Interaction of heparin cofactor II with neutrophil elastase and cathepsin G.

We investigated the interaction of the human plasma proteinase inhibitor heparin cofactor II (HC) with human neutrophil elastase and cathepsin G in order to examine 1) proteinase inhibition by HC, 2) inactivation of HC, and 3) the effect of glycosaminoglycans on inhibition and inactivation. We found that HC inhibited cathepsin G, but not elastase, with a rate constant of 6.0 x 10(6) M-1 min-1. Inhibition was stable, with a dissociation rate constant of 1.0 x 10(-3) min-1. Heparin and dermatan sulfate diminished inhibition slightly. Both neutrophil elastase and cathepsin G at catalytic concentrations destroyed the thrombin inhibition activity of HC. Inactivation was accompanied by a dramatic increase in heat stability, as occurs with other serine proteinase inhibitors. Proteolysis of HC (Mr 66,000) produced a species (Mr 58,000) that retained thrombin inhibition activity, and an inactive species of Mr 48,000. Amino acid sequence analysis led to the conclusion that both neutrophil elastase and cathepsin G cleave HC at Ile66, which does not affect HC activity, and at Val439, near the reactive site Leu444, which inactivates HC. Since cathepsin G is inhibited by HC and also inactivates HC, we conclude that cathepsin G participates in both reactions simultaneously so that small amounts of cathepsin G can inactivate a molar excess of HC. High concentrations of heparin and dermatan sulfate accelerated inactivation of HC by neutrophil proteinases, with heparin having a greater effect. Heparin and dermatan sulfate appeared to alter the pattern, and not just the rate, of proteolysis of HC. We conclude that while HC is an effective inhibitor of cathepsin G, it can be proteolyzed by neutrophil proteinases to generate first an active inhibitor and then an inactive molecule. This two-step mechanism might be important in the generation of chemotactic activity from the amino-terminal region of HC.

We investigated the interaction of the human plasma proteinase inhibitor heparin cofactor II (HC) with human neutrophil elastase and cathepsin G in order to examine 1) proteinase inhibition by HC, 2) inactivation of HC, and 3) the effect of glycosaminoglycans on inhibition and inactivation. We found that HC inhibited cathepsin G, but not elastase, with a rate constant of 6.0 X lo6 Mm1 min-'. and not just the rate, of protelysis of HC. We conclude that while HC is an effective inhibitor of cathepsin G, it can be proteolyzed by neutrophil proteinases to generate first an active inhibitor and then an inactive molecule. This two-step mechanism might be important in the generation of chemotactic activity from the amino-terminal region of HC.
During the acute inflammatory response, neutrophils respond to a variety of endogenous and exogenous stimuli by migrating to sites of injury and releasing substances that damage pathogens as well as host tissue (1). Among the hydrolytic enzymes released are serine proteinases. These enzymes are believed to be regulated in uiuo by members of the serine proteinase inhibitor (serpin) family of proteins. Interestingly, neutrophils and their released products can also inactivate many serpins not only by oxidizing a critical methionine residue in a,-proteinase inhibitor (2, 3) but also by cleaving peptide bonds in an exposed loop near the reactive sites of several other serpins (5-8). The balance between proteinase activity and proteinase inhibition is critical; disruption of the balance might contribute to disease states such as emphysema and rheumatoid arthritis. We are interested in the properties of the plasma proteinase inhibitor heparin cofactor II (HC),' which is a member of the serpin family. HC, like its homologue antithrombin III, inhibits thrombin in a reaction whose rate is dramatically increased by heparin and other glycosaminoglycans (9). This property suggests a potential role for thrombin inhibition by HC in viuo at sites rich in glycosaminoglycans, such as vessel walls and exposed basement membranes. The physiological importance of HC might also depend on the production of leukocyte chemoattractants from the HC protein, as we recently reported (10).
We investigated the interaction of HC with two major neutrophil proteinases, elastase and cathepsin G. Because HC is a potent inhibitor of chymotrypsin (ll), it might inhibit cathepsin G, which has similar hydrolytic specificity (12). On the other hand, neutrophil elastase, which inactivates the related serpins antithrombin III, az-antiplasmin, al-antichymotrypsin, and Cl inhibitor (4, 5, 8), might also inactivate HC. Sie et al. (13) have shown that neutrophil extracts degrade and inactivate HC, but the degrading activity was not identified. It is known that neutrophil elastase proteolytically inactivates antithrombin III in a reaction that is accelerated by heparin (8, 14); however, the effect of glycosaminoglycans on HC inactivation by purified neutrophil proteinases is not known.
In this study, we found that HC inhibits cathepsin G, but not elastase, and that both cathepsin G and elastase inactivate HC by a proteolytic reaction that is accelerated by heparin and dermatan sulfate. Additionally, we found that the initial product of the reaction between HC and neutrophil proteinases is a degraded form of HC that retains its inhibitory properties. A preliminary report has appeared in abstract form (15).

RESULTS
Proteinase Inhibition-Inhibition of amidolytic activity of neutrophil elastase and cathepsin G by HC was investigated using chromogenic substrates for each proteinase. Minimal inhibition of elastase was detected (50% inhibition of 25 nM elastase by 7 PM HC) and was not further characterized.
In contrast, HC inhibited cathepsin G with a second-order rate constant of 6.0 X lo6 M-' min-' (Table I). Assuming a plasma concentration of 1 ELM for HC, this translates to a half-time for cathepsin G inhibition of 10 s (t,,* = l/k+i[HC]), which may have relevance in uiuo (24). The first-order dissociation rate constant for inhibition was also determined: 1.0 x 10d3 min-' (equivalent to a half-life of about 11 h). Heparin and dermatan sulfate decreased the inhibition rate constant slightly and increased the dissociation rate constant, indicating less stable inhibition (Table I). A bimolecular complex containing HC and cathepsin G was not visible by SDS-PAGE. However, during gel filtration of mixtures of HC and cathepsin G on Sephacryl S-200, cathepsin G activity eluted with the HC peak (not shown).
Inactivation of HC-Both neutrophil elastase and cathepsin G at catalytic concentrations inactivated HC, as measured by loss of thrombin inhibition activity (Fig. 1). Neutrophil elas- tase was more effective than cathepsin G, since a smaller amount of elastase was able to inactivate HC in a shorter period of time. This may be partially due to inhibition of cathepsin G, but not elastase, by HC.
Since other members of the serpin family of proteins exhibit a dramatic increase in heat stability following limited proteolysis (25), the heat stability of HC was investigated. Native HC precipitated from solution when incubated at 58 "C ( Fig.  2). When HC was treated with elastase or cathepsin G until no thrombin inhibition activity remained, HC was stable in solution at 58°C.
The two proteinases generated HC derivatives of similar mobility in SDS-PAGE.
Neither proteinase cleaved the reactive site bond: Leu444-Ser (26). No other sequences were unambiguously identified in cathepsin G digests, but elastase digestion also produced a new aminoterminal sequence beginning with Asp39 (Table II).
Relationship between Cathepsin G Inhibition and HC Znactiuation-In order to directly compare HC inactivation with cathepsin G inhibition, both reactions were followed under identical conditions of time and concentration. As shown in Fig. 4A, both inhibition and inactivation occurred in the same interval. The calculated second-order rate constants were 6.2 X lo6 M-' min-' for cathepsin G inhibition and 1.6 x lo6 M-r min-' for HC inactivation.
These rate constants are probably overestimations since the reactions are simultaneous: some apparent cathepsin G inhibition might result from the reaction in which HC is inactivated, and some apparent inability of HC to inhibit thrombin might be the result of cathepsin G inhibition.
This was also demonstrated by intercepts of greater than unity in the replot (Fig. 4B). When the data were normalized to give intercepts of unity, the rate constants became 3.6 X lo6 M-' min-' for cathepsin G inhibition and 1.4 X lo6 M-r min-1 for HC inactivation.
Under sin G inhibition was 3.5 X 10e3 min-' (half-life of 3.3 h; not shown), thus demonstrating that inactivation of HC by cathepsin G did not occur following dissociation of cathepsin G. The stability of cathepsin G inhibition by HC was further demonstrated by the inability of excess thrombin to promote cathepsin G dissociation from HC (not shown). The possibility that neutrophil elastase contaminants in the cathepsin G preparation were responsible for HC inactivation was discounted hy demonstrating that 125 nM al-proteinase inhibitor (titrated with elastase) added to 500 nM cathepsin G did not affect cathepsin G inhibition, HC inactivation by cathepsin G, or HC proteolysis by cathepsin G (not shown).
Effect of Heparin and Dermatan Sulfate on HC Znactiuation-Because glycosaminoglycans such as heparin and dermatan sulfate alter the rate of thrombin inhibition by HC, and because heparin has been reported to accelerate the rate of inactivation of antithrombin III by elastase (8), the effect of glycosaminoglycans on the inactivation of HC was investigated. Following incubation of HC with neutrophil elastase or cathepsin G in the presence of varying amounts of heparin or dermatan sulfate (0.1 Kg/ml to 3 mg/ml), thrombin inhibition was measured. Increasing heparin concentrations initially afforded slight protection of HC from inactivation by neutrophil proteinases, but above about 10 fig/ml, heparin accelerated HC inactivation (Fig. 5A). In contrast, dermatan sulfate at all concentrations caused only a slight increase in the rate of HC inactivation (Fig. 5B). The effect of glycosaminoglycans on HC inactivation was not correlated with the effect of glycosaminoglycans on the activity of each proteinase as assayed by small chromogenic substrates. For example, HC inactivation was increased at high concentrations of dermatan sulfate, where both elastase and cathepsin G activities were depressed (Fig. 50). Furthermore, the concentration dependence of the heparin effect on HC inactivation was similar for neutrophil elastase and cathepsin G, even though the amidolytic activity of these enzymes responded differently to various heparin concentrations ( The inactivation of HC at high concentrations of heparin and dermatan sulfate resulted in altered patterns of degradation of HC protein as visualized by SDS-PAGE (Fig. 6). When heparin was included, a band similar to band III predominated, and HC activity was greatly reduced. When dermatan sulfate was included, HC activity was reduced, although proteolysis appeared to be diminished relative to controls in the absence of glycosaminoglycan.
A band similar to band II was apparent in samples containing dermatan sulfate, but this raises the possibility that this species was not identical to band II, which retains inhibitory activity. It is possible that glycosaminoglycans alter the susceptibility of HC to proteolysis such that inactivation at the reactive site can occur without extensive proteolysis in the amino-terminal region of the HC molecule.

DISCUSSION
This study was undertaken to determine whether HC inhibits the hydrolytic activity of two major neutrophil proteinases or is inactivated by them. Our results demonstrate that HC inhibits cathepsin G under physiological conditions, but that both cathepsin G and elastase can proteolyze and destroy the thrombin inhibition activity of HC. The inhibition of cathepsin G by HC might be important in uiuo as a supplement to cY1-antichymotrypsin inhibition of cathepsin G (27). The less stable inhibition obtained in the presence of glycosaminoglycans most likely reflects binding of the proteinase to the polyanion, also noted by others (28). The absence of a covalent SDS-stable HC-cathepsin G complex is similar to results with HC-chymotrypsin (11) and al-proteinase inhibitor-trypsin (29). However, the failure to detect the sequence beginning with Ser445 suggests that cathepsin G does not actually hydrolyze the Leu444-Ser bond. Our inhibition measurements differ significantly from those of a previous study which reported an inhibition rate constant of only 1.4 X lo4 Mm1 min-', possibly due to different reaction temperatures (30). The mechanism of cathepsin G inhibition is discussed further below.
The inactivation of HC by neutrophil elastase or cathepsin G was accompanied by increased heat stability, which is consistent with a conformational change caused by proteolytic cleavage of the reactive site peptide loop of HC. This is the first report of the heat stability phenomenon in HC, which has been well documented for other members of the serpin family (25,31). Proteolysis of the reactive site peptide loop renders other serpins inactive as proteinase inhibitors (31). Sequence data provided strong evidence that the inactivation of HC by neutrophil elastase and cathepsin G was accomplished by proteolysis of the reactive site loop at Val*"'-Gly (the Ps-Ps bond in the nomenclature of Schechter and Berger (32)).
Both neutrophil elastase and cathepsin G also hydrolyze HC near the amino terminus, at the Ile"'-Phe bond, as shown by sequence analysis. We conclude that this cleavage produces band II in SDS-PAGE, which retains thrombin inhibition activity. This is not unexpected since the reactive site of HC is located near the carboxyl terminus of HC (26). We have previously identified a degraded form of HC (with an amino terminus corresponding to Asn4') that contains an intact reactive site (16). The results of HC activity assays and SDS-PAGE analysis indicate that the Ile66-Phe bond is cleaved before the Va14"g-Gly bond. Neutrophil elastase and cathepsin G generate the same proteolytic fragments from HC. HC differs from antithrombin III and a2-antiplasmin in that proteolysis of HC by neutrophil proteinases initially generates a degraded form of the inhibitor that retains activity. Only Cl inhibitor appears to share this feature (6). The effect of heparin and dermatan sulfate on the inactivation of HC by neutrophil proteinases is not correlated with the effect of the glycosaminoglycan on proteinase activity (as assessed with small chromogenic substrates) but is consistent with glycosaminoglycan binding to HC to alter its susceptibility to proteolysis.
The accelerated inactivation of HC in the presence of high concentrations of glycosaminoglycans might involve a ternary complex consisting of glycosaminoglycan, inhibitor, and proteinase (similar to the ternary complex formed during thrombin inhibition by HC and glycosaminoglycan (33)). The greater effect of heparin compared to dermatan sulfate parallels the tighter binding of heparin to HC (9), but might also reflect a different binding mechanism. Our data show that the heparin effect at least is markedly concentration-dependent.
The protective effect of heparin at low concentrations is at present unexplained, although it is possible that this reflects a heparin-induced decrease in proteinase activity, which was observed with elastase but which was not detected when cathepsin G was assayed with a synthetic peptide substrate. Sie et al. (13) reported that 1 pg/ml heparin protected HC from inactivation by neutrophil lysates, while 10 pg/ml dermatan sulfate had no effect; higher glycosaminoglycan concentrations were not tested. Our results extend the observations of Sie et al. (13) and demonstrate that HC shares with antithrombin III the property of heparinaccelerated inactivation by neutrophil elastase. The results of SDS-PAGE indicate that glycosaminoglycans do not simply increase the rate of conversion of HC to band II and then band III, but probably promote proteolysis near the reactive site, thus rendering HC inactive with respect to thrombin inhibition.
terminal portion of HC (lo).' The results of this paper demonstrate that proteolysis at Ile@-Phe occurs without diminishing proteinase inhibition activity near the carboxyl terminus. Although it is not known whether heparin or dermatan sulfate influence the production of chemotactic activity, glycosaminoglycans might play a role in defining the susceptibility of HC to proteolysis with or without inactivation of HC.