Introduction

In eukaryotic organisms, programmed cell death (PCD) is an important mechanism to control normal growth and development as well as defense responses to a variety of biotic and abiotic stresses. In plants, PCD is essential for a number of developmental processes. The best-known examples include the specification of unisexual floral organs, the formation of the tracheary elements and senescence 1, 2. Moreover, an impressive body of evidence obtained from studies on plant-pathogen interactions, particularly those derived from genetic studies on Arabidopsis, has provided substantial insight into the PCD mechanism in plant cells 3, 4, 5, 6, 7. One of the most commonly observed PCD forms triggered by pathogen infection is the so-called hypersensitive response (HR), which involves the formation of reactive oxygen intermediates (ROIs) and nitric oxide, and largely relies on the salicylic acid signaling pathway 3, 8, 9. However, little is known about the biochemical mechanism of PCD in plant cells.

Sphingolipids are a diverse group of lipids that contain a relatively large hydrophobic moiety, known as ceramides that include a sphingoid or long-chain base (LCB) amide-linked to a fatty acid. Sphingolipids are not only essential components of cellular membranes, but also act as second messengers to regulate stress responses, cell proliferation and apoptosis 10, 11, 12, 13. De novo biosynthesis of sphingolipids is initiated by the condensation of serine and palmitoyl-CoA to produce 3-ketosphinganine (3-KDS) 10, 13. This reaction is catalyzed by serine palmitoyltransferase (SPT; EC 2.3.1.50), a heterodimeric complex consisting of two subunits of LCB1 and LCB2 localized in the endoplasmic reticulum (ER) 14. 3-KDS is first reduced to sphinganine or dihydrosphingosine (dh-sph) by 3-KDS reductase, and dh-sph is then converted into ceramides by (dihydro) ceramide synthase.

In higher plants, several studies suggested that fumonisin B1 (FB1) and Alternaria alternate lycopersici (AAL) toxin, two natural compounds that specifically inhibit ceramide synthase activity, cause apoptotic cell death in various species 15, 16, 17, 18, 19, 20. In Arabidopsis, FB1-induced PCD depends on multiple signaling pathways, including those of jasmonate, ethylene, and salicylic acid 19. More direct evidence was obtained from genetic studies. Arabidopsis ACD11 encodes a sphingosine (sph) transfer protein, and a mutation in this gene causes a lesion mimic phenotype characteristic of apoptosis 21. Similarly, mutations in ACD5 cause a spontaneous cell death phenotype 22. The ACD5 wild type (WT) allele encodes a ceramide kinase, and a recombinant ACD5 protein has high specificity for ceramides but not other sphingolipids 23. These observations suggest that the maintenance of sphingolipid homeostasis is important for properly regulating apoptosis in plant cells. The mechanism of sphingolipid-regulated PCD in plant cells, however, remains unclear.

In addition to a regulatory role in PCD, sphingolipids are also suggested to modulate stress responses. Recent studies revealed that sphingosine-1-phosphate (S1P) is involved in the regulation of the abscisic acid (ABA) pathway 24, 25, and S1P action appears to be mediated by physical interactions of a heterotrimeric G protein and a putative G protein-coupled receptor 25, 26. Although a direct link between the ABA pathway and PCD has not been shown, the fact that many of the phytohormone-regulated physiological activities such as seed development and stress responses may directly or indirectly involve PCD implies a possible interplay between the two pathways, in which S1P or other sphingolipids may act as critical mediators.

Here, we report the identification and characterization of the Arabidopsis fumonisin B1 resistant11-1 (fbr11-1) mutant, which displayed a reduced sensitivity to the apoptosis-inducing agent FB1. Molecular analysis indicated that FBR11 encodes an LCB1 of SPT. The FB1-induced PCD phenotype is correlated with massively increased concentrations of several sphingoid bases. A pharmacological approach showed that whereas dh-sph is a potent PCD inducer, its phosphorylated form dihydrosphingosine-1-phosphate (dh-S1P) acts specifically as an anti-apoptotic molecule. Moreover, the antagonistic effects of these compounds appear to be executed through the regulation of the ROI level, thereby controlling an apoptotic cell death program in plant cells.

Materials and Methods

Plant materials, growth conditions and genetic screen for fbr mutants

The Col-0 ecotype of Arabidopsis thaliana was used in this study. Unless otherwise indicated, plants were grown under a 16 h light/8 h dark cycle at 22 ºC in soil or on an MS medium 27 containing 3% sucrose and 0.8% agar.

A binary vector pER16 28 was used for the generation of T-DNA lines by standard methods 29, 30. Approximate 12 000 independent lines were screened for fbr mutants as described 20, 31. Briefly, pooled T2 seeds (20 lines/pool; approximate 10 seeds/line) were germinated and grown on MS medium containing 1 μM FB1 for 1–2 weeks. Putative fbr mutants were identified by visual inspection and then transferred onto a fresh MS medium without FB1.

Chemicals and treatment

FB1, paraquat and sphingoid bases (except phyto-S1P) were purchased from Sigma Inc., China and Hong Kong. Phyto-S1P was purchased from Avanti Polar Lipids (Alabaster, AL, USA). The following solvents were used for the preparation of stock solutions: water (paraquat), methanol (S1P), methanol/tetrahydrofuran/water (60/30/10%; v/v; phyto-S1P and dh-S1P) and dimethyl sulfoxide (DMSO; all others). Chemical treatment was carried out by directly germinating seeds on MS medium supplemented with appropriate chemicals or by spraying (2 ml per 90-mm Petri dish) plants germinated and grown on MS medium. The same concentrations of solvents were used as controls in all experiments. All experiments were repeated at least 5 times with commercial reagents and 2–3 times with homemade chemicals, and similar results were obtained. All data presented are representative results obtained by the use of commercial reagents.

Detection of cell death, superoxide, hydrogen peroxide and callose

Detection of superoxide, hydrogen peroxide and callose was performed by staining leaves with nitroblue tetrazolium (NBT), 3,3-diaminobenzidine (DAB) and aniline, respectively 32, 33, 34. Cell death was examined by Evans Blue staining as described 33 with minor modifications. Briefly, leaves samples were vacuum-infiltrated in 0.1% Evans Blue (w/v; Sigma) for 15 min and then maintained for 8 h under vacuum. After the staining, leaves were washed for three times (15 min each wash) with a phosphate-buffered saline containing 0.05% (v/v) Tween 20. All experiments were repeated at least 5 times, and at least 10 leaves collected from multiple seedlings (4-5-week-old) were inspected in each experiment. Penetration of the phenotype was usually higher than 80% (protoplast assay) and 90% (leaf assay) in these experiments, respectively.

Detection of nuclear DNA fragmentation in protoplasts was performed as previously described 31.

Analysis of sphingolipids

Measurement of sphingolipids was performed by Electrospray Ionization/Mass spectrum/Mass spectrum analysis on a Thermo Finnigan TSQ 7000 triple quadrupole mass spectrometer, operating in a Multiple Reaction Monitoring positive ionization mode as described 35. Briefly, samples collected and frozen in liquid nitrogen were dried by lyophilizing at −50 ºC, and then ground into fine powder, which was then fortified with the internal standards (IS; C17 base D-erythro-sphingosine, C17 sphingosine-1-phosphate, N-palmitoyl-D-erythro-C13 sphingosine, heptadecanoyl-D-erythro-sphingosine and C6-Phyto-ceramide). Fifteen milligrams of dried powder were extracted with the ethyl acetate/iso-propanol/water (60/30/10%; v/v) solvent system. After evaporation and reconstitution in 100 μl of methanol, the samples were injected on the Surveyor/TSQ 7000 LC/MS system and gradient-eluted from the BDS Hypersil C8, 150 × 3.2 mm, 3 μm particle size column, with 1.0 mM methanolic ammonium formate/2 mM aqueous ammonium formate mobile phase system. Peaks corresponding to the target analytes and internal standards were collected and processed using the Xcalibur software system. Quantitative analysis was based on the calibration curves generated by spiking an artificial matrix with the known amounts of the target analyte synthetic standards and an equal amount of internal standards. The target analyte/IS peak areas ratios were plotted against analyte concentration, which were normalized to their respective ISs and compared to the calibration curves, using a linear regression model.

Molecular manipulations

All molecular manipulations were carried out according to standard methods 36. The T-DNA tagged genomic sequence in the fbr11-1 genome was identified by Thermal Asymmetric Interlaced-PCR (TAIL-PCR) as previously described 37, 38. An FBR11 genomic clone was obtained by PCR using PWO DNA polymerase (Roche Diagnostics Hong Kong, Hong Kong). Primers used in PCR were LCBCOMF (5′ GGT CGA CGG GAG ATA GGA GGA AGA AGA CTG ATT GA 3′) and LCBCOMB (5′ CCC TAG GGG ATT CTC AAC TCC ATT AAC GTC GAG GT 3′). The FBR11 (At4g36480) genomic clone included a 1.5-Kb promoter sequence, starting from the 5′ UTR of At4g36470 (in a head-to-head configuration with FBR11), to ensure inclusion of the entire promoter sequence. The PCR product, digested with SalI and AvrII, was cloned into the XhoI and SpeI sites of pER8 39. The resulting construct was transformed into A. tumefaciens strain GV3101, which was used for transformation of Arabidopsis by vacuum infiltration 29. DNA Southern and RNA Northern blotting analyses were carried out as described previously 38.

Analysis of gene expression by real time-PCR was performed essentially as previously described 40. Primers used in the experiments described in Figure 3A and 3B were (all sequences are from 5′- to 3′-end):

Figure 3
figure 3

Molecular characterization of FBR11. (A) Schematic representation of the FBR11 gene. Exons and introns are represented by filled boxes and lines, respectively. Untranscribed and untranslated regions (UTR) are indicated as dashed lines. The insertion site and the orientation of the T-DNA insertion (arrow indicates the orientation of the left border or LB) in the fbr11-1 genome are shown. Insertion site for the right border (RB) is unclear (indicated by a question mark). UTR sequences were derived from the cDNA sequence (GenBank accession AY120759). (B) Quantative real-time RT-PCR analysis of FBR11 expression in WT and fbr11-1. RNA prepared from three-week-old seedlings germinated and grown on MS medium was used for the synthesis of the first strand cDNA using oligo-dT as a primer. PCR was carried out using primer pairs as shown in panel A (P1 through P4; the insertion site of LB was between P1 and P2). Relative expression levels were normalized to that of Actin7. Mean values of two independent experiments were shown in the graph. Bars represent standard errors. (C) Molecular complementation of the fbr11-1 phenotype. An FBR11 WT genomic DNA fragment (including the FBR11 promoter sequence) was cloned into a binary vector, and the resulting construct was transformed into fbr11-1. T2 transgenic plants were grown on MS medium supplemented with 1 μM FB1 for 2 weeks. From left to right in each panel: WT, fbr11-1, and fbr11-1 carrying an FBR11 transgene. Fourteen independent transgenic lines were tested, and similar results were obtained. Bar, 2 mm. (D) Rescue of the cell death and superoxide generation phenotypes in fbr11-1. T2 transgenic plants as described in panel C were treated with 0.02% methanol (left) or 2 μM FB1 (right) by spraying. Leaves were collected for assays of cell death (24 h) and superoxide accumulation (6 h) as described in Figure 1. Three independent transgenic lines were tested and similar results were obtained. Comp: complemented transgenic plants.

P1: GCA GAG TCA GTA GCT TGA AGA TGT

P2: CGG CAA CAA CAC TAA GCT ACT TGA

P3: ACC AGA GAC TTA GCA GTT CAG GA

P4: CAA TAG GTG ATT CCC GGT TGC TTG

Results

Identification and genetic analysis of the fbr11-1 mutant

The fbr11-1 mutant was identified in a genetic screen for fumonisin B1-resistant mutants as described 20, 31. Whereas the growth and development of WT plants were severely inhibited by FB1, the fbr11-1 mutant plant displayed substantial resistance to the toxin (Figure 1A). However, in the absence of FB1, the fbr11-1 mutant plant did not have detectable morphological alterations under different light or temperature conditions (Figure 1B and data not shown).

Figure 1
figure 1

The fbr11-1 mutant phenotype. (A) Two-week-old WT (left) and fbr11-1 seedlings germinated and grown on MS medium containing 1 μM FB1. (B) Two-week-old of the same plants germinated and grown on MS medium. (C) Cell death induced by FB1. Three-week-old WT and fbr11-1 plants were treated with methanol (0.1%; control) or 2 μM FB1 for 24 h by spraying. Leaves were then detached from treated plants and stained with Evans Blue. Dead cells were stained blue. (D) The deposition of callose induced by FB1. Plants were treated as described in (C) for 24 h, and leaves were harvested and stained with aniline. Callose deposition was visible as blue fluorescence. (E) The accumulation of superoxide induced by FB1. Plants were treated as described in (C) for 6 h, and leaves were collected and then stained with NBT. Superoxide accumulation was shown as blue precipitates. (F) The accumulation of hydrogen peroxide induced by FB1. Plants were treated as described in (C) for 6 h, and leaves were collected and then stained with DAB. Hydrogen peroxide accumulation was shown as dark brown precipitates. Bar, 2 mm (A and B).

In a cross between fbr11-1 and WT plants, all tested F1 progeny (33) were sensitive to FB1, indicating that the mutation was recessive. In F2 progeny obtained from self-pollinated F1 plants, the mutation segregated in a 1:3 ratio (FB1 resistant:sensitive =123:450, χ2=3.423; p<0.05), suggesting that the mutant phenotype was caused by a mutation in a single nuclear gene.

The programmed cell death phenotype of fbr11-1

Compared with massive cell death induced by FB1 in WT plants, essentially no cell death was detected in fbr11-1 (Figure 1C) as revealed by Evans Blue staining. This suggests that the mutant phenotype was likely caused by the failure of a toxin-evoked cell death program. Moreover, the FB1-induced cell death was accompanied by substantial deposition of callose in leaves of WT plants, which was not observed in leaves of fbr11-1 plants (Figure 1D). To further investigate the molecular basis of the fbr11-1 phenotype, we examined the accumulation of ROIs in both WT and fbr11-1 leaves under different conditions. As shown in Figure 1E, the generation of superoxide (stained by NBT) was strongly induced by FB1 in WT plants. By contrast, no superoxide accumulation was detected in fbr11-1 under identical assay conditions. Similar results were obtained when stained with DAB (staining for hydrogen peroxide) (Figure 1F). Most experiments described hereafter were carried out by both NBT and DAB staining, and similar results were obtained. For concise reasons, we only show NBT staining data, and collectively refer to these two species as ROIs. Data presented in this section suggest that the absence of cell death in the mutant was likely caused by the lack of ROI production.

Nuclear DNA fragmentation is a hallmark of apoptotic cells. To reveal whether nuclear DNA fragmentation occurred in FB1-treated cells, we employed the TUNEL method to examine WT and fbr11-1 protoplasts treated with or without the toxin (Figure 2). In untreated protoplasts derived from both WT and fbr11-1 leaves, each cell contained a single condensed nucleus as revealed by staining with Hoechst 33342, a non-specific DNA dye, and no DNA fragmentation was detected in these protoplasts. Upon treatment with FB1, significant DNA fragmentation was detected by TUNEL staining of WT protoplasts. By contrast, under the same conditions, fbr11-1 protoplasts showed a phenotype indistinguishable from that of untreated controls (Figure 2). The TUNEL positive signals were routinely detected in WT protoplasts within 8–10 h after exposure to FB1, whereas cell death was only detectable after a longer treatment of 22–24 h. On the other hand, in the leaflet assay, ROI generation could be detected 3–6 h after the FB1 treatment but was barely detectable within 3 h of the treatment. Cell death induced by the compound was again detected 22-24 h after the treatment. The above results suggested that both ROI generation and DNA fragmentation occurred prior to cell death, which are characteristics of a controlled apoptotic program rather than a necrotic cell death effect. Based on the data presented above, we concluded that the fbr11-1 mutant was incapable of initiating PCD induced by FB1, presumably resulting from the failure to generate ROIs under these conditions.

Figure 2
figure 2

Nuclear DNA fragmentation in WT and fbr11-1 protoplast induced by FB1. Protoplasts prepared from WT or fbr11-1 leaves were treated with methanol (0.005%; control) or 50 nM FB1 for 10 h. DNA molecules were then stained with Hoechst 33342 (blue signals in the upper panels), and 3′-OH groups were stained by the TUNEL method (orange signals in the lower panels). The experiment was repeated for three times (n>100 in each set of samples).

Molecular cloning of the FBR11 gene

Genetic analysis indicated that the fbr11-1 mutant genome contained a single T-DNA insertion (kanamycin resistant:sensitive = 246:85, χ2=0.04933; p<0.05). When transferred from an FB1 medium to a kanamycin medium, all 147 fbr11-1 seedlings showed resistance to the antibiotic, suggesting that the T-DNA was tightly co-segregated with the mutation. To identify the FBR11 candidate gene, TAIL-PCR 37 was used to isolate the genomic sequences flanking to the Left Border (LB) of the T-DNA. DNA sequencing analysis indicated that the LB was inserted in At4g36480 approximately 260 base pairs downstream from the stop codon (Figure 3A). PCR analysis indicated that the putative At4g36480 open reading frame (ORF) remained intact. However, a real-time quantative reverse transcription (RT)-PCR analysis revealed that the FBR11 expression was substantially reduced in fbr11-1 (Figure 3B).

To verify the identity of FBR11, we carried out a molecular complementation experiment. A 4.7-kilo base pairs (Kb) WT genomic DNA fragment, which encompassed the promoter region, the coding sequence and a portion of the 3′ UTR of At4g36480, was cloned into a binary vector pER8 39. The resulting construct was then transformed into fbr11-1 plants. The At4g36480 transgene was able to fully restore the sensitivity of fbr11-1 to FB1, including the growth arrest phenotype (Figure 3C), the cell death phenotype as well as the accumulation of ROIs (Figure 3D). These results indicate that At4g36480 represents the FBR11 gene.

FBR11 encodes an LCB1 subunit of serine palmitoyltransferase

FBR11 is a single copy gene in the Arabidopsis genome. A full-length FBR11 cDNA clone was identified from the database (accession number: AY120759). Comparison of the cDNA and genomic sequences revealed that FBR11 contains 12 introns (Figure 3A). An ORF within the FBR11 gene encodes a polypeptide of 482 amino acid residues, with a predicted molecular mass of 53.14 KDa and a pI of 7.23. Sequence comparison revealed that FBR11 encodes a putative LCB1 subunit of SPT, catalyzing the formation of 3-ketosphinganine during sphingolipid de novo synthesis (Figure 4). During the course of this study, Chen et al. 41 reported the characterization of LCB1/FBR11. Coexpression of LCB1/FBR11 and LCB2 cDNA was able to complement the long-chain base auxotrophy of S. cerevisiae lcb1Δ and lcb2Δ single and double mutants, indicating that both LCB1 and LCB2 are functional subunits of SPT. Moreover, lcb1-1, a stronger mutant allele than fbr11-1, was suggested to cause abnormal embryo development 41.

Figure 4
figure 4

A proposed pathway of plant sphingolipid de novo synthesis. Myriocin and FB1 are competitive inhibitors of SPT and ceramide synthase, respectively (Modified from 13).

Measurement of sphingolipids in Arabidopsis

The fact that FBR11 encodes an LCB1 subunit of SPT prompted us to ask whether sphingolipid metabolism has been altered by the fbr11 mutation. To address this question, we compared levels of several species of sphingolipids in fbr11-1 with those of WT plants. Total lipids were extracted from 3-week-old seedlings, and sphingolipids were then subjected to tandem Mass Spectrum (MS/MS) analysis using authentic standards. In WT plants, among the assayed sphingolipids, the most predominant sphingoid bases are phyto-sph and phytosphingosine-1-phosphate (phyto-S1P), followed by dh-sph and dihydrosphingosine-1-phosphate (dh-S1P) (Table 1). Similarly, the main ceramides also appeared to be the phyto- and dihydro-types. Notably, whereas phyto-C16-, phyto-C24- and phyto-C24:1-ceramides were the major species in Arabidopsis, phyto-C14, dhC16-, phyto-C20- and C24:1-ceramides as well as all C18-derivatives appeared to be less abundant (Table 1). We note that our measurements of dh-sph and phyto-sph levels were comparable with those of a previous study on tomato, duckweed and tobacco calli 17. Under our assay conditions, most non-phyto/dihydro-type sphingoid bases and ceramides were present only in trace amounts (Table 1). S1P, for example, was found to be approximately 18-fold less abundant than phyto-S1P. Again, in agreement with an earlier report on Commelina communis, in which S1P was estimated at 5-46 pg per g dry weight 24, the compound was approximately 10-40 pg per g dry weight in our assays.

Table 1 Measurement of sphingolipids in wild type and fbr11-1 mutant plants

Sphingolipid metabolism in fbr11-1

To monitor possible effects of the fbr11-1 mutation on ceramide metabolism, we compared several sphingolipid species in the mutant with those found in WT plants. It appeared that the mutation did not cause substantial changes among the tested sphingolipids (Table 1). We then reasoned that because fbr11-1 was resistant to FB1, a comparison of sphingolipids between the mutant and WT plants treated with the apoptosis-inducing compound might provide clues concerning the molecular mechanism of plant PCD. In both WT and fbr11-1 plants, FB1-treatment caused a massive increase of several major sphingoid bases (Table 1). Substantially altered levels in several species could be observed upon treatment with FB1 for 1 h (see Note 2 of Table 1). In WT plants, the most significant alterations were found in several sphingoid bases, including dh-sph (77-fold), phyto-sph (12-fold), phyto-S1P (162-fold) and dhS1P (144-fold). Under the same conditions, however, the FB1-resistant fbr11-1 mutant plants accumulated less of these sphingoid bases (23-, 6-, 53-, and 39-fold, respectively; Table 1). In contrast to those of free sphingoid bases, levels of ceramides were marginally altered in all cases, suggesting that FB1-induced PCD is likely related to the accumulation of sphingoid bases or an altered balance among these compounds. Moreover, compared to that of WT, the substantially altered sphingoid base levels in the mutant treated by FB1 suggest a role of FBR11 in sphingolipid metabolism.

Activation of PCD in Arabidopsis by free sphingoid bases

Altered sphingolipid levels in WT and fbr11-1 treated with FB1 suggested that sphingoid bases may play an important role in regulating PCD in Arabidopsis. To ask if these compounds are capable of triggering PCD in Arabidopsis, we treated WT plants with different sphingoid bases, followed by monitoring ROI generation and cell death in detached leaves. As shown in Figure 5, while dh-sph, phyto-sph and sph were able to induce ROI production and cell death, their phosphorylated derivatives, dh-S1P, phyto-S1P and S1P, did not have the PCD-inducing effect under the assay conditions. Moreover, the PCD-inducing effect of dh-sph, phyto-sph and sph was dose-dependent in both WT and fbr11-1 leaves (Figure 5B–5D). We reproducibly observed that dh-sph was less potent than phyto-sph, consistent with a previous finding that phyto-sph displays greater phototoxicity than dh-sph 42. The fact that WT and fbr11-1 responded to exogenous dh-sph, phyto-sph and sph in an indistinguishable manner suggested that the perception and subsequent amplification of a death signal were not affected by the mutation. We conclude from these data that free sphingoid bases, but not their phosphorylated derivatives, can trigger PCD in Arabidopsis, presumably by regulating ROI production.

Figure 5
figure 5

ROI generation and cell death induced by free sphingoid bases. (A) PCD-inducing effects of free sphingoid bases and their phosphorylated derivatives. WT Col-0 plants were treated with 1 μM of each compound as indicated on the top of the panel for 24 h (top row; cell death) or 6 h (bottom row; ROIs), respectively. See Figure 1 for other technical details. (B-D) Dose-dependent effects of free sphingoid bases on the induction of cell death and ROI production. WT Col-0 and fbr11-1 plants were treated with different concentrations of dh-sph (B), phyto-sph (C) or sph (D) as described in panel (A). The concentrations of the free sphingoid bases in the experiments are indicated on the top of each panel. See Figure 1 for other technical details.

Inhibition of dh-sph-induced ROI generation and PCD by dh-S1P

The above data indicated that the phosphorylated forms of sphingoid bases are incapable of inducing PCD in plant cells. Moreover, among the analyzed free sphingoid bases, dh-sph is the only compound that displayed an altered level in both FB1- and paraquat-treated plants (Table 1 and data not shown). Therefore, alterations of dh-sph concentration and/or the ratio between this sphingoid base and its phosphorylated form may play a role in the regulation of PCD in Arabidopsis. To test this hypothesis, we treated WT seedlings with 2 μM dh-sph combined with its phosphorylated form dh-S1P at different concentrations. Figure 6A shows that ROI generation and cell death induced by dh-sph were efficiently blocked by dh-S1P in a dose-dependent manner.

Figure 6
figure 6

Dh-S1P specifically blocks ROI generation and cell death induced by dh-sph. (A) Dh-S1P blocks ROI generation and cell death induced by dh-sph. WT Col-0 plants were treated with a solution containing 2 μM dh-sph and various concentrations of dh-S1P as indicated on the top of the panel for 24 h (top row) or 6 h (bottom row), respectively. See Figure 1 for other technical details. (B) Dh-S1P specifically inhibits ROI generation and cell death induced by dh-sph, but not by phyto-sph or paraquat. WT Col-0 plants were treated with a solution containing 1 μM dh-S1P along with 1 μM dh-sph, 1 μM phyto-sph or 3 μM paraquat as indicated on the top of the panel for 24 h (top row) or 6 h (bottom row), respectively. See Figure 1 for other technical details.

Whereas this result suggested an anti-apoptotic role of dh-S1P, an alternative possibility could be that ROI generation and cell death might also be regulated by the ratio of a free sphingoid base and its phosphorylated form. To distinguish these two possibilities, we then tested if dh-S1P also had a protective effect against ROI generation and cell death induced by other treatments. When treated by phyto-sph or paraquat, neither ROI production nor cell death could be reduced by dh-S1P (Figure 6B). These observations suggest that the balance between dh-sph and its phosphorylated form dh-S1P plays an important role in the control of ROI-modulated PCD in Arabidopsis.

Discussion

In this study, we have presented genetic, biochemical and molecular evidence showing that FBR11 encodes an LCB1 of SPT, and sphingoid bases are directly involved in the regulation of PCD in plant cells, presumably through controlling ROI production. First, FB1 causes a variety of cellular and molecular alterations characteristics of apoptotic cells, including the accumulation of ROIs and fragmentation of nuclear DNA. Second, mutation in the FBR11 gene renders the mutant resistant to FB1, and the anti-apoptotic phenotype of the mutation is correlated with the cellular and molecular alterations commonly associated with plant PCD. These observations suggest that FBR11 is involved in the regulation of the PCD signaling pathway. Third, molecular and genetic analyses indicate that FBR11 encodes an LCB1. Notably, alterations of the sphingolipid level (1 h after FB1 treatment) are prior to ROI production (3–6 h) and cell death (22–24 h). These observations suggest that an altered sphingolipid level may lead to ROI production, thereby causing cell death. Finally, we showed that whereas dh-sph, phyto-sph and sph were capable of efficiently inducing ROI generation and cell death, dh-S1P specifically blocked the apoptotic effect induced by dh-sph, suggesting that free sphingoid bases and their phosphorylated derivatives play an important role in regulating ROI generation, thereby determining the cell fate.

FBR11 encodes an LCB1 of SPT, which consists of two subunits, LCB1 and LCB2. It has been suggested that LCB2 acts as the enzymatically active center, and LCB1 appears to stabilize the heterodimer 14. The lcb1 mutations caused a substantially reduced level of the LCB2 protein in both a Chinese hamster ovary cell mutant line 43 and yeast cells 44, thus leading to lethality in both cases. Similarly, the lcb1-1 mutation causes an embryo-lethal phenotype, indicating an essential role of FBR11 in plant growth and development 41. However, a striking phenotype of fbr11-1 appears to be its resistance to FB1. As competitive inhibitors of ceramide synthase, FB1 and AAL toxin both caused markedly increased levels of several sphingolipid species 15, 17, 45. This might account for the cell death phenotype in a number of cases 16, 17, 18, 19, 45, 46. Ceramide synthase was proposed to be a multi-subunit complex, of which members of the yeast longevity assurance gene (LAG1) family, including the tomato ASC1 gene 47, are the only components identified thus far 12, 48. Whereas the dominant ASC1 allele confers FB1- and AAL toxin-resistance to tomato 15, 49, overexpression of the mammalian LAG1 homologs UOG1 50, TRH1 and TRH4 51 renders 293T cells resistant to FB1. Conversely, the recessive asc1 isogenic line is sensitive to the toxins 15, 49. However, the toxin-susceptibility of asc1, which appears to be correlated to an elevated level of dh-sph and phyto-sph 17, 45, can be partially relieved by myriocin 45. Similarly, myriocin is also able to enhance the resistance to FB1 in Arabidopsis (our unpublished data). Consistent with these observations, the fbr11-1 mutation, which presumably causes a reduced SPT activity, also leads to resistance to FB1. Thus, it appears that a decreased SPT activity, either by mutations or an enzymatic inhibitor of SPT, can partially compensate for a dysfunctional ceramide synthase, thereby reducing or preventing the generation of the death signal(s) in plant cells.

Because an impaired ceramide synthase causes PCD, the death signal(s) are likely the substrates of the enzyme. The currently known substrates of the enzyme include dh-sph and sph 10, 12, 13, which accumulate to substantially high levels upon treatment with FB1 or AAL toxin in a variety of plant species 17, 21, 45 (see also Table 1) as well as several mammalian cell lines 14, 52. The observation that FB1 causes massively increased sphingoid bases suggests that these compounds are key signal molecules that control plant PCD. In the FB1-treated plants, the phosphorylated sphingoid bases phyto-S1P and dh-S1P increase at a significantly higher level compared to their free bases (Table 1). However, in contrast to that observed in Saccharomyces cerevisiae 53, these phosphorylated sphingoid bases are incapable of activating PCD in Arabidopsis. These observations suggest that different mechanisms may operate in yeast and plants. Instead, several free sphingoid bases are able to efficiently and specifically induce ROI generation and cell death. Therefore, the FB1-resistant phenotype of fbr11-1 is presumably due to a reduced level of free sphingoid bases (see also below).

Upon FB1 treatment, levels of several free sphingoid bases increased remarkably, although a lower level of these compounds was found in fbr11-1. These results suggest that an elevated level of free sphingoid bases alone is not sufficient to account for the PCD-inducing effect of the toxin. The observation that dh-S1P specifically blocks ROI generation and cell death induced by dh-sph, but not by phyto-sph and paraquat, implies that a proper balance between a free sphingoid base and its phosphorylated form is also important for the control of plant PCD. Consistent with these findings, C2-ceramide induces, but its phosphorylated form partially blocks, PCD in Arabidopsis protoplasts 23. Similar observations were also made in animal cells, in which the balance between sph and S1P, but rather than their absolute concentrations, was thought to determine the cell fate and other physiological consequences 54. It however should be pointed out that both the increased levels of free sphingoid bases and an altered balance between a free sphingoid base and its phosphorylated form may play a role in the regulation of plant PCD. Currently we are unable to clearly distinguish these two possibilities or determine which factor plays a more dominant role.

Sphingolipid-regulated PCD in plant cells has long been appreciated, but the biochemical mechanism of the regulation remains largely unclear. We show that ROI generation is prior to the onset of cell death but after the sphingolipid alterations in FB1-treated WT plants. Conversely, no ROI accumulation was observed in the FB1-resistant fbr11-1 mutant plants. Moreover, when treated with dh-sph, phyto-sph and sph, ROI generation also appears to be required for cell death. Lastly, the anti-apoptotic effect of dh-S1P is also correlated with its inhibition of ROI generation. These observations suggest that the sphingolipid-mediated cell death is coupled with the generation of ROIs, which, in turn, may activate a downstream PCD pathway. Among the currently known components involved in the ROI-mediated signaling, the structurally related zinc-finger proteins LSD1 and LOL1 presumably act to monitor the cellular ROI levels 32, 55, 56, whereas the serine/threonine protein kinase OXI1 appears to link an oxidative burst signal with diverse downstream responses 57. Moreover, overexpression of the Arabidopsis Bax Inhibitor-1 renders the transgenic plants insensitive to ROIs induced by a number of stimuli 58. Although possible interactions among these signaling components remain unknown, it will be of interest to examine the sphingolipid-regulated PCD with respect to these genetic loci. Despite the critical regulatory role of sphingolipids in the control of ROI generation and PCD as revealed by this study, the biochemical mechanism of this regulation unfortunately remains to be elucidated. Moreover, identification of the cellular targets of sphingolipids will be a key to fully understanding the complicated regulatory machinery that controls PCD in plant cells, and will likely shed light on the mechanisms of other sphingolipid-regulated signaling activities in plant cells.