Many challenges confront researchers attempting to express recombinant proteins. Post-translational modifications, quantity produced, and recapitulating function are chief among them. These and other issues will be discussed at CHI’s “Difficult to Express Proteins” conference to be held later this month. Cutting-edge solutions include new ways to optimize gene sequences, novel fusion proteins or tags, and enhancements to functional assays for membrane proteins.
Many lines of evidence have demonstrated that changing the protein coding sequence of a gene can dramatically affect its expression. According to Mark Welch, Ph.D., director of gene design at DNA2.0, “the key is to know which codons to change. Although researchers previously had to isolate and clone genes of interest, we can now tailor-make genes synthetically. This allows us to optimize genes based upon what organism will be used for expression.”
Dr. Welch notes that the field is currently lacking systematic studies showing what gene characteristics are optimal. “There’s a lot of protein-expression folklore involved in design principles, yet also an absence of uniform experimental data. There are a confounding number of variables among examples in the literature that differ in the proteins expressed, vectors, host strains, etc. In each example you generally are presented with only two data points, a natural gene and a synthetic version. There has been no reliable way to combine results from these often contradictory experiments.”
To address this issue, DNA2.0 embarked on a detailed analysis in which it designed and independently synthesized about 40 genes each for two proteins, a DNA polymerase, and a single chain antibody. Each set systematically sampled codon usage and other variables thought relevant to protein expression.
“We found that synonymous codon variation produced more than a 40-fold difference in expression levels. We identified sequence characteristics that correlated with expression by combining multivariate regression methods along with genetic algorithms. Contrary to popular assumptions, we found that codon preferences did not correlate with the codon bias found in natural host genes.”
Dr. Welch suggests that DNA2.0’s strategy can be used in any expression system. It already has been successfully applied to expression in mammalian, yeast, plant, and fungal hosts. “Systematic variation and modeling provide a more reliable way to improve gene design for any host. In every study we’ve obtained dramatic improvements relative to prior optimization strategies. This adds great value to our genes, particularly for clients expressing high-value proteins where even a small boost in expression can mean considerable savings in time and money.”