September 15, 2013 (Vol. 33, No. 16)

Richard A. A. Stein M.D., Ph.D.

Advances in DNA biology, particularly those of the past three decades, catalyzed the emergence and expansion of experimental approaches to manipulate gene expression. Subsequently, these advances impacted protein science, a field that found itself increasingly positioned at the juncture of technology, science, and art.

A fundamental prerequisite for generating any construct, as part of cloning and protein engineering efforts, regardless of the experimental system, is the ability to grow cultures under controlled and reproducible conditions. Process scouting devices, which most frequently are disposable devices such as shake flasks, spinner flasks, or microtiter plates, are routinely used in research at the earliest stages of bioprocess development. One of their disadvantages is that they are not equipped with sensors.

Although temperature and agitation are the most frequently measured variables in experiments performed in these devices, critical culture parameters, such as pH, oxygen level, and carbon dioxide level, cannot be accurately and safely monitored. This is partly because conventional probes are often costly and difficult to use. Nevertheless, many processes that are foundational to large, industrial-scale experiments, such as the generation, testing, and selection of clones, are performed early in bioprocess development, making it necessary to collect real-time experimental information on additional parameters. Moreover, inadequate monitoring can also hinder the subsequent integration of the parameters from various stages of an experiment, when protein expression is scaled up or down.

“We are trying to make bioprocessing more intelligent and provide a greater degree of measurement at all stages,” said Govind Rao, Ph.D., professor of chemical, biochemical, and environmental engineering and director of the Center for Advanced Sensor Technology at the University of Maryland, Baltimore County (UMBC). A major effort in Dr. Rao’s lab is focused on developing noninvasive, disposable sensors to actively monitor culture parameters during growth.

“Most biological processes are very complex, and many parameters change during the growth of a culture,” added Dr. Rao. Recently, Dr. Rao and colleagues described the use of triple disposable noninvasive optical sensors that can be positioned inside culture flasks, and revealed that pH, oxygen level, and carbon dioxide level can be dynamically monitored during E. coli fermentation to collect information that would not be routinely available from shake flasks.

According to a paper co-authored by Dr. Rao in Biotechnology Progress in 2012, the sensitive element of the disposable noninvasive optical sensors is a thin, luminescent patch affixed inside the flask. The paper also noted that small electronic devices for excitation and fluorescence detection are positioned outside the shake flask for noninvasive monitoring.

This work marked the first time disposable noninvasive sensors were used to measure dynamic changes in these parameters, over time, in shake flasks. “We have now started working on next-generation technologies, and our goal is to bring the whole bioprocess to the bedside or to the point of care, and take personalized medicine to a potentially different level,” said Dr. Rao.

Next-generation bioprocessing technologies target expression of biologics on demand, and there is a DARPA-sponsored effort under way at UMBC. “These efforts are paving the way toward next-generation bioprocessing, in which very well controlled protein expression can be performed at a small scale, in compact and low-cost systems, promising to change the entire biomanufacturing paradigm,” said Dr. Rao.

Generating Membrane Proteins

Approximately 30% of all genes from all genomes sequenced to date encode membrane proteins, and their biological importance is additionally underscored by the fact that the largest class of membrane proteins, the G protein-coupled receptor (GPCR) superfamily, is the target of approximately 50% of the existing therapeutic agents. Nevertheless, most membrane proteins remain structurally elusive. This is vividly reflected by the fact that, in mid-2011, they represented only about 1% of all protein structures deposited in the Protein Data Bank.

“We have been using commercial cell-free systems to generate membrane proteins, which are the most difficult proteins to make irrespective of the expression system,” said Shuguang Zhang, Ph.D., associate director of the Center for Biomedical Engineering and head of the Laboratory of Molecular Design and the Center for Bits and Atoms at the Massachusetts Institute of Technology. Cell-free protein expression has been used for over 50 years, but it is still a challenge to use this technique to generate membrane proteins. One difficulty, among many, is the need to identify a detergent that does not interfere with the in vitro translation.

“The most important thing, when making a membrane protein in a cell-free system, is to find the right detergent,” said Dr. Zhang. “But screening conditions for membrane proteins is a very expensive and time-consuming process.” Dr. Zhang and colleagues reported the production of several GPCRs, at the milligram scale, by taking advantage of Brij-35, a nonionic polyoxyethylene detergent. By using microscale thermophoresis and circular dichroism, Dr. Zhang and colleagues further revealed that several of these proteins were properly folded and that they bound their small-molecule ligands.

More recently, Dr. Zhang and colleagues reported that short, lipid-like peptide surfactants were comparable to Brij-35 in their ability to solubilize and functionally stabilize several different olfactory receptors purified from a commercial E. coli cell-free protein expression system. Further innovations and creative technologies are needed to accelerate membrane protein studies and thereby speed up drug discovery.

Russell-Body Assays

Recombinant immunoglobulins have attracted significant attention for multiple therapeutic applications, but one of the technical challenges accompanying their wide-scale use is the pronounced heterogeneity that can be seen, in terms of their behavior, when they are overexpressed. While certain recombinant immunoglobulins are relatively soluble, others aggregate inside the endoplasmic reticulum, forming electron-dense structures known as Russell bodies, named after the 19th century pathologist who first described them in myeloma cells.

Even though they are not exact counterparts, Russell bodies and prokaryotic inclusion bodies are analogous and phenotypically similar. However, while inclusion bodies are cytosolic and mostly contain the overexpressed insoluble protein, Russell bodies are inside the endoplasmic reticulum and, to a large extent, contain resident endoplasmic reticulum proteins.

“Even though the Russell bodies were first described about 120 years ago, nobody has associated these protein aggregates with some potential usefulness and value in biotechnology for recombinant antibody expression and production,” said Haruki Hasegawa, Ph.D., cell biologist at Amgen. Capitalizing on the formation of Russell bodies, Dr. Hasegawa and colleagues developed an assay to identify recombinant immunoglobulin clones that are predisposed to aggregation.

“This phenotypic assay reflects the aggregation propensity, and becomes important at the earliest stages of antibody screening,” said Dr. Hasegawa. The assay helps identify, at early research stages, clones that later might turn out to be unsuitable for producing recombinant protein, saving substantial efforts at subsequent steps during development. “Starting with the right clone ensures that every subsequent effort to improve expression will increase the yield, and this is, therefore, by far the most important thing to being with.”

Previous studies revealed that relatively minor changes, such as mutating a single light chain residue, sometimes make a significant difference in terms of solubility, and determining whether the aggregation propensity of a specific construct will worsen or improve during protein engineering is another application where the Russell body assay is very informative.

“This phenotypic assay can guide protein engineering efforts to improve the expression fitness and make sure that one can obtain large antibody titers, which subsequently can be further increased with culture optimization,” said Dr. Hasegawa.


Capitalizing on the formation of Russell bodies, scientists at Amgen have developed an assay to identify recombinant immunoglobulin clones that are predisposed to aggregation. [Used with permission of Amgen Inc.]

Combining Synthetic Antibody Technologies and Structure-Based Design

“Our main goal is to generate proteins that are well folded and can be used for in vitro selection, but most large proteins do not always fold well and may be unstable,” said Sachdev Sidhu, Ph.D., professor at the Banting and Best Department of Medical Research, University of Toronto. To overcome this limitation, a strategy that Dr. Sidhu and colleagues are using is to perform structural analysis to identify smaller, structured protein domains, which can be expressed as a native fold that can be used as a scaffold to engineer desired binding specificities against full-length, endogenous proteins. The ability of these smaller, engineered domains to bind full-length, native cellular proteins is subsequently used to assess their functionality.

A key research effort in Dr. Sidhu’s lab is the synthesis of recombinant antibodies in bacterial and mammalian expression systems. “Bacteria have the advantage of growing very rapidly and enabling the construction of large combinatorial libraries, and after using phage display to perform engineering and design, we use mammalian systems to produce full-length antibodies, which are the closest to the natural ones,” said Dr. Sidhu.

Tailored mutagenesis facilitates the engineering of functional antigen-binding sites with reduced hydrophobicity, which in turn helps generate stabilized, less aggregation-prone proteins. “This illustrates the advantage of combining synthetic antibody technologies with structure-based design,” said Dr. Sidhu.

By using this strategy, Dr. Sidhu and collaborators identified antibodies that target uncleaved and cleaved Ebola virus envelope glycoprotein, and dynamically characterized viral fusion and entry, key steps during viral pathogenesis. “The bigger challenge in biology is to generate antibodies to antigens that are more difficult to express, such as membrane proteins, and developing those expression systems is currently the limiting factor for us and for a lot of groups,” said Dr. Sidhu.


Researchers at the University of Toronto constructed a synthetic antigen-binding fragment library designed to directly compare the contributions that CDR-H3 and CDR-L3 make to antigen recognition. They used a single optimized human framework and diversified the three heavy-chain CDRs and CDR-L3.1. The backbones of the heavy- and light-chain variable domains are shown as tubes. The frameworks are colored gray and the CDR loops are colored as follows: CDR-L3 (purple), CDR-H1 (yellow), CDR-H2 (orange) and CDR-H3 (red). [Jarrett Adams/University of Toronto]

Single-Domain Antibodies

“As a research lab, we have been involved in basic discovery and translational science, and making the transition to commercial use is currently one of the major gaps in the field,” said Mitchell Ho, Ph.D., chief of the antibody therapy section at the National Cancer Institute. Dr. Ho and colleagues are focused on the development of therapeutic antibodies for several types of malignant tumors. This process relies on phage display in E. coli, followed by the transition to mammalian expression systems to generate the constructs of interest.

“In recent years, we became particularly interested in single-domain antibodies,” said Dr. Ho. The lack of the immunoglobulin light chain provides single-domain antibodies with unique functional characteristics because the heavy chains are able to better recognize unique and novel hidden epitopes that would not be accessible for whole immunoglobulin G molecules.

Research efforts in Dr. Ho’s laboratory led to the synthesis of a human heavy-chain, variable-domain antibody, HN3, against glypican-3, which promises to provide a new therapeutic approach for liver cancer. Additionally, Dr. Ho and colleagues recently reported the synthesis of SD1, the first human single-domain antibody that targets tumors expressing mesothelin, a plasma membrane differentiation antigen. This could improve the therapeutic prospects of patients with mesothelin-expressing tumors, which include mesothelioma, ovarian cancer, and pancreatic and lung adenocarcinoma.

Current efforts in Dr. Ho’s laboratory are focused on synthesizing immunocytokines, which consist of antibodies fused to cytokines and are promising therapeutic candidates. “For some of these immunocytokines, expression in mammalian systems is too low, but baculovirus expression systems in insect cells work well, and we showed that the products, in addition to being expressed at high levels, are soluble and active,” said Dr. Ho.

Enhancing Expression Systems

“Significant improvements that we have seen in biologics manufacturing over the years are based on advances in cell-line development and media/process development, but also due to the focus of such efforts on a single cell substrate platform, CHO cells,” said Shyamsundar Subramanian, Ph.D., principal scientist and head of vaccine expression systems and process development, Merck & Co. “Unfortunately, no universal platforms are available yet for vaccines, even if one is restricted to protein antigens.”

Partly due to the variety and diversity of vaccine antigens, this shortcoming creates challenges in the a priori choice of cell substrates, and thus requires the systematic and efficient screening of expression platforms for vaccine antigens. Dr. Subramanian and colleagues have, in the past few years, focused on identifying the most suitable expression systems for vaccine candidates.

The systematic analysis of several expression systems for expressing virus-like envelope particles and bacterial toxins from Clostridium difficile revealed that insect cell platforms are the most successful. “This is the first time that Merck is using insect cells to produce human vaccine candidates,” said Dr. Subramanian. Vaccine candidates such as virus-like particles and bacterial toxins exhibit some levels of cytotoxicity in mammalian expression systems. Moreover, the use of bacterial expression systems can impact product quality and purifiability.

Baculovirus-mediated, transient overexpression in insect cells has turned out to be a particularly attractive option. One of the key lessons from the expression of various antigens is that certain expression systems work better, depending on the protein of interest. “We cannot expect a single expression system to always succeed, but we are trying to focus our attention on a few platforms that are promising,” said Dr. Subramanian.

Hastening Biomedical Progress

Protein expression has become one of the most frequently performed experimental procedures in the life sciences. The synthesis of proteins or protein fragments is a virtually indispensable ingredient of most biomedical endeavors, and represents a critical determinant of their much-anticipated success. With the increasing number of available genomes, and the vast datasets that can be generated with the omics platforms, we can expect that interest in more robust and reliable protein expression, purification, and characterization approaches will only grow in the years to come.

Previous articleBruker Nabs Prairie, Bolstering Its AFM Line
Next article25 Top-Paid Research Institute Leaders