Metals affect the structure and activity of human plasminogen activator inhibitor-1. I. Modulation of stability and protease inhibition

Abstract

Human plasminogen activator inhibitor type 1 (PAI-1) is a serine protease inhibitor with a metastable active conformation. Under physiological conditions, half of the inhibitor transitions to a latent state within 1–2 h. The interaction between PAI-1 and the plasma protein vitronectin prolongs this active lifespan by ∼50%. Previously, our group demonstrated that PAI-1 binds to resins using immobilized metal affinity chromatography (Day, U.S. Pat. 7,015,021 B2, March 21, 2006). In this study, the effect of these metals on function and stability was investigated by measuring the rate of the transition from the active to latent conformation. All metals tested showed effects on stability, with the majority falling into one of two types depending on their effects. The first type of metal, which includes magnesium, calcium and manganese, invoked a slight stabilization of the active conformation of PAI-1. A second category of metals, including cobalt, nickel and copper, showed the opposite effects and a unique vitronectin-dependent modulation of PAI-1 stability. This second group of metals significantly destabilized PAI-1, although the addition of vitronectin in conjunction with these metals resulted in a marked stabilization and slower conversion to the latent conformation. In the presence of copper and vitronectin, the half-life of active PAI-1 was extended to 3 h, compared to a half-life of only ∼30 min with copper alone. Nickel had the largest effect, reducing the half-life to ∼5 min. Together, these data demonstrate a heretofore-unknown role for metals in modulating PAI-1 stability.

Introduction

Plasminogen activator inhibitor type 1 (PAI-1) is the primary regulator of the two serine proteases, tissue-type (tPA) and urokinase-type (uPA) plasminogen activator, that initiate fibrinolysis and pericellular proteolysis.^1–6 PAI-1 inhibits tPA and uPA in a suicidal fashion typical of a serine protease inhibitor (serpin).^7–9 The reactive center loop (RCL) in PAI-1 mimics the peptide substrate of the protease and is cleaved; however, the acyl intermediate formed between the protease and the RCL is long lived because of a conformational change that occurs within the serpin, effectively inactivating the protease.¹⁰ The availability of the RCL is, therefore, key to serpin activity.

PAI-1 spontaneously switches from an active to latent form when the uncleaved RCL is translocated to the interior central beta sheet in a conformational change analogous to that occurring after protease cleavage of the RCL.¹¹ The half-life (t_1/2) of the active conformation of human PAI-1 has been determined on several occasions by different groups, with widely varying results that range from ∼1.5 h to over a day. When measured at 37°C and pH 7.4, the resulting t_1/2 is typically 1–2 h, depending on buffer conditions.^12–21 Because the inhibitory function of PAI-1 is tied to the active conformation, the transition to the latent form is a mechanism for regulating the antiprotease activity and constraining it to a limited timeframe. Association with the plasma protein vitronectin increases the t_1/2 of PAI-1 by ∼50%.^{12,13,15,17,21} This is a high affinity interaction with a K_d between 0.1 and 1 nM.^22–27 Most of the active PAI-1 in plasma is bound to vitronectin.²⁸

Some among our group previously showed that recombinant wild-type PAI-1 binds to immobilized metal chromatography (IMAC) columns that have been charged with transition metals.²⁹ However, beyond use as a convenient purification strategy, the relevance of these interactions has not been investigated. In this study, the effect of the period 2 and 3 alkaline earth metals (magnesium and calcium) and several period 4 transition metals (manganese, iron, cobalt, nickel, and copper) on the stability of PAI-1 has been evaluated. Tests for the effects of metals in combination with vitronectin on PAI-1 activity were also pursued. These results point to a novel and unanticipated role for transition metals in regulation of PAI-1 structure and function.

Results

Relative affinity of PAI-1 for immobilized transition metals

In work that has been previously disclosed for a US patent,²⁹ we demonstrated that the stable 14-1b mutant form of PAI-1 binds to a variety of metals on IMAC columns that can be exploited for purification of human PAI-1. A sample elution profile of the stable 14-1b mutant PAI-1 purification on a nickel-charged HiTrap™ chelating HP column is shown in Figure 1, Panel A. Bound PAI-1 elutes from the column upon application of an imidazole gradient, detected as a relatively pure protein in fractions C and D analyzed by SDS-PAGE, as shown in the inset. Adequate purification, with only minor contaminating bands, was also observed on columns charged with Zn²⁺, Cu²⁺, and Co²⁺, whereas PAI-1 binding to a Mn²⁺-charged column was incomplete. Figure 1, Panel B, shows the imidazole elution range for PAI-1 binding to the various immobilized metals, with Ni²⁺ exhibiting the most binding of PAI-1 and requiring higher imidazole concentrations for elution. Thus, using the IMAC approach, more PAI-1 from crude lysates binds to the Group 9–12 metals (cobalt, nickel, copper, and zinc) than the Group 7 metal, manganese. These results suggest tighter binding to the transition metals.

Figure. 1.

PAI-1 binding to immobilized transition metals in IMAC. In panel A, the stable PAI-1 mutant 14-1b³⁰ from an overexpressed E. coli lysate was loaded onto a HiTrap™ chelating HP column charged with nickel, washed with 50 mM NaH₂PO₄ and 100 mM NaCl pH 6.6, and then eluted with a linear imidazole gradient. Absorbance at 280 nm was monitored during the wash and gradient showing elution of two peaks. Samples of these peaks were collected at points A through D in the chromatogram, and the fractions were separated by SDS-PAGE (inset) in the lanes labeled accordingly. The gel indicated that PAI-1 eluted in the second peak (fractions C and D). A similar approach was taken with manganese, cobalt, copper, and zinc immobilized to the matrix. Panel B shows the concentration range of imidazole needed to elute PAI-1 from the given metal resulting in relative affinity information. It should be noted that the majority of the PAI-1 protein did not bind to the column charged with manganese.

Metals affect the stability of PAI-1, both in the absence and presence of vitronectin

Because of the coordination chemistry of the IMAC resin, protein binding to the metal ligand occurs via contributions from a small number of amino acid functional groups in the protein that contribute to coordination of the metal; thus, it is not clear from IMAC alone whether an intrinsic metal binding site with a full coordination sphere is present in a bound protein. Thus, although the IMAC results suggested specific binding, the presence of an intrinsic metal-binding site in PAI-1 required additional investigation. First, to further evaluate metal effects on the structural and functional properties of PAI-1, activity assays were pursued. The effect of metals on the half-life for conversion of PAI-1 from the active to latent conformation (t_1/2) in isolation or in the presence of added vitronectin was measured. Active PAI-1 was quantified by measuring the level to which PAI-1 inhibits tPA as a function of time; over the time course, as PAI-1 relaxes to the latent conformation, less inhibition of tPA is observed. To embark on studies to evaluate the effects of metals on PAI-1 function and stability, we first established that metals did not interfere with the protease activity of tPA (data not shown). Furthermore, because metal chloride was used as the metal source and chloride has been shown to affect PAI-1 stability,¹¹ NaCl was used as a chloride control. For all of the metals tested, the kinetic data fit well to a single exponential [Fig. 2(a,b)], directly yielding k_lat, the rate constant for the conversion of active PAI-1 to the latent conformation. Half-lives calculated from k_lat for active PAI-1 in the presence and absence of vitronectin are shown in Figure 2(c) and summarized in Table I. Experiments with copper required a different buffer without BSA to avoid background sequestration of copper via the Biuret reaction (see Methods). Data testing effects of copper on PAI-1 activity in the presence and absence of vitronectin are given in Table II.

Figure. 2.

Kinetic assay showing modulation of PAI-1 stability by metals and vitronectin. Panel A is a semi-log plot showing a single exponential fit to the loss of PAI-1 inhibitory activity in the presence of MgCl₂ or NiCl₂ compared to PAI-1 alone. Panel B shows a semi-log plot with a similar fit; however, in this case PAI-1 plus vitronectin and MgCl₂ or NiCl₂ is compared to PAI-1 plus vitronectin without addition of metals. Panel C gives the t_1/2 for PAI-1 with or without vitronectin in the presence and absence of all metals used in the initial screen (data from Table I). All kinetic data were determined in buffer containing 1% BSA and represent the mean and standard error for three separate experiments (see Methods).

Table I.

Metal Effects on the Stability of PAI-1 in the Presence and Absence of Vitronectin

Additivea	VN	klat (s−1)	t1/2 (h)	Fold stabilization with VN
None	−	1.69 (± 0.04) × 10−4	1.14 (± 0.03)
	+	1.27 (± 0.02) × 10−4	1.51 (± 0.03)	1.33 (± 0.04)
NaCl	−	1.71 (± 0.05) × 10−4	1.12 (± 0.03)
	+	1.11 (± 0.04) × 10−4	1.73 (± 0.05)	1.54 (± 0.06)
MgCl2	−	1.45 (± 0.04) × 10−4	1.33 (± 0.04)b
	+	1.14 (± 0.02) × 10−4	1.69 (± 0.03)	1.35 (± 0.04)
CaCl2	−	1.55 (± 0.06) × 10−4	1.24 (± 0.05)b
	+	1.17 (± 0.02) × 10−4	1.64 (± 0.03)	1.32 (± 0.03)
MnCl2	−	1.58 (± 0.01) × 10−4	1.22 (± 0.01)b
	+	1.04 (± 0.02) × 10−4	1.85 (± 0.03)	1.52 (± 0.03)
CoCl2	−	13.9 (± 0.1) × 10−4	0.138 (± 0.001)b
	+	0.63 (± 0.02) × 10−4	3.0 (± 0.1)c	22.0 (± 0.9)
NiCl2	−	20 (± 1) × 10−4	0.095 (± 0.07)b
	+	0.497 (± 0.004) × 10−4	3.87 (± 0.03)c	40.8 (± 3.0)

^aMetals were included at a 5 mM concentration.

^bSignificantly different from associated PAI-1 chloride control (P < 0.05).

^cSignificantly different from associated PAI-1 + VN chloride control (P < 0.05).

Table II.

Copper Effects on the Stability of PAI-1 in the Presence of the VN and SMB Domain

Additive	Cofactor	klat (s−1)	t1/2 (h)	Fold stabilization
NaCl	None	2.28 (± 0.08) × 10−4	0.85 (± 0.03)	NA
	VN	1.33 (± 0.04) × 10−4	1.45 (± 0.04)	1.71 (± 0.08)
	SMB	1.18 (± 0.03) × 10−4	1.64 (± 0.04)	1.94 (± 0.08)
CuCl2	None	3.77 (± 0.14) × 10−4	0.51 (± 0.02)a	NA
	VN	0.63 (± 0.06) × 10−4	3.08 (± 0.27)b	6.03 (± 0.58)
	SMB	0.74 (± 0.05) × 10−4	2.62 (± 0.17)c	5.13 (± 0.39)

^aSignificantly different from PAI-1 chloride control (P < 0.05).

^bSignificantly different from PAI-1 + VN chloride control (P < 0.05).

^cSignificantly different from PAI-1 + SMB chloride control (P < 0.05).

All of the metals tested influence PAI-1 stability, although some effects were modest while others are quite substantial (Tables I and II and Fig. 2). These metal effects fall primarily into one of two types. Type I includes magnesium, calcium, and manganese. When compared to the t_1/2 of PAI-1 controls measured in NaCl, each of these Type I metals has a slight stabilizing effect on the active form of PAI-1 but provides no additional stabilizing effect on the t_1/2 of PAI-1 when it is bound to vitronectin. Metals grouped as Type II include cobalt, nickel, and copper. Each of the Type II metals invokes a destabilizing effect on PAI-1 and also shows a marked stabilizing effect on the t_1/2 of the PAI-1/vitronectin complex. Thus, the Type II metals confer a vitronectin-dependent modulation of PAI-1 stability. Nickel shows the largest modulation, with a 40-fold difference in the t_1/2 observed in the absence and presence of vitronectin. The t_1/2 of the complex with copper is six times longer than that observed for isolated PAI-1 with copper, and it is two times longer than the PAI-1/vitronectin control with sodium chloride added. In some cases, metals have been shown to induce protein aggregation.^31–34 To determine whether the unusual effects of metals on PAI-1 activity were simply the result of protein oligomerization, sedimentation velocity experiments using analytical ultracentrifugation were performed. From this analysis (data not shown), PAI-1 was observed to remain monomeric in the presence of added metals.

To complement the kinetic assay, direct visualization of various species formed upon mixing of PAI-1 with tPA was accomplished via a series of gel experiments to ensure that the loss of tPA inhibition measured in the kinetic assay using chromogenic substrate reflects the conversion of PAI-1 to the latent conformation. As shown in the scheme in Figure 3(a), PAI-1 and the protease bind and form an initial noncovalent complex, and attack of the protease on the serpin in the RCL gives a transient covalent intermediate (P--tPA), which is rapidly converted to the inactive acyl-enzyme (P—tPA) via a large conformational rearrangement. In some cases, water attacks the transient intermediate before the conformational change occurs, and the protease dissociates from the complex leaving cleaved PAI-1 (P_C). As PAI-1 converts to the latent form (P_L) over time, it cannot bind the protease and proceed along this pathway.

Figure. 3.

SDS-PAGE gel showing vitronectin/metal modulation of reaction products formed upon inhibition of sc-tPA by PAI-1. Panel A is a reaction scheme showing the fates of PAI-1 before and after its reaction with tPA, where P_A is active PAI-1, P_L is latent PAI-1, P·tPA is the Michaelis complex, P--tPA is the transient covalent intermediate, P—tPA is the stable complex in which tPA has been captured, and P_C is the cleaved form of PAI-1 in which hydrolysis of the transient complex is accomplished. PAI-1 alone (Panel B) or PAI-1 plus vitronectin (Panel C) were mixed with the given metal chloride and then incubated at 37°C for the time shown (0 min or 60 min). Samples were mixed with excess sc-tPA and then run under reducing conditions on 4–12% gradient SDS-PAGE.

The various species in this reaction scheme (P—tPA, P_L, and P_C) are detected in SDS gels, and Figure 3(b,c) illustrate the relative amounts of these products that form over the 60-min time period in the presence of the various metals. When assayed immediately after the addition of any metal to PAI-1, the serpin forms significant amounts of 1:1 complex [P—tPA, Fig. 3(b)]; however, the presence of Type II metals, especially copper, induces a conformation of PAI-1 that gives more substrate form [P_C, Fig. 3(b)]. This correlation extends to experiments using metals with PAI-1 in the presence of vitronectin [Fig. 3(c)]. After incubation for 60 min at 37°C, PAI-1 mixed with Type I metals retains some activity, as indicated by the presence of the 1:1 complex and/or cleavage of PAI-1 as a substrate for the protease [Fig. 3(b)]. In contrast, PAI-1 mixed with Type II metals is essentially completely converted to the unreactive latent form (P_L) within the same time frame [Fig. 3(b)]. This trend, in which there is a large increase in the amount of latent PAI-1 observed at 1 h when the Type II metals are present, agrees with the destabilization observed in the steady state kinetic assay. In the presence of vitronectin, these results with Type II metals are reversed so that PAI-1 is stabilized and more complex persists at the 60-min time point [Fig. 3(c)]. However, after the 1-h incubation at 37°C, PAI-1 mixed with the Type I metals and vitronectin exhibits significantly more latent form than PAI-1 mixed with any of the Type II metals and vitronectin [Fig. 3(c)]. These results are consistent with the demonstration from the kinetic assays that the Type II metals confer a vitronectin-dependent stabilization on PAI-1.

From these analyses, the increase in the rate of loss of PAI-1 upon addition of the Type II metals appears to occur at the same time that the latent form of the protein increases in the reaction mixture. What is not obvious from these studies is whether there is oxidation of PAI-1 that occurs upon addition of these metals that is responsible for the loss of activity. Although the possibility of oxidative damage cannot be ruled out from these experiments, it is considered unlikely from the results with added metals and vitronectin. In the case of PAI-1 in the presence of the Type II metals and vitronectin, there is a substantial stabilization of the protein so that it retains activity for a longer period of time. This would not be expected if the protein incurred oxidative damage from addition of the metals.

Metal effects on PAI-1 stability when complexed with the SMB domain are comparable to effects of full-length vitronectin

A considerable body of work recently published by our group^35–37 has established a more extensive binding interface between vitronectin and PAI-1 than had previously been recognized. These studies have shown that binding sites exist in addition to the well-characterized interactions between PAI-1 the N-terminal SMB domain of vitronectin.^35–37 Therefore, we were interested to see whether the same dramatic effects on metal modulation of PAI-1 stability observed with full-length vitronectin were exhibited by the SMB domain. Furthermore, the comparison of data for the SMB domain vs. intact vitronectin was useful to ascertain whether some of these observed effects are caused by a vitronectin/metal interaction in addition to a PAI-1/metal interaction. For these reasons, the stability of the PAI-1/SMB domain complex was compared in the presence of metals using the SDS-PAGE method and the kinetic assay.

Kinetics of PAI-1 inactivation were measured using copper to evaluate the rate of conversion of PAI-1 to the latent form with and without the SMB domain. As is the case in the presence of full-length vitronectin, the data for the rate of conversion of the active form to the latent conformation of PAI-1 fit well to a single exponential equation [Fig. 4(a)] and resulted in k_lat values and half-lives similar to the corresponding values with full-length vitronectin [Table II and Fig. 4(b)]. Nonetheless, some differences are worthy of note. First, the SMB domain confers a modestly greater stabilization of active PAI-1 compared to full-length vitronectin. Second, the PAI-1/SMB complex in the presence of CuCl₂ is somewhat less stable than the PAI-1/vitronectin complex in the presence of CuCl_2. These differences are consistent with a relatively minor contribution from the more extensive PAI-1/vitronectin interaction surface that extends beyond the sequence encompassed by the SMB domain.^35–37 Aside from these nuances, data on the complexes with the SMB domain largely recapitulate the metal effects on PAI-1 observed with full-length vitronectin. These findings indicate that the majority of the stabilizing effect of vitronectin results from the intermolecular interactions with PAI-1 via the high-affinity binding site in the SMB domain. The fact that similar metal effects on PAI-1 stability are observed with vitronectin and the much smaller SMB domain suggests that metal effects are largely accounted for by binding to PAI-1, and it is less likely that vitronectin/metal interactions contribute to a great extent to the observed effects.

Figure. 4.

Kinetic assay evaluating the SMB domain in modulating metal effects on PAI-1 stability. Panel A is semi-log plot showing a single exponential fit to the loss of PAI-1 activity in the presence of the SMB domain and CuCl₂, compared to PAI-1 in the presence of the SMB domain with only NaCl added. Panel B is a comparison of the t_1/2 for PAI-1 with or without vitronectin in the presence of NaCl or CuCl₂ (data from Table II) and PAI-1 with or without the SMB domain in the presence of NaCl or CuCl₂ (data from Table II). All kinetic data represent the mean and standard error for three separate experiments.

The kinetic assay with PAI-1 and the SMB domain was accompanied by an analysis of reaction products in gels with inclusion of various metals (Fig. 5). From the results of mixing tPA and PAI-1 with the series of metals in the absence or presence of the SMB domain [Figs. 5(a) and 5(b), respectively], it is apparent that the SMB domain gives similar results as those observed with intact vitronectin [Fig. 3(c)]. In summary, PAI-1 forms a 1:1 complex with tPA to a similar extent with the SMB domain compared to full-length vitronectin for all of the metals tested [compare Figs. 3(c) and 5(b)]. This is observed at both the 0- and 60-min time points. PAI-1 mixed with Type I metals is stabilized over time to a similar, modest extent in the presence of the SMB domain. Again, most striking are the differences with the Type II metals, where the conversion to the latent conformation is dramatically influenced by the presence of vitronectin or the SMB domain. PAI-1 mixed with Type II metals is essentially completely converted to the unreactive latent form (P_L) by 60 min without the SMB domain. With Type II metals and the SMB domain, as observed with full-length vitronectin, the opposite is observed so that PAI-1 is stabilized over the same time frame, and there is a notable reduction in the amount of latent PAI-1 observed on gels at the 1-h time point. These results agree with the stabilization by the SMB domain observed in the kinetic assay.

Figure. 5.

SDS-PAGE analysis of the effects of the SMB domain compared to full-length vitronectin in modulating the inhibition of tPA by PAI-1 in the presence of metals. PAI-1 alone (Panel A) or PAI-1 plus the SMB domain (Panel B) were mixed with the given metal chloride and then incubated at 37°C for the time shown (0 min or 60 min). Samples were mixed with excess sc-tPA and then run under reducing conditions on 4–12% gradient SDS-PAGE. In Panel C, the amount of cleaved band [P_C, Fig. 3(a)] seen at the 0 min-time point for each condition was quantified and then normalized relative to that for PAI-1 with NaCl (see methods). In Panel D, the latent band [P_L, Fig. 3(a)] at the 0-min time and 60-min time points was quantified and then the 0-min time point was divided by the 60-min time point to give the fold increase (see Methods). For Panels C and D, measurements are plotted for PAI-1 alone (blue bars), PAI-1 plus vitronectin (magenta bars), and PAI-1 plus the SMB domain (green bars). The data for PAI-1 plus vitronectin used in these gel assays are shown in Figure 3(b). All values shown in Panels C and D represent the average of measurements from three separate gels with error bars showing the standard deviation.

The results using the SDS-PAGE assay, as quantified in Figures 3 and 5(c,d), offer a more direct comparison of reaction products after mixing PAI-1 and tPA at 0- and 60 min. Figure 5(c) shows the relative amounts of the substrate form of PAI-1, apparent as the cleaved product (P_C), which is formed immediately after mixing of PAI-1 with the various metals. Gels run on samples at the 0-time point reveal effects of metals on the conformation of PAI-1 that are not apparent from the solution kinetic assays using chromogenic substrates. From this analysis, it is apparent that the Type II transition metals favor partitioning of the reaction pathway so that more of the substrate conformation forms, regardless of whether vitronectin or the SMB domain is present.

Figure 5(d) documents the accumulation of latent PAI-1 at the 1 h time point relative to the amount initially present in the reaction mixture for all metals in the absence and presence of full-length vitronectin or the SMB domain. With Type I metals, the relative amount of latent PAI-1 at 60 min increases as expected over time and is essentially the same with either the SMB domain or vitronectin present. The amount of latent PAI-1 also increases at 60 min, without added vitronectin or the SMB domain with Type I metals. Although the absolute amounts of P_L at 1 h are fairly comparable under all three conditions, the amounts of P_L with vitronectin or the SMB domain are increased in a relative sense because there is less P_L at the 0-time point with Type I metals and vitronectin or the SMB domain. Once again, most striking in Figure 5(d) are the effects of Type II metals, where the difference between the relative amount of P_L that accumulates in 60 min is markedly reduced in the presence of the SMB domain or vitronectin, as observed in the solution kinetic assays. There are some minor differences in the relative amounts of cleaved [Fig. 5(c)] and latent PAI-1 [Fig. 5(d)] that form in the presence of the Type II metals comparing the SMB domain and vitronectin. Such observations parallel differences comparing the SMB domain and full-length vitronectin observed with the solution-based kinetic assays. For example, in the presence of cobalt, more of the cleaved form of PAI-1 is detected with vitronectin than with the SMB domain [Fig. 5(c)]. Overall the gels substantiate what is observed in the solution-based kinetics and provide additional details regarding effects of metals on PAI-1 structure. These opposing effects of Type I and Type II metals on PAI-1 conformation are further evaluated in the companion paper.³⁸

Discussion

This work provides the first evidence for distinct effects of metals on the structure and function of PAI-1. Profound consequences of metal binding are invoked on the structural transition of PAI-1 from the active to latent conformation, and the effects of vitronectin as a cofactor that stabilizes PAI-1 are dependent on the presence of metals. Metals have two kinds of effects on PAI-1. Type I metals invoke a 10–20% increase in the stability of PAI-1, so that the conversion of the active to latent conformation occurs more slowly. Type I metals have no apparent enhancement of PAI-1 half-life when it is complexed with vitronectin. More substantial effects are observed with the Type II metals, where marked destabilization of PAI-1 occurs and the conversion to the latent conformation occurs within minutes. In these cases, vitronectin reverses the situation so that the PAI-1/vitronectin complexes in the presence of Type II metals are stabilized to a much greater extent than is observed in the absence of added metal.

What properties characterize Type I and II metals?

Metal binding to proteins is a function of size, polarizability, and hybridization geometry. These considerations of the physical and chemical properties of metals have been considered a biological context using a survey of X-ray structures.³⁹ From this analysis, classes of metal-binding sites in proteins were categorized according to whether the metals are oxygen-seeking versus nitrogen/sulfur seeking. Oxygen is smaller and less polarizable than nitrogen or sulfur.³⁹ Such class designations describe the tendency of a metal/ligand complex to have either ionic bond (class A) or partially covalent bond (class B) character.³⁹ The Type I metals Ca²⁺ and Mg²⁺ are both class A, whereas Mn²⁺ is in the Borderline class between A and B. The Type II metals Ni²⁺, Co²⁺, and Cu²⁺ also are all within the Borderline category, but they tend to be more class B-like.³⁹ Also, although a single metal often exhibits various geometries, each has preferred coordination states. For magnesium and manganese, the geometry tends to be octahedral. In contrast, the coordination geometry of the Type II metals is almost always tetrahedral. In the case of IMAC, the metal ligand is bound by functional groups both on the column resin and the protein; however, an intrinsic metal-binding site with a full complement of coordinating centers appears to be present within PAI-1. It is thus unclear whether residues contributing to binding in IMAC also comprise the intrinsic metal-binding site. Differences in polarizability of the two types of metals, their oxygen-seeking or nitrogen-seeking properties, and their coordination geometries suggest that two different metal binding sites could exist in PAI-1. It should be noted, however, that there are cases in which a single metal-binding site on a protein can accommodate different geometries and can provide different side-chain or backbone atoms to accommodate different metals at the same site. For example, crystallographic databases show that magnesium and manganese can bind to the same site, although manganese can tolerate nitrogen ligands better than magnesium.⁴⁰ The question of whether there are two physically distinct metal-binding sites on PAI-1, or a common site with multiple binding modes must await further study and is addressed in part in the companion study.³⁸

What are biological roles for metals and PAI-1?

The role of metals in biology is an intensively studied area that spans across all domains of life. Elegant systems for metal homeostasis have developed because of the need to regulate appropriate bioavailable metal stores and to avoid the harmful oxidative effects that can arise at elevated concentrations. Work over the years has demonstrated a role for metals in the severity of the inflammatory response, tumor progression, diabetes, atherosclerosis, and neurological disease. In each of these pathologies, PAI-1 and/or vitronectin have been implicated. Interestingly, the most prominent impacts of vitronectin on stabilizing PAI-1 occur with metals that have growing importance in a broad range of physiological settings.^41–43

A relevant interaction in vivo must be defined by a binding affinity that falls within a physiological range in terms of metal bioavailability. Thus, depending on local concentrations and affinities, these metals may play key roles in modulating the regulatory role of PAI-1 and its cofactor, vitronectin. Because this venture into metal effects on PAI-1 was unprecedented so that affinities were not known, high concentrations of metal were used in these kinetic protocols to ensure saturation conditions. Although the levels in the kinetic assays exceed physiologically relevant metal concentrations,⁴⁴ the concentration of free metal is substantially less than the total of 5 mM used due to significant metal:buffer interactions. For example, the free concentration in these assays using copper, the transition metal that is the most relevant for physiology as the third most abundant trace metal in the body, is likely less than 4 μM. A companion study was pursued with the goal of measuring the affinity of PAI-1 for metals and characterizing mechanistic aspects of conformational changes and metal binding.³⁸ This additional work shows that PAI-1 affinity for these metals corresponds to concentrations of bioavailable metal stores in vivo.

Do metals affect the conformation of PAI-1 to adopt varied biological roles?

With the understanding about PAI-1 structure that has come from structural studies using X-ray crystallography has come an appreciation of conformational transitions in its mechanism for inactivating proteases.^{11,27,30,45–48} Changes in serpin conformation that occur upon targeting proteases, binding and entrapment via a long-lived acyl intermediate have been intensely studied.^49–51 In particular, investigations into serpins of all kinds, including antithrombin-III (the main inhibitor of coagulation proteases) and the prototype in the family, anti-trypsin, have focused on the way in which RCL flexibility drives the conformational transitions key to this elegant mechanism.^52–58 Clearly, metals have a direct affect on RCL conformation, as the Type II transition metals promote the conversion of active PAI-1 to its latent conformation via RCL rearrangement. Interestingly, vitronectin tempers this conformational relaxation in PAI-1 so that the RCL is maintained in its surface exposed and active orientation for a much longer period of time.

In addition to its role in stabilizing PAI-1, an important role for vitronectin appears to be localizing PAI-1 to the site where it is needed. Consistently, we have shown that PAI-1 is targeted to the fibrin matrix of blood clots exclusively via vitronectin.⁵⁹ In addition to maintaining the delicate balance required between coagulation and thrombolysis, PAI-1 regulates the plasminogen activation system when it operates in pericellular proteolysis that affects tissue remodeling, wound healing, and cancer progression. As such, immunological approaches have identified vitronectin as the major PAI-1-binding protein in the ECM of cultured endothelial cells.⁶⁰ The co-localization of the two proteins in vivo is well documented.^61–65 Appropriate control of PAI-1 function is vital not only to hemostasis but also for regulation of many activities in tissues, including inflammation, neurological activity, angiogenesis, and tumor growth.^66–68 The ability of metals to modulate PAI-1 activity and conformation reveals a mechanism by which PAI-1 may be regulated beyond transcription and opens important new lines of investigation.

Materials and Methods

Materials

PAI-1 cloned into the expression vector pET 24d and the PAI-1 14-1b mutant²⁰ were a gift from Grant Blouse, Henry Ford Health Sciences Center Detroit, MI. Single chain (sc-tPA), two chain tPA (tc-tPA), and the mouse antihuman vitronectin 1E934 antibody were purchased from Molecular Innovations Inc. Novi, MI. Spectrozyme tPA was purchased from American Diagnostics Inc. Stamford, CT. Protease inhibitor cocktail P8465 was purchased from Sigma Aldrich Corp. St. Louis, MO. SP Sepharose Fast Flow FF, chelating Sepharose FF, DEAE Sephacel, blue Sepharose, heparin Sepharose, high-resolution Sephacryl S100/S200, and NHS-activated Sepharose FF were purchased from GE Healthcare Piscataway, NJ. The Pierce BCA assay kit was purchased from Thermo Scientific Rockford, IL. Rosetta2(DE3)pLysS and Rosetta-Gami2(DE3)pLysS competent cells, the pET 32b expression vector, and restriction grade thrombin were purchased from EMD Biosciences Gibbstown, NJ. The cloning vector pCR2.1 was purchased from Invitrogen Inc Carlsbad, CA. The QuikChange XL II kit for mutagenesis was purchased from Stratagene Inc Cedar Creek, TX. All other reagents were of analytical grade and used without further purification.

PAI-1 expression and purification

Expression of recombinant PAI-1 used the pET24d vector containing PAI-1 in the E. coli Rosetta2(DE3)pLysS strain grown in TB media containing 50 μg/mL kanamycin and 34 μg/mL chloramphenicol. Expression of PAI-1 was induced by adding 1 mM IPTG, and PAI-1 was purified using ion exchange on SP-Sepharose, IMAC on nickel charged resin, and gel filtration on Sephacryl S-100 resin. Protein identity was confirmed by western blot and MALDI-MS. PAI-1 concentration was determined at 280 nm using ɛ₂₈₀ = 0.93 mL mg⁻¹ cm⁻¹,⁶⁹ and a molecular weight = 43760 g/mol.

Chromatography of PAI-1 14-1b using immobilized transition metals for IMAC

Lysates of E. coli overexpressing the stable 14-1b mutant form of PAI-1³⁰ were passed over 1 mL HiTrap™ chelating HP columns charged with Ni²⁺, Co²⁺, Cu²⁺, Zn²⁺, or Mn²⁺, washed with 50 mM NaH₂PO₄ and 100 mM NaCl pH 6.6, and then eluted with a 10 column volume linear imidazole gradient.

Preparation of human vitronectin proteins

Vitronectin was purified from ∼3 L of fresh frozen human plasma as described previously^70–72 with minor modifications. Vitronectin concentration was determined at 280 nm using ɛ₂₈₀ = 1.0 mL mg⁻¹ cm⁻¹,⁷³ and a molecular weight = 62,000 g/mol.⁷⁴ The sequence encoding the first 47 amino acids of vitronectin (the SMB domain and the RGD sequence) was cloned into the pET32b vector as a thioredoxin fusion protein using standard protocols for PCR amplification and subcloning. The Stratagene QuikChange ® XL II protocol was used to introduce the stop codon within the coding sequence for vitronectin and also to change the enterokinase cleavage site on the pET32b vector to a thrombin cleavage site. The resulting plasmid was transformed into Rosetta-gami 2(DE3)pLysS cells and selected by growth on LB agar containing 50 μg/mL ampicilin, 34 μg/mL chloramphenicol, and 12.5 μg/mL tetracycline. Cells were resuspended in 50 mM NaH₂PO₄, 500 mM NaCl, and 20 mM imidazole pH 7.0 and lysed by sonication in an ice bath. Cell debris was removed via centrifugation. The lysate was loaded onto a nickel Sepharose FF column, and proteins were eluted with a 20–500 mM imidazole gradient. Fractions containing the thioredoxin-SMB fusion protein were pooled and dialyzed against 20 mM Tris, 150 mM NaCl, and 2.5 mM CaCl₂. A total of 15 U of thrombin was added to ∼400 mg of fusion protein and incubated with shaking at room temperature overnight. After cleavage, the N-terminus of this recombinant SMB domain construct has the four amino acid extension: GSAM. The protein was again passed over the nickel column, and the flow through was captured, concentrated, and chromatographed on a 580 mL S-100 gel filtration column equilibrated in 50 mM NaH₂PO₄, 300 mM NaCl, and 1 mM EDTA.

Correctly folded SMB domain was isolated from the mixture via affinity chromatography on a column coupled with the 14-1b mutant of PAI-1, an inherently stable form of PAI-1 that contains four mutations that prevent it from converting to the latent form.³⁰ Aliquots of the SMB domain purified by S-100 chromatography were added to the affinity column, followed by washing with 50 mM NaH₂PO₄, 300 mM NaCl, and 1 mM EDTA pH 7.4. Bound SMB domain was eluted from the affinity column with the same buffer at pH 4.0. RP-HPLC showed one peak for the affinity purified SMB domain, and protein identity was confirmed by MALDI-MS. Activity was established by an assay that used surface plasmon resonance (SPR). In this assay, using a mixture of the SMB domain and PAI-1 at 1:1 stoichiometry (25 nM:25 nM), the SMB domain gave complete inhibition of PAI-1 binding to a vitronectin-coated SPR chip. The SMB domain concentration was determined at 276 nm using ɛ₂₇₆ = 4500 M⁻¹ cm⁻¹,⁷⁵ and a molecular weight = 5678.26 g/mol (calculated from the amino acid sequence including four disulfide bonds).

PAI-1 anti-protease activity assay

The inhibitory activity of the PAI-1 preparation was tested by titration with tc-tPA. The relative amount of functional tPA was determined using a chromogenic assay by addition of 1 mM Spectrozyme tPA.¹³ The release of free p-nitroaniline upon cleavage of the substrate by tPA was monitored at 405 nM. PAI-1 showed complete inhibition of tPA function between 1.0 and 1.25 equivalents, consistent with essentially fully active inhibitor. The activity of PAI-1 was routinely checked by SDS-PAGE to measure complexed, cleaved, or latent PAI-1 form upon addition of either tPA or uPA. From these analyses (described in detail below), it was observed that typical PAI-1 preparations were ≥85% active, consistent with the chromogenic assays.

PAI-1 stability assay

The rate of conversion of PAI-1 to the latent form was determined as previously described¹³ with the following alterations. A sample of 100 nM PAI-1 in the absence of presence of 150 nM human vitronectin or the SMB domain was equilibrated in 100 mM Tris-HCl, 1% BSA, and 1 mM EDTA pH 7.4 at 37°C. Metal chloride was added to a final concentration of 5 mM metal and 10 mM chloride. In the case of NaCl, final concentrations were 10 mM sodium and chloride. Over time, aliquots were mixed with tc-tPA at a final concentration of 50 nM PAI-1 and 60 nM tPA. The relative amount of active PAI-1 was determined by its ability to inhibit functional tPA using 1 mM Spectrozyme tPA as a substrate and monitoring the release of p-nitroaniline by absorbance at 405 nM. Relative PAI-1 activity was normalized and fit to a single exponential to determine k_lat, the rate constant for conversion to the latent form. Data were collected for ∼10 PAI-1 half-lives. All data are the average of three separate experiments.

Experiments in the presence of copper required alterations to the above protocol. BSA was excluded from the buffer to avoid copper (I) formation via the Biuret reaction. Control experiments were repeated in the absence of BSA for comparison. Also, the initial copper solution (10 mL of 10 mM CuCl₂) was titrated with 35 μL of 6M NaOH (20 mM total) to restore the pH to 7.4.

The formation of the tPA/PAI-1 complex, as well as unreactive/latent and cleaved PAI-1 in the presence or absence of metals, was visualized by reducing SDS-PAGE. Because of overlapping migration of BSA with vitronectin in the gels, BSA was excluded from the buffer; the final buffer components were 100 mM Tris-HCl and 1 mM EDTA pH 7.4 at 37°C. A concentrated metal stock was added to a sample of PAI-1 alone, PAI-1 mixed with human vitronectin or PAI-1 mixed with the SMB domain, in a final volume of 55 μL and final concentration of 5 mM metal chloride (10 mM NaCl), 4 μM PAI-1, and 6 μM vitronectin or SMB. The sample was then either mixed immediately with excess sc-tPA (i.e. 6.6 μL of 47 μM) or incubated at 37°C for 60 min and then mixed with the sc-tPA. Note that single chain tPA was used in the gel assays, in spite of its relatively lower activity compared to two-chain tPA, because the single chain form is more easily separated from vitronectin compared to tc-tPA in SDS-PAGE. After protease addition, the samples were incubated for 10 min at room temperature, and the reaction was stopped by the addition of 20 μL of reducing SDS-PAGE loading dye. Samples were boiled for 5 min, and then 12 μL was electrophoresed on a 4–12% gradient gel.

Gel bands for the cleaved and latent forms of PAI-1 were quantified using Bio-Rad Quantity One™ software with digital images of gels recorded on a Biorad ChemiDoc XRS photodocumentation system. Visible bands and a region corresponding to the gel background were delineated using the selection tool. The software was used to normalize the intensity of the bands relative to the intensity of the selected background, yielding the adjusted-intensity-volume. Results from quantification of the cleaved band at the 0-min time point were normalized based on a value of 1 for the control with PAI-1 and NaCl. Data from the latent PAI-1 band at both 0-min and 60-min time points were analyzed to determine a relative increase over this time period by dividing the adjusted-intensity-volume measurements for the 0-min time point by that for the 60-min time point. Only cleaved and latent bands were analyzed because it was not possible to quantify directly the amount of PAI-1-tPA complex with this approach due to the diffuse migration of this species on SDS-PAGE. This is due to the heterogeneity of tPA resulting from variable glycosylation, which also is apparent for the migration of the isolated sc-tPA protein in the electrophoretic field. Therefore, to establish the amount of active PAI-1 in a given sample, the amount of cleaved and unreacted/latent bands on the gel upon the reaction with tPA was determined by densitometry and compared to the amount of PAI-1 in a control sample on the same gel without tPA. The difference in the total amount of PAI-1 detected as unreacted or cleaved protein comparing the sample +/− tPA was attributed to the amount of active PAI-1 that formed a complex with tPA. This amount was typically ∼85% of the total.

Glossary

Abbreviations:

BCA

bicinchoninic acid

BSA

bovine serum albumin

DEAE

diethylaminoethyl

DTNB

dithiobis-2-nitrobenzoic acid

EDTA

ethylenediaminetetraacetic acid

HPLC

high pressure liquid chromatography

IMAC

immobilized metal affinity chromatography

IPTG

isopropyl-(-d-thiogalactopyranoside

LB

Luria broth

MALDI-MS

matrix assisted laser desorption ionization mass spectrometry

NHS

N-hydroxysuccinimide

PAI-1

plasminogen activator inhibitor type 1

PBS

phosphate buffered saline

PCR

polymerase chain reaction

RCL

reactive center loop

RP-HPLC

reverse phase HPLC

sc-tPA

single chain tPA

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

serpin

serine protease inhibitor

SMB

somatomedin B

SPR

surface plasmon resonance

t1/2

half-life

TB

terrific broth

tc-tPA

two chain tPA

tPA

tissue-type plasminogen activator

Tris

tris(hydroxymethyl)aminomethane

uPA

urokinase-type plasminogen activator.