Advertisement

Plasminogen Activator Inhibitor-1 Regulates Tumor Growth and Angiogenesis*

Open AccessPublished:September 07, 2001DOI:https://doi.org/10.1074/jbc.M105980200
      Elevated expression of plasminogen activator inhibitor-1 (PAI-1) in tumors is associated with a poor prognosis in many cancers. Reduced tumor growth and angiogenesis have also been reported in mice deficient in PAI-1. These results suggest that PAI-1 may be required for efficient angiogenesis and tumor growth. In the present study, we demonstrate that PAI-1 can both enhance and inhibit the growth of M21 human melanoma tumors in nude mice and that this appears to be due to PAI-1 regulation of angiogenesis. Quantitative analysis of angiogenesis in a Matrigel implant assay indicated that in PAI-1 null mice angiogenesis was reduced ∼60% compared with wild-type mice, while in mice overexpressing PAI-1, angiogenesis was increased nearly 3-fold. Furthermore, addition of PAI-1 to implants in wild-type mice enhanced angiogenesis up to 3-fold at low concentrations but inhibited angiogenesis nearly completely at high concentrations. Together, these data demonstrate that PAI-1 is a potent regulator of angiogenesis and hence of tumor growth and suggest that understanding the mechanism of this activity may lead to the development of important new therapeutic agents for controlling pathologic angiogenesis.
      uPA
      urokinase-type plasminogen activator
      PAI-1
      plasminogen activator inhibitor type-1
      FGF-2
      fibroblast growth factor-2
      uPAR
      urokinase-type plasminogen activator receptor
      IP
      intraperitoneal
      CAM
      chorioallantoic membrane
      PBS
      phosphate-buffered saline
      TUNEL
      terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling
      Angiogenesis is the process of new blood vessel formation from pre-existing ones and occurs in healthy individuals during wound healing and ovulation (
      • Folkman J.
      ). In cases of disease, however, aberrant blood vessel formation may lead to abnormalities such as diabetic retinopathy, chronic inflammation, and arthritis (
      • Carmeliet P.
      • Jain R.K.
      ). The growth and metastasis of solid tumors is also dependent upon neovascularization, since in the absence of angiogenesis, tumors may grow only to ∼1–2 mm in diameter, and metastases are thought to be unable to spread (
      • Folkman J.
      ,
      • Carmeliet P.
      • Jain R.K.
      ).
      The process of angiogenesis requires the migration of endothelial cells and smooth muscle cells in response to an angiogenic stimulus. Hence, tissue remodeling by plasmin degradation of the extracellular matrix is likely to be important (
      • Carmeliet P.
      • Schoonjans L.
      • Kieckens L.
      • Ream B.
      • Degen J.
      • Bronson R.
      • De Vos R.
      • van den Oord J.J.
      • Collen D.
      • Mulligan R.C.
      ). Urokinase-type plasminogen activator (uPA)1 and tissue-type plasminogen activator both activate the zymogen plasminogen to plasmin, and localization of uPA to the leading edge of migrating cells is thought to initiate a proteinase cascade, which degrades the extracellular matrix and promotes cell migration (
      • Johnsen M.
      • Lund L.R.
      • Romer J.
      • Almholt K.
      • Dano K.
      ).
      Plasminogen activator inhibitor-1 (PAI-1) is the main inhibitor of plasminogen activators (
      • Fay W.P.
      • Shapiro A.D.
      • Shih J.L.
      • Schleef R.R.
      • Ginsburg D.
      ,
      • Carmeliet P.
      • Kieckens L.
      • Schoonjans L.
      • Ream B.
      • Van Nuffelen A.
      • Prendergast G.
      • Cole M.
      • Bronson R.
      • Collen D.
      • Mulligan R.C.
      ). It exists in multiple conformational states, including active, latent, and cleaved forms, and both latent and cleaved PAI-1 are thought to be biologically inactive. PAI-1 also interacts with a number of non-proteinase ligands, including the cell adhesion protein, vitronectin (
      • Lawrence D.A.
      • Palaniappan S.
      • Stefansson S.
      • Olson S.T.
      • Francis-Chmura A.M.
      • Shore J.D.
      • Ginsburg D.
      ,
      • Deng G.
      • Royle G.
      • Wang S.
      • Crain K.
      • Loskutoff D.J.
      ). PAI-1 binds to vitronectin with a high affinity; however, when in complex with a proteinase, its affinity for vitronectin is markedly decreased (
      • Lawrence D.A.
      • Palaniappan S.
      • Stefansson S.
      • Olson S.T.
      • Francis-Chmura A.M.
      • Shore J.D.
      • Ginsburg D.
      ). We have previously shown that PAI-1 bound to vitronectin inhibits vascular smooth muscle cell migration by a mechanism that is independent of its ability to inhibit proteinase activity, but instead acts by preventing binding of the αVβ3 integrin to vitronectin (
      • Stefansson S.
      • Lawrence D.A.
      ). Furthermore, in the chick chorioallantoic membrane (CAM) assay, PAI-1 inhibits angiogenesis via two overlapping mechanisms that depend on either its ability to inhibit proteinase activity or to block integrin binding to vitronectin (
      • Stefansson S.
      • Petitclerc E.
      • Wong M.K.
      • McMahon G.A.
      • Brooks P.C.
      • Lawrence D.A.
      ). Latent PAI-1, which lacks both these functions, was unable to inhibit angiogenesis in this assay.
      Since PAI-1 is a potent inhibitor of cell migration in vitro(
      • Stefansson S.
      • Lawrence D.A.
      ,
      • Kjoller L.
      • Kanse S.M.
      • Kirkegaard T.
      • Rodenburg K.W.
      • Ronne E.
      • Goodman S.L.
      • Preissner K.T.
      • Ossowski L.
      • Andreasen P.A.
      ) and angiogenesis in vivo (
      • Stefansson S.
      • Petitclerc E.
      • Wong M.K.
      • McMahon G.A.
      • Brooks P.C.
      • Lawrence D.A.
      ), it is striking to note that increased expression of PAI-1 is associated with a poor prognosis in many cancers (
      • Nekarda H.
      • Siewert J.R.
      • Schmitt M.
      • Ulm K.
      ,
      • Pedersen H.
      • Brunner N.
      • Francis D.
      • Osterlind K.
      • Ronne E.
      • Hansen H.H.
      • Dano K.
      • Grondahl-Hansen J.
      ). PAI-1 has also been suggested to modulate tumor growth and angiogenesis, since both tumor development and vascularization were reduced in PAI-1-deficient mice (
      • Bajou K.
      • Noel A.
      • Gerard R.D.
      • Masson V.
      • Brunner N.
      • Holst-Hansen C.
      • Skobe M.
      • Fusenig N.E.
      • Carmeliet P.
      • Collen D.
      • Foidart J.M.
      ,
      • Gutierrez L.S.
      • Schulman A.
      • Brito-Robinson T.
      • Noria F.
      • Ploplis V.A.
      • Castellino F.J.
      ). However, in contrast to these results, the growth of bib16 murine melanoma tumors was unaffected by the levels of PAI-1 expression in either PAI-1 null or PAI-1 overexpressing mice (
      • Eitzman D.T.
      • Krauss J.C.
      • Shen T.
      • Cui J.
      • Ginsburg D.
      ), while overexpression of PAI-1 in PC-3 prostate carcinoma cells inhibited both tumor growth and angiogenesis (
      • Soff G.A.
      • Sanderowitz J.
      • Gately S.
      • Verrusio E.
      • Weiss I.
      • Brem S.
      • Kwaan H.C.
      ). Thus, the exact role of PAI-1 in regulating tumor growth and angiogenesis has not been clearly defined. Therefore, we have examined whether PAI-1 can affect tumor growth through the regulation of angiogenesis, and we demonstrate that PAI-1 can both inhibit and enhance the growth of M21 human melanoma tumors, apparently by regulating angiogenesis.

      MATERIALS AND METHODS

      Proteins and Reagents

      The constitutively active PAI-1 (14-1b) and the enzyme-linked immunosorbent assay for functional human PAI-1 (HPAIKT) were from Molecular Innovations. This form of PAI-1 is identical to wild-type PAI-1 with respect to proteinase inhibition and binding to vitronectin, but has a half-life of 145 h, compared with 2 h for wild-type PAI-1 and, therefore, will maintain its biological activity during the extended period of these experiments (
      • Berkenpas M.B.
      • Lawrence D.A.
      • Ginsburg D.
      ). The partially inactivated (92% latent) wild-type PAI-1 was isolated as described previously (
      • Kvassman J.-O.
      • Shore J.D.
      ).

      In Vivo Tumor Growth Assay

      The M21 tumor growth assay was performed as described previously (
      • Petitclerc E.
      • Stromblad S.
      • von Schalscha T.L.
      • Mitjans F.
      • Piulats J.
      • Montgomery A.M.
      • Cheresh D.A.
      • Brooks P.C.
      ). Briefly, subconfluent cultures of human M21 melanoma cells were harvested and washed in serum-free RPMI, and 3 × 106 cells were injected subcutaneously into the flanks of 6-week-old nude mice. After 3 days, the mice were given daily intraperitoneal (IP) injections of 100 or 10 μg of the PAI-1 variants or PBS. Tumor growth was monitored every week for 3 weeks by an individual blind to the experimental conditions, and tumor volume was determined by measuring the length and the width of the growing tumor using a caliper. At the end of each experiment, the mice were euthanized and the primary tumor harvested free of adherent skin and tissues and processed for histochemistry. For each condition at least two independent experiments were performed.
      Angiogenesis was quantitated in tumor sections by counting the number of branch points in fields from sections stained with antibodies against von Willebrand factor. For each section, 10 random fields at × 400 magnification were counted, with four to five sections per treatment being counted in a blind manner.

      In Vivo Matrigel Angiogenesis Assay

      The Matrigel implantation assay was based on the method of Passaniti (
      • Passaniti A.
      • Taylor R.M.
      • Pili R.
      • Guo Y.
      • Long P.V.
      • Haney J.A.
      • Pauly R.R.
      • Grant D.S.
      • Martin G.R.
      ) with modifications. Matrigel is a squamous cell carcinoma basement membrane matrix composed primarily of collagen IV and laminin. To determine whether Matrigel itself contains the proteolytic factors required for matrix remodeling, an aliquot of Matrigel was analyzed in an immunoblot and SDS-polyacrylamide gel electrophoresis zymography (
      • Yepes M.
      • Sandkvist M.
      • Wong M.K.
      • Coleman T.A.
      • Smith E.
      • Cohan S.L.
      • Lawrence D.A.
      ). This analysis demonstrated an absence of plasminogen and uPA, and only small amounts of tissue-type plasminogen activator that only became apparent after prolonged incubation of the zymographic film (data not shown). Thus, any factors required for proteolytic remodeling of the matrix must therefore be provided by the host and not the Matrigel. Prior to injection fibroblast growth factor-2 (FGF-2) was incubated with 10 units/ml heparin for 5 min then diluted in Matrigel (Becton Dickinson, phenol red-free) on ice for a final concentration of 250 ng/ml FGF-2 and 0.0025 unit/ml heparin. PAI-1 samples or buffer alone were then added and the samples kept on ice until injection. Wild-type, PAI-1-deficient (originally obtained from Dr. Peter Carmeliet and back-crossed onto C57/B6J for 8 generations) or PAI-1 transgenic C57/B6J mice (
      • Eitzman D.T.
      • Krauss J.C.
      • Shen T.
      • Cui J.
      • Ginsburg D.
      ) between 6 and 12 weeks old were subcutaneously injected under anesthesia in the ventral midline region with 0.5 ml of Matrigel alone or with Matrigel containing FGF-2 ± PAI-1. Alternatively, mice were injected with Matrigel containing FGF-2 then treated with daily IP injections of PBS or PAI-1. After 5–7 days the mice were euthanized and the Matrigel implant harvested, washed twice with PBS, and examined macroscopically for signs of hematoma. Any implant with visible signs of hematoma was excluded from analysis. The remaining implants were then immediately frozen on dry ice and lyophilized overnight. The weight of the dry Matrigel was determined and the implants resuspended in 0.4 ml of 0.1% Triton X-100 for 1 h, disrupted by vigorous pipetting and centrifuged at 14,000 ×g for 15 min to remove particulates. The concentration of hemoglobin in the supernatant was then determined directly by absorbance at 405 nm (
      • Gigli I.
      • Ruddy S.
      • Austen K.F.
      ) and compared with a standard curve of purified hemoglobin (Sigma).

      Histochemistry

      For histochemical analysis of frozen tumor samples, tissues were embedded in OCT and snap-frozen in a 2-methylbutane dry ice slurry and stored at −80 °C until processing. Sections were cut at 5 μm and fixed for 10 min in methanol/acetone (1:1) and rinsed in PBS. For histochemical analysis of formalin-fixed Matrigel samples, tissues were embedded in paraffin and cut at 5 μm. The frozen or formalin-fixed samples were then analyzed either by hematoxylin-eosin staining or by immunohistochemistry as described previously (
      • Stefansson S.
      • Petitclerc E.
      • Wong M.K.
      • McMahon G.A.
      • Brooks P.C.
      • Lawrence D.A.
      ,
      • Yepes M.
      • Sandkvist M.
      • Wong M.K.
      • Coleman T.A.
      • Smith E.
      • Cohan S.L.
      • Lawrence D.A.
      ). Briefly, the tumors and implants were treated with 0.3% H2O2 for 30 min to exhaust endogenous peroxidase activity, then stained with rabbit anti-human von Willebrand factor (Dako) followed by detection with mouse-absorbed biotinylated goat anti-rabbit IgG1 and horseradish peroxidase-labeled streptavidin according to the manufacturer (Vector Laboratories) and developed with the chromogen 3,3-diaminobenzidine tetrahydrochloride (Sigma) for 4 min. Mayer's hematoxylin was used as a counterstain. Tumor sections were also stained with biotinylated murine anti-human collagen IV (Dako) or biotinylated murine anti-smooth muscle cell actin (Sigma) and detected with the ABC Elite kit as described by the manufacturer (Vector Laboratories). The sections were developed and counterstained as described above. For immunofluorescence staining of tumor sections, slides were treated as above and then double stained with rabbit anti-murine vitronectin (Molecular Innovations) and rat anti-murine PECAM-1 (PharMingen) and detected with Flour 488 goat anti-rabbit IgG (Molecular Probes) and Texas Red rat anti-mouse IgG (Molecular Probes). The nuclei of the tumor cells were stained with 4,6-diamidino-2-phenylindole dihydrochloride (Molecular Probes). Apoptosis of tumor sections was determined by TUNEL reactivity using the Apoptag Kit (Oncor, Gaithersburg, MD) as described previously (
      • Yepes M.
      • Sandkvist M.
      • Wong M.K.
      • Coleman T.A.
      • Smith E.
      • Cohan S.L.
      • Lawrence D.A.
      ). Microscopy and photography were carried out using a Nikon Eclipse E800 microscope attached to a Spot RT slider digital camera with a Kodak KAI-2092 cooled color chip.

      RESULTS AND DISCUSSION

      PAI-1 Inhibits M21 Tumor Growth and Angiogenesis

      Because the growth of solid tumors requires an adequate blood supply, agents that block angiogenesis are likely to inhibit tumor growth. Since we have shown previously that PAI-1 is a potent inhibitor of vascular cell migration in vitro (
      • Stefansson S.
      • Lawrence D.A.
      ) and angiogenesis in the chick CAMin vivo (
      • Stefansson S.
      • Petitclerc E.
      • Wong M.K.
      • McMahon G.A.
      • Brooks P.C.
      • Lawrence D.A.
      ), we decided to test whether PAI-1 treatment could inhibit tumor growth in mice. M21 human melanoma cells (3 × 106) were injected subcutaneously into the flanks of nude mice, and allowed to establish for 3 days, after which the mice were treated with daily IP injections of PAI-1 (100 μg) for 21 days. A sensitive enzyme-linked immunosorbent assay for active human PAI-1 demonstrated that 4 h after injection, plasma PAI-1 levels ranged between 32 and 122 nm, with a mean of 61 nm. This is ∼3000-fold higher than the reported normal murine plasma level of 0.02 nm (
      • Eitzman D.T.
      • Krauss J.C.
      • Shen T.
      • Cui J.
      • Ginsburg D.
      ). Treatment with active PAI-1 decreased M21 tumor growth by ∼66% (p = 0.01) (Fig.1). As a negative control, mice were also treated with partially inactivated PAI-1 (∼92% latent). However, this treatment unexpectedly increased tumor growth significantly (∼46% relative to PBS-treated animals (p = 0.05)). Quantitation of active recombinant PAI-1 in plasma 4 h after injection of mice with this partially inactivated PAI-1 showed a range of 0.03 to 0.11 nm, with a mean of 0.07 nm. This suggested that low levels of active PAI-1 may promote tumor growth.
      Figure thumbnail gr1
      Figure 1Quantitative analysis of M21 human melanoma tumor growth in vivo. M21 human melanoma cells (3 × 106) were injected subcutaneously into nude mice, allowed to establish for 3 days, and then treated daily with IP injections of PBS or with 100 μg of fully active PAI-1 or partially inactivated PAI-1 (92% latent). The data represent the mean of 10 mice per condition, and S.E. are shown. ○, PBS; ●, 92% latent PAI-1; ■, active PAI-1.
      To see if the effect of PAI-1 on tumor growth might be due to its regulation of angiogenesis or to a direct effect on the tumor cells, histological sections of the various tumors were examined. Hematoxylin-eosin staining showed that treatment of tumors with PAI-1 did not appear to alter the extent of necrosis within the tissue (Fig.2, A–C). Likewise, TUNEL staining for apoptotic cells indicated that the extent of apoptosis was very low in the healthy tissue and was mostly confined to areas of dead tissue and was not altered by the different treatments (Fig. 2,D–F). Finally, staining for the proliferative cell nuclear antigen, a marker for dividing cells, did not show any obvious differences between the three treatments (data not shown). However, it is possible that some differences in the rates of tumor cell proliferation might not be observed by this analysis. This is because the relative number of proliferating cell nuclear antigen-positive cells per field can only indicate the proportion of tumor cells that are in similar phases of the cell cycle, but cannot determine whether the tumor cells are progressing through the cell cycle at the same rate. Nonetheless, together with the results of the TUNEL staining, these data suggest that PAI-1 addition does not appear to directly affect M21 tumor cell viability, and thus implies that the differences in the tumor growth rates are likely to result from indirect effects of PAI-1 on tumor growth.
      Figure thumbnail gr2
      Figure 2Histological analysis of M21 tumors.Tumor sections treated with PBS (A, D,G), partially inactivated PAI-1 (B, E,H), or fully active PAI-1 (C, F,I) were stained for the following: hematoxylin-eosin (A—C), TUNEL reactivity (D–F), immunohistochemical staining for von Willebrand factor (G–I). The original magnification for all panels is × 400.
      To examine whether PAI-1 treatment influenced angiogenesis in the tumors, sections were stained with antibodies against von Willebrand factor (Fig. 2, G–I). Sections from PBS-treated animals demonstrated a multibranched pattern of vessels. In contrast, tumor vessels from animals treated with partially inactivated PAI-1 appeared more branched, whereas tumor vessels from animals treated with fully active PAI-1 appeared longer with a decreased number of branch points. Quantitation of branch points within the tumor sections showed that the fully active PAI-1-treated tumors had significantly fewer branch points per field relative to control tumors, while treatment with partially inactivated PAI-1 significantly increased branch point number (TableI). This mirrored the results obtained in tumor growth experiments and suggested that the differences in tumor volume observed with the PAI-1 treatments may be due to PAI-1 altering the extent of angiogenesis in the tumors.
      Table IQuantitation of branch points from von Willebrand factor-stained tumor sections
      TreatmentBranch points/fieldS.E.T test
      PBS28.091.60
      Active PAI-124.791.200.05
      Latent (92%) PAI-133.991.870.01
      To see if the alterations in angiogenesis were associated with structural differences in the tumor vessels, immunohistochemical examination of specific vessel markers was performed. These data are shown in Fig. 3 and demonstrate that apart from the extent of branching, the overall architecture of the vessels was similar in each case. For example, collagen IV was observed in the basement membrane of all the vessels, while as expected smooth muscle actin was only observed in the larger vessels. Immunofluorescence staining also localized vitronectin to the basement membrane of the vessels (Fig. 3, G–I). In this analysis vitronectin is shown in green, juxtaposed to the endothelial cells shown in red, and surrounding the tumor nodules, which are labeled with 4,6-diamidino-2-phenylindole dihydrochloride to show the nuclei. Together, these results suggest that both the formation of the capillary basement membrane and the recruitment of smooth muscle cells into larger vessels are similar in each case and that PAI-1 likely regulates tumor growth by controlling the extent of angiogenesis and not the structure of the vessels formed.
      Figure thumbnail gr3
      Figure 3Immuno-analysis of vessel architecture in M21 tumors. Tumor sections treated with PBS (A,D, G), partially inactivated PAI-1 (B,E, H), or fully active PAI-1 (C,F, I) were stained for the following: immunohistochemical staining for collagen IV (A–C), immunohistochemical staining for smooth muscle cell actin (D–F), immunofluorescence analysis of vitronectin (green) (G–I), endothelial cells (red), and tumor cells (blue). The original magnification for all panels is × 400.

      PAI-1 Regulates Angiogenesis in Murine Matrigel Implants

      Since we have shown previously that fully latent PAI-1 (<1% active) does not affect angiogenesis in the chick CAM (
      • Stefansson S.
      • Petitclerc E.
      • Wong M.K.
      • McMahon G.A.
      • Brooks P.C.
      • Lawrence D.A.
      ), and since two recent reports have suggested that PAI-1 is necessary for angiogenesis (
      • Bajou K.
      • Noel A.
      • Gerard R.D.
      • Masson V.
      • Brunner N.
      • Holst-Hansen C.
      • Skobe M.
      • Fusenig N.E.
      • Carmeliet P.
      • Collen D.
      • Foidart J.M.
      ,
      • Gutierrez L.S.
      • Schulman A.
      • Brito-Robinson T.
      • Noria F.
      • Ploplis V.A.
      • Castellino F.J.
      ), we suspected that the ∼8% active PAI-1 present in the partially inactivated PAI-1 may be responsible for the stimulation of tumor growth and angiogenesis observed in the mice. Therefore, to test the hypothesis that active PAI-1 can stimulate angiogenesis at low levels but is inhibitory at high concentrations, the Matrigel implant assay was used (
      • Passaniti A.
      • Taylor R.M.
      • Pili R.
      • Guo Y.
      • Long P.V.
      • Haney J.A.
      • Pauly R.R.
      • Grant D.S.
      • Martin G.R.
      ). Matrigel, either with or without FGF-2, was injected subcutaneously into the abdomen of mice and harvested after 5–7 days. This assay permits analysis of the extent of FGF-2-induced angiogenesis into the matrix by either histological examination or by direct measurement of the amount of hemoglobin present in the implant. Quantitative comparison of the number of vessel branch points in von Willebrand factor-stained sections with the hemoglobin content of similar implants gave nearly identical results, indicating that hemoglobin content within an implant accurately reflects the extent of angiogenesis (data not shown). Analysis of Matrigel implants that did not contain growth factor showed very few cells, either within or around the implant, indicating that Matrigel itself is not angiogenic (Fig.4 A). In contrast, sections containing 250 ng/ml FGF-2 (Fig. 4 B) showed many highly branched structures throughout the implant. Closer examination of these sections at high magnification clearly demonstrates that the structures are functioning vessels, since red blood cells can be seen within the vessel (Fig. 4 C).
      Figure thumbnail gr4
      Figure 4Histological analysis of Matrigel implants from wild-type and PAI-1-deficient mice. A, untreated;B, C, D, G, treated with FGF-2 only; F and I, treated with FGF-2 + 1 μm fully active PAI-1; E and H, PAI-1-deficient mice (PAI-1−/−) treated with FGF-2 only.A–F, hematoxylin-eosin staining; G–I, immunohistochemical staining for von Willebrand factor. Aand B are at × 20 original magnification, while C–I are at × 400 original magnification. wt, wild-type.
      To determine the effect of PAI-1 on angiogenesis in Matrigel implants, we examined FGF-2-treated implants from wild-type mice and compared these to implants from PAI-1−/− mice or from wild-type mice treated with 1 μm PAI-1. Staining of sections with hematoxylin-eosin or with anti-von Willebrand factor demonstrated many highly branched vessels migrating into the center of the FGF-2-treated implant of wild-type mice (Fig. 4, D and G). In contrast, while vessels were formed in the PAI-1-deficient mice, they were located primarily in the periphery of the implant and appeared less branched compared with the wild-type mice (Fig. 4, Eand H). Similarly, sections from implants treated with 1 μm PAI-1 showed that far fewer cells had infiltrated into the implant and that these were located almost exclusively on the perimeter (Fig. 4, F and I), with the interior essentially devoid of cells (data not shown). However, unlike the controls or the PAI-1−/− mice, the cells in the PAI-1-treated implants were not organized into vessels and did not appear to form lumens (Fig. 4, F and I). Thus, PAI-1 seems to be important for regulating both the invasion and formation of the vessels. In the absence of PAI-1, endothelial cells can organize into vessels, but their ability to invade the implant appears to be reduced, whereas when excess PAI-1 is present both vessel invasion and formation is impaired.
      Quantitative analysis of angiogenesis in wild-type and the PAI-1-deficient mice indicated that angiogenesis in the PAI-1-deficient mice was decreased by 63% relative to wild-type mice (Fig.5 A). These results are also consistent with previous reports that angiogenesis is impaired in PAI-1−/− mice (
      • Bajou K.
      • Noel A.
      • Gerard R.D.
      • Masson V.
      • Brunner N.
      • Holst-Hansen C.
      • Skobe M.
      • Fusenig N.E.
      • Carmeliet P.
      • Collen D.
      • Foidart J.M.
      ,
      • Gutierrez L.S.
      • Schulman A.
      • Brito-Robinson T.
      • Noria F.
      • Ploplis V.A.
      • Castellino F.J.
      ) and suggest that although angiogenesis may proceed in the absence of PAI-1, it appears to be less efficient. In contrast, quantitation of angiogenesis in transgenic mice overexpressing PAI-1 showed a nearly 3-fold increase in angiogenesis compared with controls (Fig. 5 A). The average plasma level of active PAI-1 reported for these mice is 59 ng/ml (∼1.2 nm), which is ∼60-fold higher than the level in wild-type littermate controls, which was 0.02 nm (
      • Eitzman D.T.
      • Krauss J.C.
      • Shen T.
      • Cui J.
      • Ginsburg D.
      ). Taken together with the results of the PAI-1−/− mice, these data suggest that levels of PAI-1 slightly above normal may stimulate angiogenesis.
      Figure thumbnail gr5
      Figure 5Quantitative analysis of angiogenesis in Matrigel implants. A, hemoglobin content from implants containing FGF-2 in wild-type, PAI-1-deficient (*, p < 0.05), or PAI-1 transgenic mice (*, p < 0.05).B, hemoglobin content in implants containing FGF-2 with increasing concentrations of fully active PAI-1. In both panelsn ≥ 4 for each condition, and the mean ± S.E. are shown.
      To directly examine the effect of different PAI-1 concentrations on angiogenesis, C57/B6J mice were injected with Matrigel containing increasing amounts of PAI-1, and the hemoglobin content of the implants was determined after 5 days (Fig. 5 B). While low doses of PAI-1 (0.1 and 1.0 nm) increased angiogenesis in the implants by up to 3-fold relative to FGF-2 alone, PAI-1 concentrations of 10 nm or higher were able to inhibit angiogenesis by nearly 100%. This confirms our hypothesis that PAI-1 at low doses can promote angiogenesis, whereas at higher concentrations it is inhibitory, and suggests that the reason that the partially inactivated PAI-1 stimulated tumor growth and angiogenesis was likely due to the 8% active PAI-1 present in the preparation. However, it is also possible that the systemic administration of PAI-1 yields different results from local delivery or that latent PAI-1 may have an as yet unidentified proangiogenic activity. This latter possibility could be related to earlier observations suggesting that two other inactive serpins, latent or cleaved antithrombin III (
      • O'Reilly M.S.
      • Pirie-Shepherd S.
      • Lane W.S.
      • Folkman J.
      ) and pigment epithelium-derived factor (
      • Dawson D.W.
      • Volpert O.V.
      • Gillis P.
      • Crawford S.E.
      • Xu H.
      • Benedict W.
      • Bouck N.P.
      ), have anti-angiogenic activities. To test these possibilities, FGF-2-containing Matrigel implants were injected into wild-type mice, and the mice were given daily IP injections of PBS or either 100 μg of the same partially inactivated PAI-1 preparation used in the M21 tumor experiments containing ∼8% active PAI-1 or this same preparation after it had been incubated at 37 °C to convert the majority of the active protein to the latent form (<0.5% active PAI-1). After 7 days of treatment, the Matrigel implants were harvested and the hemoglobin content measured as before (Fig. 6). Similar to the M21 tumors, the partially inactivated PAI-1 stimulated angiogenesis by ∼1.5-fold relative to PBS-treated mice (Fig. 6). However, mice treated with fully latent PAI-1 (<0.5% active protein) showed no increase in angiogenesis compared with control mice. These data provide further evidence that low doses of active PAI-1 can promote angiogenesis and strongly suggest that the stimulation of M21 tumor growth and angiogenesis observed with partially inactivated PAI-1 treatment was due to the ∼8% active PAI-1 present in that preparation and not to any proangiogenic activity in the latent PAI-1. These results also suggest that treating M21 tumors with low doses of active recombinant PAI-1 would increase tumor growth by increasing angiogenesis. To see if this would be the case, M21 tumors in nude mice were treated with daily IP injections of 10 μg of PAI-1 and the size of the tumors measured over a 3-week period as before. This dose of PAI-1 corresponds approximately to the level of active PAI-1 injected with the partially inactivated PAI-1 preparation (10 versus 8 μg). The results of these experiments demonstrated that treatment of mice with this low dose of PAI-1 increased M21 tumor growth by 48% relative to the PBS control (data not shown), a result that exactly mirrors the increase in the growth of tumors treated with partially inactivated PAI-1 (Fig. 1). Thus, treatment of tumors with low doses of PAI-1 can stimulate tumor growth most likely by stimulating angiogenesis.
      Figure thumbnail gr6
      Figure 6Quantitative analysis of angiogenesis in implants treated with inactive PAI-1. Hemoglobin content in wild-type mice treated systemically for 7 days with IP injections of either PBS, partially inactivated PAI-1 (92% latent) 100 μg/day, or fully latent PAI-1 (>99.5% latent) 100 μg/day (*, p< 0.05 relative to fully latent PAI-1) is shown. For each condition,n ≥ 6, and the mean ± S.E. are shown.
      The observation that PAI-1 can either enhance or inhibit tumor growth and angiogenesis depending upon its concentration may also explain why some studies have shown that PAI-1 is necessary for tumor growth (
      • Bajou K.
      • Noel A.
      • Gerard R.D.
      • Masson V.
      • Brunner N.
      • Holst-Hansen C.
      • Skobe M.
      • Fusenig N.E.
      • Carmeliet P.
      • Collen D.
      • Foidart J.M.
      ,
      • Gutierrez L.S.
      • Schulman A.
      • Brito-Robinson T.
      • Noria F.
      • Ploplis V.A.
      • Castellino F.J.
      ,
      • Bajou K.
      • Masson V.
      • Gerard R.D.
      • Schmitt P.M.
      • Albert V.
      • Praus M.
      • Lund L.R.
      • Frandsen T.L.
      • Brunner N.
      • Dano K.
      • Fusenig N.E.
      • Weidle U.
      • Carmeliet G.
      • Loskutoff D.
      • Collen D.
      • Carmeliet P.
      • Foidart J.M.
      • Noel A.
      ), while others indicate that PAI-1 either has no affect (
      • Eitzman D.T.
      • Krauss J.C.
      • Shen T.
      • Cui J.
      • Ginsburg D.
      ) or is inhibitory (
      • Soff G.A.
      • Sanderowitz J.
      • Gately S.
      • Verrusio E.
      • Weiss I.
      • Brem S.
      • Kwaan H.C.
      ). These conflicting results may be a consequence of the novel regulatory role that PAI-1 plays in angiogenesis, and while the mechanism of these seemingly opposing activities is not clear, our previous studies of the effect of PAI-1 on angiogenesis in the chick CAM may provide some insight. In these studies, we demonstrated that high doses of recombinant PAI-1 inhibited angiogenesis in the CAM via two distinct but apparently overlapping pathways. The first was dependent upon PAI-1 inhibition of proteinase activity, while the second was independent of PAI-1's anti-proteinase activity and instead functioned through PAI-1 binding to vitronectin, which in turn blocked access of vascular cell integrins to vitronectin in the matrix (
      • Stefansson S.
      • Lawrence D.A.
      ,
      • Stefansson S.
      • Petitclerc E.
      • Wong M.K.
      • McMahon G.A.
      • Brooks P.C.
      • Lawrence D.A.
      ). A number of studies from other groups have also suggested that both uPA and its receptor, uPAR, can modulate cellular adhesive properties and motility as well as promote angiogenesis (
      • Wei Y.
      • Lukashev M.
      • Simon D.I.
      • Bodary S.C.
      • Rosenberg S.
      • Doyle M.V.
      • Chapman H.A.
      ,
      • Yebra M.
      • Parry G.C.N.
      • Stromblad S.
      • Mackman N.
      • Rosenberg S.
      • Mueller B.M.
      • Cheresh D.A.
      ,
      • Fibbi G.
      • Caldini R.
      • Chevanne M.
      • Pucci M.
      • Schiavone N.
      • Morbidelli L.
      • Parenti A.
      • Granger H.J.
      • Del Rosso M.
      • Ziche M.
      ,
      • Webb D.J.
      • Nguyen D.H.
      • Gonias S.L.
      ,
      • Webb D.J.
      • Nguyen D.H.
      • Sankovic M.
      • Gonias S.L.
      ,
      • Ossowski L.
      • Aguirre-Ghiso J.A.
      ,
      • Preissner K.T.
      • Kanse S.M.
      • May A.E.
      ). Based on these studies and the data presented here we suggest a speculative model for the regulation of angiogenesis by PAI-1 where PAI-1 bound to vitronectin in the provisional angiogenic matrix temporarily blocks αvβ3- and/or αvβ5-mediated adhesion, which in turn prevents the migration necessary for efficient vessel formation. However, since the activated vascular cells express uPA and uPAR, then the cells should be able to remove PAI-1 from the matrix and expose the cryptic integrin adhesion sites on vitronectin. Finally, since uPAR expression is localized to the leading edge of the migrating cells (
      • Grondahl-Hansen J.
      • Lund L.R.
      • Ralfkiær E.
      • Ottevanger V.
      • Dano K.
      ), this permits cells expressing uPA and uPAR to preferentially form new adhesion complexes at the apex of the moving angiogenic front, thus imparting directionality to vascular cell migration.
      On a molecular level, this suggests that the association of integrins with the uPA·uPAR complex (
      • Wei Y.
      • Lukashev M.
      • Simon D.I.
      • Bodary S.C.
      • Rosenberg S.
      • Doyle M.V.
      • Chapman H.A.
      ,
      • Yebra M.
      • Parry G.C.N.
      • Stromblad S.
      • Mackman N.
      • Rosenberg S.
      • Mueller B.M.
      • Cheresh D.A.
      ,
      • Ossowski L.
      • Aguirre-Ghiso J.A.
      ,
      • Tarui T.
      • Mazar A.P.
      • Cines D.B.
      • Takada Y.
      ) may serve to correctly position this complex in close proximity to the cryptic vitronectin RGD adhesion site that becomes accessible upon binding of uPA to PAI-1 and subsequent dissociation of the PAI-1·uPA complex from vitronectin (
      • Stefansson S.
      • Muhammad S.
      • Cheng X.F.
      • Battey F.D.
      • Strickland D.K.
      • Lawrence D.A.
      ). Thus, uPA/uPAR may serve to escort the integrin to the appropriate high affinity binding site within the matrix. This may also explain the apparent contradiction between the observation that high doses of PAI-1 can block tumor growth and angiogenesis (Figs.1 and 5), whereas PAI-1-deficient mice appear to show a reduced angiogenic response (Fig. 5) (
      • Bajou K.
      • Noel A.
      • Gerard R.D.
      • Masson V.
      • Brunner N.
      • Holst-Hansen C.
      • Skobe M.
      • Fusenig N.E.
      • Carmeliet P.
      • Collen D.
      • Foidart J.M.
      ,
      • Gutierrez L.S.
      • Schulman A.
      • Brito-Robinson T.
      • Noria F.
      • Ploplis V.A.
      • Castellino F.J.
      ,
      • Bajou K.
      • Masson V.
      • Gerard R.D.
      • Schmitt P.M.
      • Albert V.
      • Praus M.
      • Lund L.R.
      • Frandsen T.L.
      • Brunner N.
      • Dano K.
      • Fusenig N.E.
      • Weidle U.
      • Carmeliet G.
      • Loskutoff D.
      • Collen D.
      • Carmeliet P.
      • Foidart J.M.
      • Noel A.
      ). We predict that angiogenesis will only proceed efficiently when PAI-1 levels are within an optimal range. In PAI-1 excess, the vitronectin matrix remains saturated with PAI-1, and integrin adhesion is blocked. Conversely, in PAI-1 deficiency, the uPA/uPAR-facilitated polar localization of integrins may not occur.
      Finally, our results suggest that PAI-1 is likely to play an important role in regulating tumor growth and angiogenesis, and future studies further characterizing the mechanism of this activity will very likely lead to the development of important new therapeutic agents for controlling angiogenesis in different pathologic situations.

      REFERENCES

        • Folkman J.
        Nat. Med. 1995; 1: 27-31
        • Carmeliet P.
        • Jain R.K.
        Nature. 2000; 407: 249-257
        • Carmeliet P.
        • Schoonjans L.
        • Kieckens L.
        • Ream B.
        • Degen J.
        • Bronson R.
        • De Vos R.
        • van den Oord J.J.
        • Collen D.
        • Mulligan R.C.
        Nature. 1994; 368: 419-424
        • Johnsen M.
        • Lund L.R.
        • Romer J.
        • Almholt K.
        • Dano K.
        Curr. Opin. Cell Biol. 1998; 10: 667-671
        • Fay W.P.
        • Shapiro A.D.
        • Shih J.L.
        • Schleef R.R.
        • Ginsburg D.
        N. Engl. J. Med. 1992; 327: 1729-1733
        • Carmeliet P.
        • Kieckens L.
        • Schoonjans L.
        • Ream B.
        • Van Nuffelen A.
        • Prendergast G.
        • Cole M.
        • Bronson R.
        • Collen D.
        • Mulligan R.C.
        J. Clin. Invest. 1993; 92: 2746-2755
        • Lawrence D.A.
        • Palaniappan S.
        • Stefansson S.
        • Olson S.T.
        • Francis-Chmura A.M.
        • Shore J.D.
        • Ginsburg D.
        J. Biol. Chem. 1997; 272: 7676-7680
        • Deng G.
        • Royle G.
        • Wang S.
        • Crain K.
        • Loskutoff D.J.
        J. Biol. Chem. 1996; 271: 12716-12723
        • Stefansson S.
        • Lawrence D.A.
        Nature. 1996; 383: 441-443
        • Stefansson S.
        • Petitclerc E.
        • Wong M.K.
        • McMahon G.A.
        • Brooks P.C.
        • Lawrence D.A.
        J. Biol. Chem. 2001; 276: 8135-8141
        • Kjoller L.
        • Kanse S.M.
        • Kirkegaard T.
        • Rodenburg K.W.
        • Ronne E.
        • Goodman S.L.
        • Preissner K.T.
        • Ossowski L.
        • Andreasen P.A.
        Exp. Cell Res. 1997; 232: 420-429
        • Nekarda H.
        • Siewert J.R.
        • Schmitt M.
        • Ulm K.
        Lancet. 1994; 343: 117
        • Pedersen H.
        • Brunner N.
        • Francis D.
        • Osterlind K.
        • Ronne E.
        • Hansen H.H.
        • Dano K.
        • Grondahl-Hansen J.
        Cancer Res. 1994; 54: 4671-4675
        • Bajou K.
        • Noel A.
        • Gerard R.D.
        • Masson V.
        • Brunner N.
        • Holst-Hansen C.
        • Skobe M.
        • Fusenig N.E.
        • Carmeliet P.
        • Collen D.
        • Foidart J.M.
        Nat. Med. 1998; 4: 923-928
        • Gutierrez L.S.
        • Schulman A.
        • Brito-Robinson T.
        • Noria F.
        • Ploplis V.A.
        • Castellino F.J.
        Cancer Res. 2000; 60: 5839-5847
        • Eitzman D.T.
        • Krauss J.C.
        • Shen T.
        • Cui J.
        • Ginsburg D.
        Blood. 1996; 87: 4718-4722
        • Soff G.A.
        • Sanderowitz J.
        • Gately S.
        • Verrusio E.
        • Weiss I.
        • Brem S.
        • Kwaan H.C.
        J. Clin. Invest. 1995; 96: 2593-2600
        • Berkenpas M.B.
        • Lawrence D.A.
        • Ginsburg D.
        EMBO J. 1995; 14: 2969-2977
        • Kvassman J.-O.
        • Shore J.D.
        Fibrinolysis. 1995; 9: 215-221
        • Petitclerc E.
        • Stromblad S.
        • von Schalscha T.L.
        • Mitjans F.
        • Piulats J.
        • Montgomery A.M.
        • Cheresh D.A.
        • Brooks P.C.
        Cancer Res. 1999; 59: 2724-2730
        • Passaniti A.
        • Taylor R.M.
        • Pili R.
        • Guo Y.
        • Long P.V.
        • Haney J.A.
        • Pauly R.R.
        • Grant D.S.
        • Martin G.R.
        Lab. Invest. 1992; 67: 519-528
        • Yepes M.
        • Sandkvist M.
        • Wong M.K.
        • Coleman T.A.
        • Smith E.
        • Cohan S.L.
        • Lawrence D.A.
        Blood. 2000; 96: 569-576
        • Gigli I.
        • Ruddy S.
        • Austen K.F.
        J. Immunol. 1968; 100: 1154-1164
        • O'Reilly M.S.
        • Pirie-Shepherd S.
        • Lane W.S.
        • Folkman J.
        Science. 1999; 285: 1926-1928
        • Dawson D.W.
        • Volpert O.V.
        • Gillis P.
        • Crawford S.E.
        • Xu H.
        • Benedict W.
        • Bouck N.P.
        Science. 1999; 285: 245-248
        • Bajou K.
        • Masson V.
        • Gerard R.D.
        • Schmitt P.M.
        • Albert V.
        • Praus M.
        • Lund L.R.
        • Frandsen T.L.
        • Brunner N.
        • Dano K.
        • Fusenig N.E.
        • Weidle U.
        • Carmeliet G.
        • Loskutoff D.
        • Collen D.
        • Carmeliet P.
        • Foidart J.M.
        • Noel A.
        J. Cell Biol. 2001; 152: 777-784
        • Wei Y.
        • Lukashev M.
        • Simon D.I.
        • Bodary S.C.
        • Rosenberg S.
        • Doyle M.V.
        • Chapman H.A.
        Science. 1996; 273: 1551-1555
        • Yebra M.
        • Parry G.C.N.
        • Stromblad S.
        • Mackman N.
        • Rosenberg S.
        • Mueller B.M.
        • Cheresh D.A.
        J. Biol. Chem. 1996; 271: 29393-29399
        • Fibbi G.
        • Caldini R.
        • Chevanne M.
        • Pucci M.
        • Schiavone N.
        • Morbidelli L.
        • Parenti A.
        • Granger H.J.
        • Del Rosso M.
        • Ziche M.
        Lab. Invest. 1998; 78: 1109-1119
        • Webb D.J.
        • Nguyen D.H.
        • Gonias S.L.
        J. Cell Sci. 2000; 113: 123-134
        • Webb D.J.
        • Nguyen D.H.
        • Sankovic M.
        • Gonias S.L.
        J. Biol. Chem. 1999; 274: 7412-7420
        • Ossowski L.
        • Aguirre-Ghiso J.A.
        Curr. Opin. Cell Biol. 2000; 12: 613-620
        • Preissner K.T.
        • Kanse S.M.
        • May A.E.
        Curr. Opin. Cell Biol. 2000; 12: 621-628
        • Grondahl-Hansen J.
        • Lund L.R.
        • Ralfkiær E.
        • Ottevanger V.
        • Dano K.
        J. Invest. Dermatol. 1988; 90: 790-795
        • Tarui T.
        • Mazar A.P.
        • Cines D.B.
        • Takada Y.
        J. Biol. Chem. 2001; 276: 3983-3990
        • Stefansson S.
        • Muhammad S.
        • Cheng X.F.
        • Battey F.D.
        • Strickland D.K.
        • Lawrence D.A.
        J. Biol. Chem. 1998; 273: 6358-6366