Understanding Functional Protein Production: Synonymous Codon Contributions and Development of a Cell- Based Protein Folding Screening Tool

Proteins participate in many important functions in the cell, from serving as catalysts in biochemical pathways to providing structural integrity. To carry out their functions, proteins must fold into complex, three-dimensional structures. If proteins misfold or aggregate, their function is impaired, which can lead to disease, such as Alzheimer’s and Parkinson’s disease. Protein folding to the native state can be promoted or disrupted by both amino acid identity and synonymous codon usage. Early work by Christian B. Anfinsen showed that the amino acid identity was important for directing small proteins to their native state. However, the information contained in amino acid identity alone is not always sufficient to drive larger, more complex proteins to their native state. The inability of larger proteins to refold in vitro suggests that cellular context is crucial for directing proteins to fold properly. In vitro conditions lack chaperones, cytoplasmic crowding, and varying translation elongation rates. Elongation rate can be modulated by synonymous codons through their variable decoding rates and can therefore influence the co-translational folding process of the nascent protein.

Progress in the field of protein folding has included the understanding that a sequence containing more rare codon substitutions is more likely to misfold, and that exchanging hydrophobic amino acids with polar ones is likely to significantly disrupt folding. Despite these significant advances, we are still unable to build predictive tools for protein folding. One way to build a reliable predictive tool for protein folding is to couple large experimental datasets with machine-learning, but this has been difficult to accomplish due to a lack of tools that can map thousands of protein sequences to a folding outcome in the cellular environment. Additionally, synonymous codons are known to participate in many aspects of functional protein production and disentangling the impact of synonymous codons on folding from their other effects in the cell has proven challenging, in large part due to a lack of folding models.

In Chapter 2, I describe my use of the previously established synonymous codon folding model chloramphenicol acetyltransferase III (CAT) to probe synonymous codon effects on functional protein production in E. coli. We define functional protein production as the net effect of transcription (mRNA synthesis), mRNA half-life, translation (protein synthesis) and the probability of a protein folding correctly to its active, functional structure. Although many synonymous codon substitutions did not significantly disrupt CAT protein folding in vivo, I discovered a novel role of synonymous codons in functional protein production. Specifically, synonymous codon substitutions in the cat gene upregulated the basal expression levels of an upstream gene which shares a promoter with cat (known as an overlapping, divergent promoter, ODP). The upregulation of the neighboring gene, tetR, which is a transcription factor that binds to the shared promoter region, repressed expression of several synonymous cat mutants. Increased repression by TetR resulted in decreased accumulation of CAT which resulted in decreased E. coli fitness when cells were challenged with the antibiotic cam, due to insufficient accumulation of CAT to fight off the antibiotic. This is a significant finding given that (1) ODP’s are widespread in E. coli, which means this form of regulation may be pervasive and (2) synonymous codons may be under an additional, previously unrecognized selective pressure to regulate expression of neighboring genes.

In Chapter 3, I describe the development of a protein folding screening tool (folding reporter) that can capture changes to folding upon amino acid mutations in the context of the cellular environment. This screening tool enables the rapid screening of many mutants at once for folding perturbations. Rapid in vivo screening is necessary for mapping folding information to amino acid sequence to couple with machine-learning and for building predictive tools that can determine if a protein is likely to fold to its native state from amino acid sequence alone. An additional promising application of the folding reporter is to use it as a discovery tool for synonymous codon folding models to study synonymous codon effects experimentally. Having the capability to rapidly measure protein folding perturbations in the cell will significantly expand our understanding of protein folding by providing large amounts of in vivo relevant data that can be used to predict folding outcomes and to identify new folding models.