The α-Macroglobulin Bait Region

The amino acid sequence of a 90-residue segment of human pregnancy zone protein containing its bait region has been determined. Human az-macroglobulin, human pregnancy zone protein, and rat al-macroglobulin, a%-macroglobulin, and al-inhibitor 3 variants 1 and 2 constitute a group of homologous proteins; but the sequences of their bait regions are not related, and they differ in length (32-53 residues). The a-macroglobulin bait region is located equivalently with residues 666-706 in human tupmacroglobulin. In view of the extreme sequence variation of the bait regions, the evolutionary constraints for these regions are likely to differ from those of the remainder of the a-macroglobulin structure. The sites of specific limited proteolysis in the bait regions ad human pregnancy zone protein and rat a~-macrogl~bulin, az-macroglobulin, and alinhibitor 3 variants :L and 2 by a variety of proteinases differing in specificity have been determined and compared with those identified earlier in human a2-macroglobulin. The sites of cleavage generally conform to the substrate specificity of the proteinase in question, but the positions and nature of the P4-P4’ sites differ. Most cleavages occur in two relatively small segments spaced by 6-10 residues; and in each case, bait region cleavage leads to a-rnacroglobulin-proteinase complex formation. The rate at which a given proteinase cleaves a-macroglobulin bait regions is likely to show great variation. Possible !structural features of the widely different bait regions and their role in the mechanism of activation are discussed.

8 To whom correspondence should be addressed.
The active site of an aM-bound proteinase is accessible to small substrates and inhibitors. This feature and the ability of aMs to form complexes with a variety of proteinases from all classes  distin~ish aMs from most proteinase inhibitors.
The human azM subunit contains an exposed stretch called the bait region (Harpel, 1973;Barrett and Starkey, 1973) located near the middle of the polypeptide chain which is uniquely sensitive to cleavage by proteinases. Complex formation with a variety of proteinases is initiated by specific cleavage in that region, and the sites of cleavage have been determined for many proteinases Hall et al., 1981;Mortensen et al., 1981b;Virca et al., 1983;Sottrup-Jensen and Birkedal-Hansen, 1989). Bait region cleavage triggers conformational changes in the azM subunits resulting in the generation of tight-fitting binding sites for proteinases in the tetrameric structure. Maximally two proteinase molecules can be bound; and conceivably, each azM dimer contains one binding site.
The conformational changes initiated by bait region cleavage also cause activation of internal thiol esters formed from Cys-949 and Glx-952 in each subunit of azM (Sottrup-Jensen et al., 1980, 1984b. The activated thiol esters provide azM with a potential for covalent cross-linking of the activating proteinase through c-lysyl ( p~t e i n a~) -~-g l u t~y l (azM) bonds and also for binding of other nucleophils present at activation , 1981cSalvesen and Barrett, 1981).
As a further result of bait region cleavage, previously concealed recognition sites for receptors on a variety of cells including fibroblasts, macrophages, and hepatocytes (Debanne et al., 1975;Van Leuven et al., 1979;Gliemann et al., 1983) become exposed causing rapid clearance and degradation of azM-proteinase complexes from the circulation (Ohlsson, 1971;Imber and Pizzo, 1981). These features have been investigated in detail for human crzM and are shared by most members of the aM family.
Overall, the sequences of human aZM and PZP and of the rat proteins azM, aIM, and alIB are strongly related. Strik-G. Eggertsen and G. H. Fey, unpublished data.

15781
ingly, however, and first indicated from sequence analysis of human PZP (Sottrup-Jensen et at., 1984c), rat anM (Gehring et al., 1987), and rat a& (Braciak et al., 1988), the stretches presumed to function analogously to the human aZM bait region are dissimilar and of different length.
To further investigate the role of specific limited proteolysis in the activation of aMs, we have determined the bait region sequence of human PZP and report the sites of cleavage in that bait region and those of rat a2M and alM and the two isoforms of a113 by a set of proteinases differing in specificity. In conformity with results obtained earlier on human a,M (discussed by Roberts and Hall (1983)), we find that each bait region of different primary structure (32-53 residues) contains one or more sites at which specific limited proteolysis leading to complex formation and activation of the internal thiol esters takes place. Some clustering of cleavage sites in two major areas in each bait region is apparent. The dissimilar bait regions suggest different evolutionary constraints for different parts of the aM structure.

DISCUSSION
Amino Acid Sequence of PZP Bait Region-The partial bait region sequence determined earlier (Sottrup-Jensen et al., 1984c) was completed by analyzing overlapping CNBr fragments and tryptic peptides obtained from PZP or the PZPchymotrypsin complex (Figs. 1-3 and Table 1). The sequence of a 90-residue segment containing the bait region is shown in Fig. 4.
Sequence Diversity of a M Bait Region-The gross structure of the human a2M subunit and the extent of overall sequence similarity among human azM and rat aZM, alM, and are illustrated in Fig. 5. Scores of pairwise identity range from 75% for human a,M versus rat a2M to 55% for rat alM versus rat a& In the four aM sequences shown, 41% of all residues are conserved (56% when chemically similar residues are included). The aMs evidently constitute a family of homologous proteins resulting from divergent evolution. The bait region sequence of each aM is the only major segment of dissimilar sequence (protein and DNA level). Fig. 6 shows an alignment of a 47-residue stretch containing the human ~z M bait region with corresponding stretches in PZP and the rat aMs including the two variants of rat aI13. From this, we define the bait region of an aM as the stretch of highly variable sequence corresponding to the segment located between residues 666 and 706 in human azM flanked by segments of strongly conserved sequence. As shown in Fig. 6, the bait regions also differ in length, spanning 39 residues in human azM, 49 residues in human PZP, 53 residues in rat alM, 32 residues in rat azM, and 52 or 53 residues in rat a& variants 1 and 2, respectively.
It is likely that the evolutionary constraints of the bait regions, which constitute the segments involved in proteinase recognition, differ from those of the aM gross structure. This is reminiscent of the ovomucoid proteinase inhibitors (Laskowski et al., 1987) and the plasma proteinase inhibitors related to al-antitrypsin (Hill and Hastie, 1987). With regard to the evolution of the reactive sites and areas of proteinase contact, these protein families are thought to represent cases Portions of this paper including ("Materials and Methods," "Results," Figs, 1-4 and 7, Tables 1-5, and Footnote 4) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. of positive darwinian selection. The finding that the bait region sequence of rat aZM is encoded as a separate exon' indicates that the bait regions constitute elements of distinct genetic origin and that exon shuffling might be important in the evolution of aMs (Braciak et al., 1988).
Structural Features of (YM Bait Regions-With the exception of PZP, whose bait region contains only 3 charged residues, the bait regions are highly charged (human anM, 9 residues; rat alM, 6 residues; rat a2M, 8 residues; rat a113 variants 1 and 2,12 and 13 residues, respectively). Apart from the presence of one or more negative charges near the COOHterminal boundary, there appear to be no common patterns of charges. About one-third to one-half of the residues are hydrophilic. Furthermore, apart from rat a& and rat alia variant 2, Gly residues and, with the exception of rat azM, Pro residues are abundant. In examining the sequences using the procedures of Chou and Fasman (1978) and Garnier et al. (1978) (data not shown), no common pattern of a-helix and @-sheet is indicated. Several of the bait regions contain clustered hydrophobic residues, and hydropathic plots (Kyte and Doolittle, 1982) (Fig. 7) indicate the presence of a hydrophilic region separating two hydrophobic segments which largely correspond to the two major areas of proteolytic cleavage identified in each aM bait region (Fig. 6). The bait regions have been suggested to be irregular and flexible structures (Roberts and Hall (1983) and Sottrup-Jensen (1987)) in which even relatively hydrophobic stretches are readily accessible to cleavage (Fig. 6). It is presently not known whether the bait regions in spite of their sequence diversity assume a common gross structure.
Recent 'H NMR studies of the resonances of aromatic residues in human aZM (Gettins and Cunningham, 1986; Arakawa et at., 1986) support the contention that its bait region is flexible. Although restrictions in motility do exist, residues 683-700 appear to constitute a highly flexible surface M. Hattori, unpublished data.   Other spectroscopic studies suggest that the bait region of human a2M may be relatively compact, and it has been proposed from secondary structure predictions that the bait region forms a loop in which residues 680-685 and 695-700 form a short antiparallel &sheet looping out from the cuzM structure (Gettins et aL, 1988). However, that structure is not predicted from secondary structure analysis of other aMs (data not shown); and presently, there is no experimental evidence in this regard,  Virca et al. (1983) as summarized by Roberts and Hall (19831, we find that cleavage generally takes place at residues which are compatible with the known specificity of each proteinase. Evidently, each bait region contains one or more peptide bonds which can be cleaved by a wide selection of proteinases. However, due to the diversity of the bait region sequences, the P,4-P4' residues at the site(s) of cleavage differ greatly among aMs.

Localization of Cleauuge Site for Proteinases in Bait Region
In PZP-Staphylacoccw aurew proteinase and rat a2Mtrypsin, azM-chymotrqpsin, and rat cYzM-elastase, only one cleavage, which obviously constitutes the activating event, occurred. However, in most other aM-proteinase complexes, multiple cleavages were identified; and these could be the result of a single primary cleavage (activating event) followed by one or more secondary cleavages, or they could be the result of random cleavages, each giving rise to activation of the CUM. Clear-cut examples of sequential cleavage are provided by PZP-ch~otrypsin and rat CUIM-chymot~sin, in which a single cleavage occurring in the NHz-terminal part of their bait regions is followed by secondary cleavages toward the COOH terminus see ("Results"). In contrast, trypsin apparently cleaves rat alM randomly at two nearby sites, as also seen for rat alM-human fibroblast collagenase (Sottrup-Jensen and Birkedal-Hansen, 1989). The identification of several adjacent or closely spaced cleavage sites, e.g. in PZPelastase and PZP-thermolysin and in rat azM-papain, suggests that random cleavage may occur in many cases.
The maximal distances between sites cleaved by the proteinases studied here are 20 residues (rat CU~M), 25 residues (human a2M), 26 residues (rat al13), 31 residues (human PZP), and 39 residues (rat qM). Whether proteolytic cleavage may occur at any position within these segments is not known. Most cleavage sites identified are located in two segments spaced by 6-10 residues, largely corresponding to the two major areas of cleavage (residues 681-686 and 696-700) identified earlier in human azM (Mortensen et aL, 1981b). In PZP which contains a refractory Arg-Pro bond, very slow cleavage by trypsin and thrombin apparently took place at one or more of the chymotrypsin-sensitive bonds, suggesting that nonspecific cleavage may happen in some instances.
As judged from the high rate of interaction between human apM and proteinases such as trypsin, chymotrypsin, and pancreatic and neutrophil elastases (overall second-order rate a-Macroglobulin Bait Region Cleavage Sites constant for complex formation, >1 X lo7 M-' 5-l) (Barrett and Salvesen, 1979;Christensen and Sottrup-Jensen, 1984;Virca and Travis, 1984;Bieth and Meyer, 1984), the bait region can serve as an excellent substrate. Qualitative data (see "Results") indicate that the bait regions of PZP and the rat aMs may also be rapidly cleaved by several proteinases.
This suggests that segments of the bait regions are readily accessible at the surface of the aM structure (Roberts and Hall, 1983). However, for other proteinases including thrombin, factor X,, and urokinase, the human a2M bait region is a relatively poor substrate (second-order rate constants, <1 X lo4 M" s-') (Downing e t al., 1978;Ellis et al., 1982;Straight et al., 1985); and in some cases, e.g. factor XII, and activated complement factor Cls, no reaction has been demonstrated (Sim et al., 1979;Chan et ul., 1977). As discussed earlier (Sand et ab, 1985;Sottrup-Jensen and Birkendal-Hansen, 1989) and evident from the present data (Tables 2-5), rates of interaction with a given proteinase vary among cuMs. The structural features responsible for this are not known. Activation of aMs by Specific Limited Proteolysis-The aMs are homologous multidomain proteins activated by specific limited proteolysis. Activation generally results in gross conformational changes, activation of internal reactive thiol esters, and exposure of receptor recognition sites. The fact that the activating cleavage can occur at different positions in a relatively long stretch of residues of unique sequence in aMs is intriguing.
In considering different ways in which an "activation signal" can be generated, the following may be considered transfer of newly generated NHZor COOH-terminal charges to specific internal positions, disruption of bait region structure exposing underlying parts of structure, and release of strain ("spring effect"). At present, no experimental data allow a distinction between these possibilities. In view of the similarity of the aM bait regions and the reactive site loops of the proteinase inhibitors related to al-antitrypsin, which show considerable sequence variability and which, in several cases, contain multiple sites for proteolytic cleavage (Carrell and Owen, 1985;Carrell et al., 1987;Benda et al., 1987;Vissers e t al., 1988), a mechanism based on the relief of strain (Loebermann et al., 1984) may be appealing.
However, in contrast to what was implicitly stated in the "classical" trap hypothesis of Barrett and Starkey (1973), we suggest that the conformation of a native cuM does not depend on the bait region being strained. Rather, the native aM structure is specified and maintained by interactions between a number of other domains. Perhaps the flexible bait region shields underlying structural elements; and upon cleavage of at least one peptide bond, the structure is forced into a new conformation by upsetting a delicate balance in this metastable assembly of domains. The fine details of this system are still elusive.