A Role for Sphingomyelin-Rich Lipid Domains in the Accumulation of Phosphatidylinositol-4,5-Bisphosphate to the Cleavage Furrow during Cytokinesis

Abstract

Cytokinesis is a crucial step in the creation of two daughter cells by the formation and ingression of the cleavage furrow. Here, we show that sphingomyelin (SM), one of the major sphingolipids in mammalian cells, is required for the localization of phosphatidylinositol-4,5-bisphosphate (PIP₂) to the cleavage furrow during cytokinesis. Real-time observation with a labeled SM-specific protein, lysenin, revealed that SM is concentrated in the outer leaflet of the furrow at the time of cytokinesis. Superresolution fluorescence microscopy analysis indicates a transbilayer colocalization between the SM-rich domains in the outer leaflet and PIP₂-rich domains in the inner leaflet of the plasma membrane. The depletion of SM disperses PIP₂ and inhibits the recruitment of the small GTPase RhoA to the cleavage furrow, leading to abnormal cytokinesis. These results suggest that the formation of SM-rich domains is required for the accumulation of PIP₂ to the cleavage furrow, which is a prerequisite for the proper translocation of RhoA and the progression of cytokinesis.

INTRODUCTION

After chromosome segregation, the cell divides by the formation and ingression of a cleavage furrow at the plasma membrane, followed by separation into two cells. Several proteins are required for the formation and ingression of the cleavage furrow. The Rho-type GTPase RhoA is a key regulator of the furrow formation and ingression. RhoA regulates the ingression of the contractile ring and the completion of cytokinesis by activating its effectors (13). The translocation and activity of RhoA are regulated by the ECT2-MKLP1 complex in a microtubule-dependent manner (18, 46).

Several lines of evidence suggest that specific lipids are involved in cytokinesis (5, 30, 39). Phosphatidylinositol-4,5-bisphosphate (PIP₂) accumulates at the inner leaflet of the cleavage furrow during cytokinesis in mammalian cells (10), whereas phosphatidylethanolamine is exposed to the outer leaflet (6, 7). The inhibition of PIP₂ production blocks the recruitment of RhoGTPase to the site of cytokinesis, resulting in a defect of cytokinesis (33, 45). Cholesterol is concentrated at the cleavage furrow during cytokinesis in animal cells (31). The depletion of cholesterol or the inhibition of its synthesis impairs cytokinesis (8, 9). Sphingolipids are also involved in cytokinesis. The inhibition of sphingolipid biosynthesis induces the formation of multinuclear cells due to a defect in cytokinesis in yeast (38). Sphingolipids are required for the completion of cytokinesis in germ cells and protozoan parasites (2, 11, 32). However, little is known about the role of sphingolipids in this cytokinetic event.

Sphingomyelin (SM) is a major sphingolipid, comprising approximately 10% of the total phospholipids in mammalian cells. Together with cholesterol, SM forms specific liquid-ordered lipid domains in model membranes (24, 25). The existence and function of these domains in biological membranes are a matter of debate (17, 23). Recently, we developed methods for observing SM in vivo using lysenin, an earthworm protein that binds specifically to SM-rich domains (16, 19, 42).

In this study, lysenin was used to analyze the physiological role of SM in cytokinesis. Our results indicate that SM-rich domains in the outer leaflet are required for the enrichment of PIP₂ in the inner leaflet of the plasma membrane, which in turn recruits RhoA for proper cytokinesis.

MATERIALS AND METHODS

Lipid probes.

pQE30-EGFP-lysenin-161-297, expressing the nontoxic EGFP-lysenin, was constructed by replacing Venus in pQE30-Venus-lysenin-161-297 (19) with PCR-amplified enhanced green fluorescent protein (EGFP). pQE30-lysenin-161-297, expressing the nontoxic lysenin, was constructed by removing EGFP from pQE30-EGFP-lysenin-161-297. pQE30-EGFP-PH, expressing the EGFP-PH domain, was constructed by replacing lysenin-161-297 in pQE30-EGFP-lysenin-161-297 with the PH domain of human PLCδ 1, which was obtained from HeLa cell cDNA by PCR amplification. Recombinant proteins were expressed in Escherichia coli strain JM109 and purified using HisTrap FF crude columns (GE Healthcare, England). Purified nontoxic lysenin was labeled with an Alexa 647 labeling kit (Invitrogen, CA). Enzyme-linked immunosorbent assay (ELISA) was carried out as described previously (19). Anti-mCherry antibody (TaKaRa Bio, Japan) and anti-His antibody (Qiagen, CA) were used as primary antibodies for ELISA.

Cell culture and drug treatments.

HeLa cells were grown at 37°C in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, CA) supplemented with 10% fetal bovine serum. For synchronizing cells, HeLa cells were synchronized with 40 ng/ml nocodazole (Sigma-Aldrich, MO) for 3 h, and mitotic cells were harvested by shake-off. The harvested cells were plated in a poly-d-lysine-coated dish (BD, NJ) and incubated in the presence of 40 ng/ml nocodazole for an additional 30 min. Nocodazole then was washed out, and the cells were incubated under various conditions. For SMase experiments, after the nocodazole wash, HeLa cells were treated with 2.5 IU/ml of Staphylococcus aureus SMase (Sigma-Aldrich, MO). Treatment with the CERT inhibitor HPA12 was done as described previously (43). LLC-PK1 cells were grown at 37°C in Medium 199 (Invitrogen, CA) supplemented with 5% fetal bovine serum.

Cell staining.

For staining the SM-rich domain, HeLa cells were incubated in DMEM supplemented with 10% lipoprotein-deficient serum containing 10 μg/ml of nontoxic EGFP-lysenin. To label the cholesterol-rich domain, HeLa cells were incubated in DMEM supplemented with 10% fetal bovine serum containing 10 μg/ml of EGFP-domain 4 of theta toxin (D4). For the immunostaining of RhoA, cells were fixed with 10% trichloroacetic acid as described previously (18).

Expression of histone H2B, MKLP1, PH, synaptojanin, TubbyC, PIP5Kβ, C. elegans RhoA (CeRhoA), lysenin, D4, tubulin, and MRLC in cells.

The coding sequences for human histone H2B, MKLP1, the PH domain of PLCδ 1, synaptojanin, the Tubby domain (TubbyC), and phosphatidylinositol 4-phosphate 5-kinase β (PIP5Kβ) were obtained from HeLa cell cDNA by PCR amplification. The coding region of Caenorhabditis elegans RhoA was obtained from the C. elegans cDNA library by PCR amplification. Amplified EGFP, mCherry, Dronpa (1), and PAmCherry1 (37) were cloned into expression vector pcDNA-DEST40 (Invitrogen, CA), generating DEST40-EGFP, DEST40-mCherry, DEST40-Dronpa, and DEST40-PAmCherry1, respectively. The coding sequences were cloned into pDsRed-Monomer-N1, pEGFP-N1 (TaKaRa Bio, Japan), DEST40-EGFP, DEST40-mCherry, DEST40-Dronpa, or DEST40-PAmCherry1. Fragments of lysenin and D4 were cloned into DEST40-mCherry. For expressing EGFP-tubulin and EGFP-myosin II regulatory light chain (MRLC), pEGFP-Tub (TaKaRa Bio, Japan) and pEGFP-MRLC (18) were used, respectively. HeLa or LLC-PK1 cells were transfected with the expression vectors by Lipofectamine LTX (Invitrogen, CA) and were cultured in the presence of 1 mg/ml G418 (Nacalai Tesque, Japan) for 14 days. Stable clones were selected.

Confocal microscopy.

Time-lapse microscopic observation was carried out on the FV 1000 confocal microscope with a 60×, 1.1-numerical-aperture PlanApo objective lens (Olympus, Japan) equipped with an environmental chamber maintained with humidity at 37°C and 5% CO₂. Images were captured using FV10-ASW software (Olympus, Japan).

PALM/dSTORM imaging.

Photoactivated localization microscopy and direct stochastic optical reconstruction microscopy (PALM/dSTORM) imaging was carried out on a prototype Carl Zeiss microscope as described previously (27, 28). For staining the SM-rich domain, LLC-PK1 cells were incubated in Medium 199 supplemented with 5% lipoprotein-deficient serum containing 10 μg/ml of Alexa Fluor 647-lysenin. Cells were incubated for 10 min and fixed with 4% paraformaldehyde and 0.2% glutaraldehyde for 30 min before observation.

Polyphosphoinositide analysis.

Cells were labeled for 4 h with 0.1 μCi/ml [³³P]orthophosphoric acid in phosphate-free DMEM containing 40 ng/ml nocodazole, and mitotic cells were harvested by shake-off. The harvested cells were plated in a poly-d-lysine-coated dish (BD, NJ) and incubated a further 30 min. Cells were washed twice with ice-cold phosphate-buffered saline (PBS) and then scraped on ice in 2.4 N HCl solution. Lipid extraction was performed as described previously (21). The lipids were separated on high-performance thin-layer chromatography (HPTLC) plates that were first premigrated in a methanolic solution of potassium oxalate (1%, wt/vol) using the solvent system chloroform-acetone-methanol-acetic acid-water (80:30:26:24:14, vol/vol). The radioactive spots identified by comparison with lipid standards were quantified with a Fujifilm Imaging Plate using a BAS 5000 Bioimaging Analyzer (Fuji Film, Japan).

Sphingolipid analysis.

Cells were labeled for 24 h in DMEM with 10% fetal bovine serum containing l-[U-¹⁴C]serine (1 μCi/ml). For the last 4 h, 40 ng/ml of nocodazole was added to synchronize cells. Cells were washed twice with cold PBS and then scraped on ice in 2 mM EDTA. Aliquots of cell extract were taken for protein quantification, and lipid extraction was performed according to Bligh and Dyer (3). Lipids were separated on HPTLC plates with a solvent mixture of methyl acetate-n-propanol-chloroform-methanol-0.25% KCl (25:25:25:10:9, vol/vol) (43). Radioactive lipids were quantified with a BAS 5000 Bioimaging Analyzer.

Cholesterol analysis.

Aliquots of cell extract were taken for protein quantification, and lipid extraction was performed according to Bligh and Dyer (3). Stigmasterol (4 μg) was added as an internal standard. The lipid extracts were applied to HPTLC plates developed in hexane-diethyl ether-acetic acid (80:20:2, vol/vol/vol). The free cholesterol (FC) spots, revealed under UV after samples were sprayed with primuline solution, were collected and then extracted with a mixture of methanol-water-hexane (2:1:2). After centrifugation, the upper hexane phase was collected, dried under nitrogen, and then analyzed with a Shimadzu GC-14AH gas chromatograph (Shimadzu, Japan). An Econo-Cap EC-5 (Alltech Associate Inc., IL) capillary column (30 m by 0.32 mm; 0.25 μm) was used with helium as the carrier gas, and the oven temperature program was from 290 to 320°C at 2°C/min and with isothermal holding at 320°C for 10 min. Cholesterol was quantified using stigmasterol as the internal standard.

RESULTS

SM-rich domains are concentrated in the outer leaflet of the cleavage furrow and are required for cytokinesis.

To observe the localization of SM during cytokinesis, we stained the cells with a labeled lysenin. First, we stained the SM in the outer leaflet of the plasma membrane with the exogenously added EGFP-lysenin. We found that the fluorescence intensity of EGFP-lysenin increased around the region of the contractile ring and the midbody (Fig. 1A; also see Movie S1 in the supplemental material). To exclude the possibility that the increased fluorescence intensity is due to the closely apposed membranes in the furrow region, we compared EGFP-lysenin staining to that of a lipophilic dye. We stained the mitotic cells with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiIC₁₈) as a nonspecific membrane probe. We found that DiIC₁₈ was evenly distributed on the plasma membrane (Fig. 1B). Quantitative analysis of the fluorescence intensity indicates that, compared to DiIC₁₈, EGFP-lysenin is significantly accumulated in the cleavage furrow (Fig. 1B and C). We examined whether SM was localized to the inner leaflet of the plasma membrane during cytokinesis by expressing mCherry-lysenin in HeLa cells (Fig. 1D). Although intracellular dots and faint cytoplasmic staining were observed, the inner leaflet of the plasma membrane did not display any evident staining. ELISA confirmed that both EGFP-lysenin and mCherry-lysenin specifically bound to SM (Fig. 1E). These results demonstrate that SM-rich domains stained by lysenin are selectively concentrated to the outer leaflet of the cleavage furrow during cytokinesis.

Fig 1.

SM-rich domains are concentrated in the outer leaflet of the cleavage furrow. (A) SM is concentrated in the cleavage furrow. HeLa cells stably expressing histone H2B-DsRed (red) were stained with EGFP-lysenin (green). The pictures were taken at the indicated times. Bar, 5 μm. (B) DiIC₁₈ is evenly distributed on the plasma membrane. The upper portion shows HeLa cells stably expressing histone H2B-DsRed (red) stained with EGFP-lysenin (green) and DiIC₁₈ (red). (C) Quantitative analysis of fluorescence intensity. The fluorescence intensity of EGFP-lysenin and DiIC₁₈ was measured in the cleavage furrow and the polar membrane. The intensities were normalized to the intensity at the polar region. Data are means ± SD (n > 20). (D) SM-rich domains are localized in the outer leaflet of the cleavage furrow. HeLa cells stably expressing mCherry-lysenin (red) were incubated with purified EGFP-lysenin (green). Bar, 5 μm. (E) EGFP-lysenin and mCherry-lysenin bind specifically to SM. Purified protein of EGFP-lysenin from E. coli (green) and cell lysate of HeLa cells expressing mCherry-lysenin (red) were assayed for ELISA. Data are means ± SD (n = 3).

We estimated the concentration of SM at the cleavage furrow. First, we made uniform giant unilamellar vesicles (GUVs) containing several concentrations of SM (0 to 60%) and egg phosphatidylcholine (PC) as standards. To make a standard curve, we stained the GUVs with EGFP-lysenin and quantified the fluorescence intensity. We found that the fluorescence intensity linearly increased when the SM concentration was increased from 0 to 60% (Fig. 2A). We stained the mitotic cells with EGFP-lysenin. To avoid overestimation due to the apposed two membranes, we measured the fluorescence intensity of one membrane around the furrow region. Using a standard curve, we estimated that the concentrations of SM in the outer leaflet at the cleavage furrow and the pole region were 37% ± 13% and 10% ± 6% (n > 50), respectively (Fig. 2B).

Fig 2.

Estimation of the concentration of SM at the cleavage furrow. (A) Fluorescence intensity of EGFP-lysenin in giant unilamellar vesicles (GUVs). GUVs containing several concentrations of palmitoyl SM (0 to 60%) and egg PC were made and stained with EGFP-lysenin. Fluorescence intensity was measured at 37°C. Data are means ± SD (n > 15). Bar, 5 μm. (B) Fluorescence intensity of EGFP-lysenin in cells. HeLa cells stably expressing histone H2B-DsRed (red) were stained with EGFP-lysenin (green). The fluorescence intensity of EGFP-lysenin was measured around the furrow region (square) and the polar region (circle). Data are means ± SD (n > 50).

To address the possibility that the accumulation of SM is required for cytokinesis, we depleted SM by sphingomyelinase (SMase) treatment. We found that in 50% of cells (n > 30) the cleavage furrows did not progress properly but regressed, resulting in binucleated cells in 3 h (Fig. 3A and B; also see Movie S2 in the supplemental material). In control cells, no defect in cytokinesis was observed, suggesting that the addition of EGFP-lysenin did not affect cytokinesis. This treatment decreased SM to 19.7% ± 2.6% of the control level (n = 3) (Fig. 3C). These results suggest that SMase treatment inhibits proper cytokinesis.

Fig 3.

SM is required for proper cytokinesis. (A) Depletion of SM results in regression of the cleavage furrow. HeLa cells stably expressing histone H2B-DsRed (red) were stained with EGFP-lysenin (green). The cells were incubated in the presence of SMase. Bar, 5 μm. (B) Quantification of the phenotype. Cells were observed for 3 h (n > 30) and classified into 3 groups. Gray, black, and white colors indicate cells with normal cytokinesis, cells without nuclear division, and cells with a regressed furrow, respectively. (C) SMase treatment decreases the amount of SM in HeLa cells. Cells were labeled with l-[U-¹⁴C]serine for 24 h and incubated in the presence of 40 ng/ml nocodazole for the last 4 h. The mitotic cells were collected and incubated with 2.5 IU/ml of SMase for 1 h. The lipids were extracted and separated on HPTLC plates.

SMase treatment may alter the gross integrity of the plasma membrane. To rule out the possibility that the abnormal cytokinesis was due to effects other than the decrease of SM, we tested whether the SMase-induced regression was recovered by adding back exogenous SM. As observed above (Fig. 3B), the cleavage furrows were regressed in 47% of the SMase-treated cells (n > 40) (Fig. 4A and D). The regression was not significantly recovered by adding exogenous PC after SMase treatment. The cleavage furrows were still regressed in 44% of the cells (n > 40) (Fig. 4B and D). However, the regression was observed only in 7% of cells when exogenous SM was added after SMase treatment (n > 40) (Fig. 4C and D). These results suggest that the abnormal cytokinesis is induced by the decrease of SM. We also confirmed that the increase of ceramide after SMase treatment (Fig. 3C) did not induce the abnormal cytokinesis. We added C₆ or brain ceramide to HeLa cells and observed the phenotype. The addition of ceramide did not affect the ingress of the cleavage furrow (Fig. 4E), suggesting that the ceramide increase does not cause the abnormal cytokinesis. From these results, we conclude that SM is required for cytokinesis in the cells.

Fig 4.

Abnormal cytokinesis is due to the SM decrease. (A) The cleavage furrows regress in the SMase-treated cells. HeLa cells stably expressing histone H2B-DsRed were incubated with 2.5 IU/ml of SMase for 1 h. The cells were washed and incubated in DMEM supplemented with 10% lipoprotein-deficient serum for an additional 3 h. (B) The regression phenotype is not suppressed by adding exogenous PC. After treatment with SMase for 1 h, the cells were washed and incubated in medium containing 20 μM egg PC (Avanti Polar Lipids, AL) for an additional 3 h at 37°C. (C) The regression phenotype is suppressed by adding exogenous SM. After treatment with SMase, the cells were washed and incubated in the medium containing 20 μM brain SM (Avanti Polar Lipids, AL) for an additional 3 h at 37°C. (D) Quantification of the phenotype. Gray, black, and white colors indicate cells with normal cytokinesis, cells without nuclear division, and cells with regressed furrows, respectively. (E) The addition of ceramide does not cause defects in cytokinesis. HeLa cells stably expressing histone H2B-DsRed (red) were incubated with 20 μM of brain ceramide (Avanti Polar Lipids, AL). Bars, 5 μm.

We next observed the cytoskeleton in the SMase-treated cells. Cytokinesis involves cleavage furrow formation followed by the appearance of the midbody and eventually cell division. To examine whether the midbody is formed before regression in the SMase-treated cells, we observed tubulin in living HeLa cells stably expressing EGFP-tubulin. In control cells, EGFP-tubulin was localized to the central spindle at anaphase (Fig. 5A, 30 min) and to the midbody at a late stage of cytokinesis (Fig. 5A, 100 min). Similarly to control cells, EGFP-tubulin localized to the central spindle in the SMase-treated cells. However, during further incubation, EGFP-tubulin did not concentrate in the midbody.

Fig 5.

Cytoskeleton dynamics in the SMase-treated cells. (A) The cleavage furrow is regressed before forming the midbody in the SMase-treated cells. HeLa cells stably expressing both histone H2B-DsRed (red) and EGFP-tubulin (green) were observed in the absence (upper) or presence (lower) of SMase. (B) EGFP-MRLC is accumulated in the cleavage furrow. HeLa cells stably expressing both histone H2B-DsRed (red) and EGFP-MRLC (green) were observed in the absence (left) or presence (right) of SMase. Bars, 5 μm.

We also examined myosin II by stably expressing EGFP-fused myosin II regulatory light chain (MRLC) in HeLa cells. Both in control and SMase-treated cells, EGFP-MRLC was accumulated to the site of furrow ingression (Fig. 5B). These results suggest that the cleavage furrow normally forms but regresses before the completion of the midbody formation in SMase-treated cells.

Accumulation of PIP₂, but not cholesterol, is abolished by depletion of SM.

SM is postulated to form specific lipid domains together with cholesterol (17, 23, 36). Cholesterol has been shown to accumulate in the cleavage furrow in sea urchin eggs (31). We next examined whether cholesterol is concentrated at the cleavage furrow in the cells treated with SMase. Using domain 4 of theta toxin (D4) (35), we stained the cholesterol-rich domain in both the outer and inner leaflets of the plasma membrane in living cells during cytokinesis (Fig. 6A). To observe the cholesterol-rich domains in the outer leaflet of the plasma membrane, exogenous recombinant EGFP-D4 was added to the cells. For the staining of the cholesterol-rich domains at the inner leaflet of the plasma membrane, we expressed mCherry-D4 in the cells after plasmid transfection. The cholesterol-rich domains stained with EGFP-D4 on the outer leaflet were accumulated at the site of the furrow ingression in control cells (Fig. 6A, upper). In contrast, mCherry-D4 fluorescence was evenly distributed to the inner leaflet of the plasma membrane throughout cytokinesis (Fig. 6A, lower). These results indicate that the cholesterol-rich domains in the outer, but not the inner, leaflet of the plasma membrane selectively accumulate in the cleavage furrow during cytokinesis. We tested whether treatment with SMase affects the enrichment of cholesterol in the outer leaflet of the cleavage furrow (Fig. 6B). In the SMase-treated cells, the cholesterol-rich domains stained with EGFP-D4 were still concentrated in the cleavage furrow before regression, whereas mCherry-D4 stained all around the inner leaflet of the plasma membrane, as observed in the control cells. Quantitative analysis confirmed that EGFP-D4 was concentrated to the cleavage furrow in both control cells and SMase-treated cells (Fig. 6C). We confirmed by ELISA analysis that both EGFP-D4 and mCherry-D4 selectively bound to cholesterol (Fig. 6D). These results suggest that SM is not required for the accumulation of cholesterol in the outer leaflet of the cleavage furrow.

Fig 6.

SM is not required for the accumulation of cholesterol in the cleavage furrow. (A) Cholesterol-rich domains are concentrated in the outer leaflet of the cleavage furrow. HeLa cells transiently expressing mCherry-D4 (red) were incubated with purified EGFP-D4 (green). (B) Depletion of SM does not affect the accumulation of cholesterol-rich domain. The cells described above were incubated in the presence of SMase. Bars (A and B), 5 μm. (C) Quantitative analysis of fluorescence intensity. The fluorescence intensity of EGFP-D4 was measured in the cleavage furrow (square) and the polar region (circles). The intensities were normalized to the intensity at the polar region. Data are means ± SD (n > 20). (D) EGFP-D4 and mCherry-D4 bind specifically to cholesterol. Lipid binding properties of purified protein of EGFP-D4 from E. coli (green) and cell extracts from HeLa cells expressing mCherry-D4 (red) were assayed by ELISA. Data are means ± SD (n = 3).

PIP₂ accumulates in the inner leaflet of the cleavage furrow during cytokinesis in mammalian cells (6, 10). We observed the localization of PIP₂ during cytokinesis by stably expressing the PH domain of PLCδ 1 (6, 10) fused to mCherry. As reported, PIP₂ stained with mCherry-PH was highly concentrated in the inner leaflet of the cleavage furrow (Fig. 7A, upper; also see Movie S3 in the supplemental material). In addition, SM in the outer leaflet and PIP₂ in the inner leaflet are colocalized during cytokinesis (Fig. 7B and C). Interestingly, such PIP₂ accumulation was not observed in the SMase-treated cells (Fig. 7A, lower; also see Movie S4 in the supplemental material). Quantitative analysis of fluorescence intensity confirmed that PIP₂ was not accumulated in the furrow in the SMase-treated cells (Fig. 7D). We were not able to detect PIP₂ staining in the outer leaflet of the plasma membrane in control or SMase-treated cells using exogenously added EGFP-PH domain (data not shown). We measured the total amount of PIP₂ when the cells were treated with SMase. Quantitative analysis of PIP₂ indicates that the total amount of PIP₂ was not significantly decreased by the SMase treatment (95.5% ± 4.0% of control; n = 3) (Fig. 7E). These results suggest that SM at the plasma membrane is required for the accumulation of PIP₂ in the inner leaflet of the cleavage furrow.

Fig 7.

Depletion of SM abolishes PIP₂ accumulation in the cleavage furrow. (A) Treatment with SMase does not concentrate PIP₂ into the furrow. Cells expressing mCherry-PH (red) were incubated in the absence (upper) or presence (lower) of SMase, and then pictures were taken at the indicated times. Bars, 5 μm. (B) SM and PIP₂ are colocalized to the cleavage furrow during cytokinesis. Cells expressing mCherry-PH (red) were stained with EGFP-lysenin (green), and then pictures were taken at the indicated times. Bar, 5 μm. (C) Fluorescence intensity of EGFP-lysenin (green) and mCherry-PH (red) was measured along the line. (D) Quantitative analysis of fluorescence intensity. The fluorescence intensity of mCherry-PH was measured in the cleavage furrow (square) and the polar region (circles). The intensities were normalized to the intensity at the polar region. Data are means ± SD (n > 20). (E) SMase treatment does not decrease the amount of PIP₂. Cells were labeled with [³³P]orthophosphoric acid. The mitotic cells were incubated with or without SMase for 1 h.

Depletion of SM inhibits the recruitment of RhoA to the cleavage furrow.

It has been shown in Saccharomyces cerevisiae that PIP₂ is one factor that is required for the translocation of RhoGTPase to the site of the contractile ring formation; the depletion of PIP₂ results in a mislocalization of RhoGTPase in the furrow (45). RhoA localization was observed by immunofluorescence microscopy in HeLa cells fixed with trichloroacetic acid (Fig. 8A). As reported previously (44), RhoA was localized to the cleavage furrow and the midbody in control cells. In contrast, in SMase-treated cells, RhoA was no longer localized to the cleavage furrow. To exclude the possibility that RhoA was not observed in SMase-treated cells due to the trichloroacetic acid fixation, we studied its localization in living cells. Since EGFP-tagged C. elegans RhoA (CeRhoA) was shown to be properly localized to the cleavage furrow in mammalian cells (46), we analyzed the EGFP-CeRhoA localization in living HeLa cells. As reported, EGFP-CeRhoA was localized to the contractile ring (Fig. 8B, upper; also see Movie S5 in the supplemental material). In the cells treated with SMase, EGFP-CeRhoA was less concentrated in the cleavage furrow (Fig. 8B, lower; also see Movie S6 in the supplemental material). Quantitative analysis of fluorescence intensity confirmed that smaller amounts of RhoA accumulated in the cleavage furrow in SMase-treated cells than in control cells (Fig. 8C).

Fig 8.

RhoA is less accumulated in the cleavage furrow in SMase-treated cells. (A) RhoA is not concentrated at the cleavage furrow in SMase-treated cells. Cells were incubated in the absence (left) or presence (right) of SMase for 60 min. Cells were fixed with trichloroacetic acid and stained with anti-RhoA antibody. (B) EGFP-CeRhoA weakly localizes in the cleavage furrow after SMase treatment. HeLa cells stably expressing both histone H2B-DsRed (red) and EGFP-CeRhoA (green) were observed in the absence (upper) or presence (lower) of SMase. (C) Quantitative analysis of fluorescence intensity. The fluorescence intensity of EGFP-CeRhoA was measured in the cleavage furrow (square) and the cytosol (circle). The intensities were normalized to the intensity at the cytosolic region. Data are means ± SD (n > 20). (D) MKLP1 is normally localized to the central spindle in SMase-treated cells. LLC-PK1 cells stably expressing histone H2B-DsRed (red) and EGFP-MKLP1 (green) were observed in the absence (upper) or presence (lower) of SMase. The pictures were taken at the indicated times after the SMase incubation. Bars, 5 μm.

In addition to PIP₂, MKLP1 is reported to be required for the translocation of RhoA to the cleavage furrow (18, 46). We thus checked the possibility that RhoA did not localize to the furrow due to the abnormal localization of MKLP1. Since we were not able to obtain HeLa cells stably expressing EGFP-MKLP1, we stably expressed EGFP-MKLP1 in LLC-PK1 cells (Fig. 8D). As reported previously (18, 26, 46), EGFP-MKLP1 was accumulated at the central spindle at late anaphase. Furthermore, in SMase-treated cells, EGFP-MKLP1 was still localized to the central spindle, indicating that SMase treatment does not alter the localization of MKLP1. Taken together, the results suggest that in the SMase-treated cells, RhoA accumulation to the cleavage furrow was hindered due to the abnormal localization of PIP₂ and not the MKLP1 mislocalization.

Cholesterol is essential for SM accumulation in the cleavage furrow.

We examined how SM is accumulated in the cleavage furrow. First, we tested if the localization of SM depended on PIP₂ (Fig. 9A). It has been shown that the overexpression of the PH domain or Tubby domain (TubbyC) interferes with PIP₂, and the overexpression of synaptojanin reduces the amount of PIP₂ (10). In the cells transiently overexpressing one of these domains, we examined whether EGFP-lysenin was accumulated in the cleavage furrow. Biochemical analysis confirmed that the transient overexpression of mCherry-synaptojanin reduced PIP₂ to 49% ± 11% of the control level (n = 3). Even in the cells expressing these proteins, EGFP-lysenin staining persisted at the cleavage furrow. These results indicate that PIP₂ is not required for the accumulation of SM to the cleavage furrow. In these cells, EGFP-CeRhoA was less concentrated in the cleavage furrow than in control cells (Fig. 9B), indicating that PIP₂ is required for the accumulation of RhoA to the cleavage furrow.

Fig 9.

PIP₂ is required for the accumulation of RhoA but not SM. (A) EGFP-lysenin is accumulated in the cleavage furrow when PIP₂ is interfered with. SM was stained with EGFP-lysenin (green) in the cells overexpressing mCherry-PH (red), mCherry-synaptojanin (red), or mCherry-TubbyC (red). The intensities were normalized to the intensity at the polar region. Data are means ± SD (n > 20). (B) PIP₂ is required for the accumulation of RhoA in the cleavage furrow. mCherry-PH (red), mCherry-synaptojanin (red), or mCherry-TubbyC (red) was overexpressed in HeLa cells stably expressing EGFP-CeRhoA (green). The intensities were normalized to the intensity at the cytosolic region. Data are means ± SD (n > 20). Bars, 5 μm.

Second, we examined whether cholesterol was required for the localization of SM, since cholesterol facilitates the formation of SM-rich domains in model membranes (16, 40). Methyl-β-cyclodextrin (MβCD) removed cellular cholesterol and thus inhibited the cellular staining with D4 (Fig. 10B). We found that treatment with MβCD abolished the accumulation of mCherry-lysenin staining at the cleavage furrow (Fig. 10B). Treatment with MβCD reduced the amount of cholesterol (Fig. 10D), whereas it did not significantly reduce the total amount of SM (Fig. 3C). Lysenin binds SM only when the local concentration of SM is high (16). These results suggest that cholesterol in the outer leaflet of the plasma membrane is required for the SM accumulation in the cleavage furrow. Consistently, the cleavage furrow was regressed in 65% of the MβCD-treated cells (n > 30), as had been observed after SMase treatment (Fig. 3A and B).

Fig 10.

Depletion of cholesterol reduces the staining of SM-rich domains. (A) Both SM- and cholesterol-rich domains are concentrated in the cleavage furrow. HeLa cells stably expressing histone H2B-DsRed (red) were incubated with EGFP-D4 (green) and mCherry-lysenin (red). (B) Depletion of cholesterol with MβCD reduces the staining of SM- and cholesterol-rich domain-specific probes. The cells were incubated in 10 mM MβCD, and the pictures were taken at the indicated times. (C) Depletion of SM with SMase does not reduce the staining of cholesterol-rich domain-specific probes. The cells were incubated with SMase, and the pictures were taken at the indicated times. Bars (A to C), 5 μM. (D) Depletion of cholesterol with MβCD reduces the total amount of cholesterol. The mitotic cells were incubated with MβCD or SMase for 1 h. The lipids were extracted, and the amount of cholesterol was measured. Data are means ± SD (n = 3).

A transbilayer colocalization between the SM-rich domains in the outer leaflet and PIP₂-rich domains in the inner leaflet.

Several observations suggest that a significant pool of PIP₂ associates with the detergent-resistant membrane (DRM) fraction (4, 15, 22), in which SM is also included. Our results indicate that SM and PIP₂ are colocalized to the cleavage furrow during cytokinesis (Fig. 7B). These observations raise the possibility that SM and PIP₂ colocalize during the course of the whole cell cycle. However, there is no direct evidence that SM-rich domains in the outer leaflet overlap PIP₂ in the inner leaflet of the plasma membrane, since the sizes of these domains are below the resolution of conventional light microscopy. To examine the respective localization of SM and PIP₂ in the plasma membrane with high resolution, we used photoactivated localization microscopy (PALM) and direct stochastic optical reconstruction microscopy (dSTORM). We observed the lipids at the apical face of the plasma membrane at interphase (Fig. 11A). The Dronpa-labeled PH domain was expressed in LLC-PK1 cells to visualize PIP₂ at the inner leaflet, and SM at the outer leaflet was stained by the exogenous addition of Alexa 647-labeled lysenin. At the apical plasma membrane, small clusters of SM and PIP₂ were observed which were colocalized (Fig. 11A, left). We then estimated the clusters of SM and PIP₂ by spatial statistical analysis (Fig. 11B, left). The curves of the L(t) value were found to be above the higher confidence lines, with peaks at approximately 250 nm for both stainings, indicating that SM and PIP₂ form similar-sized domains. Both domains disappeared when cells were treated with SMase (Fig. 11A, center). The peak of the L(t) curve of PIP₂ was not observed (Fig. 11B, center). It has been reported that treatment with HPA12, an inhibitor of CERT, selectively decreased the amount of cellular SM (43). Although HPA12 inhibited normal cell growth, we found that the cleavage furrows were also regressed when the cells were treated with HPA12 (data not shown). We then examined the distribution of PIP₂ by PALM after HPA12 treatment (Fig. 11A, right). We found that HPA12 treatment prevented the formation of PIP₂ domains. The spatial statistical analysis revealed that the L(t) value was dramatically decreased by HPA12 treatment (Fig. 11B, right). We confirmed that under this condition, the amount of SM was decreased to 22.5% ± 3.8% of the control level (n = 3).

Fig 11.

Transbilayer colocalization between the SM-rich domains in the outer leaflet and PIP₂-rich domains in the inner leaflet of the plasma membrane. (A) PALM/dSTORM images of Dronpa-PH (green) and Alexa 647-labeled lysenin (red). LLC-PK1 cells expressing Dronpa-PH were stained with Alexa Fluor 647-labeled lysenin. (Center) Cells were treated with SMase for 1 h. (Right) Cells were treated with HPA12 for 48 h. (Lower) Magnified images. (B) Ripley's L-function analysis of Dronpa-PH (green)- and Alexa Fluor 647-labeled lysenin (red). Blue and magenta indicate higher and lower confidence envelopes of 99%, respectively. (C) PALM/dSTORM images of Dronpa-PH (green)- and Alexa Fluor 647-labeled lysenin (red). LLC-PK1 cells expressing Dronpa-PH were incubated with 2.5 IU/ml of SMase for 1 h at 37°C. The cells were washed and incubated for 1 h at 37°C. The cells were incubated for an additional 1 h at 37°C without exogenous lipids (left), with 50 μM brain SM (center), or with 50 μM egg PC (right). After incubation, the cells were washed and stained with Alexa Fluor 647-labeled lysenin. The lower panels show magnified images. Bars, 2 μm.

We then examined whether the addition of exogenous SM recovered the lack of SM and PIP₂ clusters in the SMase-treated cells (Fig. 11C). We found that the SM clusters formed again by adding exogenous SM after SMase treatment (Fig. 11C, center). Furthermore, as observed in control cells (Fig. 11A, left), the PIP₂ clusters appeared in the inner leaflet just beneath the SM clusters that located in the outer leaflet of the plasma membrane. We confirmed that the addition of exogenous PC did not induce the formation of the PIP₂ clusters in the SMase-treated cells (Fig. 11C, right). These results indicate that, at interphase, the clusters of SM and PIP₂ are colocalized at the plasma membrane and that SM is required for the formation of the PIP₂ clusters.

We next observed the distribution of SM and PIP₂ in the cells undergoing cytokinesis. We expressed PAmCherry1-PH in LLC-PK1 cells and stained the cells with Alexa 647-labeled lysenin. Using PALM/dSTORM analysis, we found that PH and lysenin stainings were localized in the cleavage furrow (Fig. 12B, left). Dronpa-labeled CeRhoA was also localized in the cleavage furrow. Since these proteins were highly concentrated, individual clusters were not identified in the midsection. On the other hand, we found several clusters of CeRhoA, PH, and lysenin that were colocalized around the cleavage furrow in the apical section (Fig. 12B, right). These results suggest that the clusters of SM and PIP₂ remain colocalized at the plasma membrane during cytokinesis.

Fig 12.

Transbilayer colocalization of SM, PIP₂, and RhoA at the cleavage furrow. (A) Diffraction-limited image of Dronpa-CeRhoA. (B) PALM/dSTORM images. LLC-PK1 cells expressing Dronpa-CeRhoA and PAmCherry1-PH were stained with Alexa Fluor 647-labeled lysenin. The midsection (left) and apical section (right) were observed. Green, red, and blue indicate Dronpa-CeRhoA, PAmCherry1-PH, and Alexa Fluor 647-lysenin, respectively. Bars, 2 μm.

PALM/dSTORM observation indicates that PIP₂ is localized just beneath the SM-rich domains. This result raised the possibility that PIP₂ is synthesized at the inner leaflet of SM clusters. Thus, we examined the localization of phosphatidylinositol 4-phosphate 5-kinase β (PIP5Kβ), which produces PIP₂ from phosphatidylinositol 4-phosphate (6, 41). PALM/dSTORM analysis revealed that PIP5Kβ formed small domains in the inner leaflet of the plasma membrane and that the domains of PIP5Kβ were localized to the inner leaflet on the opposite side of SM-rich domains (Fig. 13, left). In the SMase-treated cells, neither domain was observed (Fig. 13, right). These results suggest that PIP5Kβ is localized on the inner leaflet on the opposite side of the SM-rich domain at the plasma membrane.

Fig 13.

PIP5Kβ colocalizes with SM-rich domain. PALM/dSTORM images of Dronpa-PIP5Kβ (green) and Alexa Fluor 647-labeled lysenin (red). LLC-PK1 cells expressing Dronpa-PIP5Kβ were stained with Alexa Fluor 647-labeled lysenin. (Right) cells were treated with SMase for 1 h. The lower panels show magnified images. Bars, 2 μm.

DISCUSSION

In this study, we found that SM-rich domains are concentrated to the cleavage furrow at cytokinesis. The depletion of SM from the plasma membrane inhibits the completion of cytokinesis. In the SM-depleted cells, PIP₂ is no longer concentrated to the furrow. We analyzed the distribution of SM and PIP₂ by high-resolution microscopy. We found that the clusters of both outer leaflet SM and inner leaflet PIP₂ are colocalized at the plasma membrane, and that SM is required for the formation of the PIP₂ clusters. From these results, we conclude that the SM-rich domain in the outer leaflet of the plasma membrane is required for the enrichment of PIP₂ in the inner leaflet of the plasma membrane, which is needed for the recruitment of factors required for cytokinesis, such as RhoA.

We found that cholesterol in the outer leaflet of the plasma membrane is required for the SM accumulation at the cleavage furrow. This observation is consistent with the results using model membranes in which cholesterol facilitates the formation of SM-rich lipid domains (40). Cholesterol is accumulated to the cleavage furrow in the SMase-treated cells (Fig. 10C), suggesting that SM is not required for cholesterol accumulation in the furrow. Since we could not detect the cholesterol accumulation before the onset of anaphase, factors involved in cell cycle progression or cytoskeleton may be needed for the cholesterol accumulation.

The accumulation of RhoGTPase to the cleavage furrow is induced both by MKLP1 (18, 46) and PIP₂ (45). We found that SMase treatment did not affect the localization of MKLP1 (Fig. 8D), whereas PIP₂ was not concentrated in the furrow anymore (Fig. 7A and D) in the treated cells. We also found that RhoA was less concentrated in the cleavage furrow when PIP₂ was interfered with (Fig. 9B). These results confirm the importance of PIP₂ in the proper localization of RhoA but do not exclude the possibility that MKLP1 is also involved in the translocation of RhoA. One possibility is that MKLP1 determines the site of RhoA translocation and PIP₂ works as a platform for the stable attachment of RhoA to the plasma membrane and the completion of cytokinesis.

In PALM images, some clusters of SM localized next to the clusters of PIP₂ or did not include any clusters of PIP₂. We speculate that this phenomenon is caused by the following reasons. One reason is the chromatic aberration. When we stained SM with Dronpa-lysenin and Alexa 647-lysenin simultaneously, SM-rich domains stained with Dronpa-lysenin and Alexa 647-labeled lysenin were adjacent to one another. This result implies that the problem is caused by the chromatic aberration and suggests that SM and PIP₂ are indeed in very close proximity. Another reason is the low expression of Dronpa-PH. When we transiently overexpressed Dronpa-PH, most clusters of Alexa 647-lysenin included Dronpa-PH. This result implies that some clusters of SM could not be detected with clusters of PIP₂ by PALM due to the low expression of Dronpa-PH in the stable cell line.

Previously we showed by using immunoelectron microscopy that SM forms small domains on the outer leaflet of the plasma membrane (19). Recently the detailed characterization of PIP₂ domains in the inner leaflet of the plasma membrane was reported (12). However, the transbilayer colocalization of SM domains and PIP₂ domains has not been achieved. Using superresolution fluorescence microscopy, we demonstrate here that PIP₂ forms clusters in the inner leaflet just beneath the SM clusters in the outer leaflet of the plasma membrane (Fig. 11). In addition, our result indicates that clusters of PIP5Kβ are localized to the inner leaflet on the opposite side of SM-rich domains (Fig. 13). These results suggest that PIP₂ clusters are localized just beneath the SM clusters, since PIP₂ is produced at the inner leaflet of the SM clusters.

We found that the localization of PIP₂ is restricted around the clusters of SM and PIP5Kβ (Fig. 11 and 13). There are several possible reasons for this observation. First, SM, PIP5Kβ, and PIP₂ may have direct or indirect interactions. Due to interactions, the diffusion of PIP₂ may be restricted in the membrane. Second, PIP5Kβ is activated by RhoGTPase, which is located in the PIP₂ domain. It has been shown that RhoGTPase positively regulates the activity of PIP5Kβ (41). These regulations may enhance the formation of PIP₂ clusters at the inner leaflet of SM-rich domains. Third, the mobility of PIP₂ may be restricted due to protein fences. In fact, several lines of evidence suggest that there are diffusion barriers in the plasma membrane (20, 29, 34). Fluorescence correlation spectroscopy (FCS) and fluorescence recovery after bleaching (FRAP) analyses suggest that a protein fence limits the diffusion of PIP₂ (14). Further experiments are needed to understand how PIP₂ remains in the clusters.