Land plants have evolved a variety of mechanisms to enable them to grow and reproduce in environments where water is in short supply. Resurrection plants are a group of desiccation-tolerant plants that have adapted to environments where the rainfall is extremely sporadic. These plants are capable of withstanding severe drought stress, retaining less than 5% of their total water and then reviving completely and recommencing normal metabolism and growth within approximately 24 h of watering ( Gaff 1989). In the majority of higher plants only the mature seed and the pollen grain have this characteristic, and some of the drought-responsive genes isolated from the vegetative tissue of resurrection plants ( Bartels & Nelson 1994; Blomstedt et al. 1998a , b; O’Mahony & Oliver 1999) are also believed to be associated with desiccation tolerance in the embryos of many angiosperms ( Dure 1993; Bray 1993; Bartels & Nelson 1994).
Several regulatory proteins which are thought to mediate gene expression in response to drought or cold treatment via abscisic acid (ABA)-dependent pathways and ABA-independent pathways ( Jonak et al. 1996 ; Shinozaki et al. 1998 ) have been identified in desiccation-sensitive plant species such as Arabidopsis and alfalfa, and include protein kinases, protein phosphatases and transcription factors ( Jonak et al. 1996 ; Meskiene et al. 1998b ; Shinozaki et al. 1998 ).
ABA plays a key role in the induction of desiccation tolerance in the monocotyledonous resurrection plant Borya constricta ( Gaff & Loveys 1984) and in the dicotyledonous resurrection plant Craterostigma plantagineum ( Ingram & Bartels 1996). A C. plantagineum gene of unknown function (CDT-1), isolated by Furini et al. (1997) , is thought to act downstream of ABA and can induce desiccation tolerance when over-expressed in callus tissue. In addition, drought-responsive genes encoding a root-specific myb-like product (cpm7) and two members of the homeodomain–leucine zipper (HD-Zip) family have been isolated from C. plantagineum, and identification of the target genes for these DNA-binding proteins will elucidate their role in drought tolerance in this resurrection dicot ( Frank et al. 1998 ; Iturriaga et al. 1996 ).
We have selected Sporobolus stapfianus as a model system for examining the molecular mechanisms of desiccation tolerance in a monocotyledonous representative of the resurrection plants, with the expectation that the findings will have relevance for manipulating drought resistance in many crop and pasture species. The physiological features of the tolerance of this plant to drought have been studied extensively (reviewed in Gaff 1997; Oliver 1996). Desiccation tolerance in the leaf tissue of S. stapfianus is thought to be induced via an ABA-independent process as the plant dries down to 60% relative water content (RWC). Leaves detached from the plant after this stage of drying retain viability, whereas leaves detached before this stage are not viable on further drying ( Gaff & Loveys 1984). Analyses of proteins extracted at various stages of desiccation from the tolerant S. stapfianus and a desiccation-sensitive member of the Sporobolus genus, S. pyramidalis, indicate that there are two phases of novel protein synthesis in S. stapfianus associated with the development of desiccation tolerance. The first phase occurs at mild to moderate drought stress (85–50% RWC) and the second at extreme drought stress at around 40% RWC ( Kuang et al. 1995 ).
Genes whose transcripts are present at very low levels during the early stages of dehydration and whose expression profile alters during severe drought stress may play a significant role in the ability of S. stapfianus to survive severe desiccation. To isolate these types of genes, a technique was used which selects ‘cold plaque’ clones ( Hodge et al. 1992 ). These ‘cold plaques’ were distinguished by their inability to be visualized during differential hybridization with cDNA probes synthesized from mRNA of either fully hydrated or moderately desiccated leaf material of S. stapfianus. This suggests that the cold plaques represent gene transcripts which are not present in the probe population at high enough levels to allow detection and therefore may be constitutively present at low levels in the cell or may require specific conditions to be induced. Analysis of a large number of cold plaques present in a library prepared from a range of desiccated S. stapfianus tissue resulted in the identification of several low-abundance transcripts expressed specifically during severe drought stress.
MATERIALS AND METHODS
Plants of the desiccation-tolerant species S. stapfianus Gandoger and the related desiccation-sensitive S. pyramidalis Beauv. species were grown in a constant-temperature room at 28 °C and 16 h day/8 h night and harvested as previously described ( Blomstedt et al. 1998a ). In addition, S. stapfianus leaf tissue detached from a fully hydrated plant was allowed to dry to the desired RWC in a constant-temperature room ( Blomstedt et al. 1998a ). Drying of leaf tissue detached from S. stapfianus does not confer desiccation tolerance and hence this material was used as an additional desiccation-sensitive sample. Treatment with ABA consisted of spraying leaves and watering fully hydrated plants with 2·5 m M ABA in water containing 0·2% Triton X-100 as a surfactant. Control plants were treated with water/Triton X-100 only.
cDNA library construction and screening
A cDNA library, constructed from mRNA of S. stapfianus at four stages of desiccation, 83%, 68%, 56% and 10% RWCs, using a ZAP-cDNA synthesis kit from Stratagene (La Jolla, California, USA) was screened as previously described ( Blomstedt et al. 1998b ). A large number of plaques that did not hybridize with either the fully hydrated or the 83% RWC probe were detected. These ‘cold plaques’ were selected for analysis. Secondary screening was conducted on amplified inserts following T3 and T7 primed PCR. The PCR products were run on 0·8% agarose gels, transferred to Hybond N+ (Amersham) and probed with cDNA prepared from S. stapfianus leaf material at different levels of RWC to determine the expression profile of the corresponding gene during desiccation. Clones deemed to be of interest were excised according to the instructions supplied by Stratagene, and the resulting plasmids designated pSDG (Sporobolus drought gene) clones.
RNA isolation and Northern analysis
Total RNA was isolated using a phenol extraction/LiCl precipitation method as detailed in Blomstedt et al. (1998a) . Poly(A+) RNA was isolated from the total RNA using Dynabeads Oligo (dT)25 magnetic beads according to the manufacturer’s instructions (Dynal, Oslo, Norway). Transcript levels in S. stapfianus were determined in both desiccation-tolerant tissue as well as desiccation-sensitive tissue. The latter was obtained by harvesting tissue from fully hydrated plants and then allowing the tissue to dehydrate. Tissue was also harvested from a desiccation-sensitive related species S. pyramidalis and included in the Northern analysis. Northern analysis was performed using total RNA as detailed in Blomstedt et al. (1998a) .
Plasmid DNA was isolated from selected clones using an alkaline lysis method ( Sambrook, Fritsch & Maniatis 1989) and sequenced using an ABI automated sequencer and the ABI dye terminator sequencing kit according to the manufacturer’s instructions (Perkin-Elmer). Analyses of sequence data and database searches were conducted utilizing programs accessed through the Australian National Genomic Information Service (ANGIS). Sequences isolated in this study have been entered into the EMBL nucleotide sequence database, accession numbers AJ242804, AJ242803, AJ242802, AJ242801, AJ242805, AJ242806 for clones SDG43c, SDG37c, SDG137c, SDG134c, SDG50c and SDG69c, respectively.
We have previously reported the isolation of several S. stapfianus genes whose transcripts increase in abundance in the early stages of water loss and remain at relatively high abundance throughout the dehydration process ( Blomstedt et al. 1998a , b). While many of the conventional differential screening techniques are suited to those genes whose transcripts are produced in abundance, they do not easily allow direct isolation of genes with low transcript levels. In the present study, we have utilized the cold plaque strategy to detect lowly transcribed genes represented in a cDNA library prepared from S. stapfianus undergoing dehydration, with the aim of identifying medium- to low-level transcripts which alter in abundance at severe stages of drought stress.
In the primary screen, 155 cold plaques were selected. Almost half of the PCR-amplified inserts from these cold plaques were still not detected after probing, and are likely to represent genes expressed at very low abundance at all stages of water stress. However, tertiary screening of a selected subset of clones indicated that several cold plaques represented genes expressed at medium to low levels whose transcripts alter in abundance at severe stages of drought stress. Further characterization indicated that increases observed for several of the transcripts occur only in desiccation-tolerant tissue during severe drought stress, and thus these genes may be of particular interest in the study of desiccation tolerance in S. stapfianus.
Genes which exhibited transcript accumulation specifically in drought-stressed desiccation-tolerant tissue were found to encode an eIF1 translation initiation factor, two drought stress-inducible glycine-rich proteins, a tonoplast-intrinsic protein (TIP) and an early light-inducible protein (ELIP). A serine/threonine phosphatase type 2C gene was also isolated whose transcript was present during moderate drought stress but dropped below detectable levels during severe water deficit.
This is the first report suggesting that a gene encoding an eIF1 translation initiation factor may have a role in the drought stress response of plants. Although accumulation of protein initiation factor eIF1 transcripts during environmental stress has not previously been reported, accumulation of transcripts encoding translation elongation factor eEF1A during cold treatment have been reported in barley and maize ( Dunn et al. 1993 ; Berberich et al. 1995 ) and during wounding and exposure to low oxygen levels in potato ( Morelli, Shewmaker & Vayda 1994; Vayda, Shewmaker & Morelli 1995). It has been proposed that an increase in eEF1A may allow rapid synthesis of proteins required to respond to those stresses ( Kidou & Ejiri 1998). The appearance of higher levels of eIF1 transcripts during the mid stage of dehydration in S. stapfianus coincides with the phase of novel protein synthesis previously reported to occur at this level of drought stress in this resurrection plant ( Kuang et al. 1995 ). It is possible that SDG134c encodes a novel isoform of eIF1 that is capable of operating during severe drought stress; however, this seems unlikely as there are only two amino acid differences between the S. stapfianus sequence and that of maize and only one difference with the rice sequence. In the desiccation-tolerant tissue of S. stapfianus, the presence of this transcript persists, with the highest levels being observed in completely dehydrated plants. It is generally assumed that in desiccation-tolerant plants many of the molecular components essential for re-initiation of metabolic activity have survived the desiccation process and are present in the dried tissue. It is likely that rapid protein synthesis is also required during the rehydration process which leads to replacement or repair of damaged or lost components and full metabolic activity being achieved within several hours.
The protein encoded by SDG137c contains several glycine-rich GHGG repeat motifs, is highly hydrophilic and does not appear to contain an N-terminal signal sequence. A number of GRPs, some of which lack signal sequences, have been shown to be expressed in response to a variety of environmental stress conditions including drought and low temperature ( Luo et al. 1992 ; Showalter 1993) and may be associated with stabilizing membranes against damage during dehydration ( Thomashow 1998).
The glycine-rich motifs found in SDG137c are similar to those occurring in the C-terminal domain of the drought-responsive SDG43c gene product which has several glycine-rich GGHG repeat motifs interspersed with a KFK repeat motif. The N-terminal domain of SDG43c, however, contains 11 proline-rich repeats which are identical (GYPPQ) or similar (GAYPPPP) to those found in a previously isolated S. stapfianus drought-responsive gene, SDG7 ( Blomstedt et al. 1998a ). Plant proteins containing GYPPQ repeats had not previously been linked to drought stress. Similar GYP repeat regions are found in the cytosolic domain of several membrane-associated animal proteins ( Döring, Schleicher & Noegel 1991; Ovchinnikov et al. 1988 ). The N- and C-terminal domains of the SDG43c protein are separated by a run of alanines which form the basis of a central hydrophobic domain of 21 amino acids. This hydrophobic region is likely to form a helical structure according to the Chou and Fasman prediction ( Chou & Fasman 1974).
This SDG43c gene product has 66% identity at the amino acid level with a glycine- and proline-rich protein (GPRP) encoded by a gene isolated from Arabidopsis ( Marty et al. 1996 ). SDG43c also has a similar level of identity to an unpublished carrot sequence (EMBL X72383) which appears to encode a truncated version of a similar protein. Neither of these dicot sequences have yet been associated with drought stress. There appears to be no extracellular signal sequence encoded by SDG43c, nor is there one encoded by the Arabidopsis GPRP; however, Marty et al. (1996) have shown that the Arabidopsis GPRP interacts with membranes, presumably via the central hydrophobic domain. This has led to the suggestion that the GPRP may have a role in vesicle-associated intracellular transport or in linking cellular components through the plasma membrane with the cell wall ( Marty et al. 1996 ). Several studies have proposed that tyrosine residues in repeat regions similar to those occurring in SDG7 and the N-terminus of SDG43c may be involved in the formation of cross-links with GRPs and other cell wall components to reduce cell wall elasticity in response to various stresses, including water deficit ( Bradley, Kjellbom & Lamb 1992; Meier et al. 1992 ; No et al. 1997 ). This presents an attractive hypothesis for a proposed role for the S. stapfianus GPRP in drought tolerance, whereby interaction of the GPRP with the plasma membrane and cell wall may help to maintain the integrity of the cell during dehydration or rehydration.
In S. stapfianus, the SDG69c ELIP transcript is not detected in fully hydrated plants or in plants undergoing mild to moderate drought stress, and the transcript is only detected at low levels as the plant is exposed to very severe water deficit. ELIPs are homologous to the cbr gene isolated from Dunaliella bardawil ( Lers, Levy & Zamir 1991). This gene has been associated with zeaxanthin-related non-photochemical fluorescence quenching under stress conditions ( Braun et al. 1996 ). A number of studies suggest that ELIPs bind the carotenoid zeaxanthin during drought stress and provide a mechanism for non-photochemical quenching of fluorescence without the requirement for a pH gradient across the thylakoid membrane. The quenching of excess light energy during water-deficit conditions may help prevent oxidative damage to the photosynthetic apparatus during dehydration ( Casper, Eickmeier & Osmond 1993).
The expression patterns of ELIPs in response to drought stress and the application of ABA suggest that there are significant differences in ELIP gene regulation, not only between drought-sensitive and drought-tolerant plants but also between some drought-tolerant resurrection species. Drought stress-induced ELIP transcript accumulation appears to be significantly greater in C. plantagineum than S. stapfianus. Differences in regulation of ELIP transcript levels between species may reflect differences in the biological and morphological mechanisms which protect these plants from photoinhibitory damage during dehydration. Several studies have suggested that some resurrection plants, including S. stapfianus, initiate a partial systematic disassembly of the photosynthetic apparatus at an early stage in the dehydration process, thus avoiding potential photoinhibitory damage ( Casper et al. 1993 ; Di Blasi et al. 1998 ). In S. stapfianus, non-photochemical quenching was observed not to increase significantly until a RWC of 40% was reached ( Di Blasi et al. 1998 ). This finding correlates well with the water-deficit level at which the ELIP transcript is observed to increase ( Fig. 2d). It is noteworthy that field-dried S. stapfianus accumulates a high level of a water-soluble pigment that colours the outer leaf an intense purple–black which may protect the photosynthetic apparatus from excessive light intensities, whereas much less pigmentation is observed in drying C. plantagineum (light purple to grey).
Genes encoding water-specific or solute channel proteins have been shown to be inducible by osmotic stress in pea ( Guerrero, Jones & Mullet 1990), rice ( Liu et al. 1994 ), Arabidopsis ( Höfte et al. 1992 ; Yamaguchi-Shinozaki et al. 1992 ) and C. plantagineum ( Mariaux et al. 1998 ). A comparison of the expression patterns of γTIP transcripts during dehydration of S. stapfianus and the resurrection dicot C. plantagineum reveals significant differences. In C. plantagineum, the expression of the γTIP is transient, reaching a maximum after only 1 h of stress and then decreasing to less than control levels after 4 h ( Mariaux et al. 1998 ). By contrast, the S. stapfianus transcript accumulates progressively throughout the dehydration process and reaches a very high level in air-dry tissue. This suggests that these genes are regulated differently in the two desiccation-tolerant species, a situation which may reflect a difference in function. Alternatively, the differences in the expression patterns in the two resurrection species may merely have resulted from differences in the experimental treatment of sample tissues. High-level accumulation of the γTIP transcript in S. stapfianus is confined to desiccation-tolerant tissue and suggests that the S. stapfianus SDG50c γTIP may be required for the cell to accommodate the large deviations in osmotic potential that are experienced during dehydration and rehydration and to facilitate the rehydration process by raising the water permeability across the tonoplast membrane.
The S. stapfianus SDG37c gene encoding the type 2C protein Ser/Thr phosphatase (PP2C) has identity with a PP2C gene, MP2C, from alfalfa ( Meskiene et al. 1998a ) as well as the Arabidopsis PP2C genes ABI1 and ABI2 ( Leung et al. 1997 ; Meyer, Leube & Grill 1994). The Arabidopsis ABI PP2Cs play a role in signal transduction associated with stress-related ABA-dependent gene expression, whereas the alfalfa PP2C acts in an ABA-independent pathway. The PP2C activity of the Arabidopsis proteins can negatively regulate the ABA signal transduction pathway controlling both induction and repression of gene transcription in response to ABA ( Sheen 1998).
The alfalfa PP2C, MP2C, was found to deactivate a mitogen-activated protein kinase (MAPK) pathway ( Meskiene et al. 1998b ). This stress-activated MAP kinase (SAMK) pathway is activated in response to drought stress as well as wounding, cold and touch, and may act independently of ABA to allow rapid induction of cold- and drought-related gene expression ( Jonak et al. 1996 ). Jonak et al. (1996) have also suggested that the SAMK pathway may induce ABA synthesis by activating synthetic enzymes either directly or via induction of ABA synthesis genes. The expression of alfalfa MP2C follows a peak in SAMK activity, suggesting that it acts as a negative regulator of the SAMK pathway. Using transcription and translation inhibitors, Meskiene et al. (1998b) have shown that inactivation of the SAMK pathway in alfalfa requires de novo expression of the MP2C gene which is thought to result from the activity of the SAMK pathway itself and thus provide a mechanism to attenuate the stress signal by a negative feedback mechanism.
SDG37c has significantly closer identity to alfalfa MP2C than to any of the Arabidopsis PP2Cs. The identity between SDG37c and MP2C is 58%, which is 8% higher than between the Arabidopsis PP2Cs, ABI1 and AtPP2C ( Kuromori & Yamamoto 1994), which have been shown to have partially redundant functions ( Sheen 1998). Notably, transcription of the ABI1 and ABI2 genes is induced by ABA ( Leung et al. 1997 ), whereas the SDG37c gene is not. While the shared identity between the S. stapfianus PP2C and the alfalfa PP2C suggests that the SDG37c gene product may act as a negative regulator in an ABA-independent MAPK pathway in S. stapfianus, the pattern of SDG37c transcript accumulation in S. stapfianus indicates that these genes may not have completely orthologous functions. The alfalfa MP2C gene operates in a drought/cold-activated pathway and is induced following stress to attenuate the stress response, whereas the S. stapfianus SDG37c transcript is present at detectable levels in fully hydrated plants and does not appear to accumulate during cold-treatment. It is likely however, that more detailed experiments to detect transient increases in SDG37c transcript levels are required. SDG37c transcripts are not detectable below 44% RWC, which suggests that this gene product does not operate during severe water deficit. This may indicate that the signalling pathway with which SDG37c may be associated is permanently activated during continuing severe water loss, or alternatively that the pathway has ceased operation at this level of severe water deficit. The latter hypothesis may be more likely given that the homologous transcript levels in the desiccation-sensitive species S. pyramidalis drop below detectable levels at a much earlier stage of water loss (around 80% RWC, unpublished observation). Experiments are underway to examine transcript levels during rehydration and to obtain direct evidence of water-deficit related MAPK-dependent gene expression in S. stapfianus.
In the majority of higher plants, only the seed and pollen have the ability to withstand complete desiccation, and a number of the genes activated in the early stages of dehydration in resurrection plants are similar to those expressed in the desiccating seed of most plants ( Blomstedt et al. 1998a , b). This has led to the hypothesis that resurrection plants may possess regulatory genes which allow the desiccation processes of the seed and pollen to occur in the vegetative tissue. Previous gene expression studies in this laboratory have also shown that many of the early responses to mild dehydration are similar in desiccation-sensitive and desiccation-tolerant plants, with several genes showing a fairly rapid increase in abundance in response to moderate water stress ( Blomstedt et al. 1998a , b). However, several of these genes have elevated transcript levels which persist throughout the entire drying process in the desiccation-tolerant S. stapfianus, but not in comparative desiccation-sensitive tissues ( Blomstedt et al. 1998a ). This suggests that these gene products may be required to protect the plant from severe dehydration or to allow successful rehydration and the restoration of metabolic activity.
The desiccation-tolerant tissue of S. staphianus appears to retain the ability to increase transcript levels of specific genes even at extreme levels of dehydration, with several of the drought-responsive genes identified in this study showing significant expression only at severe stages of desiccation. The comparative studies suggest that this ability is largely restricted to desiccation-tolerant tissue and that these gene products may be associated with the ability of S. stapfianus to withstand levels of extreme water loss. The putative functions of these genes also provide some understanding of the molecular mechanisms which may be associated with conferring desiccation tolerance in this resurrection species. The alteration in transcript levels of these genes, which the present study indicates can occur almost throughout the entire dehydration process, suggests that there may be several unique key regulatory genes capable of operating at successive levels of water loss to confer desiccation tolerance in the vegetative tissue of this resurrection plant. Current studies using those SDG genes expressed specifically in desiccation-tolerant tissue at severe stages of water loss are aimed at identifying regulatory genes controlling their expression and addressing the question of whether similar mechanisms operate during the desiccation stage of seed development.