PP2A Methylation is Necessary for Broad Stress Tolerance

Background: LEUCINE CARBOXYL METHYL TRANSFERASE 1 (LCMT1) transfers a methyl group from the methyl donor S-adenosylmethionine (SAM) to the catalytic subunit of PROTEIN PHOSPHATASE 2A (PP2A). This post-translational modication of PP2A is manifested throughout eukaryotes from yeast to plants and animals. Although highly conserved, the importance of the methylation is poorly understood. Since Arabidopsis plants with knocked out LCMT1 grow and develop fairly normally, we decided to search for conditions that may reveal the benets of this regulation. We compared the effects of various stressful conditions on Arabidopsis wild type (WT) and a lcmt1 mutant possessing only non-methylated PP2A. Results: Seedlings were grown in Petri dishes for 5-12 days, or in rock wool and soil for up to 7 weeks. A signicant increase in sodium concentration was found for lcmt1 relative to WT, but this was not linked with stressful conditions. Plants were exposed to variable levels of the chelator EDTA, iron, zinc, aluminium, heat, and hydrogen peroxide. The lcmt1 mutant was clearly more sensitive than WT to all the various stresses, as demonstrated by effects on seedling root growth and on shoots of rosette stage plants on rock wool. When omitting EDTA, expression of genes known as signature genes for iron deciency, FIT1, bHLH100, IMA1, IRT1 was strongly enhanced in lcmt1. Although an iron starvation response was induced, Fe homeostasis was apparently maintained by slowed growth in lcmt1 and the Fe level related to tissue dry weight was not changed. Among genes induced in lcmt1 were also the Zn induced gene ZIF1, and heat shock protein HSP90-1. Concentrations of non-iron transition metals, Cu, Mn and Zn, increased signicantly in response to lack of EDTA for both lcmt1 and WT tissue, and especially the growth of lcmt1 was strongly hampered. Conclusions:

metals may be abundant in the soil, they are not always available for the plant. For example, under aerobic conditions iron is mostly present as Fe(III) which easily forms insoluble compounds in the environment. Plants secrete various substances to improve uptake of micronutrients, e.g. nicotianamine, phytochelatins, organic acids and certain phenolics (Clemens, 2019;Seregin & Kozhevnikova, 2020). To facilitate uptake, the chelator EDTA is added to Murashige-Skoog (MS) medium (Murashige & Skoog, 1962) and Hoagland solution (Hoagland & Arnon, 1950). In Arabidopsis, Fe(III) is rst reduced by FRO2 (FERRIC REDUCTION OXIDASE/REDUCTASE 2) at the root surface, and then the main transporter IRT1 (IRON-REGULATED TRANSPORTER 1) brings Fe(II) into the cell. IRT1 also enables transport of zinc, manganese, copper, cadmium and cobalt ions. Various chaperons, storage proteins and low-molecularweight ligands assure ion homeostasis inside the cells (de Abreu-Neto, Turchetto-Zolet, de Oliveira, Bodanese Zanettini, & Margis-Pinheiro, 2013; Fukao et al., 2011;Seregin & Kozhevnikova, 2020). Since micronutrient homeostasis is essential for all eukaryotes, and we had observed the importance of EDTA for growth of the lcmt1 mutant in early experiments, we decided to test if stress by non-optimal levels of iron and zinc would be aggravated in the lcmt1 mutant. Another metal, aluminium, is a common constraint to crop growth, especially in acidic soil which makes up about one third of the ice-free land area of the world (Horst, Wang, & Eticha, 2010), and the addition of abundant aluminium was tested for plants grown in rock wool. Reactive oxygen species (ROS) are mainly products of basic metabolism and generated during respiration in the mitochondria and reactions linked with photosynthesis in the chloroplasts and peroxisomes. At low levels, ROS are important in signalling pathways, but excess ROS can damage cellular components and lead to destruction of the cell unless scavenging mechanism are active. H 2 O 2 is normally generated in peroxisomes during photorespiration and in microbodies during lipid catabolism as a side-product of fatty acid oxidation and can diffuse between different compartments of the cell. ROS are also common intermediates resulting from many different types of stress, including metal stress and heat stress ( Heat shock transcription factors can be rapidly induced and lead to induction of other protective components, especially HEAT SHOCK PROTEINS (HSPs) (Ohama et al., 2017;Samakovli et al., 2020). Temperature uctuates throughout the day and night, and the global trend of increasing temperature will also have a great in uence on plants. Temperature strongly in uences catalytic reactions, including enzymatic reactions, and high or low temperatures may disrupt the balance in metabolism. High temperature will also increase membranes uidity and affect various compartments in the cell. HSPs are necessary for tolerance to heat and other stresses, and HSPs are ubiquitous in animals, plants and microorganisms (Samakovli et al., 2020 Although a phenotype is visible with more narrow leaves and early owering, the lcmt1 knockout mutant shows growth and development similar to wild type (Creighton et al., 2017). This raises the question why LCMT and PP2A methylation is conserved also in plants. Under what conditions may LCMT be bene cial to plants? In order to elucidate this question, we set off to study the phenotype of Arabidopsis with mutated lcmt1 under various stress conditions, especially stresses caused by disturbing the nutrient balance, oxidative stress, and heat stress.

Material And Methods
Seedlings for observing phenotype Seeds of WT and lcmt1 (At1g02100, SALK_079466, (Alonso et al., 2003)) obtained from the European Arabidopsis Stock Centre in Nottingham UK, and genotyped as described in Creighton et al. (Creighton et al., 2017). Plants were grown in certi ed chambers for gene modi ed plants and handled accordingly, and no permissions were necessary to collect such samples. Seeds were surface sterilized by calciumhypochlorite/ethanol and sown in Petri dishes containing ½ MS medium (Murashige & Skoog, 1962) salts and 1% sucrose. The plates were placed in the dark at 4°C for three days to ensure even germination and thereafter placed at 22°C in 16 h light and 8 h dark cycles. After 5 d at 22 o C, seedlings were transferred to new plates, plates with control MS medium and plates with medium lacking EDTA or containing various compounds to be tested. Roots were measured using imageJ (https://imagej.nih.gov/ij/download.html).

Rosette stage plants
Plants were cultivated in soil for three weeks in a 16 h light/8 h dark regimen and given only water, then plants were moved to rock wool and grown in short days 8 h light /16 h dark (to avoid transition to owering) cycles with Hoagland solution containing FeCl 3 or FeSO 4 as iron source, and with or without EDTA.

RNAseq
After sowing on ½ MS medium salts and 1% sucrose and omitting EDTA, the Petri dishes were placed at 4 o C for three days as usual followed by 5 days at 22 o C before harvesting. At this stage WT and lcmt1 seedlings still looked similar and were considered suitable for harvest and comparison of gene expression. RNA was extracted using RNeasy Plant Mini Kit (Qiagen, Chatsworth, CA, US), and treated with RNAase free (Qiagen)DNase. Euro ns Genomics Europe Sequencing GmbH (Konstanz, Germany) carried out the RNA-seq, three samples for both lcmt1 and WT, averages of the three samples are presented in supplementary Table S2.

Element content
Plant tissue was powdered in liquid nitrogen using a mortar and pestle, then dried for 48 h at 60 o C, then 3 days in a desiccator before analysed by ICP (Inductive coupled plasma mass spectrometry) at NIBIO (Norsk institutt for bioøkonomi, Kjemisk laboratorier), Ås, Norway. For seedlings around 300 shoots were harvested for each sample to make the minimum 100 mg dry weight. For rosette stage plants about 20 rosettes were used for one sample

Results
Visual effects of excluding EDTA from the growth medium In early experiments, we observed negative effects on seedlings and rosette plants when omitting the EDTA chelator from the growth medium, and the effects were strikingly stronger for lcmt1 as compared with WT (Fig. 1). After germination of seedlings on complete MS medium, seedlings were transferred to fresh MS medium or MS medium lacking EDTA. This resulted in chlorosis of shoots and impairment of root growth (Fig. 1). Apparently lcmt1 was less resistant to the stress exerted by lack of EDTA, with strongly inhibited root growth (Fig. 1b). To explore effects of stressful conditions on a later developmental stage, plants were cultivated in soil for three weeks, then moved to rock wool with Hoagland solution with or without EDTA and containing FeCl 3 or FeSO 4 as iron source ( Fig. 1c-f). Both WT and lcmt1 grew well with Hoagland prepared with EDTA in combination with FeCl 3 or FeSO 4 (Fig. 1c, e). When plants were given FeCl 3 without EDTA both WT and lcmt1 became chlorotic (Fig. 1d), and growth of lcmt1 was strongly hampered. When plants were given FeSO 4 as iron source without a chelator ( Fig. 1f), also both WT and lcmt1 became chlorotic, and the stress symptoms were again more severe for lcmt1 (Fig. 1d). Clearly, as for seedlings, omitting EDTA stressed the plants and lcmt1was more affected than WT.

Iron-EDTA
To further get insight into different stress responses of lcmt1 seedlings, the concentration of Fe-EDTA (FeNa 2 EDTA) was varied from 0 to 500 mM in the agar medium ( Fig. 2, Supplemental Fig. S1). Optimal seedling growth was observed at 25 and 50 mM (concentrations in half-strength and regular MS medium) for both WT and lcmt1. Shoots of lcmt1 generally had lower fresh weight than WT. At standard concentration (25 mM Fe-EDTA) fresh weight of lcmt1 shoots was 70% of WT fresh weight. To be able to compare changes in lcmt1 and WT related to the different treatments, all data was standardized to fresh weight and root elongation at 25 mM set to one (Fig. 2). After growth for six days with different iron concentrations, shoot weight of WT seedlings was lower by approximately 45 and 70% when exposed to 350 and 500 mM Fe-EDTA, respectively, but no signi cant difference in decreases was seen for lcmt1 as compared with WT. However, from 200 mM and higher Fe-EDTA there was a negative effect on root elongation in lcmt1 as compared with WT. At 500 mM, roots of both WT and lcmt1 stopped growing (Supplemental Fig. S1). The results showed that high concentrations of Fe-EDTA impaired root growth in lcmt1 more strongly than in WT (Fig. 2b). Interestingly, at zero Fe-EDTA root growth was also more strongly affected in lcmt1 than in WT.

Zinc and other micronutrients
The effects of excluding EDTA from the agar medium in the presence and absence of various micronutrients were tested and showed that fresh weight of shoots was severely reduced in response to lack of EDTA or lack of all micronutrients (Fe, Zn, B, Mn, Mo, Cu, Co, I) (Fig. 3). Addition of 16 mM zinc (standard concentration) led to increased fresh weight, but only when comparing media containing EDTA. A ten times higher, 160 mM, concentration of zinc did not have positive effects. In conclusion, when avoiding the micronutrients, fresh weight of both WT and lcmt1 was more than 50% reduced, but there were no obvious differences between the responses of lcmt1 and WT regarding fresh weight (Fig 3a). On the other hand, growth of roots was more strongly impaired in lcmt1 compared with WT in all media lacking micronutrients and/or EDTA (Fig. 3b).

Aluminium
Shoot fresh weight of seedlings grown on agar decreased when Al 2 (SO 4 ) 3 was added in concentrations of 0.4 or 0.6 mM, but with no signi cant difference in response between WT and lcmt1 (Fig. 4a). Roots of lcmt1 grew signi cantly less in the presence at concentrations 0.2, 0.4, and 0.6 mM as compared with WT ( Fig. 4b, Fig. S2.).
Heat stress Seedlings were exposed to 37 o C for 6-48 h and compared with non-exposed controls. After 18 h or longer time at 37 o C the seedlings gained less shoot fresh weight during the subsequent 6 days, but there was no difference between WT and lcmt1 (Fig. 5a, Fig. S3). Negative effects on root growth were observed after 12 h of heat stress (Fig. 5b). After 24, 36 and 48 h, the effects on root growth were stronger for lcmt1 than for WT (Fig. 5b, Fig. S3).
Heat stress was also tested with rosette plants grown in soil for 5 weeks. Plants were exposed to heat (37 o C) for 0, 8, 18, 24 h, then grown at 22 o C. Pictures were taken i) immediately after treatment, ii) after 1 week and iii) after 2 weeks (Fig.6, Fig. S4). Both WT and lcmt1 grew well after 8 or 18 h heat treatment ( Fig S4). After 24 h heat treatment, inspecting the plants after 1 week of recovery showed more green leaves for WT in three repeats (Fig. 6, Fig. S4), however, after the second week all plants died indicating that the shoot meristem was impaired in both lcmt1 and WT.
ROS are common metabolic intermediates and a result of many different types of stress, including metal stress and heat stress. Experiments con rmed that also ROS, in the form of H 2 O 2 added to the agar medium at 0.5-2.5 mM), resulted in stronger inhibition of root growth in lcmt1 than in WT (Supplemental Fig S5).

Element analysis
Element analysis was performed for shoots of seedlings grown in agar in Petri dishes without or with EDTA for 12 days, and for shoots of 6-week-old rosette stage plants grown rst on soil for three weeks, then on rock wool for 3 weeks without or with EDTA, or without and with iron. When comparing seedlings of lcmt1 and WT, the differences in element contents were generally less than 10% or too variable to become signi cant (Table 1 and supplemental Table S1). When comparing seedlings grown without or with EDTA (pooling WT and lcmt1) there was a clear difference for zinc content; seedlings grown without EDTA having 48% more zinc than seedlings with EDTA. For iron and sulphur, when omitting EDTA, there was a 12% decrease or increase, respectively (Table 1).
For 6-week-old plants on rock wool, as for seedlings, signi cant differences between element contents in rosettes of lcmt1 and WT were generally not found, but with one interesting exception; sodium content was elevated by 64% in lcmt1. The elevated sodium level for lcmt1 persisted also for plants grown without iron, on soil, or with extra aluminium (Supplemental Table S1). As for seedlings, when grown without EDTA, rosette plants had increased levels of zinc (increased by 89%) ( Table 1). For the rosette stage plants, other micronutrients (Cu, Mn, Mo, Ni) also showed elevated levels in the absence of EDTA. From the elevated elements, especially Cu, Mn and Zn were present at high concentrations 40, 185 and 59 mg/g DW, respectively (data based on Table S1). Such high metal levels may have in uenced several metabolic reactions and led to impaired growth and chlorosis (Fig. 1d, f). Other micronutrients responding signi cantly when omitting EDTA were molybdenum and nickel, but these represented only 8 and 3 mg/g DW, and would hardly be harmful. Apparently, in the absence and presence of EDTA, homeostasis mechanisms kept iron constant ( Table 1, Table S1), but growth was impaired. When iron was excluded from the nutrient solution (but EDTA was kept), the level of iron in rosettes decreased by 41% (Table 1, bottom row). Altogether the element analysis showed clear effects of excluding the chelator in the nutrient solution, especially by increased levels of the micronutrients Cu, Mn and Zn, however, signi cant differences related to the lcmt1 mutation was found only for sodium, with increased levels in lcmt1 plants. Table 1.Relativeelement contents in shoots of seedlings on agar or shoots of plants on rock wool. Seedlings: seeds were sown in Petri dishes, grown for 12 days in MS medium without or with EDTA. Rosettes: plants were rst grown for 3 weeks in soil, then transferred to rock wool for 3 more weeks with different nutrient solutions. The Hoagland solution was made without or with EDTA, or without and with iron. Data are given for i) an element in lcmt1 seedlings as percentage of that element in WT seedlings; ii) an element in seedlings grown in the absence of EDTA as percentage of that element in seedlings grown in the presence of EDTA; iii) an element in lcmt1 rosettes as percentage of that element in WT rosettes; iv) an element in rosettes grown in the absence of EDTA as percentage of that element in rosettes grown in the presence of EDTA; v) an element in rosettes grown in the absence of iron as percentage of that element in rosettes grown with iron. P values are given for all comparisons. Percentages that were signi cantly different from 100 with p< 0.05 by TTEST are high-lighted in red,

RNAseq
After sowing and three days at 4 o C, WT and lcmt1 seedlings were germinated and grown under stressful conditions in Petri dishes by omitting EDTA from the growth medium during 5 days at 22 o C, then harvesting. WT and lcmt1 seedlings still looked very similar and were considered suitable for comparison of changes in gene expression caused by lack of EDTA. Some seedlings were left to grow for another week and con rmed the strong growth inhibition on lcmt1 when EDTA was omitted for some more days (as in Fig. 1b). Almost 3 000 genes showed signi cantly different expression levels in lcmt1 and WT, 1528 genes were up-regulated and 1366 genes were down-regulated in lcmt1 in comparison with WT. When the limit was set to 2x different expression levels, 520 genes were up-regulated and 653 genes were down-regulated in lcmt1 as compared with WT. Several genes (155) were more than 5x downregulated in lcmt1 ( Table 2, Supplemental Table S2 with original data).  Tables S3-S8). Genes up-regulated in lcmt1 were enriched in iron homeostasis genes, synthesis of cutin and various stress-annotated genes. Iron homeostasis genes were further inspected (Fig. 7). These genes included ve BHLH transcription factors and the BRUTUS E3 ligase involved in regulation of iron homeostasis regulation, two genes directly involved in uptake of iron from the environment i.e. IRT1 and FRO2, the vacuolar localized metal transporters FRD3 and NRAMP4, and the oligopeptide transporter gene encoding a phloem speci c iron transporter gene OPT3. We manually added three IMA genes (IRONMAN1,2,6). Other IMA genes, IMA 3 and 4 were also 4-5 times higher expressed in lcmt1, but are not included in the gure. IMA genes were recently shown to be crucial for regulation of iron transport and homeostasis (Grillet, Lan, Li, Mokkapati, & Schmidt, 2018). We also added genes involved in chelating or storage of iron. Three NAS genes (1,2,4) encoding enzymes that synthesise the chelator nicotianamine were upregulated, whereas the iron storage protein FER1 was downregulated in lcmt1, indicating less storage capacity. The PP2AA3 gene, used as a control, was constant. Many stress related genes were differentially expressed in lcmt1 and WT, they were either up (97) or down (128) regulated in lcmt1. Transporter genes were also differentially expressed, both up (41) and down regulated (128) in lcmt1 (Table 3). Genes involved in photosynthesis, i.e. light harvesting, were strikingly downregulated in lcmt1. Also, genes encoding hem-binding proteins many of which are cytochromes, were down-regulated in lcmt1. The ZINC-INDUCED FACILITATOR (ZIF1) was expressed at a high level and 2.9 times higher in lcmt1 than in WT seedlings (Table S2). ZIF1 is important for transportation of nicotianamine and Zn into the vacuole and critical for both Zn and Fe homeostasis (Haydon et al., 2012). The heat shock protein HSP90-1 known to be involved in many different types of stresses was 2.2 times higher in lcmt1 than WT (Table S2).

Discussion
The various stressful conditions tested on Arabidopsis seedlings in Petri dishes all resulted in stronger inhibition of primary root growth in lcmt1 than in WT. WT also showed impaired root growth in response to the different treatments, but lcmt1 was always more susceptible to the stresses. These stresses were: lack of chelator, elevated levels of iron, zinc and aluminum, heat, and oxidative stress. Seedling shoot weight was in uenced by all stresses tested, and the decrease in shoot fresh weight was the same in lcmt1 and WT. This may be explained by the relatively short time period for the seedlings' exposure to stress (6 days) because plants in rock wool under stress for a longer time (three weeks) showed clear differences between shoot growth of lcmt1 and WT, with lcmt1 growth being much stronger impaired (Figs. 1). The fact that lcmt1 showed decreased primary root growth, regardless of the type of stress imposed, indicates that a basic function of PP2A methylation is involved.

Elements
Total element analysis revealed that one element, sodium, accumulated differently in rosette stage plants of lcmt1 and WT, with higher levels in lcmt1 (plus 64 %), and this was reproducible throughout different experiments regardless of the presence of chelator or iron, or growth in rock wool or soil (Table 1 and   Supplemental Table S1). Other elements did not vary systematically between WT and lcmt1. Omitting EDTA for three weeks had a pronounced effect on element contents for both genotypes, with high accumulation, especially of the micronutrients copper, manganese, and zinc (Table 1 and Table S1).
Interestingly, the level of iron was not much in uenced by omitting EDTA, homeostasis was apparently upheld for this element (but growth impaired). The concentration of copper, manganese and zinc, all together, made up a concentration two times that of iron in plants growing without EDTA, but in plants with EDTA these elements summed up to a concentration lower than iron (by 17%). The accumulation of copper, manganese and zinc may have given an ion imbalance and replaced iron in important proteins and cofactors leading to impaired plant growth. Apparently, WT, with methylated PP2A, is better to cope with this imbalance than lcmt1.
Native PP2A isolated from mammals contains iron and zinc in its active site, while recombinant PP2A catalytic subunit was found to also have manganese. Furthermore, manganese binding was associated with binding of the PP2A chaperon PTPA. During in situ activation of PP2A in mammals, binding of the PTPA chaperon and manganese apparently led to PP2A being active not only towards phosphorylated Ser/Thr residues, but also towards phosphorylated tyrosyl residues. After binding of the PTPA chaperon, LCMT1 methylates PP2A, which leads to release of the chaperon causing speci city of PP2A towards phosphorylated Ser/Thr residues only (Guo et al., 2014). Assuming the same activation mechanism applies to plants; since PP2A is not methylated in the lcmt1 mutant, PP2A may be more in uenced by an imbalance of the micronutrients and speci city towards substrates may decrease because of tighter binding of the PTPA chaperon. Our hypothesis is that high levels of manganese in the plant tissue, as was a result of omitting the chelator EDTA, may be especially damaging in the lcmt1 mutant by promoting less speci city of PP2A.

Transporters and protein cycling
Many of the genes involved in iron homeostasis were highly induced in lcmt1seedlings, and appeared as an enriched gene ontology group, about 13 times enriched ( Table 3, Fig. 7). The lcmt1 seedlings responded as if iron-depleted, however, the element analysis did not reveal lack of iron. The iron transporter IRT1 is present in the plasma membrane as a complex with FRO2 and the proton pump AHA2. could be a way that LCMT1 impinges on the IRT1 and iron homeostasis. Methylated PP2A has previously been found to be more frequently localized to endosomes than non-methylated PP2A (Wu et al., 2011).
Possibly, lack of methylated PP2A, as in the lcmt1 mutant, could lead to unbalanced IRT1 endo/exocytosis, but these mechanisms remain unclear. The gene expression analysis pointed to a range of transmembrane transporters differentially expressed in lcmt1 and WT, and points to lcmt1 being of importance for membrane functions ( Table 3, Table S7).

Sodium
Element levels were very similar in lcmt1 and WT except for sodium, which was increased by 64% in samples of rosette shoots of lcmt1 relative to WT ( Table 1). The NHX1 and NHX2 are the main contributors to transport of sodium and potassium into vacuoles in Arabidopsis and essential for ion balance in plants (Bassil, Zhang, Gong, Tajima, & Blumwald, 2019). The NHX1 transcript level was more than 10 times higher than the other vacuolar NHX transcripts, indicating its importance for ion homeostasis in the present experiments. NHX1 was signi cantly upregulated (by factor 1.7) in lcmt1 relative to WT as revealed in the RNAseq data (Table S1) and pointing to the possibility that the lcmt1 mutation inferred with sodium balance. However, the accumulation of sodium in lcmt1 may possibly also be linked with other genes, and the transporters of sodium are still not fully identi ed (Bassil et al., 2019).

Conclusion
The mutant lcmt1 was always more susceptible to various stress treatments than WT. The methylation of PP2A apparently is of fundamental value for coping with stress and appears to be important for tolerating an imbalance in the micronutrients.  WT and lcmt1, respectively, and 75 mm elongation of primary roots for both WT and lcmt1. Data are averages of 30 seedlings for each treatment and genotype, SE is given, a star indicates signi cant difference between WT and lcmt1 by TTEST, for p < 0.05.

Figure 3
Comparison of WT and lcmt1 growth on agar medium with or without micronutrients and EDTA. a) Shoot relative fresh weight; b) Primary root relative elongation. After germination on the standard half strength Page 22/26 MS medium for 5 days, seedlings were transferred to the different media, half MS medium with and without EDTA, media lacking micronutrient with or without EDTA (no E), and media with addition of Zn at 16 mM or 160 mM but with no other micronutrient. Data were normalized to the values at 25 mM (control medium with EDTA) set to one. Data are based on 45 seedlings for each treatment and plant type, SE is given as vertical bars. A star indicates signi cant difference between WT and lcmt1, p < 0.05 by TTEST. Figure 4 Effects of heat stress on WT and lcmt1seedlings. a) Shoot relative fresh weight; b) Primary root relative elongation. After germination on the standard half strength MS medium for 5 days, seedlings were transferred to fresh medium and exposed to 37oC for 0, 6, 12, 18, 24, 36 and 48 h. After the heat treatment, plates were placed back to 16 h light/8 h dark cycles at 22oC for 6 days before assessment.
Data are based on 45 seedlings for each treatment and plant type, SE is given. A star indicates signi cant difference between WT and lcmt1, p < 0.05 by TTEST.

Figure 6
Effects of heat stress on WT and lcmt1 rosette plants. Plants were grown in soil for 5 weeks, then treated with heat (37oC) for 24 h, and placed back at 22oC. Pictures were taken immediately after treatment, after 1 week and 2 weeks (more repeats are given in Fig. S4).

Figure 7
Expression of genes involved in iron homeostasis. Expression levels in lcmt1 are given relative to expression in WT. Seedlings were grown on ½ MS medium without EDTA for 5 days before harvesting whole seedling for RNAseq. The genes included are iron homeostasis regulator genes ( ve bHLH transcription factors, BTS and IMA genes), genes directly involved in uptake from the environment (IRT1, FRO2), genes encoding vacuolar metal transporters (FRD3, NRAMP4), gene encoding a phloem iron transporter OPT3, three nicotianamine synthesis (NAS) genes, a gene encoding the iron storage protein FER1, and the reference gene PP2AA3 (for more details see