June 1, 2014 (Vol. 34, No. 11)

Improving the Production of Recombinant Proteins in Cells

The production of recombinant proteins for therapeutic and industrial applications represents a significant activity of the biotechnology industry. Mammalian cellular systems have become popular for this purpose due to their capacity to properly produce and secrete proteins in a stable and functional form.

However, these systems are not perfect. Some of the difficulties surrounding the use of mammalian cell systems include stably integrating expression constructs at high levels, incorrect processing and/or modification of the desired protein product (e.g., the case of human proteins produced in nonhuman cell systems), and metabolic phenomena that can limit protein production.

These days, new genome-modification tools are being used to engineer cell systems to overcome some of these drawbacks. First, the PiggyBac™ transposon vector has been developed as an efficient tool for enhancing the stable integration of transgenes encoding recombinant proteins into the genome of the expression host. Second, a rapidly evolving set of engineered, targeted double-strand DNA nucleases allow the precise inactivation and/or modification of any genetic locus.

Better Gene Expression

Producing cells that stably express a recombinant protein relies on introducing an expression plasmid into the cell that must then be integrated into the genome, a time-consuming and costly process. This integration takes place at low efficiency, and even then, the vast majority of traditionally transfected cells do not produce sufficient levels of recombinant protein, possibly because the plasmids integrate in the 90% of the genome that is inactive heterochromatin. The previously mentioned piggyBac transposon system overcomes these problems.

The piggyBac transposon exists in nature as a gene encoding a transposase flanked by inverted terminal repeat (ITR) sequences. The transposase protein recognizes the ITRs, catalyzing the excision of the transposon from its locus in the genome and inserting it elsewhere at a random TTAA site.

The transposase can be supplied in trans and will, in fact, excise and re-insert any DNA sequence flanked by ITRs. Such artificial transposons can carry payloads in excess of 250 kb, which allows the inclusion of multiple genes as well as selection markers. These transposons are often constructed with chromatin insulator sequences to protect expression from position effects (silencing of the payload genes due to chromatin structure). This protection is further enhanced by the predilection for insertion into active chromatin.

The transposase has been optimized for use in mammalian cells and engineered to produce versions that are hyperactive (Super piggyBac, or sPBo) and that are capable of excision but not insertion (Excision-only transposase, or PBx).

Finally, the piggyBac transposase has a unique and useful quality among transposases: excision of a transposon from an insertion site regenerates the original sequence without any alterations.

The piggyBac transposon is thus an excellent system for the production of stably transfected cells. By titrating relative amounts of transposon and transposase, one can generate integration at high copy numbers, leading to increased expression of the protein.

Targeted Engineering of Host Systems

Beginning with the zinc finger nucleases and continuing with TAL effector nucleases (TALENs) and CRISPR/Cas9 to the newly developed NextGEN™ CRISPR, the past decade has seen a revolution in gene targeting due to the development and evolution of targeted double-stranded DNA nucleases, with continual improvements in simplicity, fidelity, and efficiency.

As the name implies, these allow the introduction of double-stranded DNA breaks to almost any site within the genome, and such breaks are then repaired either via error-prone nonhomologous end joining (NHEJ), where the DNA is re-joined with the inclusions of small insertions and/or deletions (indels) that can inactivate (knockout) a gene if they occur in a coding region, or via homologous repair (HR). This latter procedure, where a homologous sequence (e.g., the sister chromatid) is used as a template for repair, can also be harnessed by providing an artificial exogenous repair template to incorporate (knock-in) desired alterations into the genome at the break site. These nucleases can and have been used to re-engineer mammalian cells for improved protein expression.

One example of such an improvement was performed in CHO cells, which are routinely used for the large-scale manufacturing of proteins. However, in the case of antibodies, protein function can be altered by the glycosylation of the molecule.

For example, the different glycan patterns can have a significant effect on the pharmacokinetics as well as efficacy of the antibody. One such example is the presence of the fucose moiety in the core glycan structure. Antibodies without fucose display an increased cellular ADCC activity toward target cells.

Using XTN™ TALENs, we were able to target Fut8, the alpha1,6-fucosyltransferase gene, thereby inactivating the fucosylation system and generating a genetically modified CHO cell line for producing afucosylated human therapeutics. Two rounds of transient transfection with expression plasmids encoding XTNs targeting exon 10 of Fut8, combined with a proprietary selection system for nonfucosylated cells, resulted in the generation of a pool of cells in which 80% of the cells lacked fucosylated cell-surface proteins, demonstrated by FACS analysis, and 78% of the Fut8 alleles had been inactivated.

From these pools, clonal cell lines were isolated and further characterized. In Figure 1, using a cell surface detection method, the presence of fucose was assayed using flow cytometry. As expected, the parental CHO isolate demonstrated a clear unimodal population of cells containing fucose (Figure 1A). Likewise, the FUT8 knocked out pools and isolated clones exhibited the expected bi-modal and unimodal populations (Figures 1B and 1C) with the clone showing no detectable fucose on the cell surface.


Figure 1. (A) Parental CHO isolate demonstrated a clear unimodal population of cells containing fucose. (B) FUT8 knocked out pools and isolated clones exhibit the expected bi-modal and unimodal populations. (C) FUT8 knocked out clone shows no detectable fucose on the cell surface.

Importantly, antibodies expressed from these host cells were afucosylated and demonstrated the expected increased ADCC activity. Figure 2A demonstrates that antibody purified for multiple FUT8 knocked out subclones show only the G0 and G1 peaks using a standard HPLC based assay. For each of these antibodies, there is also a 6–10× shift in the IC50 value compared with the parental molecule in an ADCC assay (Figure 2B).

Although the Fut8 knockout project utilized a selection system, the efficiency of gene targeting using XTN TALENs is high enough that one can simply screen cells for the desired mutation. For example, in another gene knockout project in HELA cells, two rounds of transient transfection with XTN TALENs at 90% transfection efficiency followed by single-cell cloning and analysis identified four clones out of about 50 on a single 96-well plate that carried indels in the target gene, and in one of these both alleles were inactivated.

Given that the NextGEN CRISPR promises to be even more efficient than XTN TALENs, producing useful knockouts in cell lines should be within easy reach.


Figure 2. (A) Antibody purified for multiple FUT8 knocked out subclones show only the G0 and G1 peaks using a standard HPLC-based assay. (B) Antibodies show a 6–10× shift in the IC50 value compared with the parental molecule in an ADCC assay.

Jack Crawford ([email protected]), Carlisle Landel, Eric Ostertag, Ken Miller, and Valeriya Steffey work at Transposagen Biopharmaceuticals. Steve Sherwood, Namitha Rao, Lan Yang, Chris Simonsen, Rene Pagila, Hangjun Zhan, and Oren Beske are employed by Aragen Bioscience.

Previous articleExome Sequencing May Miss Pathogenetic Variants
Next articleTop 14 Molecular Diagnostics Firms