FM-dyes as experimental probes for dissecting vesicle trafficking in living plant cells


Dr Béatrice Satiat-Jeunemaitre. Tel.: +33 1 69 82 37 98; fax: +33 1 69 82 33 55; e-mail:


FM-dyes are widely used to study endocytosis, vesicle trafficking and organelle organization in living eukaryotic cells. The increasing use of FM-dyes in plant cells has provoked much debate with regard to their suitability as endocytosis markers, which organelles they stain and the precise pathways they follow through the vesicle trafficking network. A primary aim of this article is to assess critically the current status of this debate in plant cells. For this purpose, background information on the important characteristics of the FM-dyes, and of optimal dye concentrations, conditions of dye storage, and staining and imaging protocols, are provided. Particular emphasis is placed on using the FM-dyes in double labelling experiments to identity specific organelles. In this way, staining of the Golgi with FM4-64 has been demonstrated for the first time.

1. Introduction

The membrane-selective FM-dyes, FM4-64 and FM1-43, belong to a class of amphiphilic styryl dyes developed by Betz and co-workers (Betz et al., 1992, 1996). FM-dyes fluoresce significantly only when they are in a hydrophobic environment, e.g. a lipid-rich membrane, and were originally derived from DASPMI, a membrane-potential probe used to study mitochondria (Bereiter-Hahn, 1976). Besides having been extensively used in studies on animal cells (e.g. Cochilla et al., 1999), FM-dyes are being increasingly used for studying vesicle trafficking and organelle organization in fungal cells (Vida & Emr, 1995; Hoffmann & Mendgen, 1998; Fischer-Parton et al., 2000; Read & Hickey, 2001; Atkinson et al., 2002; Hickey et al., 2002) and plant cells (Carroll et al., 1998; Battey et al., 1999; Belanger & Quatrano, 2000a,b; Kubitscheck et al., 2000; Emans et al., 2002; Bolte et al., 2004). FM-dyes are commonly believed to be unable to cross membranes because of their amphiphilic nature, and being anchored in the outer leaflet of the bilayer. They are thought to enter the cell primarily by endocytic vesicles invaginated from the plasma membrane (Betz et al., 1992; Read & Hickey, 2001). These dyes have thus been widely used as endocytosis markers (Betz et al., 1992, 1996; Vida & Emr, 1995; Fischer-Parton et al., 2000; Ueda et al., 2001; Atkinson et al., 2002; Emans et al., 2002; Geldner et al., 2003; Grebe et al., 2003). After internalization, the dyes are distributed to different organelle membranes, probably primarily via the vesicle trafficking network and thus components of secretory pathways can become labelled (Betz et al., 1996; Belanger & Quatrano, 2000a; Fischer-Parton et al., 2000; Read & Hickey, 2001). Finally, because they are membrane-selective and stain many organelle membranes, the FM-dyes can be used as excellent vital markers to monitor organelle organization and dynamics (e.g. Hickey et al., 2002; Hickey & Read, 2003).

Despite their usefulness, the FM-dyes are stimulating much debate, particularly with regard to which organelles they stain, the precise pathways of intracellular transport they follow and how good they are as endocytosis markers in plants. Indeed, there is not yet a reliable picture of endocytosis in plant cells (Hawes et al., 1995; Battey et al., 1999), even though they have essentially the same molecular machinery for endocytosis as yeast or mammalian cells. However, clear evidence for the functioning of individual components of this molecular machinery in plant cells remains scant. Moreover, it is known that vesiculation of the plant plasma membrane can occur as a result of osmotic shock (i.e. a non-specific process), leading to some confusion about the exact nature of what constitutes endocytosis in plants (Hawes et al., 1995). The primary aim of this article is critically to assess the use of FM-dyes to analyse vesicle trafficking and organelle organization in living plant cells.

2. What staining and imaging protocols are best for FM-dyes?

2.1. Which dye to use?

The abbreviation ‘FM’ stands for the chemist’s name, Fei Mao, who developed the FM-dyes from the related probe, dimethylaminostyrylmethylpyridiniumiodine (DASPMI) (Bereiter-Hahn, 1976; Betz et al., 1992). FM1-43 and FM4-64 (Molecular Probes Inc., Eugene, OR, U.S.A.) are both membrane-selective fluorescent dyes (Fig. 1). They insert into one leaflet of a membrane lipid bilayer via their lipophilic tails (two aliphatic chains) with the pyridinium dicationic head anchored at the membrane surface. The amphiphilic nature of the dyes is believed to prevent them from freely crossing from one lipid leaflet of a membrane bilayer to the other. The only way this is thought to occur is by endocytosis, exocytosis or formation of other vesicles (Read & Hickey, 2001). FM4-64 and FM1-43 differ slightly in their chemical structure, and these structural differences can result in different patterns of membrane staining (Fischer-Parton et al., 2000; Hickey et al., 2002). For studies on plant and fungal cells, FM4-64 is usually preferred to FM1-43 due to its superior brightness, greater contrast, higher photostability, and its red-shifted emission facilitating concomitant use with the green fluorescent protein (GFP) (Fig. 2; see also section 2.4).

Figure 1.

Chemical formulae of FM1-43 and FM4-64.

Figure 2.

Emission spectra of FM dyes and other fluorescent molecules. From left to right: grey line: HDEL-mGFP4 spectrum (peak 505 nm) in ER of BY-2 cells. Solid bold line: FM1-43 spectrum (peak 550 nm) in BY-2 cells. Broken bold line: FM1-43 spectrum (peak 580 nm) in CHAPS. Solid bold line: FM4-64 spectrum (peak 670 nm) in BY-2 cells. Broken bold line: FM4-64 spectrum (peak 760 nm) in CHAPS. Grey line: chlorophyll (peak 685 nm) in Arabidopsis leaves.

2.2. Dye concentration and storage

FM-dyes may modify membrane fluidity even at micromolar concentrations, as has been shown for another endocytosis marker dye, TMA-DPH (Rodes et al., 1995). Furthermore, the toxicity of the dye is dependent upon cell density because this influences the dye–membrane lipid ratio. Therefore, both FM-dye and cell concentrations should be controlled. FM-dyes have a limited solubility in aqueous solutions and thus are usually dissolved in dimethylsulphoxide (DMSO), as recommended by the manufacturers. However, DMSO is an organic solvent that acts on lipid membranes and can cause lipid and polysaccharide extraction. It is thus essential to keep the DMSO concentration to a minimum (below 0.1%) and to perform appropriate controls, wherever possible, to test the potentially deleterious effects of this solvent. However, we have found that FM4-64 is sufficiently soluble in water to produce a 17 mm stock solution.

FM-dyes are typically kept as small aliquots of 16.4–20 mm stock solution frozen at −20 °C. During experiments, aliquots should be held on ice in darkness as far as possible because the dyes are unstable at room temperature and are light-sensitive. The dyes can be used over a wide range of final concentration, from 1 µm (Carroll et al., 1998) to 50 µm (Ueda et al., 2001) (see also Vida & Emr, 1995; Fischer-Parton et al., 2000; Kim et al., 2001; Emans et al., 2002). The optimal dye concentration to use will be the lowest for which (i) the fluorescence signal is adequate for imaging, and (ii) the cell organization or function (e.g. organelle morphology or cell growth rate) is not compromised. These parameters need to be empirically determined for each cell type studied.

For studies on BY-2 cells, we have performed flow cytometry of FM4-64-stained protoplasts (2 × 10−5/mL) from a 3-day-old tobacco BY-2 cell culture in order to assess the quantity of FM4-64 necessary to saturate the plasma membrane (data not shown). Between 5 and 15 µm, the fluorescence intensity per protoplast increased. Between 17 µm and 40 µm, the intensity reached a plateau, suggesting a saturation process (i.e. maximal staining of the plasma membrane). We therefore chose to stain cells with 17 µm FM4-64, a concentration at which membranes are well contrasted against diffuse staining within the cytoplasm (Fig. 3). No differences were detected, either in the kinetics or in the pattern of staining, in BY-2 cells when using 17 µm FM4-64 solutions made from a 17 mm stock solution in which the dye was originally dissolved in either DMSO or water.

Figure 3.

BY-2 cells. Scale bars = 5 µm. (a) Five minutes after adding stain: the plasma membrane is immediately stained, outlining the typical shape of the cell files of BY-2 cells. Inset: contiguous plasma membranes are stained but not the cell wall between two adjacent cells. (b) Ten minutes after adding stain: thickenings are often seen bulging from inside the plasma membrane. (c) Thirty to 60 min after adding stain: numerous submicrometre-sized fluorescent organelles are seen moving along cytoplasmic strands throughout the cells. (d) Sixty minutes after adding stain: vacuolar membranes retain dye fluorescence (arrows).

Figs 3–6. Optical sections of BY-2 cell lines stained with FM4-64: wild-type (Figs 3 and 4), expressing HDEL-mGFP4 (Fig. 5) or ST-GFP (Fig. 6). The images were colour coded green (for GFP) and red (for FM4-64) giving yellow co-localization in merged images. The oil-immersion objectives used were 40× (NA 1.25), providing a resolution of 160 nm in the xy-plane and 330 nm along the z-axis, and 63× (NA 1.32), providing a resolution of 150 nm in the xy-plane and 290 nm in the z-axis (pinhole 1 Airy unit). Cells were loaded with agitation by adding 17 µm FM4-64 from a 17 mm stock solution in water to their culture medium.

It has been claimed that FM4-64 may disrupt the subcellular organization in fungal cells (Torralba & Heath, 2002). However, this now appears erroneous, and appears to be due to the effect of combining FM4-64 with the vacuolar selective dye DFFDA during cell staining (Read & Kalkman, 2003). In plant cells, no effects of dye cytotoxicity have been reported so far. FM4-64 at 32 µm did not inhibit cell division or cell cycle progression in BY-2 cells (Kutsuna & Hasezawa, 2002).

2.3. Dye loading

Dye loading is achieved by immersing and often agitating the biological material in the dye-containing medium. As stated before, FM-dyes only fluoresce significantly when they are in a hydrophobic environment (e.g. lipid-rich membranes). The fact that the fluorescence quantum yield of the FM-dyes is far higher in a lipid environment than in water provides good image contrast between the cell and surrounding medium. Therefore, it is not necessary to wash the dyes out from the medium bathing a stained cell. This reduces potential problems which can arise from mechanically perturbing the cells under study. However, when attempts are made to quantify the internalization of FM4-64, or to perform pulse-chase experiments, then washing out free dye may be necessary (Kim et al., 2001; Ueda et al., 2001; Kutsuna & Hasezawa, 2002). Staining of the plasma membrane can be substantially reduced by extended washing, leaving the intracellular compartments visible. FM4-64 is reasonably photostable, and 50 frames in a time-course of cells pulsed-labelled with FM4-64, and imaged by confocal microscopy, typically results in only ∼10% photobleaching. Variations in physical parameters (e.g. room temperatures, aeration, or illumination) from one experiment, or time-point, to another may generate differences in the kinetics of dye internalization and distribution within a cell. Sampling stained cells from cell populations at regular time intervals is much easier than collecting image time courses of individual cells, and this approach has been used in this study.

2.4. Imaging the dyes

Epifluorescence microscopes equipped with the appropriate filters (e.g. fluorescein long-pass emission set for FM1-43; fluorescein or rhodamine long-pass emission set for FM4-64), and ideally coupled to a cooled CCD camera, can be used to observe and analyse the dye staining in cell suspensions or plant tissues. However, imaging the dynamics of dye localization within a cell has highly benefited from confocal laser scanning microscopy (CLSM). However, high-speed image deconvolution using a CCD camera may also be an excellent alternative to CLSM.

If possible, the spectral properties of the dyes should be checked to optimize the imaging protocols. In particular, the emission spectra of FM-dyes (due to their aromatic nuclei, Fig. 1) may differ depending on their precise membrane microenvironment (Betz et al., 1992). Spectroscopy on pure solutions can thus be misleading for studies on living cells. In methanol, FM1-43 has an excitation peak at 510 nm with an emission peak at 626 nm (red), whereas in frog and murine preparations the excitation peak is ∼470 nm and the emission 610 nm (see Typically, the emission of FM1-43 is blue-shifted in BY-2 cells relative to that in the zwitterionic detergent, CHAPS, with the excitation peak in BY-2 cells being ∼550 nm (Fig. 2). Furthermore, the fluorescence properties of FM1-43 may change according to its precise membrane environment. For instance, in the presence of purothionins (basic proteins containing domains of cysteine-rich repeated motifs that affect membrane permeability), the fluorescence intensity of the axonal staining with FM1-43 was markedly increased (Mattei et al., 1998). This variation appears to be linked to a shift in the emission spectrum of FM1-43 (E. Benoit et al., personal communication). In plant cells, such variations have not been reported so far. For confocal imaging, FM1-43 is typically excited at 488 nm laser excitation and emission detected at ∼590 nm (Carroll et al., 1998; Procissi et al., 2003).

Likewise, the spectral properties of FM4-64 may vary according to the cells under study or their molecular environment (Fig. 2). For instance, emission peaks of 760 nm and 640 nm have been reported for FM4-64 bound to 2% CHAPS micelles and for yeast membranes, respectively (see and We found the peak emission for FM4-64 in BY-2 cells to be 670 nm (Fig. 2), emphasizing the importance of checking the spectral properties of the dye in stained cells. However, we have not found differences in FM4-64 emission spectra between different membranes in BY-2 cells, even if damaged (data not shown). We routinely image FM4-64 fluorescence between 625 and 665 nm.

The 488-, 514- and 532-nm laser lines of a CLSM can be used to excite either dye because of their broad excitation spectra. FM4-64 (excitation maxima 515 nm) is excited and emits at longer wavelengths than FM1-43. We commonly excite FM1-43 and FM4-64 using the 488-nm laser line.

The dyes’ fluorescence may also be imaged in combination with other fluorophores (Fig. 2). In tissue expressing GFP, FM4-64 is preferred for double labelling because its emission in biological membranes is red, which is well separated from the green emission fluorescence of GFP. In chlorophyll-containing tissue or cells, FM1-43 is preferred because it does not fluoresce in the red part of the spectrum. Rhodamine 123 can also be imaged simultaneously with FM4-64 without ‘bleed through’ of signal between the two channels (488-nm excitation; Rhodamine 123 fluorescence detected at 515–545 nm; FM4-64 fluorescence detected at > 640 nm). For the dual labelling with FM4-64 and GFP presented in this paper (Figs 3–7), the two fluorochromes were excited with an air-cooled argon laser at 488 nm and fluorescence detected at 510–540 nm and 635–680 nm, respectively, using a TCS SP2 CLSM (Leica Microsystems, Mannheim, Germany).

Figure 4.

Cell plate staining by FM4-64 in BY-2 cells. Scale bars = 5 µm. (a) Five to 10 min after adding stain: forming cell plates are labelled by FM4-64 apparently as soon as they are in contact with the lateral plasma membrane (arrow). (b) Thirty to 60 min after adding stain: forming cell plates are labelled even when not in contact with peripheral plasma membrane. (c) Differential interference contrast image of the same cell imaged in (b). (d) Thirty to 60 min after adding stain in dividing cells: fluorescent structures are gathered around the labelled cell plate.

Figs 3–6. Optical sections of BY-2 cell lines stained with FM4-64: wild-type (Figs 3 and 4), expressing HDEL-mGFP4 (Fig. 5) or ST-GFP (Fig. 6). The images were colour coded green (for GFP) and red (for FM4-64) giving yellow co-localization in merged images. The oil-immersion objectives used were 40× (NA 1.25), providing a resolution of 160 nm in the xy-plane and 330 nm along the z-axis, and 63× (NA 1.32), providing a resolution of 150 nm in the xy-plane and 290 nm in the z-axis (pinhole 1 Airy unit). Cells were loaded with agitation by adding 17 µm FM4-64 from a 17 mm stock solution in water to their culture medium.

Figure 5.

Merged pictures of BY-2 cells expressing a GFP-coupled ER marker (HDEL-GFP) and loaded with FM4-64. Scale bars = 16 µm. (a) Thirty minutes after adding stain: FM4-64-labelled organelles (red) are within cytoplasmic strands, which are distinct from the ER membranes (green). Direct observations revealed separate movement of these two structures. The yellow at the periphery of the cells arises from the close association of the cortical ER labelled with GFP and the plasma membrane stained with FM4-64. (b) Thirty minutes after adding stain: a dividing cell in which green colour-coded ER is distinct from the red colour membranes stained with FM4-64. (c) Thirty minutes after adding stain: a dividing cell at a later stage to that in (b), the two markers are closely associated in the cell plate level, but not in the periphery of the cell. (d) Twenty hours after adding stain: FM4-64 fluorescence has mostly shifted to the vacuolar compartment, and the cytoplasm exhibits typical ER-GFP staining.

Figs 3–6. Optical sections of BY-2 cell lines stained with FM4-64: wild-type (Figs 3 and 4), expressing HDEL-mGFP4 (Fig. 5) or ST-GFP (Fig. 6). The images were colour coded green (for GFP) and red (for FM4-64) giving yellow co-localization in merged images. The oil-immersion objectives used were 40× (NA 1.25), providing a resolution of 160 nm in the xy-plane and 330 nm along the z-axis, and 63× (NA 1.32), providing a resolution of 150 nm in the xy-plane and 290 nm in the z-axis (pinhole 1 Airy unit). Cells were loaded with agitation by adding 17 µm FM4-64 from a 17 mm stock solution in water to their culture medium.

Figure 6.

BY-2 cells expressing a GFP-coupled Golgi marker (ST-GFP, coded green, b,e,h) and loaded with FM4-64 (red, a,d,g) under constant agitation. Merged images are shown in (c), (f) and (i). Scale bar = 8 µm. (a–c) Five to 10 min after adding stain: the plasma membrane is clearly labelled with FM4-64 (a) surrounding the cytoplasm containing mobile Golgi bodies running throughout the cell (b). No co-localization is observed between the plasma membrane and Golgi (c). (d–e) Ten to 30 min after adding stain. FM4-64-labelled membranes have been internalized (d), and are in the same cytoplasmic strands as the Golgi (e). However, the merged image shows that the two sets of structures become only partially associated (f, see arrows showing co-localization). (g–i) Thirty to 60 min after staining: most of the Golgi are labelled with FM4-64 (yellow structures in (i)).

Figs 3–6. Optical sections of BY-2 cell lines stained with FM4-64: wild-type (Figs 3 and 4), expressing HDEL-mGFP4 (Fig. 5) or ST-GFP (Fig. 6). The images were colour coded green (for GFP) and red (for FM4-64) giving yellow co-localization in merged images. The oil-immersion objectives used were 40× (NA 1.25), providing a resolution of 160 nm in the xy-plane and 330 nm along the z-axis, and 63× (NA 1.32), providing a resolution of 150 nm in the xy-plane and 290 nm in the z-axis (pinhole 1 Airy unit). Cells were loaded with agitation by adding 17 µm FM4-64 from a 17 mm stock solution in water to their culture medium.

Figure 7.

Merged images of optical sections of protoplasts from a BY-2 cell line expressing a GFP-targeted Golgi marker (ST-GFP). Dye loading and cell imaging as in Figs 2–6. Scale bars = 8 µm. (a) Ten minutes after staining: Golgi (green) are clearly distinct from FM4-64-labelled organelles (red). (b) Thirty minutes after staining: merged image; yellow structures resulting from co-localization of FM4-64 (red) and GFP (green) suggest that some Golgi bodies have received FM4-64-labelled membranes. (c) Sixty minutes after after staining: most of the Golgi bodies (green) are labelled with the FM-dye (red) and thus appear yellow in this merged image; some organelles only labelled with the FM-dye are still visible. (d) Sixty minutes after staining: note strong labelling of the vacuolar membranes with FM4-64 (arrow).

3. Staining patterns of FM4-64 and FM1-43 are time dependent

3.1. Kinetic studies of FM-dye internalization

In plant cells as in fungal or other eukaryotic cells, the successive staining of the plasma membrane followed by different organelles is time-dependent (Read & Hickey, 2001; Ueda et al., 2001; Emans et al., 2002; Bolte et al., 2004).

In fungal cells, FM-dyes rapidly labelled the plasma membrane and have been used to measure the kinetics of membrane internalization (Atkinson et al., 2002). In order to stain fully the majority of membrane-bounded organelles in a cell, it is usually necessary to immerse cells for 1 h or more in dye. In fungal hyphae, FM1-43 and FM4-64 dyes are taken up by both apical and subapical hyphal compartments. Interestingly, differences were observed in the patterns of organelle staining obtained with each dye (Fischer-Parton et al., 2000), suggesting that they do not necessarily follow the same trafficking pathways (see section 5.4).

In higher plant cells, kinetic studies of FM1-43 (Carroll et al., 1998; Emans et al., 2002) and FM4-64 (Ueda et al., 2001; Kutsuna & Hasezawa, 2002; Bolte et al., 2004; Figs 3–7) broadly present similar staining patterns in roots (Geldner et al., 2003), protoplasts (Ueda et al., 2001; Fig. 7) or whole cells (Kutsuna & Hasezawa, 2002; Figs 3–6). In each case, staining of the plasma membrane is immediate (Fig. 3a). Moreover, the cell wall does not act as a barrier and does not slow the staining process, as kinetic studies gave similar results between whole cells and freshly made protoplasts from the same cell line (data not shown). Osmotic shock provoking the retraction of the plasma membrane in plasmolysed cells clearly showed that, for FM4-64, the dye was not retained in cell walls. However, for FM1-43, a light staining of the cell wall was observed (data not shown). After 10 min in dye-containing medium, localized fluorescent thickenings occurred on the internal side of the plasma membrane (Fig. 3b), and mobile fluorescent organelles began to be seen at the periphery of the cytoplasm. With increasing time (20–60 min), fluorescent structures were observed throughout the whole cytoplasm (Fig. 3c). In Arabidopsis protoplasts, ring-like structures sometimes emerged following the appearance of punctuate organelles (Ueda et al., 2001). A light diffuse staining of the cytoplasm was often detected as well, which may be due to hydrophobic protein adsorption. After 1 h, numerous organelle membranes were stained, and vacuolar membranes began to be labelled as well (Fig. 3d). Between 3 and 10 h (depending on the experiment), all vacuolar membranes were stained. Ten to 30 h after dye loading, the vacuolar membranes remained stained, and dye fluorescence was even sometimes observed in the lumen of vacuoles (Kim et al., 2001; Kutsuna & Hasezawa, 2002; see also Fig. 5d).

FM4-64 staining in plant cells also highlights some specific plant features, such as the nascent cell plates in a dividing cell, between the two daughter cells. Forming cell plates were instantaneously stained but appearing only if their membranes were in contact with at least one side of the mother cell (Fig. 4a), suggesting that labelling of the cell plate is via its connection with the peripheral plasma membrane. This observation is consistent with the dye being able to diffuse laterally. Forming cell plates not yet anchored on mother cell membranes did not exhibit this immediate strong fluorescence. However, with longer incubation (30–60 min), the forming cell plate became strongly stained (Fig. 4b,c), even if there was no connection with the mother plasma membrane (as shown by collecting a series of optical sections at different optical planes down the z-axis). This forming plate was often surrounded by numerous micrometre- and submicrometre-sized fluorescent structures (Fig. 4d), which could be vesicles or larger organelles similar to those described in interphase cells after this time of staining, and suggests that organelle membranes pass to or come from the forming cell plate.

The fluorescence intensity of the stained plasma membrane may be far higher than that of organelle membranes. By reducing the gain of the photomultiplier detector of the CLSM, this feature can be used to image just the plasma membrane in living cells of, for example, the shoot apical meristem, which can be followed over several days (Grandjean et al., 2004). The instantaneous and strong labelling of the plasma membrane has also been used for morphometric studies involving root hair measurements (Procissi et al., 2003).

3.2. Do the kinetics of FM-dye internalization vary?

We have already noted that variations in experimental conditions or in dye concentration from one experiment to another may generate differences in the kinetics of internalization (see sections 1.2 and 1.3). Besides the influence of these physico-chemical parameters, it is clear that the kinetics of internalization also vary with physiological conditions, both in fungi and in plant cells. In fungal cells, fast growing hyphae of Neurospora internalize FM4-64 rapidly and dye can be detected in the cytoplasm 10 s after being added (Read & Hickey, 2001). Slow growing germ tubes of Neurospora internalize FM4-64 far more slowly (G. Wright & N. D. Read unpublished observations). Spores of Magnaporthe internalize FM4-64 2–3 min after being hydrated with the dye (Atkinson et al., 2002).

Similarly, in plant cells, Carroll et al. (1998) showed that actively secreting protoplasts from maize root cap cells have a much lower rate of FM1-43 internalization compared with that in nonsecreting protoplasts. Parton et al. (2001) also reported a quantitative relationship between FM4-64 staining and growth rate within an individual pollen tube. In our laboratory, BY-2 cells from a 3-day-old culture internalized FM4-64 at a slower rate than 2-day-old cultured cells. Interestingly, the observation was the same for protoplasts from the latter cultures, suggesting that the cell wall and shape had no impact on membrane internalization. Transgenic BY-2 cell lines like BY-2/ST-GFP (Saint-Jore et al., 2002; Brandizzi et al., 2004) also exhibited different rates of FM-dye internalization compared with wild-type cells, suggesting that these different cell lines internalize membranes at different rates. Last but not least, apparently conflicting reports exist concerning drug effects upon FM-dye internalization. Brefeldin A (BFA), when applied to maize root cap protoplasts, reduced the rate of internalization of FM1-43 (Carroll et al., 1998). In BY-2 cells, BFA stimulated the overall uptake of dye but blocked further transport to targeted organelles, creating an accumulation of large vesicles labelled with FM1-43 (Emans et al., 2002). BFA may also induce larger fluorescent patches of FM4-64 to form in other cell types (Geldner et al., 2003). However, artefactual effects of BFA on FM4-64 internalization should not be discarded, as both molecules may alter the physical environment of the plasma membrane. They might even compete with each other, modifying the kinetics of membrane exchange within cells.

4. Which organelles are stained by FM-dyes?

In plant and fungal cells, as previously indicated, after the addition of FM4-64 the plasma membrane immediately becomes strongly stained followed by staining of a variety of small organelles throughout the cytoplasm, and finally by staining of vacuolar membranes. This pathway of dye distribution closely resembles the endocytic pathway mapped in earlier electron microscopy studies (Tanchak et al., 1984). Some of the stained organelles could be endocytic intermediates, in a vesicle trafficking pathway between plasma membrane and lytic vacuoles, as described in yeast cells (Vida & Emr, 1995). However, the question arises as to precisely which organelles are stained with FM4-64, and what exactly are ‘endocytic intermediates’ in plant cells. One way of examining this is to perform double-labelling experiments in which transgenic cells expressing GFP targeted to different organelles are also stained with FM4-64.

In the following sections, studies intended to identify the FM4-64-stained structures are presented together with a discussion on FM-dyes as endosomal markers.

4.1. FM4-64 and endoplasmic reticulum markers

When stained, either by using fluorescently tagged antibodies or targeted with a fluorescent protein, the plant endoplasmic reticulum (ER) appears as a fluorescent network within the cytoplasmic strands and in the periphery of the plant cell (Satiat-Jeunemaitre et al., 1999). Figure 5 shows an optical section of a BY-2 cell line expressing an ER marker (HDEL-GFP) and loaded with FM4-64. Merged images of the dual labelled cells 30 min after dye loading clearly show no co-localization of the two fluorescent signals in the ER (Fig. 5a). Red fluorescent organelles are observed moving within the cytoplasmic strands, which are filled with green ER membranes. After 20 h, the fluorescence of FM4-64 has shifted mostly to the tonoplast, or is even localized in the vacuolar lumen, as previously indicated for long-term incubations in dye (see section 3.1; Fig. 5d). In dividing cells (Figs 4c and 5b), the association between ER and FM4-64-labelled organelles appears more complex. In the cell periphery, there is clearly no localization between ER and FM4-64-labelled membranes. At the cell plate, an association is sometimes seen at the periphery of the forming cell plate (Fig. 5b), or even in the whole cell plate (Fig. 5c). This apparent co-localization is probably only due to the limits of resolution achievable with the CLSM (see Bolte et al., 2004, for a discussion). It is well known from electron microscopy studies that ER membranes and Golgi-derived vesicles are closely associated with each other at the growing margins of the forming cell plate (Samuels et al., 1995).

These observations confirm most previous studies in which the ER and nuclear envelope have been reported not to be stained by FM4-64. One exception is a study on Fucus cells, which suggested possible staining of ER membranes by the dye (Belanger & Quatrano, 2000b). Whether this differential staining reflects an inability for FM4-64 to be anchored within ER membranes or a lack of FM4-64 transport between Golgi and ER is not known.

4.2. FM4-64 and Golgi markers

After labelling with fluorescently tagged antibodies or a fluorescent protein, plant Golgi stacks appeared as hundreds of fluorescent micrometre-sized structures throughout the cytoplasm (Satiat-Jeunemaitre et al., 1999; Fig. 6b). These have been shown by electron microscopy to behave as acceptor membranes for endocytic vesicles (Tanchak et al., 1984; Hübner et al., 1985; Hillmer et al., 1986; Hawes & Satiat-Jeunemaitre, 1996). However, Golgi membranes have not been previously shown to be stained with FM-dyes. To analyse whether this occurs, we followed the staining with FM4-64 of a BY-2 cell line expressing a GFP-tagged Golgi protein (construct: sialyl-transferase-GFP, ST/GFP, see Saint-Jore et al., 2002; Brandizzi et al., 2004) in whole cells (Fig. 6) and protoplasts (Fig. 7). In ST-GFP cells, Golgi appeared as numerous green fluorescent spots dispersed throughout the cytoplasm (Fig. 6b). After 10 min incubation in FM4-64, organelles were stained that were mostly distinct from Golgi (Figs 6c and 7a). However, after 15 min, FM4-64 began to co-localize with a few Golgi stacks (Figs 6d–f and 7b). After 1 h, most of the Golgi stacks were labelled with FM4-64 (Figs 6g–i and 7c,d).

4.3. FM4-64 and ‘prevacuolar compartment’

Prevacuolar compartments (PVCs) are defined as membranous compartments in which trafficking molecules pass through from the Golgi en route to lytic vacuoles. BP80, AtPep12p and m-Rabmc are three proteins that have been co-localized in the PVCs, and traffic between the Golgi apparatus and the lytic vacuole (Paris et al., 1997; Bassham et al., 2000; Bolte et al., 2004). FM4-64 staining of cells expressing a CFP-tagged m-Rab showed that the prevacuolar compartments became labelled with FM4-64 (Bolte et al., 2004).

4.4. FM4-64 and vacuoles

The staining of the tonoplast by FM4-64 is well established and was identified in this study by the easily recognizable morphological features of the plant vacuole (Figs 3d, 5c,d and 7d). The recent paper of Kutsuna et al. (2003) confirmed the progressive FM4-64 staining of the vacuolar membrane by using a transgenic BY-2 cell line expressing GFP-AtVam3p fusion protein targeted to it.

4.5. FM4-64 and ‘endosomes’

At present, there are no markers that are unequivocally specific for plant putative endocytic intermediates, and the identity and distinctiveness of ‘endosomes’ in plant cells remain to be defined. Potential candidate endosomal markers are the GNOM protein and some Rab5 homologues.

GNOM protein, an ARF GDP/GTP exchange factor, acts as a regulator of intracellular vesicle trafficking (Geldner et al., 2003). After 30 min of dye loading, FM4-64 is closely associated with and partially co-localizes with GNOM-positive membranes in Arabidopsis root cells (Geldner et al., 2003). Double labelling experiments of GNOM with various markers for secretory compartments have shown that GNOM-positive membranes were different from membranes of the ER network, Golgi or trans-Golgi network. Geldner et al. (2003) therefore proposed that GNOM-positive membranes were not located in the secretory pathway, and suggested that they could be components of ‘endosomal’ compartments. However, the organization of plant endosomes, and of the potential recycling routes to and from endosomes, are essentially unknown in plant cells. Moreover, the partial co-localization of GNOM and FM4-64 again raises the question as to which cell compartments containing GNOM-positive membranes FM4-64 actually stains. Geldner et al. (2003) suggested that they could represent a subset of ‘endocytic intermediates’. Based on our observations from BY-2 cells, they could be Golgi stacks (see section 4.2; Fig. 6g–i). However, it is still possible that root cells and BY-2 cells act differently with regard to the internalization and intracellular distribution of FM4-64. The structures stained in Arabidopsis root cells may even be another, as yet unidentified, organelle.

Rab proteins are another family of proteins that regulate intracellular vesicle trafficking. Rab5 proteins are small GTPases known to be localized on endosomes and are involved in endocytic processes in mammalian cells and yeast cells (see References in Bolte et al., 2000). GFP-Ara7, the Arabidopsis homologue of the conventional Rab5 (Ueda et al., 2001), has been shown to label punctuate motile structures which also stain with FM4-64 (Ueda et al., 2001; Brandizzi et al., 2004). By extension with what was known in mammalian and yeast cells, Ueda and co-workers concluded that Ara7 locates to ‘early/late endosomes’ and regulates membrane fusion in the early endocytic pathway. However, recent data on Rab5 homologues using GFP and Rab-deficient mutants highlighted the functional diversity among plant homologues of mammalian Rab5 GTPases (Ueda et al., 2001; Sohn et al., 2003; Bolte et al., 2004). Furthermore, Ara7 (a putative endosomal marker) co-localizes with mRab-YFP (a PVC marker) in Arabidopsis protoplasts and tobacco epidermal cells (S. Bolte et al., unpublished data). It may well be that the PVC stained by BP80/AtPep12p/m-Rabmc and FM4-64 is the same compartment as the ‘endosomal’ compartment stained by Ara7, GNOM and FM4-64 (Bolte et al., 2004; S. Bolte et al., unpublished data). It remains an open question whether a compartment labelled by Rab5 homologues or FM4-64 can be defined as ‘endosomal’ or ‘prevacuolar’, whether Rab5 homologues act between secretory compartments (the ER, Golgi or trans-Golgi network), or in a putative endocytic pathway.

These observations, summarized in Fig. 8, lead us to a discussion on what FM-dyes have told us about vesicle trafficking in plant cells.

Figure 8.

Model of possible pathways of membrane staining by FM4-64 in plant cells. FM4-64 immediately stains the plasma membrane by becoming inserted and anchored in the outer leaflet of the plasma membrane lipid bilayer. After internalization by endocytosis, the dye becomes localized to the inner leaflet of endocytic vesicles and all other organelles, which FM4-64 subsequent stains. After 30 min of staining, several pathways of FM4-64 transport are observed: from the plasma membrane to intracellular organelles (1), between intracellular organelles (2), or from the Golgi apparatus (GA) to the cell exterior (exocytosis) or vacuole (3). FM4-64 stains putative endocytic intermediates (END), prevacuolar compartments (PVC), Golgi apparatus (GA) and the vacuolar membrane (V: vacuole). It does not stain the endoplasmic reticulum (ER) and the nuclear envelope. Staining of putative END results in the direct transport of endocytic vesicles from the plasma membrane to these organelles (1). Labelling of PVC, GA or vacuole membranes may result from direct trafficking routes between END and these compartments (2). Vesicles derived from the GA become stained and are either exocytotic (and destined for the plasma membrane or cell plate) (3) or they enter the vacuolar pathway, labelling the PVC on their way to the lytic vacuole (V) (3).

5. What have FM dyes told us about vesicle trafficking in plant cells?

5.1. FM4-64 internalization is an active process

FM1-43 and FM4-64 are not taken up into intact living cells by unfacilitated diffusion. Cold treatment and metabolic inhibitors such as sodium azide strongly inhibit FM-dye internalization in fungal cells (Vida & Emr, 1995; Hoffmann & Mendgen, 1998; Fischer-Parton et al., 2000; Atkinson et al., 2002), as in plant cells (data not shown). As with animal or fungal cells, plant cells become stained in a time-dependent pattern by these dyes, which is consistent with their internalization being primarily by endocytosis, i.e. an active process driven by ATP hydrolysis. In budding yeast, mutants defective in endocytosis are also defective in the transport of FM4-64 (Vida & Emr, 1995; Gaynor et al., 1998).

5.2. Are other studies on plant cells consistent with FM4-64 being an endocytosis marker?

It is widely believed that endocytosis is the primary mechanism by which the FM-dyes are internalized by cells. The pathways followed by FM4-64 may mirror internalization processes, which function to recycle plasma membrane by targeting the internalized membrane in vesicles to the Golgi apparatus (GA), or recycling it back to the plasma membrane. Attempts to study endocytic processes in plant cells have been summarized in several reviews (e.g. Hawes et al., 1995, 1996; Battey et al., 1999). The unequivocal study to visualize endocytosis in plant cells comes from early electron microscopy studies in which the internalization of various electron-opaque tracers via clathrin-coated vesicles (putatively by adsorptive-mediated endocytosis) was performed on protoplasts (Tanchak et al., 1984; Hübner et al., 1985; Hillmer et al., 1986). The mapped endocytic pathway(s) showed the movement of tracers from the plasma membrane to the Golgi apparatus and the vacuole. Along the endocytic route, several endocytic intermediate compartments were described (Tanchak et al., 1984; Hübner et al., 1985; Hillmer et al., 1986; Hawes et al., 1996). However, no physiological role for this clathrin-coated mediated endocytic pathway has yet been identified. By contrast, studies to demonstrate unequivocally fluid-phase endocytosis in plant cells at the light microscope level have not succeeded. Supposed membrane-impermeant fluorescent probes such as Lucifer-Yellow or FITC-Dextran, which have been routinely used in endocytosis studies in mammalian and yeast cells, appear to be poor markers for studying endocytosis in plant cells and filamentous fungi because they can be transported across membranes (Oparka et al., 1990, 1991; Cole et al., 1991, 1997). A strategy based on the biotinylation and imaging of cell surface proteins has been devised to monitor putative protein recycling at the plasma membrane in plant protoplasts (Crooks et al., 1999). As a result, Crooks et al. were able to detect the internalization and concentration of biotinylated proteins within small cytoplasmic compartments. Whether FM4-64 or FM1-43 follow similar pathways will require these dyes to be localized at the electron microscopy level. These techniques have been developed for mammalian cells (Henkel et al., 1996), but have yet to be applied to plant cells.

5.3. FM4-64 is recycled back to the plasma membrane via secretory pathway(s)

From the data described in this paper and from the discussion above, it is clear that FM4-64-labelled organelles may be components of both endocytic and secretory pathways: endocytic vesicles, endocytic intermediates en route to the Golgi or vacuole; and exocytotic vesicles derived from the Golgi and directed to the cell surface, cell plate or prevacuolar compartments. Indeed, after 15–30 min incubation in dye we often observe the passage of dye from an endocytic to a secretory pathway, as reported in other studies on plants and fungi (Belanger & Quatrano, 2000a,b; Fischer-Parton et al., 2000; Parton et al., 2001; Read & Hickey, 2001; Camacho & Malhó, 2003). In filamentous fungi, time lapse, live-cell imaging of hyphae stained with FM1-43 or FM4-64 showed a labelling of the Spitzenkörper, a multicomponent structure predominantly comprising secretory vesicles and which is located within growing hyphal tips (Fischer-Parton et al., 2000; Read & Hickey, 2001). Similarly, apical secretory vesicles of pollen tubes become stained (Parton et al., 2001). In addition, staining of the nascent cell plate with FM4-64 indicates that the FM-dye has entered a secretory pathway. However, the cell plate is also known to be rich in clathrin-coated vesicles involved in endocytosis, and these are probably involved in recycling events between the plasma membrane and endosomal or Golgi compartments (Samuels et al., 1995; Couchy et al., 2003). Thus, as in non-dividing interphase cells, in the cell plate of dividing cells there is probably a dynamic interplay of membrane exchange between the new plasma membrane and Golgi, reflecting endocytosis and secretion (Couchy et al., 2003). Studies involving FM4-64 staining highlight the regulation of the balance between endocytic and exocytic processes in plant cells. A high level of secretory activity in plants could lead to an excess of membrane at the cell surface unless they continually expand or can retrieve this membrane. Excess membrane could be retrieved by a membrane internalization, and this may be a major function of endocytosis in plants. FM-dye studies support this hypothesis, which has also been suggested previously for plants and fungi (Samuels & Bisalputra, 1990; Derksen et al., 1995; Carroll et al., 1998; Read & Hickey, 2001).

5.4. FM4-64 and FM1-43 are differently distributed to different organelles

A comparison of FM4-64 and FM1-43 staining of fungal hyphae led to distinctly different localization of these dyes (see figure 1 in Fischer-Parton et al., 2000), which suggested the possibility of different mechanisms for sorting these dyes within the vesicle trafficking network. Essential questions are to understand why FM dyes do not stain all organelle membranes, and how FM1-43 and FM4-64 stain different organelles to different extents. As FM4-64 and FM1-43 are fluorescent lipid analogues, this differential distribution of dye to different organelle membranes may arise from a lipid sorting mechanism (Read & Hickey, 2001). Although lipid sorting has been much studied in animal cells (e.g. Mukherjee et al., 1999), virtually nothing is known about it in plant cells.

5.5. Other pathways of FM-dye uptake and distribution between organelles may exist

As discussed in section 5.2, supposedly membrane-impermeant dyes such as Lucifer-Yellow and FITC-dextrans, much used to demonstrate fluid-phase endocytosis in mammalian and yeast cells, can enter plant cells by being directly transported across membranes (Oparka et al., 1990, 1991; Cole et al., 1991). Another fluid-phase endocytosis marker, TMA-DPH, which is used in animal cells (Illinger & Kuhry, 1994), is not internalized by plant protoplasts (Tagu et al., 1987; Gantet et al., 1990). In relation to FM-dyes, alternative mechanisms to endocytic internalization and vesicle trafficking may also explain how the dyes are taken up by cells and distributed among different organelles across the cytoplasm. Fischer-Parton et al. (2000) discussed the possible action of flippases, which might ‘flip’ the amphiphilic FM-dye molecules, as lipid analogues, across the plasma membrane, and then how lipid transfer proteins might distribute these molecules to different organelles. These alternative pathways of dye uptake and distribution have yet to be tested experimentally. An analysis of the kinetics and pattern of dye staining in mutants compromised in the activity of flippases and/or lipid transfer proteins would be useful (Read & Kalkman, 2003).

5.6. A model for the pathways of vesicle trafficking followed by FM4-64 in plant cells

A model is presented in Fig. 8, which summarizes the possible pathways that FM4-64 may follow in plant cells.

From the plasma membrane to the cell plate.  The immediate staining of the plasma membrane is rapidly followed by staining of the forming cell plates. The latter possibly only occurs when the forming cell plate is connected to the peripheral plasma membrane as a result of FM4-64 diffusing laterally.

From the plasma membrane to targeted organelles (Fig. 8).  Dye is internalized from the plasma membrane into the cell in endocytic vesicles, which fuse with putative endosomes. Endosomes have still to be clearly defined in plant cells. Moreover, the exact relationsip between plant ‘endosomes’, endocytic intermediates (partially coated reticulum or multivesicular bodies, see Tanchak et al., 1984; Hübner et al., 1985; Hillmer et al., 1986) and PVCs is still debatable (Jürgens & Geldner, 2002). The PVCs might function as a junction between endocytic and secretory pathways to the vacuole (see Bolte et al., 2004, for a discussion). Whatever the identity of putative intermediary structures, FM4-64-stained membranes reach the Golgi and the vacuolar compartments. Inadequate spatial and temporal resolution in all studies so far precludes definition of the exact route(s) taken by the dye. It is not clear if all dye-labelled membrane passes through the GA before reaching the vacuole, or if it can also reach the vacuole bypassing the Golgi stacks. From our observations on BY-2 cells expressing a Golgi marker, it seems that FM-stained membrane traffics through the GA before reaching the vacuolar membrane.

Vacuolar membranes as an end point.  Much of the FM-dye finally ends up in the tonoplast.

Recycling in the secretory pathway.  FM4-64 labelling of the pivotal component of secretory pathways (i.e. the GA) could theoretically lead to the recycling of dye back to the plasma membrane (Belanger & Quatrano, 2000a, b; Parton et al., 2001). This recycling pathway was illustrated here by the labelling of the nascent cell plate isolated in the cytoplasm but not yet attached to peripheral plasma membrane. Whether FM4-64 normally recycles back to the plasma membrane in plant cells is not known, although evidence has been obtained for it occurring in fungal hyphae (Read & Hickey, 2001). It is also the very process that has made the FM-dyes so useful for assessing synaptic vesicle exocytosis (De Paiva et al., 1999).

6. Concluding comments

The arrival of FM-dyes as powerful experimental tools is providing exciting new insights into vesicle trafficking in plant cells. However, the complexity of the relationships between the dynamic pattern of staining and cell physiology have to be taken into account when planning or interpreting experimental data. Moreover, plant cells may have mechanisms of internalization and intracellular distribution of membrane-selective dyes that are different from those in animal and fungal cells. Nevertheless, FM-dyes are proving very useful for visualizing membrane internalization and recycling, endocytosis, and exocytosis. Experimental observations must be analysed with considerable care, bearing in mind the multiple targets and kinetics of these dyes depending on the type and physiology of the cell being examined.

Multiparametric time-lapse imaging on functionally disrupted mutants, and localization of the FM-dyes at the EM level, will prove very useful in the future. There is also an urgent need for new endocytosis markers for live-cell imaging in plant cells. The exact pathways of endocytosis and the identification of ‘endosomes’ in plant cells at the light microscope level still remain major challenges for this field of research.


Special thanks are given to Dr Spencer Brown (ISV, CNRS, Gif sur Yvette) for his advice and for critically reading the manuscript. We thank Prof. Chris Hawes (Oxford Brookes University, U.K.) for the kind gift of ST-GFP cell lines. The IFR87 (FR-W2251) ‘La Plante et son Environnement’ together with the Conseil Général de l’Essonne (ASTRE) provided the confocal microscopy facility. We are appreciative of the technical assistance of Marie-Thérèse Crosnier (ISV) for the cell cultures and Lise-Anne Denmat (ISV) for the construction of the HDEL-GFP BY-2 cell line.