Musashi-like MSIL2/4 proteins redundantly control development in Arabidopsis.

(A) Schematic representation of the Musashi-like (MSIL) protein family in Arabidopsis thaliana. The invariant RNP1 and RNP2 motifs within the conserved RRM domains are indicated. Numbers refer to amino acid identities between the Mouse Musashi/MSI RRM1/2 domains and the corresponding domains in Arabidopsis MSIL homologs. (B) Evolutionary relationships between MSILs and related RNA-binding proteins. The scale bar indicates the rate of evolutionary change expressed as number of amino acid substitutions per site. (C) RNA-seq expression map of MSIL genes extracted from ARAPORT11. Shading is a log2 scale of transcripts per million (TPM). (D) General overview of the roots of 6-days-old wild type Arabidopsis seedlings (Col-0) or transgenic seedlings expressing either a free GFP protein or GFP-tagged versions of MSIL2/MSIL2G and MSIL4/MSIL4G. Scale bar, 10 μm. (E) Photographs of representative rosettes of Col-0, msil2/4, msil2/4-MSIL2F1 and msil2/4-MSIL4F1 plants. (F) Photographs of representative inflorescence stems of Col-0, msil2/4, msil2/4-MSIL2F1 and msil2/4-MSIL4F1 plants. Abbreviations: Mus musculus Musashi2 (MmMsi2); human heterogeneous nuclear ribonucleoprotein A1 (hnRNPA1).

RRM-dependent RNA-binding activity is essential for MSIL function in planta.

(A) Experimentally determined structures of the RRM1 domain of MmMSI2 (PDB ID: 6C8U) and the RRM2 domain of MmMSI1 (PDB ID: 5x3y), and models of the RRM domains of MSIL4 generated using the homology-modeling server SWISS-MODEL. The α-helices and β-sheets of RRM domain as well as the phenylalanine residues that contact RNA bases are highlighted. (B) Schematic representation of the mutations introduced in the RRM domains of MSIL4. (C) Binding assays of MSIL4 RRM domains on ssRNA homopolymers in vitro. Single-stranded polyA (pA), polyG (pG), polyC (pC) or polyU (pU) RNAs were subjected to binding with the His-tagged recombinant RRM domains from MSIL4 (upper part) or its mutated version (MSIL4RRM, lower part). An anti-His antibody was used for detection. In. represents the input fraction. (D) RRM-dependent RNA-binding activity is essential for MSIL2/4 function in stem. Photographs of representative inflorescence stems of Col-0 and msil2/4 mutants complemented or not with a WT version of MSIL4 (MSIL4G-8 and 3) or an rrm mutant (MSIL4GRRM-3 and 10).

MSIL2/4 protein interactomes are enriched in proteins involved in 3’-UTR binding and translation regulation.

(A) The semi-volcano plots show the enrichment of proteins co-purified with MSIL2F1 and MSIL4F1 as compared with control IPs. Y- and X-axis display adjusted p-values and fold changes, respectively. The dashed lines indicate the threshold above which proteins are significantly enriched/depleted (fold change > 2; adjP < 0.05). (B) Venn diagram showing the overlap between the MSIL2 and the MSIL4 interactomes. (C) PANTHER-based classification of the GO molecular function that are overrepresented among the MSIL-interacting proteins. (D) Confocal monitoring of the colocalization of MSIL2G, MSIL4G and PAB2R in root tips of 8-d-old seedlings were monitored after 30 min of exposure to 38°C (heat stress) or 20°C for control treatment. (E) Cycloheximide treatment inhibits the formation of MSILG and PAB2R cytosolic foci. The transgenic plants expressing the stress granule marker PAB2R was used as a control. Root of 7-day-old seedlings expressing either MSIL4G, MSIL2G, or PAB2R were monitored after 1 hour of exposure to 38°C in the presence of DMSO control treatment (DMSO) or in the presence of cycloheximide inhibitor (+CHX). Scale bar, 10 μm.

MSIL proteins regulate the molecular architecture of SCW in Arabidopsis fibers.

(A) Expression profiles of the MSIL2 and MSIL4 genes in the inflorescence stem tissues as retrieved from the Arabidopsis inflorescence stem tissue-specific transcriptome database40 (https://arabidopsis-stem.cos.uni-heidelberg.de/). The results of two replicates are shown. F, fibers; X, xylem vessels; Cx Cambium (xylem side); Cp, Cambium (Phloem side); P, Phloem; S, Starch sheath; E, Epidermis. (B) Cross-sections of Col-0 and msil2/4 stems showing vascular bundle (xy) and interfascicular fiber (if) cells stained with Toluidine blue. Representative fiber and xylary areas analyzed in the panel 4C are outlined. (C) Measurements of the SCW thickness of xylary fiber and interfascicular fiber cells. Values are means (n>400)± SEM. Data were analyzed by unpaired Student’s test. Asterisks indicate significant differences relative to Col-0; ****p<0.0001. (D) Top: lignin autofluorescence under ultraviolet (UV) using confocal microscopy in Col-0, msil2/4 mutant, and complemented msil2/4-MSIL2F1 and msil2/4-MSIL4F1 plant stem sections. xy: xylem fibers; if: interfascicular fibers. Bottom: Phloroglucinol-HCl staining of Col-0, msil2/4, and complemented msil2/4-MSIL2F1 and msil2/4-MSIL4F1 plant stem sections. Scale bar, 100 μm. (E) Boxplots represent acetyl bromide lignin content, expressed as % of dry weight. Lignin content was measured in the bottom section of mature stems (3 biological replicates per line). Significant differences between Col-0, msil2/4+MSIL2F1, msil2/4+MSIL4F1, and msil2/4 plant lines are indicated by asterisks*, P-value< 0.01 according to Student’s t test (n = 6 to 10). (F) Confocal microscopy of Calcofluor-white staining of cross sections of Col-0 and msil2/4 mutant inflorescence stems. Scale bar, 100 μm. (G) Boxplots represent the glucose yield after incubation of Col-0 and msil2/4 lignocellulosic material with cellulases either without pre-treatment (-), or after pre-treatment (+) with sodium hydroxide (NaOH).

The accumulation and activity of the glucuronoxylan decoration machinery are altered in msil2/4 mutant.

(A) MS-based quantitative comparison of Col-0 and msil2/4 inflorescence stem proteomes. Volcano plot displaying the differential abundance of proteins in Col-0 and msil2/4 cells analyzed by MS-based label-free quantitative proteomics. The volcano plot represents the - log10 (limma p-value) on y axis plotted against the log2 (Fold Change msil2/4 vs Col-0) on x axis for each quantified protein. Green and red dots represent proteins found significantly enriched respectively in msil2/4 and Col-0 Arabidopsis stems (log2(Fold Change) ≥ 1 and - log10(p-value) ≥ 2.29, leading to a Benjamini-Hochberg FDR = 1.01 %). The up-regulated proteins involved the glucuronoxylan biosynthetic pathways are indicated. (B) Gene Ontology (GO) analysis of the proteins that are significantly up-regulated in the msil2/4 mutant. The gene ontology analysis of DEGs was performed using ShinyGO v0.76 software. Lollipop diagrams provide information about GO fold enrichment, significance (FDR in log10), and number of genes in each pathway. (C) Schematic model of glucuronoxylan substitution patterns in Arabidopsis and the enzymes involved. IRX9/10/14, glycosyltransferases involved in the synthesis of xylan (Xy) backbone. ESK1, eskimo1. GUX1, glucuronic acid substitution of xylan1; GXM3, glucuronoxylan methyltransferase3 are involved in the glucuronoxylan decoration. (D) Neutral monosaccharide composition of alcohol insoluble residue (AIR) extracted from inflorescence stems of wild-type and msil2/4 mutant plants that have been pre-hydrolyzed or not with acid. Rha, rhamnose; Fuc, fucose; Ara, arabinose; Xyl, xylose; Man, mannose; Gal, galactose; Glc, glucose. (E) Left: Immunofluorescence labeling of transverse sections of Col-0 and msil2/4 stems with the LM10 and LM11 antixylan antibodies. Representative fiber and xylary regions analyzed by immunofluorescence are outlined. Xy, xylem fibers; if, interfascicular fibers. Right: Quantification of the fluorescence was done using ImageJ software and processed according to Supplementary file 5. Significant differences are indicated by asterisks*, P-value< 0.01 according to Student’s t test (n = 5 to 12)

MSILs restrain the degree of 4-O-methylation of glucuroxylan in Arabidopsis inflorescence stem.

(A) MALDI-TOF mass spectra of xylooligosaccharides generated by xylanase digestion of xylan from Col-0, msil2/4, and complemented msil2/4-MSIL2F1 and msil2/4-MSIL4F1 inflorescence stem materials. The ions at m/z 1705/1747 and 1719/1761 correspond to acetylated xylo-decapolysaccharides bearing a GlcA residu (GlcA-Xyl10-Ac4//GlcA-Xyl10-Ac5) or a methylated GlcA residue (MeGlcA-Xyl10-Ac4//MeGlca-Xyl10-Ac5). (B) Acidic monosaccharide composition of alcohol insoluble residue (AIR) extracted from inflorescence stems of wild-type and msil2/4 mutant plants that have been pre-hydrolyzed or not with acid. GalA, galacturonic acid; GlcA, glucuronic acid. (C) Relative changes in unmethylated/methylated GlcA decapolysaccharide ratio in Col-0 (black) and msil2/4 mutant (red) as controlled by the addition of external spike-in control corresponding to a pentaacetyl-chitopentaose. (D) GFP-based RNA-IP assays. Top panel (WB): western blots performed using antiGFP or antiUGPase antibodies on protein fractions from inputs or antiGFP immunoprecipitates from WT and the msil2/4 mutant plants expressing the WT (MSIL4G-3) or RRM mutant (MSIL4GRRM-3) MSIL4-GFP fusions. Bottom panel (RT) : Corresponding RT-PCR using GXM1 or GXM3 specific primers on RNA fractions.

Model of MSIL-dependent control of glucuronoxylan methylation in Arabidopsis and its consequence for SCW architecture.

In the interfascicular fiber cells of Col-0, MSIL2/4 restrain the translation of the glucuronoxylan biosynthesis enzymes, including the rate-limiting GXM3 enzyme. This activity would keep the level of glucuronoxylan methylation at an intermediate level, therefore providing a biochemical environment that favors the interactions between the glucuronoxylan and lignin polymers. In the interfascicular fiber cells of msil2/4 mutant, the translation of the glucuronoxylan biosynthesis machinery, including GXM3, is increased, leading to the deposition of an over-methylated form of glucuronoxylan that would be less prone to interact with lignin and establish SCW formation. In a non exclusive manner, MSIL2/4 could also have a positive role in lignin synthesis, whose defect in msil2/4 would lead to a decrease in lignin content. The model was inspired from Grantham et al., 2017.