Recent successes of recombinant proteins as therapeutics have greatly increased the number of companies developing such therapies. However, protein therapeutics and vaccines are frequently difficult to develop and manufacture. Non-natural (e.g., antibody fragments), chimeric, bacterial toxin-derived, or fusion proteins present even greater challenges to produce a quality product that is soluble and active as expressed.
Multiple factors can affect product yield and quality—protease degradation, protein folding/inclusion body formation, efficient secretion, and correct disulfide bond formation. Finding the best expression host quickly is critical for speed to the clinic. Microbial expression systems are particularly amenable to rapid host strain development because of the rapid doubling time of these organisms, available genome information, reliable molecular biology tools, and their ready scalability for manufacturing.
The Pfenex Expression Technology™ platform, a Pseudomonas fluorescens-based system for the expression of recombinant proteins, has been developed by DowpharmaSM (pharma.dow.com) for the expression of therapeutic proteins. Based on a systems biology approach, the company assembled useful host strains to enable proper protein folding, secretion, and/or disulfide bond formation. These, together with a set of compatible plasmids with a variety of transcription and translation regulators, as well as a collection of periplasmic secretion signals, enable construction of hundreds of host strains and gene-expression strategies to be tested in parallel to identify the optimal expression strain for a particular recombinant protein without prior knowledge of inherent limitations to a particular target protein’s accumulation as an active entity.
Development of robust, high cell density growth conditions in 96-well format was also required for efficient screening of the multitude of host strain and expression strategy options available on the Pfenex platform.
Small-scale Strain Evaluation
It is important that cell growth and expression during the strain-development stage be scalable. Scaled-down growth of P. fluorescens in 96-well plates was developed as a simple process, using a defined minimal salts medium without animal-derived components. Antibiotics are not required for plasmid maintenance during strain development steps or scale-up, as plasmids are maintained by complementation.
Growth of P. fluorescens in 96-well plates (0.5-mL culture volume in a 2-mL deep well block) performed in standard shakers with no oxygen enrichment was found to reach cell densities similar to that obtained in shake flask culture, usually 30–50 optical density units (Figure 1).
The ability to achieve such high cell densities allows for production and recovery of active protein rapidly at small scale, which can be advantageous for screening activity of multiple proteins as well as for expression of a particular recombinant protein in an array of P. fluorescens host strains. High-throughput transformation and expression of several recombinant therapeutic proteins can be readily performed using an array of host strains.
As an example, human gamma-interferon (g-interferon), which is typically expressed as insoluble protein in E. coli unless fused to another protein such as NusA, was expressed in 90 P. fluorescens host strains. Previously it was shown that soluble, active g-interferon can be expressed in wild type P. fluorescens. Within two weeks, strains were transformed with a g-interferon-encoding plasmid and analyzed for the expression of soluble target protein.
The soluble portion of cell extracts was analyzed in high-throughput format using 96-well SDS-CGE. This tool is useful as a first-line method to estimate soluble target protein expression and to rank the productivity of a number of strains. Several host strains were identified that resulted in improved soluble g-interferon expression when compared to soluble g-interferon expression in the wild type strain (Figure 2).
A variety of host strains identified showed significant improvement of g-interferon expression. We observed that growth and expression results obtained through high-throughput expression analyses are typically reproducible at the 20-L fermentation scale. That is, strains exhibiting good growth and expression of recombinant target protein at the high-throughput scale generally produce high levels of quality, active protein when grown at the 20-L fermentation scale. The ability to screen multiple host strains in high-throughput format and scalability of those results greatly reduces expression strain development time, allowing faster progression of a therapeutic protein to the clinic.
Forming the Correct Disulfide Bonds
In addition to screening multiple host strains, it is important to evaluate multiple expression strategies. Many therapeutic proteins contain disulfide bonds, which in gram-negative bacteria such as P. fluorescens are formed in the periplasmic space, a subcellular compartment between the inner and outer membranes of P. fluorescens.
As part of the Pfenex Expression Technology toolbox, several secretion signals were identified and characterized that can be fused to the target protein, directing it to the periplasm. These have been identified by bioinformatic means and then functionally tested for their ability to direct recombinant protein secretion and faithful leader removal by the enzyme signal peptidase.
Formation of correct disulfide bonds is important for protein activity, but an added advantage of secretion to the periplasm is ready recovery of enriched target protein. Most host-cell proteins—contaminants in therapeutic production—are contained within the bacterial cytoplasm. A highly enriched fraction of target protein can be readily isolated using one of a variety of scalable periplasmic release techniques, reducing the number of steps required for purification, without cell disruption.
Like many other aspects of a biological system, the nature of secretion to the periplasm is quite complex. Screening multiple secretion leaders to find the best leader-target protein fusion is essential to identify an expression strategy that provides not only target protein in the periplasmic space with the secretion leader precisely removed, but also high yield of the target protein.
High-throughput expression analysis at the 96-well scale allows rapid testing of many secretion leaders in combination with multiple host strains and expression strategy to identify the optimal combination of both for each target protein. The ability to identify the optimal regulatory elements and expression host for target protein production using high-throughput methods enables rapid progression to clinical trials. A protein produced using Pfenex Expression Technology is now in Phase II trials after completing the safety portion of a Phase I study in 2006.