OUP user menu

Reduced expansion rate of abdominal aortic aneurysms in patients with diabetes may be related to aberrant monocyte–matrix interactions

Jonathan Golledge, Mirko Karan, Corey S. Moran, Juanita Muller, Paula Clancy, Anthony E. Dear, Paul E. Norman
DOI: http://dx.doi.org/10.1093/eurheartj/ehm557 665-672 First published online: 9 February 2008

Abstract

Aims Diabetes increases the risk of atherothrombosis, but reduces the risk of abdominal aortic aneurysm (AAA). The reason for this difference is unknown. We examined the role of diabetes and glycation on AAA expansion and extracellular matrix–monocyte interactions.

Methods and results We followed 198 patients (20 with diabetes) who had 30–45 mm AAAs with yearly aortic ultrasound for 3 years. Diabetes was independently associated with reduced AAA growth (β = −0.17, P = 0.01; OR for expansion above median 0.18, 95% confidence interval 0.06–0.57). In vitro incubation of resting human monocytes with glycated bovine serum albumin or monomeric type I collagen increased matrix metalloproteinase (MMP) secretion. In contrast, exposure of activated monocytes to glycated type I collagen lattices induced a marked reduction in MMP and interleukin-6 secretion. This de-activating effect was also demonstrated in cross-linked non-glycated collagen lattices, healthy decellularized aortic media, and decellularized aortic media from diabetes patients with atherosclerosis. In contrast, decellularized aortic media from patients with atherosclerosis, but no diabetes, induced increased MMP secretion.

Conclusion These findings confirm that the progression of AAA is slower in patients with diabetes and suggest a mechanism by which the aortic media may be protected from degradation in these individuals.

  • Abdominal aortic aneurysm
  • Diabetes
  • Glycation
  • Monocytes
  • Matrix metalloproteinase
See page 581 for the editorial comment on this article (doi:10.1093/eurheartj/ehm638)

Introduction

Diabetes is an important risk factor for events associated with occlusive intimal atheroma, such as myocardial infarction, stroke, and lower limb ischaemia.1 In contrast, abdominal aortic aneurysm (AAA) is rare in patients with diabetes. Population studies demonstrate that patients with diabetes are half as likely to develop an AAA as those without diabetes.24 Aberrant matrix remodelling is important in the progression of both atherosclerosis and AAA; however, marked medial destruction is only present in AAA.59 In both intimal expansion, typical of atherosclerosis, and medial destruction, typical of AAA, monocyte–macrophages play an important role with the production of matrix-digesting proteinases, such as matrix metalloproteinases (MMPs) and pro-inflammatory cytokines, such as interleukin-6 (IL-6).510 In occlusive atherosclerosis, macrophages are limited to the intima, whereas in AAA, these cells are active in all arterial layers.510

In the healthy artery, type I collagen makes up a substantial component of the media, where it is organized into a complex three-dimensional network.11 In the advanced atherosclerotic plaque, there is substantial deposition of collagen and other extracellular matrix (ECM) protein degradation products within the intimal plaque.12 The organization of these ECM proteins is far less ordered than demonstrated within the media.12,13 Substantial in vitro data suggests that the interaction between cells present within the artery wall and ECM matrix proteins plays an important role in vessel remodelling and that the cellular responses are influenced by the structural form in which the proteins are presented.1417

One of the classical features of diabetes is the advanced glycation of ECM proteins such as collagen.18,19 Advanced glycation end-products (AGEs) have been demonstrated to play an important role in stimulating monocyte–macrophages via specific receptors for AGE (RAGE).18,19 Glycation also alters the three-dimensional structure of ECM proteins such as collagen.20

This study consisted of two parts. First, a clinical study was performed to assess the influence of diabetes on AAA expansion. Secondly, in vitro studies to assess how glycation affects monocyte–ECM interactions were carried out. We hypothesized that glycation would have differing effects on the monocyte–ECM interaction depending on the form the ECM proteins were presented in. In particular, we postulated that glycation would cause structural changes to the aortic media that inhibited monocyte activation.

Methods

Design of clinical study

The study was approved by the relevant Ethics Committees, and informed consent was obtained from participating patients. Patients with small AAAs were recruited from centres carrying out ultrasound surveillance, with demonstrated reproducible measurements techniques.21 Given the slow growth of small AAAs, which can be within the measurement error of ultrasound, we elected to specifically recruit patients in which ultrasound surveillance was available for 3 years. Entry criteria included the following:

  1. The patient had already undergone 3 years ultrasound surveillance of their AAA.

  2. The initial maximum diameter of the AAA was 30–45 mm.

  3. The patient was available for collection of clinical data and blood.

Patients were principally identified via a screening study in which a total of 12 203 men (1418, 12%, of whom had diabetes) were screened for AAA, of which 787 had an AAA measuring 30–45 mm.22,23 A detailed history, clinical examination, and blood analysis were carried out on all patients. This allowed the identification of cardiovascular risk factors including hypertension, dyslipidaemia, smoking history, coronary heart disease (CHD), and peripheral artery disease (PAD), as previously defined.21 Diabetes was defined as fasting blood glucose ≥7.0 mM, or history of, or treatment for hyperglycaemia.24 We used this definition of diabetes rather than including more subtle criteria, such as impaired fasting glucose, as we were interested in assessing patients in which there was likely to be significant glycation of aortic ECM. The characteristics of the present cohort in comparison with that of the screening cohort are shown in Table 1. Consenting patients were prospectively entered into ultrasound surveillance. Patients were excluded if they did not consent to enter the surveillance programme, if they failed to complete 3 years follow-up due to intervention, death, or loss to follow-up, or if they did not provide blood samples for assessment. Patients in the present cohort were similar to the screen-detected cohort, except that they were older and more commonly had dyslipidaemia and PAD. Patients underwent yearly ultrasound imaging during which the maximum transverse and anteroposterior diameter of the infrarenal abdominal aorta was measured by an experienced vascular sonographer at each site using a 3.75 MHz probe and their own validated ultrasound equipment (Toshiba Capasee, Philips HDI 5000, GE Logic 9). The reproducibility of aortic measurements has been previously described.21 As an assessment of advanced glycation in patients with AAA, we measured serum Nϵ-carboxymethyl lysine (CML) by ELISA (CircuLex Co. Ltd, Nagano, Japan). The limit of detection of this assay is 0.1 ng/mL and inter-assay coefficient of variation between 3 and 6%.

View this table:
Table 1

Characteristics of the present cohort compared with screening cohort

CharacteristicPresent cohortScreening cohort
No diabetesDiabetesOverallNo diabetesDiabetesOverall
Number1782019870582787
Male167 (94%)19 (95%)186 (94%)705 (100%)82 (100%)787 (100%)
Age (years)76.7 ± 5.177.3 ± 4.976.7 ± 5.073.6 ± 4.573.1 ± 4.373.6 ± 4.5*
Hypertension98 (55%)14 (70%)112 (57%)377 (53%)56 (68%)**433 (55%)
Dyslipidaemia129 (72%)19 (95%)**148 (75%)305 (43%)45 (55%)**350 (44%)*
Current smoker25 (14%)6 (30%)31 (16%)133 (19%)12 (15%)145 (18%)
Ex-smoker120 (67%)12 (60%)132 (67%)468 (66%)63 (77%)531 (67%)
PAD43 (24%)4 (20%)47 (24%)85 (12%)13 (16%)98 (13%)*
CHD70 (39%)9 (45%)79 (40%)285 (40%)40 (49%)325 (41%)
Aspirin89 (50%)9 (45%)98 (49%)351 (50%)41 (50%)392 (50%)
Statin69 (39%)10 (50%)79 (40%)195 (28%)27 (33%)222 (28%)
Aortic diameter (mm)34.0 ± 3.833.0 ± 4.434.0 ± 3.834.6 ± 4.434.1 ± 4.334.6 ± 4.4
  • PAD, peripheral artery disease; CHD, coronary heart disease.

  • *P < 0.05 comparing between cohorts. Numbers and percentages are given for nominal variables and mean±standard deviation for continuous variable. Percentages are quoted in terms of the total number of patients with or without diabetes or the overall number in the cohort in question.

  • **P < 0.05 comparing between patients with and without diabetes in the same cohort.

Design of in vitro studies

To test our hypothesis, we cultured human monocytes in the presence of the following substratum in 24-well plates:

  1. Soluble protein: we used bovine serum albumin (BSA) at 1 mg/mL concentration.

  2. Monomer type I collagen (non-lattice): 0.1 mg/mL of type I collagen was coated onto wells overnight.

  3. Lattice type I collagen: a mixture of collagen (4 mg/mL in 18 mM acetic acid) in phosphate-buffered saline (PBS) pH7.2 was neutralized with 10% 0.1 M NaOH on ice, added to 24-well plates (500 µL), and allowed to gel overnight at 37°C in a humidified incubator, as described previously.16

  4. Decellularized healthy aortic media: healthy aortic samples were collected from sheep under terminal anaesthesia.

  5. Decellularized atherosclerotic media: atherosclerotic aortic samples were collected from patients undergoing aortic bypass. The aortic medial layer was isolated by micro-dissection, and 1 cm2 squares were placed in 24-well plates and decellularized in 0.1% SDS at 37°C for 15 h, as described previously.25

Each substratum was subsequently treated with glycolaldehyde (50 mM), glutaraldehyde (1%), glycolaldehyde (50 mM) plus aminoguanidine (25 mM) (glycation control), or PBS pH 7.2 (experimental control) in order to compare the effects of glycation and cross-linking without glycation to controls. In order to investigate whether different substrata promoted and inhibited MMP secretion from monocytes, both resting and lipopolysaccharide (LPS) (5µg/mL) activated cells (106 cells/mL) were incubated with matrix for 24 h. Outcome was measured in terms of production of gelatin-digesting MMPs and the cytokine IL-6. We felt that secretion into the culture supernatant would give a better guide to likely effects in vivo. Thus we did not measure RNA or protein concentrations of MMPs, but assessed the proteolytic activity of supernatants. To test the relevance of these findings to the in vivo situation, monocytes were exposed to ECM isolated from healthy sheep aorta and atherosclerotic arteries obtained from diabetic and non-diabetic patients. Sheep aorta was employed as it was not possible to obtain fresh healthy non-aged abdominal aorta from humans. Experiments were performed in triplicate and repeated on a minimum of two occasions.

Monocytes

The human monocyte cell line, THP-1 (referred to as monocytes), was kindly provided by Dr Rajiv Khanna (Tumor Immunology Laboratory, Queensland Institute of Medical Research) and maintained at 0.5×106 cells/mL in RPMI, supplemented with 25 mM HEPES (JRH Biosciences), 1% penicillin (10 000 U/mL)– streptomycin (10 000 µg/mL)–glutamine solution (29.2 mg/mL) (Gibco) and 10% fetal bovine serum (FBS) (Invitrogen). Manipulation of cells was carried out in a laminar flow hood under sterile conditions. Endotoxin-free medium and other reagents were used for experiments. Endotoxin concentrates were demonstrated to be below detectable limits by the LAL method (<0.1 ng/µg).26 Cells were checked to ensure that they were mycoplasma-free using an immunofluorescence assay (ICN Biomedicals). Human peripheral blood monocytes (PBMs) were isolated from healthy volunteers using Ficoll–Hypaque gradients and adherence to plastic cover slips, as previously described.27 Monocyte purity was >95% assessed by morphology and CD14 staining. Cells were activated by LPS (Salmonella typhimurium, Sigma, 5µg/mL). The ability of LPS to stimulate MMP activity was confirmed in a preliminary experiment by an increase in total MMP activity.

Pre-treatment of substratum

Glycation of the different substrata was achieved with an overnight incubation (37°C) with 50 mM glycolaldehyde (Sigma), followed by incubation with 1 M glycine ethyl ester (Sigma) for 1 h to quench the reaction.28 Experimental and control proteins were rinsed four times with PBS pH 7.2 and incubated overnight with PBS pH 7.2 to ensure removal of glycolaldehyde and glycine ethyl ester. Cross-linkage of selected substrata was achieved by an overnight incubation (37°C) with 1% glutaraldehyde, followed by rinsing with PBS as described above.29 Glycation of samples was assessed by fluorescence at an excitation wavelength of 370 nm and emission of 440 nm (Perkin Elmer Luminescence Spectrometer LS50B) characteristic for the AGE, pentosidine.30 Prior to fluorescence determination samples were digested with 200 U of type 1 collagenase (Worthington Biochemical Corp.) in 50 mM Tris–HCl (pH 7.4), 100 mM NaCl, and 15 mM CaCl2 for 24 h at 37°C. For example, glycated-BSA (27.9 ± 1.2 mean fluorescence units, n = 6, P = 0.01) and glycated collagen (44.5 ± 1.3 mean fluorescence units, n = 6, P = 0.01) demonstrated a large increase in fluorescence when compared with PBS (1.0 ± 0.1 mean fluorescence units) and aminoguanidine (0.7 ± 0.1 mean fluorescence units) treated controls.

Assessment of matrix metalloproteinases and cytokines

The response of monocytes to incubation with different substrata was assessed by measuring gelatin digestion activity and IL-6 concentration in the culture supernatant. Media harvested after 24 h of culture were centrifuged (1000 g for 5 min) to remove particulate debris and stored frozen in aliquots at −80°C until analysed, as previously described.31 Samples for comparison were always run on the same gel. Zymograms demonstrated reproducible bands at 92 and 65 kDa consistent with MMP-9 and MMP-2.31 Treatment of the zymograms with EDTA completely abolished these bands, also consistent with the activity of MMPs. Gels were assessed using the ChemiDoc™ imaging system (Bio-Rad Laboratories) and QuantityOne™ 1-D Analysis Software (Bio-Rad Laboratories), and activity was quantified by band volume taking into account area and density, referred to as relative density units (RDUs) by convention.31 In selected experiments, total MMP activity was also measured by cleavage of a fluorogenic substrate, 7-methoxycourmarin-4-acetyl-Pro-Leu-Gly-Leu-(3-[2,4-dintrophenyl]-L-2, 3-diaminopropionyl)-Ala-Arg-NH2 (American Peptide Company, USA), as previously described.28 Inter-assay coefficient of variation for these assays was between 3 and 6%.

Data analysis

Nominal and continuous variables were compared between patients with and without diabetes using Fisher's exact test and Mann–Whitney U test. All continuous variables were presented as mean±standard deviation. In calculating change in aortic diameter over time, we calculated mean yearly change (by comparing maximum aortic diameter in subsequent scans) and 3-year change (by comparing initial and last scans) for each patient. We then average these between patients mainly reporting mean yearly change. Some investigators have used modelling equations, such as Markov Chain Monte Carlo methods, to estimate aortic change.32 As complete follow-up was available in this cohort, we simply calculated mean yearly change in diameter over the 3-year follow-up (Figure 1). Trend analysis revealed that change in aortic diameter over time approximated linear; therefore, the association of diabetes with mean yearly growth was assessed with linear multiple regression analysis. Binary logistic regression was used to assess determinants of above median yearly growth over 3 years. For both regression models, we included other determinants that had previously been shown to be associated with aortic growth in order to adjust for these in the analysis. These factors were initial aortic diameter, smoking history, and PAD.32 As serum glucose and the incidence of dyslipidaemia were greater in patients with diabetes, we also constructed a further logistic regression model incorporating these factors in addition to initial aortic diameter, smoking history, and PAD. Serum concentrations of glucose, lipids, and CML were associated with changes in aortic diameter using Spearman's correlation. For the experimental study, MMP-2 and -9 activity (RDU) and IL-6 concentrations (pg/mL) were recorded as mean±standard deviation. Both set-up and measurements of outcome assays were carried out by the same scientist in an unblinded manner. The analysis of results, however, was carried out by an independent investigator. Results for paired artificial substrata experiments were compared by the Wilcoxon signed rank test. Findings from unpaired decellularized aortic studies were compared by the Mann–Whitney U test. A P-value of less than 0.05 was considered statistically significant in two-sided tests throughout. As the experiments carried out were separate studies to examine prior hypotheses, no adjustments in P-value for multiple testing were carried out.

Figure 1

Mean maximum abdominal aortic diameter (mm) during 3 years of follow-up for patients with and without diabetes. Vertical bars represent standard error of the mean

Results

Diabetes is associated with slower growth rate of abdominal aortic aneurysm

Only 20 (10%) of the 198 patients in our present cohort had diabetes, an incidence similar to that in our screening detected cohort (Table 1). The clinical characteristics of the patients with and without diabetes in our present cohort were similar, except that dyslipidaemia was more common in patients with diabetes (Table 1). Serum cholesterol (4.76 ± 1.06 mM), low-density lipoprotein (2.81 ± 1.01 mM), C-reactive protein (4.49 ± 5.87 mg/L), and creatinine (96.4 ± 28.1µM) were similar in patients with and without diabetes. Fasting glucose (8.56 ± 3.36 compared with 5.30 ± 0.62 mM, P < 0.0001), serum triglyceride (1.89 ± 1.12 compared with 1.49 ± 1.00 mM, P = 0.02), and serum high-density lipoprotein (1.07 ± 0.23 compared with 1.31 ± 0.40 mM, P = 0.001) were different in patients with and without diabetes. Over the 3 years of follow-up, the mean increase in aortic diameter was 1.88 ± 2.37 and 3.60 ± 3.49 mm in patients with and without diabetes, P = 0.02 (Figure 1). Mean yearly increase in aortic diameter was 0.63 ± 0.79 and 1.20 ± 1.16 mm in patients with and without diabetes, P = 0.02. After adjusting for other factors known to be associated with AAA growth (initial aortic diameter, smoking history, and PAD), diabetes negatively predicted mean yearly growth rate (β = −0.17, P = 0.01). After adjusting for initial aortic diameter, smoking history, and PAD, patients with diabetes had an odds ratio of 0.18 (95% confidence interval 0.06–0.57) of experiencing above median growth when compared with those without diabetes. These associations were unaffected by including dyslipidaemia or serum glucose in the models. One of the main AGEs is CML.33 We found no relationship between serum levels of CML, glucose, cholesterol, high-density lipoprotein, low-density lipoprotein, or triglyceride and aortic expansion over 3 years (r = −0.03, P = 0.74; r = 0.05, P = 0.51; r = 0.01, P = 0.87; r = 0.08, P = 0.24; r = −0.03, P = 0.70; r = 0.05, P = 0.47, respectively).

Incubation of activated monocytes with glycated collagen lattices reduces secretion of matrix metalloproteinases in association with cross-linkage

We examined the effect of in vitro glycation of different proteins on the secretion of MMPs and IL-6 by human monocytes (Table 2). Incubation of resting monocytes with glycated BSA induced a 5-, 1.5-, and 1.5-fold increase in MMP-9, MMP-2, and IL-6 secretion, respectively. Incubation of resting monocytes with glycated monomer collagen caused a small increase in MMP-9 secretion, but no change in MMP-2 and IL-6. Incubation of activated monocytes with glycated monomer collagen was associated with no change in MMP secretion (Table 2). In contrast, incubation of activated monocytes with glycated collagen lattices markedly reduced secretion of MMPs and IL-6. Total MMP activity of conditioned media from activated monocytes exposed to glycated collagen lattices reduced to 20.3 ± 2.6% of that secreted from cells exposed to PBS-treated collagen lattices, P = 0.02. Incubation of activated monocytes (Figure 2A) and PBMs (B) with glycated collagen lattices markedly reduced secretion of MMP-9 by 2-fold (Table 2). A reduction in MMP-2 (marked for PBMs, Figure 2B) and IL-6 (marked for monocytes, Table 2) secretion was also found in activated monocytes and PBMs incubated with glycated collagen lattices. As the conformation of the ECM is believed to play an important part in its cellular interactions, we postulated that our findings resulted from structural changes rather than effects on advanced glycation receptors.1417 Glutaraldehyde, unlike glycolaldehyde, induces cross-links in collagen without generating advanced glycation end-products.29,30 We compared monocyte responses with glycated and cross-linked collagen lattices. Cross-linked collagen lattices had similar effects as glycated collagen lattices on activated monocytes (Figure 2A) and PBMs (B), markedly reducing MMP-9 and MMP-2 secretion (Table 2).

Figure 2

Examples of zymograms from activated human monocytes (A) and activated human peripheral blood monocytes (B) exposed to phosphate-buffered saline-treated (control), glycated, or cross-linked collagen lattices for 24 h

View this table:
Table 2

Relative matrix metalloproteinase activity in medium from monocytes exposed to glycated and cross-linked proteins

MMPTreatmentControlP-value
Glycated BSA (resting monocytes)
MMP-27.83 ± 0.534.89 ± 0.220.02
MMP-911.27 ± 1.502.37 ± 0.620.02
IL-656.15 ± 16.5835.17 ± 12.320.02
Glycated monomer collagen (resting monocytes)
MMP-27.12 ± 0.627.30 ± 0.310.10
MMP-93.45 ± 0.412.88 ± 0.470.02
Glycated monomer collagen (LPS-activated monocytes)
MMP-28.47 ± 0.427.46 ± 1.430.11
MMP-921.75 ± 0.7420.17 ± 2.190.46
Glycated collagen lattice (resting monocytes)
MMP-21.41 ± 0.371.09 ± 0.150.10
MMP-90.67 ± 0.070.86 ± 0.090.04
Glycated collagen lattice (LPS-activated monocytes)
MMP-212.91 ± 0.8216.21 ± 0.800.02
MMP-96.00 ± 0.6512.08 ± 1.510.02
IL-620.63 ± 2.5340.63 ± 6.110.02
Glycated collagen lattice (LPS-activated PBMs)
MMP-25.86 ± 0.5114.73 ± 2.060.02
MMP-99.94 ± 0.2318.76 ± 3.770.02
Cross-linked collagen lattice (LPS-activated monocytes)
MMP-212.63 ± 0.3616.21 ± 0.800.02
MMP-96.47 ± 0.3112.08 ± 1.510.02
IL-674.79 ± 13.46157.80 ± 11.640.02
Cross-linked collagen lattice (LPS-activated PBMs)
MMP-24.47 ± 0.4914.73 ± 2.060.02
MMP-93.71 ± 1.1918.76 ± 3.770.02
  • Zymography (measured in terms of RDU) and ELISA (pg/mL, R&D Biosystems) were used to measure MMP activity and IL-6 concentration in the culture supernatants. The values (six samples/condition) are presented as mean±SD and compared between treatment and control samples using the Wilcoxon signed rank test. In all instances, the control condition for the treatment was incubation with PBS pH 7.2 in place of glycolaldehyde (50 mM) for glycation and glutaraldehyde (1%) for cross-linking. Activated monocytes or PBMs were treated with 5 µg/mL LPS.

Extracellular matrix isolated from aortic media also induces changes in monocyte matrix metalloproteinase secretion

In order to assess the relevance of our in vitro findings, we exposed monocytes to ECM isolated from aortic samples. To model healthy aortic ECM, we used decellularized sheep aortic media. Compared with collagen lattices, decellularized aortic media induced minimal secretion of MMP-9 by activated monocytes (0.58 ± 0.19 compared with 4.70 ± 0.22 RDU for PBS-treated collagen lattices, n = 6, P = 0.02, Figure 3A). Cross-linking the healthy aortic media had little effect on MMP secretion (Figure 3B), but did induce a reduction in IL-6 secretion by activated monocytes (141.78 ± 66.33 compared with 42.37 ± 30.49 pg/mL in controls, n = 6, P = 0.02). To model atherosclerotic aortic media, we collected samples from patients undergoing aortic bypass for lower limb ischaemia. Patients with diabetes (mean age 64.5 ± 8.0 years, five males and one female) were of similar age and gender to those without diabetes (mean age 65.7 ± 9.6 years, five males and one female). Decellularized media from patients with diabetes induced less MMP-9 secretion by activated monocytes than samples from patients without diabetes (5.82 ± 1.56 compared with 11.39 ± 1.49 RDU, n = 6, P = 0.02) (Figure 4). MMP-2 induction was not affected.

Figure 3

Comparison of the effect of collagen lattices (A) and decellularized healthy sheep aorta (B) with or without glutaraldehyde treatment on matrix metalloproteinase production by activated human monocytes. Cells were exposed to different matrices for 24 h and conditioned media subject to zymography

Figure 4

Comparison of the effects of decellularized aortic media from diabetic and non-diabetic patients on secretion of matrix metalloproteinases by activated human monocytes. When lipopolysaccharide-activated human monocytes were exposed to aortic media prepared from diabetic patients, the secretion of matrix metalloproteinase-9 was less than when they were exposed to samples from patients without diabetes

Discussion

In this study, we confirm the findings from one earlier investigation that diabetes protects against the progression of AAA.34 A previous study34 reported that diabetes was associated with slower growth of 40–49 mm AAAs during follow-up of 106 patients. This finding was not however adjusted for other determinants of AAA growth and did not apply to a larger number of patients with 30–39 mm AAAs in their study (n = 246). Unlike this previous study, in the present investigation, diabetes was associated with reduced AAA expansion in our cohort of 198 patients after adjustment for other determinants of AAA growth. We did not find any association between AAA progression and serum lipids, glucose, or the glycation end-product CML. These findings may be due to the inability of single blood sampling tests to give an overall picture of the degree of glycaemic control. The assessment of haemoglobin A1c may have been more revealing, but this was not available to us.

Monocyte–macrophage release of proteolytic enzymes and cytokines is critically implicated in the pathogenesis of AAA.7,8 In this study, we demonstrate that the response of human monocytes to advanced glycation depends on the type of protein glycated and the form in which it is presented. MMP-9 secretion by resting monocytes was induced by glycated BSA or glycated monomeric type I collagen (non-lattice), but inhibited by glycated collagen lattices (Table 2). Marked differences were demonstrated in the effect of different ECM proteins on activated monocytes. Glycated monomeric collagen had no effect on MMP secretion, whereas glycated collagen lattices induced marked reduction in the secretion of MMP-9, MMP-2, and IL-6 by activated monocytes (Table 2 and Figure 2). Less marked but similar changes were seen in the secretion of MMP-2 and IL-6 under these conditions. Advanced glycation end-products have been demonstrated to induce a pro-inflammatory response by stimulation of specific receptors, such as RAGE, which are expressed on monocytes and vascular smooth muscle cells within atherosclerosis.19,35 Such mechanisms likely explain the effect of glycated BSA in inducing an increased secretion of MMPs and IL-6, but do not fit with the inhibitory effects of glycated collagen lattices on monocytes. Glycation of structural proteins not only acts as a target for RAGE stimulation but also induces cross-links that increase matrix stiffness.18,30 In vitro, the response of cells to the ECM is modulated by the form in which it is presented.1417 Previous investigators have demonstrated that glycation alters the interaction between ECM proteins and endothelial cells, vascular smooth muscle cells, and fibroblasts.28,36,37 The mechanisms responsible for these effects have not been elucidated, but are believed to result from structural changes in the matrix proteins.36 We demonstrated that glutaraldehyde-treated collagen lattices, which only have cross-links and no glycation, had similar effects on activated monocytes to those of glycated collagen lattices. These findings suggest that cross-linkage formation within the three-dimensional gel matrix acts to alter the interaction between monocytes and collagen.

Healthy sheep aortic media induced minimal MMP-9 production by activated monocytes in comparison with PBS-treated collagen lattices (Figure 3). This finding requires confirmation with healthy human aortic media to which we did not have access. If confirmed, the ability of aortic media to induce minimal MMP production by monocytes is likely to be an important means by which medial integrity is maintained during inflammatory responses. Such mechanisms are likely to be significant during the chronic intimal inflammation that accompanies atherosclerosis.13,19 In this regard, it was interesting to compare the effects of aortic media removed from patients demonstrated to have atherosclerosis with and without diabetes (Figure 4). These studies demonstrated that unlike the findings with healthy aorta, incubation of monocytes with decellularized atherosclerotic aortic media did induce MMP-9 production. Secretion was however abrogated in cells exposed to diabetic ECM. High concentrations of both MMP-2 and MMP-9 have been demonstrated in biopsies of human AAA, and animal studies have suggested an important role for these proteolytic enzymes in aneurysm formation.38,39 The finding that glycated ECM markedly inhibits monocyte MMP production may in part explain the protection diabetes confers against AAA formation and growth.24,34

There are number of limitations of this study. We only recruited patients in which serum, clinical data, and 3 years ultrasound surveillance were available. However, the characteristics of our present cohort were similar to those of our screening detected group, including the incidence of diabetes (Table 1). Our findings will require confirmation in other surveillance studies. We chose to study the effects of glycation in vitro in order to address specific hypotheses in a controlled situation. We considered that the chronic nature of AAA made the study of prolonged diabetic effects such as ECM glycation of most interest in this instance. We used human PBMs and aortic samples in order to relate our studies to the patient. The array of cytokines and inflammatory cells demonstrated in human AAA biopsies illustrates the complexity of the pathology, which cannot be completely modelled in vitro.3 We chose to study IL-6 because of its association with AAA.40 It was impossible to investigate the large number of other cytokines involved such as IL-10. In vivo, a large range of other effects of diabetes will be present including direct effects of glucose on monocytes and the effects of swinging glucose concentrations.41 Further investigations using animal models of diabetes will provide additional information, but will also have limitations in being directly related to human AAA.

In conclusion, this study suggests that cross-linking of type I collagen lattices, such as those present in the aortic media, acts to inhibit MMP secretion from infiltrating activated monocytes. In contrast, glycation of soluble proteins and non-lattice collagens, more likely to be encountered within the atherosclerotic intima, stimulate MMP release from resting monocytes. These findings are relevant to the mechanisms underlying arterial occlusive disease as disparate from aneurysm formation.

Funding

This work is funded by the National Institute of Health, USA (RO1 HL080010-01), NHMRC (project grant 379600), and Diabetes Australia.

Acknowledgements

J.G. and P.E.N. hold Practitioner's Fellowships from the NHMRC, Australia (431503/458505). C.S.M. is a Fellow of the National Heart Foundation, Australia.

Conflict of interest: none declared.

References

View Abstract