Anal. Chem. 1995, 67,2943-2948
Conformational Diversity and Conformational Transitions of a Monoclonal Antibody Monitored by Circular Dichroism and Capillary Electrophoresis Michael Kats,* Priscilla C. Richberg, and David Emlyn Hughes Bristol-Myers Squibb Company, Pharmaceutical Research Institute, P.0. Box 4755, Syracuse, New York 13221-4755
Four major isoforms of the BR96 antibodywere separated by micellar electrokinetic capillary chromatography. Heatinduced reversible isoform interconversions were observed at 70 "C, and after extended incubation at 80 "C, all species irreversibly transformed into a new single peak. In the presence of sodium dodecyl sulfate (1.0 mg/mL), the isoform transformations occurred at lower temperatures without altering the separation pattern. Size exclusion chromatography analysis detected no aggregation at temperatures below 80 "C. Parallel circular dichroism m e a " e n t s indicated sigdfkaut mnfomationalchanges at 70-80 "C. The parallelism between isoform transformations and secondary structure changes allows consideration of CE-separated isoforms of BR96 antibody as conformers, an equilibrium between which can be shifted by different physicochemical factors such as elevated temperatures and amphiphilic surfactants. Immunoglobulins are glycoproteins that circulate in the bloodstream of vertebrates and whose biological function is to form and transport high-affinity and high-specificitybinding sites in response to antigenic stress and to trigger a protective reaction of the competent cells upon antigen-antibody complex formation. The discovery of tumor-associated antigens and their detection by monoclonal antibodies led to the development of monoclonal antibody conjugates with antineoplastic drugs which can be selectively targeted to the tumors.' The chimeric monoclonal antibody BR96 recognizes a carbohydrate antigen expressed predominantly on carcinomas of the breast, colon, lung, prostate, and ovary and, to a lesser extent, on normal gastrointestinal epithelium. Its tumor-selective binding activity has been used successfully to target drugs to tumor cells in vivo in nude mice bearing human tumor xenografts and in athymic rats which display the antigen on normal tissues similar to that seen in humans.2 Currently, BR96 is being evaluated in clinical trials to test the therapeutic efficacy of targeting the chemotherapeutic drug doxorubicin to tumor cells. As a modification of an immunoglobulin G (IgG), BR96 antibody (IgBR96) represents a symmetrical Y-shaped glycoprotein with an average molecular mass of 150 000 Da, where about 3%of the mass is accounted for by carbohydrate moieties. Recombinant DNA technology applied to the preparation of monoclonal antibodies derived from transfected animal cells as therapeutic agents requires extensive quality (1) Pietersz, G. A. Bioconjugate Chem. 1990,I , 89-95. (2) Trail, P. A; Willner, D.; Lasch, D. J.; Henderson, A J.; Casazza, A M.; Firestone, R A; HellstrBm, I.; HellstrBm, K E. Science 1993,261, 212215.
0003-2700/95/0367-2943$9.00/0 0 1995 American Chemical Society
control. The identity of the antibodies must be verified and their purity, efficacy, and consistency carefully monitored. Despite significant progress in development of new analytical techniques, biochemical study and quality control of recombinant monoclonal antibodies or antibody-drug conjugates represent a substantial analytical challenge due to the complexity of immunoglobulin molecules and inherent mi~roheterogeneity.~-~ Widespread currently available experimental data provide abundant evidence to the fact that a microheterogeneity pattem is a result of the glycoprotein processing of a particular hybridoma cell line and the culture method involved and is not an artifact of the isolation pro~edure.4-~This variability of isoforms is believed to be a result of deamidation of light chain asparagines or differences in composition of the carbohydrate moiety attached to immunoglobulin polypeptide chain^.^^^ Analytical methods based on molecular charge differences such as ion-exchange ~hromatography,~ chromatofocusing,4isoelectric focusing (IEF'),7 and high-performance capillary electrophoresis (CE)839are often used for the determination of monoclonal antibody isoform pattems. Although the existence of antibodies in multiple conformational states is now generally accepted, owing mostly to information obtained by high-resolution X-ray crystal structure analysis,1° there is no reported analytical method that is capable of detecting, separating, and quantitating individual conformers of immunoglobulines in water solutions. Development of such a technique would enable an improved quality control of monoclonal antibodies and a better understanding of the nature of glycoprotein microheterogeneity and the mechanisms of antigen-antibody interaction. CE has proven to be a powerful tool in the separation of structurally related peptides, proteins, and glycoproteins. Single mutation, deamidation, change in charge, change in hydrophobicity or length, and secondary structure or ambient temperature alterations affect the electrophoretic mobility and under proper conditions can be detected by CE analysis.11-13 The possibility (3) Kaltenbrunner, 0.; Tauer, C.; Brunner, J.; Jungbauer, A. J chromatogr. 1993,639,41-49. (4) Jungbauer, A; Tauer, C.; Wenisch, E.; Uhl, K; Bmnner, J.; Purtscher, M.; Steindl, F.; Buchacher, A J. Chromatog. 1990,512, 157-163. (5) Patel, T. P.; Parekh, R B.; Moellering, B. J.; Prior C. P. Biochem. J, 1992, 285, 839-845. (6) Wenisch, E.;Reiter, S.; Hinger, S.; Steindl, F.; Tauer, C.; Jungbauer, A; Katinger, H.; Righetti, P. G . Electrophoresis 1990,11, 966-969. (7) Wenisch, E.; Jungbauer, A; Tauer, C.; Reiter, M.; Gruber, G.; Steindl, F.; Katinger, H. J. Biochem. Biophys. Methods 1989,18, 309-322. (8)Bennett, L. E.; Charman, W. N.; Williams, D. B.; Charman, S. A.J. P h a m . Biomed. Anal. 1994,12, 1103-1108. (9) Compton, B. J. J. Chromatog?. 1991,559, 357-366. (10) Stanfield, R L;Wilson, I. A Trends Biotechnol. 1994,12, 275-279. (11) Florance, J. R; Konteatis, 2. D.; Macielag, M. J.; Lessor, R A; Galdes A. J. Chromatogr. 1991,559, 391-399.
Analytical Chemistry, Vol. 67, No. 17, September 1, 1995 2943
of using capillary electrophoresis for protein conformational transition detection has been previously suggested.'* Analysis of a-lactalbumin type I11 revealed a conformational transition that resulted in asymmetric peaks and sigmoidal mobility plots versus temperature in the transition region. In the present work, we studied the contribution of conformational variability to the microheterogeneity of the BR96 antibody. Heat-induced changes of secondary and tertiary structures were monitored by CD measurement and, in parallel sets of experiments, by CE and size exclusion liquid chromatography (SEC) .
M
s 0 x c
2
EXPERIMENTAL SECTION
Reagents. Sodium hydroxide 10 M solution, potassium phosphate, and sodium borate crystals were obtained from Fisher Chemical (Fair Lawn, NJ). Sodium sulfate anhydrous granular powder was purchased from J.T. Baker, Inc. (Phillipsburg, NJ), dodecyl sulfate sodium salt was from Aldrich (Milwaukee, WI), and sodium azide was from Sigma Chemical Co. (St. Louis, MO). Pretreated capillary cartridges were purchased from Beckman Instruments (Palo Alto, CA). The immunoglobulinwas chimeric antibody BR96 (Bristol-Myers Squibb, Syracuse, NY). HPLC grade water from Baxter Healthcare Corp. (McGaw Park, IL) was used in the preparation of the sample and buffer solutions. CD Measurement. The CD spectra were measured with a JASCO spectropolarimeter,Model 5-600 UASCO Intemational Co., Easton, MD). A jacketed cylindrical cuvette of 0.5 mm light path length was used for the near- and far-W range measurements. Temperature regulation was carried out by continual water flow through the cell using a Neslab circulator and thermoregulator Model RTE-110 (Neslab Instruments, Inc., Newington, NH). After each temperature adjustment and stabilization within f l "C, -5 min was allowed for thermal equilibrationbetween the circulating water and the protein solution in the cell before the first scan of the spectrum was collected. The BR96 antibody concentration in 12 mM borate buffer, pH 9.4, was 0.1 mg/mL. Four spectra, taken at a time constant of 16 s and a scanning rate of 10 nm/ min, were averaged for both samples and buffer blanks. After the buffer spectrum had been subtracted, the results were converted into the molar ellipticity using the molecular mass of the BR96 antibody, 150 000 and a total number of amino acids of 1356. Protein concentrations were measured spectrophotometrically using an absorptivity of 1.44 at 280 nm. In all experiments where sodium dodecyl sulfate (SDS) was utilized, the concentration of the detergent was 1 mg/mL, unless specified otherwise. Capillary Electrophoretic Procedures. All CE separations were performed on a Beckman P/ACE 2100 capillary electrophoresis system with a personal computer IBM PS/2 utilizing P/ACE software and Microsoft Windows interface. Pretreated capillary cartridges were purchased from Beckman Instruments. Separationswere carried out with fused-silicacapillary cartridges 57 cm in length (50 cm to detector), with an internal diameter of 75 pm (Beckman Instruments, Catalog No. 338467; rinsed by the manufacturer with 0.1 M NaOH). All electrophoretic separations were performed using a 12 mM borate buffer @H 9.4) containing 25 mM sodium dodecyl sulfate (SDS). Samples were transferred (12) Wiktorowicz, J. E.; Wilson, K. J.; Shirley, B. k In Techniques in Protein Chemistry;Villafranca, J. J., Ed.; Academic Press: New York, 1991; Chapter 31, pp 325-333. (13) Wu. S.-L.;Teshima, G.; Cacia, J.; Hancock, W. S.J. Chromatogr. 1990,5i6, 115-122. (14) Rush,R. S.; Cohen, k S.; Karger, B. L. Anal. Chem. 1991,63, 1346-1350.
2944 Analytical Chemistry, Vol. 67, No. 17, September 7, 1995
-8
u 200
d 260 Wavelength (nm)
Figure I.Temperature dependence of molar ellipticity spectra of BR96 antibody in the far-UV region. (A) Molar ellipticity spectra at (a) 20, (b) 60, (c) 70, and (d) 80 O C . (B) Molar ellipticity spectra at (a) 20, (b) 70, (c) 80, and (d) 20 "C after 70 "C and (e) 20 "C after 80 "C.
to P/ACE microvials contained in the sample holder and applied to the capillary as described below. Heat treatment of the samples was carried out with a block heater (VWR Scientiiic, Catalog No.13259-005) at 10 deg intervals. After each temperature adjustment, the BR96 antibody solution, the microvial, and a sample holder filled with water were incubated for 10 min prior to CE analysis. Upon completion of each sample analysis, the capillary was cleaned with a high-pressure rinse of 0.1 M sodium hydroxide solution for 0.5 min, followed by a high-pressure rinse of the separation buffer solution for 0.5 min. Samples were injected by a positive nitrogen pressure of 6.2 x lo5 Pa (90 psi) for 5 s. The components of both control and heat-treated BR96 antibody samples were then separated at 25 "C by a voltage of 30 kV (600V/cm). Size Fkclusion Chromatography. The SE-HPLC method used a commercially available HPLC guard column (6 mm i.d., 4 cm length) and a column of 7.8 mm id., 30 cm length, both packed with TSK gel SWXL (5 pm particle size), as obtained from TosoHaas (Montgomeryville,PA). The mobile phase consisted of 100 mM sodium sulfate, 100 mM monobasic potassium phosphate, and 0.05% sodium azide, pH 6.7. The column temperature was maintained at 23 "C, the flow rate was 1.0 mL/min., and the detection wavelength was 215 nm. A Waters 625LC with a Waters 717 autosampler, a Waters 486 variable wavelength absorbance detector, and a Nelson data collector were used for all analyses. Prior to injections, the samples of the BR96 antibody solution were treated at elevated temperatures as described in the previous section. RESULTS AND DISCUSSION CD Spectra of BR96 Antibody. The CD spectra in the far-
and near-W regions of intact and heat-treated BR96 antibody with and without SDS were examined to monitor temperature-dependent conformational transformations. As can be seen from Figures 1and 2, the molar ellipticity spectrum of the intact BR96 antibody is that of a typical immunoglobulin, with a negative band at 217 nm (representative of /3-pleated sheet structure) and several smaller positive and negative bands in the aromatic region
250
310
Wavelength (nm) Figure 2. Circular dichroism spectra of BR96 antibody in the near-UV region at elevated temperatures: (a) 20, (b) 30, (c) 40, (d) 50, (e) 60, (f) 70, and (9) 20 "C after 70 "C.
between 260 and 300 nm. In Figure Us, spectra show that the molar ellipticity spectra of BR96 antibody in the far-W region were only slightly affected by temperature changes within 20-60 "C range, indicating stability of BR96 antibody secondary structure. A simiiar pattern of structural thermal stability was observed on rat serum and monoclonal IgG2a and IgGBb, but not monoclonal IgGl or IgG2c.15 CD spectra in the near-W region Figure 2) demonstrate that BR96 antibody possesses some h e structure with a negative band at 270 nm and positive bands between 280 and 310 nm. The intensity of a negative band at 270 nm increased with temperature increases from 20 to 40 "C and then decreased with further heating from 40 to 70 "C. Changes in tertiary and secondary structures of BR96 antibody induced by heating to 70 "C were virtually reversed upon cooling to 20 "C (Figures 1B and 2). A temperature increase to 80 "C produced partly irreversible transitions, and after 30 min of incubation at 80 "C, the conformational transformation was complete and irreversible. Figure 3 shows the temperature dependence of the CD spectra of BR96 antibody in the far-W region in the presence of 1.0 mg/mL SDS. In this set of experiments SDS, a known mod8er of intramolecular interactions was used to facilitate conformational transitions and to broaden the temperature interval of the transitions. The detergent signiiicantlylowers the temperature threshold for heatinduced secondary structure transformations. Considerable changes in molar ellipticity are detected at 20 "C and every consecutive temperature increase in the range 20-70 "C induced a corresponding change in secondary structure. At 70 "C, the structural transition is practically completed, and further heating to 80 "C does not cause additional changes in the CD spectra. Capillary Electrophoresis. Micellar electrokinetic capillary chromatography has proven to be a superior approach in respect to reproducibility and separation of native proteins in comparison with other te~hniques.'~J~ Figure 4 shows the electropherogram of BR96 antibody at different temperatures. S i distinct peaks (15) Rousseaux,J.; Albert, J.-P.;Loucheux-Lefebvre, M.-H. Biochim. Biophy. Acta 1982,701, 93-101. (16) Hughes, D.E.:Richberg, P.J Chromatogr. 1993,635,313-318. (17) Strege, M.A; Lagu, A. L.Anal. Biochem. 1993,210, 402-410.
LW,,
,,,
-14
,,, ,, , ,, ,
___j
200
260 Wavelength (nm)
Figure 3. Heat-inducedchanges in molar ellipticity of BR96 antibody in the presence of 1.O mg/mL SDS.
(two of which are not completely separated) can be identified for the monoclonal BR96 antibody, and this pattern remained unchanged throughout the temperature range 20-60 "C (Figures 4A and 5). Within this temperature interval, cooling to 20 "C did not induce changes in the electropherogram pattern (data not shown). Considerable changes can be noticed at 70 "C (figure 4B), when peak 1begins to transform into peaks 2 and 4. Cooling to 20 "C and maintaining this temperature for 30 min partially reversed the heat-induced transformation and returned the ratio between isoforms close to the one that originally existed at 20 "C (Figure 5). At 80 "C, the progression of peak 1 disappearance continued further, accompanied by transformation of peak 2 into peak 4 and the appearance of an additional later-eluting peak with a retention time of 5.4 min (Figure 4C). Extending the incubation time of the samples at 80 "C to 30 min caused transition of all the species into one fraction (Figure 4D). This later-eluting fraction did not originally exist in the IgBR96 solution and appeared for Analytical Chemistry, Vol. 67, No. 17, September 1, 1995
2945
H
4
.9.
A
0
B
v
Time (min.) Electropherograms of the BR96 antibody at (A) 20, (6) 70. and (C) 80 "C and (D) incubated at 80 "Cfor 30 min. Separation buffer used was 12 mM borate buffer, 25 mM SDS, pH 9.4, with detection at 214 nm. Figure 4.
the first time on the electropherogram afker a short incubation at 80 "C. The last transformation could not be reversed by cooling the solution to 20 "C (Figure 5). To better understand the nature of the heat-induced transformations of the BR96 antibody, CE analysis of the thermally stressed protein was repeated in the presence of 1.0 mg/mL (3.5 mM) SDS in the incubation solution. As can be seen from F i r e 6, the pattern of the four major fractions of BR96 antibody remained unchanged both with and without the detergent, but the ratio of peak areas was signiicantly affected even at room temperature. SDS does not cause a formation of new IgBR96 isoforms but rather facilitates heat-induced transitions between fractions, increasing the magnitude of the changes and shifting their occurrence to lower temperatures. A peak with a retention time of 4.9 min becomes predominant at 50 "C, which is 30 "C lower than without SDS; at 70 "C, the electropherogram shows only one peak. indicating that the heat-induced transition is practically completed. Without SDS, this process requires extended incubation at 80 "C p i r e 4D). Size Exclusion Chromatography. A series of size exclusion chromatography experiments was conducted to study the p o s sibilily of BR96 antibody heat-induced aggregation in the temperature range 20-80 "C. The obtained chromatograms indicate that 2946
Analylical Chemistry, Vol. 67, No.
17, September 1, 1995
100
80
I Ill " 20
30
40
50
60
-0
IO
no
80
20
Temperalure ("C) Figum 5. Relative concentrations of the BR96 antibody isoforms separated by capillary electrophoresis at different tmq"tureS.
at 20 "C the BR96 antibody eluted as a single peak, and the same profile was maintained throughout the temperature interval 20-
B
Time (min.) Figure 6. Electropherograms of the BR96 antibody at (A) 20, (B) 30, (C) 50, and (D) 70 "C in the presence of 1.0 mg/mL SDS.
20'
A
80' 30 mm
D
Tin (min.)
Figure 7. SEC of intact (A) and heat-treated BR96 antibody by incubation at 70 (B) and 80 "C(C) and extended to 30 min incubation at 80 "C (D).
60 "C Figure 7A). After incubation at 70 "C, the antibody still eluted as a single peak but with a broader and shorter profile than at lower temperatures (Figure 7B). Incubation at 80 "C for 10 min induced formation of a new, earliereluted species which
appears on the chromatogram as a shoulder of the major peak (Figure 7C). Extending the incubation time at 80" to 30 min increased the size of the shoulder at the expense of the main component (Figure 7D). CD spectra of BR96 antibody are similar to those of IgGl and IgG2,18indicating a predominantly ,&pleated secondary structure. Comparison of the results obtained by all three analytical techniques used in this study showed that heat-induced changes in tertiary structure detected in the temperature range 20-60 "C by CD measurement in the near-UV range ( F i i e 2) had no effect on IgBR96 isoform separation pattern on the CE electropherograms or SEC profile (Figures 4 and 7A,B). This means that the structural changes reflected in the near-W region of the CD spectra that are usually associated with reorientation of the aromatic amino acid tyrosine and tryptophan side groups apparently have little or no effect on the net charge or Stokes radius of the IgBR96 molecule, the parameters determining electrophoretic mobility in CE and retention time in a SEC column. Secondary structure transitions detected by CD measurements in the farW range at 70 "C (Figure 1) were accompanied by a redistribution of the protein between CEseparated isoforms (Figure 4) and (18)Johnson, P. M.; Scopes, P. M.; Tracey, B.M.; Watkins, J. Immunology 1974, 27.27-31.
Analytical Chemistry, Vol. 67, No. 77, September 7, 7995
2947
a change in BR96 antibody peak shape, as detected by SEC. The CE results indicate that heat-induced transformations occur in two steps: in the first step, an earlier-eluting isoform with a retention time of 3.5 min partly transforms into isoforms with retention times of 4.0 and 4.9 min (Figure 4C); the second step occurs as a result of prolonged incubation at 80 "C and consists of a complete transformation of all isoforms into a single denatured form of BR96 antibody. The latter condition suggests the appearance of an earlier-eluting species on the size exclusion chromatogram, which indicates that extended incubation at 80 "C of the denatured IgBR96 is accompanied by its partial aggregation. Parallelism between structural changes (detected by CD measurements) and interconversion of BR96 antibody isofoms (separated by CE technique) were observed in the presence of SDS. The amphiphilic detergent SDS interacts with proteins by means of ionic and hydrophobic bonding, resulting in changes in intramolecularsolution dynamics,net charge, and hydrophobicity. Generally, the rate of protein inactivation correlates well with the surfactant binding. The functional activity of immunoglobulins has been reported to be lost by incubation with the detergent. SDS at concentrations of 0.1-1 mM substantially diminished the ability of sheep antiserum generated against both thyroxin and methamphetamine to bind haptens, and this two-stage process was accompanied by significant changes in secondary ~tructure.~g The effect of SDS on IgBR96 manifested itself in lowering the temperature threshold for conformational transitions (Figure 3) (19) Halfman, C. J.; Dowe, R; Jay, D. W.; Schneider, A. S. Mol. Immunol. 1986, 23, 943-949.
2948 Analytical Chemistry, Vol. 67, No. 17, September 1, 1995
and isoform interconversions (Figure 6). The changes of both characteristics can be seen at room temperature in the presence of SDS, whereas without SDS, the corresponding changes are observed at temperatures at and above 70 "C. This close parallelism in heat-induced changes between BR96 antibody secondary structure and isoform interconversions provides strong evidence of a conformational nature of differences between CE separated isoforms. Variability of the carbohydrate moiety, which is considered to be a major contributor to glycoprotein microheterogeneity, cannot explain detergent or heatinduced interconversion of BR96 antibody isoforms. The second possible source of immunoglobulin heterogeneity, partial oxidation and/or deamidation, cannot explain the facilitating effect of SDS on isoform interconversion (Figure 6) and reversibility of the first phase of the heat-induced transformations upon cooling to room temperature (Figure 5), while SEC results eliminate aggregation as a possible source of the BR96 antibody diversity at temperatures below 80 "C. The parallelism between isoform transformations and secondary structure changes allows consideration of CE-separatedisoforms of BR96 antibody as conformers, equilibrium between which can be shifted by different physicochemical factors, such as elevated temperatures and amphiphilic surfactants. Received for review March 23, 1995. Accepted June 7, 1995.8 AC9502918 Abstract published in Advance ACS Abstracts, July 15, 1995.