Fig 2: PI4KB and GBF1(364–395) form ternary complexes with ARMH3 and Arl5B.A, AlphaFold3 prediction of ARMH3 in complex with PI4KB(481–502), Arl5B, and cofactors GTP and Mg2+ colored by chain. Peptides of interest are colored and labelled accordingly. B, AlphaFold3 prediction of ARMH3 in complex with GBF1(364–389), Arl5B, and cofactors GTP and Mg2+ colored by chain. Peptides of interest are colored and labelled accordingly. C, bar graphs showing differences in percent deuterium incorporation for 0.3s (3s at 0 °C) for select peptides of interest, highlighting significant decreases in exchange relative to apo PI4KB, ARMH3, or GBF1(364–395) upon formation of ARMH3 and Arl5B-containing complexes. (Technical replicates, individual data points are shown, error bars are SD, n = 3 for each condition). D, schematic outlining PI4KB-ARMH3-Arl5B and GBF1-ARMH3-Arl5B complex formation. Text in red denotes phosphorylation site on GBF1 and PI4KB, text in parentheses denotes PI4KB isoform two residue numbering.

Fig 3: Molecular basis of the PI4KB-ARMH3 interaction.A, domain schematic of the full-length PI4KB isoform two construct used in this paper. B, sum of the number of deuteron differences in ARMH3 upon binding to PI4KB over the entire deuterium exchange time course, re-analyzed with additional ARMH3 peptides from ARMH3-Arl5B and ARMH3-GBF1 experiments included. Each point represents the centre residue of an individual peptide. Peptides that met the significance criteria (defined as >0.4 Da, >5%, and p < 0.01 in an unpaired two-tailed t test at any time point) are coloured red. Error is shown of the sum of standard deviations (SDs) across all four time points (n = 3). C, predicted aligned error (pae) plot of the AlphaFold3 prediction of the ARMH3-PI4KB complex. D, AlphaFold3 prediction of ARMH3 in complex with PI4KB, coloured by domain. E, AlphaFold3 prediction of ARMH3 in complex with PI4KB coloured by significant changes in deuterium exchange in both ARMH3 (grey) and PI4KB (wheat) upon complex formation. F, zoom-in on the predicted ARMH3-PI4KB N-lobe linker interface, showing select predicted contact residues of ARMH3 and PI4KB labeled and shown as sticks. G, dose response of PI4KB binding to ARMH3. His-PI4KB was loaded onto the anti-penta His tip at 50 nM and dipped into ARMH3 (10 nM, 40 nM). All curves were fit with a partial, 1:1 binding model. The KD value reported was generated using the average KD given for each curve meeting inclusion criteria, error is reported as standard deviation (n = 4). H, schematic of the PI4KB-ARMH3 biolayer interferometry (BLI) assay, showing binding of immobilised His-PI4KB on the tip to full-length ARMH3 in solution. I, BLI association and dissociation curves of ARMH3 and PI4KB mutants compared to wild-type. Either His-PI4KB or His-ARMH3 was loaded onto the anti-penta His tip at 50 nM and dipped into ARMH3 or PI4KB at 50 nM. PI4KB RL494EA and ARMH3 D549R are representative interfacial residues; ARMH3 K53E is an ARMH3-Arl5B complex disruptor included as a control.

Fig 4: Biolayer interferometry and HDX-MS reveal canonical GTPase-effector interaction.A, domain schematics of full-length ARMH3 and Arl5. Constructs used in this paper are Arl5A(15–179) and Arl5B(15–179), which are referred to as Arl5A and Arl5B throughout this paper. B, schematic of the biolayer interferometry (BLI) assay showing binding of immobilised His-Arl5 on the tip to full-length ARMH3 in solution. C, BLI traces of Arl5B(GTPγS) and Arl5B(GDP) binding to full-length ARMH3. His-Arl5B(GTPγS) or His-Arl5B(GDP) was loaded onto the anti-penta His tip at 25 nM and dipped into ARMH3 at 100 nM. D, comparison of BLI traces generated from Arl5B(GTPγS) and Arl5A(GTPγS) interacting with ARMH3. His-Arl5B(GTPγS) or His-Arl5A(GTPγS) was loaded onto the anti-penta His tip at 25 nM and dipped into ARMH3 at 100 nM. E, dose response of Arl5B(GTPγS) binding to ARMH3. His-Arl5B(GTPγS) was loaded onto the anti-penta His tip at 25 nM and dipped into ARMH3 (15–100 nM). All curves were fit with a partial, 1:1 binding model. The KD value reported was generated using the average KD given for each curve meeting inclusion criteria, error is reported as standard deviation (n = 5). F, sum of the number of deuteron differences in ARMH3 upon binding to Arl5B(GTPγS) analyzed over the entire deuterium exchange time course. Each point represents the centre residue of an individual peptide. Peptides that met the significance criteria (defined as >0.4 Da, >5%, and p < 0.01 in an unpaired two-tailed t test at any time point) are coloured red. Error is shown of the sum of standard deviations (SDs) across all five time points (n = 3). G, selected deuterium exchange time courses that showed significant decreases in exchange upon complex formation. Error is shown as SD (n = 3). H, AlphaFold3 prediction of ARMH3 in complex with Arl5B and co-factors GTP and Mg2+ coloured by chain, important structural features of ARMH3 and Arl5B are annotated. I, predicted aligned error (pae) plot of the AlphaFold3 prediction of the ARMH3-Arl5B(GTP) complex with co-factors GTP and Mg2+. J, AlphaFold3 model of the ARMH3-Arl5B complex coloured by significant decreases in deuterium exchange in ARMH3 upon binding to Arl5B(GTPγS). Boxes highlight significant peptides at the ARMH3-Arl5B interface and distal regions of ARMH3 showing allosteric changes.