Trehalose is both a compatible solute and a carbon source in microbes

Trehalose is an alpha-1,1 glucose disaccharide lacking reducing ends which ensures chemical stability (Crowe, 2007). Its mass is indistinguishable from that of common disaccharides such as sucrose, maltose, cellobiose and laminaribiose. Trehalose is generally quantified and detected by LC-PAD or after derivatisation, using GC-FID or GC-MS (Muller et al., 2001; Roessner et al., 2000). NMR and specific trehalases have been used to confirm the identity of the compound.

Trehalose is a primary metabolite that may be synthesized through at least five pathways; only one pathway appears present in plants

Recognised trehalose biosynthesis pathways have been previously reviewed (Avonce et al., 2006; Elbein et al., 2003) and are presented in Figure 1. In summary, the TPS/TPP pathway converts glucose-6-phosphate (G6P) and UDP-glucose (UDPG) to trehalose-6-phosphate (T6P) by trehalose-6-phosphate synthase (TPS). T6P is then cleaved to trehalose and inorganic phosphate by trehalose-6-phosphate phosphatase (TPP). In the TS pathway maltose is converted to trehalose by trehalose synthase (TS). In the TreY/TreZ pathway the terminal 1,6-linkage from a starch molecule is converted to an alpha 1,1 linkage by epimerization and then the terminal disaccharide is cleaved releasing trehalose. In the TreP pathway trehalose is synthesized from glucose-1-phosphate and glucose. Finally, In the TreT pathway trehalose is synthesized from ADP-glucose (ADPG), or UDPG, and glucose. Trehalose is cleaved into two moieties of glucose by specific trehalases (TreH pathway). Plants have been shown to encode enzymes capable of trehalose synthesis via the TPS/TPP pathway (Blazquez et al., 1998; Chary et al., 2007; Pramanik and Imai, 2005; Satoh-Nagasawa et al., 2006; Shima et al., 2007; Vogel et al., 1998). Plants also contain genes that encode trehalase (Aeschbacher et al., 1999; Muller et al., 2001), but lack genes with homology to enzymes from the other pathways. This is consistent with observations from other eukaryotes where the TPS/TPP pathway is dominant, although some fungi have genes for the TreP pathway.

Figure 1.

Known trehalose biosynthesis pathways. Enzymes are labeled in blue. The TPS/TPP pathway converts glucose-6-phosphate (G6P) and UDP-glucose (UDPG) to trehalose-6-phosphate (T6P) by trehalose-6-phosphate synthase (TPS). T6P is then cleaved to trehalose and inorganic phosphate by trehalose-6-phosphate phosphatase (TPP). In the TreY/TreZ pathway the terminal 1,6-linkage from a starch molecule is converted to an alpha 1,1 linkage by epimerization and then the terminal disaccharide is cleaved releasing trehalose. In the TS pathway maltose is converted to trehalose by trehalose synthase (TS). In the TreP pathway trehalose is synthesized from glucose-1 -phosphate and glucose. Finally in the TreT pathway trehalose is synthesized from ADP-glucose (ADPG), or UDPG, and glucose. Trehalose is cleaved into two moieties of glucose by specific trehalases (TreH pathway). Plants have been shown to encode enzymes capable of trehalose synthesis via the TPS/TPP pathway (Blazquez et al., 1998; Chary et al., 2007; Pramanik and Imai, 2005; Satoh-Nagasawa et al., 2006; Shima et al., 2007; Vogel et al., 1998). Plants also contain genes that encode trehalase (Aeschbacher et al., 1999; Muller et al., 2001), but lack genes with homology to enzymes from the other pathways. Image reproduced from Avonce et al. (2006) with permission from BioMed Central.

Trehalose metabolism is ancient and eukaryotes contain genes with both TPS and TPP domains

Genes encoding enzymes from trehalose metabolism are found almost ubiquitously (Avonce et al., 2006). Eubacteria have demonstrated all five possible pathways of trehalose biosynthesis whilst Archaea appear to lack the TS pathway. Several eubacterial species have multiple pathways. Eukaryotes have predominantly the TPS/TPP pathway. Invertebrates use trehalose as their blood sugar. Vertebrates are unable to synthesize trehalose but express trehalase particularly in the intestine. Common to the eukaryotic TPS/TPP pathway is the presence of enzymes with both TPS and TPP domains. S. cerevisiae ScTPS2 has a TPP function but also contains a TPS-like domain located N-terminally (De Virgilio et al., 1993). In contrast, Arabidopsis AtTPS1 has a TPS function but also contains a C-terminal TPP-like domain (Blazquez et al., 1998; Leyman et al., 2001). In fact all AtTPS1-11 contain two domains with homology to TPS (N-terminally) and TPP (C-terminally) suggesting that these enzymes are derived from a single event of gene fusion possibly inherited from the eukaryote before chloroplast endosymbiosis (Table 1). S. cerevisiae TPS/TPP occurs as a complex of proteins including the ScTPS1 and the ScTPS2 and it has been proposed that complex formation serves to control free levels of T6P in the cell (generally 100 µM) when trehalose is being synthesized (to above 1 M) for stress protection and as carbon reserve (Bell et al., 1998).

In Arabidopsis levels of trehalose metabolites are generally at current instrument detection limits

Trehalose in wild type plant extracts has been mostly quantified by GC-FID and GC-MS where it is near the detection limit ranging from 37 ng g^-1 dry weight in whole Arabidopsis plants up to 100 ng g^-1 dry weight in Arabidopsis flowers (Muller et al., 2001; Roessner et al., 2000; Suzuki et al., 2008). T6P inhibits some yeast hexokinases such as Yarrowia lipolytica hexokinase and this biochemical property has been exploited to determine T6P concentrations in plant extracts with levels in Arabidopsis seedlings at 4.5 nmol g^-1 FW being reported (Blazquez et al., 1994; Pellny et al., 2004; Schluepmann et al., 2003). T6P has further been quantified in Arabidopsis seedlings using a procedure that includes a specific trehalase yielding 1.8 nmol g^-1 FW (Schluepmann et al., 2004) and using LC-MS yielding a range of concentrations from 0.018 nmol g^-1 FW in carbon starved to 0.482 nmol g^-1 FW in seedlings with supplied carbon under low light conditions (Delatte et al., 2009; Lunn et al., 2006; Schluepmann et al., 2004). The LC/MS methods were adapted from a method detecting sucrose-6-phosphate in plant extracts (Chen et al., 2005) and are likely the most accurate applied to T6P quantification in plant extracts until now. Data support the view that T6P has a large dynamic range in plants consistent with a function as a sugar signaling molecule. These detection methods, however, do not distinguish between metabolites derived from plant or bacterial cells. Microbial symbionts can accumulate trehalose metabolites at concentrations two to three orders of magnitude higher than plants, so contamination could be significant. Further trehalose-containing metabolites have not been characterized in plant extracts, but little effort has been made towards this in part because of the contamination issue necessitating a genetic or cell biology approach.

T6P is a charged molecule unlikely to cross membranes without proteinaceous channels. Determination of the fine subcellular distribution of trehalose metabolism in Arabidopsis as well as the cell-specific expression of some genes encoding enzymes from trehalose metabolism will require precise methods in cell biology. In vivo staining using FRET nano-sensors as described for glucose and sucrose could be used but currently lacks the sensitivity required (Deuschle et al., 2006; Lager et al., 2006; Lalonde et al., 2005). Another alternative would be to convert trehalose or T6P into a molecule not present in plants and readily recognized by specific antibodies. A solution could be the conversion to trehalose mycolate using an enzyme from either Mycobacteria or Corynebacteria (Lederer, 1976). Trehalose mycolates are relatively insoluble and can be readily detected at very high sensitivity by antibodies. Enzymes that catalyse this conversion have not been characterized, however, and it is not known if they are sufficiently specific for their substrate and if they operate at sufficiently low concentrations of substrate. Finally trehalose metabolites are known to trigger transcriptional responses and this could be exploited in the same way as it was for the auxin-responsive promoter DR5 (Ulmasov et al., 1997).

Feeding trehalose metabolites to Arabidopsis uncovers surprising effects

Feeding trehalose to Arabidopsis elicits strong responses that are not generated by osmoticum controls such as sorbitol or mannitol. In seedlings grown on 25 mM trehalose, carbon allocation is reversed with a massive accumulation of starch from carbon fixed in the cotyledons and no starch accumulation in root tips (Wingler et al., 2000). When trehalose (100 mM) is supplied, seedlings germinate and extend cotyledons, but fail to develop primary leaves and the primary root does not grow beyond 2–5 mm (Schluepmann et al., 2004). Interestingly, the root meristem is reduced in size and cells in the extension zone swell and burst. Supply of metabolisable sugars in addition to 100 mM trehalose relieves the growth inhibitory effects of trehalose suggesting that starvation causes growth arrest; it further suggests that trehalose does not affect the ability of sink tissues to metabolise carbon allocated (Schluepmann et al., 2004). Starvation in the growth zones is not due to carbon sequestration as starch since seedlings of pgm1 mutants unable to accumulate starch remain sensitive to external trehalose (Fritzius et al., 2001). Therefore trehalose may affect carbon transport or loading for transport. This effect could be at the subcellular level i.e. export from the chloroplasts to the cytosol. Alternatively the effect could be at the plant level i.e. interface between mesophyll and vascular bundles or simply involve transport inhibition in the phloem. Interestingly, trehalose competitively inhibits the sucrose-phosphate-phosphatase (SPP) of the cyanobacterium Synechocystis sp. (Fieulaine et al., 2007). Trehalose feeding could therefore potentially inhibit sucrose synthesis which would explain the strong phenotype that arises. Tobacco plants with antisense inhibition of SPP indeed accumulate starch in their leaves. They however have light-green leaves and do not display the dark green phenotype characteristic of plants grown on trehalose or accumulating T6P (Pellny et al., 2004; Schluepmann et al., 2003; Schluepmann et al., 2004). Growth of Arabidopsis seedlings expressing T6P hydrolase on 100 mM trehalose supports the idea that T6P rather than trehalose causes growth arrest (Schluepmann et al., 2004). However, T6P in the presence of sucrose stimulates growth (Schluepmann et al., 2003), so these effects are dependent on the presence of high trehalose ie. a non metabolisable carbon source. Feeding 30 mM trehalose in the presence of 1% w/v sucrose induced accumulation of sucrose and starch in two-week-old Arabidopsis seedlings and induced proteins known from stress detoxification including AtGSTF2, AtDHAR1, AtSAMS2 and AtFQR1. It is thus possible that high trehalose released from endogenous pathogens functions as elicitor (Bae et al., 2005a). Early changes in transcription after feeding trehalose seem to support this hypothesis (Bae et al., 2005b). Repressed Rubisco activase, that is redox-activated by thioredoxin f similarly to ADPG pyrophosphorylase (AGPase), and reduced levels of ATP after 24 h trehalose feeding in the presence of 1% w/v sucrose (Bae et al., 2005a) are consistent with effects expected when T6P accumulates (Pellny et al., 2004; Schluepmann et al., 2003; Schluepmann et al., 2004).

Table 1.

Summary table of the different Arabidopsis genes associated with trehalose metabolism together with AGI numbers, mutants, phenotypes, and expression patterns.

Trehalose supplied externally is likely taken up by seedlings since expression of E.coli trehalase targeted to the cytosol allows thriving growth on trehalose medium due to glucose released from trehalose cleavage (Schluepmann et al., 2003; Schluepmann et al., 2004). Seedlings therefore do not have a sufficiently high activity of trehalase to detoxify trehalose supplied at 100 mM. This holds true for all Arabidopsis accessions tested thus far, with Cvi being the only accession with some low degree of resistance to 100 mM external trehalose (Aghdasi PhD Thesis, 2007). Feeding trehalose not only affects carbon allocation from carbon fixed by cotyledons, but also seedlings in the dark that respond to trehalose feeding. These seedlings lack hypocotyl elongation but still undergo skotomorphogenesis suggesting lack of resource for extension but sugar sensing and signal transduction (Aghdasi Utrecht University PhD thesis 2007; Wingler et al., 2000).

Feeding T6P has been attempted to rescue embryos lacking AtTPS1 yet T6P appears to be not taken up by the embryos, probably because it is highly charged and therefore cell autonomous (Eastmond et al., 2002).

Validamycin A has been used as a specific competitive inhibitor of trehalase and its mode of action has been studied at the protein structure level with the E.coli periplasmic trehalase TreA (Gibson et al., 2007). Validamycin A is taken up by roots of tobacco and potato plants and readily transported to leaves (Goddijn et al., 1997). Feeding Validamycin A at 1 mM concentration leads to trehalose accumulation in tobacco and potato but only to levels of 0.021 mg g^-1 FW suggesting that flux through the trehalase step in these species is low (Goddijn et al., 1997). Validamycin A, unlike trehalose does not cause reversal of carbon allocation in Arabidopsis seedlings when supplied at 10 mM (Wingler et al., 2000) and preparations (Duchefa Biochemie, The Netherlands) above 1 mM are toxic, inhibiting root development completely but not affecting starch accumulation in the cotyledons (Schluepmann et al. unpublished). Validamycin A at 10 µM has been used as a trehalase inhibitor combined with feeding trehalose at 25 mM to uncover effects of trehalose at mM concentrations (Ramon et al., 2007; Wingler et al., 2000). Validamycin A, however is not specific for trehalase and affects the activity of other enzymes of trehalose metabolism, for example, it also inhibits trehalose synthase and maltooligosyltrehalose synthase from Bradyrhizobium japonicum and B. elkanii (Streeter and Gomez, 2006). Validamycin A has therefore been considered as a trehalose mimic. Trehazolin on the contrary was shown to specifically inhibit trehalase from these Bradyrhizobium species not affecting high activities of trehalose synthase and maltooligosyltrehalose synthase in nodules (Streeter and Gomez, 2006).

Feeding trehalose dimycolate to Arabidopsis did not affect seedlings but we are unable to tell if the relatively insoluble mycolates are taken up from the medium (Schluepmann et al. unpublished). Trehalose dimycolates are synthesized by Mycobacterium and Corynebacterium species and their mode of action during Mycobacterium infection has been studied extensively in humans and other model mammals. Specific trehalose dimycolates elicit strong inflammatory responses and trehalose dimycolates are thought to be synthesized from T6P (Ryll et al., 2001; Takayama et al., 2005).

Conclusion: In Arabidopsis, trehalose is unlikely to function as a compatible solute or a carbohydrate reserve as in microbes because trehalose concentrations are very low. Research progress is hindered by the restricted ability to visualize and quantify trehalose metabolites in planta, but is now being achieved through a combination of genetics and analytical chemistry. Open questions remain about the subcellular distribution of the metabolites and about possible derivatives of T6P or trehalose. Feeding trehalose to Arabidopsis leads to reversion of sink/source relations suggesting that it may well be an interesting target for applications in plant breeding. To research this activity, increasingly refined annotation of the Arabidopsis thaliana genome provides knowledge of the genes encoding enzymes of trehalose metabolism. In addition, expression of these genes can now be followed at the cellular level and with very high detection sensitivity. Characterization of genes encoding enzymes of trehalose metabolism and transport has only just begun.

The Arabidopsis genome has undergone radiation in T6P metabolism

Enzymes of trehalose metabolism from Arabidopsis were initially cloned by complementation of S. cerevisiae trehalose pathway mutants. In this way AtTPS1 was found to encode a TPS and TPPA and TPPB, TPPs (Blazquez et al., 1998; Vogel et al., 1998).

Homology searches in the complete Arabidopsis genome reveal a total of 11 TPS and 10 TPP homologues (Avonce et al., 2006; Leyman et al., 2001; Schluepmann et al., 2004). All of the TPS homologues also contain a TPP-like domain at the C-terminal of the protein whilst the TPP homologues are single domain proteins. Only one gene encoding trehalase is present, AtTRE1 (Muller et al., 2001). Therefore radiation at the level of T6P rather than trehalose has occurred in Arabidopsis, and this holds true for other plants with sequenced genomes (Avonce et al., 2006; Lunn, 2007; Pramanik and Imai, 2005; Satoh-Nagasawa et al., 2006; Shima et al., 2007).

T6P metabolism is indispensable in Arabidopsis development

Surprisingly, Arabidopsis knock outs of AtTPS1 are embryo lethal (Figure 2) suggesting that none of the other 10 TPS homologues can replace its function (Eastmond et al., 2002). It also suggests that trehalose metabolism is critical at that stage of development. Complementation of tps1 with inducible AtTPS1 shows that AtTPS1 is also required for root growth and transition from vegetative to floral development (van Dijken et al., 2004). tps1 can be complemented with E. coli TPS showing that T6P is required at embryo as well as at the other stages of development (Schluepmann et al., 2003). Yet, AtTPS2-6 are apparently most highly expressed in embryo and seed (Avonce et al., 2006 from Genevestigator) and given they cannot substitute for TPS1 in tps1 mutants, probably do not contribute to T6P synthesis in these tissues. We are therefore left wondering if the roles of these gene products are regulatory rather than biosynthetic, if they are present in different subcellular compartments to AtTPS1 or if they perform a different biosynthetic reaction altogether. A regulatory role is also presumed for the S. cerevisiae ScTPS3 and ScTSL1 proteins that have seemingly no biosynthetic activity but contain both domains with homology to TPS and TPP (Bell et al., 1998). ScTPS3 and ScTSL1 appear to stabilize the TPS/TPP enzyme complex that occurs in S. cerevisiae.

Gene sequences and protein targeting predict a wide subcellular distribution of trehalose metabolism

Trehalase from Glycine max is an apoplastic glycoprotein whereas trehalase from Arabidopsis appears to be plasma membrane bound with the catalytic domain on the apoplastic side of the membrane (Aeschbacher et al., 1999; Frison et al., 2007). Localization of AtTPS1 is cytosolic as determined by a combination of immunogold labeling in TEM and translational fusions with GFP (Geelen et al., 2007). AtTPS11 was found in mitochondrial fractions as identified by LCMS/MS (Heazlewood et al., 2004). In addition, several AtTPP, AtTPPA (At5g51460), AtTPPG (At4g22590) and AtTPPI (At5g10100) have a predicted subcellular location in the chloroplast (Emanuelsson et al., 2007).

The data thus infer that trehalose is broken down only after export to the apoplast. Trehalose is likely synthesized in the cytosol by a combination of TPS and TPP activities or by a bifunctional enzyme such as AtTPS6, although catalytic activity of this enzyme requires confirmation. T6P is likely generated in the cytosol and possibly transported to the chloroplasts where it then could be cleaved to trehalose by the TPP enzymes that appear likely targeted to the chloroplast. Further research is required to study the subcellular localization of enzymes from trehalose metabolism and to identify transport molecules. Such wide subcellular distribution suggests central functions, possibly including organellar feedback regulation. Indeed, T6P controls redox-regulation of AGPase, the rate-limiting enzyme of starch synthesis in chloroplasts. Because it is assumed that T6P is synthesized in the cytosol, T6P may convey information from the cytosol to regulate chloroplastic functions, such as starch synthesis. Homology searches in the complete Arabidopsis genome reveal a total of 11 TPS and 10 TPP homologs (Kolbe et al., 2005; Table 1).

Protein structures and possible functions of different genes annotated as part of trehalose metabolism in Arabidopsis

Analysis of genetic drift amongst the numerous homologues of TPS and TPP suggests that evolutionary selection retains the specificity of each of these genes: mutations are not random and maintain synonymous codons over the entire sequence of TPS and over almost all of TPP except for five codons (Avonce et al., 2006). Therefore TPS and TPP homologues must each fulfill a specific function required by selection.

Closer examination of AtTPS1-11 allowed classification of these genes into two classes according to their TPP sequence (Leyman et al., 2001). Class I TPS (AtTPS1-4) have lost the typical phosphatase boxes in the C-terminal TPP domain. Class II TPS (TPS5-11) retained the typical phosphatase boxes in the C-terminal TPP domain. Figure 3 presents an alignment of the 11 Arabidopsis TPS proteins with E.coil OtsA (TPS), OtsB (TPP) and Thermoplasma acidophilum TPP (TaTPP). The three-dimensional structures of OtsA (Gibson et al., 2002a; Gibson et al., 2002b; Gibson et al., 2004) and of TaTPP (Rao et al., 2006) are known and amino acids have been identified that are critical for the reaction with substrate. Structures from bacterial enzymes have also been superimposed onto plant sequences to predict activity of the domains from the plant proteins (Lunn, 2007).

Class I TPSs are large proteins more than 1000 amino acids long with both TPS and TPP domains where the TPP domain appears to no longer have functional phosphatase boxes (Leyman et al., 2001; Van Dijck et al., 2002; Figure 3). Sucrose phosphate synthases similarly have a non functional SPP domain which was proposed to serve the purpose of substrate channeling (Fieulaine et al., 2005), but in the case of TPS it could also serve to sequester T6P. The AtTPS1 enzyme is special because of an N-terminal extension, which is not found in other Arabidopsis TPSs and was shown to mediate regulation of TPS activity; it is also found in the SITPS1 from Selaginella lepidophylla, a desert resurrection plant that accumulates large amounts of trehalose (Van Dijck et al., 2002). A study of proteins interacting with AtTPS1 reveals that the N-terminal peptide in AtTPS1 appears to interact with KCA, a putative microtubule motor (Geelen et al., 2007). AtTPS1 also has a unique small low complexity region at the carboxy terminus (Figure 3). tps1 can be complemented by E. coli OtsA. The likely in planta function of AtTPS1 is thus synthesis of T6P (Schluepmann et al., 2003). AtTPS1 homologues have been found in every plant species thus far analysed including banana, poplar, rice, maize and sorghum. Class I TPS1 homologues were also found in prasinophyte algae thought to represent a primitive lineage of land plants. These homologues generally constitute the single Class I TPS gene except in the species of the Brassicaceae and in the moss P. patens (Lunn, 2007). Arabidopsis AtTPS2 and 4 are expressed in developing seeds but their function remains unknown. AtTPS3 is not represented in EST data and it is not detected by micro-array expression profiling. The AtTPS3 protein moreover appears to be deleted in a part of its TPP domain (Lunn, 2007 and Figure 3).

Class II TPSs are equally as large as those from Class I; the proteins contain both a TPS and a TPP domain, but the phosphatase boxes have been retained in the TPP domain (Leyman et al., 2001; Figure 3). Recombinant AtTPS5 failed to exhibit activity in vitro (Harthill et al., 2006). AtTPS5 interacts with the transcriptional co-activator MBF1c in vitro in S. cerevisiae two-hybrid assays and is thought to be required for basal thermo-tolerance of Arabidopsis (Suzuki et al., 2008). AtTPS6 was reported to complement S. cerevisiae strains deficient in ScTPS1 growing on glucose and S. cerevisiae strains deficient in ScTPS2 growing at 38°C (Chary et al., 2007) but this result required further confirmation. Phylogenetic trees constructed with Class II TPS from Arabidopsis, poplar and rice support 5 orthologous groups each including AtTPS5, AtTPS6, AtTPS7, AtTPS8-10 and AtTPS11 (Lunn, 2007).

Figure 2.

Mutants reveal that trehalose metabolism is involved at key steps in development in Arabidopsis and corn.

• A) Silique from a tps1/TPS1 heterozygous Arabidopsis plant where 1 in 4 seeds with tps1/tps1 genotype do not contain the mature embryo but instead an embryo seemingly arrested at torpedo stage. Photo courtesy A. van Dijken.

• B) Toluidine blue stained cross section of wild-type Arabidopsis seed with embryo at the mature stage; EM, embryo. Scale bar is 100 µm. Image reproduced from Eastmond et al. (2002) with permission from Blackwell Publishing.

• C) Toluidine blue stained cross section of tps1 mutant seed with embryo arrested at torpedo stage; CE, cellular endosperm. Image reproduced from Eastmond et al. (2002) with permission from Blackwell Publishing.

• D) Architecture of wild type B73 corn female inflorescence, the ear, after fertilization when seed is ripe. Image reproduced with permission from Macmillan Publishers Ltd: Nature, Satoh-Nagasawa et al. (2006).

• E) Architecture of ramosa3 (ra3) mutant ears in B73 background (left) and in mixed genetic background (right) where axillary meristems lose their determinancy and change identity leading to abnormal branches at the base of the ear. ra3 alleles have lesions in a TPP belonging to a clade specifically found in grasses. Image reproduced with permission from Macmillan Publishers Ltd: Nature, Satoh-Nagasawa et al. (2006).

AtTPS5, 6 and 7 are phosphorylated and bind 14-3-3 proteins (Harthill et al., 2006). 2-deoxy glucose feeding increases extractable SnRK1 activity and stimulates Th49 phosphorylation of AtTPS5 in Arabidopsis cells. Authors suggest that AtTPS5 may be a target of 14-3-3 binding after SnRK1 phosphorylation but only show in vitro phosphorylation of Thr49 and Ser22 from AtTPS5 by partially purified plant SnRK1. This is particularly interesting for AtTPS5 as its expression is very responsive to sucrose (Schluepmann et al., 2004). In addition, Class II AtTPS8-11 contain SnRK1 target sequences that are phosphorylated in vitro by Arabidopsis extracts in the absence of Ca²⁺ as would be expected by SnRK1 protein kinase (Glinski and Weckwerth, 2005). We thus conclude that Class II TPS likely are regulated by known nutrient signal transduction integrators such as SnRK1, but the significance and impact of this regulation on the function of the pathway is not clear.

TPP enzymes are single domain proteins and contain the conserved phosphatase boxes (Thaller et al., 1998; Vogel et al., 1998). AtTPPA and B complement S. cerevisiae defective in ScTPS2; crude extracts from these complemented S. cerevisiae make trehalose from T6P but the specific activities obtained were 5–12 times lower than with ScTPS2. The activities were specific for T6P and did not cleave sucrose-6-phosphate or G6P. In rice, OsTPP1 and OsTPP2 have been characterized biochemically: recombinant enzymes showed a high specificity for T6P and worked at higher affinity with 10 fold lower K_m compared to microbial enzymes (Pramanik and Imai, 2005; Shima et al., 2007). Maize RA3 (encoded by RAMOSA3) is a TPP shown to complement S. cerevisiae defective in ScTPS2 and its recombinant protein also cleaves T6P specifically (Satoh-Nagasawa et al., 2006). RA3 is most similar to AtTPPB and AtTPPC but the many gene duplications that occurred before and after evolution of grasses may not necessarily mean that these genes have orthologous functions. Grasses, however, do have a RA3-specific TPP clade. RA3 is 361 amino acids long and has an 80 amino acid N-terminal sequence that is not conserved amongst other TPPs followed by the TPP domain. ra3 mutants are on the same genetic pathway as RA1 and ra3 mutants have changes in RA1 expression whilst ra1 mutants are unaffected in RA3 expression. Results therefore suggest that RA3 is upstream of RA1 in the RAMOSA pathway. RA1 is a transcription factor and authors speculate that RA3 could in addition to its enzymic function also function as a regulator of transcription to explain the developmental alterations in ra3 mutants (Satoh-Nagasawa et al., 2006). Phylogenetic analyses of plant TPP sequences reveal that they are most closely related to proteobacteria and thus likely originate from the endosymbiotic ancestor of mitochondria (Lunn, 2007).

Transporters of trehalose or T6P are not yet characterized in plants. From insects, TRET1, a specific trehalose transporter that facilitates transport across membranes has been recently described (Kikawada et al., 2007). The TRET1 protein sequence is homologous to proteins from Arabidopsis encoded by At1g75220, At1g19450 and by ERD6, a sugar transporter induced by dehydration and cold treatment (Kiyosue et al., 1998). At1g75220 and At1g19450 are likely targeted through the secretory pathway to the plasma membrane or the endomembrane system. They both belong to the monosaccharide/sugar alcohol/polyol transporter superfamily but have not been functionally characterized. Presence of a transporter for import of T6P across the chloroplastic envelope has been inferred (Smith and Stitt, 2007). T6P is unlikely to be transported by a member of the above-mentioned gene family. A member of the plastidic phosphate translocator family including TPT, GPT, XPT, or PPT could be involved. GPT was demonstrated to transport G6P, but it was not tested for T6P transport (Kammerer et al., 1998; Niewiadomski et al., 2005). There are two related GPT genes in Arabidopsis, At5g54800 (GPT1) and At1g61800 (GPT2). A knockout of GPT1 is lethal, whereas a knockout in GPT2 is indistinguishable from wild type (Niewiadomski et al., 2005). A better understanding of this aspect of trehalose metabolism will be essential for elucidation of the mechanism of the role it plays in plants.

Figure 3.

Alignment of Arabidopsis TPS proteins with E.coil OtsA (TPS), OtsB (TPP) and Thermoplasma acidophilum TPP (TaTPP). Alignment was using the full-length protein sequences with MAFFT (Katoh et al., 2005; Katoh and Toh, 2008) and the figure was made using the Jalview alignment editor (Clamp et al., 2004). Red squares indicate amino acids interacting with substrates at the TPS active site of crystals from OtsA (Gibson et al., 2002b; Gibson et al., 2004), red stars indicate amino acids interacting with substrates at the active site of crystals from TaTPP (Rao et al., 2006). Red boxes mark the conserved phosphatase boxes.

Expression of the genes associated with Arabidopsis trehalose metabolism is tightly controlled

Small changes in T6P levels bring about large changes in plant metabolism and development as seen by manipulation using E. coli enzymes and by changes in the levels of endogenous enzymes (Figure 4). Therefore, tight control of genes involved in T6P metabolism is likely to be paramout.

TPS and TPP genes respond strongly and rapidly to a number of important hormones and nutrients. TPS8 is induced by cytokinin (Brenner et al., 2005). Two TPSs (TPS9 and TPS10) and four TPP genes (TPPB, in particular) are regulated by nitrate (Wang et al., 2003). Rice TPP genes are induced by ABA (Pramanik and Imai, 2005), as is TPS1 in guard cells (Leonhardt et al., 2004).

Interestingly, in contrast to sucrose-induced TPS5, TPS8 is strongly dark induced (Piippo et al., 2006). Starvation treatments induced by low sugar and in the starch-less mutant pgm1 induced several TPSs, but repressed a TPP(Thimm et al., 2004). Hypoxia, another form of starvation, increased expression of four genes associated with trehalose metabolism (TPPA, TPPB, TPPJ and trehalase) and repressed TPS11 (Liu et al., 2005). The significance of the induction of these genes and their relationship with T6P levels is not yet known. If starvation and high sugar are both capable of inducing T6P synthesis, for example, then this would imply more than one T6P target, or the same target but in different tissues with opposing function such as source and sink. A TPP from corn, RAMOSA3, for example is expressed in a very restricted cell-file and preliminary results suggest that some TPSs from Arabidopsis have very restricted expression patterns (Satoh-Nagasawa et al., 2006). This contrasts with the expression patterns of AtTPS1 and AtTPS6 that are expressed ubiquitously at all stages of development (Chary et al., 2007; van Dijken et al., 2004). Analysis of the expression of all genes associated with trehalose metabolism in Arabidopsis roots using a dataset generated from cell-sorted root protoplasts is shown in Figure 5 (Birnbaum et al., 2003; Brady et al., 2007). TPS1 transcript is enriched in cells from the stele, in particular in the maturation stage and after full differentiation, TPS8 is found highly expressed in stem cells including quiescent center and initials. TPS11 is found expressed in the root tip and is also very abundantly expressed in the stem cells. Expression of some TPP genes is also highly specific: TPPD, for example is seemingly only expressed in the columella, whilst TPPB is mostly expressed in the stele of mature zones. TPPG is the most highly expressed gene and ubiquitously expressed with highest levels found in stage 3 maturation zone tissues. Specific cell files therefore express a specific set of genes from trehalose metabolism. In the columella, for example, all TPS genes except TPS3 appear expressed as well as TPPA, D, G and J. Analysis of trehalose gene expression in roots therefore suggests that many TPS genes are simultaneously expressed and therefore may have differing roles in individual cells. Expression of the complement of TPP genes, however, appears to be more specific.

Figure 4.

Phenotypic changes typically observed when altering T6P steady states. Constitutive over-expression of E.coli TPS in Arabidopsis results in a two fold increase in T6P steady state levels, seedlings with dark green cotyledons (A, TPS) and small bushy plants with dark green and smaller leaves compared to wild type (B and C, TPS). Constitutive over-expression of E.coil TPP (converts T6P to trehalose) or expression of E.coli trehalose-phosphate-hydrolase (TPH) (converts T6P to glucose-6-phosphate and glucose) results in up to 3 fold reduced T6P steady state levels in Arabidopsis. In contrast to plants with high T6P levels, plants with low T6P levels have partially bleached cotyledons at seedling stage (A, TPP and TPH) and pale green leaves that are larger than wild type at rosette stage (B and C, TPP and TPH). Expression of E.coli trehalase yields Arabidopsis seedlings able to thrive on medium with 100 mM trehalose compared to wild type seedlings that cease growing on this medium (D). Arabidopsis expressing E.coli trehalase, however, have levels of T6P and a phenotype indistinguishable from wild type (B, trehalase). Image reproduced from Schluepmann et al. (2003) with permission from PNAS.

Figure 5.

Expression of genes associated with trehalose metabolism in protoplasts of different root tissues. Data are from (Birnbaum et al., 2003; Brady et al., 2007); tissue identities are defined in these references.

• A) Expression of the reference gene PP2A (At1g13320) did not significantly vary amongst the different tissues, whilst TRE1 encoding trehalase is mostly expressed in lateral root cap and columella at 3 fold less than PP2A. For all genes there was no difference between the sum of gene expression over all tissues and the gene expression extracted from root before protoplast release suggesting that the procedure of cell wall breakdown and cell sorting did not alter expression of the genes (control).

• B) Expression of Class I TPS is generally low with TPS1 being the most highly expressed

• C) In root tips, expression of Class II TPS is generally higher than Class I TPS except for TPS6; TPS Class II are expressed in all tissues concomitantly yet their relative expression varies: expression of TPS8 is particularly high in the quiescent center for example

• D) Expression levels of TPP genes differ amongst each other and appear in some cases to be extremely tissue specific — e.g. TPPD (columella) and TPPB (mostly stele). TPPG expression is ubiquitous and very high in the maturing root zone.

In addition to transcriptional regulation, regulation of the translation of TPPF and G by a specific conserved upstream open reading frame is likely (Hayden and Jorgensen, 2007).

Post-translational modification of the Class II TPSs by phosphorylation can regulate their activity and their binding to 14-3-3 proteins (Harthill et al., 2006). Multisite phosphorylation of Class II TPSs by both SnRK1 and Ca²⁺-dependent protein kinases seems probable (as with sucrose phosphate syntheses), giving the possibility of enzyme activation/inactivation in response to different internal and environmental cues (Paul, 2007).

Mutants in genes from the trehalose pathway have developmental defects

Plant development can be seen as a set program of events in which metabolic and developmental pathways draw on common resource pools, for example UDPG for cell wall synthesis during cell division and expansion. The maize RAMOSA3 (RA3) gene, however, confirms a role for a gene of the trehalose pathway in controlling inflorescence branching in maize (Satoh-Nagasawa et al., 2006; Figure 2). RA3 encodes a TPP and is expressed in discrete domains subtending axillary inflorescence meristems. RA3 functions through the predicted transcriptional regulator RAMOSA1 and research has yet to reveal if and how TPP activity of RA3 and RAMOSA1 are linked.

The embryo-lethal phenotype of tps1 mutants suggests a role of AtTPS1-mediated T6P production during embryo development. Eastmond et al. (2002) showed that tps1 embryos are sensitive to externally supplied sucrose. These results are consistent with sucrose sensitivity of engineered plants expressing either TPP or TPH that contain low levels of T6P (Schluepmann et al., 2003) suggesting that T6P is required for carbon utilization and growth. Further consistent with the sucrose sensitivity of plants with low T6P is the free sugar accumulation in tps1 mutant embryos (Gomez et al., 2006). In contrast, seedlings with higher steady state levels of T6P than wild type grow faster on medium with metabolisable sugars. Seedlings with high T6P levels compared to wild type also have low levels of free sugars and accumulate starch (Schluepmann et al., 2003). Detailed analysis of developing tps1 embryos reveals that these embryos develop at a much slower pace compared to wild type (Eastmond et al., 2002; Gomez et al., 2005; Gomez et al., 2006) and that development beyond torpedo stage does occur after cold stratification and incubation on agar medium for periods ranging from 1–3 weeks (Gomez et al., 2006). Furthermore, torpedo stage tps1 embryos displayed an expression profile unlike torpedo stage wild type embryos but much resembling the profile of mature wild type embryos (Gomez et al., 2006). tps1 embryos therefore do undergo the procession of gene expression patterns typically associated with the progression through the embryo development stages. Cellular differentiation also occurs according to wild type embryo patterns in the tps1 embryos with oil and protein bodies accumulating in tps1 embryos seemingly arrested at torpedo stage; these bodies are typically found in wild type embryos at maturation stages including bent cotyledon stage, tps1 embryos therefore are not affected in programs controlling metabolic and cellular development. Overall accumulation of protein and oil end products is reduced. Interestingly, tps1 embryos have thicker cell walls compared to wild type embryos at either torpedo or bent cotyledon stage. These walls contain an epitope not found in wild type embryos and recognized by the LMC2 monoclonal antibody previously described as recognizing arabinogalactan proteins found at the plasma membrane of root apices of a number of plant species (Gomez et al., 2006; Smallwood et al., 1996). However, a number of cell wall related transcripts were down regulated. Interestingly the pectin fraction of the cell walls was increased. This represents a major flux of carbon in Arabidopsis. Cell wall thickening may be a consequence of reduced synthesis of oil and protein and reduced growth. It is possible that cell wall biosynthesis is a direct target of T6P. T6P is made from G6P and UDPG; UDPG is a known precursor of cell wall components. Incidentally cotton cellulose synthase CESA1 is redox-regulated by thioredoxin as AGPase but in the opposite direction (Doblin et al., 2002; Kurek et al., 2002). Expression analysis of the CycB1::GFP fusion in tps1 revealed that the rate of cell division is much reduced in tps1 embryos (Gomez et al., 2006). It is not known how the absence of T6P synthesis influences cell division in tps1 embryos. The strong phenotype of these embryos makes it difficult to separate primary effects of T6P from further downstream effects such as accumulation of sucrose and starch.

Nevertheless, taken together the data point to an important role of trehalose metabolism in development, but not in the coordination of the succession of developmental events, rather more likely in processes of cell division.

T6P links photosynthesis and development

Tobacco and Arabidopsis plants with high T6P levels have high photosynthetic capacity per unit leaf area due to more Rubisco and chlorophyll (Pellny et al., 2004). This striking effect of perturbing T6P steady state in plants probably reflects the complex nature of the regulation of photosynthesis, which is integrated at the whole plant level and with the production of leaf area. Low T6P levels produced the opposite phenotype of lower photosynthetic capacity per unit leaf area, but paradoxically more productive plants, as leaf area and photosynthesis per plant were greater. It is likely that T6P affects early leaf development which affects leaf size and investment of photosynthetic apparatus. It is unclear if the effect on photosynthesis is linked to the impact of T6P on starch synthesis (Kolbe et al., 2005). Altered T6P content does not appear to alter NADPH/ NADP ratios. It is proposed that the impact of T6P on the redox regulation of AGPase is separate from effects on photosynthesis.

Does T6P have a metabolic function or a signaling role?

Hypotheses have been that T6P could be required as substrate for an essential reaction in metabolism or it could be required as a signaling molecule sensed and then transduced into signaling cascades. Metabolite profiling of tps1 mutant embryos or tps1 seedlings after genetic rescue of the embryos with inducible AtTPS1 revealed vast changes in metabolite steady states when compared to wild type with most changes in sucrose, fructose and glucose (Gomez et al., 2006; Schluepmann et al., unpublished). In addition, G6P levels in seedlings expressing E. coil TPS or TPP are decreased and increased respectively (Pellny et al., 2004; Schluepmann et al., 2003). This is similar to S. cerevisiae deleted in ScTPS1, where T6P was shown to inhibit hexokinase and so low T6P results in high G6P steady states (Blazquez et al., 1993). S. cerevisiae without T6P is unable to grow on rapidly fermenting sugars such as glucose and it has been suggested that this is due to de-regulation of glycolysis with hyper-accumulation of sugar phosphates and depletion of ATP (Thevelein and Hohmann, 1995). Hexokinases from plant extracts, however, are not sensitive to T6P (Eastmond et al., 2002). Depletion of ATP due to insufficient phosphate recycling is unlikely since plants with 1000-fold accumulation of sucrose-6-phosphate do not have obvious signs of phosphate limitation (Chen et al., 2005). The main arguments for T6P as a signaling molecule are its low concentration in cells, wide dynamic range combined with the large changes in metabolism and morphology in plants that arise as a consequence of relatively minor changes in the steady state levels of T6P (Figure 4). Such changes are not found when changing the steady states of the similar molecule sucrose-6-phosphate, the precursor of sucrose or the regulatory molecule fructose-2,6-bisphosphate (Chen et al., 2005; Nielsen and Stitt, 2001). We know that T6P directly or indirectly affects specific redox regulation processes in the chloroplasts such as for example the redox activation of AGPase; the effect is independent of the NADPH/NADP ratios. The effect is found in mature leaves of Arabidopsis as well as in potato tubers. T6P mediated redox-activation is required for increased starch deposition when sugars are available (Kolbe et al., 2005). In addition, sugar-induced increases in T6P correlate with redox-activation of AGPase and increased rates of starch synthesis (Lunn et al., 2006). However most of the phenotypes produced where the pathway has been genetically modified in mutant and transgenic lines cannot be explained simply in terms of the effect on starch metabolism but rather support a more central function.

T6P inhibits SnRK1 of the SNF1/AMPK group of protein kinases

Recently, T6P was reported to inhibit SnRK1 (AKIN10/ AKIN11) of the family of calcium-independent Ser/Thr protein kinases that include AMPK of mammals and SNF1 of yeast (Zhang et al., 2009). These conserved protein kinases perform a fundamental role in transcriptional, metabolic and developmental regulation in response to energy limitation and starvation of carbon source (Hardie 2007). SnRK1 consists of a heterotrimeric complex of a catalytic alpha subunit together with beta and gamma regulatory subunits, plus a number of known and likely unknown interacting and regulatory factors (Polge and Thomas 2007). It was shown that low micromolar concentrations of T6P inhibit SnRK1 activity at a site on the SnRK1 complex distinct and physically separable from the catalytic site via an unknown intermediary factor. T6P inhibited SnRK1 most strongly in seedlings and young actively growing tissues. Inhibition was also found in all other plant tissues tested except in mature leaves that lacked the intermediary factor. Striking confirmation of the effects of T6P on SnRK1 activity measured using plant extracts was made by comparing effects on gene expression in A. thaliana with altered T6P compared to plants with altered AKIN10 expression. Baena-Gonzalez et al. (2007) using a transient protoplast expression system established a 1000 or so genes that were directly regulated by SnRK1. Using seedlings Zhang et al. (2009) found that genes normally induced by SnRK1 were repressed by T6P and those normally repressed by SnRK1 were induced, strongly supporting a role for T6P as inhibitor of SnRK1 in vivo. Overall there was a strong upregulation of biosynthetic pathways by T6P. It had previously been found that T6P promotes growth of seedlings on sucrose (Schluepmann et al. 2003). These data explain this finding and support a role for T6P as promoter of biosynthetic reactions in growing tissues through inhibition of SnRK1. It is likely that a major function of SnRK1 in plants, as for SNF1 and AMPK in other organisms, is to activate catabolic processes under starvation stress conditions and to inhibit anabolic reactions that consume ATP (Hardie 2007). Given that T6P responds strongly to sucrose supply (Lunn et al. 2006), T6P would counter the starvation response mediated by SnRK1 to promote growth and biosynthetic reactions in response to sucrose availability.

The signaling function of T6P and/or AtTPS1 is likely integrated in sugar signaling networks

Increased AtTPS1 levels lead to 3–6 fold higher ABI4 expression, but on medium with glucose, expression of ABI4 in AtTPS1 over-expressors is much reduced compared to wild type (Avonce et al., 2004). Since abi4 mutants are glucose insensitive (Huijser et al., 2000; Soderman et al., 2000), this can explain the glucose- and ABA- insensitive phenotype of AtTPS1 over-expressors. In addition, hxk1, also glucose insensitive has much increased AtTPS1 expression on glucose medium. However, it is not clear if this increased AtTPS1 expression also comes with a reduced level of ABI4 expression, which if it were the case would explain hxk1 resistance to glucose (Jang et al., 1997; Jang and Sheen, 1994; Moore et al., 2003). Interestingly Ctr1, which is also glucose insensitive (Gibson et al., 2001), has increased TPS1 expression (more than twofold in rosette, log1.3) and much decreased ABI4 expression (more than eightfold, log3) suggesting that CTR1 regulates AtTPS1 gene expression which then leads to reduced ABI4 expression (Arroyo et al., 2003 and Genevestigator, mutants). To conclude, AtTPS1 may mediate hxk1 insensitivity to glucose and AtTPS1 appears to act on ABI4 expression levels.