Complex biopharmaceuticals, such as recombinant blood coagulation factors, are addressing critical medical needs and represent a growing multibillion-dollar market. For commercial manufacturing of such, sometimes inherently unstable, molecules it is important to minimize product residence time in non-ideal milieu in order to obtain acceptable yields and consistently high product quality. Continuous perfusion cell culture allows minimization of residence time in the bioreactor, but also brings unique challenges in product recovery, which requires innovative solutions. In order to maximize yield, process efficiency, facility and equipment utilization, we have developed, scaled-up and successfully implemented a new integrated manufacturing platform in commercial scale. This platform consists of a (semi-)continuous cell separation process based on a disposable flow path and integrated with the upstream perfusion operation, followed by membrane chromatography on large-scale adsorber capsules in rapid cycling mode. Implementation of the platform at commercial scale for a new product candidate led to a yield improvement of 40% compared to the conventional process technology, while product quality has been shown to be more consistently high. Over 1,000,000 L of cell culture harvest have been processed with 100% success rate to date, demonstrating the robustness of the new platform process in GMP manufacturing. While membrane chromatography is well established for polishing in flow-through mode, this is its first commercial-scale application for bind/elute chromatography in the biopharmaceutical industry and demonstrates its potential in particular for manufacturing of potent, low-dose biopharmaceuticals.

Membrane adsorbers are well established in protein chromatography. The present paper investigated for the first time the behavior of polynucleotides on these stationary phases, taking a 7.2-kb predominantly supercoiled plasmid as example. Gradient and isocratic elution was studied. In contrast to protein high-performance membrane chromatography (HPMC), isocratic elution is possible in DNA chromatography. In the case of gradient elution, much higher salt concentrations can be used in the starting buffer. Under optimized conditions, both approaches led to a splitting of the single plasmid peak into three maximums, which corresponded to the three-albeit isolated-bands in the agarose gel. Presumably the three fractions were supercoiled, nicked, and open circular plasmid DNA. Linearization of the plasmid lowered the adsorption energy, and the linearized plasmid eluted earlier than the nonlinearized one. The HPMC experiments were compared to similar ones performed using a conventional packed-bed anion-exchange column (BioScale Q2, 7 x 52 mm, 10-micron porous particles) and a novel monolithic-type anion-exchange column (UNO Q1, 7 x 35 mm). The results and characteristic differences observed in these experiments were interpreted in the light of the newly developed theory of HPMC.

Continuous chromatographic separations, especially of multicomponent mixtures, constitute interesting options for biotechnological downstream processing. Taking the separation of plasmid DNA from clearified lysates on hydroxyapatite as a pertinent example, we discuss the potential of continuous annular chromatography (CAC) in comparison with conventional (preparative) batch chromatography. In CAC the column is realized in the form of a thin (5 mm, height 210 mm) slowly rotating annulus. The performance of such a CAC column is compared to that of an ("analytical") batch column of similar thickness (diameter) and length (4 x 250 mm) and that of a ("preparative") batch column of similar cross-sectional surface area and height (50 x 210 mm). The quality of the obtained plasmid as defined by the appearance of the corresponding agarose gels (native and linearized plasmid), the 260/280 ratio and the biological activity (transient transfection of HEK 293 cells) was found to be identical in all three cases. The yields are also shown to be equivalent. The loading factor is found to be the most decisive parameter for the transfer of a given separation method between the continuous and the batch columns. Under nonoptimized conditions, plate numbers tended to be lower in the continuous compared to the batch columns. This is shown to be largely due to an artifact created by the CAC design (collection of averaged fractions at the outlets) and can be overcome by optimizing the rotation speed. Surprisingly the large batch column consistently gave better plate numbers than either the small batch or the CAC column. Compared to the preparative batch column, wall effects are more pronounced in the CAC (respectively the small diameter batch column), which may translate into better bed stability but conceivably also contributes to an increase in plate height, due to the reduction in bed density usually observed in the proximity of the wall. The CAC is shown to be a powerful approach to continuous chromatography, which allows a direct and straightforward upscale of chromatographic bioseparation methods.

Annular chromatography represents a crossflow approach to chromatographic separations, that allows the continuous separation of multicomponent mixtures. The potential of the method for continuous bioseparation has been discussed for some time, however, we demonstrate for the first time the processing of a complex feed (cell culture supernatant) taken from an actual (bio)process. Moreover, while previously published applications of annular chromatography concentrated on noninteractive (gel filtration) or nonspecific (ion exchange) chromatography, we show the possibility of continuous annular affinity chromatography. In particular, a commercially available preparative continuous annular chromatography (P-CAC) system was used to purify a recombinant antibody (human IgG(1)-kappa) from CHO cell culture supernatants by (pseudo)affinity chromatography on hydroxyapatite (HA) and rProtein A. Methods developed using small (2 mL) batch columns could be directly transferred to the P-CAC, where they yielded similar results in terms of final product quality. Yields were between 87% and 92% in the case of HA and between 77% and 82% in the case of rProtein A chromatography. DNA removal was nearly quantitative in all cases. Concomitantly, the antibody fraction of the total protein content was raised by one order of magnitude in HA and by a factor of 50 by rProtein A chromatography. In addition, a novel HA material (particle diameter -120 microm) was investigated, which was compatible with expanded-bed applications. However, the final purity of the antibody thus obtained and also the yields (<70%) were less than satisfactory.