Down-regulation of Cyclooxygenase-2 by the Carboxyl Tail of the Angiotensin II Type 1 Receptor*

Background: The levels of the pro-inflammatory enzyme COX-2 require tight regulation. Results: The carboxyl tail of Angiotensin II type 1 receptor (AT1) enhances COX-2 degradation. Conclusion: We identified a novel mechanism for COX-2 regulation that is independent of receptor activation. Significance: The tail sequence of AT1 may serve as a basis for design of novel therapeutic agents that degrade COX-2. The enzyme cyclooxygenase-2 (COX-2) plays an important role in the kidney by up-regulating the production of the vasoconstrictor hormone angiotensin II (AngII), which in turn down-regulates COX-2 expression via activation of the angiotensin II type 1 receptor (AT1) receptor. Chemical inhibition of the catalytic activity of COX-2 is a well-established strategy for treating inflammation but little is known of cellular mechanisms that dispose of the protein itself. Here we show that in addition to its indirect negative feedback on COX-2, AT1 also down-regulates the expression of the COX-2 protein via a pathway that does not involve G-protein or β-arrestin-dependent signaling. Instead, AT1 enhances the ubiquitination and subsequent degradation of the enzyme in the proteasome through elements in its cytosolic carboxyl tail (CT). We find that a mutant receptor that lacks the last 35 amino acids of its CT (Δ324) is devoid of its ability to reduce COX-2, and that expression of the CT sequence alone is sufficient to down-regulate COX-2. Collectively these results propose a new role for AT1 in regulating COX-2 expression in a mechanism that deviates from its canonical signaling pathways. Down-regulation of COX-2 by a short peptide that originates from AT1 may present as a basis for novel therapeutic means of eliminating excess COX-2 protein.

Prostaglandins are bioactive lipids that function as major regulators of cardiovascular homeostasis. They are derived from a common H 2 prostaglandin endoperoxide (PGH 2 ), a metabolite of arachidonic acid (AA) 2 that is formed by the ratelimiting enzyme cyclooxygenase (COX). COXs exist in two main isoforms, COX-1 and COX-2 that reside on the luminal surfaces of the endoplasmic reticulum and the inner and outer membranes of the nuclear envelope (1). Both isoforms display similar catalytic mechanisms but differ in their expression pat-terns. COX-1 is expressed almost ubiquitously and fulfills many housekeeping functions, while COX-2 is usually absent from most tissues but undergoes a rapid and transient increase of expression by a broad range of pathological stimuli (2). As such, inhibition of its activity by non-steroidal anti-inflammatory drugs (NSAIDs) is one of the most common therapeutic targets for treatment of inflammation. However, COX-2 is also normally expressed in some tissues where it has some important physiological roles. In the kidney, the products of COX-2 catalysis increase the generation of the vasoconstricting hormone angiotensin II (AngII), which in turn down-regulates the expression of COX-2, mainly through activation of the angiotensin II type 1 receptor (AT 1 ) (3,4).
The AT 1 receptor belongs to the super-family of G proteincoupled receptors (GPCRs) that relay signals by activating heterotrimeric G proteins, followed by second-messenger-mediated intracellular signaling. Studies in the last decade showed that AT 1 signals through two distinct signaling pathways, whereby binding of ligand initiates activation of G proteins, but quickly thereafter switches to ␤ arrestin-mediated, G proteinindependent pathways (reviewed in (5)). Coupling of AT 1 to G proteins is mediated primarily through a DRY motif located in the third intracellular loop of the receptor. Mutation of this motif abrogates coupling to G proteins but ␤ arrestin recruitment and activation of the ERK MAP kinase pathway remains intact (6). In contrast, the absence of certain phosphorylation sites in the carboxyl tail (CT) of AT 1 prevents ␤ arrestin-mediated signaling while preserving the G-protein pathway (7).
Elevated levels of COX-2 are characteristic of many types of chronic ailments suggesting that tight regulation of its levels is critical for normal physiological function. Whereas the signaling cascades that lead to the induction of COX-2 are well-studied (8), there is much less information about the regulatory pathways that mediate its degradation. In the absence of its major substrate AA, COX-2 undergoes continuous turnover by shuttling from the endoplasmic reticulum to the cytosol via the ER-associated degradation (ERAD) pathway, where it is subsequently degraded by the proteasome (9). Degradation of COX-2 in the proteasome is preceded by its polyubiquitination (10), and was recently shown to be facilitated by caveolin-1 (11) and also through its interaction with the GPCRs, prostaglandin E 1 (EP 1 ), and ␤ 1 adrenergic (␤ 1 AR) receptors (12,13). Here we set to explore whether, in addition to the known negative feedback loop between angiotensin II and COX-2, there is an additional mechanism for down-regulating COX-2 expression by AT 1 , and to identify the domains of the receptor that mediate this effect.
Cell Culture and Transfection-HEK-293 cells were grown in Eagle's MEM media, supplemented with 10% fetal bovine serum and 100 units/ml penicillin and streptomycin. Transient transfections were carried out in subconfluent (70 -80%) monolayers using PolyJet (SignaGen Laboratories) at a ratio of 1:3 cDNA: PolyJet, according to the manufacturer's instructions. All samples contained the same amount of total cDNA.
His-Myc tagged AT 1 -CT was cloned using the gBlocks Gene Fragment: AACGGCGGATCCACCATGGCCAAGTCCCA-CTCAAGCCTGTCTACGAAAATGAGCACGCTTTCTTA-CCGGCCTTCGGATAACATGAGCTCATCGGCCAAAAA-GCCTGCGTCTTGTTTTGAGGTGGAGAAGCTTGGC-CTT. The fragment was designed to carry BamHI and HindIII restriction sites (bold) and a Kozak translation initiation sequence (underline). Cloning was performed using the standard manufacturer's protocol. All constructs were confirmed by restriction digestion analysis and sequenced at the core sequencing facilities of the Technion Israel Institute of Technology and Hylabs (Rehovoth, Israel).
Immunoprecipitation and Immunoblotting-Monolayers in 100-mm culture dishes were washed twice with ice-cold PBS and lysed exactly as we had done before (13). Nitrocellulose membranes containing the immuno-complexes or total cell lysate proteins were incubated with primary antibodies at a dilution of 1:500 (COX-2 and HA), and 1:250 for Ub. Proteins were visualized by a WesternBright ECL (Advansta, CA) and quantified using a CCD camera and Quantity One software (XRS, Bio Rad).
Radioimmunoassay-Cells were plated in 12-well dishes and transfected with COX-2, with either empty plasmin (pcDNA3.1) or AT 1, as indicated above. Radioimmunoassays were performed in triplicates exactly as described (13). Protein levels were measured in each well, and PGE 2 ng/mg protein was determined. Samples in each experiment were normalized against the controls that contained only COX-2, and data are presented as fold change in PGE 2 production obtained from different experiments.
Flow Cytometry-Cells were washed twice with PBS, and resuspended in 150 -200 l of PBS for cytometric analysis. All experiments were performed in triplicates. The samples were analyzed using BD FACSCanto II flow cytometer with DACS-Diva software (BD Biosciences, San Jose, CA), as described (13).
Microscopy-Cells were grown on 13-mm glass coverslips. Following transfection, cells were fixed with 4% paraformaldehyde, washed with PBS and blocked in PB buffer (1% BSA and 0.1% Triton X-100) for 5 min. Samples were then incubated with anti-COX-2 (1:200) for 1 h, washed three times with PBS, and incubated with Alexa Fluor 647 donkey anti-goat IgG (1:200) for 1 h. Following an additional three washes with PBS, samples were mounted onto glass slides using Mowiol (Sigma Aldrich) and visualized under an ApoTome.2 laser scanning confocal microscope (Zeiss) at a 63ϫ magnification. All images were acquired using the same exposure conditions.
Statistical Analysis-Experiments shown are mean Ϯ S.E. for data averaged from at least three independent experiments. To determine statistical significance, Student's t test or one-way ANOVA were used. Post-hoc analysis was performed with Tukey multi-comparison test when appropriate. p values Ͻ 0.05 were considered significant. Analyses were done using GraphPad Prism 5 software.

The AT 1 Receptor Down-regulates the Expression of COX-2-
To test whether expression of the AT 1 receptor affects COX-2, HEK293 cells were co-transfected with COX-2 together with either empty plasmid or the receptor, and analyzed for the ability of COX-2 to generate PGE 2 . Since HEK 293 cells do not express detectable amounts of either COX isoform, the data reflect only the activity of transfected COX-2 (13). As depicted in Fig. 1A, co-transfection of COX-2 with AT 1 at a ratio of 1:5 reduced PGE 2 secretion by nearly half. To find out whether this decrease is due to diminished COX-2 levels, we used flow cytometry to analyze the levels of YFP-tagged COX-2 in the absence or presence of AT 1 . Co-expression of both proteins under the same conditions as the RIA experiment resulted in a marked 80% reduction in COX-2 expression (Fig. 1B), suggesting that most of the reduction in PGE 2 secretion may be attributed to a reduction in COX-2 levels. Consistent with this, immunofluorescence microscopy showed that compared with co-expression with an empty plasmid, the expression of YFP-COX-2 is severely down-regulated in the presence of GFP-AT 1 (Fig. 1C).
To test whether AT 1 has a similar effect on the expression of COX-1, we expressed either COX-1 or COX-2 together with increasing amounts of CFP-AT 1 . The total DNA levels were kept the same at all times. As shown in Fig. 1D, while increasing the levels of AT 1 caused a marked drop in COX-2 levels, it did not have a significant effect on the levels of COX-1.
Down-regulation of COX-2 by AT 1 Is Not Mediated via Classical Signaling Pathways-We next sought to determine whether the effect of AT 1 on COX-2 is mediated via its classical signaling pathways (6). First, we expressed COX-2, AT 1 , or both in HEK 293 cells, stimulated them with the AT 1 ligand AngII, and measured COX-2 levels and phosphorylation of the ERK MAP kinase as an indication for receptor activation. Cells transfected with COX-2 alone did not show a response to AngII, indicating that they do not express significant amounts of endogenous AT 1 (Fig. 2A, first two lanes). Expression of AT 1 alone elicited a marked response to AngII stimulation, as detected by activation of ERK ( Fig. 2A, two middle lanes). Coexpression of COX-2 together with AT 1 significantly lowered COX-2 expression, and this phenomenon was not affected by the presence of AngII ( Fig. 2A, last two lanes). Time response experiments in the presence of AT 1 alone or AT 1 and COX-2 showed that there were no differences in the kinetics of ERK activation by AngII, both showing a peak response at 5-10 min of stimulation (Fig. 2B). However, ERK activation by AT 1 was dampened in the presence of COX-2 (Fig. 2C). The same reduction in the ability of AT 1 to stimulate COX-2 was observed using the catalytically inactive mutant G533A COX-2, suggesting that COX-2 protein interferes with signaling in this pathway in a manner that is independent of its catalytic activity.
To further exclude the involvement of possible downstream signaling by AT 1 in its effect on COX-2, we inhibited PKC, the major downstream protein kinase activated by AT 1 . Cells were transfected with COX-2 in the presence or absence of AT 1 and treated with different concentrations of the PKC inhibitor GFX for the full duration of transfection. As depicted in Fig. 2D, this treatment did not reverse the reduction caused by AT 1 . A similar result was obtained using OAG, another potent inhibitor of PKC (data not shown).
Coupling of the AT 1 receptor to G-proteins and ␤ arrestin is mediated via specific motifs on the receptor. Thus, G protein coupling to AT 1 is abolished by mutating the highly conserved DRY sequence to AAY, and association with ␤ arrestin is defec- tive in AT 1 with a TSTS/A substitutions in the CT of the receptor (6,14). To test whether these motifs are involved in the effect of AT 1 on COX-2, we tagged the mutant receptors with cyan fluorescent protein (CFP) and co-expressed them together with YFP-COX-2 (Fig. 2E). Flow cytometry measurements showed that the wild type and mutant receptors expressed to similar extents (Fig. 2E, right panel), and both significantly lowered COX-2 expression. AT 1 Promotes Degradation of COX-2 via the Proteasome-Our previously published data showed that the prostaglandin EP 1 receptor down-regulates COX-2 expression by enhancing its ubiquitination and subsequent proteasomal degradation (13). To test whether proteasomal degradation is involved in the effect of AT 1 on COX-2, cells expressing YFP-tagged COX-2, alone or with CFP-AT 1 , were treated overnight with or without the specific proteasome inhibitor MG132. Blockade of proteasomal activity abolished the effect of AT 1 , resulting in more than double the amount of COX-2 (Fig. 3A). Treatment with MG132 also caused a parallel reduction in the levels of CFP-AT 1 (Fig. 3B). Similar results were observed using untagged proteins (Fig. 3C), thus excluding the possibility that the observed effect is an artifact of the fluorescent tags. A doseresponse experiment with increasing concentrations of MG132 showed that COX-2 recovery was observed as soon as the levels of the receptor began falling at concentrations as low as 0.1 M MG132 (Fig. 3D).
We next tested whether COX-2 and AT 1 interact with each other. For this, cells were transfected with each protein alone or together, and samples were subject to immunoprecipitation. To enable detection of a possible interaction, we used a transfection ratio of 1:1 COX-2: AT 1 that was found in dose-titration experiments to have a minimal effect on COX-2 expression (Fig. 1D). As shown in Fig. 4, A and B, only cells that expressed both proteins showed the reciprocal protein in coprecipitates.

The Carboxyl Tail of AT 1 Down-regulates COX-2
Next, we measured the levels of ubiquitinated COX-2 in the presence or absence of AT 1 . COX-2 was immunoprecipitated from all samples, and membranes were probed first for ubiquitination levels and then for the presence of COX-2 and AT 1 . Under conditions of 1:1 co-expression AT 1 did not cause a significant reduction in COX-2 but the levels of its ubiquitination were elevated compared with those of COX-2 alone (Fig. 4C).
The Effect of AT 1 on COX-2 Is Mediated via Its Carboxyl Tail (CT)-The CT of most GPCRs has a critical role in their interactions with intracellular proteins. To determine whether the effect of AT 1 on COX-2 involves its CT, we overexpressed COX-2 with either a wild type receptor or a truncated receptor mutant that lacks its entire cytosolic tail (⌬324; (7)). As demonstrated by both Western blot (Fig. 5A) and flow cytometry (Fig.  5B) analyses, the effect of AT 1 onCOX-2 was completely abolished in the absence of its CT.
We next reasoned that if the CT is required for down-regulation of COX-2 expression by AT 1, then the inhibitory effect of AT 1 on COX-2 may be mimicked by the CT amino acid sequence itself. To test this, we appended the amino acid sequence (325-359) of the CT to CFP or c-Myc and followed its effect on COX-2 expression. As depicted by fluorescent microscopy (Fig. 6A) and Western blotting (Fig. 6B) co-expression of the CT of AT 1 together with COX-2 in HEK 293 cells, significantly down-regulates the expression of the latter. To test whether wild type AT 1 or the CT lower the levels of endoge-   nously expressing COX-2, we tested their effect on NIH3T3 fibroblasts that express endogenous COX-2 following serumstimulation (15). Cells were transfected with empty vector, wild type AT 1 , or CT-Myc and stimulated 1 day later with 20% serum. As shown in Fig. 6C, both wild type and CT lowered endogenous COX-2 expression in these cells, in a similar manner to the effect they had on the HEK 293 cells that were transfected with COX-2. Flow cytometry analysis showed both wild type AT 1 and CT cause a reduction of ϳ50% in COX-2 expression (Fig. 6D).

DISCUSSION
The data presented herein show that AT 1 down-regulates COX-2 expression in a mechanism that is not mediated by classical signaling pathways. Agonist stimulation of AT 1 promotes its coupling to G␣q, thereby initiating PLC-dependent activation of PKC (6). Immediately thereafter, the receptor is phosphorylated by G protein-coupled receptor kinases (GRKs) on distinct serine/threonine site located on its carboxyl terminus (14), thereby initiating a second wave of ␤ arrestin-dependent signaling (6). However, our data show that AngII-mediated activation of the receptor, or inhibition of PKC activity, do not reverse its effect on COX-2 expression. Moreover, AT 1 mutants that are defective in their ability to engage with G proteins (DRY/AAY) or ␤ arrestin (TSTS/A) do not cause a significant reversal of the receptor on COX-2 expression. A role for ␤ arrestin cannot be ruled out completely because although the ⌬324 AT 1 mutant that lacks the CT of the receptor has no effect on COX-2, it was shown to maintain a certain degree of phosphorylation-independent ␤ arresting recruitment (6). Nonetheless, since AT 1mediated decrease in COX-2 is observed in the absence of any ligand, it is more likely that the CT of AT 1 interacts with other currently unknown scaffold proteins that promote degradation of COX-2.
One of the main findings of this study is that the CT sequence of AT 1 mimics the effect of the full receptor. This suggests that the tail region has a major role in down-regulating COX-2 but whether it interacts with the same machinery as the full receptor remains to be determined. The tail region of AT 1 contains motifs that bind different molecules such as proteins of the JAK/STAT pathway (16), and ATRAP (17) that may be involved in regulating COX-2 expression. Other GPCR-associated protein candidates that may regulate COX-2 localization and subsequent degradation include GTPase ARF4 that binds to a sorting signal on the CT of rhodopsin, and the E3 ligase Nedd4 that binds to the ␤ 2 adrenergic receptor and promotes its degrada- tion (18). Identification of additional proteins that are involved in the mechanism of COX-2 down-regulation by GPCRs requires further investigation and is ongoing. As opposed to its highly inducible nature in most tissues, COX-2 is constitutively expressed in the cortex of the mammalian kidney, particularly in the macula densa and the thick ascending limb of Henle (19), where it generates prostaglandins that raise the levels of renin. Elevated renin (e.g. due to salt depletion) or inhibition of the angiotensin-converting enzyme (ACE) cause a significant increase in COX-2 expression thus constituting positive feedback loop between renin and COX-2 (19,20). In contrast, the end product of renin, AngII, negatively regulates the expression of COX-2 (3,4). This effect was shown to involve receptor signaling since administration of ACE inhibitors or angiotensin receptor blockers to rodents in vivo caused a marked elevation in COX-2 expression (3). Interestingly however, mice with a genetic depletion of AT 1 (Agtr1a Ϫ/Ϫ , Agtrb Ϫ/Ϫ ) also display significantly higher levels of COX-2 in their macula densa, thus providing support to our hypothesis that the actual presence of the AT 1 receptor may be required to keep COX-2 expression at bay.
In summary, we found that the AT 1 receptor plays an important role in facilitating COX-2 degradation, thus constituting an additional feedback loop that does not depend on classical signaling pathways. These findings are in line with our previous work that demonstrated a similar effect for EP 1 and ␤ 1 adrenergic receptors in accelerating COX-2 degradation by enhanced ubiquitination (12,13,21). Together these studies suggest that the mechanism of receptor-mediated regulation of COX-2 expression is common to other GPCRs, thus endowing them with an additional function that deviates from accepted signaling paradigms. We further posit that this type of regulation may constitute a physiological means of controlling normal COX-2 turnover, and that pathological conditions that involve changes in the expression of GPCRs may affect COX-2 as well. Lastly, the ability of a short amino acid sequence of the CT to downregulate COX-2 expression may present a basis for novel therapeutic approach for eliminating excess COX-2 protein.