Golgi-localized phosphatidylinositol 4-kinase β mediates Rab11a activation and trafficking to promote ciliogenesis

Abstract

Ciliogenesis requires an orchestrated interaction between the Golgi apparatus and centrioles via vesicle trafficking, yet this process is still poorly understood. Phosphatidylinositol 4-kinase β (PI4KB) is a conserved kinase that localizes to the Golgi for generating phosphatidylinositol 4-phosphate, an important lipid component related to cilium formation. Here, we demonstrate a previously uncharacterized mechanism of PI4KB in regulating Rab11a to enable proper ciliogenesis. PI4KB kinase activity maintains the normal vesicle density around the Golgi and Rab11a localization to centrioles in ciliogenesis. Inhibition of PI4KB activity leads to the reduced centriole localization but accumulation of Rab11a on the Golgi. We identified that the activation of Rab11a relies on PI4KB activity, while inactive Rab11a-guanosine diphosphate stably associates with Golgi and fails to undergo outward delivery. Autosomal-dominant nonsyndromic sensorineural hearing loss mutations in PI4KB abnormally intensify PI4KB interaction with Rab11a, leading to the aberrant subcellular Rab11a localization and defective ciliogenesis. Collectively, our study delineates a critical role for PI4KB in post-Golgi vesicle formation and ciliogenesis.

Fig. 1.. PI4KB is required for ciliogenesis.

(A) Immunofluorescence (IF) of motile cilia in tracheal epithelium labeled with acetylated α-tubulin (ac-tub; green) in WT and Pi4kb heterozygous mice (Het-1 and Het-2). Scale bar, 10 μm. (B) Quantification of fluorescence intensity in brackets in (A). n, mouse number; au, arbitrary units. (C and D) Scanning electron microscopy (SEM) of tracheal cilia in WT and heterozygous mice and quantification. Scale bar, 10 μm. Random regions [(C), top] were subdivided into nine grids [(C), bottom], and grids containing cilia were quantified. n, field number. (E and F) IF and quantification of motile cilia in choroid plexus labeled with acetylated α-tubulin (green). Arrows indicate cilia tips. Brackets indicate the cilia region. Scale bar, 10 μm. n, mouse number. (G and H) IF and quantification of primary cilia in renal tubules labeled with Arl13b (green). Scale bar, 10 μm. n, mouse number. (I) Western blot (WB) of PI4KB in tissues from 8-week-old WT mice and embryonic day 15 MEFs. IB, immunoblot. (J) WB of PI4KB in WT (control), KO, and rescue cells that are KO cells expressing GFP-PI4KB. (K and L) IF and quantification of ciliation in WT, KO, and rescue cells. Cilia were labeled with Arl13b (red) and PI4KB or GFP-PI4KB with anti-PI4KB (green). Scale bars, 10 μm. n, cell number from three experiments. (M and N) IF and quantification of ciliation in cells treated with PI4KB kinase inhibitor (KI). Cilia were labeled with Arl13b (green) and γ-tubulin (γ-tub; red). Insets show cilium morphology. Note the reduced ciliation rate in KI cells. Scale bars, 10 μm. n, cell number from three experiments. All data are presented as the mean of three independent experiments or animals. Statistical significance is based on a two-tailed t test (*P ≤ 0.05; **P ≤ 0.01); ns indicates no significant difference.

Fig. 2.. Subcellular localization and role of PI4KB in ciliogenesis.

(A) IF showing PI4KB (red) and GM130 (green) in RPE cells. Insets show colocalization details. Scale bars, 25 μm (main) and 5 μm (insets). (B) IF and quantification of GFP-PI4KB on cilium (top) and PI4KB on basal bodies (bottom). Cilia (red) were labeled with anti–acetylated tubulin. Centrioles (green) in bottom panel were labeled with anti–γ-tubulin. Peak values show the fluorescence intensities on the yellow lines. Neither exogenous nor endogenous PI4KB has clear localization to cilium or basal bodies. Scale bars, 5 μm. (C) Schematic and IF images of full-length PI4KB and truncations. GFP-PI4KB^1–305 and GFP-PI4KB^306–801 colocalization with Golgi (GM130; red) is shown, with arrows indicating the Golgi colocalization of GFP-PI4KB^1–305. Scale bar, 10 μm. (D) Quantification of Golgi-colocalized truncated PI4KB proteins. Black boxes indicate percentage of discernible Golgi colocalization. (E) Quantification of ciliogenesis rates in PI4KB KO cells rescued with either full-length or truncated GFP-PI4KB. (F) Removal of CP110 from mother centrioles in PI4KB-KI–treated and control cells. Insets show cilia (red) and CP110 (green). Scale bars, 10 μm. h, hours. (G) Percentage of cells with CP110 present as one dot on two centrioles in (F). (H) IF showing IFT20 (green) localization in PI4KB-KI and control cells. Centrioles were labeled with γ-tubulin antibody. Dashed lines mark cell outlines, and arrows indicate IFT20 at centriole. Scale bars, 5 μm. (I) Quantification of cells with clear IFT20 localization at centriole. The data are presented as the mean of three independent experiments. n equals the cell number from three independent experiments. Statistical significance is based on a two-tailed t test (*P ≤ 0.05; **P ≤ 0.01).

Fig. 3.. PI4KB activity is required for the pericentriolar vesicle accumulation and peri-Golgi vesicle formation.

(A) IF of vesicles labeled with SmoM2-GFP (green) accumulating around centrioles in control cells but not in PI4KB-KI–treated cells in early ciliogenesis. Centrioles were labeled with CEP164 (magenta) antibody. Scale bars, 25 μm. (B) The fluorescence intensity along the dashed lines a′ and b′ in (A). (C) WB showing the expression levels of SmoM2-GFP in control and PI4KB-KI cells. (D) TEM images showing the pericentriolar vesicles in control and PI4KB-KI–treated cells. Scale bar, 50 nm. (E) Schematic drawing showing the that vesicle density around centrioles is reduced in PI4KB-KI cells. (F) Quantification of vesicle numbers in (D). n equals the centriole number. (G) TEM images showing the peri-Golgi vesicles in control and PI4KB-KI–treated cells. Zoomed-in images showing the vesicles are indicated by arrows. Colored asterisks show the corresponding zoomed-in regions. Scale bars, 50 nm. (H) Quantification of vesicle numbers in (G). n equals the Golgi number. (I) Schematic image showing the position of TEM sectioning and step size on Golgi in (G). Quantification was performed by quantifying the average number of vesicles within the zone by red dashed line. (J) IF images showing the IFT20 (green) labeled with anti-IFT20 antibody around Golgi in control and PI4KB-KI–treated cells. Zoomed-in images show the detailed localization of IFT20 on the Golgi. Scale bars, 5 μm. (K) The fluorescence intensity along the dashed lines in (J) with red (a′) representing control cells and blue (b′) for PI4KB-KI–treated cells. The data are presented as the mean of three independent experiments. Statistical significance is based on a two-tailed t test (**P ≤ 0.01).

Fig. 4.. PI4KB activity is required for the normal subcellular distribution of Rab11a in ciliogenesis.

(A) IF showing Rab11a and PI4KB in control and PI4KB-KI–treated cells. Centrioles were labeled with anti-CEP164, and Rab11a was labeled with anti-Rab11a. Insets show Rab11a at Golgi region. Scale bars, 10 μm. (B) SIM images showing Rab11a (green) around centrioles (γ-tubulin, red) in control and PI4KB-KI cells. Scale bar, 500 nm. (C) Quantification of cells with Rab11a localized at centriole region. (D) WB of Rab11a expression in control and PI4KB-KI cells. (E) Schematic showing Rab11a at Golgi and centrioles: Rab11a is transported to centriole region in WT but retained at Golgi when PI4KB activity is inhibited. (F) PI4KB localization at Golgi under untreated, PI4KB-KI, and washout conditions (Golgi, anti-GM130; PI4KB, anti-PI4KB). PI4KB accumulates at Golgi upon inhibition and is restored after washout. Scale bar, 25 μm. (G) Quantification of the fluorescence intensity of PI4KB at Golgi region in (F). (H) Pull-down assays testing the binding affinity between Flag-Rab11a and GFP-PI4KB or GFP-PI4KB^D656A either with or without the presence of PI4KB-KI in HEK-293T cells. IP, immunoprecipitation. (I) Quantification of gray values of GFP-PI4KB or GFP-PI4KB^D656A bands in (H). (J) IF showing GFP-PI4KB or GFP-PI4KB^D656A (green) with Rab11a (red) at Golgi (magenta) in serum-starved cells. Rab11a shows reduced centriole-centered localization in GFP-PI4KB^D656A cells in bottom panels. Scale bars, 10 μm. (K) IF showing Rab11a localization in WT and PI4KB KO cells. Golgi was labeled with anti-GM130, and Rab11a was labeled with anti-Rab11a. Rab11a centers at centrioles in controls (arrows) but accumulates at Golgi region in KO cells. Scale bar, 5 μm. The data are presented as the mean of three independent experiments. n, cell number from three experiments. Statistical significance is based on a two-tailed t test (*P ≤ 0.05; **P ≤ 0.01).

Fig. 5.. SNHL mutations on PI4KB undermine the subcellular distribution of Rab11a and ciliogenesis.

(A) Schematic diagram showing the SNHL missense mutations in PI4KB, referred to as PI4KB^SNHL. (B) IF images showing the localization of Rab11a (red), GM130 (magenta), GFP-PI4KB, and GFP-PI4KB^SNHL (green). Arrowheads indicate that the Rab11a becomes accumulated at the Golgi region. Zoomed-in images show the details of Rab11a at the Golgi region. Scale bars, 25 μm (main images) and 10 μm (insets). (C and D) WBs showing the expression levels of GFP-PI4KB, GFP-PI4KB^SNHL, and Rab11a with the quantification in (D). The data are presented as the mean of three independent experiments. (E) Localization of Rab11a (red) in GFP-PI4KB (green)– and GFP-PI4KB^Q121R (green)–expressing cells. Rab11a was labeled with anti-Rab11a. Rab11a displays puncta that show colocalization with the SNHL missense mutation GFP-PI4KB^Q121R in the cytosol, as indicated by arrows. Scale bar, 10 μm. Fluorescence intensities along the dashed lines are shown in (b) and (c). (F) Pull-down assays exploring the binding affinity between Flag-Rab11a and GFP-PI4KB^SNHL or GFP-PI4KB in HEK-293T cells with numbers indicating the gray values. (G) IF images showing the ciliation in control and PI4KB KO cells rescued with GFP, GFP-PI4KB, GFP-PI4KB^D656A, and GFP-PI4KB harboring SNHL missense mutations (GFP-PI4KB^Q121R and GFP-PI4KB^V434G, respectively). Details of cilium and basal bodies are shown in the insets. Cells were labeled with anti-Arl13b (red) to show cilium and anti–γ-tubulin (magenta) to show basal bodies. Scale bars, 10 μm. (H) Percentage of ciliated cells from (G) and fig. S5E, with cell numbers counted as shown from the three independent experiments. Statistical significance is based on a two-tailed t test (*P ≤ 0.05; **P ≤ 0.01).

Fig. 6.. Active/inactive forms of Rab11a display two different subcellular localizations, and Rab11a activation requires PI4KB in ciliogenesis.

(A and B) IF of GFP-Rab11a^Q70L and GFP-Rab11a^S25N in ciliogenesis. Endogenous PI4KB (red) was labeled with anti-PI4KB. Insets show the colocalization of PI4KB with Rab11a^Q70L in early ciliogenesis but not Rab11a^S25N in the cytosolic region. Scale bars, 20 μm (main) and 5 μm (insets). (C) Pull-down assay using glutathione S-transferase (GST)–FIP3-C as the bait for active Rab11a in cell lysates from control or PI4KB-KI–treated cells, showing that Rab11a-GTP level was reduced in PI4KB-KI cells. These data are presented of three independent experiments. (D) IF showing the distribution of GFP-Rab11a^Q70L after 6 hours of serum starvation with or without the PI4KB-KI. Endogenous PI4KB was labeled with anti-PI4KB in red. Golgi was labeled with anti-GM130 in magenta. Numbers 1 and 2 show the Golgi region and cytosolic region, respectively. Colocalization of GFP-Rab11a^Q70L, GM130, and PI4KB in two regions are shown in insets. Zoomed-in images show the merged channels at the Golgi region. Scale bars, 20 μm (main images) and 5 μm (insets). (E) Quantification of GFP-Rab11a^Q70L fluorescence intensity at the Golgi region at T0 and 6 hours (T6) after serum starvation. n equals the cell number from three independent experiments. (F) IF images showing the distribution of GFP-Rab11a^S25N after 6 hours of serum starvation with or without the presence of PI4KB-KI. Endogenous PI4KB was labeled with anti-PI4KB in red. Golgi was labeled with anti-GM130 in magenta. Zoomed-in images show the detailed colocalization of GFP-Rab11a^S25N, GM130, and PI4KB at the Golgi regions. Scale bars, 20 μm (main images) and 5 μm (insets). Statistical significance is based on a two-tailed t test (*P ≤ 0.05; **P ≤ 0.01). h, hours.

Fig. 7.. Rab11a has stronger binding affinity to PI than PI4P, and inhibition of PI4KB affects TF endocytosis cycling to ERC.

(A and B) Lipid strip (100 pmol per spot) overlay assay, showing that Rab11a, Rab11a^S25N, and Rab11a^Q70L bind predominately to PI. Lipid abbreviations in PIP strip diagram (A). LPA, lysophosphatidic acid; LPC, lysophosphocholine; PE, phosphatidylethanolamine; PC, phosphatidylcholine; S1P, sphingosine 1-phosphate; PA, phosphatidic acid; PS, phosphatidylserine; dye, blue marker dye. Numbers in (B) indicate the gray values of the spots normalized to S1P. The exposure time is indicated. (C and D) IF images and timeline of transferrin (TF) endocytosis assay. TF was conjugated to Alexa Fluor 568 in red, Rab11a was labeled with anti-Rab11a in green, and centrioles were labeled with anti-CEP164 in magenta in (C). Arrow in zoomed-in indicates the accumulation of TF at the endosomal recycling compartment (ERC) in control cell. The dashed arrow indicates the absence of clear accumulation of TF at ERC in PI4KB-KI cells. Scale bars, 25 μm. The diagram in (D) shows the timeline in the assay. Note that the cells were serum starved for 24 hours to induce cilia. (E and F) Quantifications of percentage of cells that have the ERC accumulation of TF (E) and Rab11a (F). n equals the cell number from three independent experiments. Data are presented as the mean of three independent experiments. Statistical significance is based on a two-tailed t test (**P ≤ 0.01). h, hours.

Fig. 8.. PI4KB activity is required for the post–Golgi vesicle secretion and centriolar accumulation of active Rab11a in ciliogenesis.