IMPROVED ION-EXCHANGED HPLC METHOD IN mAb USING pH GRADIENT AND ITS COMPARISON WITH clEF

Ion-exchange chromatography (IEX) and capillary isoelectric focusing (cIEF) are two common methods used for molecular charge heterogeneity analysis of therapeutic proteins. The two methods are complementary to each other and each method has its own advantage and disadvantages. For example, a sample injected on an IEX column can be easily fraction collected but, sometimes, it lacks the ability to provide good separation of charge variants, such as those present in monoclonal antibody (mAb) samples. While cIEF has much better resolution than IEX, the separated peaks cannot be collected. Recently, pIsep, a new ion chromatographic technique developed by CryoBioPhysica, has been successfully applied to analyze charged heterogeneities in mAbs. The pIsep separation utilizes an external pH gradient and the charged variants are resolved based on pI differences. In this study, an acidic variant co-eluted with main peak on traditional salt-gradient cation-exchange chromatography (CEX) is separated by pIsep and the fraction was collected for further characterization. The collected fractions were used to characterize the correlation between CEX and cIEF peaks and the results from both methods are consistent.

Keywords: acidic variant; capillary isoelectric focusing; cation-exchange chromatography; CEX; cIEF; mAb; pH-gradient

Ion exchange chromatography (IEX) and capillary isoelectric focusing (cIEF) are useful analytical tools for the analysis of therapeutic protein variants based on their molecular charge heterogeneities. Both methods have been widely used and reported to characterize charge variants in mAb.[[[1]]] IEX separation relies on surface charge-charge interactions between the mAb protein and the charges immobilized on the resin of column. For example, in traditional cation exchange chromatography (CEX) positively charged ions on exposed mAb surface bind to a negatively charged functional group on column and are eluted by a salt gradient.[[7]] In contrast, cIEF separation is based on net charge heterogeneity of mAb. When a mAb protein is placed in carrier ampholytes creating a pH gradient and subjected to an electric field, it will initially move toward the electrode with the opposite charge. During migration through a pH gradient, the mAb protein is eventually focused into sharp bands where their pI is equal to the pH.[[8]]

While both methods provide complementary information, CEX also has an advantage as compared to cIEF that fractions can be collected for further characterization, and, in contrast, it is difficult and nearly impossible to collect fractions from cIEF. However, CEX sometimes may not provide a satisfactory resolution that is needed to separate all charge variants whereas cIEF gives a much better resolution. In order to bridge between these two methods, relatively new reagents from CryoBioPhysica used in CEX to separate mAb protein based on pI was developed and was first reported by Tsonev and Hirsh.[[9],[10]] This is called the pIsep HPLC method and utilizes a gradient chromatofocusing mechanism, namely, the pH gradient is introduced by external mixing of two solutions with different pH. With good controllability of the ratio of the two different pH buffers, the limitations of pH-gradient slope is eliminated in the method.[[11]] The pIsep HPLC technique also utilizes a low conductivity buffer, which prevents salt-elution effects from skewing the selectivity.[[12]] With these advantages, pIsep creates a pH gradient with good linearity and the mAb protein charge variants are resolved mainly based on their pI differences in addition to conformation. Proteins that bind to a traditional ion exchange column in a suitable pH environment are eluted when pH of the mobile phase reaches the protein's pIs, which results in the charged-based variant separation. Recently, a similar pH gradient-based ion-exchange chromatography method for mAb has been successfully validated by Rea et al.[[13]]

We have used both CEX and cIEF methods for determination of charge variants in mAb characterization. In this report, we demonstrate the utility of the pIsep HPLC method for mAb analysis and illustrate the advantage of this method for further characterization of mAbs, providing the missing link between the traditional salt-gradient CEX and cIEF.

A mAb (IgG1) used in these experiments were produced in mice myeloma NS0 cells, and purified to ≥99% by the Bioprocess Research and Development group at Merck Research Laboratories (Merck & Co., Inc., West Point, PA, USA). Sample concentrations were measured using UV/Vis spectrophotometry with a known extinction coefficient.

Phosphoric acid (1 M), hydrochloride acid (1 M), and sodium hydroxide (1 M) were purchased from Beckman Coulter (Fullerton, CA, USA). All pI markers and methylcellulose solutions were obtained from Convergent Bioscience (Toronto, Canada). Bio-Lyte ampholytes (pH 3–10 and pH 8–10) were bought from BIO-RAD Laboratories, Inc. (Richmond, CA, USA). Sodium phosphate monobasic and sodium chloride were purchased from Thermo Fisher (Waltham, MA, USA). Acidic and basic buffers (proprietary formulation) for pIsep HPLC were purchased from CryoBioPhysica (Rockville, MD, USA).

CEX experiments were conducted on a Dionex HPLC system with Chromeloen software version 6.7. Dionex GP40 pump and AD 250 UV detector were employed along with a Thermo AS3500 autosampler. A Dionex ProPac WCX-10, 4 × 250 mm column was used. The mobile phase A was 10 mM sodium phosphate, pH 7.0 and the mobile phase B was 10 mM sodium phosphate with 1 M sodium chloride, pH 7.0. Samples were diluted to 1 mg/mL in the mobile phase A and were injected into ProPac WCX-10, 4 × 250 mm column and detected at 280 nm wavelength. The mAb samples were separated using a gradient of 4%–20% mobile phase B in 20 min at 1 mL/min flow rate. In addition, a shallow gradient of 4–12% mobile phase B in 60 min at 1 mL/min was also run using the same column to improve separation of acidic variants.

The cIEF run was conducted on Convergent Bioscience iCE280 Analyzer with an Alcott liquid chromatography autosampler. The separation column was a 50 mm long, 100 µm ID × 365 µm OD silica capillary coated with fluorocarbon. Samples were diluted to 0.25 mg/mL in a solution of 0.35% methylcellulose containing a mixture of pH 3–10 and 8–10 carrier ampholytes. Two pI markers of 7.6 and 9.5 were also added to the sample solution mixture. The catholyte consists of 0.1 M NaOH in 0.1% methylcellulose and the anolyte is 0.08 M phosphoric acid in 0.1% methylcellulose. Samples were focused for 1 min using a voltage of 1500 V followed by a period of 8 min at 3000 V. The absorbance of the focused proteins is detected at 280 nm.

The pIsep HPLC runs were conducted on a Dionex HPLC system with Chromeloen software version 6.7. Dionex GP40 pump and AD 250 UV detector were employed along with a Thermo AS3500 autosampler. A Dionex ProPac WCX-10, 4 × 250 mm column was employed. Buffers for pIsep were supplied as a kit consisting of two stock concentrates, 5 mL of 500-fold concentrate low pH buffer A and 50 mL of 100-fold concentrate high pH buffer B. To prepare 500 mL acidic buffer or mobile phase A, 1 mL buffer A stock and 5 mL buffer B stock are added to 490 mL of water. To prepare 500 mL the basic buffer or mobile phase B, 5 mL buffer B stock is added to 490 mL of water. The pH of the mobile phase A was adjusted to 2.4 with 1 M HCl and mobile phase B to10.75 with 1 M NaOH. The final volume was adjusted to 500 mL for each buffer. Sample was diluted to 1 mg/mL, injected into ProPac WCX-10, 4 × 250 mm column, and detected at 280 nm wavelength. The mAb samples were separated using a gradient of 60%–95% mobile phase B (corresponding to pH 7.9–10 gradient) in 20 minutes at 1 mL/min flow rate at room temperature.

Fractions (1 mL each) were collected from pIsep HPLC run and pH was measured immediately after the collection using a pH meter (Accument ARIS, Fisher Scientific). Fractions were also collected as acidic variants (A1, A2, and A3), K0, basic variants (K1 and K2) in total 12 runs. After each run the fractions were immediately neutralized with 0.1 N HCl. All collected fractions were concentrated using Millipore filter (iCon concentrator, Pierce prod # 89994). The concentrated fractions were further analyzed using CEX and cIEF.

In salt gradient CEX, the best resolution in terms of charge variant separation normally occurs when the buffers operate at a pH close to mAb pI values.[[14]] However, in practice this is difficult to achieve since the mAb needs to first bind to the CEX column. Using the traditional CEX method with short salt gradient, we observed at least 2 small acidic variants (A1 and A2), the co-elution of major acidic variant (A3) and main peak (K0), and 2 additional basic peaks (K1 and K2) (see Figure 1a). In the longer salt gradient CEX, the separation is slightly better but still not enough to provide reliable quantitation of A3 (see Figure 1b). The peak assignments of K0, K1, and K2 are mAb with zero lysine, one lysine, and two lysine residues at that C-terminal end of the heavy chain.[[1]] These assignments for this particular mAb have been confirmed with carboxypeptidase enzyme treatment (data not shown). The A1 and A2 acidic variants have been assigned to the deamidation in the variable region of the mAb.[[15]] The A3 acidic variant is difficult to assign due to the fact that its peak is close to K0; hence, it is very difficult to collect with high purity.

Graph: FIGURE 1 (A) Chromatogram of salt gradient CEX for mAb using short gradient, illustrating three acidic variants (A1, A2, A3), main peak (K0), and two basic variants (K1, K2). (B) Similar chromatogram using longer salt gradient provides better separation, but it is still not possible to have adequate resolution between A3 and K0 variants.

In order to understand further the CEX surface charge heterogeneities, we performed cIEF experiment to evaluate its net total charge heterogeneities. The cIEF electropherogram profile is shown in Figure 2 illustrating that this particular mAb has a similar profile to the CEX separation. The main peak (K0) has pI of 8.9, basic peaks of (K1 and K2) have pI of 9.0 and 9.1, while the two acidic peaks (A1 and A2) are 8.7 and 8.8, respectively. Similarly as in CEX, the K0, K1, and K2 peak assignments correspond to mAb with zero, one, and two lysines at the C-terminal end of the heavy chain and have been confirmed using carboxypeptidase enzyme treatment (data not shown). The cIEF data demonstrate a good correlation with those from CEX with respect to main peak (K0) and two basic variants (K1 and K2). However, there are only two acidic variants were detected in cIEF while there are three acidic peaks in CEX. This difference may be due to the fact that in cIEF total charge is detected while in CEX surface charge and conformational heterogeneities are separated. Based on the intensity, it is suspected that A3 in CEX corresponds to A2 in cIEF; however, the cIEF technique is being hampered by not being able to collect fractions for further analysis, hence conclusive evidence cannot be reached. The data indicates that a better separation is achieved by cIEF. However, the inability to collect charge variant fractions has been an issue in further advancing this particular mAb characterization. Hence, an alternative HPLC-based method is evaluated to improve A3 and K0 separation in salt gradient CEX.

Graph: FIGURE 2 Electropherogram of mAb from cIEF experiment using a mixture of carrier ampholytes of 3–10 and 8–10.5 pH range. The charge heterogeneities profile is similar to those observed with CEX (Figure 1) except that it is missing one acidic variant in comparison with CEX.

Based on a similar concept to cIEF, pIsep is a powerful technique for the separation of proteins that have close charge properties by using a pH gradient. A significant feature in pIsep is the flexibility of control for slope of the outlet pH gradient. The upper and lower limits of the pH gradient are established by buffers A and B as well as the rate of increase of the proportion of buffer B. In order to separate different variants of mAb that have a pI range from 7.9 to 10 using pIsep, the pH gradient was optimized using software that was provided by CryoBioPhysica Company. The pISep HPLC chromatogram is illustrated in Figure 3A indicating that it has three acidic variants (A1, A2, and A3), main peak (K0), and two basic variants (K1, and K2). This overall chromatogram profile is quite similar to salt-gradient CEX but with a significant separation improvement between A3 and K0 and slightly reduced, but acceptable separation between A1 and A2. Unlike with salt gradient CEX, pIsep HPLC allows the protein to be focused in narrower bands during the pH-gradient elution, providing higher resolution than that usually observed in salt-gradient elution with CEX. This result provides the possibility of further characterize the A3 peak, but first this A3 from pIsep HPLC needs to be assessed and compared with A3 from salt gradient CEX and A2 from cIEF to understand its origin and the presence of any correlations.

Graph: FIGURE 3 (A) Chromatogram of mAb from pIsep HPLC experiment. The pH gradient elution was performed from pH 8 to 10. The profile is quite similar to salt gradient CEX shown in Figure 1. (B) The pH gradient curve obtained by measuring the pH off line every 1 min. Range used in Figure 3A is shown by Range Δ.

A good linearity for pH is critical for chromatofocusing type of separations.[[[16]]] In pIsep HPLC the pH linearity is governed by controlling an outlet pH gradient slope. For demonstration of linearity, fractions (1 mL each) were collected during the whole pIsep HPLC run and the pH was measured in each fraction. The fraction versus pH data is plotted in Figure 3B indicating that pH gradient is indeed linear. According to the pH plot generated from fractioning on pIsep, the pIs of the four major variants (A3, K0, K1, and K2) are in the pH range of 9.0–9.3 which is 0.2 pH unit higher than pI range of 8.8–9.1 tested by cIEF. The pI difference observed from the two methods may be due to the functional nature of external pH gradient and internal pH gradient, as well as the accuracy of the pH measurement. Nevertheless, the four major variants are eluted within 0.2 pH unit from both pIsep and cIEF, indicating a comparability of the two methods and the relative positions of the different variants.

In order to evaluate further regarding this new pH gradient HPLC method with respect to quantitation, we have calculated the % charge variants observed in pIsep HPLC and compared with cIEF using ten different lots of mAb. The data are listed in Table 1 and they show that pIsep HPLC gives comparable results with cIEF for analytical quantitation of mAb charge variant isoforms.

TABLE 1 Comparison of % Charge Variants Analyzed by pIsep and cIEF

In the case of this particular mAb, by using pIsep HPLC, the separation of A3 acidic variant from the K0 main peak is greatly enhanced therefore we can collect fractions from the mAb sample for further characterizations such as mass spectrometry (MS), ELISA, and/or other analytical tools. We have collected all acidic variants (A1, A2, and A3), main peak (K0), and the two basic variants (K1 and K2) from pIsep HPLC experiment. The collected fractions were then re-injected into both salt gradient CEX and cIEF methods. The data for cIEF and CEX are shown in Figure 4A and 4B, respectively. The expected K0, K1, and K2 peaks in pIsep HPLC correspond well with K0, K1, and K2 in both cIEF and CEX. And, finally, it is clear that A3 acidic peak in pIsep HPLC corresponds to A2 acidic peak in cIEF and A3 acidic peak in CEX (note that this fraction provides a split peak in CEX between A3 and K0 due to poor resolution). This result confirms that A3 peak in salt gradient CEX is the same as A2 peak in cIEF. Note that A1 and A2 fractions were collected together from pIsep HPLC, but they failed to show in both cIEF and CEX which is likely caused by insufficient materials. Efforts to further characterize this A3 from CEX or pIsep are in progress using mass spectrometry.

Graph: FIGURE 4 Collected peak fractions from pIsep HPLC (Figure 3A) were reinjected into (A) cIEF; the peak labeled A2 was collected from the peak labeled A3 in pIsep HPLC (Figure 3A); (B) short salt gradient CEX HPLC.

In summary, pIsep method, using a pH gradient similar to chromatofocusing, greatly extends the operational capabilities of conventional chromatofocusing and also provides a very useful preparative tool for protein charge variant fractionation. A similar approach to characterize peaks in cIEF using preparative immobilized pH gradient (Agilent OFFGEL 3100 Fractionator) was published by Meert et al.[[19]] The pIsep HPLC method has been successfully applied to bridge between CEX and cIEF applications by linking the distribution of various charge variants. The pIsep method mitigates the shortcomings of salt gradient CEX and combines the positive performance elements (such as recovery and resolution) of both ion exchange chromatography and isoelectric focusing techniques. It could increasingly be used more in the characterization of therapeutic monoclonal antibody development.

Since CEX and cIEF are used as standard analytic tools for determination of different charge variants of mAb, it is very important to understand the relationship between the two methods. We have demonstrated a good correlation between CEX and cIEF methods for the major charge variants, A3, K0, K1, and K2 in this mAb through pIsep HPLC. Similar approach can be applied to both facilitate the correlation of CEX and cIEF methods or for direct product characterization by pIsep method.

We also would like to thank our colleagues in Cell Culture Fermentation and Bioprocess Purification groups who provided us with all mAb materials. We thank Dr. Michael W. Washabaugh for his support of the work.