Glucosylceramide Biosynthesis is Involved in Golgi Morphology and Protein Secretion in Plant Cells

Authors

  • Su Melser,

    1. Université V. Segalen Bordeaux 2, Laboratoire de Biogenèse Membranaire, CNRS UMR 5200, 146, rue Léo Saignat, 33076 Bordeaux Cedex, France
    Search for more papers by this author
  • Brigitte Batailler,

    1. Université V. Segalen Bordeaux 2, Plant Imaging Platform of BIC (Bordeaux Imaging Center), INRA Bordeaux Aquitaine BP81, 71 Avenue Edouard Bourlaux, 33883 Villenave d’Ornon Cedex, France
    Search for more papers by this author
  • Martine Peypelut,

    1. Université V. Segalen Bordeaux 2, Plant Imaging Platform of BIC (Bordeaux Imaging Center), INRA Bordeaux Aquitaine BP81, 71 Avenue Edouard Bourlaux, 33883 Villenave d’Ornon Cedex, France
    Search for more papers by this author
  • Christel Poujol,

    1. Université V. Segalen Bordeaux 2, Photonic Imaging Platform of BIC (Bordeaux Imaging Center), Institut des Neurosciences François Magendie, Université V. Segalen Bordeaux 2, 33077 Bordeaux Cedex, France
    Search for more papers by this author
  • Yannick Bellec,

    1. Laboratoire Biologie Cellulaire, INRA, 78000 Versailles Cedex, France
    Search for more papers by this author
  • Valérie Wattelet-Boyer,

    1. Université V. Segalen Bordeaux 2, Laboratoire de Biogenèse Membranaire, CNRS UMR 5200, 146, rue Léo Saignat, 33076 Bordeaux Cedex, France
    Search for more papers by this author
  • Lilly Maneta-Peyret,

    1. Université V. Segalen Bordeaux 2, Laboratoire de Biogenèse Membranaire, CNRS UMR 5200, 146, rue Léo Saignat, 33076 Bordeaux Cedex, France
    Search for more papers by this author
  • Jean-Denis Faure,

    1. Laboratoire Biologie Cellulaire, INRA, 78000 Versailles Cedex, France
    Search for more papers by this author
  • Patrick Moreau

    Corresponding author
    1. Université V. Segalen Bordeaux 2, Laboratoire de Biogenèse Membranaire, CNRS UMR 5200, 146, rue Léo Saignat, 33076 Bordeaux Cedex, France
    2. Université V. Segalen Bordeaux 2, Plant Imaging Platform of BIC (Bordeaux Imaging Center), INRA Bordeaux Aquitaine BP81, 71 Avenue Edouard Bourlaux, 33883 Villenave d’Ornon Cedex, France
    Search for more papers by this author

Dr Patrick Moreau, pmoreau@biomemb.u-bordeaux2.fr

Abstract

Lipids have an established role as structural components of membranes or as signalling molecules, but their role as molecular actors in protein secretion is less clear. The complex sphingolipid glucosylceramide (GlcCer) is enriched in the plasma membrane and lipid microdomains of plant cells, but compared to animal and yeast cells, little is known about the role of GlcCer in plant physiology. We have investigated the influence of GlcCer biosynthesis by glucosylceramide synthase (GCS) on the efficiency of protein transport through the plant secretory pathway and on the maintenance of normal Golgi structure. We determined that GlcCer is synthesized at the beginning of the plant secretory pathway [mainly endoplasmic reticulum (ER)] and that d,l-threo-1-phenyl-2-decanoyl amino-3-morpholino-propanol (PDMP) is a potent inhibitor of plant GCS activity in vitro and in vivo. By an in vivo confocal microscopy approach in tobacco leaves infiltrated with PDMP, we showed that the decrease in GlcCer biosynthesis disturbed the transport of soluble and membrane secretory proteins to the cell surface, as these proteins were partly retained intracellularly in the ER and/or Golgi. Electron microscopic observations of Arabidopsis thaliana root cells after high-pressure freezing and freeze substitution evidenced strong morphological changes in the Golgi bodies, pointing to a link between decreased protein secretion and perturbations of Golgi structure following inhibition of GlcCer biosynthesis in plant cells.

Plant endomembranes contain two distinct classes of sphingolipids, glycosylated inositolphosphorylceramides (GIPCs) and glucosylceramides (GlcCers) (1,2), that are enriched in detergent-resistant membranes (DRMs) (3–5). As in other eukaryotes, the first step of sphingolipid biosynthesis in plants is catalysed by serine palmitoyltransferase (SPT), a heteromeric protein complex consisting of LCB1/LCB2 subunits (6). The essential functions of sphingolipids in plant tissues have been shown by the downregulation of LCB1 or LCB2 subunit in Arabidopsis, which results in a severe reduction in plant growth (6). Indeed, sphingolipid biosynthesis is essential throughout multiple stages of plant development, as shown by the defects in gametophyte development in Atlcb2 mutants (7).

We have recently determined the intracellular distribution of lipid microdomains in Allium porrum seedlings by preparing DRMs from endoplasmic reticulum (ER- ), Golgi- and plasma membrane-enriched fractions, and probed the involvement of sterols in their formation by treating plants with fenpropimorph (Fen), an inhibitor of cycloeucalenol cyclopropyl isomerase (CPI1) in the sterol biosynthetic pathway (8,9). Whereas most DRMs were normally isolated from the plasma membrane (PM) in control plants, they were found more abundant in the Golgi of Fen-treated plants (5). Treatment with Fen triggered a drastic shift in detergent resistance towards the Golgi membranes versus PMs, which reflected a redistribution of lipid microdomains in the secretory pathway upon a reduction in sterol biosynthesis. Another observation was that h16:0-GlcCer accumulated in Golgi DRMs after treatment with Fen, indicating that together with sterols, some GlcCer species may also be critical for membrane trafficking (5). It was shown that modification of VLCFAs (Very Long Chain Fatty Acids) altered membrane trafficking in Arabidopsis, possibly by modifying specific sphingolipid pools (10).

An extensive fenestration of the Golgi cisternae takes place in Fen-treated A. porrum seedlings (9). In addition, downregulation of LCB2 in Arabidopsis leads to a lack of deposition of the intine layer in pollen, which is correlated with a destructuring of the ER and Golgi membranes in mutant pollen (7). Such changes in the morphology of endomembranes may affect the steady-state membrane flux to the PM.

The accumulation of sterols and GlcCer in DRMs from Golgi membranes in Fen-treated A. porrum plants was correlated with a decrease of the proteins PMA2 and PIP1 in the PM (5). In a similar context, an elo3-erg6 double mutant (blocked in C26-sphingolipid and ergosterol synthesis) was found to be lethal because of a defect in the association of the H+-ATPase Pma1p with lipid microdomains (11). Additional data indicate that C26-sphingolipids may even be more critical than ergosterol for Pma1p sorting to the yeast PM (12). Finally, it was reported that sorting of melanosomal proteins in the Golgi of mammalian cells requires glycosphingolipids (13).

These results strongly argue in favour of an active requirement of sphingolipid biosynthesis for the normal functioning of the secretory pathway in eukaryotic organisms. The lack of studies in plants addressing this issue prompted us to develop a study to determine whether the biosynthesis of GlcCer is required for successful protein transport in plant cells. Our findings show the following:

  • The enzyme responsible for GlcCer biosynthesis [glucosylceramide synthase (GCS)] in Arabidopsis thaliana is located at the beginning of the secretory pathway, mainly the ER.
  • PDMP (d, l-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol) is a potent inhibitor of plant GlcCer biosynthesis both in vitro and in vivo.
  • The inhibition of GlcCer biosynthesis by PDMP induces critical changes in the morphology of the plant Golgi bodies.
  • The inhibition of GlcCer biosynthesis by PDMP leads to a partial retention of secretory proteins in the ER and Golgi membranes.
  • Overexpression of protein cargo in the secretory pathway of tobacco leaf epidermal cells increases the de novo biosynthesis of GlcCer.

As it is the case with sterol biosynthesis (5), our data indicate that GlcCer biosynthesis contributes to the maintenance of Golgi morphology and the transport of secretory proteins through the plant secretory pathway.

Results and Discussion

GCS is located in the ER

The aim of this study was to investigate whether the GlcCer biosynthesis by GCS is required for efficient protein secretion in plant cells. The question is only justified if indeed GluCer is synthesized at the beginning of the secretory pathway. To determine the subcellular location of GCS, we cloned the A. thaliana cDNA coding for the GCS (At2g19880) and produced a GCS with yellow fluorescent protein (YFP) fused at its C-terminal, which we expressed in tobacco leaf epidermal cells together with Bet11-RFP, a Golgi SNARE protein. A reticulate pattern typical of the ER was found in all the cells expressing GCS-YFP, whereas no colabelling was observed with Bet11-RFP (Figure 1).The same ER location was found for the GCS alone at different levels of expression. This localization differs from the Golgi localization of the animal GCS (14). In Drosophila, GCS GlcT-1 localizes to both the Golgi and the ER (15). The targeting signal for ER localization of GlcT-1 is not clear, but the Golgi targeting signal is located at the N-terminus. We produced a GCS construct with YFP fused at its C-terminus so that the presence of a potential Golgi targeting signal would not be masked. The presence of a dilysine (KK) motif at the C-terminus of the plant GCS reinforces the notion that the plant GCS is ER localized. The localization of the plant GCS in the ER is consistent with the localization of other sphingolipid biosynthetic enzymes such as the SPT subunit LCB1, and the ceramide synthases LOH1–LOH3 within ER membranes in plant cells (6,16); and supports the conclusion that GlcCer biosynthesis by GCS takes place at the beginning of the plant secretory pathway. Because GCS is located in an early compartment of the plant secretory pathway, and involved in the biosynthesis of GlcCer found in lipid microdomains (3–5), it became legitimate to question whether the activity of this protein was somehow related to protein trafficking.

Figure 1.

GCS is located in the ER. Representative expression pattern of GCS-YFP in tobacco leaf epidermal cells 48 h after cotransformation with A. tumefaciens carrying either GCS-YFP (green panel) or Bet11-RFP (magenta panel) expression vectors. Overlay is shown in the rightmost panel. Bars = 16 µm.

Tobacco leaf epidermal cells as a model to study lipid impact on protein secretion

We aimed to verify and validate the suitability of tobacco leaf epidermal cells infiltrated with inhibitors as a reliable system to analyse the relation between GlcCer biosynthesis and protein transport. Recently, we showed that a reduction in the biosynthesis of sterols, induced by Fen, triggers an accumulation of lipid microdomains in Golgi membranes of A. porrum seedlings (5), and we took advantage of this well-known effect of Fen on the plant secretory pathway (5,9) to evaluate the protocol for inhibiting lipid synthesis in intact mature tobacco leaves. We determined that CPI1 [the main target of Fen; (8,9)] is targeted to the ER membranes of tobacco leaf epidermal cells (data not shown), as in Arabidopsis(17). Fen (up to 500 µm) did not disturb the biosynthesis of other lipids such as monogalactosyl diacylglycerol (MGDG; data not shown), and therefore we calculated sterol content using MGDG as a reference, to take into account the variation in the total amount of lipids between leaves and the variation between thin layer chromatography (TLC) plates. We measured the effect of Fen on sterol biosynthesis 48 h after infiltration of the drug into tobacco leaves. After lipid extraction and analysis using TLC, we determined that 50 µm and 500 µm Fen induced a dose-dependent decrease in the content of sterols in tobacco leaves (data not shown). To determine whether the tobacco leaf epidermal cells could be used to investigate the effect of lipid metabolism on protein transport, we expressed N-SecYFP [a secretory version of YFP; (18)] in the absence or presence of 500 µm Fen. The calculated ratio of mean fluorescence intensity at the cell surface to that of the whole cell indicated a 30% decrease in N-SecYFP secretion when Fen was infiltrated together with Agrobacterium tumefaciens carrying the secretory protein expression vector (data not shown). This decrease in N-SecYFP secretion to the apoplast was in agreement with the effect of Fen observed in leek and A. thaliana seedlings (5), and clearly indicated that tobacco leaf epidermal cells represented an appropriate model for testing the effect of GlcCer biosynthesis on protein secretion through an in vivo confocal microscopy approach.

PDMP inhibits specifically GlcCer biosynthesis both in vitro and in vivo

GCS catalyses the transfer of glucose from uridine diphosphate (UDP)-glucose to ceramide. Two main classes of compounds are used to inhibit GCS activity, either by mimicking the UDP-glucose or the ceramide. Inhibitors that mimic the UDP-glucose are represented by iminoalkylated sugar analogs such as NB-DNJ (N-butyldeoxynojirimycine) (19). The synthetic ceramide analogs (20) include P4 (d-threo-1-(3′,4′-ethylenedioxy)phenyl-2-palmitoylamino-3-pyrrolidino-1-pro panol) and the isomers PDMP and PPMP (d,l-threo-1-phenyl-2-hexa decanoylamino-3-morpholino-1-propanol). PDMP contains n-acyl phenyl groups and a morpholine ring analogue to the ceramide fatty acyl chains and competes with ceramide on the formation of the charged transition state of the enzyme/UDP-glucose/ceramide complex (19). PDMP and its homologues are considered more specific and efficient than the glucose analogs (19), and PDMP is a strong inhibitor of the purified animal enzyme (21).

A previous study had shown that the inhibitor P4 does not affect non-human GCS enzymes in in vitro assays (22), and therefore we focused on testing only PDMP and PPMP.

To test the effect of the inhibitors on plant GCS activity, we developed an in vitro assay using tobacco leaf microsomal membranes. A typical in vitro assay was performed using 30 µg of microsomal protein and 8 µm UDP-glucose, without added ceramide. Such conditions provided a linear increase in the amount of GlcCer synthesized over the time of the assay. We were unable to obtain a consistent inhibitory effect on GCS activity when PPMP was used (data not shown). PDMP at 50 µm significantly decreased GlcCer biosynthesis with time (data not shown). To assess the specificity of PDMP, we measured the activity of related biosynthetic enzymes. Under our assay conditions, we were able to measure a significant synthesis of MGDG (data not shown) that could be accounted by the fact that the plastidial MGDG synthase is able to use glucose instead of galactose and that the microsomal membrane preparations still contain significant amounts of plastid envelope membranes. The assay conditions also allowed the synthesis of steryl glucoside (SG) and acylated steryl glucoside (ASG), which provided, along with MGDG, internal negative controls in our assays (data not shown). In this assay, the syntheses of MGDG, SG and ASG were not affected by PDMP, while the synthesis of GlcCer was decreased by up to 50% over time by the addition of 50 µm PDMP (data not shown). The results suggest that PDMP is a specific inhibitor of GCS in vitro. To test the effect of PDMP in vivo, tobacco leaf epidermal cells were infiltrated with 50, 100 or 150 µm PDMP and 10 µCi of labelled acetate for 48 h. Although the higher concentrations of PDMP were to some extent more efficient in decreasing the amounts of GlcCer biosynthesized in leaves, a higher percentage of cells became apoptotic above 50 µm PDMP, as measured by the dynamics of the cells observed using confocal microscopy. Therefore, infiltrations of 50 µm PDMP were used in following experiments, which induced a 70–80% reduction in de novo GlcCer biosynthesis, as measured by the radiolabelling of lipids over the 48-h treatment (Figure 2A). The specificity of the drug was confirmed by following the synthesis of GlcCer and other sugar-containing lipids for 8, 24, 36 or 48 h after infiltration of leaves with 50 µm PDMP. A decrease of about 30% in the content of GlcCer in leaves could already be observed 8–24 h after infiltration with PDMP (p < 0.01), reaching 60% after 36–48 h (p < 0.001) (Figure 2B). As observed in vitro, the content of MGDG (Figure S1C) and SG (Figure S1D) remained unaffected in vivo. In addition, we did not find any variation in the total amounts of the different phospholipids (data not shown).

Figure 2.

Inhibition of GCS activity in vivo by PDMP. A) De novo synthesis of radiolabelled GlcCer in tobacco leaves infiltrated with 10 µCi of (1-14C acetate, 58 Ci/mmol) and with or without 50 µm PDMP; the controls were infiltrated with 0.5% DMSO. Lipids were extracted and analysed as described in the experimental section. The amount of labelled GlcCer was reported to that of labelled MGDG to take into account the variations of acetate incorporation between leaves. Results are means ± SD of three independent experiments. B) GlcCer content of tobacco leaves 8–24, and 36–48 h after infiltration with 50 µm PDMP, or 0.5% DMSO (control, C). GlcCer content of control leaves did not vary from 8 to 48 h and results are grouped in one column. Lipids were extracted and analysed as described in the experimental section. The content of GlcCer is expressed as nanogram per milligram of fresh leaf weight (FW), and represents the mean ± SD of three independent experiments.

Finally, we checked whether the biosynthesis of the complex sphingolipid GIPC could also be affected by treatment with PDMP. For this, the spot corresponding to GIPC was recovered from TLC plates developed using a solvent system to separate polar lipids (23), and long chain bases (LCBs) were analysed (24). HPLC analyses of the GIPC-derived LCBs showed that the trihydroxy species t18:0 and t18:1(8E) represented nearly 90% of the total LCB (Figure S1A), which is a typical composition for plant GIPC (1). LCBs derived from radiolabelled GIPC did not vary significantly upon PDMP treatment (Figure S1B), indicating that PDMP was specific to GlcCer biosynthesis.

Effect of inhibiting GlcCer biosynthesis on the morphology of the Golgi

The Golgi apparatus is considered a key protein sorting station in the secretory pathway. We found in a previous study that the inhibition of sterol biosynthesis by Fen induced a fenestration of the Golgi apparatus in leek cells (9), suggesting a prominent role for sterols in Golgi maintenance in plant cells. The association of sterols and GlcCer in membranes prompted us to examine whether the GlcCer biosynthesis was required for the maintenance of Golgi morphology. The examination of Golgi in infiltrated tobacco leaves was hindered by the difficulty to fix the extensively vacuolated epidermal cells correctly enough to allow observation of Golgi structure. Therefore, we focused on root cells from A. thaliana seedlings. We first confirmed that PDMP was able to inhibit GlcCer biosynthesis in vivo in Arabidopsis seedlings to a similar extent (50–60%), as it was determined for tobacco leaf epidermal cells. Ultrastructure was analysed in epidermal and cortical cells of the division zone and the beginning of the elongation zone of the root. Golgi bodies with normal cisternae organized into typical Golgi stacks were observed in the control root cells (Figure 3A). In PDMP-treated root cells, the morphology of the Golgi bodies appeared severely affected, mainly characterized by modified cisternal producing a misaligned stack, often surrounded by swollen vesicles (Figure 3B). Morphometric analyses revealed that the thickness of the stack of the Golgi bodies (cis to trans distance) was unaffected, but the cisternal thickness as well as the stack width were significantly decreased in the PDMP-treated root cells (p < 0.01 and p < 0.05, respectively) (Figure 3C). In addition, the number of vesicles/vesicular structures in the vicinity of the remaining Golgi stacks in PDMP-treated root cells was increased (p < 0.05). The data strongly suggest that in plant cells, the Golgi partially disaggregates into vesicular structures upon inhibition of GlcCer biosynthesis by PDMP.

Figure 3.

Effect of PDMP on Golgi morphology. Typical appearance of Golgi bodies in control Arabidopsis seedlings that were incubated for 48 h in culture medium (A) or treated with 10 µm PDMP in culture medium (B). Bars in (A) and (B) = 100 nm. C) Morphometric quantification of the Golgi bodies as determined with the ImageJ software.

Effect of inhibiting GlcCer biosynthesis on the secretion of marker proteins

To investigate the effect of GlcCer synthesis on protein secretion, we transformed tobacco leaves with A. tumefaciens carrying the plasmid for expression of the soluble secretory protein N-SecYFP or the integral membrane protein marker PMA4 [an H+-ATPase of the PM; (25)], in the absence or presence of 50 µm PDMP. Fluorescence of protein markers was quantified using Metamorph software in 20–25 cells per condition (+ or – PDMP) per experiment, and the experiments were repeated at least four times.

Under normal conditions (in the absence of PDMP), N-SecYFP was highly secreted to the apoplast (Figure 4A), as it has been observed previously (18). In contrast, the fusion protein was partly retained in the intracellular membranes in the presence of PDMP (Figure 4B). The ratio Fs/Fi of mean fluorescence intensity at the cell surface (Fs) to the mean fluorescence intensity of intracellular membranes (Fi) shows clearly an important decrease in the presence of PDMP (Figure 4C, p < 0.01). In addition, the calculated ratio Fs/Ft of mean fluorescence intensity at the cell surface (Fs) to the mean fluorescence intensity of the whole cell (Ft) confirmed a 35–40% decrease in the fluorescence of N-SecYFP at cell surface (Figure 4C, p < 0.01). The decrease was accompanied by a concomitant 30–35% increase in the mean fluorescence intensity of intracellular membranes (Fi/Ft, Figure 4C, p < 0.05), indicating that the inhibition of GlcCer biosynthesis by PDMP effectively decreased the delivery of N-SecYFP to the cell surface.

Figure 4.

Effect of PDMP on N-SecYFP secretion to the apoplast. Representative fluorescence pattern in tobacco leaf epidermal cells 48 h after transformation with A. tumefaciens carrying the N-SecYFP expression plasmid in the (A) absence or (B) presence of 50 µm PDMP. Bars in (A) and (B) = 20 µm. C) Quantification of the mean fluorescence intensity at the cell surface Fs relative to that of the fluorescence intensity in intracellular membranes Fi (Fs/Fi) and to that of the whole cell Ft (Fs/Ft), and quantification of the mean fluorescence intensity in intracellular membranes Fi relative to that of the whole cell Ft (Fi/Ft).

We then examined the effect of GluCer synthesis on the transport of the membrane protein PMA4. When PMA4 was expressed in the absence of PDMP, the fluorescence was present only at the surface of the cell (Figure 5A), as expected for a PM marker. However, PDMP-treated cells presented, in addition to the normal PM labelling, fluorescent punctae reminiscent of Golgi bodies (Figure 5B,C). In addition, an ER-like pattern could also be observed (data not shown), although less frequently. To identify the nature of the punctate labelling in PDMP-treated PMA4 cells, we coexpressed PMA4 together with Bet11, a Golgi SNARE marker (18,26). Bet11 was chosen as a Golgi marker because it has been shown to label the trans Golgi network in protoplasts from Arabidopsis suspension-cultured cells (26), without interfering with the ER-Golgi traffic of membrane proteins (18), and also because it does not cycle between the ER and the Golgi as the K/HDEL receptor ERD2 or the animal sialyltransferase (ST) (27). Under normal expression conditions, PMA4 was found in the PM as expected, and Bet11 was found in typical dots corresponding to Golgi bodies [Figure 5D; (18)]. However, in the presence of PDMP, PMA4 colocalized with Bet11, suggesting that PMA4 was partially retained in the Golgi on its way to the PM (Figure 5E). The ratio Fs/Fi of mean fluorescence intensity at the cell surface (Fs) to the mean fluorescence intensity of intracellular membranes (Fi) shows clearly an important decrease in the presence of PDMP (Figure 5F, p < 0.01). The quantification of the mean fluorescence intensity of PMA4 at the cell surface (PM) compared to that of whole cells indicated a 30–35% decrease in PM mean fluorescence intensity (p < 0.05) and a simultaneous 25–30% increase in mean fluorescence intensity in intracellular membranes (p < 0.05) (Figure 5F), indicating that the inhibition of GlcCer biosynthesis reduced the transport of PMA4 to the PM. This is a similar situation as that observed in animal cells, where an absence of GlcCer led to a transport defect of membrane proteins at the Golgi (13). In a complementary approach, we performed a kinetic of expression of PMA4-GFP in the absence or presence of PDMP, and measured the amounts of the fusion protein by western blot using antibodies against PMA4 in Golgi and PM fractions prepared from the infiltrated tobacco leaves (Figure S2). Two independent experiments clearly confirmed that a considerable portion of fusion protein was retained in the Golgi fraction after 24 and 48 h of PDMP treatment.

Figure 5.

Effect of PDMP on PMA4 transport to the PM. Representative fluorescence patterns in (A) control cells or (B–C) PDMP-treated cells are presented. PMA4-GFP together with Bet11-RFP in tobacco leaf epidermal cells:(D) control or (E) PDMP-treated. Bars in (A) and (B) = 20 µm, bar in (D) = 5 µm and bar in (E) = 15 µm. F) Quantification of the mean fluorescence intensity at the cell surface Fs relative to that of the fluorescence intensity in intracellular membranes Fi (Fs/Fi) and to that of the whole cell Ft (Fs/Ft), and quantification of the mean fluorescence intensity in intracellular membranes Fi relative to that of the whole cell Ft (Fi/Ft).

In an attempt to bypass GCS, we tried to rescue GlcCer biosynthesis in PDMP-treated tobacco leaves by the addition of glucosylsphingosine as performed in Sprong et al. (13). To evaluate the feasability of such an approach in plant cells, we infiltrated tobacco leaves with PDMP, 10 µCi of labelled acetate, and with and without glucosylsphingosine for 48 h to measure de novo lipid synthesis. Unfortunately, we were not able to compensate the effect of PDMP on GlcCer biosynthesis by the addition of glucosylsphingosine. This result can be explained either by a low diffusion of glucosylsphingosine, which appears unlikely compared to the penetration of PDMP, or by the fact that glucosylsphingosine is not taken as a substrate in our system.

Contrary to PMA4, Bet11 did not seem affected by the PDMP treatment (Figure 5E). As ceramides and GlcCer are synthesized in the ER [(16); Figure 1], we examined the effect of PDMP on ER-to-Golgi transport with ERD2 and ST markers (18,27). Like Bet11, both markers appeared to be unaffected by PDMP under our experimental settings, using a confocal microscopy approach (Figure S3A–D).

In animal cells, transit of vesicular stomatitis virus-G (VSV-G) membrane protein from the ER to the cis Golgi was delayed 1.5-fold after treatment with 75 µm PDMP (28), suggesting the need for a certain level of GlcCer biosynthesis in the ER-to-Golgi transit of a membrane protein destined to the cell surface. However, unlike N-Sec and PMA4, ERD2 and ST were not affected in their transport to the Golgi, pointing to a lesser need for GlcCer in the ER-to-Golgi transport of proteins destined to the Golgi in plant cells, compared to proteins ultimately destined to the cell surface.

Overexpression of secretory protein cargo induces an increase in GlcCer and sterol biosynthesis

The observed intracellular retention of secretory proteins in tobacco leaf epidermal cells was positively correlated to the inhibition of GlcCer biosynthesis by PDMP. As GlcCer was necessary for the transport of proteins to the cell surface, we investigated whether an increase in the load of secretory proteins in plant cells was also accompanied by an increase in GlcCer biosynthesis. To test such a correlation, we overexpressed either the soluble secretory N-SecYFP or the membrane PMA4, and compared them to the overexpression of ERD2 [which does not traffic beyond the Golgi; (18,27)] and to the non-secretory cytosolic protein ranBP1 [the Ran-binding protein RanBP1a; (29)]. Lipids normally present in the membranes of the secretory pathway were analysed and quantified after radiolabelled acetate incorporation during the time of overexpression (i.e. 48 h). The results are presented in Figure 6. We did not find any increase in the amounts of phospholipids (Figure 6A), SG (Figure 6C) or free fatty acids (Figure 6E), upon overexpression of any of the four proteins tested. A significant twofold increase in the biosynthesis of GlcCer was observed when the secretory N-Sec and the integral membrane PMA4 proteins were overexpressed (Figure 6B). A similar increase in sterol levels was observed during N-Sec and PMA4 overexpression, although a lower increase could also be seen with ERD2 and RanBP1 (Figure 6D). The data strongly support the proposal that secretion of an increased load of the secretory proteins N-Sec and PMA4 beyond the Golgi induced a de novo biosynthesis of sterols and of GlcCer in particular. It has been recently shown that the de novo formation of ER export sites is induced by membrane cargo (30). Our results indicate that delivery of secretory proteins beyond the ER/Golgi also requires additional biosynthesis of specific lipids involved in the secretory pathway (5). Unlike N-Sec and PMA4, overexpression of ERD2 did not stimulate GlcCer biosynthesis in tobacco leaves (Figure 6B), again pointing to a lesser need for GlcCer in the ER-to-Golgi transport of proteins as seen in Figure S3A–D.

Figure 6.

Overexpression of secretory proteins induces an increase in GlcCer and sterol biosynthesis. Tobacco leaf epidermal cells were incubated with radiolabelled acetate for 48 h. The acetate solution was either infiltrated alone, or with Agrobacterium carrying N-Sec, PMA4, ranBP1, or ERD2 expression constructs to examine the effect of protein overexpression on de novo lipid biosynthesis in the plant secretory pathway. A) Labelling in phosphatidyl choline (PC), phosphatidyl serine (PS) and phosphatidyl ethanolamine (PE). B) Labelling in GlcCer, p < 0.01 in control versus N-Sec or PMA4. C) Labelling in steryl glucoside. D) Labelling in sterols, p < 0.01 in control versus N-Sec or PMA4, p < 0.05 in control versus RanBP1 or ERD2. E) Labelling in free fatty acids (FFAs).

Altogether, our results clearly point out a requirement of GlcCer biosynthesis for post-Golgi targeting of proteins. This is in agreement with a previous study in animal cells, which was the first to show that a defect in GlcCer induced an accumulation of transported membrane proteins in the Golgi (13). RNAi suppression of the Arabidopsis LCB2a and LCB2b genes has been shown to induce strong defects in the endomembrane system of pollen (7). Disturbing the metabolism of phosphatidic acid (31) or sterols (9) also leads to critical structural modifications in the morphology of the Golgi. Several studies in various eukaryotes have similarly correlated sphingolipid and/or sterol metabolism with protein assembly, transport and endomembrane dynamics (5,7,9–13,32), and it was recently proposed that GlcCer may exert specific functions at the cytosolic surface of the ER and Golgi membranes in protein sorting and transport events in animal cells (33 and references therein). It was suggested that such functions may be strongly regulated by the glycosphingolipid transfer protein FAPP2 (33), controlling GlcCer concentration in membrane domains. Recently, the A. thaliana glycosphingolipid transfer protein (GLTP1) was found to regulate vesicular transport to the PM/tonoplast during root hair development (34), probably by supplying GlcCer to membrane microdomains. When placed in the context of previous reports, our data support the conclusion that GlcCer biosynthesis is essential in the maintenance of Golgi morphology, membrane dynamics and protein trafficking in plant cells.

Material and Methods

Cloning of GCS and CPI1

A cDNA clone of A. thaliana GCS (At2g19880) was obtained from Riken (Japan). cDNA of CPI1 (At5g50375, 43) was obtained from RNA isolated from roots and leaves of 3-week-old A. thaliana seedlings. RNA was purified with the RNeasy Plant Mini Kit (Qiagen) and reverse transcribed with the StrataScript First-Strand Synthesis System (Stratagene). CPI1 cDNA was amplified with the primers 5′-atg tca gga tct tct tca ccg agc ttg-3′ and 5′-tca att gga gaa cca tgg taa tcc aga ctg-3′. The polymerase chain reaction (PCR) product was purified with the QIAquick Gel Extraction Kit (Qiagen), and cloned into pGEM-T Easy (Promega) in the Escherichia coli TOP10F’ strain. Plasmids were then purified with the GenElute Plasmid Mini Prep Kit (Sigma), and controlled by digestion with EcoRI and sequencing.

Sequences of GCS and CPI1 were reamplified with specific primers to add the restriction sites XbaI (5′) and SalI (3′) and suppress the stop codons. The PCR products were purified and cloned as described above. The resulting plasmids were digested with XbaI and SalI, and the sequences were introduced into the expression vector pVKH18En6-YFP in which the YFP is fused to the C-terminus of the GCS or CPI1 constructs. E. coli DH5α strains were transformed with the expression vectors to produce the plasmids used to transform the A. tumefaciens strain GV3101 used in the infiltration experiments. Additional constructs used in this study were ERD2-GFP and Bet11-RFP from Ref. (18), N-SecYFP from Ref. (35) and PMA4-GFP kindly provided by B. Lefebvre (25).

Plant material and transient expression systems

Four-week-old tobacco (Nicotiana tabacum cv. Xanthi) greenhouse plants grown at 22–24°C were used for A. tumefaciens-mediated transient expression (35,36). A. tumefaciens was cultured at 28°C, until the stationary phase (approximately 24 h), washed and resuspended in infiltration medium [50 mm MES, 0.5% (w/v) glucose, 2 mm Na3PO4, 100 µm acetosyringone (Aldrich) pH 5.6]. The bacterial suspension was inoculated using a 1-mL syringe without a needle by gentle pressure through a <1-mm hole punched on the lower epidermal surface (36). Transformed plants were incubated under normal growth conditions for 2 days at 22–24°C.

Confocal microscopy and fluorescence quantification

Transformed leaves were analysed 48 h after infiltration. Confocal imaging was performed using a Leica TCS SP2 confocal microscope fitted with a 63× oil immersion objective, numerical aperture (NA) 1.4. For imaging expressions of green fluorescent protein (GFP) and YFP constructs, excitation lines of an argon laser at 488 and 514 nm were used alternately with line switching using the multitrack facilities of the microscope. Imaging settings were as described (18,27,36), and postacquisition image processing was performed with the Leica LCS and ImageJ software. For fluorescence quantification, 25–30 z-sections were collected at 1.2–1.5 µm intervals using identical acquisition settings between control and treated cells and with 1 airy unit pinhole. Following acquisition, the background noise was subtracted from each image using a background Region Of Interest (ROI) and processed by applying a 5 × 5 pixel median filter to reduce high-frequency noise. Then an arithmetic projection was made of the 3D stack of images. The image was then thresholded for light objects and outline/cut tools were used to measure separately the whole cell and the section corresponding to the cell surface. Average fluorescence intensity, total fluorescence and cell area were measured for each cell using Metamorph's Integrated Morphometry Analysis (BIC platform, Université V. Segalen Bordeaux 2). No significant variation of cell area and total fluorescence per cell were observed under our experimental conditions (data not shown).

Transmission electron microscopy (TEM) and morphometric analyses

Thirteen-day-old A. thaliana seedlings were incubated for 48 h in liquid 1 / 2 Murashige & Skoog media (MS media) supplemented with 10 g/L sucrose. PDMP at 10 µm (PDMP-treated roots) or an equivalent volume of DMSO (control roots) was added to the media. Small root sections were flash-frozen at 2000 bars using an EM-Pact High Pressure Freezer (Leica Microsystems) and the samples were kept in liquid nitrogen. The root sections were transferred to vials containing substitution medium, consisting of 2% osmium tetraoxide in anhydrous acetone precooled to −90°C. Substitution was performed in an Automated Freeze Substitution System (Leica), with temperature increase at a rate of 4°C/h from −90°C to 0°C. Samples were rinsed with acetone and then maintained in 1% tannic acid-acetone medium. The acetone was progressively replaced by Epon resin over 48 h at room temperature and the samples were embedded in gelatine capsules, with polymerization carried out at 60°C for 24 h. Sections of 1–1.5 µm were coloured with toluidine blue to control the fixation of the samples. Ultra-thin sections (50 nm thick) were then produced and stained with 7% ethanolic uranyl acetate for 15 min, then with 1% aqueous lead citrate for 5 min.

Micrographs were taken at 80 kV on an FEI CM10 transmission electron microscope equipped with an AMT × 60 digital camera (Elexience). Morphometric analyses were performed on Golgi bodies (n = 50) from cortical and epidermal cells in the division/early elongation zone on three separately treated roots per treatment. Analysis was performed using ImageJ software.

Isolation of microsomal membranes from N. tabacum leaves

Leaves from 4–6-week-old tobacco plants (N. tabacum cv. Xanthi), grown in a growth chamber at 22–24°C under 16/8 h day/night conditions, were homogenized in the presence of 10 mm KH2PO 4 (pH 8.2) with 0.5 m sorbitol, 5% (w/v) PVP40, 0.5% (w/v) BSA, 2 mm salicylhydroxamic acid and 1 mm phenylmethylsulphonyl f luoride (PMSF). After filtration, the homogenate was subjected to successive centrifugations at 1000 ×g for 10 min, 10 000 ×g for 15 min and 150 000 ×g for 90 min. The resulting microsomal pellet was suspended in the appropriate buffer for testing GlcCer biosynthesis in vitro. Protein concentrations were determined according to the Bradford procedure (37) using BSA as a standard.

In vitro enzymatic assays for GCS activity

Tobacco microsomal membranes were resuspended with 500 µL of Tris–HCl buffer (pH 7). GCS activity was determined according to a modified protocol from Nakayama et al. (38) and Lynch et al. (39).

Microsomal membranes were incubated at 0.75 µg/mL in Tris–HCl buffer containing 2.5 mm KCl, 1 mm MgCl and 50 µm ATP, pH 7.0. The concentration of UDP-glucose [14C(U)] (200 mCi/mmol) was set at 2 µm and the total concentrations of UDP-glucose were adjusted from 2 to 128 µm with non-labelled UDP-glucose. Incubations at different UDP-glucose concentrations were for 90 min at 30°C, and the incubations as a function of time used 30 µg of protein and 8 µm of UDP-glucose. Reactions were stopped by adding 2 mL of chloroform:methanol (1:1, v/v).

For measuring the effect of PDMP on GCS activity, a 10-mm stock solution of PDMP in DMSO was used, and final concentrations of DMSO of 0.5% were used in all assays. DMSO alone at this concentration (and up to 1%) did not perturb the biosynthesis of GlcCer by GCS.

In vivo labelling of lipids in tobacco leaf epidermal cells

Ten microcurrie of [1-14C] acetate (58 Ci/mmol; Amersham) was dissolved in 500 µL of infiltration medium, and leaves were infiltrated as described earlier.

The labelled acetate solution was either infiltrated alone, with Agrobacterium carrying the plasmid constructs to investigate the effect of protein overexpression on de novo lipid biosynthesis in the plant secretory pathway, or in combination with different concentrations of PDMP to determine the effect of the drug on de novo lipid biosynthesis. As indicated earlier, PDMP was prepared from a stock solution in DMSO, and concentrations of DMSO not exceeding 0.5% were used in all experiments.

Lipid analyses

Lipids from microsomal membranes were extracted by chloroform:methanol (2:1, v/v) for 30 min at room temperature, and then washed 3× with 0.9% NaCl. The solvent was evaporated and lipids were dissolved in an appropriate volume of chloroform:methanol (1:1, v/v).

Lipids from tobacco leaves were extracted by propan-2-ol/hexane/water (55:20:25, v/v) as in Ref. (1). Polar lipids were analysed by HPTLC (60F254 plates, Merck), using the solvent system A: methyl acetate:n-propanol:chloroform:methanol:0.25% aqueous KCl (25:25:25:10:9, v/v) according to Ref. (23). Under these conditions, GlcCer and SG do not separate. These two lipid species were separated by HPTLC using solvent system B: chloroform:methanol (85:15, v/v) (40). For analysis of radiolabelled GIPC-derived LCBs, the origin was recovered from the HPTLC plates developed in solvent system A, hydrolysed by incubating in 4 n HCl in methanol for 16 h at 70°C. The products of hydrolysis were developed using solvent system C: chloroform:methanol:2 m NH4OH (40:10:1, v/v) and analysed using a Storm PhosphorImager (GE Healthcare) and ImageQuant software (Applied Biosystems). Neutral lipids were analysed by HPTLC using solvent system D: hexane:ethylether:acetic acid (90:15:2, v/v) (5). Lipids were identified by comigration with known standards and quantified by densitometric analysis (5,41) using a TLC scanner 3 (CAMAG) after primuline staining (42).

The different families of lipids had to be analysed on different HPTLC plates. To take into account any variation between plates, several known amounts of the same standards were deposited on each plate before staining and the same lipid was measured on different solvent systems.

LCB separation, derivatization and separation by HPLC were performed as described previously (24).

Acknowledgments

We thank the IFR 103 (IBVM, INRA Bordeaux) and the Bordeaux Imaging Centre (BIC), University Victor Segalen Bordeaux 2 for the confocal microscope facilities at the Plant Imaging Platform and the Photonic Imaging Platform. We also thank C. Cheniclet and V. Rouyere (Plant Imaging Platform of BIC) for their help with transmission electron microscopy (TEM). We thank J. J. Bessoule for critical reading and helpful discussions. This work was financed by University Victor Segalen Bordeaux 2, CNRS and the Conseil Régional d’Aquitaine. S. M. is the recipient of a PhD fellowship from ANR (BLAN07–1_182875), France

Ancillary