bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 1 The phosphate starvation response recruits the TOR pathway to regulate growth in Arabidopsis cell cultures

Thomas Dobrenel1,5,6, Sunita Kushwah1, Umarah Mubeen2, Wouter Jansen3, Nicolas Delhomme4, Camila Caldana2, Johannes Hanson1,6 1Umea˚ Plant Science Centre, Department of Plant Physiology, Umea˚ University, S-90187 Umea,˚ Sweden 2Max-Planck-Institut fur¨ Molekulare Pflanzenphysiologie, Am Muhlenberg¨ 1, 14476 Golm, Germany 3Molecular Plant Physiology, Institute of Environmental Biology, Utrecht University, 3584CH Utrecht, The Netherlands 4Umea˚ Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences 5Current address: Umea˚ Plant Science Centre, Department of Forest Genetics and Plant Physiology, Swedish University of Agricultural Sciences 6Correspondence to: [email protected] and [email protected]

ABSTRACT In eukaryotes, TOR (Target Of Rapamycin) is a conserved regulator of growth that integrates both endogenous and exogenous signals. These signals include the internal nutritional status, and in plants, TOR has been shown to be regulated by carbon, nitrogen and sulfur availability. In this study, we show that in Arabidopsis the TOR pathway also integrates phosphorus availability to actively modulate the cell cycle, which in turn regulates the intracellular content of amino acids and organic acids. We observed a substantial overlap between the phenotypic, metabolic and transcriptomic responses of TOR inactivation and phosphorus starvation in Arabidopsis cell culture. Although phosphorus availability modulates TOR activity, changes in the levels of TOR activity do not alter the expression of marker genes for phosphorus status. These data prompted us to place the sensing of phosphorus availability upstream of the modulation of TOR activity which, in turn, regulates the cell cycle and primary metabolism to adjust plant growth in plants. Keywords: TOR kinase, Cell division, Phosphorus starvation, Cell culture !

INTRODUCTION modulation of cell division in Arabidopsis. More recently, the characterization of a TOR-dependent phosphorylation The antibiotic rapamycin was described in 1991 as having site in the E2F cell cycle transcription factors and the the capacity to arrest cell division and block cell cycle pro- regulation of CycB1;1 expression by TOR expression (Li gression in yeasts (Heitman et al., 1991). Its binding target, et al., 2017; Xiong et al., 2013), as well as a genetic link conveniently called Target Of Rapamycin (TOR for short), between the TOR pathway and the YAK1 kinase resulting is conserved among all eukaryotic organisms (Dobrenel in inhibition of expression of the SIAMESE-RELATED- et al., 2016a) and forms heteromeric protein complexes to cyclin-dependent kinase inhibitors (Barrada et al., 2019; exert protein kinase activity resulting in the modulation Forzani et al., 2019), have further confirmed the existence of various pathways including protein translation, nutri- of this link. It has also been shown that TOR inactivation ent assimilation, and autophagy, as well as cell division, in well-fed Arabidopsis seedlings results in the repression acting as a master regulator of growth. In plants, the TOR of expression of genes involved in the cell cycle (Caldana complex includes the proteins TOR, RAPTOR (Regulatory et al., 2013). This is consistent with the findings of Xiong Associated protein of TOR) and LST8 (Lethal with SEC13 et al. (2013) that carbon-starved seedlings fail to re-initiate 8) (Deprost et al., 2005; Moreau et al., 2012). growth after being supplied with sugar in the absence of Strong expression of TOR in constantly dividing cells TOR activity and that cell cycle-related genes, which are (Menand et al., 2002) and drastic arrest of growth in re- induced after 2 hours of exogenous sugar supply, require sponse to inactivation of the TOR complex components TOR activity for their expression to be induced. (Deprost et al., 2005, 2007; Menand et al., 2002) supported an early hypothesis of a link between TOR activity and the bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 2

In addition to the sugar status, it has been shown highly specialized leaf mesophyll cells. This diversity com- that the activity of the TOR complex is modulated by a plicates the study of cellular responses to nutritional and broad range of signals both exogenous, like environmental environmental signals. Cell culture systems have been used stresses, and endogenous, such as plant hormones (Do- extensively in the past to study, for example, the cellular brenel et al., 2016a). While many studies have focused on response to sucrose starvation (Nicola¨ı et al., 2006). More the link between TOR and the sugar signaling pathway or recently, a similar approach has proven to be successful in light perception (Dobrenel et al., 2013; Pfeiffer et al., 2016; studying plant cell differentiation into tracheary elements Xiong et al., 2013), recent publications have highlighted and photosynthetic cells (Dubreuil et al., 2018; Pesquet a link with other nutrients (Canellas et al., 2018; Couso et al., 2010) or investigating sugar signaling pathways et al., 2019; Dong et al., 2017). Couso et al. (2019) showed (Kunz et al., 2014), as well as for the identification of new that, in Chlamydomonas reinhardtii, a limitation in phosphate interactors and direct targets of the TOR kinase complex availability results in a decreased level of LST8 protein as (Van Leene et al., 2019). In this study, we used a cell culture well as in a reduction in TOR kinase activity, and that the system to decipher the molecular roles of carbon and phos- psr1 (Pi starvation response1) mutant, which is defective phorus nutrition in the regulation of growth, particularly in responses to phosphate starvation, is partially resistant focusing on the integration of the TOR pathway in this to TOR inhibitors. As photoautotrophic organisms, plants cellular process. use light energy to convert CO2 into carbohydrates in their We first characterized the growth of the cell line as well leaves, while nutrient uptake by roots is necessary for as the medium composition under routine conditions of the biosynthesis of amino acids and nucleotides coupled normal growth. The cells present a typical growth curve to energy derived from photosynthesis and respiration composed of a lag phase of approximately 3 days followed during the day and night, respectively. Although sulfur by an exponential growth phase finishing around the 7th and glucose are assimilated in different organs, Dong et al. day of culture (which is the day on which cells are routinely (2017) showed that the modulation of the TOR signaling subcultured) and the cells then experience a rapid decrease pathway by sulfur availability is operated via the regu- in their growth rate and reach maximal biomass after 10 lation of glucose metabolism. Based on this and on the days of culture (figure 1A). It is important to note that, meristematic expression of TOR, we hypothesize that the although growth is not suddenly arrested after the end of signaling pathway(s) resulting in the modulation of TOR the exponential phase, only a modest biomass increase is activity in response to these nutritional levels may require observed after days 7-8, which correlates with the com- intercellular communication. plete exhaustion of sucrose from the medium (figure 1B). In this paper, we focus on understanding the tight Biomass accumulation after the exponential growth phase interconnection between perception of the nutritional sta- may then be supported by the presence of glucose and tus and modulation of TOR activity at the cellular level. fructose in the medium, potentially deriving from the con- We particularly focus on carbon nutrition as a source of version of sucrose by apoplastic invertases (supplemental energy and the phosphate nutrition that is indispensable figure 1), as well as by the nitrogen remaining in the for the production of nucleic acids and necessary to support medium since 1/5 of the initial pool of nitrate is still present growth. To do so, we took advantage of an Arabidopsis in the medium at this stage (figure 1B). cell suspension system, which was previously successfully For this study, we used 6-day-old cell cultures. This used to dissect the sugar signaling pathways and the es- corresponds to the stage before the cells exit the exponen- tablishment of photosynthesis (Dubreuil et al., 2018; Kunz tial phase and initiate multi-nutrient starvation responses, et al., 2014). Employing a similar system has proven to be although the medium is already nearly depleted of sucrose useful for the identification of new direct targets of the TOR (figure 1B, supplemental figure 2, supplemental table 2). complex (Van Leene et al., 2019). We believe that using a We diluted the cells at a 1:1 ratio in double strength MS system of rapidly dividing cells mimics the meristematic medium with or without sucrose and with or without phos- TOR responses better than whole seedling or plant-based phorus and combined this treatment with a simultaneous systems where meristematic activity is considerably diluted application of a 2 µM concentration of the TOR inhibitor among other cell types. Moreover, the use of this system AZD-8055 (figure 1C). This created control conditions in enables time-resolved experiments as well as the replicates which cells were fed with non-limiting amounts of nutri- necessary for a systems biology approach. To better un- ents and conditions in which cells were starved of phos- derstand the cellular effects of nutrient starvation and TOR phorus or sugars, as well as conditions in which TOR was inactivation, we used a systems-based approach combining chemically inactivated under plethoric or starved condi- metabolomics, transcriptomics and lipidomics. With this tions. As expected, we observed that biomass accumulation comprehensive approach, we were able to show that the was reduced in a sucrose- or phosphorus-free medium TOR pathway integrates the cellular phosphate level to when compared to the plethoric conditions (figure 1D). actively modulate the cell cycle and growth in constantly Interestingly, this reduction in biomass accumulation was dividing cells. This is in contrast to sugar availability, at the same level as that in the cells cultivated in the pres- which controls growth in meristematic cells through an ence of AZD-8055 under plethoric nutritional conditions. apparently TOR-independent pathway. However, when AZD-8055 treatment was combined with sugar starvation, growth was totally abolished, a much RESULTS more severe effect than that observed after the individual sugar starvation or AZD-8055 treatments. In contrast to TOR inactivation, phosphate starvation or carbon star- the combined sugar starvation and AZD-8055 treatment, vation all result in decreased growth the application of AZD-8055 did not have any additive Similar to other multicellular eukaryote organisms, vascu- effect to that of phosphorus starvation in inhibiting growth. lar plants contain a large diversity of cell identities, rang- This suggests that the regulation of growth in response to ing from dividing non-differentiated meristematic cells to chemical inactivation of the TOR pathway and phosphorus bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 3

A D 300% Fig. 1. Cell culture growth parame- A = 521,2 mg/ml Plethoric ters. (A) Growth curve with determi- 500 Sugar Starvation nation of growth parameters (A: max- Phosphate Starvation imal growth value, λ: period of lag 400 and µ: maximum growth rate) calcu- 200% DMSO lated with the Grofit package. Dots AZD-8055 300 represent experimental data and the red line represents the theoretical µ = 103,7 mg/ml/day growth curve corresponding to the 200 cell biomass (mg/ml) cell biomass 100% calculated parameters. (B) Evolution

growth induction to growth induction T0 (%) of medium composition during the 100 cell culture growth period revealed by determination of nitrate and su- λ = 3,1 days 0 0% crose contents. (C) Experimental de- days of 0 24 48 72 0 2 4 6 8 10 sign: 6-day-old cells were diluted in culture hours post treatment fresh medium with or without sucrose

Nitrate and phosphate and concomitantly B 60 100 E Sucrose DMSO AZD treated with 2 µM AZD-8055. Cells were harvested over a timecourse 50 0 3 6 12 24 0 3 6 12 24 hrs 80 and samples obtained were analyzed P by a multi-omics approach. Arrows 40 60 represent harvest timepoints, green T diamonds represent metabolomics 30 P/T and lipidomics analysis, blue dia- 0,7 0,6 0,6 2,4 1,5 1,0 1,9 1,0 1,0 40 1,8 monds represent biomass measure- 20 ment, red diamonds represent tran- 20 scriptomic analysis. (D) Evolution of 10 F DMSO AZD

Nitrate concentration (mM) in the medium (mM) Nitrate concentration biomass increase after dilution in Sucrose concentration (mM) in the medium (mM) concentration Sucrose 0 3 6 12 24 0 3 6 12 24 hrs fresh medium. (E-F) Evolution of 0 0 days of 0 2 4 6 8 10 P RPS6 phosphorylation determined culture by measurement of the level of phos-

C T phorylated RPS6 (P) and of the total Nutrient re-supply: DMSO / AZD: RPS6 level (T) by western blot analy- Plethoric DMSO (mock) P/T 0,7 0,7 0,7 0,8 0,8 0,9 1,0 1,0 1,0 1,3 Sugar Starvation AZD-8055 2µM sis under plethoric conditions (E) or Phosphate Starvation after phosphate starvation (F). The phosphorylation ratio (P/T) was de- hours post treatment T0 24 48 72 termined for 3 independent replicates and normalized with respect to T0. 0 1 2 3 4 5 Arrowheads indicate the position of days of culture the 35 kDa marker.

availability may, at least partially, overlap, while in con- accumulation in our cell culture based system is similar to trast, the regulation of growth by modulation of the TOR that previously reported in plants. pathway and by sugar availability seem to be independent. To compare the effect of TOR inactivation with that of To assay TOR activity in the samples, we determined nutrient insufficiency, we then investigated the metabolic the level of phosphorylation of the S6 ribosomal protein profile of the cells after the phosphorus and sugar limi- (RPS6), which has been shown to be a proxy for TOR tation treatments as well as after chemical inactivation of activity (Dobrenel et al., 2016b). As expected, cells sub- TOR by application of AZD-8055. Hierarchical clustering of cultured with fresh medium showed increased TOR activ- the metabolic profiles showed a clear dichotomy between ity within the first 24 hours after the treatment (figure 1E, the samples, with the first group containing the non-treated supplemental figure 3). Conversely, AZD-8055-treated cells samples, the sugar-limited samples, and the early time- showed a clear reduction in the level of phosphorylated points during AZD-8055 treatment (3 hours, independently RPS6 protein. However, increased TOR activity was not of nutritional level) and the early timepoints of phosphorus observed in phosphorus-starved cells, in which the RPS6 limitation (3, 6 and 12 hours) (figure 2, supplemental table phosphorylation level was seemingly unaffected by AZD- 4). In the second group, on the other hand, we found all the 8055 treatment (figure 1F, supplemental figure 3). Overall, later AZD-8055 treated samples (independently of nutri- this confirms the link between phosphorus availability and tional level) and the late samples treated with phosphorus TOR activity, possibly placing perception of the phospho- limitation (from 24 hours of treatment onwards). This clear rus pool upstream of the modulation of TOR activity. separation between samples is largely supported by an elevated level of amino acids after chemical inactivation of TOR and a prolonged period of phosphorus limitation. TOR inactivation and phosphate starvation induce sim- In contrast, these metabolites show decreasing levels over ilar changes in amino acid levels time in the other conditions, including the sugar-limitation It has previously been shown that TOR inactivation in treatment. Interestingly, while the amino acid contents fol- plants grown under favorable conditions induces major lowed the same pattern of accumulation after AZD-8055- metabolic reprogramming, characterized mostly by a gen- dependent TOR inhibition and phosphate starvation, the eral accumulation of amino acids (Caldana et al., 2013; accumulation of TCA cycle intermediates seemed to be Dobrenel et al., 2013; Moreau et al., 2012; Ren et al., 2012). specific to the AZD-8055 treatment (independently of nutri- We investigated the metabolic profile of the cells after the tional level); it was not observed in response to phosphate application of AZD-8055 using a GC-MS approach (sup- starvation (supplemental figure 5). plemental figure 4). We observed a general accumulation It is important to note that, in the sugar-limited samples, of amino acids in the AZD-8055-treated samples, thus con- although sugar-limitation and AZD treatment were per- firming that the effect of TOR inactivation on amino acid formed at the same time, the cells needed approximately bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 4 A Nutrition: No starvation Sugar starvation Phosphate starvation AZD-8055: DMSO 2µM AZD-8055 Timepoint: T0 3h 6h 12h 24h 48h 72h

Nutrition AZD-8055 TCA cycle org. ac. and AA and derivatives Others 48 48 6 12 3 3 3 0 3 3 3 6 6 4848 48 48 6 6 6 derivativesSugars and Timepoint 7272 2412 24 12 7224 722472 72 1212122424 Lipids A phenylalanine A asparagine A threonine A tyrosine A glutamine T isocitrate T citrate T glyceric acid T aconi�c acid A glutamic acid A arginine S scyllo-inositol T malic acid T fumaric acid A isoleucine A valine A 2-amino-adipic acid S sucrose A ornithine O allantoin A cysteine A serine A his�dine O dehydroascorbic acid dimer S chiro-inositol Fig. 2. The metabolic response to O hydroxybenzoic acid A pyroglutamic acid TOR inactivation mimics phospho- L linoleic acid L adipic acid rus limitation. The metabolic pro- L octadecanoic acid A spermidine file was determined by GC-MS after O ethanolamine phosphorus limitation, sugar limita- O squalene L stearic acid tion and/or TOR inactivation (AZD- O campesterol L arachidonic acid 8055). (A) Hierarchical clustering L eicosanoic acid analysis showing the relative abun- A GABA T alpha-kg dance of each metabolite over the S xylulose T succinate timecourse in the different conditions log2(FC) after normalization to T0. The letters on the right correspond to the differ- -3 -2 -1 0 1 2 3 ent metabolic categories (B) Evolu- B tion of relative content of 3 metabo- phenylalnine valine sucrose lites representative of the clusters 3 3 identified in A (annotated in bold and 2 0 2 indicated by green arrowheads). The 1 -2 1 x-axis is the time in hours and the y- 0 -4 axis is the abundance relative to T0 -1 0 0 20 40 60 0 20 40 60 0 20 40 60 after log2 transformation. (n=3)

24 hours to completely exhaust their endogenous sucrose TOR inhibition leads to distinct shifts in the lipidome, pool. This suggests that up to 24 hours, the endogenous under phosphorus or sugar starvation levels of sugars were enough to maintain the pool of su- crose and that there was actual sugar starvation only from Since it has been previously reported that TOR inactivation this timepoint onwards (figure 2B). We can thus divide as well as phosphorus limitation induces an elevation the cellular response into two phases: the first 24 hours, in the level of tri-acyl glycerides (Caldana et al., 2013; corresponding to a cellular response to an exogenous lim- Couso et al., 2019); that the pho1 (Pi homeostasis1) mutant itation, and the later timepoints, corresponding to actual which is deficient in Pi transport has reduced levels of sugar starvation. phospholipids (Rouached et al., 2011), and that phosphorus limitation leads to strong overexpression of several genes Based on the similarity of the response to chemical involved in lipid metabolism (Morcuende et al., 2007), we inactivation of TOR and to phosphorus limitation at the investigated the lipid content of the cell cultures during metabolomic level as well as on the absence of any additive the timecourse of response to TOR inactivation in the effect between these two treatments, these data confirm the context of nutrient limitation. We observed that the overall previously mentioned hypothesis of a link between phos- lipid content of the cells at the beginning of the exper- phorus availability and TOR activity. This further suggest iment (compared to the other samples) shows elevated that TOR activity is regulated in response to phosphorus levels of non-phosphorus glycoglycerolipids (digalacto- availability and that TOR regulates the level of amino syldiacylglycerol DGDG and sulfoquinovosyldiacylglyc- acids, based on the metabolic response to TOR inactiva- erol SQDG) as well as of tri-acyl glycerides (TAG) while tion compared with phosphorus limitation. In contrast to most of the phosphoglycerolipids (phosphatidylcholine phosphorus starvation, sugar starvation responses could PC, phosphatidylethanolamine PE, phosphatidylinositol PI not be linked to any increase in the amino acid level; rather and phosphatidylglycerol PG) were at rather low levels there was a decrease. Indeed, prolonged carbon starvation compared to those of the other samples (supplemental results in depletion of most amino acids and TCA cycle figure 6A). intermediates, which may be possibly consumed to pro- Interestingly, when we followed the evolution of the vide energy. Overall, this supports the hypothesis that the lipid complement in the different conditions over time, we regulation of growth after nutrient re-supply follows two noticed that the nutritional level had a strong influence (at least partially) independent routes: one which is sugar whereas there was only a modest effect of TOR inactivation dependent and a second which is phosphorus- and TOR- (figure 3A, supplemental figure 6B). The samples clustered dependent. in three main groups, reflecting the nutritional status of bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 5 A Nutrition: No starvation Sugar starvation Phosphate starvation AZD-8055: DMSO 2µM AZD-8055 Timepoint: T0 3h 6h 12h 24h 48h 72h

Nutrition AZD-8055 7224 72 4848 666 12 12 12 3 3 3 3 6 48 72242424 12 72 72 4848 12 12 2424 366 3 0 72 48 Timepoint DAG DGDG LysoPC MGDG

PC

PE

PG PI SQDG

TAG

log2(FC)

-3 -2 -1 0 1 2 3

B 6 a

a

b Fig. 3. Phosphate starvation and a a b a sugar starvation have opposite ef- a a a a fects on the accumulation of lipids, b bc c c which is not affected by TOR inactiva- a a a tion. (A) Hierarchical clustering anal- TAG level FW) (units/mg a b b c ysis showing the relative abundance bc bc c c c bc of each lipid over the timecourse in c c the different conditions after normal- c c ization with respect to T0. The names on the right correspond to the differ- d d d d ent lipid categories (B) Evolution of TAG content. The letters above the AZD-8055 0 - -+ + - + - -+ + - + - -+ + - + - -+ + - + - -+ + - + - -+ + - + histograms represent the result of an Pi 0 + + + +-- + + + +-- + + + +-- + + + +-- + + + +-- + + + +-- Suc 0 + +++ - - + +++ - - + +++ - - + +++ - - + +++ - - + +++ - - ANOVA test followed by a post-hoc Time 3h 6h 12h 24h 48h 72h Tukey analysis by timepoint (n=3)

the medium, with clear separation between the late sam- increased by the late timepoints. Interestingly, contrarily to ples of sugar starved cells in one group, the samples of what we expected based on the literature (Caldana et al., phosphorus starved cells in a second group and the other 2013; Couso et al., 2019), TOR inactivation by application of treatments in a third group. The sugar-starved cells were the inhibitor AZD-8055 did not induce any increase in the characterized by a severe depletion in diacylglycerols and TAG level; if anything it tended to repress the late increase triacylglycerols (DAG and TAG), while these lipids were in the TAG level after prolonged phosphorus limitation. found at an elevated level in the phosphorus-limited cells. The phosphoglycerolipids (PC, PE, PI and PG), in contrast, Phosphate starvation acts upstream of the TOR path- showed a gradual increase over time in the absence of way to regulate the cell cycle nutrient limitation but remained at a low level (and even As we have seen previously, there is a clear difference at tended to decrease) during phosphorus limitation. These both the metabolomic and the lipidomic levels between the variations are even more striking when we compare the early response to nutrient limitation and AZD-8055 treat- samples timepoint by timepoint (supplemental figure 6C). ment (3 to 12 hours) and the late response (after 24 hours). As mentioned, the level of TAGs was strongly influ- To better understand the biological processes influenced by enced by nutrient availability. In the absence of any nutrient the nutrient status and TOR at the metabolite and lipid limitation, the TAG level greatly decreased during the first levels as well as biomass accumulation, we performed 12-24 hours after nutrient resupply before increasing grad- transcriptomic analysis of the cells after 6 and 24 hours ually (figure 3B). This late restoration of the TAG pool did of treatment. not occur in the sugar-starved cells, in which the level of As expected, the re-initiation of growth stimulated by TAG further decreased at the late timepoints. On the other the supply of fresh nutrients was accompanied by drastic hand, in the presence of phosphorus limitation, the level transcriptional reprogramming, with more than a third of of TAG remained high at the early timepoints and further the total transcripts being deregulated as early as 6 hours bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 6

A 0-20 20200-20 0-20 20 0-20 20 0-20 20 0-20 20 0-20 20 0-20 20 0-20 20 B 0 6 24 Time (h) - 020 0-20 -Suc 20 0-20 -Suc 0 + + + + - - + + + + - - Suc 0 + + - - + + + + - - + + Pi 0 - + - + - + - + - + - + AZD CKL1 - 020 0-20 -0,543 20 0-20 0,408 -Pi -Pi CYCP2;1 CYCA3;3 CKL7 CYCA3;4 - 020 0-20 -0,448 0,811 AZD 20 0-20 0,427 0,787 AZD KRP3 KRP5 CYCP4;3 CYCH;1 -Suc -Suc E2Fc - 020 0-20 0,172 0,698 0,958 20 0-20 0,800 0,704 0,885 Dpa AZD AZD Rb CYCD2;1 CDKD;1 -Pi -Pi -0,521 0,958 0,919 0,839 0,8870,446 0,956 0,887 CDKB2;1 AZD AZD CYCA;1 DEL3 CDKB1;1 CYCD3;2 0-20 20200-20 0-20 20200-20 0-20 20200-20 0-20 20200-20 0-20 20200-20 0-20 20200-20 0-20 20200-20 0-20 20200-20 0-20 20200-20 0-20 20200-20 CYCB2;3 CYCA1;1

- 020 0-20 KRP7 DEL1 WEE1 CYCA3;1 Seedl. 0,597-0,248 0,6700,756 0,0980,685 0,463 0,560 0,6270,461 CYCA3;2 AZD CYCD5;1 E2Fa CYCD4;1 -Suc -Pi -Suc -Pi -Suc -Pi AZD -Suc -Pi AZD CYCA2;3 AZD AZD AZD AZD CYCB1;2 CYCA2;4 CYCB2;2 -1 0-0,5 0,5 1 CYCB2;1 6 hours 24 hours of treatment CYCB1;3 CYCB2;4 Pearson correlation CKS2 CYCB1;4 CYCB3;1

0 6 24 Time (h) CDKB1;2

+ + + + - - C + + + + - - CDKG;2

0 Suc Dpb

+ + - - + + + + - - + + KRP4

0 Pi CYCT1;3

- + - + - + - + - + - + CDKE;1 0 AZD KRP1 AT5G65360 DEL2 E2Fb AT1G09200 CKL8 AT3G27360 CKL11 H3.1 AT5G10400 CKL9 AT5G65350 CDKD;2 CYL;1 AT5G10390 CYCL1;1 AT4G40030 CYCD7;1 H3.3 AT4G40040 CYCB2;5 AT5G10980 CYCT1;1 CKL13 AT1G13370 H3.3- CYCT1;5 like AT1G75600 CDKF;1 CenH3 AT1G01370 CDKC;1 CDKC;2 log2(fold change to T0) CDKA;1 CKL6 -1,5 -1 -0,5 0 0,5 1 1,5 CKL5 KRP2 CDKG;1 0 6 24 Time (h) CYCC1;1

D CYCD3;3 + + + + - -

0 + + + + - - Suc CYCP3;1

CKL12 + + - - + +

0 + + - - + + Pi SIAMESE

CDKB2;2 - + - + - +

0 - + - + - + AZD CKL4 CYCB1;1 AT5G43350 CYCA2;2 AT5G43370 CDC25 CKS1 AT5G43360 CYCT1;4 AT2G38940 KRP6 AT2G32830 CYCB1;5 AT5G43340 CYCD1;1 AT3G54700 CYCD4;2 CDKD;3 AT1G20860 CKL10 AT1G76430 CYCT1;2 AT3G26570 CKL15 AT5G14040 CYCC1;2 SDS AT3G48850 CKL2 AT2G17270 CYCP1;1 AT5G46110 CKL3 AT5G33320 CYCD6;1 AT5G54800 CYCD3;1 CYCA1;2 AT1G61800 CKL14 AT5G17640 CYCP4;1 AT1G73010 CYCP3;2 AT1G17710 CYCP4;2 log2(fold change to T0) log2(fold change to T0)

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3

Fig. 4. The transcriptomic response to TOR inactivation correlates with the transcriptional reprogramming after phosphorus limitation. (A) Correlation of transcriptional response with sugar limitation (-Suc), phosphorus limitation (-Pi) and/or TOR inactivation (AZD) after 6 hours of treatment (top left) or 24 hours of treatment (top right), and comparison with 8-day-old Arabidopsis seedlings treated with 1 µM of AZD-8055 for 24 hours (cell cultures are represented along the x-axis and seedling data along the y-axis). Only genes with a FDR < 0.01 were kept for this analysis (B) Expression profiles of cell cycle related genes (Menges et al., 2005) normalized to the expression profile in the T0 samples. Genes are clustered based on their responses to the different treatments (C) Expression profiles of phosphate transporters (Poirier and Bucher, 2002), relative to their expression values in the T0 samples (D) Expression profiles of H3-encoding genes, relative to their expression values in the T0 samples.

after the treatment (supplemental figure 7A, supplemental corresponding to several metabolic pathways including table 6). A similar number of genes was found to be deregu- lipid metabolism (supplemental figure 7D-G, supplemental lated after 24 hours of treatment and there was considerable material 1, supplemental table 7). Interestingly, some GO overlap between the deregulated genes at the two time- terms linked to developmental processes were found to points (supplemental figure 7B-C). Among the genes that be enriched among both induced and repressed genes at were up-regulated, we found strong overrepresentation of 6 and 24 hours after nutrient re-supply. Although these genes linked to cell division and the cell cycle, which must GO terms were present in these four datasets, they were be related to the re-activation of biomass accumulation, more significantly enriched in the genes upregulated after as well as several terms linked to gene expression such 24 hours of growth, which contained GO terms linked to as nucleotide and amino acid biosynthesis, transcription, embryonic and post-embryonic development. ribosome assembly, translation. In contrast, among the It is to be noted that the transcriptional reprogramming genes that were down-regulated, there was a prevalence observed after the nutritional resupply also occurs in all the of genes linked to various stresses and programmed cell different conditions (supplemental figure 7A, supplemental death. We also found an overrepresentation of GO terms table 6). However, this reprogramming is less extensive un- bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 7

P i -starvation Phosphate level induced genes Lipid pool Pi Pi Pi Pi Pi Pi Pi Pi Pi Pi Pi Pi

O Phospholipids Glycerolipids AZD-8055 N N HO N N N Sugars O O

TORC1

TCA Cell cycle cycle

Amino acids

Cell growth Transcripts

Growth

Metabolites/ Lipids

Fig. 5. Proposed model for the phosphate TORC1 signaling pathway in proliferating Arabidopsis cells. Phosphate availability regulates TORC1 which, in turn, modulates the expression of genes involved in the cell cycle. Maintenance of cell cycle activity contributes to depletion of the TCA cycle intermediates (and further to a diminution in amino acid levels) which is fueled by the sugar pool through glycolysis. Colored boxes (brown for transcriptomic data, green for metabolomic and lipidomic data and blue for growth measurement) correspond to the representative data presented in this paper (black lines = plethoric nutrition, turquoise lines = phosphate starvation, pink lines = sugar starvation, dashed lines = AZD-8055 treatment). In the signaling cascade, arrows represent induction and activation, T-shaped arrows represent repression and the arrow with a dot at the end represents regulation. der phosphorus limitation, suggesting that a subset of the 8055 in seedlings and the response to 6 hours of phospho- transcripts that are deregulated after nutrient re-supply are rus starvation in cell cultures. unaffected when phosphorus limitation is applied. When comparing the transcript profile after nutrient limitation As expected based on the high number of genes over- or AZD-8055 treatment to the corresponding non-treated lapping between the responses to phosphorus limitation timepoint, up to 7000 genes were found to be differentially and TOR inactivation, enrichment of many gene ontology regulated after 6 hours of phosphorus limitation, rising to terms was shared (supplemental figure 8D, supplemental 8000 after 24 hours (supplemental figure 8A). While similar table 8). Among the induced genes, there was overrep- numbers were found for the AZD-8055 treated samples, resentation of genes linked to stresses, lipid catabolism, there were far fewer genes that were deregulated after phosphorus metabolism and protein phosphorylation as sugar limitation (1693 genes deregulated after 6 hours, 2286 well as carbohydrate metabolism. The repressed genes after 24 hours). The genes deregulated after phosphorus were mostly enriched for terms such as DNA conforma- limitation or AZD-8055 treatment showed a considerable tion and organization, transcription and translation, cell overlap and correlation but only genes deregulated after cycle and embryonic development. Most of these terms 24 hours of sugar limitation overlapped with them (with were not found to be significantly enriched among the only 3.6% after 6 hours of treatment but up to 47% after genes that were differentially expressed in response to 24 hours of treatment) and showed a positive Pearson sugar limitation. The sugar starvation induced genes were correlation, although it remained modest (figure 4A, sup- enriched for the terms response to stresses (mostly after plemental figure 8B-C). We also compared the transcrip- 24 hours), cell wall organization (only after 6 hours), car- tomic data obtained from the cell culture subjected to bohydrate metabolism, cell growth and photosynthesis. We nutrient starvation and/or AZD-8055 treatment to the tran- also observed a marked shift between the two timepoints scriptomic response of seedlings treated with AZD-8055 with, for example, the response to photosynthesis being for 24 hours, and observed a strong Pearson correlation mostly enriched among the repressed genes after 6 hours with the response to a 6-hour AZD-8055 treatment of cell but among the induced genes after 24 hours, similar to cultures. This further validated the biological relevance of what was observed for the GO terms corresponding to our transcriptomic data. Interestingly, we also observed a carbohydrate metabolism. It is important to note here that strong Pearson correlation between the response to AZD- the various terms linked to the cell cycle were not enriched among the genes deregulated after sugar limitation. bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 8

When focusing on the genes deregulated 24 hours after although it was thought to be a pseudogene. Its very low treatment (which is the first timepoint for which we have expression level might explain why it escaped detection in evidence of growth inhibition in these three treatments), we previous analysis. Interestingly, the genes coding the core found only a few genes commonly deregulated (402 genes TOR complex (e.g. the genes with the highest expression commonly induced and 676 genes commonly repressed) level, namely TOR, LST8-1 and RAPTOR.3G) were found (supplemental figure 8, supplemental table 6). The up- to be consistently induced after TOR inactivation by ap- regulated genes did not show any strongly significant GO plication of AZD-8055, although with a very modest fold term enrichment except for some terms linked to lipid change (supplemental figure 10B). Similar induction was metabolism and amino acid or organic acid metabolism also found for these three genes 6 hours after phosphorus (but the adjusted p-value was still quite high) (supplemen- limitation. This suggests that the repression of TOR activity tal figure 8E, supplemental table 9E). In contrast, genes in response to phosphorus limitation does not involve commonly down-regulated showed marked GO enrich- transcriptional repression of the expression of the genes ment for terms linked to transcription and translation (sup- coding for the TOR complex and that a negative feedback plemental figure 8E, supplemental table 9L) which could mechanism might exist to enhance the expression of these then be potentially linked to the absence of growth re- genes. initiation. Interestingly, genes related to cell cycle activi- In order to better understand the link between the TOR ties were found to be commonly repressed after 24 hours pathway and the response to phosphorus starvation, we of AZD-8055 treatment or 24 hours of phosphate starva- specifically investigated genes responsive to the phospho- tion (supplemental figure 8E, supplemental table 9I). It rus level (Poirier and Bucher, 2002; Thibaud et al., 2010). is also worth noting that genes corresponding to the GO We observed that, as expected, the phosphate transporters terms linked to phosphorus metabolism are found not were mostly transcriptionally repressed after nutrient re- only among the genes commonly induced after phosphorus supply, except when it was accompanied by phosphorus starvation or AZD-8055 treatment but also among the genes starvation, conditions under which these genes are not specifically induced after AZD-8055 treatment (supplemen- transcriptionally deregulated, compared to T0. However, tal figure 8E, supplemental table 9C), suggesting that the it is strikingly noticeable that regulation of their expression TOR pathway plays indeed a role in the modulation of the is seemingly unaffected by chemical inactivation of TOR response to phosphate starvation and that modulating it (figure 4D). Interestingly, we found similar results for genes might be sufficient to mimic phosphate starvation. that have been previously shown to be transcriptionally In cell cultures, cell division is a key determinant of deregulated in response to phosphorus limitation (supple- growth. We therefore focused on genes directly related to mental figure 11), with the pattern being particularly obvi- the regulation of the cell cycle (Menges et al., 2005) and ous for genes that have been described as being induced observed that many of them were deregulated after some by phosphorus limitation, such as genes involved in Pi of the treatments (figure 4B). Among these genes, two recycling, signaling and sensing. The lack of response of groups particularly caught our attention. The first group these genes to AZD-8055 treatment highlights the existence is composed of genes that were induced in response to of a phosphate-dependent but TOR-independent pathway. nutrient re-supply but repressed when it was accompanied This further prompts us to believe that the TOR signaling by phosphorus limitation or TOR inactivation. It consists pathway is recruited by the phosphate signaling pathway mainly of A-type and D-type cyclins. The second group is and is therefore downstream of it. similar but the deregulation occurs only after 24 hours of treatment and this group contains mostly B-type cyclins. DISCUSSION Surprisingly, the expression of these genes was unaffected Growth in multicellular organisms is the combined result by sugar limitation and starvation and their expression of cell expansion, including synthesis of cellular content, levels was comparable to what was observed in absence and cell division. It has been shown that in plants, TOR of any nutrient limitation. We also focused on the genes (Target Of Rapamycin), a conserved master regulator of coding for the Histone3 protein family and observed a growth in eukaryotic organisms, is expressed mostly in similar trend (figure 4C). They are induced as early as after meristematic tissues (Menand et al., 2002). It is therefore 6 hours of nutrient resupply but not affected or repressed believed that TOR controls growth through the modulation in response to phosphorus deficiency or TOR inactivation. of cell division in plants (for review see Ahmad et al., 2019). Desvoyes et al. (2020) have recently shown that three genes This is supported by several studies showing that TOR are particularly important in idetifying the stage of the cell directly phosphorylates the E2FA and E2FB transcription cycle in a given cell. Based on the expression profile of these factors involved in the activation of the cell cycle (Li et al., three genes, it appears that the cell cycle is re-initiated as 2017; Xiong et al., 2013). This link between TOR and the early as 6 hours after nutrient resupply (even in the absence cell cycle was later strengthened by the characterization of of sugar in the medium) (supplemental figure 9). However, the kinase YAK1 (Yet Another Kinase1) as a direct target phosphate starvation and/or chemical inactivation of TOR of TOR and principal intermediate in the regulation of contributes to keeping their expression at a low level. cell proliferation (Barrada et al., 2019; Forzani et al., 2019; Since it has been shown that nutritional status influ- Van Leene et al., 2019). ences the expression of the TOR gene in maize (Canellas To investigate the direct effect of TOR inactivation, we et al., 2018), we focused our attention on the expression therefore focused this study on a heterotrophic cell culture of the genes coding for the components of the TOR com- system where growth is mostly the result of cell division, plex (supplemental figure 10A). As previously reported similar to what is observed in meristematic tissues. There (Moreau et al., 2010), the expression of RAPTOR.5G was would be a substantial risk that, if performing the experi- found to be lower than that of the gene RAPTOR.3G. ments in whole seedlings, the effect in cells not expressing However, contrary to published work (Moreau et al., 2012), TOR would overshadow the effect in meristematic tissue. we managed to detect some expression for the gene LST8-2 As the focus of the investigation was cell proliferation, bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 9 we chose to use a system of uniform and dividing cells scriptomic shift. This leads us to hypothesize that sugar in culture. Since TOR integrates the nutritional status to limitation is sensed primarily in photosynthetic tissues and modulate growth (for review Dobrenel et al., 2016a), we requires long-distance signaling to modulate meristematic compared the effect of the TOR-specific inhibitor AZD-8055 activities through the sequestration of phosphate from the to deprivation of sugar or phosphorus, both of which have meristems. This signal could possibly involve the transport already been shown to have an effect on the modulation of sugars from mesophyll cells to the meristems and would of the TOR pathway (Couso et al., 2019; Xiong et al., 2013). explain why suspension cultures which present the features Our data suggest that, in proliferating cells, the phosphorus of meristematic cells have lost the ability to establish a level is the key nutrient content integrated to recruit the signaling response to sugar starvation. TOR pathway to modulate growth. In summary, we have shown that TOR signaling is more We showed a strong overlap between the response to intertwined with the response to phosphate starvation than phosphate starvation and TOR inactivation at the transcrip- with the response to changed sugar levels in dividing tomic level as well as at the metabolic level, suggesting plant cells. By using a cell culture system, where all cells an interconnection between these two pathways in meris- are able to divide, we employ a system that resembles tematic cells. We also showed that genes involved in the meristematic cells in whole plants and opens up possi- primary response to phosphate starvation are unaffected by bilities of experiments that are otherwise impossible to TOR inactivation while, in contrast, phosphate starvation perform in isolated meristems. Such an approach allowed results in repression of RPS6 phosphorylation which is us to measure both the transcript and the metabolite level a marker for TOR activity (Dobrenel et al., 2016b). This changes over time in meristematic cells, which are where prompts us to place sensing of the phosphorus level up- TOR expression resides, giving us access to what is most stream from the regulation of the TOR pathway. likely to be the direct response to TOR inactivation by In this study, we showed that sensing of phosphate modulating the nutrient availability in the medium or by starvation reduces the TOR activity level which, in turn, direct pharmacological treatment. This approach may shed represses the transcription of genes necessary for re-entry further light on the metabolic regulation controlled by the into and/or progression of the cell cycle. With that in mind, master regulator TOR. we suggest that phosphate starvation (and consequently TOR inactivation) in meristematic cells actively blocks cell proliferation by regulating the expression of the cell cycle MATERIAL AND METHODS genes (figure 4B-C, supplemental figure 9). This would Cell culture maintenance and experiments then repress various anabolic reactions such as protein and cell wall synthesis and DNA replication and lead The cell line used is derived from root explants of Ara- to the accumulation of soluble metabolites (figure 2A), bidopsis thaliana Col-0 already described (Dubreuil et al., including the amino acids that are found to be accumulated 2018; Kunz et al., 2014). Cells were grown at 25°C in in response to both phosphate starvation and AZD-8055 darkness with constant orbital agitation in full-strength treatment (figure 5). Interestingly, we observed a strong MS salts with vitamins (Duchefa, Haarlem, The Nether- overlap between our transcriptomic data and previously lands) supplemented after autoclaving with filter-sterilized published transcriptomic datasets for TOR inactivation, sucrose to a final concentration of 3% , and subcultured particularly for 6 days of TOR inactivation in Arabidopsis every 7 days by dilution in fresh medium at 1:10 (v/v). seedlings (Caldana et al., 2013), TOR inactivation in sugar- Cells were collected by filtration under vacuum application starved seedlings (Xiong et al., 2013) and 24 hours of TOR and extensively rinsed with ice-cold sterile distilled water. inactivation (Dong et al., 2015) (supplemental figure 12A). Prior to cell filtration, an aliquot of medium was harvested This further confirms the hypothesis that the previously for analysis (see below). Cell density was calculated by published transcriptomic response to TOR inactivation is measuring the fresh weight of the filtered cells and cell based, at least for an important part of it, on a cell au- samples were immediately snap-frozen in liquid nitrogen. tonomous response, although Dong et al. (2015) already For the nutrient resupply and starvation experiments, 6- showed that there is a poor overlap between all these day old cells were diluted at 1:1 in 2X minimal Murashige datasets. and Skoog medium (containing only vitamins and mi- We observed that sugar starvation has the opposite cronutrients) and macronutrients were provided as indi- effect compared to TOR inactivation. The transcriptomic cated in the supplementary table 1. In short, the medium reprogramming was rather modest in comparison to the used contained, or did not, a final concentration (after response to the other conditions. Similarly, the metabo- dilution by the cells) of 3% sucrose and contained, or did lite content was not as strongly affected as it was after not, 1.25 mM of KH2PO4. AZD-8055 dissolved in DMSO AZD-8055 treatment or phosphate starvation. This weaker was added to some cultures immediately after the addition response may be caused by the presence of intracellular of macronutrients, to a final concentration of 2 µM, while an sucrose (figure 2B) up to 12 hours after the beginning of equivalent volume of DMSO was used as a mock treatment. the extracellular sucrose starvation. However, it should be noted that growth limitation was observed as early as Determination of medium composition 24 hours after the treatment (figure 1D). This result is in line with another study showing that sugar starvation in cell An aliquot of medium harvested before cell filtration was culture results in few genes being transcriptionally dereg- cleared by centrifugation and immediately snap frozen in ulated although it is sufficient to arrest growth (Nicola¨ı liquid nitrogen. The total nitrate content was determined et al., 2006) (supplemental figure 12B). This is surprising by spectrophotometry according to Hood-Nowotny et al. as Xiong et al. (2013) observed that, in sugar-deprived (2010). The soluble sugar content (glucose, fructose, su- seedlings, exogenous application of sugars recruits the crose) was determined by enzymatic assay according to TOR pathway to re-initiate growth through a drastic tran- Stitt et al. (1989). bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 10

Plant material and growth conditions or with 50% of the values below 1,000 arbitrary counts Arabidopsis thaliana wild type Col-0 (reference N60000) were removed from the dataset. The remaining peaks were seeds were used. Plants were grown on half-strength then assigned to annotated lipid species using an in-house- Murashige and Skoog medium with vitamins and MES generated lipid database for Arabidopsis (Hummel et al., buffer at pH 5.9 and 10 g/L plant agar (Duchefa, Haarlem, 2011). The data were normalized to the internal standard NL) for 8 days under long day (16h light/8h dark) fluores- and with respect to sample fresh weight. For some lipid cent light (100 µE.m-2.s-1) conditions. Seeds were chlorine species, more than one peak was detected with the same gas sterilized for four hours and subsequently pipetted m/z and identical adducts but different retention times. onto solid growth medium using 0.1% agar solution before In these cases, we added the letter A, B, C, or D to the stratification in the dark at 4°C for 2 days. Plants were compound name, depending on their elution order. grown for 7 days before 24 hour treatment with 1 µM AZD-8055 or mock treatment. The 8 day old seedlings were Transcriptomics analysis harvested and snap frozen in liquid nitrogen. Treatment of cell samples Total RNA was isolated using a QIAGEN RNeasy Plant Metabolomic profiling and lipid quantification Mini Kit according to the manufacturer’s instructions. Metabolomic profiling RNA was quantified with a Nanodrop ND-100 Samples from three independent biological replicates, spectrophotometer, and RNA quality was assessed with two technical replicates each, were subjected to using an Agilent 2100 bioanalyzer (Agilent Technologies). metabolomic profiling according to Kusano et al. (2011). Libraries were prepared with an Illumina TruSeq Stranded In short, metabolites were extracted from 25 mg of mRNA kit with poly-A selection. For the sequencing, frozen cell powder in a solvent composed of chloro- clustering was done by ’cBot’ and samples were sequenced form:methanol:water (3:1:1) containing stable isotopes of on NovaSeq6000 (NovaSeq Control Software 1.6.0/RTA reference metabolites (see Kusano et al. (2011) for the v3.4.4) with a 2x151 setup using ’NovaSeqXp’ workflow list). Extracted metabolites were then derivatized with in an ’S1’ mode flowcell. Bcl to FastQ conversion methoxyamine hydrochloride in pyridine and MSTFA and was performed using bcl2fastq v2.19.1.403 from the finally diluted in heptane. Peaks were acquired with a GC- CASAVA software suite. The quality scale used was TOF-MS according to Law et al. (2018) and retention indices Sanger / phred33 / Illumina 1.8+. Data pre-processing were calculated based on separation of an n-alkane (C8- was performed following the guidelines described in 1 C40) series. Metabolites were automatically detected from Epigenesys . Briefly, the quality of the raw sequence the chromatograms based on annotation of an in-house data was assessed using FastQC2, v0.11.4. Residual UPSC library and a posteriori manually curated against ribosomal RNA (rRNA) contamination was assessed the UPSC library and the Golm library (Schauer et al., and filtered using SortMeRNA (v2.1; Kopylova et al., 2005). The matrix of peak areas obtained for the confirmed 2012) with the settings −−log −−paired in −−fastx−−sam metabolites was normalized for instrument sensitivity (by −−num alignments 1 and using the rRNA sequences calibrating with the internal standards) and with respect to provided with SortMeRNA (rfam-5s-database-id98.fasta, the fresh weight. Statistical analysis were performed with rfam-5.8s-database-id98.fasta, silva-arc-16s-database- the SIMCA 13.0.3 and RStudio v. 1.2.5019 (using R v. 3.6.2) id95.fasta, silva-bac-16s-database-id85.fasta, silva-euk- software packages. 18s-database-id95.fasta, silva-arc-23s-database-id98.fasta, silva-bac-23s-database-id98.fasta and silva-euk-28s- Lipid extraction and UPLC/MS analysis database-id98.fasta). Data were then filtered to remove Lipid extraction and analysis were performed for three adapters and trimmed for quality using Trimmomatic independent biological replicates as previously described (v0.39; Bolger et al., 2014, settings TruSeq3-PE-2.fa:2:30:10 (Salem et al., 2016). Briefly, lipids were extracted from SLIDINGWINDOW:5:20 MINLEN:50). After the two 20 mg of homogenized tissue by suspending the ma- filtering steps, FastQC was run again to ensure that no terial in 1 ml of pre-cooled (-20°C) MTBE extrac- technical artefacts had been introduced. Read counts were tion solution (methanol:methyl tert-butyl-ether [1:3; v/v]) obtained using salmon (v0.14.1, Patro et al., 2017) with spiked with 0.5 µg.ml-1 of 1,2-diheptadecanoyl-sn-glycero- non-default parameters –gcBias –seqBias and using the 3-phosphocholine. The samples were incubated for 30 min ARAPORT11 cDNA sequences as reference (retrieved on an orbital mixer at 4°C followed by sonication for from the TAIR resource; Berardini et al., 2015; Cheng 10 min in an ice-cooled sonication bath. After addition et al., 2017). The salmon abundance values were imported of 500 µl of methanol:water (1:3, v/v) to induce phase into R (R Core Team, 2019) using the Bioconductor separation, the samples were vortexed and centrifuged for (v3.10; Gentleman et al., 2004) tximport package (v.1.12.3; 5 min at 20,000×g at 4°C. An aliquot of 500 µl was collected Soneson et al., 2015). For data quality assessment (QA) from the upper phase containing the lipids and dried in a and visualization, the read counts were normalized using vacuum concentrator. The pellet was resuspended in 250 a variance stabilizing transformation as implemented in µl acetonitrile:2-propanol (7:3, vol/vol) of which 2 µl was DESeq2. The biological relevance of the data - e.g. similarity subjected to UPLC/MS analysis (Salem et al., 2016). of biological replicates - was assessed by Principal Component Analysis (PCA) and other visualization Lipid annotation and statistical analysis methods (e.g. heatmaps), using custom R scripts, available at https://github.com/nicolasDelhomme/arabidopsis- The UPLC/MS data were processed using ToxID (Version nutrition-tor. Statistical analysis of gene and transcript 2.1.2; Thermo) with mass error 15 ppm, retention time (RT) window 0.05 minutes. To remove noise and contaminants, 1. http://www.epigenesys.eu/en/protocols/bio-informatics/1283- data for every lipid species with an average peak height guidelines-for-rna-seq-data-analysis lower than the average peak height of the method blanks 2. http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 11 differential expression (DE) between conditions was primary antiphospho-RPS6 antibody generated for this ar- performed in R using the Bioconductor DESeq2 package ticle as well as a primary antibody directed against mam- (v1.26.0; Love et al., 2014), with the model ∼ Conditions malian RPS6 (Cell Signaling Technology #2317S). For the using the T0 data as reference to account for the timepoint, western blots, protein contents were normalized using the the nutrition supplied and TOR inactivation or with the primary anti-tubulin antibody (Agrisera AS10 680). model ∼ Conditions using the non-treated timepoint data as reference to account for the nutrition supplied and TOR ADDITIONAL INFORMATION inactivation at a given timepoint. FDR adjusted p-values The transcriptomic data are available from the European were used to assess significance; a common threshold Nucleotide Archive (ENA4) under the accession number of 1% was used throughout together with a cutoff for PRJEB42585. Workflows and scripts to preprocess and ana- logarithmic expression fold change set at 0.5 to identify lyze the data have been made available in a git repository deregulated genes, as suggested by Schurch et al. (2016). with the DOI 10.5281/zenodo.46080285. All expression results were generated in R, using custom scripts. ACKNOWLEDGEMENTS Gene ontology enrichment was performed with the The authors would like to thank the Swedish Metabolomics BINGO plugin of Cytoscape (v. 3.7.2) and further analyzed Center (SMC), and especially Jonas Gullberg, Inga-Britt and displayed in Gephi (v. 0.9.2) or in R with the ggplot2 Carlsson, Krister Lundgren and Hans Stenlund, for their package. excellent technical assistance. The authors are also thankful to Anne¨ Michaelis (MPIMP) for the excellent technical Treatment of the seedling samples assistance for lipid profiling analysis. We are also grateful A TRIzol® Plus RNA Purification Kit from Thermo Fisher to the Umea˚ Plant Science Centre Bioinformatics facility Scientific was used to isolate RNA from frozen seedling (UPSCb) for their help regarding the transcriptomic anal- samples according to the protocol provided. The RNA sam- ysis. The authors acknowledge support from the National ples were DNase I treated to remove remaining genomic Genomics Infrastructure in Stockholm funded by Science DNA and cleaned up with the PureLink™ spin column- for Life Laboratory, the Knut and Alice Wallenberg Foun- based RNA isolation technology (Thermo Fischer). dation and the Swedish Research Council, and SNIC/Upp- RNA samples were sequenced on an Illumina HiSeq sala Multidisciplinary Center for Advanced Computational 2500 instrument using a standard protocol by Macrogen Science for assistance with massively parallel sequencing (Korea). RNA libraries were made using Illumina TruSeq and access to the UPPMAX computational infrastructure. technology and standard protocols. Raw sequencing reads We gratefully acknowledge Daria Chrobok (DC SciArt) for were aligned to the Arabidopsis genome (TAIR10) using her help in preparing the figures. This work was supported TopHat v2.0.13 (Trapnell et al., 2009) with the parame- by grants from the Knut and Alice Wallenberg Foundation ter settings: ‘bowtie1’, ‘no-novel-juncs’, ‘p 6’, ‘G’, ‘min- and the Swedish Governmental Agency for Innovation intron-length 40’, ‘max-intron-length 2000’. On average Systems. We thank Bio4Energy, a Strategic Research Envi- 97.0% (94.0 – 98.1%) of the raw reads could be aligned ronment appointed by the Swedish government, as well as to the genome per biological replicate. This represents the Kempe foundation, for supporting this work. an average of 51.1 (40.7 – 73.3) million mapped reads. Aligned reads were summarized over annotated gene mod- LIST OF SUPPLEMENTAL MATERIAL els using HTSeq-count v0.6.1 (Anders, 2010)3 with settings: The supplemental tables are available upon request at ‘−stranded no’, ‘−i gene id’. Sample counts were depth- [email protected] or [email protected] adjusted by means of the statistical computing environ- ment R (R Core Team, 2013) using the median-count-ratio Supplemental Figure 1. Evolution of glucose and fructose content in method available in the DESeq R package version 3.2.5 the cell culture medium during the growth period. and the DESeq2 package version 1.4.5 (Anders and Huber, 2010). For each comparison of three test replicate samples and Supplemental Figure 2. Analysis of metabolite profiles identifies three major phases occurring during the growth of a culture. (A) Principal three control replicate samples, a DeSeq dataset of raw Component Analysis showing evolution of the metabolomic profile of counts for all 33540 genes in the 6 samples of interest the cells highlighting gradual changes over the timecourse (black ar- was generated. Log2 fold changes and Benjamini Hochberg row) as well as the switch between the optimal phase of exponential FDR adjusted p-values were generated using the DeSeq growth and the beginning of starvation (red arrow). (B) Cluster analysis of metabolite content during the timecourse of growth. The relative function from the DESeq2 package, with cooksCutoff set metabolite content was measured in cells during progression of the to false. culture starting from the subculture point (day 0), averaged per day Lists of genes with significantly changed expression and normalized with respect to the maximal abundance. Metabolites are separated into categories represented by the color coding on the were generated using an absolute log2 fold change set at right. (n = 4) above 0.5 and a Benjamini Hochberg FDR adjusted p-value of below 0.01 as cutoffs. Supplemental Figure 3. Evolution of the RPS6 phosphorylation by mea- surement of the phosphorylated RPS6 level (A and C) and of the total Western blots RPS6 level (B and D) by western blot under plethoric conditions (A-B) or after a phosphate starvation (C-D). Each gel picture corresponds to For quantification of the phosphorylation status of the a different biological replicate. Numbers above the bands correspond RPS6, total protein extracts were subjected to western blot to hours after treatment. Numbers on the right hand correspond to size analysis as described in Dobrenel et al. (2016b) using the markers (kDa). M is a mixed sample used for quantification standard- ization.

3. Anders S. 2010. HTSeq: Analysing high-throughput sequencing 4. https://www.ebi.ac.uk/ena/browser/home data with python. URL https://htseq.readthedocs.io/en/master/ 5. https://dx.doi.org/10.5281/zenodo.4608028 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 12

Supplemental Figure 4. Metabolomic reprogramming after TOR inac- Supplemental Figure 11. Expression profiles of phosphate starvation tivation in 4-day-old exponentially growing cells. (A) Principal Compo- regulated genes. Expression profiles of the genes listed in Thibaud et nent Analysis plot for the metabolite profile at different time points of al., (2010), normalized to their expression in the T0 samples. Annota- 2 µM AZD-8055 treatment (in red) or mock treatment (in blue) (numbers tions of genes on the right-hand side are from the same article (an- correspond to the time after treatment in hours) (B) Evolution of the notations in red and blue correspond to, repectively, genes described concentrations of three selected amino-acids (y-axis in arbitrary units) as upregulated after phosphate starvation and genes described as (C) Hierarchical clustering analysis of metabolite changes during the downregulated after phosphate starvation). kinetics of TOR inactivation. Relative metabolite contents are normal- ized to T0. The color coding on the right represents the metabolic categories (red for TCA cycle and organic acids, green for amino acids and derivatives, orange for sugars and derivatives, black for lipids and purple for others). Supplemental Figure 12. Comparison of our datasets with previously published datasets. (A) Comparison of genes found to be deregulated in response to AZD-8055 treatment in our experiment to the previously published data. Colored circles correspond to our dataset (red for Supplemental Figure 5. Hierarchical clustering analysis showing the upregulated genes and blue for downregulated genes). Stars represent relative abundance of each metabolite over the timecourse under dif- datasets for which more than one third of the genes that they found to ferent conditions after normalization to T0; similar to figure 2 but with be deregulated are shared with our datasets. (B) Comparison of genes the samples separated by treatment levels. deregulated in response to 6 or 24 hours of sugar starvation in our experiment to the previously published work of Nicola¨ı et al. (2006)

Supplemental Figure 6. Summary of the lipidomic profile of the cell culture samples during TOR inactivation and nutrient limitation. (A) For each lipid, the values are normalized with respect to the maximal value. (B) For each lipid, the values are normalized to T0 (for the color coding Supplemental Material 1. Gene Ontology term enrichment of biological of the conditions, see figure 3A). (C) For each timepoint, the values are processes affected after 6 hours (A-B) or 24 hours (C-D) of nutrient normalized to the corresponding untreated sample (values depicted are resupply for the genes that are induced (A,C) or repressed (B,D) com- log2 transformed fold changes). pared to the T0 condition. Only terms with an adjusted p-value > 0.05 were kept for the analysis.

Supplemental Figure 7. Growth re-initiation after nutrient re-supply is accompanied by extensive transcriptomic reprogramming. (A) Number of genes transcriptionally deregulated compared to the beginning of the Supplemental Table 1. Medium composition for nutrient resupply (con- experiment (B-C) Comparison of genes upregulated (B) or downregu- centration given for the medium after dilution with the cells) lated (C) after 6 or 24 hours of nutrient re-supply. (D-G) Gene Ontology term enrichment of the biological processes affected after 6 hours (D- E) or 24 hours (F-G) of nutrient re-supply. (D and F) Induced genes, (E and G) repressed genes. Colors of the GO bubbles correspond to the log transformed adjusted p-values and their sizes correspond to the Supplemental Table 2. Metabolite changes during growth kinetics number of occurrences in the dataset. Only terms with an adjusted p- value > 0,05 were kept for the analysis. To improve visualization, GO terms with more than 2000 occurrences in the genome were removed.

Supplemental Table 3. Log2(fold change) in the evolution of metabolite content in 4-day old cells treated with 2 µM AZD-8055 Supplemental Figure 8. Nutritional limitation or TOR inactivation re- sult in considerable transcriptomic reprogramming. (A) Number of genes transcriptionally deregulated compared to the corresponding non-treated timepoint. Bars above zero represent upregulated genes and bars below zero represent downregulated genes. Numbers asso- Supplemental Table 4. Quantification of metabolites over time in re- ciated with each bar correspond to the actual numbers of deregulated sponse to medium re-supply (NPS), and phosphate (NS) and sucrose genes. Stars correspond to the gene sets used for panels B and C. (B- (NP) starvation, in the presence or absence of the TOR inhibitor AZD- C) Venn diagrams showing the overlap of deregulated genes between 8055 (AZD and DMSO respectively) different conditions after 6 hours of treatment (B) and 24 hours of treat- ment (C) when compared to the corresponding non-treated timepoints. Turquoise = phosphorus limitation, pink = sugar limitation, black = AZD- 8055 treatment. (D) Bubble plot of the Gene Ontology terms enriched in the deregulated genes represented in panel A. Gray bubbles represent Supplemental Table 5. Quantification of lipids over time in response GOs with an adjusted p-value > 0,05, other colors are on a scale to medium re-supply (NPS), and phosphate (NS) and sucrose (NP) (shown below) indicating the log transformed adjusted p-value. The starvation in the presence or absence of the TOR inhibitor AZD-8055 sizes of the bubbles correspond to the number of occurrences in the (AZD and DMSO respectively) dataset normalized to the number of occurrences in the genome. (E) As in D for the different overlaps among the Venn diagrams presented in C (+ when the condition is included, - when the condition is excluded, 0 when the condition is ignored). Supplemental Table 6. List of deregulated gene expression for each of the pairwise comparisons. Log fold changes as well as adjusted p- Supplemental Figure 9. Expression profile of cell cycle marker genes values are displayed. (Desvoyes et al., 2020): CDT1a was shown to be specifically expressed in cells in G1, CycB1;1 was shown to be specifically expressed in cells in late G2 phase and early in mitosis while the histone HTR13 is expressed predominantly during S phase and early G2 phase. Supplemental Table 7. List of Gene Ontology terms enriched among the genes deregulated after nutrient re-supply compared to the T0 condition (only GO terms with an adjusted p-value ¡ 0.05 were kept). (A) Supplemental Figure 10. Expression profiles of TOR complex mem- Genes transcriptionally induced after 6 hours of nutrient resupply. (B) bers. (A) Absolute expression of genes coding for components of the Genes transcriptionally repressed after 6 hours of nutrient resupply. (C) TOR complex. (B) Expression of genes coding for components of the Genes transcriptionally induced after 24 hours of nutrient resupply. (D) TOR complex, relative to their expression values in the T0 samples. Genes transcriptionally repressed after 24 hours of nutrient resupply. Circles represent individual values, red bars represent averages. The (xx corresponds to the number of GO occurrences in the dataset, horizontal solid lines represent the T0 averages and the horizontal X to the number of genes in the dataset, nn to the number of GO dashed lines represent averages of the corresponding non-treated occurrences in the annotated genome, N to the size of the annotated samples. genome) bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 13

Supplemental Table 8. List of Gene Ontology terms enriched among plants owing to the differential expression of TOR. Jour- the genes deregulated between the different conditions compared by nal of Plant Growth Regulation, 38:216–224. timepoint. (A) Genes upregulated after 6 hours of phosphate starva- Cheng, C. Y., V. Krishnakumar, A. P. Chan, F. Thibaud- tion. (B) Genes downregulated after 6 hours of phosphate starvation. (C) Genes upregulated after 6 hours of sugar starvation. (D) Genes Nissen, S. Schobel, and C. D. Town downregulated after 6 hours of sugar starvation. (E) Genes upregu- 2017. Araport11: a complete reannotation of the Arabidop- lated after 6 hours of AZD-8055 treatment. (F) Genes downregulated sis thaliana reference genome. Plant Journal, 89(4):789–804. after 6 hours of AZD-8055 treatment. (G) Genes upregulated after Couso, I., M. Perez-Perez, M. Ford, E. Martinez Force, 6 hours of phosphate starvation and AZD-8055 treatment. (H) Genes downregulated after 6 hours of phosphate starvation and AZD-8055 L. Hicks, J. Umen, and J. Crespo treatment. (I) Genes upregulated after 6 hours of sugar starvation 2019. Phosphorus availability regulates TORC1 signaling and AZD-8055 treatment. (J) Genes downregulated after 6 hours of via LST8 in Chlamydomonas. The Plant Cell, 32(1):69–80. sugar starvation and AZD-8055 treatment. (K) Genes upregulated after 24 hours of phosphate starvation. (L) Genes downregulated af- Deprost, D., H. N. Truong, C. Robaglia, and C. Meyer ter 24 hours of phosphate starvation. (M) Genes upregulated after 2005. An Arabidopsis homolog of RAPTOR/KOG1 is 24 hours of sugar starvation. (N) Genes downregulated after 24 hours essential for early embryo development. Biochemical and of sugar starvation. (O) Genes upregulated after 24 hours of AZD-8055 Biophysical Research Commununications, 326(4):844–50. treatment. (P) Genes downregulated after 24 hours of AZD-8055 treat- ment. (Q) Genes upregulated after 24 hours of phosphate starvation Deprost, D., L. Yao, R. Sormani, M. Moreau, G. Leterreux, and AZD-8055 treatment. (R) Genes downregulated after 24 hours of M. Nicolai, M. Bedu, C. Robaglia, and C. Meyer phosphate starvation and AZD-8055 treatment. (S) Genes upregulated 2007. The Arabidopsis TOR kinase links plant growth, after 24 hours of sugar starvation and AZD-8055 treatment. (T) Genes EMBO downregulated after 24 hours of sugar starvation and AZD-8055 treat- yield, stress resistance and mRNA translation. ment. (xx corresponds to the number of GO occurrences in the dataset, Reports, 8(9):864–70. X to the number of genes in the dataset, nn to the number of GO Desvoyes, B., A. Arana-Echarri, M. D. Barea, and C. Gutier- occurrences in the annotated genome, N to the size of the annotated rez genome) 2020. A comprehensive fluorescent sensor for spatiotem- poral cell cycle analysis in Arabidopsis. Nature Plants, 6(11):1330–1334. Supplemental Table 9. List of Gene Ontology terms enriched among Dobrenel, T., C. Caldana, J. Hanson, C. Robaglia, M. Vin- the treatment-specific genes deregulated after 24 hours of treatment when compared to the 24 hour control conditions. (A-G) Upregulated centz, B. Veit, and C. Meyer genes, (H-N) Downregulated genes. (A,H) specific to phosphate star- 2016a. TOR signaling and nutrient sensing. Annual vation, (B,I) common to phosphate starvation and AZD-8055 treatment Review of Plant Biology, 67:261–285. but absent from sugar starvation treatment, (C,J) AZD-8055 treatment Dobrenel, T., E. Mancera-Martinez, C. Forzani, M. Az- specific, (D,K) common to phosphate and sugar starvation but absent from the AZD-8055 treatment, (E,L) common to the 3 treatments, (F,M) zopardi, M. Davanture, M. Moreau, M. Schepetilnikov, common to the AZD-8055 treatment and sugar starvation but absent J. Chicher, O. Langella, M. Zivy, C. Robaglia, L. A. from phosphate starvation, (G,N) specific to sugar starvation. (xx cor- Ryabova, J. Hanson, and C. Meyer responds to the number of GO occurrences in the dataset, X to the number of genes in the dataset, nn to the number of GO occurrences 2016b. The Arabidopsis TOR kinase specifically regulates in the annotated genome, N to the size of the annotated genome) the expression of nuclear genes coding for plastidic ribo- somal proteins and the phosphorylation of the cytosolic ribosomal protein S6. Frontiers in Plant Science, 7:1611. REFERENCES Dobrenel, T., C. Marchive, M. Azzopardi, G. Clement,´ M. Moreau, R. Sormani, C. Robaglia, and C. Meyer Ahmad, Z., Z. Magyar, L. Bogre,¨ and C. Papdi 2013. Sugar metabolism and the plant Target Of Ra- 2019. Cell cycle control by the TARGET OF RAPAMYCIN pamycin kinase: a sweet operaTOR? Frontiers in Plant signalling pathway in plants. Journal of Experimental Science, 4:93. Botany, 70(8):2275–2284. Dong, P., F. Xiong, Y. Que, K. Wang, L. Yu, Z. Li, and Anders, S. and W. Huber R. Maozhi 2010. Differential expression analysis for sequence count 2015. Expression profiling and functional analysis reveals data. Genome Biology, 11(10):R106. that TOR is a key player in regulating photosynthesis Barrada, A., M. Djendli, T. Desnos, R. Mercier, C. Robaglia, and phytohormone signaling pathways in Arabidopsis. M.-H. Montane,´ and B. Menand Frontiers in Plant Science, 6:677. 2019. A TOR-YAK1 signaling axis controls cell cycle, Dong, Y., M. Silbermann, A. Speiser, I. Forieri, E. Linster, meristem activity and plant growth in Arabidopsis. De- G. Poschet, A. Allboje Samami, M. Wanatabe, C. Sticht, velopment, 146(3):1–14. A. A. Teleman, J. M. Deragon, K. Saito, R. Hell, and Berardini, T. Z., L. Reiser, D. Li, Y. Mezheritsky, R. Muller, M. Wirtz E. Strait, and E. Huala 2017. Sulfur availability regulates plant growth via 2015. The Arabidopsis information resource: Making and glucose-TOR signaling. Nature Communications, 8(1):1174. mining the ”gold standard” annotated reference plant Dubreuil, C., X. Jin, J. Barajas-Lopez,´ T. Hewitt, S. Tanz, genome. Genesis, 53(8):474–85. T. Dobrenel, W. Schroder,¨ J. Hanson, E. Pesquet, Bolger, A. M., M. Lohse, and B. Usadel A. Gronlund,¨ I. Small, and A.˚ Strand 2014. Trimmomatic: a flexible trimmer for Illumina se- 2018. Establishment of photosynthesis through chloro- quence data. Bioinformatics, 30(15):2114–2120. plast development is controlled by two distinct regula- Caldana, C., Y. Li, A. Leisse, Y. Zhang, L. Bartholomaeus, tory phases. Plant Physiology, 176(2):1199–1214. A. R. Fernie, L. Willmitzer, and P. Giavalisco Forzani, C., G. Duarte, J. Van Leene, G. Clement,´ S. Huguet, 2013. Systemic analysis of inducible TARGET OF RA- C. Paysant-Le-Roux, R. Mercier, G. De Jaeger, A.-S. Lep- PAMYCIN mutants reveal a general metabolic switch rince, and C. Meyer controlling growth in Arabidopsis thaliana. Plant Journal, 2019. Mutations of the AtYAK1 kinase suppress TOR 73(6):897–909. deficiency in arabidopsis. Cell Reports, 27(12):3696–3708 Canellas, L., N. Canellas, T. Soares, and F. Olivares e5. 2018. Humic acids interfere with nutrient sensing in bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 14

Gentleman, R. C., V. J. Carey, D. M. Bates, B. Bolstad, phorus. Plant, Cell & Environment, 30(1):85–112. M. Dettling, S. Dudoit, B. Ellis, L. Gautier, Y. Ge, J. Gentry, Moreau, M., M. Azzopardi, G. Clement,´ T. Dobrenel, K. Hornik, T. Hothorn, W. Huber, S. Iacus, R. Irizarry, C. Marchive, C. Renne, M. Martin-Magniette, L. Tacon- F. Leisch, C. Li, M. Maechler, A. J. Rossini, G. Sawitzki, nat, J. Renou, C. Robaglia, and C. Meyer C. Smith, G. Smyth, L. Tierney, J. Y. H. Yang, and J. Zhang 2012. Mutations in the Arabidopsis homolog of 2004. Bioconductor: open software development for com- LST8/GbL, a partner of the TARGET OF RAPAMYCIN putational biology and bioinformatics. Genome Biology, kinase, impair plant growth, flowering, and metabolic 5(10):R80. adaptation to long days. The Plant Cell Online, 24(2):463– Heitman, J., N. Movva, and M. Hall 481. 1991. Targets for cell cycle arrest by the immunosuppres- Moreau, M., R. Sormani, B. Menand, B. Veit, C. Robaglia, sant rapamycin in yeast. Science, 253(5022):905–909. and C. Meyer Hood-Nowotny, R., N. H.-N. Umana, E. Inselbacher, 2010. Chapter 15 - The TOR Complex and Signaling P. Oswald-Lachouani, and W. Wanek Pathway in Plants. The Enzymes, Volume 27:285–302. 2010. Alternative methods for measuring inorganic, Nicola¨ı, M., M. Roncato, A. Canoy, D. Rouquie,´ X. Sarda, organic, and total dissolved nitrogen in soil. Soil Science G. Freyssinet, and C. Robaglia Society of America Journal, 74(3):1018–1027. 2006. Large-scale analysis of mRNA translation states Hummel, J., S. Segu, Y. Li, S. Irgang, J. Jueppner, and during sucrose starvation in Arabidopsis cells identifies P. Giavalisco cell proliferation and chromatin structure as targets of 2011. Ultra performance liquid chromatography and translational control. Plant Physiology, 141(2):663–673. high resolution mass spectrometry for the analysis of Patro, R., G. Duggal, M. I. Love, R. A. Irizarry, and C. Kings- plant lipids. Frontiers in Plant Science, 2:54. ford Kopylova, E., L. Noe, and H. Touzet 2017. Salmon provides fast and bias-aware quantification 2012. SortMeRNA: fast and accurate filtering of ribo- of transcript expression. Nature Methods, 14(4):417–419. somal RNAs in metatranscriptomic data. Bioinformatics, Pesquet, E., A. Korolev, G. Calder, and C. Lloyd 28(24):3211–3217. 2010. The microtubule-associated protein AtMAP70- Kunz, S., E. Pesquet, and L. Kleczkowski 5 regulates secondary wall patterning in Arabidopsis 2014. Functional dissection of sugar signals affect- wood cells. Current Biology, 20(8):744–9. ing gene expression in Arabidopsis thaliana. PLoS One, Pfeiffer, A., D. Janocha, Y. Dong, A. Medzihradszky, 9(6):e100312. S. Schone, G. Daum, T. Suzaki, J. Forner, T. Langenecker, Kusano, M., P. Jonsson, A. Fukushima, J. Gullberg, E. Rempel, M. Schmid, M. Wirtz, R. Hell, and J. U. M. Sjostrom, J. Trygg, and T. Moritz Lohmann 2011. Metabolite signature during short-day induced 2016. Integration of light and metabolic signals for growth cessation in Populus. Frontiers in Plant Science, stem cell activation at the shoot apical meristem. Elife, 2:29. 5:e17023. Law, S. R., D. Chrobok, M. Juvany, N. Delhomme, P. Linden, Poirier, Y. and M. Bucher B. Brouwer, A. Ahad, T. Moritz, S. Jansson, P. Garde- 2002. Phosphate transport and homeostasis in Arabidop- strom, and O. Keech sis. The Arabidopsis book, 1:e0024–e0024. 2018. Darkened leaves use different metabolic strategies R Core Team for senescence and survival. Plant Physiology, 177(1):132– 2019. R: A language and environment for statistical com- 150. puting. R Foundation for statistical computing, Vienna, Li, X., W. Cai, Y. Liu, H. Li, L. Fu, Z. Liu, L. Xu, H. Liu, Austria. T. Xu, and Y. Xiong Ren, M., P. Venglat, S. Qiu, L. Feng, Y. Cao, E. Wang, D. Xi- 2017. Differential TOR activation and cell proliferation ang, J. Wang, D. Alexander, S. Chalivendra, D. Logan, in Arabidopsis root and shoot apexes. Proceedings of the A. Mattoo, G. Selvaraj, and R. Datla National Academy of Sciences, 114(10):2765–2770. 2012. TARGET OF RAPAMYCIN signaling regulates Love, M. I., W. Huber, and S. Anders metabolism, growth, and life span in Arabidopsis. Plant 2014. Moderated estimation of fold change and disper- Cell, 24(12):4850–4874. sion for RNA-seq data with DESeq2. Genome Biology, Rouached, H., A. Stefanovic, D. Secco, A. Bulak Arpat, 15(12):550. E. Gout, R. Bligny, and Y. Poirier Menand, B., T. Desnos, L. Nussaume, F. Berger, D. Bouchez, 2011. Uncoupling phosphate deficiency from its major ef- C. Meyer, and C. Robaglia fects on growth and transcriptome via PHO1 expression 2002. Expression and disruption of the arabidopsis TOR in Arabidopsis. The Plant Journal, 65(4):557–570. (TARGET OF RAPAMYCIN) gene. Proceedings of the Salem, M. A., J. Juppner, K. Bajdzienko, and P. Giavalisco National Academy of Sciences, 99(9):6422–7. 2016. Protocol: a fast, comprehensive and reproducible Menges, M., S. M. de Jager, W. Gruissem, and J. A. Murray one-step extraction method for the rapid preparation of 2005. Global analysis of the core cell cycle regulators polar and semi-polar metabolites, lipids, proteins, starch of Arabidopsis identifies novel genes, reveals multiple and cell wall polymers from a single sample. Plant and highly specific profiles of expression and provides Methods, 12:45. a coherent model for plant cell cycle control. The Plant Schauer, N., D. Steinhauser, S. Strelkov, D. Schomburg, Journal, 41(4):546–66. G. Allison, T. Moritz, K. Lundgren, U. Roessner-Tunali, Morcuende, R., R. Bari, Y. Gibon, W. Zheng, B. D. Pant, M. G. Forbes, L. Willmitzer, A. R. Fernie, and J. Kopka O. Blasing,¨ B. Usadel, T. Czechowski, M. K. Udvardi, 2005. GC-MS libraries for the rapid identification of M. Stitt, and W.-R. Scheible metabolites in complex biological samples. FEBS Lett, 2007. Genome-wide reprogramming of metabolism and 579(6):1332–7. regulatory networks of Arabidopsis in response to phos- bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. 15

Schurch, N. J., P. Schofield, M. Gierlinski, C. Cole, A. Sher- stnev, V. Singh, N. Wrobel, K. Gharbi, G. G. Simpson, T. Owen-Hughes, M. Blaxter, and G. J. Barton 2016. How many biological replicates are needed in an RNA-seq experiment and which differential expression tool should you use? RNA, 22(6):839–51. Soneson, C., M. I. Love, and M. D. Robinson 2015. Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences. F1000Research, 4:1521. Stitt, M., R. M. Lilley, R. Gerhardt, and H. W. Heldt 1989. Metabolite levels in specific cells and subcellular compartments of plant leaves. Methods in Enzymology, 174:518–552. Thibaud, M. C., J. F. Arrighi, V. Bayle, S. Chiarenza, A. Creff, R. Bustos, J. Paz-Ares, Y. Poirier, and L. Nussaume 2010. Dissection of local and systemic transcriptional responses to phosphate starvation in Arabidopsis. The Plant Journal, 64(5):775–89. Trapnell, C., L. Pachter, and S. L. Salzberg 2009. TopHat: discovering splice junctions with RNA- Seq. Bioinformatics, 25(9):1105–1111. Van Leene, J., C. Han, A. Gadeyne, D. Eeckhout, C. Matthijs, B. Cannoot, N. De Winne, G. Persiau, E. Van De Slijke, B. Van de Cotte, E. Stes, M. Van Bel, V. Storme, F. Impens, K. Gevaert, K. Vandepoele, I. De Smet, and G. De Jaeger 2019. Capturing the phosphorylation and protein inter- action landscape of the plant TOR kinase. Nature Plants, 5(3):316–327. Xiong, Y., M. McCormack, L. Li, Q. Hall, C. Xiang, and J. Sheen 2013. Glucose-TOR signalling reprograms the transcrip- tome and activates meristems. Nature, 496(7444):181–186. bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

20 glucose fructose

15

10

5 Sugar content in the medium (mM) content in the medium Sugar

0 0 2 4 6 8 10

days of culture

Supplemental figure 1. Evolution of glucose and fructose content in the cell culture medium during the growth period. bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B 0 10 1 2 3 4 5 Growth initiation 6 7 5 8 9 Days of culture TCA cycle and organic acids Aand deriva � ves AA Sugars and deriva � ves Others 10 Lipids 11 0 1 2 3 4 5 6 7 8 9 10 11 valine isoleucine 0 proline Optimal growth cysteine alanine

Comp. 2 (18%) glycine threonine allothreonine -5 Starvation & Death beta-alanine Glc-6-P Fructose-6-phosphate o-phosphoethanolamine fumaric acid sucrose -10 eicosanoic acid scyllo-inositol nonanoic acid -5 0 5 tryptophan phenylalnine Comp. 1 (38%) 2-amino-adipic acid aspar�c acid ornithine arginine tyrosine glutamine pyroglutamic acid asparagine spermidine glucuronic acid homoserine malic acid ethanolamine isocitrate citric acid glyceric acid Dehydroascorbic acid dimer leucine uracil campesterol putrescine myris�c acid glutamic acid linoleic acid oleic acid linolenic acid succinic acid benzoate palmi�c acid octadecanoic acid alpha-KG GABA

0 0,5 1 Metabolite abundance normalized to maximal abundance

Supplemental figure 2. Analysis of metabolite profiles identifies three major phases occurring during the growth of a culture (A) Principal Component Analysis showing evolution of the metabolomic profile of the cells highlighting gradual changes over the timecourse (black arrow) as well as the switch between the optimal phase of exponential growth and the b eginning of starvation (red arrow). (B) Cluster analysis of metabolite content during the timecourse of growth. The relative metabolite content was measured in cells during progression of the culture starting from the subculture point (day 0), averaged per day and normalized with respect to the maximal abundance. Metabolites are separated into categories represented by the color coding on the right. (n = 4) bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A

180 180 180 130 130 130 100 100 100 70 70 70 DMSO AZD DMSO AZD 55 DMSO AZD 55 M 55 M M 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 40 0 3 6 12 24 0 3 6 12 24 40 40 35 35 35 25 25 25 Phospho-RPS6 15 15 15

B

180 180 180 130 130 130 100 100 100 70 70 70 DMSO AZD DMSO AZD 55 DMSO AZD 55 M 55 M M 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 40 40 40

35 35 35 25 25 Total RPS6 25

15 15 15

C 180 180 130 180 130 100 130 100 100 70 DMSO AZD 70 DMSO AZD DMSO AZD 70 M M 55 M 0 3 6 12 24 0 3 6 12 24 55 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 0 3 6 12 24 55 40 40 40 35 35 35 25 25 Phospho-RPS6 25

15 15 15

D 180 180 180 130 130 130 100 100 100 DMSO AZD 70 DMSO AZD 70 DMSO AZD 70 M M M 0 3 6 12 24 0 3 6 12 24 55 0 3 6 12 24 0 3 6 12 24 55 0 3 6 12 24 0 3 6 12 24 55 40 40 40 35 35 35

Total RPS6 25 25 25

15 15 15

Supplemental figure 3. Evolution of the RPS6 phosphorylation by measurement of the phosphorylated RPS6 level (A and C) and of the total RPS6 level (B and D) by western blot under plethoric conditions (A-B) or after a phosphate starvation (C-D). Each gel picture corresponds to a different biological replicate. Numbers above the bands correspond to hours after treatment. Numbers on the right hand correspond to size markers (kDa). M is a mixed sample used for quantification standardization. bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B (x104) 0 4 DMSO Ile AZD-8055

5 3 1212

12 4 8 2 5 1 12 12 0 2 3 1 8 4 3 5 12 0 4 8 12 Comp.2 (19%) 6 12 12 (x104) 0 − 5 6 Val 1 6

2 4 −5 0 5 10 Comp.1 (40%) 2

C DMSO AZD-8055 (2µM) 0 4 8 12 Time (in hours) 0 1 2 3 4 5 6 8 12 0 1 2 3 4 5 6 8 12 4 ethanolamine (x10 ) tryptophan 10 Phe oleic acid linoleic acid 8 linolenic acid benzoate 6 myristic acid nonanoic acid 4 palmitic acid octadecanoic acid eicosanoic acid 2 Dehydroascorbate dimer glucuronic acid 0 4 8 12 spermidine scyllo-inositol Time in hours glyceric acid homoserine malic acid campesterol Glc-6-P Fructose-6-phosphate fumaric acid o-phosphoethanolamine arginine uracil proline glycine GABA alpha-KG alanine putrescine aspartic acid cysteine isoleucine leucine tyrosine valine succinic acid sucrose beta-alanine citric acid isocitrate pyroglutamic acid glutamine glutamic acid 2-amino-adipic acid ornithine asparagine log2 (ratio to T0) phenylalnine -3 0 +3 allothreonine threonine

Supplemental figure 4. Metabolomic reprogramming after TOR inactivation in 4-day-old exponentially growing cells. (A) Principal Component Analysis plot for the metabolite profile at different time points of 2 μM AZD-8055 treatment (in red) or mock treatment (in blue) (numbers correspond to the time after treatment in hours) (B) Evolution of the concentrations of three selected amino-acids (y-axis in arbitrary units) (C) Hierarchical clustering analysis of metabolite changes during the kinetics of TOR inactivation. Relative metabolite contents are normalized to T0. The color coding on the right represents the metabolic categories (red for TCA cycle and organic acids, green for amino acids and derivatives, orange for sugars and derivatives, black for lipids and purple for others). bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Nutrition: No starvation Sugar starvation Phosphate starvation AZD-8055: DMSO 2µM AZD-8055 Timepoint: T0 3h 6h 12h 24h 48h 72h

Suc 0 + + + + + + + + + + + + + + + + + + + + + + + + ------Pi 0 + + + + + + + + + + + + ------+ + + + + + + + + + + + AZD-8055 0 ------+ + + + + + ------+ + + + + + ------+ + + + + + Time (h) 0 Nutrition AZD-8055 Timepoint TCA cycle org. ac. and AA and derivatives Others 0 3 6 12 48 3 6 48 3 6 48 3 6 48 3 6 48 3 6 48 derivativesSugars and

24 72 1224 72 1224 72 1224 72 1224 72 1224 72 Lipids A phenylalanine A asparagine A threonine A tyrosine A glutamine T isocitrate T citrate T glyceric acid T aconi�c acid A glutamic acid A arginine S scyllo-inositol T malic acid T fumaric acid A isoleucine A valine A 2-amino-adipic acid S sucrose A ornithine O allantoin A cysteine A serine A his�dine O dehydroascorbic acid dimer S chiro-inositol O hydroxybenzoic acid A pyroglutamic acid L linoleic acid L adipic acid L octadecanoic acid A spermidine O ethanolamine O squalene L stearic acid O campesterol L arachidonic acid L eicosanoic acid A GABA T alpha-kg S xylulose T succinate log2(FC)

-3 -2 -1 0 1 2 3

Supplemental figure 5. Hierarchical clustering analysis showing the relative abundance of each metabolite over the timecourse under different conditions after normalization to T0; similar to figure 2 but with the samples separated by treatment levels. bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. A Suc 0 ++++++++++++++++++++++++------B Suc 0 ++++++++++++++++++++++++------Pi 0 ++++++++++++ ------++++++++++++ Pi 0 ++++++++++++ ------++++++++++++ AZD-8055 0 - - - -+-- +++++ - - - - +-- +++++ - - - - +-- +++++ AZD-8055 0 - - - -+-- +++++ - - - - +-- +++++ - - - -+-- + + + + + Time (h) 0 Time (h) 0 Nutrition DAG AZD-8055 0 3 6 12 24 48 72 3 6 12 24 48 72 3 6 12 48 72243 6 12 24 48 72 3 6 12 24 48 72 3 6 2412 48 72 Timepoint

DAG

DGDG

DGDG

LysoPC

MGDG LysoPC

MGDG

PC

PC

PE

PE

PG

PG PI

PI

SQDG SQDG

TAG TAG

relative abundance log2(FC)

0 0,5 1 -3 -2 -1 0 1 2 3 Supplemental figure 6. Summary of the lipidomic profile of the cell culture samples during TOR inactivation and nutrient limitation. (A) For each lipid, the values are normalized with respect to the maximal value. (B) For each lipid, the values are normalized to T0 (for the color coding of the conditions, see figure 3A). (C) For each timepoint, the values are normalized to the corresponding untreated sample (values depicted are log2 transformed fold changes). bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is Supplemental figure 6 (cont.) made available under aCC-BY-NC-ND 4.0 International license.

C Suc + + + + + + + + + + + + + + + + + + + + + + + + ------Pi + + + + + + + + + + + + ------+ + + + + + + + + + + + AZD-8055 ------+ + + + + + ------+ + + + + + ------+ + + + + + Time (h)

log2(fold change)

-2-3 0-1 321 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B 6000 5302 5053 5150 T6 T24 4639 4775 4685 4614 4332 4066 4000 3553 3258 2165 2259 3043 2107

2000

0

-2000 C T6 T24

3036 2414 -4000 3486 4096 3844 26151747 2766 4362 4443 4438 4283 4292 5381 4817 -6000 6 hours 24 hours No limita�on Phosphorus limita�on Sugar limita�on DMSO AZD-8055

D Primary metabolism E Protein / lipid / ion DNA replica�on transport Lipid metabolism

Amino acid Carbohydrate DNA repair metabolism metabolism resp. to abio�c stress Protein matura�on

Protein Nucleoside biosynthesis transport Development Differen�a�on Response to AA and nucleo�de abio�c stresses Polysaccharide biosynthesis metabolism Cell division Mitosis Nucleo�de catabolism Transcrip�on GTPase Transla�on DNA PCD-related modifica�on processes Cell wall Ribosome Response to biogenesis assembly Development hormones

Response to F Mitosis Cell cycle G Nucleo�de catabolism hormones Microtubule organiza�on GTPase Primary Transport metabolism Response to Polysaccharide bio�c stress metabolism Response to abio�c stress DNA synthesis Transcrip�on Transport and modifica�on Defense Transla�on Post-transla�onal modifs. Protein catabolism Development PCD Senescence

Lipid Protein Nucleo�de metabolism phosphoryla�on biosynthesis

Small molecule Amino acid and metabolism nucleo�de (mostly catabolism) biosynthesis Developmental processes - log10 (adj.p.value)

0 20

Supplemental figure 7. Growth re-initiation after nutrient re-supply is accompanied by extensive transcriptomic reprogramming. (A) Number of genes transcriptionally deregulated compared to the beginning of the experiment (B-C) Comparison of the genes upregulated (B) or downregulated (C) after 6 or 24 hours of nutrient re-supply. (D-G) Gene Ontology term enrichment of the biological processes affected after 6 hours (D-E) or 24 hours (F-G) of nutrient re- supply. (D and F) Induced genes, (E and G) repressed genes. Colors of the GO bubbles correspond to the log transformed adjusted p-values and their sizes correspond to the number of occurrences in the dataset. Only terms with an adjusted p-value > 0,05 were kept for the analysis. To improve visualization, GO terms with more than 2000 occurrences in the genome were removed. bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A 5956 B Up Down 6000 5446 4857 4556 4057 4000 1490 2506 3173 1622 1220 989 2866 2939 1279

2000 25 36 1190 36 72 59 73 709 576 816 0

984 1096 -2000 C Up Down

-4000 3604 3495 3821 3901 2048 1866 4834 4995 2210 5359 1343 2313 -6000 5579 853 6 hours 24 hours 402 676 No limita�on Phosphorus limita�on Sugar limita�on 102 197 59 140 489 DMSO AZD-8055 221

Supplemental figure 8. Nutritional limitation D or TOR inactivation result in considerable trans- Induced Repressed criptomic reprogramming. GO:0006950 Response to stress (A) Number of genes transcriptionally GO:0009628 Response to abio�c s�mulus GO:0042221 Response to chemical s�mulus deregulated compared to the corresponding GO:0009725 Response to hormone s�mulus non-treated timepoint. Bars above zero GO:0006810 Transport represent upregulated genes and bars below zero represent downregulated genes. Numbers GO:0016042 Lipid catabolic process GO:0071555 Cell wall organiza�on associated with each bar correspond to the GO:0071554 Cell wall organiza�on or biogenesis GO:0042545 Cell wall modifica�on actual numbers of deregulated genes. Stars correspond to the gene sets used for panels B GO:0071103 DNA conforma�on change GO:0006260 DNA replica�on and C. GO:0006325 Chroma�n organiza�on (B-C) Venn diagrams showing the overlap of GO:0031497 Chroma�n assembly GO:0006139 Nucleobase, nucleoside, nucleo�de deregulated genes between the different and nucleic acid metabolic process conditions after 6 hours of treatment (B) and 24 GO:0010467 Gene expression GO:0006350 Transcrip�on hours of treatment (C) when compared to the GO:0016070 RNA metabolic process corresponding non-treated timepoints.

GO:0042254 Ribosome biogenesis Turquoise = phosphorus limitation, pink = sugar GO:0006412 Transla�on limitation, black = AZD-8055 treatment. GO:0006413 Transla�on ini�a�on (D) Bubble plot of the Gene Ontology terms GO:0006793 Phosphorus metabolic process enriched in the deregulated genes represented GO:0006468 Protein amino acid phosphoryla�on in panel A. Gray bubbles represent GOs with an GO:0007049 Cell cycle adjusted p-value > 0.05, other colors are on a GO:0051726 Regula�on of cell cycle GO:0051325 Interphase scale (shown below) indicating the log GO:0000278 Mito�c cell cycle transformed adjusted p-value. The sizes of the GO:0022402 Cell cycle process GO:0000279 M phase bubbles correspond to the number of GO:0022403 Cell cycle phase occurrences in the dataset normalized to the GO:0044262 Cellular carbohydrate metabolic process number of occurrences in the genome.

GO:0016049 Cell growth (E) As in D for the different overlaps of the Venn diagrams presented in C (+ when the condition GO:0015979 Photosynthesis is included, - when the condition is excluded, 0 GO:0009790 Embryonic development when the condition is ignored). GO:0009791 Post-embryonic process AZD AZD AZD AZD ilimita � on Pi limita � on Pi ilimita � on Pi limita � on Pi Pi limit. + AZD Pi limit. + AZD Pi limit. Pi limit. + AZD Pi limit. + AZD Pi limit. log10(p-value) frequency (%) ua limita � on Sugar limita � on Sugar ua limita � on Sugar limita � on Sugar Sugar + AZD limit. Sugar + AZD limit. Sugar + AZD limit. Sugar + AZD limit. 60 20 T6 T24 T6 T24 40 40 60

20 80 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Supplemental figure 8 (cont.)

E

GO:0006950 Response to stress GO:0009628 Response to abio�c s�mulus GO:0042221 Response to chemical s�mulus GO:0009725 Response to hormone s�mulus

GO:0006810 Transport GO:0051179 Localiza�on

GO:0016042 Lipid catabolic process GO:0071555 Cell wall organiza�on GO:0071554 Cell wall organiza�on or biogenesis GO:0042545 Cell wall modifica�on GO:0034440 Lipid oxida�on GO:0030258 Lipid modifica�on GO:0019395 Fa�y acid oxida�on GO:0006664 Glycolipid metabolic process GO:0006643 Membrane lipid metabolic process GO:0006633 Fa�y acid biosynthe�c process GO:0006631 Fa�y acid metabolic process GO:0006629 Lipid metabolic process

GO:0071103 DNA conforma�on change GO:0006260 DNA replica�on GO:0006325 Chroma�n organiza�on GO:0031497 Chroma�n assembly GO:0055086 Nucleobase-containing small molecule metabolic process GO:0006139 Nucleobase, nucleoside, nucleo�de and nucleic acid metabolic process GO:0010467 Gene expression GO:0006350 Transcrip�on GO:0016070 RNA metabolic process

GO:0042254 Ribosome biogenesis GO:0006412 Transla�on GO:0006413 Transla�on ini�a�on GO:0043687 Post-transla�onal protein modifica�on

GO:0006793 Phosphorus metabolic process GO:0006468 Protein amino acid phosphoryla�on GO:0006796 Phosphate-containing compound metabolic process

GO:0007049 Cell cycle GO:0051726 Regula�on of cell cycle GO:0051325 Interphase GO:0000278 Mito�c cell cycle GO:0022402 Cell cycle process GO:0000279 M phase GO:0022403 Cell cycle phase GO:0051301 Cell division

GO:0044262 Cellular carbohydrate metabolic process Frequency (%) GO:0009744 Response to sucrose GO:0005975 Carbohydrate metabolic process GO:0005991 Trehalose metabolic process

Photosynthesis GO:0015979 log10(p-value) GO:0010206 Photosystem II repair

GO:0009790 Embryonic development GO:0009791 Post-embryonic process

GO:0007623 Circadian rhythm

GO:0016265 Death GO:0008219 Cell death GO:0006916 Nega�ve regula�on of apopto�c process GO:0006915 Apopto�c process

AZD-8055 + 0 0 - + + - + + - + 0 0 - + + - + + - Pi starva�on 0 + 0 + + - + + - - 0 + 0 + + - + + - - Sugar starva�on 0 0 + - - - + + + + 0 0 + - - - + + + + bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Supplemental figure 9. Expression profile of cell AT2G31270 cycle marker genes (Desvoyes et al., 2020): CDT1a (CDT1a) was shown to be specifically expressed in cells in G1, CycB1;1 was shown to be specifically expressed in cells in the late G2 phase and early in mitosis while the histone HTR13 is predominantly expressed during S phase and early G2 phase. . . . 2.0 1.5 1.0 0.5

AT4G37490 (HTR13) . . . . 3.0 2.5 2.0 1.5 1.0

AT5G10390 (CYCB1;1) . . . . . 3.0 2.5 2.0 1.5 1.0 0.5 Time post treatment (hours) bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A

AT1G50030 AT3G18140 AT2G22040 AT3G08850 AT5G01770

0 -+-+-+-+-+-+ 0 -+-+-+-+-+-+ 0 -+-+-+-+-+-+ 0 -+-+-+-+-+-+ 0 -+-+-+-+-+-+ AZD 0 ++- -++++- -++ 0 ++- -++++- -++ 0 ++- -++++- -++ 0 ++- -++++- -++ 0 ++- -++++- -++ Pi 0 ++++- -++++- - 0 ++++- -++++- - 0 ++++- -++++- - 0 ++++- -++++- - 0 ++++- -++++- - Suc 0 6 24 0 6 24 0 6 24 0 6 24 0 6 24 Time (h) TOR LST8-1 LST8-2 RAPTOR.3G RAPTOR.5G gene

B

0 6 24 Time (h) + + + + - -

0 + + + + - - Suc + + - - + +

0 + + - - + + Pi - + - + - +

0 - + - + - + AZD

At1g50030 (TOR) 0,4

0,2

0

-0,2 log2(fold-change)

-0,4

At3g18140 (LST8-1) 0,4

0,2

0

-0,2

-0,4

At3g08850 (RAPTOR.3G) 0,4

0,2

0

-0,2 o2fl-hne log2(fold-change) log2(fold-change)

-0,4

Supplemental figure 10. Expression profile of TOR complex members. (A) Absolute expression of the genes coding for components of the TOR complex. (B) Expression of genes coding for the components of the TOR complex, relative to their expression value in the T0 samples. Circles represent individual values, red bars represent averages. The horizontal solid lines represent the T0 averages and the horizontal dashed lines represent averages of the corresponding non-treated samples. bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

AT5G01220 AT3G02870 AT3G17790 AT5G64000 AT3G52820 AT2G11810 AT1G73010 Pi recycling AT3G03540 AT3G02040 AT5G20410 AT3G05630 AT4G33030 AT2G27190 AT4G00550 AT2G45130 AT5G20150 Pi signaling and sensing AT2G26660 AT1G68740 AT2G32830 AT2G38940 Pi recovery AT1G20860 AT3G58810 AT5G04950 AT1G23020

AT4G19690 Metal homeostasis AT3G46900 AT4G19680 AT1G09790 AT3G58060 AT5G03570 Response to metal AT2G28160 AT2G03260 Pi sensing AT1G74770 AT1G49390 AT1G72200 Metal binding AT2G35000 AT3G12900 AT4G30120

0 - + - + - + - + - + - + AZD

0 + + - - + + + + - - + + Pi

0 + + + + - - + + + + - - Suc

0 6 24 Time (h)

log2(fold change to T0)

-3 -2 -1 0 1 2 3

Supplemental figure 11. Expression profile of phosphate starvation regulated genes. Expression profiles of the genes listed in Thibaud et al., (2010), normalized to their expression in the T0 samples. Annotations of genes on the right-hand side are from the same article (annotations in red and blue correspond to, respectively, genes described as upregulated after phosphate starvation and genes described as downregulated after phosphate starvation). bioRxiv preprint doi: https://doi.org/10.1101/2021.03.26.437164; this version posted March 28, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. A 6 hours of AZD treatment 24 hours of AZD treatment

101 37 151 76

409 2765 367 3567 359 4706 328 4918 Ren et al., 2012

43 23 55 33

131 2823 3581 4802 4961

(3 days) * 73 119 63 Caldana et al.,Caldana 2013

160 41 159 61

2706 209 3563 4698 4933

(6 days) * * * 114 210 94 Caldana et al.,Caldana 2013

342 2524 791 354 797 4197 706 523 2813 694 4503 517 * * Xiong et al., 2013 *

796 614 583 885 698 4159 582 615 4379 787 2070 3021 * * * Dong et al., 2015 *

B 6hrs 24hrs

1222 407 1747 39

25 93

Nicolaï 109

Supplemental figure 12. Comparison of our datasets with the previously published datasets. (A) Comparison of genes found to be deregulated in response to AZD-8055 treatment in our experiment to the previously published data. Colored circles correspond to our dataset (red for upregulated genes and blue for downregulated genes). Stars represent the datasets for which more than one third of the genes that they found deregulated are shared with our datasets. (B) Comparison of genes deregulated in response to 6 or 24 hours of sugar starvation in our experiment to the previously published work of Nicolaï et al. (2006)