Causes and effects of N-terminal codon bias (Goodman, Church and Kosuri). Science

Influential approach Combine 2 Techs

• oligo synthesis on a programmable microarray – diverse
• These authors combine with next gen sequencing – 10-200 million unique sequences in a pool
• use the next gen sequencing to measure how biology changes the diversity of the synthesized library.
• Kosuri (former Church post-doc, started OpenWetWare)

This paper’s application

• differential codon use. up 20 fold bias in ecoli between synonymous codons
• varies between genes, can be used to affect expression.
• recent observation: usage of codons varies as a function of position in the gene (along gene length).
• Start codon proximal sequence enriched for rare codons (in flies, worms, yeast and E coli).

experimental design

• diverse library of sfGFP, with different 5′ sequence: strong vs weak promoter, strong vs weak RBS, the ATG, the 10 amino-acids from 137 native e coli genes, and 13 different re-codings of this n-terminal frequency.
• 2 promoters 4 RBS, 137 genes 13 codings = 14,248 plasmids
• do with oligo-array library and create 14,248 distinct E coli strains in 1 shot.
• Get 600,000 clones (average 50x coverage), could imagine doing more.
• do next gen sequencing to for the number of ecoli strains that have each plasmid.
• use ratio of mCherry to GFP to reduce effects of extrinsic noise (ribosomes conc etc).
• sort library via FACs on GFP/RFP
• each insert /plasmid is a unique barcode, can ask how many copies of this barcode do you find in each expression bin.
• Have a binned expression measurement for each of 14K sequences.
• can see how N-terminal codon bias affects expression on a 14K
• different reshuffled recodings span a range of free energies of secondary structure.
• rare codon, increase expression (on average ~2 fold, at most 100 fold)
• most rare Arg codon get a 2x more expression.
• how do rare codons increase protein expression? Reject the slow translation initation reduces traffic jams (reducing density) to increase overall speed. — Don’t see correlation with RBS strength. (other groups have also rejected this model by showing different codons translated at the same rate)
• coding variants which change secondary structure stability
• rare codons tend to destabilize secondary structure (just by straight-up correlation plot).
• what mattered was the secondary structure not the abundance. All swaps that don’t actually change commonality of the structure but do change secondary structure, see a correlation.
• look at all ones that keep deltaG constant but do change abundance, see no effect.
• max change observed in deltaG see ~10 fold change in expression level.
• rare codons are AT enriched and more expressed and less secondary structure (rare codons tend to have A in wobble base)

what’s new?

• previous demonstration of secondary structure had 28 or 150 constructs, now 14K constructs.