Sphingolipids involvement in plant endomembrane differentiation: the BY2 case


  • Anne Aubert,

    1. Laboratoire Dynamique de la Compartimentation Cellulaire, CNRS UPR2355/IFR87, Institut des Sciences du Végétal, Centre de Recherche de Gif (FRC3115), 91198, Gif-sur-Yvette Cedex, France
    Search for more papers by this author
    • These authors have equally contributed to this work.

  • Jessica Marion,

    1. Laboratoire Dynamique de la Compartimentation Cellulaire, CNRS UPR2355/IFR87, Institut des Sciences du Végétal, Centre de Recherche de Gif (FRC3115), 91198, Gif-sur-Yvette Cedex, France
    2. Pôle de Biologie Cellulaire, Imagif, Centre de Recherche de Gif, (FRC3115), CNRS, IFR 87, 91198, Gif-sur-Yvette Cedex, France
    Search for more papers by this author
    • These authors have equally contributed to this work.

  • Claire Boulogne,

    1. Laboratoire Dynamique de la Compartimentation Cellulaire, CNRS UPR2355/IFR87, Institut des Sciences du Végétal, Centre de Recherche de Gif (FRC3115), 91198, Gif-sur-Yvette Cedex, France
    Search for more papers by this author
  • Mickael Bourge,

    1. Laboratoire Dynamique de la Compartimentation Cellulaire, CNRS UPR2355/IFR87, Institut des Sciences du Végétal, Centre de Recherche de Gif (FRC3115), 91198, Gif-sur-Yvette Cedex, France
    2. Pôle de Biologie Cellulaire, Imagif, Centre de Recherche de Gif, (FRC3115), CNRS, IFR 87, 91198, Gif-sur-Yvette Cedex, France
    Search for more papers by this author
  • Susana Abreu,

    1. Laboratoire Dynamique de la Compartimentation Cellulaire, CNRS UPR2355/IFR87, Institut des Sciences du Végétal, Centre de Recherche de Gif (FRC3115), 91198, Gif-sur-Yvette Cedex, France
    Search for more papers by this author
  • Yannick Bellec,

    1. Institut Jean-Pierre Bourgin, unité mixte de recherche 1318, INRA-AgroParisTech, Centre de Versailles-Grignon, 78026 Versailles cedex, France
    Search for more papers by this author
  • Jean-Denis Faure,

    1. Institut Jean-Pierre Bourgin, unité mixte de recherche 1318, INRA-AgroParisTech, Centre de Versailles-Grignon, 78026 Versailles cedex, France
    Search for more papers by this author
  • Béatrice Satiat-Jeunemaitre

    Corresponding author
    1. Laboratoire Dynamique de la Compartimentation Cellulaire, CNRS UPR2355/IFR87, Institut des Sciences du Végétal, Centre de Recherche de Gif (FRC3115), 91198, Gif-sur-Yvette Cedex, France
    2. Pôle de Biologie Cellulaire, Imagif, Centre de Recherche de Gif, (FRC3115), CNRS, IFR 87, 91198, Gif-sur-Yvette Cedex, France
    Search for more papers by this author

(fax +33 (0)1 69 82 3355; e-mail bsj@isv.cnrs-gif.fr).


Sphingolipids play an essential role in the functioning of the secretory pathway in eukaryotic organisms. Their importance in the functional organization of plant cells has not been studied in any detail before. The sphingolipid synthesis inhibitor fumonisin B1 (FB1), a mycotoxin acting as a specific inhibitor of ceramide synthase, was tested for its effects on cell growth, cell polarity, cell shape, cell cycle and on the ultrastructure of BY2 cells. We used cell lines expressing different GFP-tagged markers for plant cell compartments, as well as a Golgi marker fused to the photoconvertible protein Kaede. Light and electron microscopy, combined with flow cytometry, were applied to analyse the morphodynamics and architecture of compartments of the secretory pathway. The results indicate that FB1 treatment had severe effects on cell growth and cell shape, and induced a delay in cell division processes. The cell changes were accompanied by the formation of the endoplasmic reticulum (ER)-derived tubular aggregates (FB1-induced compartments), together with an inhibition of cargo transport from the ER to the Golgi apparatus. A change in polar localization of the auxin transporter PIN1 was also observed, but endocytic processes were little affected. Electron microscopy studies confirmed that molecular FB1 targets were distinct from brefeldin A (BFA) targets. We propose that the reported effects of inhibition of ceramide biosynthesis reflect the importance of sphingolipids during cell growth and establishment of cell polarity in higher plant cells, notably through their contribution to the functional organization of the ER or its differentiation into distinct compartments.


The functional organization of eukaryotic cells by a distinct set of endomembranous compartments is a key process in evolution. The important roles of organelles such as the endoplasmic reticulum (ER), Golgi apparatus (GA), endosomes and vacuoles/lysosomes in cell growth, cell division and cell polarity processes include vectorial transport activities along a cell axis from ER to plasma membrane, or from plasma membrane into the cell, and imply specific protein machineries (see references in Képès et al., 2005). Membrane lipids have been described to act both as structural components of the endomembranes and as signaling molecules in the secretory pathway (Wolf et al., 1999; Wisniewska et al., 2003; Raffaele et al., 2009b). Hence, it is clear that they play a significant role in processes such as subcellular compartmentation and transport/trafficking.

Various studies are currently under way to uncover the functional organization of lipids in parts of the endomembrane systems in eukaryotic organisms. The general outcome is a complex picture in which the biosynthetic steps of membrane lipids involved different cell compartments. Results also indicate that membrane lipids do not mix homogenously, and that their interactions with other lipids or proteins produce distinct domains within the membranes. This spatial arrangement of membrane lipids in distinct micro- or even nanodomains, rich in sphingolipids and sterols, the so-called lipid rafts, are expected to play a role in protein sorting throughout the endomembrane system in all eukaryotic cells (Laloi et al., 2007; Klemm et al., 2009).

These main features of lipid compartmentation are also shared by plant cells. However, plant cells also face specific challenges with respect to environmental and evolutionary conditions/constraints and, therefore, their endomembrane system and lipid homeostasis exhibit specialized features (respectively Hawes and Satiat-Jeunemaitre, 2005; Moreau et al., 2007; Bouttéet al., 2006; Brown et al., 2010). For example, the lipid composition of the chloroplast membrane, as well as lipid gradients from the ER to the plasma membrane or vacuoles, was demonstrated to be plant-specific using biochemical techniques (Yoshida and Uemura, 1986; Moreau and Cassagne, 1994; Moreau et al., 1998; Block et al., 2007). Hitherto the cell biology of plant lipids, i.e. their in cellulo compartmentalized synthesis, transport and subcellular interactions, have remained poorly understood. The impact of phytosterols on the regulation of cell polarity, cell division and on endomembrane organization has now been demonstrated in a number of elegant studies (Willemsen et al., 2003; Bouttéet al., 2007a). In contrast, studies on the role of sphingolipids in cell development are still in their infancy.

Compared with glycerolipids (i.e. phospholipids and glycolipids), which form the main lipid constituents of the plant endomembranes, sphingolipids and sterols are present in relatively small quantities. The metabolism of sphingolipids in plants has many features in common with that in other organisms. Ceramide is the basic backbone for all sphingolipids, such as the major complex sphingolipids glucosylceramides (GluCer) and glycosylphosphoinositol ceramides (GIPCs) (Sperling and Heinz, 2003; Markham et al., 2006).

The sphingolipid biosynthetic pathways and their cellular compartmentalization along the secretory pathway is complex (see Raffaele et al., 2009b for a review), although it has been shown that the enzymes are mostly localized in the ER (Marion et al., 2008; Melser et al., 2010). The first step in ceramide biosynthesis is carried out by the enzyme complex serine-palmitoyltransferase (Chen et al., 2006; Bach and Faure, 2010). This enzyme is made of two subunits, LCB1 and LCB2, in charge of synthesizing the long chain bases (LCBs, or sphingoid bases), which are specific to sphingolipids. These LCBs associate with very long chain fatty acyl-CoA (VLCFA-CoA) to form sphingolipids, which is orchestrated by ceramide synthases. Synthesis of complex sphingolipids such as GluCer or GIPCs is dependent on GC synthase and IPC synthases, and constitute two different biosynthetic pathways (Figure 1).

Figure 1.

 Cell compartmentation of sphingolipid synthesis in higher plant cells, outlining the potential pathway targeted by fumonisin B1 (FB1) and brefeldin A (BFA). ER, endoplasmic reticulum; GC, glucosylceramide; GIPC, glycosylinositol phosphorylceramide; LCBs, long chain bases; VLCFA, very long chain fatty acids component.

Previous reports on sphingolipid function in plants demonstrate that interfering with their synthesis has consequences for the normal functioning of the endomembrane system (Coursol et al., 2003, 2005; Liang et al., 2003; Zheng et al., 2005; Melser et al., 2010). However, the possible impact of sphingolipids on the typical architecture of the plant endomembrane system has not yet been studied in any detail. To address these questions, we have used BY2 cells (Nagata et al., 1992) expressing different specific GFP-tagged markers for plant cell compartments. Their short cell cycle (11–15 h), polar ribbon-shaped organization, ease of DNA transformation with GFP constructs and the fact that they do not possess chloroplasts make BY2 cells extremely suitable for these studies. We reported previously that interfering with sterol biosynthesis may have serious consequences for the cellular organization of BY2 cells (Mérigout et al., 2002). Therefore, it is of particular interest to analyse the effects of sphingolipid inhibitors on plant endomembrane organization in a similar way, and in the same model system. The effects of fumosinin B1 (FB1, a mycotoxin acting as a specific inhibitor of ceramide synthase; Tsegaye et al., 2007) were analysed with the morphodynamics and ultrastructural features of each compartment of the secretory pathway.

Our results combine cytometry, bio-imaging and electron microscopy approaches (Brown et al., 2007), and provide compelling evidence for a role of sphingolipids during cell growth and the establishment of cell polarity in higher plant cells through their contribution to the functional organization of the endomembrane system.


Fumonisin B1 affects cell growth and delays the G2/PM phase

The effects of 1 μm FB1 were tested on 3-day-old wild-type BY2 cells growing either in liquid cell suspension or on solid agar medium (exponential growth phase). FB1 treatment for 24 h resulted in a decrease of the pack cell volume of cultured cells in liquid medium by one-third compared with the control (data not shown). Similarly, the growth of the cell population was strongly inhibited when cells were grown on FB1-containing solid medium (Figure 2a).

Figure 2.

 Effects of fumonisin B1 (FB1) on BY2 cell growth.
(a) Three-day-old BY2 cells were streaked on solid medium without (left) or with 1 μm FB1 (right). Photographs were taken after 15 days. Scale bar: 1 cm.
(b) Histograms summarizing cell cycle analysis by flow cytometry of control (white column) and FB1-treated (black column) BY2 cells (mean of three experiments). Top: 24-h treatment (induces a significant decrease in cell population in the S and G2/M phases). Bottom: 72-h treatment results in a significant decrease of cells in the G1 phase. Note the decrease in S phase in control cells, and a still active DNA synthesis in treated cells after 72 h of FB1 treatment. *Student’s t-test.
(c) Mitotic index of 3-day-old BY2 cells after 24, 48 and 72 h of treatment with FB1 (mean of four experiments). After 24 h of treatment, there was a significant decrease in the mitotic index in treated cells (black column) compared with control cells (white column). Over the next few days, mitotic indices in control cells (white column) progressively decreased, whereas in treated cells a significant level of mitotic activity remained after 48 and 72 h of treatment.

Flow cytometry analyses revealed that FB1 does not target one phase of the cell cycle in particular (Figure 2b). After 24 h of FB1 treatment, the number of treated cells in G2/M phase increased compared with control cells (Figure 2b, top). The same was observed in cells that were treated with FB1 for 72 h (Figure 2b, bottom). These results may reflect the fact that treated cells tend to accumulate or enter the G2/M phase more slowly (higher value and longer times). On the other hand, 72 h of treatment with FB1 showed an increased number of cells in the S phase (Figure 2b, bottom). These results show that the growth decrease in the BY2 cell culture corresponds to a general slowdown of the cell cycle, possibly related to FB1 effects on the cell division processes occurring during the M phase.

The mitotic index values in BY2 cells from 4–6-day-old cultures (Figure 2c) confirmed that FB1 has a strong effect on division processes. In control cells, the mitotic index value regularly decreased over time, and at day 7, no mitosis was recorded (not shown). In FB1-treated BY2 cells, there was an initial strong decrease in the mitotic index value after 24 h of FB1 treatment. However, surprisingly, the mitotic index rose again, and reached higher values than in control cells at the same age. Light microscopy observations confirmed that mitotic cells were still recorded after 6 and 7 days in FB1-treated cultures. These results suggest that FB1-treated cells may be delayed in entering the cell division processes, and would stay in the mitotic phase for a longer period of time.

To verify whether these FB1 effects are related to effects on BY2 sphingolipid synthesis, lipid compositions were analysed by HPLC (Figure S1). The results showed increased levels of DHS and 1,4-anhydro-t18:0, presumably because these are substrates of ceramide synthases; however, the unsaturated LCBs, which are found in complex sphingolipids downstream of the ceramide synthase step, decreased. Considering the large changes observed in LCB profile, in particular with the accumulation of DHS as a hallmark of ceramide synthase impairement, we assume that FB1 inhibited ceramide synthesis in BY2 cells.

Fumonisin B1 affects cell division and cell shape in a cytoskeleton-independent manner

BY2 cells are typically organized as a multicellular ribbon (Figure 3a). This typical morphology was strongly affected by FB1 treatment (Figure 3b–d). After 24 h of treatment cells tended to swell (Figure 3b), suggesting a change in the cell growth axis. A fluorescein di-acetate test showed that the cell viability was not affected, even after 72 h of treatment (data not shown). The typical ribbon shape was still recognizable, albeit made up of enlarged, swollen cells (Figure 3c). FB1 also induced unusual division planes, as two adjacent cells on a radial plane were regularly observed (Figure 3d).

Figure 3.

 Effects of fumonisin B1 (FB1) on BY2 cell shape and division planes, by differential interference contrast (DIC). Note the enlargement of the cells after 24 (b) and 72 h (c) of treatment with FB1. The typical ribbon shape of control BY2 cells is altered (a), and the cells exhibit abnormal division planes (d and c, arrow). Scale bar: 10 μm.

FM4-64 was used as a cell plate marker in order to study possible cell plate defects in more detail (Bolte et al., 2004). FB1-treated cells often exhibited cell plates with unusual profiles (wavy profiles in Figure 4b,d; abnormal division planes in Figure 4c), and a significant number of incomplete cell plates was observed, compared with control cells (Figure 4a). These later observations are actually in accordance with the observed delay in the cell progression during the mitotic stage described above.

Figure 4.

 FM4-64 labeling of plasma membrane and cell plates in BY2 cells. (a) Control cells with linearly growing cell plate. (b–d) Abnormal cell plate patterns induced by fumonisin B1 (FB1), exhibiting either wavy patterns (b, d), or incomplete/fragmented cell plates (c, d). Scale bar: 10 μm.

The cytoskeleton is known to be directly involved in these biological processes, and it was subsequently investigated whether FB1 altered the actin and/or microtubule networks. The immunolabeling of actin did not reveal changes in actin polymerization or in the actin 3D pattern (data not shown). Similarly, the immunolabeling of microtubular arrays did not detect changes in microtubule polymerization: the same long tubular structures organized in a cortical network, as well as the alignment of short structures in the phragmoplast on each side of the growing cell plate in dividing cells, were observed in control and treated cells (Figure 5a–d). However, the typical helical 3D organization of the microtubules in interphase cells was disrupted (Figure 5b), and the 3D structure of the phragmoplasts in treated cells was not as straight as in control cells (Figure 5c), reminding us of the altered cell plate profiles shown by FM4-64 staining, such as the wavy patterns (Figure 5d).

Figure 5.

 Immunofluorescence labeling of microtubules in control (a and c) or fumonisin B1 (FB1)-treated cells (b and d). Note the altered 3D helical organization of the microtubular pattern in interphase cells (b), and the unusual wavy profile of the phragmoplast in dividing cells (d). Scale bar: 10 μm.

The loss of the typical helical pattern could relate to changes in cell growth axis, and/or to defects in some anchoring processes of microtubules at the plasma membrane caused by the sphingolipid synthesis alterations investigated below.

Fumonisin B1 affects the organization of the plasma membrane

The effects of FB1 on a BY2 cell line expressing PIN1-GFP (Bouttéet al., 2006) were investigated. In control cells, PIN1-GFP shows polar distribution, highlighting the radial membranes of the cells (Figure 6a). Cell plates in dividing cells were also visible in the PIN1-GFP line (see Bouttéet al., 2006). After treatment with FB1, the distribution of PIN1-GFP in the cell division planes was altered (Figure 6b,c), as well as in the growing/developing cell plate (Figure 6e). Moreover, FB1-treated cells may exhibit a loss of polar distribution in PIN-GFP, as fluorescence was often redistributed over the whole cell surface (Figure 6d). Finally, some FB1-treated cells exhibited intracellular PIN1-GFP-labeled aggregates, often positioned in close proximity to the nucleus (Figure 6f).

Figure 6.

 The effects of fumonisin B1 (FB1) on PIN1-GFP BY2 cell lines. Control cells (a) and FB1-treated cells (b–f), showing defects in plane division and polar labeling (b, arrow; c, differential interference contrast, DIC, image), redistribution of PIN-GFP to longitudinal membranes (d, arrows) and unusual wavy profiles of cell plates (e). PIN1-GFP was also observed as aggregates in perinuclear areas (f, arrow). Scale bar: 10 μm.

FB1 effects on the redistribution of PIN1-GFP polar patterns could be related to specific alterations of secretory, endocytic and recycling events (Bouttéet al., 2006, 2007b; Dhonukshe et al., 2007). To explore this hypothesis, a disruption of sphingolipids in BY2 cells expressing distinct markers for endomembrane compartments was undertaken.

Endoplasmic reticulum and Golgi markers are affected by FB1

The localization of GFP-HDEL, a marker for the ER, as well as of ST-GFP, a Golgi marker, was affected by FB1 treatment (Figure 7). Firstly, control cells expressing GFP-HDEL showed the typical extensive fluorescent tubular network previously described for the ER (Mérigout et al., 2002). This was mainly visible near the nuclear and cortical zone of the cells (Figure 7a). After several hours of FB1 treatment, ER tubules thickened and the meshwork started to fragment (Figure 7b). At this stage, GFP labeling of the cell plate in dividing cells was also detected, even at the end of telophase (Figure 7d). After 24 h of treatment the ER network was GFP-labeled, but, in addition, labeled aggregates were seen in the vicinity of nuclear membranes (Figure 7c).

Figure 7.

 The effects of fumosinin B1 (FB1) on 3D fluorescent patterns of GFP-HDEL (a–d) and ST-GFP BY2 cell lines (e–h). Scale bar: 10 μm.
(a) In control cells, GFP-HDEL labeled the endoplasmic reticulum (ER) meshwork. (b) After 5 h of treatment with FB1, the ER meshwork is still recognizable dispersed throughout the cortical cytoplasm. (c) After 24 h of treatment with FB1, GFP-HDEL labeling is concentrated in perinuclear aggregates. Note the fluorescent nuclear membranes. (d) After 8 h of treatment with FB1, dividing cells with a highly fluorescent cell plate wrapped in a fluorescent meshwork are visible. Labeling the nuclear membrane.
(e) In the control cell, ST-GFP labeled the Golgi stacks, giving a punctate pattern throughout the cytoplasm. (f) After 5 h of treatment with FB1, a thickening of fluorescent units is observed. (g) The ST-GFP fluorescence is concentrated in perinuclear aggregates and nuclear membranes. These features are somehow similar to the ones described in the 24-h treatment of GFP-HDEL cell lines (compare with Figure 6c). (h) After 24 h of treatment with FB1, dividing cells exhibit a fluorescent cell plate, with fluorescent aggregates at the cell plate extremities; there is strong labeling of the nuclear membrane.

The ST-GFP labeling in the Golgi apparatus was typically detected as numerous 1-μm fluorescent bodies distributed over the cytoplasm (Figure 7e). The first changes were observed after 5 h of treatment with FB1, when larger, globular structures were detected (Figure 7f). After 24–48 h of treatment, all the fluorescence was concentrated in perinuclear aggregates, and nuclear membranes were also GFP-labeled (Figure 7g). In dividing cells, GFP-labeled cell plates were visible, as well as fluorescent nuclear membranes in telophase cells (Figure 7h). The latter observation was reminiscent of GFP-HDEL labeling of the ER and nuclear membrane, and suggested that at least part of the ST-GFP fluorescence was located or relocated to the ER.

The photoactivable ST-Kaede BY2 cell line was used to further discriminate the dynamics of these aggregates formation.

FB1 affects trafficking from the ER to the GA

ST-Kaede fluorescence can be converted from green to red upon exposure to UV light (Brown et al., 2010). In control cells, at 24 h post-photoconversion, the red (photoconverted) and green (newly synthesized) fluorescence was present on the same Golgi stacks (Figure 8a–c). This clearly showed that most of the newly synthesized ST-Kaede (green) is transported to existing Golgi stacks (red), as previously described by Brown et al. (2010). FB1 was added either 4 or 24 h before the photoconversion, in order to analyse the effects of FB1 on ST-Kaede trafficking from ER to the Golgi. When cells were UV-illuminated after 4 h of treatment with FB1, no effect of FB1 was detected; at 24 h post-conversion (i.e. after 28 h of FB1 treatment), the Golgi stacks were labeled with both red and green fluorescence (Figure 8d and f), suggesting that FB1 did not have severe effects on the existing Golgi stacks, and that transport of newly-synthesized ST-Kaede into the Golgi continues at a certain level. However, green-fluorescent aggregates, like those observed in the FB1-treated ST-GFP-expressing cells, were also observed (compare Figure 8e with Figure 7g), and these structures constitute the largest part of the green fluorescence. This indicates that most of the newly synthetized ST-Kaede ended up in perinuclear aggregates.

Figure 8.

 The effects of fumosinin B1 (FB1) on the 3D fluorescent pattern of ST-Kaede. Cells were treated for short (4 h) or long (24 h) periods with FB1 (d–h and g–i, respectively), were photoconverted from green to red and then observed 24 h later. In this plate, the fluorescence detected in the red channel (a, d, g) is therefore associated with ST-Kaede proteins in place before photoconversion, meanwhile the green fluorescence (b, e, h) is associated with newly synthesized ST-Kaede during the 24 h period post-photoconversion. Scale bar: 10 μm.
(a–c) Control ST-Kaede cells. (a) Red ST-Kaede Golgi stacks in place at the time of photoconversion. (b) Newly synthesized green ST-Kaede labeling of Golgi stacks. (c) The merged picture suggests that most of the newly synthesized Kaede proteins have reached pre-existing Golgi compartments.
(d–f) Treatment with FB1 for 4 h. (d) Red ST-Kaede labeling of Golgi stacks in place at the time of photoconversion. (e) Newly synthesized ST-Kaede proteins accumulate in fluorescent aggregates and form punctuate structures. (f) Merged pictures: newly synthesized proteins have reached pre-existing Golgi stacks or have accumulated in non-Golgi structures.
(g–i) After 24 h of treatment with FB1. (g) Red ST-Kaede labeling of Golgi stacks in place at the time of photoconversion. (h) Newly synthesized ST-Kaede proteins accumulate in fluorescent aggregates; they do not form punctate structures. (i) Merge pictures: newly synthetized proteins accumulate in non-Golgi structures.

When cells were UV-illuminated after 24 h of treatment with FB1, fluorescent aggregates were already present at the time of the photoconversion. At 24 h post-conversion (Figure 8g–i), distinct red fluorescent punctate structures are observed, embedded in a more diffuse red fluorescent area in the perinuclear zone (Figure 8g). These red punctate structures, which are thought to be Golgi stacks, had to be the Golgi stacks already present at the time of photoconversion, and have remained resistant to FB1 treatment. The diffuse perinuclear fluorescence contained red and green fluorescence, reflecting accumulation of ST-Kaede in this area both before and after photoconversion (Figure 8h–i). This suggests that newly synthetized ST-Kaede failed to reach the Golgi stacks, and instead accumulated in an unknown structure around the nucleus.

As a whole, these results suggest that sphingolipid deficiency impacts the transport from the ER to the Golgi.

Endocytosis is slightly affected by FB1

In order to further discriminate between specific FB1-induced alterations of exocytosis versus endocytosis pathways, we followed the internalization of FM4-64 in BY2 cells expressing PIN1-GFP (Figure 9a–f) and ST-GFP (Figure 9g–l). FM4-64 internalization processes still took place under FB1 treatment (Figure 9e,k). At the same time, FB1 induces PIN1-GFP and ST-GFP aggregates, as previously described (Figure 9d,j, respectively). Merged pictures (Figure 9f,l) show, however, that FM4-64 did not label such aggregates. These results suggest that in our experimental conditions and in BY2 cells, FB1 affects the exocytosis pathways, but only slightly effects the endocytosis pathway.

Figure 9.

 Effects of fumosinin B1 (FB1) on FM4-64 internalisation in PIN1-GFP (a–f) and ST-GFP BY2 (g–l) cells.
(a–c) PIN1-GFP control cell: (a) typical fluorescent PIN1 pattern; (b) internalization of FM4-64; (c) the merged picture shows the partial localization of FM4-64 and GFP fluorescence. Scale bar: 10 μm.
(d–f) PIN1-GFP-treated cells: (d) FB1-induced perinuclear aggregates; (e) internalization of FM4-64; (f) the merged picture shows no co-localization of FM4-64 with FB1-induced aggregates. Scale bar: 10 μm.
(g–i) ST-GFP control cells: (g) typical fluorescent Golgi pattern; (h) internalization of FM4-64; (i) the merged picture shows the partial co-localization of FM4-64 with GFP-labeled organelles. Scale bar: 10 μm.
(j–l) ST-GFP-treated cells: (j) FB1-induced aggregates; (k) internalization of FM4-64; (l) no co-localization of FM4-64 with ST-GFP aggregates. Scale bar: 10 μm.

Ultrastructural analyses reveal the formation of ER-derived membranous compartments and a transport blockage from the ER to the GA.

Severe alterations of internal membranes and unusual proliferations of tubular membranes were observed (compare Figure 10a with Figure 10c–e). These unusual tubular structures radiate from a clear central zone (Figure 10c,e–f), and may correspond to the fluorescent aggregates seen by confocal laser scanning microscopy. In contrast, the ultrastructure of the Golgi stacks was not altered by FB1 (Figure 10b and d), even after 48 h of FB1 treatment (Figure 10f), confirming the light microscopy observations on ST-Kaede photoconverted cell lines. Typical ER structures were also detected in the cytoplasm (Figure 10e), and neither nuclear or vacuolar membranes were altered by the drug. Taken together, this suggests that the membranous aggregates may represent newly formed ER-derived structures, named FB1-induced compartments.

Figure 10.

 Effects of fumosinin B1 (FB1) on the cell ultrastructure. (a) Control with typical plant cell endomembrane features. (b) Golgi stack in control cells.
(c, d) Micrographs of cell ultrastructure after 24 h of treatment with FB1 at different electron microscope magnifications. Golgi stack and endoplasmic reticulum (ER) tubules are recognizable: nuclear and vacuolar membranes appeared unaffected, and the Golgi stack morphology is also unaffected.There is an extensive tubular membrane network (tn) within the cytoplasm.
(e, f) Higher magnifications often outline the peculiar radiating pattern of this network from a 0.5–1 μm square electron-transparent area (*).
Key: er, endoplasmic reticulum; ga, Golgi apparatus; n, nucleus; v, vacuole.
Scale bars: a, c, e, f, 1 μm; b, d, 200 nm.

We hypothesized that the altered fluorescent patterns of the Golgi marker ST-GFP observed by light microscopy could be related to an altered trafficking of the protein from the ER to the GA. Immunogold labeling of ST-GFP in both control and FB1-treated cells confirmed this hypothesis (Figure 11a), as a significant decrease in gold labeling on the Golgi stacks of FB1-treated cells is observed (Figure 11b). In control cells, gold particles were rarely seen in the cytoplasm or ER lumen, whereas in treated cells ‘clouds’ of gold labeling were detected in FB1-induced compartments, i.e. areas of tubular membrane profusions induced by the drug (Figure 11c,d). This demonstrates that ST-GFP is trapped in membranous structures that are probably ER-derived.

Figure 11.

 Immunogold labeling of Golgi stacks in ST-GFP cell lines by anti-GFP antibodies (cryofixation by high-pressure freezing). Scale bar: 200 nm.
(a) Control cells.
(b) Quantification of gold particles on 60 Golgi stacks in control cells (white column) or in fumosinin B1 (FB1)-treated cells. Note the strong decrease of immunolabeling in treated cells.
(c, d) Immunolabeling of the FB1-induced tubular meshwork by anti-GFP, illustrating the trapping of ST-GFP molecules within the area. Scale bar: 500 nm.

This labeling therefore explained the observed formation of fluorescent aggregates after FB1 treatment labeled with ST-GFP, ST-Kaede or PIN-GFP (Figures 6–8), i.e. proteins on their way to anterograde transport being blocked in ER-derived structures. The fact that no visible difference in ultrastructural features of plasma membrane or endocytic compartments was detected also confirmed our previous hypothesis that endocytosis dynamics were not primarily affected by the drug.

FB1-induced compartments are distinct from BFA-induced reorganization of the ER–GA complex

The next question was to know how specific the FB1-induced modifications were compared with some other ER–GA-disrupting drug, such as BFA (Satiat-Jeunemaitre et al., 1996). After treatment with BFA, the Golgi stacks of BY2 cells underwent profound reorganization, leading to the formation of ER–Golgi hybrid compartments (Figure 12c; Langhans and Robinson, 2007), and the typical membrane profusions induced by FB1 treatment (Figure 12b) were never observed. Interestingly, when FB1-treated cells were treated with BFA for 1 h, the cells exhibited the two types of abnormal figures: ER/Golgi hybrid compartments typical of BFA treatment; and ‘FB1-induced compartments’ made of a profusion of tubular and swollen membranes (compare Figure 12a and d). These results outline a specific sensitivity to each drug, confirming fully distinct molecular targets. They also confirm that the formation of FB1-induced compartments can be regarded as a ‘de novo’ ER-derived compartment, rather than a reorganization of previous compartments, as found in BFA-treated cells.

Figure 12.

 Brefeldin A (BFA) and fumonisin B1 (FB1) have distinct effects on the endomembrane network. (a) Control cell. (b) FB1-treated cells with typical tubular membrane profusion. (c) BFA-treated cell with endoplasmic reticulum (ER)–Golgi apparatus (GA) hybrid structure. (d) FB1- and BFA-treated cell showing both ER–GA hybrid structures (arrow) and membrane profusion (arrowhead). Key: er, endoplasmic reticulum; ga, Golgi apparatus; tn, tubular membrane network; v, vacuole. *Electron-transparent area. Scale bar: 200 nm.


These results show that, in BY2 cells, an FB1-sensitive sphingolipid pathway impairs several cell functions, interfering with the membrane trafficking pathways by: (i) introducing a delay in the progression through the cell cycle; (ii) introducing defects in cell plane positioning, cell plate growth and polar axis establishment; (iii) inducing the ER-derived membranous tubular network, i.e. FB1-induced compartments; and (iv) blocking the transport of cargo molecules from the ER to the GA.

FB1 effects outline a low sphingolipid turnover and a cell ability to compensate sphingolipid deficiency

The peculiarity of FB1 treatment compared with other pharmacological agents disturbing the organization of cell membranes, such as BFA (Satiat-Jeunemaitre and Hawes, 1992; Satiat-Jeunemaitre et al., 1996; Mérigout et al., 2002; Ritzenthaler et al., 2002; Couchy et al., 2003; Langhans and Robinson, 2007) or nordihydroguiaretic acid (NDGA) (Mérigout et al., 2002), is the time period needed to detect any FB1 effects on cell growth. The low turnover of sphingolipids could explain these slow changes in subcellular and cellular morphological features. On the other hand, biochemical changes have been reported to occur long before any morphological changes are observed (Abbas et al., 1994; Spassieva et al., 2002). This may indicate that plants are able to adjust their growth mechanisms to compensate for a reduction in ceramide synthesis, a process already noticed in the case of the partial suppression of the early step of sphingolipid synthesis (Chen et al., 2006).

Impact of sphingolipids deficiency on cell polarity

A deficiency in sphingolipids has a strong impact on the establishment of a polar axis, and may lead to the mislocalization of the PIN1 proteins. Lateral diffusion of PIN labeling, together with a disorganization of the microtubular array, over the cell surface has previously been described when BY2 cells were turned into protoplast (Bouttéet al., 2006). In this study, the observations suggest that the mechanisms involved in the maintenance of cell polarity could be related to the fine organization of the plasma membrane. In our study, such effects could indeed be related to a profound alteration of plasma membrane organization. PIN1 proteins have been reported to be associated with lipid rafts in plants (Titapiwatanakun et al., 2009), and lipid rafts are known to be enriched in sterols and sphingolipids (Mongrand et al., 2004; Raffaele et al., 2009a). Moreover, proteins and sphingolipids associate to form microdomains in the trans-Golgi network (Barz and Walter, 1999). Therefore, the redistribution of PIN1 proteins may be related either to a change in raft composition or to a defect in protein sorting along the secretory pathway.

Sphingolipids are involved in the orchestration of anterograde flows

Our results support the hypothesis that sphingolipids are essential to maintain anterograde flows, as ceramide deficiency causes an inhibition of ER to GA transport of cargo molecules. Such effects of sphingolipid deficit were also seen in animal cells where it was shown that transport of GPI proteins from ER to the GA was dependent on sphingolipids (Barz and Walter, 1999). Formations of abnormal cell plate or defects in cell plate growth under FB1 are also good indicators for involvement of sphingolipids in the regulation of exocytosis. The fundamental question of how this blockage is related to changes in ultrastructural organization remains.

In FB1 treatment, the deficiency of sphingolipids and alterations of subcellular processes are associated with a profusion of ER-derived tubular membrane networks. As FB1 may induce apoptosis-like cell death in plants (Asai et al., 2000; Shi et al., 2007), such an occurrence of membranes could be related to the formation of apoptotic-like bodies along the secretory pathway. These modifications are fully distinct from the ones induced by the ER-to-GA transport inhibitor BFA, where Golgi membranes and ER membranes fuse and create a hybrid structure (Langhans and Robinson, 2007). FB1 did not change the endomembrane reactivity to BFA: as these ER-derived structures are full of newly synthesized cargo molecules, they suggest that, without sphingolipids, ER membranes are not able to favor vesicle shuttle between the ER and the GA, but favor some tubular expansions.

As a whole, this study demonstrates specific roles for sphingolipids in endomembrane organization. Therefore. the availability of the sphingolipid pool could be a limiting factor in the ‘differentiation’ of ER membranes towards other compartments (Figure 13). Modification of the sphingolipid molecular signature would therefore interfere with specific membrane protein-associated molecular machinery in charge of vesicular mobility or fusions that could characterize ER-to-GA traffic or GA morphogenesis.

Figure 13.

 Diagram summarizing the role of fumonisin B1 (FB1)-sensitive biosynthetic pathways of ceramides in plant membrane flows.
By affecting sphingolipid synthesis within the endoplasmic reticulum (*ER), FB1 (partially) blocks the transport of cargo molecules from the ER to the Golgi apparatus (GA), which accumulate in FB1-induced ER-derived compartments in a reversible manner. Sphingolipids play a role in ER ‘differentiation’ by affecting the lipid microdomain structure within the Golgi and plasma membrane, without direct consequences to their structural architecture, but impacting on the late fusion trafficking events involved in cell polarity maintenance at the cell plasma membrane or in cell plate formation.

Experimental Procedures

Biological material

Nicotiana tabacum cv. Bright Yellow 2 (BY2) suspension-cultured cells were grown in modified MS medium as previously described (Couchy et al., 2003). Four BY2 transgenic cell lines were used in this study: ST-GFP as a Golgi complex marker (Saint-Jore et al., 2002); ST-Kaede (Brown et al., 2010); GFP-HDEL as an ER marker (Mérigout et al., 2002); and PIN1-GFP as a cell polarity marker Bouttéet al., 2006). Experiments were performed on 3-day-old BY2 cultures exhibiting a high mitotic index.

Drug treatments

Fumosinin B1 (FB1 F1147; Sigma-Aldrich, http://www.sigmaaldrich.com) was added to the suspension culture from a 10 mm aqueous stock solution, to reach a 1 μm final concentration.

Treatments with BFA in a final concentration of 10 μg ml−1 (Sigma-Aldrich) were performed for 1 h as previously described (Mérigout et al., 2002).

Cytological, immunocytological approaches and probes for light microscopy

DNA was stained with Hoechst 33342 (2 μg ml−1). FM4-64 dye was added to the cell suspension as described in Bolte et al. (2004). Indirect immunofluorescence experiments were performed as previously described (Hawes and Satiat-Jeunemaitre, 2001). Anti-tubulin antibodies (Interchim, http://www.interchim.com) were used at a dilution of 1:100. An Alexa Fluor 488-conjugated anti-mouse antibody (Invitrogen, http://www.invitrogen.com) was used as a secondary antibody at a dilution of 1:400.


Imaging by light microscopy.  Cells expressing GFP or Kaede fusions were imaged using confocal microscopy (Leica SP2; Leica, http://www.leica.com). Single and dual color imaging were performed as previously described (Brown et al., 2010). Transmission images were taken simultaneously in Nomarski mode differential interference contrast (DIC). UV-violet light exposure was used for the photoconversion of Kaede protein.

Electron microscopy

The BY2 cells were processed as described in Hawes and Satiat-Jeunemaitre (2001), except that the osmium post-fixation step was replaced by a mixture of 1% osmium and 1.5% potassium ferrocyanide. Specimens were embedded in epoxy resin (Agar low-viscosity premix kit medium, Oxford Instruments, Saclay, France) and polymerized for 16 h at 60°C. Selected pictures are representative of observations performed on 20 sections resulting from three experiments. Alternatively, embedding through an automative microwave tissue processor (AMW) device (Leica) was used, following the manufacturer’s instructions. For immunogold labeling, BY2 cells were high-pressure frozen (EMPACT2; Leica) and freeze substituted (AFS2; Leica). Specimens were then infiltrated and embedded in LRWhite resin. Immunogold labeling was performed using rabbit GFP antibodies (AB290; Abcam, http://www.abcam.com), at a dilution of 1:400.

Ultrathin sections (70–90 nm, Ultracut UC6; Leica) were post-stained with aqueous 2% uranyl acetate/lead citrate, as described by Hawes and Satiat-Jeunemaitre (2001). Grids were examined under a JEOL 1400 TEM operating at 120 kV (JEOL, http://www.jeol.com). Images were acquired using a post-column high-resolution (11 megapixels) high-speed camera (SC1000 Orius; Gatan, http://www.gatan.com).

Flow cytometry

Cell cycle analysis of isolated nuclei.  Isolated nuclei from a 2–5-day-old cell culture were stained with propidium iodide (IP; 50 μg ml−1) (Coba de la Peña and Brown, 2001), and data were acquired (excitation at 532 nm; emission through a 590-nm long-pass filter) on a CyFlow SL cytometer (Partec SL, http://www.partec.com). Cell cycles were analysed using MultiCycleAV software (P. Rabinovitch, University of Washington).

Lipid analysis

The analysis of total long chain bases of sphingolipids was performed by HPLC after fluorescent derivatization, as described before (Bach et al., 2008).

Statistical analysis

Data are means ± SEs of a minimum of three independent experiments. Differences between means were evaluated by Student’s t-test with P < 0.05 being taken as the level of significance (*P < 0.05).


The BY2 cell lines expressing ST-GFP were kindly provided by C. Hawes (Oxford, UK). We thank Marie Noëlle Soler and Spencer Brown for their help in imaging and flow cytometry approaches. Anaïs Carpentier established the protocol for AMW processing of plant cells. Thanks are also due to Cynthia Dupas and Karim Hdidou for maintenance of the BY2 culture. BSJ gratefully acknowledges funding from the IFR87 and from the ANR ‘Sphingopolar’ (ANR-07-BLANC-0202). SA was financed by the EC Lifelong Learning program. This work used the facilities of the cell biology unit of the Imagif platform of the Centre de recherche de Gif sur Yvette (http//http://www.imagif.cnrs.fr), supported by the Conseil Général de l’Essonne.