Endothelial plasminogen activator inhibitor; Mr1; Nexin, plasminogen activator inhibitor type 1; PAI; PAI1; PAI-1; Planh; PLANH1; Planh1; Plasminogen activator inhibitor 1; Plasminogen activator inhibitor type-1 (PAI-1); Plasminogen activator inhibitor-1; Serine (or cysteine) peptidase inhibitor, clade E, member 1
The temporal and spatial control of a complex extracellular proteolytic cascade regulates the fundamental processes of tissue remodeling, inflammation, thrombosis, and wound repair. Growth factor and cytokine-induced release of two plasminogen activators (PAs; urokinase-type PA, uPA; tissue-type PA, tPA) converts plasminogen to plasmin, a trypsin-like protease that in turn activates several matrix metalloproteinases to implement extracellular matrix (ECM) restructuring (Van den Steen et al. 2001). The broad substrate specificity of plasmin requires strict control of its activity which is mainly accomplished by several plasminogen activator (PA) inhibitors (PAIs). PAIs belong to the superfamily of SERPINs (serine protease inhibitors) since they target serine proteinases including uPA and tPA (Law et al. 2006). Early discovered SERPINs were originally identified as inhibitors of serine proteinases and mainly involved in blood coagulation, fibrinolysis, and complement activation (i.e., antithrombin III, α1-antichymotrypsin, C1-inhibitor, α2-antiplasmin) (Law et al. 2006).
The development of biochemical assays that could monitor specific PAIs individually as well as advances in the purification (Schleef et al. 1988) and cloning of plasminogen activator inhibitor type-1 (PAI-1; SERPINE1; NCBI_NM000602.4) led to the discovery of four additional PAIs: PAI-2 (placental-type PAI; SERPINB2; NCBI_NM002575.2), PAI-3 (Protein C inhibitor; SERPINA5; NCBI_NM000624.5), protease nexin-1 (PN-1; SERPINE2; NCBI_NM006216.3), and neuroserpin (SERPINI1; NCBI_NM005025.4). The predominant physiological role for PAI-1 in vivo is the negative regulation of fibrinolysis. Tissue injury rapidly activates the coagulation cascade resulting in the formation of a fibrin clot to minimize blood loss from injured vessels. As the vessel wall undergoes repair, endothelial cells at the site of injury secrete tPA, stimulating the conversion of plasminogen to plasmin to initiate degradation of the cross-linked fibrin in the thrombus. While fibrin clearance, controlled by PAI-1 released from platelets embedded in the clot, was originally the main function assigned to PAI-1, research over the last four decades led to the discovery of a multifaceted palette of PAI-1 activities, including fine regulation of pericellular proteolysis, ECM remodeling, and cell motility, that collectively influence tissue repair and tumor metastasis.
PAI-1 is a soluble plasma protein synthesized by various cell types including splenocytes, adipocytes, hepatocytes, platelets, megakaryocytes, macrophages, smooth muscle cells, and endothelial cells (Dellas and Loskutoff 2005). In vivo murine studies confirmed high PAI-1 transcript abundance in heart, lung, aorta, muscle and, in particular, in adipose tissue. PAI-1 gene expression is upregulated in the liver in response to endotoxin exposure and in the adipose tissue of obese individuals. While platelets contain significant concentrations of PAI-1, this pool appears to be released only upon platelet activation during thrombosis where it likely generates locally a PA-inhibitor concentration sufficient to protect the newly formed thrombi from premature lysis by circulating PAs. The widespread tissue distribution of PAI-1 as well as in cultured endothelial cells in vitro suggested the endothelium as a prominent source of PAI-1 production. Immunohistochemistry and in situ hybridization, however, failed to confirm these findings instead highlighting vascular as well as nonvascular smooth muscle cells from multiple tissues as the alternative origin of tissue PAI-1 (Fearns et al. 1996).
PAI-1 deficiency in humans results in a rare heritable autosomal recessive bleeding disorder (Fay et al. 1997) that is managed clinically through oral antifibrinolytic agents such as tranexamic acid, an analog of lysine that occupies 4–5 lysine receptor sites on plasminogen or plasmin, thus, blocking the subsequent binding of plasmin to fibrin. Development of a PAI-1 gene-deficient murine model (Carmeliet et al. 2016) has facilitated research on the pathology of PAI-1 deficiency in particular and PAI-1 physiology in general.
PAI-1 is a member of the “immediate-early” or “early response” gene set and transcribed as two distinct mRNAs, 2.3 kb (half-life 2.5 h) and 3.2 kb (half-life 0.85 h). These transcripts differ only in their 3′-UTR, in which the longer transcript contains AU-rich elements (AREs) that regulate function mRNA stability. In vitro studies using cultured cell lines confirmed that PAI-1 expression can be influenced by a diverse group of growth factors, cytokines, and hormones (i.e., TNFα, TGFβ, lipopolysaccharide, interleukin-1, VLDL, LDL, thrombin, EGF, PDGF, bFGF, dexamethasone, insulin, estradiol, phorbol-myristate acetate (PMA) (Dimova and Kietzmann 2008) in which TGFβ1 is the most well-studied inducer of PAI-1 in vivo. TGFβ1 binding to cell surface receptors stimulates the phosphorylation and subsequent nuclear translocation of receptor-associated Smads that, upon complexing with various coactivators, initiate gene transcription through interaction with Smad-binding elements (SBEs) in the PAI-1 promotor (see Fig. 1).
The PAI-1 mRNA transcript also encodes a 21–23 aa signal peptide that is cleaved during protein maturation. The secreted, active PAI-1 protein is a single-chain cysteine-free glycoprotein (potential N-glycosylation at Asn209, Asn265, Asn329) consisting of 379–381 aa residues with an approximate molecular weight of 48 kDa. Three beta-sheets and nine alpha-helices, with a carboxyterminal reactive center loop (RCL; P6’-P16) containing the target domain for PAs (P1-P1’), comprise the global structure of PAI-1 (Law et al. 2006). A unique characteristic that distinguishes PAI-1 from other SERPINs is its ability to switch spontaneously from an active protein to a stable latent form that cannot bind to PAs. Since the active configuration represents the least thermodynamic favorable conformation of PAI-1, its biological half-life is relatively short (1–2 h, 37 °C, pH 7.4). For a more detailed review of SERPIN structures see review by Law et al. (Law et al. 2006). Extended stability of the active form of PAI-1 in vivo and for research purposes in vitro can be achieved in two ways: (a) In Vivo Stabilization: Upon secretion into the circulation, PAI-1 interacts with the N-terminal Somatomedin B domain of the plasma glycoprotein vitronectin (VN). VN is present in relatively high concentration in plasma (200–300 μg/ml) and is a major component of the extracellular matrix. Binding to VN stabilizes PAI-1 in its active conformation approximately twofold to threefold without affecting the ability of VN-bound PAI-1 to interact with, and inhibit, its target PAs. Mechanistic details and consequences of this interaction are discussed in more detail below. (b) In Vitro Stabilization: Using random mutagenesis approaches, several single and multiple residue mutations in human PAI-1 resulted in generation of PAI-1 variants with a >72-fold half-life (approximately 145 hours) at 37 °C (N150H, K154T, Q319L, M354I; 14-1B) compared to the wild-type protein (Berkenpas et al. 1995). This stabilized variant of PAI-1 represent a valuable alternative to the rapidly deactivating native PAI-1 for laboratory studies of PAI-1 function in vivo and in vitro.
Highlighting the translational utility of such investigations is the growing appreciation that pathological levels of PAI-1 impact fundamental, PAI-1-dependent processes including pericellular proteolysis, ECM remodeling, and cell motility. Persistently upregulated PAI-1 expression, in fact, is a major contributing factor to several chronic diseases such as obesity, dystrophic disorders, and progressive fibrosis involving the renal, pulmonary, and hepatic systems (Ghosh and Vaughan 2012). In several human cancers, overexpression of PAI-1 is a prognostic biomarker correlated to poor disease outcome. In human breast malignancies with an axillary node-negative diagnosis, PAI-1 (and uPA) constitutes a clinically important level-of-evidence-1 (LOE-1) biomarker pair critical to decisions regarding the need for adjuvant therapy (Duffy 2002).
PAI-1: Physiological Functions
Role of PAI-1 in Fibrinolysis
Regulation of Extracellular Proteolysis by PAI-1
Initially, PAI-1 is secreted into the circulation in an active “inhibitory” conformation. Without stabilization of this active configuration in vivo through interaction with VN, PAI-1 will spontaneously convert into a latent form and rapidly be cleared from the circulation. During this transition, the N-terminus of the RCL inserts into beta-sheet A, leading to a dislocation of the P1–P1’ bond; in this configuration, the RCL is inaccessible to the target PAs. In the active state, PAI-1 presents as a “pseudo substrate” to tPA or uPA. PAI-1 and PAs initially form a 1:1 stoichiometric noncovalent docking complex subsequently forming a covalent ester bond between the SERPIN and serine proteinase after cleavage of the “bait” P1–P1′ bond (Arg346-Met347) in the RCL sequence (Law et al. 2006). The N-terminal arm of the cleaved RCL is inserted into beta-sheet A and the C-terminal segment, carrying the proteinase, is repositioned to the opposite side of the inhibitor. In this process, the proteinase acquires an inactive conformation, and the SERPIN rendered antiproteolytically inactive, thus giving rise to the often used designation for PAI-1 as a “suicide inhibitor” (Wind et al. 2002).
Regulation of Cell Adhesion by PAI-1
Cell adhesion is mediated through interactions between various ECM proteins and several cell surface molecules (i.e., integrins, uPAR, GAGs). The regulatory function of PAI-1 on cell adhesion is mediated mainly through its interaction with receptor-bound uPA (uPA:uPAR). In this scenario, binding of pro-uPA to uPAR, (a) allows for the most efficient activation of uPA by plasmin and (b) changes the conformation of uPAR facilitating interaction between the uPA:uPAR complex and several ECM-engaged integrins (containing β1, β2, and β3 subdomains) (Czekay and Loskutoff 2009). Inhibition of uPA in uPA:uPAR:integrin complexes by PAI-1 results in: (a) covalent interaction between uPA and PAI-1; (b) conformational modifications in uPAR and associated integrins leading to decrease in their affinity for ECM binding sites; (c) accessibility of a previously hidden epitope on PAI-1 that promotes binding to the endocytic receptors, LRP-1 and VLDL-R (Stefansson et al. 1998). Upon membrane clearance, uPA:PAI-1 complexes undergo lysosomal degradation, while integrins, uPAR, and endocytic receptors are recycled back to the cell surface.
This uAP:uPAR-dependent process regulates integrin activity and the subsequent detachment of ECM-bound integrins from various matrices by PAI-1 (Czekay et al. 2011). Conformational changes in uPAR and integrins would allow for a sufficient decrease in binding affinities for their respective ECM substrates to allow a reduction in cell attachment force, even without subsequent endocytic clearance by LRP-1. However, if cointernalization of uPAR and integrins should occur via LRP-1 or VLDL-R, it would not only support cellular motility by recycling cell adhesion receptors (integrins) but also uPAR. The novelty in this mechanism is that it provides a means to co-redistribute a cellular adhesion component and an ECM degradation component to support stromal invasion.
Regulation of Cell Motility by PAI-1
Controlled cycles of cell adhesion, detachment, and reattachment are hallmark features of the migratory phenotype. Directed cell movement is required for such fundamental events as embryonic morphogenesis, injury-related tissue remodeling, inflammatory cell homing, and cancer metastasis. Although multiple microenvironmental stimuli regulate directed motility in vivo, occupation of the PAI-1 binding site in the SMB domain of VN prevents the reattachment of surface-tethered uPAR as well as integrins to VN, successfully interfering thereby with cellular reattachment to this ECM protein. Inhibition of reattachment to the VN matrix could force a switch to more stable adhesion sites by engaging integrin binding to other ECM components, thus slowing cell motility (i.e., during wound healing progression), or “redirect” migration away from a VN-rich environment (i.e., during tumor cell dissemination from a highly vascularized and VN-rich tumor microenvironment).
In a sequence of elegant experiments, Providence et al. (2008) confirmed the actual fate of pericellular PAI-1 during cell locomotion in the presence of VN (Providence et al. 2008). Using a quantifiable scratch wound healing assay in vitro, PAI-1 deposition into human keratinocyte migration trails was imaged using a PAI-1/GFP chimeric protein as a “reporter”. PAI-1-directed cell migration into the “wound” site was reduced upon addition of a function-neutralizing antibody to PAI-1. Inhibiting the formation of PAI-1:uPA:uPAR:integrin complexes or preventing the complex binding to LRP-1 similarly attenuated cell motility (Providence et al. 2008). Since uPAR and integrins are inactivated in this process, it appears that LRP-1 is the potential signaling receptor involved in PAI-1 dependent cell motility.
PAI-1 as “Signaling”-Ligand
Endocytic clearance of PAI-1:uPA:uPAR:integrins quaternary complexes renders the adhesion receptors uPAR and integrins inactive. However, binding of this complex to LRP-1 affects Rho-GTPase-dependent motility and cell proliferation, suggesting activation of intracellular signaling events through and by this receptor. LRP-1 is a two-chain, approximately 600 kDa, member of the LDL-receptor family involved in endocytic clearance of at least 30 structurally diverse ligands, including uPA:PAI-1, tPA:PAI-1, PAI-1, thrombin:PAI-1, uPA:PN1, thrombin:PN1, and other SERPIN:proteinase complexes (Strickland et al. 2002). There are four ligand-binding domains in the extracellular α-chain of LRP-1. The cytoplasmic tail of the transmembrane β-chain has two NPxY motifs and numerous tyrosine residues, suggesting a role in signaling through an ability to interact with multiple adaptor and scaffolding proteins (for detailed review of structure and function of LDL receptor family members (Strickland et al. 2002). PAI-1 signaling through LRP-1 seems to be PAI-1 function-independent since active, latent, and catalytically inactive cleaved PAI-1 all bind LRP-1 and initiate cellular migration into 3D collagen matrices. PAI-1/LRP-1 complexing activates the Jak (Janus kinase 2)/Stat (Signal transducer and activator of transcription protein) type-1 signaling pathway in rat smooth muscle cells and endothelial cells (Degryse et al. 2004). Although the exact mechanism is unknown, PAI-1:LRP-1 signaling is accompanied by nuclear translocation of active Stat1 and a rapid but transient reorganization in the actin cytoskeleton resulting in microfilament polarization. Blocking PAI-1/LRP-1 interaction or using LRP-1 binding-deficient mutants of PAI-1 inhibited cytoskeletal reorganization, Jak/Stat1 signaling, and the motile response. PAI-1 variants unable to complex with uPA or VN had no such effects.
Active PAI-1 is cleared in vivo from the pericellular environment through LRP-1 and VLDL-R in a complex with uPA:uPAR and integrins, whereas latent PAI-1 and cleaved PAI-1 will remain embedded in the ECM. Whether or not these extracellular reservoirs of “inactive” PAI-1 can be released by invading tumor, inflammatory, or wound-activated cells through progressive local ECM degradation and then impact local cell motility through receptor signaling remains to be determined.
In this context, SERPINB2 (PAI-2) PA-binding leads also to clearance of PAI-2:PA by LRP-1 or VLDL-R but without inducing signaling events similar to those initiated by PAI-1:receptor complexes. Unlike PAI-1, PAI-2 alone cannot bind to either receptor and fails to expose a cryptic receptor binding site upon complexing with uPA (Cochran et al. 2011). Receptor binding of the uPA:PAI-2 dimer, rather, is mediated through an epitope in uPA. This epitope, however, does not induce signaling through either LRP-1 or VLDL-R. Since PAI-2:PA binding has no effect on cell motility or proliferation, this difference in LRP-1/VLDL-R signaling by PAI-1 complexes vs. PAI-2 complexes could potentially explain why elevated PAI-2 is a marker for favorable outcome in human breast malignancies while increased PAI-1 levels indicate a poor prognostic outcome (Croucher et al. 2008).
PAI-1 and Regulation of Syndecan Signaling
PAI-1-induced, LRP-1-mediated keratinocyte migration is also regulated through syndecans. The syndecan family represents a group of transmembrane cell surface signaling receptors that interact with various components of the ECM, which facilitates their turnover. Following injury, wound margin keratinocytes synthesize and deposit unprocessed laminin-332, supporting syndecan-1 binding through the LG4/5 domain. PAI-1, which is also expressed by the wound-edge epithelium, stabilizes this interaction by preventing plasmin-initiated proteolytic processing of laminin-332 and syndecan-1 shedding (Subramanian et al. 1997). The presence of vitronectin at the wound edge can augment this process through its ability to bind PAI-1 and extend the half-life of active PAI-1 as well as engage syndecan-1. PAI-1, by titrating pericellular levels of active plasmin, promotes syndecan-1 dependent migration on unprocessed laminin-332 by preventing cleavage of the syndecan binding site LG4/5. As the proteolytic environment matures, PAI-1 and vitronectin are endocytosed and degraded. Syndecan-1 binding is lost due to proteolytic processing of laminin-332, as well as syndecan-1 ectodomain shedding; α3β1 binding to processed laminin-332 begins to attenuate keratinocyte migration and initiate hemidesmosome formation (Goldfinger et al. 1999).
PAI-1 Signaling Through Toll-Like Receptor Type-4
A more recently described receptor system capable of PAI-1-specific transmembrane signaling involves the human Toll-like receptor type-4 (TLR4). TLR4 signaling is a major contributor to the inflammatory response by the innate immune system and has been linked to the pathophysiology of diabetes mellitus, renal allograft injury, and septic shock. The severe coagulation and organ failure associated with sepsis appears attributed to the significant increase in vascular PAI-1 expression (up to 20–30 fold) due to bacterial lipopolysaccharide (LPS) and various inflammatory cytokines. In patients with septic shock, expression of PAI-1 is closely correlated with disease severity and patient mortality. Current data implicate PAI-1 as a potential activator of TLR4 signaling, at least in macrophages (Gupta et al. 2016). In PAI-1-deficient macrophages, secretion of TNFα and macrophage inflammatory protein-2 (MIP2) was significantly reduced. Function-blocking TLR4 antibodies also attenuated release of TNFα and MIP2 from PAI-1-macrophages treated with wild-type PAI-1 or various PAI-1 mutants that lacked binding capacity for PAs, LRP-1, or VN, respectively. Latent PAI-1, in contrast, was unable to stimulate macrophage cytokine release. These findings suggest a possible feed-forward loop responsible for disease severity in which LPS, or other bacterial toxins, initiates PAI-1 expression during the early stages of septic shock; the induced PAI-1 further activates TLR4 signaling to prolong and exacerbate intravascular coagulation resulting in compromised organ function and mortality.
Aberrant PAI-1 expression is the hallmark of various metabolic diseases. The development of synthetic PAI-1 inhibitors has focused on their potential utility as antithrombotic agents or as a treatment option for acute and chronic fibrotic disease (Brown 2010). Over the last 20 years, several low molecular weight PAI-1 antagonists have been developed and tested in vitro, in animal models in vivo, and in clinical trials (Brown 2010; Simone et al. 2014). Indeed, the available data suggest that inhibitory approaches directed to PAI-1 SERPIN activity may well be a promising therapeutic strategy to limit vascular fibrosis, reduce tumor aggressiveness, reduce airway remodeling in a model of chronic asthma, and promote regenerative tissue repair. While the mechanism(s) underlying these positive findings are unclear, recent studies indicate that at least one PAI-1 small molecule inhibitor, Tiplaxtinin or PAI-039, inhibits PAI-1-induced AKT activation and Stat1-phosphorylation in vascular smooth muscle cells in vitro and attenuates vascular neointima hyperplasia in vivo (Ji et al. 2016; Simone et al. 2015). It is tempting to speculate, therefore, that migrating PAI-1 signaling, by functional blockade, may have clinical implications for a spectrum of human diseases in which an increase in PAI-1 levels is a causative pathogenic factor.
- Czekay R-P, Wilkins-Port CE, Higgins SP, Freytag J, Overstreet JM, Klein RM et al. PAI-1: an integrator of cell signaling and migration. Int J Cell Biol. 2011;201: Article ID 562481. 9 pages. doi: 10.1155/2011/562481.
- Fearns C, Samad F, Loskutoff DJ. Synthesis and localization of PAI-1 in the vessel wall. In: Vadas M, Harlan J, editors. Vascular control of hemostasis, advances in vascular biology. Victoria: Gordon and Breach Publications; 1996. p. 207–26.Google Scholar
- Wind T, Hansen M, Jensen JK, Andreasen PA. The molecular basis for anti-proteolytic and non-proteolytic functions of plasminogen activator inhibitor type-1: roles of the reactive centre loop, the shutter region, the flexible joint region and the small serpine fragment. Biol Chem. 2002;383:21–36.PubMedCrossRefGoogle Scholar