1314
Anal. Chem. 1989, 61, 1314-1317
Analytical Potential of Protein A for Affinity Chromatography of Polyclonal and Monoclonal Antibodies B r u c e J o n Compton,* M a r y A n n Lewis, F r a n c e s Whigham, J e n n i f e r Shores Gerald,’ a n d George E. Countryman Fermentation Development Laboratories, Bristol-Myers Company Industrial Division, P.O. Box 4755, Syracuse, New York 13221 -4755
Proteln A based rapkl afanlty chromatographyfor quantttation of various lmmunoglobuilns of class G (IgG) is descrlbed. Threeminute analysls udng either cltrate or phosphate buffers and detection wtth 220- or 280-nm uttravlolet absorptlon was found to be optimum for quantttation of I@ from 0.25 to 250 pg of IgG ontolumn with a percent relative standard devlatlon ( % RSD) of 2-3% RSD. The method has a detection limit estimated to be 100 ng of IgG oncolumn. I t has been used to analyze a varlety of IgG-containlng samples from such dlverse sources as hybrldoma selectlon, media cultivation, and purlkatlon studies. Gradient elution studlee and the relationship of IgG elution to IgG isoelectrlc point ( P I ) are also descrlbed.
INTRODUCTION It is well-known that many products derived from monoclonal antibody (mAb) hybridomal technology are in the process development and scale-up stage of commercialization. Clearly, rapid, accurate, automated analysis of samples related to mAb development efforts is needed. One mAb immunoglobulin class frequently developed is class G (IgG). Determinations of these large molecular weight proteins (MW approximately 150 000) are normally accomplished by using immunoassay techniques in conjunction with latex coagulation, or radio-, fluoro-, or enzyme-linked development methods ( I ) . These methods are well suited and widely used in hybridomal selection since they are highly sensitive (ng/mL concentration range), but they are, in general, difficult to automate and often lack sufficient accuracy for process optimization studies. Additional methods involve a variety of electrophoretic, immunodiffusion, and immunonephalometric methods ( 2 ) . These methods are 1-2 orders of magnitude less sensitive than the various immuno-linked methods previously mentioned. In general, they are also ill-suited for rapid automated analysis. For instance, radial immunodiffusion (RID) assays are widely used for following hybridomal production and mAb purifications, but they may take up to 3 days to execute. Our approach to implementation of large-scale routine mAb analysis has been the use of affinity chromatography based on high-performance liquid chromatographic (HPLC) equipment. Automated HPLC systems are well developed commercially and as such are regarded as reliable for routine usage. A classic means of studying and isolating antibodies, particularly those of the IgG class, has been the use of protein A from various Staphylococcus species or more recently as a recombinant product (3-5). Studies on the interactions of protein A with IgG classes and subclasses have progressed to a state where much is known and understood regarding the stoichiometry and physiochemical interactions (6-9) of protein A and IgG’s. Additionally, protein A or the more recently commercially available protein G can be used with sufficient
* Author to whom correspondence should be addressed.
On leave from Johns Hopkins Medical School, Baltimore, MD. 0003-2700/89/0361-1314$01.50/0
selectivity to allow IgG class and subclass separations (10-12). One interesting aspect of the literature on protein A as seen in numerous reviews (3-5) is the general omission of studies on the use of protein A for rapid, reproducible quantitation of IgG’s; in other words, the analytical chemistry potential of protein A binding to IgG has not, with a few exceptions (13, 14), been extensively examined or appreciated. We report here on the use of protein A for rapid (less than 3 min), automated IgG analysis of various polyclonal and monoclonal IgGs. Some fundamental studies on rapid affiiity chromatography recently have been published (15-1 7), but our intent is to illustrate the use of commercial instrumentation and materials for rapid, reliable automated IgG analysis for production-related samples. For instance, over 20 000 IgG-related samples have been analyzed by us with this method over the last year by using one instrument and little analyst intervention. A correlation of IgG elution time to the isoelectric point (PI) of IgG is also demonstrated. EXPERIMENTAL SECTION Apparatus and Materials. Protein A bound to a rigid macroreticular polymer support was commercially obtained (AffiPrep protein A, Bio-Rad, Richmond, CA). One 25-mL bottle was adequate to pack 20 4.6 mm i.d. X 5 cm HPLC stainless steel blanks (Type LTS 5 cm, No. 820-05, Keystone Scientific, State College, PA). Column packing was accomplished by pipetting the slurry directly into the column, running eluent through the column for a few minutes, and then adding more packing to fill any column voids. Various mobile phases such as citrate, borate, 2-amino-2(hydroxymethy1)-1,3-propanediol(Tris), and orthophosphate buffers were studied as mentioned and prepared with either sodium or chloride counterions from concentrated HCl or 19 N NaOH. Most studies and routine analyses were conducted with 25 mM sodium citrate (pH 7.6 i 0.1) as loading buffer, and elution was done using 25 mM citric acid (pH 2.4 A 0.1) and 280-nm detection. Highsensitivity analysis was done using a 25 mM sodium phosphate (pH 7.6 0.1) loading buffer and 100 mM sodium phosphate (pH 2.4 i 0.1) elution buffer with 220-nm detection. In general, the loading buffer (high pH) was made by using the elution buffer (low pH) and adjusting its pH to 7.6 f 0.1 with 19 N sodium hydroxide. All chemicals were of reagent grade and were from Sigma Chemical Co. (St. Louis, MO) or Aldrich Chemical Co. (Milwaukee, WI) except Dulbecco’s phosphate buffered saline (DPBS), which was from Gibco (Grand Island, NY), and water was deionized. Monoclonal antibodies produced by Bristol-Myers Industrial Division (Syracuse, NY) or Oncogen (Seattle, WA), polyclonal murine IgG (Zymed Laboratories, Inc., South San Francisco, CA), and polyclonal bovine IgG (Sigma) were diluted with DPBS and used as obtained. The HPLC analysis system consisted of a low-pressure proportioning valve gradient system, diode array spectrophotometric detector, and microprocessor controller and data system (Model 1090M with PV5 option, Hewlett-Packard, Palo Alto, CA) and was used without modification. Procedure. Measurement of IgG Concentration. Samples were collected and centrifuged to remove cells, and aliquots of supernatants were stored frozen. IgG’s can precipitate at high concentrations (>5 mg/mL), especially in unbuffered or ill-suited buffer solutions. At low concentrations, many proteins adsorb
*
0 1989 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 61, NO. 13, JULY 1, 1989
to plastic and glass surfaces. Low-concentration IgG standards were often stabilized with bovine serum albumin (10 mg/mL) solutions of DPBS. Once rapidly thawed in tap water, samples were centrifuged to clarity and analyzed immediately or stored refrigerated at 4 "C for not more than 48 h. Routine analysis of samples containing 10-1000 pg/mL IgG involves equilibration of the protein A affinity chromatography system with 90% sodium citrate buffer-10% citric acid buffer (flow rate 3 mL/min, column temperature 25 "C), followed by blank runs with DPBS to determine if the system base line was appropriate at 280-nm detection. Elution involved a step gradient at 1min to 10% sodium citrate buffer and 90% citric acid elution buffer. This mixture of loading and elution buffer results in actual pH values of 6.0 and 2.8, respectively. A 250-pL aliquot of IgG standard was injected in triplicate, and peak areas were determined by standard chromatographic integration methods. For standards at a concentration of 100 pg/mL, the RSD of the peak areas should be less than 2%. Samples were analyzed by using a 250-pL injection on-column. Sample preparation involved adjusting sample to pH 6.5 or greater as elaborated on later. Additionally, ascites fluid was diluted with DPBS such that IgG levels were below lo00 pg/mL. Unbound material is eluted from the column prior to elution of IgG. Bound IgG eluted slightly after the elution buffer breakthrough and was quantitated by using the same method as that for the standard. For high-sensitivity analysis of samples below 10 pg/mL concentration, a 25 mM sodium phosphate (pH 7.6 & 0.1) loading buffer and 100 mM sodium phosphate (pH 2.4 f 0.1) elution buffer and 220-nm detection were used.
-
u-3
j
m
T
me
( r n i n . )
Flgure 1. Comparison of elution of 21 p g of mAb murine IgG, BM1 with 25 mM (A) sodium citrate (pH 7.6)-citric acid (pH 2.6), (8)sodium phosphate (pH 7.6)-sodium phosphate (pH 2.6), and (C) Tris HCI (pH 7.5)-Tris Ki (pH 2.6) loading-elution buffer combinations and 280-nm detection. Peak shape may be a reflection of the pH elution profiles shown in Figure 2 for the various buffer combinations.
4.9
RESULTS AND DISCUSSION The concept of executing affinity chromatography using protein A is uncomplicated. Protein A binds at or above physiological pH (pH 7.2) to the Fc region of IgG molecules (3-5) in a pseudoimmunological response. The mode of binding probably involves hydrophobic interactions since many protein A loading systems use salt concentrations, often as high as 3 M sodium chloride, along with buffers such as Tris, borate, or phosphate (7-11)to maintain appropriate pH. Elution traditionally occurs a t low pH (
