Angiotensin II activates NF-κB through AT_1A receptor recruitment of β-arrestin in cultured rat vascular smooth muscle cells

Abstract

Activation of the angiotensin type 1A receptor (AT_1AR) in rat aorta vascular smooth muscle cells (RASMC) results in increased synthesis of the proinflammatory enzyme cyclooxygenase-2 (COX-2). We previously showed that nuclear localization of internalized AT_1AR results in activation of transcription of the gene for COX-2, i.e., prostaglandin-endoperoxide synthase-2. Others have suggested that ANG II stimulation of COX-2 protein synthesis is mediated by NF-κB. The purpose of the present study was to examine the interrelationship between AT_1AR activation, β-arrestin recruitment, and NF-κB activation in the ability of ANG II to increase COX-2 protein synthesis in RASMC. In the present study we utilized RASMC, inhibitors of the NF-κB pathway, β-arrestin knockdown, radioligand binding, immunoblotting, and immunofluorescence to characterize the roles of AT_1AR internalization, NF-κB activation, and β-arrestin in ANG II-induced COX-2 synthesis. Ro-106-9920 or parthenolide, agents that inhibit the initial steps of NF-κB activation, blocked ANG II-induced p65 NF-κB nuclear localization, COX-2 protein expression, β-arrestin recruitment, and AT_1AR internalization without inhibiting ANG II-induced p42/44 ERK activation. Curcumin, an inhibitor of NF-κB-induced transcription, blocked ANG II-induced COX-2 protein expression without altering AT_1AR internalization, ANG II-induced p65 NF-κB nuclear localization, or p42/44 ERK activation. Small interfering RNA-induced knockdown of β-arrestin-1 and -2 inhibited ANG II-induced p65 NF-κB nuclear localization. In vascular smooth muscle cells, internalization of the activated AT_1AR mediated by β-arrestins activates the NF-κB pathway, producing nuclear localization of the transcription factor and initiation of COX-2 protein synthesis, thereby linking internalization of the receptor with the NF-κB pathway.

angiotensin II (ANG II) stimulates acute and chronic vascular responses, resulting in physiological and pathophysiological sequelae. Activation by ANG II of the angiotensin type 1A (AT_1A) receptor (AT_1AR), the predominantly expressed isoform, especially in vascular smooth muscle, produces G protein-mediated intracellular signals, including activation of tyrosine kinases (3, 12, 19), nuclear signaling, and protein synthesis (extracellular matrix protein with vascular remodeling), events that are central to development of atherosclerosis and hypertension (see Ref. 25 for review).

Cyclooxygenase-2 (COX-2), an inducible enzyme responsible for metabolizing arachidonic acid to PGG₂, has been implicated in many pathologies, including pain, inflammation, and cancer (see Ref. 13 for review). Previously, we showed that internalization and nuclear membrane localization of the activated AT_1AR result in transcriptional activation and protein synthesis of COX-2 (20–22). Others have shown in a vascular smooth muscle cell culture model that ANG II induces transcription for COX-2 via involvement of NF-κB and cytoplasmic kinases, including Pyk2, MEKK4, and p38 (8, 15, 23).

The DNA-binding protein and activator of transcription NF-κB is a major mediator of inflammatory responses. NF-κB consists of a family of transcription factors, including NF-κB1 (p50, derived from the p105 precursor), NF-κB2 (p52 and the p100 precursor), p65 (RelA), c-Rel, and RelB. It is the p50/p65 dimers that bind to DNA and activate transcription. Additional auxiliary nuclear proteins may interact with the dimers, allowing for specificity of high-affinity DNA binding. The p50/p65 dimer is held in an inactive form within the cytoplasm through binding with one of the IκB family of inhibitory proteins (α, β, and γ). Activation of the NF-κB pathway, initiated by specific receptor activation, results in phosphorylation of IκBα through the actions of a specific IκB kinase, IKK. IκBα is subsequently ubiquitinated and targeted for proteosomal degradation, resulting in release of the p50/p65 dimer from inhibition and movement of the dimer into the nucleus and subsequent binding to DNA. This pathway is commonly referred to as the classical or canonical NF-κB activation pathway (27).

The ability of ANG II to increase the expression of IL-6, monocyte chemoattractant peptide 1, insulin-like growth factor I receptors, and other inflammatory mediators from various tissues is mediated through NF-κB activation (1, 18). As mentioned above, COX-2 expression in vascular smooth muscle cells initiated by exposure to ANG II is also mediated by NF-κB (15). In addition to initiation of signaling through G proteins, newer evidence shows that G protein-coupled receptors (GPCRs) can activate specific G protein-independent secondary signaling via mediation of β-arrestin-scaffolded intermediaries, such as p42/44 ERK and JNK (2). Since our previous data indicate that internalization of the AT_1AR is necessary for induction of COX-2 protein synthesis and others suggest that NF-κB is mediating COX-2 expression, we hypothesize that β-arrestin-mediated intracellular trafficking of the AT_1AR activates NF-κB nuclear localization and subsequent COX-2 protein expression.

MATERIALS AND METHODS

Reagents.

The NF-κB inhibitors parthenolide and curcumin were obtained from Axxora (San Diego, CA) and Ro-106-9920 from Tocris Biosciences (Ellisville, MO). Stock solutions (10 mmol/l) of the inhibitors were made using DMSO as vehicle. ANG II was obtained from Sigma-Aldrich (St. Louis, MO) and dissolved in deionized water as a 1 mmol/l stock solution. All tissue culture reagents were obtained from Invitrogen (Carlsbad, CA).

Cell culture.

Rat aortic vascular smooth muscle cells (RASMC) were prepared from isolated segments of rat thoracic aorta, as previously described (26). Aortas were removed from 150-g male Sprague-Dawley rats after exsanguination under isoflurane anesthesia, cut into ∼5-mm segments, and placed cut-edge-down onto plastic cell culture petri dishes and covered with a drop of Dulbecco's modified Eagle's medium (4,500 mg/l glucose) containing 20% FBS and 1% antibiotic-antimycotic-amphotericin B (Fungizone). After ∼5 days, explanted cells were maintained in high-glucose DMEM supplemented with 10% FBS and 1% antibiotic-antimycotic-amphotericin B and used between passages 3 and 12. Animal procedures were approved by the Medical University of South Carolina Institutional Animal Care and Use Committee.

Human embryonic kidney (HEK-293) cells (American Type Culture Collection) stably expressing wild-type AT_1AR or AT_1AR/green fluorescent protein (AT_1AR/GFP) constructs were maintained as previously described (21) using MEM supplemented with 10% FBS, 1% antibiotic-antimycotic-amphotericin B, and G418 (400 μg/ml).

Radioligand binding assays.

Binding studies employing ¹²⁵I-ANG II were performed as previously described (22). For receptor internalization studies, confluent monolayers of RASMC were exposed to ¹²⁵I-ANG II (∼200,000 cpm; PerkinElmer) at 4°C in binding buffer (50 mmol/l Tris, pH 7.4, 100 mmol/l NaCl, 5 mmol/l MgCl₂, 5 mmol/l KCl, and 1% BSA). Subsequently, cells were treated as follows. Group A was washed with cold saline buffer to remove unbound radioligand and then solubilized in 0.1% SDS and 0.1 mol/l NaOH for determination of total receptors available. Group B was acid-washed (150 mmol/l NaCl and 50 mmol/l glycine, pH 3.0) for 5 min and solubilized, and associated radioactivity was counted to determine nonspecific binding. Group C was washed with cold saline to remove unbound ¹²⁵I-ANG II, exposed to binding buffer at 37°C for 5 min to allow for internalization, acid-washed to remove surface-bound radioligand, and solubilized for determination of associated radioactivity. The amount (percentage) of internalized receptor was then determined as follows: [(C − B)/(A − B)] × 100.

Laser-scanning confocal microscopy.

HEK-293 cells stably expressing the wild-type AT_1AR or the AT_1AR/GFP construct were plated onto collagen-coated 25-mm glass coverslips (Invitrogen), placed in six-well plates, and maintained in selection medium [MEM supplemented with 10% FBS, 1% antibiotic-antimycotic-amphotericin B, and G418 (400 μg/ml)]. For transfection of receptor-expressing HEK-293 cells with β-arrestin-2/red fluorescent protein (RFP) or GFP, FuGENE HD was used according to the manufacturer's instructions. Prior to study, receptor-expressing HEK-293 cells were treated with 100 μmol/l cycloheximide in MEM with 0.1% BSA for 3 h to inhibit new protein synthesis. RASMC were plated onto collagen-coated glass coverslips embedded in 35-mm petri dishes (MatTek, Ashland, MA) for 24 h, the medium was removed and replaced with Hanks' balanced salt solution and 20 mmol/l HEPES (pH 7.4), and vehicle (DMSO) or inhibitor was added. Cells were incubated for 60 min at 37°C and then for an additional 60 min in the presence or absence of ANG II (100 nM). Cells were washed with ice-cold PBS, fixed with 4% formaldehyde-PBS for 10 min at room temperature, and washed again with PBS. For AT_1AR/GFP/HEK-293 cells, the DNA-specific fluorescent dye DRAQ5 (0.5 μmol/l; Alexis) was added, and cells were imaged. For RASMC, cells were permeabilized with 100% ice-cold methanol for 10 min at 0°C and then washed several times in PBS at room temperature. Blocking of nonspecific proteins was performed with 2% goat serum in PBS. Cells were washed again in PBS, and the primary antibody anti-p65 NF-κB (mouse monoclonal, 1:50 dilution in blocking buffer; Santa Cruz Biotechnology, Santa Cruz, CA) was applied for 2 h at room temperature. After they were washed again, the cells were incubated with the secondary antibody coupled to Alexa Fluor 588 (goat anti-mouse, 1:50 dilution in blocking buffer; Invitrogen) for 45 min at room temperature. After additional washes in PBS, DRAQ5 was added to visualize nuclei. Confocal microscopy was performed using a laser-scanning microscope (LSM 510 META, Carl Zeiss, Thornwood, NY) equipped with a ×60 objective; the laser wavelengths were as follows: 488-nm excitation and 505- to 530-nm emission for GFP, 543-nm excitation and 560- to 615-nm emission for Alexa Fluor 588 and RFP, and 647-nm excitation and 700-nm emission for DRAQ5.

Colocalization and endosome size were quantitated utilizing Volocity software (PerkinElmer). Colocalization was assessed using Pearson's correlation coefficient: $PCC=[\sum (x_i-x_{avg})\times (y_i-y_{avg})]/\sqrt{[\sum {({\mathit{x}}_i-{\mathit{x}}_{\mathrm{a}\mathrm{v}\mathrm{g}})}^2\times {({\mathit{y}}_i-{\mathit{y}}_{\mathrm{a}\mathrm{v}\mathrm{g}})}^2]}$. Vesicle size was determined using the following constraints: object intensity of 60–255, minimum object size of 0.2 μm², and ≥0.1-μm separation between objects.

Immunoblotting.

Confluent monolayers of RASMC were serum-deprived (0.1% BSA) for 24–48 h. Cells in cell culture medium (6-well plates, COX-2 expression studies) or Hanks' balanced salt solution and 20 mmol/l HEPES at pH 7.4 (12-well plates, p42/44 ERK activation assays) were exposed to vehicle (DMSO) or inhibitor for 60 min at 37°C, and 100 nM ANG II was added. At the end of the incubation period with ANG II (3 h for COX-2 and 5 min for p42/44 ERK), the cells were washed twice in ice-cold PBS; then 1× SDS-PAGE sample buffer containing β-mercaptoethanol was added, and the sample was boiled for 5 min. Samples were separated by SDS-PAGE (4–20% gradient), transferred to nitrocellulose, and probed for the presence of the COX-2 protein using a rabbit polyclonal antibody (1:500 dilution; Millipore), a mouse monoclonal antibody to β-actin (1:5,000 dilution; Sigma-Aldrich), a rabbit polyclonal antibody to phosphorylated p42/44 ERK or total p42/44 ERK (1:2,000 dilution; Cell Signaling Technology, Beverly, MA), or rabbit polyclonal antibody to β-arrestin-1 and -2 in 0.5% Tween 20-Tris-buffered saline (pH 8.0). Blots were stripped of antibodies between probing for phosphorylated and total proteins. Protein bands were detected by addition of CDP-Star reagent (New England Biolabs, Ipswich, MA) and visualized by exposure of the nitrocellulose to radiographic film (X-OMAT, Kodak). Quantitation of the visualized protein bands was performed by densitometric scanning of the exposed radiographic film (Kodak Molecular Imaging Software, Rochester, NY). The density ratio of phosphorylated to total protein was used as an indicator of kinase activation.

β-Arrestin recruitment assay.

Recruitment of β-arrestin-2 to activated AT₁Rs was detected utilizing a commercially available kit (PathHunter eXpress β-arrestin GPCR kit 93-0312E2, DiscoverX, Fremont, CA). The assay utilizes β-galactosidase enzyme fragment complementation technology, in which the GPCR of interest (AT₁R) is fused in frame with a fragment of β-galactosidase and coexpressed in Chinese hamster ovary (CHO-K1) cells stably expressing a fusion protein of β-arrestin-2 and the NH₂-terminal deletion mutant of β-galactosidase. Activation of the AT₁R initiates binding to β-arrestin-2 fusion protein, allowing formation of the active β-galactosidase. The active enzyme is then detected by providing substrate, resulting in generation of a luminescent product. Recruitment of β-arrestin-2 to the AT₁R is then quantitated by detection of a luminescent signal. For these assays, the cells provided with the kit were grown in 96-well plates for 48 h and exposed to vehicle or inhibitors for 30 min at 37°C and then to 100 nM ANG II for 90 min. Luminescence was detected following addition of substrate according to the manufacturer's protocol.

Small interfering RNA transfection.

RASMC were plated into wells of a six-well plate. After 24 h, the cells were ∼70% confluent and transfected using DharmaFect reagent (8 μl; Thermo Scientific) with a control scrambled small interfering RNA (siRNA; Negative Control #1 siRNA, Ambion) or individual siRNA targeting rat β-arrestin-1 (70 nmol) and β-arrestin-2 (30 nmol) (Santa Cruz Biotechnology) according to the manufacturer's protocol. After 24 h, the cells were passaged into 35-mm petri dishes containing collagen-coated glass coverslips and used for detection of p65 NF-κB nuclear localization by immunofluorescence, as described above. To quantify the transcription level of β-arrestin-1 and -2, quantitative real-time PCR was carried out (CFX-96, Bio-Rad). Total RNA was isolated from cells using the High Pure RNA Isolation Kit (Roche). Specific primer pairs used for amplification were as follows: 5′-GAGGACAAGAAGCCACTGA-3′ (sense) and 5′-GACTGAGCACGGAAGGTT-3′ (antisense) for rat ArrB1, 5′-GATGGTGTGGTGCTTGTG-3′ (sense) and 5′-CTGGTAGGTGGCGATGAA-3′ (antisense) for rat ArrB2, and 5′-CATTCTTCCACCTTTGAT-3′ (sense) and 5′-CTGTAGCCATATTCATTGT-3′ (antisene) for rat GAPDH. The expression level of β-arrestin-1 and -2 from each sample was normalized using the mRNA expression levels of a housekeeping gene, GAPDH. RT-PCR analysis indicated an average 48% and 72% decrease in mRNA for β-arrestin-1 and -2, respectively (n = 3).

Calcium measurements.

RASMC were plated into 96-well clear-bottom black plates (Costar) at a density of 6 × 10⁴ cells/well 24 h after transfection with siRNA, as described above. After 24 h, cells were incubated with the calcium-sensitive fluorescent probe Calcium-5 according to the manufacturer's directions (Molecular Probes) for 60 min at 37°C. At the end of the incubation, the cells were placed into a fluorometric imaging plate reader (FLIPR Tetra, Molecular Devices) and exposed to ANG II (4 wells per sample). The FLIPR Tetra is a high-throughput optical screening tool for cell-based fluorometric assays. Increases in intracellular free calcium were reflected by increases in detected fluorescence. The calcium ionophore A23187 (3 μM) was used to quantify maximum attainable fluorescence.

Statistics.

Values are means ± SE from the indicated number of studies (n), and comparisons were made using the Microsoft Excel data analysis package employing ANOVA with post hoc t-test where indicated. Significance was tested at the 95% level.

RESULTS

Previously we demonstrated that ANG II induction of COX-2 protein expression in vascular smooth muscle cells is a time- and concentration-dependent event (22) that is dependent on AT_1AR internalization. Others have shown that NF-κB mediates the ability of ANG II to initiate COX-2 protein expression (15). To examine the role of AT_1AR internalization and NF-κB activation in ANG II-induced COX-2 expression, we first utilized two NF-κB pathway inhibitors, Ro-106-9920 and parthenolide, to confirm the involvement of NF-κB in the activation of COX-2 expression in RASMC. Ro-106-9920 prevents ubiquitination and subsequent degradation of the NF-κB inhibitory protein IκBα, thereby keeping IκB from releasing the p50/65 dimer (24). Parthenolide inhibits IκB kinase, preventing the phosphorylation and release of IκBα (17). To demonstrate the effectiveness of these compounds at inhibiting the NF-κB pathway, RASMC were exposed to Ro-106-9920, parthenolide, or vehicle and then stimulated with ANG II or TNFα, a well-characterized activator of NF-κB (Fig. 1A). Immunofluorescence using an anti-p65 NF-κB antibody was then employed to examine the effects of these compounds on the nuclear localization of NF-κB. Unstimulated RASMC demonstrate cytoplasmic distribution of p65 NF-κB. Exposure of these cells to TNFα or to ANG II caused movement of p65 to the nuclear area. This effect could be blocked by Ro-106-9920 or parthenolide, demonstrating the effectiveness of these compounds at inhibiting the NF-κB pathway.

Fig. 1.

Inhibition of p65 NF-κB nuclear translocation stimulated by ANG II and TNFα by Ro-106-9920 or parthenolide in rat aorta vascular smooth muscle cells (RASMC). A: RASMC were pretreated with the NF-κB inhibitor Ro-106-9920 (Ro), the NF-κB inhibitor parthenolide (Parth, 10 μM, 30 min), or vehicle (control), exposed to 100 nM ANG II or TNFα (2 ng/ml) for 60 min, and prepared for immunofluorescence with an antibody to p65 NF-κB (stained red). NF-κB is distributed in the cytoplasm in control (unstimulated) cells. Stimulation by ANG II or TNFα produces nuclear accumulation of red NF-κB over the DRAQ5-stained blue nucleus (purple); pretreatment with Ro-106-9920 or parthenolide blocks nuclear localization of p65 NF-κB. Representative images are from 4 similar studies. B: p65 NF-κB nuclear translocation stimulated with 100 nM [sarcosine¹,Ile⁴,Ile⁸]ANG II (SII-ANG II) for 30 min. Representative images are from 2 similar studies. Scale bars, 10 μm.

The unique, non-G protein-activating, β-arrestin-dependent biased agonist [sarcosine¹,Ile⁴,Ile⁸]ANG II (SII-ANG II, 50 μM) (28) also produced nuclear localization of p65 NF-κB in these cells (Fig. 1B). SII-ANG II was shown previously to stimulate AT_1AR internalization and COX-2 expression independent of G protein activation and intracellular calcium mobilization (20), thus suggesting that activation of the NF-κB pathway is linked to AT_1AR internalization and is G protein-independent.

We previously demonstrated that chemical inhibitors of GPCR internalization prevent ANG II-induced COX-2 expression in RASMC (22), a reported NF-κB-mediated effect (15). Since Ro-106-9920 and parthenolide prevented the nuclear localization of p65 NF-κB initiated by ANG II, we determined if these compounds reduced internalization of the AT_1AR. RASMC pretreated with Ro-106-9920 demonstrated a significant reduction in the internalization of the AT_1AR in response to ANG II stimulation (Fig. 2A). Approximately 70% of cell surface AT_1ARs were internalized after 5 min of exposure to ANG II in control cells. In cells preexposed to Ro-106-9920, only 20% of the cell surface receptors were internalized. Parthenolide also reduced the AT_1AR internalization initiated by ANG II, although to a lesser extent than did Ro-106-9920.

Fig. 2.

Inhibition of angiotensin type 1 receptor (AT_1AR) internalization and β-arrestin recruitment by Ro-106-9920 and parthenolide. A: RASMC were pretreated with 10 μM Ro-106-9920 or parthenolide for 30 min at 37°C, and radioligand binding was assayed using ¹²⁵I-ANG II to measure the amount of internalized AT_1AR after stimulation with 100 nM ANG II for 5 min at 37°C. Values are means ± SE; n = 4. *P < 0.05 vs. untreated. B: laser-scanning confocal microscopy (LSCM) images of AT_1AR/green fluorescent protein (GFP)/human embryonic kidney (HEK-293) cells exposed to vehicle or 10 μM Ro-106-9920 for 30 min and stimulated with 100 nM ANG II for 60 min at 37°C. Cells were fixed with formaldehyde and prepared for LSCM. Receptor is stained green, and nuclei are stained with DRAQ5 (red). Representative images are from 2 similar studies. Scale bars, 10 μm. C: inhibition of β-arrestin-2 recruitment to the AT₁R by losartan (Los) and Ro-106-9920. Chinese hamster ovary (CHO-K1) cells expressing human AT₁R were exposed to losartan or 10 μM Ro-106-9920 for 30 min and then to 100 nM ANG II or 100 nM SII-ANG II for 90 min. Recruitment of β-arrestin-2 to the activated receptor was detected by luminescence. Values (means ± SE from 4 studies) represent fold increase in luminescence over basal. *P < 0.05 vs. unstimulated (Unstim). #P < 0.05 vs. ANG II alone. &P < 0.05 vs. SII-ANG II alone.

The effect of Ro-106-9920 on AT_1AR internalization was visually confirmed using a HEK-293 cell line stably expressing AT_1A/GFP receptors and laser-scanning confocal microscopy. Unstimulated cells demonstrated plasma membrane localization of AT_1A/GFP receptors. On exposure to ANG II, the plasma membrane-localized receptors internalized to the nuclear membrane area (Fig. 2B). Pretreatment of these cells with Ro-106-9920 followed by ANG II stimulation resulted in decreased receptor (green vesicle) internalization; the green receptors aggregated at the plasma membrane surface.

An initial step in the internalization process of GPCRs is recruitment of β-arrestin to the activated receptor. Using a commercially available β-arrestin recruitment assay (PathHunter Express, DiscoverX), we tested the effects of Ro-106-9920 on ANG II-activated recruitment of β-arrestin-2 to the human AT₁R expressed in CHO-K1 cells. ANG II, as expected, produced a robust increase in recruitment of β-arrestin-2 to the AT₁R. This increase was reduced to basal values by pretreatment with the AT₁R antagonist losartan. Similarly, pretreatment with Ro-106-9920 at a concentration known to inhibit IκBα degradation (24) dramatically inhibited recruitment of β-arrestin to the activated receptor (Fig. 2C). To document that the effects of Ro-106-9920 did not involve a G protein-mediated event, we again utilized SII-ANG II. SII-ANG II stimulates β-arrestin-dependent AT_1AR internalization and COX-2 protein synthesis independent of intracellular free calcium release in RASMC (20). SII-ANG II promoted recruitment of β-arrestin-2, which was inhibited by losartan and also by Ro-106-9920 (Fig. 2C), thus demonstrating that NF-κB is involved in the recruitment of β-arrestin-2 to the activated AT₁R.

Biochemical data presented in Fig. 2C indicate that Ro-106-9920 inhibited recruitment of β-arrestin-2 to the AT_1AR following ANG II stimulation. To confirm these findings, we utilized live-cell imaging with laser-scanning confocal microscopy of HEK-293 cells stably expressing AT_1AR/GFP and transiently expressing β-arrestin-2/RFP to visualize and quantify the effects of Ro-106-9920 on colocalization of β-arrestin-2 with the AT_1AR. In these cells, stimulation by ANG II produced an approximately fourfold increase in colocalization of the AT_1AR with β-arrestin-2. Pretreatment with Ro-106-9920 significantly inhibited this colocalization (Fig. 3), visually confirming the results presented in Fig. 2C.

Fig. 3.

Inhibition of AT_1AR-β-arrestin colocalization by Ro-106-9920. A: representative live-cell laser-scanning confocal images of HEK-293 cells stably expressing AT_1AR/GFP and transiently expressing β-arrestin-2/red fluorescent protein (RFP). Cells were exposed to DMSO (untreated) or 10 μM Ro-106-9920 for 60 min and then stimulated with 100 nM ANG II. z-Stack images (0.2 μm/slice) were taken before (0 min) and 60 min after addition of ANG II. Images are composites of acquired z stacks. In cells exposed to DMSO, ANG II stimulation produced colocalization of “green” receptors with “red” β-arrestin-2, yielding “yellow” endosomes. In cells exposed to Ro-106-9920 followed by ANG II stimulation, internalized green receptor appears not to localize with red β-arrestin-2. Scale bars, 10 μm. B: colocalization summary of studies shown in A. Pearson's correlation coefficient (PCC) is shown for AT_1AR/GFP and β-arrestin 2/RFP for cells exposed to DMSO (control), Ro-106-9920 alone, DMSO with ANG II (ANG II), or Ro-106-9920 followed by ANG II (Ro/ANG II). Values are means ± SE of number of observations in parentheses. *P < 0.05 vs. all other groups (by ANOVA).

Recruitment of β-arrestins to activated AT_1AR results in endosomal trafficking of the receptor-arrestin complex. Inhibition of NF-κB by Ro-106-9920 treatment prevented recruitment and colocalization of β-arrestin with the AT_1AR. Therefore, we next examined the effect of the NF-κB inhibitor on the formation of endosomes containing β-arrestin. In HEK-293 cells stably expressing AT_1AR and transiently expressing β-arrestin-2/GFP, β-arrestin-2 is dispersed throughout the cytoplasm. On stimulation with ANG II, β-arrestin-2 is aggregated into cytoplasmic endosomes. In cells pretreated with the NF-κB inhibitor and subsequently exposed to ANG II, β-arrestin-2 is still found in endosomes, but these endosomes are significantly larger than those in cells stimulated by ANG II without Ro-106-9920 pretreatment (Fig. 4). This suggests that inhibiting activation of NF-κB alters the normal formation of β-arrestin-containing endosomes.

Fig. 4.

Effect of Ro-106-9920 on ANG II-stimulated β-arrestin-2 trafficking. A: representative LSCM images of HEK-293 cells stably expressing AT_1AR and transiently expressing β-arrestin-2 GFP. Cells were exposed to DMSO (control) or 10 μM Ro-106-9920 for 60 min and then stimulated with 100 nM ANG II. After 60 min, cells were fixed with 4% formaldehyde and prepared for imaging. In unstimulated (control) or Ro-106-9920-pretreated cells, β-arrestin-2 GFP is distributed throughout the cytoplasm. After ANG II stimulation, β-arrestin-2 GFP is aggregated into recycling endosomes within the cytoplasm. Scale bars, 10 μm. B: summary of surface area measurements of β-arrestin-2 endosomes from confocal images in A. Values are means ± SE from 13 different images from 4 different studies. *P < 0.05 vs. ANG II alone.

The data presented above, in particular in Figs. 2–4, indicate that β-arrestin is involved in activation of the NF-κB pathway by ANG II. To confirm this involvement, we employed silencing RNA to decrease the expression of β-arrestin-1 and -2 in RASMC and then determine the consequence of their decreased expression on ANG II and TNFα activation of NF-κB. As shown in Fig. 5A, transfection of RASMC with siRNA directed to β-arrestin-1 and -2 reduced protein expression by an average of 44.7 ± 9.3% (n = 7). This decreased expression of β-arrestin-1 and -2 did not alter the ability of TNFα to stimulate nuclear localization of p65 NF-κB. However, in cells depleted of β-arrestin-1 and -2, ANG II was unable to stimulate p65 nuclear localization (Fig. 5B). To ensure that AT_1AR-G protein coupling was not altered by β-arrestin depletion, we examined the ability of ANG II to stimulate increases in intracellular free calcium, a G_q/phospholipase C-dependent event. In cells transfected with control scrambled siRNA or siRNA directed to β-arrestin-1 and -2, ANG II produced a significant increase in intracellular free calcium (Fig. 5, C and D). Thus, as expected, decreased β-arrestin-1 and -2 expression does not alter AT_1AR-G protein coupling. However, these data do support a role for β-arrestins in GPCR activation of the NF-κB pathway, demonstrating another difference between the canonical and noncanonical pathways for NF-κB activation.

Fig. 5.

Effect of β-arrestin-1 and -2 silencing on ANG II- and TNFα-induced nuclear localization of p65 NF-κB and stimulation of intracellular free calcium in RASMC. A: representative immunoblots of whole cell lysates from RASMC transfected with control (Con) scrambled small interfering RNA (siRNA) or siRNA targeting β-arrestin-1 and -2 (β-arr1 and β-arr2). β-Actin is included as an internal control. B: RASMC transfected as described in A, stimulated with 100 nM ANG II or TNFα (2 ng/ml) for 60 min, and prepared for immunofluorescence with an antibody to p65 NF-κB (stained red). NF-κB is distributed in the cytoplasm in unstimulated cells. As seen in Fig. 1, stimulation by ANG II or TNFα produces nuclear accumulation of red NF-κB over the DRAQ5-stained blue nucleus (purple). In cells in which β-arrestin-1 and -2 have been knocked down, p65 NF-κB is no longer localized to the nucleus after ANG II stimulation. TNFα stimulation is not affected by loss of β-arrestin-1 and -2. Scale bars, 10 μm. Representative images are from 4 similar studies. C: representative traces from ANG II-stimulated elevations in intracellular free calcium from RASMC transfected with control siRNA or siRNA targeting β-arrestin-1 and -2. Elevations in intracellular calcium were determined over time using a calcium-sensitive fluorescent dye and shown as relative fluorescent units (RFU). ANG II (100 nM) was added at 10 s. D: 100 nM ANG II-induced maximum elevations in intracellular free calcium in RASMC transfected with control siRNA or siRNA targeting β-arrestin-1 and -2. ANG II response is compared with maximum response initiated with the calcium ionophore A23187 (3 μM). Values are means ± SE from 3 similar studies.

Having documented the interrelationship of β-arrestin and NF-κB in ANG II-stimulated translocation of p65 to the nucleus, we next determined the role of NF-κB in the ability of ANG II to increase expression of COX-2. Ro-106-9920 inhibited ANG II-mediated expression of COX-2 in a concentration-dependent manner (Fig. 6, A and C). ANG II produced an approximately sixfold increase in COX-2 protein expression, which was reduced to baseline values by pretreatment of the cells with 10 μM Ro-106-9920. Ro-106-9920 alone had no effect on COX-2 expression. Similarly, parthenolide inhibited the increase in COX-2 protein expression produced by ANG II (Fig. 6B). These studies demonstrate that ANG II-induced COX-2 protein expression is mediated by activation of NF-κB.

Fig. 6.

Inhibition of ANG II-induced cyclooxygenase-2 (COX-2) protein synthesis by Ro-106-9920 and parthenolide. A and B: RASMC were pretreated with the NF-κB inhibitor Ro-106-9920 (0, 1, 5, or 10 μM) or 10 μM parthenolide for 30 min at 37°C and then exposed to vehicle (−) or 100 nM ANG II (+) for 3 h at 37°C. Cell lysates were prepared, and immunostaining for COX-2 protein was performed. β-Actin was used as an internal control. C: summary of densitometric scanning of COX-2 immunoblots in A and B. Values are means ± SE; n = 4. #P < 0.05 vs. control; +P < 0.05 vs. ANG II-stimulated cells (by ANOVA with post hoc t-test).

We utilized an additional inhibitor of the NF-κB pathway to explore involvement of NF-κB in internalization of the AT_1AR and ANG II-induced COX-2 expression. Curcumin disrupts the interaction of the p50/65 NF-κB dimer with NF-κB recognition motifs on targeted DNA, thus preventing initiation of transcription. Therefore, this compound prevents the nuclear activation by NF-κB without altering the nuclear localization of p65 NF-κB. To demonstrate that this is the case, immunofluorescence was used to show that curcumin did not inhibit the nuclear localization of p65 initiated by ANG II in RASMC (Fig. 7). However, curcumin significantly inhibited the increase in COX-2 protein in RASMC exposed to ANG II (Fig. 7, B and C). Next, we examined the effect of curcumin on AT_1AR internalization. Pretreatment of RASMC with curcumin did not alter the ANG II-induced internalization of the AT_1AR as determined by radioligand binding assays (Fig. 7D). These findings provide further evidence linking activation and internalization of the AT_1AR to activation of the NF-κB pathway and subsequent initiation of COX-2 protein synthesis.

Fig. 7.

Effects of curcumin on ANG II-induced p65 NF-κB nuclear localization, COX-2 protein synthesis, and AT_1AR internalization in RASMC. A: RASMC were pretreated with the NF-κB inhibitor curcumin (Cur, 10 μM) or vehicle for 30 min, exposed to 100 nM ANG II for 60 min, and then prepared for immunofluorescence with an antibody to p65 NF-κB (stained red). NF-κB is distributed in the cytoplasm in control, unstimulated cells. Stimulation by ANG II produces nuclear accumulation of red NF-κB over the DRAQ5-stained blue nucleus (purple). Pretreatment with curcumin had no effect on nuclear localization of p65 NF-κB. Representative images are from 4 similar studies. Scale bars, 10 μm. B: RASMC pretreated with 10 μM curcumin for 30 min at 37°C and exposed to vehicle (−) or 100 nM ANG II (+) for 3 h at 37°C. Cell lysates were prepared, and immunostaining for COX-2 protein was performed. β-Actin was used as an internal control. C: summary of densitometric scanning of COX-2 immunoblots in B. Values are means ± SE; n = 4. *P < 0.05 vs. Control and Cur/ANG II (by ANOVA with post hoc t-test). D: RASMC pretreated with 10 μM curcumin for 30 min at 37°C and subjected to radioligand binding assay using ¹²⁵I-labeled ANG II to measure amount of internalized AT_1AR after stimulation with 100 nM ANG II for 5 min at 37°C. Values are means ± SE; n = 4.

To ensure that the effects seen with Ro-106-9920 and parthenolide to blunt the stimulation of COX-2 expression by ANG II were not the result of inhibiting other signaling pathways initiated by ANG II, we determined if Ro-106-9920, parthenolide, or curcumin inhibits the specific G protein-mediated signaling intermediary p42/44 ERK. Phosphorylation of the intracellular enzyme p42/44 ERK is a well-characterized effect of ANG II in vascular smooth muscle. ANG II activates p42/44 ERK in a time- and concentration-dependent manner in RASMC (22). Treatment of RASMC with the NF-κB inhibitors Ro-106-9920, parthenolide, or curcumin did not inhibit the ability of ANG II to increase phosphorylation and activation of p42/44 ERK (Fig. 8). This activation was, however, inhibited by the AT_1AR antagonist losartan.

Fig. 8.

Effect of NF-κB inhibitors on ANG II-induced p42/44 ERK activation. RASMC were pretreated with an NF-κB inhibitor (10 μM), vehicle (DMSO), or the AT_1AR antagonist losartan (Los, 10 μM) and exposed to 100 nM ANG II for 5 min. Cell lysates were prepared and analyzed by anti-phosphorylated p42/44 ERK immunoblotting to determine activation of p42/44 ERK. A: representative immunoblots showing increased p42/44 ERK phosphorylation (p-p42/44 ERK) by ANG II, which was inhibited by losartan, but not by the NF-κB inhibitors. ttl-p42/44 ERK, total p42/44 ERK. B: summary of densitometric scanning of phosphorylated p42/44 ERK/total p42/44 ERK immunoblots in A. Values are means ± SE; n = 4. *P < 0.05 vs. Unstim; +P < 0.05 vs. ANG II alone.

DISCUSSION

In the present study, using pharmacological inhibitors of the NF-κB pathway and siRNA directed to β-arrestin-1 and -2, we examined the interrelationships between ANG II activation and internalization of the AT_1AR with NF-κB activation and subsequent COX-2 protein expression. We demonstrate that 1) ANG II-induced AT_1AR internalization, NF-κB nuclear localization, and COX-2 protein expression are inhibited by Ro-106-9920 and parthenolide, which are reported to block the ubiquitination and degradation of IκBα, the initial step of the NF-κB activation pathway; 2) reduced β-arrestin-1 and -2 expression resulted in inhibition of ANG II-stimulated nuclear localization of the p65 subunit of NF-κB without altering receptor-G protein coupling; 3) ANG II-induced COX-2 protein expression, but not AT_1AR internalization, is inhibited by curcumin, an inhibitor of NF-κB-DNA interaction; and 4) the NF-κB inhibitors did not alter the ability of ANG II to activate p42/44 ERK.

The AT_1AR is a well-characterized GPCR. The ability of the AT_1AR to transmit ANG II's signal from outside to inside a cell rests mainly on the activation of specific G proteins, such as G_q. These intracellular signals include elevations in intracellular free calcium; protein phosphorylation, including p42/44 ERK and myosin light-chain kinase phosphorylation; and activation of nuclear transcription and translation. However, recent evidence suggests that specific cellular signaling pathways can be activated independently of G proteins; rather, they rely on the β-arrestin-dependent internalization of the receptor itself to initiate signaling. The rapid (<10 min) and reversible activation of p42/44 ERK by ANG II is a G protein-mediated event, while the sustained activation of p42/44 ERK (>15 min) has been attributed to this β-arrestin/internalized receptor “signalsome.” Our previous studies show that internalization of the AT_1AR and, possibly, its nuclear localization result in the initiation of prostaglandin-endoperoxide synthase-2 (PTGS-2) gene transcription and in RASMC COX-2 protein synthesis, thus supporting a role for a signalsome in activation of protein transcription (21, 22). The β-arrestin signalsome may also be key to nuclear signaling activated by GPCRs. Under basal conditions, β-arrestin has been shown to bind IκKα (29), while a recent publication demonstrated, in an engineered cell system, that β-arrestin-1, through a nuclear localization sequence within its structure, localizes to the nucleus and activates NF-κB on activation of the bradykinin receptor (14). Inhibition of the degradation of IκBα through treatment with Ro-106-9920 may maintain the p50/p65-IκBα complex possibly bound to β-arrestins and, thus, alter receptor-arrestin colocalization and the normal trafficking of the signalsome, thereby producing defective (enlarged) arrestin-containing recycling endosomes. These altered endosomes may keep NF-κB from entering the nucleus and activating transcription. These exciting findings provide a potential mechanism for the nuclear localization for several GPCRs, including, as we have shown previously, the AT_1AR (21). In addition, our current findings are the first to demonstrate GPCR activation of NF-κB in a β-arrestin-dependent mechanism in a physiologically relevant cell model.

NF-κB is a well-characterized transcription factor known to be a central mediator in vascular inflammation. Its pathway of activation from cytoplasm to nucleus, with binding to NF-κB consensus sequences and initiation of transcription of mediators of inflammation, such as interleukins, cytokines, matrix metalloproteinases, and COX-2, has been well documented. In numerous cell types and tissues, including vascular smooth muscle, ANG II activates NF-κB (Fig. 9) (5, 11, 16, 33). While it is certain that, in vascular smooth muscle cells, ANG II activates NF-κB, the mechanism for this activation has not been fully explained. Cui et al. (7) demonstrated ANG II-initiated phosphorylation of Ser⁵³⁶ RelA (p65 NF-κB), one of the initiating steps in the NF-κB activation pathway, in a time-dependent manner. This was followed by modest degradation of IκBα (7). Douillette et al. (9) also demonstrated phosphorylation of RelA by ANG II but without subsequent degradation of IκBα. We have also seen phosphorylation of RelA stimulated by ANG II without degradation of IκBα (data not shown), similar to the lack of IκBα degradation seen with the dopamine receptor activation of NF-κB (31, 32). Brasier (4) and Choudhary et al. (6) showed that the ability of ANG II to initiate this phosphorylation is mediated by the small monomeric G protein RhoA (4, 6). A recent review suggests that ANG II activates the pathway in hepatocytes by G protein-mediated activation of PKC, resulting in subsequent PKC-dependent phosphorylation of a large multiprotein scaffold termed the “CBM signalsome” (a multimeric CARMA3-Bcl10-MALT1 complex) (4). Activation of this complex produces degradation of IκBα and the resultant nuclear translocation of RelA/p65 NF-κB. However, as explained by Brasier, this appears not to be the case in vascular smooth muscle, where degradation of IκBα does not occur and phosphorylation occurs at Ser⁵³⁶, rather than Ser²⁷⁶. Activation of RhoA and involvement of the cell cytoskeleton as discussed below may be the discriminating factors in the difference between NF-κB activation in hepatocytes and NF-κB activation in vascular smooth muscle.

Fig. 9.

Proposed pathway for ANG II-induced activation of COX-2 protein synthesis in RASMC: 1) ANG II interacts with its receptor, producing activation of its respective G protein (presumably G_q); 2) phosphorylation of the receptor occurs and attracts the β-arrestin complex to the receptor; 3) the receptor-arrestin-NF-κB-activated complex is directed to early endosomes after release of phosphorylated IκBα; 4) the endosomal complex migrates to the nuclear area, where, through the putative β-arrestin-1 nuclear localization sequence (14), the complex binds to the nuclear membrane; and 5) p50/65 enters the nucleus, binds to the NF-κB consensus sequence on prostaglandin-endoperoxide synthase-2 (PTGS-2), and initiates transcription for the COX-2 protein. Targets of inhibition by Ro-106-9920, parthenolide, and curcumin are indicated.

In our current study, the findings of which are summarized in Fig. 9, the NF-κB inhibitor Ro-106-9920 or parthenolide prevented ANG II-induced COX-2 expression, demonstrating NF-κB mediation. In addition, these inhibitors significantly reduced the ANG II-induced β-arrestin recruitment, colocalization of the AT_1AR with β-arrestin, and internalization of the AT_1AR, thus linking receptor internalization to the NF-κB pathway. It would be reasonable to question the specificity of these inhibitors, in that they may demonstrate other actions on the ANG II-induced signaling pathways, which may also be involved in the induction of COX-2 expression. Ro-106-9920 was initially demonstrated to be a specific inhibitor of IκBα ubiquitination and degradation, with no effect on AT₁R-ligand binding or PKC activation but a minor inhibitory effect on phospholipase C (33% inhibition at 10 μM) (24). The effects of Ro-106-9920 on AT_1AR internalization do not appear to be attributable to alteration of the state of ubiquitination for the AT_1AR (data not shown). Parthenolide was originally characterized as an inhibitor of the NF-κB pathway as a consequence of its inhibition of IκB kinase, the enzyme that phosphorylates IκB (17). Subsequently, it was shown that parthenolide may also inhibit interaction between p65/NF-κB and NF-κB/DNA binding domains (17). This secondary effect of parthenolide may explain the findings in the current study showing complete inhibition of ANG II-induced COX-2 expression and partial, yet significant, reduction of ANG II-induced AT_1AR internalization. Curcumin has been used by many investigators as an agent to inhibit the interaction of NF-κB with NF-κB/DNA recognition motifs (30). In the present study, curcumin effectively inhibited ANG II-induced COX-2 protein expression without altering the p65 NF-κB nuclear localization response. However, similar to the other inhibitors used in current study, curcumin had no effect on the ability of ANG II to activate p42/44 ERK, suggesting that the effects of these inhibitors were specific to the NF-κB pathway.

In addition to the data with Ro-106-9920 and parthenolide, the loss of ANG II-stimulated p65 nuclear localization by decreased β-arrestin-1 and -2 expression, to our knowledge, is the first demonstration of such an association and could provide the explanation for the manner in which a GPCR activates the NF-κB pathway. Previous studies show a regulatory association between an activated GPCR-β-arrestin complex and IκBα (10, 29), while nuclear localization of β-arrestin-1 was needed for bradykinin activation of NF-κB (14). Therefore, internalization of the activated receptor, mediated through the actions of β-arrestin, associates with the p50/p65-NF-κB complex through a β-arrestin-IκBα signalsome complex. The described involvement of RhoA in mediating phosphorylation of p65/NF-κB (4, 7) may correspond to RhoA's involvement in the cytoskeletal rearrangement in the internalization of the receptor. How RhoA itself is activated is not clear.

In summary, through the utilization of pharmacological inhibitors of the NF-κB pathway, we provide support for the concept that internalization of the activated AT_1AR activates the NF-κB pathway, producing nuclear localization of the transcription factor and initiation of COX-2 protein synthesis.

GRANTS

This project was supported by Dialysis Clinic, Incorporated, National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-055524 (L. M. Luttrell), National Center for Research Resources Grant 1S10 RR-027777-01 (FLIPR Tetra, L. M. Luttrell and T. A. Morinelli), and the Research Service of the Veterans Affairs Medical Center of Charleston, SC.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

T.A.M., L.L., and M.E.U. are responsible for conception and design of the research; T.A.M., M.-H.L., R.T.K., and L.P.W. performed the experiments; T.A.M., M.-H.L., R.T.K., and L.P.W. analyzed the data; T.A.M., M.-H.L., and L.P.W. interpreted the results of the experiments; T.A.M. and L.P.W. prepared the figures; T.A.M. drafted the manuscript; T.A.M., L.L., and M.E.U. edited and revised the manuscript; T.A.M., L.L., and M.E.U. approved the final version of the manuscript.