Original cloning in tomato and recognition of the existence of a gene family

Back in 1993, a fruit cDNA library derived from tomato (Solanum lycopersicum) ripe fruit was screened by differential hybridisation with transcripts from tomato leaves under normal or water-stress conditions. As a result, a cDNA clone corresponding to an mRNA that is abundant in both stressed leaves and ripe fruits was isolated (Iusem et al. 1993). In parallel, the abscisic acid (ABA)-dependence of this transcript was demonstrated in leaves using a tomato mutant deficient for ABA synthesis (unpublished data). Soon later, a genomic clone with significant sequence identity to the original cDNA was isolated and named ABA-stress and ripening (Asr2) (Rossi and Iusem 1994), while the gene originally discovered in 1993 was referred as Asr1. This finding sheds light on the recognition of a new gene family rather than different alleles of the same locus.

Soon thereafter, Amitai et al. identified a third member of this family, Asr3 (sequence deposited in GenBank, unpublished data) and Rossi et al. (1996) reported that the three known tomato sequences originate from different but tightly linked loci located on chromosome 4. Ten years later, our lab found a fourth member (Asr4) in tomato (Frankel et al. 2006). In addition, wild Solanum species related to commercial tomato and chronically exposed to arid environments, namely, Solanum chilense and Solanum peruvianum, display a fifth paralogous gene, termed Asr5 (Fischer et al. 2011).

Further evidence on the organization of Asr genes in families in different species

Over the past 20 years, Asr genes have been reported in many plant species other than tomato, ranging from ancient gymnosperms to monocots and dicots (Table 1). With regard to monocots, the landscape of the Asr gene family in rice (OsAsr) is rather different and is composed of six members (Philippe et al. 2010). In contrast to tomato Asr genes, the rice genes are located on different chromosomes, except for OsAsr3 and OsAsr4, which are tightly linked on chromosome 1 (Philippe et al. 2010). Strikingly, OsAsr6 is the most divergent of the rice paralogs but is in turn very similar to the orthologous tomato Asr4 gene, particularly in its size, which encodes for a larger protein (Philippe et al. 2010).

Table 1 Asr genes discovered across the Plantae kingdom
Full size table

As far as maize (Zea mays) is concerned, its Asr gene (ZmAsr) family consists of nine paralogous genes, thus being larger than in tomato (four members, Frankel et al. 2006), banana (four members, Henry et al. 2011) and rice (six members, Philippe et al. 2010).

Other monocot species, such as those of the genus Musa (plantain and banana), possess a four-member Asr family, much like tomato. These Asr genes are similarly clustered on the same chromosome (Henry et al. 2011).

Regarding dicots, an Asr ortholog was also reported in wild potato (Solanum chacoense) and named DS2 (Silhavy et al. 1995). Single Asr genes have also been found in other Solanaceous plants, for example, in Nicotiana species (Yang et al. 2012). Halophyte plants such as Salicornia brachiata, which grow luxuriantly on salty shores, also contain Asr genes, which are induced by high salinity in such extreme environments (Jha et al. 2009).

Gene structure and evolution

Tomato Asr genes are short with a very simple structure: two exons separated by an intron (Gonzalez et al. 2011). In particular, the Asr2 genomic sequence is 73 % homologous to Asr1 (Rossi and Iusem 1994), with an AT-rich regulatory region that has been studied in depth (Rossi et al. 1998; Rossi and Iusem 1995). For example, the Asr2 promoter has a putative ABA-response motif, and intriguingly, its 3′ untranslated region (3′ UTR) shares 92 % homology with an intron from the polygalacturonase gene, whose expression increases during fruit ripening (Rossi and Iusem 1995).

The general simple structure of tomato Asr genes with only two exons separated by an intron is also shared by other species, for example, those of the genus Musa (plantain and banana) (Henry et al. 2011) and the halophyte plant S. brachiata (Jha et al. 2009).

When members of the commercial tomato Asr gene family were compared with their orthologous genes from different Solanum wild species, Asr1 showed a very low level of non-synonymous substitutions throughout its entire sequence, indicative of a strong purifying selection (Fischer et al. 2011; Frankel et al. 2006). In contrast, Asr2 showed a higher ratio of non-synonymous to synonymous substitutions when compared with species inhabiting arid regions, implying the action of positive, adaptive selection on this gene during evolution (Frankel et al. 2003). Similarly, the Asr2 gene showed intra-specific variations in wild tomato populations that were subject to different rainfall regimes (Giombini et al. 2009). However, Fisher et al. found no evidence for adaptive evolution in the Asr2 coding sequence, in contrast to Asr4 (Fischer et al. 2011). Nevertheless, the same author recently found that the Asr2 regulatory region was indeed subject to positive selection in Solanum chilense, a species adapted to very dry environments (Fischer et al. 2013).

Interestingly, orthologous genes were much more alike in pair-wise comparisons (Asr1 vs. ci21a; Asr2 vs. ci21b; and Asr4 vs. DS2) than were paralogous genes (Asr1 vs. Asr2; ci21a vs. ci21b; etc.) (Frankel et al. 2006). One conclusion from that analysis was that the Asr family is ancient (300 million years ago), and its origin derives from gene duplications occurring before the divergence of the tomato and potato branches. Strikingly, Asr3 does not have a potato ortholog. A similar evolutionary history is inferred from the rice and maize Asr genes, as they cluster together and exhibit more similarity between orthologs than between paralogs (Frankel et al. 2006).

Unlike tomato, rice Asr genes are located on different chromosomes, though OsAsr1 and OsAsr2 are tightly linked in chromosome 1, most likely as a consequence of a recent duplication event. OsAsr6 is also located on chromosome 1 but is distant from the first two paralogs. Finally, OsAsr3, OsAsr4 and OsAsr5 are located on different chromosomes. This divergence between locations could be the result of duplication events taking place on chromosome segments or even involving whole genomes (Philippe et al. 2010).

Another evolutionary lesson arose from data regarding the common bean (Phaseolus vulgaris), which has two Asr genes, each of which have a different evolutionary history. The first, Asr1, underwent little nucleotide variation between wild and cultivated bean, indicative of a strong purifying selection, whereas Asr2 seems to have been the target of adaptive selection (Cortes et al. 2012).

To present an overall picture of the relatedness of ASR proteins in terms of sequence, a phylogenetic tree is displayed in Fig. 1 (Frankel et al. 2006). Another evolutionary study on this protein family has been performed in depth by Henry et al. (2011), focusing on Musa species (plantain and banana). This analysis revealed that, unlike tomato ASRs, those from Musa cluster together rather than with ASRs from other species, suggesting that they were the product of recent duplication events.

Fig. 1
figure 1

Neighbor-joining tree of ASR proteins from different plant species (taken from Frankel et al. 2006). The numbers in nodes are bootstrap values (only values >70 % are shown). Bar length indicates p-distance. The tree is rooted with the Ginkgo-Pine group (gymnosperms). Permission to reproduce this figure was obtained under License Agreement Number 3313060861664 between N. D. I. and Elsevier publishing company

Full size image

Epigenetics

Tomato is the only species in which epigenetic studies on Asr genes have been conducted. Asr1 and Asr2 revealed epigenetic marks on both DNA cytosines and histones (Gonzalez et al. 2011, 2013). For example, in leaves, Asr1 exhibits H3K27me3 marks along with cytosine methylation in all methylation contexts (CG, CNG and CNN), with the latter context (asymmetric methylation) being atypical within the body of a non-repetitive gene such as Asr1 (Gonzalez et al. 2011). Interestingly, CNN methylation and H3K27me3 decrease after water deprivation, concomitant with an increase in Asr1 mRNA levels (Gonzalez et al. 2011). In roots, CNN methylation in the regulatory region decreases under water stress, along with a drop in H3K9me2 within the regulatory and coding regions of Asr2 (Gonzalez et al. 2013).

Expression analyses

In tomato, Asr1 transcript was found in stressed leaves and fruit pericarp (Iusem et al. 1993). The ASR1 protein levels also increase upon to the osmotic stress induced by polyethyleneglycol (PEG) (Amitai-Zeigerson et al. 1995). Both Asr1 and Asr2 expression levels increased in dehydrated leaves (whole organ), but only Asr2 levels rose in roots (Maskin et al. 2001). A deeper examination of the cell types involved revealed Asr1 mRNA in leaf vascular tissue irrespective of stress, whereas Asr2 mRNA was observed in phloem companion cells, expanding to mesophyll cells (Maskin et al. 2008). In addition, Asr1 expression levels were high and constant throughout fruit ripening (Maskin et al. 2001). The ASR1 protein also accumulates in tomato seeds, an organ subjected to physiological desiccation during final maturation drying (Maskin et al. 2007).

Consistently, ASR proteins are also crucial in pollen, especially during the drying stage, where protection against drying conditions is vital for subsequent grain maturation (Wang et al. 2013). In contrast to tomato, the orthologous potato (Solanum tuberosum) water stress-inducible Asr gene (named DS2) is insensitive to ABA (Dóczi et al. 2002).

Of the six rice Asr genes, at least OsAsr5 is induced by gibberellins (GA), as well as by ABA. This could imply that Asr5 is not only involved in stress tolerance but also in plant growth, particularly in leaf expansion (Takasaki et al. 2008). OsAsr5 is also implicated in aluminum (Al) tolerance (Arenhart et al. 2013b). In the presence of Al ions, all six OsAsr genes increase their expression level, particularly OsAsr5, whose induction turned out to be the highest in Al-resistant cultivars (Arenhart et al. 2013b). Furthermore, the encoded OsASR5 protein is present in the chloroplast probably as an inactive transcription factor that could be released to the nucleus in response to Al to regulate genes related to photosynthesis (Arenhart et al. 2012). On the other hand, OsAsr1 is induced by ABA, a situation that in turn leads to higher tolerance to water and osmotic stress (Joo et al. 2013a). Similar to tomato, rice Asr genes exhibit non-overlapping expression patterns in different tissues and have different responses to ABA and water stress, as demonstrated by qRT-PCR and GUS staining (Perez-Diaz et al. 2014). For example, OsAsr1 is expressed mostly in leaves, whereas OsAsr3 is preferentially expressed in roots. Together, these two proteins are the most abundant ASR proteins in rice, being present in a variety of tissues. In addition, they are induced by abiotic stress but stimulated differentially by plant hormones (ABA or GA) and by different sugars, such as sucrose and glucose (Joo et al. 2013b).

In wheat, another important cereal crop, a proteomic study revealed that an ASR protein is up-regulated during water stress (Bazargani et al. 2011), whereas in the large maize Asr family, ZmAsr1 has the highest expression levels (Virlouvet et al. 2011).

Studies conducted with other species showed that, for example, in pinus (Pinus taeda), the only orthologous gene analyzed (named lp3) is expressed mostly in roots under water-deficit conditions (Padmanabhan et al. 1997), while in lily (Lilium longiflorum), the orthologous protein is present at least in developing pollen, induced by ABA and osmotic stress, i.e., PEG (Huang et al. 2000).

When tobacco plants were engineered with the pinus Asr (named lp3) coding sequence driven by its own promoter, the exogenous lp3 regulatory region was widely recognized by the heterologous transcription machinery in root and shoot meristematic regions, most cell types in leaves except the petiole, trichomes, root hairs, stems, pistils of developing floral buds, developing ovary and embryos, and developing seeds. Moreover, such a promoter proved to display elements that respond to hormones including methyl jasmonate and ABA in the recipient species. As with many LEA genes, lp3 expression decreased after seed germination, most likely as a consequence of methylation (Wang et al. 2002).

Subcellular localisation

Tomato ASR1 was first reported solely as a nuclear protein (Iusem et al. 1993) but was later detected in both the cytosol and the nucleus (Kalifa et al. 2004a). Consistent with this compartmentalized distribution, a nuclear localisation signal (NLS) was putatively identified (Kalifa et al. 2004a) but turned out to be non-functional upon experimental testing (Ricardi et al. 2012). In tobacco plants, ASR was also found in the nucleus (Yang et al. 2012).

Interestingly, during the early stages of pollen maturation, the native ASR protein from lily translocates from the cytosol into the nucleus (Yang et al. 2008). However, in this case, the existence of a functional NLS was concluded based on the fact that when this signal was artificially removed, the protein was found only in the cytosol (Wang et al. 2005).

The use of heterologous in vivo systems such as Saccharomyces cerevisiae and Nicotiana benthamiana showed that tomato ASR localizes mostly in the cytosol and usually forms granules that are readily visible using confocal microscopy (Bermudez-Moretti et al. 2006; Ricardi et al. 2012; Urtasun et al. 2010). A different pattern of subcellular distribution was observed in another heterologous system consisting of tobacco expressing the pinus Asr (named lp3), which revealed that the exogenous LP3 protein localizes exclusively in the nucleus, where it might protect against dehydration (Wang et al. 2002).

Protein structure

Tomato is the species in which ASR proteins have been analyzed most deeply at the structural level. In particular, ASR1 from this species turned out to be a small, highly charged 13-kDa polypeptide rich in Gly (7 %), Ala, (13 %), Glu (15 %), His (15 %), and Lys (17 %) with an isoelectric point of 7.9 (Iusem et al. 1993). It exhibits two distinct domains: a DNA-binding domain located at the carboxy-terminal end, and a zinc-binding domain capable of binding two Zn atoms at the amino-terminal end (Rom et al. 2006).

Recombinant ASR1, purified via Ni2+-affinity chromatography, is an intrinsically unfolded monomer in vitro, as shown by several biophysical methods. However, in the presence of Zn2+ ions, it undergoes a conformational transition from unfolded to folded as it gains more α-helix and β-strand domains, which implies a more highly ordered polypeptide structure (Goldgur et al. 2007). In addition, the expected solubility of ASR1 at high temperatures in the presence of Zn2+ was confirmed by microcalorimetry, where the peak of heat absorbance was detected between 70 and 80 °C (Goldgur et al. 2007), in line with the observed solubility at 90 °C of the ASR ortholog from lily (Wang et al. 1996).

Recently, soybean ASR (GmASR) has also been shown by in vitro intrinsic fluorescence spectroscopic assays (circular dicroism) to bind metal ions like Fe3+, Ni2+ and Cu2+ (Li et al. 2013).

To the best of our knowledge, the only in vivo structural study on an ASR protein is the one reporting ASR1 to self-assemble as homodimers and even displaying more complex quaternary structures (Ricardi et al. 2012).

Classification of ASR proteins

ASR proteins have been independently proposed by Caramelo and Iusem (2009), Battaglia et al. (2008) and Hunault and Jaspard (2010) to constitute a novel group of the LEA (Late Embryogenesis Abundant) superfamily based on: (1) their accumulation in seeds (see “Expression analysis” section), (2) some of their basic structural features mentioned in the previous section and (3) their expression under drought conditions. In this respect, Battaglia and colleagues’ group 7 corresponds to LEAPdb class 12 [PF 10714] in the classification of Hunault and Jaspard (http://forge.info.univ-angers.fr/~gh/Leadb/leaclasses.png). However, the same latter authors do not include ASR proteins in the LEA family in their recent bioinformatic study (Jaspard et al. 2012). Considering that definition of LEA proteins is not always clear-cut, it might be more practical to comply with the Sanger Institute’s Pfam database, which classifies ASRs under the ABA/WDS (ABA/water-deficit stress) category (PF02496). Besides, as ASR proteins have no sequence similarity to the accepted groups of LEA proteins and as their function as transcription factors (see next section) also distinguishes them from the known LEA proteins, we support the view that ASRs are classified as an important group of ABA/WDS proteins with distinct and unique functions and roles.

Molecular function and possible physiological roles

The elusive function of ASR proteins stimulated interesting research through both in vivo and in vitro experiments. For example, ASR1 was reported to bind in vitro to chromatin in a zinc-dependent fashion (Kalifa et al. 2004a). The physical association between DNA and both ASR1 monomers and dimers was directly observed in vitro by Atomic Force Microscopy at the single molecule level (Maskin et al. 2007) and most likely involved the binding to the short consensus DNA sequence previously determined by SELEX (Kalifa et al. 2004a).

It is worth mentioning that the grape (Vitis vinifera) ASR ortholog named VvMSA was claimed to bind to the enhancer of a sugar transporter gene (VvHT1), although the reported regulation was not clear (Cakir et al. 2003). Therefore, a transcription factor function was later proposed (Cakir et al. 2003), as was an ASR protein-mediated cross-talk between ABA and sugars (Carrari et al. 2004).

As far as their role as transcription factors is concerned, it was not until very recently that in vivo (preserving the native chromatin structure) genome-wide Chromatin Immunoprecipitation (ChIP) data were reported in rice (Arenhart et al. 2013a) and tomato (Ricardi et al. 2014). In rice, some of the putative targets of ASR5 in roots are related to Al tolerance like the ABC transporter STAR1 (Arenhart et al. 2013a). On the other hand, the targets of tomato leaf ASR1 turned out to be gene encoding proteins involved in water transport like aquaporins and cell wall remodeling proteins, namely cellulose synthase and glucanases (Ricardi et al. 2014). However, ASR1 does not seem to interact with sugar transporter genes (Ricardi et al. 2014), as reported in grape (Cakir et al. 2003).

In addition to direct targets, some ASR proteins seem to regulate different target genes, albeit indirectly, as revealed by double hybrid assays in yeast. From example, the grape ortholog interacts in vivo with a drought responsive element (DRE)-binding transcription factor to form a heterodimer (Saumonneau et al. 2008). Similarly, tobacco ASR in vivo binds to a transcription factor involved in leaf senescence and floral development (Yang et al. 2012).

Interestingly, ASRs (supposedly the cytosolic pool) might also act as chaperone-like proteins. For example, tomato ASR1 protected reporter enzymes against freezing or heat denaturation in vitro (Konrad and Bar-Zvi 2008). Similar results were found with ASRs from plantain (Dai et al. 2011) and lily (Hsu et al. 2011).

Also worthy of note is the fact that ASR from soybean exhibited a novel biochemical function: an antioxidant activity in vitro, but no model of underlying mechanism was formulated (Li et al. 2013). On the other hand, maize Asr genes seem to be involved in regulating the biosynthesis of branched-chain amino acids such as Val, Leu and Ile (Virlouvet et al. 2011).

At the physiological level, the function of ASRs as sugar transporter regulators was observed in transgenic potato and tobacco plants in which endogenous expression of Asr was artificially silenced. In these cases, glucose transport was altered as a consequence of the changes in the expression pattern of sugar transporter genes, which, in turn, were indirectly caused by the loss of ASR (Dominguez et al. 2013; Frankel et al. 2007).

Physiological effects of Asr genes in Arabidopsis and yeast

Classical model organisms lacking endogenous Asr genes but carrying foreign ones have also been tested at the physiological level. For example, tomato Asr1 was used to transform yeast (S. cerevisiae) strains that lack the MAP kinase hog1 and are, therefore, deficient in the high-osmolarity glycerol pathway. This strain is unable to grow under high salt conditions, but its resistance increased notably when carrying Asr1 through an unknown mechanism (Bermudez-Moretti et al. 2006).

On the other hand, transgenic Arabidopsis plants carrying and overexpressing the Asr gene from lily exhibit reduced sensitivity to ABA, diminished levels of dormancy and increased resistance to salt, osmotic, drought and cold stresses (Yang et al. 2005). Another report on the same species claims that, when transformed with tomato Asr1, individuals exhibited a phenotype similar to abi4 (ABA-insensitive 4) mutants, namely, no sensitivity to ABA-triggered inhibition of seed germination. This behavior resulted from competition between the endogenous transcription factor ABI4 and the exogenous ASR1 for binding to a cis-regulatory DNA sequence (Shkolnik and Bar-Zvi 2008). Arabidopsis plants were also engineered with the plantain (Musa paradisiaca) Asr transgene, which conferred resistance to osmotic stress (Dai et al. 2011).

Phenotypes of transgenic crop plants eventual biotechnological applications

There have been several studies on the heterologous expression of Asr genes in crop plants. The first was carried out by Wang et al. (2002), who produced tobacco expressing the pinus Asr coding sequence driven by its own promoter. However, no phenotypes were reported therein except for expression data described in the corresponding section of this review.

In another study also with tobacco, plants carrying the Asr1 cDNA from tomato driven by the constitutive 35S promoter showed improved tolerance to osmotic stress, likely due to decreased uptake of Na+ ions from the environment (Kalifa et al. 2004b). Leaves from these plants turned out to lose less water than wild type leaves, suggesting regulation at the stomatal level. Similar results were obtained in transgenic tobacco plants bearing the Asr cDNA from the halophyte S. brachiata (Jha et al. 2012). Analogously, when wheat Asr was overexpressed in tobacco, plant tissues showed higher water content and higher catalase and superoxide dismutase [enzymes involved in reactive oxygen species (ROS) detoxification] activities, resulting in a promising high tolerance to water stress (Hu et al. 2013).

In maize, plants constitutively overexpressing an additional copy of their own Asr1 gene exhibited a higher foliar senescence ratio than wild type plants under drought conditions (Jeanneau et al. 2002). The resulting ASR1 accumulation also caused an increase in biomass and yield (Virlouvet et al. 2011), which could be attractive for biotechnological purposes. Also appealing is the case of transgenic rice plants overexpressing an extra copy of their endogenous Asr1, in which a higher tolerance to cold stress was observed in terms of photosynthetic efficiency (Kim et al. 2009).

Potato plants have been the target of genetic transformation as well and those overexpressing tomato Asr1 showed an increase in the concentration of sugars in tubers, allowing for bigger tubercles but reducing their number. Conversely, the silencing of endogenous ci21a in potato plants led to smaller tubers with lower glucose concentrations (Frankel et al. 2007).

Concluding remarks and perspectives

The widespread ASR proteins, which have been studied by many groups worldwide, are only a component of the drought stress response, albeit an important one, making them attractive for interdisciplinary research. As a matter of fact, over the last 20 years, the field has drawn the attention of physiologists, biochemists, biophysicists, evolutionary biologists and bioinformaticians as well as experts on genomics and plant biotechnology.

Making a critical appraisal of the progress made up to date on the subject, in vivo structural approaches are still scarce (Ricardi et al. 2012) and should be considered for future studies to deepen our understanding of the chaperone-like functions of ASR proteins by revealing details of the changes in their 3-D structure resulting from stress (Caramelo and Iusem 2009). Concerning the role of ASRs as transcription factors, the recent and long-awaited breakthrough on the discovery of their target genes (Ricardi et al. 2014; Arenhart et al. 2013b), combined with further research, surely will gain insight into both the early environmental stress-sensing molecular events triggered by ABA and the late physiological adjustments that finally confer drought tolerance.

From an applied perspective, despite a natural skepticism towards engineering drought tolerance using a single transgene (Caramelo and Iusem 2009), we were able to compile several promising examples of ASR proteins that improve drought resistance in the laboratory. Nevertheless, it is imperative to perform field tests to ascertain the agronomic usefulness of the reported transgenic plants.

Finally, we express our apologies to those whose work could not be cited due to space limitations.