January 1, 2014 (Vol. 34, No. 1)
Angelo DePalma Ph.D. Writer GEN
Approximately half of approved recombinant biotherapeutics are expressed in Escherichia coli. During downstream processing, it is necessary to remove the organism’s outer membrane, which consists of lipopolysaccharide (LPS), also known as endotoxin.
LPS is involved in a diverse spectrum of immune-response pathologies.
Standard methods for removing this toxic component include activated carbon, modified sepharose resins, anion exchange chromatography, surfactants, and ultrafiltration. These methods, which are employed by processors during the final phases of downstream processing, all have their drawbacks. At best they add a process step at significant cost; at worst they affect recovery of active protein.
An alternative approach, the production of protein in endotoxin-free E. coli cells, was described at a recent Informa Life Sciences BioProduction conference in Dublin. In addition to revealing innovations concerning E. coli, the conference highlighted advances with respect to CHO strains.
The E. coli work that involved removing endotoxin was carried out by a team of scientists representing the University of Michigan, Research Corporation Technologies (RCT), Research Center Borstel, and Lucigen. This group essentially eliminated the need to remove endotoxin during downstream processing.
The scientists developed a technology called ClearColi™, cell strains derived from E. coli. These E. coli mutants lack outer membrane agonists for activating the human Toll-like receptor hTLR4/MD-2.
Of the many ligand-binding receptors on the surfaces of human immune system cells, hTLR4/MD-2 specifically recognizes endotoxin. After binding, the receptor dimerizes, which induces a signal cascade that activates a transcription factor, NF-κB, which controls transcription of several proinflammatory cytokines that are harmful to humans.
The scientists achieved the construction of ClearColi cells by blocking the synthesis of endotoxically active LPS from the precursor lipid IVA (four-A) through multiple gene deletions. An additional mutation enables viability in the presence of lipid IVA.
Lipid IVA consists of four fatty acid chains, whereas the mature LPS, which induces the hTLR4/MD-2 pathway, has six acyl chains. By contrast, precursor lipid IVA does not induce the hTLR4/MD-2 pathway, so it is endotoxically inactive.
A key feature of ClearColi cells, according to Uwe Mamat, Ph.D., a senior research scientist at the Research Center Borstel, is the presence of a so-called ABC transporter with relaxed substrate specificity. “Usually, the LPS transporter is highly specific for the hexa-acylated LPS, but barely moves lipid IVA out of the cell,” explained Dr. Mamat. “The transporter’s ability to transfer lipid IVA out to the cytoplasm is critical for maintaining cell viability.
“The ClearColi system is a versatile technology,” Mamat continued. “We have tested ClearColi cells with a number of model proteins, all of which were endotoxically inactive.”
Fab’ fragment expression within the periplasm of E. coli is desirable on several levels. Primarily, the periplasm’s oxidizing environment benefits formation of disulfide bonds, which the reducing nature of the cytoplasm does not. Additionally, the periplasm constitutes as much as 40% of the cell’s volume, proteolytic activity is lower than within the cytoplasm, and proteins that are toxic within the cytoplasm can accumulate with no deleterious effects.
Mark Ellis, principal scientist at UCB Celltech, has described techniques for generating periplasmic Fab’ yields of 5 g/L, without compromising product quality or cell viability.
Interestingly, high concentrations of Fab’ do not by themselves compromise viability. “That arises from a lack of resources, when cells deplete sugars and amino acids,” said Ellis, who also spoke at the Dublin conference.
Particularly when under pressure to overexpress a foreign protein, cells may not have access to molecules that enable them to divide. “The cells are consuming resources faster than normal.”
Although Ellis uses fed-batch cultures, the only feed ingredient is glycerol. One could also feed amino acids, but that would result in rapid and premature accumulation of biomass. Instead, Ellis knocked out protease genes specific to the periplasm, an approach that leads to higher protein accumulation.
Similar techniques, he noted, are available for proteins that accumulate within the cytoplasm. In addition, Ellis co-expresses a gene for the isomerase DsbC, which is required for forming disulfide bonds in Fab’. “There is only a finite amount of DsbC in the cell,” he explained. “Once cells start using it to make Fab’, levels are insufficient for the cell’s own disulfide-containing proteins.”
CHO Genome Yielding Results
First isolated more than 50 years ago, Chinese hamster ovary (CHO) cells have become the workhorse mammalian cell line for producing biopharmaceuticals. According to R&D Pipeline News, sales of biopharmaceuticals expressed in CHO approached $60 billion in 2011, accounting for more than half the year’s total sales of biotherapeutics.
Now that the CHO genome data is being deciphered, biopharmaceutical companies have the opportunity to modify CHO strains to commercial advantage. For example, CHO genome data has guided Nicolas Mermod, Ph.D., who directs the Institute of Biotechnology at the University of Lausanne, to overcome some of the principal factors that limit protein yields in CHO cells.
When plasmids enter cells they must first penetrate the nucleus. From there they must find a favorable chromosomal locus, insert as multiple copies, and overcome potentially unfavorable chromatin environments.
Generally, transgenes insert at one or few points in the chromosome of a given cell. “But most loci are not very favorable,” Dr. Mermod explained. “If the transgene enters an unfavorable region, protein expression will be low.” Similarly, a hundred copies of the transgene inserted into favorable loci will express more than one copy. “Both these events depend on uncontrolled DNA repair and recombination events that occur within the cell.”
Another challenge involves effects related to CHO cell and protein subtypes. CHO cells are genetically unstable due to their five decades of accumulating genetic mutations—they are, after all, cancer cells. In a typical population, some cells will be fit to produce proteins, and others will be less fit. “Some proteins are inherently more difficult to express than others, depending on the cell,” Dr. Mermod added. “Some mutations may make a CHO subtype less prone to express one protein, but not another protein.”
That may be why CHO is generally less productive than E. coli. (Although E. coli strains present fewer production-related bottlenecks, they have problems of their own, such as unfavorable glycosylation.) It is also the reason why biopharma companies generate hundreds or thousands of clones, and employ automated techniques to screen for winners.
Some groups have attempted to direct transgenes into “friendly” chromosomal loci, but this approach is limited by the difficulty in introducing multiple transgene copies to specific loci.
Dr. Mermod’s approach involves understanding the mechanisms by which transgenes integrate into chromosomes, and exploiting epigenetic DNA elements known as matrix attachment regions (MARs). MARs regulate switching between active and inactive gene expression. In addition, noted Dr. Mermod, “MARs synergize with specific recombination pathways to mediate very high expression.”
Thanks to knowledge of the CHO genome, Mermod and co-workers, together with scientists at the Swiss Institute of Bioinformatics, were able to locate the integration sites and to determine the DNA sequence at the junction between the transgene and the chromosome. Analyzing the sequence at the fusion point unveiled favorable mechanisms that allowed researchers to reengineer the cell recombination pathways in ways that cause transgenes to home in on more-favorable loci, with increased and more reliable transgene expression.
“We are blocking recombination pathways that we don’t want, in favor of those we desire,” asserted Dr. Mermod. “But we don’t know how the ones we want select favorable regions on the chromosome. One possibility is that chromatin is more open in ‘friendly places,’ and that the DNA more accessible to particular recombination proteins is also more accessible to transcription proteins that drive the synthesis of mRNA—the first step toward protein expression.”
Dr. Mermod’s group is also working with scientists at Selexis to tackle expression problems that are independent of plasmid-chromosome interactions. The researchers have demonstrated reasonable success at improving yield by improving protein secretion.
“This restored production for some difficult-to-express proteins, and even increased the production of others that are more easily expressed. But we have not yet solved this problem for all proteins,” conceded Dr. Mermod.
Silence of the RNAs
Noncoding RNAs, which influence gene expression in myriad ways, might prove useful in reengineering CHO cells for improved viability and protein expression. This idea is behind some of the research that is being carried out by Niall Barron, Ph.D., program leader for mammalian cell engineering at Dublin City University.
Reengineering efforts to create CHO cells show inherent improvement, not just better protein expression, began about 10 years ago. “Some success has been realized, particularly in improving viability in culture,” observed Dr. Barron. “But generally the engineering strategies have disappointed.”
One possible explanation is that CHO strains, which were originally consisted of attachment-dependent cells, evolved to accept suspension cell culture conditions. During this transition, the CHO genome evolved significantly.
Dr. Barron became interested in the prospect of advanced, multigene engineering methodologies, but was not interested in modifying each gene individually. He believed that going after a handful of proteins through small interfering RNA (siRNA) would be fruitful, which it was.
“MicroRNA has the ability to target multiple genes downstream of its target, depending on the context. We were hoping to affect expression of hundreds of targets,” Dr. Barron explained.
Before the CHO sequence became available, Dr. Barron’s group relied on human, mouse, and rat genomic platforms and counted on conservation of certain sequences to carry over to CHO. Since then, scientists have identified approximately 300 CHO microRNAs.
Several strategies exist for effecting desirable changes in the CHO genome. One is to overexpress a decoy transcript, a genetic “sponge.” By binding to one end of the target gene, the sponge either blocks translation of that mRNA to protein or destabilizes the transcript.
A transient approach involves inhibition of microRNA by short, complimentary siRNA directly against it. A variant consists of permanently integrating a sponge molecule.
Another technique uses short hairpin RNA that ultimately converts into an siRNA against the target microRNA—similar to the transient situation, but now stably expressed in cells. Finally, one might delete microRNA from genome completely.
“Any of those approaches is feasible. In a biopharm setting, the last approach is most desirable. But none of these engineering technologies have been through the regulators yet,” Dr. Barron cautioned.
His group actually used the stably expressed decoy approach targeting the CHO microRNA-7, with good success. In a nonoptimized fed-batch culture, cell density and longevity improved, while yield of a model protein doubled. Dr. Barron has patented the targeting of microRNA-7 for these purposes, and continues to work on noncoding RNA gene silencing.
In commercial settings, Dr. Barron recommended an even simpler approach: continuous feeding of siRNA into CHO culture, almost as media supplement. “This would not require any serious regulatory changes or resubmissions, as a genetically engineered cell line would. And it could apply to existing processes.”