HXK1 as a Glucose Sensor

Glucose is a universal carbon and energy source, by far preferred by organisms ranging from unicellular microbes to plants and animals. While sucrose is the most important transport sugar in plants, most regulatory effects can be ascribed to glucose. In many organisms, HXKs and other sugar kinases are the most ancient, conserved sugar sensors (Johnston, 1999; Stulke and Hillen, 1999; Rolland et al., 2001; Rolland et al., 2006; Claeyssen and Rivoal, 2007). A regulatory role for plant HXKs in plant sugar sensing was first suggested by the marked effects of various sugars on photosynthetic and glyoxylate cycle gene expression in Chenopodium and cucumber cell cultures (Krapp et al., 1993; Graham et al., 1994) and maize mesophyll protoplasts (Sheen, 1990). The reduction in photosynthesis related gene expression by low amounts of glucose (1–10 mM) but not glucose-6-phosphate (G6P) in a maize protoplast transient expression system, supported a signaling role for HXK (Jang and Sheen, 1994). Other substrates for HXK phosphorylation, like fructose, mannose and 2-deoxyglucose (2-dGlc) also trigger repression. Moreover, the observed repression can be blocked by the HXK-specific competitive inhibitor mannoheptulose (Jang and Sheen, 1994). Conversely, sugars that cannot be transported into the cell (L-glucose), or that are taken up but cannot be phosphorylated by HXK (6-deoxyglucose and 3-O-methylglucose), as well as intermediates of glycolysis and sugar phosphates (delivered into the protoplasts by electroporation), do not cause the same repression (Jang and Sheen, 1994). The possibility that depletion of Pi and ATP caused the reduced gene expression is excluded by the inability of P_i and ATP to relieve the glucose repression (Graham et al., 1994; Jang and Sheen, 1994).

The hypothesis that HXK acts as a glucose sensor was further substantiated by the characterization of transgenic Arabidopsis HXK sense and antisense lines (Jang et al., 1997). Wild type Arabidopsis seedlings grown under continuous light on MS medium containing 6% (approximately 333 mM) glucose exhibit developmental arrest with repression of cotyledon greening and expansion, leaf development, and hypocotyl and root elongation. Plants carrying the antisense construct of HXK1 are hyposensitive to glucose repression, while plants overexpressing HXK1 are hypersensitive (Fig. 2). No apparent differences are observed with the same concentration of mannitol or 3-O-methylglucose as osmotic controls (Jang et al., 1997). Further evidence for HXK1 as a sugar sensor is provided by the isolation of two gin2 (glucose insensitive 2) mutants mapped to the HXK1 gene (Moore et al., 2003). Importantly, two catalytically inactive HXK1 alleles (G104D and S177A) can mediate developmental arrest on 6% glucose and glucose repression of chlorophyll accumulation and photosynthetic gene expression in the gin2 mutant background. These results unambiguously show that HXK1 is a true glucose sensor in plants and that glucose signaling can be uncoupled from glucose metabolism (Moore et al., 2003).

Figure 2.

Arabidopsis WT, glucose insensitive (gin) and glucose oversensitive (glo) seedling growth phenotypes on MS medium containing high (6%) exogenous glucose levels. Wild type seedlings are affected in cotyledon greening and expansion and show reduced hypocotyl and root elongation.

The gin2 mutants provide a valuable tool to investigate the physiological importance of glucose sensing and signaling during normal plant growth, without the manipulation of exogenous sugars. In wild type plants, high light conditions (200–300 μE) can boost photosynthesis and sugar production, leading to accelerated growth with earlier leaf senescence. However, gin2 plants remain small and dark green and show little cell expansion, supporting the role of HXK1 in promoting cell expansion and growth in roots, leaves and inflorescences under enhanced light conditions (Moore et al., 2003). In transgenic tomato plants, displaying higher sugar sensitivity than Arabidopsis plants, HXK1 overexpression inhibits growth and chlorophyll accumulation and promotes senescence (Dai et al., 1999; Granot, 2007). Thus, HXK1-mediated signaling can promote or repress growth, depending on the intrinsic glucose levels and glucose sensitivity of the plants.

In the Arabidopsis genome, six HXK and HXK-like (HKL) genes can be found, serving a variety of physiological functions and likely controlled by tissue-specific expression patterns, sub-cellular localization and protein complex formation (Claeyssen and Rivoal, 2007; Granot, 2007; Karve et al., 2008). Proteomic and GFP fusion analyses indicate that several plant HXKs are located on the outer mitochondrial membrane, where a completely functional glycolytic metabolon can be found (Giege et al., 2003; Kim et al., 2006; Balasubramanian et al., 2007; Damari-Weissler et al., 2007; Granot, 2007; Karve et al., 2008). In addition, plant HXKs are found in plastids (Wiese et al., 1999; Olsson et al., 2003; Giese et al., 2005; Cho et al., 2006a; Kandel-Kfir et al., 2006; Granot, 2007). Unique in moss, the chloroplast stromal HXK accounts for 80% of the total hexose kinase activity and is responsible for glucose-mediated growth and formation of caulonemal filaments (Olsson et al., 2003). Several recent papers also report diverse functions for mitochondrial HXKs in processes ranging from programmed cell death to actin filament reorganization and glucose-mediated gene expression (Kim et al., 2006; Balasubramanian et al., 2007; Damari-Weissler et al., 2007). It will be interesting to determine how plant HXKs integrate glycolysis and cell death without the proapoptotic protein, associated with the mitochondrial glucokinase in mammalian cells (Danial et al., 2003). The functional significance of HXK interactions with porin, actin and voltage-dependent anion channels deserves further investigation (Kim et al., 2006; Balasubramanian et al., 2007).

HXK1 is also found in high molecular weight complexes in the nucleus (Cho et al., 2006b). Apparently, HXK1 regulates transcription by direct binding to the promoter of glucose-repressed genes. Chromatin immunoprecipitation experiments with HXK1 show that this complex specifically binds to cis-regulatory elements upstream of the CAB2 (chlorophyll a/b binding protein 2) and CAB3 coding regions. This nuclear activity requires two HXK1 unconventional partners, HUP1 and HUP2, components shown to be part of the plant vacuolar H⁺-ATPase (VHA-B1) and the 19S regulatory particle (RPT5B) of the proteasome complexes, respectively (Sze et al., 2002; Yang et al., 2004a; Cho et al., 2006b). Interestingly, both interaction partners have closely related homologues that do not interact with HXK1 in Arabidopsis (Cho et al., 2006b). In addition, vha-b1 and rpt5b mutants display similar seedling and adult plant phenotypes as observed for gin2 plants (Moore et al., 2003), confirming the vital role of their interaction with HXK1 in glucose signaling (Cho et al., 2006b)(Fig. 1).

The HXK-Independent Pathway

Analysis of sugar-regulated gene expression and growth of wild type, 35S-HXK1 and 35S-antiHXK1, and 35S-ScHXK2 (yeast Saccharomyces cerevisiae HXK2) Arabidopsis plants confirms the existence of both HXK1-dependent and -independent signaling pathways (Xiao et al., 2000). Interestingly, the effect of glucose on the expression of genes encoding AGP, CHS (chalcone synthase), PAL (phenylalanine ammonialyase) and ASN (asparagine synthetase) is similar in wild type plants and transgenics, indicating that this effect is independent of HXK1 (Xiao et al., 2000). Furthermore, seed germination can be inhibited by glucose in concentrations below those known to inhibit early seedling development. This inhibition operates on imbibed seeds and is independent of HXK function (Dekkers et al., 2004). Interestingly, the HXK-dependent and HXK1-independent pathways can cooperate in glucose regulation, acting at different levels. For example, glucose repression of the grape monosaccharide transporter VvHT1 expression is dependent on HXK1, while its post-transcriptional glucose regulation is not (Conde et al., 2006).

Since sugars are extensively metabolized, downstream metabolites are also potential signaling molecules. Glucose induction of two pathogenesis-related genes (PR1 and PR2) is higher in plants overexpressing HXK1 but absent in 35S-antiHXK1 plants, consistent with the requirement of HXK for signaling. However, this induction is enhanced both in plants overexpressing HXK1 and plants overexpressing yeast ScHXK2, suggesting that the induction of PR1 and PR2 is dependent on HXK catalytic activity rather than on its regulatory function. This indicates that the regulatory signal is possibly generated by downstream metabolic intermediates (Xiao et al., 2000). Genome-wide expression analyses with wild-type and gin2 mutant plants will reveal more genes regulated by HXK1-dependent and HXK1-independent sugar signaling under various growth conditions.

Disaccharide Sensing

Sucrose is the main transported sugar in plants. Although sucrose can be substituted by the hexose products in many cases of sugar effects, a number of studies have provided compelling evidence for sucrose-specific regulation in gene expression and growth. Sucrose-induced expression has been reported for genes encoding UDP-glucose pyrophosphorylase, anthocyanin biosynthesis enzymes, and a putative chloroplast glucose-6-phosphate/phosphate translocator (Wenzler et al., 1989; Yokoyama et al., 1994; Ciereszko et al., 2001; Teng et al., 2005; Gonzali et al., 2006; Solfanelli et al., 2006). Conversely, sucrose-specific repression has been shown for the transcription of a proton-sucrose symporter gene in sugar beet (Chiou and Bush, 1998; Vaughn et al., 2002) and for carrot somatic embryo radicle elongation (Yang et al., 2004b). Also, the marked effects on leaf shape, caused by the overexpression of the homeodomain leucine zipper transcription factor gene ATHB13, are specifically induced by sucrose (Hanson et al., 2001).

Specific disaccharide signaling, independent of metabolism, is also demonstrated using non-metabolizable sucrose analogs, such as palatinose and turanose. These analogs affect the expression of α-amylase in barley embryos and AGP in potato (Loreti et al., 2000; Tiessen et al., 2003), and induce starch synthesis and sucrose degradation in potato tubers (Fernie et al., 2001). However, they may activate novel pathways different from sucrose-signaling as they could be perceived as a more general stress-related signal rather than a specific non-metabolizable sucrose substitute (Sinha et al., 2002; Roitsch et al., 2003).

Another line of fascinating research suggests that sucrose regulates the translation of the ATB2 protein. The ATB2 gene encodes a group S basic region leucine zipper transcription factor (bZIP11), characterized by an unusually long 5′ untranslated region (UTR). Point mutations and deletions in this 5′ UTR have shown that a highly conserved small upstream open reading frame (uORF) encoding 42 amino acids is responsible for sucrose induced repression of translation. This uORF is also conserved in the 5′ UTRs of other S-class bZIP transcription factors of Arabidopsis and many other plant species. Expression of a 5′ UTR-GUS (β-glucuronidase) fusion construct throughout the whole plant, indicates that this sucrose regulation mechanism is functional in all plant tissues (Rook et al., 1998; Wiese et al., 2004, 2005). To demonstrate the specificity of sucrose in its repression mechanism, several other sugars, such as glucose, fructose, galactose, maltose and raffinose, were tested, but none induces repression of translation to the same extend as sucrose. Intriguingly, ATB2 transcription is induced both by sucrose and glucose (Wiese et al., 2004), providing a mechanism of rigorous fine-tuning of ATB2 protein levels in response to sucrose (Fig. 1). Sucrose may also act as a signal, controlling specific developmental processes not affected by hexose sugars. In Vicia faba embryos, high hexose levels are associated with active division during early embryo development, while sucrose accumulation is correlated with the switch from active cell division to cell expansion, differentiation and reserve (starch) accumulation (Weber et al., 1996). It will be important to uncover how the sucrose signal is perceived and transduced to regulate the diverse responses.

Cell Surface Receptors

In yeast and mammals, extracellular sucrose and glucose can be perceived by G-protein-coupled-receptors (GPCRs) linked to heterotrimeric guanine nucleotide-binding proteins (G-proteins) (e.g. Lemaire et al., 2004; Chandrashekar et al., 2006). In such a system, the extracellular signal molecule (ligand) specifically binds to the GPCR, resulting in a conformational change that stimulates the exchange of GTP for GDP at the guanine-nucleotide binding site on the interacting G-protein. This initiates intracellular signaling by dissociation of the α- and the β/γ-subunits and their interaction with a variety of downstream effectors. G-proteins act as molecular switches and intrinsic GTPase activity catalyses the return to the inactive heterotrimer. In many cases, GTPase activity and desensitization is accelerated by a new class of regulators of G-protein signaling (RGS) proteins (Srinivasa et al., 1998).

One Arabidopsis gene encodes an unusual hybrid seven-transmembrane domain protein with a C-terminal RGS domain, RGS1 (Chen et al., 2003). The rgs1 mutants are insensitive to 6% glucose, while overexpressors are hypersensitive. Growth assays with different sugars and sugar analogs suggested that RGS1 may function as a glucose sensor, independently of HXK (Chen and Jones, 2004)(Fig. 1). Glucose appears to alter the interaction of RGS1 with GPA1 (the only Gα subunit in Arabidopsis) in vivo, leading to an increased GTP hydrolysis activity of GPA1 and sugar-specific signaling events (Johnston et al., 2007). In addition, Arabidopsis gpa1 mutants are hypersensitive to inhibition of germination by sugar and ABA, confirming their potential role in sugar signaling (Ullah et al., 2002; Pandey et al., 2006). Interestingly, THF1 (THYLAKOID FORMATION1), a protein found in the outer membrane and stroma of plastids, interacts with GPA1 in root hair cells when plasma membrane and plastid membrane are in close proximity (Huang et al., 2006). The thf1 mutant seedlings show altered sensitivity to glucose and THF1 protein levels are rapidly degraded by glucose, suggesting a role for this protein in the sugar response (Huang et al., 2006)(Fig. 1).

Besides GPA1, three Arabidopsis genes encode one Gβ (AGB1) and two Gγ subunits (AGG1 and AGG2)(Weiss et al., 1994; Mason and Botella, 2000; Jones and Assmann, 2004). A loss of function mutation in the Gβ subunit (agb1) affects multiple processes, including hypersensitivity to glucose. Although the gpa1 and agb1 mutants often have opposite phenotypes, they seem to share the same function in glucose signaling (Ullah et al., 2003; Wang et al., 2006; Temple and Jones, 2007). Interestingly, a mutation in a Golgi-localized hexose transporter (sgb1, suppressor of G beta1) can reverse glucose hypersensitivity in the agb1 mutants (Wang et al., 2006). In addition, agg1 and agg2 mutants are also hypersensitive to glucose (Trusov et al., 2007). As agg1 mutants are sensitive to mannitol and more susceptible to infection, these proteins are likely involved in different cellular processes (Trusov et al., 2007).

In budding yeast, novel hexose transporter-like genes, SNF3 and RGT2, encode extracellular glucose sensors that are responsible for the regulation of functional hexose transporter gene expression. These catalytically inactive transporters possess a long cytoplasmic domain for interaction with downstream signaling components (Johnston, 1999; Johnston and Kim, 2005). The plant disaccharide transporter SUT2, containing a similar extended cytoplasmic N-terminal and central loop domain, has been proposed to be a putative sucrose sensor (Lalonde et al., 1999; Barker et al., 2000). In addition, some monosaccharide transporters also have extended cytoplasmic loops (Kleinow et al., 2000). Further research is required to demonstrate that these putative receptors are actively involved in sugar sensing.

Sugar-Signaling Mutants

Two major strategies have been used to screen for sugar signaling mutants in Arabidopsis. A simple approach is based on the effect of high levels of external sugars on seedling growth and development. For example, gin (glucose insensitive) mutants develop normally on MS medium supplemented with 6% (approximately 333 mM) glucose (Zhou et al., 1998), while sis (sugar insensitive) and sig (sucrose insensitive growth) mutants are selected on media containing 300 mM and 350 mM sucrose, respectively (Laby et al., 2000; Pego et al., 2000)(Fig. 2). Conversely, glo (glucose oversensitive), gss (glucose-supersensitive), and sss (sucrose-supersensitive) mutants show developmental arrest in the presence of otherwise non-inhibiting sugar concentrations (Table I) (Smeekens, 2000; Rolland et al., 2002; Leon and Sheen, 2003; Gibson, 2005). Remarkably, many mutants isolated in the different screens are allelic to mutants in hormone biosynthesis and signaling. Since these screens are done with seedlings, the information on sugar-hormone interactions in the signaling networks is necessarily limited to the early developmental stages.

To ensure uniform sugar responses, the initial mutant isolation screens were mostly performed on media containing high glucose or sucrose concentrations, raising concerns about the physiological relevance of sugar regulation. Moreover, the phenotypes could also be influenced by osmotic stress and other nutrients in the media. The analyses of wild-type and gin2 seedlings show that the presence of high nitrate concentrations (40 mM nitrate) in full MS medium can antagonize the glucose effects. When the nitrate effect can be minimized, the sugar effect can be observed at more physiological concentrations (Martin et al., 2002; Moore et al., 2003; Cho et al., 2006b). Indeed, 2% (110 mM) glucose without MS medium is sufficient to cause glucose repression of RBCS, CAB and CAA (carbonic anhydrase) gene expression and chlorophyll accumulation in whole seedlings (Moore et al., 2003; Cho et al., 2006b). For HXK1-mediated growth promotion in seedlings, 0.2% (11 mM) glucose in a 0.1× MS medium is sufficient to provide the glucose signal requirement in vivo (Moore et al., 2003; Cho et al., 2006b). In isolated maize and Arabidopsis mesophyll protoplasts, concentrations between 1 to 20 mM are sufficient to cause specific glucose but not sucrose repression in various assays (Jang and Sheen, 1994; Moore et al., 2003; Yanagisawa et al., 2003). Recent microarray studies show that 15 mM sucrose can modulate the expression of a large number of Arabidopsis genes in carbon-deprived seedlings grown in liquid culture (Osuna et al., 2007). Interestingly, changes in light intensity are correlated with endogenous glucose levels and plant growth phenotypes (Moore et al., 2003). The use of mutants (e.g. pgm, phosphoglucomutase) with different endogenous sugar levels has further substantiated the physiological relevance of the sugar sensing and signaling network in Arabidopsis (Bläsing et al., 2005; Gonzali et al., 2006). It would be interesting to examine the effects of pgm at various endogenous sugar levels without the presence of high nitrate (Bläsing et al., 2005).

Other more elaborate mutant screens are based on altered responses to sugar repression or activation of gene expression in transgenic reporter lines. For example, the sun (sucrose uncoupled) mutants show high activity of a sucrose-repressed plastocyanin (PC) promoter fused to a luciferase reporter in the presence of 3% sucrose (Dijkwel et al., 1997). The isi (impaired sucrose induction) mutants loose their ability to induce ApL3 gene expression on sucrose-containing medium. The regulatory sequence of the ApL3 gene was fused to a negative selection marker. Hence, seedlings surviving this screen are impaired in sucrose induction of the starch biosynthesis gene promoter (Rook et al., 2001). Both sugar insensitive and hypersensitive mutants are also obtained based on the sugar-inducible expression of a β-amylase gene (Mita et al., 1997a; Mita et al., 1997b). Other screens include the use of the sugar inducible patatin promoter in front of a GUS reporter gene (rsr, reduced sucrose response mutants) (Martin et al., 1997) or the ApL3 promoter in front of a luciferase reporter gene (hsr, high sugar response mutants) (Baier et al., 2004) (Table I).

Complex sugar signaling mutants are also identified in unrelated genetic screens. For example, the hypersenescence1 (hys1) mutant was isolated based on an early senescence screen and displays a sugar hypersensitive phenotype (Yoshida et al., 2002). Analyses of the double mutant hys1 gin2 reveal HXK1-dependent action of HYS1 in glucose-mediated developmental arrest but HXK1-independent function of HYS1 in senescence. The hys1 mutant is allelic to cpr5 (constitutive expression of pr genes5) with elevated salicylic acid (SA) levels. HXK1 and HYS1 act synergistically in the negative regulation of SA accumulation (Aki et al., 2007). Another unique sugar mutant, prl1 (pleiotropic regulatory locus1), displays complex hormone and sugar response phenotypes and is recently found to exhibit severe defects in plant defense (Nemeth et al., 1998; Palma et al., 2007). PRL1 interacts with the C-terminal region of the Arabidopsis SnRK1 protein kinases KIN10 and KIN11, acting as a negative regulator (Bhalerao et al., 1999). This C-terminal part of KIN10 and KIN11 also interacts with both the SCF ubiquitin E3 ligase subunit SKP1/ASK1 and the SKP1/ASK1-binding 26S proteasome (Farras et al., 2001). In addition, PRL1 is associated with CUL4, a subunit of the E3 ligase, in vivo. KIN10 degradation is slower both in the prl1 and cul4cs mutants, suggesting a link between SnRK1 and the ubiquitin/26S proteasome complex (Lee et al., 2008). Intriguingly, the glucose hypersensitive phenotypes found in the dark-grown hsr8 (high sugar response 8) mutant are suppressed by prl1, indicating that PRL1 acts as a positive regulator for the specific glucose effects observed in etiolated hsr8 seedlings. Under light, hsr8 does not show overt phenotypes whereas prl1 displays hypersensitive glucose repression in developmental arrest (Li et al., 2007), suggesting complex effects of glucose regulation in different growth conditions. Map-based cloning reveals that HSR8 is allelic to MUR4, encoding an enzyme involved in arabinose synthesis. The studies indicate that failure of arabinose synthesis and consequent changes in cell wall composition could have a strong impact on sugar signaling (Li et al., 2007). It remains an important challenge to link complex sugar phenotypes to specific protein-protein interactions and signal transduction pathways.

Interplay Between Abscisic Acid and Sugar Signaling

A first indication for crosstalk between sugar and ABA came from the observation that several glucose insensitive mutants (gin1/isi4/sis4 and gin5/isi2/sis3) (Table II) are allelic to ABA biosynthesis mutants (aba2 and aba3) with lower endogenous ABA levels and reduced seed dormancy (Arenas-Huertero et al., 2000; Laby et al., 2000; Rook et al., 2001; Cheng et al., 2002) (Table II). The extensive crosstalk is further corroborated by the inability of glucose to repress the CAB1 and plastocyanin (PC) genes in the gin1/aba2 and gin5/aba3 backgrounds. Also, overexpression of HXK1 in the gin1 and gin5 mutants does not alter glucose-insensitivity, indicating that ABA acts downstream of HXK1 (Zhou et al., 1998; Arenas-Huertero et al., 2000)(Fig. 3). Interestingly, ABA accumulation and transcript levels of several ABA biosynthesis genes are significantly increased by glucose, providing a mechanistic link at the molecular level (Cheng et al., 2002). Except for ABA2, which is specifically upregulated by glucose but not by ABA (Seo and Koshiba, 2002), several ABA biosynthesis genes (ABA1, ABA3, NCED3 and AAO3) show a synergistic effect of glucose and ABA in their induction (Cheng et al., 2002). While expression of the APL3 gene is not induced by ABA (Rook et al., 2001), its sucrose-induction is significantly enhanced by ABA, indicating that sugar responses are dependent on an ABA-controlled responsiveness of the tissue (Rook and Bevan, 2003). It is suggested that, through induction of ABA synthesis, high concentrations of sugars would make the seedlings also more responsive to the sugar signal (Rook and Bevan, 2003). More studies are needed to reveal whether ABA biosynthesis is directly involved in the sugar signal transduction cascade or indirectly stimulated by sugars, and modulating sugar-responsiveness. Since many of these mutants were isolated in screens on high amounts of glucose, glucose-specific (likely ABA-mediated) osmotic effects should also be considered.

Figure 3

Overview of the elaborate genetic interactions between glucose and hormone signaling in Arabidopsis thaliana.

Glucose signaling mediated by HXK1 leads to increased ABA levels and induces both ABA synthesis (ABA1-3 and AAO3 gene expression) and ABA (ABI3-5, ABI8, ABF2-mediated) signaling to control seedling development. Other ABA-insensitive signaling mutants (abi1-1, abi2-1) do not exhibit the gin phenotype, suggesting distinct ABA signaling pathway for Glc signaling and the vegetative stress response. The glucose insensitive (gin) phenotype is mimicked by ethylene precursor treatment of wild-type plants and is also displayed in constitutive ethylene biosynthesis (eto1) and constitutive ethylene signaling (ctr1) mutants, whereas the ethylene-insensitive mutants etr1-1, ein2, ein3 as well as mkk9 exhibit glucose hypersensitivity (glo phenotypes). Glucose and ethylene signaling converge on the regulation of the ETHYLENE INSENSITIVE3 (EIN3) TF to differentially modulate its protein stability. The bifurcate and antagonistic CTR1 and MKK9 pathways are both critical in determining ethylene-signaling specificity through two MAPK phosphorylation sites with opposite effects on EIN3 stability. In addition, HXK-signaling positively and negatively interacts with auxin and cytokinin signaling, respectively. More details on hormone and sugar interactions are presented in the text.

In addition to mutants in ABA biosynthesis, several sugar-response mutants are affected in ABA signaling. The gin6 mutant contains a T-DNA insertion in the promoter of the ABI4 (ABSCISIC ACID INSENSITIVE 4) gene (Arenas-Huertero et al., 2000). In screens for mutants insensitive to sucrose, sun6, sis5 and isi3 were found to be allelic to ABI4 (Huijser et al., 2000; Laby et al., 2000; Rook et al., 2001) (Table II). ABI4 encodes a transcription factor of the APETALA 2 (AP2) domain family, involved in seed development and germination (Finkelstein et al., 1998). ABI4 expression is induced by high glucose levels and this induction is abolished in the gin6 mutant (Arenas-Huertero et al., 2000; Arroyo et al., 2003). The gin6 mutant also shows insensitivity to ABA during seed germination, suggesting that cis-acting elements are required for the ABA and high glucose regulation of the ABI4 gene expression. Interestingly, the glucose-insensitive abi4 phenotype cannot be rescued by the addition of ABA, suggesting that ABI4 acts downstream of the ABA-mediated glucose signaling cascade (Arenas-Huertero et al., 2000)(Fig. 3). However, detailed analysis of glucose effects on seed germination shows that there is a differential sensitivity to different glucose levels at distinct developmental stages (Price et al., 2003). The same low glucose levels are inhibitory for seed germination but stimulatory for seedling growth and development (Yuan and Wysocka-Diller, 2006). In addition, ABA is not required for sugar-dependent induction of leaf senescence, although leaf senescence can be triggered independently by application of ABA and sugars (Pourtau et al., 2004). The complexity of these data suggests caution when analyzing tissue-and developmental stage-specific crosstalk between sugars and hormones.

In addition to abi4, mutation of ABI5, encoding a transcription factor belonging to a large basic leucine zipper (bZIP) gene family, similarly confers a glucose-insensitive phenotype, albeit more moderate (Arenas-Huertero et al., 2000; Finkelstein and Lynch, 2000; Laby et al., 2000). Conversely, overexpression of ABI5 and some of the closely related bZIP transcription factors (ABF2, ABF3 and ABF4) confers hypersensitivity to sugars (Brocard et al., 2002; Kang et al., 2002). All of these transcription factors bind to the ABA responsive cis-element ABRE (Finkelstein et al., 2005). However, abf3 and abf4 loss-of-function mutants are not insensitive to glucose, but display defects in salt and dehydration responses, suggesting that they may only play a minor role in glucose-mediated signaling (Kim et al., 2004a). Interestingly, the ABI5 protein is only expressed in vegetative tissue shortly after seed germination, consistent with a role in monitoring environmental conditions during early seedling development (Lopez-Molina et al., 2001; Brocard et al., 2002). Furthermore, ABI5 gene expression is activated by glucose in an ABA-dependent manner (Cheng et al., 2002), and, unlike ABI4, highly inducible by ABA and mannitol (Brocard et al., 2002). Still, both ABI4 and ABI5 respond to the glucose analog 2-deoxy-glucose, suggesting a common HXK-dependent expression regulation (Arroyo et al., 2003)(Fig. 3).

Two other ABA signaling components may also be involved in the sugar response. Overexpression of the transcription factor ABI3 leads to glucose hypersensitivity at the seedling stage (Finkelstein et al., 2002). Expression of ABI3 is induced by 6% glucose in an ABA-dependent manner, albeit to lower levels than ABI4 and ABI5 (Cheng et al., 2002), and loss-of-function abi3 mutants are insensitive to the glucose-induced inhibitory effect on seed germination (Yuan and Wysocka-Diller, 2006). Based on glucose-induced ABI3 expression, altered glucose-mediated gene expression in the abi3 mutants and glucose insensitive phenotypes of abi3 mutant alleles, it is concluded that ABI3 is an essential component in the regulation of glucose insensitivity (Dekkers et al., 2008)(Fig. 3). In addition, loss-of-function mutation of ABI8, encoding a small plant-specific protein with unknown function, confers highly stunted root growth, which can be partially rescued by glucose (Brocard-Gifford et al., 2004) (Fig. 3).

While several ABA-signaling transcription factors appear to be involved in sugar-mediated responses of early seedling development, two dominant ABA-insensitive mutants, abi1 and abi2, respond normally to high concentrations of external glucose (Arenas-Huertero et al., 2000). Considering the complex nature of ABA signaling (Finkelstein and Rock, 2002) and the ABA-glucose interactions (Leon and Sheen, 2003; Price et al., 2003; Yuan and Wysocka-Diller, 2006), more detailed molecular analyses will be necessary to determine the genetic and spatio-temporal network linking ABA with the mechanisms conducting sugar signaling.

Interaction with Ethylene Signaling

Interactions between the ethylene and sugar-signaling pathways were first suggested by the finding that the ethylene precursor 1-aminocyclopropane-1-carboxylic acid (ACC) prevents inhibition of cotyledon greening and expansion at high concentrations of glucose in wild type seedlings (Zhou et al., 1998; Gibson et al., 2001). In addition, the ethylene insensitive mutants etr1, ein2 and ein3 display glucose hypersensitivity (Zhou et al., 1998; Cheng et al., 2002; Yanagisawa et al., 2003). Conversely, the ethylene overproduction mutant (eto1) and the constitutive ethylene triple response mutant (ctr1/gin4/sis1) are glucose insensitive (Zhou et al., 1998; Gibson et al., 2001; Cheng et al., 2002)(Table II). However, consistent with the idea that the crosstalk between sugar and hormone-signaling pathways is strongly dependent on glucose levels and developmental stage, the ctr1 mutant is sensitive to low glucose-mediated seed germination (Price et al., 2003).

To further elucidate the role of GIN1 (later identified as ABA2) in the ethylene signal transduction cascade, etr1 gin1 and ein2 gin1 double mutants were generated and analyzed. These double mutants display a glucose insensitive phenotype, indicating that ethylene affects glucose signaling partially through ABA to promote germination and seedling development (Zhou et al., 1998; Ghassemian et al., 2000; Cheng et al., 2002). Moreover, protein stability of the transcription factor EIN3 (ETHYLENE INSENSITIVE 3), a key regulator in ethylene signaling acting downstream of ETR1, EIN2 and MKK9, is differentially regulated by glucose and ethylene (Chang et al., 1993; Alonso et al., 1999; Yanagisawa et al., 2003; Yoo et al., 2008). In maize and Arabidopsis mesophyll protoplasts, glucose enhances the proteasome-dependent degradation of EIN3 proteins in the nucleus. This glucose-induced degradation is dependent on HXK1, as it is no longer detected in gin2 mutants. Transgenic plants overexpressing EIN3 are insensitive to glucose, whereas ein3 mutants are hypersensitive (Yanagisawa et al., 2003). The degradation of EIN3 proceeds much faster in seedlings treated with glucose and the presence of the ethylene precursor ACC delays EIN3 degradation in Arabidopsis plants. The glucose hypersensitivity observed in etr1, ein2, ein6 and mkk9 mutants and the allelism of gin4 with ctr1 can also be explained by these findings (Rolland et al., 2002; Leon and Sheen, 2003; Yoo et al., 2008) (Fig. 1 and 3).

Other Interactions between Sugar and Hormone Signaling

The observation that gin2 mutants show growth defects under high light suggests a link between glucose signaling and hormones important in cell elongation, such as auxin, gibberellins and brassinosteroids (Moore et al., 2003). Tissue cultures assays show that gin2 mutants are indeed defective in auxin-induced cell proliferation and root formation. Since auxin concentrations are similar in wild type and gin2 hypocotyl, and exogenously applied IAA (indole-3-acetic acid) does not induce cell proliferation in gin2 hypocotyls, gin2 mutants are likely deficient in auxin signaling and/or uptake (Moore et al., 2003). Furthermore, the auxin-resistant mutants, axr1, axr2 and tir1, are insensitive to growth inhibition on 6% glucose. Remarkably, the catalytically inactive HXK1 alleles restore auxin-mediated cell proliferation in gin2 tissue cultures, demonstrating that glucose sensing by HXK acts synergistically with auxin signaling (Moore et al., 2003). Another aspect of sugar and auxin interactions is revealed by the hookless1 (hls1) mutant that enhances sugar-induced gene expression, which can be partially eliminated by IAA (Ohto et al., 2006).

The delayed leaf senescence found in gin2 plants links sugars with another plant hormone, cytokinin (Moore et al., 2003). In tissue culture, gin2 mutants are hypersensitive to cytokinins for shoot regeneration. The addition of cytokinins to seedlings grown on 6% glucose also eliminates the developmental arrest (Moore et al., 2003). Because cytokinins promote ethylene biosynthesis and ethylene counteracts the effects of glucose, the effect of cytokinins could be indirect. As cytokinins still promote greening in the ethylene insensitive mutants etr1 and ein2, ethylene and cytokinin act independently in response to glucose. Furthermore, constitutive cytokinin signaling transgenic plants can overcome the glucose repression response, supporting the antagonistic relationship between cytokinin and sugar signaling (Moore et al., 2003). A recent analysis of the hys1 mutant reveals that its hypersensitivity to glucose is HXK1-dependent but cannot be prevented by adding ethylene or cytokinin (Aki et al., 2007), suggesting broad HXK1 effects beyond ethylene and cytokinin signaling.

Several studies support a close link between sugars and hormones in the control of cell division and growth. Cyclin D3 (CycD3), important in the control of G₁ to S-phase transition during the cell cycle, is induced by cytokinins in the presence of sucrose (Riou-Khamlichi et al., 1999). In the absence of sucrose, addition of auxin or cytokinin no longer induces its expression. In contrast, sucrose alone is able to induce expression of this cyclin gene, suggesting that sucrose acts upstream of hormones or synergistically with hormones in regulating CycD3 expression (Riou-Khamlichi et al., 2000; Hartig and Beck, 2006).

Another hormone that interacts closely with sugars is gibberellin (GA). Glucose can affect GA-mediated α-amylase expression in barley embryos (Perata et al., 1997). Interestingly, several studies show that glucose and GA responses share the same promoter elements in rice (Morita et al., 1998; Chen et al., 2002). Three MYB transcription factors from rice bind the glucose and GA response element TATCCA, present in all promoters of α-amylase genes isolated from cereals, have been identified. All three MYBs are involved in the glucose-regulation and cooperate with a GA-regulated transcription factor to regulate expression in the absence of GA. This suggests that these three MYB transcription factors play roles in both glucose- and GA-mediated regulation of α-amylase gene expression (Lu et al., 2002). In addition, RGL2 (RGA-like) and SPY (SPINDLY), two negative regulators of GA signaling may be involved in glucose repression of seed germination as rgl2 and spy mutants germinate normally in the presence of inhibiting glucose concentrations (Yuan and Wysocka-Diller, 2006). Recently, it is shown that GA represses the sucrose activation of anthocyanin biosynthesis genes, whereas ABA and jasmonate promote the effect of sucrose (Loreti et al., 2008).

Finally, several recent studies also link SA and brassinosteroids (BR) with sugar signaling. The sugar hypersensitivity of the bls1 (brassinosteroid, light and sugar1) mutant can be repressed by exogenous BR application, suggesting crosstalk between BR and glucose signaling (Laxmi et al., 2004). The sucrose- and glucose-induced expression of the PR2 and PR5 genes (Xiao et al., 2000; Thibaud et al., 2004) are abolished in NahG mutants with reduced SA levels, suggesting a role for SA in sugar-induction of PR2 and PR5 expression (Thibaud et al., 2004).

Interactions between Sugars and Other Nutrients

Carbon and nitrogen metabolism are tightly linked and need to be coordinated in order to maintain proper plant growth and development. Carbohydrates provide energy and the carbon backbone for nitrate and ammonium assimilation in the biosynthesis of amino acids. Nitrogen is essential for incorporation in nucleic acids, amino acids and chlorophyll. The integration of sugar and nitrogen metabolism as well as sensing and signaling enables plants to respond to acute changes in internal and external nitrogen concentrations or carbon supplies by regulating the expression of genes involved in nitrogen transport, assimilation and metabolism (Coruzzi and Zhou, 2001; Lejay et al., 2003).

It has been clearly illustrated that Arabidopsis seedling growth, lipid mobilization and photosynthetic gene expression is profoundly influenced by the availability of sugars and nitrogen (Martin et al., 2002; Moore et al., 2003). Recently, the osu1 (oversensitive to sugar1) mutant, deficient in a putative methyl-transferase, was isolated based on enhanced anthocyanin accumulation and root inhibition under imbalanced sugar and nitrogen conditions (Gao et al., 2008). Interestingly, genes encoding diverse transporters for nitrogen and other nutrient uptake are up-regulated by sugars (Lejay et al., 1999; Lejay et al., 2003; Lejay et al., 2008). Several studies have demonstrated the genome-wide expression regulation by nitrogen (nitrate and ammonium), sugars and their various combinations (Wang et al., 2003; Palenchar et al., 2004; Price et al., 2004; Scheible et al., 2004; Gutiérrez et al., 2007). Although nitrogen and sugars each specifically regulate the expression of a subset of genes, sucrose or glucose are more potent signals, regulating the majority of the genome responses (Palenchar et al., 2004; Price et al., 2004), and nitrogen appears to play an important role in modifying sugar responses (Palenchar et al., 2004). A recent microarray study supports the idea that nitrogen and sugars are sensed through sugar- and nitrogen-responsive pathways in Arabidopsis roots, illustrating an interconnection at the level of signaling or metabolism (Gutiérrez et al., 2007).

An informative system to study the effects of nitrogen and sugars on plant growth and development is based on the formation of lateral roots since their number and position are strongly dependent on nutritional cues. High sugar concentrations repress lateral root formation, and the lateral root initiation1 (lin1) mutant shows lateral root initiation even in repressing environmental conditions (high external sucrose/low external nitrate) (Malamy and Ryan, 2001). The lin1 mutant is deficient in the high affinity nitrate transporter, NRT2. Since expression of NRT2 is induced by low exogenous nitrogen and by sucrose under low nitrogen conditions (Lejay et al., 1999), NRT2 is a negative regulator of lateral root development in the signal transduction pathway sensing environmental status (Malamy and Ryan, 2001; Little et al., 2005).

Another nutrient closely interacting with sugar signaling is inorganic phosphate (Pi). Pi is an essential component in many plant metabolites and plays an important role in energy balance and carbon assimilation. Since Pi is often limiting in soils, plants have evolved specific mechanisms to sense and adapt to Pi starvation. Interestingly, changes in internal Pi concentrations are often linked with altered source-sink distribution and carbohydrate content. Characterization of the phosphorus-deficient pho3 mutant (mutated in the SUC2 gene, encoding a major sucrose-proton symporter) shows that reduced levels of Pi in rosette leaves is associated with increased levels of sucrose, anthocyanins and starch (Zakhleniuk et al., 2001). In addition, knock-out of the MYB-related transcription factor PHR1, involved in the response to Pistarvation, results in altered Pi allocation between roots and shoots, and reduced levels of starch and sugars (Nilsson et al., 2007). Microarray analyses of Pi and/or sucrose treated Arabidopsis leaf segments show that nearly 150 genes are synergistically or antagonistically regulated by both stimuli (Müller et al., 2007). Interestingly, expression of several known sucrose-regulated genes including CHS and β-AMY (β-amylase) can be modulated by Pi starvation and vice versa, suggesting a tight crosstalk between Pi and sucrose-regulated gene expression (Müller et al., 2005; Müller et al., 2007). Sugars are essential in regulating Pistarvation mediated gene expression and root architecture, and it has been suggested that this response is partially mediated by HXK1 (Karthikeyan et al., 2007). However, the sugar regulation of CHS and β-AMY expression is HXK-independent (Müller et al., 2005). These studies demonstrate the complexity of crosstalk between sugar and phosphate signaling and more analyses will be necessary to reveal the molecular nature of their interactions. An overview of Arabidopsis genes mediating sugar sensitivity is summarized in Table III.

Transcription Cascades

One way for plants to respond to diverse sugar signals is to rapidly modulate gene expression. Using a variety of methods and assays, selected genes have been found to be specifically regulated by sugars (Sheen, 1990; Graham et al., 1994; Sheen, 1994; Mita et al., 1995; Koch, 1996; Rook et al., 1998; Xiao et al., 2000). Microarray technology now enables genome-wide identification of specific sets of sugar-regulated genes in Arabidopsis seedlings and adult leaves (Palenchar et al., 2004; Price et al., 2004; Bläsing et al., 2005; Li et al., 2006; Gutiérrez et al., 2007; Osuna et al., 2007). Unraveling the physiological functions of these Arabidopsis genes will bring new insight into our understanding of the sugar regulation and signaling network.

During the diurnal cycle, changes in sugar levels are accompanied by alterations in transcript levels (Gibon et al., 2004; Thimm et al., 2004; Bläsing et al., 2005). These alterations are strongly amplified in the starchless pgm mutant, which contains higher sugar levels than wild-type plants during the day and very low sugar levels at the end of the night, suggesting that the transcriptional changes observed during the diurnal cycle are caused by sugars (Bläsing et al., 2005). Using the pgm mutant, it was shown that these changes in sugar levels can even override circadian behavior (Usadel et al. 2008). Interestingly, the extensive gene expression reprogramming promoted especially by carbon depletion can be effectively blocked by the addition of sucrose or glucose (Contento et al., 2004; Gibon et al., 2004; Thimm et al., 2004; Bläsing et al., 2005; Baena-González et al., 2007). Genes involved in diverse metabolic pathways are activated or repressed by sugars. Importantly, many genes encoding transcription factors and signaling components are also highly regulated by sugars or sugar starvation, providing a molecular link of sugar signaling to multiple regulatory pathways and hormone signaling (Price et al., 2004; Li et al., 2006; Osuna et al., 2007; Baena-González et al., 2007). It is important to note that some of these microarray experiments were carried out in the presence of high nitrogen levels and may not reveal a comprehensive coverage of sugar-regulated gene expression (Martin et al., 2002; Moore et al., 2003). Future studies aim at understanding the functional relations and kinetics of sugar-regulated gene expression will faciliate the elucidation of the transcriptional cascades and signaling networks central to the control of plant growth and development.

Protein Kinases and Phosphatases

In all eukaryotic cell types, both protein kinases (PK) and phosphatases (PP) have established roles as central regulators of signal transduction. The first indication for the involvement of phosphatases in plant sugar signaling came from the use of PP1 and PP2A phosphatase inhibitors, mimicking sugar-induced repression of photosynthetic gene expression in maize protoplasts and Chenopodium cell cultures (Sheen, 1993; Ehness et al., 1997) and blocking sucrose-induced expression of genes encoding sporamin, β-amylase and the small subunit of AGP in sweet potato petioles (Takeda et al., 1994). Together with glucose- and stress-related stimuli, these PP inhibitors also activate the expression of a stress-inducible invertase and PAL genes, while rapidly activating a putative mitogen-activated protein kinase in Chenopodium cell cultures (Ehness et al., 1997). The differential effects of the PK inhibitor staurosporine on stress-and glucose-related responses suggest the involvement of different PKs in these processes (Ehness et al., 1997). Similar pharmacological approaches have also shown the involvement of distinct PPs in dark- and sugar starvation-induced DIN (DARK INDUCED) gene expression (Fujiki et al., 2000). Studies with Ca²⁺ channel blockers (LaCl₃), EGTA and calmodulin inhibitors have also shown the involvement of Ca²⁺ signaling in sugar induction of sporamin and β-amylase genes in sweet potato (Ohto and Nakamura, 1995). It is possible that Ca²⁺ signaling mediates protein phosphorylation (Boudsocq and Sheen, 2008) in some of the sugar signaling pathways. For example, sucrose can induce cytosolic calcium fluxes and increases expression of a plasma membrane-associated calcium-dependent protein kinase (CDPK) from tobacco (Iwata et al., 1998; Furuichi et al., 2001). The future challenges are to connect various sugar sensors and signaling pathways to the functions of specific PKs and PPs and sugar-regulated target genes.