Proteolytic Inactivation of CY- 1-Anti-chymotrypsin SITES OF CLEAVAGE AND GENERATION OF CHEMOTACTIC ACTIVITY*

The effect of several microbial and mammalian pro- teinases on the inhibitory activity of human plasma a-1-anti-chymotrypsin (a- 1-Achy) has been tested. Most of these enzymes caused rapid inactivation of the inhibitor by cleavage at single sites within the reactive-site loop between Pa Lys and P3’ Leu, with additional cleavages also being detected in some cases near the NH, terminus of the native protein. In contrast, two of the enzymes tested failed to inactivate a-1-Achy, although they could cause removal of peptides near the NH, terminus. Studies of neutrophil chemotaxis re- vealed that native or NH,-terminally truncated a-1-Achy was not stimulatory. However, testing of two enzymatically inactivated forms of the inhibitor (a-1-Achy’), cleaved at widely different positions within the reactive-site loop, indicated that they had become potent chemoattractants at concentrations within the na-nomolar range. Competition studies using proteolyti- cally inactivated a-1-proteinase inhibitor suggested that the chemotactic activity of the two inactivated serpins was probably mediated by the same mecha- nism. The physiological relevance of this chemotactic activity is underscored by the results of plasma elimi- nation studies which indicate that a-1-Achy’ a-2-macroglobulin, plasminogen purified from one of by sequential affinity chromatography Human plasmin by activation plasminogen Assay of Enzyme Activity-Trypsin amidase activity was measured with Bz-L-Arg-pNA (final concentration 1.0 mM) in 0.2 M Tris-HC1, 2 mM CaC12, pH 8.0. HNE and cat G activities were measured in 0.1 M Tris-HC1, 0.5 M NaC1, pH 7.4, using Suc-L-Ala-L-Ala-L-Ala-pNA and Suc-L-Ala-L-Ala-L-Pro-L-Phe-pNA (final concentrations 0.5 mM), respectively. The reactions were monitored by following the change in Alas at 25 'C. The proteolytic activities of bacterial metalloproteinases and porcine trypsin were measured with hide powder azure (Calbiochem) ( h a 1 substrate concentration 1% w/v) in 0.1 M Tris-HC1, 2 mM CaCI,, pH 7.5. Active-site Titrations-Measurement of the activities of human and porcine trypsins, as well as human plasmin, were performed with p-nitrophenyl-p'-guanidinobenzoate (20). Active-site titrated porcine trypsin was also used to determine the inhibitory activity of a-1-PI and a-2-macroglobulin. These inhibitors were used as secondary standards to determine the activities of HNE and cat G (a-1-PI) and bacterial metalloproteinases (a-2-macroglobulin). The cells were then assayed for a chemotactic response to 1 nM a-l-PI', 1 nM a-1-Achy' or 10 nM N-formyl-L-methionyl-L-leucyl-L- phenylalanine (FMLP).

states which correlate with a genetic deficiency in specific serpins have previously been described, the best example being that of a-1-PI where a deficiency state has been correlated with the development of pulmonary emphysema (3).
One of the more interesting serpins is a-l-anti-chymotrypsin (a-1-Achy), an important acute phase plasma protein whose concentration can rise up to $fold during inflammatory episodes (4, 5). The regulatory function of this inhibitor probably involves the control of neutrophil cathepsin G (cat G) (6), although it can also inhibit human pancreatic elastase I1 (7) and mast cell chymase (8). An inherited, partial, heterozygous a-1-Achy deficiency state for this serpin has been described (9) which has been associated with liver and lung function abnormalities. However, no homozygotes have yet been identified suggesting that such a defect might be lethal. More recently, a role in the development of Alzheimer's disease has been suggested because of the presence of this inhibitor in the brain amyloid plaque deposits (10).
Kress and co-workers (11)(12)(13)(14) have previously shown that a-1-Achy is more sensitive to proteolytic inactivation by microbial and snake venom proteinases than are other serpins. In these studies inactivation of the inhibitor by a proteolytic cleavage within the inhibitor reactive-site loop was suggested but no structural data were presented for confirmation. In this report we describe the kinetics of inactivation of a-1-Achy by bacterial proteinases, porcine and human trypsins, human plasmin, and human neutrophil elastase (HNE), as well as their cleavage sites within the reactive-site loop of this inhibitor. To assess the physiological importance of these proteolytic cleavages we have also determined the plasma elimination time of the modified forms of a-1-Achy. Finally, we have followed the generation of neutrophil chemotactic activity as native a-1-Achy is converted proteolytically into inactive forms.
Since it is confusing to describe the different forms of a-l-Achy generated by the proteinases used here, the following nomenclature will be used. Native a-1-Achy will refer to the fully active inhibitor. a-1-Achy' will refer to the form of the inhibitor inactivated by a proteolytic cleavage within the reactive-site region. a-1-Achy which is cleaved at the amino terminus will be referred to specifically.
Active-site Titrations-Measurement of the activities of human and porcine trypsins, as well as human plasmin, were performed with p-nitrophenyl-p'-guanidinobenzoate (20). Active-site titrated porcine trypsin was also used to determine the inhibitory activity of a-1-PI and a-2-macroglobulin. These inhibitors were used as secondary standards to determine the activities of HNE and cat G (a-1-PI) and bacterial metalloproteinases (a-2-macroglobulin).
Proteolytic Inactivation of a-1 -Achy-Constant amounts of inhibitor (final concentration 7.5 M) were incubated with individual proteinases in 0.05 M Tris-HC1, 1 mM CaC12, pH 7.4, at 37 "C in a final volume of 200 pl. The molar ratio of a-1-Achy to proteinase was adjusted so that proteolytic inactivation could be followed during a 10-120-min incubation period. Controls consisted of inhibitor and buffer only. At given time intervals aliquots were removed and tested for inhibitory activity toward cat G by addition of an excess of enzyme. After a 5-min preincubation at 25 "C, the mixture was assayed for residual cat G amidolytic activity as described above. The effectiveness of inhibitor cleavage was confirmed by SDS-polyacrylamide gel electrophoresis (21) using a 7-20% acrylamide (w/v) linear gradient. For chemotaxis and plasma elimination studies, proteolytically cleaved serpins were separated from proteinases by Mono Q fast protein liquid chromatography.
Sequence Anulysis of Modified a-1-Achy-a-1-Achy was incubated with a given proteinase for a suitable time period in molar ratios ranging from 1:l to 3000:1, as described above, until residual cat G inhibitory activity had decreased to about 10%. Proteinase activity was then inhibited by addition of either 1.0 mM diisopropyl fluorophosphate or 20 mM o-phenanthroline, depending on the enzyme being tested, and the digestion mixture was either used directly for NHz-terminal sequence analysis or subjected to Mono Q fast protein liquid chromatography in 20 mM Tris-HC1, pH 8.0 to separate the inactivated proteinase from cleaved inhibitor. The latter was then lyophilized and desalted on Sephadex G-25. NHn-terminal sequence analysis was performed on either the digestion mixture or the purified a-1-Achy' using an Applied Biosystem 470A gas-phase sequenator and the program designed by the manufacturer. In the case of a-l-Achy' obtained from HNE digests, small peptide fragments were obtained by reverse-phase high performance liquid chromatography using an acetonitrile/trifluoroacetic acid gradient as described previously in analyzing the cleavage products of P. aeruginosa elastaseinactivated a-1-PI (22). Peptides were detected by monitoring at 214 nm, the individual fragments collected and dried under vacuum, and sequence analysis performed.
Plasm Elimination St~dies-'~~I-labeled proteins (1.0 pg), radioiodinated by the lactoperoxidase/glucose oxidase method (Bio-Rad) were injected into the lateral tail vein of CD-1 mice. Blood samples (25 11) were obtained at timed intervals by retroorbital puncture and 1251 measured in a y-counter. The initial time point, taken between 5 and 10 s after injection, was considered to represent 100% radioactivity in the circulation. Each labeled protein was tested at least in duplicate.
Chemotaxis-Chemotactic activity was determined in modified Boyden chambers by the micropore-filter leading front assay (23), using neutrophils obtained from peripheral blood of healthy donors and separated on Ficoll-Hypaque density gradients (Mono-Poly resolving medium, Flow Laboratories) (24). Briefly, 1.2 X lo6 neutrophils/ml in RPMI medium 1640 (GIBCO) were placed in the upper compartment of the Boyden chamber and separated from chemo-attractants placed in the lower compartment by a cellulose nitrate micropore filter with a mean pore diameter of 5 pm (Sartorius). After 1 h the chambers were disassembled and the membranes were stained with hematoxylin. Chemotaxis was determined as the mean depth in micrometers of penetration into filter of the leading two cells. Experiments were performed in triplicate. In competitive experiments neutrophils were incubated in 10 nM of either native a-1-PI, a-l-Achy, or proteolytically inactivated forms of each inhibitor for 30 min at ambient temperature, then washed three times with medium. The cells were then assayed for a chemotactic response to 1 nM a-l-PI', 1 nM a-1-Achy' or 10 nM N-formyl-L-methionyl-L-leucyl-Lphenylalanine (FMLP).

RESULTS
Enzymatic Inactivation of a-1 -Achy-Incubation of a-l-Achy with each of the three microbial metalloproteinases, the two pancreatic trypsins, or HNE resulted in a concentrationand time-dependent decrease in its inhibitory activity toward cat G (Fig. 1). On an individual basis, a-1-Achy was inactivated most efficiently by P. aeruginosa elastase, followed by S. mrcescem 56-kDa metalloproteinase, human pancreatic trypsin, and S. aureus metalloproteinase. HNE and porcine pancreatic trypsin were less effective, causing inactivation of a-1-Achy at inhibitor:enzyme molar ratios 1 order of magnitude higher. On the other hand, a-1-Achy was totally resistant to inactivation by either clostripain or the 5' . aureus V8 proteinase, which cleaved only at the amino terminus (Table  I), and only slowly inactivated by human plasmin (at a molar ratio of 51, 40% of the inhibitory activity was retained after 2 h) (data not shown). In all cases inactivation was accompanied by limited proteolysis, as demonstrated by the appearance of a small peptide of M, near 5,000 ( Fig. !?A). Despite the removal of this peptide an increase in the rate of migration of the larger a-1-Achy fragment in SDS-PAGE was only observed in the case of the trypsin-treated inhibitor (Fig. 2B, lane 2). This same pattern of a-1-Achy versus a-1-Achy' migration in SDS-PAGE has been described previously (25). Coincidentally, AT 111, in its proteolytically cleaved form, has a mass which is 5 kDa lower than the native protein, yet migrates more slowly in SDS-PAGE than the native inhibitor (25). At the present stage we can only speculate that a change in protein tertiary structure is responsible for this unusual  difference in the electrophoretic mobility between native and cleaved serpins. Sequence Analysis of Enzymatically Inactivated Forms of a-1 -Achy-Amino-terminal sequence analysis of a-1-Achy after treatment with each of the three bacterial metalloproteinases gave identical results (Table I), yielding 2 residues/cycle. One residue in each cycle corresponded to the amino terminus of the native inhibitor. This was 2 residues longer than originally reported (26), beginning with the sequence NHn-His-Pro-. However, this result is consistent with the amino terminus predicted from the cDNA sequence reported earlier (27). It is also in agreement with the sequence of an a-1-Achy isoform recently described by Lindmark et al. (28). The second sequence represented the amino terminus of the new peptide generated by proteolytic cleavage within the reactive site of a-1-Achy which was treated with either human or porcine trypsin also gave 2 residues/cycle. One sequence was due to proteolysis between Arg-18 and Gly-19 and is apparently responsible for the faster migration of trypsin-cleaved a-l-Achy in SDS-polyacrylamide gel electrophoresis (Fig. 2), since metalloproteinase-modified a-1-Achy, in which cleavage is specifically within the reactive-site loop, migrates at the same rate as the native inhibitor. However, the second set of residues identified represents the sequence occurring after a cleavage site within the reactive-site loop (Table I). Significantly, clostripain could only cleave after Arg-18, since there were no other Arg residues accessible for proteolysis by this enzyme within the reactive-site loop.
Treatment of a-1-Achy with HNE yielded two major se-a-l-Achy.
quences per cycle. One of the major sequences again represented the amino terminus of native a-1-Achy. The other major sequence was found to occur within the reactive-site loop and was confirmed by isolation and structural analysis of individual peptide fragments, as described under "Methods." In summary, the proteinases examined cleaved a-1-Achy within the reactive-site region between the P6 Lys and the PB' Leu, or within the first 19 residues of the amino terminus of the native protein. Significantly, after proteolytic cleavage in these regions there was no further degradation of the various forms of a-1-Achy', even after prolonged incubation at relatively high enzyme concentrations (Fig. 2B, lane 7). Plasma Elimination Studies-Cleavage of a-1-Achy within the reactive-site loop with HNE, near the amino terminus with clostripain, or at both sites with trypsin had no effect on the rate of clearance of such modified forms of the inhibitor from the circulation, relative to that of the native protein (Fig. 3). Such results were obtained despite the fact that such cleavage causes major conformational changes in the structure of the inhibitor (29), similar to that reported earlier for a-l-PI (30). The site within the reactive-site loop where cleavage occurs is also apparently not significant since a form of a-l-Achy' obtained by cleavage between the P?' Arg and the Pa' Thr by a trypsin-like proteinase from Porphyromonus gingivalis was removed at the same rate as the HNE and trypsin cleaved protein.' This slow removal of various forms of a-l-Achy' is in sharp contrast to the rapid, receptor-mediated plasma elimination of the a-1-Achy-cat G complex, which is shown for comparison (Fig. 3). The slow elimination of a-l-J. Potempa and A. Mast, unpublished observation.
Achy' indicates that it is not recognized by the receptormediated pathways, which rapidly remove serpin-proteinase complexes from the circulation (31)(32)(33), further increasing the potential for their accumulation in the acute phase state.

Neutrophil Chemotactic Activity of Various Forms of a -l -
Achy-The extraordinary sensitivity of a-1-Achy to conversion to a-1-Achy' by proteolytic inactivation within the reactive-site loop, together with the slow clearance of each from the circulation, suggests that large amounts of a-1-Achy' could accumulate in infected andlor inflamed tissues. in vivo. Since

FIG. 2. SDS-polyacrylamide gel electrophoresis of different
a-1-Achy/proteinase reaction mixtures. Enzymes and inhibitor were incubated as described under "Experimental Procedures" and aliquots removed at specific times for analysis by gel electrophoresis after treatment with 1% SDS. A, a-1-Achy with P. ueruginosa elastase  this inactivation results in a major rearrangement in protein structure we investigated whether it might also lead to the generation of chemotactic activity, as previously described for various forms of a-1-PI' (34,35). In these experiments two forms of a-1-Achy' obtained by inactivation of the native protein with either trypsin or the S. aurew metalloproteinase were used because these proteinases cleave at widely separated regions within the reactive-site loop (Table I). In addition, trypsin also cleaves a-1-Achy near the amino terminus, yielding a form of a-1-Achy' which has been modified in two places. As shown in Fig. 4, both modified forms had potent neutrophil chemotactic activity. Significantly, neutrophils preincubated with either form of modified a-1-Achy or with P. aeruginosa treated a-1-PI did not migrate toward these same proteins in Boyden chamber experiments. However, similar pretreatment had no effect on the neutrophil chemotactic response to FMLP (Fig. 5), indicating that this latter compound was being recognized by a separate receptor on the neutrophil surface. Controls using either native a-1-Achy or a-1-PI had no chemotactic activity, and preincubation of neutrophils with these proteins had no effect in reducing cell migration stimulated by the proteolytically modified inhibitors (Figs. 4 and 5).

DISCUSSION
Members of the serpin family which have retained their inhibitory activity are all characterized by the presence of an exposed reactive-site loop that is sensitive to proteolysis by enzymes with which it cannot readily form stable complexes   * (dark striped bars), native a-1-Achy (stippled bars), assay medium control (solid burs), as described under "Experimental Procedures," and assayed for a chemotactic response to 1 nM of either or-1-Achy' or a-l-PI*, or 10 nM FMLP. The data represent an average of triplicate determinations, and error burs represent the standard error of the mean. Spontaneous migration was 25 -t 2 Nm. (11)(12)(13)(14)36). The crystallographic model of cleaved forms of serpins (29,30) reveals that proteolysis within the loop is accompanied by a major change in protein tertiary structure and the transition from a "stressed" (S) structure to a more ordered, thermodynamically stable "relaxed" (R) form of the molecule (37). This conformational change appears to occur in other serpins with inhibitory activity (38, 39) and may provide a physiological switch that regulates inhibitory activity at inflammatory foci (37). However, from the data described above it is likely that the inactive forms of inhibitors have alternate functions which are of far more importance during acute conditions.
Previously, native human a-1-Achy had been shown to have at least three functions, all of which appeared to involve the regulation of chymotrypsin-like proteinases, including cat G, mast cell chymase, and lymphocyte cell surface proteinases (2, 8,40,41). Cathepsin G can not only degrade connective tissue components (42,43), it can also act as an effective angiotensin-converting enzyme (44) and an inactivator of bradykinin (45). Therefore, by regulating this activity a-l-Achy may control both connective tissue breakdown and/or smooth muscle contraction at an inflammatory locus, although this has not yet been proven. In a different sense a-l-Achy may also act as an immunoregulator, since incubation of the native inhibitor with killer lymphocytes leads to the loss of both killing and antibody-dependent cytolytic activity (40,41).
Recently, a function for HNE-inactivated a-1-Achy in the production of IL-6 and, indirectly, acute phase protein synthesis was demonstrated (46). We now have data which indicate an additional role for a-1-Achy' in neutrophil chemotaxis, the results obtained indicating that this activity can be rapidly generated when the native inhibitor is cleaved by a wide array of bacterial and endogenous human proteinases to which the body may be exposed during pathological states (e.g. pancreatitis (trypsin), severe inflammation (HNE and cat G ) , and sepsis (bacterial proteinases)). Under homeostatic conditions, most proteolytic activities are tightly regulated by endogenous inhibitors; however, this is markedly altered at sites of infection and/or inflammation. Two mechanisms are recognized as being responsible for this change. In the first, an oxidative burst of activated phagocytes eliminates most of the anti-elastase activity due to a-1-PI. This facilitates the second mechanism which involves neutrophil elastase cleavage and inactivation of other serpins (47), including a-1-Achy. The latter event can also easily be executed by pathogenderived proteinases which are generally poorly controlled by endogenous inhibitors, with a-1-Achy being exceptionally sensitive to this type of limited proteolysis. The neutrophil chemoattractant activity of a-1-Achy (actually a-1-Achy') results only when the native inhibitor is cleaved within the reactive-site loop, proteolysis at other sites having no effect. a-1-PI' has already been shown to be a human neutrophil chemoattractant (34), and the results shown here for a-l-Achy', together with preliminary data on inactive forms of both AT I11 and C1 inhibitor, indicate that this may be a property of most proteinase-inactivated serpins. Heparin cofactor 11, however, is an exception, possessing neutrophil chemoattractant activity which is located in the amino-terminal region of the native protein and only detected after proteolysis (48). Cleavage of the reactive-site region of this inhibitor does not result in the generation of chemotactic activity.
Since all forms of a-1-Achy' have equal chemotactic activity, regardless of the position of peptide bond cleavage within the reactive-site loop which leads to their formation, it is likely that the exposure of new sites on the surface of the modified protein resulting from an S to R shift after peptide bond cleavage is the mechanism utilized in the generation of this activity rather than regions within the reactive-site loop. This is supported by the fact that the carboxyl-terminal peptide arising by enzymatic cleavage is apparently re-folded into a @-pleated sheet structure and at least partially buried within the modified protein (29,30).
It has been suggested, however, that the carboxyl-terminal peptide of a-1-PI, generated by enzymatic inactivation with macrophage elastase (34), is responsible for chemotactic activity, and this has been partially confirmed in the case of a-1-PI where the amino acid sequence directly involved in recognition of the specific receptor for serpin inhibitory complexes or cleaved inhibitors was pinpointed as a pentapeptide domain residing downstream from the reactive site (amino acid residues 370-374 (Phe-Val-Phe-Leu-Met)) (49, 50). It is highly possible that a similar neodomain is involved in chemotaxis, not only by a-1-PI' but also by a-1-Achy', since in the serpin superfamily this particular region is highly conserved. Such a suggestion is supported by data presented here which show that a-1-Achy' and a-1-PI' can block the chemotactic effects of each other, indicating that they probably interact with the same receptor on neutrophil cell surfaces, although it is difficult to understand how this can happen if, in both cases, the carboxyl-terminal peptide is buried within the modified protein. One possible explanation which has been suggested is that the dynamic structure of cleaved or complexed serpins in solution may differ slightly from those obtained through crystallographic analysis (50).
The chemotactic activity of a-1-Achy. may have great physiological importance, since it provides a mechanism for the attraction of neutrophils to inflammatory sites. However, it should be pointed out that the same receptors used by C5a could still be involved in a-1-Achy' binding, although this has not been investigated by us. In any case, the activity which we describe here is almost certainly important in the lung where studies of bronchial secretions of patients with chronic bronchitis have shown that the concentrations of a-1-Achy are higher than predicted by simple diffusion from blood plasma (51,52). This suggests that a-1-Achy may be synthesized within the lung, possibly by alveolar macrophages (53).
However, it has been recently shown that lung secretions containing a-1-Achy are low in inhibitory activity against chymotrypsin-like proteinases (54). This may indicate that a portion of the inhibitor has been proteolytically inactivated and, therefore, is partially responsible for the accumulation of neutrophils within the lung and in bronchial secretions in inflammatory states (55).
The rate of elimination of a-1-Achy' from the blood was studied to determine whether it is removed from the circulation by the same receptor which eliminates a-l-Achy-proteinase complexes (56). Such a rapid removal would presumably decrease its effectiveness as a neutrophil chemotactic factor. This was not the case, however, the clearance of the modified inhibitor being essentially identical with that of native a-1-Achy and consistent with that found for cleaved forms of AT 111, a-1-PI, and a-2-antiplasmin (25,57). With the exception of the latter inhibitor, each of these serpins in complex with its target proteinase is removed from the circulation by the same pathway (31-33, 56) utilizing a hepatic receptor referred to as serpin receptor 1 (56). However, a-2antiplasmin-proteinase complexes are eliminated by a separate pathway utilizing serpin receptor 2 (31)(32)(33). Since a-1-Achy' is not apparently recognized by serpin receptor 1, this receptor must be different from that utilized for the generation of neutrophil chemotaxis. Indeed, the data obtained confirm the distinct structural differences which must exist between modified and proteinase-complexed serpins (25) because both forms of a-l-PI are chemotactic (34, 58) but only the complex is rapidly removed through a specific receptor. Finally, it is likely that these results, while being obtained in a murine system, are applicable to humans since both a-1-PI (59) and AT I11 (60) in complex with their target proteinases are rapidly removed from the human circulation.
Among members of the serpin family, a-1-Achy would appear to be an excellent candidate for signaling during physiological stress, based on the following reasons. (a) It is extremely sensitive to proteolytic inactivation. (b) Both a-l-Achy:cat G complexes and a-1-Achy' can induce IL-6 synthesis and, indirectly, acute phase protein synthesis (47). (c) The plasma elimination time of the modified inhibitor insures a long-lasting stimulatory action on cells producing IL-6. (d) a-1-Achy' can act as a neutrophil attractant at inflammatory sites where it would be expected to exist in this form in high concentration and for long periods of time. It is tempting to suggest that permanent stimulation of IL-6 in individuals with malignant disease may be occurring because of their abnormally high circulating levels of a-1-Achy (61), most of which is in an inactive form: together with other acute phase plasma proteins. Indeed, the tumor cells can themselves act as a source of inactive a-1-Achy (62,63), the conversion presumably occurring in situ. However, the importance of this mechanism in promoting tumor growth remains to be established.