The SnRK1 complex, a metabolic regulator of nutrient deficit in plants, consists of a catalytic α-subunit and two regulatory subunits, β and γ that exist as several isoforms. To obtain insight into the developmental and stress conditions that regulate the expression of these regulatory subunits, four β subunits MtAKINβ1-β4, and three γ subunits MtAKINβγ, MtSNF4b and MtAKINγ were identified and characterized in seeds of M. truncatula. Their transcripts were found to accumulate differentially in vegetative and seed tissues and appear to be differentially modulated during germination and the imposition of stress. MtAKINγ and MtAKINβ3 showed identical patterns of expression upon osmotic shock, whereas transcripts of MtAKINγ and MtAKINβ1 were strongly up-regulated upon starvation of the radicles. Addition of glucose during the starvation process reversed this effect. MtAKINβ2 and MtSNF4b were specifically induced upon the re-induction of desiccation tolerance in germinated, desiccation-sensitive radicles, whereas only MtSNF4b expression was repressed by an inhibitor of abscisic acid synthesis. A second γ subunit, MtAKINβγ, was transiently expressed early during the induction of desiccation tolerance, and its expression could be modulated by blocking the respiratory ATP production by cyanide. The transcriptional regulation of the subunit isoforms showing identical expression profiles in germinating seeds appears to be under the control of different factors.
The Snf1/AMP-activated protein kinase (AMPK) family is essential for metabolic regulation in eukaryotes (for review, see Hardie, Carling & Carlson 1998; Kemp et al. 1999). In mammals, AMPK controls metabolic enzymes in response to stresses that affect cellular energy supply, including nutrient limitation, hypoxia, heat shock and exercise. In yeast, the Snf1 kinase plays an essential role in metabolic adaptation to different carbon sources and in response to environmental stress such as salt stress and heat shock (Honigberg & Lee 1998; Ashrafi, Farazi & Gordon 1998; Ashrafi et al. 2000). Like its yeast counterparts, several lines of evidence suggest that the SNF1-homologue in plants, SnRK1, has the potential to regulate carbon metabolism through both gene expression and direct control of enzyme activity. A SnRK1 cDNA expressed in rye endosperm was found to complement a yeast snf1 mutant by restoring the glucose repression function (Alderson et al. 1991). Antisense expression of a SnRK1 sequence in potato resulted in the loss of sucrose-inducibility of sucrose synthase gene expression in leaves and in the reduction of sucrose synthase gene expression in tubers (Purcell, Smith & Halford 1998). In addition, spinach SnRK1 was found to inactivate in vitro sucrose phosphate synthase (Sugden et al. 1999). Besides its involvement in nutrient stress adaptation, the yeast Snf1 is also involved in developmental processes such as sporulation (Honigberg & Lee 1998), life span and ageing (Ashrafi et al. 2000). In plants, the expression of a barley SnRK1 sequence in the anti-sense orientation resulted in the arrest of pollen development at the binucleate stage, thereby producing male sterile plants (Zhang et al. 2001). This developmental arrest is thought to originate from the inability of the pollen to respond to their carbon status.
In developing seeds, sucrose serves not only as a substrate for energy production and accumulation of starch and/or lipid reserves but also functions as a protective molecule in the dry state (for review, see Hoekstra, Golovina & Buitink 2001). During the acquisition of desiccation tolerance, sucrose levels increase dramatically prior to maturation drying (Blackman, Obendorf & Leopold 1992; Black et al. 1999). Non-reducing sugars, together with stress proteins such as heat shock proteins and late embryogenesis abundant (LEA) proteins are involved in stabilizing macromolecular structures in the dry state, thereby protecting macromolecules and retaining the functional integrity of membranes upon desiccation and rehydration (Hoekstra et al. 2001; Buitink, Hoekstra & Leprince 2002). Furthermore, it has been proposed that a concerted down-regulation of energy metabolism during desiccation is required to ensure survival in the dried state (Leprince & Hoekstra 1998; Leprince et al. 2000). As such, developing seeds must have evolved mechanisms to regulate their metabolism accordingly. Although the mechanisms of protection in the dry state are well characterized, many of the signals and signalling pathways that lead to the activation and repression of genes involved in desiccation tolerance of seeds remain to be elucidated. Abscisic acid (ABA)-related signalling pathways play an important role in the acquisition of desiccation tolerance by inducing the expression of genes associated with dehydration such as LEA proteins (Ingram & Bartels 1996). In contrast, the signalling cascades leading to the accumulation of sucrose or repression of metabolism associated with desiccation tolerance in seeds are largely unknown. The possible involvement of the SnRK1 in regulating carbon metabolism prompted us to investigate whether the complex plays a role in desiccation tolerance.
As a first step to understand the role of the SnRK1 complex in Medicago truncatula seeds, we identified and characterized four isoforms of the β subunit and the homologues of the Pv42/LeSNF4, AKINβγ and AKINγ subunits in M. truncatula. Their expression profiles were then studied in seeds in relation to germination, loss of desiccation tolerance, osmotic stress and starvation. To ascertain the transcriptional specificity of subunit isoforms to desiccation tolerance, we developed a physiological model system in which desiccation tolerance is re-induced in germinated, desiccation-sensitive radicles by submitting them to an osmotic shock for several days (Bruggink & van der Toorn 1995; Leprince et al. 2000).
MATERIALS AND METHODS
Plant material and treatments
Seeds of Medicago truncatula Gaertn. (cv Paraggio, Seedco, Australia) were imbibed on filter paper (Whatman, no. 1) in distilled water at 20 °C in the dark. At different time intervals, percentage of germination was determined by counting the number of seeds that showed visible radicle protrusion. Data represent the average of three independent replicates of 100 seeds each. To determine the percentage of desiccation-tolerant seeds during imbibition, 50 seeds were removed at different intervals of imbibition and dried at room temperature under a flow of air at 42% relative humidity achieved by a saturated K2CO3 solution. The final water content of the radicles after drying was determined gravimetrically by determination of the fresh weight and subsequent dry weight after 2 d at 96 °C in an oven, and was 0.09 g H2O g−1 dry weight (DW). After 3 d of drying, seeds were re-imbibed on filter paper and considered desiccation tolerant when the radicle protruded from the seed coat or resumed growth.
To re-induce desiccation tolerance in germinated, desiccation-sensitive radicles of seeds of M. truncatula, 20-h-imbibed, germinated seeds with a protruded radicle length between 2.6 and 2.9 mm (referred to as 3 mm in the text) were selected and submitted to an osmotic treatment by incubating them in a polyethylene glycol (PEG) 8000 solution with a water potential of −1.7 MPa at 10 °C in the dark. Furthermore, germinated seeds with a radicle length of 5 mm were selected and submitted to the same osmotic potential. The osmotic potential was calculated according to Michel & Kaufmann (1973). To avoid hypoxic conditions, the volume of PEG solution was adjusted so that seeds were not completely submerged. At different time intervals, seeds were taken from the PEG solution and rinsed thoroughly and dried for 2 d at room temperature under a flow of air at 42% relative humidity. Seeds were then re-imbibed on wet filter paper at 20 °C in the dark. Seeds were considered desiccation tolerant when the radicles resumed growth after drying and rehydration. Each time point represents 150–300 seeds originating from three to five independent experiments, and the average and standard deviation were determined.
For the starvation experiment, 100–150 germinated radicles with a length of 2.6–2.9 mm were excised and imbibed for 24 h on a filter paper with either sterilized distilled water or in the presence of 60 mm glucose (Sigma, St Louis, MO, USA). For inhibition of synthesis of proteins or ABA, 100–150 seeds with radicles between 2.6 and 2.9 mm were incubated in PEG solution containing either 100 µm cycloheximide (Sigma) or 200 µm 1-methyl-3-phenyl-5-(3-trifluoromethyl-(phenyl))-4-(1H)-pyridinone (fluridone; Duchefa, Haarlem, The Netherlands) with an osmotic potential of −1.7 MPa for 24 h at 10 °C in the dark. Fluridone was prepared as a stock solution of 40 mm in 1 mL 95% EtOH and 0.5 mL Tween 20 (Sigma). For the treatment with KCN, 100–150 seeds with 2.6–2.9 mm germinated radicles were incubated for 24 h in a PEG solution, then 2 h in PEG solution containing 0.5 mm KCN. After the treatments, seeds were rinsed briefly after which the radicles were excised, frozen in liquid nitrogen and stored at −80 °C until the extraction of RNA.
Database homology search for MtAKINβ, MtAKINγ and MtAKINβγ cDNAs
The TIGR Medicago truncatula Gene Index (MtGI) database (http://www.tigr.org) was searched for homologues of the β and γ subunit protein sequences. Both protein and nucleotide sequences were compared to the NCBI database using the blast network service. Alignments were performed using multalin (Corpet 1988). A phylogenetic tree was constructed with Treeview (Page 1996), from a multiple sequence alignment generated using CLUSTALW (http://www2.ebi.ac.uk/ClustalW). Predicted molecular weights were calculated using ExPASy software (http://tw.expasy.org/tools/pi_tool.html).
RNA isolation and northern blot analysis
Total M. truncatula RNA was isolated according to Verwoerd, Dekker & Hoekema (1989) from leaves and roots from 6-week-old-plants, 30 cotyledons of imbibed seeds, 60–150 radicles from seeds at different intervals of imbibition, and from 60 to 150 radicles from germinated seeds with a size of 2.6–2.9 mm and 5 mm at different time intervals of PEG treatment (−1.7 MPa, 10 °C). After excision, all material was directly frozen into liquid nitrogen to prevent a wounding response.
Total RNA aliquots (15 µg) were denatured at 50 °C in a glyoxal mixture according to Sambrook & Russel (2001) for 1 h, electrophoresed on a 1.2% denaturing agarose gel in BPTE buffer (Sambrook & Russel 2001) and transferred to Hybond-N membrane (Amersham, Piscataway, NJ, USA). Pre-hybridization and hybridization were carried out at 65 °C in 0.5 m phosphate buffer, pH 7.2, 7% (w/v) sodium dodecyl sulphate (SDS) and 1 mm ethylenediaminetetraacetic acid. Filters were probed with 32P randomly labelled cDNA. The cDNA probes of the different isoforms of the β, βγ and γ subunits were polymerase chain reaction (PCR)-generated, and gene-specific amplification was verified by sequencing (MWG-Biotech, Ebersberg, Germany). Forward and reverse primers used and length of PCR fragment obtained were as follows: MtAKINβ1, 5′-gcagaatgagcctcatggta-3′, 5′-ctcccatggcaacaacagat-3′, 542 bp; MtAKINβ2, 5′-ttccttccgcaaatatgga-3′, 5′-gatcgcgaaacatgtgacaa-3′, 435 bp; MtAKINβ3, 5′-cacaactagtggttgacttg-3′, 5′-ctggtctttcgagaaggttt-3′, 415 bp; MtAKINβ4, 5′-caggttcctgtgattgtcat-3′, 5′-cacatgtgtagaactaggga-3′, 528 bp, MtAKINγ, 5′-cctgaatctgggaaggtcat-3′, 5′-ccagggatatccgagcatat-3′, 727 bp; MtAKINβγ, 5′-ggaggtcaggaagtccttca-3′, 5′-ggagtccatccactccaact-3′, 632 bp; MtSNF4b, 5′-cagcactgcgagagtatgaa-3′, 5′-agtccacaagctaaggcctt-3′, 390 bp. An 18S rRNA probe (427 bp) was obtained from M. truncatula by reverse transcriptase (RT)-PCR using 5′-ccaggtccagacatagtaag-3′ and 5′-gtacaaagggcagggacgta-3′ as forward and reverse primers, respectively, to check for equal loading. Membranes were washed at 65 °C for 1 h in 2× SSC, 0.1% SDS (w/v) and than at room temperature for 10 min in 1× SSC, 0.1% SDS (w/v) (Sambrook & Russel 2001). Membranes were exposed either to film (Hyperfilm MP, Amersham Pharmacia Biotech) at −80 °C or to a phosphor screen (Eastman Kodak Company, New York, NY, USA). Phosphor screens were analysed using a phosphoimager (Molecular Imager FX Pro; Bio-Rad, Hercules, CA, USA) and quantification of the relative level of the mRNAs was performed using Quantity One software (Bio-Rad).
mRNA purification, RACE and cDNA cloning
To obtain and confirm sequences of the different subunit isoforms, standard molecular biology techniques were used (Sambrook & Russel 2001). For those isoforms for which TC/ESTs were found to encode the full length cDNA (β1, β3, and γ), primers were designed at the 5′- and 3′- terminal regions. RT-PCR was performed using mRNA from radicles from non-germinated seeds, and the PCR fragments were sequenced (MWG-Biotech AG). Full length cDNAs of MtAKINβ2, MtAKINβγ and MtSNF4b were obtained by rapid amplification of cDNA ends (RACE). Total RNA aliquots (400 µg) from non-germinated seeds of M. truncatula imbibed for 16 h were used to purify mRNA using the PolyATract mRNA isolation system III kit, following the manufacturers protocol (Promega, Madison, WI, USA). Subsequently, full-length 5′- 3′- adapted-ends cDNAs were obtained from purified mRNA using the GeneRacer Kit for RNA ligase-mediated and oligo-capping RACE (Invitrogen, Groningen, The Netherlands) according to the manufacturer's instructions. Specific primers were designed in the 5′ region of each EST encoding the three isoforms and PCR was performed to obtain the 3′- end. Amplified cDNAs were cloned using the TOPO TA cloning kit for sequencing (Invitrogen) according to the manufacturer's instructions. Recombinants were screened using internal primers and positive clones were confirmed by DNA sequencing.
Identification of β subunits of the SnRK1 complex in Medicago truncatula
To identify β subunit isoforms that are expressed in M. truncatula, the TIGR Medicago truncatula Gene Index was screened with the A. thaliana AKINβ1 protein sequence (AJ132315; Bouly et al. 1999). Four different β subunits were identified in the EST database of M. truncatula (Fig. 1). One tentative consensus (TC36504) encompassing 7 ESTs encoded for the full length MtAKINβ1 cDNA (AY247271), predicting a protein of 276 amino acids with a molecular weight of 30 kDa that showed 65% identity and 90% similarity with AKINβ1 of A. thaliana. One EST (AL388804) of 553 bp encoded for a sequence that presented similarity with the C-terminal domain of AKINβ2 (AJ132316; Bouly et al. 1999). In addition, a second EST clone (BF649596) of 622 bp was found, the sequence of which showed high similarity with the N-terminal domain of the AKINβ2. Primers were designed in the uncoding regions of both cDNAs and a cDNA sequence of 1286 bp was obtained, predicting a protein of 289 residues with a molecular weight of 32 kDa (AY247272). This MtAKINβ2 showed 69% identity and 91% similarity with AKINβ2 from A. thaliana. A third isoform (MtAKINβ3, AY247273) was identified from several ESTs, and one of them (BQ141266) corresponded to a full length cDNA analogous to AKINβ3 from A. thaliana (AF491295) encoding a protein of 13 kDa with 117 residues containing only a short part of the conserved KIS (Kinase Interaction Sequence) domain. The MtAKINβ3 protein has 61% identity and 88% similarity with AKINβ3. A fourth group of ESTs was identified encoding a β subunit that could not be found in the A. thaliana database, and will be referred to as MtAKINβ4 (AY247274). The sequence of two ESTs (BI265191 and BG583135) revealed a partial cDNA of 973 bp with the 5′ upstream stop codon missing. Figure 1 shows the alignment of the different MtAKINβ subunits with other plant β subunits. The conserved KIS and ASC (Association with SNF1 Complex) domains, which in yeast bind to the catalytic subunit SNF1 and SNF4, respectively, are underlined in Fig. 1 (Jiang & Carlson 1997).
Identification of γ and βγ subunits in Medicago truncatula
One class of γ subunits is represented by Pv42 (U40713, Lakatos et al. 1999) from Phaseolus vulgaris and LeSNF4 (AF419320) from tomato (Bradford et al. 2000) (Fig. 2a). The screen of the M. truncatula TIGR EST database using the LeSNF4 protein revealed one EST (BG450380) corresponding to a partial cDNA encoding a protein with the C-terminal end missing. The full length cDNA that was obtained using-3′ RACE predicted a protein of 381 aa. This protein has been named MtSNF4b (AY247270) in order to avoid confusion with AtSNF4 (Kleinow et al. 2000). The alignment of MtSNF4b with Pv42 and LeSNF4, as well as the A. thaliana homologue AtSNF4b that is present in the Arabidopsis NCBI database (AC007591) shows that the amino acid sequence is remarkably conserved between species (Fig. 2a).
Lumbreras et al. (2001) reported a new class of SnRK1 γ subunits for which the structural organization of domains is hybrid between the β and γ subunit. Designated as βγ, this second class is similar to the γ subunit but contains an additional N-terminal region corresponding to part of the KIS domain found in SIP1/SIP2/GAL83 of yeast and animal β subunits (AMPKβ). Using the amino acid sequence of the ZmAKINβγ1 of maize (AF276085) as screening bait, we identified one M. truncatula EST clone (BF639214) encoding the N-terminal part of the KIS domain showing a high homology with the maize counterpart. A 3′-RACE strategy using primers designed in the 5′ uncoding region was used to clone and sequence the full length cDNA (AY247268). It predicts a protein of 485 aa with a theoretical molecular weight of 53.6 kDa. The sequence of AKINβγ appears to be highly conserved amongst different plant species (Fig. 2b). MtAKINβγ shows 64 and 71% identity with ZmAKINβγ and AKINβγ from maize and A. thaliana (AF439826), respectively.
A third class of a putative γ subunit was found by screening of the M. truncatula TIGR EST database with the AKINγ protein (AJ132317, Bouly et al. 1999) from A. thaliana. One tentative consensus (TC51759) consisting of 23 ESTs was identified containing four CBS (cystathionine-β-synthase) domains that are characteristic of the γ subunit (Hardie et al. 1998). The cDNA was 1198 bp long, predicting a protein of 420 aa (AY247269). This MtAKINγ has 67% identity and 89% similarity with AKINγ. A multiple sequence alignment was performed on the SNF4/AMPKγ-type subunits of plants, yeast and mammals together with the γ domain of the plant βγ subunit (Fig. 3). In plants, three clusters of γ subunits could be identified (Fig. 3). One cluster is represented by AKINβγ/MtAKINβγ, a second cluster by LeSNF4/Pv42/MtSNF4b and a third one by AKINγ and MtAKINγ. The plant γ subunits that were found to functionally complement a yeast SNF4 deletion mutant are indicated by an asterisk (Kleinow et al. 2000; Lumbreras et al. 2001; Bradford et al. 2000, 2003).
Tissue-specific expression of the isoforms of the SnRK1 subunits in Medicago truncatula
Expression of MtAKINβγ, MtSNF4b, MtAKINγ and MtAKINβ genes was studied in leaves and roots of 6 week-old plants and in radicles and cotyledons of imbibed non-germinated seeds by northern-blot analysis using subunit-specific DNA probes (Fig. 4). MtAKINβγ mRNA was mostly found in the radicles of the embryo and hardly detectable in vegetative tissues (Fig. 4a). In contrast, MtAKINγ transcripts were mostly abundant in leaf tissues, although a prolonged exposure revealed the presence of mRNA in seed tissues (data not shown). Transcripts of MtSNF4b were mainly detected in seed tissues (Fig. 4a).
Transcripts of MtAKINβ1 and MtAKINβ2 were present in all tissues analysed. Steady-state levels of MtAKINβ1 were the highest in leaves whereas the transcript abundance of MtAKINβ2 was the lowest in root tissues (Fig. 4b). Transcripts of MtAKINβ3 were detected in all tissues (Fig. 4b) whereas the transcripts of the fourth β subunit, MtAKINβ4, were barely detectable by northern analysis and mostly found in leaves.
Isoforms of β and γ genes are differentially expressed during germination and loss of desiccation tolerance
To determine the loss of desiccation tolerance in radicles of M. truncatula during imbibition and subsequent germination, seeds were removed from the imbibition medium at different time intervals, dried rapidly to 0.09 g H2O g−1 DW, then placed back on wet filter paper. Upon rehydration, those seeds that germinated or those for which the radicle resumed growth were considered desiccation tolerant. Desiccation tolerance of the seed lot was maintained during the first 16 h of imbibition at 20 °C (Fig. 5a). Thereafter, the percentage of desiccation-tolerant seeds decreased concomitantly with the increase in the percentage of germinated seeds (Fig. 5a). After approximately 30 h of imbibition, the whole seed population had germinated and all radicles were desiccation-sensitive. The steady-state levels of the subunit transcripts were monitored in the radicles during seed imbibition and subsequent germination (Fig. 5b). At 18 h of imbibition, those seeds for which the radicle had just protruded were selected for analysis. Germinated seeds imbibed for 24 and 30 h exhibited an average radicle length of 3 and 5 mm, respectively. Transcript levels of MtAKINβ1 appear to be constitutively regulated during germination, when compared with 18S (Fig. 5b). In contrast, the mRNA levels of the other two β subunits, MtAKINβ2 and MtAKINβ3, as well as those of the γ subunits MtAKINβγ and MtSNF4b increased initially upon imbibition, then subsequently decreased upon germination (Fig. 5b). The expression profile of MtAKINγ was yet completely different from that of the other subunits. The mRNA levels transiently increased and peaked at radicle emergence (18 h of imbibition). Transcript levels of MtAKINβ4 could not be detected by northern analysis (data not shown). As we were interested in those subunits, the expression of which are present in seed tissues, MtAKINβ4 was omitted from further studies.
Differential expression of the SnRK1 subunits in relation to desiccation tolerance and osmotic stress
It has been reported that germinated, desiccation-sensitive radicles of several horticultural crops can be rendered desiccation tolerant by exposing them to an osmotic stress brought about by slow drying or an osmoticum such as PEG (Bruggink & van der Toorn 1995; Leprince et al. 2000). Here, we transferred this physiological system to seeds of M. truncatula. Germinated seeds with a radicle length of 3 mm are desiccation sensitive as they do not survive drying to 0.09 g H2O g−1 DW (Figs 5a & 6). However, by placing these seeds at 10 °C in a PEG solution with a water potential of −1.7 MPa, desiccation tolerance can be re-induced progressively (Fig. 6, open symbols). After 24 h incubation in the PEG solution, approximately 76% of the seeds had regained their tolerance to drying, whereas 90% of seeds with desiccation-tolerant radicles were obtained after 48 h. The re-induced desiccation-tolerant seeds developed into normal plants (data not shown).
Considering that the re-induction of desiccation tolerance is achieved by submitting the seeds to a low water potential, changes in gene expression could be related not only to the induction of desiccation tolerance, but also to the osmotic stress response. To be able to discriminate between these two physiological responses, gene expression was studied in germinated seeds with longer radicles (5 mm) that were submitted to the same PEG treatment. These radicles remained desiccation sensitive throughout the PEG incubation (Fig. 6, closed symbols). Increasing the incubation time or manipulating the osmotic potential of the PEG solution did not result in the induction of desiccation tolerance in the 5 mm germinated radicles (data not shown). However, 5 mm germinated radicles were found to be tolerant to the osmotic stress as normal seedlings were produced after transferring the seeds from the PEG solution to water.
To investigate the expression pattern of the SnRK1 complex in relation to desiccation tolerance, steady-state levels of the transcripts of the different β, βγ and γ subunit isoforms were studied during the induction of desiccation tolerance in 3 mm germinated radicles of M. truncatula seeds, brought about by the PEG treatment (Fig. 7). Gene expression was also investigated during a similar osmotic treatment in 5 mm germinated radicles that remained desiccation-sensitive. The MtAKINβγ transcript level in 3 mm germinated radicles increased rapidly upon transfer of the seeds to the PEG solution and showed a six-fold increase at 6 h of incubation, prior to the onset of desiccation tolerance (Fig. 7a & b). Between 6 and 24 h of incubation, transcript levels decreased and remained constant at a level slightly above that in untreated radicles. In contrast, the PEG incubation did not induce a significant increase in MtAKINβγ transcript levels in 5 mm germinated radicles that remained desiccation-sensitive (Fig. 7a & b). The difference in expression patterns of MtAKINβγ in the 3 and 5 mm PEG-treated radicles was confirmed by two additional independent northern blots originating from tissues collected from separate physiological experiments.
Upon the induction of desiccation tolerance in the 3 mm germinated radicles, transcripts of one β and one γ subunit isoform were found to increase in a similar manner. Both MtAKINβ2 and MtSNF4b increased around 24 h of PEG incubation, concomitantly with the re-induction of desiccation tolerance (compare Fig. 7c with Fig. 6). In contrast, the transcript levels of both subunits remained unchanged in germinated radicles of 5 mm length that were submitted to the same osmotic potential. Two other β and γ subunit isoforms (MtAKINβ3 and MtAKINγ) were identified based on the similarity of their expression profile in response to the osmotic treatment (Fig. 7e & f). In 3 mm germinated radicles, both transcripts increased transiently after 1 h of PEG incubation. A second peak in mRNA abundance appeared at around 24 h. A similar pattern was found in 5 mm germinated radicles, with the exception that the amplitude of the second increase was two-fold higher than in 3 mm radicles (Fig. 7e & f). The expression level of MtAKINβ1 decreased slightly during the PEG incubation, whereas the transcripts of MtAKINβ4 were not detected (data not shown).
MtAKINγ and MtAKINβ1 are specifically up-regulated upon starvation
There is accumulating evidence showing that SNF1 in yeast and SnRK1 in plants are associated with the regulation of sugar utilization in the response to nutritional stresses (Halford & Hardie 1998; Hardie et al. 1998; Zhang et al. 2001). These data prompted us to examine whether specific isoforms of the β and γ subunits in germinating radicles responded to a depletion of metabolites. Upon starvation, brought about by excising the radicles from the cotyledons of germinated seeds and incubating them 24 h on a filter paper with water, the expression of both MtAKINβ1 and MtAKINγ was strongly induced (Fig. 8). In contrast, the level of transcripts in excised radicles incubated for 24 h in the presence of 60 mm glucose did not increase (Fig. 8). Interestingly, MtAKINβ3, the subunit isoform exhibiting a similar expression pattern as MtAKINγ during the osmotic shock treatment, did not respond to the starvation treatment. Transcript levels of MtAKINβ2 and MtSNF4b remained constant upon the starvation and/or glucose treatment, whereas MtAKINβγ transcripts decreased in excised radicles in the presence and absence of glucose (Fig. 8).
Differential transcriptional regulation of SnRK1 subunit isoforms
To gain further insight into the regulation of the different subunit isoform genes during the re-induction of desiccation tolerance, we examined whether protein synthesis, ATP homeostasis and inhibition of ABA synthesis affected the transcript levels in germinated radicles during the PEG treatment. The effect of protein synthesis inhibition was studied by incubating seeds with 3 mm germinated radicles in the PEG solution in the absence and presence of 100 µm cycloheximide (Fig. 9, CHX). Without cycloheximide, MtAKINγ and MtSNF4b transcripts increased considerably upon 24 h of PEG incubation, whereas the presence of cycloheximide inhibited the mRNA accumulation. In contrast, cycloheximide strongly increased the transcript levels of βγ, β1 and β2 subunits in PEG-treated radicles. The transcript level of MtAKINβ3 decreased slightly upon the inhibition of protein synthesis, but so did 18S (Fig. 9).
To investigate whether the expression of the isoforms that are up-regulated in re-induced desiccation-tolerant tissues is linked to ABA synthesis, seeds with 3 mm germinated radicles were imbibed for 24 h in a PEG solution in the presence of 200 µm fluridone, an inhibitor of carotenoid and ABA synthesis. This high concentration was necessary to inhibit re-induction of desiccation tolerance (Buitink et al. 2003). MtAKINβ2 and MtSNF4b (i.e. the subunits that were uniquely expressed in desiccation-tolerant tissues, Fig. 7) were found to be differentially influenced by an inhibition of the ABA synthesis during PEG incubation (Fig. 9, Flu). Up-regulation of the MtSNF4b gene during PEG incubation was inhibited by the presence of fluridone. In contrast, MtAKINβ2 expression did not respond to the inhibition of ABA synthesis (Fig. 9). MtAKINβγ trancripts that accumulated early during PEG incubation were found to increase in the PEG-treated radicles in the presence of fluridone, suggesting that it is negatively regulated by ABA or by other factors downstream of the ABA response.
In animals, AMPK is considered a fuel gauge that is activated by changes in ATP/AMP ratio and acts by switching on/off ATP consuming and producing pathways. It has been hypothesized that SnRK1, the plant homologue, plays a similar role (Hardie et al. 1998). Therefore, we tested whether the subunit isoform gene expression in 3 mm radicles responded to changes in the ATP homeostatis during the PEG incubation by blocking the respiratory ATP production. Germinated seeds were incubated for 24 h in the PEG solution and subsequently submitted for 2 h to 0.5 mm KCN to rapidly bring the cellular adenylate energy charge to a low value. Only the mRNA abundance of MtAKINβγ, the hybrid that was specifically expressed prior to the induction of desiccation tolerance, was found to increase upon the cyanide treatment (Fig. 9, KCN). Genes of all other subunits, including MtAKINβ1 and MtAKINγ that were strongly induced by starvation, did not respond to the cyanide treatment.
Among the four β subunits that were characterized, MtAKINβ1 and β2 showed expression patterns similar to the Arabidopsis homologues examined by Bouly et al. (1999). Both the Arabidopsis AKINβ1 and MtAKINβ1 are highly expressed in vegetative tissues whereas only low expression levels were observed in reproductive organs. Likewise, AKINβ2 and MtAKINβ2 transcripts were mainly detected in reproductive organs. MtAKINβ3 contains only a short part of the conserved KIS domain (Fig. 1), as does its Arabidopsis homologue (AF491295). Finally, a fourth β subunit gene has been identified, MtAKINβ4. To our knowledge, this β subunit has not been previously characterized in other plant tissues and screening of the Arabidopsis databases did not reveal a sequence with significant similarity to MtAKINβ4. Three γ subunits were identified based on the presence of the four conserved CBS domains (Fig. 3). The maize homologue of MtAKINβγ and the tomato homologue of MtSNF4β are able to suppress the yeast Δsnf4 mutation (Bradford et al. 2000, 2003; Lumbreras et al. 2001). In Arabidopsis, the AKINβγ gene (AF439826) produces both a protein with a KIS domain, referred to as AKINβγ, and a smaller protein without the KIS domain, called AtSNF4 (Lumbreras et al. 2001; Kleinow et al. 2000). In M. truncatula, we were unable to detect cDNA clones without the KIS domain using a 5′-RACE approach (data not shown). In addition, northern analysis with a radio-labelled cDNA probe corresponding to the γ part of the AKINβγ did not reveal two bands of different size, suggesting that the differential splicing that occurred in Arabidopsis does not take place in M. truncatula. MtAKINγ was tentatively identified as a third γ isoform by the observation that the Arabidopsis homologue was found to interact with both the α and β subunits (Bouly et al. 1999) despite its inability to complement a yeast SNF4 deletion mutant. There are other plant proteins that show some similarity with SNF4 and bind to SNF1. One example is the SnIP1, a protein family that contains a short, hydrophobic motif that bears some resemblance to the recognition site that is present in known substrates of SnRK1 (Slocombe et al. 2002). However, considering that our selection for the putative γ subunits was based on the presence of the four CBS domains, this family was not incorporated in this study. Our results show that in M. truncatula, the transcripts of the four β and three γ isoforms accumulate differentially in vegetative and seed tissues and appear to be differentially modulated during development and upon stress.
Two γ subunits (MtSNF4b, MtAKINβγ) and one β subunit (MtAKINβ2) were found to be up-regulated during the induction of desiccation tolerance in germinated radicles by osmotic treatment. Interestingly, MtSNF4b and MtAKINβ2 exhibited similar expression patterns: their the transcript levels decreased during the loss of desiccation tolerance upon germination and increased only in germinated radicles that had become desiccation tolerant by the PEG treatment (Figs 5 and 7). Like MtSNF4b, LeSNF4 mRNA was also found to rapidly disappear upon imbibition of tomato seeds (Bradford et al. 2003). Moreover, conditions that prevented the completion of seed germination and maintained desiccation tolerance, such as low water potential and ABA, also maintained LeSNF4 mRNA abundance (Alvarado et al. 2000; Bradford et al. 2000, 2003). Indirect evidence also suggests that MtSNF4b expression appears to be regulated by ABA (Fig. 9, Bradford et al. 2003). Blocking the ABA synthesis during the PEG incubation of the desiccation-sensitive radicles led to a decrease both in the final percentage of desiccation-tolerant radicles and in the steady-state levels of MtSNF4b transcripts. Considering that the acquisition of desiccation tolerance involves the production of protective molecules such as sucrose as well as a possible co-ordinated down-regulation of metabolism (Hoekstra et al. 2001; Buitink et al. 2002), two intriguing questions remain to be answered: does a SnRK1 complex of a catalytic α subunit together with these two specific regulatory subunits exist in vivo and does it participate in the regulation of metabolism essential for the acquisition of desiccation tolerance?
The transcript levels of a second γ subunit, MtAKINβγ, increased transiently and peaked around 6 h, prior to the onset of acquisition of desiccation tolerance (Fig. 7a & b). This transient expression pattern was not detected in 5 mm germinated radicles that remained desiccation sensitive when submitted to the same osmotic potential, indicating that the cause of the transient MtAKINβγ gene expression is not solely due to an osmotic treatment. In addition, there could be developmental changes that make 5 mm radicles unable to acquire desiccation tolerance because they can no longer up-regulate MtAKINβγ in response to osmotic shock. These results are similar to those found in seed development of maize (Lumbreras et al. 2001). During embryo development, ZmAKINβγ genes were highly expressed around 10 d after pollination, prior to the acquisition of desiccation tolerance. As for MtAKINβγ, genes of maize βγ subunits did not respond to water deficit (Lumbreras et al. 2001). The repression of MtAKINβγ expression observed after 24 h of PEG incubation was inhibited in the presence of cycloheximide (Fig. 9), suggesting that MtAKINβγ gene expression is controlled either by a protein synthesis-dependent negative regulator, or by a negative feed-back mechanism. MtAKINβγ was also de-repressed when respiration was blocked after 24 h of PEG treatment (Fig. 9). Possibly, the increase in MtAKINβγ transcripts upon blocking the protein synthesis, a costly metabolic pathway in terms of ATP, or strongly reducing ATP production by addition of CN could mean that MtAKINβγ expression is sensitive to changes in the balance between energy demand and production. In yeast, SNF1 is a global regulator of carbon metabolism and is activated in response to low cellular glucose levels (reviewed by Gancedo 1998). In plants, SnRK1 has been proposed to be activated by high intracellular sucrose and/or low intracellular glucose levels (Purcell et al. 1998). Our results show that neither of the two γ subunits (MtAKINβγ and MtSNF4b) that have functional homologues (Lumbreras et al. 2001; Bradford et al. 2003) were expressed upon starvation. However, we did observe a considerable increase in transcript level of the putative γ subunit MtAKINγ as well as that of MtAKINβ1 (Fig. 8). The observation that transcripts of both MtAKINγ and MtAKINβ1 accumulated in germinated radicles upon starvation is in accordance with that of Bouly et al. (1999), who showed that AKINβ1 transcript level in Arabidopis leaves increased in the dark, a condition that affects carbon availability. Our starvation experiment is similar to that of Brouquisse et al. (1991) on excised germinated radicles of maize. These authors found that excision induced a sharp decrease in respiration rate, sugar contents and amounts of adenine nucleotides, whereas the adenylate energy charge, an indicator of the cellular metabolic status, remained unchanged. Glucose replenishment reversed the starvation effects on respiration (Brouquisse et al. 1991). In M. truncatula radicles, the starvation-induced gene expression of MtAKINγ and MtAKINβ1 was repressed when excised radicles were incubated in the presence of glucose (Fig. 8). Blocking the respiratory ATP production by cyanide is expected to induce a drastic decrease in adenylate contents but did not affect MtAKINγ and MtAKINβ1 gene expression (Fig. 9). Since respiration rates are mainly controlled by the sugar supply (Brouquisse et al. 1991), our data suggest that transcriptional regulation of the tandem MtAKINγ/β1 is more dependent on the sugar availability than on ATP homeostasis. Initial experiments with germinated seeds did not demonstrate significant differences in gene expression of both subunits after a 2-h incubation in different sugar solutions (data not shown). However, the time or developmental stage might not have been adequate to induce gene expression. MtAKINγ also responded to the osmotic treatment, suggesting that overlapping regulatory mechanisms exist between starvation and osmotic shock. Interestingly, a different β isoform, MtAKINβ3, was found to be up-regulated concomitantly with MtAKINγ.
Several studies have shown that different AMPK/SnRK1 complexes can exist in vivo (Thornton, Snowden & Carling 1998; Crawford, Halford & Hardie 2001; Ferrando et al. 2001). This study identifies non-catalytic subunit isoforms of the SnRK1 complex that are differentially regulated in seeds in response both to stresses leading to metabolic perturbations and developmental processes involving germination and desiccation tolerance. This study also demonstrates that some β and γ subunits can be simultaneously up-regulated. On the other hand, the transcriptional regulation of the subunit isoforms showing identical expression profiles does not appear to be mediated by the same factors (Fig. 9). If the two subunits are acting in the same complex then this could be a means of integrating different regulatory pathways. Further work is required to prove that the up-regulation of different regulatory subunits upon different stresses leads to the formation of specific SnRK1 complexes in vivo.
We thank B. Jettner (Seed-Co Australia Co-Operative Ltd, Hilton Australia) for the generous gift of Medicago truncatula cv Paraggio seeds. This work was supported in part by grants from the Ministère de l’Agriculture, de l’Alimentation, de la Pêche et des Affaires Rurales, Contrat Etat-Région 2000–06, Conseil Général de Maine-et-Loire and INRA.