cspA directed evolution for cold-temperature protein production

recombinant proteins are a >$271 billion market spanning biopharmaceuticals like insulin to industrial enzymes, but they are hard to make efficiently and are prone to misfolding and aggregation. industry already lowers culture temperatures from 37 ºC toward 15 ºC to raise both yield and quality, yet those hypothermic shifts act globally on cell physiology, slowing growth instead of targeting the genetic element regulating the protein of interest. this project asked whether that low-temperature boost could be engineered directly into the genetics rather than imposed on the whole cell.

the project targets the cspA cold-shock inducible expression system of E. coli for directed evolution. cspA has a native promoter (PcspA) and a long 5′ untranslated region that behaves as an RNA thermometer: unstable at 37 ºC but rapidly stabilizing mRNA at low temperatures. the idea is to tune this system so it amplifies the natural production gains of cold growth. using the reporter plasmid pUA66-PcspA-GFP (GFP under the native cspA regulatory element), three regions were identified for investigation: the native PcspA promoter, the 5′ UTR upstream box, and a suspected downstream box in the cspA coding sequence. the work split across two arms:

• dry lab — the salis lab RBS calculator was used to predict translation initiation rates (TIR) for sequence variants, focusing mutations on unconserved positions so cold-inducibility is preserved. upstream box variants reached up to a 2.6-fold predicted TIR increase over wild-type, while downstream box variants topped out around 1.6-fold, identifying the upstream box as the more promising bottleneck to optimize. higher-performing variants shared reduced mRNA secondary structure and stronger predicted ribosome binding.
• wet lab — the reporter plasmid was purified (an optimized miniprep raised yield to ~100 ng/µL), PCR conditions were tuned to linearize the plasmid and excise the native PcspA promoter, and transformation protocols were validated. two well-characterized anderson-collection promoters (J23100 and J23119) were chosen as drop-in replacements to test promoter swapping, and an error-prone PCR workflow was designed for the upstream and downstream boxes.

the wet lab hit a wall at assembly: inserting the short (~77 bp with overlaps) anderson promoter oligos into the linearized plasmid failed under both CPEC and NEBuilder, and lab time and space ran out before the epPCR workflow could run, so the full experimental validation was not completed. the main contribution was the outline itself: a reusable framework for applying directed evolution to cold-shock inducible expression systems. it pointed to screening natural and engineered promoter libraries, using CRISPR-based mutagenesis for more flexibility than epPCR, and training machine-learning models (e.g. EVMP and deep generative models) within design-build-test-learn cycles to predict promoter strength — all aimed at better biomanufacturing of proteins that are sensitive at 37 ºC, such as monoclonal antibodies, and at extending the same approach to other cold-shock elements and production hosts like CHO cells.