Monoclonal antibodies (MAbs) are expected to dominate the future biopharmaceutical landscape (1). Although chromatography is the mainstay of MAb downstream purification processes, more efficient and cost-effective chromatography technologies are needed to address increasing MAb titers in mammalian cell culture supernatants.

Ion-exchange membrane chromatography has been demonstrated to capture large biomolecules such as viruses and plasmid DNA with higher dynamic binding capacity than ion-exchange beaded column chromatography (2, 3). Recently, anion-exchange membrane chromatography has been viewed as a viable alternative to column chromatography for trace contaminant removal from a protein A-purified MAb (4, 5).

Here, a 0.35-mL membrane volume (MV) Mustang® Q coin was evaluated for Chinese hamster ovary (CHO) host-cell protein (HCP) and host-cell DNA removal from a protein A purified MAb. Spiking a protein A-purified MAb with CHO DNA showed more than 4.9-log removal at pH 8.0 and 4 mS/cm. It was found that the best HCP clearance was accomplished at pH 8.0 and 4 mS/cm conductivity when the HCP levels were reduced from 208 ng/mg of MAb to 4 ng/mg of MAb.

Materials

All flow-through chromatography testing with Mustang Q coins (Pall Life Sciences, MSTG18Q16) in a coin housing (MSTG18H16) was performed using an ÄKTA™ Explorer 100 from GE Healthcare. The load material was a recombinant humanized antibody produced in CHO cell supernatant and purified by protein A column chromatography. CHO DNA was isolated from a CHO cell line using tangential-flow filtration. The antibody concentration was received at either 3.1 mg/mL or 2.0 mg/ml in 3 mM citrate, 25 mM Tris pH 7.5 and 0.1 M NaCl with a measured (Thermo Orion, model number 130A) conductivity of 11 mS/cm at 20 °C. The Mustang Q membranes used in this study were in the coin (0.35-ml membrane volume) format. To reduce the risk of feedstock aggregation, the high flow-rate capability of Mustang membranes was used, with a flow rate of 10.5 ml/min (30 MV/min) to load the antibody process feed through the Mustang Q coin in its housing. A slower flow rate of 10 MV/min may be used.

In-line prefiltration upstream of the Mustang Q coin was performed using a 47-mm, 0.2-µm Supor® polyethersulfone (PES) membrane (Pall life Sciences part number 60301) in a stainless steel housing (FTK 200, Pall industrial). For large-volume feedstreams (>50 ml), an alternative 0.2-µm Mini Kleenpak™ (part number KM5EDFP2S) in-line prefilter with 20 cm² may be used. Prefiltration of the MAb feedstreams was also performed before loading as shown in Figure 1, with either a 25-mm 0.2-µm Supor Acrodisc® (part number 4612) for volumes of ≤40 mL or a 0.2-µm Supor Acrocap® (part number 4480; alternatively, KM5EKVP2S) for volumes of 100-200 mL.

Figure 1.

The following loading buffers were tested, and all conductivities were adjusted using 5 M NaCl solution:

• 25 mM BisTris hydrochloride [Bis(2-hydroxyethyl)amino-tris(hydroxymethyl) methane] pH 6.5 with 4 mS/cm conductivity

• 25 mM BisTris hydrochloride pH 6.5 with 11 mS/cm conductivity adjusted with 5 M NaCl solution

• 25 mM Tris pH 8.0 with 4 mS/cm conductivity

• 25 mM Tris pH 8.0 with 11 mS/cm conductivity.

Methods

The CHO protein levels were measured by an enzyme-linked immunosorbent assay (ELISA) with a high-sensitivity commercial kit from Cygnus Technologies (catalog number F015). According to the manufacturer, the lower limit of detection for this kit is <100 pg/mL, and the lower limit of quantitation was ∼0.7 ng/mL. The CHO DNA levels were measured by quantitative PCR at Cogenics, inc.. Commercially available CHO DNA (Cogenics) and DNA isolated from a CHO cell line were used by Cogenics to obtain calibration curves for different samples. The MAb concentration was calculated from A₂₈₀ readings using the conversion factor 1.5 mg/mL/cm.

The Mustang Q membrane coin was preconditioned with 10 MV of 1M NaOH, 10 MV of 1 M NaCl, and finally with the loading buffer until the effluent pH and conductivity were the same as that of the loading buffer before loading the sample. A 3 M BisTris solution or a 3 M Tris base solution was added to the MAb solution to either reduce pH to 6.5 or increase pH to 8.0, respectively, before loading on the Mustang Q coin. For cases in which reducing conductivity to 4 mS/cm was necessary, the MAb feedstream was diluted with water and buffer concentration adjusted with the required volume of either 3 M BisTris for pH 6.5 or 3 M Tris for pH 8.0 base solution. All MAb feedstreams were filtered before loading on the Mustang Q membrane using either a 25-mm 0.2-µm Supor Acrodisc for volumes of ≤40-mL or using a 0.2-µm Supor Acrocap for volumes of 100-200 mL.

The MAb feedstream pH-adjusted to 8.0 and diluted to 4 mS/cm was spiked with 7 mg of CHO DNA (0.7 mg/mL) before processing through a Mustang Q coin to measure CHO DNA removal from a protein A-purified MAb.

Following the sample load, the Mustang Q coin was washed with a loading buffer at the appropriate pH and conductivity until it’s A₂₈₀ reading reached the baseline. This wash was combined with the flowthrough, and the combined fraction was called the flowthrough fraction. A 4-mL aliquot of the feedstream and the Mustang Q flowthrough were saved for HCP and CHO DNA assays. Following completion of loading and the loading buffer wash, the Mustang Q coin was cleaned with 20 MV of 1 M NaCl in the appropriate loading buffer at 10.5 mL/min (30 MV/min), followed by 20 MV of 1 M NaOH.

Results and Discussion

The residual impurity clearance capacity in an antibody solution from a chromatography resin or membrane is typically determined from spiking experiments. Table 1 shows that when the antibody sample at pH 8.0 and 4 mS/cm was spiked with 7.0 mg CHO DNA (324 ng/mL DNA concentration) and processed through the Mustang Q membrane, the amount of DNA removed by the membrane corresponded to 4.9 log. This may not necessarily be an upper limit of DNA clearance because the spike load was limited by the amount of CHO DNA that was available. To validate the upper limit of DNA clearance, a higher spike load may be used.

Table 1. Quantitative PCR analysis of residual CHO cell DNA on CHO DNA–spiked antibody sample loaded on Mustang Q membrane coin

Residual CHO protein impurities in the Mustang Q membrane- processed samples were measured by ELISA assay. Table 2 shows that without any pH or conductivity adjustments (pH 6.5 and 11 mS/cm), about a sevenfold reduction in CHO proteins was observed in the antibody solution. However, when the conductivity was lowered to 4 mS/cm while the pH was held constant, a 52-fold CHO protein reduction was observed. Antibody recovery was calculated from the Mustang Q membrane in the flowthrough and wash fractions based on A₂₈₀ measurements.

Table 2. CHO cell protein analysis in Mustang Q coin flowthrough (FT) and wash

The CHO protein clearance was significantly higher at pH 8.0, 16 ng/mg, and 4 ng/mg (ppm) at 11 mS/cm and 4 mS/cm conductivities respectively. Thus, Table 2 suggests that the optimum loading conditions for CHO protein clearance with Mustang Q membranes is at pH 8.0 and 4 mS/cm conductivity.

Because contaminant levels are much lower (by more than three orders of magnitude for both DNA and proteins) than the dynamic binding capacity of the membrane, efficient use of the membrane for binding these contaminants can be made by scaling up based on antibody throughput (grams of protein processed per unit membrane volume and time). Based on results presented here, the next step would be to develop a scaled-down flowthrough process on a Mustang Q coin that would provide a maximum throughput in terms of grams of MAb per liter of membrane flowthrough at pH 8.0 and 4 mS/cm. An adequate safety margin may be defined for further scale-up. Furthermore, depending on the eventual scale, the flow-rate may need to be optimized because some chromatography systems may not be able to support flow rates of 30 MV/min with larger-scale devices. Once the throughput has been determined, an appropriate Mustang Q membrane capsule size may be selected to process a given antibody batch size that will be generated from the protein A chromatography step.

Mustang Q membrane loading at pH 8.0 with 4 mS/cm (threefold dilution of the sample) in 25 mM Tris buffer resulted in reduction of CHO cell proteins to 4 ppm. Purification of a CHO DNA-spiked MAb solution using a Mustang Q coin in the flowthrough mode at pH 8.0 and 4 mS/cm conductivity showed a 4.9-log reduction in CHO DNA.

REFERENCES