The Tie2 receptor antagonist angiopoietin 2 facilitates vascular inflammation in systemic lupus erythematosus
- 1Department of Nephrology, Hanover Medical School, Hanover, Germany
- 2Department of Gastroenterology, Hepatology and Endocrinology, Hanover Medical School, Hanover, Germany
- 3Institute of Pathology, Hanover Medical School, Hanover, Germany
- 4Department of Immunology and Rheumatology, Hanover Medical School, Hanover, Germany
- Correspondence to P Kümpers, Department of Nephrology, Hanover Medical School, Carl Neuberg Strasse 1, 30625 Hanover, Germany;
- Accepted 29 September 2008
- Published Online First 17 October 2008
Objective: To investigate the role of the angiopoietin–tyrosine kinase with Ig-like and epidermal growth factor-like domains (Ang–Tie) system in systemic lupus erythematosus (SLE). Endothelial activation is emerging as a key event for leukocyte recruitment and accelerated atherosclerosis in SLE. Recently, the endothelial-specific Ang–Tie ligand–receptor system has been identified as a major regulator of vascular responsiveness to inflammatory stimuli.
Methods: Ang1 (by immunoradiometric sandwich assay (IRMA)) and Ang2 (by ELISA) were measured in sera of 43 patients with SLE and 30 healthy controls. Expression of Ang2 was studied by immunohistochemistry in biopsies of human lupus nephritis.
Results: Circulating Ang2 concentrations were increased and concentrations of Ang1 decreased in patients with active SLE compared to healthy controls. This tendency was still present in inactive SLE, although less pronounced. Individual Ang2 concentrations correlated well with SLE Disease Activity Index (SLEDAI) score, proteinuria, double-stranded DNA (dsDNA) titre and soluble vascular cell adhesion molecule 1 (sVCAM-1). In a multivariate regression analysis, renal involvement was the only independent predictor for elevated Ang2. Serum Ang2 was identified as a strong predictor for disease activity by receiver operating characteristic (ROC) procedures and regression tree models. Protein expression of Ang2 was upregulated in glomeruli of patients with lupus nephritis.
Conclusions: These data indicate that Ang2-mediated disruption of protective Ang1/Tie2 signalling is operational in SLE. Ang2 might facilitate endothelial inflammation, permeability and contribute to premature atherosclerosis. Furthermore, circulating Ang2 may be a valuable new biomarker for disease activity in SLE. Strategies to control the deleterious effects of Ang2 may open new perspectives to prevent endothelial inflammation in SLE.
Systemic lupus erythematosus (SLE) is a multifactorial chronic autoimmune disease, in which tissues throughout the body, particularly the skin, joints and the kidneys are the targets of an ongoing inflammatory attack.1 2 In this process, the activational state of the endothelial layer is a major determinate for the initiation, localisation, extent and propagation of inflammatory damage.3 4 Endothelial activation is characterised by phenotypic changes from a quiescent, unresponsive to a responsive state. This process is associated with increased expression of luminal adhesion molecules (eg, vascular cell adhesion molecule 1 (VCAM-1)), leukocyte recruitment and disassembly of cell–cell junctions, finally resulting in loss of barrier function and tissue oedema.5 In SLE, several lines of evidence demonstrate excessive endothelial activation in response to immune complexes, anti-endothelial cell antibodies, anti-double-stranded DNA (dsDNA) antibodies, various cytokines (eg, tumour necrosis factor (TNF)α) and anti-phospholipid antibodies.6 7 8 9 10
Recently, angiopoietin (Ang)2 has emerged as a key mediator of endothelial cell activation.11 12 13 14 Ang1 and Ang2 are antagonistic ligands which bind with similar affinity to the extracellular domain of the tyrosine kinase with Ig-like and epidermal growth factor-like domains 2 (Tie2) receptor, which is almost exclusively expressed by endothelial cells.15 16 17 Non-redundant constitutively operational Ang1/Tie2 signalling maintains vessel integrity, inhibits vascular leakage, suppresses inflammatory gene expression and prevents recruitment and transmigration of leukocytes.18 19 20 21 In contrast, binding of Ang2 disrupts protective Ang1/Tie2 signalling and facilitates endothelial inflammation.3 11 15 18 22 Ang2 has been regarded as the dynamic regulator within the Ang/Tie system, since it constitutes a Weibel–Palade body-stored molecule (WPB), which is rapidly released and induced upon endothelial stimulation.13 14
To date, the role of Ang–Tie system in SLE has not been investigated. We performed a clinical study to correlate serum concentrations of Ang1 and Ang2 with clinical, laboratory and histological findings in patients with SLE.
A total of 43 German Caucasian patients with SLE were recruited from the Department of Immunology and Rheumatology and from the Department of Nephrology at Hanover Medical School (Hanover, Germany). The patients with SLE had at least four of the American College of Rheumatology (ACR) criteria for the classification of SLE.23 24 The study was performed in accordance with the declaration of Helsinki and approved by the institutional review board (no. 4373). Patients with malignant diseases and acute infections were excluded. Informed consent was obtained. Disease activity was assessed by the SLE Disease Activity Index (SLEDAI).25 According to the SLEDAI, patients in this cohort were subdivided into an “active” (SLEDAI⩾7) and an “inactive” SLE group (SLEDAI<7). Clinical and immunological characteristics of patients with SLE are shown in table 1. A total of 30 age-matched controls (52.0 (26.8–63.5) years, p = 0.8 vs patients with SLE; male/female = 14/16) were recruited from the Hanover Medical School staff. Hypertension was present in 3 (10%) controls, smoking in 3 (10%), none had diabetes (0%), 10 had a Body Mass Index (BMI)>25 (33%) and 9 had total cholesterol levels >200 mg/dl (30%).
White blood count, C-reactive protein (CRP), serum creatinine and urinalysis were measured by routine techniques. Renal function and proteinuria were quantified by 24-h urine collections. Antibodies against dsDNA were analysed by radioimmunoassay (IBL, Hamburg, Germany). Complement (haemolytic complement (CH50), C4 and C3) was quantified by nephelometry methods (Siemens, Marburg, Germany). Peripheral blood samples for quantification of angiopoietins and soluble VCAM-1 (sVCAM-1) were immediately placed on ice, centrifuged and stored at −80°C.
Angiopoietin 1 immunoradiometric sandwich assay (IRMA), and angiopoietin 2 and sVCAM-1 ELISA
Ang1 and Ang2 were measured by in-house IRMA and ELISA methods as described previously.26 27 Polyclonal, anti-human Ang1 affinity purified goat IgG and a monoclonal anti-human Ang1 mouse AB were obtained from R&D Systems (R&D, Abingdon, UK). Recombinant human Ang1 (90% recombinant, expressed in mouse NSO cells) was purchased from Sigma-Aldrich (Sigma-Aldrich, Munich, Germany). Recombinant human Ang2 monoclonal Ang2 AB and anti-Ang2 AB were purchased from R&D. Soluble VCAM-1 was measure by a commercially available ELISA (R&D) according to the manufacturer’s instructions.
Paraffin sections from three renal punch biopsies (lupus nephritis class IV with positive immunostaining for C3, C1q, IgA, IgM and IgG)28 were analysed. Tissue from two normal kidneys that were nephrectomised due to trauma served as controls. Specimens were cut into 2 μm thick sections, deparaffinised and preincubated with normal donkey serum (Jackson ImmunoResearch Laboratory, West Grove, Philadelphia, USA) to prevent non-specific binding. Antigen unmasking was performed by trypsin predigestion and microwave treatment (10 mM citric acid buffer (pH adjusted to 6.0)). The sections were incubated overnight at 4°C with primary polyclonal antibody (1:100) to human Ang2 (Santa Cruz Biotechnology, Santa Cruz, California, USA). The antibody was devoid of any crossreactivity with other proteins of the angiopoietin family. For fluorescent visualisation of bound primary antibodies, sections were incubated with a Cy3-conjugated secondary antibody (Jackson ImmunoResearch Laboratory) for 1 h. Control experiments were performed by omitting the primary antibody and replacing it with Tris-buffered saline (TBS).
Differences between patient groups and healthy controls were evaluated using non-parametric Kruskal–Wallis tests followed by two-sided Mann–Whitney U tests. Correlations between Ang1 and Ang2 serum concentrations and parameters of disease activity were calculated with the Spearman test. The Pearson correlation coefficient was calculated and linear regression analysis was performed after logarithmic transformation of Ang2 values (logAng2). Differences according to individual SLEDAI items were calculated by Mann–Whitney U test. Receiver operating characteristic (ROC) procedures and regression tree analysis were used to compare the diagnostic benefit between different laboratory parameters to differentiate between active and inactive SLE. The validity of the regression tree analysis model was controlled by minimising the 10-fold cross-validated relative error.29 30 Statistical significance was accepted at 5% probability concentrations. Data are displayed as mean (SD) unless otherwise stated. Data analysis was performed using SPSS (SPSS, Chicago, Illinois, USA) and GraphPad Prism software (GraphPad, San Diego, California, USA). Regression tree analyses were performed using CART software (Salford Systems, San Diego, California, USA).
Ang1 concentrations are decreased and Ang2 concentrations increased in active SLE
Mean serum Ang2 concentrations were markedly elevated in patients with active SLE (8.6 (11.21) ng/ml SD) compared to both, inactive SLE (1.4 (0.6) ng/ml, p = 0.010) and healthy controls (1.1 (0.6) ng/ml, p<0.001). Ang2 concentrations in patients with inactive SLE were still elevated compared to healthy controls (p<0.001) (fig 1A).
In turn, circulating Ang1 concentrations were lowest among patients with active SLE (36.0 (13.2) ng/ml) compared to healthy controls (53.28 (8.25) ng/ml, p<0.001). Likewise, Ang1 concentrations of the inactive SLE group were lower compared to healthy controls (42.9 (16.2) ng/ml, p = 0.001). The difference in Ang1 concentrations between patients with active and inactive SLE did not reach statistical significance (p = 0.142) (fig 1B). The calculated individual Ang2/Ang1 ratio, considered to better reflect the vascular microenvironment than the individual Ang concentrations, was dramatically increased in active SLE (24.5 (3.7)) compared to inactive SLE (3.7 (1.67), p<0.001) or healthy controls (2.0 (1.0), p<0.001), respectively. In addition, the Ang2/Ang1 ratio was significantly different between patients with active and inactive SLE (p<0.001) (fig 1C). Similar results for Ang1, Ang2 and the Ang2/Ang1 ratio were obtained when patients with anti-phospholipid syndrome (n = 3) were excluded from analysis. Neither Ang1 nor Ang2 concentrations were associated with regard to the respective immunosuppressive therapy regimen (table 1) (p = 0.334 and p = 0.413, respectively). In healthy controls, concentrations of Ang1 (p = 0.9), Ang2 (p = 0.23) and the Ang2/Ang1 ratio (p = 0.16) were not different between men (n = 14) and women (n = 16).
Serum Ang2 correlates with parameters of disease activity
A strong positive correlation was observed between Ang2 concentrations and SLEDAI (r = 0.66, p<0.001) (fig 2). Likewise, Ang2 concentrations showed a significant positive correlation with proteinuria (r = 0.46, p = 0.002), whereas no association was seen with renal function (r = 0.17, p = 0.355). Consistently, no Ang2 was detectable in urine samples from six healthy controls (data not shown).
There was a weak correlation of Ang2 concentrations and IgG anti-dsDNA antibody titres (r = 0.38, p = 0.01), whereas no association with CRP concentrations was present (r = 0.11, p = 0.48). C4, but not C3 and CH50, was negatively correlated with Ang2 serum concentrations (r = −0.32, p = 0.04). Ang2 has been shown to facilitate endothelial adhesion molecule expression. Of note, a strong positive correlation was detected between Ang2 and sVCAM-1 (r = 0.45, p = 0.006).
Similar results were obtained when the Ang2/Ang1 ratio was used instead of Ang2. In line with these data, circulating Ang1 did not correlate with any of the aforementioned parameters.
Excess circulating Ang2 is associated renal involvement in SLE
Next we wanted to address the organ specificity of elevated Ang2 by the means of individual SLEDAI items (table 1). Therefore, Ang2 as dependent variable was tested by multivariate regression analysis, using the following dichotomous SLEDAI items as independent variables: arthritis, renal involvement, cutaneous involvement, polyserositis, low complement, positive dsDNA titre, or leukopoenia. The variables neurological involvement, vasculitis and myositis were excluded due to limited sample sizes. Using backward elimination, renal involvement (p<0.001) was identified as being the only independent predictor of Ang2 levels. Accordingly, Ang2 levels were markedly elevated in patients with renal involvement (11.91 (2.03) ng/ml) compared to patients without renal involvement (2.03 (1.66) ng/ml; p = 0.007).
Serum Ang2 is a strong predictor for disease activity according to ROC procedures and regression tree models
To roughly estimate the potential use of Ang2 as a marker for disease activity we calculated ROC curves to identify different cut-off values for Ang2 that discriminate between patients with active SLE and patients with inactive SLE. The area under the curve (AUC) was 0.88 (0.06) (95% CI 0.76 to 1.005 p<0.001). A calculated Ang2 cut-off value of >2.0 ng/ml resulted in 89% specificity and a sensitivity of 81% in discriminating active from inactive SLE. In this regard, Ang2 outperformed the discriminatory ability of anti-dsDNA titres (AUC 0.82 (0.06) (95% CI 0.70 to 0.096, p<0.001)), C3 (AUC 0.83 (0.08) (95% CI 0.68 to 0.99, p = 0.001)) and C4 (AUC 0.79 (0.09) (95% CI 0.62 to 0.97, p = 0.002)).
Finally, we incorporated all laboratory parameters into a regression tree analysis model. According to this model, Ang2 cut-off of 2.9 ng/ml reliably separated patients with SLE with high (n = 11) from patients with low (n = 32) disease activity. The subgroup of patients with high disease activity was not subdivided further. The subgroup of patients with low disease activity was again subdivided by a C3 cut-off value of 0.84 g/litre. Our combined prediction model discriminated active from inactive SLE with 81% specificity and a sensitivity of 89%. The validity of this model was confirmed by cross-validation analysis. Figure 3 illustrates the combined prediction model for disease activity in our cohort as identified by regression tree analysis.
Renal protein expression of Ang2 is upregulated in patients with lupus nephritis
Immunohistology demonstrated distinct Ang2 protein expression in glomeruli alongside capillary loops in renal biopsies from patients with lupus nephritis (fig 4A). No staining was seen in sections of lupus nephritis in which the primary antibody was omitted (fig 4B). Similarly, no glomerular staining for Ang2 was observed in renal tissue from healthy kidneys (nephrectomy due to trauma) (fig 4C). In addition, Ang2 was abundantly detected in capillaries and within the endothelial layer of various-sized arteries (fig 4D).
In the present study, we measured serum concentrations of circulating Ang1 and Ang2 in patients with SLE for the first time. The decisive results are: (1) concentrations of circulating Ang2 are increased and concentrations of Ang1 decreased in patients with active SLE compared to healthy controls; (2) this tendency was still present in inactive SLE, although less pronounced; (3) individual Ang2 concentrations were closely associated with clinical and laboratory markers for disease activity; (4) Excess Ang2 was independently associated with renal involvement; (5) Serum Ang2 was identified being a strong indicator for disease activity; (6) Upregulated Ang2 protein expression was present in glomeruli of patients with active lupus nephritis.
The salient protective role of operational Ang–Tie signalling in vascular inflammation has been convincingly demonstrated in either Ang1 gain of function, or Ang2 loss of function animal models.14 19 21 Thurston et al reported that vessels in Ang1 overexpressing mice are resistant to plasma leakage caused by various inflammatory stimuli.19 Likewise, acute administration of Ang1 significantly reduced vascular inflammation in mice.21 Accordingly, Ang2-deficient mice did not exhibit any inflammatory response upon toxic or bacterial-induced peritonitis. Precise dissection of the underlying mechanisms revealed that Ang2 does not primarily affect endothelial cell inflammation itself, but facilitates endothelial activation and subsequent leukocyte transmigration in the presence of TNFα and various other inflammatory stimuli.14
Well in line with these data, we detected significantly elevated Ang2 concentrations and decreased Ang1 concentrations in patients with SLE with active disease. Assuming that endothelial activation in SLE represents an Ang2 dependent process, the amount of Ang2 within the circulation should presumably reflect the extent of activated endothelial surface. Indeed, we detected a strong correlation of Ang2 concentrations with SLEDAI score and sVCAM-1 in the present study.
So far, the issue of appropriate stimuli for Ang2 release in SLE has not been addressed. In SLE, excessive endothelial activation has been observed in response to immune complexes, anti-endothelial cell antibodies, anti-dsDNA antibodies, cytokines (eg, TNFα) and anti-phospholipid antibodies.6 7 8 9 10 Well in line with these findings, we detected a positive correlation of Ang2 with IgG anti-dsDNA antibody titres and a negative correlation with C3 levels. Furthermore, glomerular Ang2 expression was markedly upregulated compared to controls and resembled the endothelial pattern of immune complex deposition. These findings are in line with increased glomerular expression of Ang2 in preclinical models of glomerulonephritis.31 32 Davies et al demonstrated that inducible glomerular Ang2 overexpression in mice causes proteinuria and glomerular endothelial cell apoptosis.33 Again, we could show a close correlation between circulating Ang2 concentrations and vascular barrier function, using proteinuria as a surrogate marker for glomerular endothelial permeability. Of note, Ang2 concentrations were not associated with renal function per se, which almost excludes Ang2 elevation as a cause of moderate renal impairment. Based on our results and previous reports, we propose a model, in which dsDNA antibodies, complement and immune complexes might cause, promote and sustain endothelial activation in an Ang2 dependent fashion. Our data indicate that disruption of Ang1/Tie2 signalling by excessive Ang2 is a potential key mechanism that contributes to endothelial activation in patients with SLE.
Surprisingly, elevated Ang2 concentrations and Ang2/Ang1 ratio persisted in patients with low disease activity (SLEDAI<7). This finding implies that endothelial activation is not restricted to active disease and represents an ongoing feature during remission. Given a growing consensus on the role of vascular inflammation in promoting atherogenesis and the progression of atherosclerosis,34 35 chronic Ang2-driven endothelial activation may play an important role in the increased risk of accelerated cardiovascular diseases (CVD) in this population.36 37 38
There are several limitations to the present study. First, this study was not aimed at elucidating pathophysiological pathways of angiopoietin regulation in SLE, but was rather designed to evaluate a possible clinical significance of angiopoietins in SLE for the first time. Second, by choosing a cross-sectional design we cannot rule out excess circulating Ang2 as a consequence rather than a cause of glomerulonephritis. However, because of the fact that (i) Ang2 was absent in the urine of healthy controls, (ii) did not correlate with glomerular filtration rate and (iii) was expressed in SLE nephritis but absent in healthy kidneys, we favour the latter explanation. It would be of considerable interest to investigate the role of Ang2 in flare, and during follow-up under different immunosuppressive regimens. Finally, this pilot study was not designed to compare the diagnostic impact of Ang2 with other laboratory parameters in SLE (eg, dsDNA antibodies, complement), nor was it powered to define cut-off values for Ang2.
Circulating Ang2 had a prognostic value being at least equivalent to that of anti-dsDNA titres by using ROC procedures and was beyond that the strongest marker for active SLE (SLEDAI>7) when regression tree analysis was used (fig 3). According to SLEDAI items, Ang2 was independently associated with renal involvement. Presumably, this finding again reflects the extent of activated endothelial surface rather than indicating true organ specificity. Larger trials are needed to clarify this issue.
The results of the present study constitute the intriguing concept that Ang2 (as a new biomarker) can not only identify patients with active SLE, but might be a rational drug target for additional endothelial-targeted therapy at the same time. So far, the therapeutic efficacy of selective Ang2 blockade has already been demonstrated in murine models,14 39 40 and a phase I trial evaluating a recombinant Fc protein directed against the action of the angiopoietins in advanced solid cancer has been completed recently (http://www.clinicaltrials.gov/ct/show/NCT00102830).
We would like to thank Professor Hartmut Hecker (Institute of Biometrics, Hanover Medical School, Hanover, Germany) for excellent statistical support and Dr Ulrike Kümpers for critical reading of the manuscript. We are indebted to Katja Kniesch, Melanie Paschy and Heike Lührs for excellent technical assistance.
Competing interests None declared.
Ethics approval The study was performed in accordance with the declaration of Helsinki and approved by the institutional review board.