Advantages and potential limitations of applying AFM kymograph analysis to pharmaceutically relevant membrane proteins in lipid bilayers

Membrane proteins play critical roles in disease and in the disposition of many pharmaceuticals. A prime example is P-glycoprotein (Pgp) which moves a diverse range of drugs across membranes and out of the cell before a therapeutic payload can be delivered. Conventional structural biology methods have provided a valuable framework for comprehending the complex conformational changes underlying Pgp function, which also includes ATPase activity, but the lack of real-time information hinders understanding. Atomic force microscopy (AFM) is a single-molecule technique that is well-suited for studying active membrane proteins in bilayers and is poised to advance the field beyond static snapshots. After verifying Pgp activity in surface-support bilayers, we used kymograph analysis in conjunction with AFM imaging and simulations to study structural transitions at the 100 ms timescale. Though kymographs are frequently employed to boost temporal resolution, the limitations of the method have not been well characterized, especially for sparse non-crystalline distributions of pharmaceutically relevant membrane proteins like Pgp. Common experimental challenges are analyzed, including protein orientation, instrument noise, and drift. Surprisingly, a lateral drift of 75% of the protein dimension leads to only a 12% probability of erroneous state transition detection; average dwell time error achieves a maximum value of 6%. Rotational drift of proteins like Pgp, with azimuthally-dependent maximum heights, can lead to artifactual transitions. Torsional constraints can alleviate this potential pitfall. Confidence in detected transitions can be increased by adding conformation-altering ligands such as non-hydrolysable analogs. Overall, the data indicate that AFM kymographs are a viable method to access conformational dynamics for Pgp, but generalizations of the method should be made with caution.