Water stress can result in reducing crop yield world-wide (Boyer, 1982; Smirnoff, 1993; Gonzalez et al., 1995) and a recent UN survey has underlined the importance of water deficit in our planet and its effects for the coming generations unless urgent measures are taken. This situation is exacerbated in arid and semiarid ecosystems. Here, legumes play a central agroecological role through their ability to use atmospheric nitrogen via the symbiosis with Rhizobia, and thus it reduces the need for fertilizers, improve food security, and generally favor the environment (Rubio et al., 2002; Naya et al., 2007; Kohler et al., 2008; Araújo et al., 2015; Ochatt, 2015). Studying a legume model species is thus timely and Medicago truncatula is of particular interest given its rather short life cycle and autogamy. It has a small and almost completely annotated genome (500–550 Mbp) which is publicly available (Goodstein et al., 2012), and it is more drought tolerant than other legume crops such as pea, bean, and soybean (Motan et al., 1994; Gonzalez et al., 1998; Costa França et al., 2000; Galvez et al., 2005). In spite of this, previous studies on water stress resistance in M. truncatula mostly concerned gene transfer (Alcântara et al., 2015; Araújo et al., 2015; Duque et al., 2016). The assessment of physiological responses (Nunes et al., 2008) and their genetic mechanisms (Badri et al., 2011) is more limited.

Osmotic stress or water deficit can be defined as the absence of adequate moisture necessary for a plant to grow normally and complete its life cycle (Cabuslay et al., 2002). Resistance mechanisms can be grouped into three categories: firstly escape, which enables the plant to complete its life cycle before the most intense period of water shortage, secondly avoidance, which prevents exposure to water stress, and finally tolerance, which enables the plant to withstand stress conditions (Levitt, 1972; Zhu, 2002; Golldack et al., 2014). Some resistance mechanisms are constitutive and active before exposure to water shortage. In other cases, plants exposed to water stress alter their physiology, thereby acclimating themselves to withstand drier conditions. One of the tolerance mechanisms activated under such stress is that of mitigating osmotic stress, via the production of osmolytes such as proline, and soluble sugars, that protect cells against osmotic perturbation (Choudhary et al., 2005; Valliyodan and Nguyen, 2006; Fulda et al., 2011; Elmaghrabi et al., 2013; Deinlein et al., 2014). On the other hand, φw (water potential) is also known to induce a morphological variation in tissues subjected to osmotic stress, notably at the cellular level. Such variation is potentially useful to understand biodiversity by identifying cellular responses to stress that are not necessarily picked up by taxonomic or phylogenetic indices that consider cell shape or size in vitro (Ochatt et al., 2008; Ochatt and Moessner, 2010). It is also important for assessing the competence for regeneration in vitro (Ochatt et al., 2008) following the recovery of tissues with a novel genetic makeup obtained via in vitro selection (Elmaghrabi et al., 2013) or gene transfer (Alcântara et al., 2015).

Responses to abiotic stress factors involve a reprogramming of the expression of 1000s of genes, which in turn result in the modification of a range of cellular and physiological processes (Cushman and Bohnert, 2000; Sreenivasulu et al., 2004; Araújo et al., 2015). One example of tolerance to stress at the molecular level, is the induction of P5CS that encodes Δ¹-pyrroline -5-carboxylate synthetase involved in proline biosynthesis (Silva-Ortega et al., 2008). This gene is highly expressed in salt-and drought-tolerant plant species (Choudhary et al., 2005) and it is induced under salt and water stress in many plant species including legumes (Chen et al., 2009). The P5CS gene was also up-regulated in M. truncatula in response to salt stress (Elmaghrabi et al., 2013). The kinetics of expression of genes involved in the cell cycle in plants exposed to high levels of abiotic stress has been the object of a number of studies (Gill and Tuteja, 2010; Zhao et al., 2014; Roy, 2016). In Arabidopsis, a negative regulator of mitosis, WEE1, is strongly expressed in response to abiotic stress (De Schutter et al., 2007). Osmotic stress imposed using polyethylene glycol (PEG) also up-regulated oxidative DNA damage and consequently DNA repair enzymes both in imbibed seeds (Balestrazzi et al., 2011) and in plantlets (Macovei et al., 2010). Our recent work with M. truncatula also showed an increased expression of WEE1 and CCS52 (CELL CYCLE SWITCH PROTEIN 52, another gene involved in the cell cycle) in salt-acclimated tissues as well as expression of genes involved in salt tolerance (SOS1 encoding a Na⁺/H⁺ antiporter) and embryogenesis in vitro [SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE 1 (SERK1)] (Elmaghrabi et al., 2013).

Tissue culture has been used in the selection of water stress tolerant cell lines that have been used to regenerate plants resistant to harsh environmental conditions in a range of crops including Medicago sativa L., tomato, soybean, and wheat (Sakthivelu et al., 2008; Guóth et al., 2010; Mahmood et al., 2012). Water deficit in vitro can be imposed through treatment with PEG 6000 (Ochatt et al., 1998; Guóth et al., 2010; Rai et al., 2011; Yang et al., 2012). The adsorbant property of this inert osmolyte provokes in plant cells and tissues the same or comparable effects to those obtained by drying soil at the same φw and without any other associated detrimental effects (Michel and Kaufmann, 1973). PEG 6000 thus closely mimics soil water stress (Lu and Neumann, 1998) and induces increases in total soluble sugars which serve as an osmoticum, or can be a source of respiratory substrates (Srivastava et al., 1995; Elmaghrabi et al., 2013). Additionally, PEG was shown to stimulate somatic embryogenesis in vitro (Attree et al., 1995; Igasaki et al., 2003). PEG 6000 was also used, and at similar concentrations as here (although osmolarity was expressed in MPa rather than in mOsm/kg as in this work), in studies on PEG-induced DNA damage with M. truncatula in vitro plantlets (Macovei et al., 2010) and seeds (Balestrazzi et al., 2011).

As many legumes are grown (or have their center of origin) in regions with an arid to semi-arid climate (Smýkal et al., 2015), a number of studies have identified genes, QTLs, ESTs, and SNPs that are responsive to drought stress in several species (Jacob et al., 2016). However, the molecular basis of water stress tolerance is not fully understood in M. truncatula (Araújo et al., 2015). The aim of the current work was to examine the extent to which treatments with PEG could enhance osmotic stress tolerance potential in callus of M. truncatula. In addition, the accumulation of osmoprotectants, the effects on cell morphology (shape and size) and division competence, and the expression of WEE1, CCSS52, P5CS, and SOS1 were monitored in the PEG treatments to investigate the mechanism underlying the induced water stress responses as compared to those activated in response to salt stress (Elmaghrabi et al., 2013). The expression of SERK1 was also analyzed, given its key role on the competence for the subsequent regeneration through somatic embryogenesis of plants that may potentially carry the stress resistance trait acquired.

Osmotic stress, provoked by insufficient ground and/or rain water, is a paramount constraint for plant growth and development. Cultures of callus on media that impose water deficit is a method for generating new, more tolerant, plants. Here, we have shown that long term culture of M. truncatula calli on medium containing 10% PEG6000 to impose an osmotic stress results in the production of morphologically enhanced calli. An analysis of protective metabolite levels, cellular morphology, cell division and gene expression was undertaken to understand the effects of the imposed stress.

In this work, callus growth was not significantly different plus or minus 10% PEG over a period of 6 months, suggesting that calli on PEG acquired tolerance to osmotic (water) stress (Figure 1) probably mediated by an early osmotic adjustment which was likely associated to various modifications at the cellular level (Singh et al., 2015). In fact, both cell viability and mitotic index were higher in the PEG treated cells compared to the control indicating a healthy and proliferating culture. It is likely that this sustained viability in the PEG treated cells is due to the activation of defense mechanisms that may include an activation of DNA repair as shown previously (Balestrazzi et al., 2011). The similarity in growth rates following 1 month ± PEG further stresses that tolerance was obtained relatively rapidly. This result differs from those of Biswas et al. (2002) who found that in rice, callus proliferation in the presence of PEG was greater than the controls in some genotypes, although this was at a much lower PEG concentration (5–15 g L^-1). However in two genotypes of wheat, one drought tolerant, the other drought sensitive, water deficit decreased only slightly in the sensitive compared with the tolerant genotype under water conditions and, it did not change significantly in either the sensitive or tolerant genotype ± 400 mM PEG 6000 (100–400 mOsm; -0.976 MPa; Guóth et al., 2010). Likewise in chili pepper cultures, where there was very good growth after 12 months in 5–10% PEG8000 (0.57 MPa; Santos-Díaz and Ochoa-Alejo, 1994). Note that after 6 months of treatment, although osmolarity of the callus increased, osmolarity of the medium did not change since PEG is not metabolized. We decided to analyze both the medium and callus osmolarity as an indirect way of assessing the nutrient consumption from the medium by cells, which impacts their internal salt concentrations, as shown before with various species among which M. truncatula (Ochatt et al., 2008). An increased cell osmolarity appeared also to be a reliable early marker of embryogenic competence (Ochatt et al., 2008; Elmaghrabi et al., 2013). Thus the increased callus osmolarity and embryogenic capacity seen here are in line with previous observations.

In this work (Figure 4) PEG-induced stress resulted in a highly significant increase of the size of both nuclei and cells after 6 months of culture on selection medium. Remarkably, this was also coupled with a consistent and significant modification of the cell shape, reflected by the SC values observed, indicative of an increased elasticity of cell walls under PEG-induced osmotic stress. A similar modification of cell wall elasticity was observed in M. truncatula plants subjected to a severe drought stress (Nunes et al., 2008) and in transgenic M. truncatula lines expressing the trehalose-6-Phosphate Synthase 1 (AtTPS1) from Arabidopsis thaliana with altered response to water deficit and recovery (Alcântara et al., 2015). Taken together, these observations suggest a profound elastic modification of the cell walls of water stress tolerant cells, perhaps deriving form a modified ratio among cell wall fractions, and should be the object of future studies.

Similar levels of somatic embryogenesis were observed in the PEG and control calli, however, calli in the PEG treatment were distinctly green in color compared with the control. This might be consistent with more robust embryogenic callus in the PEG compared to the control treatments (Figures 1, 2), which is not surprising since MANA medium is not conductive to full maturation of the somatic embryos formed in M. truncatula (Ochatt et al., 2013; Ochatt and Revilla, 2016). It may also reflect the fact that M. truncatula is adapted to semiarid conditions and even under severe drought stress, pigment content is not affected (Biswal, 1997; Nunes et al., 2008). PEG also improved somatic embryogenesis in other species (Attree et al., 1995; Igasaki et al., 2003). Both control and PEG treatments resulted in similar levels of somatic embryogenesis, which was consistent with the similar expression levels of MtSERK in the two treatments. Note that SERK1 is highly expressed during embryo induction and early somatic embryo development in M. truncatula (Nolan et al., 2009) and in Araucaria angustifolia (Steiner et al., 2012). The possible stimulation of somatic embryogenesis in response to PEG treatment is consistent with other reports showing stress-induction of this process (Karami and Saidi, 2010), which in M. truncatula may be linked to increases in ABA (Nolan and Rose, 1998).

The physiological, metabolic and gene expression responses of calli to PEG-induced osmotic stress mirrored those found under salt stress treatments (Elmaghrabi et al., 2013) in some respects but not in others, as summarized in Figure 7. In contrast to NaCl treatment, PEG treatment did not result in any increase in water content of the calli compared with the control although osmolarity did increase. This could be explained by the differential modes of action between NaCl and PEG, as the high MW of PEG exerts a constant osmotic pressure but does not allow its entry across the wall and hence avoids cell plasmolysis which results in different energy costs and different effects on growth (Munns, 2002). However, soluble sugars did increase suggesting they are a useful marker of both salt and water (osmotic) stress (Figure 5). PEG also induced a high level of proline accumulation, which was far higher than the largest proline accumulation under stress induced by NaCl (Elmaghrabi et al., 2013), indicating that this might also be a component of osmotic stress tolerance in M. truncatula (Figure 5B), as has been found in other species (Deinlein et al., 2014). Validating this hypothesis would require field trials with regenerants from these cultures and goes beyond the scope of this study. However, the expression of MtP5CS in callus grown on PEG was similar to the control (Figure 6D) and hence does not correlate with increased levels of proline in the PEG treatment. This was surprising given that this gene encodes an enzyme that is central to proline synthesis and that its expression in M. truncatula cultures exposed to salt stress was elevated (Elmaghrabi et al., 2013; Figure 7). It may suggest that this enzyme is not a key regulatory step in proline biosynthesis under these conditions, or that an initial rise in MtP5CS expression early in the culture period was sufficient to elevate proline concentrations and that after 6 months culture, increased gene expression was no longer necessary. In other words, whether modifying proline metabolism and the expression of genes involved in it, such as P5CS, may or not be used for engineering drought tolerance, and which approach should be adopted for such modification to be done remains uncertain (Bhaskara et al., 2015).

More predictably, PEG did not induce MtSOS1 expression which was down-regulated compared with the control. This gene is highly expressed in salt stress conditions as it encodes a protein that functions as a membrane-bound Na⁺ antiporter and contributes to Na⁺ depletion in the cytoplasm (Smith et al., 2010; Feki et al., 2011). Therefore the PEG data indicate a different (non-ionic) pathway leading to osmotic stress tolerance compared with NaCl (ionic) tolerance (Figure 7) as reported by Elmaghrabi et al. (2013).

The expression of MtCCS52 was upregulated by the PEG treatment (Figure 6E). In Arabidopsis this gene is a regulator of ploidy level and its expression is positively correlated to endoreduplication. In M. truncatula cultures exposed to long-term NaCl treatments this gene was up-regulated (Figure 7), alongside an increase in endoreduplication (Elmaghrabi et al., 2013). Given the clear evidence for endoreduplication in the osmotic stress-resistant cultures here (Figure 3), the upregulation of MtCCS52is consistent with its role in Arabidopsis (Vanstraelen et al., 2009).

That MtWEE1 expression was more highly expressed in the PEG treatment suggests that this gene may have a role in maintaining normal growth in a treatment that mimics osmotic stress conditions. WEE1 kinase might be necessary to regulate normal cell size in the face of ion toxicity and osmotic (non-ionic) water stress although this could only be resolved by exposing calli from M. truncatula wee1 knockouts to these treatments. Alternatively, as a gene that is expressed in the DNA damage and DNA replication checkpoints, it may be induced in response to either single strand or double strand DNA breaks as it is in Arabidopsis (De Schutter et al., 2007). However, Gonzalez et al. (2004) observed high expression of LeWEE1 in tomato (Lycopersicon esculentum Mill.) which was correlated with endoreduplication during fruit development. Our results do not seem to indicate that there has been irreversible DNA damage due to the osmotic stress imposed on callus, and it would therefore be legitimate to link this to WEE1 expression and its role in replication checkpoint and DNA damage and the possibility that the PEG concentration used and the long-term culture on it resulted in priming (Singh et al., 2015). Other genes could also be involved in the process though, and, in this respect, in order to protect their gene integrity from DNA damage plants are capable of activating a specific response system that regulates the cell cycle, but also DNA repair and programmed cell death where genes such as Suppressor Of Gamma response 1 (SOG1) (Yoshiyama et al., 2014) and Breast Cancer 1 (BRCA1) (Block-Schmidt et al., 2011) are known to play a central role in DNA repair, chromosome segregation and chromatin remodeling So the increase in MtWEE1 seen here may be both linked to the increase in endoreduplication and required to protect the cells from DNA damage induced by the PEG-induced osmotic stress treatment.

The data reported in this study of responses to 10% PEG compared with no PEG controls indicates that at this level of osmotic stress it is possible to induce a high level of embryogenesis with no penalty on growth rate. This appears to be achieved by up-regulating protective mechanisms such as the production of osmoprotectant solutes and switching on the expression of MtWEE1. The increase in MtWEE1 and MtCCS52 expression may cause an increase in endoreduplication while protecting the cells against the potentially damaging effects of the osmotic stress on DNA integrity.

HR, DF, and SO designed the project and the experiments. AE performed the experiments. All authors analyzed the data, read and approved the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.