MINIREVIEW The undecided serpin The ins and outs of plasminogen activator inhibitor type 2 Robert L. Medcalf and Stan J. Stasinopoulos Australian Centre for Blood Diseases, Monash University, Prahran, Victoria, Australia Introduction The plasminogen activating cascade became a much investigated enzyme system during the early 1980s, mainly for its role in maintaining vascular patency and for its effect on the extracellular matrix in the context of wound healing and cell migration. The controlled generation of the powerful protease, plasmin, from its precursor plasminogen seemed to be a relatively straightforward process at the outset: two serine pro- teases had been identified that could specifically cleave plasminogen and produce active plasmin. These pro- teases (tissue-type- and urokinase-type plasminogen activator; tPA, uPA) were in turn specifically inhibited by plasminogen activator inhibitors (PAIs)-types 1 and 2, both of which belong to the serine protease inhibitor (serpin) superfamily. Other cofactors, such as the ser- pin alpha 2 antiplasmin, the urokinase receptor (uPAR) and fibrin, were also shown to play important roles in regulating plasmin formation and activity [1]. This may have been the general consensus in the late 1980s, but nowadays it has become clear that many of the individual components of the fibrinolytic ⁄ plasminogen activating system perform other roles that could not have been foreseen. tPA, for example, is not just a ‘plasminogen activator’; it is now widely appreciated for its role in the central nervous system [2,3]. Although it can act on its classical substrate, plasmino- gen, in this compartment, it also associates with other targets, and in some cases can even act like a cytokine to activate microglial cells without engaging its cata- lytic properties [4]. Similarly, the two plasminogen acti- vator inhibitors are now known to perform additional functions. PAI-1 can act as an accessory protein that modulates the association of the uPA receptor with in- tegrins. This association, in turn, influences cell migra- tion independently of the PAI-1 protease inhibitory activity [5,6]. Keywords gene regulation; plasminogen activator inhibitor type 2; protease inhibitor; serpin Correspondence R. L. Medcalf, Australian Centre for Blood Diseases, Monash University, 6th Floor Burnet Building, 89 Commercial Road, Prahran, 3181 Victoria, Australia Fax: +61 39903 0228 Tel: +61 39903 0133 E-mail: Robert.medcalf@med.monash.edu.au (Received 31 March 2005, accepted 13 July 2005) doi:10.1111/j.1742-4658.2005.04879.x Plasminogen activator inhibitor type-2 (PAI-2) is a nonconventional serine protease inhibitor (serpin) with unique and tantalizing properties that is generally considered to be an authentic and physiological inhibitor of uro- kinase. However, the fact that only a small percentage of PAI-2 is secreted has been a long-standing argument for alternative roles for this serpin. Indeed, PAI-2 has been shown to have a number of intracellular roles: it can alter gene expression, influence the rate of cell proliferation and differ- entiation, and inhibit apoptosis in a manner independent of urokinase inhi- bition. Despite these recent advances in defining the intracellular function of PAI-2, it still remains one of the most mysterious and enigmatic mem- bers of the serpin superfamily. Abbreviations ARE, AU-rich element; IL, interleukin; K5, keratin 5; LPS, lipopolysaccharide; ov, ovalbumin; PAI, plasminogen activator inhibitor; PAUSE-1, PAI-2-upstream silencer element-1; Rb, retinoblastoma; serpin, serine protease inhibitor; TNF, tumour necrosis factor; tPA, tissue-type plasminogen activator; TTP, tristetraprolin; uPA, urokinase-type plasminogen activator; uPAR, urokinase receptor. 4858 FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS For PAI-2, there was a strong suspicion soon after its discovery that the real function of this inhibitor had been overlooked. From a teleological viewpoint, a non-uPA inhibitory role was expected, as the majority of PAI-2 was found in a location where its intended or perhaps presumed natural target (i.e. uPA) did not even exist, that being the cell cytosol [7]. This minireview will focus on the cellular and molecular biology of PAI-2 and highlight some of the most recent findings on the role and impressive pattern of regulation of this enigmatic protease inhi- bitor. Although new data is emerging, PAI-2 is still one of the most cryptic protease inhibitors known and its role in biology and pathophysiology is still being unravelled. General biology of PAI-2 PAI-2 was defined as a placental tissue-derived uPA inhibitor over three decades ago [8], which was subse- quently verified by others [9,10]. Human PAI-2 con- sists of a single chain protein of 415 amino acids encoded by a 1900 bp PAI-2 transcript and is highly homologous with mouse and rat PAI-2. PAI-2 exists predominantly as a 47 kDa nonglycosylated intracellu- lar form [7], however a small percentage of PAI-2 is able to enter the secretory pathway by a process referred to as facultative translocation [11] and secre- ted as a 60 kDa glycosylated protein. The basis for this bi-topological distribution is due to the lack of a con- ventional hydrophobic amino-terminal signal sequence. Instead, PAI-2 possesses an inefficient internal signal sequence [12]. This bi-topological (intracellular ⁄ extra- cellular) expression pattern of PAI-2 has also been shown for a related serpin known as maspin [13] and this feature remains one of the most intriguing aspects of PAI-2 (and maspin) biology. Because the uPA inhibitory capacity of both forms of PAI-2 seems to be similar, it was considered early on that the release of high local concentrations of nonglycosylated PAI-2 from dead or dying cells at sites of inflammation may provide an immediate source of enriched uPA inhibi- tory activity [14]. While anecdotal evidence would cer- tainly support this, the growing consensus of opinion, however, is that PAI-2 possesses an as yet ill-defined intracellular role. Structural considerations Based on a number of criteria, PAI-2 has been classed as a member of the ovalbumin subfamily of serine protease inhibitors known as the ovalbumin (ov)-serpins, with ovalbumin being the prototypical member of this family [15]. Ovalbumin-serpins share a common genomic structure and all lack conven- tional signal sequences and are, for the most part, located intracellularly. Closer examination of the genomic structure of PAI-2 revealed another distinctive feature, that being an extension of exon 3 that encoded a unique domain bridging helices C and D of the pro- tein. This so-called C-D interhelical domain [16], other- wise known as the C-D loop, has since been implicated in the function of PAI-2. Glutamine residues in the C-D loop can be crosslinked by tissue transglutaminase and factor XIII to structures in trophoblasts and to fibrin [16–18]. Moreover, the C-D loop has been shown to bind noncovalently to annexins, retinoblastoma protein and a number of unidentified proteins [19,20]. Using the expressed C-D interhelical loop as bait, Fan et al. identified the b1 subunit of the proteosome as a binding partner [21]. The physiological relevance of these findings remains to be clarified, but none- theless points to diverse roles of the C-D loop in PAI-2 function. Polymerization of PAI-2 Many serpins have been shown to undergo loop sheet polymerization. Generally, polymerization occurs due to a genetic aberration, which results in serious patho- logical consequences due to conformational changes of these proteins [22]. PAI-2 is also able to polymerize, but in contrast to the other polymerizing serpins this is not a consequence of a mutation in the PAI-2 gene, nor is it associated with any known pathologies. Indeed, PAI-2 displays conformational plasticity and is the only known wild type serpin to form polymers spontaneously and reversibly under physiological con- ditions [23]. Furthermore, this is influenced by the redox status of the cell: PAI-2 can exist in either a sta- ble monomeric or a polymerogenic configuration, the latter stabilized by disulfide bonds that connect a cys- teine residue within the C-D loop to another cysteine residue at the bottom of the molecule [24]. The mono- meric form is also stabilized by binding to vitronectin while retaining its inhibitory activity. Under conditions of oxidative stress, the polymerized inactive configur- ation of PAI-2 can form but whether this has any other impact on cell function is unknown. More recently it was shown that the C-D loop within the sta- ble monomeric form of PAI-2 is mobile and that the monomeric and polymerogenic forms of PAI-2 were interchangeable [25]. Hence, not only does PAI-2 exist as a bi-topological protein, it can also exist in different conformational forms within the intracellular compart- ment. R. L. Medcalf and S. J. Stasinopoulos The undecided serpin FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS 4859 Expression pattern of PAI-2 and its role in pregnancy Under normal conditions, PAI-2 has a restricted tissue distribution pattern with expression detected at high levels in keratinocytes, activated monocytes and the placenta [26]. Lower constitutive levels of PAI-2 are also found in other cells, including cells of neuronal origin [27]. Plasma levels of PAI-2 are usually low or undetectable; however, they rise significantly in some forms of monocytic leukaemia [28]. One of the most physiologically striking observations for PAI-2 con- cerns its association with pregnancy. Plasma levels of PAI-2 increase impressively during the third trimester of pregnancy (up to 250 ngÆmL )1 ) and are maintained at these levels for up to 1 week postpartum and then rapidly decline [10]. The tissue source of plasma PAI-2 is the placenta itself. Indeed, PAI-2 is highly expressed in trophoblasts [29,30] and it was conjectured that PAI-2 acted to protect the placenta from proteolytic degradation towards the end of the gestational period and to regulate postpartum haemostasis. However, a placental associated PAI-2 sensitive protease is yet to be described. Perhaps the role of PAI-2 in the placenta is unrelated to protease inhibition. In this regard, it is interesting to point out that PAI-2 forms complexes with other placental proteins, including vitronectin [9,31], but the functional significance of this in terms of placental biology is unknown. The association of PAI-2 with pregnancy and its placenta-specific expression suggested that PAI-2 might perform a critical function during foetal devel- opment. If this were indeed the case, one would have predicted a developmental abnormality in PAI-2 – ⁄ – mice. Mice with a targeted deletion in the PAI-2 gene have been described but these mice have not as yet displayed any noticeable phenotype [32] at least under normal, nonchallenging conditions. To exclude the possibility that the lack of effect was due to redund- ancy with PAI-1, a double knock-out mouse was pro- duced that harboured a disruption at both the PAI-1 and PAI-2 loci. Still, no obvious phenotype was seen. Given the high degree of PAI-2 expression in the human placenta, it was surprising at first glance that foetal development and reproduction was undisturbed in PAI-2 – ⁄ – mice. However, no firm conclusions can be drawn from this as, unlike the human situation, PAI-2 is not found in the mouse placenta. It is indeed a strange curiosity that the presence and regu- lation of placental PAI-2 is not conserved in the mouse. However, this important data has only been presented as a statement within a review article [33] and additional supportive information would be welcomed on the presence or absence of PAI-2 in the mouse placenta. The role of PAI-2 in the skin PAI-2 expression within the skin is restricted to the upper layers of the dermis. PAI-2 has also been repor- ted to inhibit keratinocyte proliferation [34] and to play a role in keratinocyte differentiation [34]. A cleaved form of PAI-2 has been found in keratinocytes [35] implying that PAI-2 itself is a substrate for a pro- tease in these cells. To determine the consequences of dysregulation of PAI-2 on epidermal differentiation, Zhou et al. [36] produced transgenic mice that overexpressed PAI-2 in the proliferating layers of mouse epidermis and hair follicle cells by placing the PAI-2 transgene under the control of the keratin 5 (K5) promoter. Although the presence of PAI-2 had no effect on skin morphology or proliferation under normal conditions, the PAI-2 overexpressing mice were found to be highly suscept- ible to chemically induced papilloma formation. The means by which PAI-2 promoted papilloma formation is unknown, but may have been related to its reported antiapoptotic effect (see below) since cessation of tumour promoting treatment in control mice resulted in extensive apoptosis of the papilloma but not in the K5-PAI-2 transgenic mouse. Leukocyte biology Monocytes and macrophages express PAI-2 and levels are impressively increased in these cells following sti- mulation with tumour necrosis factor (TNF) [14] and lipopolysaccharide (LPS) [37,38]. Induction of PAI-2 gene expression has been associated with monocyte dif- ferentiation, at least in the U-937 monocyte-like cell system [39], suggesting a role for PAI-2 in this process. In the mouse system PAI-2 does not appear to be indispensable for leukocyte development as PAI-2 – ⁄ – mice exhibit normal leukocyte recruitment and appear to differentiate normally [32]. Novel insights into the role of PAI-2 in monocytes came from studies using THP-1 cells. Unlike all other widely used monocyte-like cell lines (e.g. U-937, K562, HL60) that express endogenous PAI-2, the THP-1 monocytic cells provided a notable exception to this rule. THP-1 cells bear many features common to regu- lar mononuclear phagocytes, but are closer in pheno- type to a mature monocyte than other monocytic cell lines (i.e. U-937, K562). Although the expression pat- tern of THP-1-derived uPA and its receptor (uPAR) is similar to that observed in other monocytic cell lines The undecided serpin R. L. Medcalf and S. J. Stasinopoulos 4860 FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS [40,41], THP-1 cells do not express a functional PAI-2 protein [40]. These authors demonstrated that THP-1- derived PAI-2 was functionally inactive while the PAI-2 transcript in these cells was truncated. The molecular basis for the aberrant production of PAI-2 in THP-1 cells is due to a translocation anomaly [42]. The complete absence of a functional PAI-2 in these cells defined THP-1 cells effectively as a human monocytic PAI-2 – ⁄ – cell line. To take advantage of this PAI-2 – ⁄ – cell line, Yu et al. [43] produced stable THP-1 cell lines that expressed either wild type PAI-2 or a PAI-2 mutant containing an alanine substitution at the P1 position (Arg380). The presence of wild type PAI-2 caused a significant decrease in THP-1 cell prolifer- ation, reduction in DNA synthesis and a phenotypic change following phorbol ester-induced differentiation. The ability of PAI-2 to alter the differentiation process was dependent on its active form as cells expressing PAI-2 Ala380 did not display these changes. This study demonstrated for the first time an intracellular role for active PAI-2 in monocytes. These results were con- sistent with the possibility that PAI-2 disrupted an intracellular protease(s) that was involved in cell prolif- eration and ⁄ or differentiation although no such target protease has been detected thus far. PAI-2 is also present at very high levels in eosino- philic leukocytes. Indeed the level of PAI-2 in these cells was shown to be the highest among all other leu- kocyte subtypes [44]. Furthermore, PAI-2 was localized to eosinophil-specific granules and shown to be still capable of inhibiting urokinase. It was suggested that PAI-2 might play a role in eosinophil mediated inflam- mation and tissue remodelling. Role of intranuclear PAI-2 A number of Ov-serpins have been detected within the nuclear compartment, including bomapin, PI-9, and maspin [45–47]. PAI-2 has also been shown to have a nuclear presence [20,45,46] yet the physiological role of PAI-2 in this compartment is unknown. However, in a study by Darnell et al. nuclear-located PAI-2 was shown to bind to retinoblastoma protein (Rb) via its CD-loop [20]. Rb is a prototypical tumour suppressor gene and critical cell cycle regulator that targets the E2F family of transcription factors [48]. PAI-2 colocal- ized with Rb and, interestingly, inhibited Rb turnover by protecting Rb from proteolysis [20]. This in turn led to an increase in Rb protein levels and Rb-medi- ated activities including the transcriptional repression of oncogenes. This is a curious finding because PAI-2 – ⁄ – mice do not appear to have any change in cell number, and it would be predicted that Rb turnover would be accelerated in PAI-2 – ⁄ – mice freeing E2Fs to mediate proliferation. Although additional evidence is required to explore the consequences of PAI-2 and Rb inter- action, these data underscore a novel and previously unsuspected intranuclear role for PAI-2. The role of PAI-2 in metastatic cancer, apoptosis and infection Cancer A number of in vivo studies have assessed the prognos- tic relevance of tumour- and stromal-derived PAI-2 in the metastatic spread of cancer of the neck, lung and breast [49–53]. The only established protease target for PAI-2, namely uPA, is strongly implicated in facilita- ting cell dissemination in the context of tumour meta- stasis and it is likely that the beneficial effect of PAI-2 seen in these studies is simply via uPA inhibition. Overexpression of PAI-2 in melanoma cells prevented spontaneous metastasis of transplanted cells in scid mice [54], while overexpression of PAI-2 in HT-1080 cells has also been shown to reduce uPA-dependent cell movement in vitro and metastatic development in vivo [55]. The ability of PAI-2 to selectively bind to cell surface bound uPA (via uPAR) and subsequently be internalized [56] has prompted studies to assess the effectiveness of PAI-2 as a delivery vehicle for isotopes ( 213 Bi) and toxins that can be targeted to uPA-bearing cancer cells This approach has provided positive out- comes at least in some preclinical studies [57–59]. Apoptosis Circumstantial evidence that first implicated PAI-2 as an inhibitor of apoptosis came from genetic associ- ation studies with BCL-2 [60]. The BCL-2 proto- oncogene was discovered over 15 years ago as the archetype inhibitor of apoptosis. Evidence that BCL-2 was playing such a role in humans came from studies in patients with follicular lymphoma. In these patients, a translocation event occurs between chromosomes 14 and 18 t(14; 18) that brings the BCL-2 gene into juxta- position with the locus of the immunoglobulin heavy chain, resulting in overexpression of BCL-2 [61]. This in turn inhibits the apoptotic process of the lym- phoma. The relevance of this to PAI-2 stemmed from the fact that the PAI-2 gene is located less than 300 mbp from the BCL-2 gene and is translocated along with BCL-2 in patients with follicular lym- phoma. PAI-2 and BCL-2 also share structural similar- ities and it was proposed that the function of PAI-2 may overlap with BCL-2. So with this background, a R. L. Medcalf and S. J. Stasinopoulos The undecided serpin FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS 4861 number of publications in the mid-1990s provided in vitro evidence that PAI-2 could inhibit TNF-induced apoptosis in HT-1080 fibrosarcoma cells [62] and HeLa cells [63]. A cleaved form of intracellular PAI-2 has been found in ND4 monocytes undergoing apoptosis [64]. In no case has an intracellular PAI-2-sensitive proteinase been identified. Other reports, however, have provided contradictory data [18]. One argument in the interpretation of the significance of PAI-2 dur- ing apoptosis concerns the level of enforced expression of PAI-2 in the model systems used. In most of these in vitro studies, PAI-2 was overexpressed in cells that either did not make PAI-2 at all or were expressed to levels that well exceeded endogenous expression levels. Under these conditions, PAI-2 may indeed inactivate one or more intracellular proteases, but whether this genuinely reflects the in vivo role of PAI-2 can be rea- sonably debated. Viral infection Evidence to suggest that PAI-2 participates in the host response to alphaviral infection is based on over- expression studies in HeLa cells. The protective effect of PAI-2 was indirect, as PAI-2 appeared increase interferon levels which then triggered an increase in the expression of a battery of antiviral genes [65]. Shafren et al. [66] also demonstrated the same PAI-2 over-expressing HeLa cells were protected from lytic infection by human picornaviruses. In this case, PAI-2 promoted the transcriptional down-regulation of sur- face expression of picornavirus receptors (decay accel- erating factor, intercellular adhesion molecule-1 and coxsachievirus-adenovirus receptor; DAF, ICAM-1 and CAR, respectively). These observations further support the growing body of evidence [42,43] that intracellular expression of PAI-2 is linked to a signal- ling pathway(s) that can reprogram gene expression. One may even speculate that PAI-2 could play a role in the innate immune response since its expression is commonly associated with inflammation and the host response to infection. PAI-2 gene expression and regulation Based on data accumulated over the past 17 years, it is evident that the PAI-2 gene expression can be induced by a wide range of agonists. Moreover the level of PAI-2 gene induction in some examples is quite extra- ordinary. Agonists of PAI-2 induction include growth factors (transforming growth factor-b, epidermal growth factor and monocyte-colony stimulating fac- tor; TGFb, EGF and M-CSF, respectively), hormones (retinoic acid, dexamethasone and vitamin D3), cyto- kines [TNFa, interleukin (IL)-1 and IL-2)], vasoactive peptides (angiotensin II), toxins (dioxin and endotoxin) and tumour promoters (phorbol esters and okadaic acid) [26,67]. PAI-2 mRNA expression is also strongly increased by the excitotoxic glutamate analogue, kainic acid in neuronal cells in vivo [27]. PAI-2 was cloned by groups that had an intent focus on the cell and molecular biology of PAI-2 [39,68,69], and by others inadvertently through differ- ential gene expression studies. For the latter, PAI-2 was identified as a TNF responsive gene in monocytes and fibroblasts [70,71] and as a dioxin responsive gene in keratinocytes [72]. Microarray studies identified PAI-2 as an inducible gene in response to IL-5 [73], factor 7 ⁄ tissue factor [74], and again by TNF [75]. Dif- ferential gene expression profiling (SAGE) of LPS-trea- ted primary human monocytes identified PAI-2 as the third most inducible gene being induced 105-fold by this agent [37]. In a microarray study to identify Lp(a) inducible genes in human monocytes, PAI-2 mRNA was found to be the most induced transcript from a screen of 8000 cDNAs [76]. These latter studies pro- vide further evidence of the diverse repertoire of agents that strongly regulate PAI-2 expression and by associ- ation, PAI-2 is likely to play a role in the biological consequences initiated by these agents. The impressive magnitude of induction by such a variety of biological agents prompted many laborator- ies to explore the transcriptional and post-transcrip- tional processes underlying PAI-2 expression. Transcriptional regulation of PAI-2 expression Run-on transcription assays provided direct evidence that the induction of PAI-2 expression in U-937 cells following phorbol ester treatment involved dramatic increases in the rate of PAI-2 transcription [39]. Similar studies in HT-1080 fibrosarcoma cells demonstrated a transcriptional component following TNF-mediated induction of PAI-2 expression [14]. These studies led to an analysis of the PAI-2 promoter [77,78]. DNase-1 protection experiments indicated that the proximal region of the PAI-2 promoter possessed a congested arrangement of cis-acting elements. Of these, only the AP1-like elements, AP1a (TGAATCA, )103 to )97) and AP1b (TGAGTAA, )114 to )108), and a cAMP responsive element (CRE)-like element (TGACCTCA, )187 to )182) [77,79] were shown to have functional activity during transcriptional regulation. Curiously, a repressor element located between )219 and )1100 was suggested to play a role during TNF induction [80]. The identification of the exact sequence within this The undecided serpin R. L. Medcalf and S. J. Stasinopoulos 4862 FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS region and trans-acting factors responsible for this activity have not been reported. Antalis et al. [81] characterized 5.1 kb of 5¢ flanking region in U937 cells by deletion analysis and found a silencer between )1977 and )1675 that acts in an orientation- and position-independent but not cell-specific manner. The silencer activity was localized to a 28 bp sequence containing a 12 bp palindrome at position )1832, CTCTCTAGAGAG, which was termed PAI-2- upstream silencer element-1 (PAUSE-1). Later analysis defined the minimal functional PAUSE-1 element as TCTN x AGAN 3 T 4 , where x ¼ 0, 2 or 4 [82]. UV-cross- linking analyses determined that the PAUSE-1 binding protein was  67 kDa, but its identity remains unknown. In their study, PAUSE-1 was not character- ized in the context of TNFa induction and it would be worthwhile to explore the relationship between PAUSE-1 and the element that selectively represses TNFa inducibility in HT-1080 cells [80]. Post-transcriptional regulation of PAI-2 expression As mentioned earlier, PAI-2 is one of the most highly regulated genes known, at least in terms of the magni- tude by which it is induced by growth factors, hor- mones, cytokines [73,83] and tumour promoters [39,84]. Although PAI-2 induction involves substantial changes at the level of transcription, post-transcrip- tional events are also important in modulating its expression. This was first revealed over a decade ago, when it was shown that the increase in PAI-2 mRNA after synergetic stimulation by phorbol myristate ace- tate and TNFa (1000- to 1500-fold) could not be accounted for by an increase in PAI-2 transcription rate alone (50-fold), suggesting that post-transcrip- tional processes influence PAI-2 gene expression [84]. The PAI-2 transcript has since proven to be a valu- able model to study post-transcriptional regulation, most notably at the level of mRNA instability. PAI-2 mRNA contains a functional nonameric (UUAUUUAUU) AU-rich element (ARE) in its 3¢-un- translated region [85]. Mutagenesis of this element par- tially stabilized the normally unstable PAI-2 mRNA, hence revealing a functional role for this motif [85,86]. This element also provides binding sites for several ARE binding proteins, including the stabilizing protein HuR [86] and the mRNA destabilizing protein tristetr- aprolin (TTP) [87]. HuR is a member of the Hu family of mRNA binding proteins and has been associated with promotion of mRNA stability [88]. TTP, on the other hand, is a potent mRNA destabilizing protein that associates with ARE elements in cytokine tran- scripts, including TNF a [89] and IL-3 [90]. Overexpres- sion of TTP in HEK 293 cells transfected with a constitutively active PAI-2 expression vector resulted in loss of PAI-2 mRNA, suggesting that TTP can indeed regulate PAI-2 expression [86]. Other cytoplas- mic and nuclear proteins also bind to the ARE with the PAI-2 3¢-UTR [85,86] but these are yet to be iden- tified. The PAI-2 transcript also possesses another instability determinant located within exon 4 of the PAI-2 coding region [91]. UV-crosslinking studies have identified two RNA-binding proteins (approximately 50–52 kDa) that specifically interact with this sequence. Taken together, the data published to date suggest that PAI-2 mRNA stability is influenced by elements located within both the coding region and the 3¢-UTR. It remains to be determined whether these instability elements in the coding region and the 3¢-UTR act in a coordinated fashion to control PAI-2 mRNA stability (Fig. 1). Conclusion PAI-2 has been implicated in many facets of biology some of which are unrelated to its ability to inhibit extracellular uPA. However, the ability of PAI-2 to reduce the metastatic potential of a number of cancers, Fig. 1. Schematic representation of regula- tory domains within the PAI-2 transcript that influence PAI-2 expression at the post-tran- scriptional level. At least two domains exist: one within exon 4 (E4) of the coding region and the other within the 3¢-UTR. Proteins that have been shown to bind to these regions in vitro are shown. See text for details. E, exon. R. L. Medcalf and S. J. Stasinopoulos The undecided serpin FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS 4863 presumably via inhibition of extracellular or cell-sur- face bound uPA, is arguably the most consistent and physiologically relevant finding to date. Nonetheless, the response of the PAI-2 gene to such a diverse reper- toire of agonists and the impressive magnitude of induction in leukocytes in response to toxins and cytokines invokes PAI-2, albeit circumstantially, with inflammation, tissue repair and possibly the innate immune response. Similarly, the evidence linking PAI-2 with apoptotic processes, Rb turnover, cell prolife- ration and differentiation is substantial and gaining momentum but more direct and physiologically focused experiments are needed in order to define its undisputed intracellular function. It is anticipated that this information will be forthcoming through a more extensive analysis of the PAI-2 – ⁄ – mice. Results from these experiments are eagerly awaited. Acknowledgements This study was supported by grants obtained by RLM from the National Health and Medical Research Council of Australia. References 1 Bachmann F (1987) Fibrinolysis. In Thrombosis and Haemostasis. (Verstraete M, ed), pp. 227–265. Leuven University Press, Leuven, the Netherlands. 2 Tsirka SE (2002) Tissue plasminogen activator as a modulator of neuronal survival and function. Biochem Soc Trans 30, 222–225. 3 Nicole O, Docagne F, Ali C, Margaill I, Carmeliet P, MacKenzie ET, Vivien D & Buisson A (2001) The proteolytic activity of tissue-plasminogen activator enhances NMDA receptor-mediated signaling. Nat Med 7, 59–64. 4 Rogove AD, Siao C, Keyt B, Strickland S & Tsirka SE (1999) Activation of microglia reveals a non-proteolytic cytokine function for tissue plasminogen activator in the central nervous system. J Cell Sci 112, 4007–4016. 5 Waltz DA, Natkin LR, Fujita RM, Wei Y & Chapman HA (1997) Plasmin and plasminogen activator inhibitor type 1 promote cellular motility by regulating the inter- action between the urokinase receptor and vitronectin. J Clin Invest 100, 58–67. 6 Czekay RP & Loskutoff DJ (2004) Unexpected role of plasminogen activator inhibitor 1 in cell adhesion and detachment. Exp Biol Med (Maywood) 229, 1090–1096. 7 Genton C, Kruithof EK & Schleuning WD (1987) Phor- bol ester induces the biosynthesis of glycosylated and nonglycosylated plasminogen activator inhibitor 2 in high excess over urokinase-type plasminogen activator in human U-937 lymphoma cells. J Cell Biol 104, 705– 712. 8 Kawano T, Morimoto K & Uemura Y (1970) Partial purification and properties of urokinase inhibitor from human placenta. J Biochem (Tokyo) 67, 333–342. 9 Wun TC & Reich E (1987) An inhibitor of plasminogen activation from human placenta. Purification and char- acterization. J Biol Chem 262, 3646–3653. 10 Kruithof EK, Tran-Thang C, Gudinchet A, Hauert J, Nicoloso G, Genton C, Welti H & Bachmann F (1987) Fibrinolysis in pregnancy: a study of plasminogen acti- vator inhibitors. Blood 69, 460–466. 11 Belin D, Wohlwend A, Schleuning WD, Kruithof EK & Vassalli JD (1989) Facultative polypeptide translocation allows a single mRNA to encode the secreted and cyto- solic forms of plasminogen activators inhibitor 2. EMBO J 8, 3287–3294. 12 von Heijne G, Liljestrom P, Mikus P, Andersson H & Ny T (1991) The efficiency of the uncleaved secretion signal in the plasminogen activator inhibitor type 2 pro- tein can be enhanced by point mutations that increase its hydrophobicity. J Biol Chem 266, 15240–15243. 13 Pemberton PA, Tipton AR, Pavloff N, Smith J, Erick- son JR, Mouchabeck ZM & Kiefer MC (1997) Maspin is an intracellular serpin that partitions into secretory vesicles and is present at the cell surface. J Histochem Cytochem 45, 1697–1706. 14 Medcalf RL, Kruithof EK & Schleuning WD (1988) Plasminogen activator inhibitor 1 and 2 are tumor necrosis factor ⁄ cachectin-responsive genes. J Exp Med 168, 751–759. 15 Remold-O’Donnell E (1993) The ovalbumin family of serpin proteins. FEBS Lett 315, 105–108. 16 Jensen PH, Schuler E, Woodrow G, Richardson M, Goss N, Hojrup P, Petersen TE & Rasmussen LK (1994) A unique interhelical insertion in plasminogen activator inhibitor-2 contains three glutamines, Gln83, Gln84, Gln86, essential for transglutaminase-mediated cross-linking. J Biol Chem 269, 15394–15398. 17 Jensen PH, Lorand L, Ebbesen P & Gliemann J (1993) Type-2 plasminogen-activator inhibitor is a substrate for trophoblast transglutaminase and factor XIIIa. Transglutaminase-catalyzed cross-linking to cellular and extracellular structures. Eur J Biochem 214, 141–146. 18 Ritchie H, Lawrie LC, Crombie PW, Mosesson MW & Booth NA (2000) Cross-linking of plasminogen activa- tor inhibitor 2 and alpha 2-antiplasmin to fibrin(ogen). J Biol Chem 275, 24915–24920. 19 Jensen PH, Jensen TG, Laug WE, Hager H, Gliemann J & Pepinsky B (1996) The exon 3 encoded sequence of the intracellular serine proteinase inhibitor plasminogen activator inhibitor 2 is a protein binding domain. J Biol Chem 271, 26892–26899. 20 Darnell GA, Antalis TM, Johnstone RW, Stringer BW, Ogbourne SM, Harrich D & Suhrbier A (2003) Inhibi- The undecided serpin R. L. Medcalf and S. J. Stasinopoulos 4864 FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS tion of retinoblastoma protein degradation by interac- tion with the serpin plasminogen activator inhibitor 2 via a novel consensus motif. Mol Cell Biol 23, 6520–6532. 21 Fan J, Zhang YQ, Li P, Hou M, Tan L, Wang X & Zhu YS (2004) Interaction of plasminogen activator inhibitor-2 and proteasome subunit, beta type 1. Acta Biochim Biophys Sin (Shanghai) 36, 42–46. 22 Lomas DA & Carrell RW (2002) Serpinopathies and the conformational dementias. Nat Rev Genet 3, 759– 768. 23 Mikus P & Ny T (1996) Intracellular polymerization of the serpin plasminogen activator inhibitor type 2. J Biol Chem 271, 10048–10053. 24 Wilczynska M, Lobov S & Ny T (2003) The sponta- neous polymerization of plasminogen activator inhibitor type-2 and Z-antitrypsin are due to different molecular aberrations. FEBS Lett 537, 11–16. 25 Lobov S, Wilczynska M, Bergstrom F, Johansson LB & Ny T (2004) Structural bases of the redox-dependent conformational switch in the serpin PAI-2. J Mol Biol 344, 1359–1368. 26 Kruithof EK, Baker MS & Bunn CL (1995) Biological and clinical aspects of plasminogen activator inhibitor type 2. Blood 86, 4007–4024. 27 Sharon R, Abramovitz R & Miskin R (2002) Plasmino- gen mRNA induction in the mouse brain after kainate excitation: codistribution with plasminogen activator inhibitor-2 (PAI-2) mRNA. Brain Res Mol Brain Res 104, 170–175. 28 Scherrer A, Kruithof EK & Grob JP (1991) Plasmino- gen activator inhibitor-2 in patients with monocytic leu- kemia. Leukemia 5, 479–486. 29 Hofmann GE, Glatstein I, Schatz F, Heller D & Delig- disch L (1994) Immunohistochemical localization of urokinase-type plasminogen activator and the plasmino- gen activator inhibitors 1 and 2 in early human implan- tation sites. Am J Obstet Gynecol 170, 671–676. 30 Black S, YuH, Lee J, Sachchithananthan M & Medcalf RL (2001) Physiologic concentrations of magnesium and placental apoptosis: prevention by antioxidants. Obstet Gynecol 98, 319–324. 31 Radtke KP, Wenz KH & Heimburger N (1990) Isola- tion of plasminogen activator inhibitor-2 (PAI-2) from human placenta. Evidence for vitronectin ⁄ PAI-2 com- plexes in human placenta extract. Biol Chem Hoppe Seyler 371, 1119–1127. 32 Dougherty KM, Pearson JM, Yang AY, Westrick RJ, Baker MS & Ginsburg D (1999) The plasminogen acti- vator inhibitor-2 gene is not required for normal murine development or survival. Proc Natl Acad Sci USA 96, 686–691. 33 Belin D (1993) Biology and facultative secretion of plasminogen activator inhibitor-2. Thromb Haemost 70, 144–147. 34 Hibino T, Matsuda Y, Takahashi T & Goetinck PF (1999) Suppression of keratinocyte proliferation by plas- minogen activator inhibitor-2. J Invest Dermatol 112, 85–90. 35 Risse BC, Chung NM, Baker MS & Jensen PJ (2000) Evidence for intracellular cleavage of plasminogen acti- vator inhibitor type 2 (PAI-2) in normal epidermal kera- tinocytes. J Cell Physiol 182, 281–289. 36 Zhou HM, Bolon I, Nichols A, Wohlwend A & Vassalli JD (2001) Overexpression of plasminogen acti- vator inhibitor type 2 in basal keratinocytes enhances papilloma formation in transgenic mice. Cancer Res 61, 970–976. 37 Suzuki T, Hashimoto S, Toyoda N, Nagai S, Yamazaki N, Dong HY, Sakai J, Yamashita T, Nukiwa T & Mat- sushima K (2000) Comprehensive gene expression pro- file of LPS-stimulated human monocytes by SAGE. Blood 96, 2584–2591. 38 Schwartz BS & Bradshaw JD (1992) Regulation of plas- minogen activator inhibitor mRNA levels in lipopoly- saccharide-stimulated human monocytes. Correlation with production of the protein. J Biol Chem 267, 7089– 7094. 39 Schleuning WD, Medcalf RL, Hession C, Rothenbuhler R, Shaw A & Kruithof EK (1987) Plasminogen activa- tor inhibitor 2: regulation of gene transcription during phorbol ester-mediated differentiation of U-937 human histiocytic lymphoma cells. Mol Cell Biol 7, 4564–4567. 40 Gross TJ & Sitrin RG (1990) The THP-1 cell line is a urokinase-secreting mononuclear phagocyte with a novel defect in the production of plasminogen activator inhi- bitor-2. J Immunol 144, 1873–1879. 41 Ragno P, Montuori N, Vassalli JD & Rossi G (1993) Processing of complex between urokinase and its type-2 inhibitor on the cell surface. A possible regulatory mechanism of urokinase activity. FEBS Lett 323, 279– 284. 42 Katsikis J, YuH, Maurer F & Medcalf R (2000) The molecular basis for the aberrant production of plasmi- nogen activator inhibitor type 2 in THP-1 monocytes. Thromb Haemost 84, 468–473. 43 Yu H, Maurer F & Medcalf RL (2002) Plasminogen activator inhibitor type 2: a regulator of monocyte pro- liferation and differentiation. Blood 99, 2810–2818. 44 Swartz JM, Bystrom J, Dyer KD, Nitto T, Wynn TA & Rosenberg HF (2004) Plasminogen activator inhibitor-2 (PAI-2) in eosinophilic leukocytes. J Leukoc Biol 76, 812–819. 45 Chuang TL & Schleef RR (1999) Identification of a nuclear targeting domain in the insertion between helices C and D in protease inhibitor-10. J Biol Chem 274, 11194–11198. 46 Bird CH, Blink EJ, Hirst CE, Buzza MS, Steele PM, Sun J, Jans DA & Bird PI (2001) Nucleocytoplasmic distribution of the ovalbumin serpin PI-9 requires a R. L. Medcalf and S. J. Stasinopoulos The undecided serpin FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS 4865 nonconventional nuclear import pathway and the export factor Crm1. Mol Cell Biol 21, 5396–5407. 47 Marioni G, Blandamura S, Giacomelli L, Calgaro N, Segato P, Leo G, Fischetto D, Staffieri A & de Filippis C (2005) Nuclear expression of maspin is associated with a lower recurrence rate and a longer disease-free interval after surgery for squamous cell carcinoma of the larynx. Histopathology 46, 576–582. 48 Harbour JW & Dean DC (2000) Rb function in cell- cycle regulation and apoptosis. Nat Cell Biol 2, E65– E67. 49 Hasina R, Hulett K, Bicciato S, Di Bello C, Petruzzelli GJ & Lingen MW (2003) Plasminogen activator inhibi- tor-2: a molecular biomarker for head and neck cancer progression. Cancer Res 63, 555–559. 50 Borstnar S, Vrhovec I, Svetic B & Cufer T (2002) Prog- nostic value of the urokinase-type plasminogen activa- tor, and its inhibitors and receptor in breast cancer patients. Clin Breast Cancer 3, 138–146. 51 Yoshino H, Endo Y, Watanabe Y & Sasaki T (1998) Significance of plasminogen activator inhibitor 2 as a prognostic marker in primary lung cancer: association of decreased plasminogen activator inhibitor 2 with lymph node metastasis. Br J Cancer 78, 833–839. 52 Foekens JA, Buessecker F, Peters HA, Krainick U, van Putten WL, Look MP, Klijn JG & Kramer MD (1995) Plasminogen activator inhibitor-2: prognostic relevance in 1012 patients with primary breast cancer. Cancer Res 55, 1423–1427. 53 Duggan C, Kennedy S, Kramer MD, Barnes C, Elvin P, McDermott E, O’Higgins N & Duffy MJ (1997) Plas- minogen activator inhibitor type 2 in breast cancer. Br J Cancer 76, 622–627. 54 Mueller BM, Yu, YB & Laug WE (1995) Overexpres- sion of plasminogen activator inhibitor 2 in human mel- anoma cells inhibits spontaneous metastasis in scid ⁄ scid mice. Proc Natl Acad Sci USA 92, 205–209. 55 Laug WE, Cao XR, Yu YB, Shimada H & Kruithof EK (1993) Inhibition of invasion of HT1080 sarcoma cells expressing recombinant plasminogen activator inhi- bitor 2. Cancer Res 53, 6051–6057. 56 Al-Ejeh F, Croucher D & Ranson M (2004) Kinetic analysis of plasminogen activator inhibitor type-2: urokinase complex formation and subsequent inter- nalisation by carcinoma cell lines. Exp Cell Res 297, 259–271. 57 Allen BJ, Tian Z, Rizvi SM, Li Y & Ranson M (2003) Preclinical studies of targeted alpha therapy for breast cancer using 213Bi-labelled-plasminogen activator inhi- bitor type 2. Br J Cancer 88, 944–950. 58 Ranson M, Tian Z, Andronicos NM, Rizvi S & Allen BJ (2002) In vitro cytotoxicity of bismuth-213 (213Bi)- labeled-plasminogen activator inhibitor type 2 (alpha- PAI-2) on human breast cancer cells. Breast Cancer Res Treat 71, 149–159. 59 Allen BJ, Rizvi S, Li Y, Tian Z & Ranson M (2001) In vitro and preclinical targeted alpha therapy for mela- noma, breast, prostate and colorectal cancers. Crit Rev Oncol Hematol 39, 139–146. 60 Silverman GA, Jockel JI, Domer PH, Mohr RM, Tail- lon-Miller P & Korsmeyer SJ (1991) Yeast artificial chromosome cloning of a two-megabase-size contig within chromosomal band 18q21 establishes physical linkage between BCL2 and plasminogen activator inhi- bitor type-2. Genomics 9, 219–228. 61 Weiss LM, Warnke RA, Sklar J & Cleary ML (1987) Molecular analysis of the t(14;18) chromosomal trans- location in malignant lymphomas. N Engl J Med 317, 1185–1189. 62 Kumar S & Baglioni C (1991) Protection from tumor necrosis factor-mediated cytolysis by overexpression of plasminogen activator inhibitor type-2. J Biol Chem 266, 20960–20964. 63 Dickinson JL, Bates EJ, Ferrante A & Antalis TM (1995) Plasminogen activator inhibitor type 2 inhibits tumor necrosis factor alpha-induced apoptosis. Evidence for an alternate biological function. J Biol Chem 270, 27894–27904. 64 Jensen PH, Cressey LI, Gjertsen BT, Madsen P, Mell- gren G, Hokland P, Gliemann J, Doskeland SO, Lanotte M & Vintermyr OK (1994) Cleaved intracellu- lar plasminogen activator inhibitor 2 in human myelo- leukaemia cells is a marker of apoptosis. Br J Cancer 70, 834–840. 65 Antalis TM, La Linn M, Donnan K, Mateo L, Gardner J, Dickinson JL, Buttigieg K & Suhrbier A (1998) The serine proteinase inhibitor (serpin) plasminogen activa- tion inhibitor type 2 protects against viral cytopathic effects by constitutive interferon alpha ⁄ beta priming. J Exp Med 187, 1799–1811. 66 Shafren DR, Gardner J, Mann VH, Antalis TM & Suhrbier A (1999) Picornavirus receptor down-regula- tion by plasminogen activator inhibitor type 2. J Virol 73, 7193–7198. 67 Dear AE & Medcalf RL (1995) The cellular and mole- cular biology of plasminogen activator inhibitor type-2. Fibrinolysis 9, 321–330. 68 Antalis TM, Clark MA, Barnes T, Lehrbach PR, Devine PL, Schevzov G, Goss NH, Stephens RW & Tolstoshev P (1988) Cloning and expression of a cDNA coding for a human monocyte-derived plasminogen acti- vator inhibitor. Proc Natl Acad Sci USA 85, 985–989. 69 YeRD, Ahern SM, Le Beau MM, Lebo RV & Sadler JE (1989) Structure of the gene for human plasmino- gen activator inhibitor-2. The nearest mammalian homologue of chicken ovalbumin. J Biol Chem 264, 5495–5502. 70 Webb AC, Collins KL, Snyder SE, Alexander SJ, Rosenwasser LJ, Eddy RL, Shows TB & Auron PE (1987) Human monocyte Arg-Serpin cDNA. Sequence, The undecided serpin R. L. Medcalf and S. J. Stasinopoulos 4866 FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS chromosomal assignment, and homology to plasmino- gen activator-inhibitor. J Exp Med 166, 77–94. 71 Pytel BA, Peppel K & Baglioni C (1990) Plasminogen activator inhibitor type-2 is a major protein induced in human fibroblasts and SK-MEL-109 melanoma cells by tumor necrosis factor. J Cell Physiol 144, 416–422. 72 Sutter TR, Guzman K, Dold KM & Greenlee WF (1991) Targets for dioxin: genes for plasminogen activa- tor inhibitor-2 and interleukin-1 beta. Science 254, 415–418. 73 Bystrom J, Wynn TA, Domachowske JB & Rosenberg HF (2004) Gene microarray analysis reveals interleukin- 5-dependent transcriptional targets in mouse bone marrow. Blood 103, 868–877. 74 Camerer E, Gjernes E, Wiiger M, Pringle S & Prydz H (2000) Binding of factor VIIa to tissue factor on kerati- nocytes induces gene expression. J Biol Chem 275, 6580–6585. 75 Jang WG, Kim HS, Park KG, Park YB, Yoon KH, Han SW, Hur SH, Park KS & Lee IK (2004) Analysis of proteome and transcriptome of tumor necrosis factor alpha stimulated vascular smooth muscle cells with or without alpha lipoic acid. Proteomics 4, 3383–3393. 76 Buechler C, Ullrich H, Ritter M, Porsch-Oezcueruemez M, Lackner KJ, Barlage S, Friedrich SO, Kostner GM & Schmitz G (2001) Lipoprotein(a) up-regulates the expression of the plasminogen activator inhibitor 2 in human blood monocytes. Blood 97, 981–986. 77 Cousin E, Medcalf RL, Bergonzelli GE & Kruithof EK (1991) Regulatory elements involved in constitutive and phorbol ester-inducible expression of the plasminogen activator inhibitor type 2 gene promoter. Nucleic Acids Res 19, 3881–3886. 78 Kruithof EK & Cousin E (1988) Plasminogen activator inhibitor 2. Isolation and characterization of the promo- ter region of the gene. Biochem Biophys Res Commun 156, 383–388. 79 Dear AE, Costa M & Medcalf RL (1997) Urokinase- mediated transactivation of the plasminogen activator inhibitor type 2 (PAI-2) gene promoter in HT-1080 cells utilises AP-1 binding sites and potentiates phorbol ester- mediated induction of endogenous PAI-2 mRNA. FEBS Lett 402, 265–272. 80 Dear AE, Shen Y, Ruegg M & Medcalf RL (1996) Molecular mechanisms governing tumor-necrosis-factor- mediated regulation of plasminogen-activator inhibitor type-2 gene expression. Eur J Biochem 241, 93–100. 81 Antalis TM, Costelloe E, Muddiman J, Ogbourne S & Donnan K (1996) Regulation of the plasminogen activa- tor inhibitor type-2 gene in monocytes: localization of an upstream transcriptional silencer. Blood 88, 3686– 3697. 82 Ogbourne SM & Antalis TM (2001) Characterisation of PAUSE-1, a powerful silencer in the human plasmino- gen activator inhibitor type 2 gene promoter. Nucl Acids Res 29, 3919–3927. 83 Medcalf RL, Van den Berg E & Schleuning WD (1988) Glucocorticoid-modulated gene expression of tissue- and urinary-type plasminogen activator and plasminogen activator inhibitor 1 and 2. J Cell Biol 106, 971–978. 84 Medcalf RL (1992) Cell- and gene-specific interactions between signal transduction pathways revealed by oka- daic acid. Studies on the plasminogen activating system. J Biol Chem 267, 12220–12226. 85 Maurer F & Medcalf RL (1996) Plasminogen activator inhibitor type 2 gene induction by tumor necrosis factor and phorbol ester involves transcriptional and post-tran- scriptional events. Identification of a functional nona- meric AU-rich motif in the 3¢-untranslated region. J Biol Chem 271, 26074–26080. 86 Maurer F, Tierney M & Medcalf RL (1999) An AU-rich sequence in the 3¢-UTR of plasminogen activa- tor inhibitor type 2 (PAI-2) mRNA promotes PAI-2 mRNA decay and provides a binding site for nuclear HuR. Nucleic Acids Res 27, 1664–1673. 87 Yu H, Stasinopoulos S, Leedman P & Medcalf RL (2003) Inherent instability of plasminogen activator inhi- bitor type 2 mRNA is regulated by tristetraprolin. J Biol Chem 278, 13912–13918. 88 Peng SS, Chen CY, Xu N & Shyu AB (1998) RNA sta- bilization by the AU-rich element binding protein, HuR, an ELAV protein. EMBO J 17, 3461–3470. 89 Carballo E, Lai WS & Blackshear PJ (1998) Feedback inhibition of macrophage tumor necrosis factor-alpha production by tristetraprolin. Science 281, 1001–1005. 90 Stoecklin G, Ming XF, Looser R & Moroni C (2000) Somatic mRNA turnover mutants implicate tristetrapro- lin in the interleukin-3 mRNA degradation pathway. Mol Cell Biol 20, 3753–3763. 91 Tierney MJ & Medcalf RL (2001) Plasminogen activa- tor inhibitor type 2 contains mRNA instability elements within exon 4 of the coding region. Sequence homology to coding region instability determinants in other mRNAs. J Biol Chem 276, 13675–13684. R. L. Medcalf and S. J. Stasinopoulos The undecided serpin FEBS Journal 272 (2005) 4858–4867 ª 2005 FEBS 4867 . Leuven, the Netherlands. 2 Tsirka SE (20 02) Tissue plasminogen activator as a modulator of neuronal survival and function. Biochem Soc Trans 30, 22 2 22 5. 3. MINIREVIEW The undecided serpin The ins and outs of plasminogen activator inhibitor type 2 Robert L. Medcalf and Stan J. Stasinopoulos Australian