SEARCH

SEARCH BY CITATION

Introduction

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

This manuscript is coauthored by M. Nesheim and L. Bajzar. It represents an attempt to relate the history of the discovery and characterization of TAFI (thrombin activatable fibrinolysis inhibitor) as it occurred in our laboratory. M.N. provided the prose and L.B. provided the vast majority of the work that led to the discovery. He also provided a check on the memory of M.N., in the hope that the following narrative is as accurate as possible.

The effect of activated protein C on fibrinolysis

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

Our studies that ultimately lead to the isolation and characterization of TAFI were initially motivated by observations of others in experiments carried out about 20 years ago. In 1985 Fletcher Taylor and Mickey Lockhart reported an experiment in which they added radiolabeled fibrinogen to whole blood and placed aliquots in a series of microcentrifuge tubes that contained a little thrombin [1]. Similar series of aliquots were also prepared with activated protein C (APC) included at several concentrations. At regular intervals over the next few hours, tubes were selected from each series and centrifuged to pellet the clots. Aliquots of the sera were then subjected to gamma scintillation counting to determine how much fibrin had been solubilized. As expected, no fibrinolysis was observed over time in the controls without APC. With APC, however, spontaneous lysis was observed, and both the rate and extent of lysis increased with increasing concentrations of APC. A year later, de Fouw and coworkers reported the results of similar experiments in which tissue plasminogen activator was included so that lysis occurred in the controls without APC [2]. Again, with APC, the rate and extent of lysis were both increased.

These experiments were very interesting. They seemed to imply that some sort of connection exists between coagulation, anticoagulation, and fibrinolysis. I began to wonder about the molecular basis for this remarkable effect and to think about means of finding it. Since the effect was demonstrated in whole blood, the possibilities were legion. The effect might be mediated via cells, for example. Alternatively, APC, being an enzyme, might activate or modify some other as yet unknown species. It could either activate the precursor of a fibrinolytic stimulator or eliminate an inhibitor. In view of so many possibilities, the task of identifying the mechanism indeed appeared daunting. Nonetheless, drawn to the problem as by the Sirens of Circe, initial efforts were undertaken to investigate the phenomenon.

One difficulty with the approaches described by Taylor and Lockhart or de Fouw and coworkers is that they were based on relatively few measurements taken at discrete, widely separated intervals. Consequently, the approach was not suited to quantifying accurately and precisely the rates and extents of lysis. I wondered whether a more dense set of data could be obtained by continual measurements made over time. How to do this with whole blood was not obvious, but if the phenomenon occurred in plasma, clotting and subsequent lysis could be monitored over time very simply by spectrophotometric measurements of turbidity. Thus, we carried out simple experiments in which dialyzed plasma was added to cuvettes containing a little thrombin (to induce clotting) and tissue-type plasminogen activator (t-PA) (to induce fibrinolysis) in the absence or presence of APC. The progress of clotting and subsequent lysis was monitored at closely spaced time intervals by measuring absorbance at 405 nm. A parameter-denoted lysis time (LT) was defined as the time at which lysis was 50% complete as inferred by the absorbance signal. These studies showed that APC promotes lysis in a cell-free system and that the effect requires Ca2+. In addition, the continuous measurement of turbidity appeared to be viable as a means of accurately and precisely monitoring and quantifying the effects.

About this time, I was in my office one day contemplating something or other. A young lad named Laszlo Bajzar entered and introduced himself. He said that he was visiting the biochemistry department at Queen's in search of a thesis advisor for his graduate studies. I was about to light up a cigarette (you could do that in your office in those days). He then asked me for a cigarette and we lit up together. I made a little mental note to myself, I like something about this guy. He asked whether I was interested in a graduate student. I said that I was. He enquired as to which projects might be available. I pointed out that we had work ongoing in prothrombin activation, especially with respect to factor Va, other studies involving heparin, and still others involving the reactions of the fibrinolytic cascade. I described them in some detail and pointed out that each was reasonably straightforward and would yield very useful information and insights. I also pointed out that we had this APC effect on fibrinolysis for which nothing was known regarding the mechanism. Pursuit of this also would comprise a PhD thesis project, but it was very risky. It would almost certainly require the isolation and characterization of at least one component, perhaps several. Because we had so little information at the time, no estimate could be made as to whether sufficient progress could be made to justify a PhD degree in a reasonable amount of time. I pointed out that the first series of projects would be much more predictable than a project involving the APC effect on fibrinolysis. I indicated to Laszlo that he should consider all these things carefully before picking a project. He indicated without hesitation that he wanted the APC project, regardless of its uncertainty because, as I recall, he wanted the puzzle and the challenge. Laszlo's recollection is somewhat different from my own; being perhaps clouded by vibrancy of youth, he chooses to believe instead that only one project option was legitimately presented as a potential thesis project and that my cautionary advice was somewhat absent. Nonetheless, in 1988 Laszlo began his thesis research into the molecular basis behind the observation that APC appears to promote fibrinolysis.

The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

Laszlo began by refining the lysis assay. The measurement was adapted to a newly acquired, temperature-controlled, microtiter plate reader that could monitor many samples at once. Since each sample might take 2 h or more to lyze, this was critical to obtaining sufficient data in a timely way. Conditions for the assay were carefully specified so that results would be quantitatively reproducible. Plasma was dialyzed against buffered saline to replace the citrate anticoagulant and to provide a reproducible pH and ionic strength. All assays involved an overall 1/3 dilution of plasma (A280 = 16.0). The response of the lysis time to the concentration of t-PA was determined to provide some indication as to how the time to lyze was related to the ‘fibrinolytic potential’. Laszlo soon found out that APC would reduce the lysis time of normal plasma, when Ca2+ was present, in a fashion that was saturable with respect to the concentration of APC. In the absence of Ca2+, APC had no effect. In addition, the lysis time in the absence of Ca2+ was relatively short, comparable to that obtained with Ca2+ at a saturating level of APC. This, in turn, suggested that one or more of the other vitamin K-dependent proteins were involved. Thus, barium citrate-adsorbed, dialyzed plasma (BAP) was prepared and studied. With it, lysis times were short in both the absence and presence of Ca2+, and APC had no effect. The materials removed by barium citrate adsorption were reconstituted and fractionated by making 0–5% (fraction A) and 5–40% (fraction B) cuts with polyethylene glycol (PEG) 8000. Neither of these alone, when added back to BAP, would restore the APC effect on fibrinolysis. That is, all samples showed a short lysis time, regardless of the presence or absence of APC or Ca2+. Both fractions added together, however, restored the observations made with normal plasma. In the presence of Ca2+, the lysis time was long in the absence of APC. In the presence of APC at a saturating level, it was short and at a value equal to that obtained in the absence of Ca2+, or in BAP, with or without Ca2+.

The effect of APC could also be reconstituted in a system in which the plasma was replaced with purified fibrinogen plus plasminogen. In addition, in this system plasmin was shown to accumulate faster and to a greater extent in the presence of APC. From these results, the conclusions were reached that the effect of APC on t-PA-mediated fibrinolysis is a Ca2+-dependent phenomenon, that it occurs in a cell-free system, and that at least two components that can be adsorbed from plasma with barium citrate are involved [3]. At this point Nils Bang came to Queen's University to hear about our work. It was his insightful question ‘are you sure APC shortens lysis time or does it prevent something else from prolonging it?’ that changed our way of looking at the question.

Further work was undertaken to identify the components in the PEG 8000 fractions A and B required for the APC-dependent potentiation of fibrinolysis. Fraction A was subjected to gel filtration chromatography. The activity was assessed by determining the lysis time of BAP supplemented with fraction B. The active material eluted very early and it was turbid. Its activity could be reproduced with phospholipid vesicles containing phosphatidylcholine plus phosphatidylserine (PCPS). Fraction B contained the vitamin K-dependent proteins. The components of this fraction were resolved by chromatography into separate pools of prothrombin, protein S, protein C, factor IX and factor X. Factor VII got away from us somehow. These were added back, singly and in all possible combinations, to BAP supplemented with either fraction A or PCPS, and the effect of APC on the lysis time was determined. The components both necessary and sufficient to restore the APC effect comprised a combination of prothrombin, FX, FIX and phospholipid vesicles. The effect therefore seemed to require the intact coagulation cascade or at least the intrinsic and common pathways. Thus, the combination of FXa and prothrombin was tried. This combination gave a prolonged lysis time which could be shortened with APC. The combination of prothrombin plus echarin also gave a prolonged lysis time, but this could not be shortened with APC. The effect also could be obtained with the BAP replaced with purified fibrinogen, plasminogen and factor V. In the purified system, the inclusion of APC was accompanied by complete inhibition of thrombin generation and substantially augmented plasmin generation. From these results the conclusions were reached that the apparent profibrinolytic effect is specifically due to the attenuation of prothrombin activation and that prothrombin activation in plasma generates a component or components that inhibit fibrinolysis. Thus, the existence of the thrombin activatable fibrinolysis inhibitor (TAFI) was postulated [4].

The studies utilizing purified components described above were both encouraging and confounding. The fact that an APC effect, that is, the shortening of the lysis time by APC in the presence of prothrombin and prothrombinase, could be observed when the only components presumed present were prothrombin, factor V, factor Xa, Ca2+, PCPS vesicles, fibrinogen, glu-plasminogen, and t-PA implied that the mechanism either involved some unknown property intrinsic to those components or was due to some trace contaminant brought in with one or more of the purified proteins. The magnitude of the APC effect in the system of purified components proved to vary from experiment to experiment as different batches of purified components were used. This was in contrast to the quantitative similarity that occurred from batch to batch of supplemented barium citrate-adsorbed plasma. Eventually, the variability tracked specifically to the particular batches of plasminogen being used. One inference was that the effect was due to some still unappreciated intrinsic property of plasminogen that was lost to varying degrees during its isolation. This was plausible because plasminogen is known to readily change confirmation [5]. The alternative possibility, that a contaminant was present, was frankly the desired one. Overloaded Coomassie blue-stained gels of the various plasminogen preparations showed them all to be about 98% pure, and no trace contaminant stood out to distinguish the preparations which showed large APC effects from those that showed small ones. Consequently, a sample of barium citrate-adsorbed plasma was rendered plasminogen deficient by chromatography on lysine–Sepharose. Fibrinolysis did not occur in this plasma. Aliquots were supplemented with various preparations of purified plasminogen, including one preparation which showed no APC effect, in the system of purified components. When these were tested, all plasminogen samples showed an APC effect on fibrinolysis, and their effects were quantitatively very similar. These results strongly suggested that the effect was due to a trace contaminant rather than an intrinsic property of plasminogen and thereby encouraged us to undertake efforts to isolate the putative substance.

The isolation of TAFI

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

In general, the isolation of a new substance by the classical, biochemical approach requires minimally three elements. One is an adequate supply of source material. Another is a good, reliable and preferably quantitative assay. The third is a good strategy accompanied by common sense and good, quantitative bookkeeping. A little luck does not hurt, either.

Based on our previous work, barium citrate-adsorbed plasma appeared to be an adequate source, so that criterion was satisfied. We thus proceeded headlong and immediately disregarded the second and third requirements. We decided on an assay in which a sample was added to plasminogen, fibrinogen, t-PA and Ca2+ and then lysis times were measured in the presence of low (6 nm) and high (60 nm) thrombin. The low and high thrombin levels were intended to simulate the respective absence or presence of prothrombin activation. We fully appreciated that very substantial differences in lysis times would be encountered from sample to sample, depending on the level of antiplasmin carried into the assay by the sample. This would presumably be accounted for, however, because each sample, in effect, served as its own control. Quantification, however, would be next to impossible with this approach. Nonetheless, we made a large effort to put together a fractionation scheme that included steps of salt fraction, ion exchange chromatography, size exclusion chromatography and the like.

After several months spent on this approach, we appeared to make almost no progress, in that the desired correlation between function and apparent protein homogeneity, as assessed by SDS–PAGE, was not forthcoming. Late one afternoon, Laszlo and I were sitting in front of the fume hood having a cigarette (smoking the vile weed of tobacco in the offices had been banned by that time) and discussing recent results. Laszlo had begun the particular preparation under discussion with 500 mL of plasma, a somewhat turbid yellowish batch of stuff. A sample, 20 µL, had been subjected to the assay, and the lysis time with 60 nm thrombin was longer than that with 6 nm thrombin, indicating the presence of the putative TAFI. Numerous fractionation steps had been carried out, after which the total sample consisted of 1.0 mL of a perfectly clear, colorless solution of protein. A 20-µL sample of this showed the same differential lysis time in response to 6.0 nm and 60 nm thrombin as that exhibited by the 20-µL aliquot of the starting plasma. The 1.0-mL sample had an A280 of about 40, just like the starting plasma. SDS–PAGE showed it to be very heterogeneous and indicated that we had a long way to go. This was frustrating. We decided that we had best take stock of where we had come with these efforts. Although the assay was only semiquantitative, we could make some estimate of the recovery of activity from beginning to end and compare it with recovery of total protein and thereby estimate the extent of purification and yield. Since the differential in lysis times with plasma and the latter fraction were about the same, levels of the active material in the 20-µL aliquots were about the same. Since the A280 values were about the same, the specific activities were about the same. Thus, the purification factor was about 1, meaning that no net purification had occurred. In addition, since we had started with 500 mL of sample and now had 1.0 mL, each equally active per unit volume, our overall yield was 0.2%. Although the final sample was clear and colorless, we had managed to fractionate the desired substance into oblivion without achieving any net purification.

As anyone who has made a serious effort to isolate a new substance knows, this is not very hard to do. We did know, however, that we were getting nowhere fast. This was discouraging. We had another cigarette and concluded that a return to square one was in order. We were going to have to return to the original observation by which the existence of TAFI was inferred and base our assay on it. The observation was that when prothrombin activation occurs in plasma, fibrinolysis is prolonged. We needed the equivalent of TAFI-deficient, barium citrate-adsorbed plasma. To simulate this, we devised an assay substrate consisting of fibrinogen, plasminogen, t-PA, antiplasmin, antithrombin, factor V and PCPS vesicles. The inhibitors antithrombin and antiplasmin were there to in effect ‘buffer’ those that would be added from assay samples. The assay then consisted of preparing two samples of the material to be assayed by diluting them into the assay substrate. Each was placed in a well in the microtiter plate. FXa was then added to one and buffer to the other, so that prothrombin activation occurred in one but not the other. The lysis times of both were measured and the differential in time was taken as a measure of the concentration of TAFI in the test sample. The assay was calibrated with barium citrate-adsorbed plasma. The response of the assay is shown in Fig. 1.

image

Figure 1. Standard curve for the assay of thrombin activatable fibrinolysis inhibitor (TAFI). The assay was performed by adding an aliquot of sample (in this case barium citrate-adsorbed, dialyzed plasma) to two microtiter plate wells, each containing prothrombin, factor V, plasminogen, phosphatidylcholine plus phosphatidylserine (PCPS) vesicles, Ca2+, thrombin, antiplasmin, thrombin and tissue-type plasminogen activator (t-PA). In addition, buffer was added to one of them and factor Xa to the other. Both wells were monitored for absorbance at 405 nm over time, and the lysis times were determined. Samples with factor Xa took longer to lyze than those without it. The differential in lysis time was taken as a measure of the quantity of TAFI in the sample [6].

Download figure to PowerPoint

For this assay to work, the plasminogen and antiplasmin had to be relatively free of TAFI. This was solved with respect to plasminogen by selecting preparations that showed no response in the assay, even in the absence of antiplasmin. Solving the problem with antiplasmin was a little trickier. Plasma antiplasmin is isolated by passing plasma over lysine–Sepharose to remove plasminogen, and then passing the plasmingen-deficient plasma over plasminogen–Sepharose to capture the antiplasmin, which is subsequently eluted with epsilon aminocaproic acid (EACA). This, however, yielded antiplasmin preparations that gave a positive response in the assay even without added barium citrate-adsorbed plasma. This problem could be minimized by carefully fractionating the antiplasmin, but getting adequate amounts for many assays became a serious problem. Fortunately, however, S. Busby of Zymogenetics had previously made available to us the cDNA for recombinant antiplasmin. With the help of A. Horrevoets, we were able to express it in good amounts and use it in the assay. Thus, we returned to square one and devised a fractionation scheme for TAFI. Its initial steps included barium citrate adsorption, salt cuts with ammonium sulfate, ion exchange chromatography on Q-Sepharose, and size exclusion chromatography on AcA-54. These led overall to a 29-fold purification with a 15% yield. We realized that if TAFI were a relatively low abundance protein (say 10 µg mL−1), a purification factor of several thousand-fold would be necessary, so we had a long way to go. However, I was content to use the brute force approach, if necessary, believing that the systematic approach would eventually get us there, even if a thousand liters of plasma were needed in the end. Laszlo, however, felt that the time had come to take advantage of the specificity of nature's own interactions. Thus, he wanted to work on affinity chromatography on plasminogen–Sepharose.

He prepared a column of immobilized plasminogen and subjected the material recovered after gel filtration to affinity chromatography. The assay told us that plasminogen–Sepharose captured the active material. Whether it could be recovered was another matter. He tried elution with EACA and a small blip of protein was eluted. Whether it was active could not be assessed in the assay, however, because the EACA severely inhibited the assay. Thus, the eluate was passed over a small (1.0 mL) DEAE cellulose column to capture the protein and remove the EACA. The bound protein was then eluted with NaCl. The eluted material was active in the assay and the yield from the previous step was typically in excess of 80%, and the purification factor from plasma was 14 300-fold. Thus, in a single step the relative purity increased by 500-fold, which exemplifies the beauty of affinity chromatography when it works. The purification scheme generally yielded about 190 µg of TAFI from 600 mL of plasma. The results of the affinity step are shown in Fig. 2. Fractions from the affinity step were subjected to SDS–PAGE (Fig. 3). The starting material (lane 1) and the flow through (lane 2) were not distinguishable. The three fractions eluted from the small DEAE cellulose column indicated a single band with an apparent molecular weight of about 60 000 (lanes 3–5).

image

Figure 2. Affinity chromatography of thrombin activatable fibrinolysis inhibitor (TAFI) on plasminogen–Sepharose. Partially purified TAFI was applied to plasminogen–Sepharose. After washing the column with buffer, TAFI was eluted with epsilon aminocaproic acid (EACA). These fractions were applied to a small DEAE cellulose column (inset) and active material was eluted with NaCl [6].

Download figure to PowerPoint

image

Figure 3. Electrophoretic analysis of fractions obtained after affinity chromatography on plasminogen–Sepharose. Fractions were analyzed by SDS–PAGE. Lane 1 was run with the starting material and lane 2 with the flow through. Lanes 3, 4, and 5 were run with the peak fractions 41–43 shown in the inset of Fig. 2.

Download figure to PowerPoint

A few more experiments indicated that this material was a proteolytic substrate for thrombin, which was consistent with its being activated by thrombin [6]. Partial sequence analysis of the material indicated that the protein was identical to a plasma carboxypeptidase precursor that had been described a few years earlier by Eaton et al. [7]. In those studies a relatively trace plasminogen binding protein had been isolated from plasma and its cDNA cloned. The amino acid sequence inferred from the cDNA sequence indicated that the protein, designated plasma carboxypeptidase B, was homologous to the carboxypeptidase B from the pancreas. They did not ascribe a physiological function to the protein, but did indicate that it could be activated by trypsin, thrombin or plasmin to a carboxypeptidase B-like enzyme. They also speculated that it might be involved in the regulation of fibrinolysis. This earlier work by Eaton et al. was very helpful to us in establishing the properties of TAFI because it focused our studies on its role as a carboxypeptidase B precursor and on the properties of the enzyme, which we designated TAFIa. We also realized that the protein had apparently been discovered in other laboratories. It appeared likely to be the unstable carboxypeptidase B-like activity of serum described early by Hendriks and coworkers [8] and Wang and coworkers [9], designated by them carboxypeptidase U (unstable). It was also probably identical to carboxypeptidase R (arginine) described by Campbell and Okada [10]. The protein also was isolated by Broze and Higuchi at about the same time [11]. They showed that it was related to the premature lysis observed in plasmas defective in the intrinsic pathway.

The thrombomodulin-dependence of TAFI activation

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

As soon as TAFI was isolated we began to investigate its activation by thrombin. We learned with the first experiments that, as macromolecular thrombin substrates go, TAFI was not among the good ones like fibrinogen, factor V or factor VIII. Instead, it was reminiscent of protein C in that regard, and we wondered whether TAFI activation might be promoted by thrombomodulin. At about that time, I had the good fortune of meeting J. Morser on a visit to Berlex Biosciences. He and his colleagues were working on Solulin [12], a well-characterized soluble, recombinant form of thrombomodulin. I told him about TAFI and he graciously supplied us with some Solulin. The first experiment with it showed unmistakably and dramatically that TAFI activation is indeed stimulated by thrombomodulin. This initiated a long and fruitful collaboration with J. Morser, M. Nagashima and their coworkers, that exists to this day.

A systematic investigation of the kinetics of activation of TAFI by thrombin in the presence of thrombomodulin [13] showed that thrombomodulin increases the catalytic efficiency of thrombin-catalyzed activation of TAFI by a factor of 1250, a value much like that measured in protein C activation [14]. The kinetics were consistent with an enzyme-central, parallel-assembly mechanism whereby thrombin can bind either TAFI or thrombomodulin to form binary complexes, which can then bind the remaining component to form the ternary thrombin/thrombomodulin/TAFI complex, in which TAFIa is generated. The increase in catalytic efficiency was shown to be due almost exclusively to an increase in kcat. Since thrombin, especially in the presence of thrombomodulin, can activate TAFI and since TAFIa suppresses fibrinolysis, the TAFI pathway appeared to provide a direct molecular link between the coagulation and fibrinolytic cascades such that activation of the former can downregulate the latter.

Closing the loop

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

As indicated above, our investigations that eventually led to TAFI were motivated by the observations of others that suggested that APC is somehow profibrinolytic through a mechanism unknown at the time. The properties of TAFI strongly suggested that its activation might account for the profibrinolytic effect of APC. Laszlo definitively showed that to be the case when he was a postdoctoral fellow in P. Tracy's laboratory at the University of Vermont [15]. He showed that APC had no profibrinolytic effect in TAFI-deficient plasma, and that the effect could be restored by adding back purified TAFI. In addition, he showed that a neutralizing monoclonal antibody to TAFI eliminated the effect of APC on fibrinolysis in normal plasma. Thus, in quantitative detail, these experiments showed that the apparent profibrinolytic effect of APC was due to its inhibition of prothrombin activation, which, in turn, prevented the generation of an inhibitor of fibrinolysis. This was a gratifying result.

Other biochemical work on TAFI

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

Considerable detail has been added to the story of TAFI since its isolation and characterization. Early on, P. von dem Borne, in collaboration with Laszlo, J. Meijers and B. Buoma, and me showed that activation of the factor XI-dependent pathway prolongs fibrinolysis through TAFI activation [16], an effect which had also been demonstrated by Broze and Higuchi [11]. M. Boffa in my laboratory prepared recombinant TAFI and described the spontaneous decay of TAFIa activity [17,18]. W. Wang showed that TAFI suppresses fibrinolysis by downregulating the t-PA-cofactor activity of fibrin [19]. Our collaborators at Berlex studied two naturally occurring isoforms of TAFI (A149T) and showed them to be functionally equivalent [20]. Wang and others in my laboratory, in collaboration with Nagashima and Morser, identified the minimal elements of structure of thrombomodulin needed for TAFI activation and showed them to be overlapping with, but more extensive than, those needed for protein C activation [21]. This was shown also by Kokame and coworkers at about the same time [22]. J. Walker characterized the kinetics of activation of plasminogen in the presence of high-molecular-weight soluble fibrin degradation products and showed that the cofactor activity of these is profoundly diminished by TAFIa [23]. A common polymorphism at amino acid 325 (Ile or Thr) was described in collaboration with G. Brouwers and coworkers in the Netherlands [24]. M. Schneider in my laboratory found that both isoforms are activated with identical kinetics by the thrombin–thrombomodulin complex, but that the Ile325 variant is twice as stable and 60% more potent as an antifibrinolytic, compared with the Thr325 variant [25]. Schneider also showed, along with our collaborators from Berlex, that the thrombomodulin dependence of TAFI activation is not determined by the amino acid residues surrounding the activation cleavage site [26]. Boffa and colleagues characterized the TAFI gene and studied regulation of TAFI gene expression [27–29]. Schneider showed that reversible inhibitors of TAFIa can both promote and prolong fibrinolysis, depending on conditions [30]. This observation is virtually identical to that reported about the same time by John Walker and Laszlo Bajzar [31]. M. Schneider added a fuller understanding of the means by which TAFIa suppresses fibrinolysis when he showed that TAFIa reduces the ability of plasmin-modified fibrin to protect plasmin from inhibition antiplasmin [32,33]. John Walker and Laszlo Bajzar subsequently showed that the antifibrinolytic effect attributed to TAFIa demonstrates a threshold that is critically dependent upon its ability to modulate the rate of formation and inhibition of plasmin [34]. E. Neill, R. Stewart and M. Schneider also described the development of a very sensitive and specific assay for measuring functional TAFIa in plasma [35]. Many others in laboratories around the world have made substantial contributions to our understanding of TAFI and its potential role in regulating the balance between fibrin deposition and removal. Because this article is not a review, they are not mentioned here. The many efforts and contributions of these workers, however, are acknowledged and much appreciated.

Lessons learned

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References

Some lessons were learned through the efforts to isolate and characterize TAFI. One lesson is to be quantitative, if at all possible. Quantifying lysis times indicated early on that APC was not directly profibrinolytic, but rather prevented the expression of some inhibitor of fibrinolysis. The quantitative evidence for this came from experiments showing that the time to lyze of clots formed in barium citrate-adsorbed plasma in the presence of APC was the same as that in its absence. This lysis time was also the same as that in normal plasma in the absence of calcium ion. This time was also short compared with that in normal plasma in the presence of calcium ion in the absence of APC, but it was the same as that in normal plasma with both calcium ion and APC. These quantitative relationships all suggested the existence of an inhibitor that was generated in the presence of thrombin.

Another lesson learned, or perhaps re-learned, was to avoid short cuts. Initial efforts to isolate TAFI were based on an assay in which differences in lysis times in the presence of low (6 nm) and high (60 nm) thrombin levels were observed. The measurements were easy to make, but were not quantitative. The use of these measurements put us on a path that led nowhere, over and over again. The abandonment of this approach and adoption of the more difficult, but quantitative approach whereby TAFI was measured by virtue of differences in lysis time in the presence and absence of prothrombin activation, eventually led us to TAFI.

Another lesson was to be persistent. Although the story of the isolation of TAFI can be told in a short time on a few pages, our search for it began in about 1986, but its isolation and characterization was not published until 1995, and a goodly portion of this time was spent in one way or another in the pursuit of TAFI.

A final lesson learned is that the pursuit of the molecular basis for a biochemical phenomenon can be incredibly gratifying. As seems typical in biochemical science, about 90% of the efforts spent on the problem ended in some frustration or confusion. About 10% of the time, however, a solid insight was gained that compensated many times over for the other frustrations. In the end a picture emerged that deepened one's appreciation of the fantastic beauty of nature. This is really fun.

References

  1. Top of page
  2. Introduction
  3. The effect of activated protein C on fibrinolysis
  4. The APC effect on fibrinolysis occurs in a cell-free system and requires prothrombin activation
  5. The isolation of TAFI
  6. The thrombomodulin-dependence of TAFI activation
  7. Closing the loop
  8. Other biochemical work on TAFI
  9. Lessons learned
  10. References
  • 1
    Taylor FB, Lockhart MS. Whole blood clot lysis: in vitro modulation by activated protein C. Thromb Res 1985; 37: 63949.
  • 2
    De Fouw NJ, Haverkate F, Bertina RM, Koopman J, Van Wijngaarden A, Van Hinsbergh VWM. The cofactor role of protein S in the acceleration of whole blood clot lysis by activated protein C in vitro. Blood 1986; 67: 118992.
  • 3
    Bajzar L, Fredenburgh JC, Nesheim ME. The activated protein C-mediated enhancement of tissue-type plasminogen activator-induced fibrinolysis in a cell-free system. J Biol Chem 1990; 265: 1694854.
  • 4
    Bajzar L, Nesheim M. The effect of activated protein C on fibrinolysis in cell-free plasma can be attributed specifically to attenuation of prothrombin activation. J Biol Chem 1993; 268: 860816.
  • 5
    Suenson E, Petersen LC. Fibrin and plasminogen structures essential to stimulation of plasmin formation by tissue-type plasminogen activator. Biochim Biophys Acta 1986; 870: 5109.
  • 6
    Bajzar L, Manuel R, Nesheim ME. Purification and characterization of TAFI, a thrombin activatable fibrinolysis inhibitor. J Biol Chem 1995; 270: 1447784.
  • 7
    Eaton DL, Malloy BE, Tsai SP, Henzel W, Drayna D. Isolation, molecular cloning, and partial characterization of a novel carboxypeptidase B from human plasma. J Biol Chem 1991; 266: 218338.
  • 8
    Hendriks D, Scharpe S, Van Sande M, Lommaert MP. Characterization of a carboxypeptidase in human serum distinct from carboxypeptidase N. J Clin Chem Clin Biochem 1989; 27: 27785.
  • 9
    Wang W, Hendriks DF, Scharpe SS. Carboxypeptidase U, a plasma carboxypeptidase with high affinity for plasminogen. J Biol Chem 1994; 269: 1593744.
  • 10
    Campbell W, Okada H. An arginine specific carboxypeptidase generated in blood during coagulation or inflammation which is unrelated to carboxypeptidase N or its subunits. Biochem Biophys Res Commun 1989; 162: 9339.
  • 11
    Broze GJ Jr, Higuchi DA. Coagulation-dependent inhibition of fibrinolysis. Role of carboxypeptidase-U and the premature lysis of clots from hemophilic plasma. Blood 1996; 88: 381523.
  • 12
    Glaser CB, Morser J, Clarke JH, Blasko E, McLean K, Kuhn I, Chang R-J, Lin J-H, Vilander L, Andrews WH, Light DR. Oxidation of a specific methionine in thrombomodulin by activated neutrophil products blocks cofactor activity. A potential rapid mechanism for modulation of coagulation. J Clin Invest 1992; 90: 256573.
  • 13
    Bajzar L, Morser J, Nesheim M. TAFI, or plasma procarboxypeptidase B, couples the coagulation and fibrinolytic cascades through the thrombin–thrombomodulin complex. J Biol Chem 1996; 271: 166038.
  • 14
    Le Bonniec BF, Guinto ER, MacGillivray RTA, Stone SR, Esmon CT. The role of thrombin's Tyr-Pro-Pro-Trp motif in the interaction with fibrinogen, thrombomodulin, protein C, antithrombin III, and the Kunitz inhibitors. J Biol Chem 1993; 268: 1905561.
  • 15
    Bajzar L, Nesheim ME, Tracy PB. The profibrinolytic effect of activated protein C in clots formed from plasma is TAFI-dependent. Blood 1996; 88: 2093100.
  • 16
    Von dem Borne PAKr, Bajzar L, Meijers JCM, Nesheim ME, Bouma BN. Thrombin-mediated activation of factor XI results in a thrombin-activatable fibrinolysis inhibitor-dependent inhibition of fibrinolysis. J Clin Invest 1997; 99: 23237.
  • 17
    Boffa MB, Wang W, Bajzar L, Nesheim ME. Plasma and recombinant thrombin-activable fibrinolysis inhibitor (TAFI) and activated TAFI compared with respect to glycosylation, thrombin/thrombomodulin-dependent activation, thermal stability, and enzymatic properties. J Biol Chem 1998; 273: 212735.
  • 18
    Boffa MB, Bell R, Stevens WK, Nesheim ME. Roles of thermal instability and proteolytic cleavage in regulation of activated thrombin activable fibrinolysis inhibitor. J Biol Chem 2000; 275: 1286878.
  • 19
    Wang W, Boffa MB, Bajzar L, Walker JB, Nesheim ME. A study of the mechanism of inhibition of fibrinolysis by activated thrombin-activable fibrinolysis inhibitor. J Biol Chem 1998; 273: 2717681.
  • 20
    Zhao L, Morser J, Bajzar L, Nesheim M, Nagashima M. Identification and characterization of two thrombin activatable fibrinolysis inhibitor isoforms. Thromb Haemost 1998; 80: 94955.
  • 21
    Wang W, Nagashima M, Schneider M, Morser J, Nesheim M. Elements of the primary structure of thrombomodulin required for efficient thrombin-activable fibrinolysis inhibitor activation. J Biol Chem 2000; 275: 229427.
  • 22
    Kokame K, Zheng X, Sadler JE. Activation of thrombin activable fibrinolysis inhibitor requires epidermal growth factor-like domain 3 of thrombomodulin and is inhibited competitively by protein C. J Biol Chem 1998; 273: 121359.
  • 23
    Walker JB, Nesheim ME. A kinetic analysis of the tissue plasminogen activator and DSPAα1 cofactor activities of untreated and TAFIa-treated soluble fibrin degradation products of varying size. J Biol Chem 2001; 276: 313848.
  • 24
    Brouwers GJ, Vos HL, Leebeek FWG, Bulk S, Schneider M, Boffa M, Koschinsky M, Van Tilburg NH, Nesheim ME, Bertina RM, Gomez Garcia EB. A novel, possibly functional, single nucleotide polymorphism in the coding region of the thrombin-activatable fibrinolysis inhibitor (TAFI) gene is also associated with TAFI levels. Blood 2001; 98: 19923.
  • 25
    Schneider M, Boffa M, Stewart R, Rahman M, Koschinsky M, Nesheim M. Two naturally occurring variants of TAFI (Thr-325 and Il-325) differ substantially with respect to thermal stability and antifibrinolytic activity of the enzyme. J Biol Chem 2002; 277: 102130.
  • 26
    Schneider M, Nagashima M, Knappe S, Zhao L, Morser J, Nesheim M. Amino acid residues in the P6-P'3 region of thrombin-activable fibrinolysis inhibitor (TAFI) do not determine the thrombomodulin dependence of TAFI activation. J Biol Chem 2002; 277: 994451.
  • 27
    Boffa MB, Reid TS, Joo E, Nesheim ME, Koschinsky ML. Characterization of the gene encoding human TAFI (thrombin-activable fibrinolysis inhibitor; plasma procarboxypeptidase B). Biochemistry 1999; 38: 654758.
  • 28
    Boffa MB, Hamill JD, Bastajian N, Dillon R, Nesheim ME, Koschinsky ML. A role for CCAAT/enhancer-binding protein in hepatic expression of thrombin-activable fibrinolysis inhibitor. J Biol Chem 2002; 277: 2532936.
  • 29
    Boffa MB, Hamill JD, Maret D, Brown D, Scott ML, Nesheim ME, Koschinsky ML. Acute phase mediators modulate thrombin-activable fibrinolysis inhibitor (TAFI) gene expression in HepG2 cells. J Biol Chem 2003; 278: 92507.
  • 30
    Schneider M, Nesheim M. Reversible inhibitors of TAFIa can both promote and inhibit fibrinolysis. J Thromb Haemost 2003; 1: 14754.
  • 31
    Walker JB, Hughes B, James I, Haddock P, Kluft C, Bajzar L. Stabilization versus inhibition of TAFIa by competitive inhibitors in vitro. J Biol Chem 2003; 278: 891321.
  • 32
    Schneider M, Brufatto N, Neill E, Nesheim M. Activated thrombin-activatable fibrinolysis inhibitor reduces the ability of high molecular weight fibrin degradation products to protect plasmin from antiplasmin. J Biol Chem 2004; 279: 133405.
  • 33
    Schneider M, Nesheim M. A study of the protection of plasmin from antiplasmin inhibition within an intact fibrin clot during the course of clot lysis. J Biol Chem 2004; 279: 133339.
  • 34
    Walker JB, Bajzar L. The intrinsic threshold of the fibrinolytic system is modulated by basic carboxypeptidases, but the magnitude of the antifibrinolytic effect of activated thrombin-activable fibrinolysis inhibitor is masked by its instability. J Biol Chem 2004; 279: 27896904.
  • 35
    Neill EK, Stewart RJ, Schneider MM, Nesheim ME. A functional assay for measuring activated thrombin-activatable fibrinolysis inhibitor in plasma. Anal Biochem 2004; 330: 33241.