Although the eukaryotic TOR (target of rapamycin) kinase signalling pathway has emerged as a key player for integrating nutrient-, energy- and stress-related cues with growth and metabolic outputs, relatively little is known of how this ancient regulatory mechanism has been adapted in higher plants. Drawing comparisons with the substantial knowledge base around TOR kinase signalling in fungal and animal systems, functional aspects of this pathway in plants are reviewed. Both conserved and divergent elements are discussed in relation to unique aspects associated with an autotrophic mode of nutrition and adaptive strategies for multicellular development exhibited by plants.
Growth in living organisms is not a simple sum of available energy and nutrients, but instead represents a regulated output that supports complex survival and reproductive strategies. In eukaryotes, a key element for this regulation is TOR (target of rapamycin). Initially characterized as a growth-enabling protein kinase in single-celled yeast, TOR has since been found in all major eukaryote groups where it integrates growth and metabolism with combinations of cues relating to nutrients, energy and stress. This conserved activity and an array of associated elements suggest an ancient regulatory pathway that has been adapted to support a wide range of cellular and organismal growth strategies. Since the discovery of TOR more than 20 years ago, models for its functionality have progressed from it serving as a simple gatekeeper, enforcing nutrient-dependent checkpoints on growth and cell division in single-celled yeast, to considerably more complex roles in multicellular species (Figure 1). Comparisons between animals and yeast highlight conserved elements both up- and down-stream of TOR that probably reflect ancestral roles of balancing cellular growth and metabolic homoeostasis, but also reveal additional players that integrate this set of core activities into multicellular programmes of development, many of which relate to cancer and metabolic disease.
A graphic summary of the generic roles of TOR in co-ordinating cellular growth and metabolic outputs with extrinsic inputs
The present review focuses on TOR from the relatively unfamiliar, but potentially illuminating, perspective offered by plants whose growth and metabolism contrasts with that of other eukaryotes. Instead of purely consumption-based modes of nutrient and energy acquisition, plants feature a complex mixture of metabolic regulation. During seedling development, plants shift from animal-like heterotrophic utilization of seed reserves to an autotrophic light-driven metabolism made possible through the presence of the chloroplast. Vegetative growth presents further complexity, with reduced carbon and energy obtained through photosynthesis in green shoot tissues traded for inorganic mineral nutrients scavenged by roots. Cell growth and division feature complex, but incompletely understood, patterns of vesicle trafficking, with increases in cell size achieved through ongoing remodelling of a cellulose-based cell wall matrix coupled with turgor-driven expansion. To compensate for their lack of mobility, plants complement a high level of environmentally tuned physiological regulation with adaptive programmes of development. Complex and variable architectures are elaborated throughout the vegetative life of the plant through the activity of localized clusters of pluripotent cells known as meristems, with the fate of cells descended from these embryo-like tissues guided by a suite of signalling pathways that have no obvious animal equivalents.
To provide context for a discussion of TOR signalling in plants, an overview of key findings from more intensively studied animal and fungal models is first provided. The functionality of key elements in plants is then considered with respect to both conserved and niche-specific growth and metabolic outputs. Largely unexplored features of upstream inputs are then surveyed in relation to both conserved and plant specific signal transduction pathways. More extensive background on TOR signalling is offered in a number of reviews, including several offering insights into plant-related aspects [1–9].
The emergence of TOR as a master integrator of growth and metabolism
Knowledge of TOR traces back to the mid-1970s with the characterization of a novel compound from Streptomyces hygroscopicus, a soil bacterium collected from Easter Island . Named rapamycin, for the Polynesian name for Easter Island, Rapanui, this macrolide compound was first distinguished as an antifungal agent, but then later as a potent immunosuppressant and growth inhibitor. Insights into the cellular targets of rapamycin emerged later through analyses of yeast mutants resistant to its effects, with mutations compromising the interaction of rapamycin with a large protein that became known as target of rapamycin, or TOR [11–14].
Comparative studies group TOR into a family of serine/threonine kinases known as PIKKs (phosphoinositide 3-kinase-related kinases). Although the kinase domains of PIKKs share many conserved features and have overlapping responses to chemical inhibitors with that of their namesake, PI3K (phosphoinositide 3-kinase) , all members of this family show a preference towards protein rather than lipid substrates [16–18]. Orthologues of TOR [also referred to as FRAP for FKBP (FK506-binding protein)–rapamycin-associated protein, RAFT1 for rapamycin- and FK506-binding protein target 1, or RAPT1 for rapamycin target 1] are found in all major eukaryotic groups, including metazoans, mammals and plants, and can be recognized by a complex, but well-conserved, domain organization (Figure 2). A series of HEAT [named after huntingtin, elongation factor 3, a PP2A (protein phosphatase 2A) subunit and TOR1] repeats, which contain a ~40-amino-acid sequence that forms two antiparallel α-helices, extend from the N-terminus and have been shown to mediate protein–protein interactions and membrane associations [19,20]. Downstream lie the ~800 amino acids of the FAT domain [named after FRAP, ATM (ataxia telangiectasia mutated) and TRRAP2 (transformation/transcription domain-associated protein 2)] that consists of HEAT and TPR (tetratricopeptide repeat) repeats, and which, together with the FATC (C-terminal FAT) domains, contributes to protein interactions and kinase activation . The FAT domain is followed by the FRB domain (for FKBP12–rapamycin-binding domain), which lies just upstream of the kinase domain and is targeted by the inhibitory FKBP12–rapamycin complex.
Domain organization of TOR, RAPTOR and LST8 proteins in representative species
The conserved structural complexity of TOR kinase across eukaryotes suggests a well-entrenched functionality that is supported by a large body of genetic, biochemical and pharmacological evidence. In both fungi and animals, the highly specific inhibitory effects of rapamycin on TOR has revealed its role in channelling diverse signalling inputs into coherent growth responses. Early studies showed that many aspects of rapamycin-induced growth arrest mimic starvation responses, including G0 cell division arrest and down-regulation of protein synthesis [1,22]. Conversely, activation of TOR kinase by nutrients up-regulates the capacity for protein synthesis through phosphorylation of translation-related proteins, including S6K (ribosomal S6 kinase). This sensitivity of these and other responses to rapamycin has become a standard criterion for implicating TOR in complex signalling processes.
Modulation and targeting of TOR activity reflects its association in stable protein complexes
Although rapamycin initially seemed to offer a ‘silver bullet’ to selectively kill TOR activity to explore its functionality, subsequent analysis of TOR-deficient mutants in yeast revealed additional targets whose phosphorylation by TOR is resistant to rapamycin. This disparity later became understood with the discovery that TOR acts in two functionally distinct complexes: TORC1 and TORC2 [23,24]. The rapamycin-sensitive TORC1 is distinguished by the presence of RAPTOR (regulatory-associated protein of mTOR, where mTOR is mammalian TOR), which is thought to influence the activity of the kinase and its substrate specificity [25–29]. Similar to TOR, comparisons of the domain organization of RAPTOR proteins show a high degree of conservation across all major eukaryotic groups, including a RNC/C domain (for RAPTOR N-terminal conserved/putative caspase) and a series of HEAT and WD40 repeats (Figure 2).
TORC2, by contrast, lacks RAPTOR, but contains other distinct accessory proteins, including RICTOR (rapamycin-insensitive companion of mTOR), which confers a distinct functionality on the complex [3,24]. TORC1 and TORC2 also feature a common element, LST8 (lethal with sec-13 protein 8), which like RAPTOR, has been described in all major eukaryotic groups [30–32] (Figure 2). Consisting primarily of seven WD40 repeat elements, the precise function of LST8 in TORC1 and TORC2 is not entirely clear. Recent structural studies of mammalian cell-line-derived material suggest that LST8 may help to stabilize the organization of domains critical for full TOR kinase activity, and that the positioning of LST8 next to the FRB domain may effectively limit and/or facilitate access to the catalytic site of TOR . This interpretation is in line with in vivo studies, where LST8 is required for nutrient- and RAPTOR-dependent activation of TOR [34,35].
Since their initial characterization in yeast, TORC1 and TORC2 have been shown to act in a largely conserved manner across fungal and animal species as regulatory switches that target specific growth-associated processes. TORC1 activity has been linked to regulation of protein synthesis, ribosome biogenesis, autophagy, lipid synthesis, mitochondrial metabolism and cell division . By contrast, TORC2 outputs have been linked to actin cytoskeleton organization and cellular polarization [23,37,38]. Additional work has tied TORC2 activity to cell-cycle progression, anabolism and cell survival [39–42]. Studies in yeast also indicate partial redundancy between complexes for some functions [1,43].
TOR-DEPENDENT PROCESSES IN PLANTS
Compared with animals and yeast, relatively little work has been done on TOR signalling in plants. Early suggestions for the pathway in plants came from studies that described insulin-like signalling responses, including rapamycin-sensitive phosphorylation of S6K, one of the most prominent targets of mTOR [44,45]. However, until recently, similar inhibitor-based strategies had not been investigated in the widely used model plant Arabidopsis, in part because of its relative insensitivity to rapamycin. Instead, more gene-focused analyses have been encouraged by extensive genome-based resources, including full genome sequence, nearly saturated indexed collections of insertion mutants, facile strategies for overexpression and RNAi-based inhibition, as well as extensive transcriptomic datasets. More recently, strategies for chemical inhibition of TOR have been developed. By one approach, Arabidopsis is sensitized to rapamycin through expression of native or heterologous forms of FKBP12, which rapamycin depends on for its inhibitory ternary interaction with TOR [46–49]. In a second approach, specific inhibitors of mTOR that block the ATP-binding pocket of the kinase domain have been shown to act as potent inhibitors of growth in plants, but their specificity for TOR relative to related kinases has not been fully characterized in plants [48,50,51].
Growth-related TOR outputs in plants
The first functional analysis of TOR in Arabidopsis offered support for a growth-related function in plants, showing that embryos homozygous for likely null insertion alleles of AtTOR (Arabidopsis thaliana TOR) arrest at an early stage of development . It was shown further that expression of a TOR:GUS reporter translational fusion was focused in actively growing vegetative tissues, suggesting a role for TOR in regulating post-embryonic growth as well as a significant degree of post-transcriptional regulation, given a more ubiquitous distribution of TOR transcripts (Figure 3). Subsequent studies have confirmed this role, with decreased TOR activity achieved by RNAi or chemical inhibition leading to reduced rates of both vegetative shoot and root growth and reductions in cell size [47,48,53–56]. By contrast, artificially increased TOR transcript levels have the opposite effect [47,53], suggesting that, similar to animals and fungi, TOR activity rather than nutrient levels can be limiting for growth. Reduced TOR activity is also linked to altered cell wall biosynthesis, decreased root hair development, altered patterns of branching, and decreases in the number of dividing cells in meristematic tissues [48–50,56,57]. Mutations disrupting Arabidopsis ROL5, which encodes a small G-protein belonging to the Ras superfamily, are hypersensitive to rapamycin and have defective cell walls, further supporting a role for TOR in contributing to cell wall integrity . Although these reports document changes in cellular growth processes that track TOR activity, no consistent changes in tissue organization have been described.
TOR expression during vegetative development
Metabolic regulation by TOR in plants
Similar to animals and fungi , the influence of TOR activity on growth in plants is accompanied by complex changes in metabolism, with TOR activity promoting production of metabolites for immediate growth versus storage forms for longer-term growth and reproductive strategies. Support for this type of activity is seen in the metabolite and transcriptional profiles of Arabidopsis in which RNAi or chemical inhibition of TOR or mutation of the TORC1 element LST8 leads to accumulation of starch, TCA (tricarboxylic acid) cycle intermediates and triacylglycerols, and associated changes in intermediate metabolism gene expression profiles [47,48,53,55,59]. Similar responses are also seen with rapamycin treatment in the single-celled green algal model Chlamydomonas . In both models, accumulation of storage metabolites mimics growth arrest responses triggered by nitrogen limitation, suggesting a role for TOR in rerouting of carbon when low levels of other classes of nutrient preclude growth. Changes in TOR activity are also seen to affect nitrogen metabolism with altered levels of glutamine, nitrate and genes associated with its metabolism, although some conflicting results between experiments have been attributed to different carbon nutrient input levels . Similarly, inactivation of the TAP42 (2A phosphatase-associated protein of 42 kDa) plant homologue TAP46, a PP2A-associated protein and a substrate of TOR, leads to activation of enzymes involved in nitrogen remobilization, such as glutamine synthase 1, and to the inhibition of enzymes involved in nitrate assimilation . Large increases in the levels of free amino acids observed with decreased TOR activity have been linked to decreased rates of protein synthesis, but may also reflect increased protein recycling through derepression of autophagy. Decreases in stress-induced production of galactinol and raffinose, which have been proposed to act as scavengers for ROS (reactive oxygen species) produced in actively growing tissues, have also been observed with inhibition of TOR activity [7,47,59].
Although these types of TOR-mediated metabolic responses have been linked to lifespan extension in animals [62,63], their relevance to plants remains unclear. If the transition to reproductive development is considered a sign of aging in plants, delays in flowering seen in plants with reduced TOR activity  might suggest parallels with animal systems. It should be noted, however, that the transition to reproductive development in plants is subject to multiple inputs, including nutrient levels, temperature and stress, all of which act to decouple flowering from a strictly chronologically determined output . Moreover, it could be argued that certain responses to TOR limitation, such as cellular senescence and remobilization of nutrients [53,65], constitute an accelerating aging phenotype, albeit one manifest at an organ- and tissue-specific level.
MOLECULAR ASPECTS OF TOR FUNCTIONALITY
The rapid increase in the number of full genome sequences across a diverse range of eukaryotes has promoted comparative approaches towards understanding TOR signalling. For plants, these comparisons highlight a considerable degree of conservation with certain TOR signalling elements that have been functionally characterized in other eukaryotic models, but also draw attention to missing elements, whose absence must be reconciled with the unique evolutionary history and life strategies of plants (Figure 4). Graphics that compare the functionality of these elements are presented in Figures 5(a) and 5(b).
Representation of TOR signalling components in Homo sapiens, Schizosaccharomyces pombe, Saccharomyces cerevisiae, Chlamydomonas reinhardtii and Arabidopsis thaliana
Comparison of mammalian and plant TOR signalling pathways
Similar to animals and fungi, the family of PIKKs in plants to which TOR belongs includes ATM, ATR (ataxia telangiectasia- and Rad3-related), TRRAP and SMG (suppressor of morphogenesis in genitalia), all of which share conserved FAT, PI3K and FATC domains [18,66] and, which like TOR, appear to moderate cellular responses to extrinsically determined factors, including stress. The FATC domain, whose integrity in TOR is essential in both fungi and mammals, appears dispensable in Arabidopsis, with no clear phenotype associated with its absence . In this same study, more severely disrupted alleles that also lack the kinase domain could be rescued by transgenic expression of the kinase domain alone, suggesting the potential for autonomous kinase domain function, or intragenic complementation between partial proteins.
Plants exhibit further distinctions in terms of the interaction of TOR to form complexes with accessory proteins. Despite the essential nature of TORC2 in animals and fungi, no equivalents for TORC2-specific proteins, such as RICTOR, are apparent in plant genome databases. Given the prominent role of TORC2 in supporting polarized growth in these other groups, its absence from plants could be linked to divergent strategies for driving cellular growth, which in plants feature distinct cytoskeletal dynamics and turgor-driven cell expansion. By contrast, TORC1 function in plants is well supported, with clear equivalents of RAPTOR evident [67,68] and evidence for interaction with canonical downstream targets . If TOR does play a role in supporting polarized growth in plants, it might rely on mechanisms similar to those recently described in yeast in which cytoskeletal elements mediate regulation of TORC1 [70,71] or could alternatively rely on still unknown novel types of TOR complex, as suggested by recent precedents in protozoan models [72,73].
Some evidence for specialization between duplicate forms of RAPTOR in Arabidopsis is seen in the gene pair AtRAPTOR3g and AtRAPTOR5g [67,68]. Moderate levels of AtRAPTOR3g transcripts can be detected across most vegetative tissues, with plants homozygous for disrupted forms showing a range of growth defects, including sporadic embryonic arrest, moderately decreased rates of vegetative growth, increased branching and delayed flowering. RAPTOR5g expression, by contrast, is more limited in vegetative tissues, with the highest levels in specific gametogenic and embryonic tissues. AtRAPTOR5g has no obvious loss-of-function phenotype by itself, but double mutants with AtRAPTOR3g loss-of-function alleles grow very slowly (A. Larking, D. Rexin and B. Veit, unpublished work). If these double mutants represent null alleles for RAPTOR, they would indicate a capacity for TORC1-independent growth that can be contrasted with the animal and fungal systems, where TORC1 is essential for survival.
In addition to the TORC1-specific RAPTOR, plants harbour genes that encode LST8-like proteins, whose counterparts are found in both TORC1 and TORC2 of fungi and metazoans. Before its identification as a TORC component, LST8 was shown in yeast to support trafficking of amino acid permeases from the Golgi apparatus to the plasma membrane . Further work showed that LST8 is essential for TOR-mediated RTG (repression of retrograde) signalling . In response to mitochondrial stress, TOR-mediated phosphorylation of transcription factors Rtg1/3p and Gln3p is relaxed, leading to compensatory changes in carbon and nitrogen metabolism [31,32]. LST8-like proteins from both Arabidopsis  and the green alga Chlamydomonas  share the same well-conserved domain organization as seen in non-photosynthetic counterparts, can complement LST8-deficient yeast and show specific binding with TOR. Chlamydomonas LST8 and TOR are found together in high-molecular-mass membrane-associated complexes, with GFP-tagged LST8 localizing to motile endosome-like particles in Arabidopsis.
Changes in growth and metabolism associated with deficiency of LST8 in Arabidopsis are complex. Similar to RAPTOR-deficient plants, plants deficient for LST8 are viable, but grow more slowly, branch more frequently and display characteristic alterations in metabolite profiles. Unlike the RAPTOR-deficient plants, however, LST8-deficient plants show growth defects that are strongly exacerbated in long days, suggesting a specific role for LST8 in mediating metabolic adjustments to changes in photosynthetic inputs. The possibility that photoperiod-dependent phenotypes of LST8-deficient plants reflect a TOR-independent activity of LST8 is suggested by precedents in Schizosaccharomyces pombe, where mutations in the LST8 counterpart lead to a suite of changes generally not tied to reductions in TOR activity, including diploidization, genome instability and decreased tubulin levels . Alternatively, these light-dependent phenotypes might reflect a specific role for LST8 in supporting a link between the chloroplast outputs and TOR, analogous to its role in supporting retrograde signalling with the mitochondrion .
DOWNSTREAM TARGETS AND PROCESSES OF TOR KINASE IN PLANTS
Similar to animals and fungi, plants undergo profound changes in their translational profiles  as part of growth and metabolic regulation, with TOR acting at several levels to regulate translational outputs. Increases in the capacity for protein synthesis that accompany the onset of anabolic growth processes are reflected in increases in rRNA levels that closely track levels of TOR activity . Similar to what has been reported in yeast  and mammalian species , ChIP experiments suggest binding of the kinase domain of TOR with promoter regions of rRNA genes, though it is unclear whether this is a direct interaction. Additional changes in translational outputs in response to decreased TOR levels are also reflected in reduced loading of polysomes [51,53,79].
The best-supported examples of TOR-mediated translational control in plants are seen with S6K, which, like its animal counterparts, is a well-characterized substrate of TOR. The majority of studies on plant S6K have exploited Arabidopsis, which contains two closely related forms, AtS6K1 and AtS6K2. Early work demonstrated that the ribosomal protein S6 is efficiently phosphorylated by AtS6K  and that AtS6K2 could substitute for human p70S6K in human cell lines with respect to S6 phosphorylation . Later work showed that, similar to the case in animals, S6K plays significant roles in mediating growth responses in plants, including those involving the growth-promoting hormones auxin and cytokinin [81–83]. Like many other members of the AGC kinase family to which it belongs, plant S6K is also a target for PDK1 (phosphoinositide-dependent kinase 1)-mediated phosphorylation  that contributes to full activation of S6K.
Several studies have confirmed that chemical inhibition or knockdown of TOR gene activity leads to decreased S6K phosphorylation at defined sites equivalent to their animal counterparts [48,51,56,79,85]. Additional support for TOR-mediated phosphorylation of S6K is seen in studies demonstrating physical interactions between TOR, RAPTOR and S6K proteins in transient overexpression assays in Nicotiana benthamiana.
Kinase activity of immunoprecipitated AtS6K is decreased by osmotic stress in a manner that is sensitive to RAPTOR and TOR stoichiometry , suggesting that TORC1-mediated signalling plays a role in this regulation. Studies in whole Arabidopsis plants reinforce this view, suggesting that TOR-dependent down-regulation of S6K activity is integral to adaptive osmotic stress responses [53,69].
Similar to animal  and fungal  systems, TOR-activated S6K contributes to translation-mediated growth responses in plants by several mechanisms. In both animals  and plants [48,69], TOR-activated S6K exhibits increased kinase activity towards ribosomal protein S6. Recent work in mouse models suggests that activated S6 contributes to growth through up-regulating transcription of mRNAs for proteins that contribute to ribosome biogenesis . In Arabidopsis, growth is tightly coupled to S6 levels [47,89]. TOR-activated S6K in plants has also been show to affect translation by promoting reinitiation of translation of mRNAs that contain multiple ORFs. Cauliflower mosaic virus-encoded TAV (transactivating/viroplasmin protein) promotes TOR-dependent activation of S6K, which in turn phosphorylates eIF3H (eukaryotic initiation factor 3H) to stabilize the association of the ribosome with mRNA upon transiting stop codons, thus enabling translational reinitiation at downstream ORFs that are essential for viral replication . A subsequent study  suggests that this form of TOR-mediated translational control may also be relevant to a large number of plant mRNAs in which small ORFs (μORFs) lie upstream of the primary ORF [90,91].
Other post-transcriptional mechanisms
In animal models, two factors associated with newly processed RNAs, the splicing factor AS2/ASF and the exon junction complex-associated SKAR (S6K1 ALY-REF-like), both contribute to translational outputs in a TOR-dependent manner. Although equivalent forms of regulation remain undefined in plants, many associated elements are well-conserved . Similarly, it is unclear whether plants have an equivalent of 4E-BP (eIF4E-binding protein), a well-defined target of mTOR, that, when phosphorylated, is released from eIF4E to up-regulate m7GTP cap-dependent translation initiation . Plants nevertheless present clear examples of cap-dependent translation that is supported by elements shared with animals and fungi. The possible existence of widely diverged or convergent forms of 4E-BP in plants is suggested by precedents in yeast [94,95] as well as TOR-dependent phosphorylation of human 4E-BP when transfected into plant protoplasts .
Novel forms of TORC1-mediated regulation that are likely to have relevance in plants have emerged from analyses of factors that regulate the transition from mitotic to meiotic development in S. pombe, which offered the first functional analysis of a RAPTOR-like protein in eukaryotes, but also led to the identification of a novel class of targets for direct phosphorylation by TOR. Genetic screens for mutations that suppressed a dominant form of Mei2p, a meiosis-promoting RNA-binding protein, recovered truncated forms of Mip1 (Mei2-interacting protein 1) [29,96]. Previous studies had shown that Mei2p regulates meiosis via a novel mechanism that requires its physical interaction with Mip1 as well as specific non-coding polyadenylated transcripts [96–98]. Mip1 was also shown to be essential for mitotic growth, which subsequently became understood in terms of its equivalence to the TORC1 element RAPTOR. More recent analyses indicate that Mei2p is a direct substrate for phosphorylation by TOR, which targets Mei2p for ubiquitin-mediated turnover, thereby preventing the onset of determinant meiotic development in favour of continued mitotic growth .
Although no obvious counterparts for Mei2p are found in animals, plants harbour a diverse family of Mei2-like genes that contribute to both vegetative and reproductive development. One of these, first defined by the TERMINAL EAR1 gene of maize , but later characterized in other plants [101–105], is expressed in meristematic regions where it is thought to promote more indeterminate patterns of growth. A second distinct group of Mei2-like genes are expressed more widely during vegetative and reproductive growth, and have been shown to contribute to both meiotic processes and the subsequent mitotic growth of resulting haploid gametophytes [106–108]. Physical interactions between AtRAPTOR3g and specific regions of the Mei2-like protein AML1 (Arabidopsis Mei2-like) have been shown in yeast two-hybrid studies .
Regulation of PP2A
Similar to other eukaryotes, regulation of protein phosphatase activity enables another set of TORC1-dependent outputs in plants. This mode of regulation was first described in yeast with the characterization of TAP42, which, when phosphorylated by TORC1, binds to PP2A to regulate its activity. Like yeast and animals, plants rely on PP2A to regulate a diverse range of processes involving serine/threonine-phosphorylation-dependent changes in protein activity . Initial evidence that plants use a similar mechanism was obtained with two-hybrid screens against PP2A, which yielded a gene encoding a TAP42-like protein that was termed TAP46 . RNAi-mediated knockdown of TAP46 produces a range of responses that mimic those obtained by inhibition of TOR activity, including reduced translation, autophagy and nitrogen remobilization . These data, together with experiments showing TOR-dependent phosphorylation of TAP46 in vitro, support the view that the protein is a direct target of TORC1. It is unclear whether the association of TAP46 inhibits or activates PP2A, and whether TAP46 plays a role in targeting this activity to particular substrates, as has been suggested in yeast.
Genomic and functional analyses suggest that autophagy relies on mechanisms that are largely conserved across eukaryotes. In yeast and mammals, TOR is thought to inhibit autophagy by phosphorylation of ATG13, blocking its activation of ATG1, a kinase whose activity promotes formation of autophagosomes. More recently, a second TOR-regulatory input that restricts autophagy has been described in which ULK1 (unc-51-like autophagy-activating kinase 1), the mammalian equivalent of ATG1, is also phosphorylated by TOR [112–114]. This modification of ULK1 effectively blocks an activating phosphorylation by AMPK (AMP-activated protein kinase). As activation of AMPK typically reflects reduced energy status, TOR mediated phosphorylation of ULK1 would provide a means for TOR to temper the activation of autophagy responses triggered by reduced energy levels. Functional counterparts to key autophagy elements have been described in both the single-celled alga Chlamydomonas as well as Arabidopsis [54,115,116], and results indicate that autophagosome formation is controlled in a similar manner via TOR and AMPK . It was demonstrated further that TOR-dependent regulation of autophagy occurred either downstream or independently of ROS-related signalling inputs that are often associated with stress .
Although TOR activity is likely to influence transcriptional processes in plants by a variety of mechanisms, two novel and relatively direct mechanisms have been described with conserved features that suggest the possibility of animal equivalents. By one mechanism, S6K-mediated phosphorylation of the conserved RBR1 (RETINOBLASTOMA-RELATED 1) protein triggers its nuclear localization where it inhibits the activity of E2F transcription factors, and thus represses expression of cell-cycle-promoting genes . This activity would provide a means to explain TOR-dependent increases in cell size promoted by nutrient-rich conditions.
Studies point towards a novel and more direct mechanism by which TOR may regulate cell division via phosphorylation of E2Fa. Physiological experiments suggest that many aspects of growth triggered by glucose depend on TOR activity , with the transcriptional activation of specific cell division-related targets by E2Fa playing a key role . Additional experiments suggest that E2Fa is a direct substrate of TOR, showing that it can be phosphorylated in vitro with TOR-containing immunoprecipitates, with analyses of non-phosphorylatable forms of E2Fa in planta suggesting that this TOR-mediated phosphorylation of E2Fa promotes cell proliferation.
UPSTREAM REGULATORY INPUTS FOR TOR IN PLANTS
In contrast with clear-cut examples of growth-related output mechanisms for TOR that appear to be broadly conserved across eukaryotic groups, upstream inputs appear to be more varied. Some degree of conservation could be expected for mechanisms that emerged early in common single-celled ancestors for tuning TOR outputs in response to generic inputs such as energy, nutrient and stress. By contrast, less similarity would be expected for mechanisms that arose later to communicate taxon-specific inputs, including those supporting distinct modes of nutrient acquisition, as well as separately evolved programmes for multicellular development. Further differences would probably reflect distinct repertoires of potential signalling elements between newly diverged major eukaryotic taxa that might support emerging forms of regulation. From this perspective, comparisons highlight a number of distinctions between taxa in how TOR activity is regulated.
Uncertain links for a conserved energy/stress-response module
Snf1 (sucrose-non-fermenting 1)/AMPK represents an ancient regulatory module that communicates cellular energy levels and associated forms of stress to downstream targets, including TOR . In well-characterized fungal and animal systems, Snf1/AMPK kinases are activated by low ratios of ATP to ADP and AMP, leading to the regulation of a large number of downstream targets, including the repression of TOR. Plants have clear counterparts of Snf1/AMPK [119–121], e.g. SnRK1 (Snf1-related kinase 1), which have a conserved heterotrimeric organization, can complement yeast Snf1 and show related target specificities, but their link with TOR remains largely undefined . Some type of SnRK1-mediated linkage between energy and TOR activity in plants would be consistent with the activation of TOR observed in response to light or sugar, with both blocked by inhibitors of mitochondrial ATP production .
In animals and many fungi, the SnRK1-mediated link between energy and TOR activity is partly indirect, mediated by the interaction of TSC1/2 (tuberous sclerosis complex 1/2) with Rheb (Ras homologue enhanced in brain) . Phosphorylation of TSC1/2 by SNF1/AMPK in response to low energy stress up-regulates the GTPase-activating activity of TSC1/2, which in turn converts a GTP-bound form of Rheb, a potent activator of TOR, to an inactive GDP-bound form. This indirect mode of regulation enables the integration of additional inputs, such as growth-factor-dependent phosphorylation of TSC1/2, which counteracts the AMPK-dependent repressive activity of TSC1/2 towards TOR. However, as plants lack any clear counterpart of TSC1/2 or Rheb, they may rely on other types of linkages such as that described in mammalian systems in which AMPK represses TOR activity directly through an inhibitory phosphorylation of RAPTOR . A similar lack of clear functional counterparts of Rheb and TSC1/2 in budding yeast, but not fission yeast, suggests that the activation of TOR by Rheb represents part of a regulatory module that has co-evolved with TSC1/2 .
The possibility that other types of GTPase might provide a Rheb-like activity to activate TOR has also been suggested by the analysis of TCTP (translationally controlled tumour protein), a GEF (guanine-nucleotide-exchange factor) that is common to most eukaryotic groups and whose activities have been linked to TOR outputs. In Drosophila, experiments suggest that TCTP is necessary for Rheb-dependent TORC1 outputs . However, subsequent studies have cast doubt on this model, showing a lack of functional interaction between TCTP and Rheb by several approaches [125,126]. In plants, which lack clear equivalents of Rheb, disruption or knockdowns of an Arabidopsis TCTP-like protein nevertheless show a range of growth-related phenotypes similar those reported for TOR-deficient plants . More recent studies have extended these results, with complementation of function shown between Arabidopsis and Drosophila TCTPs . Yeast two-hybrid analysis revealed specific binding of AtTCTP with a subclass of RAB GTPases, consistent with models in which non-Rheb GTPases may act in parallel or independently of Rheb to influence TOR activity.
Regulation of TOR activity by nutrients
Early work in yeast revealed rapamycin-sensitive TOR outputs for a diverse range of nutrient inputs, including carbon, nitrogen and phosphorus , but underlying mechanisms remain largely obscure, especially in plants. Unlike animals and many fungi where nutrients are acquired in more complex organic forms, plants rely largely on a tightly regulated uptake and assimilation of simple forms directly from the environment . Through ATP and reducing power obtained from light, carbon dioxide absorbed from the atmosphere is converted into sugars that serve both as starting material for synthesis of biomolecules and as a mobile energy currency. Nitrogen is typically secured through active uptake of inorganic forms via roots. In favourable light environments, inorganic forms, such as nitrate, are transported to shoot tissues and then converted into glutamine with energy and reductant from light-harvesting reactions. When light is more limited, the assimilation and conversion of nitrogen into organic forms depends on energy provided by the regulated utilization of sugar and starch reserves.
TOR-dependent growth in plants is tightly coupled to nitrogen availability, with the growth-promoting effects of inorganic nitrogen largely blocked by rapamycin . Whether this TOR-dependent growth reflects a direct response to inorganic nitrogen rather than organic derivatives in plants is unclear. Unlike animals, both plants and yeasts can use inorganic nitrogen to synthesize organic forms such as amino acids. In yeast, the regulation of TOR activity in response to nitrogen signal plays a central role in controlling the assimilation of nitrogen in the form of ammonium , but equivalent TOR-dependent mechanisms in plants are not well defined. In both animals and yeast, activation of TOR can also be achieved by activation of RAG (Ras-related GTP-binding) proteins by organic forms of nitrogen. In mammalian models, the amino-acid-dependent binding of RAGs to RAPTOR promotes the localization of TORC1 to a lysosomal domain, where activation by Rheb is thought to occur . In yeast, which contain vacuoles in place of lysosomes, a similar regulatory mechanism has been identified at the surface of yeast vacuoles , but it remains unclear whether a similar regulatory mechanism is conserved in the green lineage. No obvious counterparts to RAG/GTR GTPases can be found in full genome sequences from plants (B. Veit, unpublished work). Like Rheb, RAG/GTR GTPases group within the Ras subfamily of the Ras GTPase superfamily, which appears absent from plants . Localization requirements for TOR activation in plants that would offer further tests of a RAG/GTR-like mechanism are also not well defined. Alternative mechanisms defined in human cell culture and yeast models [132–134], in which amino acids activate TOR via the class III PI3K Vps34 (vacuolar protein sorting 34), also remain untested. Recent work in yeast has also highlighted a role for glutamine in supporting the sustained activation of TOR . This type of mechanism could be viewed as an attractive model for nitrogen-dependent TOR regulation in plants given the pivotal role of glutamine in the assimilation of inorganic nitrogen via the GOGAT (glutamine oxoglutarate aminotransferase) pathway.
With respect to carbon nutrition, recent work highlights an important role for sugar, showing that TOR-dependent growth in starch-depleted seedlings is tightly coupled to photosynthesis, and that TOR-dependent growth can be efficiently rescued in dark-grown seedlings by glucose . Knockouts of HEXOKINASE1, a key element of sugar-sensing mechanisms in plants, have relatively little effect on TOR outputs. Instead, inhibitor studies suggest that growth responses are more indirect, requiring glycolysis and mitochondrial activity. This dependence might reflect the activation of TOR via a SnRK/AMPK-based mechanism that senses ATP derived from the oxidative metabolism of glucose. The link with mitochondrial outputs might also be explained by increased production of specific TCA cycle intermediates such as 2-oxoglutarate (α-ketoglutarate), which have been shown to influence TOR activity in animals and fungal systems [135,136]. This type of regulation might also have relevance to responses to nitrogen, where glutamine-dependent inputs into the TCA cycle would influence signalling to TOR .
Distinctions in lipid-dependent mechanisms for TOR activation
Analyses of TOR in animals and fungi have highlighted membrane-based activation mechanisms , but their relevance to plants remains unclear given distinct lexicons of lipid-based signalling molecules [137,138]. For animals, receptor-activated class I PI3Ks play key roles in the activation of TOR via production of PtdIns(3,4,5)P3, which in turn leads to a membrane-based activation of PDK. Plants, although lacking class I PI3Ks as well as detectable levels of PtdIns(3,4,5)P3, do, however, exhibit gene activities that suggest possible alternative mechanisms for phospholipid-dependent activation of TOR. Counterparts of PTEN (phosphatase and tensin homologue deleted on chromosome 10), which in animals dampen PtdIns(3,4,5)P3-dependent responses through a PtdIns(3,4,5)P3 3-phosphatase activity  can be found in plants, but with activities against a distinct range of phosphoinositides. AtPTEN1, although displaying an in vitro phosphatase activity comparable with human PTEN, also exhibits an affinity for PtdIns3P. In vivo, specific expression of AtPTEN1 plays an essential role during pollen development , where its activity can be linked to autophagy and changes in the levels and localization of PtdIns3P . Two other related proteins, AtPTEN2a and AtPTEN2b, have been shown to have a wider range of expression across tissues and activity against a broader range of substrates, including PtdIns3P, PtdIns(3,4)P2 and PtdIns(3,5)P2, but not PtdIns(3,4,5)P3 . The distinct phosphoinositide specificity of plant PTENs compared with their animal counterparts is paralleled by differences in the binding and activation profiles of PDK, with plant forms binding a broader range of phosphoinositides as well as showing strong activation by PA (phosphatidic acid) as opposed to PtdIns(3,4,5)P3 [143,144].
Additional analyses have highlighted the potential relevance of these other phospholipid species to the regulation of TOR. Although both plants and fungi lack the class I PI3K for production of PtdIns(3,4,5)P3, they carry the related, but more simply structured, class III PI3K Vps34, which catalyses the formation of PtdIns3P. Initially identified in yeast , Vps34 has the hallmarks of a more ancient form of PI3K, including a near ubiquitous distribution across eukaryotes and contributions to a number of endomembrane-focused processes ranging from vesicle trafficking to autophagy. Evidence that Vps34 may mediate regulation of TOR activity has emerged from cell culture and yeast models [132,146], although in Drosophila and certain human cell lines, this form of regulation may have been lost or is obscured by parallel signalling inputs. In plants, although no direct link between TOR and Vps34 has been demonstrated, knockdown phenotypes for both genes that relate to growth and metabolism align well [141,147–149].
An intriguing aspect of models for Vps34-mediated activation of TOR with particular relevance to plants relates to PA. Evidence from yeast and mammalian cell culture systems suggests that previously described roles of Vps34 for amino-acid-dependent activation of TOR may occur via the localized production of PA mediated by PLD (phospholipase D) [133,134,150]. PtdIns3P generated by Vps34 is proposed to anchor PLD to the lysosome surface via a PX (phox homology) lipid interaction domain. This membrane-bound PLD would offer a means to produce PA, which itself has been shown to activate TOR [151,152], through binding the FRB domain . Additional work suggests an extension of this model in animal systems in which Rheb-dependent activation of TOR is mediated through Rheb-mediated activation of PLD [150,154]. Thus activation of TOR via a Vps34-dependent production of PA could be viewed as an ancestral mechanism, with Rheb-mediated regulation of PLD representing an additional mitogen/serum-sensitive regulatory module.
In plants, a possible role for PA in mediated regulation of TOR in response to stress is beginning to emerge starting with regulated changes in PA levels triggered by several forms of stress. Towards understanding potential sources of PA, Arabidopsis contains at least 12 PLDs, with ten classed as C2-PLDs (Ca2+-dependent PLDs), which appear specific to plants, and two others, containing PH/PX-PLDs (pleckstrin homology/PX PLDs), that show affinities with the PLDs of mammals . A functional analysis of one of the C2 class PLDs, PLDα3, suggests that its activation by hyperosmotic and salt stress enhances root growth and glucose sensitivity via PA-mediated activation of TOR . Under these conditions, PLDα3-deficient plants accumulate less PA, have lower levels of AtTOR transcripts and phosphorylated S6K, whereas plants overexpressing PLDα3 have higher levels of AtTOR transcripts and phosphorylated S6K. Together, these observations support models in which that PA derived from PLD activity contributes to TOR regulation. Such a mechanism would provide an attractive model to link TOR activity to a diverse range of stress-and hormone-related signalling outputs that rely on a diverse set of PLD genes found in plants.
More recently, a further example of lipid-mediated regulation has emerged in yeast where TOR activity depends on a third species of phosphoinositide, PtdIns(3,5)P2 . With both KOG1, the yeast equivalent of RAPTOR in TORC1, and its substrate S6K showing significant affinity for PtdIns(3,5)P2, a model has been suggested for PtdIns(3,5)P2 promoting TORC1 co-localization with its activators and substrates. Although factors that control levels of PtdIns(3,5)P2 are not completely understood, its synthesis is thought to depend on both Vps34, and the phosphoinositide kinase Fab1p/PIKfyve, which together catalyse the sequential phosphorylation of a phosphoinositde precursor. Although no links between PtdIns(3,5)P2 and TOR signalling have been seen in plants, Arabidopsis lines with reduced FAB1 activity exhibit enlarged vacuoles and other membrane-trafficking defects that can be linked to pollen sterility and impaired responses to auxin [158–160].
Intercellular signalling inputs to TOR
Although plants exhibit some commonality with animals with respect to intracellular signal transduction pathways, significant differences are expected for how signals would be communicated across the plasma membrane and relayed to TOR. Surveys of plant genomes highlight the lack of RTKs (receptor tyrosine kinases)  and well-conserved seven-pass GPCRs (G-protein-coupled receptors) [162,163], which in animals play key roles in activating TOR through a variety of phosphorylation-dependent mechanisms. In their place, plants feature a unique repertoire of hormonal signal transduction pathways that have no clear equivalents among other eukaryotes . By exploiting a versatile capacity for primary and secondary metabolism, plants fashion a wide range of small signalling molecules, from the relatively simple hydrocarbon ethylene to more complex hormones derived from amino acid, nucleotide, lipid, carotenoid, terpenoid and sterol precursors. Additional work has pointed to important roles of similarly diverse small peptide and RNA signalling molecules in mediating intercellular communication [165,166].
Receptor types for these signalling molecules are similarly varied. Auxin, gibberellin, jasmonate and strigolactone hormones rely on soluble receptors that target SCF (Skp, Cullin, F-box) complex-mediated regulation of turnover of proteins, many of which act as transcription factors to evoke hormone-associated responses. Membrane-associated receptors include bacterial two-component-like systems for ethylene and cytokinin, as well as RLK (receptor-like kinase)-type receptors for brassinosteroid and small peptides. RLKs are represented by hundreds of functionally diverse members in plants, compared with a handful of genes in animals, and which appear to be completely lacking in fungi . The highly diversified families of RLKs  found in plants could be expected to play especially significant roles in communicating extracellular signals, given the lack of RTK and GPCR families and downstream class IA and IB PI3Ks that support analogous forms of communication in animals. Although the vast majority of RLKs have not been functionally characterized, recent work suggests that at least some RLKs may rely on heterotrimeric G-proteins as intracellular effectors, in contrast with animals and fungi where this linkage is restricted to interactions with GPCRs .
Relatively little is known of how these fundamental hormone-response pathways might interface with TOR-dependent outputs. Transcriptional reporters for early responses to the growth-promoting hormones auxin and cytokinin are unaffected by treatments that impair TOR-dependent growth responses, suggesting that TOR-dependent checkpoints lie downstream of early auxin responses. Early work has shown that the expression of a TOR:GUS translational fusion is enhanced by auxin treatment [4,170]. In a separate study, auxin has been shown to activate TOR as monitored by Torin-sensitive phosphorylation of S6K . This up-regulation of S6K could be expected to reinforce auxin responses further, since many genes that encode auxin-response factor transcription factors contain μORFs whose inhibition of reinitiation is overcome by S6K-mediated phosphorylation of eIF3H, as described above.
Functional analyses of TOR signalling in animals and fungal models, together with fully sequenced eukaryotic genomes, have presented a list of potential plant TOR signalling elements whose roles have only begun to be tested in well-developed genetic models such as Arabidopsis. Studies to date have validated roles for TORC1-mediated control of protein synthesis and turnover in response to nutritional and stress-related cues, supporting the ancestral character of this regulated output. More recent work in plants, however, has revealed novel facets of TOR-mediated regulation, including phosphorylation-mediated regulation of E2F and RBR transcription factors by TOR and S6K respectively, as well as novel forms of TOR-regulated translation reinitiation. In addition, recent work in fission yeast showing that the RNA-binding protein Mei2p is a direct target for TORC1-mediated phosphorylation suggests that the diversified family of Mei2-like genes in plants may play key roles in mediating TOR outputs.
An intriguing, but poorly understood, aspect of TOR signalling in plants relates to how this ancient mechanism has been adapted to support unique strategies for nutrient acquisition and metabolism. With every green cell having the potential to make its own sugar via photosynthesis, it is perhaps not surprising that plants lack elements such as insulin receptors, which in animals serve to integrate TOR outputs with a centrally managed glucose economy. In their place, novel linkages with TOR in plants may help integrate their growth and metabolism with chloroplast-associated outputs. The uptake and assimilation of other nutrients, such as inorganic nitrogen, is likely to present further distinctions in how nutrient levels are coupled to TOR activity. The apparent absence of RAG/GTR systems described in animals and fungi draws attention to other, perhaps more basic, mechanisms that would draw on a more widely conserved set of players. Alternative mechanisms for TOR activation that rely on the more ubiquitous Vps34, PLDs and other agents for lipid modification offer promising candidates in this respect, and may also have relevance to a diverse set of extracellular inputs linked to growth and stress responses.
Understanding how TOR activity is coupled to multicellular programmes of development in plants presents some of the most exciting challenges. As might be expected for independently evolved mechanisms selected to support fundamentally distinct growth strategies, many differences are seen between plants and animals, yet these are somehow coupled with TOR activity in plants to achieve coherent, but flexible, growth programmes, as well as adaptive patterns of metabolic regulation. With respect to cellular growth, the apparent lack of TORC2 in plants focuses attention on alternative mechanisms that plants might use to co-ordinate polarized growth processes. Whether these rely on TORC1, novel TOR complexes or largely TOR-independent forms of signalling remain key questions. Linkages between TOR with extracellular inputs, including growth- and stress-related signalling factors also remain poorly understood. Clearly, a more complete understanding of these elements promises new insights into this ancient signalling pathway that are relevant to plants, but may also reveal significant aspects that have gone unnoticed in other eukaryotic models.
This work was supported by funding from AgResearch, the Royal Society of New Zealand and Organisation for Economic Co-operation and Development (OECD) to B.V., and Direction des Sciences du Vivant-Commissariat à l’Energie Atomique to C.R. and by grant ANR2011BSV6-01002 to C.R. and C.M.
AMP-activated protein kinase
ataxia telangiectasia mutated
eukaryotic initiation factor
huntingtin/elongation factor 3/PP2A subunit/TOR1
lethal with sec-13 protein 8
Mei2-interacting protein 1
phosphoinositide 3-kinase-related kinase
protein phosphatase 2A
phosphatase and tensin homologue deleted on chromosome 10
regulatory-associated protein of mTOR
Ras homologue enhanced in brain
rapamycin-insensitive companion of mTOR
reactive oxygen species
receptor tyrosine kinase
ribosomal S6 kinase
2A phosphatase-associated protein
translationally controlled tumour protein
target of rapamycin
transformation/transcription domain-associated protein
tuberous sclerosis complex
unc-51-like autophagy-activating kinase 1
vacuolar protein sorting 34