Biotech Research Institute, LG Chem, Taejeon, South Korea.
Codon optimization for high-level expression of human erythropoietin (EPO) in mammalian cells
Codon bias has been observed in many species. The usage of selective codons in a given gene is positively correlated with its expression efficiency. As an experimental approach to study codon-usage effects on heterologous gene expression in mammalian cells, we designed two human erythropoietin (EPO) genes, one in which native codons were systematically substituted with codons frequently found in highly expressed human genes and the other with codons prevalent in yeast genes. Relative performances of the re-engineered EPO genes were evaluated with various combinations of promoters and signal leader sequences. Under the comparable set of combinations, mature EPO gene with human high-frequency codons gave a considerably higher level of expression than that with yeast high-frequency codons. However, the levels of EPO expression varied, depending on the alternate combinations. Since the promoters and the signal leader sequences that we used are known to be equally efficient in gene expression, we hypothesized that the varied expression levels were due to the linear sequence between the promoter and the coding gene sequence. To test this possibility, we designed the EPO gene with hybrid codon usage in which the 5'-proximal region of the EPO gene was synthesized with yeast-biased codons and the rest with human-biased codons. This codon-usage hybrid EPO gene substantially enhanced the level of EPO transcripts and proteins up to 2.9-fold and 13.8-fold, respectively, when compared to the level reached by the original counterpart. Our results suggest that the linear sequence between the promoter and the 5'-proximal region of a gene plays an important role in achieving high-level expression in mammalian cells.
Gene. 199(2-Jan): 293-301.
In this study, the codon usage of a human gene, EPO, has been optimized according to the codon references in highly expressed human and yeast genes (See Fig.1 for the details). Generally, GC ending codons are more preferred in human highly expressed genes whereas AT-ending codons are preferred in yeast. The synthetic genes were expressed in human 293T cells. A very interesting observation was that that the hybrid genes (using the yeast version of the N-terminal part) showed dramatic enhancement of the protein expression. Therefore, the authors believed that the GC composition downstream of the start codon was very important as in the 5'-UTR. However, this strength of this experiment was compromised by the confusing design of the experiment. Firstly, no expression data was reported for the wild-type gene. It appears that the author was using CMVp-CD5L-EPOh gene as a control of the baseline expression. However, using a irrelevant leader sequence from CD5 dramatically compliactes the interpretation of the data. Secondly, it is not clear why the gene is optimized toward yeast codon preference since it is expressed in human cell. Ironically, using the leader sequence with codon optimized for yeast sequence seems to be a better solution. Lastly, no sequence of the 5-UTR available. This limits other researchers to confirm their hypothesis of interaction between promoter and 5'proximal region of the coding sequences. Nevertheless, this observation in this study remains very interesting and inspiring for further testing of what codon optimization really means.
In the paper, base 9 in the natural gene is C instead of G, base 42 is C instead of G