Cation-exchange chromatography is the third most used industrial method for antibody purification after anion-exchange and protein A affinity chromatography. It is most commonly used as an intermediate step but continues to attract attention as a capture method. This offers obvious cost and cleaning advantages over protein A but also imposes some sacrifices, all of which are discussed in a number of recent articles (1,2,3,4,5). Whichever application may be intended, end users seek a common set of performance characteristics. They include high capacity and recovery, high purification factor, a high degree of process control, high lot-to-lot reproducibility, and easy cleanability and sanitizability.

(Click to enlarge)

Chromatography media vendors have responded by introducing a continuum of new cation exchangers to better serve those applications. Given that they all strive to offer the ideal product, it might be expected that their capabilities would converge over time. That may happen, but the present selection of cation exchangers actually differ more from one another than they did 20 years ago. This results from their being synthesized with a greater diversity of particle architectures, from a greater diversity of polymers, and mostly exploiting one of many grafting techniques to increase ligand density and/or accessibility. Each product thus represents a unique mosaic of performance characteristics, with some features closely approaching the intended ideal and some less so.

PRODUCT FOCUS: MONOCLONAL ANTIBODIES AND OTHER PROTEINS

PROCESS FOCUS: PURIFICATION, CAPTURE, INTERMEDIATE PURIFICATION

WHO SHOULD READ: PROCESS DEVELOPMENT AND MANUFACTURING

KEYWORDS: CATION-EXCHANGE CHROMATOGRAPHY, PURIFICATION, MABS

LEVEL: BASIC TO INTERMEDIATE

Commercial performance data can help identify promising candidates for evaluation but typically do not include comprehensive competitive comparisons with other products. Users are left with that task. Unfortunately, no standardized approach has been suggested to meet this need. This article describes a limited set of basic experiments, the results from which can be used to support an objective determination of which exchanger best suits the needs of a particular application.

Capacity

Capacity has been a major focus for most recent-generation cation exchangers, with several now claiming dynamic binding capacities (DBCs) for IgG in excess of 100 g/L. As with most things, how you measure capacity influences the results that you get, yet vendor data are seldom accompanied by the procedures used to obtain them. It becomes impossible to draw valid conclusions about relative performance. Besides that, data from different sources are almost certainly obtained with samples that differ from one another, with no indication of how accurately any of them may represent your own materials. The “Estimation” box offers a comprehensive procedure for measuring dynamic capacity and includes options for evaluating elution and cleanability characteristics.

The two most common ways to determine dynamic capacity differ in the way an antibody is applied. The easier method uses off-line equilibration of the antibody to loading conditions. This method has two disadvantages: An excess amount of antibody has to be prepared for each set of conditions, and if antibody solubility is even minimally impaired under the selected conditions, it may invalidate the results and clog the column(s) being evaluated. The longer a prepared sample awaits loading, the more likely this is going to be an issue.

Antibody application by in-line dilution (ILD) is more complicated but avoids both limitations (6). Fully soluble antibody is applied through one pump, diluent buffer through another. They meet at the mixer. Residence time in diluted conditions is limited to the transit time from mixer to column, which may range from less than a second to a few seconds, depending on flow rate and the configuration of the chromatograph. After one experimental series has been completed at a particular dilution, another can be run at another dilution without a need to change the primary feed stream. This facilitates evaluating the effects of conductivity. It concurrently alters the feed concentration at the column inlet, which can create a source of error. But the error factor is equivalent across the exchangers being evaluated, so data remain suitable for comparing the performance of multiple exchangers. The process advantages and disadvantages of ILD are discussed in reference 7,.

One key requirement for comparing different exchangers is that a consistent breakthrough increment be used throughout a series of experiments. Breakthrough is generally understood to reflect the column load at which a column becomes saturated and subsequently loaded material simply flows through. The situation is more complicated on porous particle ion exchangers because loaded protein begins to break through long before the column approaches saturation. The mobile phase does not flow through the particles; it takes the path of least resistance: between the particles. Binding requires that proteins enter particle pores as they pass by particle surfaces. Thus, a given protein molecule may be swept through many strata of a column before it encounters an opportunity to bind, even though many unloaded binding sites remain. Every protein-binding event leaves less opportunity for the next protein to bind, to the point that some applied protein eventually fails to bind and exits the column. UV signal rises as an increasing concentration of protein “breaks through.” Breakthrough values for dynamic capacity determination are ultimately arbitrary. For industrial applications, values of 1, 5, or 10% are commonly reported. For experimental characterization of binding kinetics, 50% breakthrough is commonly used. As shown in Figure 1, the breakthrough increment strongly influences the capacity value you obtain. Pick a value that makes sense for your application and use that value consistently throughout your experiments.

Estimation of Dynamic Capacity, Resolution, Recovery, and Cleanability: Sample Application by In-Line Dilution

This procedure is written for an ÄKTAexplorer 100 (GEHealthcare Life Sciences, www.gelifesciences.com) plumbed with dark green PEEK tubing for applications up to 20 mL/min, but it can be adapted to any chromatograph with gradient capability, possibly with some manual intervention (16). The reference also provides a procedure for bulk loading. Refer to Figure 8 for representative results.)

1. For proteins, set the monitor to 280 and 300 nm. Monitoring at 300 nm will provide an on-scale trace if proteins exceed the range at 280 nm.

2. Plumb inlet A1 with feed stream.

3. Plumb inlet A2 with equilibration buffer.

4. Plumb inlet B1 with diluent buffer.

5. Plumb inlet B2 with elution buffer.

6. Plumb the sample injection valve with a 50-mL superloop.

7. Attach columns to be evaluated to the column selection valve.

8. Equilibrate Line A2. (Gradient, 0% B)

9. Equilibrate line B2. (Gradient, 100% B)

10. Equilibrate line B1. (Gradient, 100% B)

11. Fill the superloop with 2M guanidine-HCl, pH ~5.

12. Equilibrate all columns to be evaluated to line A2, then set the system so that all columns are off line. Zero the monitor.

13. Set gradient to 0% B. Set A inlet to A1, and equilibrate line A1. Continue flowing until the effluent reaches a constant UV absorbance. This may require 30–40 mL of feed during the first cycle, but about half that for subsequent cycles. This can be done at 5–10 mL/min. Collect the effluent.

14. Set the in-line dilution factor with the gradient maker. For example, to create an in-line dilution factor of 1:2, set the gradient maker to 66.7% B.

15. Continue flow until UV absorbance stabilizes. This value represents the 100% value from which% breakthrough determinations will be calculated.

16. Reduce flow rate to desired flow rate for the first column to be evaluate.

17. Set the system to simultaneously make a chart mark and put column of choice in line.

18. Continue loading sample until the desired breakthrough value is achieved. Measure the UV absorbance at the lowest point in the breakthrough curve. This equals 0% breakthrough. Calculate the difference in absorbance units between the 0% and 100% values. To determine the value for 5% breakthrough, multiply the difference between 0% and 100% by 0.05, then add this amount to the UV absorbance at 0% breakthrough. Calculate different breakthrough increments, if desired, accordingly. Measure the volume of sample from the chart mark to the point where absorbance reaches the target breakthrough value. Multiply the volume by the product concentration in the diluted feed stream.

19. If additional media are to be evaluated, make another chart mark as the next column is put in line.

20. When the last column has been evaluated, take the column out of line.

21. Stop flow. Set the A inlet to A2 and the gradient to 0% B. Re-initiate flow and collect the solute rinsed from the lines. Continue until UV signal reaches baseline. This can be done at 5–10 mL/min to save time.

22. Set the gradient to 100% B. Switch to inlet B2, and run B2 through the system until conductivity stabilizes at target value.

23. Reduce the flow rate to an appropriate value and put column 1 in line until it has eluted. The shape and width of the elution peak may provide worthwhile characterization of column performance characteristics. Collect the effluent.

24. If a second column is being evaluated, switch column 1 off line, and column 2 in line. Repeat as necessary to accommodate additional columns.

25. With the last column eluted and off-line, set the system (flow path) to inject, which will introduce the guanidine. Continue to flow until a steady conductivity value indicates that guanidine concentration has reached equilibrium. Re-zero the monitor.

26. Put column 1 in line until guanidine has brought UV absorbance to baseline. The size of any peak displaced by the guanidine may provide an indication of retarded elution. If the guanidine peak is substantial it may be worthwhile to pause the system for a period of time then re-initiate flow. Displacement of additional UV absorbance will tend to confirm retarded elution, likely due to slow diffusion of solute from deep within the pore structure of porous particle media.

27. If a second column is being evaluated, switch column 1 off line, and column 2 in-line. Repeat as necessary to accommodate additional columns.

28. Switch sample injection valve back to load.

29. If it is desired to evaluate another set of conditions, for example a different conductivity or pH value, replumb inlet B1 to the new diluent, then repeat instructions 10–29.

30. When all conditions have been evaluated, set the system to 0% B (inlet A2) and flush the system with A2. Pause the system and put the A1 inlet into buffer A2. Flush the system to clear the sample. Collect the effluent.

Figure 1: (Click to enlarge)

Higher capacity values are generally obtained at lower pH because it increases the positive charge on protein amino residues. Capacity also increases with lower conductivity because there is less competition between dissolved ions and proteins for charged groups on the exchanger, but there are limits. When capacity is plotted over a range of conditions, there is usually an inflection point from which capacity decreases with decreasing pH or conductivity. The conductivity at which this point occurs and the intensity of the drop vary considerably from one antibody to the next (Figure 2, 3). The overall phenomenon has been attributed to bound antibodies decreasing the exchanger's surface electronegativity (8,9). It probably also involves antibody solubility because IgGs tend to precipitate at low pH and conductivity (6). Precipitates, or transient aggregates on their way to becoming precipitates, are larger and have lower diffusion constants than single antibodies. Slower diffusion reduces the efficiency of diffusive mass transport into particle pores, reducing binding capacity.

Figure 2: (Click to enlarge)

Figure 3: (Click to enlarge)

With either explanation, the ability to achieve high binding capacities at moderate pH and conductivity values is beneficial. Feed stream dilution or diafiltration requirements will be reduced, and the antibody is more likely to remain fully soluble for the duration of a long sample-loading interval. Some vendors have suggested that their cation exchangers are “salt tolerant.” The term is understood to be relative, because even physiological conductivity is sufficient to prevent high-capacity binding for the majority of IgGs. For example, CX5 exhibits progressively higher capacity than CX3 with increasing conductivity, but by the time it reaches about half-physiological conductivity, capacity is only about 55 mg/mL (Figure 3). The conductivity curves for CX4 and CX5 are more nearly parallel, suggesting similar “salt tolerance” (Figure 2).