truncated rhodopsin lacking the phosphorylation sites restored normal photoreceptor function. These results suggest the existence of a family of receptor phosphatases involved in the regulation of G protein–coupled signaling cascades.

G protein–coupled receptors (GPCRs) mediate responses to a wide range of extracellular stimuli. These receptors contain seven putative transmembrane domains and clustered serine and threonine residues in the cytoplasmic COOH-terminus (1). The β-adrenergic receptor (2) and the light receptor rhodopsin (3) are GPCRs that share common desensitization and inactivation mechanisms. Ligand- or light-dependent activation of these receptors results in the activation of a GPCR kinase (GRK) that phosphorylates several residues in the COOH-terminus of the receptor (4). Phosphorylation results in a slight decrease in receptor activity, but also causes the receptor to become a high-affinity substrate for arrestin (5-9). Arrestin binding terminates the active state of the receptor by preventing its coupling to G protein (5, 6, 10). Although receptor dephosphorylation is essential for completing the signaling cycle (11, 12), the molecular identification of a GPCR phosphatase has proven difficult (12). We have usedDrosophila phototransduction, a phospholipase C–mediated, calcium-regulated G protein–coupled pathway for a genetic dissection of GPCR function and regulation (13).

Several mutations that lead to light-dependent retinal degeneration have been isolated in Drosophila (14-16). Most of these define genes important for photoreceptor cell signaling, suggesting that photoreceptor cell integrity depends on the normal functioning of this pathway (17). The rdgC(retinal degeneration C) gene encodes an unusual type of serine-threonine phosphatase, consisting of an NH2-terminal domain that has high sequence similarity to the catalytic domain of protein phosphatases 1, 2A, and 2B and a COOH-terminal domain containing multiple EF-hand calcium-binding motifs (18). On the basis of its primary structure, RDGC has been proposed to be a calcium-regulated phosphatase (18) and has been shown to be required for efficient dephosphorylation of rhodopsin in vitro (19).

To determine whether RDGC functions in vivo as a rhodopsin phosphatase, we analyzed rdgC mutants biochemically, physiologically, and genetically. We examined the light-dependent phosphorylation of rhodopsin in wild-type andrdgC mutant photoreceptor neurons. Unlike vertebrate opsin, most invertebrate photopigments are not bleached after light activation but can be photoconverted between the rhodopsin (R) form and a thermally stable, active metarhodopsin (M) form (20). The major rhodopsin in the fly retina, rhodopsin 1 (Rh1), has R and M forms that absorb light maximally at 480 nm (blue) and 580 nm (orange), respectively (21). Thus, blue or orange light can be used to shift between the active and inactive states of rhodopsin.

As expected, in white-eyed control flies, Rh1 was phosphorylated in a blue light–dependent manner, whereas subsequent exposure to orange light promoted its dephosphorylation (Fig. 1, A, B, and C). In contrast, Rh1 was hyperphosphorylated inrdgC mutants, and it remained in this state even after exposure to orange light (Fig. 1B). This result is consistent with the loss of a major rhodopsin phosphatase activity in rdgCmutant photoreceptor cells. No other proteins displayed altered light-dependent phosphorylation profiles in rdgCflies (22).

Figure 1

In vivo phosphorylation of retinal proteins. (A) Autoradiogram of SDS–polyacrylamide gel electrophoresis (PAGE) of32PO4 in vivo labeled retinal proteins from flies that were exposed to blue light (B) or orange light (O) for 15 min (37). Blue light promoted phosphorylation of Rh1, whereas orange light promoted dephosphorylation (w, white-eyed control flies). (B) The experiment in (A) was repeated with a 20-s pulse of light to quantitatively examine the rdgCphenotype. Upper panel: Autoradiogram of SDS-PAGE of32PO4 in vivo labeled retinal proteins. B denotes flies exposed to 20 s of blue light; BO denotes flies exposed to 20 s of blue light followed by 20 s of orange light (37). ninaE represents a null mutation in the structural gene for Rh1, rdgC is a mutation in the RDGC phosphatase gene, and Rh1Δ356 is the truncation of the last 18 residues of the COOH-terminal tail of rhodopsin. The results are representative of three independent experiments. Lower panel: The same gel blotted and probed with antibodies to Rh1. The truncation of rhodopsin results in a faster migrating polypeptide. (C) Histogram of the relative extent of 32P incorporation into Rh1. Samples were normalized to control flies exposed to blue and orange light sequentially (BO) and corrected for the amount of rhodopsin loaded. Data are means ± SEM of triplicate determinations.

Because most GPCRs are phosphorylated by GRKs at a series of serine and threonine residues on the COOH-terminal tail of the receptor (23), we predicted that the lack of a GPCR phosphatase would lead to hyperphosphorylation of the COOH-terminal region of rhodopsin. Thus, we generated transgenic flies that express a truncated rhodopsin molecule,Rh1Δ356. This mutation eliminates the last 18 amino acid residues that include the serines and threonines in the Rh1 cytoplasmic tail. This truncated rhodopsin was expressed in a ninaE mutant background such that the only rhodopsin present in these photoreceptors was the one directed by the transgene (Fig. 1B, bottom panel). The truncated receptor was expressed in near normal amounts, and the cells displayed normal light responses (24). As predicted, rhodopsin was not hyperphosphorylated in the Rh1Δ356; rdgC flies (Fig. 1B).

The finding that Rh1 was hyperphosphorylated inrdgC mutants suggested that improper rhodopsin function may underlie the retinal degeneration phenotype. Some forms of light-dependent retinal degeneration in Drosophila are the result of uncontrolled signaling and thus are suppressed by second-site mutations that prevent signaling. For instance, light-dependent retinal degeneration induced by a mutation in arrestin (Arr2) is suppressed by a null mutation in the effector phospholipase C (encoded by norpA) (16). Like arrestin mutants, therdgC degeneration is completely light-dependent (Fig. 2, A and B) because mutant flies that are grown in the dark have morphologically normal photoreceptor cells (15). However, this degeneration is not prevented by norpA, whereas it is suppressed by loss-of-function mutations in rhodopsin (25) or mutations that reduce rhodopsin amounts (26). Thus,rdgC acts at or downstream of rhodopsin, but upstream ofnorpA.

Figure 2

Rh1Δ356suppresses rdgC light-dependent retinal degeneration. Shown are cross sections (1 μm thick) through adult retinas after 6 days of light exposure (38). (A) Control flies showing normal retinal morphology, with ommatidial clusters organized as a patterned array. (B) Retinas from rdgC flies show marked degeneration. (C) Dgq mutants that lack the α subunit of the G protein that couples rhodopsin to PLC do not show light-dependent degeneration. (D) Dgqis unable to suppress the degeneration of the Dgq; rdgCdouble mutants. (E) Rh1Δ356transgenic flies display normal retinal morphology. (F) Rh1Δ356 drastically suppresses rdgC degeneration inRh1Δ356; rdgC double mutants. If kept in the dark, all genotypes display normal retinal morphology (39). Scale bar, 40 μm.

Two lines of evidence demonstrate that hyperphosphorylated rhodopsin is the cause of the light-dependent retinal degeneration. We genetically mapped the site of action of RDGC by generating double mutants between rdgC and other components of the phototransduction cascade. InDrosophila, photoconversion of rhodopsin activates a Gαq, encoded by the Dgq gene, which in turn stimulates a phospholipase C. A mutation in Dgq (27) does not protect rdgC flies from retinal degeneration, becauseDgq; rdgC double mutants still degenerate (Fig. 2, C and D). This indicates that degeneration does not require the activation of components immediately downstream of rhodopsin. We also examined the effect of theRh1Δ356 transgene on light-induced,rdgC-dependent retinal degeneration.Rh1Δ356 by itself does not affect retinal morphology (Fig. 2E). However, if the COOH-terminal hyperphosphorylation of Rh1 is directly responsible for retinal degeneration in rdgC, then elimination of the phosphorylation target sites in Rh1Δ356; rdgC double mutants should prevent the degeneration. Indeed,Rh1Δ356 acts as a strong suppressor ofrdgC degeneration (Fig. 2F).

In wild-type photoreceptor cells, termination of the light response results from the concerted action of regulatory events at multiple steps (3, 11, 17, 28). Because rdgC mutants accumulate phosphorylated rhodopsin, we expected alterations in the kinetics of photoreceptor cell inactivation. Examination of photoreceptor light responses by electroretinograms (ERGs) and intracellular recordings showed that rdgC mutants exhibited a notable decrease in the rate of deactivation relative to control flies (Fig. 3, A and B). The deactivation rate of rdgC mutant cells was less than 10% the rate of control cells (time to 85% deactivationt 85 = 1.78 s versus 0.17 s; Fig. 3C). If the deactivation defect of rdgC is attributable to the hyperphosphorylation of the serine and threonine residues in the COOH-terminal tail of rhodopsin, then theRh1Δ356 truncation should restore normal physiology to the mutant cells. Recordings from control flies expressing Rh1Δ356 showed that these photoreceptors displayed normal kinetics of activation and deactivation. As predicted, Rh1Δ356 suppressed the deactivation defect of rdgC mutants (Fig. 3A). Thus, continued phosphorylation of Rh1 is responsible for the defects in deactivation kinetics.

Figure 3

rdgC mutants have defective deactivation kinetics. (A) ERGs of w,rdgC, Rh1Δ356, andRh1Δ356; rdgC mutant flies (40). Photoreceptors were given a 0.5-s pulse of orange light.rdgC mutants display strong defects in the deactivation kinetics (arrow), and this phenotype is suppressed by theRh1Δ356 truncation. Traces from 15 independent measurements were averaged. (B) Intracellular recordings of light-activated responses from control andrdgC photoreceptors. This recording configuration demonstrates the photoreceptor specificity of the phenotype; note the strong deactivation phenotype (arrow). For control flies,t50 = 34 ± 2 ms; for rdgCflies, t50 = 136 ± 22 ms (n = 26). Data are means ± SEM. (C) Histogram of the deactivation time for the different ERG phenotypes. For control flies,t85 = 0.17 ± 0.03 s; forrdgC flies, t85 = 1.8 ± 0.3 s; for Rh1Δ356 flies,t85 = 0.16 ± 0.03 s; forRh1Δ356; rdgC flies,t85 = 0.32 ± 0.03 s (n = 15). Data are means ± SD.

Arrestin binding is required for termination of the active state of GPCRs (5, 6, 29, 30). In vertebrate photoreceptors, the formation of the rhodopsin-arrestin complex is largely determined by the phosphorylation state of rhodopsin (5, 6, 8-10,31). Because rdgC mutants accumulate hyperphosphorylated rhodopsin, the deactivation defect may result from a defect in the rhodopsin-arrestin interaction. A manifestation of this interaction in vivo is the prolonged depolarizing afterpotential (PDA) (16, 32), a sustained photoresponse that occurs when a substantial amount of rhodopsin has been photoconverted from R to the active M state (Fig.4A). A PDA results when metarhodopsin is produced in excess of free arrestin. Wild-type photoreceptors require ∼20% conversion of R to M to trigger a PDA. This amount of rhodopsin isomerization approximately equals the total number of arrestin molecules in the photoreceptor cell (16). Mutants expressing small amounts of arrestin enter a PDA with very little light (because only a small amount of M would readily saturate arrestin availability), whereas mutants that reduce amounts of Rh1 to less than those of arrestin prevent entry into a PDA (there could never be an excess of M over arrestin). In dark-raised rdgC mutants, the amounts of arrestin and rhodopsin are the same as in the wild type (22). Like arrestin mutant photoreceptors, rdgCphotoreceptors entered a PDA with approximately one-eighth as much light as did photoreceptors in control flies (Fig. 4B) (16). In contrast, Rh1Δ356; rdgC double mutants displayed a normal PDA. Thus, RDGC is required for normal rhodopsin function, and the deactivation defect may be attributable to impaired arrestin function that results from rhodopsin hyperphosphorylation.

Figure 4

Defects in the PDA response of rdgCflies. (A) ERG recording of a white-eyed control fly showing a prototypical PDA response. O, orange light (M → R conversion); B, blue light (R → M conversion). (B) Histogram of relative amount of blue light required to enter a PDA for the different genotypes. Values are normalized to control flies and represent means ± SEM of triplicate determinations.Arr2, a known mutant with a PDA defect (16), is shown for comparative purposes.

Extensive research in other systems had shown that receptor phosphorylation is required for arrestin binding; however,Rh1Δ356 transgenic flies displayed normal deactivation physiology. This contrasts to vertebrate rhodopsin, in which a truncation of the COOH-terminal tail abolished arrestin binding (9) and led to severe defects in deactivation (29). Drosophila rhodopsin could be a special case, but this is not likely given the conservation of GPCR function in different systems. Instead, the COOH-terminus of rhodopsin may function as an autoinhibitory domain for arrestin binding. In this case, the lack of the COOH-terminus in Rh1Δ356 would eliminate the need for receptor phosphorylation and would explain the rescue of rdgC phenotypes by theRh1Δ356 truncation.

Also, photoreceptor cell degeneration resulted directly from an excess of phosphorylated rhodopsin. Because rhodopsin dephosphorylation is essential for photoreceptor cell integrity, rhodopsin phosphatase mutations in vertebrates may also lead to retinal dysfunction and may account for some of the human retinal degenerative disorders. Indeed, these results could also help to explain the retinal degeneration seen in human retinitis pigmentosa patients and in transgenic mice that express a Lys296 → Glu missense mutation in rhodopsin (which eliminates the chromophore binding site and leads to light-independent receptor activation in vitro) (33,34). However, the Lys296 → Glu mutation does not lead to degeneration by constitutively activating phototransduction; instead the mutant opsin was found to be constitutively phosphorylated and inactivated by deactivation mechanisms (34, 35).

Although there are hundreds of GPCRs, there are only a few G proteins, receptor kinases, and arrestins. RDGC is also expressed in the mushroom bodies of the fly brain (18), which suggests that this phosphatase participates in different pathways. The recent finding of RDGC homologs in the nervous systems of humans and mice (36) indicates that the conservation of GPCR function probably extends to receptor phosphatases.

  • * Present address: Institute of Biotechnology and Department of Molecular Medicine, University of Texas Health Science Center, San Antonio, TX 78245, USA.

  • To whom correspondence should be addressed. E-mail: czuker{at}


Stay Connected to Science

Navigate This Article