Drought- and desiccation-induced modulation of gene expression in plants


Correspondence: Dorothea Bartels. Fax: + 49 228 732689; e-mail: dbartels@uni-bonn.de


Desiccation is the extreme form of dehydration. Tolerance of desiccation is acquired by seeds and in resurrection plants, a small group of angiosperms. Desiccation tolerance is the result of a complex cascade of molecular events, which can be divided into signal perception, signal transduction, gene activation and biochemical alterations leading to acquisition of tolerance. Many of these molecular processes are also observed during the dehydration of non-tolerant plants. Here we try to give an overview of the gene expression programmes that are triggered by dehydration, with particular reference to protective molecules and the regulation of their expression. Potential transgenic approaches to manipulating stress tolerance are discussed.


As sessile organisms, plants have developed adaptive strategies to cope with environmental stress. Understanding the responses of plants to their external environment is of importance with respect to basic research, but it is also an attractive target for improving stress tolerance. In any organism, one of the fundamental properties of adaptive mechanisms is that they must be inducible in response to the external threat, to ensure that the resources are used when required. Accordingly, plant cells have evolved to perceive different signals from their surroundings, to integrate them and to respond by modulating the appropriate gene expression.

Definition of water stress and desiccation tolerance

For the purpose of distinction, water stress in general is a mild form of water deficit, which is a condition where the water status of plants undergoes relatively small changes (Bray 1997). Desiccation means losing most of the protoplasmic ‘free or bulk’ water and surviving with only the ‘bound water’– water associated with the cell matrix. Desiccation tolerance is the result of a dynamic process and appears to be mediated by protective systems that prevent lethal damage. Desiccation of seeds is considered to be a necessary prerequisite for the completion of the life cycle of most higher plants. This allows the plant to store seeds; this is generally interpreted as a strategy to enable seed survival and ensure better dissemination of the species. Plants are very sensitive to desiccation during the vegetative phase of their life cycle, and very few plants acquire desiccation tolerance in the vegetative tissues. These include a small group of angiosperms, termed resurrection plants (Gaff 1971).

Plants have evolved two major mechanisms to cope with water deficit: stress avoidance and tolerance. Stress avoidance may be the formation of seeds before drought conditions prevail; avoidance is also achieved by specialized adaptations in the plant architecture. Morphological adaptations are, for example, the development of specialized leaf surfaces to decrease the rate of transpiration, the reduction of leaf area, sunken stomata or an increase in root length and density to use water more efficiently. A complex trait, water stress tolerance appears to be the result of the co-ordination of physiological and biochemical alterations at the cellular and molecular level: i.e. the accumulation of various osmolytes and late embryogenesis-abundant (LEA) proteins coupled with an efficient antioxidant system. Many of these mechanisms have been characterized. They have been found to exist both in desiccation-tolerant and non-tolerant plants.

Approaches for studying desiccation tolerance

An important contribution to our understanding of the mechanism of desiccation tolerance is derived from ‘resurrection plants’, which can survive even with < 5% of their total water in the vegetative tissues and are able to regain normal metabolism and growth within several hours of rewatering. Molecular studies of resurrection plants have been performed with a few species, such as Craterostigma plantagineum (Phillips & Bartels 2000) and Sporobolus stapfianus (Neale et al. 2000).

Desiccation-tolerant plants do not lend themselves easily to genetic approaches. However, DNA activation tagging is a possible tool to isolate dominant mutants, and this was applied successfully in callus of C. plantagineum. The CDT-1 mutant from C. plantagineum was isolated in this way, but the function of the CDT-1 gene is not yet understood. The desiccation tolerance of C. plantagineum calli is dependent on the exogenous application of abscisic acid (ABA). The ectopic expression of CDT-1 conferred desiccation tolerance in the absence of ABA and resulted in the constitutive expression of several ABA- and dehydration-inducible genes (Furini et al. 1997).

Mittler et al. (2001) recently reported ‘partial plant dormancy’, a novel stress-responsive strategy in Retama raetam, an evergreen stem assimilating desert C3 plant. Dormancy is not induced in all plant parts: it is induced in stems of the upper canopy but not in those of the lower canopy. During the dry season, the upper canopy has low levels of antioxidative, photosynthetic and other housekeeping proteins compared with the lower canopy. In contrast to the protein levels, the transcript levels encoding the proteins were similar in the upper and lower canopies. Therefore, the dormancy was mediated through post-transcriptional suppression of gene expression as well as through suppression of photosynthesis. This dormancy was correlated with low relative water content (RWC), but not with extreme high temperature, excess light or low relative humidity, which are other common environmental constraints in deserts.

The phytohormone ABA has been shown to mediate the response to environmental stresses such as dehydration, salt and low temperature (Leung & Giraudat 1998). Increased endogenous ABA levels in response to dehydration have been reported in many physiological studies. Dehydration probably serves as a trigger for the accumulation of ABA, which in turn activates various stress-associated genes that are thought to exert protective mechanisms. Indeed, the characterization of ABA-deficient and -insensitive mutants from non-tolerant plants affected in stress tolerance has contributed towards an understanding of the regulation of gene expression in desiccation tolerance (Leung & Giraudat 1998; further references in the contribution by E. Bray, this issue). These mutants have led to the isolation of genes involved in ABA synthesis and various components involved in signalling during stress conditions.

Overview of the up-regulated genes and gene products

What distinguishes dehydration-tolerant plants from non-tolerant plants at the molecular level? Characterization of numerous genes induced by dehydration emphasizes that the answer lies not within a single gene. Recent microarray analysis on Arabidopsis plants using 1300 cDNAs have revealed that approximately 44 genes were up-regulated in response to dehydration, of which 30 are reported to be novel (Seki et al. 2001). Fully hydrated plants were allowed to dehydrate for 2–10 h for the expression analysis. However, every up-regulated gene does not necessarily have a role in adaptation: some might be induced because of damage caused by stress (Zhu 2000). Increasing evidence indicates that the genes responding to dehydration can be categorized into two classes, based on their response in terms of time-scale. Some respond immediately – within seconds or minutes – while others are responding later, in hours, days or even weeks. This allows us to speculate that the early responsive genes may provide initial protection and amplification of signals while the genes that are responding later may be involved in adaptation to stress conditions. In the following section, we summarize the characteristics of prevalent groups of genes that accumulate under water-stress conditions in many plant species.

LEA proteins

LEA are a major group of proteins that typically accumulate during the late stages of embryogenesis or in response to dehydration, low temperature, salinity or exogenous ABA treatment – indicating their responsiveness to cellular dehydration. LEA proteins are widely distributed among monocot and dicot species, and many different forms have been isolated molecularly. LEA proteins are characterized by a biased amino acid composition, by their high hydrophilicity and solubility in water and often by their solubility after boiling. The homology among different LEA proteins, the presence of conserved protein domains, their ubiquity and the developmental specificity of their expression implies a fundamental role in desiccation tolerance. It has been proposed that these proteins may play a role in protecting cytoplasmic structures during dehydration.

LEA proteins have been divided into groups based on predicted biochemical properties and motifs with sequence similarities (Dure et al. 1989; Ingram & Bartels 1996; Cuming 1999). A Characteristic feature of group 1 LEA proteins is a 20-amino-acid motif. The wheat Em protein represents this group and homologues have been identified in a wide range of plant species. Group 2 LEA proteins, also referred to as dehydrins, are the best-studied LEA proteins. Dehydrins are characterized by a lysine-rich 15-amino-acid motif (K-segment; EKKGIMDKIKEKLPG), predicted to form an amphipathic α-helix, a tract of contiguous serine residues and a conserved motif containing the consensus sequence DEYGNP in the N-terminal part. Group 3 LEA proteins share a characteristic repeat motif of 11 amino acids. Dure et al. (1989) predicted that the 11-amino-acid peptide forms an amphipathic α-helix with possibilities for intra- and inter-molecular interactions. Group 4 LEA proteins are characterized by a conserved N-terminus, predicted to form α-helices and a diverse C-terminal part with a random coil structure. Group 5 LEA proteins contain more hydrophobic residues than the other groups; they are not soluble after boiling and are likely to adapt a globular structure (Cuming 1999).


Under non-stress conditions, plants keep their water balance by adjusting the water conductance of their tissues. Vascular tissues and guard cells play an important role in this process. A significant component in cellular water transport are aquaporins (Maurel & Chrispeels 2001). Aquaporins are a complex family of channel proteins that facilitate the transport of water along transmembrane water potential gradients. Aquaporins can regulate the hydraulic conductivity of membranes and potentiate a 10–20-fold increase in water permeability (Maurel & Chrispeels 2001). Aquaporins have a potential role in plant water relations: the regulation of expression and activity are modulated by dehydration. Several genes that encode aquaporins are up-regulated by dehydration, for example rd28 from Arabidopsis (Yamaguchi-Shinozaki et al. 1992) or the tomato-ripening-associated membrane protein (TRAMP) (Fray et al. 1994). In C. plantagineum, several aquaporins were up-regulated by dehydration; some are inducible both by dehydration and ABA, while others are inducible by drought alone, suggesting the involvement of ABA-dependent and -independent signalling pathways (Mariaux et al. 1998).

Lipid transfer proteins

Trevino & O’Connell (1998) have reported the induction of three nsLTPs (non-specific lipid transfer proteins) in Lycopersicon pennellii by drought conditions. The gene products are implicated in cuticle biosynthesis. Presumably, the induction of nsLTP in epidermal-cell-specific expression represents an adaptive response to drought stress, through which the plant may be able to reduce water loss by increasing cuticle thickness.

Protection of photosynthetic structures

The physical properties of the photosynthetic apparatus are of crucial importance in desiccation-tolerant plants. The photosynthetic apparatus is very sensitive and liable to injury, and needs to be maintained or quickly repaired upon rehydration (for a review see Godde 1999). Membrane proteins are of particular importance for the functionality of the photosynthetic apparatus. In C. plantagineum, three genes expressed preferentially upon desiccation were found to encode chloroplast-localized stress proteins (DSP) (Schneider et al. 1993). Immunological studies revealed that two proteins, DSP22 and DSP34, were located in the thylakoids, while DSP21 was found in the stroma. These proteins were proposed to play a role in the maintenance of chloroplast structures. Furthermore, it was shown that the synthesis of DSP22 is dependent on the extent of photoinhibitory damage (Alamillo & Bartels 2001). Pruvot et al. (1996) also reported the accumulation of a chloroplastic drought-induced protein (CDSP34) in the thylakoids of Solanum tuberosum. It was suggested that this protein stabilizes the thylakoid membranes. Rey et al. (1998) have characterized dehydration-induced CDSP32, a stromal thioredoxin-like protein from potato, whose expression is independent of ABA. They showed that transcript levels were up-regulated under severe stress conditions. In animal cells, thioredoxin has been shown to be involved in repair or protection mechanisms by regenerating proteins inactivated by oxidative stress (Fernando et al. 1992). Rey et al. (1998) hypothesized that CDSP32 might play a role in the preservation of native protein structures by reducing intermolecular disulfide bonds.

Proteinase inhibitors

Proteolytic activity is high during stress conditions and it has been shown to coincide with programmed cell death (Fukuda 1996). In both plants and animals, cysteine proteases are involved in programmed cell death, which can be prevented by cysteine proteinase inhibitors (Solomon et al. 1999). In plants and other organisms, protease activity can be regulated by specific protease inhibitor proteins. Plants possess many protease inhibitor genes, which are mostly known for their function in the defence against herbivores. Solomon et al. (1999) have shown that the inhibition of the cysteine proteases through ectopic expression of cystatin, an endogenous cysteine protease inhibitor gene, inhibited induced cysteine protease activity and blocked programmed cell death. BnD22 from rapeseed or WSCP from cauliflower represent members of the family of Kunitz-type proteinase inhibitors, which are induced by dehydration and salinity (Downing et al. 1992; Nishio & Satoh 1997). It is conceivable that some of the dehydration-inducible genes encoding proteinase inhibitors are likely to protect the proteins by inhibiting the activity of proteases, some of which have been described as being also induced by water deficit.

Oxidative stress

A consequence of many environmental stresses – including dehydration – is oxidative stress, i.e. the accumulation of reactive oxygen species (ROS), which damage cellular structures (Smirnoff 1993). Under optimal conditions, leaves are equipped with sufficient antioxidant enzymes and metabolites to cope with ROS. The accumulation of enzymes such as superoxide dismutases, ascorbate peroxidases, catalases, glutathione-S-transferases (GST) and glutathione peroxidases has been observed during stress conditions. The capacity of the antioxidative defence system determines the fate of the cell and whether the cell continues to function or suffers photo-oxidation (Foyer, Descourvieres & Kunert 1994).

Proteins involved in repair

Besides the loss of water, modifications of protein residues such as deamination or oxidation result in the disruption of native protein conformation or protein activity. Therefore, repair mechanisms may be involved in the genetic programme of desiccation tolerance. In accordance with this, enzymes that are capable of repairing stress-damaged proteins have been identified and are likely to contribute to water deficit tolerance. For instance, the synthesis and activity of l-isoaspartyl methyltransferase (which converts modified l-isoaspartyl in damaged proteins back to l-aspartyl) were shown to be enhanced in wheat seedlings subjected to water deficit (Mudgett & Clarke 1994). The repair aspect of the desiccation tolerance mechanism seems to be prevalent in desiccation-tolerant mosses. Gametophytic cells of desiccation-tolerant mosses undergo substantial disruption of cellular integrity. However, the cells do not die and return to a normal appearance (Oliver & Wood 1997). In the desiccation-tolerant moss Tortula ruralis, the synthesis of specific proteins (rehydrins) is induced immediately upon rehydration, and it has been suggested that some of these proteins are involved in repair mechanisms.

Accumulation of osmolytes and soluble sugars

The accumulation of compatible solutes or osmolytes under osmotic stress is well known in many organisms. Osmolytes are synthesized in response to osmotic stress and do not interfere with normal cellular biochemical reactions. They help to maintain an osmotic balance under dehydration conditions (for a review see Bray, Bailey-Serres & Weretilnyk 2000). There are several examples of the accumulation of osmolytes contributing to the relatively high water content necessary for growth and cellular metabolism. Osmolytes include sugars, polyols, proline, quaternary ammonium compounds and tertiary sulfonium compounds. The increased synthesis of osmolytes is achieved by modulating genes encoding enzymes of the osmolyte biosynthetic pathway. For instance, simultaneous up-regulation of P-5-C synthase (P-5-CS) and down-regulation of the proline dehydrogenase (ProDH) gene leads to proline accumulation during water stress (Yoshiba et al. 1997). Furthermore, a transporter (LeProT1) involved in proline and glycinebetaine was also induced by water stress (Schwacke et al. 1999). This illustrates the complexity of the regulation of the proline level. Glycine betaine, another potent osmolyte, is described in detail in this issue by N. Murata.

The mechanism of protection provided by these osmolytes is not understood. Often the accumulation is not sufficient to account for osmotic adjustment. Therefore, it is reasonable to speculate that the accumulated osmolytes function not only in osmotic adjustment: other mechanisms are also conceivable (e.g. scavenging ROS) (Hong et al. 2000). The ability of osmolytes to enhance stress tolerance was shown by over-expression in transgenic plants (see later).

A common observation in the desiccation process is the accumulation of soluble sugars. Apart from their role in osmotic adjustment, the protective effects of sugars may also include protein stabilization (Carpenter, Crowe & Arakawa 1990). An example for sucrose accumulation during desiccation is given in the desiccation-tolerant plant C. plantagineum: dehydration induces the conversion of 2-octulose, an eight-carbon sugar, to sucrose (Bianchi et al. 1991). This conversion is correlated with increases in the gene expression for sucrose synthase (Sus) and sucrose phosphate synthase (Sps) (Ingram et al. 1997; Kleines et al. 1999). SUS and SPS are considered to be key enzymes involved in sucrose synthesis/metabolism. It is a general observation that the expression of Sus gene(s) is up-regulated during dehydration/osmotic stress conditions (Pelah et al. 1997; Dejardin, Sokolov & Kleczkowski 1999). Evidence of a role for SPS in sucrose accumulation during water stress was provided by antisense expression of SPS in potato plants, in which the decreased SPS activity completely suppressed the water-stress-induced stimulation of sucrose synthesis (Geigenberger et al. 1999). Thus SUS and SPS in plants have been shown to be vital for acclimation to dehydration.

Regulatory genes

Many genes that respond to dehydration have been described, and it is thought that gene products may be involved in dehydration tolerance (Ingram & Bartels 1996; Shinozaki & Yamaguchi-Shinozaki 1997; Bray 1997). However, such individual genes have only minimal effects on conferring tolerance. The identification of molecular switches and regulatory genes would provide a better tool for plant improvement strategies. A potential example has been set by Jaglo-Ottosen et al. (1998) and Kasuga et al. (1999), who overexpressed a transcription factor gene, leading to the activation of several down-stream stress-responsive genes and resulting in enhanced stress tolerance. Therefore, the knowledge of other regulatory genes and an understanding of their mode of action will provide an important starting point for the improvement of crop plants. Studies on dehydration-induced transcription factors in plants are just emerging, and for most of the identified transcription factors, the target genes are yet to be identified. Only a few of the dehydration-induced genes themselves encode transcription factors. It is presumably the interaction between these factors and pre-existing factors that ultimately determines the response leading to gene expression and stress adaptation. The complexity is shown in Fig. 1. To activate or repress transcription, transcription factors must be located in the nucleus, bind DNA and interact with the basal transcription apparatus. Accordingly, dehydration may affect any one, or a combination, of these processes. Several classes of transcription factors will be discussed with respect to their role in dehydration and desiccation.

Figure 1.

Biochemical, anatomical and morphological changes induced by dehydration leading to adaptation. Pathways that are commonly affected when plants are subjected to dehydration are indicated. Question marks indicate less well-known processes.

Basic-region leucine zipper proteins

Basic leucine zipper (b-ZIP) proteins contain a DNA-binding domain rich in basic residues and adjacent to a leucine zipper dimerization domain. Several b-ZIP factors that bind ABA-responsive elements (ABREs) have been cloned as candidates for ABA-responsive transcription factors that induce gene expression by dehydration and/or ABA. These proteins include the wheat EmBP1 (Guiltinan, Marcotte & Quatrano 1990), the tobacco TAF-1 (Oeda, Salinas & Chua 1991) and the rice OSBZ8 (Nakagawa, Ohmiya & Hattori 1996) and OsZIP-1a (Nantel & Quatrano 1996). TRAB1, an ABRE-binding b-ZIP factor, was isolated from rice that interacts with VP1, providing evidence that VP1 functions through interaction with a factor that binds to ABRE, thereby regulating ABA-induced transcription (Hobo, Kowyama & Hattori 1999). Choi et al. (2000) have also isolated a family of Arabidopsis ABA-responsive element-binding b-ZIP factors (ABFs), and reported that ABF2, ABF3 and ABF4 function in osmotic stress signalling. Independently, Uno et al. (2000) isolated two cDNAs encoding basic leucine zipper-type ABRE-binding proteins (AREB1 and AREB2); their transcription is up-regulated by drought, NaCl and ABA. Both proteins activated transcription of a reporter gene driven by ABRE and require ABA for their activity. The Arabidopsis ABI5 gene involved in the ABA-insensitive phenotype was also shown to encode an AREB-related b-ZIP transcription factor (Finkelstein & Lynch 2000).

Myb-like proteins

Myb proteins appear to be a conserved family of ubiquitous transcription factors. The Myb motif comprises three imperfect repeats, forming a helix–turn–helix-related motif. In each repeat, there are three conserved tryptophan residues to every 18–19 amino acids, which promote secondary structure, thus rendering a functional Myb-domain. Plant myb-related genes comprise a large family and are likely to participate in a variety of functions (Meissner et al. 1999). The dehydration-induced expression of Arabidopsis, Atmyb-2 has been identified as a positive regulator of rd22 gene expression (Abe et al. 1997). Myb-related genes, cpm7 and cpm10– whose expression is up-regulated by dehydration and ABA, respectively – were cloned from C. plantagineum (Iturriga et al. 1996). Since the expression of cpm7 is restricted to roots, which represent a primary sensor of water deficit in plants, it is most likely that the encoded product functions as a transcription factor interacting with genes responsive to dehydration.

Myc-like bHLH proteins

Myc-like proteins contain the basic helix–loop–helix (bHLH) domain, which is composed of two subdomains: the basic region (as found in b-ZIPs) responsible for DNA binding and the helix–loop–helix (HLH) region responsible for dimerization. The Arabidopsis rd22 BP1 gene, encoding a Myc homologue transcription factor, was shown to be induced by dehydration, high-salt conditions and ABA, which activates the downstream rd22 gene (Abe et al. 1997). The expression of rd22 BP1 precedes the rd22 gene expression, underlining the hierarchic relationship.

HD-ZIP proteins

Homeodomain-leucine zipper (HD-ZIP) genes encode proteins that have only been identified in plants so far and are thought to regulate development and responses to environmental cues. HD-ZIP proteins are characterized by the presence of a DNA-binding homeodomain with a closely linked leucine zipper motif. The activity of HD-ZIP resides primarily in the specific DNA-binding property of the homeodomain and the ability of the leucine zipper to mediate protein–protein interaction with other HD-ZIPs. From C. plantagineum, two HD-ZIP family II genes (CPHB-1 and CPHB-2) were isolated and shown to interact in a yeast two-hybrid system (Frank et al. 1998). Both transcripts were inducible by dehydration, but showed different responses to ABA. The expression of CPHB-1 was not inducible by ABA; in contrast, the transcript level of CPHB-2 increased during ABA treatment. This difference leads to the suggestion that CPHB-1 and CPHB-2 act in different branches of the dehydration-induced network. No target gene for HD-ZIP has yet been identified for certain. Potential HDE consensus DNA-binding sequences were identified in two drought-responsive promoters from C. plantagineum, suggesting that these genes may be regulated by HD-ZIPs. Three genes of this class are shown to be responsive to dehydration in Arabidopsis. Söderman, Mattsson & Engström (1996) have isolated ATHB-7 from Arabidopsis and shown that low transcript levels were present in all plant organs. However, the accumulation of transcripts was up-regulated by water stress and ABA treatment. This up-regulation  is mediated by ABA, since no induction was shown in the mutant (aba-3). Similarly, Athb-6 and Athb-12 are reported to be water-stress-responsive in an ABA-dependent signalling pathway (Söderman et al. 1999; Lee & Chun 1998).

AP2/EREBP domain

An inspection of dehydration- and cold-regulated genes in Arabidopsis led to the discovery of the C-repeat motif CCGAC, which is also part of the dehydration-responsive element (DRE) first described for the promoter of the dehydration- and low-temperature-regulated gene rd29A (Yamaguchi-Shinozaki & Shinozaki 1993). Liu et al. (1998) isolated two cDNA clones, DREB1A and DREB2A, as DRE-binding proteins that specifically interact with the DRE motif in the promoter region of the rd29A gene. The C-repeat was observed in several cold-regulated genes, and a cDNA clone (CBF-1) that encodes a C-repeat/DRE-binding factor was again isolated (Stockinger, Gilmour & Thomashow 1997). Analysis of the amino acid sequences indicates that the DREB and CBF proteins have an AP2/EREBP domain and other characteristics of transcriptional activators. This hypothesis was confirmed by ectopic expression of the genes. DREB1/CBFs function in cold-responsive gene expression, whereas DREB2 is involved in drought-responsive gene expression. Over-expression of the DREB1A cDNA under the control of the 35S CaMV promoter or the stress-inducible rd29A promoter in transgenic plants gave rise to expression of the stress-inducible DREB1A target genes and increased tolerance to freezing, drought and salt stress (Liu et al. 1998; Kasuga et al. 1999). These results indicate that at least two independent families of C-repeat/DRE-binding factors function as transcription factors in separate signal transduction pathways. In a recent study using cDNA microarray analysis, 12 stress-inducible genes were identified as potential targets of DREB1A, including six novel genes (Seki et al. 2001).

Signal perception

In animal systems, many of the extracellular signals are perceived by membrane-associated receptor kinases. These receptor kinases are composed of an extracellular ligand-binding domain, a transmembrane domain and a cytosolic kinase domain. Binding of an extracellular domain activates the cytoplasmic kinase domain, thus transducing an extracellular signal(s) into intracellular targets. The receptor-like protein kinase gene, RPK1, from Arabidopsis is induced rapidly by dehydration, high salt and low temperature, suggesting that RPK1 may function in the transmission of environmental stress signals (Hong et al. 1997).

At present, the primary site for sensing water stress is unknown, but it can be recognized by cellular osmosensors. These are characterized well in yeast and bacterial mutants with altered perception of osmotic stress. Osmosensors are two-component systems, which contain a histidine kinase as sensor and a response regulator that relays the phosphorylation signal and leads to transcriptional activation of downstream genes (Wurgler-Murphy & Saito 1997). Recently, the Arabidopsis AtHK1 (Urao et al. 1999) gene was isolated and shown to act as a homologue of the yeast osmosensor SLN1. The ability of AtHK1 to complement the sln1 yeast mutant defective in osmosensitivity indicates that AtHK1 may function as an osmosensor in plants; however, the exact role is as yet unknown.

Signal transduction

Knowledge of intracellular signalling pathways leading to adaptive changes is still scarce. Changes in protein phosphorylation were observed when plants were exposed to water deficit, suggesting reversible protein phosphorylation as a regulator (Conley, Sharp & Walker 1997). In mammalian and yeast cells, one of the main pathways that transduce signals from the cell membrane to the nucleus and leads to the expression of the appropriate genes is the mitogen-activated protein kinase (MAPK) cascade. Several reports of numerous protein kinases with close sequence similarities to MAPKs and other kinases belonging to the MAPK cascade have been identified in plants in response to dehydration/ABA, suggesting that the MAPK cascade is involved in stress signalling in plants (Mizoguchi et al. 1996; Mikolajczyk et al. 2000).

A role for calcium signalling in plants during environmental stress conditions has been demonstrated unequivocally (Knight, Trewavas & Knight 1997). Isolation of drought- and high-salt-induced expression of calcium-dependent protein kinases (CDPKs) in plants provides indirect evidence for a role of calcium signalling (Urao et al. 1994). Using different Ca2+ channel blockers, Knight et al. (1997) have demonstrated that the expression of drought-responsive genes was inhibited during mannitol treatment in Arabidopsis, implicating the importance of calcium signalling in gene expression. Furthermore, a gene (AtCBL1;Arabidopsis thaliana calcineurin B-like protein, a calcium-binding protein) was induced in response to drought, wounding and cold stress (Kudla et al. 1999). Calcinuerin B-like proteins are implicated in a variety of signalling pathways in animals (Tong, Shepherd & Jahr 1995) and in adaptation to salt stress in yeast and plants (Nakamura et al. 1993; Pardo et al. 1998; Liu & Zhu 1998).

Phospholipid signalling plays an important role in diverse early signalling cascades in animal cells. In plants, phospholipid signalling has been implicated in various developmental processes and environmental adaptations involving phospholipases A2, C and D. The isolation of dehydration-responsive Arabidopsis genes encoding a phosphotidylinositol-specific phospholipase C (AtPLC1; Hirayama et al. 1995) and a phosphotidylinositol-4-phospahte 5-kinase (PIP5K; Mikami et al. 1998) indicate a role for phospholipid signalling in the dehydration response. Further evidence was provided by Frank et al. (2000), who isolated two phospholipase D cDNAs from C. plantagineum: CpPLD1 is expressed constitutively and is likely to be involved in early responses to dehydration, producing the second messenger phosphotidic acid to amplify the signal after its perception, while the dehydration-responsive CpPLD2 may be involved in phospholipid metabolism and membrane rearrangements. Both PLDs possess a calcium-binding domain, suggesting that PLD activity might be regulated by calcium ions. However, the downstream targets of phosphatidic acid in plants have yet to be identified.

Down-regulation of genes

Although the primary interest is in the identification of the up-regulated genes during water stress, the down-regulation of gene expression also contributes to the adaptation of plants to stress. For instance, ProDH gene expression is down-regulated during water stress, resulting in the accumulation of proline (Yoshiba et al. 1997). In C. plantagineum, studies have revealed that transcripts encoding proteins relevant to photosynthesis are down-regulated and have been estimated to represent 36% of the total number of genes altered during the dehydration process (Bockel, Salamini & Bartels 1998). The exact function of several down-regulated genes is largely unknown. The logical hypothesis is that some of the genes are down-regulated because of the fact that their product might not be suited to the new physiological condition caused by dehydration stress (Chandler & Robertson 1994).

Improving tolerance to dehydration stress by genetic engineering

Apparently, plants employ multiple mechanisms to ensure dehydration tolerance. At present, our knowledge of the metabolic changes that contribute to dehydration tolerance is incomplete, but information about the biochemical processes contributing to dehydration tolerance is essential for successful engineering of dehydration tolerance in crop plants. The genetic approaches outlined below offer new ways to evaluate the contribution of individual genes to dehydration tolerance. The past decade has seen the prospect of using genetically modified plants and testing their potential to modulate tolerance through the alteration both of osmolyte levels and enzymes that scavenge reactive oxygen species or transcription factors, and of components involved in signal transduction.

Overproduction of proline in transgenic tobacco enhanced root biomass and flower development and helped the cells to maintain water potential, thus enhancing water tolerance (Kavi Kishor et al. 1995). The explanation for this observation is not yet understood. The authors originally suggested higher osmotic potentials in the leaf sap, but this was questioned by other scientists (see Letters to the Editor, Plant Physiology110, 1051–1053). When the same gene was introduced in rice, the transgenic plants showed an increase in biomass under stress conditions (Zhu et al. 1998). These results strengthen the notion that the over-accumulation of proline by expressing P-5-CS is likely to enhance the performance of plants during water stress and salinity.

Fructans are polyfructose molecules that are produced by many plants and bacteria and they may play a role in the adaptation to osmotic stress because of their high-soluble nature. The Sac B gene from Bacillus subtilis, encoding levan sucrase, was introduced into tobacco under the transcriptional control of the 35S promoter. The transformed plants accumulated fructan and drought treatment resulted in 33% more fresh weight than in the control plants (Pilon-Smits et al. 1995). Transgenic sugar beet plants expressing Sac B accumulated fructans and performed better compared with control plants during dehydration stress (Pilon-Smits et al. 1999).

Trehalose, a non-reducing disaccharide of glucose, is known as a reserve metabolite in yeast and fungi. Biochemical studies have also shown that trehalose stabilizes proteins and membrane lipids. Tobacco plants transformed with the gene encoding the trehalose synthase subunit (TPS1) of yeast trehalose synthase accumulated trehalose and exhibited a drought-tolerant phenotype when the detached leaves were air-dried (Holmstrom et al. 1996). When tobacco was transformed with bacterial trehalose-6-phosphate synthase and trehalose-6-phosphate phosphatase, the leaves of the plants had a better photosynthetic efficiency and a higher dry weight accumulation under drought stress than controls (Pilon-Smits et al. 1998). Transgenic tobacco expressing the IMT1 gene from Mesembryanthemum crystallinum accumulated the methylated sugar alcohol, d-ononitol, and showed tolerance to drought and salt stress (Sheveleva et al. 1997). Therefore, the transfer of osmolyte-synthesizing genes into plants confirmed their role in stress adaptation despite the fact that the effect of individual genes is often rather small.

Xu et al. (1996) and Sivamani et al. (2000) produced rice and wheat plants transgenic for the barley LEA gene HVA1. Transgenic plants accumulated HVA1 protein in both leaves and roots and showed tolerance to drought and salinity.

Recently, our understanding of the role of ROS-scavenging systems in plant stress tolerance has increased through the use of transgenic approaches manipulating the level of antioxidant enzymes. A number of such experiments demonstrated that the enhancement of the ROS-scavenging systems can provide partial protection from oxidative damage, indicating that this strategy could be used to improve tolerance. Transgenic alfalfa plants harbouring MnSOD from Nicotiana plumbaginifolia performed better during drought stress under field conditions (McKersie et al. 1996). Ectopic expression of the alfalfa gene MsALR in transgenic tobacco plants provided tolerance to multiple stresses, including drought stress with decreased amounts of lipid peroxidation-derived reactive aldehydes (Oberschall et al. 2000). Therefore, it is conceivable that decreasing oxidative stress offers another avenue for providing protection against environmental stresses.

Improving stress tolerance by manipulating regulatory genes proved to be a promising tool, as demonstrated by Jaglo-Ottosen et al. (1998) and Kasuga et al. (1999). Over-expression of CBF1 increased cold-regulated (COR) gene expression and thereby promoted tolerance to freezing in Arabidopsis. Similarly, over-expression of two transcription factors belonging to the same gene family (DREB1 and DREB2) improved the tolerance to dehydration (Liu et al. 1998). DREB1A over-expression under the transcriptional control of the 35S promoter has detrimental effects on plant growth under normal conditions, although these plants performed better under stress conditions than non-transformed plants. Over-expression of DREB1A driven by a stress-inducible promoter caused minimal negative effects under normal growth conditions, and enhanced stress tolerance was observed. The over-expression of this gene activated the expression of downstream genes such as LEA genes and P-5-CS, which are involved in stress tolerance, indicating that the over-expression of a transcription factor seems to be a powerful approach for manipulating stress tolerance. Furthermore, employing abiotic-stress-inducible promoters to drive the expression of transgenes has an additional advantage over the constitutive promoter: it induces the transgene expression when desired.

The manipulation of components involved in signalling cascades seems to be another possible strategy for improving tolerance to multiple stresses (Saijo et al. 2000). A stress-induced rice gene encoding a calcium-dependent protein kinase (OsCDPK7) was expressed ectopically in rice plants, resulting in enhanced levels of stress-responsive genes such as rab16A (group 2 LEA protein), salT (glycine rich protein) and wsi18 (group 3 LEA protein) in response to salt and drought but not to cold (Saijo et al. 2000). Therefore, it was suggested that mechanisms of cold tolerance and salt/drought tolerance are different from each other, sharing OsCDPK7 as a common component.

Future prospects

A broad spectrum of genes are expressed on exposure to dehydration. However, the significance of these genes cannot be understood without knowledge of their function. Our understanding of the function of these genes is far from complete. Given the advancements in gene isolation, the research priorities are shifting from cloning to characterization of the gene products and identification of their roles in planta. The major challenge ahead is to assess the relative contribution of each gene to dehydration tolerance, and to eliminate those that do not measurably affect the level of stress tolerance. This can be achieved using gain of function and loss of function mutants. However, the genetic approach is not very realistic in model plants that express tolerance to desiccation. Therefore, Arabidopsis will serve as a reference for functional gene analysis and for gene expression studies. Gene functions will be addressed in Arabidopsis mutants and the hypothesis derived from this study will then be tested in the model plant and in transgenic plants. The availability of the Arabidopsis genome sequence will allow us to identify genes that may be unique to a plant with an extreme phenotype.

The past decade has witnessed the utilization of transgenic approaches for experimental purposes, mainly in experimental model plants but not in agriculturally important plants (particularly with reference to the improvement of drought tolerance). Most of the transgenic approaches used single genes and often constitutively active promoters, which lead only to marginal stress improvement in most cases, if any at all. The limited success with single gene transfers is not surprising, given that multiple genes act in concert during dehydration tolerance. Therefore, a major challenge in plant biotechnology will come with the need to introduce sets of genes in order to express quantitative traits determined by multiple genes. Current knowledge has broadened the possibilities for genetically engineering plants with multiple genes. Recently, the transfer of multiple genes has been demonstrated successfully by Potrykus and coworkers, who introduced three enzymes that are required for the β-carotene biosythetic pathway in rice plants (Ye et al. 2000). Alternatively, sequential transformation or sexual crosses between transgenic plants harbouring transgenes of interest are other means of introducing multiple genes, although this is a time-consuming process.

The routine application of micro arrays, specialized yeast two-hybrid interaction screens and transcript expression profiling using marker technologies such as amplified fragment length polymorphosis (AFLP) will allow us to identify pathways and sets of genes involved in a particular plant response to dehydration. This will give us a much better understanding of the interaction and molecular relationship of genes within the network of stress genes. This may also provide us with a novel guide towards choosing genes for biotechnological approaches to manipulate stress tolerance. The current phase of research takes mainly transcriptional changes into account, and post-transcriptional modifications have been largely neglected. Proteomic approaches will be necessary in the near future to reveal the plasticity of gene expression by protein modification.


S.R. gratefully acknowledges the postdoctoral fellowship from the Alexander von Humboldt Foundation. The work in the lab. of D.B. is supported by the DFG Schwerpunkt-programm Molekulare Analyse der Phytohormonwiskung.