|Home | About | Journals | Submit | Contact Us | Français|
Urokinase-type plasminogen activator (uPA) is expressed at increased levels in stenotic, atherosclerotic human arteries. However, the biological roles of uPA in the artery wall are poorly understood. Previous studies associate uPA with both acute vasoconstriction and chronic vascular remodeling and attribute uPA-mediated vasoconstriction to the kringle—not the catalytic—domain of uPA. We used an in vivo uPA overexpression model to test the hypothesis that uPA-induced vasoconstriction is a reversible vasomotor process that can be prevented—and uPA fibrinolytic activity preserved—by: 1) removing the growth factor and kringle domains; or 2) anchoring uPA to the endothelial surface. To test this hypothesis we constructed adenoviral vectors that express: wild-type rabbit uPA (AduPA); a uPA mutant lacking the NH2-terminal growth-factor and kringle domains (AduPAdel); a mutant lacking catalytic activity (AduPAS→A), and a cell-surface anchored mutant (AdTMuPA). uPA mutants were expressed and characterized in vitro and in carotid arteries in vivo. uPAS→A had no plasminogen activator activity. Activity was similar for uPA and uPAdel, whereas AdTMuPA had only cell-associated activity. AduPAS→A arteries were not constricted. AduPA, AduPAdel, and AdTM-uPA arteries were constricted (approximately 30% smaller lumens; P ≤ 0.008 vs AdNull arteries). Papaverine reversed constriction of AduPA arteries. uPA-mediated arterial constriction is a vasomotor process that is mediated by uPA catalytic activity, not by the NH2-terminal domains. Anchoring uPA to the endothelial surface does not prevent vasoconstriction. uPA catalytic activity, generated by artery wall cells, may contribute to lumen loss in human arteries. Elimination of uPA vasoconstrictor activity requires concomitant loss of fibrinolytic activity.
Urokinase-type plasminogen activator (uPA) is a multifunctional protein expressed by cells of the artery wall. uPA is present at increased levels in atherosclerotic human arteries, in proportion to disease severity (1, 2). However, the biological roles of uPA in the artery wall are incompletely understood. Increased vascular uPA expression would enhance fibrinolysis, which could prevent venous and arterial thrombosis (3–6). However, uPA can also promote arterial constriction (7, 8) and is associated with constrictive (negative) remodeling (9), both of which cause lumen loss. Definition of the mechanisms through which uPA causes arterial constriction and constrictive remodeling could improve our understanding of vascular biology and might reveal improved therapeutic approaches which preserve uPA fibrinolytic activity but eliminate uPA vasoconstrictor activity.
Vasoconstrictor activity of uPA was first reported by Haj-Yehia et al (7). Catalytically active tc-uPA—but not inactive sc-uPA—enhanced phenylephrine-induced vasoconstriction in vitro. This observation would suggest that uPA catalytic activity causes vasoconstriction; however, the authors instead localized uPA contractile activity to the kringle domain, an NH2-terminal segment of uPA (3). Subsequent experiments by this group, using mice injected with wild-type (wt) and mutant (kringle-deleted) human uPA, also implicated the kringle domain in uPA-mediated vasoconstriction (10). However, in these experiments, uPA vasoconstrictor activity was inhibited by PAI-1. Because PAI-1 inhibits uPA catalytic activity and does not interact with the uPA kringle domain (11), inhibition of uPA-mediated vasoconstriction by PAI-1 suggests that the uPA catalytic site—not the kringle domain—mediates vasoconstriction,
In contemporaneous experiments aimed at investigating the role of uPA in atherosclerosis and further developing uPA-mediated gene therapy (12), we also discovered uPA-mediated arterial constriction (8). Arteries of hyperlipidemic rabbits infused with an adenoviral vector expressing rabbit uPA were significantly constricted, with lumen loss of over 40%. Here we report use of this model of arterial overexpression of species-homologous uPA to uncover the mechanisms through which uPA causes arterial constriction. We test whether uPA-mediated arterial constriction is independent of hyperlipidemia and whether constriction is due to increased vascular tone or constrictive (negative) remodeling. We also identify the domain of uPA that is responsible for arterial constriction and test whether uPA-mediated arterial constriction can be avoided—and fibrinolytic activity preserved—by anchoring uPA to the endothelial cell surface.
We used a wild-type (wt) rabbit uPA cDNA (8) as a template to generate mutant rabbit uPA cDNAs. Rabbit uPA that lacks the NH2-terminus was constructed by using the Transformer Site Directed Mutagenesis Kit (Clontech, Mountain View, CA) to delete the bases encoding amino acids 1–123 of the mature rabbit uPA protein. The primer used for mutagenesis (5′-CGTGAGCGACTCCGAACTCATCCAAGAGTGC-3′) deletes the growth factor (i.e., receptor-binding) domain and virtually all the kringle domain, leaving the signal peptide, the connecting peptide, and the catalytic domain. Rabbit uPA protein encoded by this construct is similar to both a human uPA mutant (13), and a naturally occurring variant known as “low-molecular weight uPA” (14). We generated a uPA mutant that lacks catalytic activity by using the Quikchange XL mutagenesis kit (Stratagene, La Jolla, CA) to introduce a T→G mutation at position 1132, changing the active-site serine to alanine (3). We used this highly specific genetic approach—rather than a pharmacologic approach using purified PAI-1—to test the role of uPA catalytic activity because PAI-1 also has interactions with uPA that are active-site independent (10,11). We constructed a cell-membrane bound uPA mutant by replacing the uPA stop codon with a sequence from the rabbit polymeric immunoglobulin receptor (15). This C-terminal sequence encodes a 21-amino acid transmembrane domain and a 19-amino acid cytoplasmic tail that targets protein expression to the cell membrane. We introduced two mutations (His→Ala and Val→Ala) that enhance apical vs basolateral cell-surface targeting (16). We previously reported kinetic characterization of plasminogen activation by human uPA with this transmembrane anchor (17).
We used five first-generation adenoviral vectors: AduPA (expresses wt rabbit uPA) (8); AdCMVNull (does not express a transgene) (18); AduPAdel (expresses the NH2-terminal deletion mutant); AduPAS→A (expresses the Ser→Ala mutant); and AdTMuPA (expresses the membrane-anchored mutant). The three new vectors (AduPAdel, AduPAS→A, and AdTMuPA) were constructed by ligating each of the mutant uPA cDNAs into pΔE1sp1A-derived shuttle plasmids (19). The shuttle plasmids were transfected into 293 cells along with the large ClaI fragment of dl327 (20). The mutant sequences were confirmed by sequencing the shuttle plasmid or DNA from the recombinant vectors. Vectors were propagated, titered, and stored as described (21). Purified vector stocks were from 2.6 × 1012 to 1.1 × 1013 particles/ml. All had fewer than 1 E1A-containing genome in 106 vector genomes, determined by polymerase chain reaction.
293 cells were infected by incubation for 24 hours with AduPA, AdNull, AduPAdel, or AduPAS→A at 80 vector particles/cell. The cells were rinsed with phosphate-buffered saline (PBS) and incubated with serum-free Medium 199 (M199). Conditioned medium (CM) was collected after 6 hours and stored at −80 °C. Plasminogen activator (PA) activity in CM was measured using human plasminogen (American Diagnostica, Stamford, CT) and the plasmin substrate S-2251 (Diapharma, West Chester, OH) (22). uPA protein in CM (40 μl) was detected by non-reducing SDS-PAGE (10% gels) and western blotting onto Hybond ECL membranes (Amersham Bioscience, Piscataway, NJ). uPA was detected with goat anti-rabbit uPA (#398, American Diagnostica; 1:1000), peroxidase-linked rabbit anti-goat IgG (#A5420, Sigma; 1:2000 or 1:4000) and enhanced chemiluminescence (ECL or ECL Plus, Amersham). Human single chain uPA (scuPA; American Diagnostica) was a positive control for the PA activity and western blot assays.
Because uPA-uPAR interactions are species-specific (23), we used rabbit vascular smooth muscle cells (VSMC) (24) to investigate cell-surface interactions of wt and mutant rabbit uPA. VSMC were grown to confluence in Dulbecco’s Modified Eagle Medium with 10% fetal calf serum. To release endogenous receptor-bound uPA, 0.1 M Glycine buffer pH 3.0 was added for 3 minutes, removed, and cells were rinsed with PBS (with 0.1 mM CaCl2 and 1 mM MgCl2). VSMC were then incubated with CM from vector-infected 293 cells (or M199 only as a negative control) for 1 hr at 37 °C. The medium was removed, cells were rinsed 3 times with PBS, lysed, and the cell membranes were isolated by temperature-induced phase separation of the lysate (25). Protein in extracts was measured with the DC protein assay (BioRad, Hercules, CA) and equal amounts of each sample were analyzed by polyacrylamide gel electrophoresis and western blotting, as above. Equal protein loading and transfer was confirmed by Ponceau Red stain of membranes. We used extracts of bovine endothelial cells stably expressing a glycosylphosphatidylinositol-anchored human uPA mutant (LUK+ASN cells) as a positive control (25).
To measure cell-surface activity of wt-uPA, uPAdel, and AduPAS→A, rabbit VSMC (3 × 104 cells/well) were plated in a 96-well plate, and grown to confluence (2 days). VSMC were rinsed with PBS, treated with acidic Glycine buffer, rinsed, and incubated for 1 hr with CM from uPA-expressing 293 cells or M199 alone. Cell surface uPA activity was measured by treating VSMC with plasmin (0.2 μM) for 10 minutes to convert single- to two-chain uPA, addition of aprotinin (10 μg/ml) to inhibit plasmin, addition of the uPA-specific substrate S-2444 (1.15 mM) and measurement of OD405 over time (25). Before adding uPA-containing CM to VSMC, the concentration of uPA in the CM was measured by western blotting. uPA concentration was equalized in all CM samples used in an individual experiment by dilution of the more concentrated samples.
To verify membrane targeting and cell-associated activity of TM-uPA, bovine aortic endothelial cells were transduced with AdNull, AduPA, or AdTMuPA (5000 particles/cell), or exposed to M-199 alone. Cells were washed with PBS and acidic Glycine buffer, lysed, and subjected to temperature-induced phase separation and western blotting, as above. CM was collected from other aliquots of transduced cells and assayed for uPA activity with S-2444, as above. Additional aliquots of transduced cells, in a 96-well plate, were washed with PBS, then incubated in M-199 without phenol red with human plasminogen and S-2251. OD405 was measured over time.
Carotid artery gene transfer was performed on normal-chow-fed, male, specific-pathogen-free New Zealand white rabbits (3.0–4.0 kg, Western Oregon Rabbit Company, Philomath, OR). This protocol achieves gene transfer almost exclusively to luminal endothelium (24). Briefly, common carotid arteries were dissected from base of neck distally to the crossing of the pharyngeal nerve, about 2 cm in length. Vectors were infused for 20 minutes in an isolated vessel segment at 8 × 1011 particles/ml. For measurements of uPA mRNA and protein, arteries were harvested 3 days later, and processed as described below. For measurements of arterial structure, arteries were perfusion-fixed in situ at physiologic pressure 7 days after gene transfer (24). For measurement of endothelial integrity and vascular reactivity, arteries were harvested 7 days after vector infusion, without fixation.
Papaverine-induced vasodilation was performed in vivo on carotid arteries that were either norepinephrine-treated or untreated (26). In these experiments, we also measured mean arterial pressures to determine whether local, uPA-mediated arterial constriction affected systemic blood pressure. A 22-gauge arterial catheter was inserted into the femoral artery, connected to a pressure transducer (Living Systems Instrumentation, Burlington, VT) and mean arterial pressure was recorded. Arteries were then exposed surgically and bathed in PBS (37 °C) for 5 minutes. Some arteries were pre-constricted by topical application of norepinephrine (10 μg/ml; Abbott Laboratories, Abbott Park, IL) at 37 °C for 2 minutes, followed by rinsing with PBS and application of papaverine (5 mg/ml; American Regent Laboratories, Shirley, NY) for 4 minutes. Artery diameters in vivo were measured using a digital camera and computer-assisted image analysis. Mean diameter was calculated from 3 measurements spaced 3 mm apart. The Office of Animal Welfare of the University of Washington approved all animal protocols.
uPA protein and PA activity in CM of explanted arteries were measured by western blotting and S-2251 assay, as above. We also measured direct uPA activity in CM collected over 4 hours by treating CM with plasmin and aprotinin, followed by addition of S-2444 and then measuring OD405 over time, as above. uPA activity in CM was normalized to wet artery weight. Because TM-uPA is not secreted, we used an in situ activity assay and explanted arterial segments to measure PA activity of TM-uPA and control arteries (8). Artery segments were removed, rinsed, and incubated in 96-well plates at 37 °C in M-199 without phenol red with Glu-plasminogen (0.9μM and S-2251 (0.8 mM). Plasmin generation was measured with a spectrophotometer as the increase in OD405. Readings were made with arteries removed from the wells. PA activity was calculated with reference to scuPA standards with normalization to wet arterial weight.
We measured uPA mRNA by quantitative RT-PCR 3 days after gene transfer. RNA was extracted with the RNeasy Mini kit (Qiagen, Valencia, CA). Samples were digested with DNase I, and 100 ng RNA was used as a template for RT-PCR using Taqman one-step RT-PCR master mix reagent kit (Applied Biosystems, Foster City, CA). Forward and reverse primers for uPA were 5′-TACGAAAACATACCATGCCCA-3′ and 5′-TGCACATAGCACCAGGGTATTC-3′. The probe sequence was 5′-CACAATTACTGCAGGAACCCAGACCACCA-3′. GAPDH mRNA was measured using forward and reverse primers 5′-TCATTGACCTCCACTACATGGTCTA-3′ and 5′-CGCTCCTGGAAGATGGTGAT-3′. The probe was 5′-TCCAGTATGATTCCACCCACGGCAA-3′.
Perfusion-fixed arteries were stored overnight in 10% paraformaldehyde, rinsed in PBS, and stored in 70% ethanol. Arteries were then cut transversely into 6–8 segments and processed into paraffin. Morphometry was performed on 2–6 sections per artery (usually 4–6; sections were omitted from analysis if they were torn or folded), with each section spaced approximately 3–4 mm apart. Sections were stained with Movat pentachrome, and images recorded with a digital camera. An observer blinded to the identity of the specimens used computer-assisted planimetry to measure the lengths of the internal elastic lamina (IEL) and the external elastic lamina (EEL) as well as the intimal and medial areas. The lumen area was determined by calculating the area within the IEL (assuming circular arteries in vivo; area within IEL = IEL circumference ÷ 4π), then subtracting the intimal area, measured on the same section (8).
Transduced arteries were removed, rinsed, and embedded in OCT. Frozen sections were stained with an antibody to human CD31 (1:30; Dako, Carpenteria, CA) and bound antibodies detected with the Vectastain ABC kit (Vector Laboratories, Burlingame, CA) and aminoethyl carbazole substrate (Invitrogen, Carlsbad, CA). Primary antibody specificity was confirmed by substituting an isotype-matched antibody (eBiosciences, San Diego, CA). Slides were counterstained with hematoxylin.
Freshly harvested arteries were trimmed of surrounding fat and connective tissue, cut into 3 mm rings, and connected to a force transducer in a myograph organ bath. Arteries were equilibrated under an initial tension of 30 mN in Physiological Salt Solution (PSS) aerated with 95% O2 and 5% CO2 at 37 °C. PSS contains (in mM): NaCl 119; KCl 4.7; MgSO4 2.4; KH2PO4 1.2; CaCl2 3.3; NaHCO3 25; EDTA 0.03; Dextrose 6. Relaxation to acetylcholine (1 nM – 10 μM) and sodium nitroprusside (1 nM – 10 μM) was measured after steady-state precontraction with phenylephrine (3 μM) (27).
Normally distributed groups with equal variances are reported as mean ± SEM and were compared with the unpaired t test. Groups that were not normally distributed or that had unequal variances are presented as median (25–75%) range and were compared with the Mann-Whitney rank-sum test. Rabbit body weights were compared by one-way ANOVA. Two-way ANOVA was used to discern separate as well as interdependent effects of the uPA NH2-terminus and active site on vasoconstriction. Two-way repeated measures ANOVA was used to analyze vascular reactivity data. Analyses were performed with the SigmaStat program.
293 cells infected with AduPA, AduPAdel, or AduPAS→A expressed uPA proteins of the expected sizes and in similar amounts (Fig. 1A). PA activity in CM from cells infected with AduPA and AduPAdel was also similar [3.6 (2.5–4.2) and 2.4 (2.3–2.7) IU/ml/mg, respectively; P = 0.2; Fig. 1B]. PA activity in CM from both untransduced and AdNull cells was less than 0.1% of these values [0.0014 (0.0013–0.0017) and 0.0019 (0.0016–0.0021) IU/ml/mg, respectively]. PA activity in CM from AduPAS→A cells was essentially identical to these control values [0.0014 (0.0012–0.0016) IU/ml/mg].
Rabbit VSMCs were acid treated to release receptor-bound endogenous uPA and incubated either with 293 cell CM that contained similar amounts of wt uPA, uPAdel, or uPAS→A or with CM from AdNull-transduced 293 cells. Membranes isolated from VSMC incubated with wt uPA (from AduPA) and uPAS→A had more uPA than membranes isolated from VSMC incubated with CM from AdNull- or AduPAdel-transduced 293 cells (Fig. 2A). Cell-associated uPA activity in acid-treated VSMC incubated with CM from AduPA-, AduPAdel- and AdNull-transduced 293 cells was highest in cells incubated with CM containing wt-uPA (P ≤ 0.03 vs AduPAdel and AdNull; Fig. 2B). Surprisingly, CM containing uPAdel also modestly increased cell surface PA activity. Detection of cell-bound uPAdel in this assay (Fig. 2B) but not in membrane extracts (Fig. 2A) likely reflects a weak, cell-surface interaction of uPAdel. This interaction—which must be independent of the uPA receptor (uPAR) because uPAdel lacks the uPAR-binding domain—appears relatively stable to rinsing of the cell surface with culture medium but is likely disrupted by the membrane purification protocol.
Rabbit carotid arteries transduced with the uPA-expressing vectors secreted uPA proteins of the expected sizes and in similar amounts (Fig. 3A and Fig. S1). PA activity—measured as rate of plasmin generation—was significantly elevated in CM from AduPA or AduPAdel arteries [0.023 (0.012–0.029) and 0.0033 (0.0014–0.067) IU/ml/mg, respectively, vs. 6.3 (5.6–19) ×10−7 IU/ml/mg in CM from AdNull arteries; Fig. 3B]. This large increase in PA activity is measured by assaying plasmin activity in a small volume of plasminogen-containing serum-free medium, in which uPA accumulates over several hours. Secreted uPA activates plasminogen in the absence of plasmin inhibitors (normally present in vivo), amplifying the apparent difference in uPA activity between AduPA and AdNull arteries. In contrast, the dose of AduPA infused here increases artery wall direct uPA activity by only about 7-fold, which is within the range reported for diseased human arteries (2, 8). PA activity was not elevated in CM from uPAS→A arteries [2.1 (1.7–4.8) × 10−6 IU/ml/mg; P = 0.1 vs AdNull arteries]. Despite a generally higher amount of uPA protein in CM from AduPAdel vs AduPA arteries (Fig. 3A and Fig. S1), PA activity of CM from AduPAdel arteries was only 14% of PA activity in CM from AduPA arteries. The disproportionately higher PA activity in CM of AduPA vs AduPAdel arteries was likely due to more efficient conversion of endothelial cell-bound, endogenous rabbit plasminogen to plasmin by uPAR-bound wt uPA than by uPAR-independent uPAdel (28). Plasmin generated in this manner could accumulate in medium conditioned by AduPA arteries, increasing the apparent PA activity. In support of this hypothesis, assay of the same CM samples with the uPA-specific chromogenic substrate S-2444—which measures uPA activity directly and not plasminogen activation—revealed that uPA activity in CM from AduPAdel arteries was 2–3-fold higher than uPA activity in CM from AduPA arteries arteries (1.9 ± 0.41 × 10−4 vs 8.3 ± 1.4 × 10−5 IU/ml/mg; P = 0.02; Fig. 3C). This increase was likely due to the higher level of uPA protein in CM of AduPAdel vs AduPA arteries (Fig. S1).
We initially observed uPA-mediated arterial constriction in hyperlipidemic rabbits, in which endothelial function may already be abnormal (29). Here we tested whether AduPA also constricted arteries in normolipidemic rabbits. One week after gene transfer, AduPA arteries had a 15–20% decrease in IEL and EEL length and a 35% decrease in lumen area compared to AdNull arteries (Fig. 4A–C). Mean arterial pressure in AduPA vs AdNull-infused rabbits was the same (52 ± 5 vs 50 ± 6 mmHg; n = 8–9; P = 0.6).
We next investigated whether uPA-mediated arterial constriction was due to altered vasomotor tone. We first established an in vivo model of reversible vasoconstriction. Diameters of untransduced carotid arteries were measured in situ at baseline, after norepinephrine-mediated vasoconstriction, and after subsequent application of papaverine. Papaverine completely reversed norepinephrine-induced constriction, returning the arteries to their baseline diameters (1.5 ± 0.054 mm at baseline; 1.2 ± 0.057 mm after norepinephrine; 1.5 ± 0.049 mm after papaverine; Fig. S2). We then tested whether papaverine would also reverse arterial constriction in uPA-overexpressing arteries. Arteries transduced with AduPA or AdNull were exposed after 1 week, and papaverine was applied followed by perfusion-fixation in situ. Papaverine-treated AduPA arteries were not constricted: IEL length, EEL length, and lumen area did not differ between papaverine-treated AduPA arteries and papaverine-treated AdNull arteries (3.9 ± 0.18 vs. 4.1 ± 0.090 mm for IEL; 4.4 ± 0.14 vs. 4.4 ± 0.081 mm for EEL; 1.1 ± 0.12 vs 1.1 ± 0.10 for lumen area; P ≥ 0.4 for all; Fig. 5 and data not shown).
Compared to AdNull arteries, infusion of AduPA or AduPAdel caused constriction, with shorter IEL and EEL lengths, and smaller lumens (Fig. 6A–C; P ≤ 0.05 for all comparisons of AduPA and AduPAdel to AdNull). In contrast, AduPAS→A arteries were not constricted (P ≥ 0.4 for all comparisons to AdNull arteries). These differences in arterial size were not due to differences in rabbit size because the weights of rabbits receiving the 4 vectors did not differ (3.2 ± 0.2; 3.5 ± 0.3; 3.4 ± 0.3; and 3.4 ± 0.2 kg for AduPA, AduPAS→A, AduPAdel, or AdNull-infused rabbits; P > 0.5 by one-way ANOVA). Two-way ANOVA confirmed that uPA catalytic activity (in wt uPA and uPAdel) was associated with decreased IEL length, EEL length, and lumen area (P < 0.001 for all), that an intact uPA NH2-terminus (in wt uPA and uPAS→A) was not associated with any of these features (P ≥ 0.3 for all) and that the association of catalytic activity with constriction was independent of the NH2-terminal domains (P = 0.9 for interaction). Intimal growth in AduPA and AduPAdel arteries was minimal, accounting for only about 10% of the lost lumen area [Fig. 6C–D; compare 0.05–0.06 mm2 of intimal area including endothelium (D) to 0.4 mm2 of lumen loss (C)]. Intimal area, medial area, and intimal/medial area ratio did not differ significantly between AduPA or AduPAdel arteries and AdNull arteries (Fig. 6D–F).
Membrane-anchored uPA (TM-uPA) expressed in cultured endothelial cells was cell membrane-localized, not secreted, and significantly increased cell surface plasminogen activation (Fig. 7A–C). AdTM-uPA was infused in vivo at a concentration that yielded levels of uPA mRNA similar to those obtained with AduPA (Fig. S3A). Explanted AdTMuPA arteries secreted no detectable uPA (Fig. S3B), suggesting that cell surface localization of TMuPA was maintained in vivo. Moreover, AdTMuPA arteries had significantly elevated vessel-associated PA activity establishing that TMuPA was active in vivo (Fig. S3C). Infusion of AdTMuPA caused arterial constriction with lumen loss (Fig. 8). As with AduPA and AduPAdel arteries (Fig. 6D), intimal growth in AdTMuPA arteries was minimal [0.036 (0.018 – 0.056) mm2; n = 10], accounting for <10% of the mean 0.4 mm2 lumen loss (Fig. 8B), with the remainder due to constriction.
To determine whether uPA-overexpressing arteries were constricted due to loss or dysfunction of endothelial cells, we harvested AduPA and AdNull arteries 7 days after vector infusion, stained for CD31, and tested vascular rings for responsiveness to acetylcholine and nitroprusside. All arteries had intact endothelium and responded equally well to both agents (Fig. S4).
We used an in vivo gene transfer system to investigate mechanisms of uPA-mediated vasoconstriction and to determine whether targeted modifications of uPA could eliminate vasoconstriction while preserving fibrinolytic activity. Our major findings were: 1) overexpression of uPA in endothelium causes arterial constriction via a papaverine-sensitive vasomotor pathway; 2) catalytic activity—not the NH2-terminal growth factor and kringle domains—is required for uPA-mediated vasoconstriction; 3) a transmembrane anchor localizes uPA to the endothelial surface but does not prevent vasoconstriction. These data suggest a role for uPA catalytic activity in the regulation of vascular tone and cast doubt on the general feasibility of approaches that aim to preserve uPA-mediated fibrinolysis while blocking uPA-mediated vasoconstriction (7, 10).
Previous studies suggested both acute vasoconstriction and constrictive (negative) remodeling as potential mechanisms for lumen loss in arteries with elevated uPA expression. For example, Higazi and colleagues reported acute vasomotor effects of tc-uPA both in vitro and in vivo (7, 10). Other experiments support a role for uPA—and its primary substrate plasminogen—in the more chronic process of arterial remodeling (9, 30, 31). We found that uPA-mediated vasoconstriction was completely reversed by papaverine (Fig. 5). Based on this observation and the fact that constrictive (negative) remodeling involves structural changes in the artery wall that are not acutely reversible (26, 32), uPA-mediated constriction 7 days after gene transfer is an acute vasomotor effect. In comparison, rabbit arteries that constrict in response to reduced blood flow gradually transition from vasoconstriction to constrictive remodeling. They dilate completely to papaverine at 3 days, only 50% at 7 days, and not at all at 14 days (26). It is possible that if uPA was expressed beyond 7 days arterial constriction would no longer be papaverine-responsive, consistent with constrictive (negative) remodeling. Longer studies, using viral vectors that express uPA at high levels for several weeks, will be required to investigate this possibility.
Our data associate vasoconstrictor activity with the uPA catalytic domain with a high level of confidence (Fig. 6; P < 0.001 by 2-way ANOVA). Specifically, point mutation of the active site serine (in uPAS→A) eliminated uPA-mediated vasoconstriction; whereas, deletion of the growth factor and kringle domains (in uPAdel) had no effect on uPA-mediated vasoconstriction. These results exclude a role for uPA receptor binding in uPA-mediated vasoconstriction and are also incompatible with an existing model of uPA kringle-mediated, active site-independent vasoconstriction (7, 10). We are uncertain how to reconcile these data, although there are at least five ways in which our experimental approaches differ. First, all of the data supporting kringle-mediated vasoconstriction were generated in rodents (7, 10). It is possible that uPA acts by different mechanisms in rodents and rabbits. Second, all the data that implicate the kringle as a vasoconstrictor were generated using human uPA and rodents or rodent tissues (7, 10). It is possible that human uPA has nonphysiologic effects when applied at high doses to rodent tissues. Third, uPA may act by different mechanisms in different vascular beds. Aortic rings were used in the studies that identify the kringle as a vasoconstrictor (7, 10); whereas, we studied carotid arteries. Fourth, data that directly show vasoconstrictor activity for the uPA kringle were generated exclusively in vitro (7); whereas our data were generated in vivo. In vitro experiments may not accurately model in vivo physiology. Fifth, other studies tested sc-uPA and tc-uPA separately in vitro whereas our in vivo system likely includes both sc-uPA (produced by the vectors) and tc-uPA (generated in vivo by interaction of sc-uPA and plasminogen). Our system may be more physiologically relevant, because both sc-uPA and tc-uPA are present in vivo.
The finding of uPA-induced arterial constriction (7, 8, 10) raises the question of why uPA-induced vasoconstriction has not been reported after injection of pharmacologic doses of uPA into humans (33). A likely explanation is that humans who receive uPA (or tPA) infusions to enhance fibrinolysis also receive direct vasodilators, including nitrates and calcium-channel blockers (34, 35). It is also possible that after intravascular infusion, uPA either does not penetrate the blood vessel wall adequately or does not persist for a sufficient period of time to trigger vasoconstriction.
Although arterial constriction has not been reported after therapeutic uPA infusion in humans, both animal and human in vivo studies support the existence of uPA-mediated arterial constriction. Clowes et al. noted papaverine-responsive arterial constriction 2 weeks after balloon injury of rat carotids (36), a model in which vascular uPA expression is upregulated (37). In humans, vascular uPA activity is directly correlated with severity of human coronary artery disease (2). This correlation could be due—at least in part—to constrictive effects of uPA that worsen arterial stenoses. Further delineation of the mechanisms of uPA-mediated vasoconstriction may provide new approaches that prevent development of arterial stenoses. Downstream mediators through which uPA might trigger vasoconstriction include: plasmin-mediated activation of matrix metalloproteinases with proteolytic activation of a vasoconstrictor or degradation of a vasodilator (38–40); activation or inactivation of a cell-surface protease-activated receptor (41); a low-density lipoprotein receptor-related protein-mediated pathway leading to intracellular calcium release (42); or stimulation of angiotensin signaling. Our finding that TM-uPA causes constriction, combined with the likelihood that cell surface-generated plasmin acts only at the cell surface (28) suggests that the critical substrate(s) through which uPA and plasmin induce constriction are on the endothelial surface.
Our study has implications for efforts aimed at improving the therapeutic efficacy of uPA by modifying the noncatalytic domains or by anchoring it to the cell surface, as part of gene- or cell-based therapies (12, 43, 44). If potentially harmful side effects of uPA such as vasoconstriction were mediated by the kringle or growth factor domains or by actions on an off-target cell type (such as VSMC) they might be avoided by domain deletion, co-infusion of a domain-specific antagonist, or attachment of uPA to the primary cellular target (e.g., the endothelium). Unfortunately, our data suggest that neither noncatalytic domain-blocking nor cell surface targeting strategies are currently feasible approaches to improve the therapeutic efficacy of uPA.
We thank Margo Weiss for administrative assistance, May Han for help with uPA mutagenesis, Shannon Graf and Hardeep Bhugra for assistance with surgeries, and Corey Sullivan for help with planimetry.
This work was supported by a grant from the National Institutes of Health (HL080597). Drs. Massey and Buckler were supported by NIH T32 HL07828. Dr. Massey was supported by a grant from the EVEREST Foundation and Dr. Dichek was supported by the Locke Charitable Trust.
This article is not an exact copy of the original published article in Thrombosis and Haemostasis. The definitive publisher-authenticated version of Thrombosis and Haemostasis 2009; 102(5): ISSN: 0340-6245; DOI: 10.1160/TH09-03-0161 is available online at: http://www.schattauer.de/en/magazine/subject-areas/journals-a-z/thrombosis-and-haemostasis/contents/preprint-online/sep-15th-2009/issue/special/manuscript/12044/show.html.