Advanced Glycation End Products Modulate the Maturation and Function of Peripheral Blood Dendritic Cells

The peptides adrenocorticotrophic hormone (ACTH) fragment 4–10, angiotensin I (AI), and MG (40% aqueous solution) were purchased from Sigma (Poole, U.K.).

Peptide (2.4 mmol/l) was dissolved in sodium phosphate buffer (1 ml, 0.02 mol/l, pH 7.8) containing MG (500 mmol/l) under aseptic conditions. The reaction mixture (RM) was incubated at 37°C for 24 h in a 2-ml glass vial and stored in the reaction vessel at 4°C. An endotoxin test revealed insignificant contamination in these reagents (Charles River Endosafe LAL test kit; Wilmington, MA).

The RMs (2 μl) were diluted 1:5 with distilled water, further diluted with 0.1% trifluoroacetic acid in water, mixed with freshly prepared α-matrix (α-cyano-4-hydroxycinnamic acid, 100 mg/ml in 0.1% trifluoroacetic acid in water/acetonitrile/methanol, 1:1:1), and applied to the target plate. Samples were analyzed on the Micromass TofSpec 2E MALDI-MS.

Heparinized human peripheral blood was obtained after informed consent was granted, diluted with medium (Dutch modified RPMI 1640 containing 100 units/ml penicillin/streptomycin and 2 mmol/l l-glutamine [Sigma]), layered onto an equal volume of Ficoll Paque (Amersham Biosciences, Chalfont St. Giles, U.K.), and centrifuged (626g_av, 30 min, 20°C). Mononuclear cells were aspirated from the interfaces, washed twice (medium containing 10% FCS [Sigma]), spun at 600g for 5 min), counted (trypan blue exclusion), and suspended at 3.5 million/ml.

For binding studies using MALDI-MS, and as stimulator cell populations for mixed leukocyte reactions, DCs were enriched from nonadherent mononuclear cell populations, after overnight incubation, by centrifuging at 600g over a 14.5% wt/vol metrizamide gradient (Sigma).

Freshly isolated peripheral blood mononuclear cells (PBMCs) were exposed to sodium phosphate buffer (5 μl/ml), peptide solution (2.4 mmol/l solution in sodium phosphate buffer [5 μl/ml]), or minimally glycated peptide RM at 2 and 5 μl/ml. Samples were incubated (37°C, 5% CO₂ in air, humidified incubator) for 2.5 h unless otherwise stated.

Antibodies to CD14-FITC (MφP9) and CD11c-PE (Leu-MS) (BD Biosciences, San Jose, CA); CD40-PE (MCA1590) and CD86-PE (MCA1118) (Serotec Oxford, U.K.); CD80-PE (L307–4), CD83-PE (HB15e), and CD45-PE (HI30) (BD-Pharmingen, San Diego, CA); and CD3-ECD (UCHT 1), CD19-ECD (J4.119), and HLA-DR-PC5 (Immu-357) (Coulter Immunotech High Wycombe, U.K.) were used. Isotype-matched controls were obtained from the same manufacturers.

Monoclonal mouse anti-human RAGE was purchased from R&D systems (Abingdon, U.K.), and polyclonal anti-mouse Ig was purchased from BD-Pharmingen.

Antibody labeling was performed on cell samples centrifuged (265g_av, 5 min) and resuspended in PBS supplemented with 0.2% BSA and 0.02% sodium azide (fluorescence-activated cell sorting buffer). Aliquots of 50 μl cell suspension were labeled with monoclonal antibodies (3 μl each antibody, 15 min, room temperature), washed (2 ml fluorescence-activated cell sorting buffer, 265g_av, 5 min), and resuspended and fixed in 1% paraformaldehyde in PBS (500 μl). Data were acquired uncompensated (∼40,000–100,000 events) in the lymphocyte and monocyte light scatter region, using a Beckman Coulter EPICS XL-MCL flow cytometer, and analyzed using WinList 5.3D software (Verity). Single-color cell samples stained with either CD8 or CD3 were also acquired for objective compensation in WinList. Relative cell counts were obtained by simultaneous acquisition of Flow-count fluorospheres (Coulter Immunotech). Dead cells and debris were excluded on the basis of the light scatter (forward scatter/side scatter). Using four-color analysis, the HLA-DR⁺ DCs and monocyte/macrophage populations were defined by excluding CD3⁺ (T-cells) and CD19⁺ (B-cells) and differentiating on CD14 expression. CD14-FITC was detected in fl (fluorescence channel) 1, CD3- and CD19-ECD in fl 3, HLA-DR-PC5 in fl 4, and other phycoerythrin-conjugated markers in fl 2.

Percentage positive events for cell surface markers on gated populations were determined by histogram subtraction of values for isotype-matched controls using Winlist software.

Stimulator DC populations were metrizamide enriched as described previously. Either sodium phosphate buffer or AI solution at 5 μl/ml, or one of two concentrations of AI-AGE RM (2 or 5 μl/ml), was added to the cells (at 100,000/ml). Suspensions were incubated for 2.5 h (37°C/5% CO₂ in air).

Freshly isolated PBMCs from a different donor were suspended in medium containing 10% FCS at 5 million/ml. Triplicate cultures containing 2,500–50,000 PBMCs per well were set up in Terasaki plates. Low-density cells (LDCs) were added at a concentration of either 1,000 or 500 cells per well to give a final culture volume of 20 μl. Control wells contained PBMCs alone, PBMCs + Concanavalin A (5 μg/ml), and exposed LDCs alone. Plates were cultured over saline (37°C/5% CO₂ in air) for 3 days, wells were pulsed with [³H]thymidine (1 μg/ml; specific activity 37 kBq/μl) for 2 h, and cells were harvested onto damp filter paper. Papers were washed with saline, 5% trichloro acetic acid, and methanol. When fully air-dried, incorporation of [³H]thymidine was assayed using a Molecular Dynamics phosphor screen and Imager SI.

PBMCs (3.5 million/ml) were treated with buffer, AI, or AI-AGE RM (2 μl/ml each); placed in a sterile six-well plate; and incubated at 37°C/5% CO₂ in air. At times of 2, 4, 24, 48, and 69 h after addition of controls or AGE, 250-μl aliquots were analyzed using flow cytometry. The same protocol was followed for culture with ACTH (fragment 4–10) and ACTH-AGE. Addition of 50-μl count beads to each sample before acquisition allowed calculation of relative cell numbers.

ACTH (fragment 4–10) RM analyzed by mass spectrometry was found to contain the starting peptide and a number of adducts at higher masses (Fig. 1A). PBMCs were incubated with sodium phosphate buffer or ACTH (fragment 4–10) solution as controls or with one of two concentrations of ACTH (fragment 4–10)-AGE RM.

DCs were defined as DR⁺/(CD3 and CD19)⁻/CD14⁻ cells and monocytes as DR⁺/(CD3 and CD19) ⁻/CD14⁺, excluding cells of other lineages (Fig. 2). Expression of other markers was then assessed (e.g., CD83; Fig. 2). Phenotypes of DCs exposed to either control were similar. Samples incubated with either concentration of AGE RM showed a similar percentage expression of CD11c, CD40, and CD86 (data not shown) to that of the control cells, but a complete lack of significant CD83 expression (Fig. 3). Similar trends of surface marker expression were seen in monocyte populations. Cell aggregation and some cell death were seen in samples incubated with ACTH-AGE RM, but not in controls.

The desired MG-H adduct was seen in the AI-AGE RM as appropriate—54 μm higher than starting peptide (Fig. 1B). Within PBMCs exposed to AI-AGE RM or controls, CD11c, CD40, and CD86 expression on DCs was analogous. Surface CD83 was comparable on control samples, poorly expressed on 2-μl/ml AGE RM-exposed cells, and absent from the 5- μl/ml sample (Fig. 4A). Similar trends were again seen in CD14⁺ monocyte populations. AI-AGE-exposed cells showed only slight aggregation, and gated DCs and monocytes by flow cytometry showed little evidence for cell loss at the 2.5-h time point.

Isolated PBMCs were cultured with controls or AI-AGE RM at 37°C/5% CO₂ in air, and samples were removed at 2, 4, and 22 h. Lack of CD83 on AGE-incubated cells was again apparent after 2 h. Cells exposed to 2 μl/ml AGE RM showed an increased level of CD83 after 4 h and a higher percentage than either control after 22 h. The 5 μl/ml AGE RM-treated cells were without surface CD83 after 4 h but expressed more than the 2-μl/ml sample at 22 h (Fig. 4B).

Among the DC population, expression of surface CD83 was inversely proportional to concentration of AGE RM added to culture, within standard errors (Fig. 5A). Monocyte expression of CD83 did not drop until a higher concentration of AGEs had been added, and the decrease was less gradual than for those in the DC region (data not shown).

PBMCs were isolated and incubated at 37°C/5% CO₂ in air for 2 h so that they would express a reference level of CD83 (just over 30% positive). These cells were divided into four equal parts, to which buffer, AI, or one of the two concentrations of AI-AGE RM was added. Cells were incubated for a further 2 h. DCs within control samples showed an increase in percentage of CD83 expression from the reference level, whereas those incubated with AGEs showed no such upregulation (Fig. 5B). In the CD14⁺ population, controls again showed a rise from the reference level, whereas AGE-exposed cells showed a reduction, falling to 65% from 80% expression in the reference sample.

PBMCs were isolated and exposed to controls or AGEs and to controls or AGEs with added lipopolysaccharide (1 μg/ml) to upregulate surface markers. Cultures were left for 2, 4, or 24 h at 37°C/5% CO₂ in air. Cells were stained with monoclonal antibodies to CD40, CD80, CD45, and HLA-DR alongside the usual cocktail for population discrimination.

Neither DCs nor monocytes showed a significant difference in percentage expression of any markers after AGE exposure, except for lower CD80 expression (∼50% less) after 24 h compared with controls. CD80 was not expressed significantly after 2 or 4 h.

Cells treated with AI or buffer equally stimulated allogeneic lymphocytes (Fig. 6). LDCs exposed to 2 μl/ml AGE RM produced a low level of allogeneic stimulation, but proliferation was significantly higher than that in responders alone (P = 0.05), whereas those incubated with 5 μl/ml AGE showed no stimulatory capacity (P = 0.78). Concanavalin A responses are shown as a high positive control. Survival of DCs, measured by flow cytometry, was similar in AI-AGE- and buffer-exposed cultures.

PBMCs were cultured for 3 days with buffer, peptide, or AGE peptide, and aliquots were removed at various intervals. Relative numbers of DCs within all control cultures remained similar throughout culture (Fig. 7). Within AGE-exposed samples, numbers of DCs started to increase after 4 h of culture. Cells exposed to AI-AGE reached population plateau after 24 h of culture, whereas ACTH-AGE-cultured cells peaked in number at 24 h.

Monocytes within control cultures again remained similar throughout. Within the AI-AGE-exposed sample, monocyte numbers rose initially before dropping to almost zero after 24 h.

All changes were calculated relative to added count beads and thus were indicative of real changes in cell number.

Isolated LDCs were exposed to monoclonal anti-RAGE antibody before AI-AGE RM. Whole-cell samples were prepared for analysis by MALDI-MS with blank (control) cells and cells exposed to just anti-RAGE antibody or AI-AGE RM. The use of whole cells allowed visualization of cell surface components in the absence of a large mass of intracellular components. Added AI-AGE was seen to bind quickly to the cell surface from the supernatant and was evident for several hours. Within 24 h, AI-AGE was not detectable in either the supernatant or on the cell surface. Cells exposed to anti-RAGE antibody alone looked equivalent to control cells (Figs. 8A and C). Added AGEs could be seen equivalently on the cells with and without pretreatment with anti-RAGE antibody (Figs. 8B and D). Similar results were found when blocking was performed with the anti-CD83 antibody.

PBMCs were isolated and exposed to buffer, AI, or AI-AGE RM before double-layer staining with monoclonal mouse anti-human RAGE and fluorescein isothiocyanate-conjugated polyclonal anti-mouse Ig antibodies or just anti-mouse Ig layer. Background anti-mouse Ig staining was used as a control for histogram subtraction, allowing identification of cells positive for anti-RAGE antibody. In each sample, CD14⁻ DCs did not express RAGE, whereas a high percentage (85–91%) of the CD14⁺ monocytes did (data not shown). Expression of RAGE in monocytes was similar in each sample.

This study shows for the first time that AGEs modulate the maturation and function of peripheral blood DCs and reduce their stimulatory potential. Such effects may provide a mechanistic link between build up of AGEs in diabetes and some immunological consequences of the disease.

Minimally glycated proteins are biologically significant, making it appropriate to investigate such models in vitro (2). We prepared two peptides, minimally glycated with MG, which had profound effects on PBMCs. We chose ACTH fragment 4–10 and AI as our starting peptides, as each contained only one arginine residue with a free amino group. This choice predicted that each peptide molecule would host only one AGE adduct, reducing complicating factors in studying its effect. The AGE adduct MG-H was detected by mass spectrometry as the major product, 54 mU above the starting peptide, within the AI-AGE reaction mixture. However, the ACTH (fragment 4–10) RM was a less simple mixture, the spectrum containing many different peaks above that of the start material, including the +54-mU peak expected; another site on this peptide may be reactive other than the considered arginine.

Both AGE peptides caused a reduction of CD83 expression in peripheral blood DCs and monocyte/macrophages, defined here as the DR⁺/(CD3 and CD19)⁻/CD14⁻ and DR⁺/(CD3 and CD19) ⁻/CD14⁺ populations, respectively. ACTH-AGE changed other surface markers, but we suggest that this was due to its toxicity, as shown by cell death and deterioration. Because AI-AGE was less toxic to cells and the mass spectrum showed only one adduct present, it was used predominantly for most tests.

Putative DCs in PBMCs cultured with AI-AGE showed a lower percentage expression of CD83 than corresponding buffer- or peptide-exposed cells after 2.5 h at 37°C/5% CO₂ in air. The AGE-induced lack of CD83 was not a permanent effect, and recovery, like the initial effect, was dose dependent. The effect on monocytes differed, with reduction in CD83 expression occurring after a higher dose of AGEs. This difference suggests a varied response to AGEs, depending on cell type. We were initially unsure whether the effect of AGEs on CD83 was a downregulation, or a lack of upregulation, because freshly isolated PBMCs express no CD83, but studies indicated that it was probably the latter. Within monocytes, there was evidence of real downregulation of CD83, again suggesting a differing response according to cell type.

CD83 is quoted to have an average receptor turnover time of 90 min (12) and may show effects before other surface receptors with slower recycling systems. Samples incubated for longer times, both with and without lipopolysaccharide in the culture, showed little change in other receptors except for a lack of upregulation of CD80 after culture for 24 h. The AGE-related phenomena may not be CD83 specific.

Preventing surface expression of CD83 can inhibit DC-mediated T-cell stimulation (22). Loss in the capacity of DCs treated with AGEs to stimulate primary T-cell responses could be related to the loss of both CD83 and the costimulatory molecule CD80, particularly during the critical early events of lymphocyte activation by DCs. AGEs do not appear to bind directly to CD83 and seem to be consumed by the cells, suggesting that other indirect signaling pathways may contribute to functional changes.

Culturing peripheral blood cells with AGEs for >4 h caused an explosion in the DR⁺/(CD3 and CD19)⁻/CD14⁻ population. This effect on putative DCs was observed twice with AI-AGE and repeated once with ACTH-AGE to prove that the effect was not peptide specific. It is hard to prove a significant change in populations to account for our observations because of the small numbers of DCs compared with other cells in culture. However, known maturation of CD14⁺ monocytes into DCs in culture (23), together with loss of monocytes, make it likely that these are the source of the increased DC population in culture.

The most well-defined receptor for AGE is RAGE. We found that DC populations did not express RAGE, whereas monocytes expressed it highly. The effects of AGEs on DCs cannot therefore be RAGE mediated, as also suggested from the lack of blocking seen using anti-RAGE antibody in the spectrometric assay of AGE-exposed LDCs. Studies of alternative known AGE-binding sites such as macrophage scavenger receptors type I and II, galectin-3, and other scavenger receptors such as CD36, plus intracellular signaling pathways, will be required to determine the route by which the effects of AGEs are mediated.

Most long-term complications of diabetes can be related to vascular damage to either large vessels or microcirculation. Mortality associated with diabetes can be ascribed principally to large-vessel disease (24). Vessel disease in the diabetic and nondiabetic population is associated with recruitment and activation of monocytes and a subset of T-cells, with increased production of tumor necrosis factor-α, leading to raised levels of C-reactive protein and interleukin-6, both well-defined markers of inflammation (18,25,26). Specific vascular DCs are increased in number at sites of vascular damage and may modulate T-cell activation (27). Damage of vascular endothelium, leading to this inflammation, may be caused by one or more of the following factors: modified LDL, free radicals, hypertension, and infectious microorganisms, such as cytomegalovirus, Chlamydia pneumoniae, Helicobacter pylori, and herpes simplex virus (18,28,29). AGEs must now be considered as another possible factor. Although AGEs are endogenously produced products, binding to RAGE leads to a signaling cascade for activation of nuclear factor-κB and increased expression of vascular cell adhesion molecule-1 and interleukin-6 (30), which in turn cause inflammation and tissue damage. From our studies, it is unlikely that RAGE is the sole mechanism by which AGEs mediate their effects on the immune system. These long-lasting modified proteins can also stimulate a more classic immune response, because antibodies to AGEs are present in diabetic sera and correlate with related complications (31). DCs are likely to be involved in the initiation of these responses.

AGE peptide increased numbers of DCs, but these DCs did not upregulate the maturation marker CD83 or the costimulatory molecule CD80. The reduced capacity of AGE-exposed DCs to stimulate primary T-cell proliferation is compatible with the concept that the DCs remained immature in the presence of AGEs. Increased production of immature DCs that fail to mature and migrate appropriately into tissues could perhaps predispose to accumulation of such cells in vascular epithelium and consequent local innate inflammatory effects. Although we presently have no data to show a direct function of DCs in initiating such inflammation, suggestion of reduced CD83 (32), increased DC numbers (19), and impaired stimulatory capacity of DCs (20,21) in patients with diabetes support the idea that effects reported here may have some counterpart in vivo. In normal subjects, the MG-H adduct is present on up to 2% of all arginine residues (33). Levels in diabetes are likely to be higher because of raised MG concentrations. Assuming 10–20% conversion of AI to AI-AGE in our RM (judging by mass spectrum peak intensity), we can predict that addition of RM to cell culture, containing serum and other proteins, will result in a total arginine-MG-H concentration comparable to that in diabetic subjects.

In conclusion, changes to the maturation, phenotype, and function of DCs have been shown in response to AGEs. We have identified a novel pathway that may affect immunological activity in patients, particularly those with diabetes, who have high circulating levels of AGE proteins.

The authors thank Penny Bedford, and other members of the Antigen Presentation Research Group, for their help and support. We also appreciate the advice on peptide modification provided by Professor Paul Thornalley, University of Essex.