Hageman factor substrates. Human plasma prekallikrein: mechanism of activation by Hageman factor and participation in hageman factor-dependent fibrinolysis.

Two molecular forms of prekallikrein can be isolated from pooled normal human plasma. Their approximate molecular weights by sodium dodecyl sulfate-gel electrophoresis are 88,000 and 85,000. The two bands observed are shown to represent prekallikrein by functional, immunochemical, and structural criteria. Both forms are cleaved by activated Hageman factor, they appear to share antigenic determinants, they are not interconvertible upon incubation with activated Hageman factor or kallikrein, and the ratio of kinin-generating, and plasminogen-activating activities of the preparations are independent of the relative proportion of each band. Activated Factor XII converts prekallikrein to kallikrein by limited proteolysis and two disulfide-linked chains designated kallikrein heavy chain (Mr = 52,000) and kallikrein light chains (Mr = 36,000 or 33,000) are formed. The active site is associated with the light chains as assessed by incorporation of [3H]diisopropyl fluorophosphate. No dissociable fragments were observed in the absence of reducing agents. However, kallikrein could digest prekallikrein to diminish its molecular weight by 10,000. In addition, two factors capable of activating plasminogen to plasmin have been isolated; one is identified as kallikrein. The second principle fractionates with Factor XI and is demonstrable in normal and prekallikrein-deficient plasma.

ROBERT MANDLE, JR., AND ALLEN P. KAPLAN From the Allergic Diseases Section, Laboratory of Clinical Inuestigation, National Institute ofAllergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland 20014 Two molecular forms of prekallikrein can be isolated from pooled normal human plasma. Their approximate molecular weights by sodium dodecyl sulfate-gel electrophoresis are 88,000 and 85,000. The two bands observed are shown to represent prekallikrein by functional, immunochemical, and structural criteria. Both forms are cleaved by activated Hageman factor, they appear to share antigenic determinants, they are not interconvertible upon incubation with activated Hageman factor or kallikrein, and the ratio of kinin-generating, and plasminogen-activating activities of the preparations are independent of the relative proportion of each band. Activated Factor XII converts prekallikrein to kallikrein by limited proteolysis and two disulfide-linked chains designated kallikrein heavy chain (M, = 52,000) and kallikrein light chains (M, = 36,000 or 33,000) are formed. The active site is associated with the light chains as assessed by incorporation of PHldiisopropyl fluorophosphate.
No dissociable fragments were observed in the absence of reducing agents. However, kallikrein could digest prekallikrein to diminish its molecular weight by 10,000. In addition, two factors capable of activating plasminogen to plasmin have been isolated; one is identified as kallikrein.
The second principle fractionates with Factor XI and is demonstrable in normal and prekallikrein-deficient plasma.
However, published studies on the ability of kallikrein to convert plasminogen to plasmin are. inconclusive.
Colman first reported that kalli-* This is Paner II in a series on "Haeeman Factor Substrates." The preceding paper in this series is Ref. 114 Kaplan and Austen subsequently demonstrated a Hageman factor-activatable plasma factor called plasminogen proactivator, which upon activation, was able to convert plasminogen to plasmin (11). These authors proposed that plasminogen proactivator was responsible for the plasminogen-converting activity in prekallikrein samples. This postulate was challenged by Laake and Vennerod based upon their inability to separate prekallikrein from plasminogen proactivator activity and they concluded that kallikrein and plasminogen-activating activities are functions of the same molecule (12).
We have examined the mechanism of prekallikrein activation by HF and the ability of purified kallikrein to convert plasminogen to plasmin. Our findings can be summarized as follows. (a) Two molecular species of prekallikrein can be isolated from pooled normal human plasma. (b) Activated Hageman factor converts both forms of prekallikrein to kallikrein by limited proteolysis. The single chain of prekallikrein is cleaved at a single point and the two kallikrein chains formed are linked by one or more disulfide bonds. (c) Purified kallikrein directly converts plasminogen to plasmin. (d) An additional plasminogen-activating principle, which fractionates with Factor XI, and is distinct from prekallikrein, can be isolated from either normal plasma or prekallikrein-deficient plasma.  (23). SDS-polyacrylamide gel electrophoresis was run using the same buffer system as modified by King and Laemmli (24) with the exception that 4 M urea was added to the sample buffer. Electrophoresis was carried out in tubes ( Hageman factor fragments were prepared from Hageman factor which had activated during purification (9) or were isolated from acetone-activated human plasma (19) utilizing sequential chromatography on QAE-Sephadex, SP-Sephadex, and Sephadex G-100 as previously reported (9). A functionally pure reagent was obtained containing no detectable plasmin, plasminogen, prekallikrein, or Factor XI activity.

MATERIALS
In some preparations trace contamination with thrombin was detected; hirudin was added to these preparations to a final concentration of 2.0 units/ml which completely inactivated the thrombin and had no effect upon HF,. Preparations were quantitated by bioassay on a guinea pig ileum (9) and adjusted such that 5 pl of Hageman factor fragment generated 10 ng of bradykinin following a P-min incubation at 37" with 0.2 ml of fresh EDTA plasma. one minor band can be seen. These two bands were observed in all our prekallikrein preparations; however, the relative proportion of each band varied. Purified prekallikrein was next fractionated on a column (2.5 x 150 cm) of Sephadex G-150 (superfine) in order to determine whether these proteins could be resolved by an additional technique which reflects differences in molecular size. The prekallikrein was located by the amidolytic assay as described under "Materials and Methods." Samples were then taken across the peak and electrophoresed on SDS-polyacrylamide gel electrophoresis as shown in Fig. 2. The slower migrating band on SDS-polyacrylamide gel electrophoresis eluted earlier on gel filtration, again suggesting a difference in molecular size. However, both slower and faster species were contained within a single functional peak.
In order to investigate the antigenic relationship of the two proteins seen in our prekallikrein preparations, antibody to prekallikrein was reacted with a preparation having nearly equal amounts of the faster and slower migrating components. Only one precipitin arc was seen upon immunoelectrophoresis (Fig. 3) and a single line was obtained upon double diffusion against a wide range of antibody and antigen concentrations.
We next examined the change in SDS-polyacrylamide gel electrophoresis pattern when prekallikrein was activated to kallikrein by HF,. In nonreduced samples there was no significant change in mobility when prekallikrein was converted to kallikrein.
Both samples showed two bands and the relative proportion of these bands did not change. In the next experiment 1311-prekallikrein and 1251-kallikrein were mixed and electrophoresed on the same SDS-gel. The gel was then sliced and counted. As shown in Fig. 4, 1311-prekallikrein and lZ51kallikrein counts were superimposable. Fig. 5 compares SDSpolyacrylamide gel electrophoresis patterns of prekallikrein and kallikrein in reduced and nonreduced samples. Reduced prekallikrein exhibited a slower relative mobility, presumably due to increased unfolding upon reduction of intrachain disulfide bonds. However, the standards exhibited this same change in mobility and the calculated molecular weights of reduced and nonreduced prekallikrein did not differ. Although it is not clear in this gel, the double-banded pattern of prekallikrein was not altered by reduction (see Fig. 7). Reduced kallikrein preparations had three new bands with faster mobility, while nonreduced kallikrein had two bands that were not distinguishable from prekallikrein. Thus prekallikrein is activated to kallikrein without significant change in molecular weight. It also follows that the double-banded pattern observed cannot be attributed to kallikrein contaminating our prekallikrein preparations. Fig. 6 shows a schematic diagram of this banding pattern including an estimated molecular weight and our nomenclature for each band. The molecular weight of prekallikreins I and II was accounted for by the sum of the molecular weights of the heavy chain and light chains I and II, respectively. Thus, the molecular weight difference between prekallikreins I and II was reflected in the molecular weight difference between kallikrein light chains I and II. Thirty micrograms of kallikrein were then incubated with 10m3 M [3H]iPr2PF, reduced, and subjected to SDS-polyacrylamide gel electrophoresis.
When the stained gel was assessed by radioautography (Fig. 71, both light chains incorporated the radiolabel and no [3H]iPr2PF was incorporated into the heavy chain. Kinetics of Prekallikrein Activation by HF, -One-half milliliter of prekallikrein (0.8 mglml) plus 1251-prekallikrein were mixed with a suboptimal concentration of HF, (see "Materials and Methods") and incubated at 37". Samples for polyacrylamide gel electrophoresis analysis were then withdrawn at time intervals up to 1 h and added to SDS buffer containing 2% SDS, 0.2% mercaptoethanol, 8 M urea, and 0.02 M EDTA. Samples for amidolytic assays (20 ~1) were added to 1 ml of cold Tris/imidazole buffer, and assayed within 5 min of sampling. The results of SDS-polyacrylamide gel electrophoresis analysis are shown in Fig. 8. Controls included prekallikrein incubated at 4" or 37" with buffer for 60 min in the absence of HF, and a 0-min control in which HF, was added after the SDSbuffer and the mixture then incubated for 60 min. These controls were not different, indicating that prekallikrein was not cleaved in the absence of HF, and that SDS buffer immediately stopped the reaction. In prekallikrein samples incubated with HF, a progressive loss of protein from the prekallikrein region was accompanied by the appearance of kallikrein heavy and light chains. Both prekallikrein I and II appeared to be converted to kallikrein at the same rate and the proportion of stained protein appeared to be conserved between prekallikrein I and II and kallikrein light chains I and II. The gel was then sliced and counted for 1251, The per cent of the total sample radioactivity that was associated with the heavy chain was calculated for each sample and this percentage was divided by the value for the 60-min sample. Fig. 9 compares the rate of cleavage as reflected by the rate of the heavy chain formation with the genesis of amidolytic activity. Although similar, the kinetic curves are not identical. The generation of amidolytic activity appeared to precede cleavage.
We therefore examined the possibility that the observed bond cleavage is caused by the kallikrein generated rather than by HF,.  Under these conditions, Trasylol did not inhibit HF, but completely inactivated kallikrein in less than 2 min as assessed by kinin generation or amidolytic activity. The SDS-polyacrylamide gel electrophoresis shown in Fig. 10 represents one such experiment. No significant difference in cleavage rate with or without Trasylol could be detected during a l-h time course. Therefore the cleavage observed was not caused by the kalli-cient plasma and upon activation not only digested HMW kininogen to liberate bradykinin, but also activated plasminogen to yield the fibrinolytic enzyme plasmin. Since plasminogen proactivator was previously reported to be similar to, but separable from prekallikrein (91, we attempted to ascertain whether the two bands observed on SDS-polyacrylamide gel electrophoresis might represent two different Hageman factor substrates or two molecular forms of prekallikrein.
Preparations containing different quantities of prekallikrein at various stages of purification as well as purified prekallikrein with disproportionate amounts of prekallikrein I and II were assayed for amidolytic activity, bradykinin-generating ability, and plasminogen-activating activity. A linear relationship between bradykinin generation and p-nitroanilide liberation was observed in all preparations (Fig. 12). The correlation was highly significant (r = 0.78, p < 0.001) indicating that both assays are reflecting the same functional entity. When amidolytic activity or bradykinin-generating activity was compared to plasmin generation, a significant correlation was again observed (r = 0.99, p < 0.001) with the exception of the initial QAE-Sephadex effluent (Peak I) samples (Fig. 13). The data points from Peak I samples were pooled and compared to the data from the 12 remaining samples (see "Materials and Methods"). Peak I sample had significantly more (t = 8.78, p < 0.001) fibrinolytic activity relative to amidolytic activity, suggesting that it contained an additional fibrinolytic factor. All other samples tested had the same relative amount ofpnitroaniline, bradykinin, and plasmin-generating activity. The ability of the inactivator of the first component of complement (Ci INH) to inhibit the amidolytic and plasmino-krein generated.
To further investigate the ability of kallikrein to digest prekallikrein, '251-labeled prekallikrein was incubated with HF,-activated prekallikrein from which the HF, had been removed by ion exchange chromatography on QAE-Sephadex. The concentration of kallikrein chosen was twice the amount of activatable prekallikrein.
As can be seen in the autoradiogram (Fig. ll), kallikrein digested prekallikreins I and II to yield prekallikreins whose molecular weights were diminished by 10,000 (last two gels). This cleavage was inhibited by addition of Trasylol to the incubation mixture.
Relationship of Prekallikrein to Plasminogen Proactivator -The prekallikrein utilized in the aforementioned studies corrected the functional abnormalities in prekallikrein-defi- I  I  I  I  I  I   01   0  10  20  30  40  50  60  TIME (min.) FIG. 9. Kinetics of prekallikrein activation by HF,. All points are compared to the 60-min values which are defined as 100% activation. Open circles represent appearance of amidolytic activity. The closed circles are the rate of cleavage assessed by the increase in radiolabel in the kallikrein heavy chain position on SDS-polyacrylamide gel electrophoresis.

FIG. 10. Autoradiogram
of '251-labeled prekallikrein activated in the presence or absence of 2000 units/ml of Trasylol. The first five gels from left to right have prekallikrein plus HF, plus 'I'rasylol incubated for 0, 5, 10, 20, and 45 min. The next five gels contain prekallikrein plus HF, plus buffer incubated 0, 5, 10, 20 and 45 min. The final sample is a prekallikrein control.
FIG. 11. Autoradiogram of 1251-labeled prekallikrein plus nonlabelod kallikrein. The samples from left to right are: prekallikrein plus kallikrein plus 2000 units/ml of Trasylol incubated for 0,30, and 60 min; followed by prekallikrein plus kallikrein incubated for 0,30, and 60 min without any Trasylol. in preliminary experiments to be at twice the concentration necessary to inhibit all the detectable bradykinin generation or amidolytic activity. The inactivated mixture and kallikrein control were then fractionated on 5-ml columns of QAE-Sephadex equilibrated in 0.003 M phosphate buffer, pH 8, to remove excess Cl INH. The effluents were assayed for amidolytic and plasminogen-activating activity. The initial fall-through of the control column had both plasmin-generating andp-nitroaniline-liberating activity while neither activity was demonstrable in the QAE-Sephadex fall-through of the Ci INH-kallikrein sample (Fig. 14).
The previous experiments (Fig. 13) suggest that kallikrein is one, but not the only factor in Peak I which functions as a plasminogen activator. We next fractionated 200 ml of normal and prekallikrein-deficient plasma on a column ( amidolytic, kinin-generating, and plasminogen-activating activity. No detectable amidolytic or kinin-generating activity was detectable in the effluent from prekallikrein-deficient plasma before or after activation with HF,. Amidolytic or kinin assays were sensitive to less than 10% of the activity recovered in the effluent of normal plasma. However, the efIluents obtained from the normal and prekallikrein-deficient plasma contained plasminogen-activating activity which could be increased by the addition of HF,. The prekallikrein-deficient ellluent contained approximately one-third of the plasminogen-activating activity of the normal plasma efIluent. The QAE-Sephadex effluent obtained from the prekallikrein-deficient plasma was applied to a column (5 x 40 cm) of SP-Sephadex equilibrated in 0.003 M phosphate buffer, pH 6, and the column eluted with a sodium chloride gradient (11,26). The eluate was then assayed for prekallikrein, Factor XI, and plasminogen-activating activity. A single peak of plasminogen-activating activity which superimposed the Factor XI peak was observed (Fig. 151, while no peak of plasminogenactivating activity was found corresponding to the elution position of prekallikrein. The normal chromatogram had a by guest on March 24, 2020 http://www.jbc.org/ Downloaded from prekallikrein peak as well as a Factor XI peak and plasminogen-activating activity was associated with each peak.

DISCUSSION
Two forms of prekallikrein, designated prekallikrein I and II, could be isolated from pooled normal human plasma. Prekallikreins I and II had apparent molecular weights of 88,000 and 85,000 in SDS-polyacrylamide gel electrophoresis and were distinguished by reduced or nom-educed SDS-polyacrylamide gel electrophoresis as well as gel filtration under nondissociating conditions. Activated Hageman factor converts prekallikrein to kallikrein by limited proteolytic digestion. The single chain of prekallikrein is cleaved and two disulfidelinked chains designated kallikrein heavy and light chains are formed. In the absence of reducing agents, the active enzyme did not differ from the proenzyme in terms of size or charge. Kallikreins I and II consisted of one heavy chain (M, = 52,000) and one light chain (M, = 36,000 or 33,000) demonstrable in reduced SDS-polyacrylamide gel electrophoresis. The difference in molecular weight of prekallikreins I and II was reflected in kallikrein light chains I and II. The evidence that both bands represent prekallikrein is indirect, but persuasive: (a) both forms are activatable by HF, and yield similar cleavage patterns; (b) they contain the same antigenic determinants; (cl they are not interconvertible upon activation with HF, or upon digestion of prekallikrein by kallikrein; (d) 13H1iPr,PF is incorporated into both light chains, indicating that each contains an active site serine; (e) plasminogenactivating activity is always proportional to the amount of kallikrein function (bradykinin generation or amidolysis) regardless of the ratio of prekallikreins I and II; (f, neither band is seen when prekallikrein-deficient plasma is fractionated and no plasminogen proactivator function is found in the usual position of prekallikrein on SP-Sephadex. The cleavage of prekallikrein concomitant with activation was shown to be a direct interaction between HF, and prekallikrein. Neither the cleavage pattern nor the kinetics of cleavage were altered by the presence of high concentrations of Trasylol. However, kallikrein appeared able to digest prekallikreins I and II without activation to yield prekallikreins that were diminished by M, = 10,000. The rates of activation, when assayed by generation of an amidolytic site, were slightly faster than when cleavage was used as a measure of activation. Although this might suggest that cleavage occurs subsequent to activation, the addition of the SDS buffer to the sample for polyacrylamide gel electrophoresis analysis instantaneously stopped the reaction, while the functional assay required an additional incubation. The latter determination showed significant activity even at the earliest time point. It is also possible that the trace of '29prekallikrein used to detect cleavage may not precisely reflect the cleavage rate of the nonradiolabeled material; or that some fraction of our prekallikrein is cleaved at a slower rate by HF, and does not produce a functional molecule. Although the mechanism of activation of human prekallikrein has not previously been reported, similar activation mechanisms of bovine and rabbit prekallikrein have been described. No change in molecular weight was seen upon conversion of bovine prekallikrein to kallikrein, and a single bond cleavage was identified such that upon reduction, a heavy and light chain were formed (27). Activation of rabbit prekallikrein was reported to occur with release of a small peptide of M, = 11,000. The resultant kallikrein was smaller and more acidic than prekallikrein (28). A later study, how-ever, did not confirm such peptide formation as part of the activation mechanism and the product was a kallikrein composed of two disulfide-linked chains whose total molecular weight was the same as prekallikrein (29). It is possible that the peptide released was caused by kallikrein digestion of prekallikrein or kallikrein. We have also reinvestigated the relationship of prekallikrein to plasminogen proactivator and have concluded that the Hageman factor-dependent plasminogen proactivator originally described by Kaplan and Austen (11) was prekallikrein or the prekallikrein derivative formed by kallikrein digestion, or both. We could not detect any antigenic difference in the various forms of prekallikrein and, although prekallikreindeficient plasma does possess a plasminogen activator in theyglobulin fraction, further fractionation revealed no plasminogen proactivator activity in the usual position of prekallikrein. The identity of prekallikrein and plasminogen proactivator was first suggested by Laake and Vennerijd who were unable to separate kinin-generating and esterase activities from plasminogen-activating activity (12). We also found the activities to fractionate together, particularly when iF'r,PF was incorporated into the isolation procedures. Plasminogen proactivator was reported to have a molecular weight of 10,000 less than prekallikrein and it is possible that the peak of fibrinolytic activity observed was a result of cleavage of prekallikrein by kallikrein (11). The identification of a peak of plasminogenactivating activity in the y-globulin fraction of prekallikreindeficient plasma also suggested that prekallikrein and plasminogen proactivator might not be identical (30). These results are at variance with other reports which indicated that the y-globulin effluent obtained from prekallikrein-deficient plasma contains no plasminogen activator or proactivator activity (31). However, upon further investigation of normal plasma, a second peak of plasminogen-activating activity was identified which eluted in a position similar to Factor XI rather than prekallikrein (26). We have recently reported that SP-Sephadex chromatography of the y-globulin effluent obtained from prekallikrein-deficient plasma yields only this second peak of fibrinolytic activity (32). As shown in Fig. 15, the activity was found superimposed upon Factor XI. It is not clear whether this activity is a property of Factor XI or represents an unrelated fibrinolytic factor. Further studies are in progress to distinguish these possibilities.