Journal: Journal of molecular biology

Article Title: Identification of Iron-Sulfur (Fe-S) Cluster and Zinc (Zn) Binding Sites Within Proteomes Predicted by DeepMind's AlphaFold2 Program Dramatically Expands the Metalloproteome.

doi: 10.1016/j.jmb.2021.167377

Figure Lengend Snippet: Figure 1. Schematic illustration of our ligand-search protocol and examples of identified metal binding sites in AlphaFold2 structures. (a) a flowchart outlining the method. For the protein illustrated in the center of the flowchart, two “regions” are highlighted; these are distinct from one another, and are treated as separate entities when enumerating all possible ligand superpositions. (b) an example of an identified 4Fe-4S cluster coordinated by four cysteine residues (ligand type “4Fe-4S Cys4”; UniProt accession: P08201). (c) an example of a cluster of three closely spaced Zn binding sites coordinated by a total of eight cysteine residues; this cluster is formed by three successful additions of the ligand type “Zn Cys4” (UniProt accession: P05100). In all images of proteins in this figure, the protein is colored by each residue’s pLDDT score with dark blue indicating a “very high confidence” prediction by AlphaFold2 and dark red indicting a “low confidence” prediction.

Article Snippet: DeepMind’s AlphaFold2 software has ushered in a revolution in high quality, 3D protein structure prediction.

Techniques: Binding Assay

Journal: Journal of molecular biology

Article Title: Identification of Iron-Sulfur (Fe-S) Cluster and Zinc (Zn) Binding Sites Within Proteomes Predicted by DeepMind's AlphaFold2 Program Dramatically Expands the Metalloproteome.

doi: 10.1016/j.jmb.2021.167377

Figure Lengend Snippet: Figure 6. Example 4Fe-4S cluster binding sites annotated in UniProt that are not identified in this work. In each image protein chains are shown as transparent cartoons colored using AlphaFold20s pLDDT (i.e. confidence) scores on a spectrum of red (low confidence) to blue (high confidence), coordinating residues are shown as licorice, and iron and sulfur atoms in the 4Fe-4S cluster are shown as yellow and pink spheres, respectively. (a) an example in which AlphaFold2 builds two 4Fe-4S sites out of residues annotated in two different UniProt sites but swaps two cysteine residues between each of the sites: the two UniProt sites are shown with their carbon atoms colored in either cyan or green and the residue numbers for the corresponding sites are displayed above the image with arrows to indicate the cysteines that are swapped. (b) an example in which AlphaFold2 confidently builds a “4Fe-4S Cys4” site as “4Fe-4S Cys3” site and rejects the annotation of the fourth residue. (c) an example in which AlphaFold2 does not build the UniProt-annotated residues into a 4Fe-4S binding site but instead confidently builds the residues in an a helix (opaque blue cartoon).

Article Snippet: DeepMind’s AlphaFold2 software has ushered in a revolution in high quality, 3D protein structure prediction.

Techniques: Binding Assay, Residue

Journal: Journal of molecular biology

Article Title: Identification of Iron-Sulfur (Fe-S) Cluster and Zinc (Zn) Binding Sites Within Proteomes Predicted by DeepMind's AlphaFold2 Program Dramatically Expands the Metalloproteome.

doi: 10.1016/j.jmb.2021.167377

Figure Lengend Snippet: Figure 7. Example 2Fe-2S cluster binding sites annotated in UniProt that are not identified in this work; visualization scheme is the same as used in Figure 6. (a) An example in which AlphaFold2 confidently builds disulfide bonds in the annotated UniProt site. (b) An example in which AlphaFold2 confidently builds a potential site that is too constricted to coordinate a 2Fe-2S cluster. (c) An example in which AlphaFold2 builds the fourth coordinating residue in a position that is more than 5 A from the three other coordinating residues. (d) An example in which AlphaFold2 builds a potential 2Fe-2S site with a different fourth coordinating residue (carbon atom colored green) than that which is annotated in UniProt (red arrow).

Article Snippet: DeepMind’s AlphaFold2 software has ushered in a revolution in high quality, 3D protein structure prediction.

Techniques: Binding Assay, Residue

Journal: ACS Omega

Article Title: A Structural Network Analysis of Neuronal ArhGAP21/23 Interactors by Computational Modeling

doi: 10.1021/acsomega.2c08054

Figure Lengend Snippet: Modeled structure of the RhoGAP domains of ArhGAP21, ArhGAP23, and DmRhoGAP19D. (a) Schematic representation of ArhGAP21 and ArhGAP23 (orange box: PDZ domain, blue box: PH domain, green box: RhoGAP domain). UniProt and GenBank IDs are as follows: ArhGAP21: Q5T5U3-1/NM_20824.4; ArhGAP23: Q9P227-1/NM_001199417.2; and DmRhoGAP19D: Q9VRA6/NM_134552.6. (b) Modeled structure of the RhoGAP domain by homology modeling. RhoGAP domains are colored from the N-terminus (blue) to the C-terminus (red). (c) The modeled structure of the ArhGAP21 domain was superimposed over the RhoGAP domain of ArhGAP23. (d) Modeled structure of the RhoGAP domain generated by AlphaFold2. (e) The modeled structure of the ArhGAP21 RhoGAP domain was superimposed over the RhoGAP domain of ArhGAP23. The modeled structure was evaluated by PROCHECK and ERRAT and visualized by the UCSF Chimera software program, version 1.15 ( http://www.cgl.ucsf.edu/chimera ).

Article Snippet: The RhoGAP domains of ArhGAP21, ArhGAP23, and RhoGAP19D retrieved from InterPro were also modeled ( Figure d) using the AlphaFold2 software program, and the two human RhoGAP domains showed similar folding structures (RMSD: 0.452 Å) ( Figure e).

Techniques: Generated, Software