Cesium tolerance is enhanced by a chemical which binds to BETA-GLUCOSIDASE 23 in Arabidopsis thaliana

Cesium (Cs) is found at low levels in nature but does not confer any known benefit to plants. Cs and K compete in cells due to the chemical similarity of Cs to potassium (K), and can induce K deficiency in cells. In previous studies, we identified chemicals that increase Cs tolerance in plants. Among them, a small chemical compound (C17H19F3N2O2), named CsToAcE1, was confirmed to enhance Cs tolerance while increasing Cs accumulation in plants. Treatment of plants with CsToAcE1 resulted in greater Cs and K accumulation and also alleviated Cs-induced growth retardation in Arabidopsis. In the present study, potential target proteins of CsToAcE1 were isolated from Arabidopsis to determine the mechanism by which CsToAcE1 alleviates Cs stress, while enhancing Cs accumulation. Our analysis identified one of the interacting target proteins of CsToAcE1 to be BETA-GLUCOSIDASE 23 (AtβGLU23). Interestingly, Arabidopsis atβglu23 mutants exhibited enhanced tolerance to Cs stress but did not respond to the application of CsToAcE1. Notably, application of CsToAcE1 resulted in a reduction of Cs-induced AtβGLU23 expression in wild-type plants, while this was not observed in a high affinity transporter mutant, athak5. Our data indicate that AtβGLU23 regulates plant response to Cs stress and that CsToAcE1 enhances Cs tolerance by repressing AtβGLU23. In addition, AtHAK5 also appears to be involved in this response.

www.nature.com/scientificreports/ Understanding the mechanism and the regulation of Cs entry and plant response to Cs is important as it will facilitate phytoremediation of Cs-contaminated soils. Genetic modification of plants or the application of a chemical that induces Cs-tolerance can be used as strategies to control Cs toxicity in plants. Regarding the latter approach, several chemical compounds that function as Cs-tolerance enhancers (CsTolen) and/or Csaccumulators were identified in previous studies through the screening of synthetic chemical libraries comprising 20,000 small organic compounds 13,17 . The application of one of the CsTolen chemicals, CsTolen A, interrupted Cs uptake into plants by binding with Cs ions outside of cells, thus, rendering the plants more Cs-tolerant by reducing Cs accumulation 17 . Application of methyl cysteinate, a derivative of a sulfur-containing cysteine, was found to enhance cesium accumulation in treated plants 13 . Sulfur-containing metabolites, such as glutathione, were also found to alleviate Cs stress in Arabidopsis 18 .
The long and rod-shaped endoplasmic reticulum (ER) in Arabidopsis has been extensively studied 19 . Two types of ER bodies have been identified in Arabidopsis; a constitutive ER body and an inducible ER body. The former has been named due to their visible detection in epidermal cells of cotyledons, hypocotyls, and roots of young seedlings of Arabidopsis, and contains large amounts of PYK10/BETA-GLUCOSIDASE 23 (AtβGLU23) proteins, while the latter is induced in cells by wounding or methyl jasmonate treatment in rosette leaves and contains AtβGLU18 20,21 . Jasmonic acid is a phytohormone that regulates various physiological processes related to plant growth and development, and plant response to abiotic and biotic stresses, such as wound response and insect/pathogen attacks [22][23][24] . Exogenous application of methyl jasmonate induces the formation of ER bodies, suggesting that ER bodies mediate plant defense response to biotic and abiotic stress. The biogenesis of constitutive ER bodies is controlled by the NAI1.2 transcription factor 25 . The biogenesis of inducible ER bodies, however, is dependent on unidentified transcription factors that downregulate the expression of TONSOKU-ASSOCIAT-ING PROTEIN 1 (TSA1), a NAI2 homolog, and AtβGLU18, which is located downstream of the jasmonic acid signaling pathway 26 . βGLU proteins, as well as ER bodies, have been proposed to play an integral role in plant immunity [27][28][29] . The expression levels of TSA1 and AtβGLU18 were reported to be significantly reduced in the jasmonic acid biosynthesis mutant aos (allene oxide synthase) and the jasmonic acid-insensitive receptor mutant coi1 (coronatine insensitive1) [30][31][32] . In a previous study using aos and coi1 mutant plants, we reported that jasmonate biosynthesis and signaling are involved in Cs response 33 . Cs treatment increases the levels of jasmonates 18 and methyl jasmonate suppresses Cs-induced expression of AtHAK5 18 .
In the present study, a chemical which enhances Cs tolerance in plants was found to bind to AtβGLU23. Furthermore, the role of AtβGLU23 in Cs-stress response was investigated, as well as its relation to other previously mentioned factors, such as AtHAK5 induction and the jasmonic acid dependent pathway, during Cs stress in planta.
Phenotype quantification. The fresh weight of aerial plant parts was determined (n > 60). Primary root lengths were measured in digital images of primary roots using ImageJ (n > 60) software 37 , and statistical differences between sample means were evaluated with a one-way ANOVA followed by a Bonferroni's multiple comparisons test using Prism software version 5 (GraphPad Software, San Diego, USA).

Identification of direct targets of a small molecule (CsToAcE1). Photo-cross-linking of CsToAcE1
with agarose beads, CsToAcE1-PALC (photoaffinity-linker-coated) agarose beads, was conducted as previously described 38 . Eleven-day-old Arabidopsis seedlings were sampled and total proteins were isolated in an extraction buffer [50 mM Tri-HCl pH 7.6, 100 mM NaCl, 1 mM EDTA, 100 mM MgCl 2, 1% Triton-X, anti-protease (Roche, Basel, Switzerland)]. The extracted proteins were incubated with the chemical-linked or non-linked control beads in the presence of 10 mM CsCl for 18-24 h at 4 °C at a low rotary speed. The incubated beads were precipitated by centrifugation at 4000 rpm, 4 °C for 1 min, and were washed with the extraction buffer on ice. The proteins bound to the beads were eluted in 1 × Laemmli sample buffer at 25 °C for 30 min and were then boiled for 5 min. The eluates were separated in a 10% SDS − PAGE gel, and the bands were visualized using Coomassie blue staining. Selected protein bands were excised from the SDS − PAGE gel and subjected to modified in-gel trypsin digestion using sequencing-grade trypsin (Promega, Wisconsin-Madison, USA). The digestion mixture was separated on a nanoflow LC instrument (Easy nLC) (Thermo Fisher Scientific, Waltham, USA) using a nanoelectrospray ionization spray column (NTCC analytical column, C18, φ75 µm × 100 mm, 3 µm; Nikkyo Technos Co., Tokyo, Japan) coupled to a Q-Exactive mass spectrometer (Thermo Fisher Scientific) equipped with a nanospray ion source. MS and MS/MS data were acquired using the data-dependent top 5 method. The resulting MS/MS data were searched using Mascot search (Matrix Science, London, UK) with the following parameters: Gln → pyro-Glu (N-term Q), oxidation (M), carbamido-methyl (C), and Hex (W) 39  www.nature.com/scientificreports/ Elemental analysis. Three biological replicates of whole, 8-day-old plantlets (40-60 seedlings pooled per replicate) were collected, rinsed in Milli-Q water, and dried in an oven at 65 °C for 3-4 days. Subsequently, 2 ± 0.1 mg of dried sample was weighed and extracted as described in our previous study 17 . The concentrations of K and Cs were determined with a flame atomic absorption spectrometer AAnalyst 200 (PerkinElmer) or by inductively coupled plasma mass spectrometry (NexION® 300 ICP-MS System, Perkin Elmer, Waltham, USA). The concentrations of K and Cs were calculated based on standard curves generated for each element. Statistical differences between sample means were evaluated with a one-way ANOVA followed by a Bonferroni's multiple comparisons test using Prism software version 5 (GraphPad Software, San Diego, USA).

Reverse transcription quantitative PCR (RT-qPCR).
Total RNA was extracted from samples using Trizol (Thermo Fisher Scientific), and cDNA synthesis was performed with MMLV reverse transcriptase (Invitrogen, Carlsbad, USA) and oligo(dT), according to the manufacturer's instructions. RT-qPCR was performed in a Max3000P qPCR system (Agilent Technologies, Santa Clara, USA) using THUNDERBIRD SYBR qPCR mix (TOYOBO, Osaka, Japan). PCR conditions were as follows: first denaturation at 95 °C for 1 min; 40 cycles of denaturation at 95 °C for 15 s, followed by annealing at 60 °C for 30 s; denaturation at 95 °C for 1 min; and dissociation for a melting curve. RT-qPCR assays utilized three biological replicates for each sample. The primers used were as follows: AtβGLU23 (forward, 5′-CAA TGA GCC ATG GGT TTT CT-3′ and reverse, 5′-CGT ATC CTG ATC GTC CGT CT-3′), AtHAK5 (forward, 5′-GAG ACG GAC AAA GAA GAG GAACC-3′ and reverse 5′-CAC GAC CC TTC CCG ACC TAA TCT-3′) 40 , and β-tubulin2 (TUB2) (forward 5'-GCC AAT CCG GTG CTG GTA ACA and reverse 5′-CAT ACC AGA TCC AGTTC CTC CTC CC-3′), the latter of which was used as a reference gene 40 . A Tukey's comparisons test and student t-test were performed using Prism to determine the statistical significance relative to a control (K).
β-glucosidase activity assay.. One hundred mg of 8-day-old Arabidopsis seedlings were homogenized by grinding the sample in 0.5 ml of 50 mM sodium phosphate buffer (pH7.0). The homogenate was filtered through a 5.0 µm hydrophilic, polyvinylidene difluoride membrane (Millipore Co., Burlington, USA) 41 . The concentration of protein in each filtered, total extract was determined using a Bradford protein assay (Bio-Rad, Hercules, USA). β-glucosidase activity was assessed using a β-glucosidase assay kit (Abcam, Waltham, USA) following the manufacturer's protocol. Briefly, 20 µl of filtered total extract was mixed with the working reagent (200 µl of assay buffer with 8 µl of 4-nitrophenyl-β-d-glucopyranoside (β-NPG) substrate) and incubated at 37 °C for 6 h. β-glucosidase activity was monitored as the change in OD 420 every hour using a spectrophotometer (Beckmann, Oklahoma City, USA). One unit of enzyme was calculated based on catalysing the hydrolysis of 1 µmole of β-NPG per min at pH7.0. The assay was repeated three times and one representative data is presented.

Results
Identification of CsToAcE1 cellular target proteins. Several novel chemical compounds were identified as Cs-stress tolerance-inducing or Cs-accumulation enhancing chemicals in our previous chemical screening analysis 13,17 . Among the chemicals identified in the previous study, C 17 was classified as a Cs-Tolerance Inducer and Cs-Accumulation Enhancer, and designated as CsToAcE1 (Fig. 1A). A modified Hoagland plant media was used to adjust the concentration of K and Cs in the medium. Treatment of Arabidopsis plants with 0.3 mM CsCl (K + Cs) triggered growth retardation, however, the application of CsToAcE1 attenuated the Cs-induced growth retardation (Fig. 1B).
In the present study, CsToAcE1-binding target proteins were isolated to elucidate how CsToAcE1 functions in plant cells to improve Cs tolerance, and enhance the accumulation of Cs under Cs stress conditions, at the same time. A pull-down assay using total protein extracts from Arabidopsis plants grown on media supplemented with 0.3 mM CsCl recovered a band around 60 kDa (red box), which was only visible in the eluates derived from the CsToAcE1-beads (Fig. 1C, lane 1 and 2). These bands were excised and subjected to peptide sequencing analysis to identify the proteins 38 . A protein extracted from these bands (red box in lane 1 and lane 2) was identified as Arabidopsis thaliana BETA-GLUCOSIDASE 23 (AtβGLU23, At3g09260, gi|15232626, UniProtKB-Q9SR37, https:// www. unipr ot. org/ unipr ot/ Q9SR3 7) with 35.3% sequence coverage and a score of 1057 (Fig. S1).
AtβGLU23 is involved in plant response to Cs stress. Two mutant lines of AtβGLU23 were used to assess the role of CsToAcE1-binding AtβGLU23 protein in Cs uptake and plant response to Cs stress. One of the mutants, originally named pyk10-1 (CS69080), contains a T-DNA insertion in close proximity to the first exon of AtβGLU23 (PYK10-1) and no transcript is detected in the mutant 33,34 . Another mutant, named leb-2, contains a single nucleotide change (CCT → TCT) in the first exon and this change results in an amino acid substitution (P41S) on PYK10 (Dr. Ikuko Hara-Nishimura, personal communication). Transcripts of AtβGLU23 were detected in the leb-2 mutant, however, the amount and size of AtβGLU23 protein were altered, similar to the leb-1 mutant which is a single amino acid change mutant and (C29Y) had the fewer and larger ER bodies 35 . In the present study, the two mutants were designated as atβglu23-1 (pyk10-1) and atβglu23-2 (leb-2). The growth of 8-day old Col-0 and atβglu23 mutant plants in response to 0.3 mM CsCl was compared (Fig. 2). Both atβglu23-1 and atβglu23-2 mutants exhibited less Cs-induced growth retardation and milder aerial chlorosis relative to the negative effects observed in wild-type plants (Fig. 2A). The alleviation of Cs-induced growth retardation in the mutants suggested the possibility that AtβGLU23 was involved in the response of Arabidopsis plants to Cs stress, even in the absence of the application of CsToAcE1. Therefore, Cs and K levels were measured in wild-type and mutant plants utilizing an atomic absorption spectrophotometer to determine if the phenotype of the atβglu23 mutants and wild-type plants was associated with alterations in the concentration of K or Cs (Fig. 2B) www.nature.com/scientificreports/ levels in wild-type, Col-0 and atβglu23 mutants were comparable in the absence of the Cs treatment. Notably, Cs and K accumulation in the atβglu23 mutants was higher than the level of accumulations observed in Col-0 plants in the presence of Cs. These observations suggest that the lack of intact AtβGLU23 conferred tolerance to Cs stress in Arabidopsis plants, while accumulating higher levels of Cs, as well as K. The data also indicate that AtβGLU23 is associated with plant response to Cs stress by altering K and Cs accumulation.  www.nature.com/scientificreports/ (Fig. 3A). Treatment with CsToAcE1 under Cs stress condition resulted in lower reduction in the fresh weight and root growth relative to the untreated control in the wild-type plants, (Fig. 3B,C), while the CsToAcE1 treatment had no effect on the growth of the Cs-stressed atβglu23-2 mutant plants (Fig. 3B,C). Elemental analysis revealed that the mutation in the AtβGLU23 gene resulted in higher K accumulation under Cs-stress conditions, which was consistent with our initial data (Figs. 2B and 3D). Interestingly, the CsToAcE1 treatment resulted in improved K accumulation in Col-0 plants exposed to Cs stress to a level that was equivalent to that of atβglu23-2 plants, while the K content in the atβglu23-2 mutant was not further enhanced by the CsToAcE1 treatment (Fig. 3D). Cs content in Col-0 plants increased in response to the CsToAcE1 treatment, however, CsToAcE1 treatment of the atβglu23-2 mutant did not induce any further increase in Cs levels beyond the elevated Cs levels observed in the absence of the CsToAcE1 treatment (Fig. 3E). No synergetic or antagonistic effect of the CsToAcE1 treatment was observed in the response of atβglu23-2 mutant plants to Cs stress. The determination that CsToAcE1 does not further enhance Cs tolerance or Cs accumulation in the absence of AtβGLU23 in the atβglu23-2 mutant suggests that AtβGLU23 is the target protein of CsToAcE1.

CsToAcE1 regulates AtβGLU23 and AtHAK5 expression. Analysis of AtβGLU23 expression in wild-
type Col-0 and mutant plants indicated that AtβGLU23 expression in both shoots and roots of Col-0 plants increased approximately two-fold in response to the Cs-stress treatment, relative to control plants (no Cs stress). AtβGLU23 expression in the presence of Cs, however, was attenuated by the CsToAcE1 treatments in Col-0 shoots. AtβGLU23 expression was also attenuated by the CsToAcE1 treatments regardless of the presence or absence of Cs in Col-0 roots (Fig. 4). In contrast, the AtβGLU23 expression in athak5-2 roots was not attenuated by the CsToAcE1 treatment (Fig. 4). These data indicate that CsToAcE1 has little effect on AtβGLU23 expression in the athak5-2 roots. www.nature.com/scientificreports/ The expression of AtHAK5, a K-deficiency marker, which was previously shown to be induced by the Cs treatment 5 , was induced by the Cs treatment in Col-0 shoots and roots but only in the roots of atβglu23-2 (Fig. 5). However, the expression of AtHAK5 was not attenuated by CsToAcE1 in both Col-0 and the atβglu23-2 roots and was increased by CsToAcE1 in the presence of Cs in Col-0 shoots unlike AtβGLU23 expression (Figs. 4 and 5). Our previous study demonstrated that a jasmonic acid biosynthesis mutant (aos) and a jasmonic acid-insensitive mutant (coi1-16) exhibited greater tolerance to Cs stress than wild type plants and that Cs induces jasmonate biosynthesis and signaling 32 . Therefore, the expression of AtβGLU23 in aos and coi1-16 was analyzed to determine the involvement of jasmonic acid signaling in AtβGLU23 expression (Fig. S2). Results indicated that AtβGLU23 expression was upregulated in aos and coi1-16 mutants in response to the Cs treatment, which was similar to its expression in Col-0 plants. Unlike Col-0, however, the increased AtβGLU23 expressions by Cs treatments were not attenuated by CsToAcE1 in roots of athak5-2 mutant and the jasmonate-related mutants (Fig. S2 and Fig. 4).
CsToAcE1 negatively regulates β-glucosidase activity. β-glucosidase activity was measured in wildtype and atβglu23-2 plants to better understand how CsToAcE1 is involved in the regulation of AtβGLU23. β-glucosidase activity was induced by Cs-stress, while the CsToAcE1 treatment reduced the Cs-induced β-glucosidase activity in wild-type plants (Fig. 6). Overall enzyme activity was reduced in atβglu23-2 plants, consistent with the past data on another atβglu23 mutant 41 . In contrast to the β-glucosidase activity observed in wild-type plants, the increase in β-glucosidase activity by Cs-stress in atβglu23-2 plants was attenuated and hardly any reduction in enzyme activity was observed in atβglu23-2 plants in response to the CsToAcE1 treatment (Fig. 6). Error bars indicate standard errors (n = 3). Statistical differences relative to a control (K) were determined using a Tukey's comparisons test and statistical differences between no CsToAcE1 treated and CsToAcE1 treated were determined using student t-tests (N.D., no difference; *p < 0.05; **p < 0.01; ***p < 0.001). www.nature.com/scientificreports/ In summary, Cs stress increases AtβGLU23 transcript levels and enzyme activity, and induces growth inhibition. A Cs tolerance-enhancing chemical, CsToAcE1, binds to AtβGLU23, which is potentially mediated by jasmonates and AtHAK5, and reduces Cs-induced growth inhibition (Fig. 7).

Discussion
In the present study, AtβGLU23 in Arabidopsis was identified as the target protein of the Cs-tolerance enhancer and Cs-accumulator chemical, CsToAcE1. Interestingly, atβglu23 mutants exhibited a Cs tolerant phenotype, which was similar to that of wild-type plants treated with CsToAcE1 (Figs. 2 and 3). Reduced levels of AtβGLU23 were associated with a reduction in the growth retardation induced by Cs, and it appears that the level of Cs in mutant plants is not correlated with the level of Cs tolerance (Fig. 3). It is plausible that AtβGLU23 is involved in the response to Cs stress rather than in regulating the uptake of external Cs. CsToAcE1 may moderately suppress the negative effect of AtβGLU23 on plant growth in response to Cs, thus, conferring enhanced Cs tolerance to plants. RT-qPCR data indicate that AtβGLU23 is induced when Cs is applied and that the induced level of AtβGLU23 expression is somewhat attenuated by the CsToAcE1 treatment (Fig. 4). Furthermore, an increase in β-glucosidase activity was observed under Cs-stress conditions in wild-type plants but was alleviated by CsToAcE1. Comparatively, the induction of β-glucosidase activity by Cs stress and its alleviation by CsToAcE1 was much less in atβglu23-2 plants (Fig. 6). These results support the hypothesis that CsToAcE1 negatively regulates AtβGLU23 in plants under Cs stress. The non-responsiveness of the atβglu23 mutants to CsToAcE1 supports the premise that AtβGLU23 is the target of this chemical.
Beta-glucosidases are known to be the main components of ER bodies and AtβGLU23 (PYK10) and AtβGLU18 are the most abundant ER body proteins in Arabidopsis. AtβGLU21 and AtβGLU22 have also been suggested to regulate constitutive ER body formation, while AtβGLU18 is involved in inducible ER body formation in response to wounding and methyl jasmonate 20,21,34,42 . Several studies have suggested that the ER bodies The expression of AtHAK5 was used as a marker for K deficiency in planta. Error bars indicate standard errors (n = 3). Statistical differences relative to a control (K) were determined using a one-way ANOVA followed by a Tukey's comparisons test and statistical differences between no CsToAcE1 treated and CsToAcE1 treated were determined using student t-tests (N.D., no difference; *p < 0.05; **p < 0.01; ***p < 0.001). www.nature.com/scientificreports/   www.nature.com/scientificreports/ play a role in plant response to biotic and abiotic stress 20,[27][28][29]43 . Data obtained in our previous study suggests a link between Cs stress and jasmonate synthesis and signaling 33 . In the present study, inhibition of AtβGLU23 expression by CsToAcE1 in the presence of Cs was not observed in jasmonic acid mutants (Fig. S2), suggesting that jasmonates may play a role in CsToAcE1-AtβGLU23-mediated Cs response. In a previous study, Wang et al. demonstrated that the transcript level of AtβGLU18 was significantly reduced in aos and coi1-16, relative to basal levels expressed in wild-type plants 30 . In addition, the exogenous application of methyl jasmonate was found to induce the biogenesis of ER bodies, which in Arabidopsis requires AtβGLU protein. An interaction between jasmonic acid and AtβGLU proteins in plant response to various stresses, including Cs, has been reported 20,21 . In our present study, Cs-induced AtβGLU23 expression was partly inhibited by CsToAcE1, however, this inhibition was not observed in the athak5 roots (Fig. 4). In contrast, Cs-induced AtHAK5 expression was not inhibited by CsToAcE1 in either Col-0 or the atβglu23-2 mutant (Fig. 5). AtHAK5 protein has been reported to be primarily located in ER bodies when K is sufficient and localized to the plasma membrane under K-deficient conditions 14 . Therefore, Cs-induced K deficiency in plants may also lead to the translocation of AtHAK5 from ER bodies to the plasma membrane. Furthermore, Cs-induced AtβGLU23 expression and its downregulation by CsToAcE1 may be linked to the translocation of AtHAK5 due to the K status. Therefore, further studies are warranted to evaluate the effect of CsToAcE1 on the subcellular localization of AtHAK5 and AtβGLU23 in Arabidopsis plants. Notably, the overexpression of an Arabidopsis beta-glucosidase, AtBG1, which hydrolyzes inactive, glucose-conjugated abscisic acid to active abscisic acid, has been demonstrated to enhance tolerance to drought and salt stress 44 . Therefore, AtβGLU proteins appear to be broadly involved in plant response to a variety of stresses. In summary, we demonstrated that AtβGLU23 plays a negative role in Cs stress tolerance and CsToAcE1 attenuates the negative effects of Cs through the inhibition of AtβGLU23 activity (Fig. 7). Further studies detailing the functional mechanisms of the relationship between AtβGLU23, Cs stress, and CsToAcE1 are in progress to increase our understanding of Cs response in plants.