An Interaction between L-prostaglandin D Synthase and Arrestin Increases PGD2 Production*

L-type prostaglandin synthase (L-PGDS) produces PGD2, a lipid mediator involved in neuromodulation and inflammation. Here, we show that L-PGDS and arrestin-3 (Arr3) interact directly and can be co-immunoprecipitated endogenously from MG-63 osteoblasts. Perinuclear L-PGDS/Arr3 co-localization is observed in PGD2-producing MG-63 cells and is induced by the addition of the L-PGDS substrate or co-expression of COX-2 in HEK293 cells. Inhibition of L-PGDS activity in MG-63 cells triggers redistribution of Arr3 and L-PGDS to the cytoplasm. Perinuclear localization of L-PGDS is detected in wild-type mouse embryonic fibroblasts (MEFs) but is more diffused in MEFs-arr-2−/−-arr-3−/−. Arrestin-3 promotes PGD2 production by L-PGDS in vitro. IL-1β-induced PGD2 production is significantly lower in MEFs-arr-2−/−-arr-3−/− than in wild-type MEFs but can be rescued by expressing Arr2 or Arr3. A peptide corresponding to amino acids 86–100 of arrestin-3 derived from its L-PGDS binding domain stimulates L-PGDS-mediated PGD2 production in vitro and in MG-63 cells. We report the first characterization of an interactor/modulator of a PGD2 synthase and the identification of a new function for arrestin, which may open new opportunities for improving therapies for the treatment of inflammatory diseases.

Prostaglandins (PGs) 5 are lipid mediators formed from arachidonic acid through the action of cyclooxygenases (COXs). COXs convert arachidonic acid released from the plasma membrane to an intermediate substrate, PGH 2 , which is metabolized by specific synthases to produce PGs like PGD 2 (1,2). PGD 2 is involved in various physiological processes such as vasodilatation, bronchoconstriction (3), regulation of pain (4), and sleep (5) but is also implicated in inflammatory responses such as asthma (6) and atherosclerosis (7). PGD 2 was shown to exhibit anti-inflammatory properties as well, as increased levels of PGD 2 are observed during the resolution phase of inflammation (8 -10). Recent work by our group showed that PGD 2 displays anabolic properties in bone (11,12).
It is remarkable how very little is known about the interaction partners and the mechanisms regulating the activity of PG synthases considering their crucial physiological and pathological roles and the clinical problems associated with the long term use of COX inhibitors. Here we show that arrestin-3 interacts with L-PGDS and increases L-PGDS-mediated PGD 2 production. An arrestin-3 peptide was identified as capable of inducing PGD 2 production by L-PGDS. This is important because it shows that by identifying interacting partners of PG synthases we can not only further our understanding of the * This work was supported by a grant from Canadian Institutes of Health Research (to A. d. B. F. and J.-L. P.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Fig. S1. 1 Recipient of studentships from the Fonds de la recherche en santé du Qué bec. 2 Supported by fellowships from the Canadian Arthritis Network for part of this work. 3 Holds a Canada Research Chair in Molecular Endocrinology. 4 Recipient of "Chercheur boursier senior" salary support from the Fonds de la recherche en santé du Qué bec and the recipient of the scarcely documented regulatory mechanisms of these critical enzymes but also provide approaches to identify potentially active small molecules that can modulate their activity. This could constitute an alternative approach for the development of modulators of PG synthases.

EXPERIMENTAL PROCEDURES
Antibodies and Chemicals-The polyclonal anti-HA, polyclonal anti-myc, monoclonal anti-Arr3, and mouse anti-IgG antibodies as well as protein G-agarose were purchased from Santa Cruz Biotechnology. The anti-GST polyclonal antibody was from Bethyl Laboratories. The monoclonal anti-His antibody was from Cell Signaling Technology. The monoclonal anti-HA antibody was from Covance. The polyclonal and monoclonal L-PGDS antibodies, PGH 2 , HQL-79, and the prostaglandin D 2 EIA kit were purchased from Cayman Chemical Co. The arrestin-3 antibody blocking peptide was from Santa Cruz. The HRP-conjugated anti-mouse and the HRP-conjugated anti-rabbit were from GE Healthcare. The Alexa Fluor 488 donkey anti-mouse, immunoglobulin G (IgG), Alexa Fluor 546 goat anti-mouse, and Alexa Fluor 546 goat anti-rabbit IgG antibodies were from Molecular Probes. Selenium tetrachloride (SeCl 4 ) was from Sigma, whereas IL-1␤ was from Cedarlane.
Yeast Two-hybrid Screen-A yeast two-hybrid screen was performed following the two-hybrid system standard protocol (31). Briefly, a plasmid containing the complete cDNA of Arr3 (pAS2-1-Arr3) was transformed into the yeast strain pJ69-4␣ according to the lithium yeast transformation protocol (32). This stably transformed clone was then transformed with a human HeLa MATCHMAKER cDNA Library or with the empty pGAD424 plasmid (Clontech). Positive clones were initially selected for growth in the absence of histidine. Clones showing positive interactions were then isolated, and these interactions were confirmed by growth on quadruple selective media (Trp Ϫ , Leu Ϫ , His Ϫ , and Ade Ϫ ). pGADGH plasmids containing the library inserts from positive colonies were isolated and transformed into the DH5␣ bacterial strain. Plasmids were extracted from DH5␣ cells and transformed once more into yeast with either the bait (pAS2-1-Arr3) or the negative control (pAS2-1) and plated on quadruple selective medium (Trp Ϫ , Leu Ϫ , His Ϫ , and Ade Ϫ ) to confirm the interaction. The selected plasmids were then sequenced by dideoxy sequencing, and the identities of the clones were determined by using the NCBI BLAST alignment tool.
Immunoprecipitations-The HEK293 cells were transiently transfected with the indicated constructs and maintained as described above for 48 h. Where indicated, cells were incubated for 15 min at 37°C in the presence of 5 M PGH 2 before harvesting. The cells were then washed with ice-cold PBS and harvested in 300 l of lysis buffer (150 mM NaCl, 50 mM Tris, pH 8, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 10 mM Na 4 P 2 O 7 , and 5 mM EDTA supplemented with protease inhibitors (9 nM pepstatin, 9 nM antipain, 10 nM leupeptin, and 10 nM chymostatin; Sigma)). After incubation in lysis buffer for 60 min at 4°C, the lysates were clarified by centrifugation for 20 min at 14,000 ϫ g at 4°C. One g of specific monoclonal antibodies was added to the supernatant. After 60 min of incubation at 4°C, 40 l of 50% protein G-agarose was added followed by 3 h of incubation at 4°C. Samples were then centrifuged for 1 min in a microcentrifuge and washed three times with 1 ml of lysis buffer. Immunoprecipitated proteins were eluted by the addition of 50 l of SDS sample buffer followed by 5 min in boiling water. Initial lysates and immunoprecipitated proteins were analyzed by SDS-PAGE and immunoblotting using specific antibodies.
Recombinant Protein Production and Binding Assays-To produce His-tagged proteins, PCR fragments corresponding to the cDNA coding for full-length L-PGDS or Arr3 were inserted into the pRSETA expression vector (Invitrogen). These constructs were used to produce fusion proteins in Over-Express TM C41(DE3) Escherichia coli strain (Avidis) by following the manufacturer's instructions. The recombinant proteins were purified using nickel-nitrilotriacetic acid-agarose resin (Qiagen) as indicated by the manufacturer. The cDNA fragments coding for full-length or for different regions of Arr3 and for full-length L-PGDS were amplified by PCR and introduced into the pGEX-4-T1 vector (Amersham Biosciences) to produce the indicated glutathione S-transferase (GST)-tagged fusion proteins in the OverExpress TM C41(DE3) E. coli strain, which were purified using glutathione-Sepharose TM 4B (Amersham Biosciences) as indicated by the manufacturer. Purified recombinant proteins were analyzed by SDS-PAGE followed by Coomassie Brilliant Blue R-250 staining. Five g of glutathione-Sepharose bound GSTtagged fusion protein was incubated with 5 mg of purified histidine protein in binding buffer (10 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 10% glycerol, and 0.5% Igepal) supplemented with protease inhibitors (9 mM pepstatin, 9 mM antipain, 10 mM leupeptin, and 10 mM chymostatin) and 2 mM DTT overnight at 4°C. The binding reactions were then washed 4 times with binding buffer. SDS sample buffer was added to the binding reactions, and the tubes were boiled for 5 min. The binding reactions were analyzed by SDS-PAGE, and immunoblotting was performed with the indicated specific antibodies.
Cell Fractionation-For isolation of cytoplasmic and nuclear fractions, MEF cells were plated in 6-well plates and transiently transfected with L-PGDS-HA to facilitate its detection. Plasma membranes were then disrupted in a buffer containing 10 mM HEPES, 10 mM KCl, 0.1 mM EDTA, 1% Igepal supplemented with protease inhibitors mixture. Centrifugation for 5 min to 14,000 ϫ g at 4°C allowed recovery of the cytoplasmic fraction. Pellets were resuspended with buffer containing 10 mM HEPES, 400 mM NaCl, 1 mM EDTA, DTT 1 mM, 10% glycerol, and protease inhibitors mixture and were incubated for 45 min at 4°C. Samples were then centrifuged for 5 min at 14,000 ϫ g at 4°C to isolate the nuclear fraction in the supernatant. SDS sample buffer was added to the lysates, and samples were boiled for 5 min. Cytoplasmic (supernatant) and nuclear fractions were analyzed by SDS-PAGE, and immunoblotting was performed with the indicated specific antibodies.
In Vitro PGD 2 Production Assays-His 6 -Arr3, His 6 -L-PGDS, or His 6 alone were produced as described above and were incubated at a predetermined molar ratio in a buffer containing 1 M Tris-HCl, pH 8.0, 1 mM DTT, 0.5 M guanidine HCl (33), and 1 g/ml IgG for 10 min at room temperature in 96-well plates. PGH 2 to a final concentration of 0.5 M was added to the wells, and the reaction was performed for 1 min and then stopped with 0.4 mg/ml SnCl 2 . PGD 2 produced was then measured with the prostaglandin D 2 EIA kit according to the manufacturer's instructions. For PGD 2 production assays with peptides, the indicated peptides were added to the buffer with the same molar ratio, and PGD 2 production was measured as described above.
PGD 2 Production Assays in Cells-MEFs wt, Arr2 KO, Arr3 KO, or Arr (double knockout (dKO) were plated in 24-well plates and transiently transfected when needed with the indicated constructs. To eliminate PGD 2 production by H-PGDS, cells were preincubated with HQL-79 (a specific H-PGDS inhibitor (34) to measure only L-PGDS-mediated PGD 2 production) at 100 M for 15 min at 37°C. Where indicated, cells were incubated for 15 min at 37°C in the presence of 5 M PGH 2 or starved with DMEM without FBS for 24 h and stimulated with 5 ng/ml IL-1␤ for 16 h. The cells were lysed as described above, and PGD 2 in the lysates and in the media were measured with the prostaglandin D 2 EIA kit according to the manufacturer's instructions. For PGD 2 production assays with peptides, MG-63 cells were plated in 24-well plates, and 100 M TAT peptide was added to the cells for 1 h at 37°C (35). The assay was then performed as described above, and proteins in cell lysates were measured with a commercial kit (Bio-Rad) to normalize levels of PGD 2 .
Immunofluorescence Staining and Confocal Microscopy-For co-localization experiments, HEK293 or MEF cells were plated and transfected as described above with Arr3-GFP, COX-2, and/or L-PGDS-HA where indicated. Forty-eight hours later, 2 ϫ 10 5 cells were transferred onto coverslips coated with 0.1 mg/ml poly-L-lysine (Sigma) and further grown overnight. Cells were washed with PBS, and where in-dicated, cells were incubated for 15 min at 37°C in the presence of 5 M PGH 2 . Cells were then fixed with 4% paraformaldehyde in PBS for 10 min at room temperature, washed again with PBS, permeabilized for 20 min with 0.1% Triton X-100 in PBS, and blocked with 0.1% Triton X-100 in PBS containing 5% nonfat dry milk for 30 min at room temperature. Cells were then incubated with primary antibodies diluted in blocking solution for 60 min at room temperature. Cells were washed twice with PBS, blocked again with 0.1% Triton X-100 in PBS containing 5% nonfat dry milk for 30 min at room temperature, and incubated with appropriate secondary antibodies diluted in blocking solution for 60 min at room temperature. The cells were washed twice with permeabilization buffer and twice with PBS, and coverslips were mounted using Vectashield Mounting Medium (Vector Laboratories). Confocal microscopy was performed using a scanning confocal microscope (FV1000; Olympus) coupled to an inverted microscope with a 63ϫ oil immersion objective (Olympus), and images were processed with IMAGE-PRO PLUS 6.0. For inhibitory assays with SeCl 4 , MG-63 cells were directly transferred onto coverslips coated with 0.1 mg/ml poly-L-lysine (Sigma) and further grown overnight. Cells were washed with PBS, and where indicated, cells were incubated for 15 min at 37°C in the presence of 100 M SeCl 4 . Cells were then processed as described above.
Image Analysis-Co-localization was assessed by the examination of merged images that showed co-localized regions. In addition to the merged image, a pixel fluorogram was generated using IMAGE-PRO software. The fluorogram is a twodimensional intensity histogram of the dual-color image, indicating the distribution of all pixels within the merged image as a scattergram. Pixel values of red and green components are displayed along the x axis and y axis, respectively, where dimmer pixels in the image are located toward the origin of the scatter plot, whereas brighter pixels are located farther out. A high degree of co-localization is revealed by a diagonal distribution (at 45°) of the dots on the fluorogram. A lack of co-localization is shown by two distinct populations with a minimal overlap of dots distributed toward the red and green axes, respectively (36).

L-PGDS Directly Interacts with Arrestin-3-L-PGDS was
identified as a potential Arr3 partner during a yeast two-hybrid screen of a human HeLa cell Matchmaker cDNA library using Arr3 as bait. As shown in Fig. 1A, strong growth on Trp Ϫ , Leu Ϫ , His Ϫ , and Ade Ϫ medium was present only in yeast strain pJ69-4␣ transformed with pAS2.1-Arr3 and pGADGH-L-PGDS constructs, whereas only minor growth was observed in yeast cells transformed with pAS2.1 and pGADGH-L-PGDS, indicating that L-PGDS is interacting with Arr3. No growth was observed in yeast cells transformed with pAS2.1-Arr3 and pGADGH-H-PGDS, the other PGD 2 synthase, suggesting that Arr3 specifically interacts with L-PGDS (data not shown). Strong growth was also observed on Trp Ϫ , Leu Ϫ , His Ϫ , and Ade Ϫ medium in yeasts transformed with pAS2.1-Arr2 and pGADGH-L-PGDS (data not shown), indicating that L-PGDS also interacts with Arr2. To further confirm the interaction between L-PGDS and Arr3 and to determine whether this interaction was direct, we performed an in vitro binding assay using purified recombinant L-PGDS fused to GST and GST-L-PGDS along with purified recombinant Arr3 fused to His (His 6 -Arr3). The results presented in Fig. 1B illustrate that Arr3 binds to glutathione-Sepharosebound GST-L-PGDS and not to glutathione-Sepharosebound GST, showing that L-PGDS can interact directly with Arr3.
To investigate whether L-PGDS and Arr3 interact in a cellular context, we performed immunoprecipitation experiments in HEK293 cells transfected with pcDNA3-Arr3-myc and pcDNA3-L-PGDS-HA in the presence or absence of PGH 2 , the L-PGDS substrate. Cell lysates were incubated with a HA-specific monoclonal antibody, and co-immunoprecipitated Arr3 was detected by Western blot analysis with a myc polyclonal antibody. Our results demonstrate that Arr3-myc was modestly co-immunoprecipitated with L-PGDS in the absence of PGH 2 ( Fig. 2A). However, the L-PGDS/Arr3 coimmunoprecipitation was strongly increased by the addition of PGH 2 ( Fig. 2A). Co-immunoprecipitation experiments were also performed on endogenous proteins in the MG-63 osteoblast cell line, which produces PGD 2 endogenously (11), contrary to HEK293 cells. Cell lysates were incubated with an L-PGDS-specific monoclonal antibody or with an isotypic normal mouse IgG as a negative control, and co-immunoprecipitated Arr3 was detected by Western blot analysis with an Arr3 polyclonal antibody. As shown in Fig. 2B, endogenous Arr3 co-immunoprecipitated with endogenous L-PGDS in MG-63 cells.
Co-localization of L-PGDS with Arr3-The possible colocalization of L-PGDS with Arr3 was studied by confocal microscopy first in HEK293 cells expressing Arr3-GFP and L-PGDS-HA. Arr3 and L-PGDS were both predominantly associated with vesicular structures in the cytoplasm, whereas Arr3 was also detected at the cell membrane (Fig. 3A). The

An Arrestin/L-PGDS Interaction
scarcity of yellow pixels in the merged image (Fig. 3Ac) and analysis of the corresponding pixel fluorogram (Fig. 3Ad) show a low degree of co-localization between Arr3 and L-PGDS in basal conditions. Because PGH 2 enhanced the L-PGDS/Arr3 interaction ( Fig. 2A), co-localization of the two proteins was also studied after the addition of the L-PGDS substrate. Adding PGH 2 significantly increased the number of co-localizing pixels between L-PGDS and Arr3, particularly in the perinuclear region (Fig. 3Ao) after 60 min of treatment.
Cellular distribution of endogenous Arr3 and L-PGDS was also examined in MG-63 osteoblasts. Intriguingly, localization of Arr3 and L-PGDS was predominant in the nuclear region in basal conditions in MG-63 cells (Fig. 3B, a and b) with a high degree of co-localization between the two proteins (Fig.  3B, c and d) that was not promoted by treatment with PGH 2 (data not shown). The specificity of Arr3 detection in the nuclear region was demonstrated by using a blocking peptide (supplemental Fig. S1Ae) and by studying the localization of   S1Ba). Contrary to HEK293 cells, MG-63 cells express COX enzymes and produce significant levels of PGH 2 (37) and PGD 2 (11) endogenously. Thus, we reasoned that inhibiting L-PGDS activity could perhaps modify the intracellular distribution of L-PGDS and Arr3 in MG-63 cells. Incubation of MG-63 cells with SeCl 4 , a PGDS inhibitor (38,39), resulted in a marked redistribution of Arr3 out of the nuclear region to cytoplasmic punctates (Fig. 3Be). The change in L-PGDS distribution was less radical after SeCl 4 treatment; L-PGDS remained largely in the perinuclear region but appeared more diffused (Fig. 3Bf). Consequently, co-localization of Arr3 with L-PGDS was drastically reduced in MG-63 cells when L-PGDS activity was inhibited (Fig. 3B, g and h).
The effect of COX-2 co-expression on the localization of L-PGDS was also studied in HEK293. Expression of COX-2 alone with L-PGDS did not alter the distribution of L-PGDS (compare Fig. 3Ab to Fig. 4Ab). However, when Arr3 was coexpressed, L-PGDS redistributed to the perinuclear region in presence of COX-2 (Fig. 4Af).
The role of Arr in L-PGDS distribution was also investigated in MEFs from wild-type mice and mice lacking Arr2, Arr3, or both. Confocal microscopy of wild-type MEFs that were transfected with an HA-tagged L-PGDS construct revealed prominent localization of L-PGDS in the perinuclear region and the nucleus (Fig. 4Bb). Interestingly, the bulk of L-PGDS-HA fluorescence was associated with intracellular vesicles scattered throughout the cell in MEFs-Arr double KO (Fig. 4Bk). Distribution of L-PGDS-HA in MEFs single Arr2 or Arr3 KO was intermediate to that observed for the wildtype MEFs and MEFs-Arr double KO (Fig. 4B, e and h). Moreover, L-PGDS was more abundant in the supernatant fraction depleted of nuclei in MEFs-Arr double KO than in wild-type MEFs, as assessed by Western blot (Fig. 4C).
L-PGDS-mediated PGD 2 Production Is Modulated by Arr3-Next, we were interested in determining if Arr3 modulated PGD 2 production by L-PGDS. We performed in vitro experiments using purified recombinant L-PGDS incubated with increasing amounts of purified Arr3 in the presence of the L-PGDS substrate PGH 2 , and PGD 2 production was measured by competitive ELISA. As shown in Fig. 5A, incubation of L-PGDS with an equal molar ratio of Arr3 brought a 60% increase in PGD 2 production when compared with L-PGDS alone, whereas a 2-5-fold excess of Arr3 failed to further increase it. We next explored whether Arr3 modulated L-PGDS-mediated PGD 2 production in cells. Wild-type MEFs and MEFs Arr2 KO, Arr3 KO, or Arr double KO were incubated with PGH 2 . Fig. 5B shows that PGD 2 production was decreased by ϳ25% in MEFs lacking Arr2 or Arr3 and by 40% in MEFs lacking both arrestins compared with wild-type MEFs.
To put this new function in a physiologically relevant context, we stimulated MEFs with IL-1␤, a pro-inflammatory cytokine that can stimulate prostanoid synthesis by stimulating the expression of COX-2 (40,41). PGD 2 production was decreased by roughly 50% in MEFs lacking both arrestins, in comparison with wild-type MEFs after stimulation with IL-1␤ (Fig. 5C). Expressing Arr2 or Arr3 in MEFs Arr double KO brought the level of PGD 2 production back to that observed in wild-type MEFs (Fig. 5C), indicating that arrestins contribute to increased PGD 2 production after stimulation with IL-1␤ in MEFs.
Determination of the L-PGDS Binding Domain on Arr3-To identify the Arr3 domain involved in the interaction with L-PGDS, we generated truncated mutants of Arr3 that were introduced in the pGEX-4-T1 vector (Fig. 6A). These truncated mutants were then purified as GST fusion proteins and used in GST pulldown assays with purified His 6 -tagged L-PGDS. Protein complexes were analyzed by immunoblotting with a His-specific monoclonal antibody to detect the presence of L-PGDS. Specific interactions with L-PGDS were observed for the GST-Arr3 and GST-Arr3 1-201 constructs, indicating that L-PGDS binds to the N-terminal region of Arr3 (Fig. 6A). Very little specific L-PGDS binding to the GST-Arr3 201-409 construct was observed, close to what was observed for GST alone, throughout the different experiments that were performed. Thus, we decided to further characterize the Arr3 1-201 domain involved in L-PGDS binding by generating truncated GST mutants of this N-terminal region of Arr3 (Fig. 6B) that were used in the L-PGDS pulldown assays. Immunoblotting of the binding reactions revealed that only the region of Arr3 composed of amino acids 56 -100 was able to bind to L-PGDS (Fig. 6B), suggesting that this domain is the major L-PGDS-interacting site on Arr3.
Arr3 Peptides Increase PGD 2 Production by L-PGDS-We then assessed whether amino acids 56 -100 of Arr3 were sufficient to enhance PGD 2 production by L-PGDS in vitro. L-PGDS-mediated PGD 2 production was measured in the presence of purified His 6 -Arr3 or -Arr3 amino acids 56 -100 proteins as described above. In these experiments, the addition of full-length Arr3 increased levels of PGD 2 measured by 115%, in comparison with L-PGDS alone (Fig. 6C). Interestingly, incubation of L-PGDS with Arr3 amino acids 56 -100 enhanced PGD 2 production by 145%, demonstrating that amino acids 56 -100 of Arr3 are sufficient to modulate PGD 2 production by L-PGDS in vitro.
As schematically represented in Fig. 7A, 4 overlapping peptides of 15 amino acids corresponding to the region composed of residues 56 -100 of Arr3 were synthesized. Their effects on L-PGDS-mediated PGD 2 production was measured in vitro as above. Peptides Arr3 76 -90 and Arr3 86 -100 increased PGD 2 production by L-PGDS by roughly 100% in vitro, whereas the effects of the Arr3 amino acids 56 -70 and Arr3 amino acids 65-79 peptides were less significant (Fig. 7A). The Arr3 86 -100 peptide and its scrambled sequence were then synthesized in fusion with the YGRKKRRQRRRGGG transduction peptide of the HIV TAT protein (Fig. 7B) to facilitate peptide entry into MG-63 cells. The cells were preincubated with the TAT peptides for 1 h, and PGD 2 production was measured 15 min after the addition of PGH 2 . Quite interestingly, PGD 2 production was roughly twice as high in MG-63 cells treated with the TAT-Arr3 86 -100 peptide compared to cells treated with the control TAT-scrambled peptide (Fig. 7B).

DISCUSSION
Prostaglandin synthases recently garnered much interest because of their therapeutic potential, a fact that was rein-forced by the clinical problems associated with the long term use of COX inhibitors. However, little is known about the regulation of these enzymes and their interacting partners. In

An Arrestin/L-PGDS Interaction
the present study we have demonstrated that one of the two synthases that produce PGD 2 , L-PGDS, interacts with the scaffold protein Arr3.
A direct interaction between L-PGDS and Arr3 was suggested by our yeast two-hybrid screen and confirmed by the pulldown assays using purified recombinant proteins. Coimmunoprecipitation studies in HEK293 cells, which do not express COX enzymes (42), showed that L-PGDS and Arr3 interacted weakly in basal conditions. This interaction was, however, robustly promoted by the addition of the L-PGDS substrate, PGH 2 . Corroborating these results, confocal microscopy revealed weak co-localization between L-PGDS and Arr3 under basal conditions in HEK293 cells, but this co-localization was strongly increased in the perinuclear region after treatment with PGH 2 . Expression of COX-2 in HEK293 cells did not affect the intracellular distribution of L-PGDS. However, co-expression of Arr3 with COX-2 and L-PGDS resulted in localization of L-PGDS in the perinuclear region. Furthermore, confocal immunofluorescence microscopy and fractionation experiments showed that L-PGDS was more abundant in the cytoplasm of MEFs Arr double KO cells than in wild-type MEFs. Importantly, co-immunoprecipitation and co-localization between the two endogenous proteins was demonstrated in PGD 2 -producing MG-63 cells. Surprisingly, Arr3 demonstrated strong immunofluorescence staining in the nuclear region where it co-localized with L-PGDS in these cells. Inhibition of PGDS activity with SeCl 4 (38,39) resulted in a significant redistribution of Arr3 to the cytoplasm, whereas L-PGDS redistribution was subtle but still more diffused than in untreated cells. Taken altogether, our results suggest that 1) arrestin is involved in localizing L-PGDS in the perinuclear region when PGH 2 is present, either when added exogenously or when produced by COX-2, and 2) co-localization between L-PGDS and Arr3 in the perinuclear region appears to be regulated by L-PGDS activity.  L-PGDS has been reported to localize to the rough endoplasmic reticulum, the outer nuclear membrane, the Golgi apparatus, lysosomes, multivesicular bodies, and endocytic vesicles of various cell types (43,44). Arr3 is known to be distributed in the cytoplasm and associated to plasma and endosomal membranes. Arr3 is predominantly found out of the nucleus due to the presence of its two-leucine nuclear export signal (45). However, Neuhaus et al. (46) reported Arr3 to be localized in the nucleus of HEK293 cells and mature spermatozoa. COX-1 is located in the endoplasmic reticulum and perinuclear membranes, whereas COX-2 resides predominantly in the perinuclear envelope (47). Our results show colocalization of L-PGDS and Arr3 at the nuclear envelope and the perinuclear region (in the endoplasmic reticulum in HEK293 cells; data not shown). The limit of resolution of our confocal microscopy does not allow us to ascertain whether or not there is Arr3-or L-PGDS-associated immunofluorescence within the nucleus. The difference in Arr3 and L-PGDS localization between HEK293 and MG-63 cells under basal conditions could be due to the presence of endogenous COX, L-PGDS, and PGH 2 in MG-63 cells. Interestingly, Arr3 coexpression caused perinuclear localization of L-PGDS when COX-2 was expressed in HEK293 cells. We propose that Arr3 participates in localizing L-PGDS in the vicinity of COX enzymes where it would get access to its COX-derived substrate, PGH 2 , resulting in greater production of PGD 2 .
A cellular role was demonstrated for Arr3 in L-PGDS-mediated PGD 2 production in MEFs. MEFs lacking both Arr2 and Arr3 showed a ϳ40% decrease in PGD 2 production compared with wild-type MEFs after the addition of PGH 2 . Similarly, stimulation with the pro-inflammatory cytokine IL-1␤ resulted in ϳ50% less PGD 2 produced in MEF arrestin double KO cells than in wild-type MEFs. Transfection of Arr2 or Arr3 in MEF arrestin double KO cells brought the PGD 2 production back to wild-type MEFs levels in response to IL-1␤ treatment. These results suggest that arrestin plays a significant, but not essential role in L-PGDS-mediated PGD 2 production. Its function may reside in modulating PGD 2 production in response to particular stimuli. PGD 2 has been reported to be pro-inflammatory but also to be involved in the resolution of inflammation when produced in high concentrations (2,8,10). Further work will be necessary to determine whether arrestin-mediated increase in PGD 2 production participates in the development and/or resolution of IL-1␤-in- The binding assays were carried out using purified glutathione-Sepharosebound GST-truncated mutants of the full-length (A) or the N-terminal region (B) of Arr3 (represented schematically), which were incubated with purified recombinant His 6 -L-PGDS. The binding of L-PGDS was detected by immunoblotting (IB) using a His 6 -specific monoclonal antibody, and the GST fusion proteins present in the binding reaction were detected using a GST-specific polyclonal antibody as described under "Experimental Procedures." Blots shown are representative of three independent experiments. C, purified His 6 -L-PGDS was incubated with pRSETa flow-through (control, 100%) or equivalent molar ratio of purified His 6 -Arr3 or His 6 -Arr3 amino acids 56 -100, and PGD 2 production was stimulated by the addition of 0.5 M PGH 2 (L-PGDS substrate) for 1 min. Reactions were stopped with 0.4 mg/ml SnCl 2 , and PGD 2 was then measured with commercial enzyme-linked immunoassays as described under "Experimental Procedures." All values are the mean Ϯ S.E. from at least three separate experiments. *, p Ͻ 0.05. duced inflammation. Our data with IL-1␤ establish a new role for arrestin in linking inflammatory cytokine cell surface receptor signaling to intracellular prostaglandin synthesis.
Our study not only indicates that Arr3 could facilitate PGD 2 production by bringing L-PGDS in the vicinity of COX enzymes and, thus, of its PGH 2 substrate but that it can also increase L-PGDS enzymatic activity. Experiments performed with purified proteins in vitro showed that Arr3 stimulated L-PGDS to produce PGD 2 optimally at a molar ratio of 1:1. Identification of residues 56 -100 of Arr3 as the binding site for L-PGDS allowed us to generate two Arr3 peptides composed of amino acids 76 -90 and 86 -100 that nearly doubled PGD 2 production by L-PGDS in vitro. Interestingly, this region of Arr3 is similar in Arr2, which also interacts with and regulates L-PGDS but differs in visual arrestins and is located for the most part on a loop region that is easily accessible for protein-protein interactions (48). It was shown that the proline-rich region 88 -96 in non-visual arrestins is partly responsible for binding to the SH3 domain of Src family kinases such as c-Src (49) and Hck (50). This raises the interesting possibility that the L-PGDS-Arr3 functional interaction could be regulated by members of the Src family kinases and associated cell signaling pathways.
We were enthused when L-PGDS-mediated PGD 2 production in MG-63 cells was increased almost 2-fold by the TAT-Arr3 86 -100 fusion peptide compared with the control TAT-Arr3 scrambled peptide. One goal of our laboratory is to characterize the interacting partners of prostaglandin synthases to further our understanding of these critical enzymes but also to identify the interacting domains to derive peptides that could potentially modulate prostaglandin production. The data presented here show that this approach was successful for L-PGDS and should be applicable to other prostaglandin synthases. Increasing PGD 2 levels could be beneficial in a number of conditions. For example, PGD 2 is an endogenous sleep-promoting substance (51), and administration of selenium compounds markedly suppresses sleep (38,52). Higher levels of PGD 2 could help to resolve insomnia (53) and inflammation (54). The importance of PGD 2 in the induction and maintenance of remission from ulcerative colitis was recently highlighted (10). Its best-studied metabolite, 15d-PGJ2, stimulates peroxisome proliferator-activated receptor-␥, which in turn mediates anti-inflammatory effects (55)(56)(57)(58)(59). Our group also demonstrated that PGD 2 has a positive influence on bone anabolism (11,12). Thus, modulating the interaction between arrestins and L-PGDS could turn out to have great therapeutic value toward inflammatory and other diseases.
In conclusion, we report an L-PGDS-Arr3 interaction that modulates PGD 2 production. This is the first characterization of an interacting partner for L-PGDS and the identification of a new function for non-visual arrestins. Our findings also indicate that detailed analysis of the binding domains between a prostaglandin synthase and an interacting partner can lead to the generation of modulators of prostaglandin synthesis.