A functional, thioester-containing alpha 2-macroglobulin homologue isolated from the hemolymph of the American lobster (Homarus americanus).

An alpha 2-macroglobulin-like protease inhibitor was isolated from the cell-free hemolymph of the american lobster (Homarus americanus) by ion-exchange chromatography and gel filtration. Whereas the undissociated molecule has a molecular weight of 342,000 as determined by ultracentrifugation studies, under reducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis, the protein has a subunit molecular weight of 180,000. On the basis of this and other evidence, we conclude that the lobster protein is a dimer consisting of two disulfide-bonded monomers. The purified protein inhibits proteolytic enzymes but protects the esterolytic activity of trypsin toward low molecular weight substrates from inactivation by soybean trypsin inhibitor. The methylamine sensitivity of this activity suggests the presence of an internal thioester bond. This was confirmed by the covalent incorporation of [14C]methylamine, by the formation of Mr 55,000 and 125,000 autolytic cleavage fragments in sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and, more directly, by the amino acid sequence of a tryptic peptide containing the putative thioester region. Whereas the N-terminal amino acid sequence (22 residues) of the protein revealed an overall identity of only 18% when compared with the human protein, the sequence of the thioester-containing peptide was highly conserved, both with respect to human alpha 2-macroglobulin and to other proteins having a thioester bond. The protein showed the "slow to fast" conformational change typical in alpha 2-macroglobulins in nondenaturing gel electrophoresis after treatment with trypsin, but not after incubation with methylamine.


125,000 autolytic cleavage fragments in sodium
dodecyl sulfate-polyacrylamide gel electrophoresis, and, more directly, by the amino acid sequence of a tryptic peptide containing the putative thioester region. Whereas the N-terminal amino acid sequence (22 residues) of the protein revealed an overall identity of only 18% when compared with the human protein, the sequence of the thioester-containing peptide was highly conserved, both with respect to human a2-macroglobulin and to other proteins having a thioester bond. The protein showed the "slow to fast" conformational change typical in a*-macroglobulins in nondenaturing gel electrophoresis after treatment with trypsin, but not after incubation with methylamine. tu,-Macroglobulin (a,M)' is a high molecular weight protease inhibitor found in the plasma of vertebrates (1). It has the unique property of binding and inhibiting the great majority of endopeptidases, regardless of their specificity or catalytic mechanism (2, 3). It has been proposed that n2M inhibits the proteolytic activity of these proteases by a trapping mechanism in which the active site of the enzyme remains active and free to hydrolyze those substrates which are small enough to pass through the molecular cage with which * This work was supported by Natural Sciences and Engineering Research Council of Canada Grant A0263. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. the azM molecule envelops the protease. The caged protease may be bound convalently by transacylation from a sequestered thioester in azM. This structure is formed between cysteinyl and glutamyl side chains (4) which are separated by only two amino acids ( 5 ) . In addition to azM and several closely related protease inhibitors (6-8), a thioester is also present in the complement proteins C3 and C4 (9). Two noteworthy characteristics of the protein thiolactone ring are the covalent incorporation of primary amines such as ammonia and methylamine (10) and its ability to mediate a denaturation-induced autolysis of the polypeptide chain (11). The mechanism of trapping involves the cleavage of a peptide bond in the so-called "bait region" of the azM subunits (12, 13) which contains potential cleavage sites for most proteases and which is evidently suitably exposed. An immediate consequence of bait region proteolysis is the exposure and cleavage of the thioester bond by protease, nucleophilic amino groups, or water molecules. Thioester scission results in spectroscopically and electrophoretically demonstrable conformational changes (14, 15), which for azM, presumably results in the formation of the molecular cage around the protease. In all three human thioester-containing proteins, the same conformational changes were shown to occur when the thioester was cleaved by nucleophiles in the absence of proteolysis.
The most studied azM is that of the human. The native inhibitor is a tetramer (Mr -720,000) of identical subunits (Mr -180,000). Two dimers ( M , -360,000), each of which is composed of two covalently bound subunits, are noncovalently associated in the native molecule (16,17). Each dimer is able to bind one protease molecule (18). Functional evidence for the existence of a similar protease inhibitor in the horseshoe crab Limulus polyphemus has been published (19); and recently (20), a protein having similar protease inhibitory properties to n2M has been purified from the horseshoe crab. This molecule has a molecular weight by gel filtration of approximately 550,000 and was thought to be trimeric in structure.
In this paper, we describe the isolation and characterization of a thioester-containing protease inhibitor from the hemolymph of the lobster (Homarus americanus) with properties similar to human a2M but consisting of only two disulfidelinked subunits.   2). A yield of about 20 mg of protein was obtained from 100 ml of hemocyte-free lobster hemolymph. The purified protein was homogeneous in polyacrylamide slab gel electrophoresis under nonreducing conditions (Fig. 3). Under reducing conditions, an additional band with a lower apparent M , than the main band was visible (Fig. 3). The same phenomenon was observed for human a2M which was used as a reference protein.
Carbohydrate Analysis-Periodic acid-Schiff stain of the SDS gel showed the lobster protein to be glycosylated, and gas-liquid chromatographic analysis of a hydrolyzed sample of lobster a2M showed 5.1% (w/w) carbohydrate with mannose, N-acetylglucosamine, and fucose as the major constituent monosaccharides. Compared to human a2M, reported to contain 8.6% carbohydrate with N-acetylglucosamine, mannose, galactose, sialic acid, and fucose as the major sugars (21), the lobster protein has less carbohydrate, and galactose and sialic acid appear to be absent in the lobster protein.
Subunit Structure of a,M-SDS-PAGE under reducing conditions yielded a subunit molecular weight for lobster a2M of approximately 180,000, a value identical to that of the human species (Fig. 3). Under nonreducing conditions, the protein moved more slowly than human a2M, suggesting a higher molecular weight estimated to be approximately 440,000. The determination of the sedimentation coefficient gave, for the lobster protein, an s20,w of 11.6, which is significantly smaller than the s20,u. value of 18.2 that we obtained for the human protein. Making the assumption that, like human a2M, the lobster protein is globular and therefore that the sedimentation would be proportional to (MJ2'13, a molecular weight of approximately 360,000 would be predicted for lobster aPM. More accurate estimates of the molecular weight were obtained by sedimentation equilibrium ultracentrifugation measurements where values of 332,000 and 352,000 were obtained in runs a t 8,547 and 11,299 rpm, respectively, giving an average absolute molecular weight of 342,000. Taken together, these experiments suggest that lobster n2M is a dimeric structure consisting of two monomeric, M, 180,000 polypeptide chains linked by disulfide bonds.
Amino Acid Sequence Analysis-N-terminal amino acid sequence analysis was performed on two independent 1.6nmol samples to determine whether more than one class of subunit was present. The results are shown in Table I. The amino acid sequence was homogeneous, and the initial yield of 60% indicates that the subunits are identical.
Evidence for an Internal Thioester Bond-Since autolytic fragmentation accompanying heat denaturation is one of the hallmarks of thioester-containing proteins (9), this was the initial approach used to demonstrate the presence of a thioester structure in lobster a2M. The protein was boiled for 5 min in SDS sample buffer and then reduced with P-mercaptoethanol. The SDS-PAGE analysis depicted in Fig. 4 shows that, besides uncleaved a2M ( M , 180,000), two additional bands were present which were not evident when a2M was pretreated with methylamine. The fragments had M, values of 125,000 and 55,000, respectively, which are quite similar to the human M , 120,000 and 60,000 autolytic cleavage fragments.
Further evidence for the internal thioester in lobster a2M comes from the incorporation of ['4C]methylamine into the protein as revealed by SDS-PAGE and autoradiography of the gel (data not shown). Finally, the protein was treated with ["C)methylamine, and the protein was allowed to bind covalently to activated thiopropyl-Sepharose through the released sulfhydryl group. The complex was digested with an excess of trypsin, the residual bound peptide was eluted with buffer containing 8-mercaptoethanol, and the "C-labeled peptide was further purified by reversed-phase high performance liquid chromatography and sequenced. The sequence shown in Table I1 is consistent with that expected for an internal thioester.
Conformational Change of Lobster a2M after Reaction with Trypsin or Methylamine-Reaction of human a2M with pro-

Invertebrate Thioester Protein cYz-Macroglobulin
tease or methylamine results in a faster migrating species in polyacrylamide gel electrophoresis under nondenaturing conditions. This phenomenon is known as the "slow to fast" conformational change (15). Lobster azM migrates faster than the human species in nondenaturing gels with a migration rate corresponding to that of human half-molecules (Fig. 7). Pretreated with methylamine, lobster a2M does not show any change in migration. However, either native or methylaminetreated lobster azM does show the slow to fast conformational change after incubation with trypsin. Inhibitor Activity of Lobster a2M-Two assays were carried out to show the inhibitor activity of lobster a2M. In the amidolytic N-benzoyl-DL-arginine-p-nitroanilide assay, as described in the Miniprint Section, a corrected value for trypsin binding of 18.1 pg of azM/pg of trypsin was found. This corresponds to 0.79 mol of trypsin/mol of dimeric azM, assuming molecular weights for lobster aqM and trypsin of 342,000 and 23,800, respectively.
In the [14C]casein assay, the amount of a2M which protects [I4C]casein from cleavage by 1 pg of trypsin was about 17 pg. When corrected, this corresponds to a molar ratio of 0.75 mol of trypsin/mol of dimeric a2M, which is in good agreement with the result obtained from the amidolytic assay. Table I11 shows results obtained for a variety of proteases tested in the [I4C]casein assay. Lobster apM inhibits all the proteases tested.

DISCUSSION
Assays of lobster hemolymph with trypsin and soybean trypsin inhibitor using large and small molecular weight substrates indicated the existence of protease inhibitor activity with properties similar to those of human a2M. The purification of this invertebrate protease inhibitor was achieved using ion-exchange chromatography and gel filtration. The concentration in hemolymph was estimated by radial immunodiffusion to be about 60 mg/100 ml. Harpel and Brower (22) have commented on the seemingly high concentration of a Z M in human serum (250 mg/100 ml), pointing out that, because of the large molecular weight, this represents a concentration of only 3.5 p~, close to that of most other serum protease inhibitors. The molar concentration of a2M in lobster hemolymph is about 1.8 p~. Measurements on several batches of lobsters purchased at different times showed this concentration to be fairly constant.
SDS-PAGE under reducing conditions gave a subunit molecular weight of 180,000, identical with that of the human protein. Attempts to determine an accurate molecular weight of the native form using gel filtration and SDS-PAGE with human apM and standard proteins as references did not give results which would permit a designation of a di-, tri-, or tetrameric structure composed of M , 180,000 monomer subunits. We therefore determined the molecular weight of the native inhibitor by ultracentrifugation. The molecular weight value obtained (342,000) favors a dimeric structure for the lobster inhibitor composed of disulfide-bonded monomers. This result is consistent with the sedimentation coefficient of 11.6 S, which is close to the value reported for dimeric species of human a2M prepared by acid dissociation or reduction (23,24) and for dimeric a2M of the plaice (25). This is probably the first description of a natural dimeric aZM homologue with disulfide-bonded monomers since, in the plaice and the southern grass frog (26), the subunits are noncovalently associated. Apart from the anomalously high molecular weight estimate of 440,000 from nonreducing SDS-PAGE, which might be due to different glycosylation, no evidence was found for a trimer species such as that which occurs in the horseshoe crab (20). N-terminal amino acid sequence analysis of the inhibitor gave a single sequence, and the initial yield of 60% indicated that the dimers are composed of identical monomers. Comparison of the sequence of the lobster protein with known NH2-terminal sequences of vertebrate inhibitors revealed striking differences (Table I). Only 4 residues out of 22 are identical. The conserved proline in position 6 is located in position 7. The conserved tyrosine in position 8 is absent. Met-9 is conserved; but there is a tryptophan, usually a conservative residue, which is not found in the vertebrate proteins in position 10. With the exception of Ser-14 and Thr-20, the residues following Trp-10 differ from those in other species, including the normally conserved proline in position 13. The overall homology is therefore about 18%, which is perhaps an indication that this region is not important for the function of the inhibitor. Differences in the primary structure can be expected considering the fact that we compare proteins from species which diverged a considerable time ago. On the other hand, the partial sequence analysis of a peptide containing the thioester region which is important for the inhibitor activity of the protein is seen to be highly conserved when compared with the corresponding region in human azM (Table 11). Besides the amino acid residue in position 12, all the residues are identical, including the 2 critical amino acid residues, cysteine and, presumably in the nascent chain (9), glutamine, from which the thioester is formed. We did not confirm position 8 as glutamic acid because the yield of the phenylthiohydantoin-derivative was too low for a back hydrolysis and subsequent amino acid analysis. But the circumstantial evidence provided by the [14C]methylamine label and the elution position provides a strong argument for glutamine in position 8 of the peptide.
The autolytic cleavage of a2M which occurs at the thioester site gave, under reducing conditions on SDS-PAGE, two fragments with molecular weights which differed slightly from those of the fragments obtained from the autolytically cleaved human inhibitor. The thioester seems to be shifted by about 5000 daltons, corresponding to about 45 amino acid residues, toward the C-terminal end of the subunits. Alternatively, these differences might be related to the differences in the glycosylation of the two proteins.
Quigley and Armstrong (20) reported that the inhibitory activity of horseshoe crab a2M is lost after methylamine treatment. This indication that the thioester bond occurs in proteins in invertebrates has been fully realized with the demonstration of the sequence and of the autolytic cleavage in lobster a2M, as well as through the incorporation of methylamine and the concomitant release of a free sulfhydryl which binds to activated thiopropyl-Sepharose. Experiments with lobster a,M on gel electrophoresis under nondenaturing conditions confirmed the fact that the lobster protein is smaller than the human protein. The faster migration of the invertebrate protein is not due to differences in the isoelectric points as both proteins focused in the same region of a pH 5.0-7.0 gradient gel (data not shown). Methylamine or protease treatment leads, in human a2M, to an increase in electrophoretic mobility. The lobster protein shows the slow to fast change only after protease treatment. Methylamine treatment does not produce a conformational change detectable with nondenaturing gel electrophoresis (similar to bovine a2M (27)). In addition, methylamine-treated a2M has a tendency to aggregate and remain at the origin, and hence the bands in Fig. 7 appear faint.
Inhibition assays (N-benzoyl-DL-arginine-p-nitroanilide and [14C]casein) show the ability of lobster azM to inhibit a broad variety of proteases. The N-benzoyl-DL-arginine-p-ni-troanilide and [14C]casein assays gave values of 0.79 and 0.75 trypsin molecules bound per aZM molecule, respectively. It can therefore be concluded that the ratio is one protease molecule/inhibitor molecule and therefore different from the stoichiometry found for human a2M half-molecules (24) and for the plaice and southern grass frog (25, 26) where values in the range of 0.45 and 0.53 mol of trypsin/mol of macroglobulin were found.
In human a2M, methylamine treatment produces a slow to fast conformational change, and the protein loses completely its ability to inhibit trypsin. In the case of lobster azM, methylamine treatment does not induce a similar conformational change, but its ability to inhibit proteases is nevertheless lost. This behavior is in marked contrast to that found for bovine a2M, which also shows no shape change but remains a t least partially active following methylamine treatment. The kinetics of the conformational change accompanying proteolytic cleavage of human a Z M , as well as of C3 and C4, appears to be rate-limited by the proteolytic event per se. By contrast, the presence of an uncleaved polypeptide chain has been shown to retard, to variable degrees, the kinetics of the conformational change following methylamineinduced scission of the thioester bond in these proteins (28)(29)(30). Whereas the half-time for completion of the conformational transition at 30 "C is about 1.2 min in cuzM, it is about 13.9 min in C4 and 111 in C3 at 37 "C. As mentioned above, the tendency for the methylamine-treated a2M to aggregate makes an accurate interpretation of these observations difficult; however, lobster aZM may represent an extreme example of the above-mentioned retardation phenomenon. Nevertheless, although proteolysis will produce the slow to fast transition in methylamine-treated lobster azM, the transition does not result in trapping of the protease, suggesting that a different conformational pathway is followed when thioester cleavage precedes proteolysis in the bait region. 1 I