- Email: [email protected]
The synthesis and use of some insoluble immunologically specific adsorbents
Immunochemistry. Pergamon Press 1964. Vol. 1, pp. 219-229. Printed in Great Britain
THE SYNTHESIS AND IMMUNOLOGICALLY
USE OF SOME INSOLUBLE SPECIFIC ADSORBENTS*
N. WELIKY,H. H. WEETALL,R. V. GILDEN~and D. H. CAMPBELL Jet Propulsion Laboratory, Department of Biology, and Department of Chemistry and Chemical Engineering~, California Institute of Technology, Pasadena, California
(Received 24 April 1964)
ALTHOUGH many chemical studies of antibodies and antigen-antibody reactions require the use of purified reagents, little attention has been given to the development of methods for the removal and purification of antibodies from antiserums by specific chromatographic methods. One of the first reports was made by Campbell, Luescher and Lerman~ 1) who prepared an adsorbent by covalent coupling of bovine serum albumin (BSA) to cellulose. Although this coupling was tedious and rather complicated, a product was obtained which efficiently removed anti-BSA from rabbit antiserum and yielded a final product, after elution at pH 3.0 with HC1, that showed a high degree of purity with respect to specific precipitation with BSA at pH 8.0. The basic idea was good: it postulated that a specific reacting group anchored covalently to a bland insoluble material such as cellulose could react with its complementary material and remove it from solution. Later work by Lerman indicated the possibility of fractionating antibody with respect to reaction capacity (avidity) for a single haptenic group~Z) and removal of enzymes by placing a substrate or an inhibitor on the cellulose. Within the last few years more interest has been given to the use and properties of biological materials coupled to insoluble carriers. These include not only antigens but also enzymes. The following study was undertaken to develop newer and simpler methods for coupling specific groups to insoluble carriers for use as specific adsorbents. With the refinements currently available for determining protein structure, it has become increasingly important to develop practical methods for preparing proteins of the highest possible purity. Because of the similarities in their physical properties, it has not yet been feasible to completely separate specific antibodies from other globulins by physical techniques. The technique most commonly used is precipitation of the antibody with the homologous antigen followed by dissociation of the complex and separation of its components. Of the various methods available, the one that potentially ensures complete separation of antibodies from both serum proteins and antigen, entails coupling the antigen to an insoluble polymer before complexing with antibody. Polymer columns composed of polystyrene or cellulosel~) have been criticized * This paper, in part, presents results of one phase of research carried out at the Jet Propulsion Laboratory, California Institute of Technology, under Contract No. NAS7-100, sponsored by the National Aeronautics and Space Administration. T Present address, Wistar Institute, Philadelphia, Pennsylvania. ~ Contribution No. 3004. 219
220
N. WELIKY,H. H. WEETALL,R. V. GILDENand D. H. CAMPBELL
recently(4) because of their low capacity, high nonspecific adsorption and failure to achieve desorption under conditions that do not destroy antibody activity. The method of Campbell, Luescher and Lermanll) as modified by Gurvich~5,6~ was reported to have overcome the first two difficulties to a considerable extent, but was described as unsatisfactory with respect to the third. A further criticism has been leveled at the use of acids for dissociating antigen-antibody complexes. It has been found that dissociation of antigen-antibody precipitates is incomplete in 1.0 M acetic acid(~L A critical review of antibody isolation techniques has recently appeared (8). We have recently investigated the use of p-aminobenzylcellulose-antigen columns(9) for isolating antibody, and have developed two new procedures for synthesizing antigen columns. The behavior of serum and antibody were studied with respect to the amount of nonspecific protein retained by these columns; the degree to which complexed antibody can be dissociated and recovered; the purity of the antibody recovered; the fraction of antibody removed from serum per pass through the column; the number of passes required to exhaust serum of antibody; the effect of phosphate buffer on the behavior of antibody in this system; and the effect of lowering antibody concentration on the purity of the isolated antibody. MATERIALS AND METHODS Inorganic chemicals and organic solvents were reagent grade or equivalent. The following substances were used without further purification: crystalline bovine serum albumin (BSA), N,N'-dicyclohexylcarbodiimide (DCC), Folin-Ciocalteu phenol reagent (from California Corporation for Biochemical Research); human y-globulin (HGG), a-DNP-lysine (from Nutritional Biochemicals Corp.); p-aminobenzylcellulose (Cellex PAB) (0.4 m-equiv./g) and carboxymethylcellulose (CMC) (0.7 m-equiv./g) (from Bio-Rad Laboratories); fibroin (from General Biochemicals Inc.); and p-(p'-aminophenylazo)-phenylarsonic acid (R') (from K and K Laboratories). Polyazophenylarsonic-fibroin. To 48 mg of NaNO2 in 2.6 ml of water was added 0.8 ml of 4N HC1. The solution was added at once to 109 mg of arsanilic acid (R) in 1 ml of H20, previously brought to pH 8-9 with 1 M Na2COs. All solutions were kept between 0 and 6 ° C. After 15 min the pH was adjusted to 1-2. Excess nitrite was removed with sulfamic acid. The diazotized arsanilic acid was added slowly to 1 g of fibroin in 50 ml of HzO, keeping the pH between 7 and 9 with 0.1 N NaOH. The mixture was stirred overnight at 5 ° C. The polyazophenylarsonic-fibroin (R-fibroin) was pressed in a Buchner funnel to remove the solution and then exhaustively washed with water, dilute Na~COs, and 0.1 N HCI before use. Carboxymethylcellulose coupled to proteins.* Carboxymethylcellulose was added, with stirring, to an aqueous solution of 100 mg of protein, until the mixture thickened. A solution of 0.2 g of DCC in 0.5 ml tetrahydrofuran (THF) and 1 ml * After these experiments were completed, our supplier made changes in processing the carboxymethylcellulose. The new material only coupled satisfactorily if it was first converted completely to the acid form by stirring with 4 N HC1 and washing with water. In most cases the wet CMC was then washed with acetone, dried on a Buchner funnel and samples weighed out.
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
221
of water (2 phases) was added and then further CMC to a total of 0.60 g. After standing for 2 days at room temperature, 10 ml of water were added. The solid was dispersed, filtered and washed consecutively with large quantities of dilute Na2CO3, 0.01 M HC1 and water. After repeated decantations, columns were poured. Carboxymethylcellulose coupled to haptens.* A mixture of 0.1 g of p-(p'-aminophenylazo)-phenylarsonic acid was stirred with 2 ml H~O. CMC was added to the ~ cloudy solution, with stirring, until it thickened. Then 0.2 g DCC in 0.5 ml T H F and 0.5 ml H~O (2 phases) was added. Further addition of CMC brought the total to 0.5 g. The mixture was permitted to stand at room temperature for 2 days and treated as described for protein coupled material. CMC was added t o a mixture of 25-50 mg ~-DNP-lysine and 2 ml water until it thickened. A solution of 0-2 g DCC in 0.5 ml T H F was made, and added with 0.5 ml water. Further CMC was added to make a total of 0.5 g. After standing 2 days at room temperature, the mixture was washed exhaustively with acetone, dilute Na2CO3, and dilute HC1 until no color could be detected in the washes. Columns were poured after repeated decantations. Determination of the quantity of antigen coupled. BSA labeled with I TM was used to determine the quantity of BSA bound to CMC. The activity of the CMC-BSA was measured relative to the total activity of the BSA used in the reaction. Activity was determined with a well type scintillation counter (Nuclear Chicago). Arsenic coupled to CMC and fibroin was determined by Truesdail Laboratories, Los Angeles, Calif. Preparation of antigens and antisera. Arsanilic acid was coupled to BSA~t°) and keyhole limpet hemocyanin (KLH)(lt). Antisera to these antigens, to BSA, and to H G G were prepared by injection of New Zealand White rabbits. Sera were collected, pooled and frozen until used. Isolation of specific antibody from serum. Glass chromatographic tubes (10-12 mm O.D.) with Teflon stopcocks were indented to hold a glass wool plug. The tubes were connected to the 1 mm flow cell of a continuous flow spectrophotometer (Vanguard Instrument Co.) and the absorption at 220 m/~ recorded. Air pressure up to 15 lb/in2 was applied to the tops of the columns to speed up the flow, if necessary. Before using the columns, they were washed at pH 2.3 and pH 7-0 until no absorption was noted at 220 m~. Serum previously heated at 56 ° C for 30 min was passed through an appropriate column. The column was washed with 1 per cent NaC1 solution or 0.1 M phosphate at pH 7.0 until an optical density of less than 0.01 at 220 m~ was recorded. The antibody was eluted with 1 per cent NaC1 solution acidified with HC1 to pH 2.3, or a phosphate buffer (0.1-1.0 M) at pH 2.3. The pH of the eluate was adjusted to neutrality with 0.2 M NaOH. Determination of precipitable antibody. Mixtures of antigen and antibody in various proportions were maintained at 37 ° C for 2 hr, then at 5 ° C for 1-3 days. Precipitates washed twice with cold 0.01 M borate buffered saline at pH 8-0, were dissolved in one drop of 0.5 M NaOH and total protein determined. Antibody concentrations were calculated from precipitates formed in slight antigen excess by subtracting out the known amount of antigen added. * See f o o t n o t e on p. 220.
222
N.
WELIKY,
H. H.
WEETALL, R.
V. GILDENand D. H. CAMPBELL
In the case of anti-BSA determinations, the use of Ilal-labeled BSA permitted the subtraction of the exact quantity of BSA in the precipitate. Purity was measured by the ratio of precipitable antibody to total protein determined by the Lowry method. Determination of anti-BSA antibody by antigen binding capacity (ABC). This method~12~ was used for the rapid estimation of relative amounts of anti-BSA in various samples. T h e technique depends upon the solubility of I~I-BSA in 1/2 saturated ammonium sulfate solution, in which complexes of I~31-BSA and antibody are insoluble. Antibody samples of various dilutions were mixed with an antigen solution and precipitated. T h e antibody dilution was determined at which 33 per cent of the added antigen was found in the precipitate. T h e ABC of the sample was calculated as follows: (0.33) (~g/ml antigen added) ABC = fractional dilution at the 33 per cent reference point
Determination of protein concentration.
Protein was determined using the
procedure of Lowry~TM.
Immunoelectrophoresis~4~. Electrophoresis was carried out for 1 hr at 250 V on microscope slides in a gel made from a 1 per cent solution of agar in veronal buffer, p H 8.6, ionic strength 0.1. T h e antiserum was added to the trough and diffusion permitted to continue for 24 hr~ T h e slides were then fixed and stained with Amido Black 10B. TABLE 1. NONSPECIFIC PROTEIN ELUTED FROM CMC AND FIBROIN ANTIGEN COLUMNS*
Column CMC-BSA CMC-BSA CMC-BSA~" CMC-BSA CMC-HGG Fibroin-R Fibroin-R Fibroin-R CMC-R'
Test serum
Serum added (ml)
Protein eluted at pH 2"3 (mg/ml serum)
3"0 2"0 5'0 20"0 5"0 5"0 2'0 20"0 5.0
0"010 0'008 0"011 0'008 0"008 0.008 0"006 0'010 0.010
NRS Anti-HGG Anti-HGG NRS Anti-BSA NRS Anti-HGG NRS Anti-HGG
All columns consisted of 0.5-0.6 g CMC- or fibroin-antigen. The symbols used are NRS (normal rabbit serum), anti-BSA (rabbit antibovine serum albumin), and anti-HGG (rabbit anti-human ~,-globulin). * Nonspecific protein retained by these columns at pH 7'0 was removed, in general, by 0"1 M phosphate buffer at pH 2"3. ? Elution was with NaC1 solution acidified to pH 2.3 with HC1. EXPERIMENTAL
RESULTS
Antigens coupled to carboxymethylcellulose or fibroin were found to be capable of removing antibody from antisera prepared against the antigen. Evidence for covalent coupling of antigen to C M C rather than adsorption or entrapment was
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
223
demonstrated by omitting the DCC from the reaction of CMC with ~-DNP-lysine and with R'. All but tr~ces of these highly colored compounds were removed by washing. If DCC was present, the deep color was retained. Coupling of BSA, labeled with 1131, to CMC resulted in 93 per cent transfer of activity from the solution to the CMC. Arsenic found in the fibroin coupled to diazotized arsanilic acid amounted to 43 mg/g of CMC. Evidence concerning the effectiveness of antigens coupled to CMC and fibroin for the purification of antibody and for studying the behavior and reactions of antibody is given below. It must be emphasized that reliable results can only be obtained if the CMC or fibroin coupled to antigen is free of uncoupled antigen. These materials must be washed with water and acids or bases capable of extracting traces of the antigen, particularly haptens, each time they are used. After coupling, the CMC-antigen particles should be stirred and well dispersed in large quantities of water or buffer to ensure hydrolysis of the DCC. In the case of haptens, organic solvents may be used to extract excess hapten and unhydrolyzed DCC. TABLE 2. NONSPECIFIC PROTEIN ELUTED ON p-AMINOBENZ~YLCELLULOSE
COLUMNS
Bound antigen
Test serum
Serum added (ml)
Protein eluted at pH 2'3 (mg/ml serum)
HGG BSA BSA BSA R* R* R* R* R*
Anti-BSA NRS NRS Anti-HGG NRS NRS Anti-HGG Anti-HGG Anti-HGG
2.0 2-0 2'0 3"0 2"0 3"0 3"0 1"0 20"0
0-115 0"213 0'117 0-082 0'093 0"107 0.090 0-180 0"100
In all cases serum was washed from columns with 0'1 M phosphate at pH 7 and eluted with a phosphate buffer between 0-1 and 1"0 M. See Table 1 for test serum symbols. * A 1,5-dihydroxynaphthylazophenylarsonic acid derivative of diazotized p-aminobenzylcellulose.
Retention of nonspecific protein. Normal rabbit sera and immune rabbit sera prepared against other antigens were passed through columns composed of 0.5-0.6 g of CMC-antigen and R-fibroin. In all cases 0.011 mg or less protein per ml of serum was released at pH 2.3 (Table 1). Retention of nonspeeifie protein by columns composed of diazotized p-aminobenzylcellulose coupled to protein was found to be higher by a factor of 10 (Table 2). Acid dissociation of antibody from antigen columns. The results of experiments designed to determine the effectiveness of hydrochloric acid in 1 per cent NaC1, at pH 2-3, in dissociating these antigen-antibody complexes, and the effect of this acid treatment on the integrity of the antibody are presented in Table 3. The purity of the antibody dissociated in this manner (measured by the fraction of precipitable antibody) ranged from 90-100 per cent. Between 88 and 97 per cent of the total antibody of the original serum could be accounted for. By immunoelectrophoresis,
224
N.
WELIKY,
H. H.
WEETALL,
R. V. GILDENand D. H. CAMPBELL
the antibody was shown to have no detectable components other than 7-globulin (Fig. 1). Acid was found not to dissociate a n t i - D N P antibody from C M C - l y s i n e - D N P columns. Using a 0.6 g column, all 9-4 mg of anti-DNP antibody contained in 5 ml of rabbit serum were retained. Elution with 1 per cent NaC1 acidified to p H 2.3 with HC1 removed 1.0 mg. T h e remainder could be recovered without activity using 0.1 M NaOH. T o verify that DNP-antibody complexes from this serum did not dissociate at p H 2.3, the following experiments were performed. An antigenantibody precipitate obtained by reaction of a n t i - D N P serum with dinitrophenylated-BSA did not dissolve in 1 per cent NaC1 acidified to p H 2.3 with HC1. No protein could be detected in the supernatant liquid after centrifugation. Partial solution was realized in HC1 at p H 2.3 in the absence of added NaC1. T h e presence TABLE 3a. PURITY OF ISOLATED ANTIBODY FROM 0 " 5 - - 0 " 6 g COLUMNS
Column material
Serum added (ml)
Serum antibody (mg/ml)
Total Serum antibody (mg)
Antibody in serum effluent (mg)
5"0 5'0 5-0 5.0
3.52 2.25 3'08 3.08
17.60 11.25 15"40 15.40
12-00 7-42 8"30 --
CMC-BSA CMC-HGG Fibroin-R CMC-R'
Protein Percentage eluted at precipitable pH 2"3 antibody (mg) in eluate 3'65 3-52 5"52 9.0
90"2 97'4 101 "7 101.0
TABLE 3b. RECOVERY OF ANTIBODY CALCULATED FROM DATA OF TABLE 3a
Column material CMC-BSA CMCoHGG Fibroin-R CMC-R'
Antibody retained by column (mg)
Percentage antibody recovered from column
Percentage overall yield
Percentage* total antibody accounted for
4'60 3'83 7.10 --
79 92 78 --
21 31 36 60
89 97 90 --
* Percentage determined by the sum of the antibody in the effluent and eluate, divided by the total serum antibody. of D N P - B S A in the supernatant liquid was confirmed by the absorbance maximum at 360 m~ and its directly observable color. Because B S A - D N P is insoluble at p H 2.3 in the presence or absence of added NaC1, any solubility was due to the formation of soluble complexes with the antibody. A 1 per cent NaC1 solution containing B S A - D N P and either human or rabbit 7-globulin was found to precipitate the B S A - D N P at p H 2.3, leaving 7-globulin in solution. T o verify that the antigen columns extract antibody from low titer sera and that the antibody can be recovered in high purity, rabbit a n t i - H G G serum containing 2.9 mg of antibody per ml and rabbit anti-phenylarsonic serum containing 2.4 mg of antibody per ml were diluted to 0.29 mg/ml and 0.24 mg/ml, respectively, with normal rabbit serum. T h i r t y ml of each were passed through appropriate antigen-
FIG. 1. Immunoelectrophoretic comparison of normal rabbit to antibody isolated with acidified NaCl (bottom well) against serum. The antibody migrates in the r-globulin region.
(Facing p. 224)
serum (top well), sheep anti-rabbit
FIG. 2. Immunoelectrophoretic comparison of antibody isolated with acidified NaCl (top well) to antibody isolated with 0.1 M phosphate buffer at pH 2.3 (bottom well) against sheep anti-rabbit serum.
S y n t h e s i s and U s e of some I n s o l u b l e I m m u n o l o g i c a t l y Specific A d s o r b e n t s
225
C M C columns. T h e antibodies were eluted with 1 per cent sodium chloride solution acidified to p H 2.3 with HC1. T h e recovered antibody was 90 per cent precipitable (Table 4). T h r e e experiments were performed to determine whether proportions could be increased for larger scale operations: (a) T w o batches of C M C - B S A were prepared using 25 mg and 200 mg of BSA per 0"6 g CMC, instead of the 100 mg of BSA previously used. No antibody was eluted at p H 2.3 from B S A - C M C prepared with 25 mg BSA. T h e other column adsorbed 9.10 mg from 22.4 mg of antibody in TABLE4.
PURITY OF ANTIBODY ISOLATED FROM LOW TITER SERA*
Column antigen
Diluted serum (mg/ml)
Diluted serum added to column (ml)
Percentage precipitable antibody
0-290 0-240 0"240
30 30 30
90'0 94-7 86"3
CMC-HGG CMC-R' CMC-R'
* High titer serum was diluted 1 : 10 with normal rabbit serum for these experiments. Antibody was eluted with 1 per cent NaC1 acidified with HCI to pH 2.3. 5 ml of sera. T h e recovery from the column was 81 per cent. T h e overall yield on one pass was 33 per cent. T h e purity was 87 per cent. (b) T h e effect of increasing column length was determined for C M C - B S A and phenylarsonic-CMC by using 2 g instead of 0.5 g columns. T h e results shown in Table 5 indicate that better overall yields may be obtained by increased column size. TABLE 5. PURITY AND RECOVERY OF ANTIBODY FROM
Column material
CMC-BSA CMC-R' CMCoR'
Serum added (ml)
Total serum antibody (mg)
Percentage overall yield
Percentage antibody recovered from column
5"0 5.0 5"0
17'60 9.60 12.70
47-4 71-3 94'3
83"5 71 "3 94"3
2"0
g COLUMNS
Percentage* Percentage total precipitable antibody antibody accounted in for eluate 90'6 71 '3 94"3
91'7 96-6 101 '2
* See footnote to Table 3b.
Effect of phosphate on acid dissociation of antibody. T h e above results on acid dissociation were obtained using 1 per cent NaC1 solution acidified to p H 2.3 with HC1. If the antibody was eluted with a buffer, between 0.1 and 1.0 M in phosphate at p H 2-3, the fraction of precipitable antibody decreased by 30-90 per cent. Quantitative precipitin determinations were made after dialysis against 0.05 M phosphate buffer or 0.9 per cent NaC1. T h e phosphate eluted protein was shown to be immunoelectrophoretically identical to antibody which was 90-100 per cent precipitable (Fig. 2).
226
N. WELIKY, H. H. WEETALL, Ro Vo GILDENand D. H. CAMPBELL
In Table 3 it was shown that after a sequence of operations, including complexing of the antibody to an antigen-column followed by elution, all of the antibody could be accounted for. The antibody isolated was 90-100 per cent precipitable. Thus, eluting with phosphate buffer should not remove more nonspecific protein. To test this, 5 ml of rabbit anti-BSA serum were passed through a 0.5-0.6 g CMC-BSA column. The column was washed with 1 per cent NaC1 and eluted with 1 per cent NaC1 solution acidified to pH 2.3 with HC1. The eluate contained 4.05 mg of antibody. Further elution with 0.1 M phosphate buffer at pH 2.3 followed by 1'0 M phosphate buffer at pH 2.3 brought down less than 1 per cent additional protein. Thus the decrease of the fraction of precipitable antibody in the protein dissociated with phosphate buffer results from denaturation. Further, if nonspecific serum is passed through a column and washed with phosphate buffer at pH 2.3 no more protein is released than with 1 per cent NaC1 solution acidified by HCI (Table 1). Phosphate buffer at pH 2.3 thus releases antibody which is denatured to various degrees. Serum depletion experiments, to be described next, demonstrate that washing the columns with 0.1 M phosphate buffer at pH 7.0, and eluting with TABLE 6. D E P L E T I O N OF ANTIBODY FROM SERUM USING 0"5-0'6 g CMC-BSA COLUMNS
Serum (ml) 3'0 3.0 2.0
Total serum Percentage depletion of serum antibody* Total protein antibody recovered (mg) Pass 1 Pass 2 Pass 3 (mg) 9.81 6.54 4281"
78.6 94.9 39.7++
102.7 101.4 84.1 ,+
103.6 102.7 93'4~
10.2 6.72 --
* All values uncorrected for nonspecific protein. @ABC value of 2'0 ml of serum. ++Percentages derived from ABC values of serum effluent. phosphate buffer at pH 2.3 does not affect the amount of antibody recovered. Therefore the quantity of protein released at pH 2.3 is equivalent to the quantity of antibody released. The ABC of the serum effluent provides a direct measure of the quantity of antibody remaining in the serum. Depletion of antibody from serum. The fraction of antibody removed from serum, per pass, by an antigen column was variable. No systematic effort was made to find the cause. Nevertheless, antibody could be exhausted from the serum by several passes through an appropriate antigen column. Phosphate buffers up to 1.0 M were used for elution because they have the advantage, over acidified NaC1 solution at pH 2.3, that pH changes in the column are rapid and therefore antibody is recovered in higher concentration. Phosphate does not absorb in the ultraviolet region so that the column effluent can still be monitored spectrophotometrically and the portion containing antibody collected. The total quantity of protein released at pH 2.3 was equal to the total antibody in the serum used, if allowance was made for 10 t~g of nonspecific protein per ml of serum. In those cases where antibody in the serum effluent was measured by the ABC test, complete depletion of antibody activity was also found (Table 6).
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
227
DISCUSSION We have demonstrated that precipitable antibody can be isolated in high purity from CMC-antigen columns and fibroin-phenylarsonic columns. Contrary to previous reports, ~4,~ there did not appear to be any serious problem with retention of 'tightly bound' antibody, under the elution conditions used, except for anti-DNP antibody. The antibody complexed during the first pass should be the most tightly bound. We have shown, however, that recovery is very high. Furthermore, the serum depletion experiments demonstrate that even 'weakly bound' antibody, if present, can be complexed to the column, and recovered. In all of these experiments we made it a point to leave the antibody complexed on the column for a minimum period of time. Most experiments were completed in 4 hr or less, and none permitted to exceed 12 hr. Campbell, Luescher and Lermantl~ have previously reported that the longer antibody is permitted to remain complexed with antigen on a cellulose column, the lower the recovery of the antibody. We have also found that the presence of phosphate at pH 2.3 can cause partial loss of antibody activity, At this pH most of the phosphate is unionized or singly charged. Thus the nature of the ionic and molecular species present at this pH may have profound effects on the behavior of antibody molecules. Antibodies bound to these antigen columns can be washed with large quantities of saline at pH 5-7 without appreciable losses. In spite of this, only a portion of the serum antibody may be extracted per pass, even though the column is not loaded to capacity. This suggests that orientation and diffusion factors are important in reactions in which complexes are formed on antigen columns. Orientation and steric factors pertain both to coupled antigen and soluble antibody. Antigenic sites on macromolecules will be unavailable for binding if they are turned toward the inert carrier or if binding to one antibody molecule sterically blocks binding by others. Blocking of polymer surface interstices into which antigen molecules have wandered, by other antigen molecules or complexed antibody molecules, may further reduce the fraction of theoretically available antigenic sites. Hydrogen bonding of polar antigen groups to polar groups on the carrier may also reduce available sites. Nonrandom coupling may increase or decrease the estimated number of sites. The above considerations, together with channeling of antibody solutions between carrier particles (in the case of columns), make plausible the lack of stoichiometry between antigen and antibody if carrier coupled antigens are used. Increasing the time of contact between serum and antigen is limited by the development of irreversible binding.~l~ Further improvements in yield can be achieved by increasing surface area, increasing the ratio of coupled antigen to antibody (e.g. longer column beds) and varying coupling methods so as to improve antigen orientation and site availability. Anti-DNP antibody behaved differently from the other three antibodies with respect to isolation from its specific antigen column. Evidence obtained indicates that rabbit anti-DNP antibody does not dissociate appreciably from its antigen in 1 per cent saline solution adjusted to pH 2.3 with hydrochloric acid. The specific precipitate formed with highly substituted BSA-DNP did not dissolve. The binding at pH 2-3 was specific because neither RGG nor H G G coprecipitated with BSADNP at that pH. If no added salt was present, 10-15 per cent of the specific precipitate dissolved forming a colored solution. Because the DNP-BSA alone was
228
N. WELIKY,H. H. WEETALL,R. V. GILDENand D. H. CAMPBELL
insoluble at that pH the precipitate must have dissolved in the form of soluble complexes. Our results agree with those of Bennett and Haber(7~ in that complete dissociation in acid solution may not be assumed a priori. We did find, however, that anti-BSA, anti-HGG and anti-phenylarsonic antibodies are largely dissociated from their antigens at pH 2.3. We have developed a simple, rapid method for coupling antigens to carboxymethylcellulose, that permits not only proteins but many haptens to be bound. Nonspecific adsorption is about 0.01 mg per ml of serum, using columns composed of 0.5-0.6 g of CMC-antigen. Recovery of the complexed antibody, and precipitability of the recovered antibody is high. Coupling conditions are mild. No blocking groups are necessary to inactivate uncoupled sites on the CMC, and the CMC remains clean and white or assumes the color of the antigen. Thus, deterioration of contamination may often be clearly observed by color changes. In addition, coupling to protein antigens occurs at sites different from those which couple to diazonium compounds. We have also synthesized a hapten-coupled fibroin column with similar properties, except for color. The advantages of columns for isolating highly purified antibodies have not been fully exploited. They present a simple one-step technique as opposed to the many steps required by other procedures. Further, these columns can be re-used many times, an important consideration in cases where antigens are difficult to obtain or expensive. In general, substances with basic amino groups will couple to insoluble carboxymethylcellulose in the presence of N,N'-dicyclohexylcarbodiimide. These include bovine serum albumin, human 9,-globulin, keyhole limpet hemocyanin, ¢-DNPlysine and p-(p'-aminophenylazo)-phenylarsonic acid. The coupling is presumably through an amide bond formed between an amino group of the antigen and a carboxyl group of the CMC: cellulose
cellulose
I
O
O
I CH2 ] ÷RNH~ C=O
I
OH
DCC
CH2 -+ [ +H20 C-----O
I
HN--R
The possibility that antigen was adsorbed or entrapped rather than chemically bound to the carboxymethylcellulose was eliminated by the clear absence of coupling, if N,N'odicyclohexylcarbodiimide was omitted from the reaction mixture. Because of the mildness of the reaction conditions we have been exploring the coupling of CMC to enzymes and other biologically active molecules. These active substances coupled to carboxymethylcellulose are now being thoroughly investigated. Full advantage has not yet been taken of the use of insoluble carriers, coupled to such organic and biologically active molecules, as research tools for the purification of complementary substances and for studies of the characteristics of the bound molecules.
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
229
SUMMARY (1) We have utilized the reaction of acids and amines to form amides in the presence of N,N'-dicyclohexylcarbodiimide as a means of coupling proteins and other organic molecules with basic amino groups to carboxymethylcellulose. (2) T h e s e carboxymethylcellulose derivatives can be used for the isolation of antibody from antisera directed against either proteins or haptens. (3) T h e antibody isolated is of high purity. T h e procedure does not result in appreciable losses of antibody remaining in the serum, and reasonable yields may be expected. (4) T h e carboxymethylcellulose derivatives exhibit very low retention of nonspecific protein near p H 7 and so may be used to isolate antibody from low titer sera. (5) T h e carboxymethylcellulose derivatives are colorless or assume the color of the coupled substance. T h e y need no blocking groups after coupling, and coupling conditions are mild. Flow rates are reasonable, but in general pressure is required. (6) T h e technique offers a simple method for coupling an insoluble carrier to proteins at a site other than that resulting from coupling diazonium compounds to proteins. (7) Dissociation of rabbit anti-BSA, a n t i - H G G and anti-phenylarsonic antibodies at p H 2.3 was found to result in high recoveries of antibody bound at neutrality to specific antigens previously coupled to CMC. (8) A n t i - D N P antibody from a rabbit serum did not dissociate appreciably from its antigen at p H 2.3 whether on an antigen column or in the form of a specific precipitate. (9) Coupling of proteins and organic compounds with basic amino groups to C M C is applicable to enzymes and other biologically active molecules, in many cases without loss of activity. Acknowledgements--We thank Drs J. S. Garvey, A. Malley and A. Amkraut of the California
Institute of Technology for helpful discussions and for sera. We also thank Dr M. Litt of Allied Chemical Corp. for his contributions to a discussion of organic coupling methods. REFERENCES
1 CAMPBELLD. H., LUESCHERF. and LERMANL. S., Proc. Nat. Acad. Sci., Wash. 37, 575 (1951). ~ L ~ A N L. S., Nature 172, 635 (1953). a PUTNAMF. W. (Editor), The Plasma Proteins, Vol. I., p. 130. Academic Press, New York (1960). 4 MOt~DOALN. R. and POaTEa R. R., Biochem. Biophys. Acta 71, 185 (1963). 5 GVRVICHA. E., Kav~mR R. B. and NEZraN R. S., Biokhimiya 24, 129 (1959). 6 G ~ w c a A. E., KOZOVLEVAO. B. and TUM~a~OVAA. E., Biokhimiya 26, 803 (1962). 7 BENrq~TT J. C. and HABEa E., ft. Biol. Chem., 238, 1362 (1963). s SEHONA. H., Brit. Med. Bull. 19, 183 (1963). 9 MAL~EYA. and CAMPBelL D. H., ft. Amer. Chem. Soc. 85, 487 (1963). 10 LANDSTEINERK. and VAN DER SCHEERJ., ~¢. Exp. Med. 55, 781 (1932). ~t CAMPBELLD. H., GARWVJ. S., CRAMERN. E. and SUSSDOR~D. H., Methods in Immunology, p. 69. W. A. Benjamin Inc., New York (1963). ~ FArm R. S., ft. lnfect. Dis. 103, 239 (1958). ta LOWRy O. H., ROS~nROU~H N. J., FARR L. and RANDALLR. J., ~. Biol. Chem. 193, 265 (1951). ~a GRaBAa P. and W~LL~aMSC. A., JR., Biochim. Biophys. Acta 10, 193 (1953).
THE SYNTHESIS AND IMMUNOLOGICALLY
USE OF SOME INSOLUBLE SPECIFIC ADSORBENTS*
N. WELIKY,H. H. WEETALL,R. V. GILDEN~and D. H. CAMPBELL Jet Propulsion Laboratory, Department of Biology, and Department of Chemistry and Chemical Engineering~, California Institute of Technology, Pasadena, California
(Received 24 April 1964)
ALTHOUGH many chemical studies of antibodies and antigen-antibody reactions require the use of purified reagents, little attention has been given to the development of methods for the removal and purification of antibodies from antiserums by specific chromatographic methods. One of the first reports was made by Campbell, Luescher and Lerman~ 1) who prepared an adsorbent by covalent coupling of bovine serum albumin (BSA) to cellulose. Although this coupling was tedious and rather complicated, a product was obtained which efficiently removed anti-BSA from rabbit antiserum and yielded a final product, after elution at pH 3.0 with HC1, that showed a high degree of purity with respect to specific precipitation with BSA at pH 8.0. The basic idea was good: it postulated that a specific reacting group anchored covalently to a bland insoluble material such as cellulose could react with its complementary material and remove it from solution. Later work by Lerman indicated the possibility of fractionating antibody with respect to reaction capacity (avidity) for a single haptenic group~Z) and removal of enzymes by placing a substrate or an inhibitor on the cellulose. Within the last few years more interest has been given to the use and properties of biological materials coupled to insoluble carriers. These include not only antigens but also enzymes. The following study was undertaken to develop newer and simpler methods for coupling specific groups to insoluble carriers for use as specific adsorbents. With the refinements currently available for determining protein structure, it has become increasingly important to develop practical methods for preparing proteins of the highest possible purity. Because of the similarities in their physical properties, it has not yet been feasible to completely separate specific antibodies from other globulins by physical techniques. The technique most commonly used is precipitation of the antibody with the homologous antigen followed by dissociation of the complex and separation of its components. Of the various methods available, the one that potentially ensures complete separation of antibodies from both serum proteins and antigen, entails coupling the antigen to an insoluble polymer before complexing with antibody. Polymer columns composed of polystyrene or cellulosel~) have been criticized * This paper, in part, presents results of one phase of research carried out at the Jet Propulsion Laboratory, California Institute of Technology, under Contract No. NAS7-100, sponsored by the National Aeronautics and Space Administration. T Present address, Wistar Institute, Philadelphia, Pennsylvania. ~ Contribution No. 3004. 219
220
N. WELIKY,H. H. WEETALL,R. V. GILDENand D. H. CAMPBELL
recently(4) because of their low capacity, high nonspecific adsorption and failure to achieve desorption under conditions that do not destroy antibody activity. The method of Campbell, Luescher and Lermanll) as modified by Gurvich~5,6~ was reported to have overcome the first two difficulties to a considerable extent, but was described as unsatisfactory with respect to the third. A further criticism has been leveled at the use of acids for dissociating antigen-antibody complexes. It has been found that dissociation of antigen-antibody precipitates is incomplete in 1.0 M acetic acid(~L A critical review of antibody isolation techniques has recently appeared (8). We have recently investigated the use of p-aminobenzylcellulose-antigen columns(9) for isolating antibody, and have developed two new procedures for synthesizing antigen columns. The behavior of serum and antibody were studied with respect to the amount of nonspecific protein retained by these columns; the degree to which complexed antibody can be dissociated and recovered; the purity of the antibody recovered; the fraction of antibody removed from serum per pass through the column; the number of passes required to exhaust serum of antibody; the effect of phosphate buffer on the behavior of antibody in this system; and the effect of lowering antibody concentration on the purity of the isolated antibody. MATERIALS AND METHODS Inorganic chemicals and organic solvents were reagent grade or equivalent. The following substances were used without further purification: crystalline bovine serum albumin (BSA), N,N'-dicyclohexylcarbodiimide (DCC), Folin-Ciocalteu phenol reagent (from California Corporation for Biochemical Research); human y-globulin (HGG), a-DNP-lysine (from Nutritional Biochemicals Corp.); p-aminobenzylcellulose (Cellex PAB) (0.4 m-equiv./g) and carboxymethylcellulose (CMC) (0.7 m-equiv./g) (from Bio-Rad Laboratories); fibroin (from General Biochemicals Inc.); and p-(p'-aminophenylazo)-phenylarsonic acid (R') (from K and K Laboratories). Polyazophenylarsonic-fibroin. To 48 mg of NaNO2 in 2.6 ml of water was added 0.8 ml of 4N HC1. The solution was added at once to 109 mg of arsanilic acid (R) in 1 ml of H20, previously brought to pH 8-9 with 1 M Na2COs. All solutions were kept between 0 and 6 ° C. After 15 min the pH was adjusted to 1-2. Excess nitrite was removed with sulfamic acid. The diazotized arsanilic acid was added slowly to 1 g of fibroin in 50 ml of HzO, keeping the pH between 7 and 9 with 0.1 N NaOH. The mixture was stirred overnight at 5 ° C. The polyazophenylarsonic-fibroin (R-fibroin) was pressed in a Buchner funnel to remove the solution and then exhaustively washed with water, dilute Na~COs, and 0.1 N HCI before use. Carboxymethylcellulose coupled to proteins.* Carboxymethylcellulose was added, with stirring, to an aqueous solution of 100 mg of protein, until the mixture thickened. A solution of 0.2 g of DCC in 0.5 ml tetrahydrofuran (THF) and 1 ml * After these experiments were completed, our supplier made changes in processing the carboxymethylcellulose. The new material only coupled satisfactorily if it was first converted completely to the acid form by stirring with 4 N HC1 and washing with water. In most cases the wet CMC was then washed with acetone, dried on a Buchner funnel and samples weighed out.
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
221
of water (2 phases) was added and then further CMC to a total of 0.60 g. After standing for 2 days at room temperature, 10 ml of water were added. The solid was dispersed, filtered and washed consecutively with large quantities of dilute Na2CO3, 0.01 M HC1 and water. After repeated decantations, columns were poured. Carboxymethylcellulose coupled to haptens.* A mixture of 0.1 g of p-(p'-aminophenylazo)-phenylarsonic acid was stirred with 2 ml H~O. CMC was added to the ~ cloudy solution, with stirring, until it thickened. Then 0.2 g DCC in 0.5 ml T H F and 0.5 ml H~O (2 phases) was added. Further addition of CMC brought the total to 0.5 g. The mixture was permitted to stand at room temperature for 2 days and treated as described for protein coupled material. CMC was added t o a mixture of 25-50 mg ~-DNP-lysine and 2 ml water until it thickened. A solution of 0-2 g DCC in 0.5 ml T H F was made, and added with 0.5 ml water. Further CMC was added to make a total of 0.5 g. After standing 2 days at room temperature, the mixture was washed exhaustively with acetone, dilute Na2CO3, and dilute HC1 until no color could be detected in the washes. Columns were poured after repeated decantations. Determination of the quantity of antigen coupled. BSA labeled with I TM was used to determine the quantity of BSA bound to CMC. The activity of the CMC-BSA was measured relative to the total activity of the BSA used in the reaction. Activity was determined with a well type scintillation counter (Nuclear Chicago). Arsenic coupled to CMC and fibroin was determined by Truesdail Laboratories, Los Angeles, Calif. Preparation of antigens and antisera. Arsanilic acid was coupled to BSA~t°) and keyhole limpet hemocyanin (KLH)(lt). Antisera to these antigens, to BSA, and to H G G were prepared by injection of New Zealand White rabbits. Sera were collected, pooled and frozen until used. Isolation of specific antibody from serum. Glass chromatographic tubes (10-12 mm O.D.) with Teflon stopcocks were indented to hold a glass wool plug. The tubes were connected to the 1 mm flow cell of a continuous flow spectrophotometer (Vanguard Instrument Co.) and the absorption at 220 m/~ recorded. Air pressure up to 15 lb/in2 was applied to the tops of the columns to speed up the flow, if necessary. Before using the columns, they were washed at pH 2.3 and pH 7-0 until no absorption was noted at 220 m~. Serum previously heated at 56 ° C for 30 min was passed through an appropriate column. The column was washed with 1 per cent NaC1 solution or 0.1 M phosphate at pH 7.0 until an optical density of less than 0.01 at 220 m~ was recorded. The antibody was eluted with 1 per cent NaC1 solution acidified with HC1 to pH 2.3, or a phosphate buffer (0.1-1.0 M) at pH 2.3. The pH of the eluate was adjusted to neutrality with 0.2 M NaOH. Determination of precipitable antibody. Mixtures of antigen and antibody in various proportions were maintained at 37 ° C for 2 hr, then at 5 ° C for 1-3 days. Precipitates washed twice with cold 0.01 M borate buffered saline at pH 8-0, were dissolved in one drop of 0.5 M NaOH and total protein determined. Antibody concentrations were calculated from precipitates formed in slight antigen excess by subtracting out the known amount of antigen added. * See f o o t n o t e on p. 220.
222
N.
WELIKY,
H. H.
WEETALL, R.
V. GILDENand D. H. CAMPBELL
In the case of anti-BSA determinations, the use of Ilal-labeled BSA permitted the subtraction of the exact quantity of BSA in the precipitate. Purity was measured by the ratio of precipitable antibody to total protein determined by the Lowry method. Determination of anti-BSA antibody by antigen binding capacity (ABC). This method~12~ was used for the rapid estimation of relative amounts of anti-BSA in various samples. T h e technique depends upon the solubility of I~I-BSA in 1/2 saturated ammonium sulfate solution, in which complexes of I~31-BSA and antibody are insoluble. Antibody samples of various dilutions were mixed with an antigen solution and precipitated. T h e antibody dilution was determined at which 33 per cent of the added antigen was found in the precipitate. T h e ABC of the sample was calculated as follows: (0.33) (~g/ml antigen added) ABC = fractional dilution at the 33 per cent reference point
Determination of protein concentration.
Protein was determined using the
procedure of Lowry~TM.
Immunoelectrophoresis~4~. Electrophoresis was carried out for 1 hr at 250 V on microscope slides in a gel made from a 1 per cent solution of agar in veronal buffer, p H 8.6, ionic strength 0.1. T h e antiserum was added to the trough and diffusion permitted to continue for 24 hr~ T h e slides were then fixed and stained with Amido Black 10B. TABLE 1. NONSPECIFIC PROTEIN ELUTED FROM CMC AND FIBROIN ANTIGEN COLUMNS*
Column CMC-BSA CMC-BSA CMC-BSA~" CMC-BSA CMC-HGG Fibroin-R Fibroin-R Fibroin-R CMC-R'
Test serum
Serum added (ml)
Protein eluted at pH 2"3 (mg/ml serum)
3"0 2"0 5'0 20"0 5"0 5"0 2'0 20"0 5.0
0"010 0'008 0"011 0'008 0"008 0.008 0"006 0'010 0.010
NRS Anti-HGG Anti-HGG NRS Anti-BSA NRS Anti-HGG NRS Anti-HGG
All columns consisted of 0.5-0.6 g CMC- or fibroin-antigen. The symbols used are NRS (normal rabbit serum), anti-BSA (rabbit antibovine serum albumin), and anti-HGG (rabbit anti-human ~,-globulin). * Nonspecific protein retained by these columns at pH 7'0 was removed, in general, by 0"1 M phosphate buffer at pH 2"3. ? Elution was with NaC1 solution acidified to pH 2.3 with HC1. EXPERIMENTAL
RESULTS
Antigens coupled to carboxymethylcellulose or fibroin were found to be capable of removing antibody from antisera prepared against the antigen. Evidence for covalent coupling of antigen to C M C rather than adsorption or entrapment was
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
223
demonstrated by omitting the DCC from the reaction of CMC with ~-DNP-lysine and with R'. All but tr~ces of these highly colored compounds were removed by washing. If DCC was present, the deep color was retained. Coupling of BSA, labeled with 1131, to CMC resulted in 93 per cent transfer of activity from the solution to the CMC. Arsenic found in the fibroin coupled to diazotized arsanilic acid amounted to 43 mg/g of CMC. Evidence concerning the effectiveness of antigens coupled to CMC and fibroin for the purification of antibody and for studying the behavior and reactions of antibody is given below. It must be emphasized that reliable results can only be obtained if the CMC or fibroin coupled to antigen is free of uncoupled antigen. These materials must be washed with water and acids or bases capable of extracting traces of the antigen, particularly haptens, each time they are used. After coupling, the CMC-antigen particles should be stirred and well dispersed in large quantities of water or buffer to ensure hydrolysis of the DCC. In the case of haptens, organic solvents may be used to extract excess hapten and unhydrolyzed DCC. TABLE 2. NONSPECIFIC PROTEIN ELUTED ON p-AMINOBENZ~YLCELLULOSE
COLUMNS
Bound antigen
Test serum
Serum added (ml)
Protein eluted at pH 2'3 (mg/ml serum)
HGG BSA BSA BSA R* R* R* R* R*
Anti-BSA NRS NRS Anti-HGG NRS NRS Anti-HGG Anti-HGG Anti-HGG
2.0 2-0 2'0 3"0 2"0 3"0 3"0 1"0 20"0
0-115 0"213 0'117 0-082 0'093 0"107 0.090 0-180 0"100
In all cases serum was washed from columns with 0'1 M phosphate at pH 7 and eluted with a phosphate buffer between 0-1 and 1"0 M. See Table 1 for test serum symbols. * A 1,5-dihydroxynaphthylazophenylarsonic acid derivative of diazotized p-aminobenzylcellulose.
Retention of nonspecific protein. Normal rabbit sera and immune rabbit sera prepared against other antigens were passed through columns composed of 0.5-0.6 g of CMC-antigen and R-fibroin. In all cases 0.011 mg or less protein per ml of serum was released at pH 2.3 (Table 1). Retention of nonspeeifie protein by columns composed of diazotized p-aminobenzylcellulose coupled to protein was found to be higher by a factor of 10 (Table 2). Acid dissociation of antibody from antigen columns. The results of experiments designed to determine the effectiveness of hydrochloric acid in 1 per cent NaC1, at pH 2-3, in dissociating these antigen-antibody complexes, and the effect of this acid treatment on the integrity of the antibody are presented in Table 3. The purity of the antibody dissociated in this manner (measured by the fraction of precipitable antibody) ranged from 90-100 per cent. Between 88 and 97 per cent of the total antibody of the original serum could be accounted for. By immunoelectrophoresis,
224
N.
WELIKY,
H. H.
WEETALL,
R. V. GILDENand D. H. CAMPBELL
the antibody was shown to have no detectable components other than 7-globulin (Fig. 1). Acid was found not to dissociate a n t i - D N P antibody from C M C - l y s i n e - D N P columns. Using a 0.6 g column, all 9-4 mg of anti-DNP antibody contained in 5 ml of rabbit serum were retained. Elution with 1 per cent NaC1 acidified to p H 2.3 with HC1 removed 1.0 mg. T h e remainder could be recovered without activity using 0.1 M NaOH. T o verify that DNP-antibody complexes from this serum did not dissociate at p H 2.3, the following experiments were performed. An antigenantibody precipitate obtained by reaction of a n t i - D N P serum with dinitrophenylated-BSA did not dissolve in 1 per cent NaC1 acidified to p H 2.3 with HC1. No protein could be detected in the supernatant liquid after centrifugation. Partial solution was realized in HC1 at p H 2.3 in the absence of added NaC1. T h e presence TABLE 3a. PURITY OF ISOLATED ANTIBODY FROM 0 " 5 - - 0 " 6 g COLUMNS
Column material
Serum added (ml)
Serum antibody (mg/ml)
Total Serum antibody (mg)
Antibody in serum effluent (mg)
5"0 5'0 5-0 5.0
3.52 2.25 3'08 3.08
17.60 11.25 15"40 15.40
12-00 7-42 8"30 --
CMC-BSA CMC-HGG Fibroin-R CMC-R'
Protein Percentage eluted at precipitable pH 2"3 antibody (mg) in eluate 3'65 3-52 5"52 9.0
90"2 97'4 101 "7 101.0
TABLE 3b. RECOVERY OF ANTIBODY CALCULATED FROM DATA OF TABLE 3a
Column material CMC-BSA CMCoHGG Fibroin-R CMC-R'
Antibody retained by column (mg)
Percentage antibody recovered from column
Percentage overall yield
Percentage* total antibody accounted for
4'60 3'83 7.10 --
79 92 78 --
21 31 36 60
89 97 90 --
* Percentage determined by the sum of the antibody in the effluent and eluate, divided by the total serum antibody. of D N P - B S A in the supernatant liquid was confirmed by the absorbance maximum at 360 m~ and its directly observable color. Because B S A - D N P is insoluble at p H 2.3 in the presence or absence of added NaC1, any solubility was due to the formation of soluble complexes with the antibody. A 1 per cent NaC1 solution containing B S A - D N P and either human or rabbit 7-globulin was found to precipitate the B S A - D N P at p H 2.3, leaving 7-globulin in solution. T o verify that the antigen columns extract antibody from low titer sera and that the antibody can be recovered in high purity, rabbit a n t i - H G G serum containing 2.9 mg of antibody per ml and rabbit anti-phenylarsonic serum containing 2.4 mg of antibody per ml were diluted to 0.29 mg/ml and 0.24 mg/ml, respectively, with normal rabbit serum. T h i r t y ml of each were passed through appropriate antigen-
FIG. 1. Immunoelectrophoretic comparison of normal rabbit to antibody isolated with acidified NaCl (bottom well) against serum. The antibody migrates in the r-globulin region.
(Facing p. 224)
serum (top well), sheep anti-rabbit
FIG. 2. Immunoelectrophoretic comparison of antibody isolated with acidified NaCl (top well) to antibody isolated with 0.1 M phosphate buffer at pH 2.3 (bottom well) against sheep anti-rabbit serum.
S y n t h e s i s and U s e of some I n s o l u b l e I m m u n o l o g i c a t l y Specific A d s o r b e n t s
225
C M C columns. T h e antibodies were eluted with 1 per cent sodium chloride solution acidified to p H 2.3 with HC1. T h e recovered antibody was 90 per cent precipitable (Table 4). T h r e e experiments were performed to determine whether proportions could be increased for larger scale operations: (a) T w o batches of C M C - B S A were prepared using 25 mg and 200 mg of BSA per 0"6 g CMC, instead of the 100 mg of BSA previously used. No antibody was eluted at p H 2.3 from B S A - C M C prepared with 25 mg BSA. T h e other column adsorbed 9.10 mg from 22.4 mg of antibody in TABLE4.
PURITY OF ANTIBODY ISOLATED FROM LOW TITER SERA*
Column antigen
Diluted serum (mg/ml)
Diluted serum added to column (ml)
Percentage precipitable antibody
0-290 0-240 0"240
30 30 30
90'0 94-7 86"3
CMC-HGG CMC-R' CMC-R'
* High titer serum was diluted 1 : 10 with normal rabbit serum for these experiments. Antibody was eluted with 1 per cent NaC1 acidified with HCI to pH 2.3. 5 ml of sera. T h e recovery from the column was 81 per cent. T h e overall yield on one pass was 33 per cent. T h e purity was 87 per cent. (b) T h e effect of increasing column length was determined for C M C - B S A and phenylarsonic-CMC by using 2 g instead of 0.5 g columns. T h e results shown in Table 5 indicate that better overall yields may be obtained by increased column size. TABLE 5. PURITY AND RECOVERY OF ANTIBODY FROM
Column material
CMC-BSA CMC-R' CMCoR'
Serum added (ml)
Total serum antibody (mg)
Percentage overall yield
Percentage antibody recovered from column
5"0 5.0 5"0
17'60 9.60 12.70
47-4 71-3 94'3
83"5 71 "3 94"3
2"0
g COLUMNS
Percentage* Percentage total precipitable antibody antibody accounted in for eluate 90'6 71 '3 94"3
91'7 96-6 101 '2
* See footnote to Table 3b.
Effect of phosphate on acid dissociation of antibody. T h e above results on acid dissociation were obtained using 1 per cent NaC1 solution acidified to p H 2.3 with HC1. If the antibody was eluted with a buffer, between 0.1 and 1.0 M in phosphate at p H 2-3, the fraction of precipitable antibody decreased by 30-90 per cent. Quantitative precipitin determinations were made after dialysis against 0.05 M phosphate buffer or 0.9 per cent NaC1. T h e phosphate eluted protein was shown to be immunoelectrophoretically identical to antibody which was 90-100 per cent precipitable (Fig. 2).
226
N. WELIKY, H. H. WEETALL, Ro Vo GILDENand D. H. CAMPBELL
In Table 3 it was shown that after a sequence of operations, including complexing of the antibody to an antigen-column followed by elution, all of the antibody could be accounted for. The antibody isolated was 90-100 per cent precipitable. Thus, eluting with phosphate buffer should not remove more nonspecific protein. To test this, 5 ml of rabbit anti-BSA serum were passed through a 0.5-0.6 g CMC-BSA column. The column was washed with 1 per cent NaC1 and eluted with 1 per cent NaC1 solution acidified to pH 2.3 with HC1. The eluate contained 4.05 mg of antibody. Further elution with 0.1 M phosphate buffer at pH 2.3 followed by 1'0 M phosphate buffer at pH 2.3 brought down less than 1 per cent additional protein. Thus the decrease of the fraction of precipitable antibody in the protein dissociated with phosphate buffer results from denaturation. Further, if nonspecific serum is passed through a column and washed with phosphate buffer at pH 2.3 no more protein is released than with 1 per cent NaC1 solution acidified by HCI (Table 1). Phosphate buffer at pH 2.3 thus releases antibody which is denatured to various degrees. Serum depletion experiments, to be described next, demonstrate that washing the columns with 0.1 M phosphate buffer at pH 7.0, and eluting with TABLE 6. D E P L E T I O N OF ANTIBODY FROM SERUM USING 0"5-0'6 g CMC-BSA COLUMNS
Serum (ml) 3'0 3.0 2.0
Total serum Percentage depletion of serum antibody* Total protein antibody recovered (mg) Pass 1 Pass 2 Pass 3 (mg) 9.81 6.54 4281"
78.6 94.9 39.7++
102.7 101.4 84.1 ,+
103.6 102.7 93'4~
10.2 6.72 --
* All values uncorrected for nonspecific protein. @ABC value of 2'0 ml of serum. ++Percentages derived from ABC values of serum effluent. phosphate buffer at pH 2.3 does not affect the amount of antibody recovered. Therefore the quantity of protein released at pH 2.3 is equivalent to the quantity of antibody released. The ABC of the serum effluent provides a direct measure of the quantity of antibody remaining in the serum. Depletion of antibody from serum. The fraction of antibody removed from serum, per pass, by an antigen column was variable. No systematic effort was made to find the cause. Nevertheless, antibody could be exhausted from the serum by several passes through an appropriate antigen column. Phosphate buffers up to 1.0 M were used for elution because they have the advantage, over acidified NaC1 solution at pH 2.3, that pH changes in the column are rapid and therefore antibody is recovered in higher concentration. Phosphate does not absorb in the ultraviolet region so that the column effluent can still be monitored spectrophotometrically and the portion containing antibody collected. The total quantity of protein released at pH 2.3 was equal to the total antibody in the serum used, if allowance was made for 10 t~g of nonspecific protein per ml of serum. In those cases where antibody in the serum effluent was measured by the ABC test, complete depletion of antibody activity was also found (Table 6).
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
227
DISCUSSION We have demonstrated that precipitable antibody can be isolated in high purity from CMC-antigen columns and fibroin-phenylarsonic columns. Contrary to previous reports, ~4,~ there did not appear to be any serious problem with retention of 'tightly bound' antibody, under the elution conditions used, except for anti-DNP antibody. The antibody complexed during the first pass should be the most tightly bound. We have shown, however, that recovery is very high. Furthermore, the serum depletion experiments demonstrate that even 'weakly bound' antibody, if present, can be complexed to the column, and recovered. In all of these experiments we made it a point to leave the antibody complexed on the column for a minimum period of time. Most experiments were completed in 4 hr or less, and none permitted to exceed 12 hr. Campbell, Luescher and Lermantl~ have previously reported that the longer antibody is permitted to remain complexed with antigen on a cellulose column, the lower the recovery of the antibody. We have also found that the presence of phosphate at pH 2.3 can cause partial loss of antibody activity, At this pH most of the phosphate is unionized or singly charged. Thus the nature of the ionic and molecular species present at this pH may have profound effects on the behavior of antibody molecules. Antibodies bound to these antigen columns can be washed with large quantities of saline at pH 5-7 without appreciable losses. In spite of this, only a portion of the serum antibody may be extracted per pass, even though the column is not loaded to capacity. This suggests that orientation and diffusion factors are important in reactions in which complexes are formed on antigen columns. Orientation and steric factors pertain both to coupled antigen and soluble antibody. Antigenic sites on macromolecules will be unavailable for binding if they are turned toward the inert carrier or if binding to one antibody molecule sterically blocks binding by others. Blocking of polymer surface interstices into which antigen molecules have wandered, by other antigen molecules or complexed antibody molecules, may further reduce the fraction of theoretically available antigenic sites. Hydrogen bonding of polar antigen groups to polar groups on the carrier may also reduce available sites. Nonrandom coupling may increase or decrease the estimated number of sites. The above considerations, together with channeling of antibody solutions between carrier particles (in the case of columns), make plausible the lack of stoichiometry between antigen and antibody if carrier coupled antigens are used. Increasing the time of contact between serum and antigen is limited by the development of irreversible binding.~l~ Further improvements in yield can be achieved by increasing surface area, increasing the ratio of coupled antigen to antibody (e.g. longer column beds) and varying coupling methods so as to improve antigen orientation and site availability. Anti-DNP antibody behaved differently from the other three antibodies with respect to isolation from its specific antigen column. Evidence obtained indicates that rabbit anti-DNP antibody does not dissociate appreciably from its antigen in 1 per cent saline solution adjusted to pH 2.3 with hydrochloric acid. The specific precipitate formed with highly substituted BSA-DNP did not dissolve. The binding at pH 2-3 was specific because neither RGG nor H G G coprecipitated with BSADNP at that pH. If no added salt was present, 10-15 per cent of the specific precipitate dissolved forming a colored solution. Because the DNP-BSA alone was
228
N. WELIKY,H. H. WEETALL,R. V. GILDENand D. H. CAMPBELL
insoluble at that pH the precipitate must have dissolved in the form of soluble complexes. Our results agree with those of Bennett and Haber(7~ in that complete dissociation in acid solution may not be assumed a priori. We did find, however, that anti-BSA, anti-HGG and anti-phenylarsonic antibodies are largely dissociated from their antigens at pH 2.3. We have developed a simple, rapid method for coupling antigens to carboxymethylcellulose, that permits not only proteins but many haptens to be bound. Nonspecific adsorption is about 0.01 mg per ml of serum, using columns composed of 0.5-0.6 g of CMC-antigen. Recovery of the complexed antibody, and precipitability of the recovered antibody is high. Coupling conditions are mild. No blocking groups are necessary to inactivate uncoupled sites on the CMC, and the CMC remains clean and white or assumes the color of the antigen. Thus, deterioration of contamination may often be clearly observed by color changes. In addition, coupling to protein antigens occurs at sites different from those which couple to diazonium compounds. We have also synthesized a hapten-coupled fibroin column with similar properties, except for color. The advantages of columns for isolating highly purified antibodies have not been fully exploited. They present a simple one-step technique as opposed to the many steps required by other procedures. Further, these columns can be re-used many times, an important consideration in cases where antigens are difficult to obtain or expensive. In general, substances with basic amino groups will couple to insoluble carboxymethylcellulose in the presence of N,N'-dicyclohexylcarbodiimide. These include bovine serum albumin, human 9,-globulin, keyhole limpet hemocyanin, ¢-DNPlysine and p-(p'-aminophenylazo)-phenylarsonic acid. The coupling is presumably through an amide bond formed between an amino group of the antigen and a carboxyl group of the CMC: cellulose
cellulose
I
O
O
I CH2 ] ÷RNH~ C=O
I
OH
DCC
CH2 -+ [ +H20 C-----O
I
HN--R
The possibility that antigen was adsorbed or entrapped rather than chemically bound to the carboxymethylcellulose was eliminated by the clear absence of coupling, if N,N'odicyclohexylcarbodiimide was omitted from the reaction mixture. Because of the mildness of the reaction conditions we have been exploring the coupling of CMC to enzymes and other biologically active molecules. These active substances coupled to carboxymethylcellulose are now being thoroughly investigated. Full advantage has not yet been taken of the use of insoluble carriers, coupled to such organic and biologically active molecules, as research tools for the purification of complementary substances and for studies of the characteristics of the bound molecules.
Synthesis and Use of some Insoluble Immunologically Specific Adsorbents
229
SUMMARY (1) We have utilized the reaction of acids and amines to form amides in the presence of N,N'-dicyclohexylcarbodiimide as a means of coupling proteins and other organic molecules with basic amino groups to carboxymethylcellulose. (2) T h e s e carboxymethylcellulose derivatives can be used for the isolation of antibody from antisera directed against either proteins or haptens. (3) T h e antibody isolated is of high purity. T h e procedure does not result in appreciable losses of antibody remaining in the serum, and reasonable yields may be expected. (4) T h e carboxymethylcellulose derivatives exhibit very low retention of nonspecific protein near p H 7 and so may be used to isolate antibody from low titer sera. (5) T h e carboxymethylcellulose derivatives are colorless or assume the color of the coupled substance. T h e y need no blocking groups after coupling, and coupling conditions are mild. Flow rates are reasonable, but in general pressure is required. (6) T h e technique offers a simple method for coupling an insoluble carrier to proteins at a site other than that resulting from coupling diazonium compounds to proteins. (7) Dissociation of rabbit anti-BSA, a n t i - H G G and anti-phenylarsonic antibodies at p H 2.3 was found to result in high recoveries of antibody bound at neutrality to specific antigens previously coupled to CMC. (8) A n t i - D N P antibody from a rabbit serum did not dissociate appreciably from its antigen at p H 2.3 whether on an antigen column or in the form of a specific precipitate. (9) Coupling of proteins and organic compounds with basic amino groups to C M C is applicable to enzymes and other biologically active molecules, in many cases without loss of activity. Acknowledgements--We thank Drs J. S. Garvey, A. Malley and A. Amkraut of the California
Institute of Technology for helpful discussions and for sera. We also thank Dr M. Litt of Allied Chemical Corp. for his contributions to a discussion of organic coupling methods. REFERENCES
1 CAMPBELLD. H., LUESCHERF. and LERMANL. S., Proc. Nat. Acad. Sci., Wash. 37, 575 (1951). ~ L ~ A N L. S., Nature 172, 635 (1953). a PUTNAMF. W. (Editor), The Plasma Proteins, Vol. I., p. 130. Academic Press, New York (1960). 4 MOt~DOALN. R. and POaTEa R. R., Biochem. Biophys. Acta 71, 185 (1963). 5 GVRVICHA. E., Kav~mR R. B. and NEZraN R. S., Biokhimiya 24, 129 (1959). 6 G ~ w c a A. E., KOZOVLEVAO. B. and TUM~a~OVAA. E., Biokhimiya 26, 803 (1962). 7 BENrq~TT J. C. and HABEa E., ft. Biol. Chem., 238, 1362 (1963). s SEHONA. H., Brit. Med. Bull. 19, 183 (1963). 9 MAL~EYA. and CAMPBelL D. H., ft. Amer. Chem. Soc. 85, 487 (1963). 10 LANDSTEINERK. and VAN DER SCHEERJ., ~¢. Exp. Med. 55, 781 (1932). ~t CAMPBELLD. H., GARWVJ. S., CRAMERN. E. and SUSSDOR~D. H., Methods in Immunology, p. 69. W. A. Benjamin Inc., New York (1963). ~ FArm R. S., ft. lnfect. Dis. 103, 239 (1958). ta LOWRy O. H., ROS~nROU~H N. J., FARR L. and RANDALLR. J., ~. Biol. Chem. 193, 265 (1951). ~a GRaBAa P. and W~LL~aMSC. A., JR., Biochim. Biophys. Acta 10, 193 (1953).