The Purification of Thrombin and Isolation of a Peptide Containing the Active Center Histidine*

SUMMARY A chromatographic purification of bovine thrombin from commercial starting material is described which yields preparations judged to be essentially pure, that is, with a clotting activity of 2100 tq 2500 NIH units per mg. In addition to the two-chain form of thrombin, the purified enzyme contains a three-chain form arising from cleavage of the B chain at arginine-73 (or 76). This form must be fully active since the specific clotting activity of various preparations does not vary with the content of the three-chain form, amounting at times to 70%. Inactivation of thrombin with Na-tosyl lysyl chloromethyl ketone and Na-(p-nitrobenzyloxycarbonyl)arginyl chloromethyl ketone results in alkylation of nitrogen-3 of the active center histidine with loss of both clotting and esterase activities. A radioactive peptide has been isolated from thrombin inactivated with tritiated Na-tosyl lysyl chloromethyl ketone and shown to contain histidine-43. is a protcolytic whose re-sembles trypsin with respect to a preference for esters of argininc lysinc


SHAW
From the Biology Department, Brookhaven National Laboratory, Upton, New Yodc 11973 SUMMARY A chromatographic purification of bovine thrombin from commercial starting material is described which yields preparations judged to be essentially pure, that is, with a clotting activity of 2100 tq 2500 NIH units per mg. In addition to the two-chain form of thrombin, the purified enzyme contains a three-chain form arising from cleavage of the B chain at arginine-73 (or 76). This form must be fully active since the specific clotting activity of various preparations does not vary with the content of the three-chain form, amounting at times to 70%. Inactivation of thrombin with Na-tosyl lysyl chloromethyl ketone and Na-(p-nitrobenzyloxycarbonyl)arginyl chloromethyl ketone results in alkylation of nitrogen-3 of the active center histidine with loss of both clotting and esterase activities.
A radioactive peptide has been isolated from thrombin inactivated with tritiated Na-tosyl lysyl chloromethyl ketone and shown to contain histidine-43.
For this purpose purified thrombin was essential.
The cshromatography of thrombin OJI a rarbos~hte ioJJ es-* This research was carried out at Brookhaven National Laborat,ory uudcr the auspices of the Unites States Atomic Energy Commission.
$ Postdoctoral Fellow of the National Heart Institute. Present address, Department of Chemistry, Texas A and M University, College Station, Texas 77843.
1 lowever, the met)hod we describe, employing Bio-Rex 70 carboxylate resin (-400 mrsh) succeeds in removilig species of thrombin which have lost clotting activity and provides a convenient and reliable me:~ns of obtainiug highly active prel)aratious comparable to the httst isolated from plasma.
The thrombin thus obtained was d~owvn to contain the 1 wochain form of thrombin described by >lagnusson (12) along with an additional, fully active form cleared in the B chain.
Thrombin was inactivated by the affinity-labeling rrag;ents TLCK (4) and p-N02-ZACK (5) both of which alkylated N-3 of a histidine residue. Histidine-43 was identified as the active center hi&dine through isolation of a radioactive peptic peptide from 3H-TLCK-inhibited thrombin.
I>isulfide boll& were clruvcd and corboxymethylated a,ccording to the: method of ('restfield, RIoore, aJld Stein (16); the reaction mixture was acidified with acetic acid and applied directly to G-75 columns n-rapped in aluminum foil for simultaneous separ;ttioll of the peptide chains and desalting.
\\IICU about 2 cm of resin had settled, a RIOW flow of buffer was started and a column of 20 to 25 cm of resin Teas allowed to scttk. 'l'he column was \vxshed with 4 liters of DH 7.0 buffer before use.
Ccllex-I' and Cellex-I> resin were precycled and rcgenernted as recommended by the manufacturer.
The thrombin-coilcelltrating column (dcscribcd under "AIatcriala") was poured by means of a glass extension tube (1.5 X 100 cm) through which a I: I slurry of Cellex-1' and buffer adjusted to p1-I 7.0 was poured with the stol)caock fully opened until a bed height of 11 cm under a 100.cm head of buffer was obtained. &fore use the column was wash~~l with 400 ml of pI1 7.0 buffer.
Thrombin Chromatography--The contclrts of thirty 10,000. NIH unit vials of topical thrombin were dissolved in 100 ml of pH 7.0 buffer; the buffer was first used to rinse the vials aftel the solid had been removed.
After adjusting the $I to 6.9 with 0.1 M sodium hydroxide, the solution was applied to the Rio-Rex 70 column and rinsed in with buffer. (Occasionally the top of the resin bed cracked and sepamted as the thrombin W:IS applied in which case the first 5 cm of resin bed was stirred with buffer and allowed to resettle. This was not detrimental to the subsequent chromatography.) Elution was started with l)~l 7.0 buffer; 26.ml fractions were collected at lo-mill intervals. \\'hcn the absorbance at 280 rnp dropped below 0.05 (usually after 16 to 20 hours), the 1'1-1 8.0 buffer was introduced and clution continued at t,ho sarnc' JXtf'; 16-1111 fractions were (~01. lect,etl in polyethylene tubes at 6-min intervals. Thrombin ms obtained in the last l)cak elutetl (usually after 30 to 36 hours).
(Baughman and Laugh (9) have discsllsscd the relationship between the various methods of cxl)ressin :: clotting activities; our absorbance dues \vere uncorrected.) Concentration oJ' Dilute Throtttbin-p'l'he pooled thrombin from Bio-Rex 70 was diluted 4%fold and passed through the Cellex-I' concelltrnting column as rapitlly as it would flom (a 2-m head of buffer was used). The absorbed thrombin was eluted with 1 RI sodium chloride; 2-ml fractions xvere collected in polyethylene tubes. Fractions of specific activities >800 to 1000 NII-I units per absorbance ullit were pooled giving solutions of 10 to 14 absorbance units Ilcr ml or 5 to 7 rug per ml.
Awlmonium &zdjate l'recz'pilaiion of Thrombin--The thrombill pool in 1 M sodium cahloride was treated at 0" with 0.55 g of powdered ammonium sulfate (reagent grade) per ml in a polyethylellc beaker with gentle stirring at 0" until all had dissolvetl. After transferral to 50-ml polyethylene centrifuge tubes Iv&h washings of five l-ml portions of saturated ammonium sulfate, the suspension was c+entrifuged at 10,000 rpm for 15 min in the c~old. The prt(ail)itatrd thrombin was stored in the frozen state at -20" after SOllltiOll ill 0.05 RI SllCrOSC (5 Illl Inhibition oj Throntbin with TIJX and p-NO~-%ilC'K---I'~l~ified thrombin at :I concentration of 6.47 X IOF A\I :t~~d B sl)ecific activity of 1830 XIII units lJf>r m:;r was used.
An aliquot of the TLCKiuhibited thrombin and all of the p-NOz-%AC'K-inhibited thrombin was oxidized with performic acid (23).
Inhibition of I'hrombin with 3H-TLCK--The JH-TLCK used had a specific activity of 8.75 x lo5 dpm per pmole. Thrombin was in 1 M sodium chloride as obtained from the Cellex-P concentrating column.
When loss of esterase activity reached 99% (5 to 6 hours) the solution was dialyzed exhaustively against deionized water and lyophilized.
Peptic Digestion of Cyanogen Bromide Peptide 52-84 jrom the B Chain 3H-TLCK Thrombin-The peptide (1.28 pmoles), obtained as described in Fig. 6, was digested with 1.5 mg of pepsin at 25"in 3.5 ml of 50/, formic acid (24) for 22 hours. The reaction mixture was evaporated in a vacuum and the residue was gel filtered on Sephades G-50 (Fig. 6).

RESULTS
Thrombin has previously been purified to high specific activity by ion exchange chromatography on a carboxylatc resin (Amberlite IRC-50) (7, 8). However, the starting material for such preparations was obtained by activation of the partially lmrified zymogen, prothrombin, obtained from fresh plasma. This is generally not a convenient starting material and we turned our attention to the preparation of thrombin at high specific activity from a more convenient but relatively crude commercial source, Parke-Davis topical thrombin. Such a method would allow many other laboratories lacking facilities for fractionation of blood ready access to this important and interesting enzyme. The procedure of Baughman and Waugh (9) appeared to solve this problem, since they reported that the purification of Parke-Davis topical thrombin using columns of diethylaminoethyl-and phosphocellulose in tandem yielded preparations with specific activities equal to the highest, reported earlier by ot'her workers I  I  I  I  I  I  I  I  I   " In our hands, however, this method gave thrombin with only 50% of the expected specific activity.
Therefore, we returned to the exploration of ion exchange chromatography on carboxylate resins since this had been the method of choice in earlier work (6-8).
It was eventually found possible to obtain columns of Bio-Rex 70 ( -400 mesh) which resolved the esterase (nonclotting) thrombin from forms of thrombin with full clotting activity.
The fine particle size used and relatively large column cross section were essential to obtain thrombin of such high specific activity.
The elution pattern is similar to that obtained by Magnuason (8) except that the broad area designated esterase thrombin was not found in his preparations.
While esterase thrombin had the same amino acid composition as thrombin eluted subsequently with high specific activity (Table I) it was devoid of clotting activity.
Titration of the active center of esterase thrombin with p-nitrophenyl p-guanidinobenzoate (14) gave from 40 to 80% of the expected esterat,ic site content when the thrombin concentration was measured using the extinction coefficient of purified thrombin (29).
Although thrombin has been shown to lose clotting activity more rapidly than esterasc activity eventually leaving only a low level of esterase activity (7), this is the first isolation of a highly active csterase thrombin chrornatographically. This material has not been further investigated but could provide information about structure-activity relationships in thrombin. It amounted to about 500/, of the total initial esterase content.
The active thrombin in the last peak ( Fig. 1) was pooled and concentrated by absorption and desorption from a cellulose phosphate column shown carlier to absorb thrombin quantitatively at low ionic strength (9). From the 300 to 400,000 SlH units of crude thrombin applied to the Bio-Rex 70 column, about 40 to 50y0 of the clotting activity was recovered in purified and concentrated form from the Cellex-1' column providing 70 to 80 mg of thrombin usually with specific activities of 1,000 to 1,100 NIH units per absorbance unit. This activity is comparable to that of the most active preparations obtained in other laboratories (9). The thrombin concentration of such preparations was also measured by titration with p-nitrophenyl pguanidinobcnzoate (14) and shown to reach a level of 2,100 to 2,500 NIH units per mg based on a molecular weight of 36,000. Fully active thrombin is gcncrally considered to have a sljecific activity of about 2,000 XII-I units per mg (7-9).
The amino acid composition of I)urificd t,hrombin (Table I) ~vas con&ult from run to run.
hlagnusson (8) reported an average yield of 74 mg of thrombin with an average specific activity of 2,098 NIH units per mg of dry weight from semipurified prothrombin obtained from 20 liters of plasma.
Chrom:rtography of 300,000 units of Parkc-Davis topical thrombin as destribed, thus is coml)arable to the processing of 40 liters of whole blood.
About half of the cllottillg activity applied to the column was recovered as purified thrombin.
Purified thromhin thus obtained could be rechromatographed on P,io-Rex 70 with no change in specific activity or amino acid composition.
h third chromatography also gave no change. These results are not in accord with the report that rechromatography of thrombin separates a 75-residue peptide from thrombin w&h a 2-fold iucrease in specific activity (30). The stability of thrombin on along with the fact that thrombin loses only about 20% of its clotting activity in 48 hours if allowed to undergo autodigestion at 1.4 X 10F4 M, pH 7.0, indicates that Parke-Davis thrombin contains some enzymic constituent(s) catalyzing the formation of esterase thrombin.
The relative]) low specific activities of our preparations of stock thrombin prepared by the method of Baughman and Laugh (9) results from the failure to separate esterase thrombin from clotting thrombin since both forms adhere to and are eluted from I)hosphocellulose under the same conditions.
The amino acid compositions of esterase thrombin, stock thrombin, and purified thrombin arc in good agrccmcnt with each other (Table I) and with that of bovine thrombin calculated from the current sequence studies of Magnusson (25) or rc-1)orted by other groups (26-28).
Our preparations were also found2 to contain from 2 to 4 moles of glucosamine per mole of thrombin in agreement with the report that thrombin contains 5 to 6% of carbohydrate including sialic acid, galactose, mannose, alld glucosamine (25).
Since thrombin and trypsin have comparable specificity with respect to amino acid side chains, it was expected that TLCK and p-NOs-ZACK would give irreversible inhibition of thrombin as in the case of trypsin (4, 5). Both inhibitors completely inactivated clotting and estcrase activities of thrombin.
In each 2 We are grateful to Dr. 1)arrell Liu for determinations of gIlIcosamine and tryptophan.
case amino acid analysis indicated the loss of about one hi&dine residue per mole of inhibited thrombin. I'erformic acid oxidation of the inhibited enzyme gave 0.48 residue and 0.74 residue per mole of TLCK-and ZACK-inhibited thrombin, respectively. In order to facilitate isolation of :I pcptide coiitaining the active center histidine for coml)arison of its sequence with that of other serine proteases and to establish its location in the primary sequence, purified thrombin was labeled with tritiatcd TLCK.
One mole of 3H-TL('K was incorporated per mole of inhibited enzyme. (During subsequent degradation procedures a progressive loss of radioactivity was observed. Therefore radioactivity was not used for stoichiometry.) Gel filtration of reduced and carbosymethglated 311-'~LCK thrombin on Sephadex G-75 (Fig. 2) gave uncspected results. hlagnusson has shown (12, 25) that thrombin arises by cleavage of the sill& polypcptide chain of prothrombin in two places, releasing active thrombin in a form consisting of two chains, A and A (Fig. 3), connected by a disulfide bridge. In gel filt,ration of reduced, carboxymethylated samples of 313-TLCK-inhibited thrombin, more than the two expected chains were consistently resolved as in Fig. 2. The fragment corresponding to the 49.residue A chain in composition was inrariably found intact and unlabeled (Fig. 2, peak at right).
However, the radioactivity appeared consistently distributed between two peaks which varied reciprocally in amount from one 1)reparation to the next. The smaller cha.in accounted for 10 to 705; of the radioactivity.
It became Frc;. 2. Gel filtration 011 Sephadex G-75 of reduced and carboxymrt,hylated :LH-TI,CK-thrombi~~ (3.9 pmoles). The column (2.2 X 200 cm) was eyuilibrat,ed with -0% acetic acid and elution was carried out at 12 ml per hour. Tubes indicated by the braclcets were pooled. The disulfide bridges shown are only for illustrative purposes. The points of cross-linkage are not known. evitlwt,, as discussed below, that these represented the intact B chain and a fragment of it comprising the NHS-terminal region. The original thrombin preparation, therefore, must have L'OIIsistcd of a mixture of two forms, that described by hlagnusson, and an additional, fully active form with an int'ernal split in the B chain as shown diagrammatic:all~ in Fig. 3. The proportions of these two forms of thrombin varied from batch to batch. The results in Fig. 2, for example, indicate that the thrombin used in that experiment contained 20$'J0 of the three-&ail1 form. The fragments of the B chaili of thrombin will be referred to :IS U-l and B-2. The larger of these, B-2, was not resolved from the illh:rct B chain in the gel filtration shown in Fig. 2.3 The R-1 frngmcnt has bec~r identified as the NHz-terminal region of the li chain on the b:isis of several observations. An Nil?-terminal isolcucine residue was found as in the case of the II chain it,self (12, 25). The amino acid composition of B-1 was ill :tc*cortl with that of the first 7X residues of the partial sequence data (Fig. 4) of ~lagnusson (25), nn approximate length chosen to accommodate the content of 5 arginine residues. The precise l)oillt of t,erminntion was established by the subsequent finding n These have, however, been purified by successive gel filtrations in 3% acetic acid on G-75 collmms such as that in Fig. 2. The amino acid compositions recorded in Table II are in general agreement with those expected.   12  3  4  5  6  7  8  9  10  11  12  13  14 Ile-V~l-Gl~-Gly-Gl~-A~p-Ala-Gl~-Val-Gly-Le~-S~~-Pr~ -Trp-15  16  17  18  19  20  21  22  23  24  25  26  27  28 Gln-Val-PIet-Leu-Phe-Arg-Lys-Ser-Pro-Gln-Glu-Leu-Leu- Cys-29  30  31  32  33  34 35  36 37  38  39  40  41  42 Gly-Ala-Ser-Leu-Ile-Ser-Asp-Arg-Trp-Val-Leu-Thr-Ala-Ala- of a ('OOH-terminal arginine. Since a methionine is at position 17 of the B chain (25), cyanogen bromide cleavage of 13-l was expected to yield smaller derivatives with a composition corresponding to residues 1 through 17 and 18 through 73 and such evidence was obtained (Table II).
In addition, the carbohydrate side chain of thrombin is known to be attached to a single asparagine residue near the f\'H2-terminal of the B chain, possibly residue 50 (25).4 The B-l chain was found to contain from 2 to 4 glucosamine residues, the total content found in the thrombin preparation from which it was derived.
The position of the proteolytic split within the B chain was localized through the use of carbosypeptidase -1 and B (21). Only the latter released free amino acids, providing 1 cq of arginine.
On this basis, splits at arginine 71, 73, or 76 approxirnately accommodate the amino acid composition calculated for t,he P-1 chain (Table II).
However, of these, arginine-73 is the best candidate due to the consistent indication of 3 residues of threoninc rather than the two required by arginine-71.
Arginine-76 wvas less probably due to values of arginine a11d glutamic acid consistently 1 residue less than required by a split at that point.
Our d:tt>a thus fa\-or arginine-73; however, an additional tyrosine must be accounted for. A decision betn-ten arginine-73 and arginine-76 must await further work.
As mentionc,d above, only part of the radioactivity of 3% TLC'~-inactivatctl thrombin appeared in the B-l peak. The intact (two-chain) form of thrombin often predominated. To conserve material for degradation studies use was therefore made of the labeled B chain obtained rnixed with n-2 in the initial peak of gel filtration (Fig. 3). It was considered likely that each form of thrombin was labeled at the same histidine residue and that t)he presence of methionine at positions 17 and 84 of the B chain would permit the isolation of a radioactive cyanogen bromide fragment encompassing the same labeled histidine 4 In the numbering employed the NHz-terminal isoleucine of the B chain is rcsidur 1.     (Fig. 5, Peak 3), alanine = 3.00 residues. 0 One preparation (same procedure as in Fig. 5). 1 Cysteic acid, from air oxidation of S-carboxymethylcysteine. Q CMC, carboxymethylcysteine. residue obtained in B-1. Fractionation of the cyanogen bromide peptidcs obtained from the mixture of B and B-2 ( Fig. 5) provided a radioactive peak similar in composition to the region 18-84 of sequence (25) ( Table II) as cgpected.
The B-l pcptide and the cynnogen bromide fragmelrt of the B chain contain two histidines, namely residues 43 and 69, out of the seven present in thrombin.
To determine which of these two had been alkylated by WTLCK, furt.hclr degradation by proteolytic action was sought. The cyanogen bromide peptide from B (residues 18-84) W&S tligtsted wit)h trypsin nlld cahymotrypsin.
However, even ~)rolouged and wlwated digestions released only a small amourlt of r:ldioactiJ?ty which gave no promise of yielding a pure Ileptidc 011 chromatography.
On the other hand, subtilisin gavr extensive digrstion but, t,he labeled peljtides formed were too numerous for profitable resolution by ion exchange chromatography. I*se was finally made of pepsin to obtain a suitable digest.
Rechromatography at lower ionic strength on 1)I~GS-Sephadex gave a symmetrical lo& representing 95oj, of the radioactivity recovered in thr two llraks (only 570/, of the ratlioactivity applied was recovered).
At this point 17"/, of the radioactivity in the small peptidcs released by pepsin remained (Fig. 6, Peak 11). Amino acid analysis revealed that this material represented a fairly clean peptide corresponding to the result drscribed below.
Rechromatography on IRCdO gave further fractionation (Fig. 9). All of the radioactivity was recovered with 607, in the major peak. Analysis suggested a peptidc with the composition Thr, Alan, Cys (Table II) in which the additional presence of the alkylated histidine was indicated by radioactivity and the formation of 0.34 residue of 3-carboxymethglhistidine on performic acid oxidation. This yield of histidine oxidation product was considered adequate for 1 residue in view of the yield of 0.48 residue obtained from 'l'I,(:Kinhibited thrombin itself and earlier experiences (31). This 5. Gel filtration on Sephadex G-75 of the cyanogen bromide peptides from the mixture of the B and B-2 chains of 3H-TLCKthrombin from the first peak in Fig. 2. The column was the same as that described in Fig. 2 . 6. Gel filtration on Sephadex G-50 of the peptic digest of cyanogen bromide peptide 18-84 (the third peak in Fig. 5) from 3H-TLCK-thrombin.
The column (0.9 X 200 cm) was equilibrated with 5070 acetic acid. The flow rat,e was 10 ml per hour. The bracketed tubes were pooled. pcl)tide corresponds to residues 40 to 44 of the B chain having the sequence Thr , Ala, :lla , His, ('ys. In the isolated material, of course, the histidine was alkylated with WTLCK. Four cycles of the subtractive Edman procedure, with performic acid oxidations to give 3 carboxymethylhistidine and cysteic acid from TLCh-histidine and 5 carboxymethylcysteine, respectively, confirmed this sequence. The active center histidine of thrombin was thus identified as histidine-43. The column (0.9 X 20 cm) was equilibrated with pH 7.0 buffer prepared by adding N-ethylmorpholine (about 13.6 ml) to a solution which contained 0.1 M pyridine and 0.1 M acetic acid when made up to l-liter volume. Elution with the same buffer was carried out at a flow rate of 6 ml per hour. The bracketed tubes were pooled.
FIG. 9 (right). Ion exchange chromatography of peptic peptide on Bio-Rex 70. The column (0.9 X 20 cm) was equilibrated with buffer made by adjusting the pH of a solution of 10 ml of N-ethylmorpholine per liter of H20 to 6.5 with acetic acid. Elution, with the same buffer, was at a flow rate of 6 ml per hour.

DISCUSSIOY /
The method of purification of bovine thrombin described makes this enzyme accessible in the state of purity desirable for studies on the relationship of structure to function.
Since a large number of proteases are present in plasma, misleading results could easily be obtained particularly with the customary use of simI)le substrates in biochemical studies involving enzyme function.
The finding that thrombin may exist in more than one form in a fully active state, the difference consisting in a limited proteolytic cleavage that converts a two-chain form to a threechain form, has considerable precedence in the pancreatic proteolytic enzymes (32, 3s). It is not known whether the new form tlescribcd has physiological significance since the enzymes used experimentally for the activation of prothrombin are not necessarily the ones that function in blood clotting.
1\1ann and Batt (11) have reported that a number of molecular species are found in Parke-Davis topical thrombin when purified by ion exchange chromatography on IRC-50, followed by triethylaminoethylccllulose to remove some nonthrombin impurity. On the basis of electrophoretic analysis and chain separations, the authors conclude that their preparations contain three thrombin species that vary in proportion from batch to batch. However, since their preparations had relatively low specific activities (about 1400 NIH units per mg), it seems probable that esterase thrombin had not been removed and that this accounted for the extra heterogerleity. 5 Components A and B as designated by Mann and Batt possibly csorrespond to the forms described in this paper.
In view of the known similar esterase activities of thrombin and trypsin, it was expected that sdbstrate-derived alkylating agents such as TLCK and 1~NO*-ZACK would complex with thrombin as they do with trypsin (4, 5), leading to a subsequent specific alkylation.
The results in this regard were completely analogous to those obtained with trypsin, namely, inactivation by alkylation of N-3 of a llarticular histidinc residue. The pH rate llrofile of the esterase action of thrombin indicates the essentiality of a group with pK near 6.5 (depending on the substrate studied) (35,36). In addition, the fact that thrombin is a serine proteinase (37) with certain kinetic similarities to trypsin and chymotrypsin has strengthened the impression that a common hydrolytic mechanism is at work in all three enzymes. The observation that inactivation of thrombin results from the alkylation of histidine-43 of the B chain indicates that this residue is the active center residue whose function has been implied in the earlier kinetic studies.
It is of considerable interest that the sequence studies in progress on thrombin provide a model in which the structural relationship to the pancreatic enzymes is made clear. The additional molecular weight of thrombin over trypsin or chymotrypsin, namely about 10,000, can be accounted for by a polypeptide (the A chain) and a single carbohydrate side chain that are probably externally attached to a central globular structure formed by the B chain (25). The B chain sequence suggests collsiderable homology with chymotrypsin and therefore probably has a similar three-dimensional structure (38). In this sequence comparison histidine-43 of thrombin is homologous to histidine-57 of chymot,rypsin (25). The results with TLCK provide functional evidence supporting the sequence data.
dclinowledgments-The extensive contributions of George Latham and John Ruscica are gratefully acknowledged.