Ovostatin: A Novel Proteinase Inhibitor from Chicken Egg White I. PURIFICATION, PHYSICOCHEMICAL PROPERTIES, AND TISSUE DISTRIBUTION OF OVOSTATIN*

A proteinase inhibitor which has strong anti-colla-genase activity was found in chicken egg white. The inhibitor (PI = 4.9) was purified by poly(ethy1ene gly-col) (5.5-10%) precipitation and chromatography on Ultrogel AcA 34, DEAE-cellulose, and Sephacryl S-300. The final product was homogeneous on 5% poly- acrylamide gel electrophoresis. Stoichiometric inhibition was observed with the inhibitor and rabbit syn- ovial collagenase and thermolysin (1:l molar ratio with thermolysin). The inhibitor ran on sodium dodecyl sulfate-gel electrophoresis with reduction as a single protein band of M, = 165,000. The molecular weight of the native inhibitor was estimated to be 780,000 by sedimentation equilibrium centrifugation. Centrifu- gation analysis in 6 M guanidine hydrochloride and of the reduced sample gave Mw = 380,000 and Mw = 195,000, respectively, where 161, is the weight-aver- age molecular weight determined by equilibrium ultracentrifugation. The results indicated that the inhibitor molecule is a tetramer of identical subunits linked in pairs by disulfide bonds. Since the molecular weight and the quaternary structure of the inhibitor were similar to those of az-macroglobulin (azM) in plasma, chicken azM was isolated and compared with the inhibitor. The inhibitor was not sensitive to

Numerous protein inhibitors of proteinases have been described in many animals, plants, and microorganisms (1,2). A relatively small number of these inhibit metalloproteinases despite the fact that many biologically important metalloproteinases have been described in various mammalian tissues (see Ref. 3 for review). Several types of connective tissues (4-9) and cells (10,11) in culture produce inhibitors (Mr - 28,000) which block the activity of tissue collagenase. Recent studies by Cawston et al. (12) with the inhibitor isolated from rabbit bone culture medium showed that it also inhibits other tissue metalloproteinases that degrade gelatin and proteoglycans as well as human leucocyte collagenase and gelatinase. It does not, however, inhibit thermolysin or bacterial collagenase (12). An inhibitor with similar properties was purified from human amniotic fluid (13). Collagenase inhibitors with M, -40,000 (B,-anticollagenase) were isolated from human and dog serum (14,15). It has been also reported that platelet factor IV (16) and extracts of cartilage (17,18) and aorta (19) inhibit collagenase activity. However, specific metalloproteinase inhibitors described previously are only available in limited amounts in pure form. Therefore, using mammalian collagenase as a test enzyme, we have sought such an inhibitor in sources available in large quantities.
We have found a strong anti-collagenase activity in chicken egg white. An inhibitor protein was isolated, and a study of the mechanism of inhibition, described in our following paper (20), has shown that it is similar to that of azM1 in plasma (21). Immunochemical studies using a monospecific antibody raised against the inhibitor have indicated that the protein is synthesized in the oviduct and secreted into egg white but that it is not present in plasma or in other tissues of chickens. The inhibitor, therefore, has been named "ovostatin." In the present paper we describe the purification, physicochemical properties, and tissue distribution of this novel proteinase inhibitor and the differences between ovostatin and chicken plasma azM established by chemical and immunological criteria.

RESULTS
Purification of Ouostatin-Ovostatin was originally detected as an inhibitor of mammalian collagenase in egg white. Preliminary purification studies with gel permeation chromatography on Ultrogel AcA 34 showed, however, that fractions inhibiting collagenase activity coincided exactly with those inhibiting the proteolytic activity of thermolysin with azocasein as substrate (see Fig. 1). For convenience, therefore, the inhibitor was monitored by its capacity to inhibit thermolysin.
The results of the purification of ovostatin are summarized in Table I. A decrease in the viscosity of egg white was achieved simply by diluting it with an equal volume of H20, which made the subsequent poly(ethy1ene glycol) precipitation step effective. Poly(ethy1ene glycol) fractionation (5.5-10%) enriched ovostatin 4-fold with a 70% recovery (Fig. 4, lane 2). Analysis of the precipitate formed with 0-5.5% poly(ethy1ene glycol) by gel electrophoresis showed that about 20% of the ovostatin was precipitated in this fraction. Because most viscous materials were co-precipitated in this fraction and subsequent purification by chromatographic techniques was troublesome we discarded this precipitate.
Fifty ml of the ovostatin-rich fraction was applied to a column (4.5 X 55 cm) of Ultrogel AcA 34. The inhibitor eluted near the void volume of the column, well removed from the majority of proteins (Fig. 1). The yield at this step was about 85% with 6-fold purification. After this step, only a few other protein contaminants were detected on gel electrophoresis (Fig. 4, lane 3).
The ovostatin fraction from the Ultrogel AcA 34 separation was then applied to a DEAE-cellulose column for further purification. About 80% of the ovostatin bound to the column and was eluted with an NaCl gradient (Fig. 2). Although the recovery at this step was poor (38%), a 3-fold purification was   (1.4 pmol) was incubated with ovostatin at various molar ratios (total 300 p l ) at 23 "C for 10 min. The mixtures were then assayed at 23 "C for the residual proteolytic activity of thermolysin using azocasein as substrate (0). Ovostatin treated with 0.2 M CH3NH2 at pH 8.6 for 1 h at 23 "C was also tested for its inhibitory capacity against proteolytic activity of thermolysin (0). obtained and the major contaminants were removed (Fig. 4,  lane 4).
The final purification was achieved by gel permeation chromatography with a column of Sephacryl S-300 (Fig. 3), yielding a preparation shown to be homogeneous on polyacrylamide gel electrophoresis (Fig. 4, lane 5). About 50 mg of ovostatin was obtained with 159-fold purification from 450 ml of egg white. The overall yield of the inhibitor was 17%.
The isoelectric point of ovostatin is 4.9 (Fig. 5). The final preparation of ovostatin showed stoichiometric inhibition of rabbit synovial collagenase (Fig. 6) and thermolysin activity (Fig. 7) using reconstituted collagen fibers and azocasein, respectively, as substrates. As shown in Fig. 7, ovostatin bound to one molecule of thermolysin, but a binding ratio of ovostatin and collagenase was not established since there is no method available to titrate active collagenase. CH3NH, treatment of ovostatin at pH 8.6 did not change the electrophoretic mobility of the molecule on 5% acrylamide gel (Fig. 4, lane 6), nor did it destroy the proteinase inhibitory capacity of ovostatin (Fig. 7).
SDS-Polyacrylamide Gel Electrophoresis of Ovostatin-Analysis of purified ovostatin by SDS-gel electrophoresis in the presence of 5 M urea is shown in Fig. 8. A single protein band was resolved with or without reduction. The apparent molecular weight of ovostatin with reduction was 165,000. Ovostatin without reduction ran near the top of the gel in this system, making it inappropriate to estimate with precision the M , of unreduced ovostatin by this technique.
A single protein band was observed after reduction, suggesting that the native ovostatin molecule consists of a t least two similar polypeptide chains possibly linked by disulfide bonds. Ultracentrifugation Studies-To estimate the molecular weight of native ovostatin the protein was subjected to sedimentation equilibrium ultracentrifugation studies. The M w 3 of the native molecule was estimated to be 780,000, with precision of f 5%. Ovostatin centrifuged in 6 M guanidine hydrochloride gave a value of M u = 380,000, approximately one-half of the M w of the native molecule. M w of reduced ovostatin was 195,000, one-quarter that of the native molecule. These results together with SDS-gel electrophoretic analyses (Fig. 8) suggest that the native ovostatin molecule is a tetramer of similar subunits of which two pairs are linked by disulfide bonds (dimers) and assembled noncovalently. Observation of a single NHz-terminal sequence (see below) suggested that the subunits are identical in primary structure. M u . = 780,000 was used to calculate the molar concentration of ovostatin.
Dissociation of the two dimers noncovalently associated was demonstrated further by sedimentation rate studies in the presence of urea. Sedimentation rate analysis of native ovostatin showed a single symmetrical peak with szo,w = 16 (Fig. 9A). In the presence of 4 M urea, tetramers were partially dissociated into dimers (Fig. 9B) and in 6 M urea almost complete dissociation into dimeric forms was observed (Fig.  9C). s~, ) , ~ for the dimers and the quarter-subunits (reduced ' lii, is used for the weight-average molecular weight determined and carboxymethylated ovostatin) were 10 and 6, respectively. Stability-Ovostatin was labile at acidic pH values, but relatively stable at alkaline pH values; after dialysis at pH 4.0 for 16 h a t 4 "C, the inhibitory capacity of ovostatin was completely lost. Following similar treatment at pH 5.0, about 50% of the activity was recovered. The protein was stable in the range of pH 5.5-10. About 90% of the activity was recovered after dialysis against pH 10.5 buffer. Heating ovostatin at 55 "C for 30 min and 80 "C for 5 min destroyed about 20 and 95% of activity, respectively. However, the protein was stable at 43 "C for a t least several hours.
Comparative Studies of Ovostatin and Chicken aaM-Ovostatin has been shown to be a proteinase inhibitor with M, = 780,000. Among recognized proteinase inhibitors (see Refs. 1 and 2 for review) only a plasma protein, a2M, has a comparably large Mi,,. (28). Because of the close similarity in molecular weight and its quaternary structure (the a2M molecule is a tetramer of four identical subunits ( M , = 185,000) linked in pairs by disulfide bonds (40, 41)), the two proteins were compared chemically and immunologically. Much information is available on human azM (27, 40), but we are unaware of any reports concerning chicken azM. Therefore, studies of chicken azM were required to compare the two proteins. The purification of chicken azM was performed essentially as described by Barrett et al. (27), except that preparative isoelectric focusing was employed for the final step. The product was homogeneous on polyacrylamide gel electrophoresis ( E, double immunodiffusion. The plates contained; well A, sheep anti-(ovostatin) serum; wells I , 3, and 12, 5 pg of ovostatin; well 2, oviduct extract; well 4, pancreas extract; well 5, sheep serum (nonimmune); wells 6 and 7, egg white diluted &fold with water; well 8, heart extract; well 9, liver extract; well 10, skeletal muscle; and well 11, aorta extract. The diffusion was for 45 h at 23 "C, and the plates were washed, dried, and stained for protein as described by Barrett (38). and chicken a2M are similar, a2M ran faster. This may reflect a difference in the Stokes' radii of these proteins.
The two proteins were shown to have no immunological cross-reactivity (Fig. 12). Although several materials in egg white formed precipitin lines with rabbit anti-(chicken serum) serum, purified ovostatin was not recognized by this antiserum (Fig. 12A). Monospecific sheep anti-(ovostatin) serum showed that plasma from both cockerels and hens lack crossreactive materials (Fig. 12B). Reduced and alkylated ovostatin failed to cross-react with anti-(ovostatin) antibody raised against the native protein (Fig. 12B).
The proteinase inhibitory capacity of human ( Y~M is known to be destroyed by treating with CH3NH2 (27, 44,45). Therefore, the sensitivity of ovostatin to CH3NH2 was tested as was that of chicken a2M. The thermolysin inhibitory activity of ovostatin was not altered after treatment with 0.2 M CH3NH2 at pH 8.6 for 1 h (Fig. 7), nor was the mobility on 5% gel electrophoresis (Fig. 4, lane 6), whereas the inhibitory capacity of chicken a2M, as with human a2M (27), was destroyed by CH3NH2 (Fig. 13).
NH2-terminul Sequence of Ouostatin-Comparison of CNBr fragments of ovostatin and those of a2M showed distinct peptide maps, suggesting that the two molecules have different primary structures (Fig. 14). We then examined the NH2terminal sequences of these proteins. The sequences for the first 13 residues of ovostatin and chicken a2M are shown in Fig. 15 along with the corresponding sequence for human a2M reported by Swenson and Howard (41). The yields of phenylthiohydantoins identified in NH2-terminal sequencing of ovostatin and chicken a2M are shown in Table 11. Lysine was identified as the NH2-terminal residue of ovostatin, whereas serine was the NHz-terminal residue of chicken a2M. Considerable homology in the NH2-terminal sequences among ovostatin, chicken a2M, and human azM was noted among the first 13 residues, 5 were found to be identical with those of chicken apM and of human a2M, and 8 residues of the two a2Ms were identical in the same region (Fig. 15).
Tissue Localization of Ouostatin-Double immunodiffusion tests indicated that chicken plasma does not contain materials immunologically reactive against anti-(ovostatin) antibody. We then extended our immunological studies to examine whether or not ovostatin was present in various other chicken tissues. More sensitive "rocket" immunoelectrophoresis was employed. As shown in Fig. 16, ovostatin concentrations as low as 5 pg/ml were detected. The height of "rockets" and the ovostatin concentrations showed a good correlation (Fig.  16A); an empirical standard curve was used to quantitate ovostatin in the tissues.
The concentration of ovostatin in egg white was estimated to be 0.5 mg/ml (average of 12 eggs). The extract of oviduct contained 32 pg of ovostatin/ml (320 pglg, wet weight, of tissue). However, other tissues tested (aorta, cartilage, egg yolk, heart, intestine, liver, pancreas, skeletal muscle, and skin) contained no detectable amount of the protein (<50 pg/ g, wet weight, of tissue). Although liver and pancreas showed a long uncharacteristic streak on "rockets," double immunodiffusion plates indicated that this represented nonspecific precipitation (Fig. 16B).
Thus, immunochemical studies to localize ovostatin have indicated that the protein is synthesized by the oviduct and accumulates in egg white.
Since ovostatin has a M u of 780,000, distinctively larger than most other proteins in egg white, the purification of the inhibitor was accomplished in relatively few steps. A Purification of about 160-fold was required to obtain pure ovostatin. This agrees with the concentration of ovostatin in egg white estimated by "rocket" immunoelectrophoresis as approximately 0.5 mg/ml in a total protein concentration of about 80 mg/ml (Table I).
A 1:1 molar binding ratio of ovostatin and proteinase was established for thermolysin. Particular attention was also given to whether this ratio was an underestimation due to partial inactivation of ovostatin during the purification steps. Examination of 5 different preparations of ovostatin gave the same 1:1 binding ratio. One of these was prepared in the presence of 1 mM phenylmethylsulfonyl fluoride, 0.1 mM Ep-475 (~-trans-epoxysuccinylleucylamido-(3-methyl~butane, a cysteine proteinase inhibitor (53)) and 5 mM EDTA for the initial poly(ethy1ene glycol) precipitation step. As mentioned above, there is no method available to titrate active collagenase to arrive at an exact number for the binding ratio of the enzyme to ovostatin. However, since the M , of collagenase is 46,000 (54) the specific activity of rabbit synovial collagenase was 27,200 units/mg at 37 "C, assuming a binding ratio of 1:1. This is comparable to the specific activity of 53,000 units/mg for pig synovial collagenase (55) and 28,000 units/mg for rabbit bone collagenase (56).* In contrast, it has been reported that human azM binds to two molecules of trypsin (27, 57), elastase (58), or chymotrypsin (59), or to one molecule of plasmin (60). Unlike human aZM, Fig. 14 indicated that chicken azM binds one molecule of trypsin. We do not know, however, whether this binding ratio is due to partial inactivation of chicken aZM or due to the binding capacity of chicken azM. Future studies should resolve this point.
The native ovostatin molecule has been shown to be a tetramer of identical subunits linked in pairs by disulfide bonds. This is supported by the results of sedimentation equilibrium ultracentrifugation studies which provided values The physical properties of ovostatin resemble those of the high molecular weight globulin in egg white originally called "component 18" or "line 18" because of its position on starch gel electrophoresis (61). Later, this protein was named "ovomacroglobulin" by Miller and Feeney (62). Donovan et al. (63) reported that the native ovomacroglobulin has a M u of 640,000 with = 17.0. Low pH or 6 M guanidine hydrochloride dissociated the molecule into subunits of M w = 310,000 with s&,, = 10.7 (63). However, the M w of the reduced ovomacroglobulin was 56,000 to 68,000 (63), considerably smaller than that of the reduced ovostatin. Nevertheless, it is likely that ovomacroglobulin and the ovostatin described in this paper are identical, since there are not many other egg white proteins with molecular weights in this range and the subunit arrangement of each is similar. It has been reported that ovomacroglobulin has a wide spectrum of immunological cross-reactivity among avian egg whites (62,64). However, no function of ovomacroglobulin was described it was reported to have no inhibitory activity with trypsin using protein or ester substrates and no protection of trypsin from binding to soybean trypsin inhibitor (63). This is in contrast to our findings with ovostatin on trypsin inhibition and the protection of the enzyme from soybean trypsin inhibitor which are ' W. A. Galloway and J. J. Reynolds, unpublished work. described in the following paper (20). It is, however, quite possible that the ovomacroglobulin preparation described previously (63) might have become inactivated during purification. A Mw of 56,000 to 68,000 estimated for the reduced ovomacroglobulin are close to those of ovostatin fragments produced by metalloproteinases (20). Regardless of whether or not ovomacroglobulin is identical with ovostatin, we feel the name "ovostatin" is appropriate for the inhibitor molecule characterized from hen egg white because it indicates both a function of the protein and its tissue of origin. TWO proteins which have molecular weights and quaternary structures similar to those of ovostatin have been reported in human plasma: pregnancy-associated plasma protein A and aZM. Pregnancy-associated plasma protein A has a M, = 620,000-820,000 with four subunits of M, = 187,000 (65,66).
It is synthesized in placenta (67) but has been detected in the serum of pregnant women. Although Bischof (68) reported a weak inhibitory capacity of this protein towards plasmin and urokinase, its function as a proteinase inhibitor has not been established (66). On the other hand, azM which inhibits almost all endopeptidases (21) including mammalian collagenase (69) has a quaternary structure similar to that of ovostatin and a similar mechanism for proteinase inhibition (see Ref. 20). However, ovostatin and chicken a2M are distinct from one another in a number of respects: 1) ovostatin is insensitive to CH3NH2, whereas the proteinase inhibitory activity of chicken aZM is destroyed by CH3NH2, as established for human aZM (27,44,45); 2) there is a distinct difference in electrophoretic mobility on 5% polyacrylamide gel (pH 9.58) between the two proteins; 3) no immunochemical cross-reactivity has been noted between the two proteins; 4) distinct mapping patterns of CNBr fragments of ovostatin and those of chicken azM indicate that the two proteins have different amino acid sequences; and 5) the NHz-terminal sequences of ovostatin and azM are distinct although they show similarity.
Examples of proteins found both in egg white and in chicken plasma have been reported. The iron-binding protein ovotransferrin in egg white and serum transferrin are immunologically identical (70) and have similar polypeptide chains (71). Barrett (38) has reported that a2-proteinase inhibitor in chicken plasma is identical with ovoinhibitor; the first 31 residues sequenced were shown to be completely identical and they probably possess the same amino acid sequence (72). Despite similarities of structure and function between azM in plasma and ovostatin in egg white, our data have shown that they are different proteins. However, the NHz-terminal sequences of the two proteins are sufficiently similar (about 40% of identical residues) to suggest that the structural genes for aZM and ovostatin may have evolved from a common ancestor, but that their evolutionary divergence has led the development of distinct specificities in their action on different proteinases which are documented in the following paper (20).

EXPERIMENTAL PROCEDURES
Materials. Procollagens~e was obtained from monolayer cultures of rabbit synovial fibroblast stimulated by pharbal myristate acetate ( 2 2 ) or crystals of monosodium urate monohvdrace (23). Procollanenase was activated to collagenase either by trypsin br by 4-aminophenyl mefcuric acetate as described previously (24). Acid Soluble collagen was extracted and purified from guinea p l g skin (25). Egg white was obtained from infertile hen eggs. Chicken blood with sodium ckfrace as anticoagulant was purchased from Pel-Freeze Bialagicalq     Protein standards were as in Fig. 11 and lysozyme (14,000).
fragments of ~~o s t a t i n run without and with reaucrion. resp&tively.