SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Objective

Systemic sclerosis (SSc; scleroderma) is a systemic connective tissue disease with an extensive vascular component that includes aberrant microvasculature and impaired wound healing. The aim of this study was to investigate the presence of antiangiogenic factors in patients with SSc.

Methods

Plasma samples were obtained from 30 patients with SSc and from 10 control patients without SSc. The samples were analyzed for the ability of plasma to affect endothelial cell migration and vascular structure formation and for the presence of antiangiogenic activity.

Results

Exposure of normal human microvascular dermal endothelial cells to plasma from patients with SSc resulted in decreased cell migration (mean ± SEM 52 ± 5%) and tube formation (34 ± 6%) compared with that in plasma from control patients (P < 0.001 for both). SSc plasma contained 2.9-fold more plasminogen kringle 1–3 fragments (angiostatin) than that in control plasma. The addition of angiostatin to control plasma resulted in inhibition of endothelial cell migration and proliferation similar to that observed in SSc plasma. In vitro studies demonstrated that granzyme B and other proteases contained in T cell granule content cleave plasminogen and plasmin into angiostatin fragments.

Conclusion

Plasminogen conformation in patients with SSc enables granzyme B and granule content protease to limit the proangiogenic effects of plasmin and increase the levels of antiangiogenic angiostatin. This increase in angiostatin production may account for some of the vascular defects observed in patients with SSc.

Systemic sclerosis (SSc; scleroderma) is a systemic connective tissue disease that affects the capillaries and small arteries, resulting in reduced blood flow, tissue ischemia, and defective wound healing (1). The spectrum of vascular abnormalities in SSc extends from Raynaud's phenomenon, with reduced blood flow to the skin, finger ulcerations, and the potential for gangrene (2), progressive pulmonary hypertension, and coronary microvascular disease that can result in myocardial ischemia, contraction band necrosis, and fibrosis (3), to abnormal tissue healing leading to scarring and sclerosis (3). SSc vascular disease also affects other organs, including the kidneys and the gastrointestinal tract. Although the pathogenesis of SSc is not completely understood, the 3 main components of the disease process include autoimmunity, tissue fibrosis, and aberrant microvascular function.

Vascular homeostasis requires both proangiogenic and antiangiogenic factors. The proangiogenic factors are secreted molecules that promote endothelial cell proliferation, migration, and tubulogenesis (4, 5). Cleavage products of several extracellular proteins, including matrix proteins such as type XVIII collagen and type IV collagen, and hemostatic proteins such as plasminogen, among others, have been shown to possess antiangiogenic activity (6, 7). The activity of these factors as well as other proteins controlling angiogenesis is tightly regulated in quiescent tissue. Its alteration under pathologic conditions may result in abnormal growth of the vasculature or defective repair processes. Earlier reports have suggested the presence of antiendothelial activity in SSc, but its nature has not been defined (8–10).

Plasminogen plays a complicated role in the regulation of vascular homeostasis because of its ability to serve as a precursor of both the protease plasmin that possesses proangiogenic activity (11) as well as the angiogenesis inhibitor angiostatin (12). Plasminogen cleavage by plasminogen activators at residues Arg561–Val562 produces plasmin, which has both proteolytic and fibrinolytic activities (13). The proteolytic activity controls activation of several important angiogenic modulators such as transforming growth factor β and matrix metalloproteinase (MMP) (14, 15). Plasmin also activates and releases growth factors such as vascular endothelial growth factor and fibroblast growth factor, thereby promoting angiogenesis.

Plasminogen can also be cleaved by several proteases, including urokinase plasminogen activator (uPA), tissue plasminogen activator (16), MMP-2, MMP-3, MMP-7, MMP-9, and MMP-12 (16–19), and plasmin reductase (20, 21) within its kringle domains, producing angiostatin fragments of various sizes that possess antiangiogenic activity. Thus, alterations in plasminogen processing can have a profound effect on angiogenic homeostasis.

The present study was designed to explore the hypothesis that patients with SSc demonstrate excess antiangiogenic activity that can account for the vascular abnormalities seen in this disease. To this end, plasma samples obtained from patients with SSc and from age- and sex-matched control patients were evaluated for antiangiogenic activity. Plasma from patients with SSc, but not that from control patients, markedly inhibited the angiogenic properties of normal endothelial cells. The SSc plasma contained reduced amounts of proangiogenic plasmin and increased levels of antiangiogenic angiostatin, which was produced by proteases in T cell granule content cleaving plasminogen at kringles 3 and 4. Thus, we observed that alterations in plasminogen processing may play an important role in the pathophysiology of SSc.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Patient population.

Plasma samples (n = 30) were collected from randomly selected patients with SSc during routine appointments at the Johns Hopkins Scleroderma Center. All patients in this group either met the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for the diagnosis of SSc (22), had 3 features of the calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias syndrome, or had Raynaud's phenomenon with abnormal nailfold capillaries and SSc-specific autoantibodies. Patients were classified as having diffuse cutaneous SSc (dcSSc) if they had skin changes proximal to the elbows or knees or had involvement of the trunk; otherwise, patients were classified as having limited cutaneous SSc (lcSSc) (22). Control plasma samples (n = 10) were collected from random patients undergoing coronary angiography at the Dartmouth-Hitchcock Medical Center; control patients had no known vascular disease, including angiographically excluded atherosclerosis, and no chronic illness of any kind.

A 40-ml volume of blood was drawn from the arterial sheath and directly deposited into tubes containing heparin, for separation of plasma from whole blood. The samples were centrifuged at 1,000g for 10 minutes at 5°C in an Allegra 6R centrifuge (Beckman Coulter, Fullerton, CA) and then were aliquoted into 1.5-ml tubes and centrifuged for 1 minute at 1,000g in an Eppendorf 5417R centrifuge (Madison, WI) that was maintained at 4°C, to remove insoluble material. A 300-μl aliquot of each sample was placed into ice-cold autoclaved Eppendorf tubes to which ice-cold 100% ethanol was added to make a final 80% ethanol mixture. The samples were precipitated overnight at −20°C and then centrifuged at 14,000g at 4°C. The precipitated proteins were resuspended on ice in phosphate buffered saline.

Endothelial cell migration into a scratch wound.

Two endothelial cell types were used: human dermal microvascular endothelial cells (HDMECs) and bovine aortic endothelial cells (BAECs). Cell migration was assessed using a standard wounding assay (23–30). Following the wounding procedure of human dermal blood microvascular endothelial cells (Lonza Biologics, Portsmouth, NH), 100 μg of plasma proteins from 20 SSc samples and 10 control samples diluted in 1 ml of serum-free endothelial basal medium (EBM) was added to duplicate wells, and the extent of migration was measured 6 hours later. Cell migration in serum-free medium was used as a control.

For plasma-mixing experiments, an angiostatin standard curve was established to determine the relative amounts of angiostatin in randomly selected SSc samples. Immunoblots containing varied concentrations of angiostatin and SSc plasma were probed for angiostatin kringles 1–3 (K1–3). Densitometry was used to determine the angiostatin concentration in SSc samples. Endothelial cells were treated with equivalent amounts of angiostatin (100 μg of plasma protein), purified angiostatin (0.44 μg), SSc plasma plus control plasma (100 μg of mixed plasma protein from 4 control patients), or angiostatin (0.44 μg) plus control plasma. The migration distance after 8 hours of incubation was measured.

Endothelial cell tube formation in a collagen overlay assay.

Two endothelial cell types were used: HDMECs and BAECs. Tissue culture plates were coated with ice-cold type I collagen (Cohesion Technologies, Palo Alto, CA) at a concentration of 1.5 mg/ml and pH 7.0. HDMECs (2 × 105) were seeded onto the center of each polymerized collagen-coated well and allowed to adhere for 1 hour at 37°C. The cells were then covered with ice-cold type I collagen, placed in an incubator, and incubated overnight at 37°C. EBM containing 2% fetal bovine serum and 100 μg of plasma from patients with lcSSc (n = 10), patients with dcSSc (n = 10), or control patients (n = 10) was added to triplicate wells and incubated at 37°C for 32 hours. The number of endothelial cell enclosures was determined at 24-hour intervals using digital imaging. The experiment was performed twice.

For mixing experiments, equivalent amounts of dcSSc plasma (n = 9) and lcSSc plasma (n = 9) were combined with control plasma (100 μg of mixed plasma protein from 4 control patients) before being added to endothelial cells contained within a 3-dimensional (3-D) type I collagen matrix. Each sample was examined in triplicate in 2 independent experiments.

Chromozym assay of plasmin activity.

Equivalent amounts of plasma protein (10 μg) from SSc samples (n = 30) or control samples (n = 10) were diluted in Chromozym PL assay solution (Roche Applied Science, Indianapolis, IN), according to the manufacturer's protocol. Triplicates of each sample were dispensed into 96-well plates and incubated at 37°C in each of 3 experiments. Absorbance was read at 405 nm after each hour of incubation. A plasmin standard curve consisting of 0–0.125 IU of uPA and 0.25 IU of plasminogen was established for each 96-well plate.

Western blotting.

For plasminogen activator inhibitor 1 (PAI-1) blots, equivalent amounts of plasma proteins from 30 patients with SSc and 10 control patients were resolved on a nonreducing/denaturing polyacrylamide gel, transferred to nitrocellulose membranes, and probed with a monoclonal antibody specific for amino acid residues 110–145 in human PAI-1 (American Diagnostica, Greenwich, CT), overnight at 4°C. PAI-1 signal was amplified in a 1-hour room temperature reaction with a rabbit anti-mouse polyclonal IgG (heavy chain plus light chain) secondary antibody (Pierce, Rockford, IL). All membranes were incubated with a donkey anti-rabbit antibody conjugated to horseradish peroxidase (Amersham, Arlington Heights, IL). SuperSignal West Pico chemiluminescent substrate (Pierce) was used to detect the binding reaction. Human serum albumin was used as a lane-loading marker.

Immunoblots containing plasma from 30 patients with SSc and 10 control patients were probed with an antiplasminogen K1–3 polyclonal antibody (1 μg/ml) (EMD Biosciences, Darmstadt, Germany) in an overnight incubation at 4°C. The binding reaction was amplified with a horseradish peroxidase–conjugated donkey anti-rabbit antibody (1:5,000) in a 1-hour room temperature reaction. Detection methods were the same as those described for PAI-1. Each sample was tested ≥3 times. Quantitative measurement of the 25-kd angiostatin K1–3 fragment in patient plasma was based on detectable levels of purified angiostatin ranging from 10 ng to 150 ng (standard curve). Angiostatin levels in the samples that contained <3 ng of angiostatin (undetectable) were registered as zero.

Granzyme B cleavage of plasminogen.

Purified lysine–plasminogen and glutamic acid–plasminogen (3 μM) (Roche Applied Science) were incubated with alioquots of granzyme B (3 μM) (EMD Biosciences) added incrementally at 0, 2, and 4 hours in a buffer composed of 50 mM HEPES/KOH (pH 7.0), 10% (weight/volume) sucrose, 2 mM EDTA, and 0.1% (w/v) CHAPS. The reaction mixtures were incubated at 37°C for 4, 6, and 16 hours and then gel resolved on a 4–20% nonreducing, denaturing polyacrylamide gel. Immunoblots containing the resolved proteins were probed for plasminogen K1–3 in duplicate, as described above. Plasma proteins (12.0 μg) from control and SSc patients were incubated with granzyme B, in the same conditions as described above.

Granzyme B in control and SSc patient plasma.

Granzyme B serum levels were analyzed by enzyme-linked immunosorbent assay, using a kit from Bender MedSystems (Vienna, Austria) according to the manufacturer's directions. Measurements were made from duplicate samples, and the experiment was performed twice. Samples were quantified based on a granzyme B standard curve (2–480 pg/ml), with the mean background values subtracted.

Granzyme B and granule content cleavage of plasminogen in SSc plasma.

Plasma proteins (12.0 μg) from control and SSc patients were incubated with aliquots of granzyme B (3 μM) (EMD Biosciences), purified human granule content (0.75 μM) (31), or a combination of both added incrementally at 0, 2, and 4 hours, as described above. The reaction mixtures were incubated at 37°C for 4, 6, and 16 hours and then gel resolved on a 4–20% nonreducing, denaturing polyacrylamide gel. Immunoblots containing the resolved proteins were probed for plasminogen K1–3 in duplicates, as described above.

Granzyme B cleavage of plasmin.

Varied amounts of plasmin (0.08–0.8 units) were incubated with aliquots of granzyme B (3 μM) or human granule content (0.75 μM) (31), added incrementally at 0, 2, and 4 hours, and incubated for 16 hours at 37°C, as described for plasminogen. Immunoblots containing the reaction mixtures were probed for plasminogen K1–3 in duplicate, as described above.

Statistical analysis.

Statistical analysis was performed with a Student's 2-tailed indirect t-test, one-way analysis of variance with a post hoc least significant difference test with or without repeated measures, or with a chi-square test, as appropriate, using the SPSS 12.0.1 statistical software package (SPSS, Inc., Chicago, IL).

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Study population.

Thirty patients with either dcSSc (n = 15) or lcSSc (n = 15) were enrolled in the study (Table 1). There was a preponderance of women (87%), and the mean ± SD age was 50.8 ± 11.1 years. A full spectrum of disease was represented, including patients with extensive lung involvement and pulmonary hypertension and patients with extensive skin involvement (mean ± SD skin score 9.44 ± 7.96 [on the modified Rodnan scale]). Ten patients (50% of whom were women) with no evidence of any vascular disease including atherosclerosis (mean ± SD age 60.6 ± 10.38 years [P not significant versus the SSc group) were used as controls. There were no differences in the incidence of hyperlipidemia, hypertension, or diabetes mellitus between the groups.

Table 1. Characteristics of the 30 patients with scleroderma*
  • *

    Except where indicated otherwise, values are the number (%). ACR = American College of Rheumatology; FVC = forced vital capacity; RVSP = right ventricular systolic pressure; RHC = right heart catheterization; DLCO = diffusing capacity for carbon monoxide.

  • Raynaud's phenomenon Medsger severity score of 3 or pulmonary artery hypertension.

Women/men26/4
Age at time of sample collection, mean ± SD years50.8 ± 11.1
Disease duration at time of sample collection, mean ± SD years9.45 ± 6.8
Limited disease/diffuse disease15/15
Race 
 White22
 African American8
Fulfilled ACR criteria for scleroderma29 (93)
Skin score, mean ± SD (range 0–51)9.44 ± 7.96
Raynaud's phenomenon, Medsger severity score 
 1.Raynaud's phenomenon18
 2.Digital pitting scars8
 3.Digital ulcerations4
Interstitial lung disease (FVC <80%)14 (47)
Pulmonary artery hypertension (RVSP >40 or RHC)9 (31)
Antitopoisomerase antibody positive (n = 21)4 (19)
Anticentromere antibody positive (n = 22)5 (22)
FVC, mean ± SD percent predicted (n = 29)81.7 ± 25.5
DLCO, mean ± SD percent predicted (n = 27)76.3 ± 27
Severe vascular disease11 (36.7)
Angiostatin level, mean ± SEM ng/ml (n = 10 positive)61.9 ± 38.5
Granzyme B level, mean ± SEM ng/ml (n = 7 positive)105.1 ± 78.2

Antiangiogenic activity of SSc plasma.

To assess the effect of SSc plasma on microvascular dermal endothelial cell function, we evaluated cell migration and the ability to form vascular structures in the presence of plasma from patients with SSc or control patients. Although cell migration was stimulated by control plasma, migration in the presence of either lcSSc plasma or dcSSc plasma was markedly decreased by a mean ± SEM of 52 ± 5% (mean ± SEM 2.9 ± 0.33 mm versus 6.14 ± 0.44 mm; P < 0.001) and was similar to the migration of cells in the absence of serum (Figure 1A). Similarly, the number of complete vascular structures formed in a 3-D type I collagen matrix in the presence of either lcSSc or dcSSc plasma was a mean ± SEM 34 ± 6% less than the number of structures formed in the presence of control plasma (mean ± SEM 57 ± 5 versus 86 ± 4; P < 0.001) (Figure 1B). The effect of SSc plasma on migration and tube formation of macrovascular (aortic) endothelial cells was even more pronounced, with a mean ± SD decrease of 46 ± 14% in migration and a mean ± SD decrease of 73 ± 16% in tube formation (results not shown). Thus, SSc plasma markedly inhibited migration and tube-forming abilities of normal endothelial cells, suggesting the presence of a circulating inhibitor(s).

thumbnail image

Figure 1. A, Endothelial cell migration. Graphs show the extent of migration of human microvascular dermal endothelial cells (HMDECs) incubated overnight in serum-free medium and stimulated with additional serum-free medium (0% fetal bovine serum [FBS]), systemic sclerosis (SSc) plasma, or plasma from control patients. Each of 9 samples from each patient group was tested 6 times. Note the reduced migration in the presence of SSc plasma. B, Three-dimensional collagen tube formation. The extent of enclosed tube formation by HMDECs in the type I collagen gel was measured after 72 hours in the presence of 10% FBS, SSc plasma, or control plasma. Nine samples from each patient group were tested (n = 6). Note the reduced tube formation in the presence of SSc plasma. C, Plasmin activity. Plasmin activity in 10 μg of plasma protein from SSc and control patients was measured in a Chromozym assay. The experiment was repeated 3 times, with triplicate wells tested per sample in each experiment (SSc, n = 30; control, n = 10). Values in AC are the mean ± SEM. l-SSc = limited cutaneous SSc; d-SSc = diffuse cutaneous SSc. ∗ = P < 0.05; ∗∗ = P < 0.001 versus control, by analysis of variance. D, Plasminogen structure. Glutamic acid–plasminogen contains amino acid residues 1–77, followed by 5 kringle domains and a serine protease domain. Plasmin can cleave glutamic acid–plasminogen after residue 77 to produce lysine–plasminogen. The kringle domains can be cleaved to produce subsets with varying degrees of angiostatin activity. The serine protease domain at the carboxy terminus can be cleaved at residues Arg561–Val562 by plasminogen activators. The 2 polypeptides produced by the cleavage are rejoined by 2 disulfide bonds to form plasmin.

Download figure to PowerPoint

Plasmin levels and activity in patients with SSc and control patients.

To begin exploring the reasons for the antiangiogenic effects of SSc plasma, we examined the plasminogen system. The measured plasmin activity was 58% lower (P < 0.001) and 43% lower (P = 0.01) in the dcSSc and lcSSc groups, respectively, when compared with control plasma (Figure 1C). This difference in plasmin activity among these patient groups could be secondary to relative differences in the amounts of PAI-1, plasminogen activator, or plasminogen itself.

Immunoblots of patient plasma proteins probed for PAI-1 demonstrated no significant differences in PAI-1 levels between the groups (data not shown). Therefore, PAI-1 was not a primary contributor to the reduced plasmin activity observed in patients with SSc.

Increased plasminogen cleavage products in patients with SSc.

To determine whether plasminogen was the limiting factor behind reduced plasmin activity in patients with SSc, plasma samples from SSc and control patients were probed for the presence of intact plasminogen and its K1–3 domain using an antibody specific to residues 101–353 (Figure 1D). Densitometric analysis of the blots normalized to human serum albumin indicated that intact plasminogen levels at 90 kd were not significantly different between all groups (Figures 2A and B), but patients with SSc clearly had more glutamic acid–plasminogen relative to lysine–plasminogen. However, measurements of plasminogen cleavage products showed variability in fragments ranging from 52 kd to 58 kd and a significant increase in the amount of the 25-kd K1–3 domain (angiostatin) in both lcSSc and dcSSc sera (2.4-fold and 3.3-fold, respectively; P < 0.05 for both) compared with control (Figures 2A and B). Thus, despite similar amounts of intact plasminogen, a significantly greater amount of it was cleaved into K1–3 fragments in SSc patients than in control patients, and this was more pronounced in patients with dcSSc than in patients with lcSSc.

thumbnail image

Figure 2. Association of granzyme B with plasminogen (Plg) cleavage. Plasma samples from 30 patients with systemic sclerosis (SSc) and 10 control patients were probed for plasminogen kringles 1–3 (K1–3). A, Representative immunoblot of plasminogen in plasma samples from patients with limited cutaneous SSc (lcSSc), patients with diffuse cutaneous systemic sclerosis (dcSSc), or control patients, detected with anti–K1–3 antibody. The antibody can detect K1–3 in full-length plasminogen (90 kd) or in plasminogen cleavage products that contain only K1–3 (25 kd) or K1–3 embedded within larger fragments (52–58 kd or 38 kd). B, Quantitative analysis of plasma samples from patients with lcSSc (striped bars), patients with dcSSc (hatched bars), or control patients (open bars), probed for angiostatin K1–3 under denaturing, nonreducing conditions. Data from all experiments were combined after normalizing calculated values to those obtained with 100 ng of purified plasminogen control and human serum albumin (HSA) for equivalent lane loading. Values are the mean ± SEM. ∗ = P < 0.05 versus control, by analysis of variance. C, Representative immunoblot of control and SSc plasma samples incubated with incremental increases of 3 μM granzyme B (GrB), 0.75 μM granule content (GC), or a combination of granzyme B and granule content in a 2-, 4-, or 16-hour incubation at 37°C. Immunoblots containing the reaction mixtures were probed for plasminogen K1–3. Note that granule content cleaves plasminogen to produce 25-kd angiostatin. D, Representative immunoblot of granzyme B, added in 80-ng increments, to various amounts of plasmin and incubated for 16 hours at 37°C. Note the angiostatin at 25 kd.

Download figure to PowerPoint

Cleavage of plasminogen in SSc plasma by granzyme B and granule content.

Although several proteases can potentially cleave plasminogen in a manner that would lead to angiostatin production, granzyme B is a particularly interesting candidate given its high levels in the peripheral blood cells of some patients with SSc (32, 33). To determine whether granzyme B can cleave plasminogen, SSc and control plasma samples were allowed to interact with granzyme B or the entire granule content (the source of granzyme B) in vitro (Figure 2C). Immunoblots containing the reaction mixtures probed for angiostatin K1–3 showed that plasminogen was cleaved by granzyme B and granule content in a time-dependent manner to produce angiostatin fragments at 50–54 kd and 29–38 kd. However, granule content protease(s) other than granzyme B were required for the final plasminogen cleavage to 25-kd angiostatin.

We then explored the possibility that granzyme B or granule content may cleave plasmin, which would explain the reduced plasmin levels in SSc plasma. Varied concentrations of plasmin incubated with granzyme B were cleaved into multiple-sized fragments, including the 25-kd angiostatin (Figure 2D). The unique association between plasmin and granzyme B suggests that granule content proteases may be a factor involved in the production of angiostatin in patients with SSc. To explore this possibility, we measured K1–3 angiostatin levels and the granzyme B concentration as a representative measure of the granule content in plasma from all patients with SSc and all control patients. Detectable levels of granzyme B were measured in 7 of 30 patients with SSc and in none of the 10 control patients. Detectable levels of angiostatin were observed in 10 patients with SSc and in none of the control patients. All 7 patients with elevated granzyme B levels also had elevated angiostatin levels. Angiostatin levels in the remaining 3 patients who did not have detectable granzyme B levels were significantly lower than those in patients with detectable levels of granzyme B (mean ± SEM 76.9 ± 13.8 ng/ml versus 27 ± 5.5 ng/ml; P = 0.05). These data strongly suggested that elevated granule content and granzyme B levels are directly responsible for the generation of angiostatin in patients with SSc.

Inhibition of proangiogenic activity in control plasma by SSc plasma.

To determine whether angiostatin in SSc plasma could inhibit angiogenic functions of endothelial cells stimulated with control plasma, we mixed 100 μg of dcSSc and lcSSc plasma with 100 μg of control plasma, and the mixtures were added to wounded monolayers of aortic endothelial cells in 1 ml of EBM. The dcSSc plasma reduced the promigratory effects of control plasma, from a mean ± SEM level of 12.43 ± 0.86 mm to 5.87 ± 0.35 mm (P < 0.001). The inhibitory effect of lcSSc plasma on control plasma–stimulated migration was comparable with that of dcSSc plasma (from a mean ± SEM level of 12.43 ± 0.86 mm to 6.65 ± 0.44 mm and 5.87 ± 0.35 mm, respectively; P < 0.01 versus control for both), and the effect was more potent than that of purified angiostatin (Figure 3A). Similarly, the number of complete vascular enclosures formed by control plasma mixed with dcSSc plasma was decreased by 68.3% (mean ± SEM 12.35 ± 0.96 for dcSSc plasma plus control plasma versus 38.9 ± 1.34 for control plasma; P < 0.001), and the number of complete vascular enclosures formed by control plasma mixed with lcSSc plasma was decreased by 67.3% (12.6 ± 0.83 for lcSSc plasma plus control plasma versus 38.9 ± 1.34 for control plasma; P < 0.001). Angiostatin also reduced vascular enclosure formation but to a more limited although statistically significant effect (38%) (Figure 3B).

thumbnail image

Figure 3. Angiogenesis assays of scleroderma (SSc) and control plasma mixtures. Equivalent amounts of SSc and control plasma were mixed and compared with purified angiostatin mixed with control plasma. A, Endothelial cell migration. Graphs show migration of bovine aortic endothelial cells (BAECs), incubated overnight in serum-free medium and stimulated with the addition of serum-free medium (0% fetal bovine serum [FBS]) (open bar), plasma from patients with limited cutaneous SSc (l-SSc) plus control plasma (CS), plasma from patients with diffuse cutaneous SSc (d-SSc) plus control plasma, control plasma, or angiostatin (Ang) plus control plasma. Note reduced migration of control plasma combined with diffuse cutaneous SSc or limited cutaneous SSc plasma. B, Three-dimensional collagen tube formation assay. The extent of enclosed tube formation by BAECs in the type I collagen gel, measured after 48 hours in the presence of 10% FBS (open bar), limited cutaneous SSc plus control plasma, diffuse cutaneous SSc plus control plasma, control plasma, or angiostatin plus control plasma. Note the reduced control plasma tube formation in the presence of diffuse cutaneous SSc or limited cutaneous SSc plasma. C, Effect of angiostatin on SSc serum suppression of BAEC migration. S + A = SSc plasma plus exogenous angiostatin; S − A = SSc plasma without exogenous angiostatin; C = control plasma. Data in B and C represent 9 randomly selected patients from each patient group; each patient was tested 6 times. Values in AC are the mean ± SEM. ∗ = P < 0.05; ∗∗ = P < 0.001 versus control, by analysis of variance. D, Regulation of plasminogen and plasmin levels in patients with SSc, by granzyme B (GrB) and granule content (GC). Plasmin removes residues 1–77 on glutamic acid–plasminogen to produce lysine–plasminogen, and lysine–plasminogen is converted to plasmin. Granule content proteases cleave lysine–plasminogen into 25-kd angiostatin. Granzyme B, a granule content protease, cleaves plasmin to produce 25-kd angiostatin.

Download figure to PowerPoint

Because angiostatin is not present in plasma from all patients with SSc, we stratified the antimigratory activity of SSc plasma in all patients according to the presence or absence of angiostatin and assessed its effect on endothelial cell migration in vitro. That analysis showed that SSc serum containing angiostatin had a much more pronounced cell migration inhibitory effect than did serum not containing angiostatin, but that angiostatin does not account for the entire antimigratory effect of SSc plasma (Figure 3C). Based on the results of these experiments, we concluded that dcSSc and lcSSc plasma contains antiangiogenic factors other than angiostatin, and that angiostatin accounts for ∼75% of the overall inhibitory effect of SSc plasma.

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

In this study, we showed that exposure of normal endothelial cells to plasma from patients with SSc results in a significant inhibition of 2 proangiogenic functions: endothelial cell migration and the ability to form vascular structures in 3-D collagen gel. We also demonstrated the presence of 3 abnormalities of the plasminogen system in SSc patients: reduced plasmin activity, greater amounts of glutamic acid–plasminogen versus lysine–plasminogen, and increased amounts of plasminogen and plasmin cleavage products containing angiostatin K1–4.5 and K1–3 fragments. Proteases released from T cell granules appear to be responsible for angiostatin production and reducing plasmin levels in patients with SSc. Taken together, these results suggest that SSc is associated with a systemic state that results in inhibition of angiogenesis, and that these defects might be related, in part, to abnormalities of the plasminogen system and production of angiostatin due to plasma release of T cell granules.

Abnormal angiogenesis in patients with SSc has been suggested as a cause of some of the key features of the clinical syndrome, including abnormal wound healing (34) and progressive pulmonary hypertension. Several studies have demonstrated angiogenesis-related abnormalities, including decreased monocyte-mediated angiogenesis (35) and the presence of proangiogenic and antiangiogenic factors in SSc serum (36), including increased levels of circulating endostatin (37, 38). At the same time, SSc serum was reported to enhance the angiogenic capability of circulating mononuclear cells (39). Finally, a recent gene expression study comparing transcriptomes of microvascular endothelial cells from normal control subjects and patients with SSc demonstrated overexpression of proangiogenic transcripts and both up-regulation of suppressor genes and down-regulation of activator genes regulating cell migration in SSc endothelial cells (40). Thus, to date, there is no clear understanding of the prevalence and significance of angiogenic abnormalities in SSc (41).

In this study, we set out to examine the function of the plasminogen system in SSc. Plasminogen is the source of both plasmin, a serine protease with marked proangiogenic properties, and angiostatin, an antiangiogenic peptide. The demonstration of strong antiangiogenic activity of SSc serum that was maintained in 1:1 ratio mixing studies suggested the presence of a circulating inhibitor and prompted us to examine in detail the components of the plasminogen system. We observed both a significant decrease in plasmin activity as well as a significant increase in angiostatin levels compared with those in control patients, thereby tipping the balance toward antiangiogenesis.

To investigate factors responsible for decreased plasmin activity, we first evaluated the amount of plasminogen and the presence of its inhibitor, PAI-1. However, we detected no differences in the level of either plasminogen or PAI-1 between the groups, suggesting that neither plasminogen nor PAI-1 could account for reduced plasmin activity. It should be noted that some studies have shown elevated PAI-1 levels in fibroblasts from patients with SSc (42–45), while other studies have not (46–48). It is not clear why results differ in this regard, other than due to relatively small sample sizes in all studies and the high heterogeneity of the SSc patient population.

We then turned our attention to changes in plasminogen structure that could account for reduced plasmin generation. Plasminogen can be present in either the glutamic acid–plasminogen form (which contains residues 1–77) or the lysine–plasminogen form (in which residues 1–77 have been removed). Glutamic acid–plasminogen residues 1–77 maintain it in a compact conformation through its intramolecular binding interactions with lysine residues in kringles 4 and 5 (49), thereby reducing activation of plasminogen into plasmin (50). An examination of glutamic acid–plasminogen versus lysine–plasminogen levels in SSc and control plasma showed an abundance of full-length glutamic acid–plasminogen in SSc plasma. Thus, this shift in plasma plasminogen content from the lysine to the glutamic acid form can be one factor explaining low plasmin activity in SSc plasma.

We then evaluated the presence of angiostatin, a plasminogen-derived fragment, in plasma from patients with SSc. Individual plasminogen kringles and various subsets of kringle domains have angiostatin activity, and many are more potent angiogenesis inhibitors than the 38-kd fragment containing K1–4, which was originally identified as angiostatin (51–53). Thus, the K1–3, K1–5, and K1–4.5 (kringles 1–4 plus 85% of kringle 5) fragments exhibit more antiangiogenic activity than K1–4 (12, 54, 55). Remarkably, we detected the presence of several “angiostatins” in the SSc patient plasma, including the presence of a 25-kd form (lysine K1–3). Furthermore, patient plasma containing angiostatin was significantly more inhibitory to normal endothelial cells, suggesting the functional role of angiostatin. This was further confirmed by demonstrating a similar extent of inhibition of angiogenic endothelial cell function by the addition of purified angiostatin in a concentration similar to that observed in patients with SSc.

There are numerous reports of enzymes that cleave plasminogen at the kringle domains to produce angiostatin containing either K1–3, K1–4, and K1–4.5 or combinations of the kringle subsets. Among those reported are various MMPs (17–19) and plasmin reductase (20, 21). Granzyme B, a serine protease found in T cell granules, is strikingly associated with cleavage of autoantigens targeted by autoantibodies in patients with SSc (33, 56). In agreement with this hypothesis, we demonstrated that both granzyme B and the full granule content could indeed cleave plasminogen into various-sized fragments containing angiostatin K1–3. Proteolysis appeared to be a 2-step process in which fragments K1–4.5 underwent further proteolytic processing. Interestingly, there was a highly significant correlation between the level of granzyme B and angiostatin in SSc plasma.

Although we detected a significant correlation between plasma granzyme B and K1–3 fragment levels, the pattern of granzyme B cleavage of plasminogen implied that a second protease is required to achieve the angiostatin K1–3 fragment at 25 kd seen in plasma from patients with SSc. Furthermore, the data clearly showed that granule content, but not granzyme B, was required to achieve cleavage into a 25-kd angiostatin fragment. Thus, we conclude that both granzyme B and another protease derived from T cell granule content is necessary for angiostatin generation in patients with SSc.

It remains unclear why plasmin levels are reduced in SSc plasma. We considered that granzyme B and/or another granule content protease may cleave plasmin to regulate plasmin conversion of glutamic acid–plasminogen into lysine–plasminogen. When incubated with plasmin, both granzyme B and granule content cleaved the intact plasmin into K1–3 fragments ranging from 29 kd to 54 kd. Granzyme B cleavage of plasmin also contained a 25-kd angiostatin fragment. The elevated levels of granzyme B, and by implication, granule content in the SSc plasma, explain in part the reduction in plasmin activity observed in patients with SSc.

Thus, patients with SSc display 2 principal abnormalities of the plasminogen system: a reduction in conversion of glutamic acid–plasminogen to lysine–plasminogen and production of 25-kd angiostatin. We propose that the primary abnormality is the release of T cell granule content that leads to cleavage of both lysine–plasminogen and plasmin to angiostatin. In addition to an increase in 25-kd angiostatin production, the presence of the T cell granule content proteases in plasma, including granzyme B, leads to a reduction in the plasmin level due to its conversion to angiostatin, which leads to reduced conversion of glutamic acid–plasminogen to lysine–plasminogen (Figure 3D). Thus, the release of T cell granule content is likely the primary event that accounts for all of the observations in this study: increased angiostatic activity in SSc patient plasma, increased presence of the 25-kd angiostatin, decreased conversion of glutamic acid–plasminogen to lysine–plasminogen, and reduced plasmin activity.

The presence of an increased amount of angiostatin, therefore, is likely responsible for impaired angiogenesis in some but not all patients with SSc. Furthermore, it is interesting to speculate that in addition to contributing to the antiangiogenic activity of SSc plasma, the presence of increased amounts of angiostatin may be a marker of increased matrix degradation, which is expected in patients with SSc. Some such degradation products have been shown to possess antiangiogenic activity, including the type XV/XVIII collagen breakdown product endostatin (6, 57) and the type IV collagen breakdown product tumstatin (58), among others (7).

In summary, we demonstrated that SSc is associated with 2 significant abnormalities of the angiogenic cascade: decreased presence and activity of proangiogenic plasmin, and increased production of antiangiogenic angiostatin. Both abnormalities are generated by the proteolytic activity of granzyme B and a second protease found in granule content. The data suggest a novel mechanism contributing to vascular defects in SSc that links immune-mediated proteolysis and altered angiogenesis. This insight may provide the opportunity to investigate novel therapeutic options.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Dr. Simons 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 design. Mulligan-Kehoe, Drinane, Rosen, Wigley, Simons.

Acquisition of data. Drinane, Mollmark, Casciola-Rosen, Hummers, Hall, Wigley.

Analysis and interpretation of data. Mulligan-Kehoe, Casciola-Rosen, Rosen, Wigley, Simons.

Manuscript preparation. Mulligan-Kehoe, Hummers, Rosen, Wigley, Simons.

Statistical analysis. Mulligan-Kehoe, Hummers, Wigley, Simons.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES