RAGE ligands stimulate angiotensin II type I receptor (AT1) via RAGE/AT1 complex on the cell membrane

The receptor for advanced glycation end-products (RAGE) and the G protein-coupled angiotensin II (AngII) type I receptor (AT1) play a central role in cardiovascular diseases. It was recently reported that RAGE modifies AngII-mediated AT1 activation via the membrane oligomeric complex of the two receptors. In this study, we investigated the presence of the different directional crosstalk in this phenomenon, that is, the RAGE/AT1 complex plays a role in the signal transduction pathway of RAGE ligands. We generated Chinese hamster ovary (CHO) cells stably expressing RAGE and AT1, mutated AT1, or AT2 receptor. The activation of two types of G protein α-subunit, Gq and Gi, was estimated through the accumulation of inositol monophosphate and the inhibition of forskolin-induced cAMP production, respectively. Rat kidney epithelial cells were used to assess RAGE ligand-induced cellular responses. We determined that RAGE ligands activated Gi, but not Gq, only in cells expressing RAGE and wildtype AT1. The activation was inhibited by an AT1 blocker (ARB) as well as a RAGE inhibitor. ARBs inhibited RAGE ligand-induced ERK phosphorylation, NF-κB activation, and epithelial–mesenchymal transition of rat renal epithelial cells. Our findings suggest that the activation of AT1 plays a central role in RAGE-mediated cellular responses and elucidate the role of a novel molecular mechanism in the development of cardiovascular diseases.

www.nature.com/scientificreports/ LDL (oxLDL) receptors (LOX-1) on the cell membrane, whereby oxLDL activates AT1 and causes vascular endothelial dysfunction in mice 16 . Given these findings, we hypothesized that RAGE ligands can initiate cell signaling by interacting with the RAGE/AT1 complex on the cell membrane. Herein, we investigated whether RAGE ligands can activate G proteins and their downstream signaling pathways in an AT1-dependent manner. We also investigated whether RAGE ligand-mediated cellular responses are abolished by the inhibition of AT1 in renal epithelial cells.
Cell-based ELISA. Cells were seeded at a density of 150,000 cells per well onto 96-well transparent cellculture plates and incubated overnight at 37 °C. The following day, the cultures were transferred to serum-free conditions and the cells were further incubated for 24 h. Thereafter, the cells were fixed by 4% paraformaldehyde without permeabilization, incubated with mouse anti-V5 or rat anti-FLAG antibodies, and then incubated with HRP-conjugated mouse or rat secondary antibodies, respectively. TMB reagents (SeraCare Life Sciences, USA) were then added to each well and the colorimetric reaction was neutralized using a stopping solution (Sera-Care Life Sciences, USA). OD 450 values were measured using Multiskan Go (Thermo Fisher Scientific, USA). Measurement value of each sample was adjusted by subtracting that of negative control incubated with the corresponding secondary antibody in the absence of the first antibody.
Co-immunoprecipitation assay. Membrane proteins were extracted using the transmembrane protein extraction kit (Merck Millipore, USA). Immunoprecipitation was performed using the anti-FLAG-M2 affinity gel (Sigma-Aldrich, USA). FLAG-fusion proteins were eluted with 3 × FLAG peptide (Sigma-Aldrich, USA). Subsequently, purified proteins were separated by SDS-PAGE under reducing or non-reducing conditions, transferred to PVDF membranes, and subjected to western blotting using antibodies against FLAG and V5.
Western blotting. The cells were washed twice with PBS, and lysed using M-PER mammalian protein extraction reagent (Thermo Fisher Scientific, USA) with protease (Thermo Fisher Scientific, USA) and phosphatase (Nacalai Tesque, Japan) inhibitors. For western blot analysis, proteins were separated by SDS-PAGE and transferred onto PolyVinylidene DiFluoride (PVDF) membranes. The membranes were blocked with 5% non-fat dry milk and incubated overnight at 4 °C using the following primary antibodies: mouse V5 (Nacalai Tesque, USA), rat FLAG (Novus Biologicals, USA), rabbit ERK1/2, rabbit phospho-ERK1/2, mouse α-Tubulin (Cell Signaling Technology, USA), or mouse α smooth muscle actin (SMA) (Sigma-Aldrich, USA). Bands were detected using Chemi-Lumi One Super (Nacalai Tesque, Japan) and Chemiluminescence detection system LAS-4000 Mini (GE Healthcare, USA). www.nature.com/scientificreports/ In situ proximity ligation assay. The in situ proximity ligation assay (PLA) was conducted using the Duolink kit (Sigma-Aldrich, USA) to determine the proximity of membrane receptors using mouse anti-V5 antibody (Nacalai, Japan) and rabbit anti-FLAG antibody (MBL, Japan), as reported previously 16 . Images were acquired using the BZ-X700 fluorescence microscope. Quantitative fluorescence cell image analysis was performed using the BZ-X analyzer system (Keyence, Japan).

Quantification of cellular IP1 accumulation.
Cells were seeded at a density of 8000 cells per well onto 96-well transparent cell culture plates and incubated overnight at 37 °C. The following day, cultures were transferred to serum-free conditions and further incubated for 24 h. Thereafter, the cells were treated for 1 h with IP1 stimulation buffer mixed with the same amount of DMEM without phenol including vehicle, RAGE-BSA, HMGB-1, and angiotensin II at an indicated concentration at 37 °C, 5% CO2. Triton X was then added at a final concentration of 1%, and cell lysates were prepared after shaking the plates for 30 min. Finally, the cell lysates were transferred to 384-well white plates and IP1 levels were measured using the IP-One assay kit (Cisbio, France).The emission signals were measured at 620 and 665 nm after excitation at 320 nm, using the ARTEMIS plate reader(Furuno Electric Co. Ltd., Japan).

Quantification of cellular cAMP content.
Cells were seeded at a density of 8000 cells per well into 96-well transparent cell-culture plates and incubated overnight at 37 °C. The following day, the cultures were transferred to serum-free conditions and the cells were further incubated for 24 h. Treatment with 10 µM losartan (Sigma-Aldrich, USA) or 10 µM FPS-ZM1 (Merck Millipore, USA) started 24 h or 30 min prior to the stimulation, respectively. siRNA for RAGE or AT1a was performed 24 h prior to the stimulation. Thereafter, the cells were treated for 1 h with DMEM without phenol, 1 mM IBMX, and 1 µM Forskolin including each reagent at the indicated concentration at 37 °C, 5% CO2. Triton X was then added at a final concentration of 1%, and cell lysates were prepared after shaking the plates for 30 min. Finally, the cell lysates were transferred to 384-well white plates and cAMP levels were measured using the cAMP dynamic 2 kit (Cisbio, France). The emission signals were measured at 620 and 665 nm after excitation at 320 nm, using the ARTEMIS plate reader(Furuno Electric Co. Ltd, Japan). Statistics. All data are presented as the mean ± SEM. The statistical difference between two treatments or among multiple treatments was determined using the Student's t-test or one-way ANOVA with Bonferroni testing, respectively.

RAGE formed a complex with AT1 on the cell membrane.
To understand the signaling mechanisms triggered by RAGE stimulation, we generated CHO cells that express no endogenous AT1 stably encoding V5-tagged RAGE alone, and V5-tagged RAGE with FLAG-tagged AT1, FLAG-tagged mutated AT1 that impairs the ability to activate G protein (AT1mt) 16,17 , or FLAG-tagged AT2-the AT1 isoform (CHO-RAGE, CHO-RAGE-AT1, CHO-RAGE-AT1mt, and CHO-RAGE-AT2, respectively). The immunofluorescence images of these cell lines are shown in Fig. 1a. The similar expression levels of the corresponding receptors on the cell surface were confirmed by cell-based ELISA under non-permeabilized condition (Fig. 1b). The immuno- www.nature.com/scientificreports/ precipitation of membrane proteins with an anti-FLAG antibody revealed that RAGE is immunoprecipitated with AT1 (Fig. 1c). An in situ PLA assay conducted under non-permeabilized conditions to determine the membrane proximity of two molecules (< 30-40 nm) exhibited a similar enhancement of red fluorescence in www.nature.com/scientificreports/ CHO-RAGE-AT1 and CHO-RAGE-AT1mt, but not in CHO-RAGE-AT2, implying the specific proximity of RAGE and AT1 either with or without mutation ( Fig. 1d-1,2). These results are consistent with a previous study that demonstrated the oligomerization between AT1 and RAGE using the bioluminescence resonance energy transfer method 15 . Losartan and irbesartan did not alter the proximity between RAGE and AT1, indicating that the heterodimerized RAGE/AT1 complex is not uncoupled by ARBs (Fig. 1d-1,2).

RAGE ligands induced G protein-selective AT1 activation.
AngII binds to AT1 to initiate intracellular signaling cascades by activating specific G proteins α subunit, including Gαq/11 (Gq), Gαi/0 (Gi), and Gα12/13 12 . Herein, we assessed the activation of Gq and Gi by measuring inositol monophosphate (IP1) accumulation and the inhibition of forskolin-induced cyclic adenosine monophosphate (cAMP) production, respectively. AngII increased inositol monophosphate (IP1) levels in CHO-RAGE-AT1, but not in CHO-RAGE-AT1mt (Fig. 2a, b). However, the RAGE ligands HMGB1 and AGE-BSA did not alter IP1 levels in CHO-RAGE-AT1, suggesting that RAGE ligands cannot activate Gq signaling (Fig. 2b). By contrast, the treatment of CHO-RAGE-AT1 with HMGB1 as well as AngII decreased forskolin-induced increase in cAMP levels in a dose-dependent manner (Fig. 2c). The effect of AngII or RAGE ligands on cAMP levels was absent in CHO-RAGE, and was abrogated upon treatment with a Gi inhibitor, pertussis toxin, in CHO-RAGE-AT1 (Fig. 2d). The treatment of CHO-RAGE-AT1mt with RAGE ligands did not affect cAMP levels, supporting the dependence of this phenomenon on the AT1-G protein pathway (Fig. 2d). While the treatment of CHO-RAGE-AT2 with RAGE ligands did not affect cAMP levels, Ang II reduced cAMP levels consistent with previous studies that AngII-AT2 activates Gi (Fig. 2d) 18 . Notably, mDia1 overexpression in CHO-RAGE cells did not affect RAGE ligand-mediated cAMP reduction, indicating that AT1 is essential for the RAGE-mediated activation of G proteins (Fig. 2d). Additionally, the treatment of cells with an ARB, losartan, inhibited both RAGE ligand-and AngII-induced reduction of cAMP levels (Fig. 2e). The treatment of cells with a RAGE inhibitor, FPS-ZM1 abrogated RAGE ligand-induced reduction of cAMP levels, but not AngII-induced reduction of cAMP levels (Fig. 2e). These data indicated that RAGE ligands selectively activate the G protein pathways mediated by AT1, and that this activation is inhibited by the pharmacological blockade of AT1.

AGE-induced cellular signaling was blocked by the inhibition or knockdown of AT1 in kidney epithelial cells.
We then used NRK52E cells, rat renal proximal tubular epithelial cells, to clarify if the observed machinery is relevant in primary cells that express endogenous AT1 and RAGE. AGE-BSA increased ERK1/2 phosphorylation, and the increase at 5 or 30 min was inhibited by the treatment of olmesartan (Fig. 3a). siRNA-mediated AT1 knockdown or the treatment of losartan significantly inhibited AGE-induced ERK 1/2 phosphorylation as well as RAGE knockdown (Fig. 3b, c). AGE-induced inflammatory response assessed by NF-κB reporter assay was inhibited similarly by an ARB, olmesartan and a RAGE inhibitor, FPS-ZM1 (Fig. 3d). HMGB1 and AGE-BSA activated the Gi-dependent cell signaling as well as AngII as shown by the inhibition of forskolin-induced cAMP production (Fig. 3e). Either siRNA of AT1 or RAGE abolished the effect of these RAGE ligands on the forskolin-induced cAMP production (Fig. 3e).

ARBs abolished RAGE ligand-induced renal epithelial-mesenchymal transition.
We used the immunoblot analysis of α-smooth muscle actin (SMA), a molecular marker of mesenchymal cells 19 to further investigate whether RAGE ligands cause AT1-dependent Epithelial-mesenchymal transition (EMT) (Fig. 3f). Treatment of NRK52E cells with AGE-BSA for 72 h increased α-SMA and the treatment with ARBs (irbesartan and losartan) attenuated the AGE-induced EMT. Treatment with HMGB1 also induced EMT and ARB was shown to abrogate HMGB1-induced EMT. ARB treatment did not attenuate EMT induced by Transforming Growth Factor-β (TGF-β) (Fig. 3g).

Discussion
Previous studies have reported the interaction between RAGE and the RAS primarily at the transcriptional level [20][21][22][23] 15 . In the present study, we reported that the physical interaction between RAGE and AT1 on the cell membrane facilitates the signal transduction pathway from ligand binding with RAGE to the activation of AT1. These findings suggest that the crosstalk between RAGE and AT1 is bidirectional and far more direct than previously supposed. Notably, our data suggest that mDia1 has no effect on RAGE ligand-mediated signaling in CHO cells lacking AT1. mDia1 is a key component in RAGE signaling due to its interaction with the RAGE cytoplasmic domain 4,5 . Hence, further studies are required to understand the role of mDia1 in RAGE signaling with emphasis on its interaction with AT1. Importantly, our data suggest that AT1 activation induced upon the binding of RAGE ligands to RAGE has distinct characteristics from that induced upon AngII binding. AngII activates Gi and Gq through AT1; conversely, Gi, but not Gq, is activated by RAGE ligands via the AT1/RAGE complex. This can be explained by the recently proposed machinery of biased AT1 activation. Recent structural analyses have revealed that β-arrestin-biased agonists induce a less "open'' conformational change of AT1 compared to that induced by Ang II or other agonists with enhanced Gq coupling activity 13,14 . It was also shown that β-arrestin-biased AngII analogs preferentially activate Gi over Gq 12 . Future structural analysis will be required to clarify how the ligand-RAGE interaction induces conformational change of AT1, but the mechanism of biased activation may explain why RAGE ligands and AngII exert distinct physiological and pathophysiological effects. For example, RAGE ligands do not increase blood pressure as much as AngII because the pressor function of AngII relies We found that the pharmacological and genetic inhibition of AT1 abolished RAGE ligand-induced cellular responses in renal tubular epithelial cells. ARBs were shown to attenuate RAGE ligand-induced EMT, a crucial cellular response in the development of diabetic nephropathy 24,25 . While the protective role of ARBs in the development of diabetic nephropathy has been extensively studied, certain inherent mechanisms have been presumed to exist that modulate the direct inhibition of Ang II-mediated AT1 activation. Our current findings provide a novel molecular basis to support the role of ARBs in diabetic nephropathy by directly inhibiting the signaling pathway of RAGE. Interestingly, in line with our findings, it was reported that an ARB, valsartan, inhibited the renal injury induced by the infusion of AGE-modified rat serum albumin, while the AGE inhibitor pyridoxamine inhibited the renal injury induced by Ang II infusion 26 . Nevertheless, given that ACE inhibitors are also shown to prevent the development of diabetic nephropathy, further investigation will be required to elucidate if, and to what extent, the pharmacological inhibition of AT1 has an additive class effect on the inhibition of diabetic complications other than the blockade of Ang II signaling.