The inositol 5-phosphatase INPP5K participates in the fine control of ER organization

Abstract

INPP5K (SKIP) is an inositol 5-phosphatase that localizes in part to the endoplasmic reticulum (ER). We show that recruitment of INPP5K to the ER is mediated by ARL6IP1, which shares features of ER-shaping proteins. Like ARL6IP1, INPP5K is preferentially localized in ER tubules and enriched, relative to other ER resident proteins (Sec61β, VAPB, and Sac1), in newly formed tubules that grow along microtubule tracks. Depletion of either INPP5K or ARL6IP1 results in the increase of ER sheets. In a convergent but independent study, a screen for mutations affecting the distribution of the ER network in dendrites of the PVD neurons of Caenorhabditis elegans led to the isolation of mutants in CIL-1, which encodes the INPP5K worm orthologue. The mutant phenotype was rescued by expression of wild type, but not of catalytically inactive CIL-1. Our results reveal an unexpected role of an ER localized polyphosphoinositide phosphatase in the fine control of ER network organization.

Figure 1.

ER localization of INPP5K and ARL6IP1. (A) Domain organization of yeast INP54 and human INPP5K. Note the presence of an ER-anchoring tail at the C terminus of yeast INP54. Human INPP5K lacks such an anchoring ER and instead contains the SKICH domain. (B and C) Confocal images of COS-7 cells expressing EGFP-INPP5K alone (B) or coexpressing EGFP-INPP5K and the ER marker mCh-VAPB, an intrinsic membrane protein (C), showing the dual localization of EGFP-INPP5K in the cytosol (indicated by the diffuse fluorescence in B) and in the ER (shown by the colocalization with mCh-VAPB on tubular structures and the line scan analysis in C). Representative examples of several cells imaged in at least three independent experiments. All transfected cells exhibited the phenotype shown. Scale bars: 10 µm in B, 5 µm in C. (D and E) Confocal images of COS-7 cells expressing EGFP-ARL6IP1 (C) or coexpressing EGFP-ARL6IP1 and mCh-VAPB (D), demonstrating the localization of EGFP-ARL6IP1 in the ER (shown by the colocalization with mCh-VAPB on tubular structures and the line scan analysis in D). Representative examples of cells imaged in at least three independent experiments. All transfected cells exhibited the phenotype shown. Scale bars: 10 µm in D, 5 µm in E. (F) Representative confocal images and corresponding line-scan analysis of COS-7 cells showing that EGFP-INPP5K has a dual localization in the ER and cytosol in cells only expressing endogenous ARL6IP1, while it is primarily recruited to the ER when coexpressed with mCh-ARL6IP1 and loses its ER localization upon ARL6IP1 knockdown. Scale bars: 5 µm. The ratio of EGFP-INPP5K fluorescence intensity in the ER relative to the adjacent cytosol under the three conditions is plotted in the right panel. Data were acquired by line-scan analyses of multiple individual ER tubules from multiple cells (“Endogenous”: n = 39 ER tubules from 9 cells; “Overexpressed”: n = 39 ER tubules from 11 cells; “Depleted”: n = 33 ER tubules from 10 cells). ****, p < 0.0001 (two-tailed t test). (G and H) Representative confocal image of mouse cortical neurons grown 4 d in vitro and transfected with EGFP-INPP5K and mCh-ARL6IP1, showing their localization in the ER reticular network. Scale bars: 10 µm in G, 2 µm in H.

Figure 2.

INPP5K is recruited to the ER via the interaction with ARL6IP1. (A) Left: INPP5K domain structure and deletion constructs used for the experiments shown in B and C. Right: ARL6IP1 domain structure and predicted topology. TM, transmembrane regions. L1 to L3, three cytosolically exposed regions. Segments highlighted in blue were replaced with flexible linkers of equivalent length comprising myc tags and these constructs were used for experiments shown in F. (B–H) Representative confocal images and corresponding line-scan analysis of COS-7 cells coexpressing the EGFP-INPP5K and mCh-ARL6IP1 constructs depicted in A. (B) WT INPP5K and WT ARL6IP1 colocalize in the ER. (C) The C-terminal 17-aa segment of INPP5K is dispensable for colocalization with ARL6IP1. (D and E) Neither the 5-phosphatase domain only (INPP5K^1–361) nor the SKICH domain only (INPP5K^276–448) is sufficient to bind ER-bound ARL6IP1. (F) The INPP5K^I363T patient mutation strongly impairs the recruitment to the ER. (G and H) The recruitment of INPP5K to the ER is strongly reduced when the N-terminal region of ARL6IP1 is replaced by another sequence (G) but remains unchanged when the C-terminal region of ARL6IP1 is replaced (H). The replacement of the L1 segment, but not the L3 segment, of ARL6IP1 nearly abolishes the recruitment of WT INPP5K to the ER. Hot spots of ARL6IP1^L3-myc possibly reflect misfolded proteins. Note these hot spots are not enriched for INPP5K (see Results). Scale bars: 5 µm. (I) Plot of the ratio of EGFP-INPP5K fluorescence intensity on the ER relative to the adjacent cytosol based on line-scan analysis of individual tubules from multiple cells. Data for cells expressing INPP5K^WT and ARL6IP1^WT were the same as those shown in Fig. 1 F. (INPP5K^WT and ARL6IP1^WT: n = 39 ER tubules from nine cells; INPP5K^1–361 and ARL6IP1^WT: n = 31 ER tubules from four cells; INPP5K^276–448 and ARL6IP1^WT: n = 35 ER tubules from four cells; INPP5K^I363T and ARL6IP1^WT: n = 31 ER tubules from four cells; INPP5K^WT and ARL6IP1^L1-Myc: n = 32 ER tubules from six cells; INPP5K^WT and ARL6IP1^L3-Myc: n = 32 ER tubules from four cells). ****, p < 0.0001; n.s., not significant (two-tailed t test). (J) Extracts of HeLa cells transfected with HA-INPP5K and the indicated EGFP-tagged constructs were subjected to anti-GFP immunoprecipitation (IP) and then processed for SDS-PAGE and immunoblotting (IB) with anti-HA antibody. Left: Representative blot from three independent experiments. The bar graph on the right shows quantification of HA-INPP5K coprecipitated with EGFP-tagged proteins normalized to input (from densitometric scans of gel bands). Data are presented as mean ± SEM, n = 3; **, p < 0.01; *, p < 0.05; n.s., not significant (two-tailed t test).

Figure 3.

Preferential localization of ARL6IP1 and INPP5K, relative to other ER proteins, in peripheral ER tubules. (A) COS-7 cell coexpressing the ER membrane marker EGFP-Sec61β, mCh-INPP5K, and Myc-ARL6IP1 imaged by confocal fluorescence microscopy. Both EGFP-Sec61β and mCh-INPP5K fluorescence are present throughout the ER tubular network, but mCh-INPP5K is enriched over EGFP-Sec61β in the peripheral tubular ER (as marked by arrowheads) and is nearly undetectable in ER sheets. Scale bars: 5 µm. Insets at the bottom left corners show the nuclear envelope of a different cell from the same field, demonstrating that mCh-INPP5K fluorescence is absent from the nuclear envelope marked by EGFP-Sec61β fluorescence. Inset scale bars: 2 µm. Representative examples of several cells imaged in at least three independent experiments. (B) COS-7 cell coexpressing EGFP-Sec61β and mCh-ARL6IP1, showing the relative enrichment of mCh-ARL6IP1 over EGFP-Sec61β in the peripheral tubular ER (as marked by arrowheads) and the presence of EGFP-Sec61β, but not mCh-ARL6IP1, on the nuclear envelope. Scale bars: 5 µm. Representative examples of several cells imaged in at least three independent experiments. (C and D) Representative confocal images of COS-7 cells coexpressing mCh-ARL6IP1 and EGFP-Sec61β showing regions including the nuclear envelope (C) or peripheral ER sheets (D), respectively. Graphs show a representative example of the quantification of the normalized fluorescence intensity measured along the dashed lines as delineated in the merged fields. Note the lack of mCh-ARL6IP1 signal from the nuclear envelope (B) and from the peripheral ER sheets, except for their edges (C), while EGFP-Sec61β labels these structures (arrowheads). Scale bars: 2 µm. (E–I) Representative confocal images and respective line-scan analysis of the periphery of COS-7 cells coexpressing mCh-ARL6IP1 and other GFP-tagged proteins as indicated. While ARL6IP1 and INPP5K are homogenously present on all the ER tubules and precisely localized (E), the fluorescence of ER membrane proteins Sec61β, VAPB, and Sac1 and of the luminal ER marker ss-GFPox-KDEL declines toward the ends of the most distal tubules (F–I, arrowheads). Scale bars: 5 µm. Representative examples of several cells imaged in at least three independent experiments. (J) Quantification of normalized fluorescent intensity of mCh-ARL6IP1 and ss-GFPox-KDEL along peripheral ER tubules based on line scans as exemplified on the top. Note that the dashed line elongates beyond the ends of the ER tubules into the background. Data are represented as mean ± SD (n = 13 ER tubules from six cells). The bar graphs at bottom right show the average fluorescence intensity along 1-µm segments at the proximal and distal end, respectively. ****, p < 0.0001; n.s., not significant (two-tailed t test).

Figure 4.

ER tubules populated by ARL6IP1 and INPP5K undergo rapid motion. (A–D) Analysis of the motility of ER proteins. (A and C) Representative confocal images of COS-7 cells coexpressing EGFP-ARL6IP1 and mCh-VAPB (top fields of A) or mCh-ARL6IP1 and EGFP-Sec61β (top fields of C). In the images shown, three time points were color-coded and merged, so that the stationary ER elements staying on the same sets of pixels appear white, while the motile ER elements occupying alternate pixels appear in colors. Note the abundance of the colored ER tubules in ARL6IP1 images, relative to the VAPB and Sec61β (arrowheads). Bottom: Graphic display of motility from the same field shown above during a 5-min recording. Differences of fluorescence intensity at each pixel between subsequent time-lapse images were calculated, and these values are added up and pseudocolored (see Fig. S3 for methods). Scale bars: 5 µm. (B and D) Plots of motility index (see Fig. S3 for methods) where cumulative values of motility in several cells, calculated as in A and B, were normalized to the initial fluorescence intensity. n = 21 cells expressing EGFP-ARL6IP1 and mCh-VAPB (C), and n = 24 cells for cells expressing mCh-ARL6IP1 and EGFP-Sec61β (D). Data are represented as scattered dots with the solid black bar as mean (two-tailed t test). (E) Histograms of transport distances and velocity of the tips of motile ER tubules in cells overexpressing EGFP-ARL6IP1 (out of 81 events from five cells). All the events shown represent ER tubule movement that originated and ended during the recorded time. (F) Example of a growing ER tubule sliding along a preexisting ER tubule (arrows point to the tips of a ER tubule). Scale bar: 1 µm. (G) ER tubule extension events were captured during live-cell imaging of cells expressing YFP-α-tubulin and mCh-ARL6IP1 at the times indicated. Arrows depict the movement of ARL6IP1-postive ER tubules along microtubules. Scale bar: 2 µm. (H) Frequency of ARL6IP1-positive ER tubules that grow along microtubules marked by YFP-α-Tubulin (data from 14 cells). (I) Live-cell images of a COS-7 cell expressing YFP-α-Tubulin and mCh-ARL6IP1 upon 5 µM nocodazole treatment. Note the depolymerization of microtubules, the collapse of tubular ER network and the accumulation of bright foci containing mCh-ARL6IP1. Scale bar: 2 µm. Images are representative of three independent experiments. (J) Time-lapse images of cells expressing mCh-ARL6IP1 and GFP-CLIP170, a microtubule plus end-tracking protein. Arrowheads point to the tip of an elongating ARL6IP1-positive ER tubule. Note this tip lacks CLIP170 fluorescence. Scale bar: 2 µm. (K) Frequency of ARL6IP1-positive ER tubules tips adjacent (<1 µm) to CLIP170 puncta (data from 10 cells).

Figure 5.

Increased abundance of ER sheets upon loss of INPP5K or ARL6IP1. (A) Western blots of WT HeLa cells transfected with the indicated siRNAs showing depletion of INPP5K or ARL6IP1 proteins. (B) Top row: Representative confocal images of HeLa cells transfected with the indicated siRNAs and expressing the ER marker EGFP-Sec61β. High magnification of the regions enclosed by dashed boxes in the top row are shown in the bottom row. Scale bars: 5 µm in the top row, 2 µm in the bottom row. Solid red lines denote the edge of the cell. Note the predominant presence of ER tubules in the control cell, but the predominance of ER sheets throughout the cytoplasm of INPP5K or ARL6IP1 siRNA-treated cells. (C) Occupancy of the ER (both tubules and sheets) per unit area of each cell analyzed based on the analysis described in B for cells treated with control or INPP5K siRNA and expressing the indicated EGFP-INPP5K constructs. Automatic thresholding was applied (see Materials and methods for details) to select the total area occupied by the ER network (delineated by solid lines). The occupancy of the ER network per unit cell area serves as a proxy for the abundance of peripheral ER sheets and was calculated by dividing the pixel area occupied by the ER by that of the total cell area. Pooled data from three independent experiments are presented as scattered dots (solid black bars indicating the mean ± SD) and analyzed by two-tailed t-tests. n = 28 cells for control or INPP5K siRNA, 25 cells for INPP5K siRNA + EGFP-INPP5K^WT, 25 cells for INPP5K + EGFP-INPP5K^D192A, 21 cells for INPP5K + EGFP-INPP5K^D361G. ****, p < 0.0001; n.s., p > 0.05 (INPP5K siRNA + EGFP-INPP5K^D192A vs. INPP5K siRNA, p = 0.0541; INPP5K siRNA + EGFP-INPP5K^D361G vs. INPP5K siRNA, p = 0.057).

Figure 6.

cil-1 functions cell autonomously in PVD to regulate ER morphology. (A) Schematic drawing of PVD dendrites and ER pattern. Note that the ER tubules invade some, but not all, dendritic branches in the PVD. PVD dendrites are divided into three different segments: posterior, proximal, and distal dendrites. Based on the ER branch morphology, ER tubules are divided into three types (I, L, and T) to indicate whether the ER tubules invade just the secondary or tertiary branches. (B) Representative confocal images of transgenic animals expressing a PVD neuron marker ser-2Prom3::mCherry (PVD::mCherry) and an ER resident protein ser-2Prom3::GFP::SP12 (PVD::GFP::SP12) simultaneously. Note that the ER tubules in the branches are reduced in the cil-1 mutant. This defect is restored by reexpresssion of CIL-1 in PVD neurons (PVD::CIL-1). Arrowhead represents an ER branch. Scale bar: 10 µm. (C) Quantification of PVD proximal region ER branches. *, p < 0.05; **, p < 0.01; n.s., not significant by one-way ANOVA. Error bars indicate SEM. n = 40 for each genotype. (D) ER morphology in PVD cell body revealed by SIM imaging. The images are single confocal slices crossing the nuclear envelope at approximately its equatorial region. A roughly concentric circle (red) of 4.6 µm in diameter was drawn around the nucleus. The number of crossings between the red circle and ER tubules was counted. Scale bar: 5 µm. (E) Quantification of the number of crossings between the red circle and ER profiles in cells of the various genotypes. The number of crossings reflects the complexity of the ER. ***, p < 0.001; n.s., not significant by one-way ANOVA. Error bars indicate SEM. n > 48 for each genotype.

Figure 7.

The SKICH domain is essential for CIL-1 ER localization. (A) Representative confocal images of a transgenic worm expressing mCherry::SP12 and CIL-1::GFP driven by the hypodermal cell specific promoter. Scale bar: 10 µm. The pattern of CIL-1::GFP fluorescence is very similar to the pattern of the ER marker mCherry::SP12. Insets showing magnified view of the boxed region (scale bar: 2 µm). (B) Representative confocal images of transgenic worms expressing full-length CIL-1::GFP or a truncated from of CIL-1 lacking the SKICH domain, CIL-1(deSKICH)::GFP. While WT CIL-1 has the reticular distribution in cells expected for an ER protein, the truncated protein has a diffuse cytosolic localization. Scale bar: 10 µm.