(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Insights into the conservation and diversification of the molecular functions of YTHDF proteins [1]

['Daniel Flores-Téllez', 'University Of Copenhagen', 'Biology Department. Copenhagen', 'Universidad Francisco De Vitoria', 'Facultad De Ciencias Experimentales. Pozuelo De Alarcón', 'Madrid', 'Mathias Due Tankmar', 'Sören Von Bülow', 'Junyu Chen', 'Kresten Lindorff-Larsen']

Date: 2023-10

YT521-B homology (YTH) domain proteins act as readers of N6-methyladenosine (m 6 A) in mRNA. Members of the YTHDF clade determine properties of m 6 A-containing mRNAs in the cytoplasm. Vertebrates encode three YTHDF proteins whose possible functional specialization is debated. In land plants, the YTHDF clade has expanded from one member in basal lineages to eleven so-called EVOLUTIONARILY CONSERVED C-TERMINAL REGION1-11 (ECT1-11) proteins in Arabidopsis thaliana, named after the conserved YTH domain placed behind a long N-terminal intrinsically disordered region (IDR). ECT2, ECT3 and ECT4 show genetic redundancy in stimulation of primed stem cell division, but the origin and implications of YTHDF expansion in higher plants are unknown, as it is unclear whether it involves acquisition of fundamentally different molecular properties, in particular of their divergent IDRs. Here, we use functional complementation of ect2/ect3/ect4 mutants to test whether different YTHDF proteins can perform the same function when similarly expressed in leaf primordia. We show that stimulation of primordial cell division relies on an ancestral molecular function of the m 6 A-YTHDF axis in land plants that is present in bryophytes and is conserved over YTHDF diversification, as it appears in all major clades of YTHDF proteins in flowering plants. Importantly, although our results indicate that the YTH domains of all arabidopsis ECT proteins have m 6 A-binding capacity, lineage-specific neo-functionalization of ECT1, ECT9 and ECT11 happened after late duplication events, and involves altered properties of both the YTH domains, and, especially, of the IDRs. We also identify two biophysical properties recurrent in IDRs of YTHDF proteins able to complement ect2 ect3 ect4 mutants, a clear phase separation propensity and a charge distribution that creates electric dipoles. Human and fly YTHDFs do not have IDRs with this combination of properties and cannot replace ECT2/3/4 function in arabidopsis, perhaps suggesting different molecular activities of YTHDF proteins between major taxa.

Regulation of gene expression is essential to life. It ensures correct balancing of cellular activities and the controlled proliferation and differentiation necessary for the development of multicellular organisms. Methylation of adenosines in mRNA (m 6 A) contributes to genetic control, and absence of m 6 A impairs embryo development in plants and vertebrates. m 6 A-dependent regulation can be exerted by a group of cytoplasmic proteins called YTHDFs. Higher plants have many more YTHDFs than animals, but it is unknown whether these many YTHDF proteins carry out fundamentally different or roughly the same molecular functions, as only a small fraction of them have been studied thus far. This work addresses the origin and reasons behind YTHDF expansion during land plant evolution and reveals that most YTHDFs in flowering plants have the same molecular functions that facilitate rapid division of differentiating stem cells. Remarkably, this molecular activity is present in the most basal lineage of land plants and functions similarly across 450 million years of land plant evolution. We also identified a few plant YTHDF proteins with divergent molecular function despite their ability to bind to m 6 A. Our work provides a firm basis for further advances on understanding molecular properties and biological contexts underlying YTHDF diversification.

Funding: This work was supported by the Novo Nordisk Foundation through a Hallas-Møller Ascending Investigator 2019 grant (NNF19OC0054973) to P.B. and a grant to K.L.-L. via the PRISM (Protein Interactions and Stability in Medicine and Genomics) center (NNF18OC0033950), by the European Molecular Biology Organisation through a long-term fellowship (ALTF 810-2022) to S.v.B., by the European Research Council through a Consolidator grant (ERC-2016-CoG 726417 PATHORISC) to P.B., and by the Independent Research Fund Denmark through a Research Project2 grant (9040-00409B) to P.B. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The following authors received salary from the funders: M.D.T. (Novo Nordisk Foundation, grant NNF19OC0054973); S.v.B. (Novo Nordisk Foundation, grant NNF18OC0033950, and European Molecular Biology Organisation, grant ALTF 810-2022); J.C. (Novo Nordisk Foundation, grant NNF19OC0054973); L.A.-H. (European Research Council grant ERC-2016-CoG 726417, Independent Research Fund Denmark, grant 9040-00409B, and Novo Nordisk Foundation, grant NNF19OC0054973).

In this study, we systematically define overlaps in the molecular functions of YTHDF proteins in land plants (embryophytes). Employing functional complementation of leaf formation in triple ect2/ect3/ect4 (te234) mutants, we demonstrate that at least one member of all clades in arabidopsis, and the only YTHDF protein from a group of primitive land plants, can replace ECT2/3/4 functions. In contrast, a few late-diverging ECTs were not able to perform the molecular functions of ECT2/3/4. Based on these results, we propose an ancestral molecular role of land plant YTHDF proteins in stimulation of primordial cell proliferation, and sustained functional redundancy during the diversification process that started more than 400 million years ago (Mya). In addition, our results also support the functional divergence of a small subset of fast-evolving plant YTHDF proteins with contributions to specialization mainly from the IDRs, but also from the YTH domains.

The great expansion of YTHDF proteins in higher plants is unique among eukaryotes [ 8 ]. Phylogenetic analyses of plant YTHDF domains have established the existence of 3 clades in angiosperms, DF-A (comprising Ath ECT1, ECT2, ECT3, ECT4), DF-B (comprising Ath ECT5, ECT10, ECT9), and DF-C (comprising Ath ECT6, ECT7, ECT8 and ECT11) [ 26 ]. The fact that the eleven arabidopsis paralogs have the aromatic residues necessary for specific binding to m 6 A [ 49 ] suggests that they may all function as m 6 A readers. The unified model for YTHDF function recently proposed for the three vertebrate paralogs [ 18 , 22 , 39 ] is consistent with what had already been established for Ath ECT2, ECT3 and ECT4 in the plant DF-A clade [ 16 ], even though the three plant paralogs are more divergent in sequence than the highly similar mammalian YTHDF1-3 [ 8 , 16 , 39 ]. The two most highly expressed members in A. thaliana, ECT2 and ECT3, accumulate in dividing cells of organ primordia and exhibit genetic redundancy in the stimulation of stem cell proliferation during organogenesis [ 16 , 17 ]. The two proteins probably act truly redundantly in vivo to control this process, because they associate with highly overlapping target sets in wild type plants, and each exhibits increased target mRNA occupancy in the absence of the other protein [ 23 ]. Simultaneous knockout of ECT2 and ECT3 causes a 2-day delay in the emergence of the first true leaves, aberrant leaf morphology, slow root growth and defective root growth directionality among other defects [ 16 , 17 ] that resemble those of plants with diminished m 6 A deposition [ 50 – 52 ]. For the third DF-A clade member, ECT4, the genetic redundancy is only noticeable in some tissues as an exacerbation of ect2/ect3 phenotypes upon additional mutation of ECT4, most conspicuously seen in leaf morphogenesis [ 16 , 17 ]. Despite the strong evidence for redundant functions among these three YTHDF paralogs in the DF-A clade, the presence of many other YTHDF proteins in arabidopsis leaves open the question of whether substantial functional specialization of YTHDF proteins exists in plants.

While yeast, flies and primitive land plants encode only one YTHDF protein [ 24 – 26 , 37 , 38 ], vertebrates have three closely related paralogs (YTHDF1-3) [ 6 , 8 , 15 , 39 ] and higher plants encode an expanded family, with eleven members in Arabidopsis thaliana (Ath) referred to as EVOLUTIONARILY CONSERVED C-TERMINAL REGION1-11 (ECT1-11) [ 27 , 40 ]. It is a question of fundamental importance for the understanding of how complex eukaryotic systems use the regulatory potential of m 6 A whether these many YTHDF proteins perform the same biochemical function, or whether their molecular properties are specialized. Such molecular specialization could, for instance, arise as a consequence of differential binding specificity to mRNA targets, distinct protein-protein interaction properties, or distinct biophysical properties of IDRs, such as those related to the propensity to phase separate [ 41 , 42 ]. Alternatively, diversification of biological functions could be achieved with distinct expression patterns or induction by environmental cues, even if targets and molecular functions are the same. The studies on YTHDF specialization in vertebrate cells illustrate the importance of the topic, and, equally, the importance of addressing it rigorously [ 43 ]. Initial work on mammalian cell cultures advocated a model in which YTHDF1 would enhance translation of target mRNAs, YTHDF2 would promote mRNA decay, and YTHDF3 would be able to trigger either of the two [ 9 , 44 – 46 ]. Nonetheless, subsequent research work in mouse, zebrafish and human cell culture involving single and combined ythdf knockouts and analysis of interacting mRNAs and proteins [ 18 , 22 , 39 ], and comparative studies of human YTHDF1/2/3 structure and molecular dynamics [ 47 ] do not support functional specialization, and propose a unified molecular function for all three vertebrate YTHDFs in accelerating mRNA decay. Whether metazoan YTHDF1/2/3 are molecularly redundant or not, and what their precise molecular functions may be, are topics that continue to be debated as of today [ 48 ].

Two different phylogenetic clades of YTH domains have been defined, YTHDC and YTHDF, sometimes referred to simply as DC and DF [ 6 , 15 ]. Genetic studies show that major functions of m 6 A in development in both vertebrates and higher plants depend on the YTHDF clade of readers [ 16 – 22 ]. In all cases studied in detail thus far, plant and animal YTHDF proteins are cytoplasmic in unchallenged conditions [ 15 , 16 , 23 – 26 ], and contain a long N-terminal intrinsically disordered region (IDR) in addition to the C-terminal YTH domain [ 6 , 8 , 27 ]. While the YTH domain is necessary for specific binding to m 6 A in mRNA [ 9 – 14 ], the IDR is considered to be the effector part of the protein [ 9 , 28 – 30 ]. Nonetheless, it has been proposed that the IDR may also participate in RNA binding, because the YTH domain alone has low affinity for mRNA [ 6 ]. Indeed, the IDR-dependent crosslinks between a YTHDF protein and mRNA detected upon UV-irradiation of living arabidopsis seedlings [ 31 ] experimentally supports such a mechanism, conceptually equivalent to the contribution of IDRs in transcription factors to specific DNA binding [ 32 ]. Additionally, the IDR may be involved in phase separation. Plant YTHDF proteins can form condensates in vitro [ 16 ] and, upon stress, localize in vivo to distinct foci [ 16 , 26 ] identified as stress granules [ 26 ]. Similar properties have been reported for human YTHDFs, which engage in liquid-liquid phase separation when concentrated on polyvalent m 6 A-modified target RNA in vitro [ 33 – 36 ].

N6-methyladenosine (m 6 A) is the most abundant modified nucleotide occurring internally in eukaryotic mRNA. It is of major importance in gene regulation as illustrated by the embryonic lethality of mutants in the dedicated mRNA adenosine methyltransferase in higher plants [ 1 ] and in mammals [ 2 ]. The presence of m 6 A in an mRNA may have multiple biochemical consequences. These include changes in the secondary structure [ 3 – 5 ] and the creation of binding sites for RNA-binding proteins specialized for m 6 A recognition [ 6 , 7 ]. YT521-B homology (YTH) domain proteins [ 8 ] constitute the best studied class of m 6 A-binding proteins [ 6 , 7 , 9 ]. They achieve specificity for m 6 A via an aromatic pocket accommodating the N6-adenosine methyl group, such that the affinity of isolated YTH domains for m 6 A-containing RNA is 10-20-fold higher than for unmodified RNA [ 9 – 14 ].

Results

The phylogeny of YTHDF proteins comprises several clades in land plants The adaptation of plants to terrestrial life was accompanied by the acquisition of morphological, physiological, and genomic complexity [53–55]. Knowing how diversification of YTHDF proteins came about during the course of plant evolution is relevant to understand their distinct functions and may hint at roles they might have played in this process. However, the species included in the so far most detailed phylogenetic study on plant YTHDF proteins jump from bryophytes (mosses, liverworts and hornworts [54,56–60]) to angiosperms (flowering plants) [26]. To include the YTHDF repertoires of intermediately diverging clades of embryophytes such as lycophytes, ferns, and gymnosperms, we performed phylogenetic analyses using YTHDF proteins from 32 species widely spread across land plant evolution. We also included YTHDFs from 8 species of green algae (chlorophytes and charophytes) and, as outgroup, 3 microalgal species from eukaryotic taxa evolutionarily close to plants (Fig 1A). In the resulting phylogenetic tree, all embryophyte YTHDFs branch from a single stem that connects this monophyletic group to the more divergent green algal orthologs (Figs 1B and S1–S3), whose peculiarities are further considered in the discussion. Regarding land plants, the tree largely agrees with the DF-A/B/C clades defined on the basis of angiosperm YTHDF sequences [26], with some minor differences. We consider that the former DF-C group comprising Ath ECT6/7/8/11 [26] can be subdivided into two groups that diverged early during YTHDF radiation (Figs 1B and S1). Thus, we introduced the name ‘DF-D’ for the clade defined by one Amborella trichopoda (a basal angiosperm) YTHDF protein (Atr DFD) and Ath ECT6/7 (Figs 1B and S1). Furthermore, since the group comprising bryophyte YTHDFs did not receive a designation in previous studies, we named it DF-E (mnemotechnic for ‘Early’). It includes YTHDF proteins from bryophytes and some lycophyte orthologs that branched close to the moss subgroup (Figs 1B and S1). Finally, an additional group composed of YTHDF proteins from six species of ferns and two gymnosperms was named DF-F, as they did not fall into any of the other clades (Figs 1B and S1). Interestingly, the alignment revealed independent gains and losses of molecular traits during plant YTHDF evolution. For example, an aspartate-to-asparagine substitution in helix α1 that increases affinity for m6A [61], characteristic of DC and plant DF-B YTH domain proteins [26, 61], is also present in the fern proteins of the DF-F clade, but absent in DF-Es, DF-As and gymnosperm DF-Fs (S4 Fig), suggesting independent losses or acquisitions. Taking these observations together, we conclude that land plant YTHDFs are phylogenetically more diverse than previously appreciated. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Phylogenetic analysis of YTHDF proteins in plants. (A) Schematic representation of Viridiplantae evolution and relative position of the species used in this study. The architecture of the diagram and the age (million years, My) indicated on some nodes are a simplified version of the trees from Su et al. [59] and Bowman [62]. The length of the branches has been adjusted for illustrative purposes. (B) Phylogenetic tree of YTHDF proteins in plants (S1–S4 Datasets), with color-coding and abbreviations of species names as in A. Three YTHDF proteins from taxa outside but evolutionarily close to Viridiplantae are included as outgroup. Arabidopsis thaliana (Ath) YTHDF proteins are named after the original nomenclature for proteins containing an Evolutionarily Conserved C-Terminal Region (ECT) established by Ok et al. [40]. Proteins from other plant species adhere to the nomenclature established by Scutenaire et al. [26], with small variations reflecting additional clades (DF-E, -F) and the split of the original DF-C clade into -C and -D. Statistical support is calculated as approximate likelihood-ratio test (aLRT), and indicated by grayscale-coded spheres and values on the most relevant nodes. For simplicity, only a subset of proteins is labelled–those from one representative species of the main taxa highlighted in bold in A–, but a comprehensive representation of the same tree with all protein names and aLRT values is available in S1 Fig. The length of the branches represents the evolutionary distance in number of amino acid substitutions per site relative to the scale bar. Double lines crossing the longest branches indicate that the branch has been collapsed due to space problems, but a schematic representation of their relative lengths is shown on the lower-left corner, and not-collapsed branches can be found in S1 Fig. https://doi.org/10.1371/journal.pgen.1010980.g001

YTHDF protein diversification occurred early during land plant evolution Our phylogenetic analysis shows that plant YTHDF diversification started at least before the radiation of Euphyllophytes (plants with true leaves comprising ferns, gymnosperms and angiosperms) more than 410 Mya [59,63]. This is because bryophytes possess one or two YTHDFs in the ‘early clade’ (DF-E), while all six fern species analyzed here have extended sets DF proteins (4–9 paralogs) that spread into clades DF-F, DF-D, and the branch sustaining the DF-A and DF-B groups (Fig 1B). Furthermore, because one DF-protein from the lycophyte Isoetes taiwanensis branches from the common stem between clades DF-C and DF-D (Ita DF-CD), it is likely that a first diversification event started in the ancestral vascular plants. Thus, our analysis reveals that YTHDF radiation started early in land plant evolution and coincided with the acquisition of morphological complexity and the adaptation to diverse environments.

A functional complementation assay for plant YTHDF proteins To address the degree of functional specialization among plant YTHDF proteins in a simple manner, we set up a functional complementation assay that scores the ability of each of the eleven arabidopsis YTHDF proteins (Fig 2A) to perform the molecular functions of ECT2/3/4 required for rapid cellular proliferation in leaf primordia. Initial attempts at using the ECT2 promoter for ectopic ECT expression in primordial cells were unsuccessful, as pilot experiments with a genomic ECT4 fragment revealed that its expression was substantially lower than that of a similar ECT2 genomic fragment when driven by the ECT2 promoter (S5 Fig), perhaps pointing to the presence of internal cis-regulatory elements. We therefore turned to the use of ECT cDNAs (cECTs) under the control of the promoter of the ribosomal protein-encoding gene RPS5A/uS7B/US7Y (At3g11940 [64,65]), henceforth US7Yp, active in dividing cells [66] in a pattern that resembles the ECT2/3/4 expression domain [16,17]. Transformation of the US7Yp:cECT2-mCherry construct in te234 plants resulted in complementation frequencies of ~40–50% among primary transformants, a percentage slightly lower but comparable to that obtained with the genomic ECT2-mCherry construct under the control of ECT2 promoter and terminator regions (ECT2p:gECT2-mCherry) [16] (S6A Fig). Free mCherry expressed from a US7Yp:mCherry transgene showed no complementation (S6A Fig), as expected. Importantly, the leaf morphology and the pattern of mCherry fluorescence in US7Yp:cECT2-mCherry lines was indistinguishable from that of ECT2p:gECT2-mCherry (S6B Fig). Hence, we proceeded with expression of cDNAs encoding all of ECT1-11 fused to C-terminal mCherry under the control of the US7Y promoter (henceforth ECT1-11) in te234 mutants (Fig 2B). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Most but not all arabidopsis YTHDF proteins can replace ECT2/3/4-function. (A) Diagram showing the relative length of the N-terminal IDRs of Arabidopsis thaliana (Ath) YTHDF proteins (ECT1-ECT11) together with Marchantia polymorpha (Mpo) DFE, Homo sapiens (Hs) YTHDF2, and Drosophila melanogaster (Dm) YTHDF. Numbers indicate the length of the proteins in amino acids (aa). (B) Strategy followed for the functional assay. US7Yp:cECT(X)-mCherry constructs are introduced in ect2-1/ect3-1/ect4-2 (te234) plants, and complementation rates are estimated by the percentage of primary transformants (T1) whose first true leaves have a size (s) exceeding 0.5 mm after 10 days of growth. The construct US7Yp:mCherry is used as negative control. Examples of T1 seedlings expressing control and ECT2 constructs are shown. Dashed outlines are magnified below to show mCherry fluorescence in emerging leaves. (C) Weighed averages of the complementation percentages observed for each US7Yp:cECT(X)-mCherry construct in 2–5 independent transformations (I.T.). Letters between parenthesis indicate the DF- clade assigned to each protein. Colouring highlights ECT2 (reference) and proteins with low or negligible complementation capacity (ECT1/9/11) for fast identification along the manuscript. Low transformation efficiency (T.E.) for ECT9 can be inferred by the low number of transformants (n) highlighted in red, but detailed T.E. and raw complementation rates in each transformation can be found in S9 and S7A Figs respectively. Additional data regarding this assay is shown in S8 Fig. (D) Same as C on a single transformation using the rdr6-12/te234 background. (E-H) Rosette phenotype of 32-day old T1s bearing US7Yp:cECT(X)-mCherry constructs in the rdr6-12/te234 background, sorted according to their degree of complementation in three panels: F (high), G (residual) and H (non-complementation and enhancement). Different genetic backgrounds and control transformants (mCherry) are shown in E as a reference. The scalebar (5 cm) applies to all plants in the four panels. https://doi.org/10.1371/journal.pgen.1010980.g002

Percentage of complementation among primary transformants as a readout for functionality To design the experimental approach in detail, we considered the fact that expression of transgenes involves severe variations in levels and patterns among transformants. This is due to positional effects of the T-DNA insertion and the propensity of transgenes to trigger silencing in plants [67], and explains why only a fraction of ECT2-mCherry lines rescues loss of ECT2 function (S6A Fig, [16]). Hence, many independent lines need to be analyzed before choosing stable and representative lines for further studies. Taking this into account, we decided to perform systematic and unbiased comparisons of ECT functionality by counting the fraction of primary transformants (henceforth T1s) able to complement the late leaf emergence of te234 plants for each construct. We set a size threshold for the rapid assessment of large numbers of transformants considering that leaves longer than 0.5 mm after 10 days of growth constitute unambiguous sign of complementation, because this is the minimum size never reached by te234 or control lines (S2B Fig). Hence, we calculated complementation percentages as the number of T1s with leaf size (s) larger than 0.5 mm relative to the total number of transformants, and averaged that score over several independent transformations (Figs 2C and S7A, S8A).

Most arabidopsis YTHDF paralogs can perform the molecular function required for leaf development The results of the comparative analysis reveal a high degree of functional overlap within the arabidopsis YTHDF family, because eight out of the eleven YTHDF proteins, namely ECT2/3/4/5/6/7/8/10, complement the delayed leaf emergence of ect2/3/4 mutants with frequencies ranging from 14% (ECT6) to 39% (ECT2) when expressed in the same cells (Fig 2C). Importantly, complementation is indirect proof of m6A-binding ability, because the proteins must bind to the targets whose m6A-dependent regulation is necessary for correct leaf organogenesis in order to restore ECT2/3/4 function [16]. This conclusion is not trivial, as some eukaryotes lacking canonical m6A-methyltransferases encode YTH domain proteins [68] and, indeed, the YTH domain of fission yeast Mmi1 does not bind to m6A [69]. Thus, most, but not all, arabidopsis YTHDF proteins retain the molecular functions required to stimulate proliferation of primed stem cells. Of the three remaining proteins, ECT1 showed only residual complementation capacity, while no complementation was observed for ECT11 at this stage (Fig 2C). Finally, we could not conclude at this point whether ECT9 could replace ECT2/3/4 function or not, because this transgene produced very few transformants in the te234 background with barely visible fluorescence (Figs 2C and S7–S9). All such transformants underwent complete silencing in the following generation (T2).

Differences in complementation scores are not explained by a bias in expression levels Our T1 scoring method relies on the assumption that the US7Y promoter drives similar ranges of expression of the different ECT cDNAs. However, different silencing propensities between the ECT cDNAs might introduce bias in the results. To take this possibility into account, we first performed western blotting from three lines per transgene selected as the best-complementing in their group (S8B Fig). The results showed that the higher complementation scores of e.g. ECT2, ECT3 and ECT8 are not explained by higher expression. Rather, the reverse was true. For example, while ECT2 could complement te234 plants with the lowest expression among all ECTs, the few lines expressing ECT1 able to pass the complementation threshold had the highest expression of all ECTs, suggesting that ECT1 can only partially compensate for loss of ECT2/3/4 when overexpressed.

The same functional classes of ECTs emerge from complementation of silencing-deficient te234 Next, we tested whether the same results could be reproduced in plants lacking RNA-DEPENDENT RNA POLYMERASE 6 (RDR6), a gene essential for the biogenesis of transgene-derived short interfering RNAs (siRNAs) [70, 71]. Because siRNA-guided transgene silencing is impaired in rdr6 mutants, higher and more homogenous ranges of expression among different transgenes can be expected in this background. Hence, we constructed rdr6-12/te234 mutants that exhibited both the narrow, downward-curling leaf phenotype of rdr6 mutants [72] and the characteristic rosette morphology of te234 [16] (S10A Fig). Importantly, the delay in leaf emergence was identical in te234 and rdr6-12/te234 mutants (S10A Fig) and the complementation rates of ECT2 transgenes were much higher in the silencing-deficient plants (S6A Fig), indicating that the <100% complementation frequencies in te234 were indeed due to RDR6-dependent silencing. In the rdr6/te234 background, complementation scores of the eight functionally equivalent ECTs (ECT2/3/4/5/6/7/8/10) ranged from 59% (ECT6) to 94% (ECT2), and the differences between these ECTs were less pronounced, but followed a trend largely unchanged compared to the te234 background (Fig 2C and 2D). On the other hand, the difference between the eight ECTs with ECT2/3/4-like function and the group of ECTs with negligible complementation capacity (ECT1/9/11) was highlighted by the rdr6 mutation: ECT1 and ECT11 did not increase their scores (Fig 2D), and absence of silencing resulted in a number of ECT9 transformants with visible mCherry fluorescence, yet no complementation activity (Figs 2D and S10B), thus verifying the lack of ECT2/3/4-like function of ECT9.

Rosette phenotypes corroborate the complementation scores To verify the results of the complementation assay in the adult phenotype, we characterized rosettes of two independent lines for each transgene (Fig 2E–2H). As expected, rdr6-12/te234 plants expressing ECTs with high complementation scores resembled single rdr6-12 mutants (Fig 2E and 2F), while the ECT1 and ECT11 transformants with the biggest first true leaves at 10 DAG produced plants only slightly bigger than the rdr6-12/te234 background (Fig 2E and 2G), indicating low but residual ECT2/3/4-like function for these two proteins. Strikingly, rdr6-12/te234 plants expressing ECT9 exhibited strong developmental phenotypes (Figs 2H and S10B) that resembled those of severely m6A-depleted plants [51,52]. Such an effect was not observed for any other Ath ECT. This result suggests that ECT9 can bind to m6A targets, either exerting opposite molecular actions compared to ECT2/3/4, or simply precluding their possible regulation by remaining ECTs such as ECT5/6/7/8/10 by competitive binding. We tested whether ECT9 can outcompete endogenous ECT2/3/4 in the proliferating cells of the apex by expressing the ECT9 transgene in single rdr6-12 mutant plants. We found no evidence for this, because the transformation produced many fluorescent T1s without obvious defects (S11 Fig). Altogether, our results indicate that arabidopsis YTHDF proteins may be divided into three functional classes, i) ECT2/3/4/5/6/7/8/10 with the molecular functionality of ECT2/3/4 required for correct and timely leaf formation; ii) ECT1/11, with only residual ECT2/3/4 functions, and iii) ECT9, unable to perform such functions and toxic when present in the expression domain of ECT2/3/4 in their absence.

Validation of the functional assay by loss-of-function genetic analysis One of the predictions on genetic interactions between Ath ECT genes that emerges from our results is particularly unexpected and, therefore, suitable for the validation of our complementation assay by classical genetic analysis: Mutation of ECT1 should not exacerbate the developmental defects of te234, even if it is endogenously expressed in leaf primordia. ECT1 lends itself well to validation for two reasons. First, ECT1 is the closest paralog of ECT2/3/4 (Fig 1B), intuitively suggesting some degree of genetic redundancy. Second, ECT1 promoter-reporter fusions [40] and mRNA-seq data ([73], S12 Fig) suggest that expression levels and tissue specificity of ECT1 are comparable to that of ECT4, whose activity is easily revealed as an enhancer mutant of ect2/ect3 [16].

ECT1 and ECT2/3/4 are cytoplasmic and are expressed in the same cell types We first assessed the expression pattern and subcellular localization of ECT1 using stable transgenic lines expressing TFP translationally fused to the C-terminus of ECT1 (ECT1-TFP, S13A Fig). The pattern of turquoise fluorescence was strongly reminiscent of that of fluorescent ECT2/3/4 fusions [16] with signal at the base of young leaves, in leaf primordia, main root tips and lateral root primordia at different stages (Fig 3A–3D). In addition, confocal microscopy of meristematic root cells showed cytoplasmic ECT1-TFP signal with heterogenous texture similar to ECT2/3/4 [16] (Fig 3E). We noticed, however, that most root tips had a few cells containing distinct ECT1-TFP foci (Fig 3F) in the absence of stress. We found no trace of free TFP that could be causing imaging artifacts (S13A Fig) but note that the ECT1-TFP transgenic lines selected on the basis of visible and reproducible pattern of fluorescence may overexpress ECT1 (S13B Fig). We conclude from the similar expression pattern and subcellular localization of ECT1 and ECT2/3/4 that assessment of the possible exacerbation of te234 phenotypes by additional loss of ECT1 function is a meaningful test of our functional assay. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Endogenous ECT1 has negligible ECT2/3/4-like function, despite similar expression pattern and subcellular localization. (A-D) Expression pattern of ECT1p:gECT1-TFP (gDNA) in aerial organs (A), main root (B), lateral root primordia (C) and emerging lateral roots (D) of 10-day-old seedlings. The expression mimics the pattern observed for ECT2/3/4 (Arribas-Hernández et al., 2020). (E-F) Intracellular localization of ECT1-TFP in meristematic cells of root tips in unchallenged conditions. Although the cytoplasmic signal is largely homogenous I, sporadic foci (F) are frequently observed. S13A Fig shows the integrity of the fluorescently tagged protein in independent transgenic lines assessed by protein blot. (G) Schematic representation of the ECT1 locus. Exons are represented as boxes and introns as lines. The positions and identifiers of the T-DNA insertions assigned to the ect1-1, ect1-2 and ect1-3 alleles are marked, and the location of qPCR amplicons and hybridization probes (P) for analyses is indicated below. Genotyping of ect alleles is shown in S13C–S13F Fig. (H) Expression analysis of ECT1 mRNA in wild type and T-DNA insertion lines by qPCR. *P<0.01, **P<0.001, ***P<0.0001, and ****P<0.00001 in T-test pairwise comparisons (supporting data in S9 Dataset). Northern blot using the probe (P) marked in G detects ECT1-TFP mRNA, but the endogenous ECT1 transcript is below detection limit (S13B Fig). (I) Morphological appearance of seedlings with or without ECT1 in the different backgrounds indicated. Nomenclature and additional phenotyping of these and alternative allele combinations can be found in S14 Fig. DAG, days after germination. Scale bars are: 1 cm in A, 1 mm in B, 100 μm in C, 10 μm in D-F, 1 mm in upper panels of I, and 1 cm in the lower panels. https://doi.org/10.1371/journal.pgen.1010980.g003

Mutation of ECT1 does not exacerbate the phenotype of ect2/ect3/ect4 plants We isolated three homozygous knockout lines caused by T-DNA insertions in the ECT1 gene, ect1-1, ect1-2, and ect1-3 (Figs 3G and 3H and S13C–S13F), and crossed two of them to different ect2/ect3/ect4 triple mutants [16,17] to obtain two independent allele combinations of quadruple ect mutants, qe1234 and Gqe1234 (see S14A Fig for nomenclature). In both cases, the quadruple mutant plants were indistinguishable from the corresponding te234 and Gte234 triple mutant parentals (Figs 3I and S14B). Similarly, triple ect1/ect2/ect3 mutants were identical to double ect2/ect3 mutants, and single ect1 mutants did not show any obvious defects in ect2/3-associated traits such as leaf or root organogenesis and trichome branching (Figs 3I and S14B–S14D). Thus, despite their similar expression pattern, there is no indication of redundancy between ECT1 and the other three Ath DF-A paralogs. This unexpected conclusion agrees with the poor ECT2/3/4-like activity of ECT1 in our functional assay, thereby validating the approach.

The divergent function of ECT1 and ECT11 is caused mainly by properties of their IDRs Having established that different molecular functions are represented in the arabidopsis YTHDF family, we went on to map the molecular regions responsible for this functional divergence. First, we investigated whether lack of te234 complementation by ECT1/9/11 is due to altered target binding by the YTH domain, IDR-related effector functions, or a combination of both. Thus, we built chimeric IDR/YTH constructs between ECT2 and ECT1/9/11 (Fig 4A). We also tested an ECT8 IDR /ECT2 YTH chimera (C-8/2) to assess whether hybrid constructs can at all be functional, choosing ECT8 as a positive control due to its robust complementation of seedling and adult phenotypes (Fig 2C–2D, 2F). Expression of the constructs in te234 mutants showed that the C-8/2 chimera was fully functional (Figs 4B and S7B, S15), thus validating the approach. The ECT1- and ECT11-derived hybrids showed that their YTH domains retain m6A-binding ability, because the complementation frequencies for the C-2/1 and C-2/11 chimeras were not zero (Fig 4B), contrary to an m6A-binding deficient point mutant of ECT2 [16]. They also proved, however, that both the IDR and the YTH domain of these two proteins only retain a reduced degree of function compared to the equivalent regions in ECT2. In particular, the IDRs of both proteins scored lower than their YTH domains when combined with the other half from ECT2 (Fig 4B), pointing to divergence in the IDR as the main cause for different functionality. In wild type ECT1 and ECT11 proteins, the combination of the two poorly-performing parts is likely the cause of the minute ECT2/3/4-like activity. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Dissection of ECT1/9/11 functional regions. (A) Schematic representation of the strategy followed to express chimeras with N-terminal IDR and YTH domains of different ECT proteins. (B) Weighed averages of the complementation rates observed for each chimeric construct measured as in Fig 2B and 2C. The raw complementation rates and transformation efficiencies of each independent transformation (I.T.) can be found in S7B and S9 Figs respectively, and photographs of the scored 10-day-old T1 seedlings with mCherry fluorescence are shown in S15 Fig. (C) 10-day-old primary transformants of rdr6-12/te234 with the indicated transgenes. Dashed outlines are magnified below to show mCherry fluorescence. (D) Same genotypes as in C at 26 days after germination. Additional independent transgenic lines, developmental stages and controls for C and D can be found in S10 Fig. All transgenes are expressed from the US7Y promoter and contain mCherry fused to the C-terminus. https://doi.org/10.1371/journal.pgen.1010980.g004

The IDR of ECT9 is incapable of performing ECT2/3/4 molecular functions, but its YTH domain retains some ECT2/3/4-like function The ECT2 IDR /ECT9 YTH chimera (C-2/9) yielded normal transformation efficiencies and up to 17% of T1s had signs of complementation compared to 56% for ECT2 (Figs 4B and S7B, S9, S15), suggesting that the YTH domain of ECT9 retains at least partial functionality, including m6A-binding activity. However, the reverse construct (ECT9 IDR /ECT2 YTH , C-9/2) recapitulated the results obtained for ECT9: low transformation efficiency and low fluorescence in the few T1s recovered, and complete silencing in the next generation (Figs 4B and S7B, S9, S15). Hence, we introduced the C-9/2 chimera into the rdr6-12/te234 background and observed plants with stronger fluorescence exhibiting aberrant morphology, again similar to what was obtained with ECT9 (Figs 4C, 4D and S7C, S10B). These results confirm that ECT9 is not functionally equivalent to ECT2/3/4, and point to the IDR as the main site of functional divergence of the protein.

The IDRs of ECT1/9/11 have biophysical properties distinct from ECT2/3/4/5/6/7/8/10 We next focused on the N-terminal regions of Ath ECT1/9/11. Because protein disorder predictions using MobiDB [77] confirmed that they are all disordered (S18 Fig), we used approaches suitable for the study of IDRs. Prediction of short linear motifs (SLiMs) that may mediate interaction with effector proteins gave many possibilities that did not follow a clear pattern when compared between the different ECT classes (S18 Fig). Hence, we decided to study other properties known to be important in IDRs and focused on the propensity to form liquid condensates, the amino acid composition, and the distribution of charge. To determine whether the distinct functional classes of ECTs differed in their propensity to phase separate, we employed a recently developed coarse-grained model of IDRs with residue-specific parameters estimated from experimental observations of many IDRs [78] to simulate their tendency to remain in the condensed phase or disperse in solution. As control, we used the well-studied IDR of the human RNA-binding protein Fused in Sarcoma (FUS) [79–82]. The results showed that the IDR of ECT2 has a clear propensity to remain in the condensed phase (Fig 5B–5C and S1 Movie) with a highly favorable Gibb’s free energy change for the transition from dilute aqueous solution to condensed phase (Fig 5D), comparable to, albeit not as pronounced as for the IDR of Hs FUS (Fig 5B–5D). While the IDRs of most Ath ECT proteins behaved comparably to ECT2, the IDRs of ECT1 and ECT11 were clear outliers (Figs 5B–5D and S19A). Compared to all other ECTs, the IDR of ECT1 showed a much stronger propensity to phase separate, consistent with the spontaneous granule formation of ECT1-TFP in vivo (Figs 3F and 5B–5D, and S2 Movie). On the contrary, the IDR of ECT11 dispersed into solution much more readily (Fig 5B–5D and S3 Movie). Thus, the IDRs of ECT1 and ECT11 have markedly different biophysical properties from those of ECT2/3/4/5/6/7/8/10, perhaps contributing to their different functionality in vivo. In contrast, the IDR of ECT9 did not stand out from the rest in the simulations of phase separation propensity. We also note that the different phase separation propensity among the ECTs is driven, in the main, by their different content of sticky residues as defined by the CALVADOS model [78] (S19B Fig). In this context, it is potentially of interest that ECT1, ECT9 and ECT11 all have a higher content of Phe residues than the nearly Phe-depleted IDRs of ECT2/3/4/5/6/7/8/10 (S19C Fig). For ECT1, whose Tyr and Trp enrichment is comparable to that of ECT2/3/4/5/6/7/8/10 (S19C Fig), the additional enrichment of the similarly sticky Phe may contribute to its more pronounced phase separation propensity. We finally analyzed the charge distribution, as this may dictate properties such as compactness and shape of IDR conformations [83, 84]. We found that the IDRs of ECT2/3/4/5/6/7/8/10 show a recurrent pattern with accumulation of net negative charge towards the N-terminus, and recovery to near-neutrality closer to the YTH domain, resulting in creation of electric dipoles (Figs 5E and S19D). The IDRs of ECT1, ECT9 and ECT11 differ from this pattern, but in distinct ways: ECT1 does not show a pronounced patterning of the charged residues, while ECT9 increasingly accumulates negative charge (Fig 5E) and ECT11 increasingly accumulates positive charge (S19D Fig). It is an interesting possibility, therefore, that charge separation in the IDR is a requirement to fulfill the molecular functions of ECTs to accelerate growth in primordial cells.

The molecular function of ECT2/3/4 was present in the first land plants To test whether the common ancestor of plant YTHDF proteins indeed had a molecular function similar to that of the modern Ath ECT2/3/4, we subjected Mpo DFE (Fig 2A) to our functional assay, as bryophytes are the earliest-diverging taxon of land plants (Fig 1A, [54,56–60]). The result showed a partial, but clear capacity of two different splice forms of Mpo DFE to complement the delay in leaf emergence and morphology defects of arabidopsis te234 plants (Figs 6B and S7D, S20A–S20B). Although the complementation scores were low (Fig 6B), a thorough inspection of the fluorescence intensity in all primary transformants revealed that the transgene was only expressed very weakly in a reduced subpopulation of plants, possibly due to the tendency to trigger silencing upon heterologous expression of a long cDNA from a distant plant species [86]. We therefore introduced the best-complementing isoform into rdr6-12/te234 mutants. This approach yielded a much higher complementation score of 52% (Fig 6C), and plants very similar to those expressing ECT2 in the same background (Figs 6D and S6B). Thus, the Ath ECT2/3/4 activity needed to stimulate proliferation of primed stem cells was present in the first embryophytes, suggesting that this activity is an ancestral molecular function of YTHDF proteins in land plants. Gratifyingly, the simulations of the behavior of the IDR of Mpo DFE showed a dipole-like charge pattern and a phase separation propensity similar to those of the ECT2/3/4/5/6/7/8/10 group, further substantiating that these IDR characteristics correlate with the ability to fulfill ECT2/3/4 function in leaf formation (Fig 6E–6H). We investigated this interesting possibility more deeply by analysing the IDRs of YTHDF proteins from species representing the main clades of land plants that diverged prior to flower evolution, a moss, a lycophyte, a fern and a gymnosperm, and from the sole species in the sister group to all flowering plants, A. trichopoda [87, 88]. This analysis showed that at least one YTHDF protein with ECT2-like charge distribution and phase separation propensity is present in all these taxa (S21A Fig). In particular, this included the only YTHDF protein in the moss Ceratodum purpureus. Interestingly, when the proteins from the different taxa were analysed together, a general trend for increased phase separation propensity that again correlated with higher ‘stickiness’ (λ) was apparent in proteins of clades -A, -B and -E compared to those of -C, -D and -F (S21B Fig). We also noticed that in the species that encode multiple YTHDF proteins, some paralogs exhibit divergent IDR features similar to those seen in Ath ECT1/9/11, but this phenomenon was not clade-specific (S21 Fig). This result suggests that the IDR-based functional diversification observed in arabidopsis may apply more generally. We conclude that the molecular function of YTHDF proteins required for stimulation of primordial cell proliferation is deeply conserved in land plants, and that the biophysical properties of the IDRs of Mpo DFE and the Ath DF proteins capable of performing this function are recurrent in all the major plant taxa.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1010980

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.