Root diffusion barrier control by a vasculature-derived peptide binding to the SGN3 receptor

See allHide authors and affiliations

Science  20 Jan 2017:
Vol. 355, Issue 6322, pp. 280-284
DOI: 10.1126/science.aaj1562

Keeping roots water-tight

The Casparian strip provides a waterproofing function to plant roots, protecting them against unregulated influxes of water and minerals. The integrity of the Casparian strip depends on a receptor-like kinase. Doblas et al. and Nakayama et al. now identify the peptide ligands in the core of the root (the stele) that help regulate Casparian strip formation. The receptor is expressed on the outward-facing surface of the root endodermal cells that surround the stele. When the endodermal layer is sealed by the Casparian strip, the peptide ligands cannot reach their receptors. When the endodermal layer is breached, whether by damage or during development, the peptides reach their receptors and activate signaling that encourages lignin deposition, shoring up the strips.

Science, this issue p. 280, p. 284


The root endodermis forms its extracellular diffusion barrier by developing ringlike impregnations called Casparian strips. A factor responsible for their establishment is the SCHENGEN3/GASSHO1 (SGN3/GSO1) receptor-like kinase. Its loss of function causes discontinuous Casparian strips. SGN3 also mediates endodermal overlignification of other Casparian strip mutants. Yet, without ligand, SGN3 function remained elusive. Here we report that schengen2 (sgn2) is defective in an enzyme sulfating peptide ligands. On the basis of this observation, we identified two stele-expressed peptides (CASPARIAN STRIP INTEGRITY FACTORS, CIF1/2) that complement sgn2 at nanomolar concentrations and induce Casparian strip mislocalization as well as overlignification—all of which depend on SGN3. Direct peptide binding to recombinant SGN3 identifies these peptides as SGN3 ligands. We speculate that CIF1/2-SGN3 is part of a barrier surveillance system, evolved to guarantee effective sealing of the supracellular Casparian strip network.

The endodermis and its Casparian strip are the major extracellular diffusion barrier in the root of all higher plants and are crucial to plant nutrition (1). Individual Casparian strips fuse into a supracellular network that seals the endodermal cell layer, protecting the central vasculature. Their function is similar to tight/adherens junctions of animal epithelia, yet their molecular makeup is distinct. CASPARIAN STRIP MEMBRANE DOMAIN PROTEINS (CASPs) were the first factors responsible for Casparian strip formation to be identified (2). CASPs form stable transmembrane scaffolds [the Casparian strip membrane domain (CSD)] that localize lignin biosynthetic enzymes for Casparian strip formation (3). A forward genetic screen for endodermal barrier mutants (35) identified SCHENGEN3 (SGN3), a leucine-rich repeat receptor-like kinase (LRR-RLK) also known as GASSHO1 (GSO1). SGN3 displays a restricted localization within the endodermal plasma membrane, surrounding the forming individual CASP microdomains (4). Its mutant phenotype—discontinuous yet stable, functional, and correctly aligned CASP domains—suggests a role for SGN3 in promoting growth and fusion of CASPs. It is assisted in this by SCHENGEN1 (SGN1), a receptor-like cytoplasmic kinase whose mutant displays very similar, albeit weaker, phenotypes to SGN3 (5). Subcellular localization of SGN1 is distinct from that of SGN3, as it is observed exclusively at the outer (cortex-facing) plasma membrane domain of the endodermis. SGN3 and SGN1 distributions overlap at the outer edge of the Casparian strip domain, where they might constitute a localized signaling module of enhanced activity. To grasp the logic of this signaling pathway, it was crucial to identify a ligand for SGN3. Among the more than 200 LRR kinases in Arabidopsis, few cognate receptor-ligand pairs have been identified (6, 7). Biochemical characterization of a number of receptor-ligand pairs has revealed that all investigated LRR receptors require ligand-contacting co-receptors for high-affinity binding and activation (8).

sgn2, a third mutant with a similar phenotype to sgn3 and sgn1 (Fig. 1, A to C), provided a lead as to the identity of the SGN3 ligand. By rough mapping, whole-genome sequencing, complementation, and analysis of independent transferred DNA insertion alleles, we identified a splice acceptor mutation in the seventh intron of At1g08030 as the causal mutation for sgn2 (fig. S1). At1g08030 encodes the only tyrosylprotein sulfotransferase (TPST) in Arabidopsis; this enzyme is responsible for sulfation of peptide ligands (e.g., RGFs, PSK, PSY1) that promote root meristem maintenance and growth (7, 9). Consistent with the above, sgn2-1 shows a root growth phenotype that is similar to yet weaker than tpst-1 (hereafter referred to as sgn2-2) (fig. S1, G to I). Although we reproduced the root growth complementing activities of the known peptides (fig. S2, B to D), PSY1, PSK, and RGF peptides did not rescue the CASP localization defects of tpst/sgn2 (fig. S2E). In addition, neither mutants of RGF genes nor those of PSK or PSY receptors displayed any indication of an endodermal barrier defect (fig. S2A). This indicated that Casparian strip defects of tpst/sgn2 are not due to the lack of activity of any known and characterized sulfated peptides but rather are caused by an as yet unknown TPST substrate.

Fig. 1 A stele-produced substrate of Arabidopsis tyrosylprotein sulfotransferase is required for Casparian strip formation.

(A) Impaired diffusion barrier in sgn2-1, visualized by the presence of propidium iodide (PI) in the stele (median longitudinal view) at 20 cells after the onset of cell elongation. (B) Surface view maximum projections of the Casparian strip membrane domain (CSD) network, visualized with CASP1-GFP. (C) Surface view maximum projections of Casparian strip (CS) autofluorescence. Note the discontinuities in the mutants (arrowheads) in (B) and (C). The spiral-like signal is from a deeper-lying xylem vessel. (D to J) Analysis of the WT, sgn2-2, and sgn2-2 complemented with mCherry–tyrosylprotein sulfotransferase (TPST) under TPST (D), UBQ10 (E), WOX5 (F), SCR (G), CASP1 (H), CO2 (I), and SHR (J) promoters. (Top panels) Primary root length of the respective lines in comparison with the WT and sgn2-2. Values are relative to the WT. (Middle panels) Quantification of PI permeability in the lines described. (Bottom panels) Schematic of promoter line expression in the root meristem (hues indicate expression intensity) and summary of results. Data represent mean values and SD (error bars) (n = 10 roots). Different letters (a, b, and c) indicate statistically significant differences [P < 0.01, analysis of variance (ANOVA) and Tukey test]. Scale bars: (A) to (C), 20 μm. ep, epidermis; co, cortex; en, endodermis; st, stele.

To establish the tissue in which the unknown sulfated peptide is produced, we expressed a complementing fusion of TPST-mCherry under various constitutive or cell type–specific promoters (Fig. 1, D to J, and fig. S3). By expressing TPST-mCherry under the meristem center–specific WOX5 promoter (10), we could separate the action of TPST on root meristem activity from that of Casparian strip formation because root growth was rescued in this line, although formation of Casparian strips was not (Fig. 1F). Expression from the SCR promoter, which drives expression in meristem center, proendodermal, and endodermal cells (11), also did not rescue the Casparian strip yet did rescue root growth, similar to the WOX5 line (Fig. 1G). Expression from the CASP1 promoter, active specifically in differentiating endodermis (2), rescued neither root growth nor Casparian strip defects (Fig. 1H). The two latter experiments establish that the putative peptide is not being produced in the endodermis itself. We therefore asked whether expression in cell layers neighboring the endodermis could rescue the Casparian strip defect. No rescue was observed expressing TPST-mCherry in the cortex, the outer neighbor of the endodermis (12) (Fig. 1I). By contrast, expression from the stele-specific SHR promoter (11) rescued the Casparian strip defect of tpst/sgn2 (Fig. 1J). We therefore postulated that a stele-produced peptide promotes barrier formation in the endodermis.

Through database searches for putative, stele-expressed peptides with a DY consensus motif, we identified At4g34600 as a protein that had been reannotated as a putative small secretory peptide with an activity on lateral root emergence (13). Moreover, endodermal differentiation genes appeared to be coexpressed with At4g34600 (table S1). We synthesized sulfated and nonsulfated variants of the predicted peptide [named CASPARIAN STRIP INTEGRITY FACTOR 2 (CIF2)] (table S2). When either WOX5 or SCR promoter-complemented lines of sgn2 (used instead of sgn2 because of their improved root growth) were treated with 1 nM peptide, both propidium iodide (PI) permeability and discontinuities of the Casparian strip were rescued (Fig. 2, A and B, and fig. S4, A and B) and partial complementation was still found in the picomolar range. Starting at concentrations of 100 nM, a mislocalization of CASP1–green fluorescent protein (GFP) into ectopic microdomains was observed at the cortex-facing plasma membrane upon treatment in the wild type (WT) (Fig. 2C). In addition, an endodermis-specific overaccumulation of lignin was observed (Fig. 2D). Colocalization of CASP1-GFP with lignin revealed that the strict co-incidence of CASP1-GFP and lignin formation (Fig. 2E) was broken upon peptide treatment (Fig. 2F), leading to lignin accumulation at the edges of the CASP domain, as well as in the cell wall corners and edges between the endodermis and cortex (Fig. 2, G to I). The nonsulfated form of the peptide was also active, although with 10- to 100-fold lower effectiveness (fig. S4, C and D). Using promoter fusions of the CIF2 encoding gene, as well as its closest homolog (called CIF1), we confirmed vascular expression of both genes and observed the onset of CIF2 expression in vascular tissues just at onset of CASP1 expression in the endodermis (Fig. 2, J and K, and fig. S4E), consistent with a role of these peptides in promoting endodermal barrier integrity.

Fig. 2 CIF2 peptide is expressed in the stele and affects CASP1 localization and lignin deposition in the endodermis.

(A) Quantification of PI permeability in the WT without peptide and sgn2-2 complemented with mCherry-TPST under WOX5 promoter, grown without or with CIF2. Data represent mean values and SD (error bars) (n = 10). Different letters indicate statistically significant differences (P < 0.01, ANOVA and Tukey test). (B) Surface view maximum projections of CS staining at the 20th endodermal cell after the onset of elongation. Lines and conditions are as in (A). Discontinuities in lignin deposition in the sgn2-2 line disappear upon the addition of 1 nM CIF2 (arrowheads). (C and D) Surface view maximum projections of the CSD network, visualized with CASP1-GFP and CS-stained in the WT at the 10th endodermal cell after CASP1 expression, grown with or without CIF2. (E and F) Surface confocal sections showing CASP1-GFP, CS-stained and colocalized with or without CIF2. (G and H) Radial optical sections showing CASP1-GFP (green) localization with or without CIF2. Cleared root stained with Fuchsin (red, lignin) and Calcofluor White (blue, cell wall). (I) CASP1-GFP completely colocalizes with lignin in the WT, but ectopic lignin is seen after CIF2 treatment. Staining is as in (G) and (H). (J) Expression analysis of CIF2 and CIF1 promoters with NLS-GFP confirms stele-specific gene expression. Reporter lines were counterstained with PI (red) and nuclear GFP signals (fire LUT of the ImageJ software). Note that pCIF1 shows weaker expression than pCIF2. (K) Quantification of pCASP1 and pCIF2 onset of expression (n = 5). Error bars indicate SD. Scale bars: (B) to (D), 20 μm; (E), (F), and (I), 5 μm; (G) and (J), 25 μm.

Endodermal phenotypes of sgn2 and sgn3 are similar, and the double mutant does not enhance sgn3 phenotypic severity, suggesting that they act in a linear pathway (Fig. 3A) and making SGN3 an evident receptor candidate. When treating different sgn3 mutants with high peptide concentrations (1 or 10 μM), we found that all peptide-induced phenotypes were absent in sgn3 (Fig. 3, B to D, and fig. S5, A and B). Moreover, peptide application induces disappearance of the SGN3-mVenus signal from the plasma membrane over time scales reported for ligand-induced endocytosis (Fig. 3E). The total resistance toward 10 μM of peptide, together with peptide-induced degradation of SGN3, suggested that we had identified SGN3 ligands. For consistency with the accompanying paper from Nakayama et al. (14), we designated At4g34600 as CIF2 and At2g16385 as CIF1. We discovered that CIF2 also induces enhanced and precocious suberization in a sgn3-dependent fashion (Fig. 4F and figs. S5C and S7E). To demonstrate direct, physical interaction between SGN3 and CIF1/2, we expressed the SGN3 ectodomain from insect cell culture and purified it to near homogeneity (fig. S6A). In isothermal titration calorimetry (ITC) experiments, CIF2 was found to bind with low nanomolar affinity, whereas CIF1 bound with weaker (26 nM) affinity (Fig. 3G and table S4). No substantial binding was detected with nonsulfated CIF2 (fig. S6B). To further establish ligand binding specificity, the fluorescently-labeled SGN3LRR domain was used in microscale thermophoresis (MST) assays. In these assays, CIF2 associates with the SGN3LRR domain with midnanomolar affinity [dissociation constant (Kd) CIF2 ≈ 300 nM]. Nonsulfated CIF2 peptide also binds, albeit with a Kd in the micromolar range (≈4 μM) (fig. S6C), consistent with a low but measurable in vivo activity of nonsulfated CIF2. The discrepancies in Kd values obtained by MST or ITC might be due to fluorescent labeling steps required for MST. We observed a complete lack of binding of the flg22 peptide (binding LRRs domains of FLS2) to SGN3, further demonstrating binding specificity (fig. S6D). Nanomolar binding affinity of the SGN3 ectodomain to CIF2 in ITC assays matches the in vivo activity of externally applied peptide and is also maintained in assays at pH 5.8 (fig. S6E). Compared with previous LRR ectodomain–ligand interactions, nanomolar Kds of SGN3 with CIF2 are unusual, as they are measured in the absence of co-receptors. When measured, ligand binding of LRRs without co-receptors is about two orders of magnitude lower than for SGN3 (15). According to current paradigm, ligand activation of receptors through assembly of a high-affinity ternary complex of receptor, co-receptor, and ligand is required for signal transduction. Our findings suggest that, although probably widespread, this mechanism might not be ubiquitous and that activation of SGN3 by CIFs could represent a different mode of activation.

Fig. 3 CIF peptides are ligands of the LRR-RLK SGN3.

(A) Quantification of the number of discontinuities per 100 μm of CS in the WT, sgn2-1, sgn3-3, and the double mutant. Data represent mean values and SD (error bars) (n = 10). Different letters indicate statistically significant differences (P < 0.01, ANOVA and Tukey test). (B) Surface view maximum projections of CSD network, visualized with CASP1-GFP at the 10th endodermal cell after CASP1-GFP expression in the WT and sgn3-3 with or without CIF2 or CIF1. (C) Surface view maximum projections of CS staining in the lines and conditions described in (B). (D) Surface confocal sections showing CASP1-GFP and CS staining in the lines and conditions described in (B). (E) Surface view of SGN3-mVenus and CASP1-mCherry during 2 hours in liquid medium without and with CIF2. Note the disappearance of the double strip pattern of SGN3 surrounding the CASP domains (red) upon treatment with CIF2. SGN3 is marked with relative intensity. (F) Epifluorescence images of endodermal suberin in the WT and sgn3-3 in the absence or presence of CIF2. (G) Quantification of CIF2 (top) and CIF1 (bottom) binding affinity with SGN3LRR by isothermal titration calorimetry (ITC). The dissociation constants (Kd) and the stoichiometries (N) are indicated. Similar results were obtained in at least three independent experiments. Scale bars: (B) and (C), 20 μm; (D), 5 μm; (E), 10 μm; (F), 100 μm.

Fig. 4 SGN1 receptor-like cytoplasmic kinase is necessary for CIF2 signalization.

(A and B) Surface view maximum projections of the CSD network, visualized with CASP1-GFP and CS-stained at the 10th endodermal cell after CASP1 expression in the WT and sgn1-2, grown with or without CIF2. Note that CASP1-GFP ectopic localization is weakly produced by CIF2 in sgn1-2 (arrowheads). (C) Surface confocal sections showing CASP1-GFP and CS staining in the lines and conditions described in (A) and (B). (D) Epifluorescence images of endodermal suberin in the WT and sgn1-2 in the absence or presence of CIF2. (E) Schematic model of CIF1/2 action. Scale bars: (A) and (B), 20 μm; (C), 5 μm; (D), 100 μm.

Recently, we described a receptor-like cytoplasmic kinase, SGN1, as a putative SGN3 downstream kinase (5). Indeed, even application of 1 μM CIF2 is not able to complement the SGN3-like discontinuities of the sgn1 mutant (Fig. 4, A and B, and fig. S7, A and B). Moreover, sgn1 displays partial resistance to peptide-induced overlignification and oversuberization (Fig. 4, C and D, and fig. S7, C and D), supporting the idea that SGN1 is required to transduce CIF-activated SGN3 signals.

SGN1 is observed exclusively at the outer, cortex-facing side of the endodermal plasma membrane (5). CIF peptides, by contrast, are produced in inner, stelar cell layers. We therefore propose a model that explains both the necessity for the spatially restricted SGN3 and SGN1 colocalization and the biological meaning of the CIF-SGN3-SGN1 signaling module. Because of the cortex-facing localization of SGN1, the CIF-activated SGN3 receptor would only be able to signal through SGN1 as long as the extracellular diffusion barrier of the Casparian strip is not established (Fig. 4E). Once sealed, CIFs will be contained within the stele by the Casparian strip network and no longer able to reach the domain where SGN3 and SGN1 can colocalize and interact. This would lead to an absence, or strong reduction, of signaling and would report to the endodermis that the diffusion barrier has been established. We propose that CIF peptides represent a barrier surveillance system, evolved to ensure that CASP domain growth and lignification proceeds until complete closure and possibly to reactivate lignification and suberization whenever the endodermal diffusion barrier becomes damaged or otherwise compromised. Supporting this model are our previous observations that the prominent and unexplained overlignification and suberization syndrome observed in many different endodermal barrier mutants (4, 16, 17) is suppressed in sgn3 mutants. Our model now explains this overlignification and suberization of Casparian strip mutants: In all Casparian strip mutants, CIF peptides would escape from the stele, leading to continuous activation of the SGN3-SGN1 pathway, causing overlignification and suberization—similar to what is elicited by exogenous peptide treatments—and thus sealing the endodermal barrier. Such a plant diffusion barrier surveillance system would represent an intriguing case of convergent evolution, as findings in animal epithelia suggest the existence of similar barrier surveillance systems, based on the spatial separation of epidermal growth factor ligands and receptors toward the basolateral and apical sides of the epithelium, respectively (18, 19).

Supplementary Materials

Material and Methods

Figs. S1 to S8

Tables S1 to S4

References (2027)

References and Notes

  1. Acknowledgments: This work was supported by funds to N.G. from the European Research Council (Plant-Memb-Traff and ENDOFUN) and the Swiss National Science Foundation and to Y.B. from the Austrian Academy of Science through the Gregor Mendel Institute. V.G.D. was supported by the Fundación Alfonso Martín Escudero, M.B. by an European Molecular Biology Organization long-term postdoctoral fellowship, and S.F. by a Japan Society for the Promotion of Science postdoctoral fellowship for research abroad. We thank the Central Imaging Facility and the Genome Technology Facility of the University of Lausanne as well as the Vienna Biocenter Core Protein Technologies Facility for assistance and technical support. We thank T. Clausen for access to his ITC platform. The supplementary materials contain additional data.
View Abstract

Stay Connected to Science

Navigate This Article