Keywords

1 Introduction

The phytohormone abscisic acid (ABA), a sesquiterpenoid with a 15-carbon ring, has a critical role in a large range of physiological functions in a variety of plant tissues over the entire life span of plants. During seed development, ABA prevents vivipary and contributes to dormancy and desiccation tolerance (Robertson 1955; Meurs et al. 1992; Schwartz et al. 1997; Gubler et al. 2005). ABA also inhibits seedling development (Lopez-Molina et al. 2001). In Arabidopsis the hormone influences root architecture by preserving primary root growth and inhibiting the development of lateral roots at the meristem activation checkpoint (Zhang et al. 2010). In the shoot, ABA’s major role is to regulate stomatal aperture and thereby modulate the hydraulic conductance of plants (Kim et al. 2010; Munemasa et al. 2015; Merilo et al. 2018). ABA also plays a role in reproductive organ development, with possible functions in male sterility and fruit development (Peng et al. 2006; Oliver et al. 2007).

Furthermore, ABA is a key mediator of physiological responses to multiple environmental stresses, such as drought, cold, and pathogen attack. In general, the earliest adaptive responses of plants to various stresses are induced by an increase in transcript levels of major ABA biosynthetic genes, which in turn leads to accumulation of cellular ABA and triggers the expression of multiple stress-responsive genes (Inuchi et al. 2001; Tan et al. 2003). These molecular processes have been widely studied in the model plant Arabidopsis thaliana (L.) Heynh., where most of the ABA signaling mechanisms have been elucidated.

Abiotic stress conditions, such as water stress, increase ABA levels and thus initiate signaling pathways that induce multiple cascades of molecular and cellular responses, including, in the case of water stress, expression of stress-related genes, cellular ROS accumulation, and stomatal closure. Stomatal closure and ROS accumulation also serve as a mechanism for pathogen defense (Melotto et al. 2006; Desikan et al. 2008), thereby providing a platform for crosstalk between biotic and abiotic stress responses. However, ABA can also directly activate protein kinases, as it does in the activation of SLAC, a guard cell anion channel, leading to stomatal closure (Raghavendra et al. 2010).

For the past several decades, ABA studies have focused on the hormone’s biosynthesis and signaling pathways. Almost every ABA biosynthetic enzyme was identified quite some time ago (Nambara and Marion-Poll 2005). However, ABA’s receptors (Park et al. 2009; Ma et al. 2009) and transporters (references below) were identified only recently. One of the main reasons for this delay is that, unlike the ABA biosynthetic enzymes, the protein families involved in perception and transport are highly redundant.

Over ten different proteins of the PYR/PYL/RCAR (pyrabactin resistance1/PYR1-like/regulatory components of ABA receptor) family act as ABA receptors in different tissues and at various developmental stages (Raghavendra et al. 2010; Gonzalez-Guzman et al. 2012). The ABA transporters that have been identified to date belong to several different protein families – the ABC (ATP-binding cassette) superfamily, the NPF (nitrate transporter1/peptide transporter family) family, and the DTX/MATE (detoxification efflux carriers/multidrug and toxic compound extrusion) family. ABA transporter mutants often exhibit no apparent phenotypes under normal growth conditions, but show distinct phenotypes under a specific stress condition (e.g., pathogen exposure or drought) or during a specific developmental stage. Thus, ABA transporters may have specific functions and may act in concert with other ABA transporters but also be partially redundant with them.

Why did plants evolve to have such diverse ABA transporters belonging to so many different protein families? Are the physiological functions of these ABA transporters highly redundant or not? The answers to these questions hinge on further studies of mutants with knockouts of multiple ABA transporters. Here, we review the history of ABA transporter studies, the properties of the protein families involved in ABA transport, and the functions of ABA transporters in biological processes. We also propose some hypotheses to explain why plants have evolved so many ABA transporters.

2 Where Does ABA Come from, and Where Is It Going?

In general, ABA meets the definition of a hormone, in that it is synthesized in one place and acts at other places within the plant. However, ABA is synthesized in several places, not just one (Nambara and Marion-Poll 2005), and, as will be described below, there is at least one cell type that relies on cell intrinsic ABA biosynthesis. Furthermore, ABA, like other plant hormones, is often present in conjugated forms. Some of these conjugates can be hydrolyzed under specific conditions (Lee et al. 2006; Xu et al. 2012), releasing active ABA and triggering a signaling cascade within the cell.

Studies of ABA transport and signaling historically focused on how ABA reaches guard cells and induces stomatal closure. Early findings suggested that roots, leaves, and even guard cells can synthesize ABA (Zeevaart and Creelman 1988). Initial attempts to identify the ABA receptor suggested that the receptor was located in the plasma membrane. This implicated that ABA would be perceived at the apoplastic face of the guard cells and had not to be transported into guard cells (Zeevaart and Creelman 1988). Subsequent work revealed that the ABA receptor is a member of the PYR protein family (Park et al. 2009; Ma et al. 2009). These receptors are localized in the cytosol. Therefore this discovery changed the view where ABA is perceived and pointed out that ABA has to cross the plasma membrane to reach its receptor.

As water stress is initially perceived in the root, the ABA required for stomatal closure was hypothesized to be produced in roots and translocated into shoots before being delivered to the guard cells (Zeevaart and Creelman 1988; Jackson 1993). To test this hypothesis, Blackman and Davies (1985) performed split root experiments with maize (Zea mays L.). One part of the root system was well watered, while the other was subjected to a reduced water potential. This experimental design allows the shoot to remain unstressed while half of the root is exposed to water stress. The authors reported that plants subjected to this treatment had smaller stomatal apertures than control plants with both parts of their root systems well-watered. However, the authors could not detect any difference in ABA concentrations in the shoot between the treatment and control plants. Further experiments led to the hypothesis that a continuous supply of cytokinins might be the factor allowing maximal stomatal opening.

Other evidence challenges the hypothesis that root-to-shoot transport of ABA or cytokinins is required to control stomata opening in response to drought. For example, such a mechanism cannot account for the fast stomatal reaction observed in tall trees, and an alternative mechanism involving a hydraulic signal has been postulated several times (Christmann et al. 2007). At least for the plants investigated so far, there is no evidence that a chemical signal has to be transported from the root to the shoot to induce stomatal closure. Evidence to the contrary is mainly provided by grafting experiments. For example, Holbrook et al. (2002) grafted tomato (Solanum lycopersicum L.) shoots able to produce ABA onto mutant roots unable to do so and showed that stomatal conductance corresponded to the shoot genotype. This result was further confirmed by grafting experiments using split roots or a pressure chamber.

In a detailed study, Christmann et al. (2007) took advantage of a series of Arabidopsis mutants to gain insight into the question of the source of ABA involved in water stress responses in the shoot. These authors also confirmed that ABA is not transported from the root to the shoot; instead, shoot-derived ABA is the signal leading to stomatal closure. They also showed that a sudden change in root water availability results in a rapid change in shoot turgor pressure, which induces rapid synthesis of ABA. These experiments support the hypothesis that a hydraulic signal is the primary cause for the increase in ABA content and the resulting stomatal closure in the shoot.

Stomata close in response not only to water shortage in the soil but also to changes in atmospheric humidity. As mentioned above, guard cell intrinsic ABA synthesis has been postulated for a long time. More recently, Bauer et al. (2013) confirmed this hypothesis and showed that guard cell intrinsic ABA synthesis is important for protecting plants against changes in atmospheric water potential. Although wild-type Arabidopsis plants did not wilt when exposed to dry air, the ABA biosynthesis mutant aba3-1 did. Rescuing ABA biosynthesis specifically in guard cells of this mutant restored the capacity to close stomata in dry air and the plants did not wilt.

A recent, more comprehensive study compared the corresponding mutants of Arabidopsis, pea (Pisum sativum), and tomato and also examined the relative importance of guard cell autonomous and shoot-dependent processes (Merilo et al. 2018). Besides highlighting several important regulatory mechanisms, the authors showed that the pools of ABA synthesized by guard cells and by phloem companion cells are redundant and that their relative importance may be plant specific. Therefore, under certain circumstances, transport of ABA from the leaf tissue may also function in stomatal closure in response to dry air.

Less information is available for other cases that may involve ABA transport. This may be either because the sites of ABA synthesis and action are too close together or because classical tools have not been able to determine whether ABA must be transported or not. One such case concerns seeds and seed germination. Most plants, with the exception of those growing in tropical climates, tightly control seed germination to avoid germination under unfavorable conditions. The mechanisms regulating this process have been studied extensively, and it is now generally accepted that ABA, which represses germination, and gibberellins, which promote germination, act antagonistically to regulate the process (Holdsworth et al. 2008). Dormant seeds typically contain high ABA concentrations, while seeds with impaired ABA production germinate precociously (Robertson 1955). Seed germination is a complex process that varies at least slightly among plants. In Arabidopsis, the embryo is surrounded by an active endosperm and the testa, a dead outer layer of maternal origin. In Arabidopsis, the endosperm is responsible for most or all of the germination-repressive activity (Bethke et al. 2007). Using a seed coat bedding assay, Lee et al. (2010) demonstrated that ABA is released from the endosperm of dormant seeds and is the factor inhibiting embryo development. Hence, export of ABA from the endosperm and uptake of this hormone into the embryo must occur during seed dormancy.

Transport of ABA is likely or may occur in other ABA-dependent processes, but the routes of ABA movement remain largely unknown. For example, under water stress, more suberin is also deposited in the endodermis in an ABA-dependent manner, and lateral root primordia are repressed, but it is not known whether the ABA that regulates these processes is de novo synthesized in situ or transported from elsewhere (Deak and Malamy 2005). ABA also decreases the number of hypodermal passage cells, non-suberized cells of the exodermis (Liu et al. 2019). Finally, in the development of nitrogen-fixing nodules in the Fabaceae, ABA regulates nodule production and probably also bacteroid number (Ding et al. 2008). An ABA transporter involved in this process has been recently identified; however, we still do not have much insight into the source of the ABA delivered to the nodule meristem.

3 Which Proteins Are Involved in ABA Transport?

ABA transport is catalyzed by a large number of transporters belonging to several families (Kuromori et al. 2018). The first class of transporters implicated in ABA transport was the ABC protein family. Subsequently it was demonstrated that some members of the NPF and MATE families are also able to transport ABA. Here, we present some characteristics of the transporter families involved in ABA transport. We have only limited knowledge about the AWPM-19 (ABA-induced wheat plasma membrane polypeptide-19) family; therefore we do not discuss this membrane protein family here.

3.1 The ABC Superfamily

ABC transporters are among the oldest transport proteins and are present in all living organisms from bacteria to humans. These transporters are classified as pumps, since they are directly activated by ATP hydrolysis (Kang et al. 2011; Hwang et al. 2016). The Arabidopsis genome encodes 131 ABC transporters, and a similar or higher number has been reported for all plants analyzed so far. This number is far higher than that for humans (49) and most other organisms.

A functional ABC transporter contains two cytosolic (nucleotide binding fold, NBF) and two transmembrane (transmembrane domain, TMD) domains. These domains can be part of one large protein, a so-called full-size ABC transporter, or part of a half-size transporter, which contains only one NBF and TMD, or exist as single proteins corresponding to NBFs and TMDs (bacterial-like). In all cases, however, the proteins have to be assembled to contain two NBFs and two TMDs to act as a functional transporter.

All ABC transporters contain an ABC signature flanked by Walker A and Walker B domains, which are responsible for ATP binding on each NBF. In plants, ABC transporters have been shown to transport a plethora of compounds, such as lipids, secondary compounds, heavy metals, and many hormones. In most cases, a single ABC transporter is able to transport several to many chemically unrelated compounds. All the ABC transporters that have been described as being involved in ABA transport are full- and half-size ABC proteins of the ABCG class. This class has a so-called reverse arrangement, in which the first TMD is located at the N-terminus of the protein, followed by an NBD. So far, AtABCG25, AtABCG30, AtABCG31, and AtABCG40 in Arabidopsis, Lr34 in wheat (Triticum aestivum L.), and MtABCG20 in Medicago have been shown to transport ABA (Table 1).

Table 1 List and properties of ABA transporters described in the text
Full size table

3.2 NPFs (NRT1/PTR Family)

The name NPF (nitrate peptide family) indicates that proteins of this family were originally identified as nitrate transporters (NRTs) and peptide transporters (PTRs) (Longo et al. 2018). The first member of this protein family was isolated in Arabidopsis in a screen for resistance to chlorate, a toxic nitrate analogue. This transporter was subsequently shown to be a nitrate uptake transporter (Tsay et al. 1993). Most NPFs are proton/substrate symporters, but some are facilitators, bidirectional transporters, or proton-coupled potassium antiporters.

This class of transporters has 12 transmembrane α-helices and is part of the large major facilitator superfamily (Léran et al. 2014; Longo et al. 2018). The N-termini and C-termini are both located in the cytosol. Like ABC transporters, NPFs have been identified in all living organisms, from bacteria to fungi to humans, and, also like ABC transporters, the number of NPFs is much higher in plants than in most other organisms. Plant NPFs have been divided into eight subclasses (Léran et al. 2014), but assignment of an NPF to a particular subclass does not necessarily predict its substrate(s). Like ABC transporters, NPFs transport a plethora of substrates such as nitrate, peptides, glucosinolates, auxin, jasmonic acid, and ABA. The ABA importer AIT (ABA-importing transporters) was the first and so far only member characterized in detail (Kanno et al. 2012). However, ABA transport activity has been demonstrated or postulated for an additional 11 NPFs (Chiba et al. 2015).

3.3 MATEs

MATE transporters were originally identified as multidrug and toxic compound extrusion proteins conferring resistance against drugs in bacteria (Morita et al. 1998). In Arabidopsis, these transporters are also called DTX proteins. As in the case of ABC transporters and NPFs, they have been identified in most living cells, and plants contain a far higher number of these transporters compared to other organisms. For Arabidopsis, 56 MATEs have been reported, whereas humans have only 2 (Takanashi et al. 2014; Upadhyay et al. 2019).

Most MATEs contain 12 α-helices, with 2 long extensions at the N-terminus and C-terminus, but there are a few exceptions where more or fewer α-helices have been predicted. Furthermore, most MATEs act as proton antiporters, extruding compounds from the cytosol into either the apoplast or the vacuole. MATEs transport a wide variety of compounds, such as toxic abiotic compounds, a large range of plant secondary metabolites (e.g., flavonoids and alkaloids), citrate, salicylic acid, auxin, and ABA. It is also likely that a given MATE can transport multiple chemically unrelated compounds. Recently it was shown that two MATEs act as chloride channels and have a strong impact on stomatal movement (Zhang et al. 2017). AtDTX50 is so far the sole ABA transporter identified in this family (Zhang et al. 2014).

4 What Is the Role of ABA Transporters in Plants?

4.1 Modulation of Hydraulic Conductance in Sessile Plants

Our views on ABA transport have been changed by the results showing that during water stress ABA is not transported from the root to the shoot and subsequently to guard cells but rather is redistributed within a leaf (Christmann et al. 2007). However, even in this case, ABA has to move from the phloem parenchyma cells, the site of biosynthesis, to the guard cells. Since guard cells are not connected to mesophyll or phloem parenchyma cells via plasmodesmata, ABA must cross membranes to make this journey. Older work postulated that an ABA importer on the guard cells is not required, because apoplastic pH was reported to be 5.5 to 6.0. At this pH, a substantial proportion of the ABA (pKa 4.75) is protonated and can diffuse across membranes. However, it has now been shown that under water stress conditions the pH of the apoplastic space increases to 7.2, and consequently the proportion of protonated, freely diffusible ABA decreases. Under these conditions, only a small fraction of ABA can enter guard cells by simple diffusion, implying that guard cells need a transporter to import ABA from the apoplast to the cytosol. By contrast, for ABA export from the cytosol to the apoplast, a transporter has always been assumed to exist, since at a cytosolic pH of 7.5 only negligible amounts of uncharged, freely diffusible ABA is present.

Two plasma membrane-localized transporters of the ABC protein family were the first ABA transporters to be identified. One, AtABCG25, catalyzes ABA export from the cytosol to the apoplast (Kuromori et al. 2010); the other, AtABCG40 (Kang et al. 2010), catalyzes the uptake of ABA from the apoplast into the cytosol. Both transporters are expressed in roots and shoots. While in shoots AtABCG25 is localized predominantly to the vasculature of leaves, AtABCG40 is mainly expressed in guard cells. ABA transport by these proteins has been studied using membrane vesicles isolated from insect cells overexpressing AtABCG25 and yeast and BY2 cells expressing AtABCG40. Transport was strictly ATP-dependent, and both transporters exhibited high and specific affinity for ABA, with KMs of 0.26 and 1 μM for AtABCG25 and AtABCG40, respectively.

Thermal imaging of Arabidopsis plants overexpressing AtABCG25 revealed that these plants were warmer than the wild type, in line with increased ABA release and decreased stomatal aperture. In the case of AtABCG40, thermal imaging revealed that atabcg40 mutant plants were cooler than the corresponding wild-type plants under osmotic stress conditions or when exogenous ABA was supplied. This result indicated that the stomata of the mutant could not close efficiently in response to elevated ABA. Indeed, measurement of stomatal apertures revealed that atabcg40 mutant plants were less sensitive to ABA-induced closure. Transformation of the mutant plants with AtABCG40 completely complemented the stomatal defects. These results demonstrate that AtABCG25 acts as an ABA exporter and AtABCG40 as an ABA importer. However, the phenotypes of atabcg40 mutant plants were quite mild and not nearly as strong as those described for ABA biosynthesis mutants. This raised the question of whether diffusion could play a role or additional ABA transporters are present.

To identify additional ABA transporters, Kanno et al. (2012) devised an elegant screening method. The authors took advantage of the fact that some of the PYR/PYL/RCAR ABA receptors interact with a group of PP2C-type protein phosphatases in the presence of ABA. Using a modified yeast two-hybrid system, they transformed a yeast strain with these two proteins, ABA receptor, PYR1 and ABI1 or HAB1, PP2C-type phosphatase (Park et al. 2009; Ma et al. 2009). In the presence of 0.1 μM ABA, there was no interaction between the partners because the ABA could not enter the yeast cells. The rationale of the screen was that a transporter would allow efficient uptake of ABA into the yeast cells, leading to an interaction between the receptor and the phosphatase. A screen of two cDNA libraries led to the identification of four members of the NRT1/PTR (NPF) family that facilitated interaction between the partners. One of these candidates had previously been characterized as the low-affinity nitrate transporter NRT1.2 (Huang et al. 1999) and exhibited the most pronounced ABA transport activity when expressed in yeast and insect vesicles. This transporter, also called AIT1 by the authors, was characterized in more detail. The protein is localized in the plasma membrane and catalyzes the uptake of ABA with a KM of 5 μM.

Interestingly, although this transporter is mainly localized to the vasculature of leaves and inflorescences and not in guard cells, thermal imaging of ait1 mutants showed that inflorescence stems but not leaves were cooler than in the wild type. This result indicates that AIT1 regulates stomatal closing in inflorescence stems but not in leaves. Changes in thermal imaging in line with increased ABA uptake activity were observed in leaves only when AIT1 was overexpressed. The authors hypothesize that AIT1 acts mainly as importer at guard cells in inflorescence stems and in the cells of the vasculature resembles the localization of ABA biosynthetic enzymes to avoid loss of ABA from these cells.

The three other AITs identified in the modified yeast two-hybrid screen were not characterized in detail. In a follow-up paper, the same laboratory looked systemically for NPFs able to transport ABA and extended the method to search for transporters of other hormones, such as gibberellins and jasmonic acid-isoleucine (Chiba et al. 2015). Eleven additional NPFs emerged as candidate ABA transporters from this study, but their physiological roles merit further examination.

A MATE protein has also been identified as an ABA exporter. In an effort to understand the function of MATE/DTX proteins, the laboratory of Professor Luan produced homozygous mutants for a large number of Arabidopsis MATE genes. They observed that the dtx50 mutant was smaller and yellower than the wild type when grown in soil (Zhang et al. 2014). To obtain clues about the substrate(s) of DTX50, the authors screened a multitude of compounds to determine if any would affect the phenotype of the dtx50 mutant. They observed that the growth of dtx50 mutants was more severely inhibited than that of the wild type and other mutants in the presence of ABA. This result suggested that dtx50 mutants might accumulate more ABA and struggle to release it. Indeed, the dtx50 mutants had increased ABA and increased expression of ABA marker genes.

Using E. coli and Xenopus as heterologous expression systems, as well as wild-type and dtx50 Arabidopsis protoplasts, Zhang et al. (2014) convincingly showed that AtDTX50 is an ABA exporter exhibiting high specificity for the physiological enantiomer of ABA. As expected from the dtx50 phenotype, DTX50 is localized to the plasma membrane. Expression of DTX50 was mainly observed in veins and guard cells. The guard cell localization was quite surprising, but is consistent with the observation that dtx50 plants were more drought tolerant and that their stomata closed faster. Apparently, release of ABA from guard cells is important for fine-tuning the regulation of stomatal aperture.

All the work we have described so far used Arabidopsis as a model plant. Much less is known about stomatal regulation in monocotyledons. In a screen for drought-induced genes in rice, Yao et al. (2018) identified OsPM1 (plasma membrane protein1), a membrane protein belonging to the AWPM-19 family. It was already known that genes of this small family are strongly induced by abiotic stress (Koike et al. 1997; Rerksiri et al. 2013; Chen et al. 2015). OsPM1 is mainly expressed in mature embryos, vasculature, and guard cells and is highly induced by ABA. By measuring fluorescence resonance energy transfer in yeast expressing OsPM1, the authors showed that OsPM1 exhibits ABA uptake activity. Physiological experiments revealed that OsPM1 plays an important role in tolerating drought stress, since OsPM1-RNAi plants had a lower survival rate and OsPM1-overexpression (OE) plants had a higher survival rate compared to the wild type. Measurements of water loss and stomatal aperture (which is very difficult in rice) were in line with the survival rates observed: RNAi plants lost more water, while OE plants retained more water. Furthermore, no differences in stomatal apertures could be detected under control conditions, but significantly more stomata were closed in OE plants compared to the wild type in the presence of ABA. By contrast, RNAi plants had significantly more open stomata. All these results indicate that OsPM1 is an ABA transporter required to regulate stomatal aperture and hence drought resistance in rice.

A study on AtABCG22 is worth mentioning in this context (Kuromori et al. 2011). AtABCG22 is highly expressed in guard cells and is required for stomatal regulation. Mutant plants lacking this transporter lose much more water and are more susceptible to drought. However, AtABCG22 had no detectable ABA transport activity, and furthermore, double mutants of atabcg22 combined with the biosynthesis mutant nced3 (nine-cis-epoxycarotenoid dioxygenase 3) or with ost1 (open stomata 1) had enhanced ABA-related phenotypes. These results, together with other physiological observations presented in this work, preclude that AtABCG22 acts as an ABA transporter. Instead, the protein may either transport an unidentified factor that regulates stomatal aperture or be involved in an ABA-independent pathway of stomatal regulation.

Merilo et al. (2015) used gas exchange assays to study the roles of different ABA-related genes and ABA transporters in stomatal regulation. Under reduced air humidity, the only mutant that displayed a gas exchange phenotype was atabcg22, which, as mentioned above, lacks an ABC protein not involved in ABA transport. In response to exogenous ABA, atabcg40 and ait1 had slightly, but not significantly, longer ABA response half-times and a slower ABA-induced decrease in stomatal conductance.

These results can be interpreted in several ways. The large number of ABA importers and exporters and the relatively subtle phenotypes of loss-of-function ABA transporter mutants indicate that there is considerable redundancy among these proteins. Hence, multiple-knockout mutants should be produced to get a clearer picture. Another reason Merilo et al. (2015) might not have detected strong phenotypes is that their experiments were relatively short-term. As shown for the stomatal reaction to changes in relative humidity, the internal ABA stores of guard cells may be sufficient for the initial response. A third point is that, in our experience, plant growth conditions have a strong effect on the reactivity of guard cells. Therefore, different results may be obtained in different laboratories as a function of their growth conditions. This may also explain why some phenotypes have not been reproduced in other laboratories.

4.2 Pathogen Resistance

Numerous reports have dealt with the role of ABA in plant–pathogen interactions (e.g., Cao et al. 2011). Intriguingly, ABA can have opposite effects depending on the situation: it can either promote or inhibit the interaction between the pathogen and the plant. The effect of ABA may depend on the nature of the pathogen, the type of affected tissue, or the developmental stage of the plant, but many questions remain. One of the few genes conferring sustainable pathogen resistance in wheat is the full-size ABCG transporter Lr34res (the resistant Lr34 allele). This allele evolved from the ancestral, susceptible allele (Lr34sus) by two gain-of-function mutations after wheat domestication (Krattinger et al. 2013).

To learn more about the mechanism underlying this important trait, Krattinger et al. (2019) analyzed the transcriptomes of plants expressing the resistant and susceptible forms of the Lr34 transporter. They observed that many ABA-related genes were highly upregulated in plants expressing Lr34res, mimicking to a large extent ABA-dependent stress reactions. To verify that this deregulated expression of ABA-dependent genes has a physiological impact, Krattinger et al. (2019) performed drought stress and stomatal aperture experiments and established that Lr34res plants lost less water and had a reduced stomatal conductance compared to Lr34sus plants. Krattinger et al. (2019) also investigated whether Lr34res could act as an ABA transporter.

Indeed, Lr34res and also, surprisingly, Lr34sus were able to transport ABA in Lr34-expressing yeast cells. However, although immunoblots clearly showed the presence of the Lr34res protein, no Lr34sus protein could be detected. Furthermore, ABA accumulation assays using Lr34-expressing rice seedlings showed that Lr34res-expressing seedlings accumulated more ABA than wild-type and Lr34sus-expressing seedlings. Therefore, although both forms can transport ABA in a heterologous system (yeast), Lr34res is probably the only form that performs this function in plants because it is the only one present in detectable amounts. Nevertheless, exactly how Lr34res-mediated ABA transport confers resistance is an open question. The tissue distribution of Lr34 is still unknown, and the ABA levels in leaves of Lr34res-expressing plants are not changed. Hence, it is tempting to speculate that it is ABA redistribution, not increased levels of the hormone, that leads to pathogen resistance. The ability to visualize ABA redistribution may clarify why ABA renders plants more susceptible to pathogens in some cases and more resistant in others.

Phylogenetic analyses have revealed that there are many homologs of the ABA transporter AtABCG40 in Arabidopsis and other organisms (Hwang et al. 2016). So far, these homologs have been shown to transport of two categories of chemicals: ABA and defense molecules. AtABCG34 and AtABCG36 transport chemicals important in defense against pathogens (Khare et al. 2017; Stein et al. 2006; Lu et al. 2015; Matern et al. 2019), AtABCG30 transports ABA (Kang et al. 2015), and AtABCG31 transports ABA (Kang et al. 2015) and possibly some pathogen defense chemicals (Cho et al. 2019). Indeed, Cho et al. (2019) suggested that many of the full-size ABCGs in plants have co-evolved with the ABCGs of their oomycete pathogens in a long-term arms race.

We can imagine that some of the many AtABCG40 homologs in plants are involved in pathogen defense indirectly via ABA transport, while others evolved to directly transport different substrates for defense against specific pathogens. Future research on the identity of the substrates of these transporters will clarify the many questions on the role of ABA and ABA transporters in plant pathogen defense.

4.3 Regulation of Seed Development and Germination

As mentioned above, seed dormancy depends on a continuous supply of ABA from the endosperm to the embryo. Furthermore, some reports suggest that ABA must also be exported from the embryo to allow seed germination (Holdsworth et al. 2008). All of the studies we described earlier in the section on drought stress and guard cells have also addressed the issue of whether seed germination is affected, but with the exceptions detailed below, none has analyzed this aspect in further detail.

In the case of AtABCG25 and DTX50, the seeds of loss-of-function mutant plants were more sensitive to ABA; their germination and early seed establishment were more severely limited by an exogenous application of ABA than were seeds from wild-type plants (Kuromori et al. 2010; Zhang et al. 2014). By contrast, AIT1 and ABCG40 loss-of-function mutants were less sensitive to exogenously applied ABA compared to the corresponding wild type (Kang et al. 2010; Kanno et al. 2012). These results demonstrate that at least some of the same ABA transporters are involved in drought stress prevention as well as in germination, but the results presented could not explain exactly how these transporters act.

To shed light on this question, Kang et al. (2015) sought to identify ABC transporters that are responsible for the export of ABA from the seed coat and uptake into the embryo. First, the authors analyzed the expression of all 43 ABCG genes from Arabidopsis and established that 10 are highly expressed in seeds. Screening the corresponding mutant lines revealed that four mutants exhibited an altered germination pattern. Second, to establish if these ABCG transporters are more likely involved in ABA release from the seed coat (testa and endosperm) or ABA uptake into the embryo, Kang et al. (2015) separated the seed coat from the embryo and performed a qPCR analysis. Concomitantly the authors generated transgenic plants expressing promoter GUS (β-glucuronidase) reporter gene, uidA constructs. The two methods demonstrated that AtABCG25 and AtABCG31 are localized mainly to the endosperm, while AtABCG30 and AtABCG40 are localized mainly to the embryo. These results suggest that AtABCG25, which had already been characterized as an ABA efflux transporter, and AtABCG31 release ABA from the endosperm, whereas AtABCG30 and AtABCG40 import ABA into the embryo (Kang et al. 2015).

To gain insight into the role of these transporters in seed germination, the authors used the seed coat bedding assay established by Lee et al. (2010). When isolated embryos of the aba2-1 mutant were placed on seed coats of wild-type seeds, germination was completely inhibited. By contrast, when placed on atabcg31 and atabcg25 seed coats, aba2-1 embryos germinated, indicating that AtAGCG31 and AtABCG25 release ABA, and consequently the seed coats of mutants lacking these proteins release less ABA and have an impaired embryo germination inhibitory effect.

To analyze the function of AtABCG30 and AtABCG40, the embryos of mutants lacking these proteins and wild-type embryos were incubated on Cvi (Cape Verde islands) seed coats bedding. These seed coats release more ABA than those of the Columbia ecotype. In line with the expression analysis, germination of wild-type embryos was strongly inhibited, whereas the atabcg30 and atabcg40 embryos germinated, indicating that less ABA was taken up by these embryos. These results indicate that AtABCG30 and AtABCG40 import ABA into the embryo.

Finally, transport experiments confirmed that AtABCG25 and AtABCG31 catalyze ABA export from the seed coat (testa and endosperm), whereas AtABCG30 and AtABCG40 act as embryo-localized importers. However, and similar to ABA transport under drought stress, it can be estimated that for both ABA import and export the respective pairs of transporters are responsible for only about 50–60% of the total ABA transport activity. As the NPFs as well as the MATE DTX50 have been shown to also be localized to seeds, they may mediate at least part of the remaining transport activity. A small part of ABA import may also occur by diffusion.

A novel aspect of ABA transport and seed germination was recently reported by Pawela et al. (2019). The authors characterized in detail an ABC transporter, MtABCG20, from Medicago truncatula that was strongly induced when the plants were exposed to polyethylene glycol, which mimics water stress, and ABA. They showed that this transporter is localized to the plasma membrane, and, by introducing MtABCG20 into tobacco BY2 cells, they observed that ABA was released much faster by these cells than by those transformed with an empty vector. Furthermore, ABA transport could also be observed in vesicles isolated from these cells. These results indicate that MtABCG20 is an ABA exporter. Using promotor-GUS constructs, Pawela et al. (2019) demonstrated that MtABCG20 is localized to both roots and seeds. MtABCG20 expression increased strongly after imbibition, and the expression level remained high for more than 3 days if the seeds were kept at 4°C, while the transcript level declined to less than 50% when the seeds were kept at 24°C.

Germination of mtabcg20 was inhibited to a greater extent by ABA than was the wild type. To establish why this was so, the authors performed a detailed analysis on the site of expression. They observed that in this case the MtABCG20 ABCG transporter was localized to a region that had not previously been reported to contain a transporter, the hypocotyl–radicle transition zone. This is the zone where embryo elongates and is of central importance for germination. The exact mechanism by which embryo-localized ABA inhibits seed germination remains unclear, but apparently reduction of this ABA pool at the specific hypocotyl/radicle elongation zone is a prerequisite for efficient germination. This is in line with the observation of the authors that the ABA concentration in the embryo axis of mtabcg20 plants is more than double that in the wild type. This result is supported by the observation that much higher transcript levels of a highly ABA-induced PP2C gene (HAI2) were detected in mtabcg20 plants compared to the wild type when ABA was supplied exogenously.

In conclusion, work published so far has shown that ABCG proteins are involved in three steps of seed dormancy and germination: the release of ABA from the endosperm, the uptake of ABA by the embryo, and the release of ABA from the embryo. The first two transport processes contribute to maintaining dormancy and preventing early seed germination. The third process concerns the release of ABA from the embryo and therefore promotes embryo elongation and seed germination. This last observation highlights the notion that the ABA content in the embryo is also an important factor in seed germination, at least in some plants.

4.4 Additional Roles for ABA Transporters

As mentioned in the introduction, ABA functions in many developmental and stress-related processes. While the role of ABA has been clearly demonstrated for most of these processes, it is not always known whether ABA is synthesized directly within the target cells or transported to these cells. For instance, it remains to be established whether ABA, serving as a signal that induces the formation of suberin, which reduces water loss from the roots under drought stress (Hose et al. 2001), is produced directly in the exodermis and endodermis or transported from the root phloem parenchyma cells to these cell layers.

4.4.1 Lateral Root Production

The role of ABA in regulating lateral root formation and elongation varies between plant species. In Arabidopsis, ABA inhibits lateral root development (De Smet et al. 2003). The lateral roots of mutant plants lacking AtABCG40 are smaller when they were exposed to exogenously applied ABA. However, since the detailed localization of AtABCG40 is not known, the function of this transporter in lateral root emergence and elongation remains to be determined (Kang et al. 2010).

Work on Fabaceae indicates that, in this plant family, ABA is involved in the pre-emergence development of lateral roots (Gonzalez et al. 2015). In contrast to Arabidopsis, in Fabaceae, low concentrations of ABA increase the number of lateral root primordia and also the number of total lateral roots (Harris 2015). In line with these results, Pawela et al. (2019) showed that the loss-of-function mutants of MtABCG20 described above produce fewer lateral roots, indicating that this transporter delivers ABA to the sites of lateral root primordia formation and is involved in regulating lateral root number. In Medicago, an NPF (MtNRT1.3) has been shown to regulate lateral root formation and inhibit primary root growth (Pellizzaro et al. 2014). However, since these proteins are involved in nitrate uptake and some members also in ABA transport, it is challenging to distinguish between the effect of nitrate nutrition and ABA.

4.4.2 ABA Transporters in Nitrogen-Fixing Nodules

Most plants rely on soil nitrogen for their growth. However, some plants, such as Fabaceae and a few others, can establish a symbiosis with nitrogen-fixing bacteria, in the case of Fabaceae rhizobia. The symbiosis with nitrogen-fixing bacteria enables the plants to obtain nitrogen independently of the soil nitrogen status (Udvardi and Poole 2013). In a complex crosstalk, plants produce nodules, the organs hosting bacteria, which due to their special environment are also called bacteroids. Plants provide bacteroids mainly with carboxylates, while bacteroids release ammonia through the membrane separating the bacteria from the plant cell, the so-called peribacterial membrane, into the plant cytosol. To establish an equilibrium of these exchanges, the number of nodules has to be tightly controlled, and it has been shown that ABA inhibits nodule formation (Ding et al. 2008). ABA also affects the later stages of nodule formation. Exposure of infected roots to external ABA leads to a decreased amount of leghemoglobin, and hence it can be anticipated that the oxygen concentration will increase in these nodules and reduce nitrogen fixation (González et al. 2001).

Loss-of-function mutants for MtABCG20 produced fewer nodules compared to the wild type (Pawela et al. 2019). The rather small, but significant, changes in nodule number suggest that additional ABA transporters are involved in the regulation of nodule formation. LATD/NIP (lateral root defective/numerous infection threads, polyphenolics) could be such a candidate, since the expression of this transporter is regulated by ABA (Yendrek et al. 2010). In legumes, LATD/NIP function is required for the development of two kinds of lateral root organs, lateral roots and nitrogen-fixing root nodules. However, it remains to be established whether this protein is indeed an ABA transporter.

4.4.3 Internal Transporters Important for ABA Homeostasis

ABA glucose ester (ABA-GE) is the major ABA conjugate and to our knowledge is present in various organs of all plant species analyzed so far (Piotrowska and Bajguz 2011). ABA-GE can be hydrolyzed and serves as a source of biologically active, free ABA. ABA-GE has been reported to be localized to vacuoles (Bray and Zeevaart 1985). However, the first report demonstrating the importance of ABA-GE hydrolysis was the identification of AtBG1, a specific ER-localized ABA-GE β-glucosidase (AtBG1) (Lee et al. 2006). Plants lacking this glucosidase activity exhibit pronounced ABA-deficiency phenotypes, including sensitivity to dehydration, impaired stomatal closure, earlier germination, and lower ABA levels, demonstrating the importance of this conjugated ABA pool.

Two subsequent papers showed that the vacuolar ABA-GE pool also plays an important role in the ABA-related stress response, since two vacuolar β-glucosidases able to hydrolyze vacuolar ABA-GE exhibit a similar, although less pronounced, function as AtBG1 (Wang et al. 2011; Xu et al. 2012). The observation that the single mutants of the vacuolar β-glucosidases exhibit a weaker phenotype than those lacking AtBG1 may be due to a partial redundancy.

These results raise the question of which transporters mediate ABA-GE accumulation in the ER and vacuolar lumen and how free ABA hydrolyzed by glucosidases is exported from these two compartments. Burla et al. (2013) showed that vacuoles can take up ABA-GE via two transport mechanisms. One is directly activated by ATP and exhibits properties consistent with the action of an ABC transporter. Indeed, two vacuolar ABC transporters, AtABCC1 and AtABCC2, are able to transport ABA-GE into vesicles from yeast transgenically expressing these transporters. Vacuolar ABA-GE uptake was also shown to be driven by a proton antiporter. However, in this case the nature of the transporter remains to be elucidated.

The KM for the uptake of ABA-GE is approximately 1 mM for both transport mechanisms and hence far higher than the cytosolic concentration of this conjugate. Thus, the vacuolar uptake of ABA-GE does not have a regulatory role in ABA-GE storage. Whether ABA released after hydrolysis diffuses across the vacuolar membrane or is delivered to the cytosol by a transporter is unknown. In the acidic vacuole, a large proportion of ABA is present in the protonated, diffusible form and would probably allow efficient delivery to the cytosol. However, the pH of the ER lumen is distinctly less acidic than that of the vacuole, and diffusion of ABA to the cytosol would probably be much less efficient (Martinière et al. 2013). Therefore, a transporter that has yet to be discovered likely mediates the export of ABA from the ER to the cytosol.

5 Concluding Remarks

ABA is involved in many different environmental responses and developmental processes. Not all functions are mentioned in the transport sections; for instance, there is evidence that ABA is also involved in bud dormancy (Cooke et al. 2012; Li et al. 2018), in root hydrotropism and xylem formation, and as a factor regulating the circadian clock and senescence (Dietrich et al. 2017; Ramachandran et al. 2018; Lee et al. 2016). The members of the ABA receptor complex are present as small gene families. In Arabidopsis, there are 14 RCAR/PYR/PYL receptor proteins and 9 clade A PP2Cs (Yoshida et al. 2019). This diversity in ABA receptors greatly expands the number of possible combinations, influencing where and how ABA can act. Indeed, it has been hypothesized that this diversity allows ABA to be perceived in a wide concentration range and in different cells embedded in different tissues (Yoshida et al. 2019). The large number of potential ABA receptor combinations allows the plant to act in a versatile way to different demands during development and in response to environmental stimuli. Together, the large number of potential receptor combinations implies that ABA transport must be versatile so that the plant can respond to various developmental stages and environmental conditions.

It is striking that the phenotypes of mutants with defects in ABA transporters described so far are rather moderate compared to those lacking enzymes involved in ABA biosynthesis. This suggests that, in all the responses described so far, more than one transporter is involved in delivering ABA and activating a given process. However, if there were complete redundancy among members of the ABA transporter families, mutants lacking an ABA transporter would not have an aberrant phenotype. Hence, more than one transporter is likely required for optimal ABA action.

Fig. 1
figure 1

ABA transporters described to date function in plant development and adaptation to changing environmental conditions. For details, see text

Full size image

It is likely that, depending on the growth conditions, the absence of a given ABA transporter leads to a more or less pronounced phenotype. We hypothesize that ABA transport is mediated by several transporters that have overlapping expression in a given tissue or cell type. The presence of several ABA transporters with a limited transport activity may help fine-tune the amount of ABA delivered to a specific site. This is important, since in many of the processes where ABA is involved, ABA is not the only player but participates in a complex phytohormone network and different internal ABA concentrations may have different effects. Furthermore, the presence of several transporters may act as security, since the absence of one transporter would not necessarily lead to plant death.