Freezing Tolerance in Plants Requires Lipid Remodeling at the Outer Chloroplast Membrane

See allHide authors and affiliations

Science  08 Oct 2010:
Vol. 330, Issue 6001, pp. 226-228
DOI: 10.1126/science.1191803


Plants show complex adaptations to freezing that prevent cell damage caused by cellular dehydration. Lipid remodeling of cell membranes during dehydration is one critical mechanism countering loss of membrane integrity and cell death. SENSITIVE TO FREEZING 2 (SFR2), a gene essential for freezing tolerance in Arabidopsis, encodes a galactolipid remodeling enzyme of the outer chloroplast envelope membrane. SFR2 processively transfers galactosyl residues from the abundant monogalactolipid to different galactolipid acceptors, forming oligogalactolipids and diacylglycerol, which is further converted to triacylglycerol. The combined activity of SFR2 and triacylglycerol-biosynthetic enzymes leads to the removal of monogalactolipids from the envelope membrane, changing the ratio of bilayer- to non-bilayer–forming membrane lipids. This SFR2-based mechanism compensates for changes in organelle volume and stabilizes membranes during freezing.

Freezing tolerance in plants is a critical factor limiting the geographic distribution of wild species. It also determines the agricultural range of important crops (e.g., citrus), because freezing damage in transitional climate zones can have devastating effects on agriculture. At the cellular level, freezing damage leads to leakiness of biomembranes enclosing organelles or the cell itself. Mechanistically, this is attributed to severe cellular dehydration initiated by extracellular ice formation that results in decreased water potential across the plasma membrane (1, 2). Under these conditions the formation of non-bilayer lipid structures, such as the inverted hexagonal II (HII)–type structures, can occur, and, thus, stabilization of lamellar membrane systems within the dehydrated plant cell is a key determining factor in the survival of freezing (13). To survive mild freezing, plants like Arabidopsis require preexposure to low, nonlethal temperatures in a process known as cold acclimation (24). Cold acclimation involves transcriptional (5) and metabolic (6) changes that result in multiple mechanisms protecting against freezing damage, including increases in intracellular solutes and the accumulation of cryoprotective metabolites (7) and proteins (8). However, the phenotype of Arabidopsis plants carrying mutations in the SENSITIVE TO FREEZING 2 (SFR2) gene indicates a distinct mechanism required for freezing tolerance (911). The sfr2 mutants show extensive intracellular damage after freezing recovery, with rupture of both chloroplasts and tonoplasts, likely through fusion of destabilized membranes in these organelles. The SFR2 protein is constitutively present, and sfr2 mutants are not affected in the regulation of known cold-acclimation processes (911).

With cellular membrane damage in mind, alterations in lipid composition during cold acclimation, such as increased fatty acid unsaturation and phospholipid content, have long been correlated with enhanced freezing tolerance but may be adaptations to growth at low above-freezing temperatures rather than freezing stress per se (1, 2). More recent lipid-profiling studies in Arabidopsis have revealed that more-drastic lipid compositional changes occur during freezing, including a decrease in the chloroplast-specific lipid monogalactosyldiacylglycerol (MGDG) (12, 13).

A link between SFR2 and the reported freeze-dependent decrease in MGDG became evident in our effort to identify the enigmatic galactolipid:galactolipid galactosyltransferase (GGGT) (14). GGGT converts MGDG to di-, tri-, and tetragalactosyldiacylglycerol (DGDG, TGDG, and TeDG, respectively) by transglycosylation, resulting in the concomitant production of diacylglycerol (DAG) (15). The role of GGGT in galactolipid metabolism and the encoding gene(s) have remained unresolved since GGGT activity was first reported, but GGGT is known not to contribute significantly to DGDG synthesis in vivo during normal growth, as originally proposed (16, 17). In a reverse genetics candidate approach, we investigated new transfer-DNA mutant alleles disrupted in the SFR2 gene (sfr2-3 and sfr2-4, fig. S1) (18). SFR2 is predicted to encode a glycosyl hydrolase family 1 (GH-1) protein for which the native substrates remain unknown. We considered SFR2 to be a GGGT-encoding candidate because (i) some GH-1 family proteins are known to catalyze transglycosylation reactions similar to that postulated for GGGT (19, 20) and (ii) both the SFR2 protein and GGGT activity have been unambiguously localized to the chloroplast outer envelope membrane (9, 21). Infiltration of wild-type leaves with MgCl2 leads to TGDG and TeDG accumulation, presumably by activating GGGT. Applying this simple assay to the sfr2 mutants revealed their inability to accumulate these lipids, consistent with an apparent requirement of SFR2 for GGGT activity in vivo (fig. S2).

The sfr2-3 and sfr2-4 alleles showed similar freezing sensitivity (Fig. 1A) as observed for previously described sfr2-1 and sfr2-2 alleles (11). Testing whether freezing could also induce GGGT, we observed that wild-type plants accumulated TGDG and TeDG during freezing with a corresponding ~20 mole percent (mol %) decrease in MGDG, whereas the sfr2 mutants did not (Fig. 1, B and C). Freezing treatment increased relative amounts of DGDG independent of SFR2 (Fig. 1C). It seems possible that this increase is due to activity increases in the uridine 5′-diphosphate galactose (UDP-Gal)–dependent enzymes of DGDG biosynthesis after treatment or simply due to a relative decrease in other lipids classes. Expression of SFR2 cDNA into sfr2-3 restored accumulation of TGDG and TeDG after MgCl2 infiltration, indicating genetic complementation (fig. S3). We also found a ~7.5-fold increase in triacylglycerol (TAG) in wild-type freeze-treated plants, to 6% of total fatty acids esterified in TAGs (Fig. 1, D and E). Importantly, the sfr2 mutants showed ~50% reduction in TAG accumulation and a large decrease in 16:3 (number of carbons:number of double bonds) acyl groups esterified to TAG in freeze-treated plants compared with wild type (Fig. 1F). The presence of 16:3-containing TAG species in wild-type leaves was confirmed by mass spectrometry (fig. S4). The 16:3 acyl group is predominately found esterified to MGDG in Arabidopsis (22), indicating that DAG produced by GGGT activity is, in part, further acylated to TAG. DAG amounts did not change during freezing, although 16:3 was found specifically in freeze-treated wild-type plants (fig. S5). DAG derived from MGDG has also been shown to be converted to phosphatidic acid during freezing (12, 13). These results show that GGGT is activated as a result of freezing treatment, that the sfr2 mutants are defective in GGGT activation, and that DAG produced by GGGT during freezing is further metabolized to TAG, a nonpolar lipid not found in membranes.

Fig. 1

Freeze-induced galactolipid remodeling is not observed in sfr2 mutants. (A) Appearance of wild-type (Col2) and sfr2 mutants at 5 days after freezing treatment. (B) Thin-layer chromatogram of lipid extracts stained for glycolipids with α-naphthol reagent from cold-acclimated (CA) or freeze-treated (FT) wild-type (WT) and sfr2 plants. (C) Changes in galactolipid amounts in plants treated as in (B). *P < 0.05 or **P < 0.01 versus WT CA levels for three biological repeats. Error bars indicate SD. (D) Thin-layer chromatogram of neutral lipids visualized by H2SO4 and charring from WT and sfr2 plants treated as in (B). (E) Percent of total fatty acids esterified to TAG in plants treated as in (B); the average and standard deviation of at least three biological repeats are shown. (F) Fatty acid profile of TAGs plants treated as in (B) shown on a mol % basis of the average and standard deviation of at least three biological repeats; **P < 0.01 or ***P < 10−3 versus WT FT levels. The CA-treated control values for sfr2-3 were 44.7 ± 2.0 mol % (average ± SD) MGDG and 18.7 ± 0.7 mol % DGDG and for sfr2-4 were 43.9 ± 1.3 mol % MGDG and 17.9 ± 0.7 mol % DGDG. These were statistically not different from values for WT CA-treated plants shown in (C). The labels along the x axis indicate the fatty acid species identified.

The SFR2 protein was produced in yeast by expression of the SFR2 cDNA and localized to the isolated membrane fraction. Incubation of SFR2-containing membranes with deoxycholate-dispersed MGDG resulted in the production of DGDG, TGDG, and TeDG, whereas the same incubation of empty vector control membranes did not (Fig. 2A). In a time course using [3U-H]Gal–MGDG (23), SFR2-containing membranes converted ~60% of labeled MGDG into DGDG and oligogalactolipid products in 3 hours. This conversion was markedly reduced by the addition of EDTA equivalent to Mg2+ ions present (10 mM) (Fig. 2, B and C) as was previously observed in chloroplast envelopes (23). The fatty acids 16:3 and 18:3, specific to the plant-derived MGDG added to the assay, were present in the DAG pool only in the SFR2 protein–containing reactions, indicating DAG is being produced from MGDG by SFR2 in vitro. Proton–nuclear magnetic resonance (NMR) spectra of DGDG isolated from a scaled-up reaction showed the glycosidic linkages to be all in the β-anomeric configuration (ββDGDG), which is distinct from the βαDGDG synthesized by the UDP-Gal–dependent DGD1/2 enzymes found in wild-type leaves during normal growth (15) (fig S6). The retention of configuration from βMGDG to ββDGDG by SFR2 is consistent with all other GH-1 enzymes, which are classified as retaining β-glycosidases with a broad range of substrate specificities (24).

Fig. 2

Recombinant SFR2 in vitro activity from transgenic yeast. (A) Thin-layer chromatogram of lipid extracts stained for galactolipids with α-naphthol reagent from in vitro GGGT activity assays using deoxycholate/MGDG dispersions incubated with membrane fractions from empty vector control (EV) or SFR2 expression strain (SFR2). The in vitro reactions were stopped after 1 hour of incubation. Lipids from the reactions were extracted, chromatographed, and compared with lipids from a freeze-treated leaf sample (FT Leaf). (B) In vitro GGGT activity using labeled [3U-H]Gal–MGDG. The data are presented as the percent of 3H label found in DGDG, TGDG, and TeDG over time. Empty vector and SFR2 were assayed as in (A), and a SFR2 membrane protein assay containing EDTA at a concentration equimolar to Mg2+ present (10 mM) is also shown. Data are representative of two measurements, and the standard deviation is shown if it exceeds symbol size. (C) Percent of [3H] label found in different galactolipid species in the SFR2 assay shown in (B). (D) Fatty acid composition of diacylglycerol in empty vector control (EV) and SFR2 reactions at 3 hours. Data are presented on a mol % basis as the average and standard deviation of three biological repeats.

Many GH-1 family enzymes are known to have moderate transglycosidase activity under certain in vitro conditions, and examples for which transglycosylation predominates over hydrolysis are known in other GH families (e.g., GH-70 glucansucrases) (20). SFR2, which is the most divergent among 253 GH-1 family proteins (25), has evolved to catalyze MGDG transgalactosidation (GGGT activity) by exclusion of water as an acceptor, which could be attributed to several intrinsic features of SFR2, including the exclusory binding of acceptor galactolipids and/or the close apposition of the catalytic domain to the membrane surface. As such, SFR2 may provide insight into improving the transglycosidase activity of other GH-1 enzymes to produce oligosaccharides of medical or industrial importance (26). Orthologs of SFR2 are found in all land plants with completed genomes (9). Intriguingly, SFR2 is also conserved in species with no tolerance to freezing (e.g., tomato, maize, and rice), suggesting that SFR2 function is not restricted to freezing protection (9). There is a large overlap in the mechanisms required for freezing tolerance and dehydration because of water deficit or high salinity (3, 4), and SFR2 orthologs in these species may act on membrane stabilization during other abiotic stresses that cause cellular dehydration. Indeed, infiltration of Arabidopsis leaves with any osmotically active compound tested induced GGGT activity (table S1).

Our hypothesis for the requirement of SFR2-dependent galactolipid remodeling in freezing tolerance centers on the prevention of membrane fusion from the formation of non-bilayer HII-type structures brought about by dehydration. During dehydration, non-bilayer structures are formed at the interface of apposed membranes and are believed to initiate at the chloroplast envelope membranes during freezing (1, 2, 8). This results in fusion between bilayers (27), particularly when membranes are enriched in glycerolipid species with relatively small head groups (e.g., MGDG and phosphatidylethanolamine), because these show a higher propensity for transition to the HII phase in vitro. Previously, sfr2 mutants showed extensive chloroplast and tonoplast rupture in leaves during freezing recovery, which was proposed to arise from fusion of destabilized membranes (9). Here, we have shown that SFR2 partially converts MGDG to DGDG and oligogalactolipids, the latter of which are not prone to form HII phases in vitro. In addition to the prevention of non-bilayer–type structures, accumulation of oligogalactolipids results in an increased average thickness of the head-group domain of the bilayers and an increase in the localized concentration of hydroxyl groups per unit surface area, which enhance the repulsive hydration force between apposed bilayers during freeze-induced dehydration (28). Together, these factors could promote a sufficient distance between apposed bilayers that prevents membrane fusion. The overall phenomenon is analogous to the UDP-glucose–dependent modulation of mono- to diglucolipid ratios observed in Acheloplasma laidlawii under different abiotic stress conditions (29). One distinction is that DAG is produced by SFR2 and is further metabolized to TAG and possibly other lipid species, thereby preventing the accumulation of DAG, which can form nonlamellar phases. In this regard, the action of SFR2 provides a mechanism by which polar membrane lipids and excess membrane are removed by conversion to nonpolar lipids (e.g., TAGs) to accommodate a shrinking organelle after freezing or, more generally, osmotic stress.

Supporting Online Material

Materials and Methods

Figs. S1 to S6

Table S1


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank S. Marshall and T. Durret for help with proton-NMR and mass spectrometry analyses, respectively, and the laboratory of M. Thomashow for use of equipment and advice on the plant freezing experiments conducted in this work. This work was supported in parts by a grant from the U.S. Department of Energy (DE-FG02-98ER20305) and Michigan Agricultural Experiment Station.
View Abstract

Stay Connected to Science

Navigate This Article