- Email: [email protected]
Purification and large-scale preparation of antithrombin III
Purification Antithrombin
and Large-Scale III
Preparation
of
DAVID L. HOFFMAN, B.A. Berkeley. ~/ifoma
Cutter Biological, Miles Inc., has implemented large-scale isolation and purification of antithrombin III (AT1111 from human blood plasma. The starting material, Cohn fraction IV-l, is suspended and AT111 separated from other proteins by heparin-affinity chromatography. The concentrate is subsequently heated for 10 hours at 60°C in a solution of 0.5 M sodium citrate to inactivate the causative agents of viral hepatitis B and acquired immune deficiency syndrome. After heating, a second heparin-affinity chromatography step is employed to isolate the active AT111 and to remove heat-denatured protein. The purified solution is compounded with 0.1 M alanine and 0.15 M sodium chloride, filled in vials of 500 or 1,000 IU each, and freeze dried. The final product has a specific activity of not less than 6.4 IU AT111 per mg protein, of which over 90 percent binds to heparin on crossed immunoelectrophoresis. Preclinical and clinical testing have shown this product to be safe for prophylactic and therapeutic administration to persons with congenital deficiency of ATIII.
From Ihe Department of Plasma Manufacturing Technology, Cutter Bwloglcal. Miles Inc.. Berkeley, C-allforma. Requests for reprmts should be addressed to David L. Hoff. man. Cutter Blologlcal. Miles Inc.. 4th and Parker Streets, Berkeley. Califorma 94710.
September
he first methods reported for the isolation of antiT thrombin III (ATIII) from human plasma by conventional techniques were lengthy, with poor yields [1,2]. A significant improvement in AT111 methodology occurred when heparin bound to an insoluble matrix was introduced as an affinity adsorbent [3,4]. However, the methods described still included procedures more suited to small-scale purification, which restricted their usefulness in commercial applications. Because sizable amounts of AT111 were needed for clinical evaluation and eventual commercial use, more efficient methods were required. Therefore, we have developed a large-scale method for the preparation of a highly purified AT111 concentrate based on the use of Heparine Ultrogel. The method utilizes Cohn fraction IV-l, an unused by-product of routine plasma fractionation [5]. This method was developed and implemented under conditions usually employed in the production of plasma derivatives for clinical and commercial use in orcler to meet Food and Drug Administration requirements for biologic proclucts [61. The utility of the method to be described and the safety of the product prepared have been proved by the use of the preparation in humans [71 and animals [8-101. MATERIALS AND METHODS Starting material for isolation of AT111 was pooled human plasma. Donors were screened to confirm the absence of hepatitis B surface antigen and antibody to human immunocleficiency virus and were determined to have alanine aminotransferase levels within normal limits. Pools were thawed and centrifuged to remove factor VIII-rich cryoprecipitate. Fibrinogen was removed by ethanol precipitation and centrifugation. Vitamin K-dependent coagulation factors were removed by adsorption on diethylaminoethyl (DEAE) Sephadex. Globulins were precipitated and removed by centrifugation as fraction II+III, and fraction IV-l was precipitated by adjusting the pH of the effluent and removed by centrifugation (Figure 1). Fraction IV-1 was chosen as a source material for two reasons: (1) the use of fraction IV-l, usually a byproduct of the fractionation process, would have a negligible impact on the other plasma-derived products; and (2) fraction IV-1 was already in use as a source material for the preparation of alphai-proteinase inhibitor (A,PI) concentrates [ll]. Buffer solutions for the initial affinity chromatography were prepared from 0.01 M tris(hydroxymethyl)aminomethane (Tris)-hydrochloric acid buffer by adding sodium chloride to reach final concentrations of 0.11,0.30, and 2.0 M and adjusting the pH to between 7.5 and 7.8. Diafiltration buffer was prepared from 0.02 M phosphate buffer by adding sodium chloride to reach a final concentration of 0.15 M and adjusting the pH to 7.5. Buffer solutions for final affinity chromatography were prepared from 0.02 M phosphate 11, 1989 The American Journal of Medicine
Volume 87 (suppl 38)
JB-23s
‘SYMPOSIUM
ON ANTITHROMBIN
III I HOFFMAN
Figure 1. Cohn fractionation. F = factor; Fr = fraction; a-l-PI = alpha,-proteinase tnhibitur.
buffer by adcling sodium chloride to reach final concentrations of 0.4 and 2.0 M and adjusting the pH to 7.5. The seconcl cliafiltration buffer was 0.15 M sodium chloride, pH 7.5. Heparine Ultrogel A4R, manufactured by Reactifs IBF, was cletermined to be capable of bincling G IU AT111 per ml wet cake. Before use, the gel was dividecl into two batches, one for the initial separation of ATIII from suspenclecl fraction IV-l, am1 one for the final purification of AT111 to remove denatured protein after heating to inactivate viruses. The gel was extensively washed with 2 M sodium chloride solution and finally equilibrated in either 0.15 M sodium &loride in 0.01 M Tris buffer or 0.4 M sodium chloricle in 0.02 M phosphate buffer, as appropriate for the intenclecl use of the gel. After each use for isolation of ATIII, the gel was recycled by washing with 2 M sodium chloride and storecl at 5°C in 2 M sodium chloride to which ethanol had been added to reach a final concentration of 25 percent. The gel was used over 200 times for large-scale purification of AT111 without any cletectable decrease in aclsorbing capacity. Concentration and cliafiltration in large-scale operation were carriecl out on a custom-designed, closeclloop ultrafiltration system containing six 25-squarefoot surface area Romicon cartriclges with a molecular weight cutoff of 30,000 daltons. The ultrafilter membrane was storecl between uses in 0.5 M sodium hydroxide at 5°C. Sterile filtration was carriecl out with IO-inch or 30inch Duofine 0.2~pm prefilters and lo-inch or 30-inch Millipore Duropore 0.22~.urn sterilizing filters or their equivalent. Isolation of AT111 from fraction IV-l was carried out as follows (Figure 2). The fraction IV-l paste was clissolved in 0.01 M Tris-hydrochloric acid buffer, at a ratio of 1 kg paste to 24 liters of buffer, and the pH was adjustecl to 9.25 with sodium hydroxicle. The solution was warmed to 35°C for one hour and cooled to 5°C. The pH was adjustecl to 7.8, and the soclium chloride concentration was increased by 0.11 M. The solution was filtered to remove undissolved particulates ancl passecl over about 40 liters of Heparine Ultrogel that hacl been previously equilibrated with 0.11 M soclium chloride buffer, pH 7.8. The affinity adsorption step was carried out at a concentration equivalent to about 0.5 g of fraction IV-1 paste per 1 ml gel. The adsorption of AT111 from suspended fraction IV-1 was initially performed in a batchwise process, because 3524s
September 11, 1989
The American Journal of Medicine
the high protein and lipid content of the solution made packed column chromatography too slow to be practical. Propessively larger batch kettles were usecl as techniques evolvecl until a closecl-cell Schenk filter was modified to hold the gel loosely in cells through which the solution was pumped. These batch methocls permittecl suspended fraction IV-l flow rates through the gel three to five times those possible using a packecl column. Fortunately, the adsorption of AT111 to Heparine Ultrogel was relatively insensitive to conditions of pH, ionic strength, and temperature, allowing the range of buffers and temperatures requirecl by the more restrictive procedures involvecl in the purification of AiPI that followed. Unbouncl protein was washed from the gel with 0.3 M soclium chloride at 2o”C, ancl the ATIII was eluted with 2 M sodium chloride. The gel was typically used for two or three adsorption-elution cycles per batch. After eluate was collected, it was concentrated by ultrafiltration and diafiltered with 0.15 M sodium chloride. The purity obtained by this methocl, over 50 percent, was quite acceptable given the high yielcl, since further purification followecl. Soclium citrate was added to reach a concentration of 0.5 M, and the pH was acljusted to 7.55. The solution was heated 10 to 11 hours at GO% and transferrecl aseptically to an isolated procluction area reserved for plasma products treated by methocls proved to inactivate viruses. After cooling the solution, sodium chloride was aclclecl to increase the final concentration to 0.4 M. This solution, at 5”C, was applied to 60 liters of Heparine Ultrogel packed in a Sepragen radial-flow column equilibrated with 0.4 M sodium chloride. Unbound protein was washecl out with aclditional 0.4 M sodium chloride bufferecl solution. Purifiecl AT111 was elutecl with 2 M soclium chloricle. The gel was typically usecl for six to 12 adsorption-elution cycles per batch. After eluate was collected, it was concentrated by ultrafiltration and cliafilterecl with 0.15 M soclium chloride, pH 7.5. Alanine was aclclecl to reach a concentration of 0.1 M. The solution was sterilizecl by filtration, filled into vials, ancl freeze-dried. Biologic activity of ATIII was determinecl as heparin cofactor activity using a two-stage assay method 1121. A sample of ATIII was added to solution containing excess bovine thrombin with heparin as a catalyst, and all AT111 present was inactivatecl by combination with thrombin. The chromogenic substrate S-2238
Volume 87 (suppl 38)
SYMPOSIUM
was then added and reacted with all remaining thrombin, forming a charxteristic color. The amount of color formed was inversely proportional to the amount of biologically active AT111 originally present. Crossecl immunoelectrophoresis was performed in two dimensions in agarose gels 1131. First, the sample was placed in agarose that contained heparin. AT111 bound to heparin migrated in the first dimension more slowly than other proteins present. Second, the sample now separated by heparin affinity was passed into agarose gel impregnated with antibocly to ATIII. AT111 precipitatecl in the second gel and, after staining, was quantified to determine the ratio of heparinbound AT111 to unbound AT111 present. Sterility, safety, and nonpyrogenicity of the AT111 concentrates were determined accorcling to United States Foot1 and Drug Administration regulations. The pyrogen test was carried out using 500 IU of AT111 concentrate per kg body weight of rabbit.
ON ANTITHROMBIN
III I HOFFMAN
r Ill:.c.lltl “Ltr‘llIICr.lt‘O” oi*LILCration I FOrm”ld~LO”. F*IIL”q Frccre Dry‘nq
RESULTS
il, AT,,,
This process cliffers markedly from other purification processes described in that it employs a second heparin affinity purification step to remove all clenaturecl AT111 remaining after heating for virus inactivation. Early methods for purification of AT111 produced acceptable results after one heparin adsorption step. When heat in the presence of citrate was cletermined to be an effective method for virus inactivation, concentr&.es prepared using only one purification step often containecl significant amounts of denatured protein after heat treatment. Other early commercial processes also utilized fraction IV-l as a starting material, but, to our knowledge, none employed a second purification step to remove heat-denatured protein as a part of the virus-inactivation process. A patent has been granted in recognition of the novelty and value of this unique method of viral inactivation ant1 purification [ 141. The AT111 concentrate prepared by the method clescribecl is more than 95 percent pure. The concentration of AT111 throughout the process is summarizecl in Table I. Typical values are listed for AT111 potency, protein concentration, and specific activity throughout the purification process. The specific activity is shown increasing GO-fold after the first Heparine Ultrogel purification step. The specific activity decreases 20 to 25 percent after heat treatment, a result of protein denaturation, but is subsequently increasecl to a theoretical maximum of 8 IU/mg protein after the second Heparine Ultrogel purification. There is no appreciable change in the specific activity during freeze clrying and, in seven recent lots made, the final product has had a specific activity of 7.4 + 1 IU/mg protein. This process has been scaled up to GO kg of starting fraction IV-l with consistent yields ancl purification factors. Virus inactivation studies have been undertaken to clemonstrate the effectiveness of heating for 10 hours at Go”C in the presence of 0.5 M citrate. Experiments performed by the Bureau of Biologics of the Food and Drug Administration showed that heating under these conditions inactivated hepatitis B virus [ 151. Subsequently, using i,jl vit1.0 test methods, we have cletermined that many viruses are similarly inactivated, including human immunocleficiency virus [lG] (Table II). Additionally, stability testing has shown that the
Figure 2. Antithrombin Ill isolabon and purification. Fr = fraction; NaCl = sodium chloride; u-l-PI = alpha,-proteinase inhibitor.
TABLE I Antithrombin III Specific Activity at Major Process Points ATlll Activity* (IUlml) SuspendedCohn fraction WI Punfledconcentrate Pasteurizedconcentrate Repunfred concentrate Lyophrlued fmalproduct
Protein Concentration* hg/ml)
IO
I
Specific . (luA%;kg Protein)
0.1
;z
:
6 5
ii
z
ii
*Typical values, approxrmate.
TABLE II Antithrombin Ill Virus Inactivation Studies Heating 10 Hours at 60°C in 0.5 M Citrate
virus
FirstTime Point Where Nondetectable
Vesicularstomahhs Vaccmra Porcme parve Herpes simplex I Polio II Cytomegalowus Vrsna Sindbrs Human rmmunodefrcrency VINS
30 minutes 30 minutes 5 hours 30 mmutes 30 mmutes 30 mmutes I hour 2 hours 1 hour
J
product retains over 90 percent of its biologic activity after two years of storage [171. In conclusion, we have developed and implemented a method for purification and large-scale preparation of AT111 that usesa unique combination of heat treatment and repurification to effect viral inactivation. This method yields a solution-pasteurized product, freeze dried in vials of 500 to 1,000 units each, with a purity greater than 95 percent. It produces a single
September 11,
1989 The American Journal of Medicine
Volume 87 (suppl 38)
30-25s
SYMPOSIUM
ON ANllTHROMl3lN
Ill I HOFFMAN
band on crossed immunoelectrophoresis, has a specific activity of not less than 6.4 IU/mg protein, and contains not more than 0.004 units of heparin per IU ATIII.
ACKNOWLEDGMENT Research performed under the direction of Dr. Robert Jordan provided the basis for much of the information contained in this report. Therefore, his assistance in recalling major research milestones and his advice during preparation of this article is gratefully acknowledged.
REFERENCES 1. Abrldgaard U: Punkcahon of two progressive antdhrombrns of human plasma. Stand J Ckn Lab Invest 1967; 19: 190-195. 2 Rosenberg RD. Damus PS: The purdrcabon and mechamsm of action of human ant+ thrombinheparm cofactor. J Brol Chem 1973: 248: 6490-6505. 3. Miller-Andersson M. Borg H. Andenson LO: Purification of antdhrombrn III by affrnrty chromatography. Thromb Res 1974: 5: 439-452. 4. Thaler E. Schmer G: A sample two-step rsolatron procedure for human and bovrne antdhrombrn 111111(hepann cofactor): a comparison of two methods. Br J Haematol 1975; 31: 233-243. 5. Cohn EJ. Strong LE. Hughes WL Jr, el a/: Preparabon and properhes of serum and plasma proterns. IV. A system for the separatton Into frachons of the protein and flurds. J
36-26s
September
11, 1989 The American
Journal
of Medicine
Volume
Am Chem Sot 1946; 68: 459-475. 6. Code of Federal Regulatrons, No. 21. Parts 600-1299 (Food and Drug Admvxstratron. 1988). 7. Schwartz RS. Kavanagh E. Davies 0. Bogdanoff DA Clrnrcal experience wrth anhthrom bin Ill concentrate for prophylaxes and treatment of congenrtal and acqurred antrthrombrn III defrcrency. Am J Mad 1989; 87 (suppl 38): 3853S-38.605. 8. Emerson TE Jr, Fournel MA, Leach WJ, Redens TB: Protectron against drssemrnated rntravascular coagulahon and death by anhthrombrn Ill rn the Escher&m co/r en dotoxemrc rat. Crrc Shock 1987; 21: l-13. 9. Redens TB, Leach WJ. Bogdanoff DA, Emerson TE Jr: Synergrstrc protechon from lung damage by pretreatment wrth the combrnahon of antrthrombm Ill and alpha.1 protemase rnhrbrtor In the E co11 endotoxemrc sheep pulmonary dysfunchon model. Crrc Shock 1988, 26: 15-26. 10. Taylor FB. Emerson TE Jr. Jordan RE. Chang AK, Blrck KE: Anhthrombrn Ill prevents the lethal effects of Eschenchfa cob rnfusron In baboons. Crrc Shock 1988; 26 (3): 227-235. 11. Shearer MA, Sasagawa PK. Hern RH: Method of preparmg alpha.1 protemase mhrbrtor and anhthrombrn Ill. Unrted States Patent No. 4. 656, 254; 1987. 12. Abrldgaard U. Lre M, Odegard OR: Anhthrombrn (hepann cofactor) assay with “new” chromogenrc substrates (S-2238 and Chromozym TH). Thromb Res 19R 11: 549-553. 13. Barrowcldfe TW: Studres of hepann bmdmg to antrthrombrn Ill by crossed rmmuno. electrophoresrs. Thromb Haemost 1980; 42: 1434-1445. 14. Jordan RE. Krlpatnck J: Viral machvatron and punfrcahon of actrve proterns. Umted States Patent No. 4. 749. 783: 1988. 15. Tabor E. Murano G. Snoy P, Gerety RI: Drvrsron of Blood and Blood Products, and Drvrsron of Qualrty Control, Bureau of Brologrcs. Food and Drug Admrnrstratron: lnachvatron of hepatrhs B vrrus by heat In antrthrombrn Ill stabrlrzed wrth citrate. Thromb Res 1981; 22: 233-238. 16. Cutter Brologrcal, Miles, Inc. Unpublrshed vrrus tnactrvatron data 17. Cutter Brologrcal. Mrles, Inc. Unpubkshed stabrlrty testing data.
87 (suppl 36)
and Large-Scale III
Preparation
of
DAVID L. HOFFMAN, B.A. Berkeley. ~/ifoma
Cutter Biological, Miles Inc., has implemented large-scale isolation and purification of antithrombin III (AT1111 from human blood plasma. The starting material, Cohn fraction IV-l, is suspended and AT111 separated from other proteins by heparin-affinity chromatography. The concentrate is subsequently heated for 10 hours at 60°C in a solution of 0.5 M sodium citrate to inactivate the causative agents of viral hepatitis B and acquired immune deficiency syndrome. After heating, a second heparin-affinity chromatography step is employed to isolate the active AT111 and to remove heat-denatured protein. The purified solution is compounded with 0.1 M alanine and 0.15 M sodium chloride, filled in vials of 500 or 1,000 IU each, and freeze dried. The final product has a specific activity of not less than 6.4 IU AT111 per mg protein, of which over 90 percent binds to heparin on crossed immunoelectrophoresis. Preclinical and clinical testing have shown this product to be safe for prophylactic and therapeutic administration to persons with congenital deficiency of ATIII.
From Ihe Department of Plasma Manufacturing Technology, Cutter Bwloglcal. Miles Inc.. Berkeley, C-allforma. Requests for reprmts should be addressed to David L. Hoff. man. Cutter Blologlcal. Miles Inc.. 4th and Parker Streets, Berkeley. Califorma 94710.
September
he first methods reported for the isolation of antiT thrombin III (ATIII) from human plasma by conventional techniques were lengthy, with poor yields [1,2]. A significant improvement in AT111 methodology occurred when heparin bound to an insoluble matrix was introduced as an affinity adsorbent [3,4]. However, the methods described still included procedures more suited to small-scale purification, which restricted their usefulness in commercial applications. Because sizable amounts of AT111 were needed for clinical evaluation and eventual commercial use, more efficient methods were required. Therefore, we have developed a large-scale method for the preparation of a highly purified AT111 concentrate based on the use of Heparine Ultrogel. The method utilizes Cohn fraction IV-l, an unused by-product of routine plasma fractionation [5]. This method was developed and implemented under conditions usually employed in the production of plasma derivatives for clinical and commercial use in orcler to meet Food and Drug Administration requirements for biologic proclucts [61. The utility of the method to be described and the safety of the product prepared have been proved by the use of the preparation in humans [71 and animals [8-101. MATERIALS AND METHODS Starting material for isolation of AT111 was pooled human plasma. Donors were screened to confirm the absence of hepatitis B surface antigen and antibody to human immunocleficiency virus and were determined to have alanine aminotransferase levels within normal limits. Pools were thawed and centrifuged to remove factor VIII-rich cryoprecipitate. Fibrinogen was removed by ethanol precipitation and centrifugation. Vitamin K-dependent coagulation factors were removed by adsorption on diethylaminoethyl (DEAE) Sephadex. Globulins were precipitated and removed by centrifugation as fraction II+III, and fraction IV-l was precipitated by adjusting the pH of the effluent and removed by centrifugation (Figure 1). Fraction IV-1 was chosen as a source material for two reasons: (1) the use of fraction IV-l, usually a byproduct of the fractionation process, would have a negligible impact on the other plasma-derived products; and (2) fraction IV-1 was already in use as a source material for the preparation of alphai-proteinase inhibitor (A,PI) concentrates [ll]. Buffer solutions for the initial affinity chromatography were prepared from 0.01 M tris(hydroxymethyl)aminomethane (Tris)-hydrochloric acid buffer by adding sodium chloride to reach final concentrations of 0.11,0.30, and 2.0 M and adjusting the pH to between 7.5 and 7.8. Diafiltration buffer was prepared from 0.02 M phosphate buffer by adding sodium chloride to reach a final concentration of 0.15 M and adjusting the pH to 7.5. Buffer solutions for final affinity chromatography were prepared from 0.02 M phosphate 11, 1989 The American Journal of Medicine
Volume 87 (suppl 38)
JB-23s
‘SYMPOSIUM
ON ANTITHROMBIN
III I HOFFMAN
Figure 1. Cohn fractionation. F = factor; Fr = fraction; a-l-PI = alpha,-proteinase tnhibitur.
buffer by adcling sodium chloride to reach final concentrations of 0.4 and 2.0 M and adjusting the pH to 7.5. The seconcl cliafiltration buffer was 0.15 M sodium chloride, pH 7.5. Heparine Ultrogel A4R, manufactured by Reactifs IBF, was cletermined to be capable of bincling G IU AT111 per ml wet cake. Before use, the gel was dividecl into two batches, one for the initial separation of ATIII from suspenclecl fraction IV-l, am1 one for the final purification of AT111 to remove denatured protein after heating to inactivate viruses. The gel was extensively washed with 2 M sodium chloride solution and finally equilibrated in either 0.15 M sodium &loride in 0.01 M Tris buffer or 0.4 M sodium chloricle in 0.02 M phosphate buffer, as appropriate for the intenclecl use of the gel. After each use for isolation of ATIII, the gel was recycled by washing with 2 M sodium chloride and storecl at 5°C in 2 M sodium chloride to which ethanol had been added to reach a final concentration of 25 percent. The gel was used over 200 times for large-scale purification of AT111 without any cletectable decrease in aclsorbing capacity. Concentration and cliafiltration in large-scale operation were carriecl out on a custom-designed, closeclloop ultrafiltration system containing six 25-squarefoot surface area Romicon cartriclges with a molecular weight cutoff of 30,000 daltons. The ultrafilter membrane was storecl between uses in 0.5 M sodium hydroxide at 5°C. Sterile filtration was carriecl out with IO-inch or 30inch Duofine 0.2~pm prefilters and lo-inch or 30-inch Millipore Duropore 0.22~.urn sterilizing filters or their equivalent. Isolation of AT111 from fraction IV-l was carried out as follows (Figure 2). The fraction IV-l paste was clissolved in 0.01 M Tris-hydrochloric acid buffer, at a ratio of 1 kg paste to 24 liters of buffer, and the pH was adjustecl to 9.25 with sodium hydroxicle. The solution was warmed to 35°C for one hour and cooled to 5°C. The pH was adjustecl to 7.8, and the soclium chloride concentration was increased by 0.11 M. The solution was filtered to remove undissolved particulates ancl passecl over about 40 liters of Heparine Ultrogel that hacl been previously equilibrated with 0.11 M soclium chloride buffer, pH 7.8. The affinity adsorption step was carried out at a concentration equivalent to about 0.5 g of fraction IV-1 paste per 1 ml gel. The adsorption of AT111 from suspended fraction IV-1 was initially performed in a batchwise process, because 3524s
September 11, 1989
The American Journal of Medicine
the high protein and lipid content of the solution made packed column chromatography too slow to be practical. Propessively larger batch kettles were usecl as techniques evolvecl until a closecl-cell Schenk filter was modified to hold the gel loosely in cells through which the solution was pumped. These batch methocls permittecl suspended fraction IV-l flow rates through the gel three to five times those possible using a packecl column. Fortunately, the adsorption of AT111 to Heparine Ultrogel was relatively insensitive to conditions of pH, ionic strength, and temperature, allowing the range of buffers and temperatures requirecl by the more restrictive procedures involvecl in the purification of AiPI that followed. Unbouncl protein was washed from the gel with 0.3 M soclium chloride at 2o”C, ancl the ATIII was eluted with 2 M sodium chloride. The gel was typically used for two or three adsorption-elution cycles per batch. After eluate was collected, it was concentrated by ultrafiltration and diafiltered with 0.15 M sodium chloride. The purity obtained by this methocl, over 50 percent, was quite acceptable given the high yielcl, since further purification followecl. Soclium citrate was added to reach a concentration of 0.5 M, and the pH was acljusted to 7.55. The solution was heated 10 to 11 hours at GO% and transferrecl aseptically to an isolated procluction area reserved for plasma products treated by methocls proved to inactivate viruses. After cooling the solution, sodium chloride was aclclecl to increase the final concentration to 0.4 M. This solution, at 5”C, was applied to 60 liters of Heparine Ultrogel packed in a Sepragen radial-flow column equilibrated with 0.4 M sodium chloride. Unbound protein was washecl out with aclditional 0.4 M sodium chloride bufferecl solution. Purifiecl AT111 was elutecl with 2 M soclium chloricle. The gel was typically usecl for six to 12 adsorption-elution cycles per batch. After eluate was collected, it was concentrated by ultrafiltration and cliafilterecl with 0.15 M soclium chloride, pH 7.5. Alanine was aclclecl to reach a concentration of 0.1 M. The solution was sterilizecl by filtration, filled into vials, ancl freeze-dried. Biologic activity of ATIII was determinecl as heparin cofactor activity using a two-stage assay method 1121. A sample of ATIII was added to solution containing excess bovine thrombin with heparin as a catalyst, and all AT111 present was inactivatecl by combination with thrombin. The chromogenic substrate S-2238
Volume 87 (suppl 38)
SYMPOSIUM
was then added and reacted with all remaining thrombin, forming a charxteristic color. The amount of color formed was inversely proportional to the amount of biologically active AT111 originally present. Crossecl immunoelectrophoresis was performed in two dimensions in agarose gels 1131. First, the sample was placed in agarose that contained heparin. AT111 bound to heparin migrated in the first dimension more slowly than other proteins present. Second, the sample now separated by heparin affinity was passed into agarose gel impregnated with antibocly to ATIII. AT111 precipitatecl in the second gel and, after staining, was quantified to determine the ratio of heparinbound AT111 to unbound AT111 present. Sterility, safety, and nonpyrogenicity of the AT111 concentrates were determined accorcling to United States Foot1 and Drug Administration regulations. The pyrogen test was carried out using 500 IU of AT111 concentrate per kg body weight of rabbit.
ON ANTITHROMBIN
III I HOFFMAN
r Ill:.c.lltl “Ltr‘llIICr.lt‘O” oi*LILCration I FOrm”ld~LO”. F*IIL”q Frccre Dry‘nq
RESULTS
il, AT,,,
This process cliffers markedly from other purification processes described in that it employs a second heparin affinity purification step to remove all clenaturecl AT111 remaining after heating for virus inactivation. Early methods for purification of AT111 produced acceptable results after one heparin adsorption step. When heat in the presence of citrate was cletermined to be an effective method for virus inactivation, concentr&.es prepared using only one purification step often containecl significant amounts of denatured protein after heat treatment. Other early commercial processes also utilized fraction IV-l as a starting material, but, to our knowledge, none employed a second purification step to remove heat-denatured protein as a part of the virus-inactivation process. A patent has been granted in recognition of the novelty and value of this unique method of viral inactivation ant1 purification [ 141. The AT111 concentrate prepared by the method clescribecl is more than 95 percent pure. The concentration of AT111 throughout the process is summarizecl in Table I. Typical values are listed for AT111 potency, protein concentration, and specific activity throughout the purification process. The specific activity is shown increasing GO-fold after the first Heparine Ultrogel purification step. The specific activity decreases 20 to 25 percent after heat treatment, a result of protein denaturation, but is subsequently increasecl to a theoretical maximum of 8 IU/mg protein after the second Heparine Ultrogel purification. There is no appreciable change in the specific activity during freeze clrying and, in seven recent lots made, the final product has had a specific activity of 7.4 + 1 IU/mg protein. This process has been scaled up to GO kg of starting fraction IV-l with consistent yields ancl purification factors. Virus inactivation studies have been undertaken to clemonstrate the effectiveness of heating for 10 hours at Go”C in the presence of 0.5 M citrate. Experiments performed by the Bureau of Biologics of the Food and Drug Administration showed that heating under these conditions inactivated hepatitis B virus [ 151. Subsequently, using i,jl vit1.0 test methods, we have cletermined that many viruses are similarly inactivated, including human immunocleficiency virus [lG] (Table II). Additionally, stability testing has shown that the
Figure 2. Antithrombin Ill isolabon and purification. Fr = fraction; NaCl = sodium chloride; u-l-PI = alpha,-proteinase inhibitor.
TABLE I Antithrombin III Specific Activity at Major Process Points ATlll Activity* (IUlml) SuspendedCohn fraction WI Punfledconcentrate Pasteurizedconcentrate Repunfred concentrate Lyophrlued fmalproduct
Protein Concentration* hg/ml)
IO
I
Specific . (luA%;kg Protein)
0.1
;z
:
6 5
ii
z
ii
*Typical values, approxrmate.
TABLE II Antithrombin Ill Virus Inactivation Studies Heating 10 Hours at 60°C in 0.5 M Citrate
virus
FirstTime Point Where Nondetectable
Vesicularstomahhs Vaccmra Porcme parve Herpes simplex I Polio II Cytomegalowus Vrsna Sindbrs Human rmmunodefrcrency VINS
30 minutes 30 minutes 5 hours 30 mmutes 30 mmutes 30 mmutes I hour 2 hours 1 hour
J
product retains over 90 percent of its biologic activity after two years of storage [171. In conclusion, we have developed and implemented a method for purification and large-scale preparation of AT111 that usesa unique combination of heat treatment and repurification to effect viral inactivation. This method yields a solution-pasteurized product, freeze dried in vials of 500 to 1,000 units each, with a purity greater than 95 percent. It produces a single
September 11,
1989 The American Journal of Medicine
Volume 87 (suppl 38)
30-25s
SYMPOSIUM
ON ANllTHROMl3lN
Ill I HOFFMAN
band on crossed immunoelectrophoresis, has a specific activity of not less than 6.4 IU/mg protein, and contains not more than 0.004 units of heparin per IU ATIII.
ACKNOWLEDGMENT Research performed under the direction of Dr. Robert Jordan provided the basis for much of the information contained in this report. Therefore, his assistance in recalling major research milestones and his advice during preparation of this article is gratefully acknowledged.
REFERENCES 1. Abrldgaard U: Punkcahon of two progressive antdhrombrns of human plasma. Stand J Ckn Lab Invest 1967; 19: 190-195. 2 Rosenberg RD. Damus PS: The purdrcabon and mechamsm of action of human ant+ thrombinheparm cofactor. J Brol Chem 1973: 248: 6490-6505. 3. Miller-Andersson M. Borg H. Andenson LO: Purification of antdhrombrn III by affrnrty chromatography. Thromb Res 1974: 5: 439-452. 4. Thaler E. Schmer G: A sample two-step rsolatron procedure for human and bovrne antdhrombrn 111111(hepann cofactor): a comparison of two methods. Br J Haematol 1975; 31: 233-243. 5. Cohn EJ. Strong LE. Hughes WL Jr, el a/: Preparabon and properhes of serum and plasma proterns. IV. A system for the separatton Into frachons of the protein and flurds. J
36-26s
September
11, 1989 The American
Journal
of Medicine
Volume
Am Chem Sot 1946; 68: 459-475. 6. Code of Federal Regulatrons, No. 21. Parts 600-1299 (Food and Drug Admvxstratron. 1988). 7. Schwartz RS. Kavanagh E. Davies 0. Bogdanoff DA Clrnrcal experience wrth anhthrom bin Ill concentrate for prophylaxes and treatment of congenrtal and acqurred antrthrombrn III defrcrency. Am J Mad 1989; 87 (suppl 38): 3853S-38.605. 8. Emerson TE Jr, Fournel MA, Leach WJ, Redens TB: Protectron against drssemrnated rntravascular coagulahon and death by anhthrombrn Ill rn the Escher&m co/r en dotoxemrc rat. Crrc Shock 1987; 21: l-13. 9. Redens TB, Leach WJ. Bogdanoff DA, Emerson TE Jr: Synergrstrc protechon from lung damage by pretreatment wrth the combrnahon of antrthrombm Ill and alpha.1 protemase rnhrbrtor In the E co11 endotoxemrc sheep pulmonary dysfunchon model. Crrc Shock 1988, 26: 15-26. 10. Taylor FB. Emerson TE Jr. Jordan RE. Chang AK, Blrck KE: Anhthrombrn Ill prevents the lethal effects of Eschenchfa cob rnfusron In baboons. Crrc Shock 1988; 26 (3): 227-235. 11. Shearer MA, Sasagawa PK. Hern RH: Method of preparmg alpha.1 protemase mhrbrtor and anhthrombrn Ill. Unrted States Patent No. 4. 656, 254; 1987. 12. Abrldgaard U. Lre M, Odegard OR: Anhthrombrn (hepann cofactor) assay with “new” chromogenrc substrates (S-2238 and Chromozym TH). Thromb Res 19R 11: 549-553. 13. Barrowcldfe TW: Studres of hepann bmdmg to antrthrombrn Ill by crossed rmmuno. electrophoresrs. Thromb Haemost 1980; 42: 1434-1445. 14. Jordan RE. Krlpatnck J: Viral machvatron and punfrcahon of actrve proterns. Umted States Patent No. 4. 749. 783: 1988. 15. Tabor E. Murano G. Snoy P, Gerety RI: Drvrsron of Blood and Blood Products, and Drvrsron of Qualrty Control, Bureau of Brologrcs. Food and Drug Admrnrstratron: lnachvatron of hepatrhs B vrrus by heat In antrthrombrn Ill stabrlrzed wrth citrate. Thromb Res 1981; 22: 233-238. 16. Cutter Brologrcal, Miles, Inc. Unpublrshed vrrus tnactrvatron data 17. Cutter Brologrcal. Mrles, Inc. Unpubkshed stabrlrty testing data.
87 (suppl 36)