N6-adenosine methylation of mRNA integrates multilevel auxin response and ground tissue development in Arabidopsis Free

Methylation at the N6 position on adenosine (m⁶A) is the most common covalent modification of mRNA. m⁶A is required for numerous plant morphogenetic processes, including gametophyte and embryo development (Zhong et al., 2008; Zhang et al., 2019; Cheng et al., 2022), shoot stem cell fate, vascular formation and directional root growth (Shen et al., 2016; Ruzicka et al., 2017; Arribas-Hernández et al., 2020; Gao et al., 2021), among many others (Arribas-Hernández and Brodersen, 2020; Xu et al., 2022; Shen et al., 2023; Tang et al., 2023). The dynamic status of m⁶A marks is handled by three groups of proteins. m⁶A is introduced to mRNA by the so-called writers.

The Arabidopsis thaliana writer complex contains the proteins MTA, MTB (METHYLTRANSFERASE A and B), FIP37 (FKBP12 INTERACTING PROTEIN 37), VIR (VIRILIZER) and HAKAI (Zhong et al., 2008; Ruzicka et al., 2017), which act together with other factors (Parker et al., 2021; Zhang et al., 2022a). m⁶A can be removed by the erasers, demethylases ALKBH9 or ALKBH10 (ALPHA-KETOGLUTARATE-DEPENDENT DIOXYGENASE HOMOLOG) (Duan et al., 2017; Martínez-Pérez et al., 2017; Shoaib et al., 2021). The presence of m⁶A can be recognized by the readers, as reported for the YTH-domain proteins ECT2, ECT3, ECT4 (EVOLUTIONARILY CONSERVED C-TERMINAL REGION 2, 3 and 4) and CPSF30 (CLEAVAGE AND POLYADENYLATION SPECIFICITY FACTOR 30) (Arribas-Hernández et al., 2018, 2021; Scutenaire et al., 2018; Pontier et al., 2019; Song et al., 2021).

The plant hormone auxin coordinates developmental decisions during plant growth, including meristem activities, gravitropic and phototropic responses, lateral root, apical hook and vascular formation, and maintaining apical dominance. The involvement of m⁶A in auxin-mediated processes has been proposed in the theoretical model (Omelyanchuk et al., 2017), in connection with microRNA biogenesis (Bhat et al., 2020), or as a result of a comprehensive phenotypic analysis of m⁶A reader mutants (Arribas-Hernández et al., 2020). The presence of m⁶A on the YUCCA3 transcript, encoding a protein essential for auxin biosynthesis, has been demonstrated as required for rice male gametophyte development (Cheng et al., 2022). The connection between m⁶A and auxin has also been investigated in lateral root development (Ruzicka et al., 2017; Zhang et al., 2022a). Nevertheless, no coherent study thoroughly analyzing the auxin-dependent processes related to m⁶A in the model plant Arabidopsis thaliana has been presented so far.

The Arabidopsis root apical meristem (RAM) is organized around the quiescent center (QC), a group of sparsely dividing cells. QC is surrounded by mitotically active stem cells called initials (i.e. a stem cell niche), which divide asymmetrically to produce the structured patterns of cell files that define the architecture of the growing root (Pardal and Heidstra, 2021). The molecular mechanisms of the earliest stages following asymmetric cell division have been described in detail for ground tissue patterning. The asymmetric divisions are predominantly controlled by a well-characterized core regulatory module consisting of the transcription factors SHORTROOT (SHR) and SCARECROW (SCR), and the D-type cyclin 6 (CYCD6;1). The SHR protein is produced in the stele and migrates to the adjacent outer cell layer, where it activates the expression of SCR (Helariutta et al., 2000; Gallagher et al., 2004). Together, SCR and SHR trigger, with the contribution of auxin, expression of CYCD6;1 in the daughter cells of ground tissue initials, leading to their periclinal division and the eventual formation of endodermis and cortex (Sozzani et al., 2010; Cruz-Ramírez et al., 2012; Koizumi et al., 2012).

Here, we reveal that Arabidopsis hypomorphic writer mutants exhibit auxin-related defects among their pleiotropic phenotypes. They also display remarkable resistance to exogenously applied auxin and aberrant formation of endodermis in the primary root. We examine the molecular connections between m⁶A and auxin-dependent pathways and propose the candidate target transcripts associated with the described defects. Finally, we link the observed ground tissue defects with a perturbed auxin response and highlight the multifaceted impact of m⁶A on auxin-dependent processes.

The m⁶A writer mutants show a variety of phenotypes, including defective root vascular formation, apical dominance, and gravity response (Ruzicka et al., 2017). These phenotypes are reminiscent of auxin-related defects (Bishopp et al., 2011; Su et al., 2017; Weijers et al., 2021). Therefore, we examined further morphogenetic processes commonly associated with deficient auxin response, utilizing virilizer-1 (vir-1), a hypomorphic writer mutant that displays strongly reduced m⁶A levels. Although the hypocotyl bending towards the light source (Harper et al., 2000; Friml et al., 2002; Ding et al., 2011) was not notably altered in the vir-1 mutants (Fig. S1A), they showed defects in other processes ascribed to auxin, such as the maintenance of root meristems (Dello Ioio et al., 2008; Ruzicka et al., 2009), apical hook development (Vandenbussche et al., 2010; Zadnikova et al., 2010) and cotyledon venation patterns (Przemeck et al., 1996; Mattsson et al., 1999) (Fig. 1A-C). Altogether, it appears that defective activity of the m⁶A writers can also affect auxin-dependent processes.

We further investigated the resistance of the vir-1 mutants to exogenously applied auxins. First, we chose 2,4-dichlorophenoxyacetic acid (2,4-D), a stable and non-metabolized synthetic auxin. In a root elongation assay, vir-1 showed a remarkable resistance to this compound (Fig. 1D,E). In addition, vir-1 displayed reduced sensitivity to the native auxin indole-3-acetic acid (IAA) and to other compounds that interfere with auxin-dependent processes, such as the transportable synthetic auxin 1-napthtoxyacetic acid (NAA), and an inhibitor of auxin transport, 1-N-phenoxyacetic acid (NPA) (Fig. S1B-D). Hence, vir-1 is resistant to exogenous auxin treatment, regardless of the auxin-related compound applied.

Mutants deficient in the core subunits of the m⁶A writer complex have previously been shown to display highly similar, possibly interchangeable, defects (Ruzicka et al., 2017). Therefore, we tested their auxin sensitivity, including the mta ABI3::MTA (Bodi et al., 2012) and fip37-4 hypomorphs, inducible MTB RNAi lines, and hakai loss-of-function mutants (Ruzicka et al., 2017) (Fig. 1D-H, Fig. S2A,B). Apart from the hakai lines, which show milder phenotypes than other lines (Ruzicka et al., 2017), all genotypes consistently showed resistance to 2,4-D. Taken together, these data reveal that severe depletion of m⁶A levels leads to auxin-related defects and reduced sensitivity to exogenously applied auxins.

Next, we tested the auxin responsiveness of the mutants impaired in the m⁶A readers (Arribas-Hernández et al., 2018; Scutenaire et al., 2018) and erasers (Duan et al., 2017; Martínez-Pérez et al., 2017). The triple m⁶A reader mutant ect234 (Arribas-Hernández et al., 2020) displayed moderate resistance to the auxin treatment (Fig. 1I, Fig. S2C). We also tested mutants affected in genes encoding the eraser proteins (alkbh9b, alkbh10a and alkbh10b, or the alkb10a alkbh10b double mutant). Although their phenotypes are rather subtle (Duan et al., 2017; Martínez-Pérez et al., 2017), they exhibit hypersensitivity to abscisic acid (Shoaib et al., 2021; Amara et al., 2022; Tang et al., 2022). By contrast, we observed overall normal sensitivity to auxin in these lines (Fig. 1I). Thus, the increased presence of m⁶A marks does not seem to potentiate the sensitivity to auxin. However, the auxin response mediated by m⁶A is likely associated with the prevailing activity of the ECT2 clade of m⁶A readers.

Numerous mutants deficient at various stages of RNA processing exhibit auxin-related defects or resistance to auxin treatment (Retzer et al., 2014; Hrtyan et al., 2015; Weijers and Wagner, 2016; Soprano et al., 2017). Following a literature search, we gathered a collection of prominent RNA-processing mutants displaying defective auxin response (Table S1 and references therein), and analyzed their sensitivity to 2,4-D. We were largely able to recapitulate their altered response to auxin treatment described in the literature in our growth conditions. Notably, the m⁶A writer mutants displayed a more pronounced resistance to auxin compared with most of the lines (Fig. 2A). Auxin response is required for xylem development in the root (Bishopp et al., 2011). Thus, we also tested whether some of the lines defective in RNA processing (Table S1) display protoxylem defects, as do the m⁶A writer hypomorphs (Ruzicka et al., 2017). Indeed, these lines showed aberrant protoxylem formation (Fig. 2B), but only with a limited correlation with their auxin resistance (Fig. 2A). These findings thereby further underline the highly relevant role of m⁶A in auxin-dependent processes; however, the auxin response mediated by RNA processing has a limited correlation with the primary root protoxylem formation.

The defective auxin response of m⁶A writer mutants can be, in principle, explained by changes at one or more of four functional tiers: (1) by expression changes of some of the components required for mRNA methylation following auxin treatment (Pu et al., 2019), or by defects in (2) the auxin metabolome (Cheng et al., 2022), (3) transport or (4) signaling in the respective lines. (1) It was recently proposed that levels of the VIR protein rapidly drop following auxin treatment in quantitative proteomics experiments (Pu et al., 2019). We analyzed the root tip fluorescence of the GFP-VIR transgene in lines treated with NAA, but we detected no expression changes in the given time window (Fig. S3A), nor at any specific tissue within the primary root tip (Fig. S3B). (2) We quantified endogenous levels of IAA and its main precursors and metabolites in the roots of vir-1 and fip37-4. We generally observed very subtle changes in the respective compound levels in the lines examined. Only the amounts of the early IAA precursor anthranilate, IAA itself, and its subsequent breakdown products were usually slightly elevated in vir-1 and fip37-4 (Fig. 3A). However, the strong phenotypes described above are unlikely to be caused by such subtle and relatively inconsistent metabolic changes (Hayashi et al., 2021; Casanova-Sáez et al., 2022). (3) We also tested the ability of vir-1 to transport radioactively labeled NAA (Lewis and Muday, 2009). As the fip37-4 mutants are severely stunted (Ruzicka et al., 2017), we used the ABI3::MTA hypomorphs (Bodi et al., 2012) here instead. We found that the auxin transport rates were not markedly altered in these lines (Fig. S3C). (4) Finally, to examine whether auxin signaling could be affected, we crossed vir-1 and fip37-4 with the transcriptionally controlled auxin (signaling) response reporter DR5::GFP (Ulmasov et al., 1997; Benková et al., 2003). We observed that GFP fluorescence significantly decreased in both mutant lines (Fig. 3B-D). These data collectively suggest that auxin signaling is likely the most affected tier of the auxin response in m⁶A writer mutants.

Numerous experimental transcriptome-wide procedures have been employed to identify the mRNAs bearing m⁶A in plants. As the technical limits of the various approaches have been repeatedly discussed (Linder et al., 2015; Garcia-Campos et al., 2019; Parker et al., 2020; Zhao et al., 2022), we compiled prominent, publicly available datasets, regardless of the m⁶A-detection methodology used (Wan et al., 2015; Shen et al., 2016; Anderson et al., 2018; Bhat et al., 2020; Parker et al., 2020; Zhang et al., 2022a) (Tables S2, S3). Adjusting comparable thresholds, we found that the diverse gene lists show a relatively high overlap (Fig. 4A). To determine the candidate mRNAs involved in the m⁶A-mediated auxin response (Table S4), we chose transcripts present in at least two out of six independent datasets (Table S5). Next, we sorted those directly participating in auxin-dependent processes according to their respective functional categories (De Smet et al., 2011), using their genome-wide proportions as a reference. We observed that many mRNAs associated with auxin synthesis, metabolism and transport carry m⁶A marks, but they do not seem to show any remarkable or consistent prevalence, in contrast to those involved in auxin signaling (Fig. 4B, Tables S4, S5). To test the robustness of our approach, we analyzed additional combinations of datasets, in which some of the used gene lists were replaced by others (Hu et al., 2021; Wang et al., 2023; Tables S2, S5). We obtained a similar proportion in the respective functional categories, regardless of whether the datasets of seedling (Fig. S4A) or diverse tissue (Fig. S4B, Tables S2, S5) origin were used. In addition, in all dataset combinations, we consistently noticed a striking absence of m⁶A in the SAUR (SMALL AUXIN UP RNA) transcripts (Ren and Gray, 2015; Fig. 4B, Fig. S4, Tables S4, S5).

Among the transcripts encoding the auxin signaling components, we found three out of six mRNAs encoding the auxin TIR1 and AFB receptors and ten out of the 23 auxin response factors (ARFs) carried m⁶A marks under given selection criteria. These included, in particular, the central developmental regulators, both class A activators (ARF6, 7, 8, 19) and class B repressors (ARF2, 3, 4) (Guilfoyle and Hagen, 2007; Weijers and Wagner, 2016; Cancé et al., 2022; Caumon and Vernoux, 2023), as being enriched (Fig. 4B, Table S6). In addition, we found a number of IAA-amino acid amidohydrolases among the respective m⁶A-carrying transcripts (Table S7), perhaps correlating with the moderately changed levels of IAA and its metabolites (Fig. 3A). Taken together, this indicates that the auxin response mediated by m⁶A is likely regulated at multiple levels, arguably with a distinct role of transcripts linked with auxin signaling.

While examining confocal optical sections of 4-day-old vir-1 and fip37-4 primary root tips, we noticed occasional cell duplication in the endodermis (Fig. 5A). The formation of the cortex, including the observed expression of the cortical reporter CO2, was generally normal (Fig. 5A, Fig. S5A). Conversely, the expression of WOX5::GFP, a marker of QC identity, was noticeably perturbed in vir-1 and fip37-4 (Fig. S5B) (Ruzicka et al., 2017; Arribas-Hernández et al., 2020). WOX5 plays a role in regulating ground tissue initials; however, the central role in controlling ground tissue development is ascribed to a bi-stable switch involving the SHR-SCR transcriptional complex, acting in concert with the auxin-promoted cyclin CYCD6;1 (Cruz-Ramírez et al., 2012; Clark et al., 2020; Van den Broeck et al., 2021). We therefore crossed the SHR, SCR and CYCD6;1 reporters to the vir-1 and fip37-4 mutants, and examined their expression in regions in which the ectopic periclinal divisions of the endodermis occur. We observed the SHR::SHR:GFP signal only in the inner endodermal cells, whereas SCR::SCR:GFP was seen in both newly formed cells (with a moderately lower fluorescence in the outer endodermal cell) (Fig. 5B,C). As the movement of SHR further out of the endodermis is directly restricted by SCR (Cui et al., 2007), it therefore seems that SCR (and likely SHR) function is not affected in m⁶A writer mutants and that the ectopically dividing cells maintain their ground tissue identity. In juvenile wild-type seedlings, the expression of CYCD6;1::GFP is confined to the ground tissue initials and their daughter cells (Sozzani et al., 2010). Remarkably, in the majority of the m⁶A writer mutant seedlings, we observed the CYCD6;1::GFP signal extended across the ground tissue, marking the additional periclinal divisions of the whole root meristem, even inside the stele (Fig. 5D,E). This indicates that m⁶A is involved in the determination of ground tissue cell fate by controlling the timing of the asymmetric cell division downstream of the SHR/SCR regulatory module.

Concomitantly, in the compiled datasets of transcripts carrying m⁶A marks (Table S3), we found neither SHR, SCR nor WOX5. Only CYCD6;1 was present in two datasets (Table S8). However, local activation of the auxin response has also been shown to promote extra cell divisions of the endodermis, including ectopic CYCD6;1 expression (Cruz-Ramírez et al., 2012; Seo et al., 2021). Ground tissue defects related to the inhibition of auxin transport have also been described (Cruz-Ramírez et al., 2012; Seo et al., 2021). Although transcripts encoding the PIN7 auxin transporter (or possibly PIN2) showed m⁶A marks (Table S5), we did not observe any change in the PIN7::PIN7:GFP expression in the proximity of endodermal periclinal divisions in the vir-1 mutant (Fig. S5C). The relevant pin mutants showed normal endodermal formation (Fig. S5D), indicating that the endodermal defects of m⁶A writer mutants are likely unrelated to auxin transport. Furthermore, although the expression of DR5::GFP was generally not directly associated with ectopic duplication in the endodermal cells (Cruz-Ramírez et al., 2012; Seo et al., 2021; Fig. S5E), we tested whether the observed defects could still be associated with erratic auxin signaling (Fig. 4B). To this end, we treated both vir-1 and fip37-4 with the inhibitor of auxin signaling α-(phenylethyl-2-oxo)-indole-3-acetic acid (PEO-IAA) (Hayashi et al., 2012). Strikingly, even lower PEO-IAA doses, which barely inhibited primary root growth (Fig. S5F), led to suppression of periclinal divisions in the endodermis of writer mutants (Fig. 5F). This indicates that the correct status of mRNA methylation is required for endodermis development and involves auxin signaling, evidenced by an ectopically activated auxin response in m⁶A writer mutants. Taken together, these findings highlight the multilevel effect of m⁶A on auxin-dependent processes (Fig. S6A).

Here, we investigate how m⁶A impacts the auxin response in Arabidopsis. Recently, several models describing a possible link between m⁶A and auxin-dependent pathways have been suggested. Bhat et al. (2020) observed decreased expression of the DR5 reporter in mta knockdown lines and connected these findings with the reduced synthesis of microRNAs targeting the genes for auxin TIR1/AFB receptors. Yet, the deduced elevated expression of the auxin receptors would perhaps lead to hypersensitivity to exogenously applied auxin (Ruegger et al., 1998). Next, in rice male generative organs, it was demonstrated that m⁶A tissue specifically influences the expression of the rate-limiting factor required for auxin synthesis, YUCCA3 (Cheng et al., 2022). In contrast, overall auxin synthesis activity (including the methylation status of prominent auxin synthesis-related transcripts, particularly YUCCAs) remains generally unaffected in the Arabidopsis m⁶A writer mutants (Fig. 3A; Table S5). Thus, the proposed mechanism appears to be different in Arabidopsis roots. In addition, Zhang et al. (2022a) observed a reduced lateral root induction rate following auxin treatment in writer mutants. They suggested that it might be exerted via altered methylation status of some ARF transcripts (mostly encoding those of class A).

Our work reveals insights with respect to Arabidopsis seedlings. The hypomorphic writer mutants exhibit dramatically depleted levels of total m⁶A pools, reaching as little as 10% of those seen in the wild type (Ruzicka et al., 2017). Consistently, these lines display a variety of phenotypes, such as altered resistance to abscisic acid or salt and cold stress, vascular, trichome or flower defects, or aberrant photosynthetic activity (Bodi et al., 2012; Ruzicka et al., 2017; Hu et al., 2021; Shoaib et al., 2021; Govindan et al., 2022; Zhang et al., 2022b; Wang et al., 2023). Thus, one can assume that the auxin response can also be affected at multiple levels. Our analysis expands the list of the candidate target transcripts with those encoding the TIR1/AFB receptors, and highlights, besides those from class A, also the class B inhibitory ARFs (Cancé et al., 2022; Caumon and Vernoux, 2023). We suggest that the resistance of writer mutants to auxin can be ascribed to the reduced m⁶A rate in mRNAs linked with auxin receptors or stimulatory class A ARFs. We also observe subtle changes in internal levels of auxin and its metabolites, mirrored by the pronounced m⁶A status in transcripts encoding the IAA amino-acid hydrolases (Fig. S6A).

The exact role of auxin signaling in ground tissue formation is not well known. It has been demonstrated previously that triggering the auxin response in endodermal cells leads to activation of the CYCD6;1 reporter and exuberant divisions of endodermal cells (Cruz-Ramírez et al., 2012; Seo et al., 2021). Both vir-1 and fip37-4 exhibit abnormal periclinal divisions in the endodermis of the primary root. The corresponding CYCD6;1 misexpression not only accompanies the additional endodermis periclinal divisions higher up in the root tip, but occasionally even occurs in the stele. As interfering with auxin signaling in these lines can partially suppress these phenotypes, one can therefore assume that the auxin response is ectopically activated in some of the endodermal cells in these lines (Fig. S6B). Neither the native cortex-endodermis initial cells nor ectopically dividing cells show increased DR5 activity (Fig. S5E). It has also been recently demonstrated that the B-type ARFs are likely not capable of activating the DR5 reporter (O'Malley et al., 2016; Galli et al., 2018; Stigliani et al., 2019; Freire-Rios et al., 2020; Rienstra et al., 2023). Both class A and B ARFs display an increased presence of the m⁶A marks in their respective transcripts (Fig. 4B, Table S6). Therefore, we suggest that the ratio of ARF As and Bs, perturbed in the m⁶A writer mutants, is reflected by the ground tissue defects and represents thereby an additional detectable level of the m⁶A-mediated auxin response (Fig. S6B).

A large number of studies carried out in various plant model systems have shown that m⁶A prevailingly enhances the stability of mRNAs (Anderson et al., 2018; Wei et al., 2018; Wang et al., 2021; Cheng et al., 2022; Hou et al., 2022) or promotes their translational efficiency (Luo et al., 2020; Hou et al., 2022; Wang et al., 2023). A major portion of SAUR mRNAs (or SAUR proteins) are short-lived (Ren and Gray, 2015). Thus, it is a remarkable coincidence that the whole class of SAUR transcripts lack m⁶A. Although SAURs cannot be therefore connected with the auxin-related phenotypes described here, we suggest that they could be used as a helpful reference for analyzing transcriptomic profiles related to m⁶A.

Unless indicated otherwise, all the Arabidopsis thaliana (L.) Heynh lines were in the Col-0 background. RNA-processing mutant lines with auxin-related defects and their corresponding background ecotypes and sources are listed in Table S1; additionally, mutant lines ect2-1 ect3-1 ect4-2 (ect234) (Arribas-Hernández et al., 2020), alkbh9b (Martínez-Pérez et al., 2017), alkbh10a and alkbh10b-2 (Duan et al., 2017), alkb10a alkbh10b-2 (this work), pin3-3 pin4-101 pin7-102 (pin347) (Willige et al., 2013) and eir1-4 (pin2) (Abas et al., 2006) were used. The following marker lines were introduced by crossing into respective genotypes: WOX5::GFP (Blilou et al., 2005), CYCD6;1::GFP (Sozzani et al., 2010), SHR::SHR:GFP (Nakajima et al., 2001), SCR::GFP-SCR/scr-4 (Cui and Benfey, 2009; because of their Ws genetic background, the outcrossed plants carrying the vir-1 mutation were compared with the wild type-like segregants), CO2::GFP (Heidstra et al., 2004; Marquès-Bueno et al., 2016), DR5::GFP (Benková et al., 2003), PIN7::PIN7:GFP (contains native 3′-UTR elements; Blilou et al., 2005).

Seeds were surface-sterilized with chlorine gas for 5 h and planted on 0.5× Murashige and Skoog medium containing 1% sucrose. Plates were kept for 2 days in the cold room (4°C) before placing them into cultivation chambers (16 h:8 h light:dark photoperiod, 22°C:18°C). Plants were grown vertically for 5 days unless indicated otherwise. For microscopy and phenotype analyses, primary roots of seedlings were used unless otherwise specified. The following chemicals were used for plant treatments: 2,4-dichlorophenoxyacetic acid (2,4-D), indole-3-acetic acid (IAA), 1-napthtoxyacetic acid (NAA), 1-N-phenoxyacetic acid (NPA), propidium iodide (PI) (Sigma-Aldrich) and α-(phenylethyl-2-oxo)-indole-3-acetic acid (PEO-IAA) (Hayashi et al., 2012). Inducible RNAi expression was controlled by germinating seeds for 6 days (together with appropriate controls) on sterile media containing 5 µM 17-β-estradiol (Sigma-Aldrich).

The assays on hypocotyl bending and apical hook development were made using a constructed RaPiD chamber (Kashkan et al., 2022a preprint, 2022b). For the quantification of root, hypocotyl and apical hook growth parameters, the photographed seedlings were measured with ImageJ software (Schneider et al., 2012). Analysis of cotyledon vein formation was performed as described previously (Ruiz Sola et al., 2017). For the assay of protoxylem defects, 5-day-old seedlings were cleared and analyzed as previously described (Bishopp et al., 2011). For the analysis of marker expression by confocal microscopy, seedlings were stained in vivo with PI. For the RAM size and endodermis analyses, seedlings (6 and 4 days old, respectively) were cleared and stained with PI as described previously (Truernit et al., 2008). RAM size was determined as the number of epidermis cells between the QC and the first elongating cell. Endodermis periclinal divisions were analyzed on a series of optical cross-sections through the apical meristem of each root using ImageJ software (Schneider et al., 2012). Routine confocal microscopy was performed on Zeiss LSM 880. The fluorescent signal was quantified using the ImageJ software. Bright-field microscopy with differential interference contrast was conducted on a Zeiss Axio Imager ApoTome2. Unless stated otherwise, at least 15 seedlings were processed in each experiment, and three independent experiments were performed, giving reproducible results. Representative experiments were selected for the presentation of data.

For quantification of the auxin metabolome, the roots of 5-day-old seedlings were used as a starting material. The procedure was followed as detailed by Novák et al. (2012). Briefly, root samples were extracted with 1 ml cold phosphate buffer (50 mM; pH 7.0) containing 0.1% sodium diethyldithiocarbamate, supplemented with internal standards. After centrifugation at 20,000 rpm (23,000 g) for 10 min, 500 µl of the extract was acidified with 1 M HCl to pH 2.7 and purified by solid phase extraction using Oasis^TM HLB columns (30 mg, 1 ml). For quantification of indole-3-pyruvic acid, the remaining 500 µl of the extract was derivatized by cysteamine (0.25 M, pH 8.0) for 1 h, acidified with 3 M HCl to pH 2.7 and purified by solid phase extraction. Samples were analyzed using the HPLC system 1290 Infinity (Agilent Technologies) equipped with a Kinetex C18 column (50 mm×2.1 mm, 1.7 µm; Phenomenex) and linked to a 6460 Triple Quadrupole (Agilent Technologies). Auxin levels were quantified using stable isotope-labeled internal standards as a reference. The samples were processed in septuplicate (each 20 mg of starting tissue).

Long-distance auxin measurements were performed as described previously (Lewis and Muday, 2009) on at least ten stem segments (20 mm) of 4-week-old plants.

Six published datasets of Arabidopsis transcripts containing enriched m⁶A marks were selected for analysis: five of them (Wan et al., 2015; Shen et al., 2016; Anderson et al., 2018; Bhat et al., 2020; Parker et al., 2020; Zhang et al., 2022a) were obtained by anti-m⁶A antibody-based sequencing methods (Dominissini et al., 2012; Meyer et al., 2012), and the remaining one originated from Nanopore direct RNA sequencing (Parker et al., 2020). The datasets from Bhat et al. (2020) and Wan et al. (2015) were thresholded according to their false discovery rate values to make the set size comparable among all analyzed data (Table S2). Duplicate transcript codes, plastidic and mitochondrial genes, pseudogenes and transposable elements were removed from all lists for consistent comparison. The coordinates of methylation sites obtained from the dataset presented by Parker et al. (2020) were converted into the gene list using bedtools (v2.27.1) (Quinlan and Hall, 2010) and Araport11 genome annotation (Cheng et al., 2017) as a reference. The Bedops software package (v2.4.37) (Neph et al., 2012) was used for the subsequent file format conversions. The UpSet intersections of the datasets were plotted using the UpSet 2.0 online tool (Lex et al., 2014).

For a comparison of means between the two groups, two-tailed unpaired Student's t-test was applied. Statistical analysis of multiple groups was determined by ANOVA with subsequent Dunnett's multiple comparisons test. The statistical tests were performed in GraphPad Prism software.

We thank G. Jia, V. Pallás, L. Arribas-Hernández, P. Brodersen, S. D. Michaels, M. Kalyna, D. Kim, C. Luschnig, M. Van Lijsebettens, N. V. Raikhel, R. Tsugeki, M. Lenhard and P. Benfey for sharing published material. Microscopy work was done with the assistance of the Imaging Facility at the Institute of Experimental Botany. The work of the Imaging Facility is supported by the MEYS, project LM2023050 Czech-BioImaging, and the European Regional Development Fund, project CZ.02.1.01/0.0/0.0/18_046/0016045.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201775.reviewer-comments.pdf.