Metallophosphoesterase regulates light-induced rhodopsin endocytosis by promoting an association between arrestin and the adaptor protein AP2

Light-induced endocytosis of rhodopsin in the retina is critical for preventing photoreceptor hyperactivity and for the survival of photoreceptor cells. In Drosophila, this process is mediated by arrestin1 (Arr1). Because Arr1 lacks a clathrin-binding domain required for receptor internalization and the C-terminal sequence that interacts with the β-subunit of the clathrin adaptor protein AP2, the mechanism of how Arr1 mediates endocytosis of the major rhodopsin Rh1 is unclear. Here, using several approaches, including Arr binding and pulldown assays, immunofluorescence techniques, and EM imaging, we found that Drosophila metallophosphoesterase (dMPPE) is involved in light-induced rhodopsin endocytosis. We observed that the photoreceptor cells of a dmppe mutant exhibit impaired light-induced rhodopsin endocytosis and that this impairment is independent of dMPPE phosphoesterase activity. Furthermore, dMPPE directly interacted with Arr1 and promoted the association of Arr1 with AP2. Of note, genetic dmppe deletion largely prevented retinal degeneration in norpA (encoding phospholipase C) mutants, which were reported previously to contribute to retinal degeneration, by suppressing Rh1 endocytosis. Our findings demonstrate that Arr1 interacts with AP2 and that dMPPE functions as a critical regulator in Rh1 endocytosis and retinal degeneration.

G protein-coupled receptors (GPCRs) 3 mediate extracellular signals from neurotransmitters, hormones, cytokines, and sensory stimuli to intracellular signaling cascades (1,2). To control the signaling event, a family of regulatory molecules called arrestins uncouple the activated receptors from the G proteins to deactivate GPCRs (3,4). When the GPCRs are con-tinuously activated, the bound arrestins will recruit clathrin proteins to internalize the receptor from the cell surface through dynamin-dependent endocytosis. The internalized GPCRs subsequently undergo lysosomal degradation or are recycled back to the plasma membrane (5)(6)(7)(8). Both internalization and down-regulation of GPCRs reduce their density on the plasma membrane, thereby causing long-term desensitization of the cell to extracellular stimuli.
In Drosophila photoreceptors, light stimulation activates rhodopsin to form activated metarhodopsin, which in turn activates heterotrimeric G proteins and phospholipase C (PLC) (9). The activation of PLC leads to opening of transient receptor potential and transient receptor potential-like channels and extracellular Ca 2ϩ influx (10,11). Endocytosis of a major rhodopsin called Rh1 is mediated by two types of arrestins. The major visual arrestin Arr2 interacts with the ␤-subunit of the clathrin adaptor protein AP2 through its C-terminal tail to induce endocytosis of the photoactivated metarhodopsin in a number of visual system mutants (12)(13)(14). The minor visual arrestin Arr1 is essential for light-induced endocytosis of Rh1, even though it lacks the C-terminal AP2-binding sequence (15). Whether Arr1 also mediates Rh1 endocytosis through its interaction with AP2 remains unknown.
Drosophila metallophosphoesterase (dMPPE) is an eye-enriched protein. Our previous study showed that it acts as a phosphoesterase and mediates Rh1 deglycosylation. In this study, we show that dMPPE is involved in light-induced Rh1 endocytosis and that its role in regulating Rh1 endocytosis is independent of its phosphoesterase activity. We further reveal that dMPPE directly interacts with Arr1 and regulates lightinduced Rh1 endocytosis by promoting Arr1/AP2 association. We also show that genetic removal of dmppe suppresses retinal degeneration in both norpA and trp mutants. Our study reveals the molecular mechanism of Arr1-mediated endocytosis of Rh1 and identifies a regulator in light-induced Rh1 endocytosis and retinal degeneration.

dMPPE is involved in light-induced Rh1 endocytosis
Previous studies have identified some eye-enriched genes through microarray analysis (16,17). To investigate the potential functions of these genes in light-induced Rh1 endocytosis, we screened flies with eye-enriched gene mutations to characterize novel regulators for light-induced Rh1 endocytosis.

cro ARTICLE
These flies were first adapted to a dark environment so that most Rh1 proteins were localized in the rhabdomere (Fig. 1A). Then they were exposed to room light (700 lux) for 30 min to induce Rh1 endocytosis. Our previous study showed that dmppe mutants exhibit defects in Rh1 deglycosylation (18). In this screen, we observed much fewer endocytic Rh1 vesicles (ERVs) in the cell bodies of dmppe mutant photoreceptors compared with WT ones even after prolonged light exposure (Fig. 1, A  Our previous study showed that dMPPE mediates Rh1 deglycosylation (18). Moreover, the glycosylation status of several membrane receptors has been reported to affect their endocytosis (19). Thus, we hypothesized that defective deglycosylation may affect Rh1 endocytosis. We measured light-induced Rh1 endocytosis in an ␣-man-II mutant that also displays defective Rh1 deglycosylation (18). Our data demonstrate that the number of ERVs per ommatidium in the ␣-man-II mutant was comparable to that in WT flies ( Fig. 1D

dMPPE interacts directly with Arr1
To explore how dMPPE regulates Rh1 endocytosis, we performed pulldown and nano-LC-MS/MS analysis to identify proteins that interact with dMPPE. We found several molecules
Next, we used recombinant GST-dMPPE proteins to map the critical Arr1-interacting region. dMPPE has been predicted to contain two transmembrane domains (14 -30 aa and 312-328 aa) and both the N and C termini are localized to the intracellular side. Therefore, we generated the N-terminal deletion mutant GST-dMPPE ⌬N and the C-terminal deletion mutant GST-dMPPE ⌬C . We found that both full-length GST-dMPPE and GST-dMPPE ⌬N bound to Arr1, whereas GST-dMPPE ⌬C did not (Fig. 3C). Furthermore, the C-terminal region of dMPPE (GST-dMPPE C ) alone was sufficient for Arr1 binding (Fig. 3C). We further narrowed down the binding epitope to residues 344 -352 of dMPPE using alanine scanning mutagenesis (Fig. 3D). Taken together, these data indicate that the C-terminal intracellular tail of dMPPE is essential for Arr1 binding.

dMPPE/Arr1 interaction does not affect Rh1/Arr1 association
Given that Arr1 binding to the C-terminal tail of Rh1 is the initial step in Arr1-mediated Rh1 endocytosis (15), we further performed an arrestin-binding and -release assay to examine whether dMPPE affects Rh1/Arr1 association. Rh1 can be photoconverted between the active (metarhodopsin, M) and inactive states (rhodopsin, R) (15). In the dark-adapted WT, Arr1 does not bind to inactive R and mainly exists in the superna-tants of head homogenates. After exposure to blue light (480 nm), inactive R is photoconverted to M and most Arr1 associates with rhabdomere membranes. When M is photoconverted back to R by orange light exposure (580 nm), Arr1 is released and detected in the supernatant again (Fig. 5, A and B). In the binding and release assay, dmppe mutants showed a similar Arr1 association pattern as WT files (Fig. 5, A and B: dark, 32.5 Ϯ 6.9% versus 35.7 Ϯ 9.3%, p ϭ 0.66; blue light, 69.1 Ϯ 8.9% versus 70.1 Ϯ 2.0%, p ϭ 0.87; orange light, 40.0 Ϯ 11.3% versus 34.5 Ϯ 3.5%, p ϭ 0.46; t test). Notably, dMPPE itself did not undergo translocation and resided predominately in the membrane fraction in WT flies (Fig. 5, A and C). These results demonstrate that the dMPPE/Arr1 interaction does not affect Rh1/Arr1 association.

dMPPE promotes Arr1/AP2 association
The interaction between ␤-arrestin and the AP2 adaptor complex has been shown to be critical for GPCR endocytosis (20). In Drosophila, point mutations in the C-terminal tail of Arr2 (Arr2 K391A and Arr2 R393A ) completely disrupted its interaction with AP2 (12)(13)(14). However, Arr1 lacks the C-terminal sequences that interact with AP2 (Fig. 6A), suggesting that an alternative mechanism must exist for Arr1-mediated Rh1 endocytosis. Interestingly, AP2 was also identified as one of the dMPPE-associated proteins in our MS analysis (Fig. 3A). Our co-immunoprecipitation experiments confirmed that dMPPE can associate with both Arr1 and AP2␣ in vivo (Fig. 3B). Thus, we hypothesized that dMPPE might function as a linker between Arr1 and AP2␣. To test this, we carried out coimmunoprecipitation experiments and found that anti-Arr1 antibodies precipitated much less AP2␣ from the extracts of dmppe mutant heads than that of WT heads (Fig. 6B: 100% versus 39.4 Ϯ 15.3%; p Ͻ 0.001; t test). These data strongly indicate that dMPPE promotes the association between Arr1 and AP2␣.

Depletion of dMPPE prevents retinal degeneration in norpA mutants by impairing Arr2/AP2␣ association
Because Arr1 and Arr2 exhibit high similarity in their amino acid sequence and Arr2 was shown to bind dMPPE in our MS spectrum analysis, we further investigated whether dMPPE also

Metallophosphoesterase mediates rhodopsin endocytosis
promotes Arr2/AP2␣ association. Previous reports showed that norpA (encoding phospholipase C) mutants undergo lightdependent retinal degeneration because of abnormal accumulation of Rh1/Arr2 stable complexes (12,21). Arr2/AP2 association is essential for internalization and loss of the ␣-subunit of the AP2 adaptor complex can suppress retinal degeneration in the norpA mutant (14). Our data show that genetic removal of dmppe diminishes Arr2/AP2␣ association and suppresses Rh1 endocytosis. We found that 30-min exposure to room light induced much fewer ERVs in norpA;dmppe photoreceptors compared with norpA photoreceptors (Fig. 7A: 1.54 Ϯ 0.22 versus 0.74 Ϯ 0.20; p Ͻ 0.001; t test). EM study further revealed that the norpA;dmppe double mutant largely rescued the retinal degeneration observed in 6-day-old norpA mutants (Fig. 7B). Moreover, arrestin-binding and -release assay showed that stable Rh1/Arr2 complexes accumulated in norpA;dmppe double mutants (Fig. 7, C and D). Taken together, these data indicate that genetic removal of dmppe suppressed retinal degeneration in norpA mutants by inhibiting Rh1 endocytosis but not disrupting stable Rh1/Arr2 complex formation.

Metallophosphoesterase mediates rhodopsin endocytosis
trp (encoding transient receptor potential channel) mutants have been shown to undergo light-dependent retinal degeneration. Although depletion of Ptdlns(4,5)P 2 but not accumulation of Rh1/Arr2 stable complexes accounts for light-induced retinal degeneration in trp mutants (22), elimination of Arr2 indeed partially suppressed the retinal degeneration in trp mutants (23). Consistently, we found that genetic removal of dmppe also largely suppressed retinal degeneration in trp mutants (Fig. 7B). The molecular mechanism of dMPPE depletion in suppressing retinal degeneration in trp mutants requires further investigation.

dMPPE regulates Rh1 endocytosis
Endocytosis is a fundamental cell biological process that is important for various aspects of cellular physiology ranging from nutrient uptake and signal transduction to synaptic transmission and development (24,25). In Drosophila photoreceptors, light stimulation induces isomerization of chromophores, thereby inducing a conformational change of inactive rhodopsin to active metarhodopsin (9). Under normal physiological conditions, a fraction of metarhodopsin is internalized upon light exposure (15,26). Endocytosis of Rh1 is thought to scavenge constitutively active or damaged metarhodopsin to induce long-term adaptation and protect the photoreceptors from excessive Ca 2ϩ influx-induced cell toxicity (26,27). Moreover, endocytosis and exocytosis of Rh1 also maintains Rh1 homeostasis upon long-term light exposure (26 -28).
It has been known that C-terminal phosphorylation of Rh1 triggers its endocytosis and that Arr1 binds to phosphorylated Rh1 to initiate endocytosis (15). Structural studies have revealed that the N-and C-terminal domains of arrestin are composed of antiparallel ␤-sheets that are linked by an unusual polar core (29). A network of intramolecular interactions between charged residues buried within the polar core maintains the basal conformation of arrestin (30). When the N-terminal domain of arrestin binds to activated GPCRs, arrestins undergo a conformational change to expose their C-terminal domain. The C-terminal tail of arrestin interacts with components of the endocytic machinery, including the clathrin heavychain and the ␤-subunit of AP2, thereby facilitating GPCR internalization (31). However, Drosophila Arr2 and Arr1 do not contain a clathrin-binding domain. Although Drosophila Arr2 has been shown to interact with the ␤-subunit of AP2 (14), Drosophila Arr1 lacks the sequences needed to mediate an interaction with AP2␤. In this study, we show that dmppe mutants exhibit reduced light-induced Rh1 endocytosis and reveal that the intracellular tail of dMPPE binds directly to Arr1. Depletion of dMPPE or the Arr1-binding sites in dMPPE largely impairs Arr1/AP2 association and Rh1 endocytosis. Our study identified a novel regulator in Rh1 endocytosis and provides strong evidence that the Arr1/dMPPE interaction regulates Rh1 endocytosis.
Our co-immunoprecipitation assay revealed that Arr1 can still partially associate with AP2 in the absence of dMPPE. These data are consistent with the observation that prolonged light exposure can still induce Rh1 endocytosis in dmppe mutants. These results suggest that other molecules might be able to mediate Arr1/AP2 association. Phosphatidylinositol-bisphosphate (PI(4,5)P 2 ) is an attractive candidate (4,5). The C-terminal domain of ␤-arrestin contains a basic region that binds to PI(4,5)P 2 and Drosophila Arr2 has been shown to interact with PI(4,5)P 2 (32,33). The N terminus of the AP2 ␣-subunit can bind to PI(4,5)P 2 , and this interaction promotes the localization of AP2 to the plasma membrane and regulates GPCR endocytosis (34 -37).

Genetic removal of dmppe prevents retinal degeneration in both norpA and trp mutants
Under normal conditions, the interaction between Arr2 and Rh1 is transient, because light-triggered Ca 2ϩ influx activates CaM kinase II, which subsequently phosphorylates Arr2 to cause its release from Rh1 (38,39). In norpA mutants, photoresponses were abolished, which prevents the normal rise in Ca 2ϩ after light stimulation and results in the accumulation of Rh1/ Arr2 stable complexes and retinal degeneration (12,21). In this study, we show that genetic removal of dmppe prevents retinal degeneration in norpA mutants but does not disrupt the formation of stable Rh1/Arr2 complexes. We further demonstrated that depletion of dMPPE suppresses Rh1 endocytosis in norpA mutants. Consistent with this, inhibition of Rh1 endocytosis using a temperature-sensitive dynamin (Shibire ts ) or by removing the ␣-subunit of AP2 has been shown to suppress retinal degeneration in norpA mutants (12,14). These results support the conclusion that dMPPE regulates Rh1 endocytosis.
In summary, our study provides evidence that the dMPPE/ Arr1 interaction is critical for light-induced Rh1 endocytosis and for preventing retinal degeneration. Our findings have identified a novel regulator in Rh1 endocytosis and retinal degeneration. In addition, our results provide a linkage between Arr1 and the endocytic machinery. Three sets of independent data were averaged, and the data are presented as the mean Ϯ S.D. C, quantification of relative dMPPE in membrane fraction. Three sets of independent data were averaged, and the data are presented as the mean Ϯ S.D. n.s., no significance.

Fly genetics and light treatment
The mutant alleles of pBacCG8889 e02905 were obtained from Harvard Medical School and were crossed into a cn,bw background. WT flies used in this study were cn,bw. The mutant alleles used for each gene in this work were norpA P24 , arr2 5 , arr1 1 , and trp 9 . p[UAS::dMPPE] was generated by our lab (18). The P-element insertion line ␣-Man-II␣ LL01094 was obtained from the Kyoto stock center and genetic mosaics were induced using the FLP-FRT, p[GMR-hid] technique with an ey-FLP driver to generate mitotic clones of a single genotype in the eye (40). All flies were reared at 25°C on standard medium in a 12-h light/12-h dark cycle incubator, with 60 -80% relative humidity. To induce Rh1 endocytosis, white eye flies younger than 3 days old were dark adapted overnight and then exposed to room light (ϳ700 lux) for 30 min or 1 h.

Generation of transgenic flies
To generate p[UAS-dMPPE 164 -183A ] and p[UAS-dMPPE 344 -355A ] transgenic flies, fragments of CG8889-RA Figure 6. dMPPE promotes Arr1/AP2 association. A, alignment of Drosophila Arr1 and Arr2. The AP2-binding residues in Arr2 are highlighted in red. The C terminus of Arr1 is short and lacks these AP2-binding residues. The PIs-binding residues are highlighted in yellow. B, Arr1 co-immunoprecipitated with AP2␣ from WT head extracts, but a much lower quantity than from dmppe mutant head extracts. Fly head extracts from either WT or dmppe mutants were immunoprecipitated with anti-Arr1 antibody. The precipitates, as well as 4% of head extracts, were subjected to Western blot analysis with antibodies against dMPPE, Arr1, and AP2␣. IP, immunoprecipitated. Quantification of the relative level of AP2␣ co-immunoprecipitated by anti-Arr1 antibody is presented under the blots. Seven sets of independent data were averaged, and the data are presented as the mean Ϯ S.D. ***, p Ͻ 0.001. Each picture shows a single ommatidium. Scale bar, 1 m. C, Arr2-Rh1 binding in WT, dmppe, norpA, and norpA;dmppe flies. Arr2-Rh1-binding and -release assays were performed as described in "Experimental Procedures." D, quantification of the percentage of Arr2 bound to rhodopsin-containing membranes in the dark (D), after treatment with blue light (B), or after treatment with blue light followed by orange light (BO). Three sets of independent data were averaged, and the data are presented as the mean Ϯ S.D. ***, p Ͻ 0.001.

Co-immunoprecipitation
Two hundred fly heads were collected and homogenized on ice in 1 ml lysis buffer (10 mM imidazole, pH 7.35, 10% sucrose, 5 mM MgCl 2 , 1 mM DTT, 160 mM KCl, and 1ϫ cOmplete Protease Inhibitor mixture (Roche)) (42). After rotation for 30 min at 4°C, the lysates were centrifuged at 14,000 rpm at 4°C for 5 min. The supernatant was incubated with 1 l anti-dMPPE or 0.5 l anti-Arr2 or anti-Arr1 antibodies at 4°C for 2 h. After blocking with 4% BSA in PBS buffer at 4°C for 30 min, 50 l protein A beads (Thermo Fisher) were added and incubated at 4°C overnight. After five washes with PBS, the bound complexes were eluted with 2ϫ SDS sample buffer and subjected to SDS-PAGE and Western blotting.

Pulldown assays
All GST and MBP fusion proteins were expressed in Escherichia coli BL21 cells and purified with GSH-Sepharose (GE Healthcare) or amylose resin (New England Biolabs), respectively. To map the Arr1-binding site in dMPPE, MBP-Arr1 fusion protein-coupled beads were incubated with various purified GST-dMPPE fragments. After washing, the elution was analyzed by Western blotting.

Arrestin-binding assay
Arr1-and Arr2-binding assays were performed as described previously (38,43). Briefly, eight flies from each group and condition were collected and adapted to the dark overnight. After exposure to pure blue light (480 Ϯ 10 nm) for 30 s, fly heads were homogenized in the dark. After centrifuging at 14,000 rpm for 5 min, pellet and supernatant fractions were separated under dim red light and subjected to SDS-PAGE and Western blot analysis. Arr1-and Arr2-release assays were performed in the same manner, except that the flies were exposed to 30 s of pure blue light followed by 90 s of pure orange light (580 Ϯ 10 nm).

Immunostaining
Immunofluorescence staining was carried out as described previously (44). Briefly, fly heads were hemisected and immersed in 10% formaldehyde fixing solution for 2 h on ice. The eyes were dehydrated with acetone and embedded with LR White resin. One-micrometer sections were cut across the top one third of the eye and stained with anti-Rh1 antibody (1:300) (DSHB, catalog no. 4C5). The stained sections were examined under an LSM 700 confocal microscope (Zeiss, Germany). The activity of Rh1 endocytosis was quantitated as the number of ERVs per ommatidium. At least six sections from three flies were examined and averaged for each genotype and light treatment.

EM
EM was performed as described previously (45). Briefly, fly heads were hemisected and fixed in 0.1 M sodium cacodylate buffer (pH 7.2) with 2.5% glutaraldehyde and 4% paraformaldehyde at 4°C for 2 h. After rinsing with 0.1 M sodium cacodylate for three times, the eyes were stained with 1% osmium tetroxide for 1 h at room temperature. After ethanol dehydration and propylene oxide washing, the samples were embedded using standard procedures. The ultrathin sections (90 nm) were cut, collected, and stained with uranyl acetate followed by lead citrate. Micrographs were taken at 80 kV on a Hitachi-7650.

Identification of dMPPE-associated proteins by MS
The GST fusion proteins were expressed in E. coli BL21 cells and purified with GSH-Sepharose (GE Healthcare). Approximately 500 g of each protein was coupled to 5 ml Sepharose 4B agarose and incubated with head extracts of 1000 flies (in PBS containing 1% Triton X-100 and protease inhibitors). After three washes with PBS, the proteins were eluted and dissolved in ammonium bicarbonate buffer solution (25 mM, pH 8.0) and digested with trypsin at 37°C for 16 h. The nano-LC-MS/MS experiments were performed with LTQ Orbitrap MS (Thermo Fisher) equipped with a nanoelectrospray ion source as described previously (46). The LTQ Orbitrap instrument was operated in positive ion mode. An Agilent 1100 series LC system equipped with a reverse-phase microcapillary column (0.075 ϫ 150 mm, C18 column, Dionex) was used, with the mobile phase consisting of buffer A (0.1% formic acid in H 2 O) and buffer B (0.1% formic acid in acetonitrile). The analytical condition was set at a linear gradient from 0 to 60% buffer B in 60 min, and the flow rate was adjusted to 200 nl/min. The column was re-equilibrated at initial condition for 10 min. The MS/MS spectra acquired from precursor ions were submitted to MaxQuant.
MaxQuant was used for identification and peptide quantification. A spectrum selector was used to import spectrums from raw files. The search engine was used to cross-reference identified peptides against Drosophila UniProt fasta database. The

Metallophosphoesterase mediates rhodopsin endocytosis
mass tolerance was set to be 10 ppm for precursors and 0.01 Da for fragment ions. Oxidation on methionine and deamination on asparagine and glutamine were chosen as variable modifications. Carbamidomethyl on cysteine was set as static modification. Trypsin was set as the proteolytic enzyme with up to two missed cleavages. The precursor ion area detector in Max-Quant was used for quantification. Each protein was considered identified when at least three unique peptides were identified. The list of identified proteins is included in the supporting information.

Experimental design and statistical analysis
Immunostaining and EM studies on fly eye sections were conducted to explore the function of dMPPE in Rh1 endocytosis and retinal degeneration. Biochemical studies on fly eyes were conducted to explore the detailed mechanism.
The Western blotting and fluorescence images were analyzed using the Image J software (National Institutes of Health). Data from more than three independent experiments were averaged. Statistical analysis was performed using Prism 5.0 (GraphPad Software). All data are presented as the mean Ϯ S.D. Student's t test was used to compare the genotypes or conditions. Statistical significance was set at p Ͻ 0.05 (*); p Ͻ 0.01 (**); p Ͻ 0.001 (***); and no significance (n.s.).