TOR Is a Negative Regulator of Autophagy in Arabidopsis thaliana

Background Autophagy is a protein degradation process by which cells recycle cytoplasmic contents under stress conditions or during senescence; a basal level of housekeeping autophagy also occurs under non-stressed conditions. Although a number of genes that function in autophagy (ATG genes) have been identified in plants, the upstream components that regulate the plant autophagy pathway are still obscure. Target of rapamycin (TOR) is a negative regulator of autophagy in both yeast and animals, and homologs of TOR in plants control plant growth and protein synthesis. However, a role for TOR in regulation of autophagy in plants has not been demonstrated previously. Methodology/Principal Findings In this paper we used RNA interference (RNAi) to generate transgenic plants with reduced AtTOR transcript level. By observing monodansylcadaverine- (MDC) and GFP-AtATG8e-labeled autophagosomes, these plants were demonstrated to have constitutive AtATG18a-dependent autophagy. Reverse transcriptase-PCR also showed increased expression of some AtATG genes in the RNAi-AtTOR plants. Unlike autophagy induced by starvation or salt stress, an NADPH oxidase inhibitor did not inhibit the constitutive autophagy in the RNAi-AtTOR lines, indicating that AtTOR is either downstream of or in a parallel pathway to NADPH oxidase. Conclusions/Significance Together, our results provide evidence that TOR is a negative regulator of autophagy in plants.


Introduction
Upon exposure to stress conditions or during senescence, plants degrade cytoplasmic macromolecules inside the vacuole [1] by a process known as macroautophagy, or autophagy for short [2]. Autophagy is highly induced under several types of stress conditions [3,4,5,6,7,8,9]; a basal level of autophagy also exists in cells to constitutively remove unwanted materials [10,11]. Upon activation of autophagy, a double membrane structure called an autophagosome forms around the cargo to be degraded, and then delivers it to the vacuole where it is broken down and recycled [12]. Several specific functions of autophagy have been characterized in plants, including degradation of aggregated proteins in nutrient-starved plant cells [13] and proteins damaged during oxidative stress [6,14], and regulation of cell death during pathogen immune responses [15,16,17].
In Arabidopsis thaliana, many AuTophaGy (ATG) genes have now been characterized and found to be essential for autophagy. These have primarily been identified based on sequence similarity to yeast autophagy genes. For example, in yeast the autophagy proteins Atg2, Atg9 and Atg18 form a functional module, as Atg2 and Atg18 are required for Atg9 cycling [18]. In Arabidopsis, an Atg18 homolog, AtATG18a, has been shown to be required for autophagosome formation during multiple environmental stress conditions and also leaf senescence [5]. AtATG18a is expressed throughout the plant, with an increase in transcript level observed during conditions that upregulate the autophagy pathway [5,6]. In RNAi-AtATG18a transgenic plants with reduced AtATG18a transcript level, autophagosome formation was disrupted and the plants were more sensitive to autophagy-inducing stresses and displayed accelerated senescence [5,6,19]. Likewise, Arabidopsis ATG2 and ATG9 genes also function in autophagy, suggesting that the role of these genes may be conserved between species [5,8,10].
Although the identification and characterization of ATG genes in plants has greatly enhanced our understanding of autophagy, the upstream regulatory components in the plant autophagy pathway are still obscure. In Arabidopsis, autophagy is induced by multiple abiotic and biotic stresses, including nutrient deficiency, oxidative, salt and drought stresses and pathogen invasion [5,6,16,17,19]. A potential role for NADPH oxidase in regulation of autophagy under some conditions has been suggested by inhibitor studies; NADPH oxidase inhibitors block autophagy activation during nutrient and salt stress but not during osmotic stress, indicating that NADPH oxidase-dependent and -independent regulatory pathways may exist [19]. In yeast and animals, target of rapamycin (TOR) kinase was shown to be a negative regulator of autophagy [20,21]. Tor is a phosphatidylinositol 3kinase-related kinase that functions as a Ser/Thr protein kinase [22]. It is inactivated by rapamycin, which forms a complex with FKBP12 (FK506 binding protein); this complex binds Tor and inhibits Tor activities [23], thus being a valuable tool in determining Tor function. Tor also controls other processes that regulate growth in response to nutrient status, for example control of translation initiation by activating the ribosomal p70 S6 kinase and inhibiting the eukaryotic translation initiation factor 4E binding protein 1 (eIF-4E BP1) [24,25,26,27].
In yeast, two TOR genes have been identified; plants, mammals and other eukaryotes have only one TOR gene [21]. In both yeast and mammals, two TOR complexes exist, TORC1 and TORC2, each of which contains distinct TOR binding partners. Only TORC1 is sensitive to rapamycin [28]. Some binding partners have been identified, including Raptor [29], which binds to substrates and presents them to Tor for phosphorylation, and LST8 [30,31], which stabilizes the TOR complex. The Atg1/ Atg13 complex, which is required at an early stage of autophagy initation to induce autophagosome formation, has been identified as a Tor substrate in metazoans. Tor functions by phosphorylating Atg13 in a nutrient-dependent manner, although the relationship between and regulation of Tor, Atg1 and Atg13 is different between yeast and metazoans [32,33,34,35].
The TOR protein is conserved in plants [36] and RAPTOR homologues have also been identified in Arabidopsis [37,38]. However, disruption of the TOR gene is lethal and causes an early block in embryo development [36], impeding the analysis of TOR function in plants. In addition, Arabidopsis is insensitive to rapamycin, and this inhibitor therefore cannot be used to study the TOR pathway in this species [36,39]. Previous research with transgenic Arabidopsis plants with increased or decreased TOR expression level [40,41] showed that growth of root and shoot was correlated with TOR expression level, indicating a role in growth regulation. A recent study in the green alga Chlamydomonas reinhardtii showed that autophagy is induced upon rapamycin treatment, suggesting that regulation of autophagy by TOR may extend to photosynthetic species [42]. However, whether TOR regulates autophagy in multicellular plants has not been investigated.
In this study, our goal was to investigate whether TOR, a negative regulator of autophagy in yeast and mammals, also plays a role in plant autophagy induction. We demonstrate that RNAi-AtTOR plants with decreased expression of AtTOR have constitutive autophagy and increased expression of some ATG genes, even in the absence of stress conditions. This constitutive autophagy is dependent on the autophagy gene AtATG18a. We also show that, unlike stress-induced autophagy, an NADPH oxidase inhibitor does not inhibit the constitutive autophagy in these plants, indicating that AtTOR is most likely either downstream of or in a parallel pathway to NADPH oxidase. Taken together, our results indicate that AtTOR is a negative regulator of autophagy in Arabidopsis thaliana.

Generation of RNAi-AtTOR transgenic plants
Because disruption of the AtTOR gene (At1g50030) is embryo lethal [36], to study the role of AtTOR in autophagy, RNA interference (RNAi) was used to generate plants with reduced AtTOR transcript level. An RNAi-AtTOR construct was made by linking two inverted AtTOR gene specific regions with a 1 kb GUS spacer and driven by the Cauliflower Mosaic Virus 35S promoter [43]. Arabidopsis thaliana plants were transformed with this RNAi-AtTOR construct using the floral dip method [44] and screened for kanamycin resistance. Transformants with reduced AtTOR transcript level were identified by RT-PCR using AtTOR gene specific primers ( Figure 1A). Out of the 5 independent transgenic lines shown, named RNAi-1 to 5, lines RNAi-2 and RNAi-3 showed the greatest reduction in AtTOR transcript compared with the WT control, and were therefore selected for further experiments. All experiments were performed using T2 seedlings, which showed reproducible phenotypes that correlated with residual AtTOR expression level. Within a transgenic line, all individuals showed consistent AtTOR expression levels and phenotypes. Previous research has shown that plants with a partially silenced AtTOR gene show reduced growth (e.g. shorter root and shoot length, smaller rosette leaves) [40,41]. To confirm that the phenotype of the RNAi lines generated were consistent with previous reports, the root length and the fresh weight of seedlings were measured. WT, RNAi-2 and RNAi-3 seeds were germinated and grown on MS medium for 10 days with the plates oriented perpendicular to the ground. The RNAi plants were found to have a small but statistically significant decrease in root length when compared with WT ( Figure 1B),which is consistent with previous reports [40]. In yeast, TOR regulates growth and protein synthesis in response to nutrient availability [45,46]; therefore 10-day root length under nitrogen starvation conditions was also analyzed ( Figure 1B). The root length of RNAi seedlings was not only decreased compared with WT, but also decreased still further under nitrogen starvation conditions compared with control nutrient-rich conditions.
To measure the fresh weight, RNAi-2, RNAi-3 and WT seedlings grown on MS media were weighed at 5, 10, 15 and 20 days ( Figure 1C). From 10 days onward, both RNAi lines had a decreased fresh weight compared with the WT seedlings. To further investigate potential phenotypes of older plants in soil, RNAi-2 and WT plants grown in soil were observed throughout their lifespan. No significant differences were observed between RNAi and WT plants in seed volume, flower time and rosette size (data not shown). This differs from previous reports of reduced growth of TOR RNAi plants, possibly due to differences in growth conditions or extent of TOR silencing [40].

RNAi-AtTOR plants have constitutive autophagy
TOR is a negative regulator of autophagy in yeast and mammals [20,21]. Possible effects on autophagy in the absence of stress were therefore investigated in RNAi-AtTOR transgenic plants with decreased expression of AtTOR. Autophagy can be analyzed in Arabidopsis seedlings by staining with the fluorescent dye monodansylcadaverine (MDC), which selectively labels autophagosomes [47]. WT seeds and transgenic seeds of RNAi lines RNAi-1 to RNAi-5 ( Figure 1A) were germinated on MS plates and grown for one week, followed by MDC staining and fluorescence microscopy to visualize autophagy ( Figure 2A). In both RNAi-2 and RNAi-3 seedlings, which have the greatest reduction in AtTOR transcript level, constitutive autophagy was observed close to the root tip, seen as rapidly moving fluorescent puncta. Very few autophagosomes were seen in transgenic seedlings with less effective reduction of transcript level (RNAi-1, RNAi-4, RNAi-5) and in WT plants. This indicates that the constitutive autophagy phenotype correlates with a greater reduction in AtTOR transcript level.
To confirm the constitutive autophagy in RNAi-AtTOR plants, transgenic plants expressing the autophagosome marker GFP-AtATG8e [6] were crossed with RNAi-2 plants. Seeds from the cross were germinated on MS plates and grown for one week, then seedlings examined for autophagy activity; the autophagosomes were directly visualized under the fluorescence microscope via GFP fluorescence ( Figure 2B). Consistent with the MDC staining result, autophagosomes can be observed even under control conditions in the RNAi-26GFP-AtATG8e plants above the root tip areas, while the GFP-AtATG8e transgenic plants have a diffuse GFP signal as expected with very few autophagosomes visible. These results further confirm the constitutive autophagy in RNAi-AtTOR plants.
We considered two possibilities to explain the constitutive presence of autophagosomes in RNAi-AtTOR plants: either the decreased expression of AtTOR leads to the constant formation of autophagosomes, or the RNAi-AtTOR plants are deficient in delivery of autophagosomes to the vacuole for degradation. To distinguish between these two possibilities, and to further analyze the subcellular GFP-AtATG8e distribution and fate, the vacuolar H + -ATPase inhibitor concanamycin A (conc A) was used to inhibit the degradation of autophagic bodies in the vacuole [9,48]. Conc A inhibits trafficking of vacuolar proteins and prevents vacuolar protein degradation [9,49]. Autophagic bodies therefore accumu- late in the vacuole in Arabidopsis roots instead of being degraded after treatment with conc A [9]. Seven-day-old GFP-AtATG8e and RNAi-26GFP-AtATG8e seedlings were transferred to medium containing 1 mM conc A or dimethyl sulfoxide (DMSO) as a solvent control for 12 h, and then observed by both fluorescence microscopy using a FITC filter and differential interference contrast (DIC) microscopy ( Figure 2C). In control conditions, the vacuoles in both RNAi-26 GFP-AtATG8e and GFP-AtATG8e roots rarely contain any spherical structures in DIC images. GFP-AtATG8e roots have a diffuse cytoplasmic GFP signal while RNAi-26 GFP-AtATG8e roots contain several punctate GFP-labeled structures in the cytoplasm, consistent with Figure 2B. This indicates that the DSMO solvent did not induce autophagy. After treatment with conc A, the vacuoles in the GFP-AtATG8e roots contained some weak and diffuse GFP fluorescence, and as expected, occasionally contained a few GFP puncta corresponding to spherical structures in the DIC images, as plants have a basal housekeeping level of autophagy [10,11]. In contrast, the vacuoles in the RNAi-26 GFP-AtATG8e roots contained many GFP-AtATG8e-labeled puncta and spherical structures in the DIC images which have been shown previously to be autophagic bodies [9]. By comparing the GFP fluorescence with the DIC images in RNAi-26 GFP-AtATG8e roots after conc A treatment, most of the GFP-AtATG8e signal and the spherical structures localize inside the vacuole. This suggests that after treatment with conc A, autophagic bodies accumulate in the vacuole in the RNAi-26 GFP-AtATG8e roots instead of being degraded. The accumulation of autophagic bodies in the vacuoles of RNAi-26 GFP-AtATG8e plants suggests that autophagosomes are formed and successfully transferred into vacuoles in these plants. These results indicate that the constitutive autophagy in RNAi-AtTOR seedlings is most likely caused by the increased formation of autophagosomes, rather than a deficiency in delivery to or fusion between autophagosomes and vacuoles. These data therefore suggest that AtTOR may negatively regulate autophagy in Arabidopsis.
Some ATG genes are up-regulated in the RNAi-AtTOR plants A number of genes have been shown to be required for autophagy in Arabidopsis and several are upregulated under conditions that induce autophagy [50]. Previously, we have shown that under starvation, oxidative, salt and osmotic stresses, autophagy induction is correlated with the upregulation of the AtATG18a gene (At3g62770) [5,6]. In the RNAi-AtTOR plants, autophagy was observed even under control conditions. Therefore the expression of the AtATG18a gene was analyzed in these plants. AtATG9 (At2g31260) and the AtATG8 gene family are also essential for autophagy in Arabidopsis [8,9]; the expression of AtATG9 and several AtATG8 [AtATG8b (At4g04620), AtATG8e (At2g45170), AtATG8f (At4g16520), AtATG8h (At3g06420)] genes was also analyzed. WT and RNAi-2 seeds were germinated on control MS plates and grown for one week, RNA was extracted from 8 mm of the root tips where autophagy was seen and RT-PCR was performed. As shown in Figure 3A, in the RNAi-2 plants, where the AtTOR gene was partially silenced, the AtATG18a and AtATG9 transcript levels were increased compared with WT plants. In contrast, no increases were observed for the AtATG8b, AtATG8e, AtATG8f and AtATG8e genes. 18S RNA was used as a control for equal RNA levels in each sample. The expression level of each gene was quantified by densitometry of bands from at least three independent RT-PCR experiments, with the expression in WT plants set to 1. The mRNA level of both AtATG18a and AtATG9 was significantly higher in the RNAi-2 plants than in WT, but no difference was seen in expression of any of the AtATG8 genes tested compared with WT plants ( Figure 3B).The same RT-PCR experiment was also performed with the RNAi-3 line, with identical results to the RNAi-2 line (data not shown).
Previously, AtATG18apro: GUS transgenic plants were generated, in which the GUS reporter gene was expressed under the control of the AtATG18a promoter as an alternative way to observe the expression of AtATG18a [19]. In these plants, GUS activity is very low under control conditions, but increases substantially upon induction of autophagy. To confirm the RT-PCR result, AtATG18apro: GUS plants were crossed with RNAi-2 plants. The seeds were germinated on MS plates and grown for one week; the seedlings were then collected and submerged in GUS staining solution for 16 h, destained in 70% (v/v) ethanol for 16 h and observed with a light microscope ( Figure 3C). In the AtATG18apro: GUS plants, the roots have no visible GUS activity under control conditions. In contrast, in the RNAi-26AtATG18apro: GUS plants, GUS activity is evident throughout the roots. This indicates that the AtATG18a gene is up regulated in the RNAi-26AtATG18apro: GUS plants.
To further quantify the GUS staining results in Figure 3C, GUS activity was measured using the fluorometric substrate 4methylumbelliferyl-b-D-glucuronide (MUG) [51]. MUG can be hydrolyzed by GUS to produce the fluorochrome 4-methyl umbelliferone (MU). AtATG18apro: GUS and RNAi-26AtATG18apro: GUS seeds were germinated and grown on MS medium for 7 days with the plates oriented perpendicular to the ground. 0.05 g of root tissue was excised, followed by measurement of GUS activity by monitoring the fluorescence of MU at 0 h, 0.5 h, 1 h, 2 h, 4 h, 6 h, and 16 h ( Figure 3D). In both AtATG18apro: GUS and RNAi-26AtATG18apro: GUS plants, the fluorescence increased gradually over time. However, the fluorescence increase was much greater in the RNAi-26AtATG18apro: GUS plants. After 6 h, the fluorescence in RNAi-26AtATG18apro: GUS plants was more than 3 fold higher than in the AtATG18apro: GUS plants, and increased to almost 8 fold higher after 16 h. This again shows that the RNAi-26AtATG18apro: GUS plants have higher GUS activity than the AtATG18apro: GUS plants, which is consistent with the GUS staining results.
These results together indicate that the AtATG9 and AtATG18a genes are up-regulated in the RNAi-AtTOR plants in the absence of stress.

AtTOR-regulated autophagy is dependent on AtATG18a
Since AtATG18a is required for autophagosome formation during stress and senescence, RNAi-AtATG18a plants with reduced AtATG18a transcript level are defective in autophagosome formation [5]. To determine whether AtTOR-regulated autophagy requires AtATG18a, the RNAi-2 plants were crossed with RNAi-AtATG18a plants. The seeds were germinated on MS plates and grown for one week, followed by MDC staining. As shown in Figure 4A, RNAi-2 plants have constitutive autophagy in control conditions, while RNAi-26RNAi-AtATG18a plants have no observable autophagy. This indicates that autophagy induced upon inactivation of AtTOR requires AtATG18a, in common with stress-induced autophagy.
To further confirm the absence of autophagosome formation in the RNAi-26RNAi-AtATG18a plants, conc A was used to assess the accumulation of autophagic bodies in the vacuole. Seven-dayold RNAi-2 and RNAi-26RNAi-AtATG18a seedlings were transferred to medium containing 1 mM conc A or DMSO as a control for 12 h, and then observed by DIC microscopy ( Figure 4B). In control conditions, spherical structures were absent in both RNAi-2 and RNAi-26RNAi-AtATG18a vacuoles. After treated with conc A, spherical structures accumulated in the RNAi-2 vacuoles but not the RNAi-26RNAi-AtATG18a vacuoles. These data again suggest the constitutive formation of autophagosomes in the RNAi-2 but not the RNAi-26RNAi-AtATG18a plants. This confirms that AtTOR-regulated autophagy is dependent on AtATG18a.

NADPH oxidase inhibitor does not inhibit autophagy in RNAi-AtTOR plants
Previous results showed that autophagy is induced by nutrient deprivation, senescence, high salinity, oxidative and osmotic stresses [5,6,19]. All of these conditions elevate the cellular ROS (reactive oxygen species) levels, which can act as signal molecules to activate stress response and defense pathways [52,53]. Plasma membrane NADPH-dependent oxidase is a major source of signaling ROS. The NADPH oxidase inhibitors DPI (diphenylene iodinium) and imidazole [54] inhibit nutrient and salt stressinduced autophagy, whereas autophagy induced by osmotic stress is insensitive to these inhibitors, indicating that autophagy is regulated by NADPH oxidase-dependent or -independent pathways, determined by the induction conditions [19].
To determine whether AtTOR works downstream or upstream of NADPH oxidase in the autophagy signaling pathway, the effect of the NADPH oxidase inhibitor imidazole [54] on RNAi-AtTOR plants was tested. One week old RNAi-2 and WT seedlings, germinated on control MS plates, were transferred to liquid MS medium plus or minus 20 mM imidazole. WT seedlings transferred to liquid MS medium plus 0.16 M NaCl plus or minus 20 mM imidazole were used as positive controls. To observe autophagosomes, MDC staining was performed ( Figure 5A). For the WT seedlings, no autophagosomes were observed either with or without imidazole treatment under the control conditions, demonstrating that as expected, imidazole does not induce autophagy. Upon exposure to 0.16 M NaCl, numerous autophagosomes were seen in WT plants in the absence of imidazole, whereas in the presence of imidazole, no autophagosomes were present, confirming that imidazole inhibits autophagy induction under salt stress conditions, consistent with previous results [19]. For the RNAi-2 plants, imidazole had no effect on the constitutive autophagy seen in these lines, with numerous autophagosomes seen both in the presence and the absence of this inhibitor. This indicates that the NADPH oxidase inhibitor imidazole does not inhibit the observed autophagy in the RNAi-2 plants.
To confirm the MDC staining results ( Figure 5A), one week old RNAi-26 GFP-AtATG8e and control GFP-AtATG8e seedlings were transferred to liquid MS medium plus or minus 20 mM imidazole. GFP-AtATG8e seedlings transferred to liquid MS medium plus 0.16 M NaCl plus or minus 20 mM imidazole were used as positive controls. Autophagosomes can be directly visualized via GFP fluorescence ( Figure 5B). In the control condition, no autophagosomes were observed either with or without imidazole treatment. Autophagy was induced in the presence of 0.16 M NaCl, whereas adding imidazole inhibited the autophagy induction. In RNAi-26 GFP-AtATG8e seedlings, GFP-labeled autophagosomes were seen both with and without the imidazole. These results are consistent with the MDC staining, again suggesting that imidazole does not inhibit autophagy in the RNAi-2 plants.
To confirm that the GFP-AtATG8e structures in the RNAi-2 plants are autophagosomes, one week old RNAi-26GFP-AtATG8e seedlings were incubated in the presence of 1 mM conc A or DMSO for 12 h, each plus or minus 20 mM imidazole. Autophagic bodies were then observed by both fluorescence microscopy using a FITC filter and differential interference contrast (DIC) microscopy ( Figure 5C). In both the presence and absence of imidazole, many punctate GFP-AtATG8e signals were observed. After adding conc A, in both the presence and absence of imidazole, many punctate GFP-labeled spherical structures were observed. By comparing the GFP fluorescence with DIC images, both GFP-AtATG8e and the spherical structures in DIC images were localized inside the vacuoles. This indicates that conc A treatment led to the accumulation of autophagic bodies in the vacuoles both with and without imidazole treatment, again confirming that the NADPH oxidase inhibitor imidazole does not inhibit the constitutive generation of autophagosomes in the RNAi-26 GFP-AtATG8e plants. Together, these results suggest that AtTOR is either downstream of NADPH oxidase, or works in an NADPH oxidase-independent pathway.

Discussion
Although several autophagy-related genes have been identified and characterized in plants during the past decade, the upstream regulatory components in the autophagy pathway are still unclear. somes. B 7-day-old GFP-AtATG8e and RNAi-26 GFP-AtATG8e seedlings were transferred to liquid MS medium plus or minus 20 mM imidazole for 4 h, or GFP-AtATG8e seedlings were transferred to liquid MS medium plus 0.16 M NaCl plus or minus 20 mM imidazole for 4 h. Autophagosomes were detected by fluorescence microscopy via a FITC-specific filter. Arrows indicated GFP-labeled autophagosomes. Scale bar = 50 mm. C Seven-day-old RNAi-26 GFP-AtATG8e seedlings were transferred to liquid MS medium containing 1 mM concanamycin A (+concA) or DMSO (-concA) as a solvent control for 12 h, or liquid MS medium plus 20 mM imidazole containing 1 mM conc A or DMSO for 12 h followed by both fluorescence microscopy and DIC microscopy. Scale bar = 50 mm. doi:10.1371/journal.pone.0011883.g005 Autophagy has been reported to be activated in the unicellular green alga Chlamydomonas reinhardtii upon rapamycin treatment [42], suggesting that TOR negatively regulates autophagy in this species. However, due to the insensitivity of Arabidopsis to rapamycin and the lethality of the knockout mutant, the study of TOR function in the autophagy pathway in land plants has been difficult [36,39]. To overcome these obstacles, we generated RNAi-AtTOR transgenic plants to reduce AtTOR transcript level. These lines were first confirmed to have reduced growth phenotypes ( Figure 1B), consistent with previous research [40], and the reduced root growth phenotype was enhanced upon nitrogen starvation. TOR regulates multiple pathways to coordinate the response to nutrient availability, and decreased TOR activity during starvation is predicted to both increase the activity of the autophagy pathway and decrease growth-promoting pathways such as translation [55]. In two independent RNAi-AtTOR lines, constitutive autophagy was observed in roots under control conditions, in contrast to WT plants in which autophagosomes are rarely seen in the absence of stress (Figure 2). The autophagy-related genes AtATG18a and AtATG9 were shown to have an increased transcript level in the RNAi-AtTOR lines ( Figure 3A, 3B). AtTOR-regulated autophagy was dependent on AtATG18a, as RNAi-AtTOR6RNAi-AtATG18a plants do not show constitutive autophagy (Figure 4). Finally, an NADPH oxidase inhibitor did not inhibit autophagy in RNAi-AtTOR plants as it does under nutrient and salt stress conditions ( Figure 5), suggesting that AtTOR is either downstream of or in a pathway that is parallel to NADPH oxidase.
There is considerable evidence that TOR is an upstream regulator of autophagy in numerous species. Tor suppresses starvation-induced autophagy in the Drosophila fat body [56] and yeast Tor controls autophagy by responding to nutrient availability [32]. In the green alga Chlamydomonas reinhardtii, which is sensitive to rapamycin, inhibition of TOR by rapamycin leads to an increase in vacuole size and formation of ATG8-decorated autophagosomes, consistent with a role in regulation of autophagy [42,57]. In plants, TOR was shown previously to sense nutrient availability and regulate plant growth [40]. However, little evidence has directly shown a relationship between TOR and autophagy in multicellular plants. In this study, we provide evidence that TOR also controls autophagy, in addition to growth, in plants.
When autophagy is induced by stress conditions, autophagosomes are observed in the roots [5,6]. Autophagosomes were observed in the area close to the root tip in the RNAi-AtTOR seedlings, primarily in the division and elongation zones. TOR is most highly expressed in rapidly dividing and growing tissues such as meristems, embryos and endosperm [36]. It is therefore likely that the decreased expression of the AtTOR gene has the greatest effect on rapidly growing and expanding cells and therefore caused the induction of autophagy in these cells. It is clear from this and previous research [36,40] that TOR signaling is only partially suppressed in the RNAi lines described, and a complete disruption of TOR activity may be necessary to see the full effect of TOR inhibition on autophagy. In addition, the fluorescence assays used are technically difficult in shoot tissues due to the higher autofluorescence, making assessment of possible induction of autophagy in shoot cells problematic. It is possible that autophagy is also activated in rapidly growing shoot tissues in the RNAi lines. The GUS staining results shown in Figure 3C suggests an increased expression of AtATG18a throughout most of the root in the RNAi-AtTOR plants; this suggests that AtTOR activity is affected in upper parts of the root as well as root tips. One explanation is that the increased expression of AtATG18a in certain root regions does not necessarily correspond to the formation of autophagosomes. Further reduction of AtTOR expression level may be required for autophagosome formation, or other factors may also contribute to regulation of autophagy induction.
In general, Tor proteins function as Ser/Thr protein kinases [22]. Tor signals through a combination of direct phosphorylation of downstream targets and repression of phosphatase activity [58]. Some possible substrates have already been identified in plants, such as a meiosis signaling molecule AML1 (Arabidopsis Mei2-like 1) [59] and EBP1 (ErbB-3 epidermal growth factor receptor binding protein) [60] in Arabidopsis, and a translation regulator S6K (ribosomal p70 S6 kinase) [39,61], but their relationship to autophagy induction is not known. The Atg1/Atg13 complex has been shown to be a Tor substrate in yeast and metazoans. Tor functions by phosphorylating Atg13 in a nutrient-dependent manner, although the regulation and functions of Tor, Atg1 and Atg13 are divergent when comparing different species, complicating the generation of an overall model for TOR function [32,33,34,35]. In Arabidopsis, three putative Atg1 homologues and two putative Atg13 homologues have been identified [62], but their functions and interactions with TOR are still unknown. Our study shows that at least some AtATG genes are under the control of AtTOR in the autophagy pathway. However, to better understand and further investigate the components in the autophagy pathway, future experiments may focus on testing autophagy induction by inactivating or activating AtTOR and identifying its downstream targets.

Generation of RNAi-AtTOR transgenic plants
The RNAi-AtTOR construct was generated as described by Chuang and Meyerowitz [43]. Sense and antisense fragments of AtTOR were amplified by RT-PCR (reverse transcription-polymerase chain reaction) using gene-specific primers ( Table 1). The sense fragment, a 1-kb GUS spacer gene fragment, and the antisense fragment were ligated into the plant T-DNA binary vector pCGN and driven by the Cauliflower Mosaic Virus 35S promoter. The RNAi construct was introduced into Agrobacterium tumefaciens strain GV2260 by electroporation [63], and then into Arabidopsis thaliana Columbia-0 plants by Agrobacterium-mediated transformation using the floral dip method [44]. The expression level of the AtTOR gene in each transformant was determined by RT-PCR and homozygous T2 transformant seeds with reduced AtTOR mRNA level were used for further studies.

RT-PCR analysis of AtTOR, AtATG18a and AtATG9
Total RNA was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and followed by DNase I treatment. The final RNA concentration was determined using a NanoDrop ND-1000 spectrophotometer; 1 mg RNA was used to generate cDNAs using Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA, USA). An oligo dT primer was used for AtATG18a and the AtATG8s (b, e, f, h), gene specific primers were used for AtTOR and AtATG9 (Table 1). Gene specific primers used for PCR are also shown in Table 1. The PCRs were run for 28 cycles with annealing temperatures of 50, 55, 55 and 60uC and extension times of 1, 1, 1 and 4 min for AtTOR, AtATG18a, AtATG8s and ATG9, respectively.
RT-PCR signals were quantified by densitometry. Individual bands were analyzed using Quantity One software (Bio-Rad Laboratories; Hercules, CA) using the volume analysis function. The relative signals were calculated with the wild-type control value set as 1 for each gene individually. The results shown are an average of at least 3 independent experiments.

MDC staining and microscopy
Wild-type and RNAi-AtTOR seedlings were stained with MDC as previously described [47]. Seedlings were incubated with 0.05 mM MDC for 10 mins, washed 3 times with phosphate buffered saline (PBS) and observed using a Zeiss Axioplan II compound microscope equipped with Axio Cam HRC digital imaging system (Carl Zeiss Inc., Göttingen, Germany). MDC fluorescence was visualized using a DAPI-specific filter and GFP fluorescence was visualized using a FITC-specific filter.

Concanamycin A Treatment
Seven-day-old seedlings grown on MS plates were transferred to MS liquid medium containing 1 mM concanamycin A or dimethyl sulfoxide (DMSO) as a solvent control for 12 to 16 h in the dark. The roots were mounted in water and then observed by fluorescence and differential interference contrast (DIC) microscopy.

Fluorometric assay
Fluorometric reactions for analysis of GUS activity were performed according to Jefferson et al. [51]. 0.05 g of root tissue was homogenized in 50 ml extraction buffer (0.1 M Tris-HCl pH 7.5, 0.3 M sucrose, 1 mM EDTA, 0.1 mM PMSF) with liquid nitrogen. Plant extracts were centrifuged for 5 min at 4uC at 14,000 rpm. The supernatants were added to 1 mM MUG solution and at regular time intervals the reactions were terminated with 0.2 M Na 2 CO 3 solution. Fluorescence was then measured with a BIO-TEK Synergy HT multi-detection microplate reader, with excitation at 360 nm and emission at 460 nm. The microplate reader was calibrated with freshly made 4-methyl umbelliferone (MU) standards of 0.5 nm, 5 nm, 25 nm and 50 nm.

Inhibitor treatment
7-day-old seedlings grown on nutrient solid MS medium were transferred to MS liquid medium plus or minus 20 mM imidazole for 4 h. Table 1. Primers used for generating the RNAi-AtTOR construct and for RT-PCR analysis of AtTOR, AtATG18a, AtATG9 and AtATG8s.