Local Synovial Engagement of Angiogenic TIE-2 Is Associated With the Development of Persistent Erosive Rheumatoid Arthritis in Patients With Early Arthritis

Authors


Abstract

Objective

To examine the role of vascular endothelial growth factor (VEGF) and angiopoietin signaling in the diagnosis and disease outcome of patients with early arthritis.

Methods

Fifty patients with early arthritis (disease duration <1 year) who had not been treated with disease-modifying antirheumatic drugs (DMARDs) were monitored prospectively and were classified at baseline and after 2 years as having undifferentiated arthritis (UA), rheumatoid arthritis (RA), or spondyloarthritis (SpA). All patients underwent arthroscopic synovial biopsy at baseline. Synovial expression of VEGF, VEGF receptor, angiopoietin 1 (Ang-1), Ang-2, TIE-2, and activated p–TIE-2 was evaluated by immunohistochemistry. Serum levels of VEGF, Ang-1, and Ang-2 were measured by enzyme-linked immunosorbent assay. Secreted products of macrophages stimulated with Ang-1 and Ang-2 were measured using a multiplex system.

Results

Expression of Ang-1 was comparable between the patients with RA at baseline and patients with UA who fulfilled the criteria for RA over time (UA/RA), and it was significantly higher in patients with RA (P < 0.05) or UA/RA (P < 0.005) than in patients with SpA. TIE-2 and p–TIE-2 were more highly expressed in patients with RA (P < 0.005) or UA/RA (P < 0.05) than in patients with SpA. Ang-1 significantly enhanced the tumor necrosis factor–dependent macrophage production of cytokines and chemokines that are known to be elevated in the synovial fluid of patients with early RA. In RA, relative TIE-2 activation predicted the development of erosive disease (R2 = 0.35, P < 0.05).

Conclusion

Local engagement of synovial TIE-2 is observed during the earliest phases of RA, suggesting that TIE-2 signaling may contribute to disease development and progression or may indicate an attempt to protect against these processes. Early therapeutic targeting of TIE-2 signaling may be useful in improving outcome in arthritis.

Rheumatoid arthritis (RA) is a chronic inflammatory disease of unknown etiology characterized by synovial inflammation in multiple joints with hyperplasia of the synovial intimal lining layer, influx of inflammatory cells, and neovascularization. In a large proportion of patients, chronic inflammation results in cartilage and bone destruction, disability, and comorbidity ([1, 2]). In RA, early and aggressive treatment regimens have been shown to prevent or limit joint destruction, improve functional outcome, and decrease mortality risk ([3, 4]). The ability to make an early diagnosis and estimate prognosis in the individual patient is important in determining the choice of treatment, both in RA and in other inflammatory joint diseases. This decision-making ability is complicated by the great heterogeneity of disease activity and outcome among RA patients, which has become even more apparent since the development and application of the American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) 2010 criteria for RA ([5, 6]). Possibly, this clinical heterogeneity reflects molecular processes that are differentially activated in the RA patient subgroups. Analysis of synovial tissue, the primary site of inflammation in RA, might identify diagnostic and prognostic markers, as well as provide insight into cellular and molecular processes that contribute to pathogenesis.

Angiogenic processes likely play key roles in the initiation and perpetuation of synovial inflammation in RA, as well as in the development of joint erosions. Neovascularization allows for the influx of inflammatory cells in synovial tissue and provides nutrients and oxygen to the hyperplastic synovium ([7, 8]). In particular, vascular endothelial growth factor (VEGF) and angiopoietins 1 and 2 (Ang-1 and Ang-2), signaling via their tyrosine kinase receptors VEGFR and TIE-2, respectively, play an important role in angiogenesis and are expressed in the synovium of patients with inflammatory arthritis ([9-12]). VEGF promotes endothelial cell proliferation, and in combination with Ang-1, stabilizes new blood vessels ([13]). Ang-1 can also promote fibroblast-like synoviocyte (FLS) proliferation and matrix metalloproteinase (MMP) secretion ([14]). Ang-2 can antagonize Ang-1 activation of TIE-2, but also has unique proinflammatory capacities, such as cooperating with tumor necrosis factor (TNF) to induce expression of endothelial cell adhesion protein ([15, 16]). TIE-2 activation is prominently localized to synovial macrophages in RA, and Ang-1 and Ang-2 can also cooperate with TNF to induce distinct but overlapping patterns of inflammatory gene expression in human macrophages ([17]).

In RA, serum levels of VEGF, Ang-1, and Ang-2 are related to clinical parameters of inflammation and blood flow in inflamed joints, as measured by ultrasound ([18]). Controversy exists concerning the exact role of VEGF and Ang-1 in the development of erosions. In one study, no relationship was observed between VEGF and joint destruction in early RA ([19]). However, other studies of early arthritis patients have indicated that serum levels of Ang-1 and VEGF are related to inflammation and joint destruction at baseline and after 1 year of followup ([20, 21]). In the present study, we investigated the expression and activation of angiogenic pathways in the synovial tissue of a prospective cohort of disease-modifying antirheumatic drug (DMARD)–naive early arthritis patients in relation to diagnosis and to the development of persistent erosive disease.

PATIENTS AND METHODS

Patient population, synovial tissue acquisition and analyses, and serologic analyses

Fifty patients who had been consecutively included in our prospective early arthritis cohort (arthritis duration, defined as the time from the first swollen joint, <1 year), had been diagnosed as having RA, spondyloarthritis (SpA), or undifferentiated arthritis (UA) at inclusion, and had synovial tissue biopsy samples available for analysis were enrolled in this study. All patients had arthritis in at least 1 knee, ankle, or wrist joint, were DMARD-naive, were not taking corticosteroids, and had not received intraarticular steroid injections.

Treatment was initiated after baseline study procedures were completed, in order to start treatment during an early phase of the disease. Choice of treatment was made by the treating physician, with the aim of achieving low disease activity or remission. Diagnosis was made at baseline and after 2 years of followup according to established classification criteria for RA ([22]) and SpA ([23]). Patients were classified as having UA if no classification diagnosis could be made. In addition, patients were classified as having self-limiting, persistent nonerosive, or persistent erosive disease ([24]). Self-limiting disease was defined as no arthritis on examination in a patient who had not taken DMARDs or steroids in the preceding 3 months. Erosive disease was defined as the presence of erosions (modified Sharp/van der Heijde erosion score of ≥1) on radiographs of the hands and feet at the 2-year followup ([25]).

At study inclusion, we collected data on demographic features, disease duration, and disease activity parameters and obtained blood samples. In addition, all patients underwent arthroscopic synovial tissue biopsy sampling of a knee, wrist, or ankle joint as previously described ([26, 27]). Biopsy samples were stored, processed, stained for immunohistochemistry, and evaluated using digital image analysis as previously described ([28, 29]).

This study was approved by the Institutional Review Board at the Academic Medical Center, University of Amsterdam, and was performed in accordance with the Declaration of Helsinki. All study patients provided written informed consent prior to inclusion in the study.

Immunohistochemistry and digital image analysis

For each patient, 6–8 synovial tissue biopsy samples were pooled for immunohistochemistry. The synovial biopsy samples were snap-frozen en bloc in Tissue-Tek OCT (Miles) immediately after collection. Cryostat sections (5 μm) were cut and mounted on Star Frost adhesive glass slides (Knittelgläser). Sealed slides were stored at −80°C until used for immunohistochemistry. The sections were fixed with acetone, and endogenous peroxidase activity was blocked by immersion in 0.3% hydrogen peroxide and 0.1% sodium azide in phosphate buffered saline (PBS). Slides were incubated overnight at 4°C with primary antibody diluted in 1% bovine serum albumin/PBS. Primary antibodies used in this study were polyclonal rabbit antibodies specific for TIE-2, Ang-1, Ang-2, VEGF, VEGFR (all from Santa Cruz Biotechnology), and phosphorylated TIE-2 (Cell Signaling Technology). As a negative control, irrelevant/isotype-matched immunoglobulins were applied to the sections instead of the primary antibody, or the primary antibody was omitted. Sections were washed with PBS and incubated with swine anti-rabbit horseradish peroxidase (HRP)–conjugated antibodies (Dako), followed by incubation with biotinylated tyramide and streptavidin–HRP, and development with aminoethylcarbazole (Sigma) ([29]). Slides were counterstained with Mayer's hematoxylin and mounted in Kaiser's glycerol gelatin (Merck).

After staining, the sections were examined with digital image analysis. All sections were analyzed in random order by trained readers (DdL, GPMvdS, and MGHvdS) who were blinded with regard to the patients' clinical characteristics. The analysis was performed using a computer-assisted image analysis algorithm, as described in detail elsewhere ([30]). Images were acquired and analyzed using a Syndia algorithm on a QWin-based analysis system (Leica). Expression and/or phosphorylation of proteins was calculated as the number of positive cells per mm2 or the median integrated optical density (IOD) per mm2 of tissue, corrected for cellularity. Relative phosphorylation values were obtained by dividing the p–TIE-2 IOD by the total TIE-2 IOD.

Measurement of autoantibodies and cytokines in patient serum

The presence of IgM rheumatoid factor (IgM-RF) and anti–citrullinated protein antibody (ACPA) in patient serum that had been collected at study inclusion was measured using IgM-RF (Sanquin) and anti–cyclic citrullinated peptide 2 (Euro-Diagnostica) enzyme-linked immunosorbent assay (ELISA) kits, respectively. VEGF, Ang-1, and Ang-2 were measured using standard quantitative sandwich ELISAs (R&D Systems).

Monocyte isolation, macrophage differentiation, and analysis

Human peripheral blood mononuclear cells (PBMCs) were isolated from healthy volunteer donor buffy coats and differentiated into macrophages as previously described ([31]). Monocytes were differentiated into macrophages by culture at 5 × 105 cells per well in 24-well tissue culture plates in Iscove's modified Dulbecco's medium (IMDM; Invitrogen) supplemented with 1% fetal calf serum (FCS). Nonadherent cells were removed after 1 hour by washing with IMDM/1% FCS, and adherent cells were cultured for 7 days in IMDM/10% FCS supplemented with 100 μg/ml of gentamicin (Invitrogen) and granulocyte–macrophage colony-stimulating factor (GM-CSF) (5 ng/ml; BioSource International). On day 4, the medium was refreshed by replacing half of it with fresh IMDM/10% FCS supplemented with GM-CSF.

Differentiated macrophages were stimulated for 24 hours with medium alone or with recombinant human Ang-1 or Ang-2 (200 ng/ml; R&D Systems) in the presence or absence of TNF (10 ng/ml; Sigma-Aldrich). Cell-free supernatants were collected and analyzed for epidermal growth factor (EGF), eotaxin, fibroblast growth factor 2 (FGF-2), interleukin-12 p40 subunit (IL-12p40), IL-1 receptor antagonist (IL-1Ra), macrophage inflammatory protein 1α (MIP-1α), MIP-1β, transforming growth factor α (TGFα), and VEGF-A content using the relevant human single-plex assays (Bio-Rad) according to the manufacturer's protocol. Results were read with a Bio-Plex 200 system (Bio-Rad).

Statistical analysis

Statistical analysis was performed using SPSS software version 16.0. For comparison of differences in expression or phosphorylation of markers between the different diagnostic groups, the Kruskal-Wallis test was used. The Mann-Whitney U test was used to compare differences in expression or phosphorylation of markers between 2 outcome groups and diagnostic groups. Correlations were examined with Spearman's rank correlation coefficient. Univariable logistic regression analysis was performed to evaluate the predictive value of markers for disease persistence. The variance explained according to Nagelkerke's R2 is reported. Macrophage gene expression profiles were compared using Wilcoxon's signed rank test. P values less than 0.05 were considered statistically significant.

RESULTS

Characteristics of the study patients

Of the 50 patients included in the present study, 30 were classified as having RA at baseline and after 2 years of followup (RA/RA), 5 patients had UA at baseline but fulfilled the ACR/EULAR 2010 criteria for RA at the 2-year followup (UA/RA), 8 patients had UA at baseline that remained UA at the 2-year followup (UA/UA), and 7 patients had SpA at baseline and after the 2-year followup. Baseline characteristics of the diagnostic subgroups are shown in Table 1.

Table 1. Baseline characteristics of the study patients, according to the presence of erosive or nonerosive disease at baseline/followup*
 RA/RA (n = 30)UA/RA (n = 5)UA/UA (n = 8)SpA/SpA (n = 7)P
  1. P values for comparisons across groups were determined by Kruskal-Wallis test or chi-square test, as appropriate. P values less than 0.05 were considered significant. RA = rheumatoid arthritis; UA = undifferentiated arthritis; SpA = spondyloarthritis; DAS28 = Disease Activity Score in 28 joints; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; VAS = visual analog scale; IgM-RF = IgM rheumatoid factor; ACPA = anti–citrullinated protein antibody.
Age, median (range) years50 (37–56)50 (48–58)42 (32–53)43 (33–44)0.57
Disease duration, median (range) months5 (3–8)5 (3–10)5 (3–6)2 (1.5–5.5)0.45
DAS28, median (range)5.3 (4.4–6.3)3.4 (3.2–4.0)4.4 (2.9–4.9)3.7 (2.9–4.6)0.003
ESR, median (range) mm/hour27 (13–47)15 (14–53)38 (25–41)15 (9–43)0.80
CRP, median (range) mg/liter11.5 (4–37)11.8 (11–26)20.2 (8–36)7.3 (6.4–14.9)0.81
Joint counts, median (range)     
Tender joints (68 assessed)15 (5–23)3 (3–4)2 (1–3)4 (1–9)<0.001
Swollen joints (66 assessed)8 (5–14)1 (1–3)1 (1–3)2 (1–4.5)<0.001
Global disease activity, median (range), 0–100-mm VAS48 (25–72)44 (32–45)69 (42–80)35 (20–70)0.56
No. (%) IgM-RF positive16 (53)0 (0)0 (0)0 (0)0.01
No. (%) ACPA positive15 (50)0 (0)0 (0)0 (0)0.001

Enhanced synovial tissue Ang-2 expression in SpA and activation of the Ang-1/TIE-2 axis in RA

We first assessed serum concentrations of VEGF, Ang-1, and Ang-2 in relationship to the diagnosis in the study patients (Figure 1A). We detected serum concentrations of VEGF and Ang-1 that were similar to those in previous reports ([12, 20]). We observed no significant differences in the concentrations of VEGF, Ang-1, and Ang-2 between the different diagnostic groups (UA/UA, RA/RA, UA/RA, and SpA) (Figure 1A).

Figure 1.

Differential expression of angiopoietin 1 (Ang-1) and Ang-2 in the synovium of patients in different diagnostic groups. Serum levels (A), immunohistochemical staining (B), and synovial tissue expression (C) of vascular endothelial growth factor (VEGF), Ang-1, and Ang-2 in 4 diagnostic groups were determined: patients with undifferentiated arthritis (UA) at baseline and followup (UA/UA), those with rheumatoid arthritis (RA) at baseline and followup (RA/RA), those with UA at baseline and RA at followup (UA/RA), and those with spondyloarthritis (SpA) at baseline and followup. For the images shown in B, original magnification × 100. Results in A and C are expressed as the integrated optical density (IOD), corrected for tissue cellularity. Data in A and C are shown as box plots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and lines outside the boxes represent the 10th and 90th percentiles. Circles indicate outliers. ∗ = P < 0.05; ∗∗ = P < 0.005.

We then performed immunohistochemical analyses to evaluate synovial tissue expression of VEGF, Ang-1, and Ang-2 in the different diagnostic groups. Expression of each marker in both the synovial intimal lining and sublining layers was seen (Figure 1B). VEGF expression was observed predominantly in the synovial sublining of patients in all diagnostic groups. Ang-1 expression was observed in the intimal lining layer, the synovial sublining, and the sublining vasculature. Staining in the same synovial regions was observed for Ang-2 in all diagnostic groups. Digital image analysis revealed no significant differences in VEGF expression between the diagnostic groups (Figure 1C). Expression of Ang-1 and Ang-2 was comparable between the RA/RA and UA/RA groups. In contrast, expression of Ang-1 was significantly higher in the RA groups (both RA/RA [P < 0.05] and UA/RA [P < 0.005]) compared to the SpA group. Ang-2 expression was elevated in the SpA group as compared to the RA/RA (P < 0.005) and the UA/UA (P < 0.005) groups.

Next, we studied the expression of VEGFR and TIE-2, the receptors for VEGF and for Ang-1 and Ang-2, respectively. VEGFR expression was observed in the blood vessels of synovial tissue samples from all diagnostic groups (Figure 2A). TIE-2 expression was seen in the intimal lining layer, the synovial sublining, and blood vessels. Quantitative analysis revealed no significant difference in VEGFR expression between the different diagnostic groups (Figure 2B). Expression of TIE-2 was higher in the RA/RA and UA/RA groups as compared to the SpA group (P = 0.005 and P < 0.05, respectively) and higher in the RA/RA group as compared to the UA/UA group (P < 0.05).

Figure 2.

Differential expression and activation of TIE-2 in the synovium of patients in different diagnostic groups. A and B, Immunohistochemical staining (A) and synovial tissue expression (B) of vascular endothelial growth factor receptor (VEGFR), TIE-2, and p–TIE-2 in the 4 diagnostic groups were determined: patients with undifferentiated arthritis (UA) at baseline and followup (UA/UA), those with rheumatoid arthritis (RA) at baseline and followup (RA/RA), those with UA at baseline and RA at followup (UA/RA), and those with spondyloarthritis (SpA) at baseline and followup. For the images shown in A, original magnification × •••. Results in B are expressed as the integrated optical density (IOD), corrected for tissue cellularity. Data are shown as box plots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and lines outside the boxes represent the 10th and 90th percentiles. Circles indicate outliers. C, Levels of epidermal growth factor (EGF), fibroblast growth factor 2 (FGF-2), interleukin-12 p40 subunit (IL-12p40), IL-1 receptor antagonist (IL-1Ra), macrophage inflammatory protein 1β (MIP-1β), and transforming growth factor α (TGFα) in tissue culture supernatants from granulocyte–macrophage colony-stimulating factor–differentiated macrophages were determined by multiplex analysis after incubation of macrophages for 24 hours in medium alone or in the presence of tumor necrosis factor α (TNFα; 10 ng/ml), TNFα plus angiopoietin 1 (Ang-1; 200 ng/ml), or TNFα plus Ang-2 (200 ng/ml). Values are the mean ± SEM of 6 independent experiments. ∗ = P < 0.05; ∗∗ = P < 0.005.

As Ang-1 expression was elevated in the RA diagnostic groups as well, we examined the activation status of TIE-2 using phosphospecific antibodies. Activated TIE-2 was observed primarily in the intimal lining layer and blood vessels, but could also be seen in the synovial sublining (Figure 2A). Activated TIE-2 was significantly elevated in the RA/RA (P < 0.005) and UA/RA (P < 0.05) groups as compared to the SpA group (Figure 2B). The relative activation of TIE-2 (ratio of p–TIE-2 to TIE-2) did not differ between diagnostic groups. These data suggest that Ang-1/TIE-2 signaling is specifically activated in patients fulfilling the ACR/EULAR 2010 criteria for RA (even before the criteria are met), while Ang-2 is specifically up-regulated in SpA patients.

Macrophage TIE-2 signaling–induced cytokines, growth factors, and chemokines known to be elevated in the synovial compartment of patients with early RA

We next examined how TIE-2 activation in the synovial tissue of patients with early arthritis might contribute to the onset and progression of RA. The synovial compartment of early arthritis patients who later develop RA contains elevated levels of specific T cell–, macrophage-, and stromal cell–derived cytokines ([32, 33]). As we have previously found that TIE-2 is prominently activated in RA synovial macrophages and that Ang-1 and Ang-2 can cooperate with TNF to stimulate macrophage IL-6 production, we examined the ability of Ang-1 and Ang-2 to regulate the expression of myeloid-derived cytokines that are known to be elevated in the synovial fluid of patients with early RA ([17]). Stimulation of human PBMC-derived macrophages with Ang-1 or Ang-2 had no significant effect on macrophage production of EGF, eotaxin, FGF-2, IL-12p40, IL-1Ra, MIP-1α, MIP-1β, or VEGF-A (data not shown). Also, no significant induction of eotaxin or VEGF-A was noted following stimulation with TNF, either alone or in combination with Ang-1 or Ang-2 (data not shown).

However, while TNF alone failed to induce the production of EGF, FGF-2, IL-1Ra, or TGFα, we found that Ang-1 cooperated with TNF to stimulate significant macrophage production of FGF-2 (P < 0.05) and TGFα (P < 0.05) and that both Ang-1 and Ang-2 cooperated with TNF to induce IL-1Ra production (P < 0.05) (Figure 2C). TNF alone induced IL-12p40 and MIP-1β production (P < 0.05). IL-12p40 levels were further enhanced by Ang-1 (P < 0.05), but not Ang-2, and similar trends were observed with MIP-1β (Figure 2C).

Relationship between the relative engagement of synovial TIE-2 and the development of erosive disease in RA patients

Next, we investigated the serum and synovial tissue expression levels of these angiogenic markers in relationship to disease persistence and development of erosive disease in all patients fulfilling the ACR/EULAR 2010 criteria for RA after followup (n = 33). Two patients were lost to followup and could not be classified according to outcome. Five of these patients had self-limiting disease, 18 developed persistent nonerosive disease, and 10 developed persistent erosive disease. Table 2 shows the characteristics of the patients by RA type. Treatment was comparable between the group with persistent nonerosive RA and the group with persistent erosive RA. Patients with self-limiting RA received significantly fewer DMARDs than did patients with persistent diseases.

Table 2. Baseline characteristics of the RA patients, by type of RA*
 RA typeP
Self-limiting (n = 5)Persistent nonerosive (n = 18)Persistent erosive (n = 10)
  1. P values for comparisons across groups were determined by Kruskal-Wallis test or chi-square test, as appropriate. P values less than 0.05 were considered significant. RA = rheumatoid arthritis; DAS28 = Disease Activity Score in 28 joints; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; VAS = visual analog scale; IgM-RF = IgM rheumatoid factor; ACPA = anti–citrullinated protein antibody; DMARDs = disease-modifying antirheumatic drugs.
Age, median (range) years36 (26–46)52 (43–55)53 (34–66)0.36
Disease duration, median (range) months3.25 (2.25–4.25)4.5 (3–8)5 (3–9)0.50
DAS28, median (range)2.9 (2.4–3.4)5.6 (2.4–6.4)5.0 (4.3–5.5)0.05
ESR, median (range) mm/hour15 (8–22)27 (18–46)32 (11–70)0.48
CRP, median (range) mg/liter5.0 (3.0–11.0)15.5 (4.7–33.0)9.0 (4.0–30.8)0.52
Joint counts, median (range)    
Tender joints (68 assessed)3 (1–3)17 (8–25)7.5 (5–19)0.002
Swollen joints (66 assessed)1 (1–2)8 (5–19)7 (5–11)0.02
Global disease activity, median (range), 0–100-mm VAS45 (26–66)47 (35–72)27 (12–58)0.28
No. (%) IgM-RF positive1 (20)7 (39)6 (60)0.22
No. (%) ACPA positive1 (20)8 (44)4 (40)0.55
No. taking DMARDs    
Any DMARD216100.02
No DMARD320
1 DMARD296
2–3 DMARDs053
DMARD plus biologic agent021

Ang-1 and Ang-2 levels in serum (Figure 3A) and synovial tissue (Figure 3B) did not differ between the 3 outcome groups. Expression of TIE-2 was significantly increased in the group with erosive disease as compared to the group with self-limiting disease (P < 0.05), and p–TIE-2 was significantly increased in the groups with persistent nonerosive disease and persistent erosive disease as compared to the group with self-limiting disease (P = 0.005 and P < 0.05, respectively) (Figure 3C). The ratio of p–TIE-2 to TIE-2 was significantly lower in patients with self-limiting RA versus those with erosive RA (P < 0.005) and in those with persistent nonerosive RA versus those with persistent erosive RA (P < 0.05) (Figure 3C). All other markers were comparable between the persistent nonerosive and the persistent erosive RA groups.

Figure 3.

Association of synovial engagement of angiopoietin 1 (Ang-1)/TIE-2 signaling with persistence and progression of rheumatoid arthritis (RA). A and B, Serum (A) and synovial tissue (B) expression of vascular endothelial growth factor (VEGF), angiopoietin 1 (Ang-1), and Ang-2 in patients fulfilling RA classification criteria after 2 years of followup, with self-limiting, persistent nonerosive (persistent), or persistent erosive (erosive) RA. Persistent erosive disease was defined as a modified Sharp/van der Heijde erosion score of ≥1 on radiographs of the hands and feet at the 2-year followup. C, Synovial tissue expression of VEGF receptor (VEGFR), TIE-2, p–TIE-2, and the ratio of p–TIE-2 to TIE-2 in the 3 RA subgroups. Results are expressed as the integrated optical density (IOD), corrected for tissue cellularity. Data are shown as box plots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and lines outside the boxes represent the 10th and 90th percentiles. Circles indicate outliers. ∗ = P < 0.05; ∗∗ = P < 0.005.

No differences in synovial expression of Ang-1, Ang-2, TIE-2, p–TIE-2, or the ratio of p–TIE-2 to TIE-2 were observed between ACPA-positive and ACPA-negative patients, and neither RF nor ACPA was related to the development of erosive disease in this cohort (data not shown). Logistic regression analysis revealed that the relative activation of TIE-2, but none of the other markers we examined, was significantly related to the development of erosive disease in all RA patients (R2 = 0.35, P < 0.05).

DISCUSSION

Increasing effort is being put into the identification of biochemical markers that have potential diagnostic or prognostic value and might even serve as therapeutic targets that can guide personalized treatment decisions in RA ([34]). Among the serologic parameters, circulating ACPAs and RFs are quite specific for the diagnosis of RA and have been associated with a higher risk of developing destructive disease ([35]). Other serologic markers, such as VEGF ([20]), MMPs ([36]), macrophage inhibitory cytokine 1 ([37]), and soluble granzyme B ([38]), have also been associated with destructive disease in RA. Our results demonstrate that expression of Ang-1, and associated TIE-2 activation, is increased in the synovial tissue of early RA patients as compared to SpA patients, while Ang-2 expression is more prominent in SpA.

We have previously shown in the same early arthritis cohort that there was no clearcut relationship between cell infiltration (including infiltration by CD3+ T cells and CD68+ macrophages) and von Willebrand factor expression in the synovium on the one hand and diagnosis or disease outcome on the other ([39]). Thus, differences in the expression and/or engagement of Ang-1 and TIE-2 cannot be explained by differences in the level of synovial inflammation and vascularization. Rather, Ang-1/TIE-2 signaling appears to be specifically involved in pathogenic processes that are active in RA. In our assessment of patients fulfilling the ACR/EULAR 2010 criteria, we observed increases in the expression of p–TIE-2 in patients with persistent nonerosive disease or persistent erosive disease as compared to those with self-limiting disease, and increased TIE-2 activation (ratio of p–TIE-2 to TIE-2 expression) in RA patients with persistent erosive disease. Despite enhanced synovial TIE-2 phosphorylation in these patient groups, no differences in Ang-1 expression were observed, possibly reflecting a nonlinear relationship between Ang-1 levels and TIE-2 activation, contributions of Ang-2 to TIE-2 phosphorylation, or as-yet-unassessed potential differences in negative regulators of TIE-2 signaling, such as TIE-1 or vascular endothelial tyrosine phosphatase ([40]). Together, these data suggest that the level of p–TIE-2 expression and the ratio of p–TIE-2 to TIE-2 might be useful predictive biomarkers, as they could explain up to 39% of the variance in outcome in developing persistent erosive disease.

In murine models of RA, blockade of TIE-2 or neutralization of Ang-2 decreases not only synovial angiogenesis, but also arthritis development and severity, as well as joint destruction ([17, 41]). Our study demonstrates that an increased relative engagement of TIE-2 is prospectively related to the development of joint destruction, raising the possibility that this pathway plays an important role in joint destruction in patients with early arthritis. The exact mechanisms through which local stimulation of TIE-2 is associated with the development of persistent erosive RA remain to be elucidated and could reflect processes that are independently mediated by several cell populations. For example, Ang-1 promotes the RA FLS proliferation, survival, and invasiveness and cartilage destruction ([14]). Published data also indicate that both Ang-1 and Ang-2 can act as chemokines for FLS ([10]).

We have recently identified macrophages as prominent targets of Ang signaling in RA synovial tissue, and both Ang-1 and Ang-2 promote macrophage cytokine and chemokine production ([17]). Here, we also present data suggesting that Ang-1 could act, at least in part, by cooperating with TNF or other inflammatory cytokines to promote the activation of synovial macrophages and the subsequent release of cytokines that are elevated early in the development of RA ([32]). Direct effects of Ang-1 or Ang-2 on endothelial cells could also promote inflammation in early arthritis through induction of adhesion molecule expression by endothelial cells ([16, 42]). Last, although it is tempting to speculate that enhanced Ang-1 expression and TIE-2 ligation contribute to disease onset and joint destruction by one or more of these mechanisms, it is also possible that Ang-1 and active TIE-2 represent a protective attempt to restore synovial vascular homeostasis, as transgenic overexpression of Ang-1 can prevent inflammation in murine disease models ([43]). Therefore, a more global assessment of gene regulation in endothelial cells, FLS, and myeloid cells in response to Ang-1 and Ang-2 stimulation, as well as their contributions to tissue destruction and repair in inflammatory arthritis, is warranted.

Differences in microscopic and macroscopic synovial tissue vascularization between RA and SpA patients have been suggested by a variety of previous studies ([12, 44-46]). Ang-2, which is known to be involved in vessel destabilization and the sprouting of new vessels, is highly expressed in psoriatic arthritis as compared to RA, and increased expression of Ang-2 is associated with the distinctive appearance of tortuous vessels that are observed in SpA, but not RA, synovial tissue ([12, 46]). In the present study, we found that synovial Ang-2 expression is highly elevated in the early SpA group as compared to the other diagnostic groups. In addition to effects on blood vessel morphology, it will be of interest to determine if Ang-2 contributes to other distinct features of SpA, such as macrophage phenotype and bone remodeling. Compared to RA, synovial macrophages in SpA show a more distinct alternatively activated (M2) phenotype, which is associated with tissue remodeling ([47]). In this regard, Ang-2 has recently been shown to reinforce immunoregulatory M2 properties in human monocytes, but in comparison to Ang-1, it has overlapping, yet distinct, effects on macrophage gene expression ([48]). Recent evidence also suggests that TIE-2 is expressed on osteoblast precursors, and overexpression of Ang-1 can promote bone formation in vitro and in vivo ([49]). Similar experiments assessing the effects of Ang-2 have not been reported, and formal experimentation is needed to determine if differential involvement of Ang-1 and Ang-2 signaling to myeloid cells in RA and SpA contributes to distinct pathophysiologic characteristics of these diseases.

In conclusion, our data show that engagement of synovial TIE-2 occurs in the earliest phases of RA and predicts the onset of persistent erosive disease. These data need to be validated in an independent cohort study of early arthritis patients, including analyses of synovial biopsy samples, for which the current study provides the rationale. Conceivably, the use of combinations of the biomarkers described herein with other clinical and biologic parameters may further improve the predictive value. It is also tempting to speculate that targeting Ang-1 and TIE-2 therapeutically at the earliest stages of disease might be useful in improving outcome in RA, but additional experimental confirmation in patients with early RA is still required.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Reedquist had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. M. G. H. van de Sande, de Launay, de Hair, García, G. P. M. van de Sande, Gerlag, Reedquist, Tak.

Acquisition of data. M. G. H. van de Sande, de Launay, de Hair, García, Wijbrandts, Gerlag.

Analysis and interpretation of data. M. G. H. van de Sande, de Launay, de Hair, García, G. P. M. van de Sande, Wijbrandts, Gerlag, Reedquist, Tak.

Acknowledgments

We thank C. van der Horst (Academic Medical Center, University of Amsterdam) for help with digital image analyses, and T. Dekker, B. S. Dierdorp, and Dr. R. Lutter (Academic Medical Center, University of Amsterdam) for performing the multiplex analyses.

Ancillary