Logo: to the web site of Uppsala University

T Transcription factors (TFs) must locate and selectively bind their DNA targets while remaining responsive to cellular signals. In my thesis project, I used the lac repressor (LacI) as a model to investigate the dominant microscopic parameter for sequence specificity and study how conformation switch bias influences TF-DNA interaction phenotypes. I also developed a new methodology for high-throughput mapping of these phenotypes across a vast protein sequence space. In Paper I, a three-state model was derived, which links macroscopic association (ka) and dissociation (kd) rates to microscopic determinants, and predicts an anti-correlation between ka and kd. High-throughput kinetics on protein-binding microarrays (HT-k-PBMs) across 2,479 Lac operator variants confirmed this anti-correlation. We found that the inferred variation in recognition probability (ptot​) exceeded that of the microscopic off-rate (koff,μ) by ~1.7-fold, conclusively demonstrating that sequence specificity is governed primarily by association. In Paper II, we tested a mechanistic hypothesis that ptot​ is set by the conformational switch centered on the LacI hinge region, which was proposed as the mechanism to balance speed and stability of target binding.  We engineered two hinge-helix mutants of LacI—V52A, which increases helical propensity, and Q55N, which decreases it—and quantified their kinetic phenotypes both in vitro using HT-k-PBM and in vivo by measuring repression strength with the Miller assay and target search and binding rates using single-molecule measurements in living bacterial cells. Relative to WT-LacI, the in vitro macroscopic kinetics (ka, kd, Kd) of the engineered mutants shifted in opposite directions. This translated to distinct in-cell performances: V52A strengthened repression but with reduced DNA specificity and loss of inducibility, while Q55N weakened repression but increased specificity and retained inducibility. Notably, neither variant measurably altered the target search speed in cells relative to WT-LacI. Finally, in Paper III, we present an Optical Pooled Screening (OPS) method that combines chromosomal dual barcodes with pooled λ-Red recombineering to scale single-molecule phenotyping to many LacI variants expressed from the chromosome. In the pilot study, 5/6 strains were correctly decoded, and the expected phenotypes were recovered. We also outlined key constraints needing design refinements prior to full-scale implementation.