Sampling the conformational landscapes of transporters and receptors with AlphaFold2

Abstract

Equilibrium fluctuations and triggered conformational changes often underlie the functional cycles of membrane proteins. For example, transporters mediate the passage of molecules across cell membranes by alternating between inward-facing (IF) and outward-facing (OF) states, while receptors undergo intracellular structural rearrangements that initiate signaling cascades. Although the conformational plasticity of these proteins has historically posed a challenge for traditional de novo protein structure prediction pipelines, the recent success of AlphaFold2 (AF2) in CASP14 culminated in the modeling of a transporter in multiple conformations to high accuracy. Given that AF2 was designed to predict static structures of proteins, it remains unclear if this result represents an underexplored capability to accurately predict multiple conformations and/or structural heterogeneity. Here, we present an approach to drive AF2 to sample alternative conformations of topologically diverse transporters and G-protein coupled receptors (GPCRs) that are absent from the AF2 training set. Whereas models generated using the default AF2 pipeline are conformationally homogeneous and nearly identical to one another, reducing the depth of the input multiple sequence alignments (MSAs) led to the generation of accurate models in multiple conformations. In our benchmark, these conformations were observed to span the range between two experimental structures of interest, suggesting that our protocol allows sampling of the conformational landscape at the energy minimum. Nevertheless, our results also highlight the need for the next generation of deep learning algorithms to be designed to predict ensembles of biophysically relevant states.

Systematic identification of conditionally folded intrinsically disordered regions by AlphaFold2

“…If AlphaFold2 can recover the inherent conformational heterogeneity in the bound/modified forms of IDRs/IDPs, such predictions would be valuable to generate testable hypotheses and search for plausible structured conformations that are accessible to a given IDR/IDP. Interestingly, two recent studies proposed that either reducing the depth of the MSA or performing rational in silico mutagenesis within the MSA can drive AlphaFold2 to sample alternate conformations (Alamo et al 2021; Stein & Mchaourab 2021). It remains to be tested if this approach is generally applicable and amenable to IDRs/IDPs.…”

Application of Homology Modeling by Enhanced Profile–Profile Alignment and Flexible-Fitting Simulation to Cryo-EM Based Structure Determination

“…However, multimer predictions based on AF2 [33][34][35] do not show a performance that is as good as the monomer prediction of AF2. Furthermore, it is expected that there are cases in which application of AF2 is not straightforward, not only because of the difficulty posed by multimers but also because AF2 is designed to generate only one state of a protein [84,85]. Additionally, we performed the modeling of TPC2 based on the model generated by AF2.…”

Modeling Alternate Conformations with Alphafold2 via Modification of the Multiple Sequence Alignment

“…In a recent preprint, a different methodology for sampling protein conformational space with AF2 was described. 23 To obtain multiple conformations several alterations in the AF2 pipeline were made, there was no recycling within the AF2 module and the number of sequences of the MSA seen by AF2 was modified. In addition, the set of 8 protein targets had neither of the structures describing the conformational change in AF2’s training set.…”