Probing the subcellular distribution of phosphatidylinositol reveals a surprising lack at the plasma membrane

Abstract

The polyphosphoinositides (PPIn) are central regulatory lipids that direct membrane function in eukaryotic cells. Understanding how their synthesis is regulated is crucial to revealing these lipids' role in health and disease. PPIn are derived from the major structural lipid, phosphatidylinositol (PI). However, although the distribution of most PPIn has been characterized, the subcellular localization of PI available for PPIn synthesis is not known. Here, we used several orthogonal approaches to map the subcellular distribution of PI, including localizing exogenous fluorescent PI, as well as detecting lipid conversion products of endogenous PI after acute chemogenetic activation of PI-specific phospholipase and 4-kinase. We report that PI is broadly distributed throughout intracellular membrane compartments. However, there is a surprising lack of PI in the plasma membrane compared with the PPIn. These experiments implicate regulation of PI supply to the plasma membrane, as opposed to regulation of PPIn-kinases, as crucial to the control of PPIn synthesis and function at the PM.

Figure 1.

Fluorescent PI is not enriched at the PM. (A) Internalization of fluorescent lipids. COS-7 cells were loaded at 37°C with the indicated acyl-conjugated TopFluor lipids complexed with BSA. After 15 min, a 20-fold excess of un-complexed BSA was used to back-extract lipid remaining in the outer PM leaflet. (B) Loaded TopFluor-PI is not metabolized. COS-7 cells were loaded or loaded and back-extracted as in A. Lipids were then extracted and resolved by TLC. (C–E) Loaded TopFluor-PI labels the Golgi (C), ER (D), and mitochondria (E). COS-7 cells that had been transfected with mCherry-β4-GalT^N82 (C), iRFP-Sec61β (D), or COX8A^N29x2-mCherry (E) were loaded and back-extracted as in A. Scale bar = 20 µm in all panels; insets are 10.9 µm in C or 7.3 µm in D and E. Data are representative of three or more experiments.

Figure S1.

TopFluor-PI fluorescence distribution is not contaminated by bleed-through from transfected organelle markers. Images show a nontransfected cell imaged under identical conditions to those presented in Fig. 1. Clear mitochondrial (mito; i), Golgi (ii), and ER (III) morphology of the green fluorescence is seen even with no expression of markers for these compartments, demonstrating that they are not due to fluorescence bleed-through. Note, mitochondrial morphology is evident from the differential interference contrast image (gray). Insets are 7.3 µm and serve as scale bar.

Figure 2.

Chemically induced dimerization of a split PI-PLC induces intracellular accumulation of DAG. (A) Experimental setup: rapamycin-induced dimerization of FKBP-fused amino-terminal 187 residues of PI-PLC (cyan) with the FRB-fused carboxy-terminal 100 residues (orange) reconstitutes the active enzyme, though no visible change in the cytosolic localization of the TagBFP2/iRFP-fused enzyme fragments is observed (confocal images at right; scale bar = 20 μm). (B) DAG accumulation on cytosolic leaflets of intracellular organelles, with the greatest increases in the Golgi and endosomal/lysosomal compartments. Zero or comparatively minor changes were observed in mitochondria, peroxisomes, or the ER. Cells were expressing GFP-PKD1-C1ab to detect DAG, the indicated organelle markers fused to mCherry or mKO (magenta), FKBP-PI-PLC^N187, and FRB-PI-PLC^C100 (not shown); they were treated with 1 µM rapamycin at time 0. Inset regions are 15 µm and serve as scale bars. (C) Anchoring PI-PLC^N187 to the mitochondrial outer membrane shows transient accumulation of DAG after recruitment of PI-PLC^C100. Cells were transfected as in B, but with AKAP^N31—fused PI-PLC^N187 replacing the unanchored version in B. Inset = 30 µm and serves as scale bar. In both B and C, the curves at right show the mean change in C1ab reporter intensity at each compartment, with SEM shaded. The violin plots show AUC, with the number of cells (pooled across three to six independent experiments), the sum of signed ranks (W), and P value from a two-tailed Wilcoxon signed rank test compared to a null hypothesis AUC value of 0. mito, mitochondria.

Figure S2.

Kinetics of PI-PLC^C100-FRB-iRFP recruitment to organelle-targeted BFP-FKBP-PI-PLC^N187. (A) Schematic. (B) Summary data. Mean time constant ± 95% confidence interval is shown for each organelle-targeted construct. (C and D) Data for C1ab recruitment as shown in C (Fig. 3 A) and D (Fig. 2 C) is shown alongside that for PI-PLC^C100-FRB-iRFP from the same cells. Data are means with SEM shaded; black fits represent the mean fit for all cells to the single-phase exponential ΔIntensity = Plateau × e^−(time/τ). mito, mitochondria.

Figure 3.

Very little PI can be converted to DAG at the PM. (A) PM-specific dimerization of split PI-PLC induces very little DAG compared with PI(4,5)P₂-specific PLC. Cells were transfected with PI-PLC^C100-FRB, PM-targeted Lyn^N11-FKBP-PI-PLC^N100, and GFP-PKD1-C1ab to detect DAG. Rapamycin was used to induce PI-PLC reconstitution at the PM, or else endogenous PLCβ was activated by G_q-coupled agonist ATP (to activate endogenous P2Y receptors) or carbachol (CCh; to activate overexpressed muscarinic M3 receptors). TIRFM images are color-coded to represent fluorescence intensity relative to prestimulus levels (F_t/F_pre) as indicated. Inset region is 5 µm and serves as scale bar for images in A. The line graphs show mean F_t/F_pre with SEM shaded; P values are derived from Dunn’s multiple comparison test compared to time 0, following Friedman’s test (Friedman statistic = 49.16, PI-PLC, P = 0.045; 497.0, ATP, P < 10⁻⁴; 282.3, CCh; P < 10⁻⁴). Where not indicated, the P value from Dunn’s test is >0.05. The violin plots show AUC analysis of the line graphs with the number of cells from three independent experiments, with results of a Kruskal-Wallis test and P values from a post hoc Dunn’s multiple comparison test indicated. (B) Split PI-PLC induces translocation of Nir2 to ER-PM contact sites. Cells were transfected with PM-targeted split PI-PLC and stimulated with rapamycin or ATP as in A. Translocation of GFP-Nir2 was recorded; images show representative TIRFM images with fluorescence normalized to prestimulus levels (F_t/F_pre). Scale bar = 10 µm. The line graphs show mean F_t/F_pre with SEM shaded; the violin plots show AUC analysis of the line graphs with the number of cells from three independent experiments indicated, along with results of Wilcoxon signed rank test comparing each population to a hypothesized AUC of 0, as well as a Mann-Whitney U test comparing the differences between AUC after ATP or rapamycin stimulation.

Figure 4.

PI-PLC leads to little or no depletion of PPIn on Golgi and endosomes or PM. (A and B) Cells were transfected with PI-PLC^C100-FRB, FKBP-PI-PLC^N100 (or PM-targeted Lyn^N11-FKBP-PI-PLC^N100), and GFP-PKD1-C1ab to detect DAG, along with the indicated PPIn biosensor. Dimerization and activation of PI-PLC was induced with 1 µM rapamycin as indicated. Images are confocal sections (A) or TIRFM (B). Scale bars = 20 µm. Inset = 10 µm for P4M and 15 µm for FYVE. Line graphs show the change in compartment-specific fluorescence of C1ab (green) or the PPIn biosensor (magenta) normalized to prerapamycin levels (F_pre). The violin plots show AUC, with the number of cells (pooled across three independent experiments), the sum of signed ranks (W), and P value from a two-tailed Wilcoxon signed rank test compared to a null hypothesis AUC value of 0 (with a baseline of 1).

Figure 5.

Compartment-specific recruitment of PI4KA reveals intracellular PI pools. (A) Experimental setup: rapamycin-induced recruitment of FKBP-PI4KA^N1001 by dimerization with compartment-specific FRB to induce PI4P synthesis from endogenous PI, revealed with GFP-P4M or -P4C. (B) PI4P accumulation on cytosolic leaflets of intracellular organelles, with the greatest increases in the Golgi and mitochondria, with barely detectable increases in endosomes and ER. Peroxisomes show a large increase, but PI4P intensity also occurs outside of PMP-marked compartments. Cells expressing GFP-P4M or -P4C (green), compartment-specific FRB (magenta), and FKBP-PI4KA^N1001 (cyan) as indicated were treated with 1 µM rapamycin at time 0. Inset regions are 15 µm and serve as scale bars. The line graphs at right show the mean change in PI4P reporter intensity at each compartment, with SEM shaded. The violin plots show AUC analysis of the curves, with the number of cells (pooled across two to four independent experiments), Mann-Whitney U statistic, and P value from a two-tailed test. (C) Peroxide-mediated inhibition of ER-associated SAC1 PI4P phosphatase reveals an endogenous pool of PI that can be converted to PI4P. Experiment and data are identical to B, except cells were treated with 500 µM hydrogen peroxide in addition to rapamycin. mito, mitochondria.

Figure S3.

Kinetics of mCherry-FKBP-PI4KA^C1001 recruitment to organelle-targeted FRB. (A) Schematic. (B) Summary data. Mean time constant ± 95% confidence interval is shown for each organelle-targeted construct. (C–I) Data for PI4P biosensor (green) recruitment as shown in C (Fig. 6 A) and D–I (Fig. 5 B) is shown alongside that for mCherry-FKBP-PI4KA^C1001 (orange) from the same cells. Data are means with SEM shaded; black fits represent the mean fit for all cells to the single-phase exponential ΔIntensity = Plateau × e^−(time/τ).

Figure 6.

Recruitment of PI4K to the PM reveals a scarcity of PI. (A) Recruitment of PI4KA^N1001 shows no increase in PI4P or PI(4,5)P₂. COS-7 cells imaged by TIRFM expressing GFP-P4Mx1 or Tubby_c^R332H-mCherry together with FKBP-PI4K^N1001 and Lyn^N11-FRB were treated with 1 µM rapamycin at time 0 to induce dimerization. Scale bar = 20 µm. The curves at right show the mean change in reporter intensity, with SEM shaded. The violin plots show AUC analysis of the curves, with the number of cells (pooled across three independent experiments) and the Mann-Whitney U statistic and P value from a two-tailed test. (B) PI4KB induces PI4P increases at the mitochondria but not the PM. COS-7 cells were transfected with P4Mx1 to detect PI4P increases by confocal microscopy in conjunction with the indicated compartment-specific FRB and FKBP-PI4KB. Insets are 15 µm and serve as scale bars. Line graphs are means with SEM shaded; inset violin plot shows AUC analysis, with number of cells (pooled across three independent experiments) and the Mann-Whitney U statistic and P value from a two-tailed test. (C) PI4KB is active in the PM. COS-7 cells were cotransfected with PI(4,5)P₂ biosensor Tubby_c-GFP (green) and muscarinic M3 receptors to stimulate PLCβ-induced PI(4,5)P₂ and PI4P depletion from the PM in response to carbachol treatment. Subsequent treatment with the muscarinic antagonist atropine induces resynthesis of PI4P and PI(4,5)P₂ via endogenous PI4KA. Where indicated, the PI4KA inhibitor A1 was added at a 30-nM concentration. Images show confocal sections (scale bar = 20 µm) before stimulation, after addition of carbachol, and after atropine addition as indicated. Curves are means with SEM shaded. The violin plot shows AUC analysis for the post-atropine addition, with number of cells (three independent experiments), Kruskal-Wallis statistic, and P value from a two-tailed test. P values between individual groups shown are derived from a post hoc Dunn’s multiple comparison test. CCh, carbachol; mito, mitochondria.

Figure S4.

Kinetics of mCherry-FKBP-PI4KB recruitment to organelle-targeted FRB. (A) Schematic. (B) Summary data. Mean time constant ± 95% confidence interval is shown for PM (C) and mitochondria (D). Data for GFP-P4Mx1 PI4P biosensor (green) recruitment as shown in Fig. 6 B is shown alongside that for mCherry-FKBP-PI4KB (maroon) from the same cells. Data are means with SEM shaded; black fits represent the mean fit for all cells to the single-phase exponential ΔIntensity = Plateau × e^−(time/τ). mito, mitochondria.