September 15, 2012 (Vol. 32, No. 16)
David Daniels, Ph.D.
The challenges of protein expression, purification, and production have been identified and are being addressed in both heterologous and homologous systems.
The ultimate goal of the research effort dictates what approach can be brought to bear, whether that goal be proper crystallization for resolution of protein structure or production of therapeutic proteins for use in the clinic.
Understanding the full range of complexities for protein expression and production can only come by the exchange of ideas among colleagues and collaborators. This was the focus of the CHI’s “Bioprocessing Summit” last month.
Well-characterized protein expression systems include cell-free extracts, E. coli, S. cerevisiae, baculovirus-insect cell systems, CHO cells, and HEK293 cells. Interestingly, there is usually no universal choice of the best system to use to produce any particular protein. This typically needs to be determined empirically for each specific protein of interest.
The lab of Don Jarvis, Ph.D., professor of molecular biology at the University of Wyoming, is focused on optimizing the baculovirus-insect cell system for recombinant glycoprotein production. Protein glycosylation, which is important because it influences protein structure and function, involves three distinct steps: first, the transfer of the initial oligosaccharide precursor; second, trimming of some terminal sugars; and third, addition of different sugars to form the final oligosaccharide side-chain.
Glycosylation patterns are cell type dependent and insect and mammalian pathways are distinctly different. The Jarvis lab has been focused on identifying genes encoding the endogenous insect cell enzymes and then adding human genes to enable insect cells to synthesize human-type glycan structures.
“It turns out that insect cells are highly amenable to uptake and incorporation of DNA,” Dr. Jarvis indicated. “We have been able to co-transform Sf9 insect cells with up to seven unlinked markers plus a selectable marker. After selection, we can find clones that express all seven markers at surprisingly high frequency. Screening is particularly easy as the glycan-processing pathway is linear. Thus, we can verify the presence of terminal sialic acids in a simple and fast lectin binding assay, which confirms that all seven genes have been expressed.”
“We have found that a minimum of six genes are needed to humanize the insect cell glycosylation pathway. We are now focused on finding innovative ways to increase the efficiency of this biosynthetic process.”
Exploiting S. cerevisiae
At the University of Delaware Biotechnology Institute, Carissa Young, Ph.D., and her colleagues are asking the question, what’s preventing us from expressing our heterologous protein at high functional yields in yeast cells? The lab has taken a systematic approach to understand why proteins fail to make it to their final destinations, either the plasma membrane or secreted into the media.
To investigate protein-trafficking effects, the lab has designed expression cassettes that incorporate conventional affinity tags and fluorescent proteins used for the identification and purification of the heterologous proteins. This enables the development of rationally designed chimeras that can also be used to identify subcellular localization in numerous organelles including the ER, Golgi, and plasma membrane.
“We have taken a holistic approach to determine which elements of protein structure impact trafficking of heterologous proteins in yeast. Specifically, exit motifs play a pivotal role in ER-to-Golgi trafficking within the secretory pathway, and very few studies have assessed these effects,” shared Dr. Young.
“Using chimeric receptors, we confirmed at both the gene and protein levels the results of trafficking motifs and domain-swapping, while correlating these effects to growth rate, expression levels, and activity. Different from an ER-signal sequence located at the N-terminus, ER exit and retrieval motifs consist of a 3- to 4-amino acid sequence, generally clustered in groups of two or three at the C-terminus of the protein. Interestingly, most affinity tags used in protein work contain several trafficking motifs that can be exploited for heterologous protein expression.”
The lab is working with GPCR membrane proteins, members of the adenosine and neurokinin receptor families, as well as antibody fragments. Copy number, time of expression, and the induction and culture conditions can all impact the level of expression for the membrane protein or the secreted protein. Consequently, genes that encode these heterologous protein sequences are inserted directly into the yeast genome with a controlled copy number.
Time-course experiments evaluate samples, cells, and media in order to quantify the total level of protein expression and the degree to which antibody fragments are secreted. Expressed and purified protein levels are calculated by Western blots coupled with standards developed in-house.
Membrane Protein Expression
The challenge of overexpressing membrane proteins in cells needs to be addressed on multiple levels. First, can you make enough protein without killing the cells? Second, what detergent provides optimal extraction of the membrane protein and stabilizes the protein without aggregation? (Membrane proteins often display significant nonspecific aggregation.)
Third, concentration of the protein enables formation of the proper crystal lattice structure. At concentrations of 1 mg/mL the protein solution is so dilute that the probability of forming crystal nuclei (the initial event in crystallization) is extremely low. The majority of proteins crystallized have initial concentrations between 5 and 20 mg/mL.
At higher concentrations in homogeneous solutions the proteins have a higher probability of interacting in the proper orientation to form a proper crystal lattice (nucleation event). Unfortunately, for many aqueous and membrane proteins, as the concentration of purified protein is increased, there is often an increase in the formation of nonspecific aggregation producing a nonhomogeneous population of purified protein—not desirable for crystallization.
Self-interaction chromatography is a method that has been used to assess the best conditions to place a soluble protein that will support stability of the protein in high concentration without formation of aggregates. Development of an automated analytical self-interaction chromatography system is now being adopted for identification of buffer conditions that will support rapid identification of the optimum solubility conditions followed by identification of solution conditions with a higher probability of crystal formation.
“We’ve taken this same approach to determine the best buffer conditions to grow membrane protein crystals. We load the column with 0.75 mg of protein, allowing it to form covalent bonds to a polymer-based bead matrix in the column. This guarantees that every surface of the protein is exposed to the solvent,” said Larry DeLucas, Ph.D., director, Center for Structural Biology, University of Alabama at Birmingham.
“We then load a few µg of the same protein onto the column under 100 different buffer conditions that contain different solubilizing additives (excipients) and measure the elution retention time for the injected protein. This represents a small random sampling of the total number of possible combinations and concentrations that could be investigated.
“This experimental data along with other parameters, including total protein covalently bound to the matrix and the column void volume is input into an equation that calculates for each solution condition the second virial coefficient, a thermodynamic term that represents the sum of all protein-protein interactions. The calculated second virial coefficient values with their respective solution conditions are then input into an artificial neural network (ANN),” he continued.
“The ANN uses this input data to predict the second virial coefficient values for the complete factorial of possible solution conditions. The resulting predicted values provide an assessment of each solution condition’s ability to reduce (positive value) or increase (negative value) protein-protein interactions. For protein crystallization, we are looking for a gentle net attraction, solution parameters that force the protein molecules to display weak attraction, thereby improving the probability of forming high-quality crystals.”
From this approach the DeLucas lab has determined that lithium chloride in solutions up to 1 M can support the concentration of the cystic fibrosis transmembrane regulator protein from 0.05 mg/mL to 0.5 mg/mL in a homogeneous solution without formation of aggregates.
As a former payload specialist who flew aboard the 1992 U.S. Space Shuttle Columbia, mission STS-50, Dr. DeLucas was able to demonstrate that certain protein crystals are of higher quality when grown in a microgravity environment as compared with earth-grown crystals of the same protein. At that time, the challenge was the short duration of time in space.
Crystals grow slower in space and so the size of the crystals from the typical space shuttle 10-day mission was often too small for diffraction analysis. With the establishment of the International Space Station (ISS) the time constraints can be overcome. Dr. DeLucas will be contributing more than 100 proteins from researchers in academia and industry to a future commercial launch of the Space X Dragon laboratory, flown on the Falcon rocket.
Immortalization of Human Cells
At the University of California, San Francisco, Dieter Gruenert, Ph.D., has developed a standard approach to establish immortalized human cells that maintain the differentiated characteristics of the primary cells from which they are derived. These cell lines enable the study of normal cell function and disease mechanisms, and serve as screening platforms for therapeutic intervention. They are preferred over primary cell systems which have a limited lifespan and can be difficult to manipulate.
“We use an origin-defective SV40 plasmid in a standard transfection protocol to isolate clones of human epithelial cells with the desired phenotypes. We take specific care to capture the cells of interest at a particular stage of differentiation and then lock the cells in that stage under prescribed culture conditions,” said Dr. Gruenert.
“We have 15–20 different cell lines that are available to scientists for their research efforts. We also have hundreds of clones that have been established but not yet fully characterized prior to freezing them down.”
The key here is that the cell lines are of human origin and not animal or insect cells. Essentially each species has similarities to humans, but there are often significant molecular and metabolic differences that can influence the actions of pharmacological agents. Dr. Gruenert is adamant: “If you are focused on human disease, then use cell lines of human origin.”
Dr. Gruenert is also very sensitive to the issues of establishing immortalized cell lines. The issue of de-differentiation of cells in culture is taken seriously, and cell-line provenance is monitored on a regular basis by comparison to the primary cells and evaluation of the epithelial endpoints: keratin expression, tight junction formation and polarity, and finally neoplastic progression.
As long as the cells are maintained under the culture conditions in which they’ve been isolated, they will maintain their proper character. To that point, the lab has a workhorse cell line that has been in culture for over 20 years.
The development of human-induced pluripotent stem (iPS) cells has opened a new arena of possibilities for immortalized cells. Since the iPS cells are immortal and retain their diploid karyotype, they can potentially be directed to differentiate into any cell type within the body. This is not only significant in terms of cellular therapy through tissue repair and regeneration, it is a significant tool for the development of pharmacological and genetic therapies.
Biogen Idec has a dedicated drug development focus on production of biotherapeutic proteins for use in the clinic for neurodegenerative and autoimmune disease states. Brian Majors, Ph.D., a scientist in the cell engineering group at Biogen Idec shared the process by which his group has been able to optimize cell-line constructs they obtain from R&D for the production of the proteins destined for Phase I clinical trials.
“Our group transfects CHO cells using the expression vector containing the lead molecule characterized by the biological team. We grow up to thousands of clones and then seed 24-well deep-well plates. The focus is on defining the culture conditions that will best replicate those used in the bioreactor that will be used to produce material for the clinical trial,” said Dr. Majors.
“The protein is harvested from each of the clones after 14 days of growth. The proteins are purified in two steps: first, capture by binding protein A, and then size-exclusion chromatography for isolation of protein monomers. The purified proteins are then subjected to analytical analysis. A handful of clones that produce the best quantity and quality of product are processed further.”
The team uses a label-free method to monitor cell production of proteins during the 14-day culture period. A fiber optic tip coated with protein A is dipped into the culture media to capture and quantitate the protein in solution. The overall goal of the team’s effort is to establish a process to predict cell-line performance early in the development cycle to save time and money for the project.
At NovImmune, they are investigating possible extension of their portfolio of therapeutic antibodies by the introduction of a newly developed kappa-lambda body (κλ-body™). This fully human bispecific antibody format has typical functional and biochemical characteristics of a human IgG. The κλ-body features a common heavy chain but two different light chains (one κ and one λ), giving a different target specificity to each antibody arm.
The benefits of using a fully human format, without mutations or linkers, are numerous. They include PK profiles and Fc effector functions of therapeutic antibodies, robust productivity and stability during manufacturing and storage, and low risk of provoking immunogenicity.
The upstream processing group, headed by L. Di Grazia, further developed the novel tri-cistronic vector system for expression of the κλ-bodies in CHO cells, originally designed by the internal research group. CHO cell lines for two different κλ-body products were established, and the cell lines were observed to be genetically stable and maintain constant high-level productivity. The κλ-bodies were tested and shown to maintain biological function.
“Over the course of our work, we’ve grown these cells for over 50 generations without loss in productivity or κλ-body assembly efficiency,” said Di Grazia.
In order to manufacture batches for preclinical studies, the process has been successfully scaled up to the 100 L level using stirred bioreactors. The aim of the early cell-line screening approach is to minimize the impact of cell-line variability on downstream processing, notably with respect to the impact of varying concentrations of product-related impurities.
“To date we readily achieve multigram/liter titers for different κλ-body products using the platform USP,” noted Di Grazia. “These yields are equivalent to those achieved for our monospecific antibody controls, suggesting that κλ-body assembly is as efficient as that of therapeutic monoclonal antibodies.
“We routinely use a binding assay to verify that the purified products maintain the appropriate binding specificity as well as their ability to co-engage the two targets. This is a critical quality control assay for us. Our κλ-body products have also demonstrated favorable drug characteristics in vivo with PK properties equivalent to monospecific monoclonal antibodies.”
The κλ-body products developed to date by NovImmune have been shown to be readily amenable to a scalable platform manufacturing process. Di Grazia noted that the final product has the benefits of a monoclonal antibody plus the added functionality of multiple specificities: it is stable, has the expected half-life in circulation, a low potential for immunogenicity, and contains essential effector functions necessary for cutting-edge biotherapeutics.