Increasing Capacity Utilization in Protein A Chromatography

Currently, most monoclonal antibodies (mAbs) are produced by recombinant cell culture in batch or fed-batch mode. After completion, the fermentation harvest is clarified by filtration and/or centrifugation prior to entering downstream purification. The Protein A affinity chromatography capture step for mAbs has a transit time in the range of 1 to 2 days. The transit time eventually determines the amount of Protein A resin (i.e., the column size) required for the capture step.

Another important factor influencing the required resin amount is the dynamic capacity (DBC) of the resin, which is mainly dependent on resin properties and the loading flow rate. The DBC is determined by the feed volume (more precisely the mAb mass) that can be loaded onto the column until a certain concentration value at the column outlet is reached. In batch capture, typically the column is loaded up to 80-90% of the load volume value that would correspond to 1% of the feed concentration in the column effluent, which accounts for capacity losses of the resin with time (e.g., due to column cleaning). The larger the feed flow rate or throughput, the smaller this breakthrough elution volume is. This tradeoff leaves a large fraction of the total resin capacity—or “static” capacity—unused. Depending on the shape of the breakthrough curve, the mAb concentration profile recorded at the column outlet when the column is loaded beyond its DBC, the fraction of unused capacity can be 50% of the static capacity or even beyond.

Sequential loading processes such as twin-column capture processes aim at improving stationary phase capacity utilization by loading a first column beyond its dynamic capacity and capturing the mAb that is breaking through on a second column. The first column is thereby loaded much closer to its static capacity value.

This context is illustrated in Figure 1 where area A represents the mass that can be loaded on a single column before reaching the 1% breakthrough value. When loading two columns in series until reaching a desired breakthrough value of X% of the feed concentration, the upstream column contains the additional mass corresponding to area B and the downstream column contains the mass corresponding to area D. The total area A, B, C corresponds to the static capacity. Thus the capacity utilization of batch-capture chromatography is A/(A+B+C) and the capacity utilization of a two-column capture process (CaptureSMB, ChromaCon) is (A+B)/(A+B+C) according to Figure 1. Note that the capacity utilization of batch chromatography in reality would be even lower because the elution volume corresponding to 1% breakthrough would be multiplied with the safety factor of 80-90%, which was omitted in the explanation for simplicity.

mAb-capture case study
A case study was carried out for the capture of IgG1 with a titer of 1.1 mg/mL from clarified cell-culture harvest using a twin-column capture process (CaptureSMB, ChromaCon), operated on Contichrom Lab-10 equipment (ChromaCon) with ChromIQ software, and two 0.5-cm i.D. by 10-cm/L columns packed with Amsphere JWT-203 protein A (JSR Life Sciences). The Contichrom Lab-10 included one UV detector mounted at the outlet of each column.

To evaluate process performance in dependence of the throughput, a series of single-column experiments was carried out using maximum feed flow rates of 150-600 cm/h. The protocol for recovery and regeneration included a 5-CV (column volume) wash with binding buffer, a 5-CV wash with 1M NaCl buffer, another 5 CV wash with binding buffer, 5-CV elution with 50 mM Citrate buffer, pH 3.2, cleaning for 15 min with 0.1 M NaOH at a fixed flow rate of 100 cm/h, re-equilibration with 2 CV of elution buffer followed by 3 CV of binding buffer. A startup step with prolonged sequential loading was included to reach a cyclic steady state from the second cycle on.

The product pool concentration was determined using analytical Protein A chromatography (Poros A20 column, Life Technologies). Aggregate content was measured by size exclusion chromatography (TSK-Gel G3000SWXL, Tosoh). Host-cell protein (HCP) values were determined using a CHO-HCP ELISA kit (# F550, Cygnus Technologies). DNA content was quantified by fluorescence (Quant-iT PicoGreen dsDNA kit, Life Technologies).

The UV signals recorded at the outlet of each column are shown in Figure 2 for a run with a maximum feed flow rate of 300 cm/h. The disconnected phases (duration 61 min) and interconnected phases (duration 41 min) are clearly distinguishable. During the interconnected phase, the breakthrough of the upstream column is visible as excursion from the elevated baseline. The elevated baseline is due to the impurities in the flow-through and the excursion to the mAb break-through from the upstream column. The downstream column does not exhibit this excursion indicating that the mAb is fully adsorbed. Process analytical technologies (PAT) for monitoring and process control based on the breakthrough signals (1) are straightforward to implement and have been validated also for twin-column CaptureSMB.

For comparison, a single-column batch process with equal resin volume (0.5 i.D. by 20 cm/L column) was run with a load corresponding to 90% of the 1% breakthrough value and the recovery and regeneration protocol indicated previously.

Table 1: Yield and purity values obtained in case study for batch and twin-column capture (CaptureSMB) runs at maximum feed flow rates of 150-600 cm/h.

Yield

HCP

DNA

Aggregates

[%]

[ng HCP/ mg mAb]

[ng HCP/ mg mAb]

[%]

Batch

87-93

7000-12000

1.0-5.0

2.0-3.0%

CaptureSMB

87-95

5000-12000

2.0-4.0

1.5-2.5%

The results of the experimental series are summarized in Table 1 and indicate a high yield of the mAb at comparable purity (aggregate, DNA and host cell protein [HCP] contents) in batch and twin-column capture process chromatography. The productivity of twin-column capture process is larger than the productivity of batch single-column mAb capture. With increasing flow rates, the dynamic capacity decreases, which has a larger effect on productivity in batch operation than in twin-column capture process operation. Therefore, the relative productivity of twin-column capture process chromatography, with respect to batch chromatography, increases with increasing flow rates. (see Figure 3).

The load in terms of mg mAb per mL of stationary phase before elution is strongly dependent on the feed flow rate in batch chromatography due to the earlier breakthrough at elevated feed flow rates. In the twin-column capture process chromatography, the dependence is smaller since the load of the column to be eluted is maximized during the sequential loading phase. The buffer consumption (L buffer per g mAb produced), therefore, remains almost constant for the CaptureSMB example as the feed flow rate increases.

Performance comparison of 2-, 3- and 4-column setups
Simulations of twin-, 3- and 4-column sequential loading processes were carried out based on break-through curve simulations as described by Godawat et al. (1) and Carta and Jungbauer (2). The parameters for the J-fitting function were determined by fitting of breakthrough experiments at the four different flow rates used in the studies. The simulations were based on the 3- and 4-column configurations described by Godawat et al. (1). The performance of the sequential loading processes was compared with respect to productivity, load, buffer consumption, and product pool concentration. In Figure 4, the productivity simulation results are shown for the different processes for a titer of 3 g/L.

The productivity was computed by calculating the average feed flow rate of the process, multiplying it with the feed concentration and dividing by the total resin volume. The simulations show that the productivity of the twin-column process is larger than the one of the 3-column process and equal or larger than the one of the 4-column process, depending on the time required for recovery and regeneration. The load per cycle of the twin-column process is slightly reduced compared to the 3- and 4-column process while the buffer consumption is slightly increased. The results regarding the productivity are explained in the following section.

Operational aspects of twin-column operation
In most sequential loading processes described in the literature for capture of mAbs using Protein A chromatography, two columns are loaded in series (1, 3-4). Two columns were shown to be sufficient to accommodate the mass transfer zone using protein A stationary phases (5).

Schematic of twin-column capture process Figure 1: Schematic illustration of the twin-column CaptureSMB process. IC: interconnected phase, B: batch phase.

The CaptureSMB (ChromaCon) process comprises interconnected phases I1 and I2, where the columns are loaded and washed sequentially, and batch phases B1 and B2, where the formerly upstream column is washed, eluted, and regenerated. The formerly downstream column is continued to be loaded.

In the subsequent interconnected phase, the regenerated column is placed in the downstream position and the previously loaded column is placed in the upstream position.

The process applies a dual loading flow-rate strategy to optimize the overall process performance. In the interconnected phase, the columns are operated at maximum possible feed flow rate; in the batch phase, the column that previously was in the downstream position is continued to be loaded at a lower flow rate. The lower feed flow rate ensures that no breakthrough of feed from the single column occurs. Moreover, the lower feed flow rate leads to an improved dynamic capacity for the batch-feeding step. The duration of the batch phase is determined by the overall duration of the recovery and regeneration tasks.

The twin-column process in protein A chromatography incorporates the advantages of sequential loading using the minimum required number of columns. The twin-column process is operated by using different loading flow rates in the sequential loading step and the batch step (unless the feed titer is very low). During the sequential washing period following the sequential loading, the feed flow is zero. Thus, the process throughput is calculated using the average flow rate. It is often perceived that continuous manufacturing needs to operate at a constant feeding flow rate; this is not needed if a small reservoir of harvest material is introduced in the process, decoupling harvest flow streams and downstream process flow streams.

Process configurations with three or more columns with a constant feed flow rate have certain disadvantages. In such configurations, two columns are loaded sequentially, and in parallel, a third or subsequent column(s), performs recovery and regeneration tasks. To operate the process at a constant loading flow rate in all steps, the duration of the parallel recovery and regeneration tasks must not exceed the time until breakthrough occurs from the sequentially loaded columns, or the flow rate of the sequentially loaded columns needs to be reduced. This problem and others related to the duration of the individual steps are referred to as “scheduling constraints” (1) being accentuated by an extensively long recovery and regeneration protocol or if the protein titer increases, above 2-3 g/L. By adding more columns, the recovery and regeneration tasks can be shared among more columns, leading to an improved scheduling of the 4-column process compared to the 3-column process for the selected recovery and regeneration protocol.

Generally, the inclusion of more columns in the process has to be approached with care because more columns represent a larger resin volume and the feed stream to the system cannot be increased beyond a maximum value. Thus, deliberately increasing the number of columns when operating at maximum feed flow rate leads to the same amount of feed material processed by a larger number of columns (i.e., resin volume), which decreases the productivity (defined as mass of mAb produced per time and resin volume). Moreover, the addition of extra columns leads to additional hardware complexity and the risk of downtime due to equipment failure.

Conclusion
A  twin-column periodic countercurrent capture process has the advantage of combining an efficient sequential, countercurrent loading process with a minimal twin-column hardware configuration.

References
1. R. Godawat, et al., Biotechnol. J. 7 (12), 1496-1508 (2012).
2. G. Carta and A. Jungbauer, “Chapter 8: Effects of Dispersion and Adsorption Kinetics on Column Performance” in Protein Chromatography: Process Development and Scale-Up. (Wiley-VCH, Weinheim, Germany, 1st ed, 2010). p 253.
3. E. Mahajan, A. George, B. Wolk, Journal of Chromatography A 1227, 154-162 (2012).
4. V. Warikoo, et al., Biotechnol Bioeng. 109 (12), 3018-3029 (2012).
5. M. Angarita, et al. Prep Symposium (Boston, MA, 2013).

About the Authors
Thomas Müller-Späth, PhD,*  Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zürich, Switzerland and ChromaCon AG, Zürich, Switzerland; Monica Angarita, Ms Sc, and Daniel Baur, Ms Sc, Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zürich; Roel Lievrouw, PhD, and Geert Lissens, PhD, JSR Life Sciences, JSR Micro NV., Belgium; Guido Ströhlein, PhD, MBA and Michael Bavand, PhD, ChromaCon AG; Massimo Morbidelli, professor, Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zürich
*To whom all correspondence should be addressed, email: thomas.mueller-spaeth@chromacon.ch