A distinct pool of phosphatidylinositol 4,5-bisphosphate in caveolae revealed by a nanoscale labeling technique

Abstract

Multiple functionally independent pools of phosphatidylinositol 4,5-bisphosphate [PI(4,5)P(2)] have been postulated to occur in the cell membrane, but the existing techniques lack sufficient resolution to unequivocally confirm their presence. To analyze the distribution of PI(4,5)P(2) at the nanoscale, we developed an electron microscopic technique that probes the freeze-fractured membrane preparation by the pleckstrin homology domain of phospholipase C-delta1. This method does not require chemical fixation or expression of artificial probes, it is applicable to any cell in vivo and in vitro, and it can define the PI(4,5)P(2) distribution quantitatively. By using this method, we found that PI(4,5)P(2) is highly concentrated at the rim of caveolae both in cultured fibroblasts and mouse smooth muscle cells in vivo. PI(4,5)P(2) was also enriched in the coated pit, but only a low level of clustering was observed in the flat undifferentiated membrane. When cells were treated with angiotensin II, the PI(4,5)P(2) level in the undifferentiated membrane decreased to 37.9% within 10 sec and then returned to the initial level. Notably, the PI(4,5)P(2) level in caveolae showed a slower but more drastic change and decreased to 20.6% at 40 sec, whereas the PI(4,5)P(2) level in the coated pit was relatively constant and decreased only to 70.2% at 10 sec. These results show the presence of distinct PI(4,5)P(2) pools in the cell membrane and suggest a unique role for caveolae in phosphoinositide signaling.

Fig. 1.

Labeling of the liposome. (A) Outline of the method. Cells were rapidly frozen, freeze-fractured, and evaporated with carbon (C) and platinum/carbon (Pt/C) in vacuum. The replica of the split membrane was digested with SDS to remove noncast molecules and labeled by GST-PH. Both the cytoplasmic and exoplasmic halves of the membrane were examined. (B) Labeling of small unilamellar liposome replicas. Freeze-fracture replicas of liposomes containing 95 mol % of phosphatidylcholine (PC) and 5 mol % of phosphatidylinositol or a phosphoinositide were labeled. Only liposomes containing PI(4,5)P₂ were labeled intensely by GST-PH. A PH mutant, GST-PH(K30N, K32N), which does not bind PI(4,5)P₂, showed little labeling in the PI(4,5)P₂-containing liposome. (C) Quantification of the GST-PH labeling in the liposomes. The number of gold particles per 1 μm² of the liposome surface is shown (blue). The labeling on the convex (red) and concave (yellow) surfaces showed equivalent results.

Fig. 2.

Labeling in the flat undifferentiated area of the cell membrane. (A) Distribution of PI(4,5)P₂ in the P face of the flat undifferentiated cell membrane of the human fibroblast in culture. GST-PH was used at a concentration of 100 ng/mL (Left) or 30 ng/mL (Right). The mean L(r) − r curve compiled from 30 randomly chosen areas showed that the labeling by 100 ng/mL GST-PH was randomly distributed, whereas that by 30 ng/mL GST-PH showed weak clustering. (B) Effects of inositol trisphosphates on the PI(4,5)P₂ labeling. GST-PH (100 ng/mL) was preincubated with either 1 mM Ins(1,4,5)P₃ or 1 mM Ins(1,3,4)P₃ before applying to replicas. Ins(1,4,5)P₃ abolished the labeling completely, whereas Ins(1,3,4)P₃ did not. (C) Labeling by GST-PH (K30N, K32N) used at 100 ng/mL. Virtually no labeling was observed. (D) The clustering that was detected by 30 ng/mL GST-PH decreased when cells were treated with 5 mM methyl-β-cyclodextrin (MβCD) for 60 min to extract free cholesterol or with 1 μM latrunculin A (Lat A) for 10 min to depolymerize actin. After these treatments, the normalized nearest neighbor distance became closer to one, and the average labeling density increased (*, P < 0.01; **, P < 0.001).

Fig. 3.

Labeling in caveolae and coated pits. GST-PH was applied at a concentration of 30 ng/mL. (A) Intense labeling of PI(4,5)P₂ at the caveolar rim. The large (10-nm) and small (5-nm; arrows) gold particles label PI(4,5)P₂ and caveolin-1, respectively. Deep caveolae were fractured at the neck portion and appeared electron-lucent in the center because Pt/C evaporation did not reach the deep portion. In contrast, shallow caveolae were observed all along the contour. To measure the labeling density in relation to the distance from the caveolar center, 30 caveolae were randomly chosen from 3 independent experiments. The caveolar labeling was most intense at 30–50 nm from the center, and the labeling density was much higher than that in the flat undifferentiated membrane area (green bar). (B) Dense caveolar labeling of PI(4,5)P₂ in the mouse smooth muscle (vas deferens) in vivo. The caveolae also showed intense PI(4,5)P₂ labeling at the rim. The labeling density in the flat undifferentiated membrane is shown as a reference (green bar). (C) Labeling of PI(4,5)P₂ in the coated pit that was observed as a smooth indentation of 150–200 nm in diameter (arrowheads). The size, morphology, and lack of caveolin-1 labeling suggested that they are clathrin-coated pits. The labeling density of PI(4,5)P₂ was the highest at the rim, or 70–100 nm from the center. The labeling density in the flat undifferentiated membrane is shown as a reference (green bar). For this quantification, protein A conjugated to 5-nm colloidal gold was used so that the labeling density in the flat undifferentiated membrane (green bar) is different from A.

Fig. 4.

Labeling after Ang II stimulation. GST-PH was applied at a concentration of 100 ng/mL. (A) Representative micrographs of the PI(4,5)P₂ labeling before and 10, 40, and 130 sec after the treatment with 1 μM Ang II. (B) The time course of the labeling intensity change. The relative labeling intensity was shown by taking the labeling density of the control sample as the standard (the average ± standard deviation; **, P < 0.001). Thirty areas of the undifferentiated membrane and 30 caveolae were randomly chosen for each time point. For caveolae, the labeling at 30–50 nm from the center was measured. (C) The labeling intensity in relation to the distance from the caveolar center. The labeling density in the undifferentiated membrane of the untreated cell was taken as the standard. Thirty caveolae were randomly chosen for each time point. The labeling at 30–50 nm from the caveolar center did not change significantly at 5 sec and 10 sec after the Ang II stimulation. At 40 sec, the caveolar labeling decreased significantly, whereas the labeling in the flat membrane started to recover and was denser than that of the caveolar rim. The caveolar labeling returned to the control level at 130 sec.

Fig. 5.

Labeling after ionomycin treatment. GST-PH was applied at a concentration of 30 ng/mL. (A) Representative micrographs of the PI(4,5)P₂ labeling in the flat undifferentiated membrane before and 1 and 5 min after the treatment with 10 μM ionomycin. (B) Representative micrographs of the PI(4,5)P₂ labeling in caveolae before and 1 min after the ionomycin treatment. PI(4,5)P₂ and caveolin-1 are marked by large (10-nm) and small (5-nm) gold particles (arrows), respectively. (C) The change of the labeling intensity after the ionomycin treatment. The labeling density in the flat undifferentiated membrane of the untreated cell was taken as the standard. Twenty undifferentiated areas and 75 caveolae were randomly chosen for each time point. The labeling in caveolae, or in the zone 30–50 nm from the caveolar center, decreased significantly 1 min after the treatment, whereas that in the undifferentiated membrane showed a large decrease only at 5 min (*, P < 0.01; **, P < 0.001). Caveolae were scarcely found in the 5-min sample, probably because of the structural change induced by the high [Ca²⁺]_i.