- Email: [email protected]
Very fast analysis of impurities in immunoglobulin concentrates using conjoint liquid chromatography on short monolithic disks
Journal of Immunological Methods 271 (2002) 47 – 58 www.elsevier.com/locate/jim
Very fast analysis of impurities in immunoglobulin concentrates using conjoint liquid chromatography on short monolithic disks K. Branovic a,b, G. Lattner b, M. Barut c, A. Strancar c, Dj. Josic b, A. Buchacher b,* b
a Institute of Immunology, Inc., Zagreb, Croatia Octapharma Pharmazeutika Produktionsges. m.b.H., Oberlaaer Strasse 235, A-1100 Vienna, Austria c BIA Separations d.o.o., Ljubljana, Slovenia
Received 8 May 2002; received in revised form 1 July 2002; accepted 14 August 2002
Abstract Transferrin and albumin are often present in immunoglobulin G (IgG) concentrates and are considered as impurities. Therefore, it is important to determine their concentration in order to obtain a well-characterized biological product. Here, we describe their determination based on conjoint liquid chromatography (CLC). The established method combines two different chromatographic modes in one step: affinity and ion-exchange chromatography (IEC) combined in one column. Therefore, two CIM Protein G and one CIM quaternary amine (QA) monolithic disks were placed in series in one housing forming a CLC monolithic column. Binding conditions were optimized in a way that immunoglobulins were captured on the CIM Protein G disks, while transferrin and albumin were bound on the CIM QA disks. Subsequently, transferrin and albumin were eluted separately by a stepwise gradient with sodium chloride, whereas immunoglobulins were released from the Protein G ligands by applying low pH. A complete separation of all three proteins was achieved in less than 5 min. The method permits the quantification of albumin and transferrin in IgG concentrates and has been successfully validated. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Monolithic column; CLC, Conjoint liquid chromatography; Immunoglobulin; Albumin; Transferrin; Quantification; Validation
1. Introduction Human immunoglobulin G (IgG) preparations, manufactured from human plasma, are used to treat patients with primary and secondary immunodeficiencies (Dwyer, 1987; Dickler and Gelfand, 1996). Conventionally, immunoglobulin G preparations have * Corresponding author. Tel.: +43-1-61032-241; fax: +43-161032-330. E-mail address: [email protected] (A. Buchacher).
been produced exclusively by precipitation techniques (Curling, 1980), preferably involving cold ethanol fractionation (Cohn et al., 1946; Kistler and Nitschmann, 1962). In recent years, however, chromatographic steps have been introduced downstream of the ethanol fractionation steps (Hoppe et al., 1973; Nourichafi et al., 1993). Another method is to purify IgG directly from plasma by combining chromatographic methods (Friesen et al., 1985). The intention is to remove traces of protein contaminants such as prekallikrein activator, which are thought to be responsible for clinical side effects. Therefore, atten-
0022-1759/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 1 7 5 9 ( 0 2 ) 0 0 3 3 9 - 3
48
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
tion has focused on the purity of intravenous IgG concentrates. Two common accompanying proteins in concentrates of human immunoglobulin are albumin and transferrin. Although their presence is not associated with any known adverse reaction, their determination is important if a well-characterized biological product is required. Usually, the concentration of albumin and transferrin in immunoglobulin solutions is in the range of 0 – 1 mg/ml, depending on the manufacturing process and on the stage of the process the sample, has been drawn. This is a very low concentration compared to the 50 –100 mg/ml of immunoglobulins present in such solutions. Conventional methods for the determination of albumin and transferrin are radial immunodiffusion (RID), laser nephelometry and enzyme-linked immunosorbent assay (ELISA). RID, the oldest method for the quantification of both proteins, is simple, but timeconsuming. It takes at least 2 days before the characteristic rings are formed by precipitation of the antigen– antibody complex (Mancini et al., 1965). Laser nephelometry also detects precipitated immunocomplexes (Sternberg, 1977) and has been automated. Furthermore, only 1 h is required for testing. However, the method suffers from a low sensitivity limit, which is 0.6 mg/ml for albumin and 0.35 mg/ml for transferrin. The ELISA procedure has become a popular method for the quantitative determination of plasma proteins (Pressler et al., 1994), and concentrations as low as 1 Ag/ml can be determined with high accuracy. However, these assays usually require several hours due to multiple incubation steps. Although column chromatographic methods, especially ion-exchange chromatography (IEC) are used for the preparative isolation of immunoglobulins (Hoppe et al., 1973; Friesen et al., 1985; Nourichafi et al., 1993), neither anion-exchange HPLC nor cation-exchange HPLC can, at present, be used effectively for the determination of albumin and transferrin in immunoglobulin concentrates. This is because of the low concentration of these accompanying proteins and also the heterogeneous nature of polyclonal immunoglobulins. Fractions of the polyclonal IgG mixture show the same behavior in IEC as albumin and transferrin and elute or bind in the same way. Protein G is a cell wall protein derived from the Streptococcus, which is able to bind IgG and other plasma proteins such as albumin. It has been engi-
neered to have a higher affinity to IgG, and the binding domain for albumin has been deleted. In both analytical and preparative IgG separations (Akerstro¨m et al., 1985), the advantage of Protein G is that the ligand binds all four subclasses of human IgG (Bjorck and Kronvall, 1984). Chromatographic separation of albumin, transferrin and IgG involves several different steps. However, such methods are time-consuming, less reproducible and result in a low yield. Online methods using two or more different columns in tandem permit automation of the separation process. A combination of ionexchange and affinity columns has already been used for analytical as well as preparative separation of proteins (Qi et al., 2001). For the analysis of immunoglobulins, a combination of ion-exchange and affinity chromatography with immobilized Protein A has been reported (Fang et al., 1998), whereas tandem columns comprising two affinity columns have been used for simultaneous determination of albumin and IgG in serum (Hage and Walter, 1987). However, none of these methods is simple or fast enough to allow in-process control of IgG production. Short monolithic disk permits very fast separation of proteins in ion exchange as well as in affinity mode (Josic et al., 2001; Strancar et al., 2002). Monolithic disk chromatography has been pioneered more than 10 years ago. The advantage of this technology is the low mass transfer resistance of the disks, which allows high separation speeds. There are many applications for IgG purification using Protein A or G disks or socalled membranes (Langlotz and Kroner, 1992; Platonova et al., 1999; Zhou et al., 1999; Zou et al., 2001). It is also possible to stack two monolithic disks with different surface properties into one housing. This type of chromatography was introduced by Strancar et al. (1998) and Josic et al. (1998) and named conjoint liquid chromatography (CLC). In this way, two-dimensional chromatography can be carried out in a single step without column switching. Recently, Gupalova et al. (2002) described a multifunctional fractionation approach to recover IgG and serum albumin (SA) from mammalian plasma within 15 min. One IgG-binding protein G disk and one SAbinding Protein G disk were installed into the same cartridge. IgG and SA were captured simultaneously, but the elution was performed in subsequent steps. The SA-binding disk was removed first, and the IgG
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
was desorbed using diluted hydrochloric acid. After reinstallation of the SA-binding disk, SA was eluted. Ostryanina et al. (2002) combined four disks with different affinity functionalities in the same cartridge, thereby permitting the separation of different antibodies within a few minutes. The major aim of this study has been to develop, validate and implement a very fast, reliable and accurate method for the determination of transferrin and human serum albumin (HSA) in concentrates of human immunoglobulins. For this purpose, a system has been used that consisted of two affinity CIM monolithic disks with immobilized Protein G and one anion-exchange CIM quaternary amine (QA) monolithic disk, all stacked in a single housing.
49
2.3. Columns
2. Materials and methods
CIM monolithic disks bearing strong (quaternary amine—CIM QA disk) anion groups or Protein G ligands (CIM Protein G disk) both from BIA Separations (Ljubljana, Slovenia) were used throughout the experimental work. A CIM disk consists of a 3 12mm-ID disk-shaped poly(glycidyl methacrylate-coethylene dimethacrylate) highly porous monolithic polymer that is seated in a nonporous self-sealing fitting ring. The bed volume of a disk is 0.34 ml, and the porosity is 62%. The conjoint liquid chromatography column was constructed by stacking one CIM QA and two CIM Protein G disks in one CIM housing, all from BIA Separations. The housing was then connected to the HPLC system so that the mobile phase first rinsed the two Protein G disks and, subsequently, the QA disk.
2.1. Chemicals
2.4. Buffers
Size-exclusion PD10 columns prepacked with Sephadex G-25, 4– 15% gradient PhastGels, PhastGel sodium dodecyl sulphate (SDS) buffer strips and protein standards were purchased from Amersham Biosciences (Uppsala, Sweden). Transferrin, semisaturated with iron, AgNO3, 5-bromo-4-chloro-3-indolylphosphate (BCIP) and nitroblue tetrazolium (NBT) were purchased from Sigma (St. Louis, MO, USA). Four to twenty percent Tris-Glycine gels, TrisGlycine SDS sample buffer, wide range protein standards and prestained standards were purchased from Novex (San Diego, CA, USA). Human serum albumin (HSA) and immunoglobulin concentrates were from the production facility or the R&D Department, Octapharma (Vienna, Austria). All other chemicals of analytical reagent grade were purchased from Merck (Darmstadt, Germany).
The binding buffer used for the binding of albumin and transferrin to a CIM QA disk and of immunoglobulins to the CIM Protein G disk was 20 mM Tris – HCl buffer, pH 7.0. Asymmetry (As) elution buffer for albumin and transferrin, the binding buffer supplemented with 1 M NaCl, was used. As elution buffer for immunoglobulins, 0.1 M Glycine-HCl, 0.1 M NaCl, pH 2.0, was used. Eluted immunoglobulins were neutralized by adding 80 Al of 1 M Tris – HCl, pH 9.0 to adjust the pH to 7.0. For regeneration, disks were washed with the following buffers: 20 column volumes of 0.1 M Tris – HCl containing 0.5 M NaCl, pH 8.0 and 20 column volumes of 0.1 M acetic acid containing 0.5 M NaCl, pH 3.5.
2.2. Instrumentation
Calibration curves were prepared using known concentrations of albumin and transferrin. For this purpose, stock solutions of transferrin and albumin were prepared by dissolving albumin ( z 99% purity) and transferrin ( z 98% purity), respectively, in the binding buffer at a concentration of 5 mg/ml. From these stock solutions defined, aliquots were added to an IgG solution containing albumin and transferrin below the limit of quantification (albumin < 0.03 mg/
A chromatography work station, consisting of two biocompatible pumps, an injection valve with a 100 Al sample loop made of polyether ether ketone, a UV detector (254 nm), a conductivity detector and a controller, from Bio-Rad (Hercules, CA, USA) was used. Peak areas were integrated using the EZLogic Chromatography Analysis Software from Bio-Rad.
2.5. Calibration standards
50
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
ml, transferrin < 0.002 mg/ml). The standard solutions were then applied to PD10 columns (Amersham Biosciences) equilibrated in the binding buffer according to the following procedure: 2.5 ml standard solution were applied to the column and, subsequently, eluted using 3.5 ml of buffer (dilution factor: 1.4) resulting in a final concentration of 25 mg/ml for IgG and of 0.05 to 1 mg/ml for albumin and transferrin. These standard solutions were then passed through a 0.22 Am filter (Millipore, Bedford, MA, USA).
reducing conditions according to Laemmli (1970) using a Phast System (Amersham Biosciences) or a Novex system (Novex). Separations were carried out for 1 h at 200 V on 4 – 15% or 4– 20% gradient gels. The amount of protein applied was 500 ng/lane for Phast System gels and 5 Ag/lane for Novex gels. Protein bands were visualised by silver staining (Heukeshoven and Dernick, 1985) or using Coomassie brilliant-blue stain from Sigma. 2.9. Radial immunodiffusion (RID)
2.6. Samples Immunoglobulin concentrates containing 50 mg/ml immunoglobulin G and between 0.1 and 2 mg/ml transferrin and albumin were used. The samples were diluted 1:1.4 (1 part sample plus 0.4 part buffer) with binding buffer and passed through PD10 columns (Amersham Biosciences) equilibrated against the same buffer which resulted in an additional dilution of the sample. The final concentration of immunoglobulins was 25 mg/ml and the concentrations of transferrin and albumin were beween 0.05 and 1 mg/ml (total dilution factor is 1.42). For spiking experiments to samples with known transferrin and albumin content, aliquots of the albumin and transferrin stock solutions were added and prepared in the same way as the other samples. Before applying the samples to the CIM disks, they were passed through a 0.22 Am filter (Millipore).
The albumin and transferrin content was determined by radial immunodiffusion kits purchased from Dade Behring (Marburg, Germany) or The Binding Site (Birmingham, UK). Each test was performed according to the manufacturer’s instructions.
3. Results and discussion
One-hundred microliter of standard solution or sample equilibrated in binding buffer were injected onto the stack of disks equilibrated with binding buffer. Then a two-step gradient comprising an initial step of either 0.08 or 0.1 M followed by a second step of either 0.3 or 0.4 M NaCl, respectively, was applied at a flow rate of 4 ml/min to elute bound proteins from the ion-exchange disk. Glycine buffer containing 100 mM NaCl was applied to release bound proteins from the Protein G ligands.
For a well-characterized therapeutic product, it is important to know the concentration of each impurity. Albumin and transferrin are typical contaminating proteins in immunoglobulin concentrates. Their quantification is possible by different analytical methods such as RID, laser nephelometry and ELISA. These methods do not permit simultaneous determination of all proteins. In addition, there is no test system that is both fast and sensitive. Here, a method is described, which requires only 15 min for sample preparation and less than 5 min for separation. The conjoint liquid chromatography system was designed in a way that the immunoglobulin is captured by a Protein G monolithic disk, while impurities are bound by an anion-exchange monolithic disk placed in sequence. Then impurities are resolved by a salt gradient and immunoglobulins eluted with an acidic buffer. The chromatographic conditions for a successful separation of the impurities from the immunoglobulin were first optimized on the individual disks and then on the combination of the disks as described below.
2.8. Electrophoretic technique
3.1. Optimization of chromatographic conditions
Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out under non-
Protein affinity chromatography using CIM Protein G was chosen to separate immunoglobulins from
2.7. Chromatographic procedure
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
51
Fig. 1. Separation of immunoglobulin G from accompanying proteins on two CIM Protein G monolithic disks. Chromatograms of two different immunoglobulin concentrates (IG-1 and -2) are superimposed. Binding buffer: 20 mM Tris – HCl pH 7.0; elution buffer: 0.1 M Glycine – HCl, 0.1 M NaCl, pH 2.0; flow rate: 4 ml/min; sample concentration: 2 mg IgG/100 Al. (FT = flow through; P = eluted peak).
albumin and transferrin. Fig. 1 illustrates the separation of IgG concentrates with differing levels of impurity. The flow through (FT) and the eluate were
Fig. 2. SDS – PAGE under nonreducing conditions using 4 – 15% gradient gels. Samples were collected during the separation of two IgG concentrates (IG-1 and -2) on two CIM Protein G disks (see Fig. 1). Five hundred nanogram of protein were applied to each lane. S1 = starting material of IG-1, FT1 = flow through of IG-1, P = eluted peak of IG-1; S2 = starting material of IG-2, FT2 = flow through of IG-2, P2 = eluted peak of IG-2 (A = albumin; T = transferrin; Ig = IgG).
further analyzed by SDS-PAGE (see Fig. 2, lanes FT1 and FT2)). The main impurities of these concentrates, albumin and transferrin, could be quantitatively removed. Neither coeluted with IgG. On the other hand, immunoglobulins were completely eluted by a buffer with low pH because other accompanying proteins could not be detected by SDS-PAGE combined with silver staining; a highly sensitive method for the detection of proteins (Fig. 2, lanes P1 and P2). Next, the chromatographic stability of Protein G disks was determined by measuring the dynamic binding capacity for immunoglobulins after repeated sample applications. Regeneration of the disk was carried out after each sample application, and the dynamic binding capacity determined after 50 or 100 cycles. The dynamic binding capacity for immunoglobulins on the CIM Protein G disks was calculated at 10% breakthrough. After 100 cycles, the Table 1 Dynamic binding capacity of IgG on CIM Protein G disks (regeneration after each run was performed) Number of injections
Dynamic binding capacity Disk #1 [mg/ml]
Disk #2 [mg/ml]
1 50 100
14.7 14.5 11.7
13.1 12.0 11.7
52
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Fig. 3. Separation of transferrin and albumin on a CIM QA monolithic disk. Binding buffer: 20 mM Tris – HCl, pH 7.0; elution buffer: binding buffer containing 1 M NaCl; flow rate: 4 ml/min, sample concentration: 45 Ag transferrin and 45 Ag albumin/250 Al (FT = flow through; 1 = peak eluted with 0.08 M NaCl; 2 = peak eluted with 0.3 M NaCl).
monolithic disk still had the same resolution power, and only 10 –20% dynamic binding capacity was lost. As shown in Table 1, capacity slightly decreased with the number of cycles. After 50 applications, the capacity remained constant for disk no. 1, and for disk no. 2, it was only 10% lower. After 100 runs, the capacity of disk no. 1 was 20% lower and that of disk no. 2 was 10% lower. A strong anion-exchange CIM QA disk was selected to separate transferrin from albumin. In preliminary experiments, an attempt was made to separate immunoglobulins from their protein impurities by ion-exchange chromatography on monolithic disks. Due to the very broad range of isoelectric points of immunoglobulins (pI from 4.35 to 9.95; Prin et al., 1995), there was an overlap with the isoelectric points of albumin and transferrin, and complete resolution was not possible. Albumin and transferrin were bound to the CIM anionic disks, whereas the main fraction of the immunoglobulins passed through the ion exchanger. However, some immunoglobulins also bound to the anion-exchange disk and were coeluted with albumin and transferrin. A similar situation was observed with separations performed using the cationexchange mode. Most immunoglobulins bound to the ion exchanger, while a small fraction passed through the column together with transferrin and albumin (data not shown). Therefore, an affinity chromatog-
raphy step on a CIM Protein G disk was introduced, separating immunoglobulins from albumin and transferrin.
Fig. 4. SDS – PAGE of fractions collected during separation of transferrin and albumin on a CIM QA column. (see Fig. 3). A 4 – 15% gradient gel was used. Five hundred nanogram protein were applied to each lane. S = starting material (nonbound fraction derived from affinity chromatography on CIM Protein G column), FT = flow through, 1 = fraction eluted with 0.08 M NaCl, 2 = fraction eluted with 0.3 M NaCl, A = albumin; T = transferrin).
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
53
Fig. 5. Optimized separation of transferrin, albumin and IgG on CIM CLC monolithic column consisting of two CIM Protein G and one CIM QA disks. Chromatograms of two different runs using different step gradients are superimposed. Binding buffer: 20 mM Tris – HCl, pH 7.0; elution buffer I: 20 mM Tris – HCl, 1 M NaCl, pH 7.0; elution buffer II: 0.1 M Glycine – HCl, 0.1 M NaCl, pH 2.0; flow rate: 4 ml/min; sample concentration: 2 mg IgG/100 Al (‘‘1’’ indicates the transferrin peak eluted with 0.08 or 0.1 M NaCl; ‘‘2’’ indicates the albumin peak eluted with 0.3 or 0.4 M NaCl; ‘‘3’’ indicates the IgG peak eluted with low pH).
Fig. 6. SDS – PAGE under reducing and nonreducing conditions. A gradient gel of 4 – 20% was used. Samples were collected during the separation of an IgG concentrate on CIM CLC monolithic column (see Fig. 5). S = starting material, 1 = chromatographic peak No. 1, 2 = chromatographic peak No. 2, 3 = peak No. 3, A = albumin; T = transferrin; Ig = IgG).
54
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Table 2 Linearity of the standard curve of transferrin and albumin Transferrin [mg/ml]
Peak area for transferrin [mAU * ml]
Albumin [mg/ml]
Peak area for albumin [mAU * ml]
0.05 0.20 0.40 0.60 0.80 1.00 r = 0.9942 sy = 18.271 a = 417.83 b = 34.117
46 396 131 682 198 316 295 295 341143 466 259
0.05 0.20 0.40 0.60 0.80 1.00 r = 0.9959 sy = 9.9832 a = 273.75 b = 12.452
33 117 59 869 117 186 174 629 246 077 278 766
r: coefficient of correlation. sy: standard deviation of the response. a: slope. b: intercept.
The flow-through fraction from the Protein G affinity chromatography contained only transferrin and albumin because all immunoglobulins were captured by the Protein G disks (see Fig. 1). A total volume of 250 Al were loaded on one CIM QA disk. A linear salt gradient was applied to find the elution position of albumin and transferrin. For optimized resolution, a step gradient was investigated (Fig. 3). The highest resolution was obtained by eluting transferrin using a step gradient with 0.08 M
NaCl and albumin using a step gradient with 0.3 M NaCl. Under such conditions, albumin and transferrin are separated as demonstrated by SDS-PAGE with silver staining (Fig. 4). After completion of these optimization runs, both methods—affinity chromatography and anionexchange chromatography—were coupled together by stacking the disks into one housing. The concentration of IgG compared to transferrin and albumin was, in some samples, in the ratio of 1:500. In order to improve the limit of quantification for transferrin and albumin, two Protein G disks had to be combined with one ion-exchange disk. More sample volume could be injected without a breakthrough of IgG. Each Protein G disk had the ability to bind 5 mg IgG. Using two CIM Protein G disks and one CIM QA disk, a complete separation of immunoglobulins, albumin and transferrin could be achieved in one step in less than 5 min. Immunoglobulins were captured on the CIM Protein G disks, while albumin and transferrin were bound on a CIM QA disk. Albumin and transferrin were eluted by a salt gradient, while IgG was removed by lowering the pH of the eluting buffer. Tailing of the peaks was reduced by increasing the salt concentration of each gradient step (Fig. 5). Transferrin was eluted at 0.1 M NaCl instead of 0.08 M, and albumin was eluted at 0.4 M NaCl instead of 0.3 M. Immunoglobulins were eluted by decreasing the
Fig. 7. Linearity of the separation of transferrin, albumin from IgG on CIM CLC monolithic column. Chromatograms of six different runs with sample concentrations from 0.05 to 1.00 mg/ml albumin or transferrin, respectively, are superimposed. Binding buffer: 20 mM Tris – HCl, pH 7.0; elution buffer I: 20 mM Tris – HCl, 1 M NaCl, pH 7.0; elution buffer II: 0.1 M Glycine – HCl, 0.1 M NaCl, pH 2.0; flow rate: 4 ml/min; sample concentration: see standard preparation in the Materials and methods section.
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
55
Fig. 8. Linearity of the calibration function: regression lines of the separation of transferrin, albumin from IgG on a CIM monolithic column. The peak areas used for regression analysis are shown in Table 2.
pH value to 2.0 supplemented with 100 mM sodium chloride. Eluates were immediately neutralized with 80 Al of 1 M Tris –HCl, pH 9.0. Separated fractions were collected and analyzed by SDS-PAGE (Fig. 6). The IgG concentrate (lane S) containing all three proteins (immunoglobulins, transferrin and albumin) revealed bands with apparent molecular masses of 200, 75 and 66 kDa, respectively. In the first peak, only transferrin was identified (lane 1). In the second peak, only albumin was detected (lane 2) and only immunoglobulins in the last peak (lane 3). 3.2. Method validation Increasing amounts of transferrin and albumin standards were added to the pure immunoglobulin sample (containing 25 mg IgG/ml) to give concentrations of 0.05, 0.2, 0.4, 0.6, 0.8 and 1.0 mg/ml, respectively, before application to the monolithic disks combined in one housing. Peak areas were plotted against the amount of injected material and regression analysis was performed. First-order linear regression was obtained for transferrin and albumin with correlation coefficients of 0.994 and 0.996, respectively (Table 2). Residuals were randomly distributed. At the 95% level, it was established that the confidence interval of the intercept included ‘‘0.’’ Thus, it can be concluded that the
method was performed without constant systematic error. Chromatograms and regression lines are presented in Figs. 7 and 8, respectively. For confirmation of the lower limit of the test range, the limit of quantification was calculated according to the ICH guidelines using a specific calibration curve. Therefore, a standard curve ranging from 0.07 to 0.21 mg/ml albumin and transferrin, respectively, was prepared (Table 3). In this case,
Table 3 Determination of limit of quantification Transferrin [mg/ml]
Peak area for transferrin [mAU * ml]
0.071 60.106 0.100 72.841 0.129 85.634 0.143 98.031 0.157 105.036 0.186 114.844 0.214 128.342 r = 0.996 a = 486.03 sy = 2.46 LOQ = 0.05 mg/ml
Albumin [mg/ml]
Peak area for albumin [mAU * ml]
0.071 18.747 0.100 25.586 0.129 31.298 0.143 34.636 0.157 37.682 0.186 44.187 0.214 47.700 r = 0.997 a = 206.81 sy = 0.83 LOQ = 0.04 mg/ml
r: coefficient of correlation. a: slope. sy: standard deviation of the response. LOQ: limit of quantification = 10 sy/a.
56
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Table 4a Repeatability and intermediate precision for transferrin
Table 5a Recovery rate of transferrin
Transferrin Day Concentration [mg/ml]
Xim STD RSD [%]
Transferrin 1 0.18 0.19 0.19 0.18 0.19 0.186 0.0055 2.9
2
3
0.16 0.18 0.19 0.18 0.18 0.177 0.0104 5.8
0.15 0.14 0.15 0.16 0.15 0.150 0.0071 4.7
Measured
0.171 0.0175 10.2
Spike 2
Spike 3
the correlation coefficient for transferrin was 0.996 and that for albumin 0.997. The limit of quantification for transferrin was 0.05 mg/ml and for albumin 0.04 mg/ml as calculated by multiplying the standard deviation of the response by a factor of 10 and dividing by the slope of the calibration curve.
Table 4b Repeatability and intermediate precision for albumin Albumin
Concentration [mg/ml]
Xim STD RSD [%]
1 0.10 0.09 0.10 0.09 0.10 0.096 0.0052 5.4
0.089 0.0090 10.2
Xim: mean value per day. Xm: total mean value. STD: standard deviation. RSD [%]: relative standard deviation.
0.270 0.270 0.270 0.570 0.570 0.570 0.970 0.970 0.970
107 117 119 98 101 100 102 103 104
[%]
2
3
0.09 0.09 0.09 0.10 0.09 0.092 0.0045 4.9
0.07 0.08 0.08 0.08 0.08 0.078 0.0045 5.7
The precision of the method was tested by application of a sample, containing 0.17 mg/ml transferrin and 0.09 mg/ml albumin five times on three different days. The relative standard deviation (RSD) per day ranged from 2.9% to 5.8% for transferrin and from 4.9% to 5.71% for albumin. The intermediate precision was 10.2% for both transferrin and for albumin (Table 4a and 4b). The recovery for spiked samples was determined. An immunoglobulin sample, containing 0.17 mg/ml transferrin and 0.09 mg/ml albumin, was spiked with 0.1, 0.4 and 0.8 mg/ml albumin and transferrin, respectively. Three runs with each solution were performed; the recovery for transferrin ranged from 98% to 119% whereas that for albumin ranged from 82% to 101% (Table 5a and 5b). Table 5b Recovery rate of albumin Albumin Measured
Calculated
[g/l] Spike 1
Intermediate precision Xm STD RSD [%]
0.290 0.317 0.322 0.557 0.574 0.569 0.987 1.003 1.009
measured concentration ½mg=ml RR: Recovery Rate ¼ calculated concentration ½mg=ml 100%.
Xim: mean value per day. Xm: total mean value. STD: standard deviation. RSD [%]: relative standard deviation.
Day
RR
[g/l] Spike 1
Intermediate precision Xm STD RSD [%]
Calculated
Spike 2
Spike 3
0.169 0.191 0.178 0.483 0.482 0.484 0.743 0.814 0.729
RR [%]
0.189 0.189 0.189 0.489 0.489 0.489 0.889 0.889 0.889
measured concentration ½mg=ml RR: Recovery Rate ¼ calculated concentration ½mg=ml 100%.
89 101 94 99 99 99 84 92 82
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58 Table 6 Comparison of CLC with RID Sample
Transferrin [mg/ml]
No
RID
1 2 3 4 5 6
0.36 0.40 0.14 0.09 0.11 0.12 0.36 0.37 0.14 0.19 0.12 0.16 Mean D STD D TV t (5, 95%)
CLC
Albumin [mg/ml]
D
RID
0.04 0.05 0.00 0.01 0.05 0.04 0.02 0.0367 1.11 2.571
0.52 0.52 0.06 0.05 0.09 0.11 0.70 0.70 0.03 0.08 0.03 0.05 Mean D STD D TV t (5, 95%)
CLC
D 0.00 0.01 0.02 0.00 0.05 0.02 0.01 0.0216 1.51 2.571
Mean D: mean value of D. STD D: standard deviationpof ffi D. DA n TV: test value ¼ Amean . STD D t: significance value of the student’s distribution.
Immunoglobulin concentrates (# IGs 1 –6) were analyzed by the conjoint liquid chromatography method and by RID, which is accepted as one of the standard methods for the determination of transferrin and albumin. The contents of transferrin and albumin obtained from integration of the respective peaks were compared with the results obtained from RID (see Table 6). A difference t-test at the 95% level confirmed that there was no statistically significant difference between the methods. Preliminary experiments showed that peaks with shoulders and tails were observed, when monolithic disks were damaged. Similar observation was also reported by Hahn and Jungbauer (2001) for some prototype disks. Asymmetry (As) of the transferrin peak was chosen as a criterion for system suitability. Asymmetry was calculated as the ratio (at 10% height) of the first half peak width to the second half peak width. As acceptance criteria of 0.8 and 1.2 were set. If As were out of this range, the run was repeated. If this resulted again in an asymmetric peak, new disks were taken. The special geometry of monolithic disks permitted a very fast separation with low back pressure. The stacking of monolithic disks with different functional groups avoided column switching and made the method faster and simpler. The separation of contaminating proteins, particularly albumin and transferrin, and monoclonal antibodies from mouse ascites fluid in a tandem arrangement consisting of
57
CIM anion-exchange monolithic disk and a CIM Protein A monolithic disk has been described previously (Josic et al., 1998). These authors demonstrated that the separation of IgG from accompanying proteins was possible within only 3 min. However, no quantitative determinations were performed. Strancar et al. (1998) separated IgG from human plasma in 2 min by combining a CIM DEAE and a CIM Protein G monolithic disk in one housing, but, again, no quantification of the isolated protein was carried out. Gupalova et al. (2002) demonstrated that IgG and serum albumin (SA) from mammalian plasma can be separated within 15 min by two disks with different affinity functionalities. The elution had to be performed discontinuously because the IgG-binding disk as well as the SA-binding disk have to be treated with very acidic solutions in order to achieve desorption of the target proteins. However, the goal of this work was not a quantitative determination, but preparative work for quantitatively capturing of both proteins. The present studies demonstrate that conjoint liquid chromatography can be used as a precise and accurate analytical method, which could be validated for pharmaceutical purposes.
Acknowledgements This work was supported by the Forschungsfo¨rderungsfonds, Austria, project no. 803172.
References Akerstro¨m, B., Brodin, T., Reis, K., Bjo¨rck, L., 1985. Protein G: a powerful tool for binding and detection of monoclonal and polyclonal antibodies. J. Immunol. 135, 2589. Bjorck, L., Kronvall, G., 1984. Purification and some properties of streptococcal Protein G, a novel IgG-binding reagent. J. Immunol. 133, 969. Cohn, E., Strong, L., Hughes, W., Mulford, D., Ashworth, J., Melin, M., Taylor, H., 1946. Preparation and properties of serum and plasma proteins: IV. A system for the separation into fractions of the protein and lipoprotein components of biological tissues and fluids. J. Am. Chem. Soc. 68, 459. Curling, J.M., 1980. Methods of Plasma Protein Fractionation. Academic Press, London, UK. Dickler, H., Gelfand, E., 1996. Current perspectives on the use of intravenous immunoglobulin. Adv. Intern. Med. 41, 641.
58
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Dwyer, J., 1987. Intravenous therapy with gamma globulin. Adv. Intern. Med. 32, 111. Fang, Y., Dumont, L., Larsen, B., 1998. Real-time isoform analysis by two-dimensional chromatography of a monoclonal antibody during bioreactor fermentations. J. Chromatogr., A 816, 39. Friesen, A., Bowman, J., Bees, W., 1985. Column ion exchange chromatographic production of human immune serum globulin for intravenous use. Vox Sang. 48, 201. Gupalova, T.V., Lojkina, O.V., Pa`la`gnuk, V.G., Totolian, A.A., Tennikova, T.B., 2002. Quantitative investigation of the affinity properties of different recombinant forms of Protein G by means of high-performance monolithic chromatography. J. Chromatogr., A 949, 185. Hage, D., Walter, R., 1987. Dual-column determination of albumin and immunoglobulin G in serum by high-performance affinity chromatography. J. Chromatogr. 386, 37. Hahn, R., Jungbauer, A., 2001. Control method for integrity of continuous beds. J. Chromatogr., A 908, 179. Heukeshoven, J., Dernick, R., 1985. Simplified method for silver staining of proteins in polyacrylamide gels and the mechanism of silver staining. Electrophoresis 6, 103. Hoppe, H., Mester, T., Hennig, W., Krebs, H., 1973. Prevention of Rh-immunization. Modified production of IgG anti-Rh for intravenous application by ion exchange chromatography (IEC). Vox Sang. 25, 308. Josic, D., Schwinn, H., Strancar, A., Podgornik, A., Barut, M., Lim, Y.-P., Vodopivec, M., 1998. Use of compact, porous units with immobilized ligands with high molecular masses in affinity chromatography and enzymatic conversion of substrates with high and low molecular masses. J. Chromatogr., A 803, 61. Josic, D., Buchacher, A., Jungbauer, A., 2001. Monoliths as stationary phases for separation of proteins and polynucleotides and enzymatic conversion. J. Chromatogr., B 752, 191. Kistler, P., Nitschmann, H., 1962. Large scale production of human plasma fractions. Vox Sang. 7, 414. Laemmli, U., 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680. Langlotz, P., Kroner, K.H., 1992. Surface-modified membranes as a matrix for protein purification. J. Chromatogr., A 591, 107. Mancini, G., Carbonara, A., Heremans, J., 1965. Immunochemical quantitation of antigens by single radial immunodiffusion. Immunochemistry 2, 235.
Nourichafi, N., Geschier, C., Stoltz, J.-F., 1993. Comparison of various chromatographic supports for purifying human plasmatic immunoglobulins from Cohn II + III fraction. Biotechnol. Blood Proteins 227, 207. Ostryanina, N.D., Vlasov, G.P., Tennikova, T.B., 2002. Multifunctional fractionation of polyclonal antibodies by immunoaffinity high-performance monolithic disk chromatography. J. Chromatogr. 949, 163. Platonova, G.A., Pankova, G.A., Il’ina, I.Ye., Vlasov, G.P., Tennikova, T.B., 1999. Quantitative fast fractionation of a pool of polyclonal antibodies by immunoaffinity membrane chromatography. J. Chromatogr., A 852, 129. Pressler, T., Mansa, B., Pedersen, S., Espersen, F., Hoiby, N., Koch, C., 1994. Methodologic problems in establishing normal values for IgG subclass concentrations in a pediatric population; comparison of radial immunodiffusion and ELISA methods. Allergy 49, 772. Prin, C., Bene, M., Gobert, B., Montagne, P., Faure, G., 1995. Isoelectric restriction of human immunoglobulin isotypes. Biochim. Biophys. Acta 1243, 287. Qi, Y., Zan, Z., Huang, J., 2001. Chromatography on DEAE ionexchange and Protein G affinity columns in tandem for the separation and purification of proteins. J. Biochem. Biophys. Methods 49, 263. Sternberg, J., 1977. A rate nephelometer for measuring specific proteins by immunoprecipitin reactions. Clin. Chem. 23, 1456. Strancar, A., Barut, M., Podgornik, A., Koselj, P., Josic, D., Buchacher, A., 1998. Convective Interaction Media (CIM): polymer based supports for fast separation of biomolecules. LC GC Int. 10, 660. Strancar, A., Podgornik, A., Barut, M., Necina, R., 2002. Short monolithic columns as stationary phases for biochromatography. In: Freitag, R. (Ed.), Modern Advances in Chromatography: Advances in Biochemical Engineering/Biotechnology, vol. 76. Springer-Verlag, Heidelberg, Germany, p. 50. Zhou, D., Zou, H., Ni, J., Yang, L., Jia, L., Zhang, Q., Zhang, Y., 1999. Membrane supports as the stationary phase in highperformance immunoaffinity chromatography. Anal. Chem. 71, 115. Zou, H., Luo, Q., Zhou, D., 2001. Affinity membrane chromatography for the analysis and purification of proteins. J. Biochem. Biophys. Methods 49, 199.
Very fast analysis of impurities in immunoglobulin concentrates using conjoint liquid chromatography on short monolithic disks K. Branovic a,b, G. Lattner b, M. Barut c, A. Strancar c, Dj. Josic b, A. Buchacher b,* b
a Institute of Immunology, Inc., Zagreb, Croatia Octapharma Pharmazeutika Produktionsges. m.b.H., Oberlaaer Strasse 235, A-1100 Vienna, Austria c BIA Separations d.o.o., Ljubljana, Slovenia
Received 8 May 2002; received in revised form 1 July 2002; accepted 14 August 2002
Abstract Transferrin and albumin are often present in immunoglobulin G (IgG) concentrates and are considered as impurities. Therefore, it is important to determine their concentration in order to obtain a well-characterized biological product. Here, we describe their determination based on conjoint liquid chromatography (CLC). The established method combines two different chromatographic modes in one step: affinity and ion-exchange chromatography (IEC) combined in one column. Therefore, two CIM Protein G and one CIM quaternary amine (QA) monolithic disks were placed in series in one housing forming a CLC monolithic column. Binding conditions were optimized in a way that immunoglobulins were captured on the CIM Protein G disks, while transferrin and albumin were bound on the CIM QA disks. Subsequently, transferrin and albumin were eluted separately by a stepwise gradient with sodium chloride, whereas immunoglobulins were released from the Protein G ligands by applying low pH. A complete separation of all three proteins was achieved in less than 5 min. The method permits the quantification of albumin and transferrin in IgG concentrates and has been successfully validated. D 2002 Elsevier Science B.V. All rights reserved. Keywords: Monolithic column; CLC, Conjoint liquid chromatography; Immunoglobulin; Albumin; Transferrin; Quantification; Validation
1. Introduction Human immunoglobulin G (IgG) preparations, manufactured from human plasma, are used to treat patients with primary and secondary immunodeficiencies (Dwyer, 1987; Dickler and Gelfand, 1996). Conventionally, immunoglobulin G preparations have * Corresponding author. Tel.: +43-1-61032-241; fax: +43-161032-330. E-mail address: [email protected] (A. Buchacher).
been produced exclusively by precipitation techniques (Curling, 1980), preferably involving cold ethanol fractionation (Cohn et al., 1946; Kistler and Nitschmann, 1962). In recent years, however, chromatographic steps have been introduced downstream of the ethanol fractionation steps (Hoppe et al., 1973; Nourichafi et al., 1993). Another method is to purify IgG directly from plasma by combining chromatographic methods (Friesen et al., 1985). The intention is to remove traces of protein contaminants such as prekallikrein activator, which are thought to be responsible for clinical side effects. Therefore, atten-
0022-1759/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 1 7 5 9 ( 0 2 ) 0 0 3 3 9 - 3
48
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
tion has focused on the purity of intravenous IgG concentrates. Two common accompanying proteins in concentrates of human immunoglobulin are albumin and transferrin. Although their presence is not associated with any known adverse reaction, their determination is important if a well-characterized biological product is required. Usually, the concentration of albumin and transferrin in immunoglobulin solutions is in the range of 0 – 1 mg/ml, depending on the manufacturing process and on the stage of the process the sample, has been drawn. This is a very low concentration compared to the 50 –100 mg/ml of immunoglobulins present in such solutions. Conventional methods for the determination of albumin and transferrin are radial immunodiffusion (RID), laser nephelometry and enzyme-linked immunosorbent assay (ELISA). RID, the oldest method for the quantification of both proteins, is simple, but timeconsuming. It takes at least 2 days before the characteristic rings are formed by precipitation of the antigen– antibody complex (Mancini et al., 1965). Laser nephelometry also detects precipitated immunocomplexes (Sternberg, 1977) and has been automated. Furthermore, only 1 h is required for testing. However, the method suffers from a low sensitivity limit, which is 0.6 mg/ml for albumin and 0.35 mg/ml for transferrin. The ELISA procedure has become a popular method for the quantitative determination of plasma proteins (Pressler et al., 1994), and concentrations as low as 1 Ag/ml can be determined with high accuracy. However, these assays usually require several hours due to multiple incubation steps. Although column chromatographic methods, especially ion-exchange chromatography (IEC) are used for the preparative isolation of immunoglobulins (Hoppe et al., 1973; Friesen et al., 1985; Nourichafi et al., 1993), neither anion-exchange HPLC nor cation-exchange HPLC can, at present, be used effectively for the determination of albumin and transferrin in immunoglobulin concentrates. This is because of the low concentration of these accompanying proteins and also the heterogeneous nature of polyclonal immunoglobulins. Fractions of the polyclonal IgG mixture show the same behavior in IEC as albumin and transferrin and elute or bind in the same way. Protein G is a cell wall protein derived from the Streptococcus, which is able to bind IgG and other plasma proteins such as albumin. It has been engi-
neered to have a higher affinity to IgG, and the binding domain for albumin has been deleted. In both analytical and preparative IgG separations (Akerstro¨m et al., 1985), the advantage of Protein G is that the ligand binds all four subclasses of human IgG (Bjorck and Kronvall, 1984). Chromatographic separation of albumin, transferrin and IgG involves several different steps. However, such methods are time-consuming, less reproducible and result in a low yield. Online methods using two or more different columns in tandem permit automation of the separation process. A combination of ionexchange and affinity columns has already been used for analytical as well as preparative separation of proteins (Qi et al., 2001). For the analysis of immunoglobulins, a combination of ion-exchange and affinity chromatography with immobilized Protein A has been reported (Fang et al., 1998), whereas tandem columns comprising two affinity columns have been used for simultaneous determination of albumin and IgG in serum (Hage and Walter, 1987). However, none of these methods is simple or fast enough to allow in-process control of IgG production. Short monolithic disk permits very fast separation of proteins in ion exchange as well as in affinity mode (Josic et al., 2001; Strancar et al., 2002). Monolithic disk chromatography has been pioneered more than 10 years ago. The advantage of this technology is the low mass transfer resistance of the disks, which allows high separation speeds. There are many applications for IgG purification using Protein A or G disks or socalled membranes (Langlotz and Kroner, 1992; Platonova et al., 1999; Zhou et al., 1999; Zou et al., 2001). It is also possible to stack two monolithic disks with different surface properties into one housing. This type of chromatography was introduced by Strancar et al. (1998) and Josic et al. (1998) and named conjoint liquid chromatography (CLC). In this way, two-dimensional chromatography can be carried out in a single step without column switching. Recently, Gupalova et al. (2002) described a multifunctional fractionation approach to recover IgG and serum albumin (SA) from mammalian plasma within 15 min. One IgG-binding protein G disk and one SAbinding Protein G disk were installed into the same cartridge. IgG and SA were captured simultaneously, but the elution was performed in subsequent steps. The SA-binding disk was removed first, and the IgG
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
was desorbed using diluted hydrochloric acid. After reinstallation of the SA-binding disk, SA was eluted. Ostryanina et al. (2002) combined four disks with different affinity functionalities in the same cartridge, thereby permitting the separation of different antibodies within a few minutes. The major aim of this study has been to develop, validate and implement a very fast, reliable and accurate method for the determination of transferrin and human serum albumin (HSA) in concentrates of human immunoglobulins. For this purpose, a system has been used that consisted of two affinity CIM monolithic disks with immobilized Protein G and one anion-exchange CIM quaternary amine (QA) monolithic disk, all stacked in a single housing.
49
2.3. Columns
2. Materials and methods
CIM monolithic disks bearing strong (quaternary amine—CIM QA disk) anion groups or Protein G ligands (CIM Protein G disk) both from BIA Separations (Ljubljana, Slovenia) were used throughout the experimental work. A CIM disk consists of a 3 12mm-ID disk-shaped poly(glycidyl methacrylate-coethylene dimethacrylate) highly porous monolithic polymer that is seated in a nonporous self-sealing fitting ring. The bed volume of a disk is 0.34 ml, and the porosity is 62%. The conjoint liquid chromatography column was constructed by stacking one CIM QA and two CIM Protein G disks in one CIM housing, all from BIA Separations. The housing was then connected to the HPLC system so that the mobile phase first rinsed the two Protein G disks and, subsequently, the QA disk.
2.1. Chemicals
2.4. Buffers
Size-exclusion PD10 columns prepacked with Sephadex G-25, 4– 15% gradient PhastGels, PhastGel sodium dodecyl sulphate (SDS) buffer strips and protein standards were purchased from Amersham Biosciences (Uppsala, Sweden). Transferrin, semisaturated with iron, AgNO3, 5-bromo-4-chloro-3-indolylphosphate (BCIP) and nitroblue tetrazolium (NBT) were purchased from Sigma (St. Louis, MO, USA). Four to twenty percent Tris-Glycine gels, TrisGlycine SDS sample buffer, wide range protein standards and prestained standards were purchased from Novex (San Diego, CA, USA). Human serum albumin (HSA) and immunoglobulin concentrates were from the production facility or the R&D Department, Octapharma (Vienna, Austria). All other chemicals of analytical reagent grade were purchased from Merck (Darmstadt, Germany).
The binding buffer used for the binding of albumin and transferrin to a CIM QA disk and of immunoglobulins to the CIM Protein G disk was 20 mM Tris – HCl buffer, pH 7.0. Asymmetry (As) elution buffer for albumin and transferrin, the binding buffer supplemented with 1 M NaCl, was used. As elution buffer for immunoglobulins, 0.1 M Glycine-HCl, 0.1 M NaCl, pH 2.0, was used. Eluted immunoglobulins were neutralized by adding 80 Al of 1 M Tris – HCl, pH 9.0 to adjust the pH to 7.0. For regeneration, disks were washed with the following buffers: 20 column volumes of 0.1 M Tris – HCl containing 0.5 M NaCl, pH 8.0 and 20 column volumes of 0.1 M acetic acid containing 0.5 M NaCl, pH 3.5.
2.2. Instrumentation
Calibration curves were prepared using known concentrations of albumin and transferrin. For this purpose, stock solutions of transferrin and albumin were prepared by dissolving albumin ( z 99% purity) and transferrin ( z 98% purity), respectively, in the binding buffer at a concentration of 5 mg/ml. From these stock solutions defined, aliquots were added to an IgG solution containing albumin and transferrin below the limit of quantification (albumin < 0.03 mg/
A chromatography work station, consisting of two biocompatible pumps, an injection valve with a 100 Al sample loop made of polyether ether ketone, a UV detector (254 nm), a conductivity detector and a controller, from Bio-Rad (Hercules, CA, USA) was used. Peak areas were integrated using the EZLogic Chromatography Analysis Software from Bio-Rad.
2.5. Calibration standards
50
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
ml, transferrin < 0.002 mg/ml). The standard solutions were then applied to PD10 columns (Amersham Biosciences) equilibrated in the binding buffer according to the following procedure: 2.5 ml standard solution were applied to the column and, subsequently, eluted using 3.5 ml of buffer (dilution factor: 1.4) resulting in a final concentration of 25 mg/ml for IgG and of 0.05 to 1 mg/ml for albumin and transferrin. These standard solutions were then passed through a 0.22 Am filter (Millipore, Bedford, MA, USA).
reducing conditions according to Laemmli (1970) using a Phast System (Amersham Biosciences) or a Novex system (Novex). Separations were carried out for 1 h at 200 V on 4 – 15% or 4– 20% gradient gels. The amount of protein applied was 500 ng/lane for Phast System gels and 5 Ag/lane for Novex gels. Protein bands were visualised by silver staining (Heukeshoven and Dernick, 1985) or using Coomassie brilliant-blue stain from Sigma. 2.9. Radial immunodiffusion (RID)
2.6. Samples Immunoglobulin concentrates containing 50 mg/ml immunoglobulin G and between 0.1 and 2 mg/ml transferrin and albumin were used. The samples were diluted 1:1.4 (1 part sample plus 0.4 part buffer) with binding buffer and passed through PD10 columns (Amersham Biosciences) equilibrated against the same buffer which resulted in an additional dilution of the sample. The final concentration of immunoglobulins was 25 mg/ml and the concentrations of transferrin and albumin were beween 0.05 and 1 mg/ml (total dilution factor is 1.42). For spiking experiments to samples with known transferrin and albumin content, aliquots of the albumin and transferrin stock solutions were added and prepared in the same way as the other samples. Before applying the samples to the CIM disks, they were passed through a 0.22 Am filter (Millipore).
The albumin and transferrin content was determined by radial immunodiffusion kits purchased from Dade Behring (Marburg, Germany) or The Binding Site (Birmingham, UK). Each test was performed according to the manufacturer’s instructions.
3. Results and discussion
One-hundred microliter of standard solution or sample equilibrated in binding buffer were injected onto the stack of disks equilibrated with binding buffer. Then a two-step gradient comprising an initial step of either 0.08 or 0.1 M followed by a second step of either 0.3 or 0.4 M NaCl, respectively, was applied at a flow rate of 4 ml/min to elute bound proteins from the ion-exchange disk. Glycine buffer containing 100 mM NaCl was applied to release bound proteins from the Protein G ligands.
For a well-characterized therapeutic product, it is important to know the concentration of each impurity. Albumin and transferrin are typical contaminating proteins in immunoglobulin concentrates. Their quantification is possible by different analytical methods such as RID, laser nephelometry and ELISA. These methods do not permit simultaneous determination of all proteins. In addition, there is no test system that is both fast and sensitive. Here, a method is described, which requires only 15 min for sample preparation and less than 5 min for separation. The conjoint liquid chromatography system was designed in a way that the immunoglobulin is captured by a Protein G monolithic disk, while impurities are bound by an anion-exchange monolithic disk placed in sequence. Then impurities are resolved by a salt gradient and immunoglobulins eluted with an acidic buffer. The chromatographic conditions for a successful separation of the impurities from the immunoglobulin were first optimized on the individual disks and then on the combination of the disks as described below.
2.8. Electrophoretic technique
3.1. Optimization of chromatographic conditions
Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) was carried out under non-
Protein affinity chromatography using CIM Protein G was chosen to separate immunoglobulins from
2.7. Chromatographic procedure
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
51
Fig. 1. Separation of immunoglobulin G from accompanying proteins on two CIM Protein G monolithic disks. Chromatograms of two different immunoglobulin concentrates (IG-1 and -2) are superimposed. Binding buffer: 20 mM Tris – HCl pH 7.0; elution buffer: 0.1 M Glycine – HCl, 0.1 M NaCl, pH 2.0; flow rate: 4 ml/min; sample concentration: 2 mg IgG/100 Al. (FT = flow through; P = eluted peak).
albumin and transferrin. Fig. 1 illustrates the separation of IgG concentrates with differing levels of impurity. The flow through (FT) and the eluate were
Fig. 2. SDS – PAGE under nonreducing conditions using 4 – 15% gradient gels. Samples were collected during the separation of two IgG concentrates (IG-1 and -2) on two CIM Protein G disks (see Fig. 1). Five hundred nanogram of protein were applied to each lane. S1 = starting material of IG-1, FT1 = flow through of IG-1, P = eluted peak of IG-1; S2 = starting material of IG-2, FT2 = flow through of IG-2, P2 = eluted peak of IG-2 (A = albumin; T = transferrin; Ig = IgG).
further analyzed by SDS-PAGE (see Fig. 2, lanes FT1 and FT2)). The main impurities of these concentrates, albumin and transferrin, could be quantitatively removed. Neither coeluted with IgG. On the other hand, immunoglobulins were completely eluted by a buffer with low pH because other accompanying proteins could not be detected by SDS-PAGE combined with silver staining; a highly sensitive method for the detection of proteins (Fig. 2, lanes P1 and P2). Next, the chromatographic stability of Protein G disks was determined by measuring the dynamic binding capacity for immunoglobulins after repeated sample applications. Regeneration of the disk was carried out after each sample application, and the dynamic binding capacity determined after 50 or 100 cycles. The dynamic binding capacity for immunoglobulins on the CIM Protein G disks was calculated at 10% breakthrough. After 100 cycles, the Table 1 Dynamic binding capacity of IgG on CIM Protein G disks (regeneration after each run was performed) Number of injections
Dynamic binding capacity Disk #1 [mg/ml]
Disk #2 [mg/ml]
1 50 100
14.7 14.5 11.7
13.1 12.0 11.7
52
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Fig. 3. Separation of transferrin and albumin on a CIM QA monolithic disk. Binding buffer: 20 mM Tris – HCl, pH 7.0; elution buffer: binding buffer containing 1 M NaCl; flow rate: 4 ml/min, sample concentration: 45 Ag transferrin and 45 Ag albumin/250 Al (FT = flow through; 1 = peak eluted with 0.08 M NaCl; 2 = peak eluted with 0.3 M NaCl).
monolithic disk still had the same resolution power, and only 10 –20% dynamic binding capacity was lost. As shown in Table 1, capacity slightly decreased with the number of cycles. After 50 applications, the capacity remained constant for disk no. 1, and for disk no. 2, it was only 10% lower. After 100 runs, the capacity of disk no. 1 was 20% lower and that of disk no. 2 was 10% lower. A strong anion-exchange CIM QA disk was selected to separate transferrin from albumin. In preliminary experiments, an attempt was made to separate immunoglobulins from their protein impurities by ion-exchange chromatography on monolithic disks. Due to the very broad range of isoelectric points of immunoglobulins (pI from 4.35 to 9.95; Prin et al., 1995), there was an overlap with the isoelectric points of albumin and transferrin, and complete resolution was not possible. Albumin and transferrin were bound to the CIM anionic disks, whereas the main fraction of the immunoglobulins passed through the ion exchanger. However, some immunoglobulins also bound to the anion-exchange disk and were coeluted with albumin and transferrin. A similar situation was observed with separations performed using the cationexchange mode. Most immunoglobulins bound to the ion exchanger, while a small fraction passed through the column together with transferrin and albumin (data not shown). Therefore, an affinity chromatog-
raphy step on a CIM Protein G disk was introduced, separating immunoglobulins from albumin and transferrin.
Fig. 4. SDS – PAGE of fractions collected during separation of transferrin and albumin on a CIM QA column. (see Fig. 3). A 4 – 15% gradient gel was used. Five hundred nanogram protein were applied to each lane. S = starting material (nonbound fraction derived from affinity chromatography on CIM Protein G column), FT = flow through, 1 = fraction eluted with 0.08 M NaCl, 2 = fraction eluted with 0.3 M NaCl, A = albumin; T = transferrin).
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
53
Fig. 5. Optimized separation of transferrin, albumin and IgG on CIM CLC monolithic column consisting of two CIM Protein G and one CIM QA disks. Chromatograms of two different runs using different step gradients are superimposed. Binding buffer: 20 mM Tris – HCl, pH 7.0; elution buffer I: 20 mM Tris – HCl, 1 M NaCl, pH 7.0; elution buffer II: 0.1 M Glycine – HCl, 0.1 M NaCl, pH 2.0; flow rate: 4 ml/min; sample concentration: 2 mg IgG/100 Al (‘‘1’’ indicates the transferrin peak eluted with 0.08 or 0.1 M NaCl; ‘‘2’’ indicates the albumin peak eluted with 0.3 or 0.4 M NaCl; ‘‘3’’ indicates the IgG peak eluted with low pH).
Fig. 6. SDS – PAGE under reducing and nonreducing conditions. A gradient gel of 4 – 20% was used. Samples were collected during the separation of an IgG concentrate on CIM CLC monolithic column (see Fig. 5). S = starting material, 1 = chromatographic peak No. 1, 2 = chromatographic peak No. 2, 3 = peak No. 3, A = albumin; T = transferrin; Ig = IgG).
54
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Table 2 Linearity of the standard curve of transferrin and albumin Transferrin [mg/ml]
Peak area for transferrin [mAU * ml]
Albumin [mg/ml]
Peak area for albumin [mAU * ml]
0.05 0.20 0.40 0.60 0.80 1.00 r = 0.9942 sy = 18.271 a = 417.83 b = 34.117
46 396 131 682 198 316 295 295 341143 466 259
0.05 0.20 0.40 0.60 0.80 1.00 r = 0.9959 sy = 9.9832 a = 273.75 b = 12.452
33 117 59 869 117 186 174 629 246 077 278 766
r: coefficient of correlation. sy: standard deviation of the response. a: slope. b: intercept.
The flow-through fraction from the Protein G affinity chromatography contained only transferrin and albumin because all immunoglobulins were captured by the Protein G disks (see Fig. 1). A total volume of 250 Al were loaded on one CIM QA disk. A linear salt gradient was applied to find the elution position of albumin and transferrin. For optimized resolution, a step gradient was investigated (Fig. 3). The highest resolution was obtained by eluting transferrin using a step gradient with 0.08 M
NaCl and albumin using a step gradient with 0.3 M NaCl. Under such conditions, albumin and transferrin are separated as demonstrated by SDS-PAGE with silver staining (Fig. 4). After completion of these optimization runs, both methods—affinity chromatography and anionexchange chromatography—were coupled together by stacking the disks into one housing. The concentration of IgG compared to transferrin and albumin was, in some samples, in the ratio of 1:500. In order to improve the limit of quantification for transferrin and albumin, two Protein G disks had to be combined with one ion-exchange disk. More sample volume could be injected without a breakthrough of IgG. Each Protein G disk had the ability to bind 5 mg IgG. Using two CIM Protein G disks and one CIM QA disk, a complete separation of immunoglobulins, albumin and transferrin could be achieved in one step in less than 5 min. Immunoglobulins were captured on the CIM Protein G disks, while albumin and transferrin were bound on a CIM QA disk. Albumin and transferrin were eluted by a salt gradient, while IgG was removed by lowering the pH of the eluting buffer. Tailing of the peaks was reduced by increasing the salt concentration of each gradient step (Fig. 5). Transferrin was eluted at 0.1 M NaCl instead of 0.08 M, and albumin was eluted at 0.4 M NaCl instead of 0.3 M. Immunoglobulins were eluted by decreasing the
Fig. 7. Linearity of the separation of transferrin, albumin from IgG on CIM CLC monolithic column. Chromatograms of six different runs with sample concentrations from 0.05 to 1.00 mg/ml albumin or transferrin, respectively, are superimposed. Binding buffer: 20 mM Tris – HCl, pH 7.0; elution buffer I: 20 mM Tris – HCl, 1 M NaCl, pH 7.0; elution buffer II: 0.1 M Glycine – HCl, 0.1 M NaCl, pH 2.0; flow rate: 4 ml/min; sample concentration: see standard preparation in the Materials and methods section.
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
55
Fig. 8. Linearity of the calibration function: regression lines of the separation of transferrin, albumin from IgG on a CIM monolithic column. The peak areas used for regression analysis are shown in Table 2.
pH value to 2.0 supplemented with 100 mM sodium chloride. Eluates were immediately neutralized with 80 Al of 1 M Tris –HCl, pH 9.0. Separated fractions were collected and analyzed by SDS-PAGE (Fig. 6). The IgG concentrate (lane S) containing all three proteins (immunoglobulins, transferrin and albumin) revealed bands with apparent molecular masses of 200, 75 and 66 kDa, respectively. In the first peak, only transferrin was identified (lane 1). In the second peak, only albumin was detected (lane 2) and only immunoglobulins in the last peak (lane 3). 3.2. Method validation Increasing amounts of transferrin and albumin standards were added to the pure immunoglobulin sample (containing 25 mg IgG/ml) to give concentrations of 0.05, 0.2, 0.4, 0.6, 0.8 and 1.0 mg/ml, respectively, before application to the monolithic disks combined in one housing. Peak areas were plotted against the amount of injected material and regression analysis was performed. First-order linear regression was obtained for transferrin and albumin with correlation coefficients of 0.994 and 0.996, respectively (Table 2). Residuals were randomly distributed. At the 95% level, it was established that the confidence interval of the intercept included ‘‘0.’’ Thus, it can be concluded that the
method was performed without constant systematic error. Chromatograms and regression lines are presented in Figs. 7 and 8, respectively. For confirmation of the lower limit of the test range, the limit of quantification was calculated according to the ICH guidelines using a specific calibration curve. Therefore, a standard curve ranging from 0.07 to 0.21 mg/ml albumin and transferrin, respectively, was prepared (Table 3). In this case,
Table 3 Determination of limit of quantification Transferrin [mg/ml]
Peak area for transferrin [mAU * ml]
0.071 60.106 0.100 72.841 0.129 85.634 0.143 98.031 0.157 105.036 0.186 114.844 0.214 128.342 r = 0.996 a = 486.03 sy = 2.46 LOQ = 0.05 mg/ml
Albumin [mg/ml]
Peak area for albumin [mAU * ml]
0.071 18.747 0.100 25.586 0.129 31.298 0.143 34.636 0.157 37.682 0.186 44.187 0.214 47.700 r = 0.997 a = 206.81 sy = 0.83 LOQ = 0.04 mg/ml
r: coefficient of correlation. a: slope. sy: standard deviation of the response. LOQ: limit of quantification = 10 sy/a.
56
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Table 4a Repeatability and intermediate precision for transferrin
Table 5a Recovery rate of transferrin
Transferrin Day Concentration [mg/ml]
Xim STD RSD [%]
Transferrin 1 0.18 0.19 0.19 0.18 0.19 0.186 0.0055 2.9
2
3
0.16 0.18 0.19 0.18 0.18 0.177 0.0104 5.8
0.15 0.14 0.15 0.16 0.15 0.150 0.0071 4.7
Measured
0.171 0.0175 10.2
Spike 2
Spike 3
the correlation coefficient for transferrin was 0.996 and that for albumin 0.997. The limit of quantification for transferrin was 0.05 mg/ml and for albumin 0.04 mg/ml as calculated by multiplying the standard deviation of the response by a factor of 10 and dividing by the slope of the calibration curve.
Table 4b Repeatability and intermediate precision for albumin Albumin
Concentration [mg/ml]
Xim STD RSD [%]
1 0.10 0.09 0.10 0.09 0.10 0.096 0.0052 5.4
0.089 0.0090 10.2
Xim: mean value per day. Xm: total mean value. STD: standard deviation. RSD [%]: relative standard deviation.
0.270 0.270 0.270 0.570 0.570 0.570 0.970 0.970 0.970
107 117 119 98 101 100 102 103 104
[%]
2
3
0.09 0.09 0.09 0.10 0.09 0.092 0.0045 4.9
0.07 0.08 0.08 0.08 0.08 0.078 0.0045 5.7
The precision of the method was tested by application of a sample, containing 0.17 mg/ml transferrin and 0.09 mg/ml albumin five times on three different days. The relative standard deviation (RSD) per day ranged from 2.9% to 5.8% for transferrin and from 4.9% to 5.71% for albumin. The intermediate precision was 10.2% for both transferrin and for albumin (Table 4a and 4b). The recovery for spiked samples was determined. An immunoglobulin sample, containing 0.17 mg/ml transferrin and 0.09 mg/ml albumin, was spiked with 0.1, 0.4 and 0.8 mg/ml albumin and transferrin, respectively. Three runs with each solution were performed; the recovery for transferrin ranged from 98% to 119% whereas that for albumin ranged from 82% to 101% (Table 5a and 5b). Table 5b Recovery rate of albumin Albumin Measured
Calculated
[g/l] Spike 1
Intermediate precision Xm STD RSD [%]
0.290 0.317 0.322 0.557 0.574 0.569 0.987 1.003 1.009
measured concentration ½mg=ml RR: Recovery Rate ¼ calculated concentration ½mg=ml 100%.
Xim: mean value per day. Xm: total mean value. STD: standard deviation. RSD [%]: relative standard deviation.
Day
RR
[g/l] Spike 1
Intermediate precision Xm STD RSD [%]
Calculated
Spike 2
Spike 3
0.169 0.191 0.178 0.483 0.482 0.484 0.743 0.814 0.729
RR [%]
0.189 0.189 0.189 0.489 0.489 0.489 0.889 0.889 0.889
measured concentration ½mg=ml RR: Recovery Rate ¼ calculated concentration ½mg=ml 100%.
89 101 94 99 99 99 84 92 82
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58 Table 6 Comparison of CLC with RID Sample
Transferrin [mg/ml]
No
RID
1 2 3 4 5 6
0.36 0.40 0.14 0.09 0.11 0.12 0.36 0.37 0.14 0.19 0.12 0.16 Mean D STD D TV t (5, 95%)
CLC
Albumin [mg/ml]
D
RID
0.04 0.05 0.00 0.01 0.05 0.04 0.02 0.0367 1.11 2.571
0.52 0.52 0.06 0.05 0.09 0.11 0.70 0.70 0.03 0.08 0.03 0.05 Mean D STD D TV t (5, 95%)
CLC
D 0.00 0.01 0.02 0.00 0.05 0.02 0.01 0.0216 1.51 2.571
Mean D: mean value of D. STD D: standard deviationpof ffi D. DA n TV: test value ¼ Amean . STD D t: significance value of the student’s distribution.
Immunoglobulin concentrates (# IGs 1 –6) were analyzed by the conjoint liquid chromatography method and by RID, which is accepted as one of the standard methods for the determination of transferrin and albumin. The contents of transferrin and albumin obtained from integration of the respective peaks were compared with the results obtained from RID (see Table 6). A difference t-test at the 95% level confirmed that there was no statistically significant difference between the methods. Preliminary experiments showed that peaks with shoulders and tails were observed, when monolithic disks were damaged. Similar observation was also reported by Hahn and Jungbauer (2001) for some prototype disks. Asymmetry (As) of the transferrin peak was chosen as a criterion for system suitability. Asymmetry was calculated as the ratio (at 10% height) of the first half peak width to the second half peak width. As acceptance criteria of 0.8 and 1.2 were set. If As were out of this range, the run was repeated. If this resulted again in an asymmetric peak, new disks were taken. The special geometry of monolithic disks permitted a very fast separation with low back pressure. The stacking of monolithic disks with different functional groups avoided column switching and made the method faster and simpler. The separation of contaminating proteins, particularly albumin and transferrin, and monoclonal antibodies from mouse ascites fluid in a tandem arrangement consisting of
57
CIM anion-exchange monolithic disk and a CIM Protein A monolithic disk has been described previously (Josic et al., 1998). These authors demonstrated that the separation of IgG from accompanying proteins was possible within only 3 min. However, no quantitative determinations were performed. Strancar et al. (1998) separated IgG from human plasma in 2 min by combining a CIM DEAE and a CIM Protein G monolithic disk in one housing, but, again, no quantification of the isolated protein was carried out. Gupalova et al. (2002) demonstrated that IgG and serum albumin (SA) from mammalian plasma can be separated within 15 min by two disks with different affinity functionalities. The elution had to be performed discontinuously because the IgG-binding disk as well as the SA-binding disk have to be treated with very acidic solutions in order to achieve desorption of the target proteins. However, the goal of this work was not a quantitative determination, but preparative work for quantitatively capturing of both proteins. The present studies demonstrate that conjoint liquid chromatography can be used as a precise and accurate analytical method, which could be validated for pharmaceutical purposes.
Acknowledgements This work was supported by the Forschungsfo¨rderungsfonds, Austria, project no. 803172.
References Akerstro¨m, B., Brodin, T., Reis, K., Bjo¨rck, L., 1985. Protein G: a powerful tool for binding and detection of monoclonal and polyclonal antibodies. J. Immunol. 135, 2589. Bjorck, L., Kronvall, G., 1984. Purification and some properties of streptococcal Protein G, a novel IgG-binding reagent. J. Immunol. 133, 969. Cohn, E., Strong, L., Hughes, W., Mulford, D., Ashworth, J., Melin, M., Taylor, H., 1946. Preparation and properties of serum and plasma proteins: IV. A system for the separation into fractions of the protein and lipoprotein components of biological tissues and fluids. J. Am. Chem. Soc. 68, 459. Curling, J.M., 1980. Methods of Plasma Protein Fractionation. Academic Press, London, UK. Dickler, H., Gelfand, E., 1996. Current perspectives on the use of intravenous immunoglobulin. Adv. Intern. Med. 41, 641.
58
K. Branovic et al. / Journal of Immunological Methods 271 (2002) 47–58
Dwyer, J., 1987. Intravenous therapy with gamma globulin. Adv. Intern. Med. 32, 111. Fang, Y., Dumont, L., Larsen, B., 1998. Real-time isoform analysis by two-dimensional chromatography of a monoclonal antibody during bioreactor fermentations. J. Chromatogr., A 816, 39. Friesen, A., Bowman, J., Bees, W., 1985. Column ion exchange chromatographic production of human immune serum globulin for intravenous use. Vox Sang. 48, 201. Gupalova, T.V., Lojkina, O.V., Pa`la`gnuk, V.G., Totolian, A.A., Tennikova, T.B., 2002. Quantitative investigation of the affinity properties of different recombinant forms of Protein G by means of high-performance monolithic chromatography. J. Chromatogr., A 949, 185. Hage, D., Walter, R., 1987. Dual-column determination of albumin and immunoglobulin G in serum by high-performance affinity chromatography. J. Chromatogr. 386, 37. Hahn, R., Jungbauer, A., 2001. Control method for integrity of continuous beds. J. Chromatogr., A 908, 179. Heukeshoven, J., Dernick, R., 1985. Simplified method for silver staining of proteins in polyacrylamide gels and the mechanism of silver staining. Electrophoresis 6, 103. Hoppe, H., Mester, T., Hennig, W., Krebs, H., 1973. Prevention of Rh-immunization. Modified production of IgG anti-Rh for intravenous application by ion exchange chromatography (IEC). Vox Sang. 25, 308. Josic, D., Schwinn, H., Strancar, A., Podgornik, A., Barut, M., Lim, Y.-P., Vodopivec, M., 1998. Use of compact, porous units with immobilized ligands with high molecular masses in affinity chromatography and enzymatic conversion of substrates with high and low molecular masses. J. Chromatogr., A 803, 61. Josic, D., Buchacher, A., Jungbauer, A., 2001. Monoliths as stationary phases for separation of proteins and polynucleotides and enzymatic conversion. J. Chromatogr., B 752, 191. Kistler, P., Nitschmann, H., 1962. Large scale production of human plasma fractions. Vox Sang. 7, 414. Laemmli, U., 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680. Langlotz, P., Kroner, K.H., 1992. Surface-modified membranes as a matrix for protein purification. J. Chromatogr., A 591, 107. Mancini, G., Carbonara, A., Heremans, J., 1965. Immunochemical quantitation of antigens by single radial immunodiffusion. Immunochemistry 2, 235.
Nourichafi, N., Geschier, C., Stoltz, J.-F., 1993. Comparison of various chromatographic supports for purifying human plasmatic immunoglobulins from Cohn II + III fraction. Biotechnol. Blood Proteins 227, 207. Ostryanina, N.D., Vlasov, G.P., Tennikova, T.B., 2002. Multifunctional fractionation of polyclonal antibodies by immunoaffinity high-performance monolithic disk chromatography. J. Chromatogr. 949, 163. Platonova, G.A., Pankova, G.A., Il’ina, I.Ye., Vlasov, G.P., Tennikova, T.B., 1999. Quantitative fast fractionation of a pool of polyclonal antibodies by immunoaffinity membrane chromatography. J. Chromatogr., A 852, 129. Pressler, T., Mansa, B., Pedersen, S., Espersen, F., Hoiby, N., Koch, C., 1994. Methodologic problems in establishing normal values for IgG subclass concentrations in a pediatric population; comparison of radial immunodiffusion and ELISA methods. Allergy 49, 772. Prin, C., Bene, M., Gobert, B., Montagne, P., Faure, G., 1995. Isoelectric restriction of human immunoglobulin isotypes. Biochim. Biophys. Acta 1243, 287. Qi, Y., Zan, Z., Huang, J., 2001. Chromatography on DEAE ionexchange and Protein G affinity columns in tandem for the separation and purification of proteins. J. Biochem. Biophys. Methods 49, 263. Sternberg, J., 1977. A rate nephelometer for measuring specific proteins by immunoprecipitin reactions. Clin. Chem. 23, 1456. Strancar, A., Barut, M., Podgornik, A., Koselj, P., Josic, D., Buchacher, A., 1998. Convective Interaction Media (CIM): polymer based supports for fast separation of biomolecules. LC GC Int. 10, 660. Strancar, A., Podgornik, A., Barut, M., Necina, R., 2002. Short monolithic columns as stationary phases for biochromatography. In: Freitag, R. (Ed.), Modern Advances in Chromatography: Advances in Biochemical Engineering/Biotechnology, vol. 76. Springer-Verlag, Heidelberg, Germany, p. 50. Zhou, D., Zou, H., Ni, J., Yang, L., Jia, L., Zhang, Q., Zhang, Y., 1999. Membrane supports as the stationary phase in highperformance immunoaffinity chromatography. Anal. Chem. 71, 115. Zou, H., Luo, Q., Zhou, D., 2001. Affinity membrane chromatography for the analysis and purification of proteins. J. Biochem. Biophys. Methods 49, 199.