Fig. 1: Engineering regulatory networks for complex phenotypes in E. coli.

a The workflow for engineering regulatory networks for complex phenotypes in E. coli. We picked 82 regulators that regulate or interact with ~4000 genes in total to compose a diverse and comprehensive regulatory network library. Then, we designed, constructed, and mapped the regulatory network library containing 110,120 specific mutations expected to perturb metabolism. We screened the library for different targeted phenotypes, and we identified the mutants with desired inner metabolism for the targeted phenotypes. b Plasmid barcode‐based mapping of enriched variants across all targeted genes under different conditions. From the inner circle to the outer circle, they represent the log₂ enrichment scores of variants under 2 g/L furfural, 300 mg/L styrene, and 30 g/L acetate. The four highly enriched genes for each selection were labeled with different colors. c Mutations in RpoB shown on the structural model (PDB: 3IYD). The previously unidentified mutated residue is shown in red, and previously reported residues are shown in blue. d Growth verification of the reconstructed variants compared with nontargeting control under 2 g/L furfural. n = 3 for each curve. Error bars show mean value ± SD. e A partial structural model of LexA (PDB: 3JSO). The previously unidentified mutated residue is shown in red, previously reported residues are shown in blue, and the mutated residue that is enriched in multiple selections is shown in purple. f Growth verification of the reconstructed variants compared with nontargeting control under 300 mg/L styrene. n = 3 for each curve. Error bars show mean value ± SD. g A partial structural model of SoxR (PDB: 2ZHG). The mutated residues for the acetate experiment are shown in red, and the mutated residues for furfural experiment are shown in blue. h Growth verification of the reconstructed variants compared with nontargeting control under 30 g/L acetate. n = 3 for each curve. Error bars show mean value ± SD.