Plants are modest organisms, their basic requirements for life being largely satisfied by light, CO2 and water, with the addition of a supply of mineral nutrients in the soil. Macronutrients, which are required in comparatively large quantities, include the elements nitrogen (N), potassium (K), sulphur (S), phosphorus (P), magnesium (Mg) and calcium (Ca). Our understanding of plant nutrient requirements has led to enormous progress in agricultural food production, and most farmers in Western Europe and the USA routinely apply N, P and K as fertilizers. Nevertheless, nutrient deficiencies regularly occur even in fertilized fields as chemical and physical properties of the soil can lead to reduced mobility and absorbance or leaching of individual nutrients. Financial constraints, especially in developing countries, often necessitate that farmers prioritize some nutrients over others (e.g. N over K). As a result of such imbalanced fertilizer input, crops do not reach their full potential and soils are depleted of specific nutrients. With fertilizer prices increasing and agriculture (as well as biofuel production) moving into marginal soils, a good understanding of nutrient uptake and usage by the plant becomes ever more important. In particular, molecular information concerning the regulation of nutrient transport will be an essential prerequisite for biotechnological efforts to increase nutrient usage efficiency.
For the uptake of macronutrients and their allocation in different cellular compartments and tissues, plants employ a number of transport proteins (‘transporters’), which differ from each other not only in their tissue and membrane location but also in their mode of energization, substrate affinity and specificity (Blatt, 2004). The enormous variety of features displayed by transport proteins provides an invaluable pool for plants from which to select those transporters that are best suited to fulfil their nutritional demands in particular conditions. Approximately 1000 genes (5% of the entire genome) of Arabidopsis thaliana have known or putative functions in membrane transport (Maathuis et al., 2003). Clearly they are not all active everywhere and at all times. Rather their expression and activity are regulated in response to a number of external and internal stimuli. The degree to which plants make use of the specific features of individual transporters has been revealed in a recent study. For example, the spatial arrangement of four different members of the ammonium transporter (AMT) family within the A. thaliana root, in combination with their distinct substrate affinities, provides a highly sophisticated system for directed ammonium transport within the root tissue (Yuan et al., 2007). Differential regulation of the individual transporters by external factors further increases the complexity of such systems.
2. Function and regulation
Regulation of nutrient transporters is closely related to the functions of the transported macronutrients, which can be roughly divided into metabolic, osmotic and energetic functions (Marschner, 1995; Epstein & Bloom, 2005). Apart from K and Ca, all macronutrients are integrated into important organic compounds such as amino acids and proteins (N and S), nucleic acids (N and P), phospholipids (P) and chlorophyll (Mg). P and Mg have additional functions in energy conservation and conversion (Mg-ATP). In accordance with these functions, regulation of N, P, S and Mg transporters has the overall goal to supply the plant with essential molecules for the build-up of dry matter and energy during growth and development. K and Ca differ in several aspects from other macronutrients. They are not metabolized but remain in their ionic elementary form throughout plant growth and development. In their capacity as osmolytes they drive growth by securing a steady increase in the plant water content, and regulation of their transport is therefore again under the diktat of growth and related stimuli. However, transport of K+ and Ca2+ is not only a means for uptake and allocation of these nutrients but serves other essential functions. Notably, K+ flux counterbalances other ion fluxes, for example, fluxes of protons, thereby enabling the activity of ATP synthase, H+-ATPase and H+ co-transport systems. As a result, regulation of these systems is often linked to the regulation of K+ transport. Furthermore, the osmotic function of K+ is not limited to growth but extends to the reversible movement of plant organs, most importantly the opening and closing of stomata (Willmer & Fricker, 1996; Blatt, 2000). Consequently, K+ channels in guard cells are regulated by environmental factors such as light, humidity and CO2. Ca2+ transport is closely related to the role of cytoplasmic Ca2+ as a second messenger. Influx of Ca2+ into the cytoplasm through ion channels and its subsequent removal by Ca2+-pumps create oscillations of cytoplasmic Ca2+ that constitute a signal that is in many cases essential for linking environmental stimuli to downstream events inside the cell (Sanders et al., 1999, 2002). Consequently, regulation of Ca2+ transporters is at the centre of many cellular responses to the environment.
3. Physiology and mechanistics
Regulation of nutrient transport can be discussed from a physiological and a mechanistic point of view. In this review we do both. Section II reviews current knowledge regarding external and internal stimuli that regulate nutrient transporters and the signalling pathways involved. While it is impossible to include all existing evidence here, we present a number of examples that we consider representative for the different situations that require regulation of nutrient transport. Section III describes molecular mechanisms underlying regulation of nutrient transporters. Investigation of these mechanisms is greatly aided by techniques that directly measure nutrient movement through the respective transport proteins and high-resolution structural models on which to base structure–function analysis. The former are available for ion channels (e.g. patch clamp analysis), and the latter exist for some ion channels and pumps. Much of the information collated in Section III therefore relates to the transport of K, Ca or protons. To enhance the potential usefulness of this section in terms of our understanding of the regulation of other nutrient transporters, we present only those mechanisms that operate in more than one type of transporter or that functionally link different types of transporters. The overall aim of this review is to provide an incentive for knowledge transfer between different lines of research, for example concerning different stimuli (nutrients, metabolites and water status), nutrients (N, S, P, K and Ca), types of transporters (channels, pumps and co-transporters), and levels of regulation (transcript, protein and submolecular). The future challenge will be to experimentally establish for each stimulus–response pair a functional continuum between receptor and targeted transporter, and to understand how simultaneously occurring stimulatory inputs are integrated into distinct mechanistic outputs.
II. Stimuli and signals
1. Nutrient availability
Transcript abundances of ion transporters often vary with the concentration of their substrate in the growth medium. While some transporters are induced by a decrease in substrate concentration from high to low supply, others are induced by an increase in substrate concentration from nil to low supply. For example, abundances of transcripts encoding high-affinity sulphate (e.g. A. thaliana AtSULTR1;2 and AtSULTR1;2; Buchner et al., 2004) and phosphate (e.g. AtPT1 and AtPT2; Al-Ghazi et al., 2003) transporters rise upon removal of S or P from the growth medium. By contrast, up-regulation of members of the high-affinity nitrate transporter (NRT2) family is observed after adding small amounts of nitrate (10–50 µM) to an N-depleted medium (Krapp et al., 1998; Filleur & Daniel-Vedele, 1999). Responsiveness to external supply can even differ between homologous genes in different species. For example, ammonium transporters of the AMT family in A. thaliana (e.g. AtAMT1;1 and AtAMT1;3) show increased expression during N deficiency, whereas some AMT homologues in tomato (Lycopersicon esculentum) (LeAMT1;2) and rice (Oryza sativa) (OsAMT1;1 and OsAMT1;2) are induced by N supply (Loque & von Wiren, 2004). AMT transporters also differ in their temporal response to N deficiency, suggesting that they respond to plant nutrient status rather then external concentrations. Thus, transfer of plants to N-free medium induces the expression of AtAMT1;1 and AtAMT1;3 within 3 d, whereas induction of AtAMT1;2 and AtAMT2;1 requires more extended periods of N deficiency (Gazzarrini et al., 1999; Sohlenkamp et al., 2000).
The observation that some transporters are induced by a change from high to low nutrient supply and others by a change from nil to low supply indicates a fundamental difference in the underlying signalling pathways that allows plants to respond to different environmental situations, that is, progressive depletion of the nutrient in the soil or resupply after a period of scarcity. In both cases the regulatory events result in high expression levels of specific transporters under conditions that require their function as high-affinity systems in nutrient uptake. Whether differences observed at the mRNA level are still apparent at the protein level is often questioned. A recent study in which myc-tagged SULTR1;1 and SULTR1;2 proteins were expressed under the control of the endogenous promoters showed that the response to S starvation is not only apparent but stronger and faster at the protein level (Yoshimoto et al., 2007).
Astonishingly few transporters involved in K+ transport respond to varying K supply at the level of transcripts (Maathuis et al., 2003). Out of some 50 genes expected to have K+ transport capacity (Mäser et al., 2001) only HAK5, a putative high-affinity K+ uptake system, has consistently been reported as being induced by K starvation (Armengaud et al., 2004; Shin & Schachtman, 2004; Gierth et al., 2005). Transcript abundances of K+ channels, although responding to several environmental and hormonal stimuli (Pilot et al., 2003), are not affected by external K supply. It appears that adaptation of K+ channels to K availability and plant K status occurs primarily at the protein level (see Section III).
The molecular elements targeted by nutritional stimuli have been identified in some cases. A GATA transcription factor was found to be inducible by nitrate (Bi et al., 2005), and a 150-bp region in the promoter of AtNRT2.1 contains a GATA motif (Girin et al., 2007). The 150-bp cis-acting element is required for up-regulation of NRT2.1 by nitrate and its repression by N metabolites. The GATA motif, while necessary for regulation of NRT2.1, is not sufficient, and it has been suggested that a DOF element in the promoter regions plays an additional role in the regulation of NRT2.1. Interestingly, activity of the 150-bp region is further modulated by sucrose, and DOF elements, as well as the GATA transcription factors, play a role in regulating carbon (C) metabolism (Bi et al., 2005; Girin et al., 2007). These regulatory units are therefore possible points of integration between C and N metabolism. Analysis of mutants with deletions in the upstream region of AtSULTR1;1 identified a 16-bp sulphur-responsive element (SURE) between −2777 and −2762 that is sufficient and necessary for enhanced expression of SULTR1;1 in response to S starvation (Maruyama-Nakashita et al., 2005). Regulation of phosphate transporters and other P-responsive genes is under the control of transcription factors of the MYB-CC family such as PHR1 and PHR2, which act as positive regulators (Rubio et al., 2001; Todd et al., 2004). AtPHR1 recognizes a GnATATnC motif, the P1BS element. However, although the P1BS motif is present in the promoters of many P-regulated genes it is not overrepresented in P-regulated genes (Hammond et al., 2003). This could indicate that other promoter elements are required for P-specific responses or that P regulates PHR genes at the post-transcriptional level (Amtmann et al., 2006). For example, a recent study provided evidence for post-transcriptional regulation of PHR1 by sumoylation. SIZ1, a small ubiquitin-like modifier (SUMO) E3 ligase, transiently activates PHR1 during P starvation (Miura et al., 2005). Finally, protein degradation by ubiquitination has emerged as an important regulatory mechanism for plant adaptation to N- and P-limiting conditions (Fujii et al., 2005; Peng et al., 2007). Particularly interesting is the case of PHO2, an E2 ubiquitin conjugating enzyme, which underlies the locus of a P-hyperaccumulating A. thaliana mutant. Upon P starvation, PHO2 mRNA is degraded by a complementary micro RNA (miR399), resulting in de-repression of downstream targets including phosphate transporters PHT1.8 and 1.9 (Fujii et al., 2005; Aung et al., 2006; Bari et al., 2006; Chiou et al., 2006).
Identification of upstream events linking changes in nutrient availability to gene expression of the respective transporters has been an area of intensive research, the results of which are summarized in Fig. 1. Transcriptional regulation of sulphate transporters involves phosphorylation (e.g. by the Snf1-like Ser/Thr kinase Sac3; (Davies et al., 1999) and dephosphorylation events (Maruyama-Nakashita et al., 2004a). Production of reactive oxygen species (ROS) occurs in roots in response to K, N, P and S deprivation (albeit differing in origin and location; Shin et al., 2005; Schachtman & Shin, 2007), and is necessary for the induction of some downstream responses including de-repression of HAK5 (Shin & Schachtman, 2004). Several studies indicate a role for plant hormones in mediating between nutritional stimuli and nutrient transporters. Thus, K starvation enhances the expression of enzymes involved in the biosynthesis of ethylene (Shin & Schachtman, 2004) and jasmonic acid (Armengaud et al., 2004), and concentrations of the two hormones increase in roots and shoots of K-starved plants, respectively (Shin & Schachtman, 2004; Cao et al., 2006). However, the exact position of ethylene and jasmonate signals within the K starvation response remains to be elucidated. Expression of sulphate and phosphate transporters is repressed by cytokinin and application of cytokinin suppresses their induction by S or P starvation (Martin et al., 2000; Maruyama-Nakashita et al., 2004c; Hou et al., 2005). Signal transduction of cytokinin-dependent responses involves the histidine kinase CRE1 (Inoue et al., 2001), as cre1 mutants no longer show a response of sulphate transporters to cytokinin (Maruyama-Nakashita et al., 2004b). The observation that both cytokinin and CRE1 transcript abundances decrease during P starvation (Franco-Zorrilla et al., 2005) further supports the notion of a CRE1/cytokinin signalling pathway in nutrient responses.
2. Primary assimilates and other nutrients
Primary assimilates often exert feedback control on nutrient uptake (Fig. 1). For example, expression of the nitrate transporter NRT2.1 is repressed by high concentrations of the amino acids arginine, asparagine and glutamine (Krapp et al., 1998; Zhuo et al., 1999; Vidmar et al., 2000). In A. thaliana, induction of the ammonium transporter AMT1;1 in response to resupply of ammonium only occurs if assimilation of ammonium into glutamine is inhibited by methionine sulphoximine (Rawat et al., 1999). By contrast, in rice, glutamine induces the expression of OsAMT1;1 (Sonoda et al., 2003), indicating that the same amino acid can act as a metabolic trigger for both down- and up-regulation of AMT genes, depending on isoform and plant species (Loque & von Wiren, 2004). Sulphate transporters in barley (Hordeum vulgare) (HvST1) and A. thaliana (SULTR1;1 and 1;2) are repressed by glutathione (GSH) and cysteine (Smith et al., 1997; Maruyama-Nakashita et al., 2004c). O-acetyl-L-serine (OAS, a precursor of cysteine synthesis) overrides the negative feedback regulation of HvST1 by GSH (Smith et al., 1997). Both A. thaliana genes are also up-regulated by OAS, albeit with different sensitivities (Maruyama-Nakashita et al., 2004c).
The extent to which a plant can utilize a particular nutrient depends on the availability of all other nutrients. It can therefore be expected that nutrient transporters will also be regulated by nonsubstrate nutrients. Very few studies have directly addressed this question, but the effect is clearly visible in microarray experiments where removal of one nutrient almost always results in transcript changes of genes mediating the transport of other nutrients (Wang et al., 2002; Hammond et al., 2003; Maathuis et al., 2003; Nikiforova et al., 2003). One example that has been investigated in some detail concerns the response of sulphate transporters to N supply. It was found that low N attenuates the induction of SULTR1;1 and SULTR1;2 by S starvation, and suggested that the signal resides in the OAS pool (Maruyama-Nakashita et al., 2004b). OAS production involves an amino transfer reaction and therefore depends on the supply of N. Another example of cross-regulation is the effect of P supply on the expression of AMTs and NRTs (Wang et al., 2002; Wu et al., 2003), which may involve a systemic sucrose signal. NRT2.1 was also found to be down-regulated during K starvation (Armengaud et al., 2004), which is surprising in the light of increased sugar concentrations in K-starved roots (Amtmann et al., 2008). In this case, regulation is likely to occur through changes in organic and amino acid concentrations, which in turn may be the result of allosteric regulation of several glycolytic enzymes by K+ and pH (Amtmann et al., 2006).
3. Carbon status
Nutrient uptake is tightly linked to the C status of the plant, and indirectly controlled by environmental factors that determine the photosynthetic rate, such as light (Fig. 1). Dependence on C metabolism was established for sulphate transporters of the SULTR family. Addition of glucose and sucrose enhanced the transcriptional response of SULTR1;1 and SULTR1;2 to S starvation (Maruyama-Nakashita et al., 2004c). Conversely, depletion of C sources from the growth media attenuated the induction of these two genes in S-free medium. Lejay and co-workers (Lejay et al., 2003) tested a number of root ion transporters for regulation by photosynthesis. The A. thaliana genes AMT1.1, AMT1.2 and AMT1.3, NRT1.1 and NRT2.1, HST1, AtPT2 and AtKUP2, encoding ammonium, nitrate, sulphate, phosphate and potassium transporters, respectively, were all repressed in the dark. This repression was prevented by adding sucrose at the beginning of the dark period, indicating a link to the photosynthetic rate rather than the circadian clock. The authors found a strong correlation between the stimulating effects of light and sucrose, and measured an increase in the concentration of soluble sugars in the root tissue during the light period. Lejay and co-workers went on to investigate a possible role of known sugar signalling pathways in the regulation of these transporters. Hexokinase (HXK) has been postulated to be a sugar sensor and a regulatory element for crosstalk between C and N metabolism (Jang et al., 1997; Moore et al., 2003). Previous experiments with sugar analogues that are phosphorylated by HXK but poorly metabolized by glycolysis (i.e. 2-deoxyglucose (2-DOG) and mannose) had suggested that HXK was necessary and sufficient for the creation of sugar signals, independent of its function in sugar metabolism (Jang & Sheen, 1994). However, this was not the case for light regulation of nutrient transporters (Lejay et al., 2003). Expression of NRTs, AMTs and HST1 was repressed rather than stimulated by 2-DOG or mannose, and glucosamine, an inhibitor of HXK, decreased mRNA levels of NRT2.1 even when sucrose was applied. Further evidence against a role of HXK signalling in light regulation of transporters came from experiments with sugar sensing mutants (rsr1, sun6, gin1-1 and hxk), none of which showed an altered transcriptional response of NRT2.1 to sucrose and light. Induction of NRT2.1 by sucrose and glucose was, however, abolished in HXK antisense plants, suggesting that catalytic activity of HXK is required for sugar regulation of this transporter. Recently, the same group identified glucose-6-phosphate (G6P) as the sugar signal (Lejay et al., 2008). G6P regulation operates downstream of the metabolic function of HXK and requires in most but not all cases an active oxidative pentose phosphate pathway (Lejay et al., 2008). It should be noted that the sensitivity of NRT2.1 to nitrogen, light and sugars at the level of the transcript is reflected in high-affinity nitrate uptake but not in protein abundance at the target membrane (Wirth et al., 2007), which suggests that additional post-translational regulatory mechanisms are involved in adjusting nitrate uptake to the nutritional and metabolic requirements of the plant.
4. Water status
Over the past three decades the stomatal guard cell has risen to the status of the premier model for study of membrane transport and its regulation in plants, especially in relation to those characteristics associated with ion channels. Stomata are pores that provide the major route for gaseous exchange across the impermeable cuticle of leaves and stems (Hetherington & Woodward, 2003). They open and close in response to exogenous and endogenous signals – the most important of these being light, CO2 and the hydration status of the plant – and thereby control the exchange of gases, most importantly water vapour and CO2, between the interior of the leaf and the atmosphere.
While the Ca2+ signal predominates in regulating Cl− channels and inward-rectifying K+ channels, it is curiously absent in the control of outward-rectifying K+ channels in the guard cell, for example GORK in A. thaliana and its counterpart in Vicia (Hosy et al., 2003; Dreyer et al., 2004). Instead, current through these channels is strongly increased by increasing cytosolic pH (pHi; Grabov & Blatt, 1997) consistent with the rise in pHi evoked by ABA (Irving et al., 1992; Blatt & Armstrong, 1993). Unlike the situation for Ca2+, virtually nothing is known of the mechanism behind this rise in pHi nor of its site of action, although its kinetics are sufficiently slow (Blatt & Armstrong, 1993) to be accommodated by cation exchange and charge balancing events during solute efflux from the vacuole (MacRobbie, 2000). Figure 2 summarizes the signalling pathways mediating between plant water status and transport activity in stomatal guard guard cells.
III. Molecular mechanisms for regulation
1. Voltage gating
Membrane voltage can affect ion permeation through transport proteins in two ways. First, as an electromotive force on ion (charge) movement across the membrane, the membrane voltage acts as an electrical analogue of concentration in driving ion flux through ion channels, carriers and pumps. Secondly, membrane voltage serves as a regulatory factor in controlling ion flux. Many ion channels open and close – or gate – in response to membrane voltage, thereby controlling the flux of ions on timescales of milliseconds. In general, gating is associated with conformational changes of the channel protein coupled to a ‘voltage sensor’ that effectively control the opening of the pore (Hille, 2001). Best characterized are the Kv (Shaker) K+ channels, which incorporate a voltage sensor domain comprising a charged transmembrane helix which moves in response to changes in the membrane voltage and is thought thereby to pull directly on the closure mechanism of the channel pore (Sigworth, 2003; Dreyer et al., 2004; Tombola et al., 2005). Several members of this channel family have important functions in plant K nutrition as they mediate uptake of K+ from the soil (AKT1; Hirsch et al., 1998) or its long-distance transport in the xylem (SKOR; Gaymard et al., 1998) and phloem (AKT2/3; Marten et al., 1999). While gating of AKT2/3 is only weakly voltage-dependent, the open probability of AKT exhibits a steep increase at negative membrane potentials. Such ‘inward rectification’ gives the channel ‘valve’ type characteristics favouring uptake of K+ when an inward driving force exists while minimizing its loss when the driving force is directed outwards (Maathuis & Sanders, 1995).
Regulation of channels often targets the voltage sensitivity of the gating. Most importantly, for many plant K+ channels, the voltage dependence of gating is modulated by the availability of K+ outside (Blatt, 1991). For example, in stomatal guard cells, outward rectifying channels open only when the driving force for net K+ flux is directed outwards. This regulation ensures that K+ efflux – which is needed to drive stomatal closure and control gas exchange – can occur even when extracellular K+ varies over concentrations from 10 nM to 100 mM (Blatt & Gradmann, 1997). The ability to respond to extracellular K+ is integral to the K+ channel protein itself and therefore represents one of the very few examples in which we know of the mechanism for ‘nutrient sensing’. From a molecular standpoint, gating by K+ implies that cation binding with the channel protein stabilizes a closed conformation of the channel pore. Recent studies identified a site deep within the S6 transmembrane helix of the channel protein, adjacent to the so-called pore helix that is essential for this K+ sensitivity (Fig. 3; Johansson et al., 2006). The finding is significant, because analogous interactions between the pore helix and the S6 helix are known to affect gating of mammalian K+ channels (Alagem et al., 2003; Seebohm et al., 2003, 2006), but in animals these interactions are favoured by cation occupation of the pore. One recent study (Li et al., 2008) builds on the findings of Johansson and co-workers, demonstrating that, with a few additional mutations, the voltage dependence for SKOR gating shifts sufficiently to yield inward rectification behaviour. It is still not clear, however, whether the channel in this form still retains a coupling to external K+ concentration. So, the results of Li and co-workers leave open the most important question about the interplay in the mechanics of gating by K+ and voltage that accounts for the response of the channel to the K+ environment.
External pH is another important modulator of channel activity. K inward currents in the plasma membrane of guard and root cells are activated by acidification of the external medium (Blatt, 1992; Ilan et al., 1996; Amtmann et al., 1999). Activation is achieved by a combined effect of external pH on channel gating and transport rate. It was further demonstrated in barley roots that the pK of channel activation is close to the apoplastic pH of barley roots, which in turn is determined by the activity of the plasma membrane proton pump (Amtmann et al., 1999). Hence, regulation of K+ channels by external pH directly links K+ uptake to the activity of the primary H+-ATPase. Analysis of mutations in the potato (Solanum tuberosum) K+-channel KST1 identified a histidine residue in the outer pore region of the channel as crucial for the effect of pH on channel gating (Hoth et al., 1997).
Auto-inhibitory domains play an important role in the post-translational regulation of many transporters, including Ca pumps and antiporters (for ammonium transporters, see Section III.4). Auto-inhibitory domains in Ca pumps are closely related to those in proton pumps, and we will therefore include a description of the latter here. Auto-inhibitory domains of P-type H+ and Ca2+ ATPases reside in the C- and N-termini, respectively (Geisler et al., 2000a; Morsomme & Boutry, 2000; Baekgaard et al., 2005), and interact both intramolecularly with other parts of the pump and extramolecularly with activating proteins including 14-3-3 proteins, protein kinases and calmodulin (Fig. 4 and Section III.3).
The observations that cleavage of a C-terminal fragment by trypsin treatment led to H+-ATPase activation, and that removal of the C-terminus was necessary to achieve functional complementation of a yeast strain lacking endogenous proton pump activity provided the first evidence for auto-inhibition of plant P-type H+-ATPases (Palmgren et al., 1991; Palmgren & Christensen, 1993; DeWitt & Sussman, 1995; Baunsgaard et al., 1996). A number of point mutations that release auto-inhibition and improve the coupling ratio between proton pumping and ATP hydrolysis were subsequently identified in the tobacco (Nicotiana plumbaginifolia) H+-ATPase PMA2 (Morsomme et al., 1996, 1998). The majority of these mutations are concentrated in two regions located close to one another within the first half of the C-terminal domain. Similarly, systematic alanine scanning of the C-terminus of the plasma membrane proton pump AHA2 from A. thaliana revealed two regions (RI and RII) as being important for auto-inhibition (Axelsen et al., 1999).
Auto-inhibitory domains also feature in co-transporters. For example, CAX1, a vacuolar Ca2+/H+ antiporter of A. thaliana, contains an N-terminal regulatory region of 36 amino acid residues that interacts with a neighbouring domain, thereby inhibiting Ca2+ transport (Pittman & Hirschi, 2001). Expression of N-terminally truncated versions of CAX1 as well as versions with several point mutations within the 36-amino acid region suppresses Ca2+ sensitivity in yeast, but expression of the full-length cDNA does not (Pittman et al., 2002). The function of such regulation in planta has now been shown for the first time. Mei and co-workers (Mei et al., 2007) selected a number of N-terminal variants of CAX1 based on their inhibitory effect in yeast, and subsequently analysed the effects of these variants – expressed under the control of the 35-S promoter – on ion contents in tobacco. All transgenic lines showed high levels of CAX1 transcript but they clearly differed with respect to the concentrations of several mineral nutrients, and this difference was in accordance with the inhibitory impact of the mutations previously determined in yeast.
3. Protein–protein interaction
14-3-3 proteins 14-3-3 proteins regulate the activities of a wide range of targets via direct protein–protein interaction, which depends on the phosphorylation status of the targets (Roberts, 2003). The interaction involves binding of short amino acid motifs, containing phospho-serine or phospho-threonine, of the target protein and a conserved amphipathic region in each monomer of a dimeric 14-3-3 protein (Roberts, 2003). 14-3-3 proteins are highly conserved among kingdoms, and are involved in many cellular processes (Aitken, 1996). In plants, 14-3-3 proteins interact with metabolic enzymes (Huber et al., 2002; Comparot et al., 2003), transcription factors (Schultz et al., 1998), protein kinases (Camoni et al., 1998) and ion transporters (de Boer, 2002).
Effects of 14-3-3 proteins on ion channel currents have been observed in a number of plant systems including tomato, barley and A. thaliana (Saalbach et al., 1997; van den Booij et al., 1999; van den Wijngaard et al., 2005; Latz et al., 2007). Recent studies have provided clues to the physiological relevance of this type of ion channel regulation. For example, patch clamp experiments in barley embryos uncovered a functional link between the role of ABA in seed dormancy and 14-3-3 regulation of K uptake channels in the emerging radicle (van den Wijngaard et al., 2005). The current model is that ABA causes channel dephosphorylation (by activating protein serine/threonine phosphatases ABI1 and 2; Armstrong et al., 1995), which leads to dissociation of a 14-3-3 protein and inhibition of a K+ inward current that is required for radicle emergence (van den Wijngaard et al., 2005). The likely molecular target of this regulatory ensemble is the K+ channel AKT1, which contains C-terminal 14-3-3 binding motifs both in A. thaliana and in barley (A. H. De Boer, pers. comm.).
Ion channels in the tonoplast are also subject to regulation by 14-3-3 proteins. Latz and co-workers (Latz et al., 2007) measured opposite effects of 14-3-3 on K+currents mediated by TPK1 (Gobert et al., 2007) and TPC1 (Peiter et al., 2005). These channels differ not only in voltage dependence but also in ion selectivity, and therefore differential regulation by 14-3-3 could provide a means to alter the overall ion selectivity of the tonoplast. The future challenge is to identify the signalling pathways that link this regulation to environmental or developmental stimuli.
14-3-3 proteins might also modulate nitrate transport. NRT2.1 from tobacco and NRT2.4 from A. thaliana contain perfect 14-3-3 binding motifs, albeit in different positions (Miller et al., 2007). 14-3-3 regulation of nitrate transporters is an intriguing prospect as several key enzymes of N assimilation (e.g. nitrate reductase and glutamine synthetase) are regulated by 14-3-3 (Moorhead et al., 1996; Finnemann & Schjoerring, 2000).
Calmodulin Calmodulins (CaMs) are small Ca2+-binding proteins that can translate intracellular Ca2+ signals into a variety of cellular responses. In accordance with this function, CaMs are involved in plant responses to a large number of environmental stimuli (Snedden & Fromm, 2001). Targets of this large gene family (approx. 28 members in A. thaliana) include Ca2+-ATPases, cyclic nucleotide gated channels (CNGCs) and the vacuolar Na+/H+ antiporter NHX1 (Fig. 4).
The relative kinetics of the influx of Ca2+ and its removal by Ca2+ pumps shape the Ca2+ signal, thereby endowing it with some specificity (Allen et al., 2001; Harper, 2001). The interaction among Ca2+, CaM and Ca2+-ATPases plays an important role in this process because it creates a negative feedback loop that instigates removal of Ca2+ from the cytoplasm as soon as cytoplasmic Ca2+ concentrations start to rise (Fig. 5). Upon binding of Ca2+, CaM interacts with the N-terminus of IIB-type Ca2+ pumps and leads to increased activity. There is no consensus CaM binding site but they are usually 15–30 amino acids long and form an alpha helix containing two bulky hydrophobic residues that function as anchors for CaM (Crivici & Ikura, 1995; Yap et al., 2003). N-terminal CaM binding domains (CaMBD) have been identified in the cauliflower (Brassica oleracea) Ca2+-ATPase BCA1 (Malmstrom et al., 1997) and in the Arabidopsis Ca2+ pumps ACA8 and ACA9 (Bonza et al., 2000; Schiott et al., 2004). Interference of CaM binding with auto-inhibition involves six residues in the CaMBD including hydrophobic anchor residues, which suggests that these residues have a dual function in CaM recognition and in auto-inhibition (Baekgaard et al., 2006). Kinases are again important modulators of this regulation but in contrast to their stabilizing effect on the C-terminal 14-3-3 protein complex in proton pumps they seem to inhibit CaM action on Ca2+ pumps. For example, it was shown for the endomembrane Ca2+-ATPase ACA2 that phosphorylation of a serine site near the CaMBD by a Ca2+-dependent kinase inhibits CaM stimulation and basal activity (Hwang et al., 2000). The physiological meaning of this apparently inverse effect of intracellular Ca2+ on ACA2 via CaM and Ca-dependent protein kinase (CDPK) remains to be elucidated.
Experiments with the vacuolar H+/Na+ antiporter AtNHX1 have raised the surprising possibility that CaM also acts in the vacuole (Yamaguchi et al., 2003, 2005). Yeast two-hybrid assays showed that a CaM-like protein AtCaM15 interacts in a Ca2+- and pH-dependent manner with the C-terminus of AtNHX1. Progressive deletions of the C-terminus mapped the binding site to a region that has indeed the potential to form a positively charged amphiphilic helix. Previously the group had shown that the C-terminus of AtNHX is located in the vacuolar lumen (Yamaguchi et al., 2003), and that its deletion increases the Na+/K+ selectivity of the antiport. In accordance with these findings, CaM binding to NHX1 decreases the maximal transport rate for Na+ but not for K+. Hence, regulation of NHX1 by vacuolar CaM might provide a molecular switch between Na+ and K+ transport.
CaM-binding sites are also present in the C-termini of CNGCs. CNGCs display a wide range of substrate specificity (ranging from K selectivity to no selectivity among cations) and are involved in a number of physiological processes (Talke et al., 2003; Demidchik & Maathuis, 2007) but their specific functions remain obscure. From a mechanistic viewpoint it is interesting that in plant CNGCs the CaM-binding domain overlaps with the cyclic nucleotide-binding domain (Arazi et al., 2000; Fig. 4), and there is some evidence that binding of CaM at the C-terminus interferes with ligand binding and activation (Hua et al., 2003).
Protein kinases The importance of [Ca2+]i for plant responses to environmental stimuli is well established but how [Ca2+]i signals are translated to alterations in ion transport activities has long been uncertain, although several strings of evidence suggested that phosphorylation events were involved (Luan et al., 1993; Thiel & Blatt, 1994; Grabov et al., 1997). The molecular components of one pathway linking [Ca2+]i with transport activity were discovered in salt-oversensitive (sos) mutants of A. thaliana (Wu et al., 1996; Liu & Zhu, 1998; Halfter et al., 2000; Shi et al., 2000; Qiu et al., 2004). In the SOS pathway, a salt-induced rise in [Ca2+]i is translated into enhanced activity of a Na+/H+ antiporter (SOS1) via a calcineurin-B-like (CBL) protein (SOS3, aka CBL4) and a CBL-interacting protein kinase (SOS2, aka CIPK24). Over recent years it has become clear that CBLs and CIPKs are ubiquitous functional modules that are involved in many signalling pathways, including those regulating nutrient transport. Thus a forward screen for growth of A. thaliana plants on low K+ concentrations identified a regulon consisting of CBL1 (or CBL9), CIPK23 and AKT1 (Xu et al., 2006). Knockout mutants for CIPK23 (lks1) and double knockout mutants for the CBLs (cbl1cbl9) mimicked the akt1 phenotype (hypersensitivity to low K+ in the presence of ammonium; Hirsch et al., 1998) and displayed reduced K+ uptake compared with wild-type plants. Furthermore, co-expression of CBL1 and CIPK23 with AKT1 in Xenopus oocytes resulted in measurable K+ inward currents, while AKT1 alone did not produce currents in this expression system (Li et al., 2006; Xu et al., 2006). Based on these findings, the response of plants to low K comprises the following steps (Fig. 5). A decrease in the external K+ concentration hyperpolarizes the membrane potential and leads to Ca2+ influx through hyperpolarization-activated Ca2+ channels (Allen et al., 2001). Upon a rise of [Ca2+]i, CBL1 or CBL9 binds to CIPK23 and recruits it to the plasma membrane (in analogy to the SOS pathway; Quintero et al., 2002; Quan et al., 2007). Here the kinase domain of CIPK23 interacts with the C-terminal ankyrin domain of AKT1, thereby phosphorylating and activating the channel (Li et al., 2006; Xu et al., 2006). Dephosphorylation with subsequent deactivation is achieved by the phosphatase AIP1 (Lee et al., 2007). Two-hybrid analysis showed that several other members of the relevant gene families interact with each other, thereby creating a large network of CBL/CIPK/AKT1 complexes (Lee et al., 2007). However, which of the potential protein combinations are co-expressed and active in a particular cell type or condition in planta is unknown. For example, SOS2 specifically interacts with CBL4 (SOS3) in roots and CBL10 (SCABP8) in shoots (Quan et al., 2007). In addition to CIPK23, another CIPK homologue, CIPK9, has been found to play a role in plant adaptation to K+ deficiency (Pandey et al., 2007). cipk9 knockout plants show reduced growth on low K but unlike lks1 they do not differ from wild-type plants with respect to K+ content and K+ influx. One possibility to explain this phenotype is that CIPK9 interacts with vacuolar K+ transporters facilitating K+ release from the vacuole (e.g. TPK1; Gobert et al., 2007), thereby assisting cellular K+ homeostasis under K+ deficiency (Amtmann & Armengaud, 2007). Targets of CBL/CIPK regulation also include the vacuolar transporters NHX1 and CAX1, both of which interact with SOS2 (Cheng et al., 2004; Qiu et al., 2004). NHX1 is a Na+/H+ antiporter, which under certain conditions also transports K+ (see previous section on Calmodulin), while CAX1 is a Ca2+/H+ antiporter. Both are likely to play a role in cellular and whole-plant homeostasis of K+ and Ca2+.
Membrane transporters are often assembled from several subunits. These can either be identical (homo-oligomers) or different (hetero-oligomers). Both homo- and hetero-oligomerization can provide a means for regulation of ion transport. Trans-activation within a homomeric complex of three AMT1;1 subunits has recently been discovered through functional analysis of structural mutants based on the crystal structure of a homologous ammonium transporter from Archaeoglobus fulgidus (Loque et al., 2007). Each of the AMT subunits provides a functional pore for ammonium. The study identified specific sites within the soluble C-terminus and the pore that are required for allosteric regulation of the homomer, and provided evidence that post-translational modification (e.g. phosphorylation) of the C-terminus of a single monomer leads to a conformational change resulting in cooperative closure of all three pores in the complex. The authors suggest that this mechanism allows rapid inactivation of the multipore complex to protect against over-accumulation of potentially toxic ammonium at high external ammonium concentrations or during depolarization.
While trans-activation within homomers allows for rapid responses of a transport pathway, differential regulation of individual subunits within a heteromer has the potential to introduce a high degree of plasticity into ion transport within a single cell. In fact, regulation of AMT transporters may use both mechanisms, as several isoforms of the AMT1 family can interact with each other (Ludewig et al., 2003; Loque & von Wiren, 2004; Neuhauser et al., 2007). The best-studied example of heteromeric protein assembly concerns K+ channels of the Shaker family. Functional channels consist of four α-subunits, which in the simplest case form homo-tetramers. Biochemical experiments and yeast two-hybrid studies have revealed that interaction between the individual subunits involves three C-terminal domains. The KHA domain at the extreme C-terminus cross-interacts with a region just downstream of the hydrophobic core, and the putative cyclic nucleotide-binding domain interacts with itself (Daram et al., 1997).
The possibility of functional heteromerization between different α-subunits was uncovered by co-expression of different plant Shaker channel mRNAs in Xenopus oocytes. Dreyer and co-workers (Dreyer et al., 1997) showed that the currents produced by co-expression could not be explained by simply adding homomeric channel currents. Analysis of expression patterns in plants shows that many tissues express at least two types of Shaker channels previously shown to interact in heterologous systems (Dreyer et al., 1997; Baizabal-Aguirre et al., 1999; Pilot et al., 2001, 2003; Zimmermann et al., 2001), thus providing the opportunity for the formation of hybrid channels in planta. For example, KAT1 and KAT2 are both expressed in guard cells and interact in heterologous expression systems (Pilot et al., 2001). Formation of KAT1/2 heteromers provides a good explanation for the apparently contradictory observations that over-expression of mutant KAT1 channels affects stomatal function (Kwak et al., 2001) while KAT1 knockout does not (Szyroki et al., 2001). However, in this as in most other cases, in vivo co-localization of the subunits to the same membrane is still elusive.
The question remains whether heteromerization provides a physiological means for regulation. An interesting case is AtKC1, which does not form functional homo-tetrameric channels in heterologous expression systems, but appears to act as a modulator of AKT1 currents (Reintanz et al., 2002; Duby et al., 2008). Both genes are expressed in root hairs but knockout of AKT1 alone is sufficient to completely abolish K+ inward current, thus confirming AtKC1 as a ‘silent channel’. Disruption of the AtKC1 gene does, however, strongly suppress the inward current and alters its Ca2+ and pH sensitivity (Reintanz et al., 2002). These findings, together with recent evidence that AKT1 and AtKC1 preferentially associate in vivo (Duby et al., 2008), indicate that AKT1/AtKC1 heteromers underlie physiological K+ uptake by root hairs.
Finally, transmembrane ion and solute transport is subject to regulation via membrane traffic, if only through its impact on the population of transport proteins available at the membrane surface. Thus, exocytosis and endocytosis of ion and solute transporters serve to control transport capacity, albeit not necessarily the intrinsic kinetic characteristics for transport across the membrane. Among mammalian cells, the best-characterized model is that of Na+-coupled glucose transport via the GLUT4 transporter which cycles between the apical membrane and a pool of cytosolic vesicles in intestinal epithelial cells (Simpson et al., 2001; Ishiki & Klip, 2005). Insulin stimulates the exocytosis of GLUT4 and, as a consequence, stimulates the rate of glucose uptake. Fusion of GLUT4 vesicles depends on a number of membrane trafficking proteins, including so-called SNARE complexes of mammalian SNAP-23, Syntaxin 4 and VAMP2 proteins (Volchuk et al., 1996; Chamberlain & Gould, 2002; Williams & Pessin, 2008). In turn, GLUT4 transporters are recovered from the apical plasma membrane by endocytosis and sequestered in specialized GLUT4 vesicles before recycling.
There is no doubt that membrane targeting of nutrient transporters is important for nutrition. For example, a mutation in an SEC12-type phosphate traffic facilitator (PHF1) impairs P transport, resulting in the constitutive expression of many P starvation-induced genes (Gonzalez et al., 2005). Nitrate uptake through AtNRT2.1 depends on interaction of AtNRT2.1 with the smaller AtNAR2 protein, which carries an N-terminal ‘secretory pathway signal’ (Orsel et al., 2006). The observation that in nar2 knockout mutants green fluorescent protein (GFP)-tagged AtNRT2.1 protein is not correctly targeted to the plasma membrane suggests a role of NAR2 in trafficking (Orsel et al., 2007). The interesting question is whether reversible recruitment of transporters to the target membrane can be a means for adjusting nutrient uptake to nutrient availability and demand.
In A. thaliana, a number of integral membrane proteins have now been found to traffic to, and be recovered from, the plasma membrane (e.g. KOR1 (Robert et al., 2005), PIN1 (Geldner et al., 2001), BRI1 (Geldner et al., 2007) and FLS2 (Robatzek et al., 2006)). Of particular interest in this context is the boron transporter BOR1, which is essential for xylem loading and boron translocation to the shoot under nutrient limitation. Boron resupply leads to BOR1 endocytosis and degradation in the vacuole (Takano et al., 2005). Each of these examples entails traffic characterized to varying degrees by changes in the constitutive turnover of the integral membrane protein. Unlike GLUT4 traffic, however, on endocytosis these plant proteins enter a one-way path that leads to their sequestration in the vacuole and degradation.
Traffic of the Kv-like K+ channel KAT1 presents a different picture. Turnover of KAT1 at the plasma membrane of intact epidermal and guard cells is tightly controlled through a mechanism evoked by ABA and leads to recycling in true exchange with an endomembrane pool distinct from known degradation pathways to the vacuole (Sutter et al., 2006). The close parallel with GLUT4 traffic and its role in transmembrane solute transport is self-evident. Furthermore, studies using dominant-negative (so-called Sp2) fragments of SYP121, a plasma membrane SNARE previously shown to have a role in guard cell ion channel control (Leyman et al., 1999), have indicated that export of KAT1 to the plasma membrane is dependent on SYP121 function. Sutter and co-workers (Sutter et al., 2007) found that co-expression of the SYP121 Sp2 fragment selectively suppressed KAT1 delivery to the plasma membrane and altered its distribution and anchoring within microdomains in the plasma membrane.
Recent work from the same laboratory has yielded direct evidence for SYP121 as a key structural element determining the gating of another K+ channel (Honsbein et al., 2007). Significantly, the interaction was found to be essential for K+ channel gating and K+ uptake in the A. thaliana root. Thus, the SNARE appears to be essential for channel-mediated K+ nutrition, a function wholly distinct from any role in membrane traffic. It will be of interest now to determine whether similar SNARE interactions contribute directly to other ion transport, signalling and homeostatic functions in plant mineral nutrition.
IV. Conclusions and outlook
Transport proteins mediate ion fluxes across biological membranes that underlie nutrient acquisition, cell volume changes and transpiration. Regulation of ion transporters is essential to adjust these parameters to plant development and environmental challenges. The combination of electrophysiological, biochemical, molecular and genetic approaches has created a wealth of information on responses of macronutrient transporters to environmental stimuli, the signalling pathways leading to these responses and the molecular mechanisms involved. Many studies have provided evidence for the involvement of signalling molecules such as Ca2+, cyclic nucleotides, ROS and nitric oxide, hormones such as ABA, ethylene and cytokinin, and regulatory proteins such as kinases and phosphatases, calmodulins and 14-3-3 proteins in these processes. Nevertheless there are still substantial gaps in our understanding of the molecular mechanisms that link signalling pathways to transporter abundance and activity. The data collated in this review seem to suggest that the transport of nitrate, phosphate and sulphate is primarily regulated at the transcript level, whereas transport of K+ and Ca2+ is predominantly regulated at the protein level. Because anionic nutrients are taken up by proton co-transporters while cationic nutrients in most situations enter the cells passively through ion channels, this could indicate different strategies for regulating active and passive transport. Indeed, responsiveness at the transcript level to external nutrient supply was observed for genes encoding active transporters for K+ and Ca2+ (e.g. HAK5 and CAX; Maathuis et al., 2003; Armengaud et al., 2004). However, it is also possible that the mechanisms of regulation observed for different nutrient transporters merely reflect preferences for certain experimental approaches within research communities (e.g. mineral nutrition vs guard cell biology). Several recent publications indicate that post-translational mechanisms are also important for regulation of N, P and S transporters, and this line of research is now gaining momentum (Liu & Tsay, 2003; Orsel et al., 2006; Loque et al., 2007). Surprisingly, there is hardly any information on Mg transport and its regulation (Schock et al., 2000). Another area that still awaits progress is the identification of the receptor elements that perceive nutritional stimuli, and in many cases the exact nature of the primary stimulus itself remains to be characterized. Active research in the area of transport regulation will continue to fill these gaps, supported by the increasing willingness of scientists to investigate interdependences between different nutrients, signalling pathways and physiological processes. The success of this research relies on continuous improvements in experimental techniques. In particular, interaction screens and phospho-proteomics with membrane proteins (Nuhse et al., 2007; Lalonde et al., 2008), fluorescent indictors for a number of key ions, metabolites and signalling molecules (Looger et al., 2005; Belousov et al., 2006; Gu et al., 2006) and tools for studying recombinant proteins and their interactions in planta (Walter et al., 2004; Capanoni et al., 2007) are likely to uncover much-sought information on in vivo cellular and molecular events involved in the regulation of nutrient transporters under various environmental and nutritional conditions. Identification of the regulatory elements for nutrient uptake will not only increase our understanding of how plants adapt to conditions of nutrient shortage but also provide potential targets for future bioengineering efforts aimed at improving crop performance on marginal soils.