Extended Data Fig. 8: Possible molecular model of GS autopalmitoylation.

a, Structure of human GS, showing its bifunnel-shaped catalytic site. A schematic of the GS decamer is shown from the top and front views with individual subunits A and B labelled and coloured grey and green, respectively. On the right is a close-up of the bifunnel catalytic site that is formed between subunits A and B. The GS decamer has ten active sites, each located at the interface of two adjacent subunits. ATP enters from the top, whereas glutamate enters from below; manganese ions (Mn²⁺) are shown as grey spheres. b, Molecular dynamics simulation of palmitoyl-CoA in the catalytic cleft of GS predicts that, whereas the head of palmitoyl-CoA is tightly bound to the adenine-binding site, the tail can point in opposing directions with respect to the principal axis of the protein. The most representative structures of the two alternative conformations (A and B) observed during the long molecular dynamics simulations for palmitoyl-CoA binding to GS (in blue, seen from two different perspectives) are shown in red (A, tail bending upwards) and green (B, tail bending downwards). c, Detailed view of conformation A, which is the main conformation. The sulfur atom of palmitoyl-CoA (which is immediately adjacent to the carbon on which the nucleophilic attack occurs) (coloured yellow) approaches the highly conserved cysteine 209 (also coloured yellow), with an interatomic distance (S–S) that, during the simulations, reversibly fluctuates between 3 and 8 Å. The hydrophobic tail positions itself along grooves characterized by the presence of hydrophobic residues. Colour coding is as follows: carbon, grey; nitrogen, blue; phosphorous, gold; oxygen, red. Cysteines and serines within 5 Å of the palmitoyl tail are highlighted in yellow and orange, respectively. The hydrophobic residues around the tail are shown in green. d, Detailed view of conformation B, in which the tail is found in a buried hydrophobic cleft, with the sulfur at a distance of 5 Å or less from the conserved serines 65 and 75 and the tail occupying the site of the GS inhibitor MSO. Details of the extensive steric clash between MSO and the secondary binding pose (B) observed in palmitoyl-CoA MD simulations are shown. Palmitoyl-CoA is represented as sticks, with standard atomic colours as stated in c. MSO is shown in cyan and its position is taken from entry 2QC8 of the Protein Data Bank. Cysteines and serines within 5 Å of the palmitoyl tail are highlighted in yellow and orange, respectively. The hydrophobic residues around the tail are shown in green. e, GS immunoblotting after streptavidin pull-down of biotin-azide-clicked lysates from 16C-YA (palmitoylation probe) labelled HEK-293T cells overexpressing wild-type GLUL or GLUL with a point mutation in C209. The input shows the level of GS overexpression. A representative blot from 4 independent experiments is shown. f, g, Quantification of total sprout length (f) and number of sprouts per spheroid (g) for control and GLUL^KD ECs with or without overexpression of shRNA-resistant GLUL encoding the point mutation C209A (rGLUL^C209A-OE) (mean ± s.e.m.; n = 4 independent experiments; *P < 0.05 versus control, one-way ANOVA with Dunnett’s multiple comparison versus control). h, Schematic of protein autopalmitoylation. Upon binding of palmitoyl-CoA to the protein, free CoA (grey oval) is released and can be detected. i, Recombinant wild-type GS and GS with point mutations R324C and R341C were incubated with different concentrations of palmitoyl-CoA in a cell-free system at physiological pH. The amount of CoA released per minute was determined as a direct readout for protein autopalmitoylation. Data are mean ± s.e.m. of 3 (R324C and R341C) and 4 (WT) independent experiments. NS, P > 0.05; *P < 0.05 according to two-way ANOVA comparing the entire dose–response to the dose–response of wild-type GS. j, Different amounts of recombinant wild-type, R324C and R341C GS were incubated with a fixed concentration of palmitoyl-CoA (40 μM), and the amount of CoA released per minute was determined as readout for autopalmitoylation. Data are mean ± s.e.m. of 4 (R324C and R341C) and 5 (wild-type) independent experiments. NS, P > 0.05; *P < 0.05 according to two-way ANOVA comparing the entire dose–response to the dose–response of wild-type GS. The data for wild-type GS from i and j are also included in Extended Data Fig. 7 as stand-alone data, but are included here for comparison purposes. k, Boyden chamber migration for control, GLUL^KD, GLUL^KD + rGLUL^OE, GLUL^KD + rGLUL^R341C-OE and GLUL^KD + rGLUL^R324C-OE ECs, all under mitomycin C-treatment (mean ± s.e.m.; n = 3 independent experiments; NS, P > 0.05; *P < 0.05, one-way ANOVA with Dunnett’s multiple comparison versus control). Exact P values: GLUL^KD versus control: 0.0004; GLUL^KD + rGLUL^C209A-OE versus control: 0.0004 (f); GLUL^KD versus control: 0.0001; GLUL^KD + rGLUL^C209A-OE versus control: 0.0001 (g); R324C versus WT: 0.8228; R341C versus WT: 0.7530 (i); R324C versus WT: 0.1331; R341C versus WT: 0.0003 (j); GLUL^KD versus control: 0.0054; GLUL^KD + rGLUL^OE versus control: 0.8152; GLUL^KD + rGLUL^R341C-OE versus control: 0.3645; GLUL^KD + rGLUL^R324C-OE versus control: 0.2118 (k). For gel source images, see Supplementary Fig. 1.