The amphiphilic dye FM4–64 is used to investigate endocytosis and vesicle trafficking in living eukaryotic cells. The standing hypothesis is that it is inserted into the outer leaflet of the plasma membrane and, from there, is passed on to intracellular membrane compartments by endocytosis. We tested this hypothesis by microinjecting FM4–64 into the cytoplasm and vacuole of Nicotiana tabacum BY-2 suspension culture cells and Tradescantia virginiana stamen hair cells. We found that the dye did not label any membranes when injected into the cytoplasm, but clearly labelled the tonoplast when injected directly into the vacuole. However, because the dye is pH-sensitive, the fluorescence intensity between the plasma membrane and tonoplast varied. We conclude that FM4–64 is a specific marker for the endocytic pathway. Nevertheless, little is known about the molecular interactions of FM4–64 with these particular phospholipid membrane leaflets. We, therefore, appeal for biochemical research to determine which membrane lipids FM4–64 interacts with.
Endocytosis in plants is important for recycling the plasma membrane while depositing large amounts of cell wall material through exocytosis, as, for example, is eminent in fast, tip-growing cells such as pollen tubes and root hairs (Miller et al., 1997). On the other hand, endocytosis is used to recycle plasma membrane proteins or extracellular material for the purposes of signalling and keeping the pools of these proteins fresh and dynamic (Murphy et al., 2005; Geldner & Jürgens, 2006). Moreover, membrane recycling is essential for the growing cell plate that is laid down between the two daughter nuclei during cytokinesis. It has recently been suggested that endocytosis may also bring membrane proteins and cell wall material directly to the growing cell plate (Dhonukshe et al., 2006).
The amphiphilic styryl dye FM4–64 is widely used as a marker of the plasma membrane and the endocytic pathway in many organisms, including plants (Fischer-Parton et al., 2000; Kutsuna & Hasezawa, 2002; Bolte et al., 2004). It is fluorescent only in a lipophilic environment, for example, in phospholipid bilayer membranes. FM4–64 is thought to insert into the outer leaflet of the plasma membrane because of its lipophilic tail and highly hydrophilic, cationically charged head group (Betz et al., 1996). From there it is internalized, together with its lipid surroundings, by the natural process of endocytosis. It is hypothesized that, consequently, it can only be found in the inner membrane leaflet of the endocytic compartments: the endocytic vesicles, endosomes, Golgi bodies, prevacuolar compartments and tonoplast (Fischer-Parton et al., 2000; Bolte et al., 2004). The endoplasmic reticulum and the nuclear envelope are not labelled (Bolte et al., 2004). Because of the nature of the dye and the phospholipid bilayer, FM4–64 does not cross these membranes nor does it seem to flip-flop to the other side (Griffing, 2008; this issue of the Journal of Microscopy). Many long-chain fatty acids, however, are actively and selectively flipped from the production side, that is, the cytoplasmic side, to the other side by flippases, thereby creating the typical lipid composition asymmetry of membranes, which could be important for the distinct FM4–64 distribution (McArthur et al., 1999; Pomorski & Menon, 2006).
From our experiments with synthetic lipid vesicles, it became clear that FM4–64 is more specific in the type of lipid membrane that it binds to than commonly assumed. Specifically, 1,2-dioleoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (DOPG) vesicles were not labelled with FM4–64 (Esseling-Ozdoba et al., 2008; in preparation). This result, and the recent idea that endocytosis plays a role in cell plate formation (Dhonukshe et al., 2006), warranted us to investigate the specificity of FM4–64 as a marker of the endocytic pathway. Earlier, it was shown that microinjection FM4–64 into Lilium longiflorum pollen tubes gave a diffuse cytoplasmic staining and no peripheral staining of the plasma membrane or endocytic and vacuolar membranes that normally would be labelled if the dye were applied from the outside (Parton et al., 2001). If FM4–64 is only inserted into the outer leaflet of the plasma membrane and if this leaflet is somehow specific for FM4–64 uptake, then it should follow that applying the dye from the cytoplasmic side should not label the plasma membrane or any other internal membranes. Furthermore, if the dye is taken up in the inner leaflets of specific membrane compartments by endocytosis and further trafficking, then microinjecting FM4–64 directly into these internal compartments, such as the vacuole, should label its membrane.
Here, we show that FM4–64 does not label any membranes when injected into the cytoplasm of living plant cells. However, it does label the tonoplast when injected directly into the vacuoles of these cells. The labelling is not very strong as the fluorescence of the dye is pH-dependent. These results confirm the hypothesis that FM4–64 is incorporated specifically into the outer leaflet of the plasma membrane, and from there, is passed on to the inner leaflet of the equally specific membranes surrounding the endosomes, prevacuolar compartments and the tonoplast. We conclude that FM4–64 is indeed a reliable marker of the endocytic pathway in plant cells.
Materials and methods
For pulse labelling and microinjection of FM4–64, we used Nicotiana tabacum Bright Yellow-2 suspension culture cells (BY-2) and Tradescantia virginiana (T. virginiana) stamen hair cells. Tobacco BY-2 cell cultures were grown in 40 mL liquid medium (Murashige & Skoog inorganic salts, 3% sucrose, 200 mg/L KH2PO4, 100 mg/L myo-inositol, 1 mg/L thiamine, 0.2 mg/L 2,4-dichlorophenoxyacetic acid, pH 5.8) in constant darkness at 25°C and 100 rpm. Each week, they were diluted 27 times in 250 mL Erlenmeyer flasks. BY-2 protoplasts were prepared by mixing 10 mL of a 3-day-old culture with 50 mL of digestive solution (1% cellulase, 0.1% pectolyase, 0.45 M mannitol, pH 5.5) and incubating this for 80 min at ∼30°C and 50 rpm on a shaker in the dark. The protoplasts were then carefully pelleted by centrifugation at 1000 g and washed twice with 0.5 M mannitol. For cytoplasmic extracts, these protoplasts were mixed in 1:5 ratio with 30% Percoll in 0.6 M mannitol and centrifuged onto a 70% Percoll cushion. This method removes the large vacuole with its lytic enzymes from the cell. The collected mini-protoplasts were then washed in 0.6 M mannitol and in extract buffer [50 mM piperazine-N, N′-bis (2-ethanesulfonic acid) (PIPES), 5 mM ethylene glycol tetraacetic acid (EGTA), 2 mM magnesium chloride (MgCl2), 3 mM dithiothreitol (DTT), 1 protease inhibitor ethylene diamine tetraacetic acid (EDTA)-free Complete mini-tablet; Roche, Mannheim, Germany], homogenized and again centrifuged to remove the nuclei (Komoda et al., 2004; Murata et al., 2005). The supernatant contained up to 4 mg/mL protein and the majority of vesicles and organelles. T. virginiana plants were grown in climate chambers under 16 h of light at 23°C and 6 h of darkness at 18°C. Growing stamen hairs were excised from young flower buds of about 6 mm. The hairs were immobilized in low-gelling temperature agarose (BDH Chemicals, Poole, United Kingdom) and covered with liquid Tradescantia medium [5 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid (HEPES), 1 mM KCl, 1 mM MgCl2, 0.1 mM CaCl2, pH 7.0 with KOH].
FM4–64 (N-(3-triethylammoniumpropyl)-4- (6-(4-(diethylamino)phenyl)hexatrienyl) pyridinium dibromide; Invitrogen, Eugene, Oregon) was dissolved in DMSO at 2 mm and then diluted in the BY-2 medium, Tradescantia medium or injection buffer (5 mm HEPES, 100 mm KCl, pH 7.0). For pulse labelling, FM4–64 was diluted to 32 μM (Kutsuna & Hasezawa, 2002). For microinjection, the FM4–64 dye, with a final concentration of 5 μM, was mixed with 0.2 mg/mL FITC-dextran (4.4 kDa; Molecular Probes, Invitrogen). This allowed us to verify into which compartment of the cell we injected the dye. To assay the pH dependency of FM4–64, we either first diluted FM4–64 in citric acid-sodium phosphate buffers (pH 3–8 in 0.5 M mannitol) and then mixed these with protoplasts, or first mixed the dye with the protoplasts for 5 min and then added this to the various buffers. Fluorescence of the plasma membrane was assayed by calculating the mean peak fluorescence of nine 40 pixel-wide strips across the membrane of three confocal laser scanning microscopy (CLSM) images of the protoplasts using ImageJ (Wayne Rasband, National Institutes of Health, Bethesda, MD, USA). FM4–64 absorption spectra from 300 to 850 nm were measured with a Hewlett-Packard 8453 UV-visible diode array spectrophotometer with UV-Visible ChemStation Rev. A.02.05 software (Agilent, Amstelveen, the Netherlands) at 5 μM in citric acid-sodium phosphate buffers ranging from pH 3 to 8 in 0.5 M mannitol.
Microinjections were performed as described previously (Vos et al., 1999). In short, borosilicate needles were back-filled with 2 μL of dye cocktail in injection buffer and screwed into a needle holder, which was attached by tubing to a screw-type syringe (Gilmont, Barrington, Illinois). The needle holder was then placed into a Narishige (Tokyo, Japan) micro-manipulation system mounted on a Zeiss Axiovert microscope with differential interference contrast (DIC) optics and a Pascal CLSM system (Zeiss, Jena, Germany). The cells were injected with roughly equal amounts of dye cocktail in the cytoplasm and the vacuole and were imaged before, during and after microinjection with a 63× 1.4 NA DIC lens. FM4–64 and FITC-dextran were excited with the 543-nm line of the helium-neon laser and the 488-nm line of the argon laser, respectively. Fluorescence was recorded in a dual-scan mode as FM4–64 is excited with both laser lines (filter combination 488–543 HFT, 505–530 BP, 560 LP).
FM4–64 labelled the plasma membrane when applied from the outside of the plant cell (Figs 1 and 2). From there it was taken up into the membranes of endocytic vesicles and compartments, as shown before (Kutsuna & Hasezawa, 2002). Pulse labelling of tobacco BY-2 cells showed that when the dye was washed out after 10 min, the remaining FM4–64 had accumulated in the membrane that surrounds the vacuole, the tonoplast, by the next day (Fig. 1). The plasma membrane and other endocytic membranes were no longer labelled at that moment. When T. virginiana stamen hair cells were incubated with FM4–64, we saw that the uptake was much slower (Fig. 2). The cross walls of the hairs, in particular, were only clearly labelled after 15–30 min. As in tobacco BY-2 cells, after the pulse labelling of T. virginiana stamen hair cells, the tonoplast was labelled the next day. However, unlike with BY-2 cells, the cuticle, and possibly the cell wall of the stamen hair cells, remained labelled after 30 h. Probably, the waxes in the cuticle, which is essential for the stamen hair to prevent water loss when the flower opens (Nawrath, 2006), also strongly bind FM4–64 and slow down the labelling of the interior. In neither species were the endoplasmic reticulum and nuclear envelope labelled.
To assess if FM4–64 is specifically incorporated into the outer leaflet of the plasma membrane and, from there, is taken up into the inner leaflet of the endocytic compartments, we hypothesized that it would not label the cytoplasmic side of any membrane when injected into the cytoplasm, but that it would label the inner leaflet of the tonoplast when injected directly into the vacuole. When we injected FM4–64 together with FITC-dextran into the cytoplasm of tobacco BY-2 cells, we could not detect any specific labelling of the membranes (Fig. 3; top panel). The fluorescence of the FITC-dextran confirmed that we had indeed injected the dye in the cytoplasm of the cell. When we injected tobacco BY-2 cells directly into the vacuole, a dim but unmistakable labelling of the tonoplast was seen (Fig. 3; middle panel). We repeated the experiment with T. virginiana stamen hair cells and found the same results: injection of FM4–64 into the cytoplasm did not label any membranes, but injection into the vacuole clearly labelled the tonoplast (Fig. 4). As a control for the activity of the FM4–64 dye, we added, at the end of each experiment, the remaining injection probe to the culture medium surrounding the cells on the slide under the microscope and imaged the labelling of the plasma membrane (Figs 3 and 4; bottom panel).
As an alternative to the microinjection experiments, we tested if the membranes present in cytoplasmic extracts made from tobacco BY-2 suspension culture cells would be labelled with FM4–64. We diluted FM4–64 in extract buffer and mixed this with the cytoplasmic extract to a final dye concentration of 2 μM. None of the samples were labelled, except for a few undefined structures that are probably the outside-out plasma membrane vesicles (Fig. 5).
The tonoplast labelling of the stamen hair cells was brighter than the labelling of the tonoplast in BY-2 cells. We suspect that the acidity of the vacuole might change the fluorescence of FM4–64. We could not directly check the acidity of the vacuoles of T. virginiana stamen hair cells or tobacco BY-2 cells; we therefore investigated the influence of the pH on the FM4–64 fluorescence intensity in the plasma membrane of tobacco BY-2 protoplasts. We used the protoplasts instead of whole cells because the cell wall may have a buffering capacity (Felle, 2001). FM4–64 was added at a final concentration of 5 μM to citrate-phosphate buffers ranging from pH 3 to 8 for 5 min, after which the protoplasts were added (“buffer first” experiment) or FM4–64 was added to protoplasts first and the buffers (pH 3–8) after 5 min (“cells first” experiment). After imaging, the pixel intensities were measured and plotted against the pH of the buffer (Fig. 6a). The fluorescence of FM4–64 was clearly dependent on the acidity of the buffer. In acidic environments (pH 3–4), the dye was hardly fluorescent compared to in neutral or alkaline environments. Interestingly, the fluorescence was independent of the location of the FM4–64 at the time of buffer application as both experiments, “buffer first” and “cells first”, showed the same effect. We expect that the loss of fluorescence is due to protonation of FM4–64. We, therefore, measured the absorbance spectrum of the dye in buffers with pH values ranging from pH 3 to 8 and found a shift in the maximum absorbance peak from 497 nm to 413 nm, with an isosbestic point at 447 nm (Fig. 6b). When we added concentrated KOH to the dye in citrate-phosphate buffer with pH 3 and raised the pH to 6.5 or 11.0, the peak absorbance shifted back from 413 nm to 497 nm, suggesting that the protonation was reversible. The peak value did not reach the same values as shown before because of dilution of the sample (to about 80% and 74% dye concentration for pH 6.5 and pH 11.0, respectively).
We injected FM4–64 into the cytoplasm and vacuole of living plant cells and found that FM4–64 is specifically inserted into the outer leaflet of the plasma membrane and the inner leaflet of the tonoplast and does not label these membranes from their cytoplasmic sides (Figs 3 and 4). This confirms the hypothesis that the lipophilic domain of FM4–64 is inserted into membrane leaflets with a specific phospholipid composition (or containing specific glycolipids, cholesterol and/or (glyco-) protein domains). Only when applied from the outside does FM4–64 label the plasma membrane and, after endocytosis, the various target membranes that are, in origin, related to the plasma membrane outer leaflet (Bolte et al., 2004). It is likely that the phospholipid composition of the inner and outer leaflets of the plasma membrane, the vacuole and the endocytic membranes of the plant cell are very different in composition, just like in the animal cells (Ikeda et al., 2006). Alternatively, FM4–64 that was injected into the cytoplasm might be enzymatically inactivated, much like the de-esterification of the vital stain fluorescein-diacetate. So far, it is not known which phospholipids are the preferred targets for FM4–64 insertion. Interestingly, the cuticle in Tradescantia stamen hair cells was also intensely labelled with FM4–64. Possibly, FM4–64 also binds to waxes, higher molecular weight fats or fatty acid epoxides. If only small amounts of FM4–64 are used, the cuticle could absorb all the dye, which might erroneously suggest the lack of endocytosis. Future biochemical research should determine which membrane lipids FM4–64 interacts with.
Our conclusion is supported by the finding of Griffing (2008; this issue of the Journal of Microscopy), who showed, using the fluorescence resonance energy transfer (FRET) technique, that FM4–64 is not able to accept resonance energy from cytoplasmically expressed green fluorescent protein because the two fluorophores are physically separated. Only when the membrane was challenged with saponin, which is known to induce flip-flop of phospholipids and other membrane-bound components, was the energy transfer achieved.
Interestingly, although partly embedded in the phospholipid bilayer, FM4–64 fluorescence nevertheless depended on the acidity of the environment, probably due to protonation. The protoplasts that were labelled with FM4–64 and then incubated in buffers with pH values ranging from pH 3 to 8 lost their plasma membrane fluorescence with increasing acidity (Fig. 6a). Under these acidic conditions, the FM4–64 peak absorbance shifted from 497 nm to 413 nm in a reversible manner (Fig. 6b). This protonation may also explain the relatively lower fluorescence levels in the tonoplast as vacuoles are generally acidic (around pH 5.0–5.5). The absence of fluorescence when FM4–64 was injected into the cytoplasm is not likely caused by this pH effect as the pH of the cytoplasm is normally slightly alkaline (pH 7.2–7.5), although the pH values of 5.7–6.0 have been recorded for T. virginiana stamen hair cells (Pheasant & Hepler, 1987; Felle, 2001). Even under these slightly acidic conditions, the FM4–64 labelling of the plasma membrane of tobacco BY-2 protoplasts was almost as bright as under neutral or alkaline conditions (Fig. 6a).
The kinetics of FM4–64 uptake may differ among various plant materials (this work; Bolte et al., 2004), and therefore, some of the dye redistribution may take place via the secretory pathways from the pre-vacuolar compartment and Golgi within minutes after entering through the endocytic pathway. We conclude that FM4–64 is a specific marker of the endocytic pathway when used with some care concerning the used concentration, the acidity of the medium and the timing of imaging.
We would like to thank Henk Kieft for the assistance with the confocal microscope and Adrie Westphal for the assistance with the spectrometry. We would also like to thank Drs. A.M.C. Emons, A.A.M van Lammeren and S. Bolte for discussion and reading of the manuscript. This work was supported by a VENI grant to J.W.V. from the Netherlands Organisation for Scientific Research (NWO).