Jeffrey Perkel has been a scientific writer and editor since 2000. He holds a PhD in Cell and Molecular Biology from the University of Pennsylvania, and did postdoctoral work at the University of Pennsylvania and at Harvard Medical School.
There’s no mystery to creating protein in the laboratory. Molecular biologists have been introducing coding sequences to an in vitro transcription/translation system, or expressing these constructs in bacterial or eukaryotic cells, for a long time. Usually, you can obtain enough material to work with. On a small scale, anyway.
Sometimes, though, researchers need to scale up their operations beyond the microgram or milligram scale, for instance to advance a molecule into preclinical development. At that point, things can get dicey. It’s generally not practical to simply scale up small-scale conditions—culture volumes and reagent requirements alone can make this prohibitive. Here we provide some guidance on scaling up your target-protein production.
Start with the host
Optimization can take months of painstaking optimization, as multiple variables—vector design, cell line, growth conditions and more—must be considered. Still, there are strategies researchers can use to ensure they squeeze every last drop from their expression cultures.
The first consideration in scaling protein production is perhaps the most obvious: the host itself.
Mammalian cells, such as Chinese hamster ovary (CHO) cells, are the most likely to produce functional eukaryotic proteins, but researchers can also grow them in bacteria or plants.
Bacteria are relatively inexpensive, grow rapidly and yield large amounts of protein, says Jianjun Wang, a professor of biochemistry and molecular biology at Wayne State University School of Medicine. On the other hand, he says, “bacteria do not have complicated cellular machinery,” so the resulting proteins lack most mammalian post-translational modifications and proper folding because of missing disulfide bridges.
Plants offer the advantage of relatively low cost compared with bacterial and mammalian cells, says Somen Nandi, managing director of the Global HealthShare Initiative at the University of California, Davis—five to 10 times cheaper in some cases, he says. Plus, they carry no human pathogens, and if desired, recombinant seeds can be easily stored and recoveredon an as-needed basis. But scaling plant seed-based productions (depending on the species and method of protein production) may require additional arable land, which can present problems in space-constrained environments.
Efficient expression
According to Pranhitha Reddy, a consultant who has worked at both Amgen and Seattle Genetics, the trick to efficient protein scale-up is to consider the entire process: from host cell to expression vector to growth conditions to harvest.
One common vehicle for protein production is CHO, an adherent cell line available from the ATCC. But most companies adapt the line for their purposes, Reddy says, by acclimating them to growth in serum-free media, for instance, or engineering in a single site of gene integration for stable cell transfection and expression. In the latter case, “You are guaranteed a certain level of expression without intensive screening,” she explains. “You are targeting a ‘hot site,’ i.e., a transcriptionally active site.”
Single-site integration offers another benefit, adds Andreas Castan, staff scientist for bioprocess research and development at GE Healthcare's Life Sciences business: simplified screening. “When you target your gene of interest in one location, then you only have to separate transfected from nontransfected cells, and that will save you more than a month compared to randomly transfecting cell lines,” he says. “So that is a technique that is becoming more and more widespread in the industry.”
Likewise, large-scale expression vectors often differ from the standard reagents found in many academic labs, Reddy says.
To maximize output, researchers should consider every element of the expression construct, including the promoters and enhancers that drive expression, codon usage, splice junctions, insulators, signal sequences and more.
Maximizing growth
Wang grows his proteins in bacteria using a customized pET30 vector derivative featuring a shortened polyhistidine purification tag and a short proteinase (Factor Xa) cut site to liberate the protein from the tag.
Growth media for protein production should also be rich, Reddy says—rich enough to support 20 million cells, at least, per milliliter of culture. Overall, a healthy mammalian system should be able to produce at least 20 pg to 30 pg of protein per cell per day, she says; this is a “specific productivity” that is “on the lower end, but still acceptable” for academics to target when considering moving a product into the clinic.
Most large-scale mammalian protein-expression systems are based on stable transformation, says Reddy. But plant researchers often use transient transfection, Nandi says, via recombinant Agrobacterium infection of the plant leaves or cells. In either event, however, researchers need to screen for the highest producers. “It’s a numbers game,” Nandi says. “If you have 30 lines, you have 30 choices; if you have 3,000 lines, you have 3,000 choices to find the highest-expressing line. You have to screen them carefully.”
To get a sense of the depth of screening required to move all these tumblers into place and unlock the protein production of your chosen cells, consider a 2009 paper by Wang, demonstrating a method for dramatically improved protein expression in E. coli—up to about 0.7 gm per liter of culture using a standard shaking incubator [1]. “That’s huge,” he says. “If you grow protein in mammalian cells, if you can get 1 mg, you will be very lucky. You normally talk about micrograms.”
Among other things, Wang and his team altered growth conditions to ensure the cells stayed in the exponential growth phase rather than in the stationary phase prior to IPTG induction. They also used a “double colony selection” approach to ensure maximal protein expression by the chosen clone, improved oxygenation and ran time courses to maximize yield. In the end, they were able to produce 34 mg of unlabeled human apolipoprotein AI per 50 ml of culture.
But that’s bacteria. Eukaryotic cells are far more complicated. The physical scale-up process itself can present a bottleneck, says Castan, as researchers want to evaluate large numbers of clones under process-relevant conditions, whether in tube bioreactors, deep-well plates or micro-bioreactors. As a result, many researchers opt to outsource the work to a contract research organization or least collaborate with a more knowledgeable partner, he says.
Researchers can perform initial optimization steps in-house, Nandi says, such as identifying good chromatography media for protein purification. But beyond that—between creating and screening stable cell lines, optimizing vectors and developing a scalable, reproducible, GMP-compatible process—many researchers simply punt. The process is both expensive and laborious, Reddy notes, and few academics have either the resources or the expertise to handle it appropriately. “It’s a long process,” she says. “You can’t just take [a protein from an academic lab] and put it in the clinic.”
Reference
[1] Sivashanmugam, A, et al., “Practical protocols for production of very high yields of recombinant proteins using Escherichia coli,” Protein Science, 18:936-48, 2009. [PMID: 19384993]
Image: Dreamstime Images