Advanced glycation end products of human β₂ glycoprotein I modulate the maturation and function of DCs

β₂ Glycoprotein I (β₂GPI) or apolipoprotein H is an abundant plasma glycoprotein that binds to negatively charged phospholipids and is involved in clotting mechanisms and lipid pathways.¹  Studies performed in healthy individuals have shown a correlation between plasma β₂GPI and fasting glucose, lipids, and lipoprotein levels.²  β₂GPI plasma concentrations are strongly associated with the metabolic syndrome and cardiovascular disease in type 2 diabetic patients and could be considered as a clinical marker of cardiovascular risk.³  This glycoprotein is also the most common target for antiphospholipid antibodies frequently associated with vascular cell dysfunction,⁴  thrombotic events, and pro-atherogenic mechanisms.^5-8  In chronic disorders related to endothelial cell dysfunction, such as systemic lupus erythematosus, antiphospholipid antibody syndrome (APS), and atherosclerosis, β₂GPI plays a role as a target antigen for an immune-mediated attack, possibly influencing the progression of disease.^9-13  β₂GPI stimulates a vigorous adaptive humoral response, but also a cellular immune response.^14,15  Cellular immunity to β₂GPI exists in patients with APS¹⁴  and in healthy individuals.¹⁶  Recently, we showed that β₂GPI is a T-cell target in patients with advanced carotid atherosclerotic plaques.¹⁷  Although much is known about β₂GPI as a cofactor in autoimmune diseases, crucial information is still lacking to clarify why this abundant self-plasma protein is the target of autoimmune responses.

The molecular structure and location of the major epitopic region(s) on the β₂GPI molecule are controversial. Several studies have investigated whether the immune response is directed to native β₂GPI^8,9,18  or to cryptic or neoepitopes.^10,13  Decisive events generating cryptic or neoepitopes include β₂GPI binding to anionic surfaces such as phospholipids and oxidative modifications that alter phospholipid binding.^19-22  In a previous study, we revealed the effects of oxidative stress on β₂GPI structure and how this event renders this self-protein able to activate dendritic cells (DCs), the professional antigen-presenting cells capable of activating both innate and adaptive immunity.^23,24  Many other mechanisms may be responsible for generating cryptic structures, and multiple mechanisms receive support from the heterogeneous antigenic specificities in β₂GPI-specific antibodies and T cells. One candidate mechanism is nonenzymatic glycosylation (glycation), a process that leads to the formation of early, intermediate, and advanced glycation end products (AGEs), which are able to modify self-molecule structures and functions. Even though glycation is present physiologically and is modulated by several factors, disorders of glucose metabolism and systemic autoimmune diseases associated with inflammation and oxidative stress may favor the formation and accumulation of these products.^25-27  AGEs accumulate continuously on abundant and long-lived proteins in the extracellular matrix and are present in inflamed tissues such as rheumatoid synovia and atherosclerotic blood vessels.^28,29  Six receptors that recognize and bind AGEs have been identified.^30,31  The best characterized and most extensively studied receptor for AGEs is RAGE, a 46-kDa protein that is mainly expressed on the surface of endothelial cells, on smooth muscle cells, and on monocyte-derived DCs.^32,33  Although there is accumulating evidence that AGEs are involved in the progression of inflammatory and immune-mediated diseases,^29,34,35  further investigations are needed to clarify the role of glycation in rendering self-proteins able to activate the immune response.

Our objective in this study was to explore possible modifications of β₂GPI induced by in vitro exposure to glucose and the effects of the interaction between glucose-modified β₂GPI (G-β₂GPI) and DCs. We used immunochemical and cytofluorimetric analyses to investigate whether G-β₂GPI is able to activate monocyte-derived immature DCs (iDCs) from healthy human donors. We also sought to determine the contribution of RAGE and the signaling molecules downstream of its activation.

The supplemental materials (available on the Blood Web site; see the Supplemental Materials link at the top of the online article) contain an expanded “Methods” section. The procedures for enrollment and study protocols were fully approved by the institutional review board of the transfusion center at the Sapienza University of Rome, and all patients gave informed consent in accordance with the Declaration of Helsinki.

Human native β₂GPI was purchased from Calbiochem. Endotoxin contamination in β₂GPI preparations, as determined by the quantitative chromogenic limulus amebocyte lysate assay (QCL-1000; BioWhittaker), was < 0.05 EU/mL of protein. For all experiments involving β₂GPI, polymyxin B was added to the cell-culture medium at 10 μg/mL, concentration that completely neutralizes the activity of these amounts of lipopolysaccharide (LPS).

To study glucose-induced modification of β₂GPI, human β₂GPI was dissolved in glycation buffer solution (0.144 g/L KH₂PO₄ and 0.426 g/L Na₂HPO₄), pH 7.4, at a 10 μg/mL final concentration and immediately frozen at −80°C under sterile conditions. A highly purified preparation of human albumin (Sigma-Aldrich) treated with D-glucose or D-mannitol under the same conditions used for β₂GPI was used. β₂GPI aliquots were incubated in the dark at 37°C for different times (from 0-10 days in sealed vials) with the same concentrations (250mM) of glucose (Sigma-Aldrich) or of the nonreducing sugar mannitol (Sigma-Aldrich) used as an isoosmotic control, as described previously.³⁶  All chemicals used were of the highest available purity.

The sequence of the β₂GPI precursor (Homo sapiens; accession #NP_000033) was subjected to computer-assisted analysis to predict the presence of glycation sites using an algorithm available online (NetGlycate 1.0 Server, http://www.cbs.dtu.dk/services/NetGlycate/), as described previously.³⁷  The β₂GPI sequence was also subjected to PROSITE (http://www.expasy.ch/tools/scanprosite/) and ASC analyses (http://bioinformatica.isa.cnr.it/ASC, release 2010), according to a published protocol³⁸  to evaluate whether one of the predicted glycation sites was located close to functional sites.

Aliquots (10 μg) of glucose-treated β₂GPI (G-β₂GPI), mannitol-treated β₂GPI (M-β₂GPI), and native β₂GPI were subjected to SDS-PAGE using 2%-15% acrylamide gradient gels, as described previously.³⁹  The highly purified preparation of human albumin (Sigma-Aldrich) treated with glucose or mannitol under the same conditions used for β₂GPI was used as a positive control. Nonspecific binding of glycated proteins was minimized by saturation of plastic surfaces and checking the protein recovery after each manipulation. The protein molecular weight standards were from Novex® Sharp Pre-Stained Protein Standard (Invitrogen).

Fluorescence studies were carried out as described previously.⁴⁰  β₂GPI (20 μg/200 μL final volume in glycation buffer) was incubated at 37°C in the presence or in the absence of glucose or mannitol for 10 days in the dark, and steady-state fluorescence emission spectra were collected with a FluoroMax-2 spectrofluorometer (Jobin Yvon-SPEX) equipped with a thermostated cuvette holder using an excitation wavelength of 370 nm, equal bandwidths for excitation and emission (5/5), 1-second integration time, and data collection between 400 and 600 nm at 20°C. Emission spectra of sugar-incubated β₂GPI were obtained by subtracting the contributions of separate identical sugar solutions from the fluorescence of the sugar-β₂GPI mixture.

G-β₂GPI, M-β₂GPI, or native β₂GPI (0.5 μg) were spotted onto Immobilon-P strips. Each strip was exposed overnight to serum (diluted 1:100) obtained from 10 patients with primary APS or from 10 control healthy subjects at room temperature. Bound Abs were visualized with HRP-conjugated anti–human IgG, and immunoreactivity was assessed by the chemiluminescence reaction using the ECL system (Amersham).

Blood samples from 5 healthy blood donors from the transfusion center at the Sapienza University of Rome were used to obtain PBMCs.

Monocytes and iDCs were obtained from PBMCs as described previously.²³  Immature DCs were stimulated with 200 ng/mL of LPS (strain 0111:B4 Escherichia coli; Sigma-Aldrich) for 18 hours to obtain LPS-matured DCs. The purity of iDCs was > 95%, as assessed by flow cytometric analysis (FACSCanto using FACSDiva; BD Biosciences) of cells stained with a mixture of CD14-FITC and CD1a-PE mAbs (Pharmingen). CD4⁺ T cells and untouched CD4⁺ naive T cells were purified from PBMCs by magnetic selection using anti-CD4⁺ microbeads and the naive CD4 T-cell isolation kit II (Miltenyi Biotec), according to the manufacturer's instructions. The purity of positively selected CD4⁺ T cells and negatively selected naive CD4⁺ T cells was > 95%, as assessed by flow cytometric analysis.

Preliminary dose-response experiments (0-20 μg/mL) established that G-β₂GPI effects were dose dependent, and we chose 10 μg/mL as the optimal reagent concentration for DC stimulation. Five-day human iDCs were stimulated or not with G-β₂GPI, M-β₂GPI, glucose (5mM), mannitol (5mM), or LPS (200 ng/mL) in the presence or absence of a pretitrated concentration of anti-RAGE (25 μg/mL; Chemicon International) for different times (0-72 hours), then collected and washed. For phenotypic analysis, DCs were stained with PE-conjugated mAbs to CD1a, CD80, and human leukocyte antigen-D region related (HLA-DR), and FITC-conjugated mAbs to CD83, CD86, and CD40 (Pharmingen) and with the mouse anti–human RAGE mAb (Chemicon International) or with isotype-matched control mAb for 30 minutes at 4°C. To assess RAGE surface expression, cells were washed and stained with FITC-conjugated goat anti–mouse Ab (Sigma-Aldrich) for 30 minutes on ice. All samples were analyzed by flow cytometry on a FACSCanto using FACSDiva software (BD Biosciences).

DC culture supernatants were collected at 18 hours after stimulation with G-β₂GPI (10 μg/mL), M-β₂GPI (10 μg/mL), glucose (5mM), mannitol (5mM), or LPS (0.2 μg/mL) in the presence or in the absence of the anti-RAGE mAb (25 μg/mL). Levels of IL-12 p70, TNF-α, IL-10, IL-1β, and IL-6 were determined by ELISA (OptEIA kits; BD Biosciences) following the manufacturer's instructions. The limits of detection were as follows: IL-10, TNF-α, and IL-1β: 16 pg/mL; IL-12p70: 7.8 pg/mL; and IL-6: 2.2 pg/mL.

Cell-free lysates from DCs were stimulated with G-β₂GPI (10μg/mL) or M-β₂GPI (10 μg/mL) for 4 hours or left unstimulated and immunoprecipitated with anti-RAGE mAb (Chemicon International). The immunoprecipitates were subjected to 10% SDS-PAGE, followed by Western blot analysis with anti-β₂GPI polyclonal Ab (Affinity Biologicals).

The allostimulatory ability of stimulated and unstimulated DCs was evaluated in a standard mixed-lymphocyte reaction. Allogeneic T cells (1 × 10⁵ cells/well) were incubated with irradiated DCs for 3 days at different responder-stimulator ratios (1:4-1:64 DCs:T cells) in a 96-well round-bottom plate. On day 2, 0.5 μCi/well of ³H-methyl-thymidine (Amersham) was added to each well for 18 hours at 37°C. Net counts per minute of triplicate cultures were measured.

To find out whether G-β2GPI- and M-β2GPI–treated DCs primed naive T lymphocytes, negatively selected naive allogeneic T cells were cultured with G-β2GPI- or M-β2GPI–treated DCs at a ratio of 20:1. LPS-matured DCs were used as positive control to prime IL-4- or IFN-γ-expressing T cells. Activated T cells were expanded for 10 days with recombinant IL-2 (30 U/mL; Roche Molecular Biochemicals) added on day 5 in a 24-well plate in complete medium to obtain polyclonal T cell lines to be analyzed for cytokine expression by flow cytometry as previously described.²³ 

Immature DCs were stimulated with G-β₂GPI (10 μg/mL) or M-β₂GPI (10 μg/mL) for 4 hours or left unstimulated and successively fixed in 4% formaldehyde in PBS for 30 minutes at 4°C. After washing, cells were incubated for 1 hour with anti-RAGE mAb (Chemicon International) or polyclonal anti-β₂GPI Ab (Affinity Biologicals), followed by addition of Texas Red–conjugated anti–goat or FITC-conjugated anti–mouse Abs (Sigma-Aldrich) for 45 minutes. Images were acquired with a scanning confocal microscope (Leica) equipped with an argon ion laser. FITC and Texas Red fluorochromes were excited at 418 and 518 nm, respectively. Images were collected at 512 × 512 pixels, processed, and filtered to minimize background.

The Fast Activated Cell-based ELISA MAPK assay kit was used to monitor p38 and ERK activation according to the manufacturer's recommendations (Active Motif). In brief, iDCs were cultured and seeded in 96-well plates at 5 × 10⁴ cells/well. Cells were stimulated for different times (0-45 minutes) with G-β₂GPI (10 μg/mL), M-β₂GPI (10 μg/mL), glucose (5mM), mannitol (5mM), LPS (0.2 μg/mL), or phorbol myristate acetate (PMA; 0.2 μg/mL). The number of cells in each well was counted and normalized using the crystal violet solution. The results were expressed as arbitrary units.

To confirm the role of MAPK in G-β₂GPI-induced iDC maturation, iDCs were pre-incubated at 37°C for 30 minutes with the MAPK inhibitor PD098059 or SB203580 before using in iDC phenotypical maturation experiments.

The NF-κB (p65 and p50) transcription factor assay kit (Active Motif) was used to monitor NF-κB activation. Unstimulated DCs and DCs stimulated for 45 minutes at 37°C in 5% CO₂ with G-β₂GPI (10 μg/mL), M-β₂GPI (10 μg/mL), glucose (5mM), mannitol (5mM), LPS (0.2 μg/mL), blocking anti-RAGE Ab (Chemicon International), or control Ab (25 μg/mL) were lysed. Protein content was quantified, and activated levels of p65 and p50 subunits were determined in equal amounts of lysates using antibodies directed against the subunits bound to the oligonucleotide containing the NF-κB consensus-binding site. A HeLa cell extract was used as a positive control and NF-κB wild-type and mutated consensus oligonucleotides were used to monitor the specificity of the assay according to the manufacturer's instructions.

Mean values and SD were calculated for each variable under study. All statistical procedures were performed using GraphPad Prism software 4.0. Data were tested for Gaussian distribution with the Kolmogorov-Smirnov test. Normal distributed data were analyzed using 1-way ANOVA with a Bonferroni post hoc test to evaluate the statistical significance of intergroup differences in all tested variables. P < .05 was considered statistically significant.

The following lysine residues (Figure 1A) have been found to be potentially involved in β₂GPI protein glycation: K25, K63, K123, K129, K157, K287, K303, K306, and K336. The signal peptide^1-19  is underlined in the figure. The protein was further subjected to PROSITE and ASC analyses (see “Bioinformatic analysis”), which indicated that the following cysteine residues are likely to be involved in disulfide bond formation and to be responsible for the Sushi domain secondary structure: C23-C66, C51-C79, C84-C124, C110-C137, C142-C188, C174-C200, C205-C248, and C234-C260. This analysis indicated that at least 2 cysteine residues important for Sushi domain formation, C23 and C124, are very close to the lysine residues K25 and K123, which were previously identified as potential glycation sites. This observation strongly supports the hypothesis that glucose exposure under experimental conditions able to induce protein glycation may significantly modify β₂GPI structure and function.

Purified β₂GPI preparations, incubated with glucose or mannitol, were characterized to evaluate the generation of glycoxidation products of the protein. Using SDS-PAGE analysis under denaturing conditions, we determined the presence of the native β₂GPI and of a higher–molecular weight complex likely corresponding to the formation of β₂GPI dimers in both G-β₂GPI and M-β₂GPI (Figure 1B panel i). Protein dimers can be generated by protein oxidation in culture medium, as described previously.²³  In addition to these forms, in the G-β₂GPI, we detected an additional component with a lower mobility than that of the native protein, which was probably caused by sugar molecule addition to β₂GPI and resembled the pattern observed for the highly purified preparation of glucose-treated human albumin (Figure 1B panel ii) that has been well characterized in previous studies.³⁹  To confirm the presence of glycation products, protein preparations were further subjected to spectrofluorometric analysis to measure the nontryptophan fluorescence, which represents a known AGE fluorescence marker (Figure 1C). A significant increase of AGE fluorescence was recorded for G-β₂GPI compared with M-β₂GPI, indicating that human β₂GPI under our experimental conditions was sugar modified, and that the modification likely consisted of AGE formation (AGE-β₂GPI).

To verify whether the presence of the glycation products may affect sera reactivity with the protein, a dot-blot assay was performed using 10 serum samples from patients with APS. Densitometric analysis showed that the reactivity was stronger for G-β₂GPI than for M-β₂GPI or native β₂GPI at the same protein concentration (Figure 1D). None of the healthy subjects' sera reacted with the β₂GPI preparations.

Unstimulated DCs showed an immature phenotype (HLA-DR^low and CD83⁻) and were weakly immunoreactive for CD80, CD86, CD40, and RAGE. As expected, after 18 hours of incubation, LPS caused DCs to mature so that CD83 appeared and HLA-DR, CD80, CD86, and CD40 expression increased. RAGE expression on the DC surface remained unchanged (Figure 2A). Similarly to LPS, G-β₂GPI and M-β₂GPI, but not control sugar, induced DC maturation. Only G-β₂GPI induced a statistically significant up-regulation of RAGE (Figure 2A), and its expression remained elevated until 72 hours (P < .01; Figure 2B). Pretreatment of iDCs with a saturating concentration of the blocking anti-RAGE mAb prevented the appearance of CD83 and the up-regulation of CD86, CD40, and RAGE in response to G-β₂GPI (P = .01; Figure 2), but not in response to M-β₂GPI (Figure 2). The anti-RAGE mAb treatment did not affect cell viability, as assessed by trypan blue staining (data not shown). Cell treatment with an equal concentration of irrelevant control Ab did not affect β₂GPI-induced DC maturation (data not shown).

After 18 hours of culture, G-β₂GPI and M-β₂GPI triggered statistically significant up-regulation of IL-12p70, TNF-α, IL-10, and IL-1β secretion (Figure 3). G-β₂GPI–stimulated DCs produced higher amounts of IL-1β than did M-β₂GPI–stimulated DCs (P < .001). Unbound glucose left the up-regulation of cytokine secretion unmodified. Pretreatment of iDCs with saturating concentrations of the blocking anti-RAGE mAb prevented the up-regulation of all cytokines in response to G-β₂GPI (P = .01), but not in response to M-β₂GPI (Figure 3).

When irradiated DCs that had been prestimulated with G-β₂GPI or M-β₂GPI were tested in a mixed-lymphocyte reaction, the relatively low proliferative ability (mean counts per minute) of resting allogenic T cells achievable with unstimulated DCs significantly increased, starting from a DC:T-cell ratio of 1:4 (DC:T-cell ratio of 1:16: unstimulated vs G-β₂GPI, P = .004; unstimulated vs M-β₂GPI, P = .04) (Figure 4A).

In cell-culture experiments designed to determine whether G-β₂GPI–treated iDCs primed and polarized allogeneic naive CD4⁺CD45RA⁺ into typical Th1 or Th2 cells, cytofluorimetric analysis showed that most naive T cells cocultured with G-β₂GPI–treated iDCs turned into typical IL-4–producing Th2 cells (17%). Only a small percentage (4%) differentiated into IFN-γ–producing Th1 cells (Figure 4B). Conversely, control iDCs stimulated with LPS and M-β₂GPI–treated iDCs showed a larger number of IFN-γ–producing T cells and a smaller number of IL-4–producing cells than unstimulated iDCs (Figure 4B).

The bias toward a Th2 polarization after stimulation with G-β₂GPI–treated iDCs was supported by dose-dependent IL-6 production that was higher in culture supernatants from G-β2GPI–treated iDCs than in those from control iDCs (at 10 and 30 μg/mL: P < .001) (Figure 4C) and from M-β2GPI–treated iDCs (at 30 μg/mL: P < .001). G-β2GPI–treated iDCs also showed a dose-dependent increase of the regulatory cytokine IL-10 levels (P < .001) (Figure 4C).

To define the involvement of RAGE in the β₂GPI–iDC interaction, double-immunofluorescence labeling using confocal laser-scanning microscopy was performed on unstimulated and 4-hour–stimulated iDCs. Microscopic analysis demonstrated low expression levels of β₂GPI and RAGE in unstimulated iDCs and M-β2GPI–stimulated DCs (Figure 5A panels i and ii) and a very low antigen colocalization. Conversely, G-β2GPI–stimulated iDCs were highly positive for both β₂GPI and RAGE, and a colocalization of β₂GPI with RAGE was detected, as revealed in the merged images (Figure 5A panels iii and iv). Specificity of the Ab reaction was confirmed by the absence of staining in the background controls (supplemental Figure 1).

β₂GPI-RAGE interaction was confirmed by coimmunoprecipitation experiments. Cells were incubated with G-β₂GPI or M-β₂GPI and immunoprecipitated with anti-RAGE mAb. Western blot analysis of the immunoprecipitates showed a strict interaction between β₂GPI and RAGE, which was higher in G-β2GPI–stimulated iDCs than in M-β2GPI–stimulated iDCs (Figure 5B). The identity of the RAGE bands was confirmed by Western blot analysis (data not shown).

We examined the role of p38 MAPK and ERK activation in G-β2GPI–stimulated iDCs. Increased phosphorylation of p38 and ERK, which peaked at 30 minutes, was observed in G-β2GPI–stimulated DCs (n = 6, P = .001; Figure 6A). Glucose control sugar induced a significant activation of ERK (P = .001; Figure 6A, panel ii). Pretreatment of cells with the p38 MAPK inhibitor SB203580 prevented the phenotypic maturation of DCs (as shown by the appearance of CD83, P < .05; Figure 6B), and the increase of RAGE expression in response to G-β₂GPI (P < .001; Figure 6C), whereas the ERK MAPK inhibitor PD98059 prevented only the increase of RAGE (P < .001; Figure 6C).

In G-β₂GPI- and M-β2GPI–stimulated DCs, active p65 and p50 levels were significantly increased compared with iDCs (n = 6, P ≤ .01 and P ≤ .001, respectively, Figure 6D). These levels were only slightly lower than those obtained after LPS stimulation (LPS vs medium, P < .001). Glucose control sugar did not induce any significant NF-κB activation. Pretreatment of iDCs with saturating concentrations of the blocking anti-RAGE mAb prevented the up-regulation of active p65 in response to G-β₂GPI (P = .001; Figure 6D panel i), but not to M-β₂GPI. This assay was specific, because incubation of a HeLa extract in the presence of an unbound wild-type consensus oligonucleotide abolished binding of both subunits; conversely, incubation of the HeLa extract with mutated consensus oligonucleotide did not affect NF-κB binding (data not shown).

Our in vitro study provides new data showing the formation of an AGE-β₂GPI after in vitro exposure to glucose and the interaction between AGE-β₂GPI and DCs. Our results strongly suggest that AGE-β₂GPI is able to activate monocyte-derived iDCs from healthy donors, underlining the role of glycation in rendering self-proteins such as β₂GPI able to activate the immune response.²⁵  We have demonstrated that RAGE is engaged by human AGE-β₂GPI and is able to trigger a signaling pathway mediated by the activation of ERK, p38 MAPK, and NF-κB.

In the present study, we primarily investigated the effects of glucose treatment on β₂GPI structural modifications, bearing in mind that in human serum and tissues, many other molecules contribute to the total glycating activity. Furthermore, it is important to take into account that protein glycation is an accumulation effect, and therefore longer incubation with lower concentrations of glycating agents might be comparable to shorter incubations with higher concentrations and vice versa. Therefore, it is expected that longer exposure of β₂GPI may induce more evident structural modifications and production of fluorescent products. In our experiments, we verified the occurrence of glycation in G-β₂GPI by the appearance of high–molecular weight complexes and by the formation of a fluorescent AGE. The presence of β₂GPI dimers in both sugar-treated β₂GPI preparations may have been due to protein oxidation spontaneously occurring in the culture medium, as described previously.²³ 

Our bioinformatic analyses of the β₂GPI primary structure indicated that several potential glycation sites are present within the molecule, and at least 2 of them are very close to the cysteine residues essential to determining the secondary structure of β₂GPI, which is characterized by the presence of Sushi domains. Therefore, our structural studies indicate that glucose treatment may induce a significant misfolding effect on the β₂GPI structure, likely leading to expression of cryptic or neo-epitopes recognized by the immune system and to significant functional effects. Indeed, dot-blot experiments showed that glucose treatment may increase patients' sera reactivity with the protein, thus suggesting a possible role for this modified protein in the activation of immune system.

When we analyzed the phenotypic characteristics of DCs, the professional antigen-presenting cells capable of activating the immune response, after stimulation with G-β₂GPI, we found that the mature DC-restricted marker CD83 appeared. In parallel, G-β₂GPI up-regulated the surface molecules HLA-DR, CD80, CD86, and CD40, indicating that G-β₂GPI induced a mature DC phenotype.

Further information on DC activation induced by G-β₂GPI comes from our experiments investigating functional changes in DCs. G-β₂GPI preparations specifically stimulated DCs to secrete IL-12p70, TNF-α, and IL-1β, the cytokines that support the differentiation of Th1 cells^41,42  and form a link between innate and adaptive immunity. G-β₂GPI stimulated DCs to also secrete IL-10 and IL-6, cytokines that promote Th2 responses and inhibit Th1 polarization. In agreement with these results, our data also indicate that G-β₂GPI–stimulated DCs activated a Th2-type response by allogeneic naive T cells, characterized by IL-4 expression. This Th2 bias is consistent with the presence of β₂GPI–autoreactive T cells with a Th2 profile determined in the peripheral blood of patients with APS¹⁴  and in patients with advanced carotid atherosclerotic plaques.¹⁷ 

In addition, G-β₂GPI enhanced the capacity of DCs to stimulate T-cell proliferation in an allogenic mixed-lymphocyte reaction. Further studies will be needed to verify whether protein glycation may modify DC uptake and presentation of β₂GPI, thus inducing T- and B-cell immunogenicity.

It was not surprising that, like G-β₂GPI, M-β₂GPI induced a DC phenotypical and functional maturation. This result is in agreement with our previous findings showing that oxidized β₂GPI generated in the culture medium is able to active DCs.²³ 

Information on DC activation specifically induced by AGE-β₂GPI came from our experiments demonstrating that RAGE, the specific receptor for AGEs,^30-32  was overexpressed in DCs treated with AGE-β₂GPI, but not with M-β₂GPI. Receptor binding receives further support from our preliminary experiments showing that the changes in the percentage of phenotypically mature DCs induced by AGE-β₂GPI were dose dependent. Two different experimental approaches indicated that RAGE up-regulation is a critical event in activating DCs. The first evidence came from experiments showing that pretreatment of iDCs with saturating concentrations of the blocking anti-RAGE mAb induced a significant decrease of DC phenotypic surface-marker expression (CD83, CD86, and CD40) and of cytokine production in response to G-β₂GPI but not to M-β₂GPI. The latter derives from the results of scanning confocal microscopy analysis demonstrating that expression and colocalization of β₂GPI and RAGE were higher in G-β₂GPI–stimulated iDCs than in M-β₂GPI–stimulated iDCs. This finding was confirmed and extended by coimmunoprecipitation experiments revealing a strict interaction between G-β₂GPI and RAGE, which was higher in iDCs stimulated with G-β₂GPI than in those stimulated with M-β₂GPI.

Our findings also help to explain how AGE-β₂GPI may activate DCs. Under our experimental conditions, the interaction with RAGE is involved in the maturative effects of DCs through a signaling cascade implicating the activation of both the p38 MAPK and ERK pathways and NF-κB translocation. Our findings are in agreement with previous evidence demonstrating that AGEs are able to activate both p38 and ERK1/2 MAPK and NF-κB after binding to RAGE,⁴³  leading to an enhanced inflammatory response and local tissue injury.^33,44  Our study indicates that the inflammatory response mediated by G-β₂GPI–activated DCs is characterized by high IL-1β production, a cytokine linked to several autoinflammatory disorders.⁴⁵ 

Although enhanced in diabetes, AGE accumulation also occurs in euglycemia, with aging,⁴⁶  and in systemic autoimmune diseases,⁴⁷  albeit to lower degrees, driven by oxidative stress and inflammation. Because RAGE expression is increased in the inflammatory milieu, and so is present in patients with systemic autoimmune diseases, these patients are especially susceptible to the deleterious effects of AGEs. The AGE-RAGE interaction might act as a proinflammatory loop in these patients, thus contributing to chronic low-grade inflammation and rendering these individuals susceptible to the development of accelerated endothelial dysfunction and atherosclerosis.⁴⁸ 

Our in vitro findings also help to explain why the abundant plasma protein β₂GPI becomes immunogenic in vivo. Glycation and glycoxidation of β₂GPI can change the protein structure. It is conceivable to suggest that structurally changed β₂GPI, acting as an inflammatory stimulus, is able to activate immunogenic DCs but also to alter tolerogenic DC functions, thus leading to autoreactive T-cell responses and the development of an autoimmune response.⁴⁹  It is generally agreed that some autoimmune diseases are associated with abnormal presentation of cryptic or neo-epitopes of self-antigens by DCs. Because epitope dominance is influenced by protein structure,⁵⁰  glycation and glycoxidation events may change the molecular context of β₂GPI epitopes (for altered secondary or tertiary structure), thus permitting the efficient presentation of cryptic and neo-determinants.

In conclusion, it can be theorized from our results that a microenvironment predisposes local β₂GPI to glycation and/or oxidation, thereby initiating a local autoimmune process. Chronic oxidative stress causes an accumulation of AGEs. The generation of AGEs and augmentation of proinflammatory mechanisms in the vessel provide a potent feedback loop for sustained oxidant stress, ongoing generation of AGEs, and vascular perturbation. Our in vitro findings now call for studies in patients with chronic disorders related to endothelial cell dysfunction to verify the pathogenetic role of glycated β₂GPI as a trigger of specific humoral and cellular immune reactions. In addition to inhibiting glycation and glycoxidation reactions, antioxidant therapy may also act directly by dismantling AGE/RAGE signaling, thus preventing or reducing the complications of cardiovascular disease.

The support of the Complex Protein Mixture (CPM) Analysis Facility at Istituto Superiore di Sanità, Rome, is acknowledged. This work was supported by grant 8ABF/2 to R.R. from the Italian Ministry of Health.

Contribution: B.B., E.P., A.C., and F.F performed research; B.B., L.S., F.F., and R.R. designed research and analyzed data; and B.B., M.S., and R.R. designed research and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Rachele Riganò, Istituto Superiore di Sanità, Department of Infectious, Parasitic and Immune-mediated Diseases, 299 Viale Regina Elena, Rome, Italy; e-mail: rachele.rigano@iss.it.