During plant growth and defense, cell cycle activity needs to be coordinated with cell wall integrity. Little is known about how this coordination is achieved. Here, we investigated coordination in Arabidopsis thaliana seedlings by studying the impact of cell wall damage (CWD, caused by cellulose biosynthesis inhibition) on cytokinin homeostasis, cell cycle gene expression and cell shape in root tips. CWD inhibited cell cycle gene expression and increased transition zone cell width in an osmosensitive manner. These results were correlated with CWD-induced, osmosensitive changes in cytokinin homeostasis. Expression of CYTOKININ OXIDASE/DEHYDROGENASE 2 and 3 (CKX2, CKX3), which encode cytokinin-degrading enzymes, was induced by CWD and reduced by osmoticum treatment. In nitrate reductase1 nitrate reductase2 (nia1 nia2) seedlings, CKX2 and CKX3 transcript levels were not increased and cell cycle gene expression was not repressed by CWD. Moreover, established CWD-induced responses, such as jasmonic acid, salicylic acid and lignin production, were also absent, implying a central role of NIA1/2-mediated processes in regulation of CWD responses. These results suggest that CWD enhances cytokinin degradation rates through a NIA1/2-mediated process, leading to attenuation of cell cycle gene expression.
Chemically complex cell walls surround all plant cells and are a distinctive feature of plants. These walls form a mechanical barrier during defense but are plastic during growth, cell morphogenesis and division. This is achieved by active modulation of the wall metabolism that underpins the carbohydrate-based complex wall structure, consisting mainly of cellulose, hemicelluloses, pectins, lignin and different cell wall proteins (Höfte and Voxeur, 2017). Despite a need for coordination between plant cell wall metabolism and cell division, the processes underlying this coordination are largely unknown (Gigli-Bisceglia and Hamann, 2018). In Saccharomyces cerevisiae, the cell wall integrity (CWI) maintenance mechanism is intricately involved in this coordination (Levin, 2011), with a dedicated checkpoint monitoring CWI and regulating cell cycle progression (Kono et al., 2016; Levin, 2011).
In plants, CWI can be impaired during growth and following exposure to abiotic and biotic stress (Bacete et al., 2018; Feng et al., 2018; Wolf, 2017). CWI-dependent responses include modifications of cellular metabolism (such as osmosensitive changes in carbohydrate metabolism and phytohormone production), growth inhibition, modification of pectic polysaccharides and deposition of lignin and callose (Engelsdorf et al., 2018; Bacete et al., 2018; Cano-Delgado et al., 2003; Ellis and Turner, 2001; Hamann et al., 2009; Hématy et al., 2007; Tsang et al., 2011; Wormit et al., 2012; Zhao and Dixon, 2014). Osmotic support (mannitol, poly-ethylene glycol or sorbitol) can suppress most responses induced by cell wall damage (CWD) in a concentration-dependent manner both in yeast and Arabidopsis seedlings, suggesting that they respond similarly to CWI modifications (Engelsdorf et al., 2018; Hamann, 2015; Hamann et al., 2009; Levin, 2011). This is further supported by the observation that ARABIDOPSIS HISTIDINE KINASE1 (AHK1), which is required for osmoperception and abiotic stress responses, partially complements yeast strains deficient in osmosensing (Tran et al., 2007). In addition, MID1-COMPLEMENTING ACTIVITY1 (MCA1) from Arabidopsis partially complements a yeast strain deficient in the plasma membrane-localized Ca2+-channel protein complex MID1-CCH1, which is involved in the yeast CWI maintenance mechanism (Levin, 2011; Nakagawa et al., 2007; Paidhungat and Garrett, 1997). In Arabidopsis, MCA1 mediates Ca2+-based signaling processes, CWI maintenance and both hypo-osmotic stress and mechanoperception (Denness et al., 2011; Furuichi et al., 2016; Yamanaka et al., 2010). Ca2+-based signaling processes are required for CWI maintenance as Ca2+ signaling inhibitors prevent CWD-induced phytohormone [jasmonic acid (JA); salicylic acid (SA)] and lignin production in Arabidopsis seedlings (Denness et al., 2011). In addition to the leucine-rich repeat receptor kinases MALE DISCOVERER 1-INTERACTING RECEPTOR LIKE KINASE 2/LEUCINE-RICH REPEAT KINASE FAMILY PROTEIN INDUCED BY SALT STRESS (MIK2/LRR-KISS) and FEI2, members of the Catharanthus roseus receptor-like kinase1-like family (CrRLK1Ls) have been implicated in CWI maintenance (Engelsdorf et al., 2018; Franck et al., 2018). THESEUS1 (THE1) is required for CWI maintenance and resistance to Fusarium oxysporum f. sp. conglutinans infection (Hématy et al., 2007; Van der Does et al., 2017), and FERONIA (FER) is required for growth regulation during salt stress (Feng et al., 2018). The CWI maintenance mechanism could be conserved as homologs for MCA1 have been identified in Zea mays (NOD) and Oryza sativa (OsMCA1) and for THE1 in Marchantia polymorpha (MpTHE) (Honkanen et al., 2016; Kurusu et al., 2012; Rosa et al., 2017).
Mutations in many cell wall biosynthetic genes cause growth defects, which are possibly caused by hyperactive defense signaling (Bacete et al., 2018). Dwarf phenotypes have also been associated with loss of anisotropic cell expansion due to mutations in cellulose synthase genes (CESAs), encoding proteins that synthesize cellulose during cell wall formation and cell division (Chen et al., 2018; McFarlane et al., 2014). Cellulose biosynthesis inhibitors have become routinely used tools, because they allow CWD to be induced in a tightly controlled manner (temporally and spatially), enabling targeted studies of early and late stress responses by preventing pleiotropic long-term effects (Manfield et al., 2004; Tateno et al., 2016; Wormit et al., 2012). One of the inhibitors (C17) was identified through its ability to inhibit cell division (Hu et al., 2016). Another one, isoxaben (ISX), inhibits specifically cellulose production during primary cell wall formation in elongating and dividing cells and affects subcellular localization of the cellulose synthase complexes within minutes of application (Gutierrez et al., 2009; Heim et al., 1990; Paredez et al., 2006; Watanabe et al., 2018). ISX is frequently used because of the availability of resistance-causing mutations, such as isoxaben resistant 1 (ixr1-1), which represent efficient controls (Scheible et al., 2001). The ixr1-1 mutation resides in CELLULOSE SYNTHASE A3 (CESA3), encoding a subunit of the cellulose synthase complex that is active during primary cell wall formation. ISX-triggered cellulose reduction leads to changes in gene expression, loss of anisotropic cell expansion, lignin deposition, JA and SA accumulation and ultimately cell death (Cano-Delgado et al., 2003; Ellis et al., 2002; Hamann et al., 2009; Engelsdorf et al., 2018). ISX-induced responses, including accumulation of SA, JA, reactive oxygen species (ROS) and lignin, can be suppressed by diphenylene iodonium treatment (Denness et al., 2011). Diphenylene iodonium is an inhibitor of the flavin-containing enzymes implicated in ROS production; the specificity of diphenylene iodonium is concentration dependent and its precise mode of action in plants is unknown (Kärkönen and Kuchitsu, 2015).
Plant cell walls and the cell cycle undergo major changes in response to drought and osmotic stress (Skirycz et al., 2011; Tenhaken, 2014). Whereas the response to osmotic stress is mediated by an ethylene-dependent mechanism, cytokinins are involved in drought stress adaptation both as positive and negative regulators (Li et al., 2016; Nguyen et al., 2016; Nishiyama et al., 2012). Cytokinin application to plants and cell cultures induces expression changes in genes encoding cell wall-modifying proteins (expansins, laccases and pectin-modifying enzymes), suggesting that cytokinins can modify plant cell wall metabolism (Brenner et al., 2012). Although the precise mode of action of cytokinins is unknown, dirigent-like proteins could form a regulatory element (Paniagua et al., 2017). Their expression is regulated both by cytokinins and by different biotic stresses and they mediate stereo-selectivity during lignan biosynthesis. Cytokinins also regulate plant cell cycle progression (Schaller et al., 2014). They modulate transition from G2 to M phase by regulating CDKA/B and CYCB1 activity and from G1 to S phase by controlling CYCD3;1 expression (Scofield et al., 2013). Cytokinin amounts are regulated in different ways, including through changes in nitric oxide (NO) levels (Zürcher and Müller, 2016). NO can interact directly with trans-zeatin (tZ) type cytokinins creating nitrated and nitrosylated cytokinin species, lower both NO and tZ levels or inhibit phosphorylation of cytokinin signaling components (Feng et al., 2013; Liu et al., 2013). Cytokinins also regulate NO levels, which control CYCD3;1 induction, suggesting the existence of a regulatory loop between both molecules (Shen et al., 2013; Tun et al., 2008). Mutations in NITRATE REDUCTASE 1 (NIA1) and NITRATE REDUCTASE 2 (NIA2) that impair nitrate assimilation and NO production are genetic tools regularly used to study NO-based signaling (Chamizo-Ampudia et al., 2017; Tun et al., 2008).
Here, we combined phenotypic and genetic characterizations of CWI and cytokinin signaling components with gene expression analysis and cytokinin measurements to investigate if and how cell cycle activity in Arabidopsis thaliana seedlings is coordinated with CWI. The results suggest that a NIA1/2-dependent, turgor-sensitive process, controlling expression of cytokinin-degrading enzymes, is responsible for changes in certain cytokinins in response to CWD. These changes in cytokinin levels appear to attenuate cell cycle activity by reducing CYCD3;1 expression.
Osmotic treatments attenuate ISX-induced root cell shape changes
To analyze regulatory processes coordinating CWI and cell cycle progression, we determined when and how cellulose biosynthesis inhibition affects Arabidopsis thaliana seedling root growth and root apical meristem (RAM) organization. We used an established model system in which seedlings are grown under controlled conditions in liquid culture and cellulose production is inhibited by ISX (Hamann et al., 2009; Engelsdorf et al., 2018). We confirmed ISX specificity by determining root lengths in mock- or ISX-treated Col-0 and ixr1-1 seedlings (Fig. 1A). Mock-treated (DMSO, ISX solvent) Col-0 and ixr1-1 seedlings exhibited similar root lengths both at the start of (t0) and after 24 h of treatment. ISX inhibited root growth nearly completely in Col-0 seedlings whereas no inhibition was detectable in ixr1-1 seedlings, confirming ISX specificity. Osmoticum treatments (300 mM sorbitol, s300) were included because they suppress ISX-induced responses (Hamann et al., 2009; Engelsdorf et al., 2018). Seedlings were treated with DMSO, ISX, s300 or a combination of ISX and sorbitol (ISXs300) and root lengths were measured (Fig. 1B). Root growth was reduced by s300 whereas ISX treatment caused a nearly complete stop. ISXs300 treatment resulted in an intermediate (non-additive) growth phenotype.
To determine if and when treatment effects are detectable on the meristem level, we measured both RAM length and RAM cell number in seedling roots treated for 3, 6 or 9 h. ISX treatment reduced RAM length significantly after 9 h (Fig. 1C) and cell number after 3 h (Fig. 1D). Although s300 treatments had no consistent effects on RAM length, cell number was reduced after 6 h. ISXs300 treatment resulted in RAM lengths similar to mock whereas RAM cell numbers were similar to ISX. We then measured lengths and widths of individual cells in the transition zone (Fig. 1E). ISX treatments reduced cell length after 3 h. s300 treatment alone had limited effects on cell lengths, whereas ISXs300-treated seedlings exhibited pronounced effects after 6 h. S300 treatments had no detectable effects on transition zone cell width. In ISX-treated roots, cell width was increased from 3 h onwards, whereas cells subjected to ISXs300 co-treatment exhibited intermediate widths from 6 h onwards. These results suggested that ISX treatments inhibit RAM cell division, limit cell elongation and increase cell width in the transition zone. s300 co-treatment primarily reduced the ISX-induced increase in cell width but effects on cell elongation were limited and no impact was detectable on ISX-induced cell division arrest.
Osmotic co-treatments attenuate ISX-triggered inhibition of cell cycle gene expression
To investigate whether ISX and osmoticum treatments affect cell cycle gene expression, we performed time course experiments with seedlings expressing a pCYCB1;1::CYCB1;1 D-box-GUS reporter (Colón-Carmona et al., 1999). β-Glucuronidase (GUS) staining analysis suggested that pCYCB1;1::CYCB1;1 D-box-GUS transcription was reduced in RAMs after 9 h of ISX treatment (Fig. 2A), whereas s300 had milder effects on reporter expression (Fig. 2A). Reporter activity in RAMs exposed to ISXs300 seemed more similar to that in s300-treated than to that in ISX-treated RAMs (Fig. 2A). This effect is not restricted to the RAM as the reporter exhibited similar expression changes also in shoot apical meristems (Fig. 2A). To obtain quantitative data, qRT-PCR-based expression analysis was performed for three genes regulating different transition steps in cell cycle progression. CYCB1;1 regulates the G2-to-M transition, whereas CDKB2;1 is active in the late S to M phase (Okushima et al., 2014; Schaller et al., 2014) and CYCD3;1 is required during G1/S transition (Scofield et al., 2013). The genes examined exhibited dynamic transcript level changes with pronounced differences after 9 h (Fig. 2B). CYCB1;1, CDKB2;1 and CYCD3;1 transcript levels increased over time in mock- and s300-treated seedlings with similar dynamics (Fig. 2B). Transcript levels increased initially also in ISX-treated seedlings. However, after 3h CYCD3;1 transcript level started to decrease whereas CYCB1;1 plateaued and CDKB2;1 kept increasing slightly (Fig. 2B). CYCD3;1 and CYCB1;1 transcript levels were significantly lower in ISX- than in mock-treated seedlings after 9 h, whereas CDKB2;1 transcript levels did not differ between the treatments. In ISXs300-treated seedlings, transcript levels of the three genes were similar to mock controls after 9 h. To investigate whether the transcriptional effects depend on ISX-induced cellulose biosynthesis inhibition, Col-0 and ixr1-1 seedlings were treated for 9 h and CYCB1;1 and CYCD3;1 transcript levels were investigated (Fig. 2C). ISX-treated ixr1-1 seedlings exhibited no transcript changes and the changes caused by s300 were similar to Col-0. These results established that the effects of ISX are caused by cellulose biosynthesis inhibition. The qRT-PCR-based expression analysis performed so far used whole seedlings. To investigate whether the transcriptional effects are tissue specific (i.e. confined to shoot or root apical meristems), experiments were repeated and CYCB1;1 and CYCD3;1 transcript levels were determined in whole seedlings, shoots or roots only (Fig. S1A). Transcript levels of both genes changed in a similar manner in the different tissue types as at the whole seedling level.
We repeated the expression analysis with seedlings grown on vertically oriented plates (Fig. S1B) to investigate whether the liquid culture system influences the effects investigated. ISX, s300 and iSXs300 treatments had the same effects on CYCB1;1 and CYCD3;1 transcript levels in plate-grown seedlings as in seedlings grown in liquid culture. To investigate whether the differences in transcript levels between mock- and ISX-treated seedlings might be caused by differences in number of lateral root primordia, we quantified the number of lateral roots and lateral root primordia in mock- and ISX-treated seedlings after 6 and 12 h of treatment (Fig. S1C). No differences in the number of lateral root primordia formed were detected. These results indicate that particular growth conditions or modifications in growth dynamics are not responsible for the changes observed in the transcript levels.
ISX treatment causes displacement of CESA6 subunits from the plasma membrane (Paredez et al., 2006). We hypothesized that co-treatment with osmoticum might suppress this ISX-induced displacement, thus preventing loss of cellulose synthase activity, and tested this using a CESA6-YFP fusion protein. We determined movement speed and distribution of the fusion protein in root epidermal cells of mock-, ISX- and/or 300 mM mannitol (m300)-treated seedlings (Fig. S1D,E). m300, ISX and ISXm300 treatments all resulted in similar speed changes and displacement of CESA6-YFP fusion protein from the plasma membrane.
Transcript levels for all genes increased over time in mock-treated seedlings, possibly because fresh growth medium was added at the start of the experiments (Riou-Khamlichi et al., 2000). Differences in transcript levels became pronounced after 9 h. In ISX-treated seedlings CDKB2;1 transcript levels did not differ significantly from mock controls, whereas CYCB1;1 were slightly and CYCD3;1 markedly reduced (Fig. 2C). The effects of ISX on CYCB1;1 and CYCD3;1 transcript levels were sensitive to osmotic manipulation to a similar extent as has been reported before for other ISX-induced responses (Hamann et al., 2009; Wormit et al., 2012; Engelsdorf et al., 2018). In all experiments CYCD3;1 transcript levels seemed to be most sensitive to the manipulations.
ISX-induced changes in transcript levels require nitrate reductase activity
THE1 mediates certain ISX-induced CWD responses (Hamann et al., 2009; Hématy et al., 2007; Engelsdorf et al., 2018). Therefore, we investigated whether absence of THE1, or other genes previously implicated in CWI maintenance, affects ISX-induced reduction of CYCD3;1 and CYCB1;1 expression. We characterized knockouts (Kos) for THE1, WALL ASSOCIATED KINASE 2 (WAK2), MIK2, FEI2, and the ion channels MSCS-LIKE 2 (MSL2), MSCS-LIKE3 (MSL3) and MCA1 (Denness et al., 2011; Haswell et al., 2008; Hématy et al., 2007; Kohorn et al., 2006; Van der Does et al., 2017; Xu et al., 2008; Engelsdorf et al., 2018). Because JA signaling has been implicated both in CWI maintenance and regulation of CYCB1;1, we included a KO allele for a key JA biosynthetic enzyme (ALLENE OXIDE SYNTHASE, AOS) (Ellis and Turner, 2001; S´wia˛tek et al., 2004). To investigate a possible involvement of NO-based signaling processes, we used nia1 nia2 seedlings. We quantified CYCB1;1 and CYCD3;1 transcript levels after 9 h in mock-, ISX-, s300- and ISXs300-treated seedlings. Transcript level changes for both genes induced by the different treatments were similar in Col-0 aos, mik2, wak2, fei2, the1 and mca1 seedlings (Fig. S2A-C). In mock-treated msl2 msl3 seedlings, CYCB1;1 and CYCD3;1 transcript levels were elevated, but relative changes in the responses were similar to Col-0 (Fig. S2D) (Wilson et al., 2016). Transcript levels in mock-treated nia1 nia2 seedlings were similar to Col-0, but ISX, s300 and ISXs300 treatments caused no changes (Fig. 3A). To ensure that these effects were not due to indirect consequences on growth caused by changes in nitrogen metabolism in the nia1 nia2 seedlings, we assessed seedling root growth and ISX-induced root growth inhibition in nia1 nia2 seedlings (Fig. 3B). Root length was only slightly reduced in mock-treated nia1 nia2 seedlings compared with Col-0 controls, and ISX treatment inhibited root growth in Col-0 and nia1 nia2 seedlings similarly, suggesting that loss of NIA1 and NIA2 activity impairs neither growth per se nor ISX-induced root growth inhibition.
These results implicated NIA1 and NIA2 in CWD responses and indicated that THE1-mediated responses to ISX might also be nitrate reductase dependent. Therefore, we introduced a THE1 gain-of-function allele (the1-4) into the nia1 nia2 background (Merz et al., 2017). We analyzed JA, SA and lignin accumulation in Col-0, nia1 nia2, the1-4 and nia1 nia2 the1-4 seedlings after ISX treatment. In ISX-treated nia1 nia2 seedlings, SA, JA and lignin production were dramatically reduced compared with ISX-treated Col-0 seedlings (Fig. 3C,D). Whereas ISX-induced responses were enhanced in the1-4 seedlings compared with Col-0, responses were reduced in nia1 nia2 the1-4 seedlings compared with the1-4 (Fig. S3A-C). In nia1 nia2 the1-4 seedlings, root growth was still inhibited by ISX treatment and CYCD3;1 and CYCB1;1 transcript levels did not respond to ISX treatment, similar to nia1 nia2 seedlings.
To understand better the contributions of NIA1 NIA2 to ISX-induced responses, we performed an RNA-seq experiment with Col-0 and nia1 nia2 seedlings mock- or ISX-treated for 1 h. Principal component analysis (PCA) showed that most of the differences result from the genotypes (PC1, 90.2%) whereas treatment had minor effects (PC2, 3.4%) (Fig. 3E). Gene ontology (GO) analysis of the results from mock-treated Col-0 versus nia1 nia2 detected an over-representation of genes implicated in stress responses and primary metabolism (Table S3). Analysis of Col-0 mock versus Col-0 ISX and nia1 nia2 mock versus nia1 nia2 ISX showed pronounced over-representation of defense response genes (Tables S4 and S5). 4852 genes were differentially expressed between mock-treated Col-0 and nia1 nia2 seedlings, confirming that NIA1 and NIA2 contribute to fundamental processes (Fig. 3F). Whereas 149 genes exhibited significant transcript-level changes in ISX-treated Col-0 seedlings versus mock, only 84 genes exhibited such changes in nia1 nia2. Thirty-three genes exhibited significant differences in both ISX-treated Col-0 and nia1 nia2 seedlings compared with mock controls, suggesting that NIA1 and NIA2 are not required for their activation (Table S6). Among the 33 genes, several encode components of Ca2+-signaling elements and ERF or WRKY family transcription factors. To confirm that the transcriptomics-derived expression data are reproducible, we selected three candidate genes (At2g44840, At4g31320, At2g17330) for validation by qRT-PCR-based expression analysis (Fig. S4A,B). Similar to the RNA-seq analysis-derived data, At2g44840 expression was induced strongly in Col-0 seedlings but less so in nia1 nia2. At2g17330 transcript levels were reduced in nia1 nia2 seedlings compared with Col-0 and were not responsive to ISX. At4g31320 transcript levels were reduced in ISX-treated Col-0 but slightly induced in nia1 nia2 seedlings.
The transcriptomic analysis detected fundamental differences between Col-0 and nia1 nia2. However, results from growth assays and targeted cell cycle gene expression analysis of mock-treated seedlings detected no profound differences between nia1 nia2 and Col-0 controls. This suggests that the effects on ISX-induced responses, detected in nia1 nia2 seedlings, are specific. THE1-dependent CWI signaling and JA production were not required for the ISX-induced reduction of CYCB1;1 and CYCD3;1 transcript levels. A NIA1/2-dependent process seems to be required for the ISX-induced changes in CYCB1;1 and CYCD3;1 transcript levels and for phytohormone/lignin accumulation and also contributes to THE1-mediated processes.
tZ application rescues ISX-triggered transcript level changes
Cytokinins regulate cell cycle progression (Schaller et al., 2014). To investigate whether the effects of ISX on transcript levels are cytokinin dependent, Col-0 seedlings were treated for 9 h either with increasing concentrations of tZ, ISX alone or ISX with increasing tZ concentrations (Fig. 4A). tZ treatment alone resulted in similarly increased CYCB1;1 transcript levels regardless of concentration. CYCB1;1 transcript levels in seedlings co-treated with ISX and tZ were in between transcript levels observed in ISX- or tZ-treated ones. Compared with CYCB1;1, CYCD3;1 transcript levels were enhanced in a concentration-dependent manner by tZ. Co-treatments with ISX and tZ led to CYCD3;1 transcript levels similar to those of mock controls, revealing an effect of tZ on the ISX-induced reduction. To determine whether NIA1 and NIA2 are required to mediate the effects of ISX and tZ treatments on transcript levels, we repeated the experiments with nia1 nia2 seedlings (Fig. 4B). In mock-treated nia1 nia2 and Col-0 seedlings, CYCD3;1 and CYCB1;1 transcript levels were similar. tZ treatment at 100 nM or 1 μM induced elevated transcript levels for both genes whereas ISX treatment did not reduce transcript levels in nia1 nia2 seedlings. In seedlings co-treated with ISX and 100 nM or 1 μM tZ, transcript levels of both genes were similar to those in seedlings treated with the corresponding tZ concentrations alone. These results suggested that the effect of ISX on CYCD3;1 and CYCB1;1 transcript levels is dependent on NIA1 NIA2 whereas the tZ effect is not.
Cytokinin homeostasis is changed in ISX- and osmo-treated seedlings
Results from the tZ experiments suggested that cytokinins modulate the effects of ISX on CYCD3;1 and CYCB1;1 transcript levels. To determine whether ISX treatments affect cytokinin homeostasis, we measured total cytokinins and different isoprenoid cytokinins [including tZ, cis-zeatin (cZ) and N6-(Δ2-isopentenyl) adenine (iP)] in Col-0 seedlings mock-, ISX-, s300- or ISXs300-treated for up to 9 h (Fig. 5). Total cytokinin and tZ levels increased over time in mock-treated seedlings possibly owing to increased growth caused by provision of fresh media at the start of the time course (Fig. 5). Results for the cytokinin bases cZ/tZ and iP suggested (mostly cZ-) type-specific changes after 3 h. Levels of total cytokinins and of cZ, tZ and iP types started to show differences after 6 h, which became pronounced after 9 h. ISX-treated seedlings exhibited reduced levels of total cytokinins and tZ and iP types whereas cZ types were enhanced. Sorbitol treatments resulted in intermediate levels (between ISX and mock) with the exception of total iP type, for which levels were similar to mock-treated controls. Intriguingly, in ISXs300-treated seedlings the levels of most cytokinin types were more similar to those of s300-treated seedlings than to ISX-treated seedlings. To investigate whether the changes might be tissue specific, we analyzed cytokinin levels in aerial parts after 9 h of treatment (Fig. S5) and found similar trends to those observed in whole seedlings.
The measurements showed that ISX treatments cause reductions in the levels of the most biologically active cytokinins (tZ and iP types), whereas s300 treatment seemed to have the opposite effect (including the attenuation of ISX treatment effects). The changes in iP-type levels in ISX- and/or s300-treated seedlings appear to be closely correlated with the changes observed in CYCD3;1 transcript levels.
CKX2 and CKX3 expression is induced in an osmo-sensitive and NIA1/2-dependent manner by ISX treatments
The results from gene expression analysis, tZ treatments and cytokinin profiling suggested that ISX- and osmoticum-induced effects on CYCD3;1 and CYCB1;1 transcript levels are mediated by a cytokinin-dependent mechanism. We therefore investigated whether cytokinin signaling is regulating the ISX- and/or osmoticum-induced effects by quantifying CYCD3;1 and CYCB1;1 transcript levels after 9 h in single, double and multiple KO seedlings for cytokinin perception and signaling elements including: ARABIDOPSIS HISTIDINE KINASE 1 (AHK1), AHK2, AHK3, AHK4 (also known as CRE1), ARABIDOPSIS HISTIDINE-CONTAINING PHOSPHOTRANSFER PROTEIN 1 (AHP1), AHP2, AHP3, AHP4, AHP5 and ARABIDOPSIS RESPONSE REGULATOR 1 (ARR1), ARR10, ARR12 (Zürcher and Müller, 2016). CYCB1;1 and CYCD3;1 transcript levels were reduced in mock-treated ahk2/3, ahk3/4, ahp1/2/3/4/5 and arr1/10/12 seedlings (Fig. S6A-D), suggesting that attenuating cytokinin signaling affects basal expression of cell cycle genes. ISX-treated ahk1, ahk2/3, ahk3/4, ahk4, ahp1/2/3/4/5 and arr1/10/12 seedlings showed similarly reduced transcript levels as in Col-0. CYCB1;1 levels in s300- and ISX/s300-treated ahk2/3 seedlings were not elevated as those induced in Col-0 by such treatments, suggesting changed sorbitol sensitivity. To investigate the possibility of genetic redundancy and growth-related defects in AHK-triple KO plants that might affect the responses analyzed, we repeated some of the experiments and included LGR991, an AHK4 inhibitor/AHK3 competitor (Nisler et al., 2010) (Fig. S6C). In LGR991/ISX-treated Col-0, ahk2/3 and ahk3/4 seedlings CYCB1;1 and CYCD3;1 transcript levels were still reduced. LGR991/s300-treated Col-0 seedlings exhibited lower CYCB1;1 transcript levels than s300-treated Col-0. LGR-991 co-treated ahk2/3 and ahk3/4 seedlings still responded to the treatments but CYCB1;1 transcript level changes were less pronounced in ahk2/3. These effects were not detectable for CYCD3;1.
Next, we investigated whether the effects of tZ treatments on CYCB1;1 and CYCD3;1 expression are mediated by the established cytokinin signaling mechanism (Fig. 4). Col-0, ahk2 ahk3 and ahk3 ahk4 seedlings were mock-, ISX-, tZ- or ISX/tZ-treated and CYCB1;1 and CYCD3;1 transcript levels determined (Fig. S6E). ahk2 ahk3 seedlings exhibited lower basal transcript levels than Col-0 seedlings but responded similarly to the treatments. In ISX/tZ-treated ahk3 ahk4 seedlings, CYCD3;1 transcript levels did not recover as they did in Col-0 seedlings, suggesting that AHK3 AHK4 are required for tZ perception.
Because certain cytokinins were reduced in ISX-treated seedlings, we investigated whether active degradation is responsible. Knockout of individual cytokinin oxidases (CKXs) seems to have no phenotypic effect due to redundancy (Bartrina et al., 2011). Therefore, we used overexpression lines for CYTOKININ OXIDASE 2 (oeCKX2, localized in the endoplasmic reticulum and apoplast) and CYTOKININ OXIDASE 3 (oeCKX3, residing in the vacuole) because they enabled us to test whether manipulating cytokinin levels affects CYCB1;1 and CYCD3;1 transcript levels and to determine whether spatial requirements may exist (Werner et al., 2003). We analyzed CYCB1;1 and CYCD3;1 transcript levels in oeCKX2 and 3 seedlings. In oeCKX2 seedlings, CYCB1;1 transcript levels were lower in mock- and s300-treated seedlings than in Col-0 controls, whereas mock- and s300-treated oeCKX3 seedlings were similar to Col-0 (Fig. 6A). In ISX-treated Col-0 seedlings, transcript levels were reduced whereas reductions in oeCKX seedlings were less pronounced. In ISXs300-treated Col-0 and oeCKX3 seedlings, CYCB1;1 transcript levels were similar to those of mock-treated seedlings, whereas in oeCKX2 they were similar to those observed in ISX-treated ones. CYCD3;1 transcript levels were reduced in mock-treated oeCKX2 seedlings compared with Col-0. ISX-treated Col-0 seedlings exhibited reductions in CYCD3;1 transcript levels, which were not observed in oeCKX2 (owing to the reduction in basal expression) and less pronounced in oeCKX3 seedlings. In s300- or ISX/s300-treated Col-0 seedlings, CYCD3;1 levels were enhanced. In oeCKX2 and oeCKX3 seedlings the effects of s300 and ISX/s300 were similar to those observed in Col-0.
The impaired responsiveness of CYCD3;1 and CYCB1;1 in oeCKX lines suggested a possible role of cytokinin homeostasis/degradation in the CWD response. We performed timecourse experiments and quantified CKX2 and CKX3 transcript levels in Col-0 seedlings to investigate whether these effects are regulated at the transcriptional level. CKX2 transcript levels were significantly induced in ISX-treated seedlings after 9 h (Fig. 6B). s300- and ISXs300-treated seedlings exhibited CKX2 transcript levels similar to mock-treated controls. CKX3 transcript levels were increased after 6 h of ISX treatment, reduced in s300-treated seedlings and similar to mock controls in ISXs300-treated seedlings. To investigate whether ISX-induced changes in CKX2 and CKX3 transcript levels are dependent on NIA1 and NIA2, we performed timecourse experiments and investigated CKX2 and CKX3 transcript levels in mock- or ISX-treated Col-0 and nia1 nia2 seedlings. Whereas CKX2 and CKX3 transcript levels increased over time in ISX-treated Col-0 seedlings, such increases were absent in nia1 nia2 (Fig. 6C).
Genetic manipulation of cytokinin perception and signal translation affected osmoticum-induced changes in CYCB1;1 transcript levels but did not affect the ISX-induced changes in CYCB1;1 and CYCD3;1 transcript levels. ISX-induced CKX2 and CKX3 expression was osmosensitive and dependent on NIA1 and NIA2 activity. Manipulating CKX2 expression affected CYCB1;1 and CYCD3;1 basal transcript levels and responses to ISX and s300 treatments. These results suggest that ISX and osmoticum treatments modify CKX2 and CKX3 expression, which in turn seems to affect CYCD3;1 transcript levels preferentially.
Plant cell wall metabolism and cell cycle activity are essential components of development, immunity and abiotic stress (Zhao and Dixon, 2014; Polyn et al., 2015; Zhao et al., 2017; Bacete et al., 2018). Here, we determined how CWD (impairing CWI) affects cell cycle progression and investigated the regulatory processes.
Phenotypic analysis of seedling root growth indicated that ISX-induced growth inhibition is attenuated by osmoticum (Fig. 1B). Analysis of cell number and length showed that the osmoticum effect is restricted to attenuation of RAM length, whereas RAM cell number (cell division) was still arrested (Fig. 1C,D). Cellulose is an essential element for the completion of the cross-walls separating both daughter cells after mitosis; therefore, complete inhibition of cell division is expected and is similar to effects observed in cellulose biosynthesis mutants (Chen et al., 2018; Miart et al., 2014; Nicol and Höfte, 1998). More detailed phenotypic analysis showed that ISX limits elongation in the transition zone and increases cell width in an osmosensitive manner similar to previous studies (Fig. 1E) (Cosgrove, 2018). Cellulose biosynthesis inhibition during cell elongation results in isodiametric cell expansion caused by the interplay of turgor pressure and weakened cell walls. Here, turgor-driven isodiametric cell expansion was apparently attenuated by addition of osmoticum. Co-treatment with osmoticum seemed to suppress primarily the ISX-induced increase of transition zone cell width but not the effects on cell division or on cell elongation, suggesting that cell shape distortion could be responsible for activation of CWD responses.
Analysis of ISX-treated Arabidopsis seedlings expressing the pCYCB1;1::CYCB1;1 D-box-GUS reporter suggested that CYCB1;1 transcript levels are reduced by ISX and attenuated by co-treatment with osmoticum in both shoot and root apical meristems (Fig. 2A; Fig. S1). qRT-PCR-based expression analysis showed that CDKB2;1 transcript levels did not change, suggesting that CDKB2;1 expression is not responsive to CWD. CYCB1;1 and CYCD3;1 transcript levels were reduced by ISX treatment and similar to mock levels if co-treated with osmoticum. Based on the results of the phenotypic analysis, changes in transcript levels seem to occur in response to a change in cell shape and not because of cell division or elongation arrest. Similar effects of osmoticum and CWD on cell cycle and survival have been reported in yeast and other plant species (Iraki et al., 1989; Levin and Bartlett-Heubusch, 1992).
We performed a genetic analysis (including AOS, MIK2, WAK2, FEI2, THE1, MCA1, MSL2 MSL3) and found that genes required for ISX-induced JA/SA or lignin production are not necessary for changes in CYCD3;1 and CYCB1;1 transcript levels (Fig. S2). These results suggest a second pathway exists, regulating the effects on cell cycle gene expression. However, both pathways respond to an osmosensitive stimulus, indicated by the similar effects of osmoticum on ISX-induced JA, SA, lignin production and CYCD3;1 and CYCB1;1 transcript levels (Engelsdorf et al., 2018). As part of the genetic analysis we also examined nia1 nia2 seedlings. Differences on the transcriptome level between Col-0 and nia1 nia2 seedlings suggested that fundamental metabolic differences exist, in accordance with involvement of NIA1 and NIA2 in nitrogen metabolism (Fig. 3E,F) (Chamizo-Ampudia et al., 2017). However, CYCB1;1 and CYCD3;1 transcript levels and seedling root lengths were similar in mock-treated Col-0 and nia1 nia2 seedlings, suggesting that global metabolic differences have a limited impact on the processes examined here (Fig. 3A,B). In these seedlings, neither ISX-derived reduction of CYCB1;1 and CYCD3;1 transcript levels nor increased JA/SA/lignin levels were detectable (Fig. 3A,C,D). Simultaneously, results from the analysis of nia1 nia2 the1-4 seedlings showed that NIA1 and NIA2 are required to mediate the THE1-dependent responses, implying that NIA1/2-dependent processes are involved in both THE1-dependent and -independent processes (Fig. 3C,D, Fig. S3).
nia1 nia2 seedlings have been used as genetic tool to study NO-based signaling processes and were identified as important regulators of abiotic stress responses (Vicente et al., 2017; Xie et al., 2013). NO has also been suggested to regulate and be regulated by cytokinins, suggesting the existence of a regulatory feedback loop between the two (Tun et al., 2008; Feng et al., 2013; Liu et al., 2013; Shen et al., 2013). In Col-0 seedlings, treated jointly with increasing concentrations of tZ and ISX, ISX-induced reductions in CYCD3;1 transcript levels were alleviated in a concentration-dependent manner by tZ (Fig. 4A). In nia1 nia2 seedlings, ISX-induced reductions in transcript levels were absent whereas tZ treatment effects were still detectable, implying that the tZ-derived effects do not require NIA1 and NIA2 (Fig. 4B). This was supported by results obtained with ahk3 ahk4 seedlings (Fig. S6E). The results suggested that the effects of ISX on CYCD3;1 transcript levels and activity might be mediated through changes in cytokinin levels, similar to previously reported regulation of CYCD3;1 by cytokinins, but may not require histidine kinase-based signaling (Scofield et al., 2013). It remains to be determined whether auxin may also play a role here as previous work has implicated NO-based signaling in stem cell homeostasis through auxin (Sanz et al., 2014).
Cytokinin measurements in ISX-, osmoticum- or ISX/osmoticum-treated Col-0 seedlings showed that individual treatments had opposite effects on certain cytokinins whereas the combined treatments resulted in levels similar to mock controls (Fig. 5). Intriguingly, iP-type level changes were particularly similar to CYCD3;1 transcript level changes, suggesting they may be responsible for changes in transcript levels. Experiments with maize exposed to osmotic stress showed that osmotic stress affects cytokinin levels, but the experimental conditions differed, making comparisons difficult (Vyroubalová et al., 2009). We assessed whether cytokinin signaling is required for the effects of ISX and osmoticum treatments on CYCD3;1 and CYCB1;1 transcript levels. In several genotypes (ahk2 ahk3, ahk3 ahk4, arr1 arr10 arr12) we observed lower CYCD3;1 transcript levels in mock-treated seedlings, corroborating cytokinin function in regulating CYCD3;1 activity (Scofield et al., 2013) (Fig. S6A-D). ahk2/3 seedlings exhibited transcript level changes upon treatment with osmoticum only, implicating AHK2 and AHK3 in osmoperception, which is in line with their previously described role in drought stress perception (Tran et al., 2007). In all the genotypes examined (also in combination with LGR-991), ISX-induced reductions in transcript levels were still detectable, suggesting that intact cytokinin signaling is not required. However, expression of two genes encoding cytokinin-degrading enzymes (CKX2, CKX3) was induced by ISX treatment in a NIA1/2-dependent and osmosensitive manner. Characterization of CYCD3;1 and CYCB1;1 transcript levels in ISX- and/or osmoticum-treated oeCKX2 and oeCKX3 seedlings suggested that CKX2 and CKX3 overexpression affects ISX- and osmoticum-induced changes (Fig. 6A-C). Bearing in mind that CKX enzymes cleave the isoprenoid side chain of N6-(Δ2-isopentenyl) adenine (iP), trans- and cis-zeatin (tZ, cZ) and their ribosides (iPR, tZR and cZR), leading to adenine/adenosine and the corresponding aldehyde, both CKX2 and CKX3 could be responsible for the observed changes in cytokinin levels (Kopečný et al., 2016). The results suggest that ISX-induced CWD causes increased expression of CKX2 and CKX3 through NIA1/2-dependent processes. Increased expression of CKX2 and CKX3 could lead to reductions in certain cytokinins, which in turn reduces transcript levels of CYCD3;1 and cell cycle progression.
We integrated our results with the available information on CWI maintenance signaling into a model summarizing how plant cells respond to ISX-derived CWD (Fig. 7). CWD (generated by ISX or cell wall-degrading enzymes) activates at least three different signaling cascades. One is THE1-dependent and responsible for induction of JA/SA/lignin production. A second one activates PROPEP1 and PROPEP3 expression, leading to accumulation of PROPEP3 in ISX-treated seedlings (Engelsdorf et al., 2018). These genes encode precursors for signaling peptides enhancing immune responses and attenuating THE1-mediated signaling via their respective receptors PEPR1 and PEPR2 (Bartels and Boller, 2015; Engelsdorf et al., 2018). The third one, also THE1 independent, is responsible for the effects on CYCD3;1 and CYCB1;1 transcript levels. NIA1/2-dependent processes are required for both THE1-dependent and (THE1-independent) cell cycle responses, suggesting they form a hub for CWI-related signaling processes. These proteins seem responsible for controlling changes in CKX2 and CKX3 transcript levels and thus the cytokinin degradation rate. This is supported by the changes observed in iP-type cytokinin levels, which are inversely related to CKX transcript levels. Interestingly, reductions and increases in iP-type cytokinins are closely correlated with changes in CYCD3;1 expression. CYCD3;1 was identified as a cell cycle regulator, which is active at the G1/S stage transition (Scofield et al., 2013). Therefore, CYCD3;1 could act as integrator allowing progression through G1/S stage only if CWI is intact. This implies the existence of a cell cycle checkpoint in plants, similar to the one described in yeast, dedicated to coordination of CWI and cell cycle activity (Kono et al., 2016).
Our results show that osmoticum reduced many of the ISX-induced responses. Such effects of turgor on cell cycle activity have been observed earlier in plants and yeasts, and have been discussed as important regulators of cell cycle progression (Iraki et al., 1989; Levin and Bartlett-Heubusch, 1992; Sablowski, 2016). We speculate that the osmoticum effects occur because CWD involves plasma membrane displacement versus the cell wall, changes in surface tension of the cell wall itself or simply changes in the optimal state of the plasma membrane (stretching/shrinking). These changes could be detected by sensor proteins similar to the mechanisms in place in yeast or animal cells (Austen et al., 2015; Hamant and Haswell, 2017; Heinisch et al., 2010). Elucidation of such sensors will provide the mechanistic insights necessary to understand how CWI is monitored in plants and how downstream responses are activated.
MATERIALS AND METHODS
Plant growth and treatment
Arabidopsis thaliana genotypes used are listed in Table S1. Unless stated otherwise, all results are based on three independent experiments with at least three biological replicates each; all values shown are mean±s.d. Seedlings were sterilized and grown in liquid culture with minor modifications (Hamann et al., 2009). Thirty milligrams of seeds were sterilized by sequential incubation with 70% ethanol and 50% bleach on a rotating mixer for 10 min each and washed three times with sterile water. All experiments were performed in 6-day-old seedlings treated with 600 nM isoxaben (ISX, Sigma-Aldrich; in DMSO), 300 mM sorbitol (s300, Sigma-Aldrich) or a combination of both, in freshly prepared medium. Zeatin (1 mM stock solution) (Sigma-Aldrich) was dissolved in water. Cytokinin competitor LGR-991 was dissolved in DMSO and used at a final concentration of 20 μM. Seedlings were grown in long-day conditions (16 h light, 22°C/8 h dark, 18°C) at 150 μmol m−2 s−1 photon flux density on a IKA KS501 flask shaker at a constant speed of 130 rotations per minute.
GUS staining was performed using staining buffer containing 50 mm sodium phosphate (pH 7.0), 0.5 mm potassium ferrocyanide, 10 mm EDTA, 0.1% (v/v) Triton X-100, 2% (v/v) dimethyl sulfoxide and 1 mm 5-bromo-4-chloro-3-indolyl-d-glucuronide at 37°C for 24 h (Jefferson et al., 1987; Savatin et al., 2014). Roots and shoots imaged using a Zeiss Axio Zoom-V16 stereomicroscope.
Analysis of root growth phenotypes
Root length measurements were performed right before (0 h) and 24 h after treatments (DMSO/mock, ISX, s300 or ISXs300). Absolute RAM length was determined in primary roots treated for 3, 6 and 9 h. Root tips were stained for 5 min with propidium iodide (1 mg/ml), followed by two washing steps with MilliQ water and imaging using a LSM Zeiss 710 or a Leica SP8. RAM length [defined as the distance between the quiescent center (QC) and the first elongating cell in the root cortical cell layer], RAM size (expressed as number of cortical cells in the RAM), individual cell length and width were measured using Fiji (https://fiji.sc).
Quantification of cytokinin metabolites was performed according to a method described previously (Svačinová et al., 2012) and slightly modified as described before (Antoniadi et al., 2015). Samples were homogenized and extracted in 1 ml of modified Bieleski buffer (60% methanol, 10% HCOOH and 30% H2O) together with a cocktail of stable isotope-labeled internal standards (0.25 pmol of cytokinin bases, ribosides, N-glucosides and 0.5 pmol of cytokinin O-glucosides per sample). Extracts were purified with Oasis MCX column (30 mg/1 ml, Waters) conditioned with 1 ml each of 100% methanol and H2O, equilibrated sequentially with 1 ml of 50% (v/v) nitric acid, 1 ml of H2O and 1 ml of 1 M HCOOH. After sample application onto an MCX column, un-retained compounds were removed by a wash step using 1 ml of 1 M HCOOH and 1 ml 100% methanol, pre-concentrated analytes were eluted by two-step elution using 1 ml of 0.35 M NH4OH aqueous solution and 2 ml of 0.35 M NH4OH in 60% (v/v) methanol solution. Eluates were evaporated to dryness in vacuo and stored at −20°C. Cytokinin levels were determined using ultra-high performance liquid chromatography-electrospray tandem mass spectrometry (UHPLC-MS/MS) (Rittenberg and Foster, 1940).
Salicylic acid and jasmonic acid quantification
Lignification was investigated 12 h after start of treatments in root tips. Lignin staining, detection and quantification were performed as described by Engelsdorf et al. (2018).
Quantification of plasma membrane-localized YFP-CesA6 particles
Live cell imaging was performed as described by Ivakov et al. (2017). Col-0 seedlings were grown in flasks containing liquid MS (×0.5), and placed on an orbital shaker (80 rpm; 1.45 g). Six-day-old seedlings were treated with mock (DMSO), 600 nM isoxaben (ISX), 300 mM mannitol (m300) or a combination of isoxaben and mannitol (ISXm300), for 1 h in freshly prepared medium. Seedlings were mounted on a sample holder such that they were sandwiched in between a 1-mm-thick gel pad (1% agarose) and a cover slip. Root epidermal cells in the elongation zone were imaged with a Nikon Ti-E inverted confocal microscope using a CFI APO TIRF 3100 N.A. 1.49 oil immersion objective and evolve CCD camera (Photometrics Technology). Image acquisition was performed with Metamorph software (version 7.5). For plasma membrane CesA6 particle density measurements, an image was acquired at the cell surface, followed by an image at 0.6 µm below the surface in order to image the intracellular CesA6-containing compartments including Golgi and smaCCS/MASCs (small CESA-containing compartments/microtubule-associated CESA compartments). The ‘find maxima tool’ of ImageJ was used for particle detection within a defined area, using an identical noise threshold.
Total RNA was isolated using a Spectrum Plant Total RNA Kit (Sigma-Aldrich). Two micrograms of total RNA were treated with RQ1 RNase-Free DNase (Promega) and cDNA synthesis was performed with the ImProm-II Reverse Transcription System (Promega). qRT-PCR was performed using a LightCycler 480 SYBR Green I Master (Roche) and primers (Table S2) diluted according to manufacturer's specifications. Expression levels of each gene are shown relative to the reference gene ACT2 and determined using a modified version of the Pfaffl method (Pfaffl et al., 2002) as previously described (Ferrari et al., 2006). LinRegPCR software was used to obtain Cq values and individual PCR efficiency.
Total RNA was extracted using a Spectrum Plant Total RNA Kit (Sigma-Aldrich). RNA integrity was assessed using an Agilent RNA 6000 Pico Kit. Sequencing samples were prepared as described previously (Ren et al., 2015). Single-read sequencing was performed for 75 cycles on a NextSeq500 instrument (Illumina RTA v2.4.6). FASTQ files were generated using bcl2fastq2 Conversion Software v1.8.4. Each FASTQ file was subjected to quality control through fastQC v11.1 before technical replicates were combined and an average of 13.1 million reads was produced for each library. Reads were aligned to the A. thaliana genome (Ensembl v82) with STAR v2.4.1 in two-pass mode. On average, 96.2% of the reads aligned to the genome. Reads that aligned uniquely to the genome were aggregated into gene counts with Feature Counts v1.4.6 using the genome annotations defined in Ensembl v82. Differentially expressed genes were extracted from Col-0 and nia1 nia2 genotypes after ISX/mock treatments using DESeq2 differential expression analysis pipeline (Love et al., 2014) in R. For both genotypes, genes showing differential expression in response to ISX treatment versus mock with a 1% false discovery rate (FDR) were considered significantly differentially expressed (DE). In order to assess the variance within the samples, a principal component analysis (PCA) was performed. Regularized logarithm transformation from DESeq2 package was applied to read counts and PCA was performed for 2000 genes with highest variance using a built-in R function ‘prcomp’.
The intersection of the DE genes between the genotypes was visualized using a Venn diagram produced with Venny 2.1 (http://bioinfogp.cnb.csic.es/tools/venny/index.html). Analysis of GO categories was performed using AtCOECiS (Vandepoele et al., 2009).
Statistical significance (Student's t-test, ANOVA) was assessed using IBM SPSS Statistics v24. Statistically significant differences are indicated with different letters for the one-way ANOVA/Tukey's HSD test at α=0.05. Pairwise comparisons, and statistically significant differences according to Student's t-test are represented with *P<0.05, **P<0.01, ***P<0.001.
The authors thank Hana Martínková and Ivan Petřík for their help with cytokinin analyses and John Mansfield for comments on the manuscript.
Conceptualization: N.G.-B., T.E., L.V., S.P., T.H.; Methodology: N.G.-B., T.E., L.A., T.H.; Validation: N.G.-B., T.E.; Formal analysis: T.E., M.S., L.V., T.H.; Investigation: N.G.-B., T.E., M.S., L.V., G.A.K., A.Y., L.A., J.H.; Resources: M.S., L.A., O.N., J.H.; Data curation: L.V.; Writing - original draft: T.H.; Writing - review & editing: N.G.-B., T.E., M.S., L.V., G.A.K., A.Y., O.N., S.P., J.H., T.H.; Supervision: T.H.; Project administration: T.H.; Funding acquisition: J.H., T.H.
This work was supported by a Marie Curie Skłodowska fellowship ‘SugarOsmoSignaling’ (to T.E.), EEA Grants (7F14155 CYTOWALL to N.G.-B./T.H./J.H.), Ministry of Education, Youth and Sports of the Czech Republic (Ministerstvo Školství, Mládeže a Tělovýchovy) (LQ1601 CEITEC 2020 and LM2015062 Czech-BioImaging) and a European Regional Development Fund Project ‘Centre for Experimental Plant Biology’ (CZ.02.1.01/0.0/0.0/16_019/0000738).
Transcriptomics data have been deposited in Gene Expression Omnibus under accession number GSE109613.
The authors declare no competing or financial interests.