Purification and Characterization of a Novel Zinc-Proteinase from Cultures of Aeromonas hydrophila*

Reuerse-phase was developed with a linear acetonitrile gradient containing tri- fluoroacetic acid (0.1%, v/v). Homogeneity of these peptides was assessed by analytical reverse-phase HPLC (Vydac CIS column) using similar gradients, and purity was found to be 95% or higher. Amino acid analysis of purified peptides gave the expected amino acid ratios (+El%). Other Methods and Analyses-Zinc content was determined by Drs. B. L. Vallee and R. Shapiro, Harvard Medical School, Boston, using electrothermal atomic absorption spectroscopy. Protein concentration was determined using BCA reagent (Pierce). The amount of protein in purified AhP preparations, used for enzyme kinetics studies and metal analysis, was determined by amino acid analysis. The presence of carbohydrate was assayed for by concanavalin A-Sepharose chromatography according to Santer et al. (22). The f-(y-Glu)-Lys cross-link content was measured by the method of Harsfalvi et a/. (1).


Berger, A. (1967) Biochem. Biophys. Res. Commun.
27, 157-162). Nonpolar amino acid residues seem to be favored in the P,' and Pz' positions. The enzyme contains one atom of zinc and is inhibited by 1,lOphenanthroline, but not by EDTA. Iodoacetate, leupeptin, diisopropyl fluorophosphate, phenylmethylsulfonyl fluoride, pepstatin, and az-macroglobulin have no effect on enzyme activity. Disulfide reducing reagents, such as dithiothreitol or 2-mercaptoethanol, inactivate the enzyme completely. The partial amino-terminal sequence shows 46% identity with a zinc metallo-proteinase from a strain of Lysobacter enzymogenes and 69% identity with the LasA protein from Pseudomonas aeruginosa.
An isopeptidase capable of breaking t-(y-G1u)-Lys isopeptide bonds cross-linking protein chains has been sought for a number of years. Such an enzyme could be used to identify the protein chains, extracellular or intracellular, which are cross-linked by transglutaminase and occur in cell and tissue extracts in highly polymerized form detectable at the top of polyacrylamide gels (1). Moreover, an isopeptidase could be therapeutically important if sufficiently specific for crosslinked fibrin since it could act synergistically with plasmin in clot dissolution, especially in the heavily e-(y-G1u)-Lys crosslinked clots of dense thrombi (2).  reported the presence of an enzyme (destabilase) in the secretions of the salivary gland of the leech Hirudo medicinalis, which * This work was supported by a grant from Cortech Inc., National Institutes of Health Grant AM 34503, National Science Foundation Grant DCB 851112, and by Summer Student Fellowships supported by the Hughes Institute of Medicine and by Haverford College. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. hydrolyzed y-glutamyl-p-nitroanilide (y-Glu-pNA') and appeared to convert the y-chain dimer of cross-linked fibrin into the monomer. Although we were able to confirm the presence of a y-Glu-pNA hydrolyzing activity in the saliva and salivary gland extracts of starved leeches, we found no fibrin isopeptidase activity.* However, by using enrichment techniques, we isolated from the intestinal tract of the leech a bacterium which secretes an enzyme into its culture medium that appears to convert the y-chain dimer of fibrin into the y-monomer. This bacterium was identified as a strain of Aeromonas hydrophila. The enzyme hitherto referred to as A. hydrophila proteinase (AhP) was found not to be an isopeptidase, but rather a metallo-proteinase, which specifically cleaves the Gly-Ala peptide bond located within the sequence Gly-Gly-Ala, near the cross-link site of the y-chain dimer. Another metallo-proteinase with analogous fibrinolytic specificity has been found in puffadder snake venom (6-8). Preliminary reports of the purification and some of the properties of AhP have appeared elsewhere (40,41). In the present paper a detailed purification procedure and characterization of the enzyme are presented.

EXPERIMENTAL PROCEDURES
Enrichment Culture Isolation of Bacterium-Leeches (BioPharm) were fed once with rabbit blood and kept at 16 "C for 4-6 weeks without further feeding. Salivary gland excretions were collected according to the method of Rigbi et al. (9) with minor modifications.2 Salivary excretions, salivary glands, and intestinal contents were used as inocula for enrichment cultures on insoluble fibrin (see "Fibrin Assay of AhP") or on a variety of isopeptide-containing substrates in minimal salts under a variety of conditions. Clot-dissolving cultures were further investigated, and a bacterium was isolated which produced an isopeptidase-like effect on fibrin in the presence of EDTA. The bacterium isolated is a Gram-negative rod which was identified as a member of the species A. hydrophila by the American Type Cultural Collection and will be referred to as A. hydrophila sp.
Purification of AhP-A. hydrophila sp. was grown on buffered tryptone yeast extract broth (tryptone, 8 g/liter; yeast extract, 5 g/ liter; and NaH2P04, 10 g/liter, adjusted to pH 7.0) for 16-18 h at 30 "C. During purification, AhP activity was monitored using a fibrin assay described below. Optimal activity is obtained in the stationary phase of growth at pH 5 8. The cells and culture fluid were chilled and centrifuged to remove the bacteria. A concentrated EDTA solution, pH 7.4, was added to make the final concentration 20 mM; this reduced the total proteolytic activity in the supernatant sufficiently to proceed with measurement of AbP activity. The supernatant was concentrated 20-fold in an Amicon CH2 ultrafiltration system with a The abbreviations used are: pNA, p-nitroanilide; Ac, Nu-acetyl; AhP, A. hydrophila proteinase; LeP, L. enzymogenes proteinase; Nph, p-nitrophenylalanine; PAGE, polyacrylamide gel electrophoresis; Tfa, trifluoroacetyl group; SUC, N-a-succinyl; TNBS, 2,4,6-trinitrobenzenesulfonic acid; HPLC, high performance liquid chromatography; PEI, polyethyleneimine.
SlYlO membrane, and the protein was precipitated with ammonium sulfate added to 80% of saturation. The precipitate was resuspended in a minimal volume of 20 mM HEPES, 0.1 M NaCl, 20 mM EDTA, pH 7.4, dialyzed against the same buffer and applied onto a Bakerbond WP-PEI column equilibrated in 20 mM HEPES, 0.1 M NaC1, pH 7.4 (buffer A). AhP was not retained on the column, but a major portion of a contaminating material was adsorbed. The AhP fraction obtained from the PEI column was dialyzed for 1-3 h at 4 "C against 20 mM HEPES, 30 mM NaC1, pH 6.8 (buffer B); longer dialysis at this point resulted in some loss of AhP activity. The dialyzed material was loaded onto a CM-cellulose column equilibrated in the same buffer. AhP was eluted with a 30 mM NaCl, pH 6.8-0.4 M NaCl, pH 7.4 gradient in 20 mM HEPES. The gradient was monitored with a Markson conductance meter (model 1096). In some studies (as indicated) gelatin-agarose chromatography (10) was also employed to remove minor contaminants. The protein was adsorbed on a column of immobilized gelatin (Pierce Chemical Co.) in 0.05 M Tris, 0.02 M NaH2P04, 0.15 M NaCl, 0.5 mM phenylmethylsulfonyl fluoride, pH 7.4, and AhP was eluted with 1 M urea in the same buffer. The final product was dialyzed against buffer A, or precipitated with ammonium sulfate (80%) and then dialyzed against buffer A and stored frozen.
Polyacrylamide Gel Electrophoresis (PAGE)-SDS-PAGE was performed according to Laemmli (11). Proteins were stained with Coomassie Blue R-250 or silver stain (Bio-Rad). Cross-reactivity with the y-chain or y-chain products was assayed by immunostaining with anti-y-chain antibody3 on nitrocellulose membranes using Auroprobe BL (Jannsen) to detect the secondary antibody.
Isoelectric Focusing-Isoelectric focusing was conducted by Dr. N. B. Egen (The University of Arizona). The experiment was carried out in 4 M urea using pH 3.5-10 Ampholine PAGE-plates (12). Proteins were stained with a silver stain reagent (13).
Reuerse-phase High Performance Liquid Chromatography fHPLC)-A Waters HPLC system consisted of two model 510 pumps, an Automated Gradient Controller, a Lambda-Max model 481 LC Spectrophotometer, and a model 745 Data Module. A 300 A/5 pm Vydac or Waters Delta-Pak c 1 8 analytical column (0.46 X 15 cm or 0.39 X 15 cm, respectively) was employed with a 30-min linear CH,CN gradient (10-80%) containing 0.1% trifluoroacetic acid. The flow rate was 1 ml/min. Absorbance of eluate was usually monitored at 214 nm or occasionally at 278 nm for peptides containing nitrophenylalanine.
Amino Acid Analysis-Proteinlpeptide samples were hydrolyzed under vacuum in a nitrogen atmosphere for 24 h in 6 N HC1 or its vapor at 110 "C. Methanesulfonic acid (4 M), containing 0.2% 3-(2aminoethyl) indole (Pierce), was used for tryptophan analysis. The acid hydrolysates were analyzed using the PICO-TAG system (Waters Associates) and the standard ninhydrin system (Beckman).
Amino Acid Sequence Determination-AhP, prepared as described above including gelatin-agarose chromatography, was further purified on a Vydac C, reverse-phase column (4.6 mm X 15 cm) to remove any residual contaminants. The sequence of the first 40 amino acid residues of AhP was determined using an Applied Biosystems 477A protein/peptide sequencer and an on-line Applied Biosystems 120A HPLC system. Protein sequencing was done by W. S. Lane at the Harvard University Protein Microchemistry Laboratory (Cambridge, MA).
Preparation of Fibrin y-Chain Fragments for Sequencing-The ychain fragments of fibrin were generated by incubation of insoluble fibrin with AhP employing a protocol similar to that described below for the fibrin assay. The enzyme-treated material was reduced, carboxymethylated, and the presumed y-chain monomer then separated on CM-Sepharose (Pharmacia LKB Biotechnology Inc.) as described by Doolittle et al. (14). The y-chain fragment was electrophoresed on a 10% PAGE gel. The proteins in the gel were electrophoretically transferred to an Immobilon brand polyvinylidene difluoride membrane (Millipore), and the portion of the membrane that contained the y-chain fragment was cut out following the procedures of Matsudaira (15). The protein on the membrane was subjected to seven cycles of sequence analysis, as described by Matsudaira (15).
Fibrin Assay of AhP Activity-Insoluble fibrin clots were prepared in the bottom of small test tubes by incubating 50 pg of human fibrinogen (Kabi) with 20 pkatals of thrombin (Kabi/Helena) in 0.05 M NaC1, 10 mM CaC12 in a total volume of 10 p1 for 1 h at 37 "C. The clots can be stored frozen for future use. (There is a sufficient amount of transglutaminase in this fibrinogen preparation to convert most of the y-chain monomers to dimers and to form a-chain polymers.) AhP was added in a total volume of 25 p1 containing 20 mM HEPES, 0.1 Kindly supplied by Dr. E. F. Plow. M NaCl, pH 7.4. The reaction was stopped by the addition of the SDS-PAGE sample buffer containing SDS and 2-mercaptoethanol ( l l ) , then boiling for 2 min. Reaction products were examined by SDS-PAGE in 8.5% gels (11) and compared with soluble (uncrosslinked) and insoluble (cross-linked) fibrin controls. AhP activity was monitored through the diminution of the y-chain dimer and appearance of a band in the position of the y-chain monomer. Soluble fibrin was prepared in the presence of EDTA to prevent cross-linking by the Factor XIII. A unit of AhP was defined as the amount of enzyme which converts 50% of the y-chain dimer to the pseudo-monomer in 1 h.
Azocoll Assay of AhP Activity-Azocoll (Calbiochem) was prewashed with buffer A and suspended (1.5 mg/ml) in the same buffer. One ml of the suspended Azocoll was incubated with AhP at 37 "C on a tumbler. At the end of the reaction time, the entire mixture was centrifuged and the amount of the released dye in the supernatant quantitated at 520 nm. One Azocoll hydrolysis unit was defined as a of l.O/h in 1.1-ml volume and found to be on average 0.4 AhP units in the fibrin assay above.
TNBS Assay of AhP Actiuity-Five mM synthetic peptide substrates were incubated with 5-50 pg/ml of AhP in 0.05 M HEPES, 0.05 M NaCl, pH 7.5, for 5-20 min at 25 "C. The reaction was stopped by boiling for 15 s. The entire sample was diluted 10-fold with 0.1 M sodium borate, pH 9.2, and the free amine content was determined with TNBS according to Snyder and Sobocinski (16) with a minor modification. The absorbance of the trinitrophenyl derivatives was measured on a microplate reader (Molecular Devices) at 405 nm. The linearity of the assay was tested by changing the time of the reaction and the concentration of the enzyme. In most of the assays, no more than 10% of substrate was allowed to hydrolyze. If the extent of hydrolysis was more than 10% (but less than 20%), the substrate concentration was corrected for the loss during the reaction (17).
Intra-assay relative standard deviation was 5%. K,,, and kat values were calculated by a non-linear regression analysis program (ENZ-FITTER Elsevier-Biosoft).
Chromophoric Assay of AhP Actiuity-This continuous spectrophotometric assay was performed at 25 "C in a similar manner to that described by Hofmann and Hodges (18) for penicillopepsin. The nitrophenylalanine substrates (0.15 and 0.3 mM) and the enzyme (20-50 pg/ml) were incubated in 0.05 M HEPES, 0.05 M NaCl, pH 7.0, and the absorbance change was measured at 304-310 nm with an HP-5482A diode-array spectrophotometer. The reference absorbance was selected at 404-410 nm. The hydrolysis of nitroanilide substrates (1 mM) was monitored at 400-410 nm by measuring the absorbance of released nitroaniline. The buffer was 0.05 M HEPES, 0.05 M NaC1, pH 7.5.
Inhibition Studies-AhP was preincubated with an appropriate concentration of inhibitor in 0.05 M HEPES, 0.05 M NaCl, pH 7.0 or 7.5, at room temperature, and the residual enzyme activity was measured using one of the assays described above. The preincubation times were 75 and 15-30 min for the chelating agents and other inhibitors, respectively. Due to relatively poor aqueous solubility of 1,7-p,henanthroline, 5% MeOH was included in the assay buffer. Methanol at this concentration did not have any significant effect on the enzymatic/inhibition reaction and the subsequent TNBS assay.
Chromatography of AhP on Immobilized a2-Macroglobulin-A 1-ml column containing immobilized bovine az-macroglobulin (Boehringer Mannheim) was equilibrated with 0.1 M HEPES, 0.15 M NaCl, pH 7.0. AhP (400 pg) was applied slowly and held on the column for 15 min. When the column was then washed with the same buffer, AhP was eluted in the void volume (70% recovery of the Azocoll hydrolysis activity). The column still retained capacity to bind 300 pg of pancreatic elastase.
Preparation of Peptides-Synthetic di-, tri-and tetrapeptides were prepared via solution-phase synthesis using l-ethyl-3-(3-dimethylam-inopropy1)carbodiimide (Sigma) as the activation/coupling agent (19). Peptides were purified by a combination of (successive) anionexchange chromatography (DEAE-Sephadex, Pharmacia) and cationexchange chromatography (Dowex AG50W-X4, Bio-Rad). Homogeneity of purified peptides was assessed by thin layer chromatography on silica gel plates, and the primary structure of each peptide was further confirmed by 300 MHz'H nuclear magnetic resonance spectroscopy. Automated solid-phase syntheses of all other peptides (BioSearch 9600) were carried out by standard methods on methylbenzhydrylamine resin (BioSearch) (20). Peptides were deprotected/ cleaved from the resin with anhydrous hydrogen fluoride (21)  was developed with a linear acetonitrile gradient containing trifluoroacetic acid (0.1%, v/v). Homogeneity of these peptides was assessed by analytical reverse-phase HPLC (Vydac CIS column) using similar gradients, and purity was found to be 95% or higher. Amino acid analysis of purified peptides gave the expected amino acid ratios (+El%).

Other Methods and Analyses-Zinc content was determined by
Drs. B. L. Vallee and R. Shapiro, Harvard Medical School, Boston, using electrothermal atomic absorption spectroscopy. Protein concentration was determined using BCA reagent (Pierce). The amount of protein in purified AhP preparations, used for enzyme kinetics studies and metal analysis, was determined by amino acid analysis.
The presence of carbohydrate was assayed for by concanavalin A-Sepharose chromatography according to Santer et al. (22).
The f-(y-Glu)-Lys cross-link content was measured by the method of Harsfalvi et a/. (1).

RESULTS
Purification of AhP-AhP activity was initially measured using a fibrin assay (Fig. 1) as described under "Experimental Procedures."Later purifications were followed with the Azocoll and fibrin assays. Our early observations that AhP activity is not affected by 5-20 mM EDTA allowed us to use this chelator during initial steps of AhP purification. EDTA (20 mM) reduced the proteolytic activity in the bacterial media. An Amicon CH2 ultrafiltration system with a 10-kDa cut-off membrane was very useful, not only for concentrating the culture fluids, but also for the removal of small molecular contaminants. Subsequent purification was accomplished by anion-exchange chromatography on a Bakerbond WP-PEI column (Fig. 2), cation-exchange (CM-cellulose) chromatography (Fig. 3), and occasionally (see "Discussion") gelatinagarose chromatography. When concentrated culture fluid was applied onto the Bakerbond WP-PEI column (Fig. 2), the non-adsorbed fractions, 17-60, all showed AhP-like activity in the fibrinolytic assay. In the subsequent purifications, the non-adsorbed material was collected as one fraction. This fraction, assayed in the presence of 20 mM EDTA, showed 60-80% of the Azocoll hydrolysis activity measured in the absence of EDTA. Only this material was used for further purification. The retained material, which was eluted with increased salt concentration, contained several fractions which completely hydrolyzed fibrin clots even in the presence of EDTA. However, these fractions, assayed in the presence of EDTA, showed only 3-25% of the Azocoll hydrolysis activity in the absence of EDTA. Interestingly, DEAE-cellulose did not effect a satisfactory separation of AhP from other proteases. The non-adsorbed fraction from the anion-exchange column was briefly dialyzed to change the buffer and applied to CM-cellulose (Fig. 3). The two ion-exchange columns (WP-PEI and CM-cellulose) removed most of the contaminating proteins. The final yield of AhP was 2-4 mg of active enzyme/liter of culture medium. Specific activity was highest after CM-cellulose (Fig. 3) with 400-500 or 150-200 units/mg protein, measured in the fibrin or Azocoll assay, respectively. Specific activities of AhP at earlier steps of purification could not be measured with accuracy because contaminating proteinases, even in the presence of EDTA, interfered with the fibrin and Azocoll assays. Gelatin-agarose purification reduced specific activity by approximately 50% which may be due to tight binding of AhP to the affinity column and subsequent use of urea for elution. This step was employed in early stages of our work in order to assess whether the apparent isopeptidase activity of the partially purified enzyme could be distinguished from proteolytic activity of this preparation on Azocoll and on the a-and @-chains of cross-linked fibrin. We also employed Superose 12 (Pharmacia), Bakerbond CBX (J. T. Baker Chemical Co.) and propyl Lane I , An-doublet, BO-, and y-chains of fibrinogen. Lane 2, ndoublet, O-, and y-chains of uncross-linked fibrin. Lane 3,n-doublet, @-chain, and y-y-dimer of cross-linked fibrin. Lanes 4 and 5, reduced amounts of n-doublet and y-y-dimer, and appearance of 7'. the pseudo-y-monomer generated by the action of AhP on cross-linked fibrin. Factor XIII, XIIIa, and AhP are too low in concentration to be visible in this gel. Note that AhP also splits the n-chains; one of the breakdown products is seen at the bottom of lane 5.
aspartamide (The Nest Group) columns for the same purpose (data not shown). The fact that several different chromatographic procedures were unable to separate the above activities from each other was an early indication that AhP is a proteinase rather than an isopeptidase. The gelatin-agarose purification step was included nevertheless in the preparation of the enzyme samples used for sequencing. Unless otherwise indicated, the CM-cellulose chromatography was the final step in purification of AhP. The purified enzyme migrated as a single band in SDS-PAGE using Coomassie Blue (Fig. 4) or silver stain, and as a single tailing peak in CIS (Waters) Reverse-Phase HPLC. Isoelectric focusing (performed in 4 M urea) followed by silver staining of gels showed one strong band at pH 8.5 and two very minor closely spaced bands at pH 8.0.
Physicochemical Properties-The molecular weight of AhP Procedures"). The other non-adsorbed fractions showed little or no AhP activity. The adsorbed fractions could not be assayed for fibrinolytic activity due to interference caused by other proteases (see "Results").

FIG. 3. Purification of AhP by CM-cellulose chromatography.
The column (2.5 X 20 cm) was equilibrated with 20 mM HEPES, 30 mM NaCl, pH 6.8. The flow-through material (fractions, 17-60) eluted from the column illustrated in Fig. 2 was dialyzed against the equilibrating buffer and applied to the CM-cellulose column. The column was developed with a gradient from the equilibrating buffer to 20 mM HEPES, 0.4 M NaCl, pH 7.4, starting at fraction 32 as indicated by the dashed line. The Azocoll hydrolysis assay was carried out on an aliquot of each fraction, but note that the absorbance readings are one fraction before the actual collected fractions. The fibrinolytic assay (not shown) displayed the same profile as the Azocoll was measured by SDS-PAGE in 15% gels under reducing conditions (Fig. 4) with the apparent molecular weight being 19,000. The enzyme was also electrophoresed in the absence of 2-mercaptoethanol (Fig. 4). Under these conditions the migration distance of AhP was slightly greater (probably because the protein maintains a more compact structure when the disulfide bridges are intact), and the band was broader. AhP seems to have little or no carbohydrate as the enzyme was not retained on a ConA-Sepharose column, and its molecular weight as measured by SDS-PAGE agrees well with our gene sequence data (178 amino acid residue^).^ Metal analysis by atomic absorption spectrophotometry showed that there is 1 mol of zinc/mol of enzyme (0.94 gram-atoms of zinc/mol, average of two measurements). However, AhP was not adsorbed on a zinc chelating column (Boehringer Mannheim) .
The amino acid composition of AhP and the sequence of S. W. Jones, manuscript in preparation.  Table I and Fig. 5, respectively. The enzyme contains a large proportion of aromatic amino acid residues and a small fraction of lysine and cysteine residues. The partial amino-terminal sequence shows 46% identity with a zinc metallo-proteinase from a strain of Lysobacter enzyrnogenes (former Myxobacter) (23-25) and 69% identity with the LasA protein (LasA) from Pseudomonas aeruginosa (26, 27) (Fig. 5 ) .
Enzymatic Properties-The optimal pH for AhP activity, assessed with the substrate, Suc-Gly-Gly-Nph-CONH2, was 7.5. The enzyme was stable in 4 M urea, 5% methanol, 2% acetone, 2% dimethyl sulfoxide, 4.5 M ammonium sulfate, and 2 M NaC1. Stability in methanol, acetone, and dimethyl sulfoxide was tested by preincubating the enzyme and the reagent for 30 min and then carrying out the fibrinolytic assay in the presence of the reagent. Stability to urea, ammonium sulfate, and NaCl was tested by preincubating in the presence of the reagent for 1 h and then dialyzing against the assay buffer  Total 100.00 PICO-TAG method was used for all amino acids except cystine. Determined as cystine using standard minhydrin method. Acid hydrolysis was performed in the presence of 4 M methanesulfonic acid (Pierce). and measuring activity with Azocoll. We observed 5-8% and 30% loss of activity after 1 h of incubation of the enzyme in 0.05 M HEPES, 0.05 M NaCl, pH 7.0, at room temperature and 40 "C, respectively, and complete inactivation after 15 s of boiling in the same buffer. The enzyme activity was not inhibited by serine proteinase inhibitors such as diisopropyl fluorophosphate, phenylmethylsulfonyl fluoride, and leupeptin, nor the cysteine protease inhibitors iodoacetate and leupeptin (Table 11). Pepstatin, an aspartic protease inhibitor, and a*-macroglobulin, a nonspecific proteinase inhibitor, also had no effect on enzyme activity. However, the enzyme was inhibited by the metal chelators 1,lO-phenanthroline and 8hydroxyquinoline-5-sulfonic acid, and the sulfhydryl-containing compounds dithiothreitol and 2-mercaptoethanol. 1,7-Phenanthroline, which is not a chelator, and phosphoramidon, which inhibits some metallo-proteinases, had only a marginal inhibitory effect on AhP activity. Phosphoramidon had some effect at millimolar concentrations (Table 11); however, the effective concentration of this compound is usually in the micromolar range (28). Interestingly, 5 mM EDTA, which deactivates most metallo-proteinases including those of bacterial origin (29), did not inhibit AhP even though the effective formation constant of EDTA with zinc is 4.3 X IO' , at pH 7.5 (30). Substrate specificity of AhP was first characterized using cross-linked fibrin (Fig. 1). The enzyme causes the disappearance of the y-dimer chain with the concomitant appearance of a chain which in SDS-PAGE had a mobility exactly equal to that of the monomer. Other hydrolytic effects on the apolymer, and on the a-and @-chains were a t first judged to be due to the presence of proteolytic contaminants. Additional results such as cross-reactivity of the product of y-dimer hydrolysis with anti-y-chain antibodies and the identity of the electrophoretic mobility of the hydrolysis product with that of the y-chain suggested that AhP was indeed an t-(y-G1u)-Lys cleaving isopeptidase. However, we began to question this conclusion when we found that the AhP hydrolysis product of the y-dimer, just like the dimer itself, contained one t-(y-G1u)-Lys cross-link/chain. Furthermore, the AhPderived product of the y-dimer, when subjected to a few cycles of sequencing, revealed a primary Tyr-Val-Ala-Thr-Arg-Asp-Asn and a secondary Ala-Xxx-Gln-Ala-Gly sequence.
The primary sequence was the predicted NHz-terminal sequence of the y-chain. The secondary sequence, however, indicated that the split of the cross-linked y-chains occurred after the 2nd Gly residue in the Gly-Gly-Ala sequence, immediately adjacent to the Glu-Lys isopeptide bonds of the y-dimer (Fig.  6). It can be noted that this kind of cleavage results in two isopeptide-linked chains with the same molecular weight and amino acid content as the y-chain monomer (Fig. 6 ) . This cleavage site was confirmed using a synthetic fragment (FJ of the y-chain of fibrin, Ac-Glu-Gly-Gln-Gln-His-His-Leu-  Gly-Gly-Ala-Lys('Tfa)-Gln-Ala-CONH,. HPLC analysis of the product of AhP hydrolysis showed only one split between the P1 (31) & and P1' & residues.
Somewhat broader selectivity by the enzyme was observed for short synthetic substrates containing unbranched N-substituents in the NHz-terminal glycine residue (Table 111). HPLC analysis of the digestion products, generated from some selected substrates (Table 111), revealed that Ac-Gly and, to a smaller degree, Suc-Gly accumulated to significant levels together with the expected Ac-Gly-Gly and Suc-Gly-Gly. Polar residues, especially those with free a-amino or acarboxyl groups located in close proximity to the scissile peptide bond, were not tolerated well. The nitroanilide peptides, Suc-Gly-Gly-pNA, Suc-(a-CONHCH,)y-Glu-pNA, y-Glu-pNA, and NO-acetyl-Ne-(N-acetyl-L-y-glutamyl-a-N-methy1amide)-L-lysine-N-methylamide were not substrates. Nitroanilide peptides are often good substrates for serine and cysteine proteases. The nitroanilide substrates and all other substrates were assayed at 1 and 5 mM concentration, respectively. No hydrolysis above the detection limit (about 5 FM) was observed with the nitroanilide compounds. The reference substrate, Ac-Gly-Gly-Ala CONH, (Table 111), was also assayed as a positive control at the same concentration (1 mM). Under the same conditions the reference substrate was hydrolyzed completely.
Once we had established the proteolytic nature of AhP, our inability to remove or even decrease by purification the slower secondary effects of the enzyme on the a-polymer and the aand @-chains could be explained in terms of the cleavage of other Gly-Gly-Xxx sites present in these chains. Furthermore, we found that both the rate and pattern of fibrinogen and fibrin breakdown differed markedly. Unlike the y-chain dimer of fibrin, the y-chain of fibrinogen does not appear to be cleaved at a measurable rate. We shall report on the details of the effect of AhP on fibrinogen and fibrin elsewhere.' Suffice it to note here that the action of AhP is highly dependent on the secondary and tertiary structure of its fibrinogen and fibrin substrates.
A few synthetic substrates contained Nph in the PI' position (Table 111). The presence of Nph allowed for a continuous spectrophotometric monitoring of proteinase activity (18). The reaction was carried out at pH 7.0 since at pH 7.5 the difference between the molar absorptivities of the substrate and the products is too small for reliable measurements. Since substrate concentrations used in this assay were at least five times below the K,,, value, only kCat/Km values could be calculated from the exponential curves obtained for these sub-  Enzymatic hydrolysis of all the substrates except the nitroanilide substrates was monitored with TNBS. In case of the nitroanilide substrates the release of nitroaniline was monitored directly at 400-410 nm. The concentrations of the nitroanilide substrates and all other substrates were 1 and 5 mM, respectively (see "Results"). Vertical arrows indicate peptide bonds split. F,, F, and F,, symbols used in the text represent the peptides on the left in the same line.

A Novel
Proteinase in Cultures of A. hydrophila the specificity which AhP exhibits for the y-chain of fibrin is probably due to a conformational change occurring in the fibrinogen to fibrin transformation. That such a conformational change indeed occurs was shown by Donovan and Mihalyi (37) in their scanning calorimetric study of fibrinogen clotting and by Rowbotham et al. (38), who raised a monoclonal antibody to fibrin which does not cross-react with fibrinogen. We were not able to detect significant AhP activity in Escherichia coli, mouse liver or brain, more than 100 bacteria obtained by various enrichment techniques from soil samples, and three other strains of A . hydrophila from the American Type Culture Collection. The absence of AhP activity in other strains of A. hydrophyla sp. suggests that its production by the leech endosymbiont is an evolutionary adaptation to the bacterium's nutritional environment. We observed that the supplementation of the tPA-plasminogen clot lysis system with AhP significantly increases the lysis rate of cross-linked fibrin (41). Its role in the bacterium and in the puffadder venom may be that of accelerating the lysis of cross-linked fibrin.