Combining cysteine scanning with chemical labeling to map protein-protein interactions and infer bound structure in an intrinsically disordered region

The Mycobacterium tuberculosis genome harbours nine toxin-antitoxin (TA) systems of the mazEF family. These consist of two proteins, a toxin and an antitoxin, encoded in an operon. While the toxin has a conserved fold, the antitoxins are structurally diverse and the toxin binding region is typically intrinsically disordered before binding. We describe high throughput methodology for accurate mapping of interfacial residues and apply it to three MazEF complexes. The method involves screening one partner protein against a panel of chemically masked single cysteine mutants of its interacting partner, displayed on the surface of yeast cells. Such libraries have much lower diversity than those generated by saturation mutagenesis, simplifying library generation and data analysis. Further, because of the steric bulk of the masking reagent, labeling of virtually all exposed epitope residues should result in loss of binding, and buried residues are inaccessible to the labeling reagent. The binding residues are deciphered by probing the loss of binding to the labeled cognate partner by flow cytometry. Using this methodology, we have identified the interfacial residues for MazEF3, MazEF6 and MazEF9 TA systems of M. tuberculosis. In the case of MazEF9, where a crystal structure was available, there was excellent agreement between our predictions and the crystal structure, superior to those with AlphaFold2. We also report detailed biophysical characterization of the MazEF3 and MazEF9 TA systems and measured the relative affinities between cognate and non-cognate toxin–antitoxin partners in order to probe possible cross-talk between these systems.


Supplementary Figures
(each measuring point), an MST trace is recorded. All traces are then normalised to start at 1000. For each trace, the FNorm value for the dose -response curve is calculated from Fhot (MST laser on)/Fcold (MST laser off). The KD values for the interactions were determined employing standard data analysis with MO.Affinity analysis software (5s MST-on time for evaluation) (Wienken et al., 2010;Jerabek-Willemsen et al., 2011;Seidel et al., 2012;Chattopadhyay et al., 2022). in MazE3 is close to the predicted interface. (C-D) Overlay of binding histograms of labeled and unlabeled cells shown in red and blue respectively. In surface displayed (C) WT MazE3 (the binding of cognate WT partner is reduced after cell surface labeling (left panels), indicating that the WT cysteine is part of the interface. However, upon mutation of the cysteine residue to alanine, in MazE3-C98A there is no change in binding after labeling. The cysteine library for MazE3 was thus constructed in the Cys→Ala background. Since, C62, C71 in MazF3 are not a part of the interface, the Ala mutants did not show any decrease in binding signal after labeling, however since MazF3 C62A-C71A, had much higher expression and binding signal as compared to the WT, the final MazF3 DMS library was made in the double mutant background.         Table S3. Summary of the k-means clustering parameters used to identify interacting residues.

Supplementary Tables
Class 1 and Class 2 refer to mutant with WT like or lower binding affinity than WT respectively. AlphaFold2 and homology models respectively. The third column is the experimentally identified mutants of the MazE3 cysteine library from the vertical gate. The (┼) sign in magenta represents reduction in binding upon mutation and labeling. The orange represents reduced binding (≥15%) upon mutation, but no further reduction was observed upon labeling. The black sign represents no change in binding upon mutation or labeling. Red represent no change in binding after mutation but increase in binding after labeling (≥1.2 fold).

Predicted interacting
residues from the homology model ┼ * -: Not chosen for cysteine mutagenesis *: No read counts associated with this mutant or total read count less than the cut-off used. AlphaFold2 and homology models respectively. The third column is the experimentally identified mutants of the MazF3 cysteine library from the vertical gate. The (┼) sign in magenta represents reduction in binding upon mutation and labeling. The orange represents reduced binding (≥15%) upon mutation, but no further reduction was observed upon labeling. The black sign represents no change in binding upon mutation or labeling. Red represent no change in binding after mutation but increase in binding after labeling (≥1.2 fold).

Mutants
Predicted interacting residues from AlphaFold2 model

Identified from the DMS-FACS
┼ --: Not chosen for cysteine mutagenesis *: No read counts associated with this mutant or total read count less than the cut-off used. Table S6. List of mutants of MazE6 selected for cysteine scanning mutagenesis, and interface residues in MazEF6 AlphaFold2 or homology models, or inferred from Cys scanning. In the first and second columns, the (┼) sign indicates interface mutants selected based on the MazEF6 AlphaFold2 and homology models respectively. The third column is the experimentally identified mutants of the MazE6 cysteine library from the vertical gate. The (┼) sign in magenta represents reduction in binding upon mutation and labeling. The orange represents reduced binding (≥15%) upon mutation, but no further reduction was observed upon labeling. The black sign represents no change in binding upon mutation or labeling. Red represent no change in binding after mutation but increase in binding after labeling (≥1.2 fold).

Mutants
Predicted interacting residues from AlphaFold2 model

Identified from the DMS-FACS
┼ * -: Not chosen for cysteine mutagenesis *: No read counts associated with this mutant or total read count less than the cut-off used. AlphaFold2 and homology models respectively. The third column is the experimentally identified mutants of the MazF6 cysteine library from the vertical gate. The (┼) sign in magenta represents reduction in binding upon mutation and labeling. The orange represents reduced binding (≥15%) upon mutation, but no further reduction was observed upon labeling. The black sign represents no change in binding upon mutation or labeling. Red represent no change in binding after mutation but increase in binding after labeling (≥1.2 fold).

Mutants
Predicted interacting residues from AlphaFold2 model Predicted interacting residues from the homology model ┼ --: Not chosen for cysteine mutagenesis *: No read counts associated with this mutant or total read count less than the cut-off used.  No read counts associated with this mutant or total read count less than the cut-off used. Table S9. List of mutants of MazF9 selected for cysteine scanning mutagenesis, and interface residues either identified from the MazEF9 crystal structure or from the DMS-FACS experimental data. In the first and second columns, the (┼) sign indicates interface mutants selected based on the MazEF9 AlphaFold2 and homology models respectively. The third column is the experimentally identified mutants of the MazF9 cysteine library from the FACS. The fifth column shows the corresponding ΔASA values from the MazEF9 crystal structure (PDB ID: 6KYT). The (┼) sign in magenta represents reduction in binding upon mutation and labeling. The orange represents reduced binding (≥15%) upon mutation, but no further reduction was observed upon labeling. The black sign represents no change in binding upon mutation or labeling. Red represent no change increase in binding after labeling (≥1.2 fold).