The Arabidopsis TOR kinase links plant growth, yield, stress resistance and mRNA translation

Dorothée Deprost, Lei Yao, Rodnay Sormani, Manon Moreau, Guillaume Leterreux, Maryse Nicolaï, Magali Bedu, Christophe Robaglia, Christian Meyer

Author Affiliations

  1. Dorothée Deprost1,,
  2. Lei Yao2,3,,
  3. Rodnay Sormani3,
  4. Manon Moreau1,
  5. Guillaume Leterreux1,
  6. Maryse Nicolaï3,
  7. Magali Bedu1,
  8. Christophe Robaglia3 and
  9. Christian Meyer*,1
  1. 1 Unité de Nutrition Azotée des Plantes, Institut Jean‐Pierre Bourgin, INRA Versailles, Versailles, 78000, France
  2. 2 Agro‐Biotechnology Research Center, Beijing Academy of Agriculture and Forest Sciences, PO Box 2449, Beijing, 100097, China
  3. 3 Laboratoire de Génétique et Biophysique des Plantes, CNRS‐CEA‐Université de la Méditerranée Faculté des Sciences de Luminy, 163 Avenue de Luminy, Marseille, 13009, France
  1. *Corresponding author. Tel: +33 1 30 83 30 67; Fax: +33 1 30 83 30 96; E-mail: meyer{at}
  1. These authors contributed equally to this work


Plants, unlike animals, have plastic organ growth that is largely dependent on environmental information. However, so far, little is known about how this information is perceived and transduced into coherent growth and developmental decisions. Here, we report that the growth of Arabidopsis is positively correlated with the level of expression of the TARGET OF RAPAMYCIN (TOR) kinase. Diminished or augmented expression of the AtTOR gene results in a dose‐dependent decrease or increase, respectively, in organ and cell size, seed production and resistance to osmotic stress. Strong downregulation of AtTOR expression by inducible RNA interference also leads to a post‐germinative halt in growth and development, which phenocopies the action of the plant hormone abscisic acid, to an early senescence and to a reduction in the amount of translated messenger RNA. Thus, we propose that the AtTOR kinase is one of the contributors to the link between environmental cues and growth processes in plants.


The modulation of growth rate in response to environmental cues such as nutrient availability is essential for survival. The main function of the conserved eukaryotic target of rapamycin (TOR) kinase is to promote cell growth in response to favourable conditions. In yeast and animal cells, TOR regulates numerous biological processes, including transcription and translation of ribosomal components, which collectively contributes to cell growth. The TOR kinase exists in at least two large protein complexes that act in recruiting its various substrates (Martin & Hall, 2005; Wullschleger et al, 2006). The first complex includes the TOR (TOR1 or TOR2 in yeast), RAPTOR/KOG1 and mLST8/GβL proteins, and is sensitive to rapamycin; the second complex contains TOR (TOR2 in yeast), Rictor/AVO3 and mLST8/GβL, and phosphorylates Akt/PKB in animal cells.

As immobile organisms, plants need to constantly adjust their growth—defined as an increase in organ size—and development—production of new organs—to nutritional information such as light, water or ion availability; however, the genes and mechanisms involved in the perception and transduction of these information are so far poorly known. Several components of the TOR signalling pathway are present in Arabidopsis, and are ideal candidates for operating the link between nutritional sensing and the regulation of growth (Menand et al, 2004). Some members of the TOR complexes such as the RAPTOR (Anderson et al, 2005; Deprost et al, 2005) and mLST8/GβL proteins exist in plants. There is also evidence that possible TOR substrates and regulators, such as S6 or other AGC kinases (Turck et al, 2004) and the 3‐PHOSPHOINOSITIDE‐DEPENDENT PROTEIN KINASE 1 (PDK1), are conserved in plants (Deak et al, 1999; Bögre et al, 2003; Otterhag et al, 2006). Furthermore, it has been reported that Arabidopsis PDK1 phosphorylates S6 kinase 1 and that, as in animal cells, RAPTOR interacts in vivo both with S6 kinase and with TOR through the HEAT repeats (Otterhag et al, 2006; Mahfouz et al, 2006).

Unlike algae such as Chlamydomonas (Crespo et al, 2005), higher plants seem to be resistant to rapamycin (Menand et al, 2002); therefore, the use of rapamycin to inhibit the activity of plant TOR has been precluded. Fortunately, several large collections of insertion mutants are available in Arabidopsis, and we have previously shown that disruption of the unique AtTor gene causes, as in Drosophila (Zhang et al, 2000) or Caenorhabditis elegans (Long et al, 2002), an early stop in embryo development (Menand et al, 2002).

By using Arabidopsis T‐DNA insertion lines in which the AtTOR gene is overexpressed and RNA interference (RNAi) lines with decreased levels of AtTOR expression, we show that Arabidopsis growth, seed yield, osmotic stress resistance, abscisic acid (ABA) and sugar sensitivity as well as polysome accumulation are positively correlated with levels of AtTOR messenger RNA.

Results And Discussion

AtTOR expression level influences growth and seed yield

To investigate the role of the AtTOR kinase in post‐embryonic growth and development, we first looked for hypomorphic (viable) tor alleles in the collections of T‐DNA insertion mutants of Arabidopsis. Arabidopsis lines carrying T‐DNA insertions upstream of the AtTOR translation start (as in the putative promoter or 5′ noncoding regions) segregated viable homozygous mutants (supplementary Fig S1 online). Insertion of the T‐DNAs resulted in an increased steady‐state level of the AtTOR mRNA either in the shoots and roots—mutants G548 and G166 from the Gabi collection (Fig 1A)—or only in the roots—mutants S7817 and S7846 from the Salk collection (Fig 1D). Both T‐DNAs carry strong promoters at the border facing the 5′ leader sequence that could drive an increased transcription of the AtTOR gene. Such an alteration of transcriptional regulations following a T‐DNA insertion just upstream of the translation start has already been described for the Arabidopsis WALL ASSOCIATED KINASE‐LIKE 4 (WAKL4) gene (Hou et al, 2005). Thereafter, we tried to obtain Arabidopsis plants that underexpressed the AtTOR mRNA by introducing an AtTOR‐derived hairpin construct under the control of the 35S promoter (constitutive RNAi; supplementary information online). Few transformants were obtained, but the 35‐7, 65‐1 and 36‐6 lines showed a modest decrease in the accumulation of AtTOR mRNA (Fig 1A,D). These data indicate a strong counter‐selection for the complete constitutive silencing of AtTOR and are consistent with AtTOR being an essential gene in plants.

Figure 1.

The AtTOR gene regulates plant growth. (A,D) Expression of the AtTOR gene in the leaves (A) and roots (D) of the control line (Col0 wild type) and in RNAi‐silenced (35‐7 and 65‐1) and RNAi‐overexpressing lines (Gabi mutants G166 and G548; Salk mutants S7817 and S7846). Plants were grown in vitro for 15 days and relative expression levels were determined by real‐time quantitative PCR (RT–PCR) with the control line as a reference (100 arbitrary units) in each experiment. The mean values±s.d. of three independent experiments are shown. (B) Col0 WT plants, an RNAi‐silenced line (35‐7) and the G548 AtTOR‐overexpressing line grown in the greenhouse for 36 days. (C,F) Leaf (C) and root growth (F) curves in hydroponic cultures of control Col0 (cross), G548 (diamonds), G166 (circles), 35‐7 (squares) and 65‐1 (triangles) lines. The mean values±s.d. of three independent experiments are shown and correspond to the weight of individual plants. (E) Length of the root system in control, AtTOR‐overexpressing and AtTOR‐silenced lines. Plants were grown in hydroponic conditions for 49 days. ND, not determined; RNAi, RNA interference.

Shoot and root growth was found to be enhanced in G166 and G548 AtTOR‐overexpressing plants that were grown either in soil or in controlled hydroponic conditions (Fig 1B,C,E,F). In the S7846 and S7817 mutant lines in which overexpression of the AtTOR mRNA was restricted to the roots, only root growth was found to be enhanced (data not shown). Conversely, the partly silenced 35‐7 and 65‐1 lines showed reduced shoot and root growth (Fig 1). To gain a better understanding of the differences in leaf size, we measured the size of rosette leaves and adaxial epidermal cells (Fig 2; supplementary Fig S2 online). The area of rosette leaves increased gradually as the expression of AtTOR was augmented, and a comparable increase in epidermal cell size was observed. Interestingly, it has recently been shown that the epidermis determines shoot growth in Arabidopsis (Savaldi‐Goldstein et al, 2007). Conversely, the elongation of etiolated hypocotyls of seedlings was not found to be affected by variations in AtTOR expression (supplementary Fig S3 online), which suggests that the AtTOR kinase does not control cell elongation, at least in etiolated hypocotyls. To determine the basis of the observed increase in root size for AtTOR‐overexpressing lines, the structure of the root tip was examined by confocal microscopy (supplementary Fig S4 online). No marked changes in either the size of root meristem or organization were detected. Seed production was also higher in the plants overexpressing AtTOR in shoots, whereas it was reduced in AtTOR‐silenced plants (Table 1). This difference was due to an increased or decreased growth and size of the inflorescence, and not due to changes in either seed or silique size (data not shown). It has already been reported that decreasing the level of the TOR kinase affects body size in Drosophila and C. elegans (Zhang et al, 2000; Long et al, 2002); however, this is the first evidence to our knowledge that elevated expression of the TOR gene can also result in an increased growth rate, fitness and size.

Figure 2.

AtTOR expression is correlated with the size of rosette leaves and epidermal cells. The size of rosette leaves and adaxial epidermal cells was determined for wild type (Col0), the AtTOR‐silenced 65‐1 and 35‐7 lines, and for the G548‐ and G166‐overexpressing lines. Leaf size (right axis) was measured on leaves 5 and 6 and averaged (n=4). Epidermal cell size (left axis) was measured on the adaxial side of leaf 5 (n=3 × 4). See supplementary Fig S2 and supplementary information online for details.

View this table:
Table 1. Production of seeds by Arabidopsis lines with increased (G548 and G166) or decreased (35‐7 and 65‐1) TOR expression

AtTOR controls resistance to osmotic stress

Nutrient availability is an important determinant of root growth (Malamy, 2005). To determine the role of AtTOR in linking the perception of nutritional information to root development, we studied the influence of the level of AtTOR expression on root size in response to nitrogen starvation or excess. Nitrogen starvation results in enhanced root growth, whereas an excess of nitrate inhibits root growth (Zhang et al, 1999) as a result of osmotic stress (Deak & Malamy, 2005). We observed that overexpression of AtTOR did not modify root length in N‐limited conditions when compared with control plants (Fig 3A,B). Conversely, when plants were grown in the presence of excess nitrate, the AtTOR‐overexpressing S7817 and S7846 lines showed a longer primary root (Fig 3A,B). This indicates that increased AtTOR expression relieves the inhibition of primary root growth by excess nitrogen.

Figure 3.

AtTOR expression is correlated with sensitivity to osmotic stress. (A) Col8 wild type and S7817 seedlings were grown for 2 weeks on growth media without nitrogen (left) or on media supplemented with 100 mM nitrate (right). (B) Col8 wild type and AtTOR‐overexpressing line S7817 were grown in vitro for 15 days with different nitrogen supplements (0, 35 or 100 mM nitrate). The length of the primary root was then measured at different time points. Values represent means±s.d. of ten individual plants. The differences in root growth are only statistically significant for 100 mM nitrate. (C) Col0 wild type, AtTOR‐silenced (35‐71 and 36‐6) and AtTOR‐overexpressing (G166 and G548) lines were grown in vitro for 15 days with 35 mM KCl, and the length of the primary root was determined at the end of the culture. Values represent means±s.d. of ten individual plants. (D) Seeds of Col0 wild type and of the AtTor‐silenced 35‐7 and 65‐1 lines were sown on growth medium containing 5 mM mannitol. Representative 35‐7 and 65‐1 seedlings are shown.

To determine whether this effect was specific for nitrate or was linked to a more general osmotic effect, we used potassium chloride to change the osmotic potential of the growth medium. The primary root growth of the silenced lines 35‐7 and 36‐6 was reduced under high salt concentration, whereas the AtTOR‐overexpressing lines G166 and G548 showed a longer primary root than the control line (Fig 3C). As for nitrate, the differences in primary root growth between these lines were much smaller when plants were grown in low‐salt medium (data not shown). Seedlings of the 35‐7‐ and 65‐1‐silenced lines also showed a higher sensitivity to osmotic stress after germination (Fig 3D). From these observations, we conclude that the level of AtTOR expression is inversely correlated with the sensitivity of the primary root length to salt concentrations and that a constitutive AtTOR expression might alleviate the effect of osmotic stress on root growth.

In Schizosaccharomyces pombe, a mutation in the TOR gene resulted in cells that are hypersensitive to osmotic stress (Weisman & Choder, 2001). In Arabidopsis, the expression of S6 kinases was found to be regulated by osmotic stress (Mizoguchi et al, 1995) and Arabidopsis plants overexpressing S6 kinase were also hypersensitive to osmotic stress (Mahfouz et al, 2006).

Silencing of AtTOR expression arrests plant growth

As levels of AtTOR mRNA were only partly reduced in constitutive RNAi lines, we investigated the effect of a more complete loss of AtTOR expression. We designed an ethanol‐inducible system for the expression of double‐stranded RNA based on the alc regulon of Aspergillus (Lo et al, 2005; supplementary information online). When silencing of the AtTOR gene was conditionally induced by ethanol at bolting, the amount of AtTOR mRNA was reduced to less than 20% of that of the control alca:GUS line (data not shown), and the growth of existing leaves was almost completely stopped when compared with control plants (Fig 4A). Ethanol‐induced silencing of AtTOR expression triggered an early yellowing, which is linked to chlorophyll breakdown and is typical of leaf senescence (Fig 4A). Leaves from these plants also accumulated very high concentrations of soluble sugars (Fig 4C). Furthermore, we measured two and three times more glutamine synthetase and glutamate dehydrogenase activity, respectively, in leaves from silenced plants (data not shown). These variations in enzyme activities and levels of soluble sugar are usually associated with leaf senescence and nutrient remobilization in Arabidopsis (Diaz et al, 2005). This indicates that AtTOR activity is needed to restrain senescence and nutrient recycling.

Figure 4.

Conditional silencing of AtTOR expression by ethanol‐inducible RNA interference phenocopies in plants the action of rapamycin in animal and yeast cells. (A) After 6 weeks of culture without ethanol, Arabidopsis plants (Landsberg erecta) carrying either an alcA:RNAi‐TOR construct (control line, RNAi4), a 35S:ALCR alcA:RNAi‐TOR construct (allowing for ethanol‐inducible silencing of AtTOR, RNAi6) or a 35S:ALCR alcA:GUS construct (GUS) were grown in the greenhouse for 10 days with or without 5% (v/v) ethanol (EtOH). (B) Landsberg erecta wild type (Le) and the progeny of two independent lines carrying a 35S:ALCR alcA:RNAi‐TOR construct were sown in vitro with (10 or 50 mM) or without ethanol. Representative seedlings are shown. (C) Comparison of carbohydrate contents (log scale) in the control (Le), GUS, RNAi4 and two independent AtTOR‐silenced lines as well as in the homozygous progeny of the RNAi6 line grown with ethanol. The RNAi6 line was grown for 10 days with or without ethanol (n=8), all other lines were grown for 27 days with ethanol (n=4). The mean values±s.d. are shown. (D) Close‐up of an RNAi6‐3 plantlet grown as in (B) with 50 mM ethanol. FW, fresh weight; RNAi, RNA interference.

When grown in vitro with increasing concentrations of ethanol, the silenced lines also showed an ethanol‐dependent developmental arrest (Fig 4B,D). Silenced plantlets showed etiolated and unexpanded cotyledons, reduction and swelling of the hypocotyl and a lack of root development, as was more pronounced as ethanol concentration increased. This phenotype is reminiscent of Arabidopsis plantlets blocked by osmotic stress in a post‐germinative checkpoint through ABA signalling or with an increased level of the ABA‐insensitive 5 (ABI5) bZIP protein (Lopez‐Molina et al, 2001). ABI5 is a crucial regulator of ABA signalling during post‐germinative growth, and plantlets with silenced AtTOR also resembled Arabidopsis mutants of KEEP ON GOING, an E3 ubiquitin ligase required for ABI5 degradation (Stone et al, 2006). Furthermore, AtTOR RNAi lines were found to be hypersensitive to high sugar concentrations, whereas plants overexpressing AtTOR were less affected than the control plants (supplementary Fig S5 online). High sugar concentrations can retard seedling growth through the action of ABA (Finkelstein & Gibson, 2002) and we indeed observed a positive correlation between AtTOR expression and sensitivity to ABA (supplementary Fig S6 online). When grown on 2 μM ABA, about half of the seedlings from the silenced 35‐7 line showed the same developmental arrest as those germinated on high concentrations of mannitol (Fig 3D) or when AtTOR expression was fully silenced (Fig 4D). Together, these results indicate that ABA or osmotic stress provokes a decrease in AtTOR expression or activity. The observation that the 35‐7 RNAi line is more sensitive to osmotic stress than to ABA can be explained by the fact that osmotic stress signalling is mediated only in part by ABA (Finkelstein & Gibson, 2002). Thus, these results show that AtTOR expression is required for post‐embryonic growth and might act as a relay for ABA signalling between environmental information and the post‐germinative growth processes.

Silencing of AtTOR decreases accumulation of polysomes

Cell growth is intimately connected with ribosome biogenesis and protein translation (Wullschleger et al, 2006). We detected a marked decrease in the amount of soluble proteins in ethanol‐silenced plants—6.5 mg and 8.3 mg of soluble proteins per g fresh weight of rosette for the RNAi6.1 and 6.2 lines, respectively, compared with 15.3 mg and 13.7 mg of soluble proteins per g fresh weight for the Landsberg erecta and alca:GUS control lines, respectively. The impact of AtTOR silencing on the efficiency of mRNA translation was then evaluated. On silencing, we observed a small but reproducible reduction in the abundance of high‐molecular weight polysomes (Fig 5A). A decrease in polysome accumulation was also obtained after rapamycin treatment of yeast cells or of Arabidopsis plants expressing the yeast FKBP12 protein (Powers & Walter, 1999; Sormani et al, 2007). A marked increase in the 80S peak corresponding to the complete ribosome was observed after rapamycin treatment of yeast cells, but not after silencing of the AtTOR gene. Interestingly, changes in the expression level of the Arabidopsis EBP1 (ErbB‐3 epidermal growth factor receptor binding protein) gene, which shows a high sequence conservation to human EBP1—a nucleolar and cytoplasmic regulator of ribosome assembly and translation—affected organ growth in a manner similar to variations in AtTOR expression (Horvath et al, 2006). A higher expression of the AtEBP1 gene was indeed detected in the AtTOR‐overexpressing line S7817 when compared with Col0 after hybridization of CATMA arrays (‐bin/projects/CATdb; D. Deprost & C. Meyer, unpublished data). These preliminary results were confirmed by quantitative real‐time reverse transcriptase–PCR and the level of the EBP1 mRNA was found to be around twofold higher and lower in the AtTOR‐overexpressing and AtTOR‐silenced lines, respectively, when compared with the level of EBP1 mRNA in control lines (Fig 5B). Furthermore, the expression of the AtTOR and AtEBP1 genes was closely correlated when compared across various developmental stages (Genevestigator database; Zimmermann et al, 2004; supplementary Fig S7 online). Therefore, the EBP1 protein could be one of the targets of AtTOR in Arabidopsis and act downstream of the TOR kinase on the mRNA translation machinery.

Figure 5.

Variations in AtTOR expression affect the messenger RNA translation machinery. (A) Absorbance profile at 254 nm of ribosomes purified by ultracentrifugation on a sucrose density gradient. The ribosomal pellet fraction was prepared (I) from shoots for the ethanol‐inducible RNAi6 line, and (II) from plantlets grown in vitro in liquid medium for the partly silenced 35‐7 line. Experiments were repeated at least three times and a typical experiment is shown for each line. Black lines correspond to control wild‐type plants (Col0 and Landsberg erecta for, respectively, the 35‐7 and RNAi6 lines). The positions of the 60S ribosomal subunits, monosomes (80S) and polysomes are indicated. (B) Increased and decreased expression of the Arabidopsis EBP1 gene in, respectively, AtTOR‐overexpressing (G548 and G166) and AtTOR‐silenced (35‐7 and ethanol‐induced RNAi6‐3) lines. RNA for relative quantification by real‐time RT–PCR was obtained from leaves. The EBP1 relative expression levels were determined as in Fig 1 with the control lines as reference (Col0 for G548, G166 and 35‐7; ethanol‐treated Landsberg erecta for RNAi6‐3) in each experiment. The mean values±s.d. of three independent experiments are shown. EBP, human ErbB‐3 epidermal growth factor receptor binding protein; OD, optical density; RT–PCR, reverse transcriptase–PCR.

In plants, the inactivation of AtTOR expression by inducible silencing phenocopies the action of rapamycin in animal and yeast cells—that is, an arrest of growth linked to a reduction of mRNA translation (Wullschleger et al, 2006). The use of conditional silencing of AtTOR expression is thus an efficient tool for decreasing TOR activity and will open new perspectives for the study of the TOR signalling pathway in plants.


Our data indicate that the TOR kinase could be important in the control of plant growth by exogenous information, partly through the regulation of mRNA translation and as a target of ABA signalling. Plant growth is an important yet poorly known biological process and only a few genes have been described that positively and negatively regulate the plant body size and the rate of growth (Horvath et al, 2006). The study of the AtTOR pathway is thus important for our understanding of the basic growth control mechanisms in plants.


Constructs. For constitutive RNAi, a 300‐bp fragment of the FKBP‐rapamycin‐binding (FRB) domain of AtTOR was amplified by PCR. The DNA fragment obtained was cloned in two opposite orientations in the pHANNIBAL plasmid under the control of the cauliflower mosaic virus 35S promoter. For the alcA:RNAi‐TOR construct, a 190‐bp fragment of the FRB domain of AtTOR was amplified by PCR and cloned in pHANNIBAL, in which the 35S promoter was replaced by the alcA promoter (O. Leleu, unpublished data). Induction of the ethanol switch was achieved by adding ethanol in the growth medium or by ethanol vapour from open Eppendorf tubes filled with 5% (v/v) ethanol in some pots. The 35‐7, 36‐6 and 65‐1 lines were obtained by transforming Arabidopsis (Col0 ecotype) plants with the construct described above for constitutive RNAi.

Plant material. The four T‐DNA insertion mutants used in our study were derived from the Col8 (Salk collection, or the Col0 (Gabi collection, http://www.gabi‐ accession lines. Plants were grown in the greenhouse under standard conditions.

Polysome preparation. For the experiment using the 35‐7 RNAi line, seeds (25 mg) were grown for 10 days in 20 ml of MS/2 medium with 1% sucrose at 25°C under constant illumination and shaking. For the ethanol‐inducible RNAi lines, control and ethanol‐treated shoots were collected without flower buds. See supplementary information online for details.

Supplementary information is available at EMBO reports online (

Supplementary Information

Supplementary Information [embor7401043-sup-0001.pdf]


We thank O. Leleu, P. Laufs and Syngenta for the ethanol‐inducible RNAi lines and constructs, and H.‐N. Truong and E. Diatloff for improving the manuscript. D.D. was supported by a grant from the Ministère de la Recherche et de la Technologie, L.Y. by a doctoral grant from the Association Franco‐Chinoise de Recherche Scientifique et Technique and R.S. by a doctoral grant from the Commissariat à l'Energie Atomique. This work was partly supported by a grant from Agence Nationale de la Recherche (ANR Blanc06‐3_135436) to C.M. and C.R.