Signal Transduction in Receptor for Advanced Glycation End Products (RAGE)

Background: RAGE is implicated in diabetes complications, inflammation, and neurodegeneration. Results: Cytosolic domain of RAGE, ctRAGE, contains an unusual α-turn that mediates the mDia1-ctRAGE interaction and is required for RAGE-dependent signaling. Conclusion: A novel mechanism through which extracellular RAGE ligands regulate RAGE-mDia1 signaling is established. Significance: A novel binding interface as a target for suppression of RAGE ligand-stimulated signal transduction is identified. The receptor for advanced glycation end products (RAGE) is a multiligand cell surface macromolecule that plays a central role in the etiology of diabetes complications, inflammation, and neurodegeneration. The cytoplasmic domain of RAGE (C-terminal RAGE; ctRAGE) is critical for RAGE-dependent signal transduction. As the most membrane-proximal event, mDia1 binds to ctRAGE, and it is essential for RAGE ligand-stimulated phosphorylation of AKT and cell proliferation/migration. We show that ctRAGE contains an unusual α-turn that mediates the mDia1-ctRAGE interaction and is required for RAGE-dependent signaling. The results establish a novel mechanism through which an extracellular signal initiated by RAGE ligands regulates RAGE signaling in a manner requiring mDia1.

The receptor for advanced glycation end products (RAGE) 2 is a pattern recognition receptor that consists of three extracellular immunoglobulin domains, a transmembrane helix, and a short 42-amino acid cytoplasmic domain (1). RAGE binds diverse ligand families, including advanced glycation end products (1-3), S100/calgranulins (4), high mobility group box-1 (HMGB1) (5), amyloid-␤ peptide, ␤-sheet fibrils (6), and phosphatidylserine (7). Studies in vitro and in vivo indicate that RAGE is a signal transduction receptor for these ligand families (4,8). The deletion of the cytoplasmic domain of RAGE exerts a "dominant negative" effect in which the signal transduction response to RAGE ligands is blunted (9 -12). A number of sig-nal transduction cascades are activated upon ligand-RAGE interactions, including mitogen-activated protein kinases, phosphatidylinositol 3-kinase, Jak/STAT (signal transducers and activators of transcription) (4), and the Rho GTPases Rac-1 and Cdc42 (13). RAGE-ligand interactions evoke central changes in cellular properties, including stimulation of cellular migration and proliferation, and lead to such pathological conditions as diabetes complications, Alzheimer disease, inflammation, and cancers.
RAGE also plays a pivotal role in the atherosclerotic process (14). According to the "response to injury hypothesis" (15), both migration and proliferation of cells from media to intima are central to the atherosclerotic pathogenesis. RAGE stimulates both processes as a result of the binding of the formin homology (FH1) domain of mammalian diaphanous-1 (mDia1) to the short cytoplasmic tail of RAGE (C-terminal RAGE; ctRAGE) (13). mDia1 acts as a potent actin and microtubule polymerization factor that regulates a number of processes, including cell migration and division (reviewed by Higgs (16) and Wallar et al. (17)). Two formin homology domains of mDia1 are required for mDia1 actin polymerization activity (16). The precise mechanism by which mDia1 stimulates actin polymerization is not completely understood. However, mDia1 appears to promote polymerization from the barbed end of actin filaments in cooperation with the actin-binding protein, profilin (18,19).
The molecular details of RAGE-mDia1 interactions are required to explore specific pathways leading to RAGE-dependent pathologies. Here we present the solution structure of ctRAGE and identify the mDia1 (FH1)-ctRAGE interaction surface. In vitro binding studies reveal that two amino acids, Arg-5 and Gln-6 of ctRAGE, which are part of the mDia1-ctRAGE interaction surface, are essential for interactions with mDia1. When Arg-5 and Gln-6 are mutated to alanines, the loss of the binding epitope for mDia1 suppresses RAGE ligand-induced downstream phosphorylation of AKT and migration and proliferation of vascular smooth muscle cells (SMCs). These studies suggest a novel signaling paradigm in which extracellular cues stimulated by RAGE ligand binding are transduced to the cytoplasmic domain of the receptor via mDia1 to stimulate a fundamental signaling network.
DNA fragment coding for ctRAGE (amino acids 362-404) was amplified from human cDNA clone ID IOH12890 (Invitrogen) with primers R-CT_HindIII (TTAAAGCTTATGTTGT-GGCAAAGGCGGCAAC) and R-CT_BamHI (ATTGGATC-CTCAAGGCCCTCCAGTACTACTCTCGC) and cloned into the BamHI and HindIII sites of pTM-7 vector (21), which confers kanamycin resistance. This vector is used to direct small peptides to bacterial inclusion bodies by fusing them with a very hydrophobic protein, TrpL (21). The resulting plasmid, pTM-ctRAGE, expresses an N-terminal His-tagged fusion of TrpL-ctRAGE.
Labeling, Expression, and Purification of Wild Type and Mutant ctRAGE-To uniformly label ctRAGE, pTM-ctRAGE or pTM-R5A/Q6A-ctRAGE was transformed into Escherichia coli strain BL21(DE3) Codon ϩ (Novagen). For U-15 N labeling, cells were grown at 37°C in minimal medium (M9) containing 35 mg/liter kanamycin and 1 g/liter [U-15 N]ammonium chloride as the sole nitrogen source. For U-13 C, 15 N labeling, cells were grown at 37°C in M9 medium containing 35 mg/liter kanamycin, 1 g/liter [U-15 N]ammonium chloride, and 2 g/liter [U-13 C]glucose instead of unlabeled glucose as the sole carbon source. Cells were grown to 0.7 A 600 at 37°C, induced with 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside, and grown overnight. Cells were harvested and resuspended in 20 mM Hepes-Na (pH 7.0) buffer, containing 8 M urea, and sonicated. The lysate was centrifuged, and the supernatant was loaded onto a nickel-NTA-agarose column (Qiagen). The column was washed with 20 mM Hepes-Na buffer, pH 7.0, containing 8 M urea, and the protein was eluted with 20 mM phosphate buffer, pH 4, containing 8 M urea. Fractions containing the eluted protein were pooled and dialyzed into 10 mM sodium phosphate, pH 6.5, buffer. Up to 70% formic acid (v/v) was added to the resultant sample, and the N-terminal His tag and TrpL were removed by cyanogen bromide cleavage at room temperature for 1 h. The sample was cleared by centrifugation and dialyzed into buffer A (10 mM phosphate buffer, pH 7.0) before anion exchange chromatography on a Q column (Amersham Biosciences). The protein was eluted with a gradient to buffer B (10 mM phosphate buffer, pH 7.0, 1 M NaCl). The fractions containing eluted protein were concentrated by using Ultra-Centricones (Millipore). Purity was estimated to be Ͼ95% by Coomassie-stained SDS-PAGE.
Purification of mDia1 FH1 and Its Fragments-To express the His-MBP-FH1 fusion protein, H-MBP-FH1 was transformed into E. coli strain BL21(DE3) Codon ϩ (Novagen). Cells were grown in Luria broth to 0.7 A 600 at 37°C, induced with 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside, and grown overnight at room temperature. Cells were harvested and resuspended in 20 mM Hepes-Na (pH 7.0), 100 mM NaCl buffer and lysed by sonication. The supernatant was loaded onto a nickel-NTA-agarose column (Qiagen). The column was washed with 20 mM Hepes-Na buffer, pH 7.0, and the protein was eluted with 20 mM phosphate buffer, containing 250 mM imidazole. The fractions containing the eluted protein were dialyzed into Pre-Scission cleavage buffer (20 mM Tris, pH 7.0, 150 mM NaCl, and 1 mM DTT). MBP-FH1 was cleaved by PreScission Protease overnight at room temperature. His-MBP fusion was removed by passing the protein sample through a nickel-NTA-agarose column equilibrated with PreScission cleavage buffer. The protein was concentrated by using Ultra-Centricones (Millipore).
Site-directed Mutagenesis of ctRAGE-To doubly mutate the ctRAGE, the QuikChange II XL site-directed mutagenesis kit (Strategene) was used. Following mutagenic PCR, pTM-ctRAGE was restriction-digested with DpnI for 1 h and transformed into E. coli strain DH10B. Mutated plasmid was isolated and purified using the Mini-Prep kit (Qiagen). DNA sequencing identified plasmid pTM-R5A/Q6A-ctRAGE, which codes for the appropriate mutant ctRAGE.
To perform hydrogen-deuterium exchange experiments, [U-15 N]ctRAGE was lyophilized and reconstituted on ice in deuterated NMR buffer (10 mM potassium phosphate (pH 6.5), 100 mM NaCl, 0.02% (w/v) NaN 3 , in 100% D 2 O). The final concentration of the NMR sample was 1 mM. Deuterium exchange was monitored by a series of 15 N HSQC experiments conducted at 6 ºC, for which samples were collected every 5 min for a duration of 5 min.
To study mDia1-ctRAGE interactions, the NMR titration experiments were performed. 0.5 mM unlabeled mDia1 FH1 or the mDia1 FH1 fragments, in NMR buffer, were titrated into 100 M [U-15 N]ctRAGE in four steps to yield ctRAGE/mDia1 FH1 molar ratios of 5:1, 2:1, 1:1, and 1:2, respectively. The results of the titration were monitored by 15 N HSQC. Over the course of titration, the signal/noise ratio of the peaks that did not show any changes was kept constant by adjusting the number of scans.
For the mDia1 FH1-ctRAGE titration, we observed differential broadening of the subset of peaks characteristic of the intermediate to slow exchange regime, k off Յ ⌬, between free and bound states, where k off is the inverse of the lifetime of the bound state and ⌬ ϭ ⍀ a Ϫ ⍀ a is the chemical shift difference between the free (⍀ a ) and bound (⍀ b ) states. In this case, we can estimate the dissociation constant, K d based on the following assumptions. For binding reactions, k on is usually on the order of 10 6 M Ϫ1 s Ϫ1 . Assuming that the average change of the chemical shift is ϳ0.01 ppm, intermediate exchange will occur when the dissociation constant is Յ10 M.
Structure Calculation-Structural calculations were carried out with CYANA version 2.1 (25) using 295 distance restraints derived from 13 C-edited NOESY and 15 N-edited NOESY spectra, 33 pairs of backbone torsion angle restraints derived from TALOS (26), and two restraints for hydrogen bonds. NOEs were converted to upper limit distances using the CALIBA module in CYANA (25). The reference volume determined by CALIBA was increased 2 times before conversion in order to loosen the distance restraints. All upper limit distances for intermolecular NOEs were set to 6 Å. These experimental restraints are summarized in Table 1. To perform CYANA calculations, a single polypeptide chain was constructed for the ctRAGE molecule.
Refinement-The CYANA-generated distance and angle restraints were converted into CNS format in CCPN (27). A total of 1,000 structures was calculated, and the 200 lowest energy structures were subjected to water refinement and further analysis by PROCHECK_NMR (28). Eleven of the ctRAGE residues from the structured region 1-15 were in the most favorable regions of the Ramachandran plot, and one (Glu-11) was in generously allowed regions. There were no residues in the disallowed regions of the Ramachandran plot. The structural statistics of the 20 best structures are reported in Table 1.
Cell Lines and Materials-Wild type primary murine aortic vascular smooth muscle cells (SMC) were isolated and employed through passages 5-7. Vascular SMCs were grown in 10% FBS containing DMEM (Invitrogen).
Site-directed Mutagenesis of Full-length RAGE-The human RAGE full-length cDNA was cloned and inserted into expression vector pcDNA3.1 as described previously (29). This plasmid was used as the template to create the double mutation at R366A and Q367A within the RAGE cytoplasmic domain using the QuikChange site-directed mutagenesis kit (Stratagene). These mutations correspond to R5A and Q6A of ctRAGE, respectively. The mutated plasmid was sequenced to ensure that only the desired substitutions to alanine were present.
Transfection-Wild type primary aortic vascular SMCs were transfected with a vector or mutant human RAGE cDNA construct using the Nucleofector kit (Lonza).
Western Blot Analysis-Total SMC lysates were immunoblotted and probed with AKT-specific antibody, and phospho-AKT-specific antibody. Antibody to mDia1 was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), antibody to RAGE was obtained from Gene Tex, and antibody to GAPDH was obtained from Cell Signaling Technology. After probing with the primary antibodies, membranes were stripped and reprobed for relative total AKT protein or GAPDH where indicated. HRP-conjugated donkey anti-rabbit IgG (Amersham Biosciences) or HRP-conjugated sheep anti-mouse IgG (Amersham Biosciences) was used to identify sites of binding of the primary antibody where indicated. Blots were scanned by using an AlfaImager TM 2200 scanner with AlfaEase (AlfaImager) FC 2200 software. Results are reported as relative absorbance of test antigen to relative total protein. In all Western blot studies, at least triplicate cell lysates per group were used; results of representative experiments are shown. Where indicated, SMCs were pretreated for 1 h with 10 M LY294002, a phosphatidylinositol 3-kinase (Pl3K) inhibitor, (30) prior to stimulation with S100B.
Smooth Muscle Cell Migration and Proliferation Assays-SMC migration assays were performed using the QCM colorimetric cell migration kit (Chemicon). Cells (3 ϫ 10 5 /well) were seeded into the upper chambers fitted with a lower 8-m porous polycarbonate membrane, and the insert was placed in the lower chamber of a 24-well dish containing Dulbecco's modified Eagle's medium and no stimulant, 10 g/ml S100B (generously provided by Dr. Guenter Fritz), or 10 ng/ml non-RAGE ligand PDGF (R&D Systems) and incubated at 37°C for 5 h. Relative migration was measured according to the manufacturer's instructions. Proliferation of cultured SMCs was quantified by two experimental approaches. First, proliferation of wild type, vector, and double mutant RAGE-expressing cells was quantified according to the manufacturer's instructions using a kit by Chemicon International. Where indicated, SMCs were transfected with cDNA expressing murine mDia1, a generous gift of Professor John Copeland (University of Ottawa), alone or with ctRAGE double mutant, and the effects of mediator S100B or PDGF on migration and proliferation were assessed. In other studies, proliferation was assessed by measuring the incorporation of tritiated thymidine. In both cases, SMCs were seeded at a density of 2 ϫ 10 4 cells/well in 24-well tissue culture-treated plates and incubated in serum-free DMEM for 16 h. Cells were exposed to serum-free DMEM containing the indicated concentration of S100B or PDGF along with [ 3 H]thymidine (1 Ci/well). Cells were harvested 48 h after the incubation period, and cellular proliferation was determined based on the incorporation of tritiated thymidine. Cell counting was performed and confirmed that increased tritiated thymidine incorporation reflected an increase in cell number.
Data Analysis-Mean Ϯ S.D. values are reported. Statistical comparisons between groups were determined using one-way analysis of variance; where indicated, individual comparisons were performed using Student's t test.

C-tail of RAGE Contains a Region with Ordered Tertiary
Structure-To initiate structural analysis of the ctRAGE-mDia1 interaction, we bacterially overexpressed and purified ctRAGE, which is immediately preceded by the 25-amino acid transmembrane helix (1) (Fig. 1A). Solution NMR spectroscopy is well suited to characterize small peptides, such as ctRAGE, and their interactions. The HSQC NMR experiment provides information about the protein backbone structure and is used to assess the suitability of a protein for structure determination (22). The 15 N HSQC spectrum of ctRAGE is well resolved but has limited chemical shift dispersion (Fig. 1B), indicating that the majority of the ctRAGE residues are disordered.
Due to ctRAGE proximity to the cell membrane in vivo, it is possible that the ctRAGE structure is perturbed by an interaction with the lipids. To exclude this possibility, we titrated [U-15 N]ctRAGE with a model membrane (31), formed by DMPC/ CHAPSO bicelles. We observed no changes in the 15 N-HSQC spectrum of [U-15 N]ctRAGE after adding 0.5% DMPC/ CHAPSO bicelles (supplemental Fig. 1). This result suggests that there is no specific interaction between ctRAGE and the cell membrane.
To identify ordered regions of the ctRAGE primary structure, we assessed the flexibility of the ctRAGE backbone. The NOE, which depends on local motions of the backbone NH vectors, provides a convenient measure to identify residues participating in the folded structure: steady-state 15 N{ 1 H} NOEs in proteins are positive and close to one for residues from folded regions and small or negative for residues from disordered segments (23,(32)(33)(34).
Steady-state 15 N{ 1 H} NOE NMR experiments identified a region of ctRAGE that has limited flexibility due to structural constraints and is likely to be folded. The results (Fig. 1C) show that only the N terminus of ctRAGE (residues 2-15) possesses large positive 15 N{ 1 H} NOE values and is thus ordered. Amino acids 16 -42 exhibit small positive or even negative steady state NOEs, suggesting unrestricted local flexibility of the backbone indicative of disordered protein fragments (34,35).
Hydrogen bonds are often critical to hold the structure of small proteins together. To confirm that the N-terminal fragment of ctRAGE may contain intramolecular hydrogen bonds, we performed an amide hydrogen-deuterium exchange experiment in which the extent of proton-deuterium exchange is monitored by NMR spectroscopy (36,37). Solvent-exposed backbone amide hydrogens are usually efficiently exchanged with deuterium if a protein is dissolved in a deuterated solvent. Protein amide hydrogens involved in intramolecular hydrogen bonding are expected to be exchanged at a slower rate than solvent-exposed amide hydrogens. The local electrostatic field of the amide proton can also strongly affect the exchange rate (38). After 5 min of hydrogen deuterium exchange, only a few strong peaks corresponding to Arg-5, Gly-9, and also Glu-16, Glu-26, and Asn-28 can be seen in the 15 N HSQC spectrum of ctRAGE (Fig. 1D), suggesting that hydrogen bonds may indeed be present in the folded part of ctRAGE.
ctRAGE Folds into ␣-Turn-We used standard NMR experiments to solve the solution structure of the ctRAGE ordered region, comprising residues 2-15. Based on 295 distance constraints and 66 dihedral angle constraints obtained from NMR experiments, 20 structures with the lowest target function values were superimposed ( Fig. 2A and Table 1). The root mean square (r.m.s.) deviation of the backbone atoms of the best 20 structures was 0.9 Å (Fig. 2A); well folded proteins typically exhibit a backbone r.m.s. deviation below 0.5 Å. The average value of 0.5 for the steady-state 15 N NOEs also suggests that the backbone of ctRAGE is flexible (34), which is to be expected for a short non-cyclic peptide (35). These data suggest that the structure is dynamic and may include multiple conformers.
The structure consists of a loop stabilized by electrostatic interactions between Arg-5 and Glu-10 and between Glu-11 and Arg-12 as well as two hydrogen bonds between the backbone amide of Arg-5 and carbonyl of Glu-11, and the amide of Gly-9 and carbonyl of Arg-5 ( Fig. 2B and supplemental Figs. 2  and 3). To validate the solution structure, we made a double mutant in which Arg-5 and Gln-6 were changed to alanine to create R5A/Q6A-ctRAGE. The mutations would eliminate critical electrostatic interactions between Arg-5 and Glu-10 and disrupt the tertiary structure. Indeed, the mutations greatly affect chemical shifts of residues in the folded region of ctRAGE located far away from the mutation site (supplemental Fig. 4). Chemical shifts of the residues 15-42 were not significantly changed due to the R5A/Q6A mutation, suggesting that this region remains unstructured in the double mutant.
The pattern of hydrogen bonds classifies the structure as an ␣-turn (39). The ␣-turn corresponds to a chain reversal that involves five amino acids and is stabilized by a hydrogen bond between the carbonyl group of the first residue and the amino group of the fifth (39), such as the ctRAGE hydrogen bond between Arg-5 and Gly-9. (␣) turns are found at the C-terminal helical part of proteins and peptides and on the loop side of ␤-hairpins (40).
The protein molecular surface is critical for understanding protein-protein interactions. We constructed an electrostatic map of ctRAGE that shows that the ctRAGE molecular surface is highly charged, as is expected based on the primary structure (Fig. 2C). At the same time, there is a hydrophobic patch formed by the methylene groups of Arg-5 and Gln-6 side chains that may be important for protein-protein interactions.
In Vitro Identification of Residues Critical for mDia1 FH1-ctRAGE Interactions-To characterize interactions between mDia1 FH1 and ctRAGE, we bacterially overexpressed and  FEBRUARY 10, 2012 • VOLUME 287 • NUMBER 7 purified an MBP fusion construct of FH1. MBP was subsequently removed by using PreScission protease releasing FH1 (Fig. 3A). The mDia1 FH1 construct (Fig. 3A) is slightly larger than the construct used in the original study (13) that reported the discovery of the mDia1-RAGE interaction. mDia1 FH1 includes poorly conserved polyproline regions as well as a highly conserved region between the FH1 and FH2 domains of mDia1 (16).

Signal Transduction and RAGE
Because of the exquisite sensitivity of chemical shifts to the chemical environment, solution NMR spectroscopy is used to identify interaction surfaces between reacting molecules (22,41). Molecular binding preferentially perturbs the chemical environment of atoms in the immediate vicinity of the binding site, leading to changes in the corresponding NMR spectrum. Titrating unlabeled FH1 into uniformly 15 N-labeled ctRAGE, [U-15 N]ctRAGE, resulted in specific changes in the HSQC spectrum of ctRAGE (Fig. 3B). Due to 15 N editing of the experiment, only backbone amide protons and nitrogens of ctRAGE are present in the spectrum. When the molar ratio of FH1 to ctRAGE reached 1:1, peaks corresponding to Gln-3, Arg-4, Arg-5, and Gln-6 either disappeared or were substantially broadened when compared with the rest of the peaks. These changes suggest that residues Gln-3 through Gln-6 participate in the interaction with FH1. Peak broadening is characteristic of slow to intermediate exchange and suggests that the dissociation constant of the interaction K d is Ͻ10 M (3, 42). This result is in good agreement with the results from FH1-RAGE binding studies reporting a K d of 30 M (13).
Mapping the observed chemical shift changes onto the molecular surface of ctRAGE allowed us to identify the FH1-ctRAGE interaction surface. It consists of a hydrophobic patch formed by the methylene groups of Arg-5 and Gln-6 that is contiguous with a positively charged surface formed by Arg-4 and Arg-5 (Fig. 3C, see also Fig. 2C).  To confirm the FH1-ctRAGE interaction surface, we made a ctRAGE double mutant in which residues Arg-5 and Gln-6 were changed to alanine to create R5A/Q6A-ctRAGE (Fig. 1A). Titrating unlabeled FH1 into [U-15 N]R5A/Q6A-ctRAGE resulted only in small (Ͻ0.02 ppm) changes in the position of one amino acid residue, Arg-10. No other chemical shift changes or substantial broadening of the R5A/Q6A-ctRAGE peaks were observed, suggesting that the double mutant interacts with FH1 very weakly, with a K D well above 1 mM (supplemental Fig. 5). We conclude that Arg-5 and Gln-6 are an important part of the FH1-ctRAGE interaction surface.
The FH1 domain is proposed to be disordered (16). In this case, FH1-ctRAGE binding may depend predominantly on the primary sequence of FH1. We attempted to identify the specific region(s) of FH1 that interacts with ctRAGE by creating four deletion constructs: FH1-pep1 (amino acids 646 -660); FH1-pep2-del1 (amino acids 683-712); FH1-pep2-del2 (amino acids 714 -745), which contains short polyproline fragments; and FH1-pep2-del3 (amino acids 749 -771), which contains a part of the conserved FH1-FH2 linker (Fig. 3A). These constructs were purified under denaturing conditions and are likely to be disordered under native conditions. Based on NMR titration experiments, none of these constructs interact with ctRAGE. These results suggest that mDia1 FH1 forms a tertiary structure that is necessary for the mDia1 FH1 interaction with ctRAGE, and this structure is absent in the deletion constructs.
R366A/Q367A-RAGE Evokes Dominant Negative Response in RAGE-mDia1 Signaling-The data suggest that the binding of mDia1 FH1 to ctRAGE depends upon residues Arg-5 and Gln-6 of ctRAGE. Functional studies were performed by using full-length double mutant R366A/Q367A-RAGE, where Arg-366 and Gln-367 correspond to Arg-5 and Gln-6 of ctRAGE, respectively (Fig. 1A). To determine the potential mechanistic implications of the in vitro findings, we transfected R366A/ Q367A-RAGE or vector alone into primary murine aortic SMCs. The efficiency of transfection (ϳ65-70%) is demonstrated in supplemental Fig. 6. When vector-treated primary murine SMCs were incubated with RAGE ligand S100B, an  FEBRUARY 10, 2012 • VOLUME 287 • NUMBER 7 increase in phosphorylation of AKT was noted after 5, 10, and 20 min of incubation. In contrast, in cells transfected with the R366A/Q367A mutant, S100B failed to phosphorylate AKT over the same time course (Fig. 4, A and B).

Signal Transduction and RAGE
To test if PI3K was involved in S100B-RAGE-mediated AKT phosphorylation and the effects of R366A/Q367A-RAGE transfection, we performed the following experiment. Wild type untransfected primary murine aortic SMCs, wild type SMCs transfected with empty vector, and wild type SMCs transfected with ctRAGE double mutant (R366A/Q367A) were treated with S100B or control. SMCs were pretreated with LY294002 (10 M), or vehicle, DMSO, for 60 min and then stimulated with S100B (10 g/ml) for 5 min. As shown in Fig.  4C, pretreatment with LY294002 significantly suppressed the FIGURE 4. S100B stimulates AKT activation in vascular smooth muscles cells via ctRAGE-mDia1 FH1 interaction through Arg-5 and Gln-6. A, wild type, vector-transfected, or full-length double mutant R366A/Q367A human RAGE-transfected SMCs were stimulated with 10 g/ml S100B for the indicated times. Total lysates were subjected to Western blotting with antibodies against phospho-AKT or total AKT. Quantified levels of phosphorylated AKT to total AKT in the wild type-transfected SMCs at different times following stimulation with S100B are shown. -Fold changes are relative to control. Error bars, S.D. (*, p Ͻ 0.05). B, RAGE expression levels in the indicated transfected wild type SMCs were assessed by Western blots probed with primary anti-RAGE antibody followed by anti-GAPDH antibody as a loading control. C, wild type, vector-transfected, or full-length double mutant R366A/Q367A human RAGE-transfected SMCs were stimulated with 10 g/ml S100B for 5 min. Where indicated, cells were pretreated with 10 M PI3K inhibitor LY294002 or with vehicle, DMSO. Total lysates were subjected to Western blotting with antibodies against phospho-AKT or total AKT. Quantified levels of phosphorylated AKT to total AKT in the wild typetransfected SMCs following stimulation with S100B are shown. -Fold changes are relative to control. Error bars, S.D. (*, p Ͻ 0.05). Assays were performed in triplicate, and the results shown are representative of three independent experiments. Arg-366 and Gln-367 of full-length RAGE correspond to Arg-5 and Gln-6 of ctRAGE, respectively. NS, nonstimulated. effects of S100B in wild type and vector-transfected wild type cells on AKT phosphorylation. In R366A/Q367A-RAGE SMCs, no additional suppressive effects of LY294002 on S100B-mediated AKT phosphorylation were observed compared with vector transfection alone. Taken together, these data suggest that the effects of R366A/Q367A-RAGE on suppression of S100Bmediated AKT phosphorylation were at least in part through the action of PI3K.
To test the functional implications of R366A/Q367A-RAGE, we treated wild type and vector-and R366A/Q367A-transfected SMCs with S100B and tested migration and proliferation responses. In wild type or vector-treated cells, incubation with S100B resulted in significantly increased SMC migration and proliferation compared with control (Fig. 5, A and B). However, in R366A/Q367A-transfected cells, S100B-mediated migration and proliferation of SMCs were significantly reduced. When vector-and R366A/Q367A-transfected SMCs were treated with a non-RAGE ligand, PDGF, migration and proliferation responses of SMCs were not affected, thereby suggesting that R366A/Q367A did not cause a nonspecific reduction in migration and proliferation responses and that RAGE-mDia1 did not contribute to the promigration and proproliferation effects of PDGF (Fig. 5, A and B). Note that similar results in a distinct SMC proliferation assay employing tritiated thymidine were obtained (supplemental Fig. 7).
Finally, to determine if mDia1 is required for the effects of R366A/Q367A-RAGE, full-length murine mDia1-overexpressing SMCs and R366A/Q367A-RAGE-transfected cells cotransfected with cDNA overexpressing full-length murine mDia1 were stimulated with S100B or PDGF for 5 or 48 h, and migration and proliferation experiments, respectively, were performed. In cells overexpressing mDia1, a significant increase in S100B-stimulated migration (Fig. 5A) and proliferation (Fig. 5B) was observed compared with vector-alonetransfected SMCs. In mDia1-overexpressing SMCs, transfection with R366A/Q367A-RAGE resulted in significant reduction in migration (Fig. 5A) and proliferation (Fig. 5B) compared with vector-transfected S100B-stimulated SMCs or compared with mDia1 overexpressing SMCs (without R366A/ Q367A-RAGE). Taken together, these data suggest that mDia1-mediated enhanced S100B stimulation of migration and proliferation in SMCs is significantly suppressed by R366A/Q367A-RAGE. Western blot confirmation of overexpression of mDia1 in transfected SMCs compared with vector treatment is shown in supplemental Fig. 8.

DISCUSSION
Deducing the interactions between ctRAGE and intracellular proteins is important to understand how RAGE signal transduction is implicated in various disease states. The interaction between ctRAGE and the FH1 domain of mDia1 is clearly involved in migration and cell proliferation, two critical processes in the etiology of atherosclerosis, tumors, and immune/ inflammatory disorders. Studies in multiple cell types indicate that the short 42-amino acid cytosolic tail of RAGE, ctRAGE, is absolutely required for RAGE signaling (9 -12); however, ctRAGE contains no phosphotyrosine, -threonine, or -serine motifs that are typically used for intracellular signal propagation (Fig. 1A). Here we characterized the intracellular mechanism of RAGE signaling by studying the interaction of mDia1 with ctRAGE.
Our structural analysis revealed that ctRAGE contains an N-terminal segment that folds into an ␣-turn and a long unstructured C-terminal tail. ␣-Turns are characterized by a hydrogen bond between the first and the fifth residue in the turn (39). In globular proteins, these structures are usually exposed to solvent and protrude outward from the protein surface with a hooklike shape (40). ␣-Turns function as a proteinor ligand-interacting module in, for example, the CD4 receptor, cyclooxygenase-1, and ␣-lytic protease (39). ␣-Turns are also relevant structural domains in small peptides, particularly in cyclopeptides containing 7-9 residues (43, 44). ␣-Turns con-tain mainly hydrophilic amino acids (39). Indeed, the ␣-turn of ctRAGE consists of 4 charged amino acids, Arg-5, Arg-7, Arg-8, and Glu-10.
The FH1-ctRAGE interaction surface lies within the ␣-turn. Because the ctRAGE primary structure is well conserved beyond the ␣-turn (Fig. 1A), it is likely that mDia1 is not the only intracellular effector of RAGE signaling. The interaction surface is contiguous and contains hydrophilic and polar residues located on one side of the ctRAGE ␣-turn (Fig. 3C). Ionic as well as non-ionic interactions may be critical for mDia1 FH1 binding. The R5A/Q6A double mutant abolishes the FH1-ctRAGE interaction. No interaction was detected when short fragments of the FH1 domain, which are likely to be disordered, were used instead of the intact FH1 domain. These results suggest that the FH1 domain possesses a well defined structure required for binding to ctRAGE.
Previous cellular and in vivo studies in which novel transgenic mice were generated to express the dominant negative form of RAGE in a cell-specific manner, such as in endothelial cells, smooth muscle cells, neurons, and macrophages (9 -12), examined the role of the cytoplasmic domain of RAGE in RAGE ligand-stimulated signaling by completely deleting the intracellular domain. The present study establishes that key residues within the cytoplasmic domain are required to bind to the FH1 domain of mDia1 and that such binding is essential for RAGE ligand-induced AKT signaling, migration, and proliferation in SMCs. The observation that R366A/Q367A mutation of fulllength RAGE, which corresponds to R5A/Q6A mutation of ctRAGE, has no effect on PDGF-stimulated migration and proliferation demonstrates that the double mutant does not nonspecifically abolish proliferation and migration responses in SMCs. Although it was postulated that the complete deletion of the cytoplasmic domain exerts its effects by blocking RAGE signaling consequent to ligand binding, it is plausible that the extracellular and transmembrane domains of the receptor lacking the cytoplasmic domain serve as a sink for ligands, thereby reducing the binding to any signaling receptor. The present work supports the concept that RAGE ligands exert their biological effects by binding to the extracellular domain of the receptor and transducing a signal through the intracellular domain.
The FH1 domain forms an extended structure within mDia1, making the FH1 domain accessible for binding to ctRAGE (45)(46)(47). Indeed, ctRAGE can bind the FH1 domain within fulllength mDia1 (13). This result suggests that the ␣-turn of ctRAGE is constitutively present (i.e. even in the absence of an extracellular RAGE ligand), implying that mDia1 may also be constitutively bound. What then triggers RAGE-mDia1 signaling when ligands bind to RAGE?
Many RAGE ligands are multivalent and cause RAGE molecules to move together on the cell surface (3, 42, 48 -51). This process will lead to intracellular clustering of ctRAGE, which, in turn, will increase the local concentration of mDia1 bound to ctRAGE (Fig. 6). In isolation, mDia1 is autoinhibited by the binding of the C-terminal diaphanous autoinhibitory domain and N-terminal diaphanous inhibitory domain (45,(52)(53)(54). The upstream effector of mDia1, Rho-A, can release this autoinhibition, activating mDia1 for actin polymerization (55). It is also known that domain-domain contacts within mDia1 are likely to be quite dynamic, suggesting the possibility of intermolecular domain swapping between adjacent mDia1 molecules (45). We propose that the high local concentration of mDia1, induced by the binding of RAGE ligands, leads to an mDia1 configuration that is active (Fig. 6). It is important to note that this mode of activation is different from that of RhoA. RhoA activation leads to a severalfold increase in phosphorylation of downstream effectors as opposed to the 50% increase observed due to RAGE activation of mDia1 (Figs. 4 and 5, A and  B).
In conclusion, our findings show that the FH1-ctRAGE interaction is critical for signaling and functional parameters in primary SMCs and reveal for the first time the structural mechanism through which the RAGE cytoplasmic domain interacts with the FH1 domain of mDia1. These data identify a novel binding interface as a target for suppression of RAGE ligandstimulated signal transduction. domains of mDia1 are required for mDia1 activity. DID and DAD, N-terminal diaphanous inhibitory domain and C-terminal diaphanous autoregulatory domain of mDia1, respectively. mDia1 is autoinhibited due to the interaction between the regulatory diaphanous inhibitory and diaphanous autoregulatory domains, DID and DAD, respectively. RAGE can constitutively bind to mDia1 by using the ctRAGE-FH1 interaction. Clustering of mDia1 molecules due to extracellular RAGE (exRAGE) binding to a RAGE ligand results in intermolecular domain swapping and partial restoration of mDia1 activity.