(4 The Is01 tion of Rabbit Insulin Antibodies'#' EDWARD R. ARQUILLA AND CHARLOTTE COBLENCE Department of Pathotogy, School of Medicine, University o f Southern California, Los Angeles, California This investigation is concerned with the isolation of tabbit insulin antibodies by elution from an insoluble insulin complex. The ebtion of antibodies from insoluble anti$en antibody aggregates has been previously demonstrated (Campbell, Luescher and Lerman, '51; Isliker, '51). Campbell, buescher and Lerman ('51) prepared an insoluble antigen by conjugating bovin serum albumin to a diazobenzene-cellu ose derivative. They were able to foqm an insoluble antibody antigen aggaegate and then elute the bovine serum albumin antibodies from this preparatibn at a low pH. In the follbwing experiments the insoluble antibod3 insulin aggregate was prepared by adsorbing antibodies from rabbit insulin antiserum onto an insoluble insulin complek made by conjugating insulin to an erythrocyte stroma-cellulose mixture. The antibody insulin aggregate was isolated from proteins and other materials in serum by washing with saline. The e ect of pH, temperature and salt concentr tion on the dissociation of antibody from insulin was studied. Berson and Yalow ('59), using Pi-labeled insulin and whole human insulin antiserum, were llble to estimate the dissociation constants of insulin antibody complex. They qoncluded that insulin has two antigenic sites, each having different dissociation onstants. In these experiments the e ct of pH, temperature and salt concentr tion on the dissociation or" antibody fro insulin in an isolated system was stu 'ed. It appears that insulin like other pr tein antigens may have a multiplicity o antigenic sites, the binding properties of which are not well understood. By selectin the optimum pH, temperature and sal concentration for the dis- 1 s" i 8 ti sociation of antibody from insulin it has been possible to purify the insulin antibody from rabbit antiserum about 500fold with approximately a 30% recovery. METHODS AND MATERIALS Antigen. The antigen used in preparing the rabbit insulin antiserum was crystalline beef i n ~ u l i n .This ~ is a sample of zinc crystals of bovine insulin having 27 units per mg. There was some hyperglycemic factor present in this preparation. Erythrocyte preparations. The sheep erythrocytes used for the titration of antibody activity were stored as sheep blood in an equal part of Alsever's solution (Bukantz, Rein and Kent, '46). This blood was stored at 5°C and if not used within three weeks was discarded. Antiserum. Adult male white rabbits were immunized with alum-precipitated insulin as has previously been described by ArquilIa and Stavitsky ('56). The antiserum obtained was stored in aliquots at -60°C. Titration of insulin antibody activity. The insulin antibody titer of these sera was measured by the hemagglutination method, employing sheep erythrocytes to which insulin was conjugated with bisdiazobenzidine, (ArquiUa and Stavitsky, '56). The diluent used for titrations was verond buffered saline, pH 7.4 (Kabat and Mayer, '48), containing 5 X lo-' M MgClp M CaCL in which 1.5 mg and 1.5 X bovine serum albumin (Armour Co.) per 'This work is supported by U.S.P.H.S. grant A-3541 (Sl). * A report on this work was presented to the Annual Meeting of the American Diabetes Association in Miami Beach, June, 1960. wish to thank h.Otto K. Bebrens, Eli Lilly, for the generous supply of crystalline beef insulin. 203 204 EDWARD R. ARQUILLA AND CHARLOTTE COBLENCE packed heavy layer and a lighter fluffy layer. As much of the clear solution as possible was drawn off with a capillary pipette. The lighter layer could then be separated from the heavier layer by gently rinsing the tube with 0.11 M phosphate buffer. The remaining precipitate was also suspended in 0.11 M phosphate buffer. Both portions of the stroma were then dialyzed against distilled water and lyophilized. The lighter top fraction of the sheep erythrocyte stroma (SI) was used in making the cellulose stroma complex to which insulin was conjugated. Preparation of the cellulose stroma mixture. Eight grams of Whatman’s cellulose powder (type A ashless standard grade) were added to 2 gm of S,. This material was suspended in 50 ml of saline and homogenized in a ground glass hand homogenizer so that equal distribution of the stroma and cellulose could be obtained. This mixture was diluted to 600 ml with saline and allowed to stand at 5°C for 24 hours with constant gentle stirring. The mixture was then washed with saline an additional 5 times, each wash containing about 600 ml. Hemoglobin was present in the first 4 washes, whereas the last liter was dissolved. Prior to use the diluent was incubated at 80°C for one hour and then cooled to 2°C in a water bath. Much of the nonspecific agglutination observed at high dilutions of antiserum was eliminated with this diluent. Protein determination. Protein concentration was determined by the method of Lowry, Rosebrough, Farr and Randall (’51). Preparation of sheep erythrocgte stroma. Sheep’s blood stored with an equal part of Alsever’s solution was washed three times with saline. The packed erythrocytes were suspended with an equal volume of saline and dialyzed against 20 volumes of distilled water. This suspension was centrifuged at 18,000 g’s in a Servall angle centrifuge and the precipitate washed with water until the supernatant was a pale pink. The precipitate was then washed with 0.11 M phosphate buffer, pH 7.4, at 40,000 g’s for 30 minutes at 2°C in a Spinco ultracentrifuge until the supernatant was very faintly pink. Finally the precipitate was suspended in 0.11 M phosphate buffer, pH 7.4, and centrifuged at 18,000 g’s. Two distinct layers of precipitate formed, a tightly To 40 mg of stroma cellulose suspension (in 4 ml) add 5 mg insulin (in 2.5 ml) and 1 ml BDB (1 to 15 with phosphate buffer pH 7.4). Incubate at room temperature for 20 minutes while shaking and centrifuge at 2100 g’s. I Supernatant Precipitate L &- Wash with 20 ml saline. Add 4 ml antiserum and incubate at room temperature while shaking. Centrifuge for 30 minutes at 2100 g’s at 2°C. Discard I Supernatant J Precipitate & Absorbed antiserum. Save for titration and protein determination. Precipitate J Discard Washed with 20 rnl saline at 2°C Elute with 12 ml of 5X buffered saline, pH 3.2, at 37°C for 10 minutes while stirring. Centrifuge 30 minutes 2100 g’s at 2°C I Dialyzed against buffered saline pH 7.4 Fig. 1 Procedure for elution of rabbit insulin antibodies. 205 ISOLATION O F INSULIN ANTIBODIES two washes appeared to be grossly free of hemoglotfn. The resultant white precipitate (C was dialyzed against distilled water and lyophilized. Preparatidn of insulin conjugated stroma cellulose mixture. The CS, mixture was su pended in saline in a concentration of 1 mg per ml. Crystalline beef insulin was made up in a solution containing 2 m per ml (fig. 1). To 4 ml of the evenly uspended CS,, 2.5 ml of insulin solutiop containing 5 mg of insulin was added pnd evenly suspended. One ml of bis-diasobenzidine (BDB) was added to this mixture, and then gently agitated at room temperature for 20 minutes and centrifuged for 10 minutes at 2,100 g’s. The supernatant was discarded and the button washed twice with 20 ml of saline. This amount of insoluble insulin complex (ICSI) was used to adsorb the antibodies from 4 ml of rabbit insulin antiserum with the subsequent formation of the insoluble antibody insulin complex (AbICS;). 8 $ RESULTS Adsorption of antibodies from antiserum with insoluble insulin complex. The ability of insulin antiserum to agglutinate insulin-sensitized erythrocytes was markedly decreased following the incubation of the antisesum with ICSl (table 1). Four ml of rabbit insulin antiserum were added to an insoluble insulin complex made with 10 mg of CSI and 5 mg of insulin. This mixture was incubated for 30 minutes at room temperature with constant stirring, and then centrifuged (fig. 1 ) . The anti-insulin activity in the supernatants from 5 different antisera treated in this manner were compared with the activity of aliquots from the same antisera which had not been incubated with ICSl (table 1). TABLE 1 Adsorption of insulin antibodies from antiserum incubated unith insoluble insulin complex Antiserum A B - ,. r D E Titer antiserum not incubated Titer antiserum after. incubation 1/3200 1/420 1/5120 1/50 1/25 1 /40 1/10 1J40 Most of the insulin antibodies were adsorbed out of the rabbit antiserum by the insoluble insulin complex. Following this adsorption, the insoluble antibodyinsulin complex (AbICSI) was washed twice with 20 ml of saline. Most of the weakly bound serum proteins were removed with the first saline wash. There were a small amount of insulin antibodies present in the first saline wash. No detectable antibodies were present in the second saline wash. The small amount of insulin antibodies consistently observed in the first saline washes probably represent insulin antibodies in the residual serum which was in the liquid phase of the AbICS, (table 2). TABLE 2 Effectiveness of saline washes on insoluble antibodg insulin complex Titer Antiserum Saline wash 1 Saline wash 2 < 1/6400 1/8 I/1 Protein concentration mglml 83.0 1.26 0.038 About 80 to 85% of the total serum proteins were recovered in the adsorbed antiserum. Between 2 and 3% of the total proteins were present in the first saline wash, and about 0.5% in the second saline wash. Between 0.05 and 0.1% of the total proteins were recovered as purified insulin antibodies. What portion of the unaccountable (15 to 20% ) total serum proteins were still bound to insoluble aggregate is unknown. Factom influencing dissociation of antibody from the antibody insulin complex. The effect of pH, temperature, and salt concentration on the elution of insulin antibodies from AblCS, was investigated. The pH of the solution employed for eluting the antibodies was found to be the most critical of these three parameters. At pH levels above 3.5 the amount of antibodies dissociated decreased markedly. Variations in salt concentration and temperature had minor effect on the amount of insulin antibody activity recovered in the eluate (fig. 2). Optimal conditions for the dissociation of antibodies from AbIC& were obtained 206 EDWARD R. ARQUILLA AND CHARLOTTE COBLENCE EFFECT OF IONIC STRENGTH pH 3.2 TEMP. 0.C 512 256 i 128 W L .- & c 64 32 0 IX ox EX TIMES SALINE 2x 5x EFFECT OF TEMPERATURE pH 3.2 IONIC STRENGTH 5 X SALINE with a solution containing 9 parts of 0.77 M NaCl and one part of 1 M citric acid, pH adjusted to 3.2 with NaOH (fig. 1). A volume of this eluting solution equal to three times the amount of antiserum was added to the washed AbICSl and incubated at 37°C for 10 minutes with constant gentle shaking. The aggregate was then centrifuged, the supernatant decanted and adjusted to pH 7 with NaOH and dialyzed against saline (fig. 1). Recovery o f antibody. Using the above conditions, it was possible to recover approximately one-third of the insulin antibody activity in the eluate which contained about 1/1400th of the protein concentration of the serum from which the antibody was obtained (table 3). TABLE 3 The elution of rabbit insulin antibodies 512 Titer Protein 1/12800 80,000 1/50 1/48001 61,000 19 r/ml Antiserum Absorbed antiserum Eluate 256 128 Q) L .c c 'The observed titer was multiplied X 3 since the elution volume was 3X the volume of the antiserum used. 64 32 The antibodies in the eluate were apparently not altered to any detectable degree 0 25 37 56 by the method of elution. The antibodies TEMPERATURE were still capable of agglutinating insulinsensitized erythrocytes. If complement was added to the system, hemolysis of the EFFECT OF pH insulin-sensitized erythrocytes took place. IONIC STRENGTH S X SALINE Both the hemolysis and agglutination of TEMP. 0.C the insulin-sensitized erythrocytes by the purified antibodies were inhibited with crystalline insulin. 512 Residual binding of antibodies to in256 soluble insulin aggregate. In order to determine whether residual antibodies to I28 L insulin still remained attached to the inc sulin complex after elution, the complex c 64 was labeled with fluorescent anti-rabbit 32 gamma globulin antibody (fig. 3). It was necessary to minimize the degree I 0' of nonspecific staining of the insulin complex by adsorbing the fluorescent conjuPH gate with 10 mg/ml ICW. The adsorbed Fig. 2 Parameters for the elution of rabbit in- fluorescent conjugate still contained antibodies to rabbit gamma globulin, whereas sulin antibodies. n - I Q) .- - ISOLATION OF INSULI N ANTIBODIES 207 Fig. 3 Residual binding of antibody to insoluble insulin aggregates as demonstrated with fluorkscent anti-rabbit gamma globulin. the antibodiels to stroma and much of the fluorescent npaterials which attached nonspecifically tQ the insulin cellulose stroma complex werR removed by this prior adsorption. This ads bed fluorescent anti-rabbit gamma glob lin conjugate was then incubated for 20 minutes at room temperature with AbICS, which had been subjected to three elution similar to those described previously. he complex was then centrifuged and ashed twice with saline. A smear of the utton was made on a microscopic slide nd examined by fluorescent microscopy (fig. 3). There was marked staining of th aggregate by the fluorescent anti-rabbit gamma globulin antibody. It therefore applears that rabbit antibody remains attached to the insulin complex in spite of repealted elutions with highly concentrated salU solutions at pH 3.2. There is sohne question whether the fluorescent conjogate was staining only antibodies attached to insulin. The possibility that rabbit amma globulin other than insulin antib dy was stained by the fluorescent antibody cannot be eliminated. 7 4 1 L i DISCUSSION The use of a stroma cellulose mixture for the preparation of an insoluble antigen has versatility since once prepared it is quite stable and within a relatively short period of time almost any antigen can be conjugated to it with bis-diazobenzidine. The diazobenzene derivative of cellulose, as described by Campbell, Luescher and Lerman ('51 ) requires the preparation of the complete antigen and therefore can be used only for the isolation of one species of antibodies. The amount of stroma protein which will contaminate preparations of eluted purified antibody is an unknown but a real consideration when using AbICS1. Theoretically this danger does not exist with the use of the diazobenzene derivative of cellulose. The insoluble insulin aggregate was found to be very effective in adsorbing out antibodies from insulin antisera. In these experiments antiserum was used in excess so as to insure a complete saturation of all available insulin. Therefore, there were residual antibodies in the anti- 208 EDWARD R . ARQUILLA A N D CHARLOTTE COBLENCE sera adsorbed with ICS,. In other experiments it has been possible to remove all of the demonstrable antibodies in antisera by employing slightly larger quantities of ICS, to adsorb the antibodies from the antisera. The purified antibody which was eluted from the aggregate was stable at 0°C for at least two weeks. Dialysis against water and subsequent lyophilization resulted in complete loss of anti-insulin activity as measured by the hemagglutination reaction. Acetone precipitation of the eluate also caused the loss of the antibody activity. In preliminary experiments it has been possible to concentrate the eluate at least 100-fold by dialysis against 25% polyvinylpyrrolidone in water. Preliminary studies on the homogeneity of the isolated antibody indicate that there are three components which can be isolated by column chromatography. The characteristics of the various components are presently being investigated. It is difficult to accept the concept of Berson and Yalow (’59) that insulin has two antigenic sites, and that to each of these sites there is a single species of antibodies and that each has a different dissociation constant. It appears that insulin antibodies can be dissociated from an isolated antigen antibody aggregate at various pH’s, temperatures, and salt concentrations to varying degrees. In spite of repeated elutions at conditions where t h e optimum amounts of antibodies could be dissociated from the insoluble insulin complex, only 30% of the antibody activity was recovered and there appeared to be considerable amounts of rabbit gamma globulin remaining adherent to the antibody insulin complex as was demonstrated by the fluorescent antibody technique. Whether insulin has only two antigenic sites is still open to question. The insulin molecule does have a considerable number of amino acids (e.g., aspartic, glutamic and phenylalanine, etc.) which according to the studies of Landsteiner (‘45) are good antigenic determinants. Landsteiner (’45) has also demonstrated a number of antibodies with different binding capacities against a single antigenic site. It therefore appears likely that insulin, like other proteins, is an antigen which has a multiplicity of antigenic sites and to which there are a multiplicity of antibodies with varying binding capacities to each of these sites. SUMMARY 1. It has been possible to prepare an insoluble insulin aggregate by conjugatinq insulin to a cellulose stroma complex with bis-diazobenzidine. 2. This insoluble insulin aggregate is effective in adsorbing antibodies from rabbit insulin antiserum. 3. It has been possible to isolate the insulin antibodies by elution from this aggregate with approximately a 500-fold purification. 4. The effect of pH, temperature, and salt concentration on the dissociation of rabbit antibody from insulin permits the postulate that there are a multiplicity of antibodies to the insulin molecule which have a multiplicity of binding characteristics. ACKNOWLEDGMENTS We wish to thank Mr. James Alexander for his technical assistance. Dr. Arquilla wishes to acknowledge the kindness and consideration he received from Dr. Normand L. Hoerr during the 7 years he was associated with the Department of Anatomy at Western Reserve University. It is his feeling that without Dr. Hoerr’s direction it would not have been possible for him to complete his education and continue his academic life . LITERATURE CITED Arquilla, E. R., and A. B. Stavitsky 1956 The production and identification of antibodies to insulin and their use in assaying insulin. J. Clin. Invest., 35: 458-466. Berson, S. A., and R. S. Yalow 1959 Quantitative aspects of the reaction between insulin and insulin-binding antibody. Ibid., 38: 1 9 9 6 2016. Bukantz, S. C., C. R. Rein and J. R. Kent 1946 Studies in complement fixation. 11. Preservation of sheep’s blood in citrate dextrose mixtures (modified Alsever’s solution) for use in the complement fixation reaction. J. Lab. Clin. Med., 31: 394. Campbell, D. H., E. Luescher and L. S. Lerman 1951 Immunologic adsorbents. I. Isolation of antibody by means of a cellulose-protein antigen. Proc. Nat. Acad. Sci. U.S., 37: 575578. ISOLATION OF INSULIN ANTIBODIES i. Isliker, H. C. 1953 Purification of antibodies by means antigens linked to ion exchange resins. A n N. Y. Acad. Sci., 57: 225-238. Kabat, E. A., and M. M. Mayer 1948 Experimental Im unochemistry. C. C Thomas, Springfield, Illinois, p. 107. . 209 Landsteiner, K. 1945 The Specificity of Serological Reactions, rev. ed. Harvard University Press, Cambridge, Massachusetts. Lowry, 0. H., M. J. Rosebrough, A. L. F a n and R. J. Randall 1951 Protein measurement with folin phenol reagent. J. Biol. Chem., 193: 265-275.