Introduction

Signaling pathways initiated by plant hormones are relevant for almost every aspect of plant physiology, including growth, development and response to stress. Plant hormones belong to two groups depending on their downstream cascade: whereas auxin, gibberellic acid and jasmonic acid are perceived by specific receptors and transmitted through selective proteolysis, abscisic acid, cytokinin and brassinosteroids are perceived by cytoplasmic and/or membrane-bound receptors and then a reversible protein kinase cascade transmits the signal to cellular factors that finally affect the expression of target genes1. Abscisic acid (ABA), in particular, is fundamental at several stages of plant development such as seed formation and dormancy, stomatal closure and response to a variety of stresses2. The ABA signaling pathway has been recently established as a core formed by the soluble ABA receptors PYRABACTIN RESISTANCE/PYRABACTIN RESISTANCE-LIKE/REGULATORY COMPONENT OF ABA RECEPTORS (PYR/PYL/RCARs), which inhibit the TYPE 2C PROTEIN PHOSPHATASES (PP2Cs). This in turn favors the accumulation of the active form of SNF1-RELATED PROTEIN KINASES (SnRK2s) that modulate the expression of a large variety of ABA-responsive genes3,4.

Three main families of transcription factors have been described in ABA response: the B3-, APETALA2- (AP2), and basic leucine zipper- (bZIP) domain families, represented by ABI3, ABI4 and ABI5, respectively5,6,7. Multiple ABA-responsive genes have been reported by transcriptomic assays. Many of them contain the G-box ABA Responsive Element (ABREs) binding sites in their promoter regions8,9. Other transcription factor binding sites are present in the regulatory regions of ABA-modulated genes10. In fact, ~10–15% of all genes are deregulated after drought or salt treatments in an ABA-dependent manner9,11,12. Therefore, the task of defining the functional relevance of putative ABA target genes is difficult due to their large amount and complexity.

The Arabidopsis genome contains 15 proteins that contain a GRAM (Glycosyltransferases, Rab-like GTPase Activators, Myotubularins) domain, 9 of which were also originally annotated as “ABA-responsive protein related” because of their homology with ABA45 in barley, which is induced by ABA13. The GRAM domain is found in a wide range of organisms and consists of a ~70 amino acids region14 folded as the pleckstrin homology domain15 and mostly studied in animal myotubularin proteins that bind to phosphoinositol15,16. Among the plant GRAM domain-containing proteins that lack the feature of being related to ABA response, several have been reported such as VAD1 (VASCULAR ASSOCIATED DEATH1)17, involved in defense response in vascular tissues, and MYOTUBULARIN1 (MTM1) and MTM2, which affect the dehydration stress-responding transcriptome18. Interestingly several members of the subgroup putatively related to ABA response have been identified independently by their interaction with other proteins such as FIP1 (FH INTERACTING PROTEIN 1)19, PRSL1 (PP1 REGULATORY SUBUNIT2-LIKE PROTEIN)20, and GEM (GL2 EXPRESSION MODULATOR)21. More recently GEM-RELATED5 (GER5) has been implicated in regulating seed development and inflorescence architecture based on its expression pattern, which was shown to be in part overlapping with that of GER1 and GEM in reproductive organs22. Furthermore, ger5–1 mutants show reduced sensitivity to ABA and transcript changes in carbohydrate metabolism and catabolic processes22. The expression pattern of GRAM domain-containing proteins under various experimental settings, in particular those related to ABA, suggests that they may play functions in the perception of environmental signals and hormone signaling23. However, the potential relevance, if any, of GEM in ABA response has not been systematically explored.

Here, we focused on GEM to establish its role in ABA response by combining reporter gene expression studies and GEM expression studies in ABA signaling mutants with other molecular studies. We have found that GEM expression occurs in plant locations and stages associated with physiological increases of ABA. Accordingly, GEM is activated by ABA, in an ABRE-dependent manner, and requires some known regulators of the core signaling network of ABA. Germination assays showed that gem-1 seeds had increased dormancy levels than wild type seeds. Together, our results led us to conclude that GEM is part of the ABA signaling pathway and acts as a negative regulator of germination.

Results

GEM belongs to a subfamily of the GRAM domain-containing proteins

The GRAM domain has been found in a wide range of organisms from prokaryotes to eukaryotes. In plants, >260 proteins have been identified to contain this domain, most of which are monodomain proteins, unlike in animals in which GRAM domain is combined with a wide variety of other domains23. In particular, the Arabidopsis genome encodes 15 of these proteins. Among them, five proteins contain, in addition to the GRAM domain, another domain such as the C2 domain related to Ca2+-mediated signaling processes24,25 or the myotubularin domain18. The remaining 10 proteins contain a single GRAM domain and include VAD1, involved in hypersensitive response17, which is unrelated to the other nine proteins (Supplemental Figure 1). Based on sequence homology and phylogenetic analysis the latter set of proteins form a subfamily formed by GEM and 8 members that we propose to name GRE (GEM-RELATED; GRE1 through GRE8; Supplemental Table 1). We use GRE because the acronym GER had been already taken for GERMIN1 (GER1; AT1G72610), GER2 (AT5G39190) and GER3 (AT5G20630)26, which are proteins unrelated to the GRAM domain family. The size of these 9 proteins is also an overarching characteristic that separated them, with 210–299 amino acids, from the others larger proteins, with 598–1027 amino acids. This characteristic may be important taking into account the involvement of this domain in cell signaling. It must be noted that two of these proteins had been identified by independent studies: FIP1 (FH INTERACTING PROTEIN 1)19 and PRSL1 (PP1 REGULATORY SUBUNIT2-LIKE)20, which would correspond to GRE1 and GRE3, respectively, according to the nomenclature proposed here. Also, while this work in progress, GER5/GRE5 has been studied in detail and shown to be involved in seed development and inflorescence architecture22.

The GRAM domain of all GRE proteins is characterized by two unique motifs that differentiate them from the rest of GRAM domain proteins in Arabidopsis: (i) a highly conserved region we called CR1 (common region 1) with the amino acid motif CYL/ISTT/SxG (position 20–28 from the beginning of the GRAM domain; Fig. 1 and Supplemental Figure 2), and (ii) the common region 2 (CR2) with the amino motif DxxD/EFWFMGF, present close to the C-terminus (Fig. 1 and Supplemental Figure 2). Furthermore, we find other amino acids conserved throughout the sequence such as GGxE, just upstream of the GRAM domain (Supplemental Figure 2), and Y/HYR/K at the end of the GRAM domain, this latter lacking in GER3 sequence (Supplemental Figure 2). The role of these proteins is largely unknown, although their expression pattern under various experimental settings suggests that they may play functions in the perception of environmental signals and hormone signaling23. In this work, we focused on GEM to establish its potential role in this context.

Figure 1: Amino acid homology relationship of GEM and GER proteins.
figure 1

GRE proteins are characterized by the GRAM domain (dark grey), that includes a highly conserved common region CR1 (black) and CR2 (pale grey) towards the C-terminus.

Full size image

Expression pattern of GEM

To determine the spatial pattern of GEM expression we generated Arabidopsis transgenic plants (pGEM2774::GUS) expressing the GUS reporter gene under a 2774 bp sequence upstream of the ATG of GEM. Several independent lines were selected, showing a similar expression pattern, and the results described below are representative of them.

This gene appears to be expressed throughout several stages of vegetative and reproductive development, including roots, several stages of gametogenesis and developing seeds, although with limited expression at different developmental times. After germination, GUS expression is not detected (or at very low levels) in young seedlings (2 days, Fig. 2a) and then was detected at higher levels in the roots, particularly at the transition and differentiated regions (5 days; Fig. 2b). The same GUS expression pattern was observed in the lateral roots (Fig. 2c). The aerial parts of the seedlings did not show significant GUS staining except in the stipules of 6–7 day-old seedlings (Fig. 2d). Leaves and other regions during vegetative development did not show GUS expression, except in the hydathodes (Fig. 2e).

Figure 2
figure 2

Spatial expression pattern of GEM at the gene (a–p) and protein (q,r) level, using GUS staining and Western blotting, respectively. The pGEM2774::GUS plants were used for this study. (a,b) young seedlings. (c) Emerging lateral root. (d) Stipules. (e) Hydathodes. (f–k) Details of flowers at different developmental stages. (l-n) Pollen grain development. (o) Developing silique. (p) Detail of a mature silique. (q) Rabbit antisera were raised using a GEM-specific N-terminal peptide and used to detect protein by Western blot in extracts of wild type, gem-1 and GEMoe seedlings (14 day-old). Note that the band corresponding to GEM (black arrowhead) is not detectable in the gem-1 mutant, whereas a strong band of HA-GEM (empty arrowhead) is detected in the GEMoe plants. Molecular weight markers are indicated at the left. (r) Detection of GEM protein (black arrowheads) in various organs, as indicated. Note that the label “stipules” refers to the shoot apex containing stipules.

Full size image

At the reproductive stage (28 days), the GEM promoter is active at specific times of gametogenesis. A detailed analysis revealed a temporally limited expression pattern of GEM during pollen development (flower stage 8–10) (Fig. 2f–k). Thus, at anther stages I-IV, corresponding to pre-meiosis and meiosis, no expression was detected. At the post-meiotic stages V-VIII a clear expression was detected in anther primordia, including stages of microspore formation (Fig. 2f–h), although at later stages GEM expression is turned off. When the stamens have not yet grown over the ovary and pollination has not yet occurred GEM expression is also very apparent in the stigma (flower stage 12; Fig. 2i,j). A closer inspection revealed that GEM expression is particularly high at the tapetum layer (Fig. 2l). Then, the switch-off of GEM expression occurs concomitantly with the degradation of the tapetum layer. Expression decreases at later stages of pollen development (Fig. 2l–n). In the early stages of seed maturation (silique stages 3–5), expression is localized in all developing embryos within the embryo sacs, in places that may correspond to the endosperm layer (Fig. 2o). GEM expression is highly reduced at later stages when pigmentation of the seed coat becomes darker in the mature silique (Fig. 2p). This expression pattern is consistent with public microarray data (http://bbc.botany.utoronto.ca/efp/cgi-bin/efpWeb.cgi?primaryGene=AT2G22475&modeInput=Absolute).

To detect GEM expression at the protein level we generated rabbit polyclonal antibodies against a GEM-specific N-terminal peptide (see Methods). After purification of the polyclonal antisera with the peptide, we were able to detect GEM protein in whole cell extracts (Fig. 2q). A specific band was considered to be GEM since it was not detected in extracts of the gem-1 mutant21, which carries out a T-DNA insertion in the second intron of the GEM gene. It is worth noting that the mobility of GEM protein under our conditions (apparent molecular mass ~37 kDa) was slightly lower than expected from its amino sequence (~32.2 kDa).

The abundance of GEM mRNA and protein correlate nicely as revealed by Western blot of extracts from different organs (Fig. 2r). At the vegetative stage, GEM protein is abundant in roots, but also detectable in samples of the shoot apex containing the stipules and in the leaves. In agreement with the GUS expression pattern, GEM was also detected in flower buds and in developing siliques but its level was lower in dry seeds and virtually absent in imbibed seeds (Fig. 2r).

GEM expression is regulated by ABA

Previous studies suggested that GEM (and some GRE proteins) might be related to perception and/or signaling of hormonal response13,23. In fact, RNA in situ studies revealed that GEM, as well as GRE5/GER5 and GRE1/GER1, is expressed at the male and female reproductive organs22. We first used the pGEM2774::GUS plants to evaluate the transcriptional response of this gene to hormone treatments. We observed that the GEM promoter is strongly activated by ABA, but not by gibberellic acid (Fig. 3a). To evaluate the temporal pattern of GEM activation in response to ABA, we determined GUS activity at various times during ABA treatment and found that an increase in promoter activity was detected already 4 h after application of ABA (Fig. 3b).

Figure 3: Response of GEM expression to ABA treatment.
figure 3

(a) GEM expression is activated 24 h after treatment with ABA (100 μM) but not with GA (25 μM). (b) Time-course of activation of GEM expression by ABA (100 μM). (c) Identification of ABA-responsive elements in the GEM promoter.

Full size image

To gain further insight into the mechanisms controlling GEM expression we analyzed its promoter. A detailed in silico search using the PlantPAN software27 predicted the presence of binding sites for various transcription factors. Among them, two consensus ABA responsive elements (ABRE) were identified at −2345/2352nt (ACGTGTC) and −2394/2401nt (ACGTGGC) upstream of the translation start site (Fig. 3c). Consistent with previous data, pGEM2774::GUS showed that GEM expression increases in response to ABA in the transition and differentiated regions of the root (Fig. 3c). Since the ABRE sites are relatively far from the ATG codon, which is not common to most of these elements in other ABA responsive genes28, we sought to determine whether they were required for ABA response by using plants expressing a shorter promoter lacking the putative ABRE elements (pGEM2046::GUS) (Fig. 3c). ABA treatment of these reporter plants did not result in a detectable stimulation of GEM expression under these conditions (Fig. 3c), indicating that the region containing the ABRE sites is required for full response to ABA treatment.

GEM expression is activated by the ABA signaling pathway

The signaling cascade controlling the response to ABA, although very complex, has been delineated to a considerable extent around a basic core formed by the PYR/PYL/RCAR receptors, the PP2C phosphatases and the SNRK Kinases3. To start defining how ABA controls GEM expression, we determined the mRNA levels of GEM by qPCR in a collection of mutants affected at different stages of the ABA response. We determined the response of GEM to a 2 h treatment with 100 μM ABA and found that GEM mRNA levels were not largely affected in pyr112458, snrk2.2-2.3-2.6 and abi4-1 mutants, whereas in the case of abi1-2 and abi1-1 we observed a small increase and decrease, respectively, in GEM levels. This behavior is consistent with the known ABA-related mutant phenotypes. In addition, GEM levels in abi3-1 and abi5-1 were significantly decreased, suggesting that GEM is a putative component in the ABA response pathway (Fig. 4a).

Figure 4: GEM and GRE5/GER5 expression in ABA reposnsive mutant backgrounds.
figure 4

Relative amounts of (a) GEM and (b) GER5/GRE5 transcripts in the indicated mutant backgrounds in the absence and the presence of a treatment with 10 μM ABA for 2 hours. Values from two independent biological replicates were normalized to the ACT8 level and then to the value of GEM in each mutant. In all cases, except abi4-1 in panel a and pyr112458 and abi1-1 in panel b, differences were statistically significant with at least p ≤ 0.05.

Full size image

We also included the related GER5/GRE5 gene in this analysis since it is the most responsive to ABA29. We found that GER5/GRE5 mRNA levels in the various mutant backgrounds were different from those of GEM (Fig. 4b), which together with previous results22, reinforce the idea that they act at different stages in the ABA response.

GEM binds phospholipids

Animal GRAM domain-containing proteins have been reported to bind different phosphoinositides30,31,32. Plant proteins containing the pleckstrin homology (PH) domain, structurally related to the GRAM domain, bind phosphoinositides18,33. Therefore, we tested that possibility using GEM protein expressed in bacteria as a His-tagged protein as well as N- and C-terminal deletions (Fig. 5a) and a commercially available array of immobilized lipids. Full-length GEM protein binds strongly to PI3P, PI4P, PI5P and PA, and with less affinity to PI(3,5)P2, PI(4,5)P2 and PI(3,4,5)P3 (Fig. 5b). Neither the N-terminal moiety, nor the C-terminal that contains the GRAM domain, exhibited any detectable binding to these lipids, suggesting that binding requires the concerted action of several protein domains. It is worth noting that significant binding was detected only when the bacterially purified GEM protein was incubated with a plant extract, similar to previous reports16.

Figure 5: Identification of GEM-binding phospholipids and proteins.
figure 5

(a) Purification of bacterially-expressed full-length His-GEM proteins as well as its N-terminal and C-terminal moieties, as indicated. (b) Binding of polypeptides purified in (a) to various lipids was assessed by using membranes with lipids spotted as indicated in the left panel (LPA; lysophosphatidic acid; LPC, lysophosphocholine; PI, PtdIns; PI(3)P, PtdIns(3)phosphate; PI(4), PtdIns(4)phosphate; PI(5)P, PtdIns(5)phosphate; PE, phosphatidylethanolamine; PC, phosphatidylcholine; S1P, sphingosine-1-phosphate; PI(3,4)P2, PtdIns(3,4)diphosphate; PI(3,5)P2, PtdIns(3,5)diphosphate; PI(4,5)P2, PtdIns(4,5)diphosphate; PI(3,4,5)P3, PtdIns(3,4,5)triphosphate; PA, phosphatidic acid; PS, phosphatidylserine). Proteins bound were identified by Western blotting. (c) Identification of GEM-interacting proteins was carried by mass spectrometry. Extracts of wild type and GEMoe plants, expressing HA-GEM, were prepared and fractionated through polyacrylamide gel electrophoresis (a representative gel is shown here), as described in Methods. Proteins bands present uniquely in the GEMoe extracts were cut and subjected to mass spectrometry analysis. Molecular weight markers are shown at the left of the gel.

Full size image

To gain further insight into the GEM functional network we sought to identify in vivo GEM protein interactors. To this end, we took advantage of the GEMoe plants21 that express a HA-tagged GEM protein. Soluble protein extracts were prepared from these plants as well as from wild type (Col-0) plants, and bound to agarose-antiHA beads. After washing, bound proteins were eluted, fractionated through polyacrylamide gels (Fig. 5c), and individual bands present only in the GEMoe sample were subjected to identification by mass spectrometry. First, we identified a doublet with a mobility of ~48 kDa as HA-GEM, confirming the adequacy of our purification protocol. In addition, a prominent GEMoe-specific band with an apparent mobility of ~63 kDa was identified as PIP5K9, a phosphoinositol-4-phosphate (PI4P) kinase that converts it into PI(4,5)P2. Although at this stage, the functional relevance of the GEM-PIP5K9 interaction remains to be addressed in the future, the finding serves to reinforce the idea that GEM might participate in lipid-mediated signaling processes. This may provide a link to the relevance of GEM for ABA response since PA is able to inhibit the function of ABI134.

Seed dormancy depends on correct levels of GEM

ABA is known to be part of the pathway controlling seed dormancy3,35. Based on our data relating GEM and ABA we sought to determine if GEM participates in seed dormancy. To this end, we plated freshly-collected wild type, gem-1 and GEMoe seeds and quantified germination without stratification over a 7-day period when the wild type seeds showed a germination efficiency of ≥95% (Fig. 6a). The effect of GEM on seed germination was clearly observed by daily measurement of the germination efficiency in the absence of stratification. Under our conditions, the germination profile of gem-1 seeds was delayed compared to the wild type seeds (clearly observed at 3 days after sowing) whereas the GEMoe seeds followed a slightly advanced profile (Fig. 6a). This effect was virtually lost after 24 h of cold treatment of the seeds (Fig. 6b).

Figure 6: Relevance of GEM for dormancy and seed germination.
figure 6

(a) Germination frequency of wild type, gem-1 and GEMoe seeds, as indicated, was estimated on a daily basis. (b) Effect of the stratification time on the seed germination phenotype of wild type, gem-1 and GEMoe seeds, analyzed 3 days after sowing. Comparison with the wild type was assessed by the Student’s t test (*p ≤ 0.1; **p ≤ 0.05).

Full size image

Discussion

A total of 260 proteins of 35 species have been identified within the Viridiplantae clade containing a GRAM domain. This is comparable to the amount of GRAM domain-containing proteins identified in Metazoan species23. In spite of the identification of these proteins, our knowledge on the function of plant GRAM domain-containing proteins is very scarce and only scattered studies have been reported on some of them19,20. Here, we have focused on GEM, a GRAM domain-containing protein previously identified by its ability to interact with the DNA replication protein CDT1 and modulate root hair patterning21. We have now studied GEM in the context of its putative ABA response function previously assigned by its similarity with ABA4513. Together with other related proteins they constitute a subfamily of 9 members, including GEM, that we propose to name GRE (for GEM-RELATED) and share a GRAM domain as their main identifiable structural feature.

The GRAM domain, as well as its structural homolog domain, PH domain, is thought to be an intracellular lipid-binding and protein-binding domain involved in membrane processes during cell signaling in animals15. In plants, GRAM domain proteins have been related with hormonal and environmental perception14, although a lipid-binding activity has not been demonstrated so far. GEM constitutes the first example since it binds strongly to phosphoinositide-monophosphate (PI(3)P, PI(4)P, PI(5)P) and PA (phosphatidic acid), and to a lesser extent to PI(3,4)P2, PI(4,5)P2, PI(3,4,5)P3. It is worth noting that these interactions are detected only when recombinant GEM protein expressed in bacteria is incubated with a plant extract, suggesting that some post-translational modifications and/or association with other plant proteins are required for efficient binding to phospholipids. The same requirement in a similar assay has been previously reported for animal GRAM domain-containing myotubularins16. Quite interestingly, in this work we have identified by mass spectrometry PIP5K9, a PI4P kinase that produces PI(4,5)P2, as a GEM-interacting protein. This protein-protein interaction has been found by in vivo immunoprecipitation experiments that typically reveal high affinity interactions. This finding directly relates GEM with phospholipid metabolism, as it is the case of other GRAM domain containing protein in Arabidopsis, e.g., MTM118, which is involved in elevating the cellular level of PI(5)P in response to dehydration stress. Thus, PIP5K9 may be the protein that GEM requires for its phospholipid binding, a question that we relegate to future work. It is conceivable that GEM participates as a lipid transfer protein regulating the activity or location of signaling proteins as described about the GRAM domain of animal proteins16. In this sense, GEM, and also GRE3, has been recently reported to bind the serine/threonine phosphatase PP1c20, whose activity is regulated by PA binding36. In addition, root hair formation, where GEM has been implicated21, is regulated by membrane processes that implicate phospholipid kinases37 and phospholipid regulation38 linking lipid signaling with other GEM functions. The functional relevance of the GEM-PIP5K9 interaction remains to be addressed in the future. However, a PIP5K gene has been reported to be induced rapidly in response to drought, salt and ABA39. Together, our findings reinforces the relevance of GEM for ABA response, since PA is able to inhibit the function of ABI134, and the idea that GEM might participate in lipid-mediated signaling processes.

ABA is known to control the expression of nearly ~10% of all Arabidopsis genes, of which more than half are activated3,40. The altered expression of certain ABA or stress response genes has been also described in mutants of GRAM domain-containing proteins17. Our data indicate that GEM, as well as GER5/GRE5, behaves as early response genes to ABA treatment, although with distinctive features. We have demonstrated that the region containing the two ABRE binding motifs, identified in ~20% of ABA-responsive genes28, are required for ABA-mediated GEM expression. Given the relatively long sequence upstream the GEM transcriptional start site we can not discard that regulation of GEM expression respond to several other transcription factors. We have observed that GEM mRNA levels do not increase in response to ABA treatment in the abi5-1 mutant background. Interestingly, ABRE sequences are recognized by ABFs/AREBs, bZIP domain-containing transcription factors of which ABI5 is a member. This suggests that ABI5 might be a regulator of GEM expression in response to ABA. In further support of this conclusion, ABI5 is a target of SUMOylation by SIZ141 and GEM expression in the siz1 mutant does not respond to ABA treatment42. ABRE-mediated response to ABA appears to be a general feature of members of the GRE subfamily of GRAM domain proteins since, in addition to GEM and GER5/GRE5, the mRNAs of GRE1, GRE4, GRE6, GRE7 and GRE8 also increase after ABA treatment43. The promoters of all these genes, except GRE7 and GRE8, contain at least one ABRE-like motif. The effect of ABA on GEM expression may serve to link, at least in part, two independent observations regarding the effect of GEM and of ABA on root hair patterning21,44. Moreover, the inhibitory activity of ABA on the expression of CDT1a could also add complexity to this network given the interaction of CDT1a and GEM45,46. Based on the ground work established here, the possible relevance of this interaction in ABA response could be an attractive topic for future research.

In addition to ABI5 and SIZ1, other factors appear to be required for full GEM expression, suggesting that it is finely controlled by a complex set of coordinated pathways. Thus, the use of plants carrying mutations in various genes involved in ABA signaling response led us to show that activation of GEM expression depends on well-known factors in the pathway. The pyr1,pyl1,pyl2,pyl4,pyl5,pyl8 sextuple mutant in the receptors, the first element in the ABA signaling cascade47, has normal levels of GEM mRNA whereas those of GER5/GRE5 are reduced, pointing to other members of the receptor family as responsible for activating GEM expression. The next step depends on the PP2C-type phosphatase ABI1, a negative regulator of the SnRK2 kinases48. We found that plants carrying the abi1-1 allele, but not the abi1-2 allele, do not increase GEM expression in response to ABA. This observation is consistent with the idea that GEM is a downstream target since (i) the abi1-1 allele corresponds to a point mutation that abolishes binding to the PYR/PYL/RCAR receptors but not binding to the SnRK2 kinases, thus rendering a mutant ABI1 protein insensitive to ABA but maintaining its repressor activity on the SnRK2 kinases, and (ii) the abi1-2 allele, on the contrary, is a null T-DNA insertion mutant that leads to a constitutively activated ABA signaling49,50. These data are in agreement with transcriptomic analysis of abi1-1 and abi1-2 mutant plants29. The SnRK2 kinases are positive regulators of ABA signaling and we have found that GEM expression in response to ABA, but not GER5/GRE5 expression, is reduced in the snrk2 triple mutant51. SnRK2 kinases phosphorylate various transcription factors involved in ABA signaling, including ABI5, a member of the bZIP AREB/ABF family, the B3 type domain ABI3 and the AP2/APETALA member ABI440. As already mentioned, activation of GEM expression is abolished in the abi5-1 mutant. Furthermore, a similar result was obtained in the abi3-1 mutant, consistent with the participation of both ABI5 and ABI3 in GEM expression in response to ABA. Finally, the amount of GEM transcripts appears to be post-transcriptionally regulated by the 5′-3′-exoribonucleases XRN2 and XRN3, as deduced from the increased levels of GEM mRNA detected in the double xrn2 xrn3 mutant52. Interestingly, XRN2 and XRN3 activity is inhibited in response to stress by lipid second messengers, which levels depend on the phospholipid phosphatase ALX8/FRY/SAL1. Consistent with a role in maintaining GEM transcript stability, the alx8 mutant exhibits increased levels of GEM mRNA53, suggesting that GEM is a likely target of this post-transcriptional regulatory pathway. A different mechanism seems to apply to GER5/GRE5 since its transcript level is not increased in the alx8 mutant53.

The GEM expression pattern coincides temporally and spatially with the accumulation of ABA in several tissues and organs such as roots, the tapetum layer or endosperm during seed development54, the latter consistent with germination phenotype of gem-1 plants. Endosperm accumulates lipid and protein storage and is the main tissue responsible for seed dormancy55,56. The slight, although clear, dormancy phenotype in gem-1 plants is consistent with the hypothesis that GEM participates in ABA-mediated processes occurring during seed maturation. The increase in PA levels observed during seed development57 may be related to the ability of GEM to bind phospholipids. Thus, our results led us to tentatively place GEM as a positive effector of germination, necessary to break seed dormancy. However future studies are needed to unravel the complex role of GEM and the precise step of ABA signaling that is implicated.

Methods

Plant material and growth

Arabidopsis thaliana plants (Col-0 and Ler) were grown on ½ MS salts medium and 1% sucrose plus 0.8% agar (16 h/8 h light/dark photoperiod), at 22˚C and 70% relative humidity. Plants older than 10d were grown on soil at the same conditions. The homozygous T-DNA insertion line gem-1 (SALK_145846) and transgenic overexpressing line GEMoe have been described21. ABA-insensitive plants were provided by R. Solano (abi1-2), P.L. Rodríguez (hab1-1 abi1-2, pyr1-pyl1-pyl2-pyl4-pyl5-pyl8, snrk2.2-snrk 2.3-snrk 2.6) and O. Lorenzo (abi1-1, abi3-1, abi4-1, abi5-1).

Germination assays

For germination assays freshly harvested wild type Col-0, gem-1 and GEMoe seeds were collected at the same time from plants grown in the same conditions. For each genotype, one hundred seeds were sown in triplicate, after 0–24 hours at 4 °C in darkness, on MS medium and incubated at 22 °C under 16 h/8 h light/dark conditions. Germination was scored every 24 hours by radicle emergence through the seed coat.

Real-Time qPCR Analysis

Total RNA from seedlings and roots was extracted using Trizol reagent (Invitrogen). RT was carried out with the SuperScriptIII RT system (Invitrogen) using 500 ng of RNA as template and polydT primers. qPCR was performed in a LightCycler System using the FastStart DNA Master Green I (Roche) or an ABI Prism 7900HT SDS GoTaq qPCR master mix (Promega) with a 1:3–1:5 total cDNA dilution. The amount of actin (ACT8) mRNA was used for normalization before calculating fold changes. Data were generated from biological replicates. The following primer pairs were used to detect expression of: ACT8 (AT1G49240), forward 5′AACGACCTTAATCTTCATGCTGC3′, reverse 5′GGTAACATTGTGCTCAGTGGTGG3′; GEM (AT2G22475), forward 5′AGATAGCCTTGTCCGATGAG3′, reverse 5′TCTCCATCGTATCTTTCAAC3′; GER5/GRE5 (AT5G13200), forward 5′TCTGGCACAATCTGAAGACAGG3′, reverse 5′TGGTCAGCAGATGATGCGTAGC.

Antibody generation and Western-Blot assays

Polyclonal antibodies (anti-GEMpep) were generated using a GEM N-terminal peptide (LSDEVEIETKGSDS) of the first exon (39–53 aa) as antigen to inoculate rabbits (Biogenes, Germany). After purification of the antisera using GEM-HA beads, the antibodies were validated and tested by Western blot analysis in extracts of total protein (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM MgCl2, 0.2% Nonidet P-40). After transfering to membranes and blocking with 5% milk, a 1:5000 anti-GEMpep dilution was used. GEM protein was detected using a chemiluminescent procedure (ECLplus western-blot detection system; Amersham Bioscience).

GEM promoter:GUS fusions and transgenic plants

To generate the pGEM2774::GUS and pGEM2046::GUS constructs, the indicated fragments upstream of the translational start of GEM gene were amplified by PCR from the BAC F14M13 (GenBankAC006592) using primers 5′-GTCGACGCCAAGAAACCAAGAAAGAT-3′ paired with 5′-TTTAGGATCCACCTCAGTCTTCACTACG-3′, and 5′-TATAAAGCTTATACGAGCAAGGGCTG-3′ paired with 5′-TTTAGGATCCACCTCAGTCTTCACTACG-3′, respectively. These fragments, which also contained 41 bp of the N-terminal coding region of GEM, were cloned into pBI101 (Clontech) binary vector, fused in frame to the GUS gene, sequenced to verify the junctions of the chimeric constructs. Plants were transformed with the Agrobacterium tumefaciens strain C58CRifR using the floral dip method58. T1 seeds were plated on the germination medium containing 50 μg ml−1 of kanamycin.

GUS assay

Histochemical GUS assays were performed as described59 with slight modifications. Plant material was infiltrated by vacuum with the GUS substrate (100 mM NaPO4, pH 7.0, 0.5 mM K3Fe(CN)6, 0.5 mM K4Fe(CN)6, 0,1% Tween-20 and 4 mM X-Gluc) for a maximum of 10 min, and incubated at 37 °C in darkness. Samples were cleared by several changes of 70% ethanol. Samples were examined with an Axioskop2 plus microscope (Zeiss) with digital Coolsnap FX camera (Roper Scientific).

Binding to phospholipids

Constructs of GEM cDNA and its deletions NtGEM (coding for amino acids 1–170) and CtGEM (171–299) fused to 6xHis in a pRSET-B vector have been described21. Recombinant proteins were expressed in E. coli BL21 Rosetta after 3 hours of growth at 30 °C in liquid medium supplemented with isopropyl β-D-1-thiogalactopyranoside (IPTG) and purified using Ni-NTA beads (Qiagen). The phospholipid binding assay was done using PIP-Strips (P-6001, Echelon Biosciences, Salt Lake City, UT). The membrane was incubated with 0.5 μg of purified recombinant proteins, supplemented as indicated with 100 μg of total protein extracts of wild type plants. Immunoblot analysis was performed as described above using 1:3000 diluted mouse monoclonal anti-poly His antibody (Sigma, H1029) and 1:10000 anti-mouse antibody conjugated to horseradish peroxidase (Amersham).

Identification of GEM-interacting proteins by mass spectrometry

To identify GEM-interacting proteins we performed a co-immunoprecipitation assay using extracts of HA-tagged GEMoe plants. Protein extracts were prepared from 1 g of 14 day-old GEMoe plants grown in liquid medium. Plants were ground and homogenized in 1 ml of extraction buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM MgCl2, 0.2% Nonidet P-40) supplemented with plant protease inhibitor cocktail (Sigma) and phosphatase inhibitor cocktail (20 mM NaF, 0.1 mM Na orthovanadate, 5 mM Na pyrophosphate). The protein extract was sonicated on ice five times for 10 sec each and precleared with 50 μl of protein G-agarose for 15 min (10% flurry) (Santa Cruz Biotechnologies). After centrifugation at 3000 g, the cleared protein extract was incubated with protein G-agarose beads preincubated with anti-HA (Roche, 1:1000) with rotation for 1 h at 4 °C. Beads were washed five times with 1 ml of extraction buffer and finally proteins were eluted with 50 μl of 0.2 M glycine, pH 2.8 and immediately neutralized with 2 M Tris-HCl, pH 8.0. Proteins were fractionated by SDS-PAGE 8% long gels and transferred onto a nitrocellulose membrane. As a control, an extract of wild type plants (Col-0) was used. Protein bands detected by Coomassie Blue staining only in the GEMoe extract were extracted for peptide analysis. The molecular mass of tryptic fragments was measured with a MALDI Spectrometer and the sequence compared to the TAIR database.

Additional Information

How to cite this article: Mauri, N. et al. GEM, a member of the GRAM domain family of proteins, is part of the ABA signaling pathway. Sci. Rep. 6, 22660; doi: 10.1038/srep22660 (2016).