RESUMO
While many computational methods accurately predict destabilizing mutations, identifying stabilizing mutations has remained a challenge, because of their relative rarity. We tested ΔΔG0 predictions from computational predictors such as Rosetta, ThermoMPNN, RaSP, and DeepDDG, using 82 mutants of the bacterial toxin CcdB as a test case. On this dataset, the best computational predictor is ThermoMPNN, which identifies stabilizing mutations with a precision of 68%. However, the average increase in Tm for these predicted mutations was only 1°C for CcdB, and predictions were poorer for a more challenging target, influenza neuraminidase. Using data from multiple previously described yeast surface display libraries and in vitro thermal stability measurements, we trained logistic regression models to identify stabilizing mutations with a precision of 90% and an average increase in Tm of 3°C for CcdB. When such libraries contain a population of mutants with significantly enhanced binding relative to the corresponding wild type, there is no benefit in using computational predictors. It is then possible to predict stabilizing mutations without any training, simply by examining the distribution of mutational binding scores. This avoids laborious steps of in vitro expression, purification, and stability characterization. When this is not the case, combining data from computational predictors with high-throughput experimental binding data enhances the prediction of stabilizing mutations. However, this requires training on stability data measured in vitro with known stabilized mutants. It is thus feasible to predict stabilizing mutations rapidly and accurately for any system of interest that can be subjected to a binding selection or screen.
RESUMO
The Mycobacterium tuberculosis genome harbors nine toxin-antitoxin (TA) systems that are members of the mazEF family, unlike other prokaryotes, which have only one or two. Although the overall tertiary folds of MazF toxins are predicted to be similar, it is unclear how they recognize structurally different RNAs and antitoxins with divergent sequence specificity. Here, we have expressed and purified the individual components and complex of the MazEF6 TA system from M. tuberculosis. Size exclusion chromatography-multiangle light scattering (SEC-MALS) was performed to determine the oligomerization status of the toxin, antitoxin, and the complex in different stoichiometric ratios. The relative stabilities of the proteins were determined by nano-differential scanning fluorimetry (nano-DSF). Microscale thermophoresis (MST) and yeast surface display (YSD) were performed to measure the relative affinities between the cognate toxin-antitoxin partners. The interaction between MazEF6 complexes and cognate promoter DNA was also studied using MST. Analysis of paired-end RNA sequencing data revealed that the overexpression of MazF6 resulted in differential expression of 323 transcripts in M. tuberculosis. Network analysis was performed to identify the nodes from the top-response network. The analysis of mRNA protection ratios resulted in identification of putative MazF6 cleavage site in its native host, M. tuberculosis. IMPORTANCE M. tuberculosis harbors a large number of type II toxin-antitoxin (TA) systems, the exact roles for most of which are unclear. Prior studies have reported that overexpression of several of these type II toxins inhibits bacterial growth and contributes to the formation of drug-tolerant populations in vitro. To obtain insights into M. tuberculosis MazEF6 type II TA system function, we determined stability, oligomeric states, and binding affinities of cognate partners with each other and with their promoter operator DNA. Using RNA-seq data obtained from M. tuberculosis overexpression strains, we have identified putative MazF6 cleavage sites and targets in its native, cellular context.