June 15, 2012 (Vol. 32, No. 12)
Angelo DePalma Ph.D. Writer GEN
In small molecule drug development, it is often said that the low-hanging fruit has already been harvested. Most structures that are chemically accessible, manufacturable, and druggable have already been considered.
Somewhat analogously, the leading edge in biologics development are proteins that for one reason or other present development or manufacturing challenges. Reasons vary, from difficult transfection or poor expression, to low yields or high losses on purification.
Among the speakers at CHI’s recent “PEGS” conference was Anne London, Ph.D., investigator II and lab head at the Novartis Institutes for BioMedical Research, who spoke on evaluating and streamlining platform approaches to preclinical protein production.
Preclinical protein production is a broad discipline, with expectations and goals differing from project to project. Rapid, efficient production are common goals, but challenges arise with differing maturity of programs (e.g., initial antigen production to late-stage in vivo material) and quality specifications. Dr. London presented case studies reflecting her group’s approach to arriving at a platform capable of adapting to fit production requests, with emphasis on cell-line evaluation and impurity clearance.
“The goals for each production vary based on the protein’s intended use,” she said. Live rabbit studies require a large amount of “clean” protein with very low endotoxin levels. By contrast, endotoxins will not affect proteins used in affinity binding assays, where buffer choice and concentration may be more relevant. “Fully understanding the product’s intended use is critical to guide production to the defined endpoint.”
Preclinical production platforms are “much simpler” than a GMP biologics manufacturing platform because they entail fewer polishing steps and less validation. For example, viral clearance is not required at the preclinical stage, so virus removal and inactivation are not explored. “With the goal of simply making enough protein for the intended use, only steps needed to grow, capture, purify, and deliver the protein are considered at the preclinical stage,” Dr. London explained.
A recurring production theme is low endotoxin material. In one case with the production of a His-tagged protein, endotoxin levels were too high (>1.0 EU/mg) for the intended use after standard purification.
“Guaranteeing product quality below 1.0 EU/mg is difficult, if next to impossible, in a non-GMP setting.” The situation resulted in the evaluation of multiple endotoxin-removing membranes and filters, with the final solution being a detergent extraction. Dr. London reports spending “many hours” troubleshooting this process, but eventually the product met quality specifications.
Project scaleup is another issue facing early-stage protein production. Dr. London’s group encountered an antibody that was evaluated at small scale in multiple transient cell lines before scaling into a Wave bioreactor.
“After picking the highest expressing cell line at the small scale, the culture was scaled up in the Wave with poor results; the expression levels in the flask did not scale linearly into the Wave.” This production required further process development at Wave reactor scale to meet the expression levels seen in shake flasks. “Not every project will scale up well, so time must be budgeted to allow for cell culture process development in these rare cases,” Dr. London warned.
An Unusual Phenotype
Haruki Hasegawa, Ph.D., senior scientist at Amgen, presented an investigation on the role of secretory cargo in shaping maximum protein secretory capacity. He began by questioning the nature of a cell’s maximum secretory capacity: Is it simply due to the collective contributions of intrinsic cellular machinery?
Dr. Hasegawa studied an unusual CHO phenotype that, under optimized growth conditions, produced a model human IgG clone as rod-shaped crystals within the endoplasmic reticulum (ER) lumen. Crystal growth was accompanied by cell enlargement and multinucleation, and continued until crystals outgrew cell size and breached the cell membrane.
In this phenotype the efficiency of IgG protein synthesis and oxidative folding exceeded the capacity of ER export machinery. As a result, export-ready IgG accumulated in the ER lumen until a threshold concentration was reached to nucleate crystals. “Using an in vivo system that reports accumulation of correctly folded IgG, we showed that the ER-to-Golgi transport steps became rate-limiting in cells with high secretory activity,” Dr. Hasegawa said.
Do techniques exist for altering physicochemical properties of secretory cargo generally, for expression systems other than this phenotype? “There are two major ways to alter the physicochemical properties of secretory proteins,” Dr. Hasegawa explained. The first is to alter post-translational modifications, the second is changing the protein’s primary sequence through mutagenesis. “These two approaches are widely employed by protein engineers using various model organisms and any type of protein.”
Commenting on this work on the Faculty of 1000 website, Dr. Jesse Hay at the University of Montana wrote, “What this story tells us is that a CHO cell’s secretory capacity is not inherently limited by its ability to synthesize, oxidatively fold, and assemble functional product—it is limited by its ER export machinery.”
Gene Fusion Method
Morten Nørholm, Ph.D., senior scientist at the Novo Nordisk Foundation Center for Biosustainability at the Technical University of Denmark, presented a talk on improved expression of a poorly expressed membrane protein by fusing its gene with that of an N-terminal peptide.
Dr. Nørholm could mimic this effect by re-engineering the 5´ end of the membrane protein with “favorable synonymous mutations.” This effect is surprisingly large compared to substitutions at other locations within the gene.
Gene fusion was conducted in high-throughput fashion using a versatile molecular cloning technology, “uracil-excision”. This PCR-based cloning technology was invented 20 years ago but has been neglected due to incompatibilities with modern PCR techniques.
In 2006, Dr. Nørholm’s group published a paper in Nucleic Acids Research identifying a compatible proof-reading DNA polymerase. “Today the technology is heavily used, open-source, simple, and inexpensive,” Dr. Nørholm explained. “We hypothesize that the peptide simply replaces a suboptimal 5´-end and/or N-terminal peptide sequence of the native construct. It is not rocket science, but a very simple and useful tool indeed.”
How generally applicable is this approach to nonmembrane proteins? “We work almost exclusively with membrane proteins, so it is hard for me to say. But the effect was similar for the few soluble proteins we tested,” Dr. Nørholm explained. “The expressing organisms—bacteria—may be part of the magic due to the strict architectural constraints of the ribosome binding site in relation to the 5´ end of the coding sequence.” However, as Dr. Nørholm pointed out, there have been reports describing optimized species-independent translational leaders for cell-free expression.
Peter Rhode, Ph.D., R&D vp at Altor Bioscience, spoke on maximizing production of interleukin-15. Touted as a possible curative treatment for cancer and viral diseases, IL-15 is poorly expressed in bacterial and animal cells. Production in mammalian cells, explained Dr. Rhode, is controlled at the levels of gene expression, translation, and secretion.
“Researchers have tried optimizing codons and altering genetic elements, as well as fusing with albumin, but those approaches have not improved expression that much.” The National Cancer Institute has produced GMP-scale quantities of IL-15 in bacterial systems. The protein is excreted in inclusion bodies, which requires re-folding. But issues with protein deamination and amino acid additions have complicated large-scale production.
Altor scientists found that co-expressing an IL-15 superagonist variant with a soluble IL-15 receptor alpha-IgG1 Fc (IL-15Rα) fusion molecule leads to fully active IL-15:IL-15Rα complex in high yield from CHO cells.
Compared with the “bare” cytokine, the IL-15:IL-15Rα complex is far more active. “It is the substance that immune cells produce,” noted Dr. Rhode, “and how IL-15 is presented in the body.” The superagonist IL-15 variant alone is as much as 10 times as active as the native cytokine. When combined with IL-15Rα, potency and pharmacokinetics improve over 20-fold.
The receptor also acts as a molecular chaperone vital to proper expression and secretion of IL-15. “The limitations we’ve observed with IL-15 are likely due to the lack of this ‘partner’ molecule in animal expression cell lines. We recognized early in our development plans that co-expressing IL-15Rα could assist both in expression and efficacy.”
The IL-15 complex is a potent molecule dosed at microgram/kg compared with multiple mg/kg for monoclonal antibodies. According to Dr. Rhode a “modest-scale GMP run will provide enough material to treat several hundred patients.”
Tools of the Trade
Sabine Geisse, Ph.D., also from Novartis, is director of nuclear localization signal technology. She discussed alternatives to HEK293 and CHO cells for transient protein expression. Derived from human amniocytes, Dr. Geisse’s pet line, CAP-T®, “clearly enhances chances of success” when added to the repertoire of cell lines amenable to transient protein production.
The line’s developer, Cevec, claims “two weeks from gene to milligram or gram amounts of protein” for early-stage studies. CAP-T has an impressive scientific resume as well, with heavy testing and support from European academics.
What makes CAP-T cells special is that they are derived from nontumor human amniocytes. Very few nontumor human immortalized cells are able producers when transfected transiently. Dr. Geisse described the cells as “fast, cost-effective, efficient, stable with high-level expression, and highly reproducible.” Her group uses the cells to generate antigens, early antibody candidates, and any type of tool protein and that requires secondary modifications deferred by mammalian cells in the context of research protein production.
“The benefits, relative to CHO and HEK, are biological characteristics that can lead to higher expression levels for some proteins, or the ability to produce candidate proteins that do not express well in HEK293 and CHO cells.” The drawback, which Dr. Geisse terms “temporary,” is that CAP-T requires “additional fine-tuning and modifications to be user-friendly and applicable on large scale.”
Gregory T. Bleck, Ph.D., R&D platform lead at Catalent Pharma Solutions, described an experiment comparing his company’s GPEx® expression and cell-line engineering technology and a more traditional expression in CHO cells.
Catalent scientists produced GPEx cell lines for one easily expressed protein and three difficult molecules that were either poor producers or exhibited subpar cell-line stability. The “control” process was the in-house platform of a Catalent pharmaceutical partner. The pharmaceutical partner compared cell lines developed by GPEx to cell lines developed using their internal system.
Investigators did not observe significant differences between the two methods for the easy-to-express antibody. For the other three proteins (two antibodies and a fusion protein), GPEx outperformed their standard system in both productivity and stability.
Why the difference? “GPEx seems to result in more consistently, high-expressing, stable cell lines than other processes,” said Dr. Bleck. “We have not elucidated the exact mechanism, but we have never encountered a protein that we could not produce at least as well as a more traditional expression system.”
One possible explanation for GPEx success with difficult proteins is that unlike standard cell-line engineering, which inserts multiple copies of a gene at one location, GPEx introduces many individual gene copies in different locations on the host genome. GPEx also has the reputation for generating extremely genetically stable lines that express proteins reliably and efficiently. But in this particular study, Dr. Bleck admitted, “the reasons for GPEx superiority might be different for each protein.”
Habib Horry, Ph.D., strategic marketing manager at Polyplus-Transfection, discussed the advantages of next-generation polyethyleneimine (PEI) in transient protein expression. A cationic polymer, PEI promotes cell attachment to vessels. It is also an extremely efficient transfection agent, causing DNA to condense into positively charged particles, attach to cell surfaces, and enter through endocytosis. Once inside the cell the DNA finds its way to the nucleus and into the cell’s protein-making machinery.
During the mid-1990s, PEI became the second significant transfection agent discovered (the first being poly-L-lysine). A spinoff from a Strasbourg, France, research organization, Polyplus has since 2001 held two exclusive patents on the use of PEI for transfection of CHO cells. One patent covers the U.S., the other the rest of the world. Using PEI for transfection requires a license from Polyplus.
“The advantage of PEI is that it is efficient and low cost,” Dr. Horry said. But not every company employing the reagent for this specific use is aware of Polyplus’ exclusive market position.
PEI produces transiently transfected CHO cells that churn out between 10 and 100 mg of protein per liter of cell culture—a volumetric productivity that until about a decade ago was considered good for stably transfected cells. Using the reagent, companies can produce a few grams of protein suitable for characterization and preclinical studies. Most developers have settled on a process based on stable transfection before a molecule enters human testing.
Despite great strides in transient transfection, the technique remains somewhat disconnected from the eventual production-scale process. “Techniques learned at this stage are not directly applicable to stable clones,” Dr. Horry observed. “This still represents a gap in the development of protein therapeutics.”