Microscale to manufacturing scale-up of cell-free cytokine production--a new approach for shortening protein production development timelines

Abstract

Engineering robust protein production and purification of correctly folded biotherapeutic proteins in cell-based systems is often challenging due to the requirements for maintaining complex cellular networks for cell viability and the need to develop associated downstream processes that reproducibly yield biopharmaceutical products with high product quality. Here, we present an alternative Escherichia coli-based open cell-free synthesis (OCFS) system that is optimized for predictable high-yield protein synthesis and folding at any scale with straightforward downstream purification processes. We describe how the linear scalability of OCFS allows rapid process optimization of parameters affecting extract activation, gene sequence optimization, and redox folding conditions for disulfide bond formation at microliter scales. Efficient and predictable high-level protein production can then be achieved using batch processes in standard bioreactors. We show how a fully bioactive protein produced by OCFS from optimized frozen extract can be purified directly using a streamlined purification process that yields a biologically active cytokine, human granulocyte-macrophage colony-stimulating factor, produced at titers of 700 mg/L in 10 h. These results represent a milestone for in vitro protein synthesis, with potential for the cGMP production of disulfide-bonded biotherapeutic proteins.

Figure 1

Protein synthesis activity is dependent on the time and temperature of extract pre-incubation. a: Aliquots of the cell-free extract were incubated at various times and temperatures (▪: 37°C, •: 30°C, and ▴: 25°C) to optimize the cell-free synthesis activity. b: Aliquots of the cell-free extract prepared using a tubular bowl centrifuge (•) or from cells harvested with large scale disc-stack continuous centrifugation, followed by lysate clarification in a tubular bowl centrifuge (○) were incubated at 30°C for various times. The concentration of rhGM-CSF produced after 5 h in a 60 µL scale reaction from these various pre-incubation conditions was determined by [¹⁴C]-leucine incorporation as described in the Materials and Methods section. CVs were ≤10% and the protein samples were ≥95% soluble for all time points.

Figure 2

Yields of soluble rhGM-CSF for several gene sequences designed using (a) ProteoExpert 5′ mRNA optimization or (b) DNA 2.0 codon bias algorithms as described in the Materials and Methods section. The expression of soluble rhGM-CSF was measured by [¹⁴C]-leucine incorporation in 250 µL. CVs were ≤10%, for all constructs tested. The gene sequence of GM-CSF1 (Supplementary Methods) was used to optimize cell-free expression in scale-up experiments.

Figure 3

Combinatorial optimization of protein folding in the OCFS system. Response surface of redox potential and DsbC concentration on soluble expression of rhGM-CSF. The expression of soluble rhGM-CSF was measured by [¹⁴C]-leucine incorporation in a 96-well plate for IAM-treated extract as a function of the initial redox potential of added [GSSG]_tot = 5 mM and DsbC concentration. For reference, the redox potential of the E. coli cytosol is −270 mV.

Figure 4

The rate of soluble rhGM-CSF production is independent of scale from plates, to stirred tank reactors, to large-scale bioreactors. The initial slopes corresponds to a translation rate of ∼1 peptide bond per second per ribosome. The concentrations of rhGM-CSF were determined by [¹⁴C]-leucine incorporation at 250 µL scale, and by RP HPLC analysis at larger scales, as described in the Materials and Methods section. CVs were ≤10% and the proteins were ≥95% soluble for all time points. Lines are shown for visual purposes only.

Figure 5

Analysis of process purification steps used in the production of pharmaceutical-grade rhGM-CSF. a: Process flow diagram for protein purification (b) Cell-free reaction product pool was visualized, using non-reducing SDS–PAGE, by ¹⁴C autoradiography of leucine incorporation at 5 h that measures only protein produced, or by Sypro staining of all proteins in the extract at 10 h. Autoradiography indicates incorporation of ¹⁴C-leucine corresponding to the molecular weight of rhGM-CSF and confirms that only rhGM-CSF was produced because the T7 RNAP limits transcription to the T7 promoter on the plasmid added to the reaction, with no evidence for aberrant protein products produced from native RNAP transcription. c: Non-reducing SDS–PAGE of the process purification pools. Lane 1: Anion exchange pool. Lane 2: Tangential flow filtration pool. Lane 3: SEC final product pool.

Figure 6

rhGM-CSF is correctly folded and functional. a: Liquid chromatography–Mass spectrometry analysis of intact mass of 14604.78 Da corresponds to the expected average neutral mass of 14604.75 Da. Two minor product related impurities were observed corresponding to removal of the first two amino acids (-M and -MA). These peak intensities do not correspond accurately to the relative amount of product related impurities. b: Base peak chromatogram of a V8 protease digestion of rhGM-CSF. The two peptides with disulfide bridges labeled V7 = 9 and V8 = 11 (underlined) coincide with the theoretical masses with correct disulfide bridge formation. The V7 = 9 linkage was confirmed by tandem MS and the V8 = 11 by accurate mass. Other observed V8 digest peptides are marked accordingly. Digest peptides denoted with a plus sign indicate a missed cleavage. c: Analytical size exclusion chromatogram of purified rhGM-CSF on a Tosoh TSKgelG3000SWXL column, flow rate 0.5 mL/min. d: The effective dose 50, ED₅₀ <0.1 ng/ml of rhGM-CSF induced proliferation of human TF-1 cells, corresponds to a specific activity of 14 × 10⁶ IU/mg, comparable to commercially available E. coli-derived rhGM-CSF protein as a positive control.