Learning from halophytes: physiological basis and strategies to improve abiotic stress tolerance in crops

Abstract

Background

Global annual losses in agricultural production from salt-affected land are in excess of US$12 billion and rising. At the same time, a significant amount of arable land is becoming lost to urban sprawl, forcing agricultural production into marginal areas. Consequently, there is a need for a major breakthrough in crop breeding for salinity tolerance. Given the limited range of genetic diversity in this trait within traditional crops, stress tolerance genes and mechanisms must be identified in extremophiles and then introduced into traditional crops.

Scope and Conclusions

This review argues that learning from halophytes may be a promising way of achieving this goal. The paper is focused around two central questions: what are the key physiological mechanisms conferring salinity tolerance in halophytes that can be introduced into non-halophyte crop species to improve their performance under saline conditions and what specific genes need to be targeted to achieve this goal? The specific traits that are discussed and advocated include: manipulation of trichome shape, size and density to enable their use for external Na⁺ sequestration; increasing the efficiency of internal Na⁺ sequestration in vacuoles by the orchestrated regulation of tonoplast NHX exchangers and slow and fast vacuolar channels, combined with greater cytosolic K⁺ retention; controlling stomata aperture and optimizing water use efficiency by reducing stomatal density; and efficient control of xylem ion loading, enabling rapid shoot osmotic adjustment while preventing prolonged Na⁺ transport to the shoot.

INTRODUCTION

It is estimated that about 3·6 billion of the world's 5·2 billion ha of dryland used for agriculture suffers from erosion, soil degradation and salinization (Riadh et al., 2010). Salt-affected soils impact upon nearly 10 % of the land surface (950 Mha) and 50 % of all irrigated land (230 Mha) in the world (Ruan et al., 2010). It is estimated that in Australia alone, 67 % of agricultural land has the potential for transient salinity (Rengasamy, 2006), costing the Australian farming economy in the vicinity of A$1330 million year⁻¹ (Rengasamy, 2002). The global annual losses in agricultural production from salt-affected land are in excess of US$12 billion and rising (Qadir et al., 2008; Flowers et al., 2010).

Moreover, it is not the economic penalties per se, but the fact that agricultural food production needs to increase between 50 and 70 % (Brown and Funk, 2008; Ruan et al., 2010; Millar and Roots, 2012) by 2050 to match the projected population growth to 9·3 billion that is of a major concern. This increase cannot be achieved by merely using the currently available arable land. Moreover, climate change over the next decades is expected to decrease annual precipitation in sub-tropical regions. As a consequence, in presently irrigated areas, we can expect that good quality water will be increasingly reserved for drinking and urban use, and farmers will need to turn to the use of brackish and saline water for irrigation (Barrett-Lennard and Setter, 2010). Adding to the problem is the developing energy crises. Fossil fuels are being heavily depleted, and energy plant species (produced for biofuels) are starting to compete with food crops for productive land (Valentine et al., 2012). Also, the urban sprawl around cities and regional centres results in conversion of agricultural land into other uses. In Australia alone, this trend has already resulted in a loss of approx. 10 % of agricultural land over the last 10 years, from 456 Mha in 2001 to 409 Mha in 2009 (Millar and Roots, 2012).

The solution to these issues is 2-fold. The first (intensive) approach is to increase crop production dramatically on currently used arable land by improving plant productivity under stress conditions. This implies a major breakthrough in crop breeding for stress tolerance. Given that the diversity for stress tolerance within traditional crops (including landraces) is likely to be too narrow to achieve this goal (Colmer et al., 2005), stress tolerance genes must be identified in extremophiles and then introduced into traditional crops. As argued in this review, halophytes could be the number one choice to achieve this goal. The second (extensive) approach is to shift the focus of agricultural production to marginal (currently unproductive) land. For saline areas, this can be achieved by using alternative crop species, e.g. domestication of halophytic plants suitable for highly saline environments (Epstein et al., 1980; Flowers and Yeo, 1995; Flowers, 2004). For semi-arid and arid areas, this implies a broad use of poor quality (saline) irrigation water. Again, halophytes come into the spotlight as species capable of not only coping, but actually benefitting from saline irrigation.

The use of halophytes for saline agriculture has been the subject of numerous reviews over the last 10–15 years (Glenn et al., 1999; Flowers and Colmer, 2008; Ruan et al., 2010), so is only briefly summarized in the next section. Instead, this paper focuses on two other issues: (1) the key physiological mechanisms conferring salinity tolerance in halophytes that can be introduced into non-halophytic crop species to improve their performance under saline conditions; and (2) the specific genes that need to be targeted to achieve the above goal.

HALOPHYTES FOR SALINE AGRICULTURE: SUCCESS STORIES

Halophytes are defined as plants that naturally inhabit saline environments and benefit from having substantial amounts of salt in the growth media. Halophytes grow in a wide variety of saline habitats, from coastal regions, salt marshes and mudflats, to inland deserts, salt flats and steppes. They occur across a wide range of plant families, with the Chenopodiaceae being dominant (Flowers and Colmer, 2008). Halophytes have evolved a range of adaptations to tolerate seawater and higher concentrations of salts. These include adjustment of their internal water relations through ion compartmentation in cell vacuoles, the accumulation of compatible organic solutes, succulence, and salt-secreting glands and bladders (Flowers et al., 1986; Colmer et al., 2005; Flowers and Colmer, 2008; Shabala and Mackay, 2011).

Optimal halophyte growth is achieved at a concentration of around 50 mm NaCl for monocots, and between 100 and 200 mm for dicots (Glenn et al., 1999; Flowers and Colmer, 2008). Moreover, some halophyte species do not show significant yield reduction even when irrigated with seawater (e.g. Suaeda maritima; Greenway and Munns, 1980). This is well beyond the capability of any known conventional crop species, making halophytes ideal for ‘saline agriculture’. The latter term refers to the use of plants in soils affected by salinity to the benefit of farmers and the wider community (Barrett-Lennard and Setter, 2010; Riadh et al., 2010). The imperative to develop plants capable of growing in saline agricultural systems is growing and, in Australia alone, 26 salt-tolerant halophyte species have been identified as capable of producing products of value to agriculture (Barrett-Lennard, 2002).

This point is further illustrated by comparing the effects of salinity on growth and yield of the cereal species wheat (Triticum durum L.) and the grain-like crop species quinoa (Chenopodium quinoa Willd.) (Fig. 1). Wheat is one of the most, if not the most essential staple crop, with 220 Mha producing 704 t year⁻¹ (FAOSTAT 2011 data, http://faostat.fao.org). Wheat is classified as a salt-sensitive crop; its growth and yield are strongly suppressed by even moderate concentrations of NaCl in the growth media (Fig. 1). In a specific case, 150 mm NaCl treatment caused a >70 % reduction in shoot dry weight, and the grain yield of salt-treated plants was only 6 % of the control (Cuin et al., 2009). Quinoa is a highly nutritional seed crop from the Chenopodiacea family that originates from the Andean region of South America where it has been cultivated for at least 7000 years (Jacobsen, 2003). In addition to being gluten free, the grain has an outstanding composition of essential amino acids, is rich in vitamins (A, B2, E) and minerals, and it represents a valuable source of carbohydrates and essential fatty acids for human nutrition (Repo-Carrasco et al., 2003). Being a true halophyte, quinoa plants benefit from having sodium in the growth media, displaying optimal growth between 100 and 200 mm NaCl (Fig. 1). Moreover, some quinoa varieties can be grown with seawater salt concentrations (40 dS m⁻¹) (Jacobsen et al., 2003; Koyro and Eisa, 2008). The average quinoa grain yield in the USA is between 1·3 and 1·9 t ha⁻¹, and trials in Denmark demonstrated seed yields of 2–3 t ha⁻¹ (Jacobsen et al., 2010). This is comparable with wheat production. Similar to wheat, quinoa can be used in bread, noodles, pastry, salads and soups (Jacobsen, 2003). Given that maximum quinoa production is achieved at salinity levels where wheat production is commercially non-profitable, if possible at all (Fig. 1), quinoa can indeed be considered as an alternative cereal crop for saline agriculture.

Fig. 1.

The effect of salinity on growth and biomass accumulation of (A) halophyte (quinoa; Chenopodium quinoa ‘5206’) and (B) non-halophyte (wheat; Triticum durum ‘Towner’) species. Optimal quinoa growth is observed at NaCl concentrations between 100 and 200 mm, while wheat growth is highly suppressed by 150 mm NaCl. The photogrpah in (A) is reproduced from Hariadi et al. (2011), with permission from the Society for Experimental Botany. The photograph in (B) is courtesy of Dr Tracey Ann Cuin.

In addition to the above example of an immediate use of quinoa as an alternative cereal crop, the key areas of application of halophytes in saline agriculture are briefly outlined below.

Desalination

Halophytes are ideally suited for revegetation and remediation of salt-affected land (Flowers et al., 2010; Manousaki and Kalogerakis, 2011a, b). To give a few examples, Atriplex nummularia (old man saltbush) has been shown to achieve a biomass yield of 20–30 t ha⁻¹ year⁻¹ and accumulate between 20 and 40 % NaCl in its dry matter when irrigated with saline water (Watson and O'Leary, 1993; Ghnaya et al., 2005). Similarly, Suaeda fruticose (seablite) can remove >2·5 t of salt per hectare in a single harvest of the aerial parts of the plant each year (Chaudhri et al., 1964), while Suaeda salsa at a density 15 plants m⁻² could potentially remove between 3 and 4 t of Na⁺ per hectare, if the plants are harvested at the end of the growing season (Zhao, 1991). This ability to remove significant amounts of salt to the point that the soil can be returned to agricultural productivity makes halophytes highly promising as a phytodesalinization tool. It should be noted, however, that the ability of halophytes to do this will depend strongly on the actual salt concentrations in the soil (Barrett-Lennard, 2002) and may require a substantial amount of time. Also, halophyte performance declines dramatically under dryland conditions as a result of accumulation of high salt concentrations in the root zone (Norman et al., 2013). As such, the annual leaf yield of Atriplex species under non-irrigated conditions in a low rainfall zone in southern Australia (330–370 mm year⁻¹) ranged only between 0·4 and 0·7 t dry matter ha⁻¹ (Norman et al., 2008). Thus, the use of halophytes as a desalinization tool cannot be taken as granted and should be considered in the environmental context.

Phytoremediation

Over the last 200 years, industrialization in Europe and elsewhere has led to an enormous increase in production, use and release of traces of heavy metals into the environment (Flowers et al., 2010). Because tolerance to salt and heavy metals relies, at least in part, on common physiological mechanisms (Thomas et al., 1998), halophytes are widely advocated for phytoremediation purposes. This includes both phytostabilization and phytoextraction (Manousaki and Kalogerakis, 2011b). Phytostabilization refers to the situation when a metal-tolerant non-accumulator plant is capable of tolerating, but not translocating and accumulating, the metals in the aerial parts of the plant. Nerium oleander (an endemic plant of the Mediterranean region) is a good example of this strategy: the plant shows optimal growth and no visual toxicity symptoms at 2400 ppm Pb in the soil (Manousaki and Kalogerakis, 2011a). Such plants provide vegetation cover, stabilizing the soil and preventing metals from being mobilized or leached into groundwater. Phytoextraction is the removal of toxic metals from the soil by their accumulation in the above-ground parts of the plant, so they can be harvested and removed. Atriplex halymus and Tamarix smyrnensis have been identified as two highly promising species to deal with Pb- and Cd-contaminated soils (Lutts et al., 2004; Manousaki and Kalogerakis, 2009).

Fodder and energy crops

Halophytes could be grown on 130 × 10⁶ ha of potential arable land, and, as biomass crops, could directly sequester up to 0·7 Gt of carbon per annum (Glenn et al., 1992). Many of them have been evaluated for their productivity and developed as fodder crops (e.g. O'Leary et al., 1985; Malcolm, 1986; Glenn et al., 1998; Asad, 2002). Members of the Chenopodiaceae (e.g. various Atriplex species) have shown good yield potential and high consumptive water use when irrigated by industrial brines in arid climates in the Arizona district of the USA (Jordan et al., 2009; Soliz et al., 2011). A perennial grass Panicum turgidum produced >6 t ha⁻¹ year⁻¹ when grown in saline soil irrigated with brackish water with EC = 12 dS m⁻¹ (Khan et al. 2009). However, factors such as nutritive value and voluntary feed intake should also be considered before halophytes may compete with traditional fodder species (Norman et al., 2013). Halophytes could be also used as a new source of energy crops that will not compete with conventional agriculture for valuable resources of fertile soil and fresh water. Salt cedar (Tamarix spp.) can be used as a good example. This species can be grown under extreme desert conditions if irrigated with reclaimed sewage and brackish water, yielding between 26 and 52 t ha⁻¹ year⁻¹; a level not less than that obtained for common cash crops on arable land (Eshel et al., 2010). Growing such salt-tolerant energy crops on marginal agricultural land would help to counter concerns that the biofuel industry reduces the amount of land available for food production (Qadir et al., 2008).

These examples illustrate the efficacy of the immediate use of halophytes in salt-affected areas not suitable for traditional crops. In addition to this, halophytes could provide valuable information about the physiological traits that are crucial for handling salt stress, and also serve as a source of ‘tolerance genes’ to improve salinity tolerance in non-halophytes. Some of these are discussed in detail below.

TARGETING EXTERNAL SEQUESTRATION IN SALT BLADDERS

Found in about 50 % of all halophyte species (Flowers and Colmer, 2008), salt bladders are arguably the most remarkable anatomical feature of halophytes. Being much larger than epidermal cells, epidermal bladder cells (EBCs) represent a good possibility to sequester excessive Na⁺ away from metabolically important mesophyll cells. With the bladder's diameter being often approx. 10 times bigger than that of epidermal cells (Shabala and Mackay, 2011), each EBC could sequester 1000-fold more Na⁺ compared with epidermal cells. Having salt bladders is therefore likely to be highly advantageous for species growing under saline conditions. In addition to being the storage sites for excess Na⁺ and Cl⁻, salt bladders may also play an important role as a secondary epidermis to reduce water loss and prevent excessive UV damage when exposed to salinity (Adams et al., 1998; Shabala and Mackay, 2011).

Salt bladders are modified epidermal hairs and, as such, are classified as trichomes, along with glandular hairs, thorns and surface glands (Adams et al., 1998; Glover, 2000; Ishida et al., 2008). In halophytic grasses, they are represented by bicellular microhairs, which are present on the surfaces of leaves of most families, with the exception of the Pooideae (Amarasinghe and Watson, 1988; Marcum and Murdoch, 1992; Ramadan, 1998). Similar structures are also found in all cereal crops (Fig. 2). Importantly, their density increases with increased salinity levels (Ramadan and Flowers, 2004). However, in contrast to halophyte grasses, in cereals, these microhairs appear to not have a glandular function (Thomson et al., 1988), and their size is not big enough to sequester excess Na⁺ continuously. If one of these features could be modified, then an intriguing possibility to manipulate the salt tolerance in cereal crops would exist. How can this be achieved? Currently though, there is little understanding of the ionic and molecular mechanisms that mediate Na⁺ excretion through glands, even though modifying the number, size and shape of trichomes may be the most practical way to improve efficiency of Na⁺ sequestration in crop leaves.

Fig. 2.

Micrograph of trichomes (A, B) and salt bladders (C, D) on the abaxial surface of two glycophyte (arabidopsis and barley) and two halophyte (quinoa and Atriplex) species. Images were taken using a scanning electron microscope in environmental mode. Phototographs are courtesy of Dr Jayakumar Bose (TIA) and Dr Karsten Goemann (Central Science Laboratory), University of Tasmania.

Not much is known about the molecular mechanisms of epidermal cell patterning and salt bladder formation in halophytes, and all existing knowledge comes from arabidopsis. The arabidopsis trichome is a large single cell about 200–300 µm in length, typically with three branches, an elaborated cuticle and a suite of socket cells (Glover, 2000). Trichome formation is the result of an interaction between neighbouring epidermal cells that reinforces any initial differences in the levels of gene expression or cell cycle stage (Larkin et al., 1996; Glover, 2000). The number of branches correlates strongly with the ploidy level (normally between 32C and 64C, where C equals haploid DNA content per nucleus; Guimil and Dunand, 2007; Passardi et al., 2007), suggesting that endoreduplication plays a role in branch initiation. Trichomes are considered to grow by diffuse growth and not by tip growth (Schwab et al., 2003). In contrast to stomata, trichomes are often concentrated in the epidermal layers that overlie the vascular strands in leaves (Martin and Glover, 2007). A dramatic increase in the cell size is also a result of nuclear genome duplication without accompanying mitotic cytoplasmic division (Sugimoto-Shirasu and Roberts, 2003). However, it remains unclear how the final cell size is regulated (Guimil and Dunand, 2007).

Forward genetics experiments in arabidopsis have identified >40 genes involved in cell fate determination and trichome formation (Glover, 2000; Martin and Glover, 2007; Ishida et al., 2008; Pesch and Hulskamp, 2009; Tominaga-Wada et al., 2011). The key genes are briefly summarized in Table 1.

Table 1.

Key genes involved in cell fate determination and trichome formation

Gene	Mode of action	Role
Positive regulators
GL1, MYB23, MYB5	Encodes R2R3 type MYB transcription factors	Promotes trichome initiation; necessary for the initial enlargement of the nucleus and for the uncoupling of DNA synthesis from cytokinesis
TTG1	Encodes a WD40 repeat protein	Regulates trichome initiation and hairless cell differentiation
GL3, EGL3	Encodes basic helix–loop–helix (bHLH) transcription factors	Regulates trichome initiation and hairless cell differentiation
GL2	Encodes a homeodomain/leucine zipper transcription factor	Required for trichome development post-patterning
Negative regulators
CPC, TRY, ETC1, ETC2, ETC3	Encode R3-type MYB transcription factors	Required for lateral inhibition of neighbouring cells and correct spacing of trichomes

The patterning of the arabidopsis trichome is controlled through a cell interaction mechanism by a number of positive and negative regulators (Pesch and Hulskamp, 2009). Expression of TTG1, GL3 and GL1 occurs initially in all leaf epidermal cells (Martin and Glover, 2007). These form fate activator complexes that are required for trichome development post-patterning. The activator complex directly activates transcription of its inhibitors TRY, CPC, ETC2 and other R3 MYBs that undergo protein movement to neighbouring cells (Grebe, 2012). This inhibits the formation of the active TTG1–GL3–GL1 complex in neighbouring cells (Martin and Glover, 2007), preventing trichome development there. The activator complex also directly activates transcription of GL2, a downstream regulator of trichome fate and differentiation, as well as transcription of the mitosis inhibitor SIM required for endoreplication (Grebe, 2012). Depending on the distance that the negative regulator can travel, some epidermal cells will develop into trichomes, while others do not, so generating a pattern.

After branching, trichomes expand along the whole cell axis, with directionality of this expansion being dependent on the reorientation of microtubules and the actin cytoskeleton (Ishida et al., 2008). The trichome expansion itself is turgor mediated (Guimil and Dunand, 2007) and is presumably not different from that of any other expanding cells in either roots or shoots.

From a breeding point of view, there are four critical questions to be answered. (1) Which genes should be manipulated to increase the trichome density? (2) How can the sharp and pointy trichome structures currently present in cereals be converted (Fig. 2) into the round, ‘balloon-like’ structures that are more efficient for Na⁺ sequestration (such as found in halophytes; Fig. 2)? (3) How can the size/volume of trichomes be increased? (4) How can it be ensured that Na⁺ will actually be pumped into the trichomes?

As commented on above, the decision of whether an epidermal cell will develop into a trichome depends on the movement of R3 MYB inhibitors from currently differentiating cells into neighbouring cells. Understanding the factors controlling this transport (and suppressing this movement) may be key to answering the first question. Indeed, some halophytes develop highly dense and even multilayered patterns of salt bladders (illustrated for Atriplex in Fig. 2). This is in a sharp contrast to the far fewer and less developed trichomes currently found in cereals (Fig. 2).

Phytohormones are known to regulate trichome initiation and growth (Tominaga-Wada et al., 2011). Gibberellin (GA) controls both initiation and morphogenesis of arabidopsis trichomes (Chien and Sussex, 1996; Perazza et al., 1998), and mutants of maize that are insensitive to GA show delayed macrohair formation (Martin and Glover, 2007). Jasmonates are also involved in regulating trichome density by acting upstream of the MYB–bHLH–WD-40 complex (Kobayashi et al., 2010). In contrast, salicylic acid has a negative effect on trichome induction and reduces the effect of jasmonic acid (Traw and Bergelson, 2003). The role of cytokinins has also been documented (Greenboim-Wainberg et al., 2005). Given this multitude of hormonal factors, it is difficult to select one or a few target genes to manipulate trichome formation patterns.

Controlling trichome size may be slightly easier. To a large extent, size is determined by the number of endoreduplications, and it was shown that SIM and D-type cyclins play an integral role in controlling this process in arabidopsis (Churchman et al., 2006). Hence, there is a good chance that, by manipulating one or a few genes responsible for cyclin production, we can modify trichome volume in cereals. As for their size (spherical vs. narrow-like), the key may be in understanding the factors controlling the orientation of microtubules (Mathur and Chua, 2000). Again, hormonal control is very likely (Blume et al., 2012).

As for the last question, we know very little (if anything) about the physiological and molecular mechanisms of Na⁺ transport in trichomes. Most of the knowledge on this subject is derived from microscopy studies and lacks functional characterization. Both apoplastic and symplastic components have been suggested. Evidence for a symplastic pathway comes from the abundance of plasmodesmatal connections in some of these structures, while tissue-specific expression of some ion transporters implies that ions must be released from some of the cells (e.g. mesophyll cells) and then reabsorbed by epidermal or bladder cells (reviewed in Shabala and Mackay, 2011). The extent of leaf cutinization and the presence of cuticular pores may also play a pivotal role in water flow into trichome cells, and so affect their ion loading.

TARGETING INTERNAL Na⁺ SEQUESTRATION IN VACUOLES

Halophytes rely heavily on the use of inorganic ions (Na⁺, Cl⁻ and K⁺) to maintain shoot osmotic and turgor pressure under saline conditions (Flowers et al., 1977; Storey and Wyn Jones, 1979; Storey, 1995; Glenn et al., 1999), while glycophytes achieve this predominantly by increased de novo synthesis of compatible solutes. The three major inorganic ions, Na⁺, K⁺ and Cl⁻, account for 80–95 % of the cell sap osmotic pressure in both halophyte grasses and dicots (Glenn et al., 1999; Shabala et al., 2011), while in non-halophyte species such as wheat and barley, the contribution is typically between 50 and 70 % (Chen et al., 2007; Cuin et al., 2010). As osmolyte biosynthesis comes with a high carbon cost (between 50 and 70 mol of ATP are required to produce 1 mol of compatible solute; Raven, 1985; Shabala and Shabala, 2011), the associated yield penalties are significant; plants would have no sugars or ATP left for anything else. The strategy employed by halophytes is much more efficient. Indeed, they only need to use compatible solutes for osmotic adjustment in the cytosol, which comprises roughly 10 % of the cell volume. Metabolic Na⁺ toxicity is thought not to differ between halophytes and glycophytes (Flowers and Colmer, 2008), so Na⁺ has to be kept away from sensitive metabolic pathways, regardless of the salinity tolerance of the plant. Thus, all (or at least most) of the Na⁺ taken up for rapid osmotic adjustment must be efficiently sequestered in vacuoles.

The classical view is that vacuolar Na⁺ sequestration is achieved via tonoplast Na⁺/H⁺ antiporters (Blumwald and Poole, 1985; Barkla et al., 1995; Gaxiola et al., 1999; Apse and Blumwald, 2007), although the role of pinocytosis has also been advocated (Balnokin et al., 2007). Tonoplast Na⁺/H⁺ exchangers belong to the CPA family of cation/proton antiporters (Apse and Blumwald, 2007; Rodríguez-Rosales et al., 2008). At least six NHX isoforms have been found in arabidopsis, with their expression pattern being both tissue and stress specific (Rodríguez-Rosales et al., 2009). The operation of NHX exchangers has to be energized by vacuolar H⁺ pumps, and both tonoplast H⁺-ATPases (Ayala et al., 1996; Vera-Estrella et al., 1999; Wang et al., 2001) and PP-ases (Parks et al., 2002; Vera-Estrella et al., 2005; Guo et al., 2006; Krebs et al., 2010) are involved.

There are a large number of reports suggesting that the much greater ability of halophytes to sequester Na⁺ in their vacuoles is related to both the constitutive expression of tonoplast Na⁺/H⁺ antiporters and the stimulation of their activity under saline conditions (Barkla et al., 1995; Glenn et al., 1999). In contrast, in glycophytes, such antiporters must be activated by NaCl. Moreover, it appears that such activation occurs only in salt-tolerant glycophyte species, while in salt-sensitive plants, their expression levels are extremely low and not salt inducible (Apse et al., 1999; Zhang and Blumwald, 2001).

The idea of improving salinity tolerance in glycophytes by overexpressing tonoplast NHX Na⁺/H⁺ exchangers is rather attractive, and a substantial number of papers have been published on this (Apse et al., 1999; Zhang and Blumwald, 2001a, b). However, as commented on by some authors (e.g. Flowers, 2004), none of these attempts has delivered salt-tolerant cultivars to the farmers' fields. Why is this?

As mentioned above, the activity of Na⁺/H⁺ exchangers must be energized by the vacuolar H⁺ pumps. Thus, as the very least, overexpression of NHX exchangers in transgenic crops should be complemented by increased activity of tonoplast H⁺-ATPases or H⁺-PPases. This is rather problematic for at least two reasons. First, the available ATP pool is significantly reduced due to high ATP demand for de novo synthesis of compatible solutes. Also, a substantial part of the ATP pool is needed by the cell to restore (an otherwise depolarized) membrane potential, essential for K⁺ retention (Shabala and Cuin, 2008). These two processes compete with tonoplast H⁺-ATPase pumps for the available ATP, handicapping the activity of the tonoplast H⁺-ATPase. This leaves H⁺-PPase as the possible driver for energizing NHX activity. However, H⁺-PPase activity is known to be K⁺ dependent (Rea and Poole, 1993), and the cytosolic K⁺ pool is dramatically reduced under saline conditions. This reduction results from both depolarization- (Chen et al., 2007; Shabala and Cuin, 2008) and reactive oxygen species (ROS)-induced (Cuin and Shabala, 2007; Demidchik et al., 2010) activation of K⁺ efflux channels located at the plasma membrane. Thus, to make transgenic lines overexpressing Na⁺/H⁺ NHX exchangers fully functional under saline stress, some other conditions should be met: (1) plants should not heavily invest into the production of compatible solutes, so leaving a greater part of the ATP pool available for fuelling tonoplast H⁺-ATPase; and (2) plants should possess mechanisms for efficient K⁺ retention to enable tonoplast H⁺-PPases to function. As far as I know, none of these conditions has been considered or done in conjunction with any attempt to improve salinity tolerance via the overexpression of tonoplast Na⁺/H⁺ NHX exchangers in crops. Moreover, most modern crop varieties (e.g. barley or wheat) have been selected for their ability to reduce the accumulation of Na⁺ in the shoot. This automatically implies relying heavily on organic osmolytes for osmotic adjustment and, hence, a failure to meet the first condition.

There is one more essential condition that is always ignored by plant breeders. To confer salinity tolerance, toxic Na⁺ ions pumped into the vacuole must be prevented from leaking back into the cytosol (Pantoja et al., 1989; Shabala and Mackay, 2011). If this condition is not met, an energy-consuming futile Na⁺ cycling between cytosol and vacuole will occur, further depleting the limited ATP resources. Thus, strict control over the activity and expression of Na⁺-permeable tonoplast SV (slow vacuolar) and FV (fast vacuolar) channels, two major passive transport systems that may potentially contribute to Na⁺ back-leak into the cytosol (Hedrich and Neher, 1987; Pottosin et al., 1997, 2001; Brueggemann et al., 1999), is needed. In stark contrast to active transport systems, apart from a few early attempts (Maathuis et al., 1992), the properties of vacuolar ion channels in halophyte vacuoles remain essentially unexplored. Recently, we have addressed this issue by studying the properties of vacuolar SV and FV channels in mesophyll cells of Chenoponium quinoa (Bonales-Alatorre et al., 2013a, b). Both young and old leaves were used. The expectation was that old leaves that accumulate more Na⁺ and have far fewer epidermal salt bladders (Shabala et al., 2012) would rely heavily on a vacuolar sequestration for their optimal functioning under saline conditions, while young leaves would accumulate less Na⁺ and have the possibility to sequester it in numerous and well developed EBCs. It was found that at physiologically relevant tonoplast potentials, most FV and SV channels were functionally inactive in salt-grown old leaves, while their conductances in young leaves grown under similar conditions were several fold higher (Bonales-Alatorre et al., 2013a). This mirrored the amount of Na⁺ accumulated in the leaf mesophyll. A much smaller (if any) difference was found in plants grown under control conditions. Importantly, the number of open SV and FV channels differed between quinoa genotypes and showed a strong positive correlation with salinity tolerance (a 3- to 5-fold difference between sensitive and tolerant varieties) (Bonales-Alatorre et al., 2013b). Taken together, these results indicated that the ability of quinoa plants to control the conductivity of SV and FV tonoplast channels is essential for conferring salinity tolerance in this species. It was suggested that further progress in crop breeding for salinity tolerance could be achieved by delineating the mechanisms underlying this efficient tonoplast channel control in plants grown under saline conditions (Bonales-Alatorre et al., 2013a).

CONTROLLING STOMATAL APERTURE

Biomass accumulation is directly proportional to the amount of CO₂ assimilated by the plant over the growing season so is ultimately determined by the plant's ability to regulate the stomatal aperture. In glycophytes, the low soil water potential imposed by salinity causes a marked decline in stomatal conductance (g_s), reducing both net CO₂ assimilation (P_n) and transpiration rates (Munns 2002). The rationale behind this reduction is an attempt to minimize the water loss under the conditions of reduced water availability (‘physiological drought’) imposed by salinity. Given that halophytes grow better under saline conditions, they should exhibit less g_s reduction compared with glycophytes exposed to the same salinity treatments. Literature reports on this matter, however, are highly controversial. Experiments with isolated epidermal strips have shown that an increase in apoplastic NaCl concentrations results in a decreased stomatal aperture in Aster tripolium and Cochlearia anglica (Perera et al., 1994; Robinson, 1996). This is in sharp contrast to non-halophylic species whose stomata respond to elevations in apoplastic Na⁺ by increasing their aperture (Zeiger, 1983). In addition, some reports suggest that halophyte species may substitute K⁺ for Na⁺ in their stomata (reviewed by Shabala and Mackay, 2011).

It should also be noted that the decline in stomatal conductance in halophytes is not always accompanied by a loss of leaf water content (Redondo-Gomez et al., 2007), and the effects of NaCl and water stress on leaf gas exchange in some halophyte species are strikingly different (Ueda et al., 2003). In fact, the presence of NaCl in the growth media assists plants in alleviating the detrimental effects of osmotic stress caused by the presence of polyethylene glycol (PEG) (experiments with Atriplex halimus; Martinez et al., 2005). One possible explanation may be a role for Na⁺ in cell osmotic adjustment (discussed in a previous section). Alternatively, it could be related to the above-mentioned ability of some halophyte species to substitute K⁺ for Na⁺ in their stomata. However, the only available electrophysiological study that compares the properties of the ion channels mediating Na⁺ and K⁺ transport across the guard cell plasma membrane in a halophyte species (Aster tripolium) revealed a high selectivity of inward-rectifying KIR channels for K⁺ over Na⁺ (P_Na/P_K <0·005; Very et al., 1998). Thus, if halophytes use Na⁺ instead of K⁺ to increase turgor pressure rapidly in stomatal guard cells, its uptake is mediated by another transport system, not KIR. However, the molecular and electrophysiological identity of this transporter remains completely unknown and warrants further investigation.

Another controversy relates to the mechanisms mediating stomatal closure, the process mediated by the rapid K⁺ efflux via depolarization-activated outward-rectifying KOR channels (MacRobbie, 1998). By comparing the properties and regulation of KOR channels between a halophytic species A. tripolium and its non-halophyte relative Aster amelus, Very et al. (1998) found that outward K⁺ currents were inhibited by cytosolic Na⁺ in A. tripolium, but not in A. amelus. This suggests that once Na⁺ is accumulated in the cytosol, non-halophyte species lose their ability to close their stomata, while halophytes still possess some efficient control over stomatal movement. However, this hardly explains the fact that the extent of g_s decrease under saline conditions is much higher in non-halophytes, at any given NaCl concentration. Obviously, a much broader range of halophyte species should be involved in patch-clamp studies to reveal both the identity of Na⁺- and K⁺-permeable ion channels in guard cells and the modes of their control by internal and external factors, including salinity.

OPTIMIZING WATER USE EFFICIENCY BY TARGETING STOMATAL DENSITY

While the control of stomatal conductance is essential for maintaining the correct balance between CO₂ assimilation and transpirational water loss under saline conditions, a substantial amount of water evaporated from the leaf surface may bypass stomata and exit through the cuticle. Depending on the circumstance, the contribution of this non-stomatal component can be as high as 28 % of the total amount of water transpired through stomata (Boyer et al., 1997). Nonetheless, I am not aware of any breeding programme or research targeting genetic variability in cuticular transpiration as a physiological trait to improve salinity tolerance in crop species. Why is this so and what is the evidence that such an approach would deliver the desired outcomes?

It is traditionally assumed that as leaf expansion growth is reduced by salinity (Munns, 2002), cells become smaller in size, and this results in a larger number of cells per unit surface area (i.e. increased cell density). However, this seems not to be the case for halophytes. Salinity causes a marked (about 30 %) decrease in stomatal density in quinoa (Orsini et al., 2011; Shabala et al., 2012, 2013). A negative correlation between stomatal density and salt tolerance was reported for a related amaranth species (Omami et al., 2006). Decreases in stomatal density with increasing salinity have also been found in several highly salt-tolerant halophyte species such as Kochia prostrata (Karimi et al., 2005), Suaeda maritima (Flowers, 1985), Distichlis spicata (Kemp and Cunningham, 1981), Atriplex halimus and Medicago arborea (Boughalleb et al., 2009), and Aeloropus lagopodies and Lasiurus scindicus (Naz et al., 2010). A freshwater species of Spartina had high stomatal densities, while the salt marsh species had significantly lower stomatal densities (Maricle et al., 2009). It was argued that the observed reduction in stomatal density might represent a fundamental mechanism by which a plant may optimize water productivity under saline conditions (Adolf et al., 2012; Shabala et al., 2013). Indeed, cuticular transpiration is usually concentrated in the area surrounding stomata where there are more and larger cuticular pores (Marschner, 1995). While stomatal conductance is a dynamic process that can be rapidly regulated by ion fluxes into and out of guard cells, cuticular transpiration relies almost exclusively on the existing (passive) hydraulic permeability of the leaf surface, so cannot rapidly adjust to changing conditions. Thus, having fewer fully opened stomata will be more beneficial than having many partially opened ones, because in the latter case we will also be dealing with an increased number of cuticular pores that cannot be controlled. The qualitative model put forward in Shabala et al. (2012) suggests that a reduction in stomatal density will increase water use efficiency (WUE) under saline conditions.

As mentioned earlier, neither the cuticular transpiration component nor the stomatal density have been used as physiological indices in marker-assisted selection-based breeding programmes. However, there is some circumstantial evidence suggesting that a reduction in stomatal density is a highly conserved response employed not only by halophytes, but also by crop species as a part of their adaptation to salinity. Depicted in Fig. 3 are some data correlating the changes in stomatal density under saline condition with the observed yield reduction in two species: Chenopodium quinoa (filled symbols) and barley (open symbols). The three salt-sensitive genotypes that did not survive 320 mm salinity stress all showed a substantial (>20 %) increase in stomatal density. Salt-tolerant Numar and ZUG293 varieties (Chen et al., 2007) either did not change (ZUG293), or even decreased (Numar) their stomata density under these conditions. All four quinoa varieties depicted in the figure maintained a relatively high yield when exposed to salinity and displayed a 20–40 % reduction in stomatal density. Overall, a very strong correlation (R² = 0·83) was found between the observed decline in stomatal density and the relative plant yield. This suggests that reducing stomatal density is indeed playing a substantial role in plant adaptive responses to salinity.

Fig. 3.

The decrease in stomatal density under saline conditions correlates positively with relative plant yield. Relative changes (% control) in stomatal density were measured in barley and quinoa (as indictaed in the key) genotypes contrasting in salinity tolerance. These are plotted against relative shoot biomass (% control; fresh weight basis). Numbers next to the letter Q indicate specific quinoa varieties, as in Shabala et al. (2013).

How can stomatal density be targeted in breeding programmes? The answer depends on whether the observed reduction is related to the changes in developmental patterns under high salinity conditions or whether it is merely a consequence of the increased leaf succulence. Stomatal patterning occurs via lineage-based mechanisms (Martin and Glover, 2007). The process starts with an unequal cell division of a protodermal cell whose fate is determined either by receptor-like kinase (RLK) activity or by the positional signals from the underlying mesophyll. The products of this division are a subsidiary cell and a meristemoid. The process is repeated once again and the meristemoid becomes round and a mother guard cell. This then divides equally to form two guard cells. Similar to trichomes, the entire process is under heavy transcriptional control (Martin and Glover, 2007), with WER, TTG1 and GL2 genes also involved in control of stomatal development. It is widely acknowledged that stomata patterning is controlled by the signals emanating from mesophyll cells (Serna and Fenol, 2000) and that hormonal signals play an important role (Serna and Fenoll, 1997). Specific details of this process, however, remain elusive.

Similar to our knowledge of trichome development, all available data related to stomatal patterning comes from arabidopsis. Moreover, while the mutants with an excess number of clustered stomata [e.g. the too many mouths (tmm) mutant; Serna, 2009] are available, no arabidopsis mutants with a reduced number of stomata have been described. Thus, much more work is needed before we will be able to target stomatal density reduction in breeding programmes.

TARGETING XYLEM ION LOADING

The ability of halophytes to use Na⁺ as a cheap osmoticum to maintain cell turgor implies that substantial amounts of Na⁺ are delivered to the shoot via the transpiration stream. Concentrations of around 50 mm can be taken as a reasonable estimate for the Na⁺ xylem sap content in halophyte species (reviewed by Shabala and Mackay, 2011). Having such high xylem Na⁺ concentrations could also be essential for the formation of water potential gradients to drive water transport to the shoot (Balnokin et al., 2005; Shabala et al., 2013). A strong negative correlation (r = –0·73) has been found between xylem Na⁺ concentration and salinity tolerance among a range of quinoa genotypes grown under high salinity (400 mm NaCl) conditions for 8 weeks (Shabala et al., 2013). Interestingly, the xylem sap Na⁺ concentration was mostly below 10 mm, i.e. several fold lower than typically reported in the literature (Rozema et al., 1981; Clipson and Flowers, 1987). The most logical explanation for this is that the control of xylem loading of Na⁺ is a highly dynamic process and the xylem Na⁺ content changes dramatically as the stress progresses. The ‘ideal’ scenario for a plant would be to send the amount of required Na⁺ to the shoot quickly to achieve full osmotic adjustment rapidly and maintain the normal growth rate (hence, no yield penalties). Once this is achieved, it would be better for a plant to reduce the rate of xylem Na⁺ loading to the absolute minimum required for driving cell turgor in newly growing tissues. Thus, it is the timing aspect of the regulation of xylem Na⁺ loading that appears to be critical for salinity tolerance.

The supporting evidence for this hypothetical scenario has been obtained recently by comparing the kinetics of xylem Na⁺ loading in pea and barley, two glycophyte species contrasting in their salinity tolerance. Salt-tolerant barley plants rapidly loaded Na⁺ in the xylem, with a 2-fold elevation in xylem Na⁺ measured within 6 h of acute NaCl stress onset (Bose et al. 2013). This increase was only transient, and xylem Na⁺ remained at a steady (between 10 and 20 mm) level for the entire duration of the experiment (4 weeks). In contrast, pea plants restricted xylem Na⁺ loading during the first few days of treatment, but failed to prevent its elevation over the longer term (Bose et al., 2013). Being a salt-sensitive species, pea plants follow the exclusion strategy, so do not use much Na⁺ for osmotic adjustment in leaves. Given the very limited volume of the shoot apoplast, the slow but steady delivery of Na⁺ to the shoot, combined with its active exclusion from the mesophyll, results in a massive Na⁺ build-up in the apoplast, with xylem Na⁺ exceeding 500 mm after 4 weeks of salt stress (Bose et al., 2013). Such extreme concentrations would impose a severe osmotic stress on the leaf mesophyll, as well as a massive plasma membrane depolarization and a consequent K⁺ leak from the mesophyll (Shabala et al., 2000). This disturbance to cytosolic K⁺ homeostasis will result in activation of caspase-like proteases and endonucleases (Hughes and Cydlowski, 1999; Shabala et al., 2007; Demidchik et al., 2010), resulting in cell death. Thus, it appears that the time-dependent regulation of the rate of xylem Na⁺ loading is absolutely essential for plant salinity tolerance. Can this trait be targeted in breeding programmes?

The honest answer would be ‘not yet’. The major limiting factor is the lack of information both about the nature of the signalling factors that are behind this time-dependent regulation and about the actual membrane transporters that are involved in the process of xylem loading. Several options could be considered.

First, xylem Na⁺ loading may be a channel-mediated process. Na⁺-permeable non-selective outward-rectifying channels have been reported to be present at the xylem–parenchyma interface in non-halophyte species (Wegner and Raschke, 1994; Wegner and De Boer, 1997). Their activity is regulated by numerous factors such as apoplastic pH (Lacombe et al., 2000), polyamines (Zhao et al., 2007) and abscisic acid (Pilot et al., 2003). All these factors are known to be modulated under saline conditions. It remains to be shown whether similar channels are also present in halophyte species, how they are regulated and if they can be responsible for the observed inactivation of the xylem Na⁺ loading process as the stress progresses.

Secondly, xylem Na⁺ loading may be a thermodynamically active process that requires energy to pump Na⁺ into the xylem (De Boer and Volkov, 2003; Lun'kov et al., 2005; Shabala and Mackay, 2011). Indeed, given the highly negative membrane potential values for halophyte parenchyma cells (e.g. –130 to –140 mV in Atriplex; Anderson et al., 1977), and assuming 30 mm xylem Na⁺ concentration, cytosolic Na⁺ concentrations in parenchyma cells need to reach molar levels to make passive (channel-mediated) loading possible. Most authors doubt that such high cytosolic Na⁺ can be reached (reviewed in Munns and Tester, 2008). Most probably, both active and passive transport systems are involved, but their respective roles may differ, depending on the length of time since salinity onset. A hypothetical model depicting this process is shown in Fig. 4. The onset of salinity stress will cause a sequential depolarization in the root tissues (Wegner et al., 2011), accompanied by the progressive accumulation of Na⁺ in parenchyma cells. At the same time, xylem Na⁺ concentrations remain low. As stress progresses, parenchyma cells become more depolarized (the result of more Na⁺ entering them), and their cytosolic Na⁺ concentration becomes higher. Together, these two factors make channel-mediated xylem Na⁺ loading feasible. This Na⁺ is then sent to the shoot by the transpiration stream to be used for rapid osmotic adjustment.

Fig. 4.

A hypothetical model depicting the kinetics of xylem loading and the mechanisms involved. At initial (control) conditions (A), cytosolic Na⁺ concentrations in xylem parenchyma cells are low, and the membrane potential is too negative to allow passive xylem Na⁺ loading via non-selective cation channels (NORCs). Active Na⁺ transporters are not constitutively expressed so do not contribute to the process. The onset of salinity stress results in a substantial depolarization of root cells (Wegner et al., 2011), accompanied by the progressive accumulation of Na⁺ in the parenchyma cell cytosol, while the xylem Na⁺ concentration remains low. This enables channel-mediated xylem Na⁺ loading (B). As Na⁺ is loaded in the xylem, the xylem Na⁺ concentration becomes higher, and parenchyma cells also become repolarized, making further passive loading impossible. Further xylem Na⁺ loading may be mediated by one of two active transport systems: either SOS1 (Na⁺/H⁺ exchanger) or CCC (2Cl⁻:Na⁺:K⁺ symporter) (C). Depending on whether these transporters are constitutively expressed or are inducible by salinity, the kinetics of xylem Na⁺ will differ significantly.

As Na⁺ is loaded in the xylem, xylem Na⁺ becomes higher, and parenchyma cells become repolarized, making further passive loading impossible. This explains the drop in the xylem loading process observed in our experiments with barley (Bose et al., 2013). From here, plants have to switch to thermodynamically active xylem loading. Two possible candidates are proposed. One is a SOS1 Na⁺/H⁺ exchanger. In glycophytes, SOS1 is preferentially expressed at the xylem symplast boundary of roots (Shi et al., 2002), and SOS1 gene homologues have been cloned from many halophytes species (Maughan et al., 2009; Cosentino et al., 2010; Guo et al., 2012; Yadav et al., 2012). Importantly, SOS1 activity was reported to be inducible by salinity, in both glycophytes (Shi et al., 2002) and halophytes (Oh et al., 2010; Cosentino et al., 2010).

Another possible candidate for the thermodynamically active xylem Na⁺ loading may be a cation–Cl (CCC) co-transporter. In animals, the CCC family is essential for adequate homeostasis of the most abundant electrolytes, K⁺, Na⁺ and Cl⁻, playing key roles in cell ionic and osmotic regulation (reviewed by Delpire and Mount, 2002). CCC proteins are secondary active transporters that mediate the movement of Cl⁻ tightly coupled to that of K⁺ and/or Na⁺ across the plasmalemma (Haas, 1989). Such CCC transporters have also been found in plants (Harling et al., 1997), and strong CCC–GUS (β-glucuronidase) expression was found in xylem parenchyma cells (Colmenero-Flores et al., 2007). Due to the highly negative membrane potential (see above), Cl⁻ movement from the xylem parenchyma into the xylem will be thermodynamically passive, so can be used as a driving force to move Na⁺ against the electrochemical potential and to load it into the xylem. Importantly, this loading can be electrogenically passive (assuming 2Cl⁻:Na⁺:K⁺ stoichiometry) and also enable a concurrent loading of K⁺ into the xylem, along with Na⁺. Such a phenomenon of a concurrent increase in xylem Na⁺ and K⁺ was reported in our previous papers dealing with both halophytes (quinoa; Shabala et al., 2013) and glycophytes (barley; Shabala et al., 2010), and interpreted in the context of the essentiality of the maintenance of a xylem Na/K ratio that enables optimal osmotic adjustment in the shoot. It remains to be proven in direct experiments whether such CCC transporters are expressed in stellar tissues of halophytes, what their role is in mediating xylem Na⁺ and K⁺ loading, and how their activity is regulated as a function of time and external conditions. Once these answers are available, controlling xylem Na⁺ loading may become feasible, in practical terms, via molecular tools.

CONCLUSIONS

Halophytes are naturally ‘salt-loving’ plants that outcompete any existing traditional crop when grown in hostile saline environments. Importantly, there is nothing unique that halophytes possess that is not found in crop species. Instead, halophytes are doing everything ‘a bit better’ and possess a set of highly complementary and well-orchestrated mechanisms in place to deal with salinity stress. However, the rapid progress in molecular biology and development of various ‘omics’ has somewhat overshadowed the importance of in-depth physiological studies on regulation and co-ordination of the above mechanisms. As a result, despite the general acceptance of the lack of a ‘silver bullet’ to improve the salinity trait, hundreds of papers are submitted describing attempts to manipulate one specific gene in the hope of tackling salinity stress tolerance. In light of the above, this approach is erroneous by default. Also, for many years, efforts of breeders have been predominantly aimed at the trait involving Na⁺ exclusion from uptake (Munns and Tester, 2008). This is not what halophytes are doing. At the same time, many key features of halophytes that are highlighted in this review, such as using trichomes for external Na⁺ sequestration, reducing stomatal density or regulating efficiency and timing of xylem Na⁺ loading, have never been manipulated by breeders. This opens up novel and previously unexplored possibilities for improving salinity tolerance in crops.