Dehydration leads to different physiological and biochemical responses in plants. We analysed the lipid composition and the expression of genes involved in lipid biosynthesis in the desiccation-tolerant plant Craterostigma plantagineum. A comparative approach was carried out with Lindernia brevidens (desiccation tolerant) and two desiccation-sensitive species, Lindernia subracemosa and Arabidopsis thaliana. In C. plantagineum the total lipid content remained constant while the lipid composition underwent major changes during desiccation. The most prominent change was the removal of monogalactosyldiacylglycerol (MGDG) from the thylakoids. Analysis of molecular species composition revealed that around 50% of 36:x (number of carbons in the acyl chains: number of double bonds) MGDG was hydrolysed and diacylglycerol (DAG) used for phospholipid synthesis, while another MGDG fraction was converted into digalactosyldiacylglycerol via the DGD1/DGD2 pathway and subsequently into oligogalactolipids by SFR2. 36:x-DAG was also employed for the synthesis of triacylglycerol. Phosphatidic acid (PA) increased in C. plantagineum, L. brevidens, and L. subracemosa, in agreement with a role of PA as an intermediate of lipid turnover and of phospholipase D in signalling during desiccation. 34:x-DAG, presumably derived from de novo assembly, was converted into phosphatidylinositol (PI) in C. plantagineum and L. brevidens, but not in desiccation-sensitive plants, suggesting that PI is involved in acquisition of desiccation tolerance. The accumulation of oligogalactolipids and PI in the chloroplast and extraplastidial membranes, respectively, increases the concentration of hydroxyl groups and enhances the ratio of bilayer- to non-bilayer-forming lipids, thus contributing to protein and membrane stabilization.
Craterostigma plantagineum Hochst. (Linderniaceae) is a resurrection plant capable of equilibrating with air to 0% relative humidity and recovering full physiological activities within 24 h after rehydration (Bartels et al., 1990). C. plantagineum has been extensively studied, since it acquires desiccation tolerance by induction of a coordinated genetic programme rather than by anatomical adaptations (Bartels and Salamini, 2001). Several hundred genes are differentially expressed in response to dehydration (Rodríguez et al., 2010). The most abundant group of dehydration-induced genes are LEA (late embryogenesis abundant) genes (Bartels and Salamini, 2001). Accumulation of LEA proteins is correlated with desiccation tolerance in plants, fungi, nematodes and bacteria (Garay-Arroyo et al., 2000; Browne et al., 2002). Other mechanisms contributing to desiccation tolerance are the stimulation of antioxidant defence, accumulation of sucrose and expression of unique non-protein-coding transcripts (Bianchi et al., 1991; Hilbricht et al., 2008; Dinakar and Bartels, 2012).
Dehydration causes congestion of cytoplasmic components and cell contents become viscous, enhancing the probability of protein denaturation and fusion of apposed membranes (Hoekstra et al., 2001). Membranes belong to the first targets of degradation during dehydration. Protection of membrane integrity is essential to maintain metabolic homeostasis (Sahsah et al., 1998). Lipids and proteins are the predominant constituents of biological membranes. While some lipids, such as phosphatidic acid (PA) (Munnik and Testerink, 2009) and polyphosphoinositides (Munnik and Vermeer, 2010) play a role in cell signalling in response to dehydration, the majority of lipids have a structural role in establishing membrane bilayers. Each individual lipid has unique biophysical properties, and therefore the lipid composition of the membrane is critical to maintain the bilayer structure and to avoid membrane fusion or disruption (Webb and Green, 1991). During dehydration, the lipid composition changes to contribute to membrane stabilization (Torres-Franklin et al., 2007). One of the best documented dehydration responses is the decline in galactolipids with respect to phospholipids, and the increase in the digalactosyldiacylglycerol (DGDG) to monogalactosyldiacylglycerol (MGDG) ratio, mainly originating from the decrease in MGDG, the major chloroplast lipid (Stevanovic et al., 1992; Quartacci et al., 1997; Gigon et al., 2004; Torres-Franklin et al., 2007). These changes are accompanied by the accumulation of non-polar lipids (Navari-Izzo et al., 1990; Quartacci et al., 1997; Gigon et al., 2004). Another common response to dehydration is the decline in the degree of fatty acid desaturation (Monteiro de Paula et al., 1990; Daklma et al., 1995). It is not clear whether lipid changes are adaptive or result from uncontrolled degradation. However, an increase in catabolic enzymes and an inhibition of lipid biosynthesis has been observed in response to dehydration (Monteiro de Paula et al., 1993; Gigon et al., 2004). The alterations of lipid contents depend on stress intensity. Lipid changes occur at mild drought stress in dehydration-sensitive plants, whereas in tolerant plants, changes are mainly observed after severe dehydration (Sahsah et al., 1998). The degradation of membrane lipids and the decrease in fatty acid desaturation are less pronounced in tolerant plants, indicating higher membrane stability (Monteiro de Paula et al., 1990; Gigon et al., 2004). The importance of lipids in the adaptation to stress has also been demonstrated in transgenic plants. Murata et al. (1992) reported that chilling tolerance of tobacco was improved when the proportion of unsaturated fatty acids in chloroplast phosphatidylglycerol (PG) was high. Likewise, overexpression of ω-3 desaturases in tobacco resulted in increased tolerance to salt and drought stress (Zhang et al., 2005). The SFR2 (sensitive to freezing 2; GGGT, galactolipid:galactolipid galactosyltransferase) protein is involved in the remodelling of the chloroplast envelope membranes through conversion of MGDG into DGDG and oligogalactolipids during freezing and dehydration (Moellering and Benning, 2011). Both freezing and desiccation result in water loss from the cell, and the lipid changes introduced by SFR2 are believed to stabilize the bilayer. Thus, Arabidopsis sfr2 mutants display severe damage upon freezing, mainly due to chloroplast rupture (Fourrier et al., 2008).
Craterostigma plantagineum is a homoichlorophyllous resurrection plant in which thylakoid membranes stay intact during desiccation (Schneider et al., 1993). Despite the importance of lipids for functional membrane integrity, no data on membrane lipid composition in C. plantagineum are available. We focused on the metabolic study of lipid and fatty acid compositions in C. plantagineum during de- and rehydration, and the expression of genes involved in lipid biosynthesis. A comparative approach was carried out using plants differing in desiccation tolerance to discriminate between adaptive lipid modifications and changes due to non-specific lipid degradation.
The lipid composition of C. plantagineum is profoundly changed during desiccation/rehydration
Quadrupole time-of-flight (Q-TOF) mass spectrometry (MS) revealed that a constant level of total polar lipids at approximately 100 nmol mg−1 dry weight (DW) was maintained in C. plantagineum leaves during desiccation and rehydration, but lipid composition underwent prominent changes (Figure 1). Dehydration led to a decrease in the amount of MGDG and a general increase in phospholipids (Figure 1; ED samples). Further dehydration resulted in almost complete loss of MGDG, which was reduced by 96% from 41.9 to 1.8 nmol mg−1 DW compared with non-dehydrated leaves (Figure 1; D samples). Phospholipids continued to accumulate, with the exception of PG which decreased during desiccation. The galactolipids DGDG, trigalactosyldiacylglycerol (TGDG) and tetragalactosyldiacylglycerol (TeGDG) accumulated in desiccated leaves, resulting in a decrease of the MGDG:DGDG ratio from 1.93 ± 0.16 (control) to 0.06 ± 0.01.
The composition of membrane lipids recovered during the first 48 h of rehydration, but many lipids did not reach the values of control plants (Figure 1; ER and LR samples). MGDG was actively synthesized, while the amounts of DGDG and TeGDG declined, and TGDG remained about 18-fold higher during rehydration. The MGDG to DGDG ratio increased to 1.33 ± 0.30 during recovery. Phospholipids decreased upon rehydration, although they were still higher than in untreated samples. The PG content did not recover, but continued to decline to 55% of that of control leaves.
The Q-TOF MS/MS analysis allowed us to measure individual lipid molecular species, and therefore to follow galactolipid moieties during membrane lipid remodelling. Molecular species containing 36 carbons (36:x; in particular with six double bonds, 36:6) were predominant in MGDG from non-stressed leaves (37.4 nmol mg−1 DW, approximately 90% of total MGDG) (Figure 2). During desiccation, 90% of the MGDG decrease was attributed to 36:x molecular species, whose total amount declined to 1.8 nmol mg−1 DW.
Monogalactosyldiacylglycerol is partially converted into DGDG and oligogalactolipids in C. plantagineum during desiccation
The amounts of DGDG, TGDG and TeGDG increased during desiccation, from 21.6 to 28.4, from 0.025 to 0.745 and from 0.029 to 0.076 nmol mg−1 DW, respectively, and they decreased upon rehydration to 21.3, 0.420 and 0.060 nmol mg−1 DW, respectively (Figure 1). The increase was mainly due to the accumulation of 36:x molecular species, 7.71 nmol mg−1 DW in DGDG, 0.62 nmol mg−1 DW in TGDG and 0.029 nmol mg−1 DW in TeGDG (Figure 2), indicating that DGDG and oligogalactolipids were directly synthesised from MGDG. Therefore, about 8 nmol mg−1 DW (equivalent to about 20%) of MGDG was converted into DGDG and oligogalactolipids during desiccation. In plants, DGDG can be produced by two pathways, i.e. by UDP-Gal-dependent DGDG synthases (DGD1, DGD2) or by SFR2 which catalyses the MGDG-dependent galactosylation of MGDG and DGDG. DGDG produced by the different pathways can be distinguished by the anomeric linkages between the galactose residues. While the inner galactose (GalI) in DGDG is always in the β-configuration (due to the specificity of MGDG synthases), the second galactose (GalII) in DGDG produced by DGDG synthases is α-anomeric (αβDGDG), whereas in SFR2-dependent DGDG it is in the β-configuration (ββDGDG) (Kelly and Dörmann, 2002; Moellering et al., 2010). Nuclear magnetic resonance (NMR) spectroscopy revealed that the glycosidic linkages of DGDG isolated from non-dehydrated and desiccated leaves solely corresponded to αβDGDG (Table 1), whereas the alternative, SFR2-dependent form, ββDGDG, was not detected. Therefore, DGDG produced in C. plantagineum is synthesized by DGD1/DGD2 but not by SFR2. Analogous to DGDG, the anomeric configuration of the outermost galactose (GalIII) in TGDG is indicative for its origin from the DGD1/DGD2 (α) or SFR2 (β) pathways. We therefore isolated TGDG from desiccated C. plantagineum leaves. Analysis by NMR spectroscopy demonstrated that it corresponded to the anomeric structure βαβTGDG (Table 2, Figure S1 in Supporting Information) indicating that SFR2 is responsible for the attachment of the terminal galactose (GalIII) in TGDG.
Table 1. 1H NMR data of digalactosyldiacylglycerol (DGDG) from Craterostigma plantagineum. The 1H NMR spectra (700 MHz) of per-O-acetylated DGDG from control and desiccated C. plantagineum leaves show a single anomeric signal at δH 4.48 p.p.m. (3J1,2 8.0 Hz) for the inner galactose (GalI) indicating β-configuration and a single anomeric signal at δH 4.94 p.p.m. (3J1,2 3.6 Hz) for α-configuration of the second galactose (GalII) (Hölzl et al., 2005). Internal reference: CHCl3 (δH 7.26 p.p.m.)
Table 2. 1H NMR data of βαβTGDG from Craterostigma plantagineum leaves. The 1H NMR spectra (700 MHz) of per-O-acetylated trigalactosyldiacylglycerol (TGDG) from desiccated C. plantagineum show three distinct anomeric signals. The signal at δH 4.46 p.p.m. (3J1,2 8.0 Hz) corresponds to the innermost β-galactose (GalI), the one at δH 4.86 p.p.m. (3J1,2 3.6 Hz) to the α-galactose (GalII); both configurations are in accordance with αβDGDG (Table 1). The third and outermost galactose (GalIII) in TGDG is β-configurated [H-1III at δH 4.45 p.p.m. (3J1,2 8.0 Hz)]. Internal reference: CHCl3 (δH 7.26 p.p.m.)
Monogalactosyldiacylglycerol is hydrolysed and diacylglycerol is used for phospholipid and triacylglycerol synthesis or disassembled in C. plantagineum during desiccation
The conversion of MGDG into DGDG and oligogalactolipids does not explain the full extent of the decrease in MGDG during desiccation. Therefore, other lipid remodelling or degradation processes must be involved. The accumulation of 36:x (and in particular 36:6) molecular species was also observed in phospholipids during desiccation (Figure 2; D samples). A large proportion of the increase in the two most abundant phospholipids, phosphatidylcholine (PC) and phosphatidylethanolamine (PE), was based on 36:x, as PC increased from 15.4 to 31.0 nmol mg−1 DW (10.62 nmol mg−1 DW of the additional PC with 36:x backbone) and PE increased from 6.3 to 11.4 nmol mg−1 DW (3.64 nmol mg−1 DW of 36:x PE). The contribution of 36:x molecular species to the increases in phosphatidylinositol (PI) and phosphatidylserine (PS) was less pronounced. PI increased from 4.5 to 19.3 nmol mg−1 DW (2.82 nmol mg−1 DW of 36:x PI) and PS increased from 0.348 to 0.823 nmol mg−1 DW (0.06 nmol mg−1 DW of 36:x PS). After 48 h of rehydration, the lipid composition returned to values close to control. Only the levels of 36:6 remained low in MGDG and high in TGDG and TeGDG.
The analysis of non-polar lipids showed that triacylglycerol (TAG) accumulated from 0.146 to 3.11 nmol mg−1 DW in desiccated leaves (Figure 3; D samples) and decreased during recovery to 1.46 nmol mg−1 DW (Figure 3; LR samples). 18:x/18:3/18:3 represented 72% of the TAG accumulated during desiccation (Figure 4; D samples), hence this TAG was synthesized from 36:x diacylglycerol (DAG). No change in the amount of 36:x DAG was observed (Figure 4). In conclusion, about 50% of 36:x MGDG that was lost during desiccation was hydrolysed, and 36:x DAG was incorporated into TAG and phospholipids.
The decrease of approximately 40 nmol mg−1 DW of MGDG (mostly 36:x) during desiccation (Figure 1) can be explained by a conversion into DGDG and oligogalactolipids (approximately 8 nmol mg−1 DW; 20%) or by hydrolysis and conversion of DAG into TAG and phospholipids (approximately 20 nmol mg−1 DW; 50%). As no other lipid classes increased, the residual amount of MGDG (approximately 12 nmol mg−1 DW, 30%) was presumably disassembled. MGDG hydrolysis by action of galactolipases and β-galactosidases results in the release of free fatty acids (FFAs) and lyso-MGDG, respectively. The levels of FFAs increased from 0.63 nmol mg−1 DW in the control samples to 1.87 nmol mg−1 DW during desiccation, and decreased to 0.79 nmol mg−1 DW after rehydration (Figure S2). This accumulation of FFAs in the desiccated leaves was mostly caused by an increase in 18:3 from 0.16 to 1.35 nmol mg−1 DW. The levels of lyso-MGDG and lyso-DGDG remained low, of the order of fmol mg−1 DW. There was a decrease in 18:3 lyso-MGDG in the desiccated leaves, but this change was negligible compared with overall changes in MGDG (Figures 1 and S3). Therefore, the levels of FFAs and lyso-MGDG remain low, indicating that MGDG breakdown products do not accumulate in C. plantagineum during desiccation.
Diacylglycerol (34:x) is incorporated into PI in C. plantagineum during desiccation
The total amount of DAG decreased by 77% 2 days after withholding water (Figure 3; ED samples) and by 63% in desiccated leaves (Figure 3; D samples), and recovered to control values upon rehydration (Figure 3; LR samples). While the 36:x DAG content remained mostly unchanged, about 65% of the decrease in DAG corresponded to the decline of 34:x species (16:0/18:3, 16:0/18:2 and 16:0/18:1). This decrease in 34:x DAG was reversed upon rehydration (Figure 4). In parallel, 34:x phospholipids increased during desiccation, and this was reversed after rehydration (Figure 2). The largest accumulation in 34:x molecular species was observed for PI, which increased by 11.9 nmol mg−1 DW after desiccation (80.9% of the total PI increase). The increase of 34:x molecular species was less pronounced in the other phospholipids, 4.3, 2.1 and 0.56 nmol mg−1 DW in PC, PE and PS, respectively. Thus, 34:x moieties are mainly used for PI synthesis and to a minor extent for the other phospholipids in C. plantagineum during dehydration.
Dehydration results in the accumulation of PA in C. plantagineum
In general, phospholipid levels increased in C. plantagineum during desiccation. The largest proportional increase was observed for PA (Figure 2). The levels of PA were 10-fold higher in desiccated (1.54 nmol mg−1 DW) than in unstressed leaves (0.17 nmol mg−1 DW). The contribution of 34:x and 36:x to this increase was similar, 48% and 52% respectively. Two days after rehydration, PA decreased to 0.31 nmol mg−1 DW.
Expression patterns of C. plantagineum genes involved in lipid metabolism
Changes in lipid contents might be caused by alterations in the expression of genes encoding enzymes of lipid biosynthesis or degradation. Therefore, the expression patterns for the following enzymes were analysed: MGDG synthase (MDG); DGDG synthase (DGD); SFR2, involved in the synthesis of oligogalactolipids; PI synthase (PIS); β-galactosidases (βGAL), possibly involved in MGDG hydrolysis; diacylglycerol O-acyltransferase (DGAT) involved in TAG synthesis; phospholipase D (PLD), involved in phospholipid hydrolysis and PA production. Amino acid BLAST searches against the translated C. plantagineum transcripts (Rodríguez et al., 2010) revealed the presence of lipid metabolism enzymes with high sequence identity with A. thaliana homologues. One putative MGD, DGD, SFR2 and DGAT isoform, two putative PIS isoforms and three putative βGAL isoforms were isolated from C. plantagineum by RT-PCR (sequences shown in Figure S4). PC:DAG acyltransferases (PDAT) sequences could not be retrieved from C. plantagineum. BLAST searches with the C. plantagineum cDNA sequences were performed to identify the closest homologues in A. thaliana. Thus, the isolated C. plantagineum cDNAs were found to correspond to AtMGD1 (At4 g31780), AtDGD1 (At3 g11670), SFR2 (At3 g06510), DGAT2 (At3 g51520), βGAL7 (At5 g20710), βGAL9 (At2 g32810) and βGAL17 (At1 g72990) as best matches. CpPLD1 and CpPLD2 cDNAs were previously isolated (Frank et al., 2000).
The RT-PCR analysis showed that the CpMGD1, CpDGD1 and CpSFR2 transcripts were up-regulated at the onset of dehydration, then slightly declined, and after rehydration recovered to the basal expression level (Figure 5). The induction of expression of CpDGD1 and CpSFR2 correlated with the accumulation of DGDG and oligogalactolipids in desiccated leaves. CpMGD1 expression was also induced, but the amount of MGDG decreased during desiccation. Two genes encoding β-galactosidases possibly involved in MGDG hydrolysis, CpβGAL7 and CpβGAL9, were up-regulated during desiccation and down-regulated during recovery, whereas CpβGAL17 displayed the opposite pattern. Expression of the two CpPIS isoforms did not change during desiccation or rehydration. Transcripts of DGAT2 accumulated during dehydration and declined after watering. CpPLD1 and CpPLD2 showed opposite responses, CpPLD1 expression decreased slightly during the early dehydration and recovered during desiccation, while CpPLD2 was up-regulated during dehydration and down-regulated after rehydration.
A comparative approach to discriminate between general responses to dehydration and desiccation tolerance mechanisms
To discriminate between changes in lipid metabolism associated with common plant responses to dehydration and those involved in the acquisition of desiccation tolerance, lipid changes were examined in three species, A. thaliana (desiccation sensitive), Lindernia subracemosa (desiccation sensitive) and Lindernia brevidens (desiccation tolerant). In contrast to C. plantagineum, the total polar lipid contents decreased in the three species, the amounts were 13, 50 and 84% lower in A. thaliana, L. brevidens and L. subracemosa after desiccation compared with untreated plants, respectively (Figure 6).
Similar to C. plantagineum the three species showed a decrease in MGDG levels during dehydration, albeit less pronounced in A. thaliana (23% lower) than in L. subracemosa and L. brevidens (97 and 93% lower, respectively) (Figure 6). Both TGDG and TeGDG also increased, but not DGDG, which decreased in L. brevidens and in L. subracemosa. The same pattern was observed for 36:x, 36:6 in particular (Figures S5, S7 and S9). The levels of TAG increased in A. thaliana and L. brevidens after dehydration (Figure 7) as a consequence of the increase of 18:x/18:3/18:3 (Figures S6 and S10). 18x/18:3/18:3 TAG also slightly increased in L. subracemosa, but other TAG species decreased (Figure S8). Therefore, the conversion of MGDG into DGDG, oligogalactolipids and TAG is a common response to dehydration in all four plants.
In contrast to C. plantagineum, 36:x DAG was not converted into phospholipids in the other three plants. In general phospholipids decreased in the three plants during dehydration (Figure 6). The total levels of DAG did not change in the three species during dehydration (Figure 7). However, the levels of 16:0/18:2 and 16:0/18:3 DAG decreased in the two Lindernia species. Phosphatidylinositol only increased in the desiccation-tolerant species L. brevidens, rising from 12.2 to 15.6 nmol mg−1 DW (Figure 6), as a consequence of the accumulation of 34:x PI (Figure S9). The increase in 34:x PI in L. brevidens becomes even more obvious when compared with the other phospholipids, which decreased after desiccation. Therefore in analogy with C. plantagineum, 34:x moieties are incorporated into PI in L. brevidens during desiccation, suggesting that this response is unique to desiccation-tolerant plants.
Phosphatidic acid increased in L. subracemosa and L. brevidens during desiccation from 0.562 to 3.321 and from 0.72 to 8.46 nmol mg−1 DW, respectively, while PA decreased in Arabidopsis from 1.61 to 0.52 nmol mg−1 DW (Figure 6). Similar to C. plantagineum, the increase of PA in the Lindernia species was due to 34:x and 36:x molecular species in similar proportions (Figures S7 and S9).
One of the first symptoms of dehydration injury is the decrease in the contents of membrane lipids (Monteiro de Paula et al., 1990; Sahsah et al., 1998; Gigon et al., 2004). In C. plantagineum, the membrane lipid composition was profoundly modified but the total lipid content remained constant during desiccation/rehydration, indicating a high capacity to maintain membrane stability. A similar observation was reported for chloroplasts of the resurrection plant Boea hygroscopica (Navari-Izzo et al., 1995). In contrast, the two Lindernia species showed a decrease in the lipid content during desiccation. The stability of membrane lipids in C. plantagineum, in comparison to the Lindernia species, might be due to changes in lipid compositions and to a higher capacity for sucrose accumulation and LEA protein synthesis (Phillips et al., 2008).
The decrease in MGDG and PG indicates that the chloroplast participates in the desiccation response. In C. plantagineum, chlorophyll is degraded during dehydration, resulting in a decrease in photosynthetic activity (Dinakar and Bartels, 2012). The decrease in photosynthetic complexes protects the plant against oxidative stress (Miller et al., 2010). The decline of the surface area of the thylakoid membrane might prevent fusions of bilayers in the shrinking chloroplast. After rehydration, lipid composition is restored. Monogalactosyldiacylglycerol is synthesized within 24 h after rehydration, possibly facilitated by the accumulation of CpMGD1 transcripts during desiccation. This reconstruction of thylakoid membranes correlates with the ability of C. plantagineum to recover photosynthetic activity (Dinakar and Bartels, 2012).
The conversion of MGDG into DGDG and oligogalactolipids is a common response to dehydration
The MGDG content decreased in the four plants during dehydration, while DGDG and the oligogalactolipids TGDG and TeGDG increased. The decrease in MGDG is a common adaptation of plants to drought, osmotic stress or freezing (Monteiro de Paula et al., 1993; Gigon et al., 2004; Moellering et al., 2010). Due to the small size of the head group, MGDG molecules are cone-shaped and non-bilayer forming, and therefore form inverted hexagonal II (HII) structures, whereas DGDG and oligogalactolipids are cylindrical shaped and form lamellar bilayers (Sprague, 1987; Webb and Green, 1991). Thus, the conversion of cone-shaped MGDG into cylindrical DGDG and oligogalactolipids stabilizes the membranes during dehydration. The accumulation of DGDG and oligogalactolipids results in an increased thickness of the head groups and an increase in the concentration of hydroxyl groups, thus enhancing the repulsive hydration forces between adjacent membranes to avoid bilayer fusion (Moellering et al., 2010).
Monogalactosyldiacylglycerol, DGDG and the oligogalactolipids of C. plantagineum and of the Lindernia species contain mostly 36:6 molecular species, i.e. two 18:3 acyl groups, in agreement with the scenario that these species are 18:3 plants. In contrast, Arabidopsis, a 16:3 plant, contains a considerable proportion of 34:6 (sn1-18:3, sn2-16:3) MGDG (Heinz and Roughan, 1983; Browse et al., 1986). Digalactosyldiacylglycerol can be synthesized by two pathways. The DGD1/DGD2 pathway uses UDP-galactose and MGDG for DGDG synthesis (Kelly and Dörmann, 2002). Alternatively, the galactolipid:galactolipid galactosyltransferase SFR2 produces DGDG by transferring a galactose from one MGDG to another (Moellering et al., 2010). Both UDP-Gal-dependent and -independent enzymes can produce TGDG and TeGDG through progressive galactosylation (Kelly and Dörmann, 2002; Moellering et al., 2010). While DGD1 is responsible for the bulk of DGDG biosynthesis under normal growth conditions (Dörmann et al., 1999), the expression of DGD1 and DGD2 is induced during phosphate deprivation, when large amounts of DGDG accumulate (Kelly and Dörmann, 2002). SFR2 is activated in response to freezing (Moellering et al., 2010). In desiccated C. plantagineum, the outermost galactose (GalII) in DGDG was in the α-configuration, indicating it was synthesized by DGD1/DGD2. Trigalactosyldiacylglycerol which was of βαβ-configuration was exclusively produced by SFR2. Therefore, the conversion of MGDG into DGDG and oligogalactolipids depends on the activation of two pathways, i.e. DGD1/DGD2 (producing αβDGDG) and SFR2 (converting αβDGDG into βαβTGDG). Accordingly, the accumulation of DGDG and oligogalactolipids correlated with the up-regulation of CpDGD1 and CpSFR2 expression. The accumulation of DGDG and oligogalactolipids with distinct anomeric configurations was previously observed in different plant species (Kelly and Dörmann, 2004). In the Arabidopsis tgd1/dgd1 double mutant, the synthesis of αβDGDG is decreased due to the deficiency in DGD1 activity, while the galactolipid:galactolipid galactosyltransferase (SFR2) activity is constitutively upregulated by a mutation in the TGD1 gene encoding a lipid transporter component (Xu et al., 2003). In this double mutant, DGDG consists of αβ-DGDG and ββDGDG, while TGDG is entirely of the βββ-configuration. In most plant species and in most organs, the αβDGDG isomer is predominant, but ββ-DGDG accumulates in legume seeds to up to 26.4–32.5% of total DGDG in adzuki beans (Kojima et al., 1990). The TGDG in adzuki beans consists of two isomers, βαβTGDG and βββTGDG. Rice bran contains a different series of oligogalactolipids based on αβDGDG and ααβTGDG (Fujino and Miyazawa, 1979; Kelly and Dörmann, 2004). Therefore, the accumulation of different anomeric forms of DGDG and oligogalactolipids might depend on the relative activity of DGD1/DGD2 (producing α-anomeric bonds) versus SFR2 (β-anomeric bonds). This scenario implies that DGD1/DGD2 and SFR2 can accept DGDG substrates with different anomeric configurations. Furthermore, it is possible that the relative activities and substrate specificities of DGD1/DGD2 and SFR2 differ between the plant species, possibly reflecting adaptive processes to environmental constrains during evolution.
Craterostigma plantagineum presumably also contains a gene orthologous to AtDGD2, but no corresponding CpDGD2 sequence was retrieved. In Arabidopsis, SFR2, DGD1 and DGD2 are localized to the chloroplast envelope, where they have access to only a low proportion of MGDG while most of the galactolipids are in the thylakoids (Froehlich et al., 2001; Fourrier et al., 2008). Therefore, it is conceivable that only a limited proportion of MGDG can be converted into DGDG and oligogalactolipids by the DGD1/DGD2 and SFR2 pathways.
Acyl groups derived from MGDG hydrolysis are used for phospholipid synthesis
During desiccation, MGDG can be hydrolysed by β-galactosidases. In plants fatty acyl groups are exported from plastids in the form of FFAs during biosynthesis. However, it is unclear whether MGDG-derived DAG can be directly exported from the plastids during desiccation in C. plantagineum, or whether DAG is first hydrolysed in the plastids and the exported FFAs are used for reassembling phospholipids at the endoplasmic reticulum (ER) (Figure 8). Bhalla and Dalling (1984) confirmed the presence of β-galactosidase activity in the stroma and thylakoids of wheat mesophyll protoplasts. The expression of two β-galactosidase genes, CpβGAL7 and CpβGAL9, was upregulated in C. plantagineum during dehydration. It is conceivable that β-galactosidases contribute to the hydrolysis of thylakoid-localized MGDG not accessible to SFR2, DGD1 or DGD2, thereby stabilizing thylakoids by removing non-bilayer-forming MGDG, and reducing the surface area of the membranes.
The incorporation of 36:x moieties into phospholipids was not observed in the other three plants. Previous studies showed that desiccation led to a decrease in the galactolipid to phospholipid ratio in desiccation-tolerant plants due to degradation of MGDG, but no data on molecular species composition were presented (Stefanov et al., 1992; Stevanovic et al., 1992; Quartacci et al., 1997).
Mass spectrometry showed that 36:x DAG was also used for the biosynthesis of 18:x/18:3/18:3 TAG in C. plantagineum during desiccation. The increased biosynthesis of leaf TAG correlated with induced CpDGAT2 expression. DGAT1 and PDAT are the most relevant TAG synthases in Arabidopsis seeds (Zhang et al., 2009). However, DGAT1 and PDAT sequences could not be retrieved from C. plantagineum. 36:x DAG did not accumulate, indicating that it is rapidly converted into TAG and phospholipids. The synthesis of TAG from 36:x DAG during desiccation was also observed in the other three plants, suggesting that TAG accumulation represents a common response to dehydration. Accumulation of TAG could help to accommodate a shrinking organelle during osmotic stress by removing excess lipids from the membrane. The single hydroxyl group of DAG forms a rather small ‘polar head group’. Therefore, DAG is conical shaped and forms inverted micellar (HII) structures, introducing small areas of unstable negative curvature into bilayers and facilitating membrane fission or fusion (Goñi and Alonso, 1999). Thus, the conversion of DAG into TAG, which is deposited in oil bodies, or into phospholipids with a cylindrical shape, contributes to membrane stabilization.
Disassembly of MGDG during desiccation
As described above, the predominant fraction of MGDG is converted into DGDG, TGDG, TeGDG or hydrolysed and DAG used for phospholipid and TAG production in C. plantagineum during desiccation. However, there is still a residual decrease in MGDG by about 12 nmol mg−1 DW (30%) which cannot be accounted for. Presumably, this fraction of MGDG is hydrolysed by β-galactolipases and galactosidases, releasing FFAs, lyso-MGDG and galactosyl-glycerol. The amounts of FFAs and lyso-MGDG in desiccated C. plantagineum were very low (Figures S2 and S3). Furthermore, Bianchi et al. (1992) found no evidence for the accumulation of glucosyl-glycerol and galactosyl-glycerol in C. plantagineum, while glucosyl-glycerol accumulated in Myrothamnus flabellifolia (Bianchi et al., 1993). Therefore, galactolipid breakdown products, including FFAs, lyso-MGDG and galactosyl-glycerol, do not accumulate during desiccation in C. plantagineum, indicating that they must be rapidly metabolized. It is known that FFAs and lyso-galactolipids act as detergents, affecting bilayer integrity. Therefore, FFAs, lyso-galactolipids and galactosyl-glycerol are presumably rapidly degraded via β-oxidation and further hydrolysis.
Accumulation of PA is a common plant response to dehydration
Phosphatidic acid showed the largest proportional increase of phospholipids in C. plantagineum during desiccation. The activation of PLD and the release of 32P-labelled PA from phospholipids during desiccation in C. plantagineum was reported earlier (Frank et al., 2000). Here, we demonstrate that steady-state levels of PA increase in C. plantagineum during long-term drought stress. The expression patterns of two phospholipase D genes, CpPLD1 and CpPLD2 (Figure 5), are in agreement with the results reported by Frank et al. (2000). CpPLD1 was constitutively expressed while CpPLD2 was induced during desiccation. Thus, CpPLD1 might act in the early dehydration response, producing PA as a second messenger, and CpPLD2 might be involved in bulk phospholipid turnover during late dehydration (Frank et al., 2000). The molecular species composition of 34:x/36:x PA resembles that of PC and PE, but it is different from that of PI which mostly contains 34:x. Therefore, the bulk PA pool is closely related to PC and PE, suggesting the existence of different PA pools enriched in 34:x or 34:x/36:x during desiccation/rehydration.
Phosphatidic acid also accumulated in the two Lindernia species. It was the only phospholipid that increased during desiccation in L. subracemosa, while the PA content declined in Arabidopsis. These results agree with data of Hong et al. (2008) who showed that the steady-state levels of PA in Arabidopsis decrease during prolonged dehydration. The differences observed for accumulation of PA between the different plants could be a consequence of the dehydration regimes. C. plantagineum, L. subracemosa and L. brevidens plants were completely dried in <2 weeks, whereas Arabidopsis plants were subjected to a progressive slow dehydration regime for 3 weeks that led to a relative water content of about 27.6% (Table 3), corresponding to mild drought stress. Therefore, the lack of accumulation of PA in Arabidopsis may be due to less severe dehydration.
Table 3. Relative water content (%) of plants subjected to dehydration. Craterostigma plantagineum leaves were collected from non-dehydrated plants (C), after 2 (ED) and 14 days (D) of dehydration, and 1 (ER) and 2 days (LR) after re-watering. Arabidopsis thaliana leaves were collected from plants maintained without watering for 21 days. Lindernia brevidens and Lindernia subracemosa plants were desiccated in the jars
Values represent mean ± SD, n = 5.
90.0 ± 1.1
63.1 ± 2.2
6.0 ± 2.9
75.2 ± 5.4
86.5 ± 6.1
82.4 ± 2.4
27.6 ± 8.1
87.1 ± 3.3
5.8 ± 1.9
84.8 ± 1.2
3.9 ± 1.5
Accumulation of PI correlates with acquisition of desiccation tolerance
Fatty acyl groups (18:x) derived from breakdown of MGDG together with 16:0 from de novo synthesis presumably serve as precursors for synthesis of 34:x moieties (Figures 1 and 8). Desiccation resulted in a decline of DAG, in particular of 34:3, 34:2 and 34:1 DAG, presumably caused by the incorporation of 34:x backbones into PI. While Q-TOF MS/MS of phospholipids and glycolipids only revealed the identity of the head group and the DAG moiety, it is not easy to unravel the exact distribution of the number of carbons and double bonds to the two acyl groups. Devaiah et al. (2006) determined the acyl group distribution of the most abundant phospholipids and glycolipids in Arabidopsis via MS/MS analysis in the negative mode. For example, the PI molecular species 34:3, 34:2 and 34:1 mostly contain 16:0/18:3, 16:0/18:2, 16:0/18:1 backbones, respectively. As PI is synthesized from CDP-DAG and L-myo-inositol by PI synthase (PIS; Mueller-Roeber and Pical, 2002) at the extraplastidial membranes (Löfke et al., 2008), its molecular species composition is presumably very similar in all plant species, including 16:3 and 18:3 plants. In contrast to PI, PC and PE are derived from DAG. This might explain the different molecular species compositions of PI versus PE and PC. It is possible that 34:x molecular species are specifically enriched in the CDP-DAG pool which gives rise to PI synthesis, while the DAG employed for PC and PE synthesis represents a mixture of 34:x and 36:x.
In C. plantagineum, two CpPIS isoforms are constitutively expressed, which might explain the rapid biosynthesis of PI at the onset of dehydration. Thus, the increase in 34:x PI might be related to desiccation tolerance. This hypothesis is supported by the observation that PI was the only phospholipid whose amount increased in the resurrection plant L. brevidens, but not in the desiccation-sensitive plants L. subracemosa or A. thaliana. The proportion of PI in C. plantagineum and L. brevidens increased from 4.5 to 18.1% and from 9.5 to 26.7% after desiccation, becoming the third and the first most abundant glycerolipid, respectively.
The protective role of PI is supported by the finding that overexpression of maize ZmPIS improves membrane stability (Zhai et al., 2012). Phosphatidylinositol is the biosynthetic precursor of phosphoinositides (PIP2), which give rise to second messengers in stress signalling (Munnik and Vermeer, 2010). However, considering the strong accumulation of PI in C. plantagineum during desiccation, it is likely that PI plays an important role in the structural stabilization of the membrane rather than signalling. Phosphatidylinositol, like PC, PG and PS, has a cylindrical shape. Therefore, these phospholipids form a bilayer membrane (Cullis et al., 1985). In this regard, the accumulation of the sugar alcohol lipid PI in the extraplastidial membranes is analogous to the increase in DGDG and oligogalactolipids in the chloroplast. Phosphatidylethanolamine is the only non-bilayer-forming phospholipid in the extraplastidial membranes. In C. plantagineum, the ratio of PE to PI decreases from 1.2 in control to 0.6 in desiccated leaves, in analogy with the decrease in the MGDG to DGDG ratio in chloroplasts. Due to the unique sugar alcohol in the head group, the accumulation of PI leads to an increased concentration of hydroxyl groups which might enhance the repulsive hydration force between adjacent bilayers to avoid membrane fusion during dehydration. Furthermore, PI serves as the precursor for the synthesis of glycosyl inositol phosphorylceramides (GIPC), abundant glycosphingolipids in the plasma membrane (Wang et al., 2008). Therefore, the increase in PI in desiccation-tolerant plants might also lead to an accumulation of GIPC, resulting in an increased content of sugar head groups in the bilayer. The hydroxyl groups of membrane lipids (galactolipids, PI, GIPC) can establish hydrogen bonds with macromolecules to avoid their precipitation in the tightly packed cytoplasm of a desiccated cell. They can interact with LEA proteins, which accumulate in C. plantagineum during desiccation (Hoekstra et al., 2001; Phillips et al., 2008). The LEA proteins undergo conformational changes, thus forming an amphipathic α-helix which interacts with anionic phospholipids (e.g. PA, PI) to stabilize the membranes (Tolleter et al., 2007; Koag et al., 2009; Petersen et al., 2012).
Our results show that lipids play an important role in the acquisition of desiccation tolerance in C. plantagineum. Figure 8 proposes a model for the role of lipids in desiccation tolerance. The strongest lipid change during dehydration is the removal of MGDG, a conic-shaped non-bilayer-forming chloroplast lipid. C. plantagineum contains mostly 36:x MGDG molecular species. Thus, all lipids produced from MGDG-dependent DAG during desiccation are characterized by a high 36:x content. A fraction of MGDG is converted into cylindrical DGDG by the DGDG synthases DGD1/DGD2 and subsequently into oligogalactolipids by SFR2/GGGT in the chloroplast envelope. After cleavage of the head group, MGDG-derived 36:x DAG can be hydrolysed, and 18:x FFA exported to the ER where they are reused for the synthesis of phospholipids. Alternatively, 36:x DAG might be directly exported from plastids during desiccation. The residual amount of MGDG is presumably disassembled. 34:x moieties derived from de novo assembly of 16:0 and 18:x acyl groups at the ER, are mainly used for synthesis of PI during dehydration. Phosphatidylinositol, with a sugar alcohol head group, contributes to the stabilization of membranes and proteins. The importance of this response for the acquisition of desiccation tolerance is supported by the finding that PI accumulation was only observed in desiccation-tolerant species. Further experiments with isolated plastids are required to study the cellular compartmentalization of the different lipid pathways.
Craterostigma plantagineum plants were propagated according to Bartels et al. (1990). L. brevidens (Skan) and L. subracemosa (De Wild) plants were propagated with a light intensity of 80 μmol m−2 sec−1 at 22°C and a day/night cycle of 16/8 h for 2 months (Phillips et al., 2008). A. thaliana (L.) Heynh. (Col-0) plants were grown on potting soil/vermiculite under white light of approximately 90–110 μmol m−2 sec−1 at 22°C with a day/night cycle of 16/8 h.
For dehydration experiments, C. plantagineum leaves were collected from 2-month-old untreated plants (control, C), after 2 (early dehydration, ED) and 14 days (desiccation, D) of dehydration, and 1 (early recovery, ER) and 2 days (late recovery, LR) of re-watering. A. thaliana plants were maintained without watering for a period of 21 days. L. brevidens and L. subracemosa plants were desiccated directly in the jars for 2 weeks. The relative water contents (Table 3) were determined according to Bernacchia et al. (1996).
Extraction, separation and quantification of lipids
For lipid extraction, separation and quantification by Q-TOF MS/MS measurements see the Supporting Methods.
Nuclear magnetic resonance spectroscopy
Nuclear magnetic resonance spectroscopic measurements were performed in CDCl3 (DGDG) or CDCl3/MeOH-d4 10:1 (v/v) (TGDG) at 300 K using 3 mm microtubes on a Bruker AvanceIII 700 MHz NMR (equipped with an inverse 5 mm quadruple-resonance Z-grade cryoprobe; http://www.bruker.com/). Deuterated solvents were purchased from Deutero GmbH (http://deutero.de/). Prior to the measurements, the purified per-O-acetylated glycolipids were exchanged twice from CDCl3. Chemical shifts were referenced to internal chloroform (δH = 7.26 p.p.m.). All data were acquired using a Bruker TOPSPIN V 3.0 (http://www.bruker.com). The 1H NMR assignments were confirmed by two-dimensional (2D) 1H,1H-COSY and -TOCSY experiments. Anomeric configurations and glycosidic sequences were further validated by 2D 1H,13C-HSQC and 1H,13C-HMBC (DGDG only) experiments.
Gene expression analysis
Total RNA was isolated from leaves of plants derived from three independent experiments (Valenzuela-Avendaño et al., 2005). The cDNA derived from 100 ng total RNA was subjected to 30 cycles of RT-PCR amplification in a 20 μl reaction mixture (Dinakar and Bartels, 2012).
The following A. thaliana protein sequences were used for searching the translated C. plantagineum transcriptome database (TBLASTN; www.blast.ncbi.nlm.nih.gov/Blast.cgi) for C. plantagineum sequences encoding lipid biosynthesis enzymes (Rodriguez et al., 2010): MGDG synthases, MGD1 (At4 g31780), MGD2 (At5 g20410) MGD3 (At2 g11810); DGDG synthases, DGD1 (At3 g11670), DGD2 (At4 g00550); sensitive to freezing response 2, SFR2 (At3 g06510); PI synthases, PIS1 (At1 g68000), PIS2 (At4 g38570); DAG O-acyltransferases, DGAT1 (At2 g19450), DGAT2 (At3 g51520); PC:DAG acyltransferases, PDAT1 (At3 g44830), PDAT2 (At5 g13640). Only A. thaliana β-galactosidase genes expressed in leaves (Ahn et al., 2007) – except isoforms involved in cell wall modification (Gantulga et al., 2009) – were used in the query: β-galactosidases, βGAL7 (At5 g20710), βGAL9 (At2 g32810), βGAL10 (At5 g63810), βGAL17 (At1 g72990). TBLASTN analysis revealed one cDNA sequence each for MGD1, DGD1, SFR2, PIS1, PIS2, DGAT2, βGAL7, βGAL9 and βGAL17. Furthermore, the CpPLD1 and CpPLD2 cDNA sequences (Frank et al., 2000) were used for primer design. Sequences of oligonucleotides used for cDNA amplification are shown in Table S1. The PCR products were cloned into a pJET1.2 vector for sequencing (Figure S4).
We thank H. Peisker and C. Buchholz (University of Bonn) for technical assistance and E. Heinz (University of Hamburg) for providing lyso-MGDG and lyso-DGDG standards. F. Gasulla was supported by a contract from the Generalitat Valenciana (APOSTD/2011/071, program VALi+d) and K. vom Dorp by the Deutsche Forschungsgemeinschaft (SFB 645).