Skip to main content
. 2022 Mar 3;11:e75751. doi: 10.7554/eLife.75751

Figure 1. Alternative conformations of transporters and G-protein-coupled receptors (GPCRs) can be predicted by AlphaFold2 (AF2).

(A) Representative models of the transporter LAT1 in inward-facing (IF) and outward-facing (OF) conformations. Experimental structures shown in gray and models shown in colors. (B) Comparison of AF2 models with inactive/active or IF/OF experimental structures as a function of multiple sequence alignment (MSA) depth for GPCRs (top) and transporters (bottom), respectively. All models shown here were generated without templates. Dashed lines indicate the template modeling (TM) score between experimental structures and are shown for reference. (C) Supplementing shallow MSAs with OF templates allows AF2 to predict the OF conformation of MCT1. (D) Experimental structures superimposed over models with the greatest TM scores. Inactive/IF and active/OF cartoons shown on the top and bottom in teal and orange, respectively.

Figure 1.

Figure 1—figure supplement 1. Example principal component analysis (PCA) of ASCT2 models generated by AlphaFold2 (AF2) containing outlier models.

Figure 1—figure supplement 1.

Colors correspond to multiple sequence alignment (MSA) depth in sequences, while orange and teal diamonds refer to the outward- and inward-facing experimental structures, respectively. Left: outlier models are indicated by gray arrows and could be clearly delineated along PC2. Right: removal of these models reveal collective variables along PC2.
Figure 1—figure supplement 2. Conformational homogeneity as a function of multiple sequence alignment (MSA) depth.

Figure 1—figure supplement 2.

Conformational homogeneity was defined by classifying models as more similar to one of the two experimental structures based on their template modeling (TM) scores and calculating the fraction of models in the larger group. MCT1 and proteins in the training set (CCR5, MurJ, PfMATE, and SERT) were biased toward one specific conformation (uniformity = 1) even when using shallow MSAs. X-Axis shown on log-2 scale for clarity. Models generated using MSAs with 16 sequences were omitted from this figure.
Figure 1—figure supplement 3. Templates contribute to conformational sampling only when shallow multiple sequence alignments (MSAs) are provided.

Figure 1—figure supplement 3.

(A) PTH1R. (B) LAT1.
Figure 1—figure supplement 4. Protein targets with one conformation in the training set cannot be predicted in the alternative conformation.

Figure 1—figure supplement 4.

The following experimental structures served as references for the training and the alternative conformations, respectively: 5UIWr and 7F1Qr (CCR5), 5T77a and 6NC9a (MurJ), 3VVNa and 6FHZa (PfMATE), and 5I6Xa and 6DZZa (SERT).