The plasma membrane of focal adhesions has a high content of cholesterol and phosphatidylcholine with saturated acyl chains Free

Cell–extracellular matrix (ECM) adhesion complexes, called focal adhesions (FAs), play a pivotal role in sensing surrounding microenvironments, such as the type or rigidity of ECM, and regulating cell behaviors including cell survival, migration and differentiation (Engler et al., 2006; Frisch and Francis, 1994; Kuroda et al., 2017). Proteomic studies have revealed that FAs comprise hundreds of proteins that form a giant ‘adhesome’ (Horton et al., 2015; Kuo et al., 2011). Protein components in FA are regulated by various conditions, including FA maturation or ECM stiffness, to regulate cellular function (Ichikawa et al., 2017; Kuo et al., 2011; Laukaitis et al., 2001; Yamashita et al., 2014; Zaidel-Bar et al., 2003). In addition, recent studies have discovered that not only proteins but also lipids play an important role in cell–ECM adhesion. For example, phosphatidylinositol 4,5-bisphosphate (PIP2) produced by phosphatidylinositol phosphate kinase type 1γ at FAs binds and activates various FA proteins, including vinculin and talin proteins, to promote FA formation (Di Paolo et al., 2002; Gilmore and Burridge, 1996; Ling et al., 2002; Martel et al., 2001). Expression of phosphatidylcholine (PC) flippases increases PC flow from the exoplasmic to the cytoplasmic leaflet and suppresses FA formation and cell adhesion (Miyano et al., 2016).

A number of reports have suggested the importance of lipid rafts in FA function. Lipid rafts are highly ordered plasma membrane domains that have low fluidity and are rich in cholesterol and sphingolipids, such as sphingomyelin (SM) (Sezgin et al., 2017; Simons and Ikonen, 1997). Lipid rafts concentrate certain proteins and play an important role as a platform for intracellular signaling (Lingwood and Simons, 2010; Sezgin et al., 2017; Simons and Toomre, 2000). Some FA proteins have been shown to localize in lipid rafts that have been biochemically isolated as detergent-resistant membranes (DRMs), and disruption of lipid rafts deteriorates FA formation and cell migration (Baillat et al., 2008; Nagasato et al., 2017; Su et al., 2013; Wang et al., 2013; Yeh et al., 2017). Interestingly, lipids in the FA membrane have been reported to be less fluid than those in other membranes (Geiger et al., 1982). Furthermore, experiments using Laurdan, a reagent that varies its peak emission based on the fluidity or order of the membrane, have shown that FA membranes have low fluidity and are highly ordered (Gaus et al., 2006), which are characteristics of lipid rafts. However, not only lipids but proteins could affect the peak shift of Laurdan (Gaus et al., 2006), which makes it difficult to distinguish whether the highly ordered state of the FA membrane is caused by lipid composition or the accumulation of hydrophobic transmembrane proteins. Altogether, these reports suggest that the FA membrane has a lipid raft-like state, but the detailed lipid composition in the FA membrane has not been elucidated.

In this study, we show an enrichment of cholesterol in the FA membrane by staining intact cells with purified enhanced green fluorescent protein (EGFP)-tagged D4, a toxin-derived protein probe that specifically binds to a cholesterol-rich membrane (Maekawa, 2017; Shimada et al., 2002). Furthermore, the FA-rich fraction is isolated to extract the lipid and examine the lipid composition of the FA membrane. Biochemical quantification reveals a higher content of cholesterol and PC with saturated fatty acid chains in the lipids extracted from the FA-rich fraction than in plasma membrane fraction or the whole-cell membrane. To our knowledge, this is the first report of a biochemical analysis of FA membrane lipids. Together, our data demonstrate that lipid composition of the FA membrane is distinct from the bulk plasma membrane.

Although there are a number of reports suggesting that the FA membrane is a raft-like domain, direct observation of lipids known to be enriched at the lipid raft has not been undertaken at FAs. Therefore, we aimed to investigate the cholesterol content at FAs by visualizing cholesterol using EGFP-conjugated D4 (EGFP–D4). D4 is the cholesterol-binding domain of Perfringolysin O, a bacteria-derived pore-forming toxin (Maekawa, 2017; Shimada et al., 2002). Because the D4 domain itself has no toxicity, EGFP–D4 can be used as a probe for a cholesterol-rich membrane (>30–40 mol%) (Kishimoto et al., 2016; Liu et al., 2017; Maekawa, 2017; Shimada et al., 2002). Mouse embryonic fibroblasts (MEFs) stably expressing mCherry–vinculin were incubated with purified EGFP–D4 and visualized by total internal reflection fluorescence (TIRF) microscopy to observe the ventral plasma membrane. After 30 min incubation, EGFP–D4 was clearly localized to the FA regions visualized by mCherry–vinculin in addition to the prominent accumulation at the filopodial protrusion, as reported previously (Kishimoto et al., 2020) (Fig. 1A). Pre-depletion of cholesterol from the plasma membrane using methyl-β-cyclodextrin (MβCD) treatment clearly impaired EGFP–D4 signals (Fig. 1A). Furthermore, re-loading of cholesterol to cholesterol-depleted cells using the MβCD-cholesterol treatment led to a recovery of EGFP–D4 signals (Fig. S1A), which indicated the selectivity of EGFP–D4 for cholesterol. Interestingly, peripheral FAs showed more EGFP–D4 signals than central FAs, but after cholesterol re-loading, the difference of EGFP–D4 intensity between peripheral and central FAs became small, suggesting a difference in lipid composition between plasma membranes of peripheral and central FAs. EGFP–D4 also colocalized with other major FA proteins (e.g. talin, paxillin and α5 integrin; Fig. 1B; Fig. S1B). Membrane staining using CellMask reagents did not show the colocalization to FA using TIRF microscopy (Fig. S1A,B), indicating that the EGFP–D4 signal is specific for cholesterol, but not derived from membrane curvature. Next, we examined whether EGFP–D4 localizes with endocytosis marker proteins because caveolin-1, which localizes to cholesterol-rich membrane, is involved in endocytosis of integrin (del Pozo et al., 2005; Shi and Sottile, 2008). However, caveolin-1 as well as the clathrin light chain (CLTA) did not colocalize with EGFP–D4 or paxillin in our conditions (Fig. S1F). Because EGFP–D4 is not a membrane-permeable probe, cholesterol-rich domains observed using purified EGFP–D4 are considered to exist in the outer leaflet of the plasma membrane. To examine the distribution of cholesterol in the inner leaflet, we transfected the plasmid encoding EGFP–D4 or EGFP–D4H (D4 D434S, a mutant with higher sensitivity for cholesterol) into the cells. Interestingly, neither EGFP–D4 nor EGFP–D4H expressed in the cells showed localization to the FA regions, but were merely diffusively expressed with a similar pattern to plasma membrane (Fig. S1C,D). This could be a result of the difference in cholesterol abundance between the inner and outer leaflets of the plasma membrane (Liu et al., 2017) or due to the differences in EGFP–D4 or EGFP–D4H concentration.

Next, we performed TIRF microscopy time-lapse imaging of EGFP–D4-stained cells. In the time-lapse images of the EGFP–D4-stained cells, we focused on the FA assembly and disassembly processes. We found that mCherry–vinculin and EGFP–D4 signals increased similarly during FA assembly (Fig. 1C; Movie 3). Quantified data and kymograph analysis further support this observation (Fig. 1C,E). These results demonstrate that the cholesterol distribution in the plasma membrane is dynamically changing during FA formation. Because EGFP–D4 in the medium was washed away before observation, the assembly of EGFP–D4 observed here most likely reflects the recruitment of EGFP–D4 that had already been bound to cholesterol. To exclude the possibility of cholesterol-independent diffusion of EGFP–D4 after binding to the membrane, cells stained with EGFP–D4 were treated with MβCD. MβCD treatment resulted in the rapid loss of the EGFP–D4 signal, confirming the continuous binding of EGFP–D4 to cholesterol in live cells (Fig. S1E and Movies 1 and 2). In contrast to the assembly process, there was a significant delay in the EGFP–D4 signal decrease during the disassembly process (Fig. 1D; Movie 4). This might be because components other than vinculin remain and retain cholesterol, even after vinculin leaves. Another possibility is that mCherry–vinculin quickly escapes the TIRF plane by diffusion into the cytosol, whereas EGFP–D4 remains in the TIRF plane during diffusion in the membrane. Collectively, these results show the close relationship between FA formation and the lipid composition of FA membranes – cholesterol-rich membranes accumulate simultaneously with FA assembly.

To investigate the precise distribution of cholesterol-rich domains in FA regions, we performed super-resolution imaging of the EGFP–D4-stained cells using structured illumination microscopy (SIM) (Fig. 2). Line profiling of the SIM image revealed not only colocalization of EGFP–D4 and paxillin-positive FAs but also EGFP–D4 localization within the vicinity (∼200 nm) of FAs (Fig. 2A,B). Although the KANK-positive region, namely, the ‘FA belt’ region is near FAs (Sun et al., 2016), we found that EGFP–D4 was not colocalized with KANK2 (Fig. 2C,D). Taken together, cholesterol-rich regions do not correspond to KANK2-rich FA belts.

There are few studies on the lipid composition of FA membranes because of the lack of methods to isolate them. To biochemically analyze the lipid composition of the FA membrane, we established a method to isolate lipids from the FA-rich fraction by modifying the FA protein isolation method reported by Kuo et al. (2011). Using this method, MEFs were hypotonically shocked, followed by the removal of the dorsal plasma membrane, organelles, cytosol, and most ventral membrane using a water jet treatment (Fig. 3A), and the remaining fraction containing FAs was isolated as the ‘FA-rich fraction’. Organelle and cytosol removal after the treatment was examined by western blotting. As shown in Fig. 3B and Fig. S2A, ER (GRP94, also known as HSP90B1, the FA-rich fraction/intact cell ratio was 0.09), mitochondria (MTCO1, 0.35) and cytosol (GAPDH, 0.11) proteins were almost undetectable in the FA-rich fraction. In contrast, FA marker proteins [vinculin, 3.1 and phosphorylated paxillin (p-paxillin), 4.5] were clearly concentrated in the FA-rich fractions, which indicates the resistance of FAs to water jet treatment. Because vinculin also localizes in the cytosol, p-paxillin is more extensively concentrated in the FA-rich fraction compared to vinculin. Immunostaining analysis also showed the lack of organelle and cytosol marker proteins and the existence of FA proteins in the FA-rich fraction isolated via the hypotonic shock and water jet procedure (Fig. 3C,D; Fig. S2B,C). Although plasma membrane markers, such as Na/K-ATPase and transferrin receptor (TfR), were detected in the FA-rich fractions by western blotting (Fig. 3B), the immunostaining of the FA-rich fraction showed that TfR remained where vinculin localized (Fig. S2C,D). Given that plasma membrane proteins, including TfR, can freely move into the FA region (Shibata et al., 2012), these observations suggest that the plasma membrane in the vinculin-positive region remained but other basal plasma membrane was almost removed after FA isolation treatment. Furthermore, staining with CellMask Orange, a lipid bilayer staining reagent, revealed the existence of the plasma membrane in and around vinculin-positive region in the FA-rich fraction (Fig. 3E). EGFP–D4 signal was also detected after FA isolation (Fig. S2E). Together, these observations indicate that this method is useful for isolation of plasma membrane in the FA region.

To analyze nanogram-scale amount of lipid extracted from the fraction, we used a highly sensitive fluorometric method of biochemical quantification (Amundson and Zhou, 1999; Morita et al., 2010). To ensure that the lipids extracted from the FA-rich fractions are derived from the cell–ECM adhesion structure, we compared the amount of choline lipid (phosphatidylcholine and sphingomyelin), the most abundant lipid in a cell, extracted from the FA-rich fractions of MEFs seeded on fibronectin (FN)-coated and BSA-treated (prevention of non-specific attachment) dishes with that from cells on non-coated or only FN-coated dishes. The results of lipid quantification showed that the amount of choline lipid extracted from the FA-rich fractions of MEFs on FN-coated dishes with and without BSA blocking was 4-fold and 2.8-fold higher, respectively, than that on non-coated plastic dishes (Fig. S3A). Importantly, the amount of choline lipid correlated with the amount of p-paxillin (Fig. S3A,B), indicating that most lipids extracted from the FA-rich fractions depend on cell–ECM adhesion.

Next, we decided to compare the FA membrane with bulk plasma membranes. Because a density-gradient centrifugation method (Suski et al., 2014) did not work for the isolation of plasma membrane from MEFs, we isolated giant plasma membrane vesicles (GPMVs) using N-ethylmaleimide (NEM) and used them as a source of plasma membrane. GPMVs have been used for lipidomics analysis of plasma membrane previously (Baumgart et al., 2007; McGraw et al., 2019; Sezgin et al., 2012). We found that GPMVs were stained by membrane-staining reagent (CellMask Orange and FM 1-43), but not by ER tracker or Mito-Tracker (Fig. S3C). Thus, we decided to use GPMVs as a plasma membrane fraction. We then quantified the amount of cholesterol, the major component of the lipid rafts (Simons and Ikonen, 1997; van Meer et al., 2008), as well as choline lipids and compared the cholesterol content in each fraction by calculating the ratio of cholesterol to choline lipids (Fig. 4). The FA-rich fractions showed a higher cholesterol/choline lipid ratio compared with the GPMVs or whole-cell membrane. These results suggest that the membrane in FA region is enriched in cholesterol compared with bulk plasma membrane or whole-cell membrane.

Next, we examined whether the PC, SM or phosphatidylethanolamine (PE) in FAs has a specific fatty acid composition. Lipids extracted from the whole cells, GPMVs and FA-rich fraction were subjected to liquid chromatography (LC)–tandem mass spectrometry (MS/MS) analysis (MRM mode), after which the ratio of PC, SM and PE species was determined. The ratio of PC species that contain saturated fatty acids including PC (30:0), PC (32:0) and PC (34:0) was two times higher in the FA-rich fraction than in the GPMVs or whole-cell membrane (Fig. 5A). Accordingly, the ratio of PC species that contain unsaturated and long fatty acids was low in the FA-rich fraction and high in GPMVs. A similar trend was observed in the experiment in which weak water jet treatment, instead of GPMVs, was used to isolate the plasma membrane fraction, in which a portion of the ventral plasma membrane as well as the membrane of the FA region was isolated as plasma membrane fraction (Fig. S4). Furthermore, we found that the content of SM (34:1) species in the FA-rich fraction was high, but that of SM (42:2), which suppresses the liquid-ordered domain (Courtney et al., 2018; Vazquez et al., 2019), was low (Fig. 5B) compared to what was seen for GPMVs, although the contents of SM (34:1) and SM (42:2) in FA-rich fraction were similar to that in intact cells. This is consistent with both the above observation that the FA-rich fraction has membrane with a higher cholesterol content and the report that FA membrane has highly ordered membrane (Gaus et al., 2006). In contrast to the specific fatty acid chain composition of PC and SM in the FA-rich fraction, the ratio of PE species was only slightly changed among the fractions (Fig. 5C). Furthermore, we compared the SM content in each fraction by calculating the ratio of the total peak area of SM to that of PC. As shown in Fig. 5D, FA-rich fractions had higher SM/PC ratio compared with GPMVs or the whole-cell membrane. Taken together, these results suggest that the membrane in FA region has a higher content of cholesterol and PC with saturated fatty acid chains compared with bulk plasma membrane or whole-cell membrane.

In the present study, we hypothesized that plasma membranes in the FA region have a specific lipid composition, especially a lipid raft-like composition. To examine this hypothesis, we performed live-cell imaging using the cholesterol probe EGFP–D4 and found that cholesterol assembles with FAs. Furthermore, we established a method to isolate lipids from the FA-rich fraction. The biochemical analysis of the isolated lipids showed that the membranes in the FA-rich fraction have higher contents of cholesterol and SM as well as PC with saturated fatty acids compared to the bulk plasma membranes or total cell membranes. In contrast to previous studies that relied on rather indirect measurements, this work provides direct evidence that FA membranes have a specific lipid composition.

Plasma membrane at FA region has been reported to be less fluid than membrane in other regions (Geiger et al., 1982). Lipids with saturated acyl chain are well known to form less-fluid membrane (Ernst et al., 2016). Cholesterol has a higher affinity for lipids with saturated fatty acids than for lipid with unsaturated fatty acids and induces domain separation of the membrane containing saturated and unsaturated lipids, resulting in decreased fluidity of lipids with saturated fatty acids (Scherfeld et al., 2003). Therefore, our finding that FA membranes have high contents of cholesterol and PC with saturated fatty acids can explain less fluidity of the FA membrane.

We showed that membranes in FA-rich fraction have higher cholesterol contents than bulk plasma membranes or whole-cell membranes. Another adhesion structure, the tight junction (TJ), is also associated with lipid raft-like membrane enriched in cholesterol (Nusrat et al., 2000; Shigetomi et al., 2018). Our study further showed more PC with saturated fatty acids and cholesterol but less SM (42:2) content in the membrane of the FA-rich fraction than in bulk plasm membranes. SM (42:2) suppresses lipid-raft microdomain formation (Courtney et al., 2018; Deng et al., 2019), consistent with both our present observation that the FA-rich fraction has a membrane with a higher cholesterol content and the report that the FA membrane has a highly ordered membrane (Gaus et al., 2006). Intriguingly, the membrane fraction containing the TJ has more SM (42:2) than bulk plasma membranes (Shigetomi et al., 2018). Therefore, these observations suggest that membranes of both the FA-rich fraction and the TJ fraction have high cholesterol content, but lipid compositions, especially SM species, differ from each other.

The formation of lipid rafts or ordered membrane domains is mediated by not only lipid–lipid interactions but also lipid–protein interaction (Sezgin et al., 2017). Dimer formation of glycosylphosphatidylinositol-anchored proteins or the accumulation of the lipidated proteins can nucleate ordered membrane domains (Contreras et al., 2012; Schwarzer et al., 2014; Suzuki et al., 2012; Tulodziecka et al., 2016). Therefore, it is possible that clustering of lipid-binding or lipidated FA proteins, including vinculin and talin that associate with PIP2 (Case et al., 2015; Gilmore and Burridge, 1996; Kanchanawong et al., 2010; Kellie and Wigglesworth, 1987), contribute to the formation and maintenance of the specific lipid composition of the FA region. Indeed, vinculin, as well as actomyosin activity, both of which promote integrin-mediated adhesion, increase nanoclustering of glycosylphosphatidylinositol-anchored-protein (Kalappurakkal et al., 2019). Another candidate is a membrane protein integrin. Depletion of β1 integrin has been reported to result in decreased cholesterol and SM contents at the cellular level (Pankov et al., 2005). In addition, active and inactive β1 integrins segregate into distinct nanoclusters in focal adhesions, and ligand binding of integrin alters integrin preference from liquid-disordered to liquid-ordered membrane domains (Ge et al., 2018; Siegel et al., 2011; Spiess et al., 2018). It will be of interest to explore how the lipid composition of FA membranes is maintained.

Our experiment using purified EGFP–D4 showed the high content of cholesterol at least in the outer leaflet of the plasma membrane of FA. It is unclear whether the high content of cholesterol in the outer leaflet at FAs directly regulates the accumulation of lipids, such as PIP2, in the inner leaflet or the behavior of cytoplasmic protein. It has been reported that clustering of SM in the outer leaflet can induce the formation of lipid domains in the inner leaflet though transbilayer interaction (Raghupathy et al., 2015). Furthermore, cholesterol accumulated in the outer leaflet is required for the SM clustering in the outer leaflet at the cleavage furrow, leading to the accumulation of PIP2 in the inner leaflet (Abe et al., 2012). A very recent report of molecular dynamics simulation indicates that PIP2 in the inner leaflet can accumulate beneath a lipid raft consisting of SMs depending on the length of SM acyl chain (Li et al., 2023). Therefore, it is possible that the high content of cholesterol in the outer leaflet at FAs modulates the FA assembly through the lipid in the inner leaflet. It will be of interest to examine this possibility in the future.

The possible functions of a specific composition in the FA regions might be to regulate FA protein recruitment and turnover as well as to modulate FA-mediated signals. Recent studies have discovered that integrin activation by ligand binding triggers integrin sequestration into cholesterol-rich membrane regions (Ge et al., 2018; Wang et al., 2013). Cholesterol enhances the membrane binding of FAK (also known as PTK2) and stimulates integrin recycling (Takahashi et al., 2021). Furthermore, cholesterol promotes PIP2 clustering and visibility based on in vitro binding assay and molecular dynamics simulations, leading to the recruitment of PIP2-binding protein to plasma membrane (Lolicato et al., 2022). In addition, cholesterol depletion leads to vinculin mobilization (Nagasato et al., 2017). It is well accepted that lipid rafts function as a platform for integrating multiple intracellular signaling pathways (Simons and Toomre, 2000). Tyrosine kinases, such as Src family kinases and FAK, that localize at FAs are recruited to lipid rafts, and their activation is regulated by proteins that localize to lipid rafts (Kajiwara et al., 2014; Oneyama et al., 2008). Because the FAK–Src axis is a well-known pathway that regulates various FA-mediated signaling, including phosphoinositide 3-kinase (PI3K)-Akt and ERK proteins (Baillat et al., 2008; Mitra and Schlaepfer, 2006; Webb et al., 2004), it is reasonable to speculate that lipid raft-like specific membrane might function to regulate FA turnover and FA-mediated signaling.

Another important point of this study is that a method to isolate lipids from the FA-rich fraction would be a powerful tool for investigating the lipid composition of the FA membrane, the details of which remain largely unknown despite accumulating studies showing the importance of lipids in FA functions. It is well known that several FA proteins are activated by binding to PIP2 (Gilmore and Burridge, 1996; Martel et al., 2001). Furthermore, the alteration of transbilayer PC and PS compositions seen upon overexpression of flippases for each phospholipid is reported to affect FA formation (Miyano et al., 2016). Our method could be used to study the abundance of lipids in the FA-rich fraction, as mentioned above, to understand the FA not only from the standpoint of proteins but also from the lipidic point of view.

In conclusion, imaging analysis using the cholesterol probe EGFP–D4 indicates that cholesterol is enriched in FAs. Furthermore, we have established a method to isolate lipids from the FA-rich fraction. Biochemical lipid quantification showed that the FA-rich fraction has membranes enriched in cholesterol and SM as well as PC with saturated fatty acids. This study provides the first direct evidence that FA membranes have special lipid compositions and facilitates our understanding of the mechanisms by which cell migration, proliferation and differentiation are regulated by cell–ECM adhesions.

pCDH-EF1-EGFP-vinculin-IRES-Hygro and pCDH-EF1-mCherry-SGX3-vinculin-IRES-Hygro were previously reported (Ichikawa et al., 2017; Nagasato et al., 2017). pET28b-His-EGFP-D4 [kindly provided by Dr Toshihide Kobayashi (UMR 7021 CNRS, Université de Strasbourg, France)] was previously reported (Ishitsuka et al., 2011). Lyn-Cerulean-GAI(1-92) was constructed by replacing CFP in Lyn-CFP-GAI(1-92) (Addgene #37311) with Cerulean. Clathrin light chain [kindly provided by Dr Takahiro Fujiwara (Kyoto University, Japan)] was subcloned into pCDH-EF1-mCherry-SGX3-IRES-Hygro. pEGFP-N1-mCaveolin1-mCherry was constructed by replacing mEOS3.2 in pEGFP-N1-mCaveoin1-mEOS3.2 (kindly provided by Dr Takahiro Fujiwara) with mCherry. EGFP–D4 was subcloned from pET28b-His-EGFP-D4 into the pCDH-EF1-IRES-Hygro vector and a point mutation (D434S) was introduced in order to express EGFP–D4 and EGFP–D4H in cells.

Mouse monoclonal anti-KDEL [10C3, sc58774, 1:500 for western blotting (WB), 1:100 for immunostaining (IF)], anti-GFP (B-2, sc-9996, 1:2000 for WB), and rabbit polyclonal anti-ERK2 (C14, sc-154, 1:10,000 for WB) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Mouse monoclonal anti-MTCO1 (1D6E1A8, ab14705, 1:1000 for WB, 1:100 for IF), mouse monoclonal anti-α1 Na⁺/K⁺ ATPase (464.6, ab7671, 1:100 for IF), rabbit monoclonal anti-GM130 (EP892Y, ab52649, 1:1000 or WB, 1:100 for IF), and rabbit polyclonal anti-vinculin (ab73412, 1:100 for IF) antibodies were purchased from Abcam Biotechnology (Cambridge, UK). Mouse monoclonal anti-vinculin (hVIN-1, V9131, 1:10,000 for WB) and anti-talin (8d4, 1:100 for IF) antibodies were purchased from Sigma-Aldrich (St Louis, MO, USA). Rabbit polyclonal anti-phospho-paxillin (Tyr 118) (PP4501, 1:1000 for WB) antibody was purchased from ECM biosciences. Rabbit monoclonal anti-Akt (pan) (C67E7, #4691, 1:1000 for WB, 1:100 for IF) and anti-phospho-paxillin (#2541, 1:50 for IF) antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). Mouse monoclonal anti-transferrin receptor (H68.4, 136800, 1:1000 for WB, 1:100 for IF), anti-paxillin (5H11, 1:100 for IF) antibodies, ER tracker (E34251), Mito tracker (M7512), and FM 1-43 (T3163) were purchased from Invitrogen/Thermo Fisher Scientific (Carlsbad, CA, USA). Rabbit polyclonal anti-α5 integrin antibodies (AB1928, 1:100 for IF) were purchased from Merck (Darmstadt, Germany). Rat monoclonal anti-β1 integrin antibodies (9EG7, 550531, 1:25 for IF) were purchased from BD Biosciences (San Jose, CA, USA). Alexa Fluor 555-conjugated goat anti-mouse-IgG antibody, Alexa Fluor 633-conjugated goat anti-rabbit IgG antibody, CellMask Orange and CellMask Deep Red were purchased from Thermo Fisher Scientific (Waltham, MA, USA). MβCD, blebbistatin, phospholipase D (PLD) and choline oxidase were purchased from Sigma-Aldrich. Horseradish peroxidase (HRP) was purchased from Oriental Yeast Co. (Tokyo, Japan). AmplexRed was purchased from Thermo Fisher Scientific.

Wild-type (WT), EGFP–vinculin-expressing, and mCherry–vinculin-expressing MEFs were established as described in previous reports (Kioka et al., 2010; Nagasato et al., 2017; Yamashita et al., 2014). WT MEFs, EGFP–vinculin MEFs, mCherry–vinculin MEFs, mCherry–paxillin MEFs, and Lyn-Cerulean-GAI(1-92)-expressing MEFs were cultured in Dulbecco's modified Eagle's medium (Nacalai Tesque, Kyoto, Japan) supplemented with 10% fetal bovine serum (Gibco/Thermo Fisher Scientific) at 37°C in a humidified atmosphere of 5% CO₂. Cholesterol was depleted by incubating cells with 5 mM MβCD in medium for 30 min at 37°C as previously reported (Nagasato et al., 2017), given that this condition is sufficient for depleting cholesterol to almost 50% and a higher concentration of MβCD does not deplete more cholesterol. Cholesterol rescue was by incubating cells with a 5 mM MβCD and 0.5 mM cholesterol mixture in medium for 30 min at 37°C after cholesterol depletion.

Whole-cell lysates were prepared in lysis buffer [1% SDS and 1% cOmplete protease inhibitor cocktail (Roche) in PBS]. Equal amounts of lysates were analyzed by SDS-PAGE and western blotting as described (Kuroda et al., 2017). Uncropped images of western blots shown in Fig. 3B and Fig. S3B are shown in Fig. S5.

For EGFP–D4 purification, BL21 (DE3) was transfected with pET28b-His-EGFP-D4, and protein synthesis was induced by isopropyl β-d-1-thiogalactopyranoside. For His–EGFP purification, Freestyle 293-F cells (Thermo Fisher Scientific) was transfected with pcDNA3.1(+)-His-EGFP (Hirayama et al., 2013). The cells were collected and re-suspended in ice-cold 300 mM NaCl and 10 mM imidazole in PBS containing cOmplete protease inhibitor cocktail. After lysis by ultrasonication, cell debris was removed by centrifugation (1600 g for 45 min) and the supernatant collected. The His-tagged protein was trapped using immobilized metal affinity chromatography Ni-charged resin (Bio-Rad, Hercules, CA, USA) and eluted using 300 mM NaCl and 500 mM imidazole in PBS. The eluate was concentrated using the Amicon ultra-4 (30 k) centrifugal filter device (Merck).

Intact and FA-rich fraction samples were fixed with 4% (w/v) paraformaldehyde and 2% (w/v) glutaraldehyde in PBS. Intact cells were permeabilized with 0.2% Triton X-100 in PBS. Samples were blocked with 10% goat serum in PBS. The samples were then immunostained by incubating cells with primary antibodies overnight at 4°C and with secondary antibodies for 1 h at room temperature (RT). Membrane was stained using 1:12,500 diluted CellMask Orange in PBS for 20 min at RT. EGFP–D4 staining was performed by incubating cells in 5 µg/ml EGFP-D4 and 0.01% BSA in Hank's balanced salt solution (HBSS) for 30 min at 20°C to suppress endocytosis.

Confocal images were acquired using LSM laser scanning confocal microscope with a Plan-Apochromat 63×/1.4 NA oil immersion objective lens (Carl Zeiss, Oberkochen, Germany). TIRF images were taken with a Nikon Eclipse Ti microscope with an Apo TIRF 100×/1.49 NA oil immersion objective lens with or without a 4× magnification lens. Live TIRF images were taken using a stage top chamber (Tokai Hit, Fujinomiya, Japan) adjusted to 37°C in 5% CO₂. TIRF angles were set to observe ∼100 nm from the dish surface. Image quantifications were performed using ImageJ software.

Super-resolution images were acquired using N-SIM (Nikon, Tokyo, Japan) with an Apo TIRF 100×/1.49 NA oil immersion lens following the manufacturer's protocol as previously described (Ichikawa et al., 2017).

The method of isolation of FA-rich fraction was designed according to a previous report (Kuo et al., 2011). WT MEFs or EGFP–vinculin-expressing MEFs were seeded (24 h, 6.0×10⁵ cells) on 10-cm dishes coated with 5 µg/ml fibronectin in PBS (Sigma-Aldrich) and blocked with 1% bovine serum albumin (BSA; Sigma-Aldrich, A7638) in PBS. Cells were treated with hypotonic buffer (2.5 mM triethanolamine, pH 7.0) for 1.5 min. The nucleus, cytosol, organelles and most plasma membrane were removed using PBS in a Waterpik WJ6RW Interplak Dental Water Jet (Conair, Stamford, CT, USA). The water jet power was set to 300 ml/min for FA isolation and set to 150 ml/min for plasma membrane isolation (the ‘weak’ water jet condition). The plate surface was washed thoroughly for 2 min using a water jet. The intact and FA-rich fractions were scraped from the plates with TBS containing cOmplete Protease Inhibitor Cocktail and Phosphatase Inhibitor Cocktail 2 (Sigma-Aldrich) for western blotting and lipid quantification. For immunostaining for the FA-rich fractions, the cells were fixed with 4% (w/v) paraformaldehyde in PBS, then the dishes were cut into 4×2 cm squares.

The GPMV isolation method was designed according to previous reports (Baumgart et al., 2007; McGraw et al., 2019; Sezgin et al., 2012). Briefly, MEFs were seeded (6.0×10⁵ cells) on fibronectin-coated 10-cm dishes. After 24 h, the cells were washed and incubated with GPMV NEM buffer (5 mM NEM, 5 mM DTT, 10 mM HEPES, 150 mM NaCl and 2 mM CaCl₂) for 1 h. The buffers were collected as GPMV fractions for western blotting and lipid quantification. For live imaging, GPMVs were stained for membrane markers (1:12,500 diluted CellMask Orange in PBS or 5 µg/ml FM 1-43) or with 1 µM ER tracker or 0.5 µM Mito Tracker for 30 min at 37°C.

Lipids were extracted using the Bligh and Dyer method (Bligh and Dyer, 1959) with a ratio of 1:1 chloroform:methanol. Lipids of FA-rich fractions from three dishes were used as one sample of FA-rich fractions. For intact fractions, lipids from 1/30th of one dish were subjected to quantification. The extracted lipids were dried and re-dissolved in reaction mixture (0.1% Triton X-100 and 5 mM sodium cholate in PBS). Choline lipids and cholesterol were quantified as previously described (Amundson and Zhou, 1999; Morita et al., 2012) with modification. In brief, choline lipids were first reacted with 0.5 U/ml PLD and 0.5 mM CaCl₂ in HBSS (final concentration) and then with 0.2 U/ml choline oxidase, 0.02 U/ml HRP and 50 µM AmplexRed in PBS. Cholesterol was first reacted with HBSS and then with 0.2 U/ml cholesterol oxidase, 0.02 U/ml HRP and 50 µM AmplexRed in PBS. Each reaction was conducted for 30 min at 37°C. Fluorescence (excitation and emission of 530 nm and 590 nm, respectively) was measured using the Cytation 5 Imaging Reader (Bio Tek Instruments, Winooski, VT, USA).

Analysis of phospholipids was performed on a high-performance liquid chromatography system, Shimadzu LC-30AD (Shimadzu, Kyoto, Japan), coupled to a triple quadrupole mass spectrometer, LCMS-8040 (Shimadzu, Kyoto, Japan), as described previously (Suito et al., 2018). The spectrometer parameters were as follows: nebulizer gas flow 2 l/min, drying gas flow 15 l/min, interface voltage 4.5 kV, DL temperature 250°C, and heat block temperature 400°C. The multiple reaction monitoring (MRM) transition was [M+H]⁺→[184.1]+ for PC, [M+H]⁺→[184.1]⁺ for SM, and [M+H]⁺→[M+H - 141.0]⁺ for PE.

Data are presented as the mean±s.e.m. Statistical significance was calculated by one-way analysis of variance using the Tukey's multiple comparison test when comparing multiple groups. Statistical analyses were performed using Origin 8.6J software.

We thank Nikon Instech Co., Ltd. (Tokyo, Japan) for technical assistance with microscopy. We also thank Dr Toshihide Kobayashi (CNRS, France) for providing the EGFP–D4 plasmid. We thank Dr Shiho Ito for her advice on purification of EGFP–D4. Super-resolution imaging was partly supported by North Campus Instrumental Analysis Station, Kyoto University.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260763.reviewer-comments.pdf