VAC14 oligomerization is essential for the function of the FAB1/PIKfyve-VAC14-FIG4 complex

Abstract

The PIKfyve-VAC14-FIG4 complex synthesizes and turns over phosphatidylinositol-3,5-bisphosphate, PI(3,5)P₂, an essential signaling lipid. A medium-resolution structure revealed that VAC14 forms a star-shaped pentamer scaffold. Two legs of VAC14 bind FIG4, with one leg also occupied by PIKfyve. The significance of VAC14 oligomerization was unknown. Here, using Alphafold2 and cryogenic electron microscopy maps we generated an atomic-resolution prediction, and found that some mutations linked to pediatric neurodegenerative diseases reside in the VAC14-VAC14 interfaces. A corresponding yeast mutation, along with additional mutations, demonstrates that VAC14 oligomerization is critical for Fab1/PIKfyve function. These mutations cause defects in the generation of PI(3,5)P₂, in VAC14 localization, and in VAC14 oligomerization. Similarly, VAC14 patient mutations expressed in human VAC14 knockout (KO) cells, are defective in the formation of the PIKfyve-VAC14-FIG4 complex, as measured by pull-down assays, are defective in VAC14 oligomerization as measured by fluorescence-detection size-exclusion chromatography of cell lysates, and are defective in colocalization with VPS35-containing endosomes. These studies show that VAC14 oligomerization plays a crucial role in the regulation of PIKfyve/FAB1 and provide insights into selected patient mutations. Moreover, they suggest that small molecules that stabilize the VAC14 complex may provide an intervention for diseases linked to mutations in VAC14.

Figure 1.

Model of the human PIKfyve-VAC14-FIG4 complex using AlphaFold2 Multimer predictions fit to medium-resolution cryo-electron microscopy maps. (A-C) Cytoplasmic facing side (left) and (D-F) membrane facing side (right) of the PIKfyve-VAC14-FIG4 complex. Mutation of VAC14 residue R643 (dark blue) or R681 (red), underlie severe pediatric neurological syndromes. The predicted model suggests that VAC14-R643 interacts with VAC14-D730 (light blue) and that VAC14-R681 interacts with the backbone of VAC14-M702 (pink). To aid in distinguishing between individual VAC14 monomers, each monomer is colored alternating between purple and gray. The interfaces are not identical, and each is numbered in a counterclockwise direction (A, D) Notably residues associated with neurological syndromes, are within predicted VAC14-VAC14 interfaces. The residues are predicted to be close enough to interact with the indicated residue on an adjacent VAC14 monomer. These predictions suggest that the PIKfyve-VAC14-FIG4 complex will be less stable in the patients and this may contribute to the disease. Moreover, the structural predictions suggest that formation or stability of the pentamer is important for PIKfyve function. (B, E) Close up of the predicted interactive pairs. (C, F) The distances between potential interactive pairs in interface #4 are shown. (C) Distances between R643 and D730 within each interface range from 1.9–6.6Å. (F) Distances between R681 and M702 within each interface range from 2.0–7.5Å.

Figure 2.

Mutations at the predicted interfaces of the yeast VAC14 pentamer result in a defect in the yeast PIKfyve/FAB1-VAC14-FIG4 complex as indicated by defects in the synthesis and turnover of PI(3,5)P₂. (A) Recombinant Saccharomyces cerevisiae VAC14 (ScVAC14) fused to His₆ was purified from E. coli. An aliquot of the peak fraction was run on SDS-PAGE (left panel), and the remainder of the sample was used for negative-stain electron microscopy (right panels). Notably, the negative stain 2D class averages of ScVAC14 reveal that it forms a pentamer. (A-B) Due to data-size constraints for AlphaFold2 multimer, truncated ScVAC14-(101–880) was analyzed, and is predicted to assemble into a pentamer. Notably, the pentamer model fits within the 2D class averages. Together, these observations suggest that the ScVAC14 pentamer is similar to human VAC14, and that the VAC14-VAC14 interfaces are partially conserved. (C) Importantly, the residue mutated in the patient, VAC14-R681 is conserved with yeast ScVAC14-R708. Thus, we tested the equivalent mutation in yeast (ScVAC14-R708H*), as well as an alternate substitution, ScVAC14-R708E. In addition, other residues in the predicted human VAC14-VAC14 interfaces are conserved in the ScVAC14-VAC14 interfaces and were tested. For example, ScVAC14-R708 is predicted to form a hydrogen bond with the carbonyl oxygen of the main chain from ScVAC14-M729 (conserved with human VAC14-M702). (C-D) In addition, ScVAC14-M729 is predicted to insert into a hydrophobic patch. Furthermore, ScVAC14-R743 (conserved with human VAC14-R716) is predicted to form a hydrogen bond with the carbonyl oxygen of the main chain of ScVAC14-I731 (conserved with human VAC14-L704). Additional contacts within the predicted interfaces: the side chains of ScVAC14-L709 (conserved with human VAC14-L682) is predicted to insert into a hydrophobic patch. (E) Point mutations in the predicted ScVAC14-VAC14 interfaces impair the acute elevation and decrease in PI(3,5)P₂ levels during salt shock. vac14Δ yeasts were transformed with plasmids expressing Envy-tagged WT ScVAC14, ScVAC14 mutants, or empty vector control. Yeast cells were grown in inositol-free media with myo-³H-inositol overnight and treated with 0.9M NaCl for the times indicated. N=3. Data presented as the mean ± SD. (F) Each of the ScVAC14 mutants tested are stable in yeast. Western blot analysis indicates that the expression and stability of the VAC14 mutants are similar to wild-type ScVAC14.

Figure 3.

Mutations at the predicted ScVAC14-ScVAC14 interfaces result in enlarged vacuoles, loss of ScVAC14 localization to the vacuole, and disruption of ScVAC14 oligomerization. (A) Mutations at the predicted ScVAC14-ScVAC14 interfaces result in enlarged vacuoles and loss of ScVAC14 on the vacuole. A vac14Δ yeast strain was transformed with plasmids expressing Envy-tagged wild-type or mutant ScVAC14 (green) or a control empty vector. Yeast vacuoles were labeled with FM4–64 (magenta). Representative images are shown. Scale bar: 2 μm. (B) For each ScVAC14 mutant, percent cells with multilobed vacuoles were quantified and compared with wild-type ScVAC14. A minimum of 70 cells were counted for each strain per experiment. N=3. (C) Fraction of the vacuole membrane (FM4–64) that contains ScVAC14 (green) was quantified for wild-type yeast, or the ScVAC14-R708H or ScVAC14-M729E mutants. Note that the ScVAC14-R708E, ScVAC14-L709D and ScVAC14-R743E mutants were fully cytosolic. Data presented as mean ± SD. Statistical analysis – one-way ANOVA and Dunnett’s multiple comparisons test. ****P<0.0001, ***P<0.001. (D) ScVAC14 mutants with point mutations at the predicted interface, are defective in oligomerization. A vac14Δ yeast strain was co-transformed with plasmids expressing HA-tagged and Envy-tagged WT VAC14 or Envy-tagged VAC14 mutants or empty vectors. (The band with * is HA-tagged WT VAC14.) VAC14-HA mutants were pulled down and the amount of co-immunoprecipitated Envy-tagged VAC14 was measured. (E) The oligomerization of ScVAC14 mutants relative to wild-type ScVAC14 as accessed by the amount of ScVAC14-Envy pulled-down by ScVAC14-HA was quantified. Data presented as the mean ± SD. Statistical analysis – one-way ANOVA and Dunnett’s multiple comparisons test. Compared with WT, **P<0.01, ****P<0.0001.

Figure 4.

Mutations at the putative ScVAC14-ScVAC14 interface can be partially rescued when co-expressed with wild-type ScVAC14. (A) Co-expression of ScVAC14 mutants with wild-type ScVAC14 partially rescues the ability of mutant ScVAC14 to associate with the vacuole. A vac14Δ yeast strain was co-transformed with plasmids expressing wild-type ScVAC14-HA and Envy-tagged wild-type ScVAC14 or Envy-tagged ScVAC14 mutants (green). Yeast vacuoles were labeled with FM4–64 (magenta). Representative images are shown. Scale bar: 2 μm. (B) Fraction of the vacuole membrane (FM4–64) that contains ScVAC14 (green) was quantified. Data presented as mean ± SD. Statistical analysis – one-way ANOVA and Dunnett’s multiple comparisons test. Compared with WT, **P<0.01, ***P<0.001, ****P<0.0001. (C) ScVAC14 mutants with point mutations at the predicted interface, partially oligomerize with wild-type ScVAC14. A vac14Δ yeast strain was co-transformed with plasmids expressing HA-tagged wild-type ScVAC14 and Envy-tagged wild-type ScVAC14 or Envy-tagged ScVAC14 mutants or a control empty vector. Wild-type VAC14-HA was pulled down and the amount of co-immunoprecipitated Envy-tagged VAC14 was measured. (Representative western blot). (D) The incorporation of VAC14 mutants into a multimer relative to wild-type VAC14 was quantified. Data presented as the mean ± SD. Statistical analysis – one-way ANOVA and Dunnett’s multiple comparisons test. Compared with WT. **P<0.01, and ns, not significant.

Figure 5.

Patient mutations at the predicted interfaces of the human VAC14 pentamer do not suppress formation of enlarged vacuoles in VAC14 KO cells and are defective in their localization to VPS35-containing membranes. (A) VAC14 knockout MDA-MB-231 cells were either mock-transfected or transfected with both free GFP and the indicated FLAG-tagged VAC14 plasmids. 24 hours later, live cells were imaged. Representative images shown. Scale bar: 2 μm. (B) The percentage of cell area occupied by the vacuole area was measured. Individual cells are plotted in grey while the mean and SEM of each repeat are plotted in black. Data reported relative to the wild-type controls. Statistical significance from three independent experiments was determined using nested one-way ANOVA and Tukey’s multiple comparisons test. ***p < 0.001, ****p < 0.0001, and ns, not significant. (C) Patient mutations at the predicted interfaces of the human VAC14 pentamer are defective in their ability to localize to VPS35-containing membranes. VAC14 knockout HT1080 cells were transiently transfected with the indicated FLAG-tagged VAC14 plasmids for 6 hrs. Cells were semi-permeabilized with digitonin to remove overexpressed cytosolic FLAG-VAC14. Then cells were fixed, permeabilized with saponin, and immuno-stained with an anti-VAC14 antibody, an anti-VPS35 antibody, and corresponding Alexa-Fluor-conjugated secondary antibodies. Scale bar: 20 μm. (D) The fraction of VPS35 that colocalizes with the indicated FLAG-tagged VAC14 mutant was determined from three independent experiments. Individual cells are plotted in grey and the mean and SEM of each repeat are plotted in black. Statistical significance from three independent experiments was determined using nested one-way ANOVA and Tukey’s multiple comparisons test. ****p < 0.0001.

Figure 6.

VAC14 patient mutations are defective in oligomerization and result in defects in the stability of the PIKfyve-VAC14-FIG4 complex. (A) Fluorescence-detection Size Exclusion Chromatography of lysates from HT1080 VAC14 KO cells transiently transfected with the following plasmids: wild-type VAC14 (black), VAC14-R643W (blue), VAC14-R681H (red), VAC14-R681E (brown), and free mGreenLantern (green). Arrows indicate observed peaks of fluorescence at elution volumes of 18.2 mL, 15.8 mL, 12.9 mL, and 8.9 mL. The left-most peak likely represents the void volume. Based on standards (insert on right), the next peak, 12.9 mL, 1090 kDa is possibly the intact PIKfyve-VAC14-FIG4 complex (915 kDa) or a dimer of the VAC14 pentamer (1148 kDa). The next peak,170 kDa, is likely a monomer of the VAC14-mGreen Lantern fusion protein (115 kDa), and the last peak, 33 kDa, is free mGreen lantern (27 kDa). (B) Co-immunoprecipitation studies reveal that the VAC14 mutants are defective in forming a complex with PIKfyve and VAC14. VAC14 knockout HeLa cells were co-transfected with a plasmid expressing wild-type VAC14-FLAG or the indicated VAC14-FLAG mutant, as well as a plasmid expressing the tetracycline repressor, used to decrease the expression level of VAC14-FLAG. VAC14-FLAG was pulled down and the amount of co-immunoprecipitated PIKfyve and FIG4 was detected by western blot analysis. Representative blots of a pull-down and corresponding input were shown. (C) Quantification of incorporation of FIG4 into the complex. (D) Quantification of the incorporation of PIKfyve into the complex. N=3. Data presented as the mean ± SD. Statistical analysis – one-way ANOVA and Dunnett’s multiple comparisons test. **p < 0.01, ****p < 0.0001.