Journal of Biochemical and Biophysical Methods, 25 (1992) 285-297

285

© 1992 Elsevier Science Publishers B.V. All rights reserved 0165-022X/92/$05.00

JBBM 00971

Coupling of antibody-binding fragments to solid-phase supports: site-directed binding of F(ab') 2 fragments Martin L. Yarmush, Xiao-ming Lu and David M. Yarmush Department of Chemical and Biochemical Engineering, Rutgers University, Piscataway, NJ (USA) (Received 25 June 1992) (Revised version received 15 August 1992) (Accepted 23 August 1992)

Summary A method to covalently bind antibody fragments, via their carbo~! termini to solid supports, is presented. The strategy involves: (1) reversibly blocking all the accessible carboxyl groups on the antibody molecule with phenylhydrazine, (2) exposing the carboxyl termini of the fragment by enzymatic digestion with pepsin and (3) subsequently coupling the fragment to an appropriate support. Experiments with an anti.bovine serum albumin monoclonal antibody and C-14 phenylhydrazine revealed that the blocking step was nearly completely reversible with a dilute solution of FeCI 3. Radioiodinated blocked F(ab') 2 fragments were then coupled to an amino-functionalized Sepharose 4B column, and characterized as to their coupling capacity (mass of protein coupled/ml of bead), and antigen-binding activity. The coupling capacity of the blocked fragments was found to be 12%, half the coupling efficiency of unmodified radioiodinated F(ab') 2. The antigen-binding capacity (mol antigen bound per tool antibody coupled) for the blocked F(ab') 2, on the other hand, was found to be 1.9, which was approx. 3.5-times greater than for the unmodified F(ab') 2. Comparisons with other conventional coupling techniques were also made. These preliminary studies suggest that this technique can provide one with the means to obtain more uniform and active populations of immobilized antibody fragments.

Key words: Immunoadsorbent; Monoclonai antibody; Immobilization; F(ab') 2 fragment; Affinity chromatography

Correspondence address: M.L. Yarmush, Department of Chemical and Biochemical Engineering, Rutgers University, P.O. Box 909, Piscataway, NJ 08855, USA.

286

Introduction With the advent of monoclonal antibody (mAb) production technology, protein purification by immunoadsorption has received increased attention. This heightened interest is evident on the laboratory scale, where new proteins of importance to the basic scientist are continually being isolated, as well as on the industrial scale, where several therapeutic and diagnostic proteins are being produced [4]. The more widespread use of immunoadsorption has also piqued an awareness of the important factors involved in the design of efficient and economical immunoadsorbent systems, including issues such as the yield of active antibody upon immobilization. There are various well-documented coupling chemistries that eifect satisfactory immobilization of antibodies to a support. In all cases, immobilization is dependent on the reactive sites on the matrix as well as the functional groups on the antibody. For example, one of the most popular coupling chemistries, CNBr activation [2,7], utilizes the OH groups on a typical support and primarily the e-amino grou~s of lysine on the antibody. However, because the iysine groups are distributed throughout the antibody molecule, coupling via most conventional methods leads to a randomly oriented antibody population with a portion of the immobilized species unable to bind antigen. The overall consequence is an immunoadsorbent which in many cases has greatly reduced binding efficiency [15,17]. Some recent reports have described the oriented coupling of antibodies onto supports through the oligosaccharide moiety of the antibody. By coupling the oxidized carbohydrate moieties to matrices with hydrazide (Hz) groups, Hoffman and O'Shannessy have demonstrated a vastly increased binding capacity of polyclonal antibodies for its antigen relative to the random coupling obtained by using N-hydroxysuccinimide (NHS)-activated matrices [8]. The carbohydrate technique gave results favorable for polyclonal antibodies but less so for mAbs. Similarly, Matson anL, Little [9] have also demonstrated the superiority of antibody immobilization via the carbohydrate region onto Hz-containing supports over immobilization onto either cyanogen bromide (CNBr) or NHS-containing supports. However, the carbohydrate-coupling procedure proved to be only marginally better than the coupling with CNBr-Sepharose and NHS-Affi-Gel 10 when used for mAbs. Thus, it would appear from these experiments that mAb coupling efficiencies and binding capacities with hydrazide chemistry are somewhat variable and that the true benefits of hydrazide chemistry may need to be determined for each and every mAb. In this paper we describe initial results using a different approach t~ward oriented coupling. Our strategy involves: (1) reversibly blocking all the accessible carboxyl groups on an antibody molecule, (2) producing a carboxyl-group-protected F(ab') 2 fragment by enzymatic digestion (which would have two exposed carbo~l groups at the C-terminals), and (3) subsequently coupling this fragment to an appropriate support. Using a well-characterized mAb, we demonstrate below (1) the feasibility of this approach and (2) a preliminary comparison with other more conventionally prepared immunoadsorbents.

287

Materials and Methods

Chemicals and reagents Phenylhydrazine (Phz, specific gravity- 1.098), ammonium sulfate, ferric chloride, trizma hydrochloride (tris(hydroxymethyl)aminomethane hydrochloride), sodium acetate, glycine, sodium chloride and 1-ethyl-3-[3-dimethyl-aminopropyl]carbodiimide hydrochloride (EDAC)were obtained from Sigma (St. Louis, MO). Ethylenediamine (cerified)was purchased from Fisher Scientific (Fair Lawn, N.I) and trichloroacetic acid (TCA) was obtained from EM Science (Cherry Hill, NJ). [14C]phenylhydrazine HC! (14C-Phz) and Nal~I were obtained from ICN Radiochemicals (Irvine, CA). Affi-Gel Hz hydrazide gel was obtained from Bio-Rad Laboratories (Richmond, CA). CNBr-activated Sepharose 4B and AH-Sephax,ose 4B were from Pharmacia LKB (Piscataway, NJ). Milli-Q-treated water was used in all solutions. Bovine serum albumin (BSA) monomer was obtained from Sigma and shown to be free of dimers or larger aggregates by high-performance liquid chromatography (HPLC). Pepsin was a product of Worthington Bi~chemical (Freehold, NJ). These two proteins were used without further purificatioa. Monoclonal antibody production and purification The anti-BSA mAb, 9.1 [12], was produced in spinner flasks by conventional techniques :[5]. Purification of the mAb was performed in the following manner. 200 g (NH4)2SO4 was added to 800 ml clarified bioreactor solution while maintain' ing thorough stirring. The precipitate was spun at 300x g for 5 min at i5 ° C, redissolved in 80 ml water, and dialyzed first against water (2000 ~ml) for 4 h at 4°C and then against phosphate-buffered saline with 0.02% sodium azide, (pH 7.2; PBSA) for the same Jength of time. A small amount of precipitate formed during dialysis was removed by centrifugation before applying the solution to a BgA-Sepharose 4B column. The mAb was eluted with glycine buffer (0.1 M, pH 2.5) and dialyzed immediately against PBSA. This solution was then concentrated to a volume of 15 ml by using a micoconcentrator with 30 kDa cut-off (Amicon, Danvers, MA) and eluted through a Sephacryl S-300 size-exclusi0n column (98 x 2.5 cm, I.D.)with PBSA. The fractions corresponding to the antibody monomer peak (as detected by online UV absorbance at 280 nm) were collected and concentrated to 4.3 mg/ml and stored at 4°C. The purity of the mAb was verified with isoelectric focusing (IEF) using silver staining (PhastSystem; Pharmacia LKB) and HPLC analysis using a TSK G300SW, 300 x 7.5 I.D. column (Toso Haas, Philadelphia, PA). IEF revealed four sharp bands centered about an average of pl 7.0 [12]. Radiolabeling and radioactivity measurements Both mAb and F(ab') 2 fragments (purified as described below) were radiola: beled with ~25I according to the ICI method [10]. Iodinated mAb and F(ab') 2 preparations were extensively dialyzed against PBS, eluted through separate Sephacryl S-300 columns (60 x 1.5 cm, I.D.) and tested for incorporation of the iodine by precipitation with 10% TCA. Greater than 99.5% of the radioactivity was present in the precipitate for all preparations used in this study.

288 Radiolabeled ~4C-Phz samples (described below) were added to 3.3 ml of scintillant cocktai| and counted in a Beckman LS 7000 scintillation counter (Beckman Instruments, Fullerton, CA). ~I-labelled antibody or antibody fragments were counted in a Beckman Gamma 5500B counter. Protein concentrations were determined by absorbance measurements at 280 nm using an extinction coefficient of 1.4 ml/mg per cm for the mAb and 0.66 ml/mg per cm for BSA using a Beckman DU-70 UV-visible spectrophotometer.

Chemical modifications S:ep i. Blocking COOH groups of 9.1 mAb. 'The 9.1 mAb stock solution (4.3 rag/m!) was dialyzed against phosphate-buffered saline (PBS; 10 mM phosphate; pH 7.2) and aliquoted in equal amounts (1.16 ml). Aliquo~s of 5.8 ~1 of the 14C-gaped Phz stock solution (0.923 rag, 7.8 mCi/mmol) was mixed completely with 45 ~! phenylhydrazine (molar latio of the 'cold' to 'hot' phenylhydrazine was 70:1), and the resulting mixture was pipetted into the sample tubes along with 6.3 mg of EDAC. A Phz: mAb molar ratio of approx. 1500: 1, and an EDAC:protein molar ratio of approx. 1000:1 was chosen based on prior reaction optimization experiments with BSA and other proteins. The reaction was maintained under an inert gas atmosphere (helium) to prevent oxidation of Phz, and the pH was maintained at around 4.8 (± 0.3) by adjustment with 0.01 N HCI during the first 30 rain of the activation. Experiments were also performed under more stable pH conditions using 0.1 M acetate buffer (pH 4.8) which had the potential of reacting with EDAC to form acetylated protein and phenylhydrazide. Both buffered and unbuffered 9.1 mAb solutions were found to be blocked in a similar manner. Buffered solutions appeared clear whereas unbuffered solutions showed slight cloudiness. After 4 h, unreacted Phz was removed (described below) by applying the reactant mixture to a PD-10 column (Pharmacia LKB)which had been pre-equilibrated with protein (to reduce non-specific absorption). Control tubes contained mAb without Phz and EDAC. All reaction and purification steps used for the test samples were also performed on the controls. Step 2. Purification of the blocked mAb. Excess Phz and EDAC were removed using a PD-10 column which had been equilibrated with 0.1 M acetate buffer (pH 4.8). The collected volume of modified mAb was, on average, 2.4 ml. Three aliquots (0.1 ml) were counted, and the number of accessible COOH groups was calculated from the measured radioactivity. Step 3. Digestion of modified mAb and controls. The blocked mAb and control solutions were concentrated down to ml using a Centriton-30 microconcentrator (Amicon) and transferred to polypropylene tubes. 100/~1 of a 2 mg/ml pepsin stock solution in acetate buffer (0.1 M, pH 4.8) was added to each tube. The digestion was carried out in a water bath at 37 °C overnight and was quenched by adding 0.5 ml Tris=HCl (2 M, pH 8.0) to each digestion tube. Step 4. Purification of F(ab')z fragments. Digested mAb Was applied to a size-exclusion column ($ephacryl S-300, 68 × 1.5 cm) equilibrated with acetate buffer (0.1 M, pH 4.8). Fractions containing F(ab') 2 fragments w e r e collected and pooled (18 ml), and then concentrated to a volume of about 3.3 ml using a

289

Centricon-lO microconcentrator (Amicon). Three aliquots (0.1 ml) were taken from each sample and counted. IEF was used to monitor both the coupling (blocking reaction) and removal (deblocking reaction) of Phz. The covalent attachment of Phz reduces the number of negative charges and hence the net charge of the protein becomes more positive causing it to migrate toward the anode. The subsequent removal of Phz should then result in.the protein migrating back to a position corresponding to its original p I. To calculate the number of Phz molecules on F(ab')2, it is necessary to determine the extinction coefficient of the F(ab') 2 fragments. Protein concentration analysis for both modified and unmodified F(ab') 2 were performed with a modified Lowry protein assay using BSA as the standard (protein determination per procedure 690; Sigma). The mean extinction coefficients of blocked and unblocked F(ab') 2 were found to be 1.19 and 1.15 ml/mg per Cm, respectively, assuming a Mr of 100 kDa for the F(ab')2 fragments. Step 5. Deblocking of pHz from F(ab')2. For the deblocking study, 0.l-ml aliquots of the blocked F(ab') 2 solution were tranferred to polypropylene tubes. After addition of ferric chloride (6/~l of 0.5 M) to each solution, the tubes were gently shaken three times over a period of 30 rain, and their contents applied to a PD-10 column to remove the ferric chloride and Phz groups. Three aliquots were taken from each deblocked mAb sample and counted to determine the deblocking efficiency of the ferric chloride step.

Step 6. Coupling of F(ab') 2 to functionalized beads and adsorption capacity determination. The coupling of 9.1 anti-BSA mAb to Affi-Gel Hz and F(ab') 2 to CNBr-activated Sepharose was performed according to instructions supplied by the respective suppliers. The immobilization was quantified using tzSI-labeled 9.1 mAb for the Affi-Gel Hz reaction and lzSI-labeled F(ab') 2 fragment for the CNBr-activated Sepharose reaction. Both radiolabeled, blocked F(ab'): and unblocked F(ab') 2 were also reacted with AH-Sepharose 4B beads (beads which contain amine functionalities) using excess EDAC. For determination of immunoadsorbent capacity, 2 ml of support material containing immobilized antibody or F(ab')2 fragment protein was incubated for 1 h in tubes with an equal volume of 0.5 mg/ml BSA solution in PBSA, and then packed in a height-adjustable glass column, 8 x 1 cm (Ommnifit, Atlantic Beach, NY). The packed beads were rapidly washed with two column-volumes of PBSA to remove excess BSA, and then with 0.5 M NaCl to remove non-specifically bound BSA. The residual specifically bound BSA was then eluted with 0.1 M glycine (pH 2.5) and the eluted protein was monitored with an HP 1046A fluorescence detector (excitation wavelength 280 nm; emission wavelength 350 nm) connected to an HP !090 HPLC system (Hewlett Packard, Palo Alto, CA).

Results The steps leading to immobilization of F(ab')2 fragments via their carbox'yl termini are presented schematically in Fig. 1. By chemically protecting the accessi-

290

A

I.IOOC~~COOH + fheayxHOOOCOOH PHOC~COPH C

D ÷ feenj

OCCO

O~C ~0

.,~~+Wmt,,

NOOCCOOH F(W)z hqment -$ol•ble eubodilmId¢

OPH

NH! N.He _l . . . . ,__ ~ ' ~ " " -'"

OOCO

NHNH

" Oriented Fleb')z Frasmnt

Coupled F(ab'z)~egmnt

Debloc~e~,

Fig. 1. Scheme portraying the different steps needed to immobilize an F(ab') 2 fragment via the carboxvI termini onto supports containing amine groups.

ble carboxyl groups with Phz (step A) and then enzymatically digesting the antibody with pepsin, we were able to generate two accessible carbo~l groups at the fragment C-terminus (step B). The COOH-protected fragment was then bound to an amine-c0ntaining support (step C), and the immunoadsorbent was reacted with FeC! 3 to remove the protecting group (step D). Solutions containing 5 mg of anti-BSA mAb 9.1 were reacted with 14C Phz, and enough EDAC was added so that EDAC:mAb ratio was 1000: 1. After removing the unreacted components, mAb was found to contain 0.55-0.64 ,ttmol of t4C-phz, which corresponded to approx. 17-19 Phz molecules bound per mAb (Table 1). TABLE 1 Results from chemical modification procedures on mAb 9.1 Yield of blocked F(ab') 2 fragment (%, tool basis) a Yield of deblocking step

(%) COOH groups modified on mAb COOH groups modified on ]~ab°)2 fragment Efficiency of FeCi3 deblocking (%) a Starting material was 5 mg of mAb 9.1.

60

9o 18 5 94

291 10_

6 .*,_ 4

O e~

m 0

.n

f J

Coupling of antibody-binding fragments to solid-phase supports: site-directed binding of F(ab')2 fragments.

A method to covalently bind antibody fragments, via their carboxyl termini to solid supports, is presented. The strategy involves: (1) reversibly bloc...
2MB Sizes 0 Downloads 0 Views