Send to printer »

Feature Articles : Mar 1, 2008 (Vol. 28, No. 5)

Optimizing Transient Gene Expression

Applications Expected to Move Beyond Discovery and the Preclinic to Clinical Realm
  • K. John Morrow Jr.

Transient gene expression represents an appealing and complementary alternative to the development of stable cell lines for the expression of bioengineered proteins, according to experts in the field.

Yves Durocher, Ph.D., project leader from the National Research Council Biotechnology Research Institute in Montréal, Canada, and his colleagues  optimized the process of transient gene expression, examining a variety of  parameters. “Classic transient expression protocols involve short-term production of proteins over a span of days to a couple of weeks,” says Dr. Durocher. “Applying plasmid DNA with a transfection reagent to cells grown in a T-flask will result in micrograms of protein, but this level can now be scaled up to much larger yields.”

Applications for these pilot levels of recombinant proteins run the biotechnology gamut, including studies on toxicity, stability, affinity, and activity of the product. Mammalian cells are the vehicle of choice, and while CHO cells have predominated in the past, the human HEK293 line is gaining wide acceptance.

There are a variety of nonviral transfection reagents including DEAE Dextran, calcium phosphate, lipids, and polyethylenimine. These techniques are thought to be far superior to the use of viruses, which require a BSL2 facility and are cost ineffective on a large scale.

Historically, the three prominent cell lines used for large-scale gene transfer have been COS, HEK293, and CHO cells. Optimization of the cell line early on in protocol development is a critical issue since validation of multiple cell lines is an unattractive option.

The genetic variant of the HEK293 line that expresses the Epstein Barr nuclear antigen (HEK293E) allows active propagation of plasmids carrying the appropriate origin of replication. This results in augmented levels of protein expression and makes this 293 cell line the optimum choice for Dr. Durocher’s protocol.

The preference of transfection-promoting agents was determined largely by cost considerations. For a 10 L transfection protocol, 293Fectin ran $4,000, XtremeGENE came to $5,000, whereas polyethylenimine was less than a dollar. Given that their performance is roughly equivalent, the decision was not a difficult one.

Polyethylenimine, discovered as an efficient gene-transfer reagent in the mid 1990s, is a highly positively charged branched or linear polymer that complexes with DNA and is thought to gain entry to the cell by endowing the complex with an overall positive charge. It is known to move into the nucleus of the cell once taken up by endosomes.

Whereas today transient gene expression is used for discovery and the preclinical phases of development, Dr. Durocher envisions that in the future there will be a myriad of applications in clinical studies as demands arise for recombinant proteins occupying a specific therapeutic niche. “These could include biotherapeutics for personalized medicine, vaccines for possible pandemics, orphan drugs, and a range of recombinant viruses for gene therapy,” he suggests.

In fact, HEK293 cells have already been approved in the past for the production of recombinant viruses used in clinical trials. Many of them were even produced by large-scale transfection. “There is no reason why these cells could not be used for the production of biotherapeutics,” he concludes. “We could face a range of recombinant proteins but have only one cell line to validate when using this technology.”

“We addressed the early IgG expression issues by optimizing our in-house transient expression system,” says Lekan Daramola, Ph.D., head of antibody format technologies at Medimmune.

Dr. Daramola emphasized that the path of drug development, from early-stage research through preclinical development, requires ever-increasing amounts of antibodies, from microrgrams to grams, when preclinical and clinical studies are undertaken. The choice of expression system used in the different stages of drug development is dependent on a number of factors including the quantity required, the quality of material required for study, and the speed of delivery. Transient expression can currently be used to provide up to gram amounts in a relatively short time. The in-house transient expression system used by Medimmune is based on cotransfection of the heavy and light chain vectors into HEK293E cells. Yields of up to 200 mg/L have been obtained.

Stable CHO cell lines are generally used to provide clinical material, so Dr. Daramola and his colleagues are investigating the use of CHO cells for transient expression, a more auspicious choice and one in which similar glycosylation patterns to the CHO stable cell line can be expected.

There are a range of causes of poor transient gene expression, which may negate its advantages if they are not properly addressed. These include failures in vectors, cell type, transfection reagents, transfection efficiency, and cell culture process. Dr. Daramola emphasizes that many IgGs may be inherently poor expressors because of their DNA sequences. Medimmune initiated a pilot study in which a range of different issues were studied using three different antibodies. Three poorly expressing IgGs from different projects were transfected into HEK293E cells using 293Fectin as a transfecting agent.

Samples were taken at two days post-transfection, and the RNA and protein production levels were analyzed by Northern/RT-PCR and Western blotting/ELISA respectively. As expected, the control antibodies produced much more protein than the poor expressing antibodies.

By swapping heavy chains and light chains in control versus poor expressors, however, it was demonstrated that the light chain genes of the poor expressors were crippling the overall IgG expression. Northern blotting and RT-PCR analyses of the RNA transcripts showed a correlation between poor expression and the presence of a truncated light chain transcript.

Truncated light chain transcripts may be due to the presence of cryptic splice sites within the gene, which can greatly reduce protein synthesis. To confirm this hypothesis, silent mutations were introduced to remove the splice donor site from the sequence, this was accompanied by a large increase (up to 20-fold) in the transiently expressed IgG. In all three of the cases tested, there was a dramatic increase in protein production.

The identification of missplicing may not be all of the poor expression issues. Transient expression yields can be greatly enhanced by optimizing the transfection system and removing undesirable signals from the gene. Because manipulations such as codon optimization and removal of cryptic splice sites may bring improvements in some but not all cases, the question arises, what and when to optimize?

“We would say that optimization should be started as soon as poor expression issues are identified, since inherently poor expressors may require more investigations and may not always respond to simple manipulations,” Dr. Daramola notes.

Maximizing Production in CHO Cells

Achievement of high yields during transient gene expression poses a number of challenges, according to Henry Chiou, Ph.D., technology area manager at Invitrogen. “The culture media, degree of cellular contact, the function of the plasma membrane, and capacity for endocytosis all need to be considered.”

Dr. Chiou’s group elected to optimize the CHO line. Because of the wealth of information regarding the CHO lines, it is known that their glycosylation patterns are similar to human glycosylation, and for this reason, they are the logical choice for the production of human therapeutic proteins.

“In our investigations, we were guided by various parameters that affect protein yield,” he continued. “We have found that transfection efficiency, transgene copy number, the level of gene expression of the transformants, and the persistence of the expression phenotype are all crucial values.” Dr. Chiou and his group employed the Freestyle MAX transfection reagent, which was optimized for large-scale transient protein production in CHO cells. This reagent is a cationic lipid developed for transfection by Invitrogen.

Cationic lipids contain a long lipid chain and a positively charged head, so when they complex with negatively charged DNA, the negatively charged phosphate groups are neutralized and the DNA collapses into a greatly reduced volume. As such, it can be transfected into a recipient cell and its genetic information expressed.

Because the lipoplexes are extremely complex, it is not known how the molecules are arranged with respect to one another nor is it clear whether they can be rigorously analyzed. For this reason their performance is studied empirically through procedures that Dr. Chiou describes, namely, transfection frequencies of 80% can be achieved with Freestyle MAX, as measured by the uptake of the green fluorescent protein determinant.

Protein production in the initial lab bench studies using transfected cells increased rapidly, reaching 60 mg/mL of IgG after just six days. It then slowly declined over the course of weeks.

Scale up of production involves the use of shaker or spinner flasks and then to either the Wave or a fixed-tank bioreactor, raising the volume from 2 to 100 liters.

Optimum performance requires that cell viability be maintained and that the time and method used to introduce the transfection complex into the reactor be carefully monitored. For this reason the growth rates of the cells, the agitation rates, and the kinetics of cellular attachment were scrupulously followed throughout the optimization process. Over extended periods, the Wave bags maintained the highest level of green fluorescent protein positive cells.

“In our scale-up studies, although production of antibody was at about 37 mg/mL at day six in the bioreactor, it rose to over 60 mg/ml at day 11,” Dr. Chiou stated, “which is when peak antibody production (>60 mg/mL) was reached at the 30 mL scale.”

“The take-home message was that we could generate as much protein yield per mL of culture at the 10 liter scale as we could at the 30 mL scale, provided the culture conditions are optimally adjusted,” he concluded.

HEK and CHO Cheek to Jowl

Gerald Casperson, Ph.D., an associate fellow in biotherapeutics at Pfizer, uses HEK293f cells adapted to suspension culture in serum-free medium. Dr. Casperson carries out his transfection protocol with the Invitrogen 293fectin lipid reagent designed to work in the 293 Freestyle medium.

In a transient expression platform in 1 liter flasks, the HEK293 line generated as high as 170 mg/mL of antibody. “Our fast expression system reduced time to deliver milligram quantities of mAbs by eight weeks,” Dr. Casperson states, “so there’s less work for cell biologists, and with our raw conditioned medium in a very clean condition, there’s less work for biochemists.”

Indeed, Dr. Casperson claims that, while the stable CHO production times run to 12 weeks, with the transient HEK293f system the collection of medium was concluded in one week. His group was able to express a wide range of proteins including kinases, receptors, and enzymes.

To provide an additional jump start to the growth performance of the HEK293 cells, the investigators added sodium butyrate to the culture medium. This compound has been known for years to be a regulator of gene expression in cell culture and its addition resulted in a doubling or more of antibody production.

Improving CHO Performance

Approaches to optimization are numerous, according to Alison Mastrangelo, Ph.D., group leader at Lonza Biologics. Her work includes  adopting pools of transfected cells as a path to increasing protein yields. “Transient gene expression is quick, but it can be complicated,” Dr. Mastrangelo states.

In vitro protein yields can be improved by various additives including serum, which provides good yields but causes difficulties in subsequent purification. The Lonza group attempted to eliminate serum by substituting bovine serum albumin, which could not restore the transient output achieved with serum, nor could addition of lipids and insulin improve the output of proteins by the cultures.

On the other hand, auspicious results were obtained with insulin-like growth factor (IGF) in the form of Novozymes’ LONG R3 IGF-I. This 83 amino acid analog of human IGF-I is composed of the human IGF-I sequence with the substitution of an Arg for the Glu at position 3 (hence R3) and a 13 amino acid extension peptide at the N-terminus. Its action is the same, but it is endowed with a much longer half-life.

Combining the factor with mild hypothermia (32°C) further increased final antibody concentrations fourfold to 12.5 mg/L. “Simple bioprocess strategies can have significant synergistic effects on transient expression,” Dr. Mastrangelo adds. “Strategies that retain cells in transfected and transcriptionally active states can lengthen the duration of expression, boosting overall transient mAb production.”

Pooling cultures allow greater overall production in a shorter time than that obtained with a stable cell line in a bioreactor. The pooled transfection method is capable of quickly generating recombinant product comparable to the generic fed-batch and fermentation operations. There is consistency in the proteins generated from flasks derived from different transfections, so the average of material produced by many cell lines compares with the output of a single cell line.
“The pooled transfection method can yield hundreds of grams of material in a time frame between transient expression and stable cell line generation,” Dr. Mastrangelo states.

Unresolved Questions

For all its virtues, transient gene expression can’t deliver the quantities that permanent cell lines can produce, at least not yet. “There is a lot of labor put into stable cell line development to achieve gram per liter yields,” notes Dr. Chiou. “First of all, high-expressing clones are selected from a pool of surviving transfectants after selection and gene amplification.”
“These are individually optimized for culture conditions that will provide the highest yield. This may involve assessment of metabolic needs, optimization of trace elements, and addition of supplemental components such as amino acids and glucose feeds. This entire process will take many months and in some cases over a year,” he says.

“So the attractiveness of transient production is that it is quick,” Dr. Chiou continues, “From the time you start a seed culture of cells and begin to prepare the quantity of DNA needed until the time you are ready to transfect the cells is about 10 days. From transfection to the time when you can harvest protein requires only 4–14 days, depending on the culture.”

Dr. Chiou feels that to generate kilos of an antibody by transient transfection is not within reach, yet. Since the highest reported IgG yields by transient transfection are in the range of ~400 mg/L, a kilogram of protein would require scaling up the culture and transfection to 2,500 liters.

“I do know of at least one major pharma that does run some transient production at the 1,000 liter scale, but experience and conditions at that scale are still generally unexplored territory,” he continued. “It would require an order of magnitude increase in yield and/or a lot more experience at the >1,000 liter culture scale to get to kilogram production, but I don’t doubt that we could get there.”

The other aspect of raising performance focuses on the reagents. DNA-cationic complexes or lipoplexes are Byzantine molecular structures, irregular and heterogeneous. This makes them difficult to dissect and characterize, which in turn complicates standardization and development of a protocol that will be repeatable and consistent. Alternatives such as viral mediated entry, however, have drawbacks including cost and biosafety, which are limiting factors for large-scale procedures.

A lot of fundamental investigation is aimed at acquiring a better understanding of these molecules, but the work is difficult. When imaging methods are applied to lipoplexes, one sees a pastiche of amorphous globs that resembles a plate of spaghetti and meatballs, a somewhat less than informative picture.

Fortunately, a number of analytical techniques are being applied to the lipoplex problem, including dynamic light scattering, analytical ultracentrifugation, gel electrophoresis, circular dichroism, and fluorescence spectroscopy. These studies coming from basic science laboratories are producing insights into the molecular innards of the lipoplexes and should add to the purely empirical understanding accumulating through commercial laboratories.