ROS production by localized SCHENGEN receptor module drives lignification at subcellular precision

Production of reactive-oxygen species (ROS) by NADPH oxidases (NOXs) impacts many processes in animals and plants and many plant receptor pathways involve rapid, NOX-dependent increases of ROS. Yet, their general reactivity has made it challenging to pinpoint the precise role and direct cellular targets of ROS. A well-understood ROS target in plants are lignin peroxidases in the cell wall. Lignin can be deposited with exquisite spatial control, but the underlying mechanisms have remained elusive. Here we establish a full kinase signaling relay that exerts direct, spatial control over ROS production and lignification within the cell wall. We show that polar localization of a single kinase component is crucial for pathway function. Our data indicates that an intersection of more broadly localized components allows for micrometer-scale precision of lignification and that this system is triggered through initiation of ROS production as a critical peroxidase co-substrate.


27
processes in animals and plants and many plant receptor pathways involve rapid, NOX-dependent 28 increases of ROS. Yet, their general reactivity has made it challenging to pinpoint the precise role 29 and direct cellular targets of ROS. A well-understood ROS target in plants are lignin peroxidases 30 in the cell wall. Lignin can be deposited with exquisite spatial control, but the underlying 31 mechanisms have remained elusive. Here we establish a full kinase signaling relay that exerts 32 direct, spatial control over ROS production and lignification within the cell wall. We show that 33 polar localization of a single kinase component is crucial for pathway function. Our data indicates 34 that an intersection of more broadly localized components allows for micrometer-scale precision 35 of lignification and that this system is triggered through initiation of ROS production as a critical 36 peroxidase co-substrate. 37 As in animals, NADPH oxidase-produced ROS in plants is important for a multitude of 38 processes and the number of NADPH oxidase genes (10 in Arabidopsis, called RESPIRATORY 39 BURST OXIDASE HOMOLOGs, RBOHs, A-J) suggests a high complexity of regulation of ROS 40 production in plants. Among its many roles, ROS-dependent regulation of plant cell wall structure and 41 function is considered to be among its most critical 1 . The cell wall is the nano-structured, sugar-based, 42 2 pressure-resisting extracellular matrix of plants and NOXs are thought to be the predominant ROS 43 source in this compartment (also termed apoplast) 1 . 44 A staggering number of kinases have been shown to regulate plant NOXs and the activation 45 mechanism of NOX-dependent ROS production is well established, especially in response to microbial 46 pattern-recognition by immune receptors 2 . However, the specific role and direct molecular targets of 47 ROS during microbial pattern-recognition have remained elusive 3 . The same applies for the central 48 role of ROS in tip growing cells, such as root hairs or pollen tubes, where ROS is thought to be part of 49 an intricate oscillation of cell wall stiffening and loosening, aimed at allowing cell wall expansion 50 without catastrophic collapse 4,5 . In this case, ROS is proposed to be important for counteracting cell 51 wall loosening pH decreases, but it is again unclear what direct targets of ROS would mediate cell wall 52 stiffening. Cell wall lignification by apoplastic peroxidases, can therefore be considered as the most 53 well-established role of ROS, where the peroxidases themselves are the direct "ROS targets", using it 54 as a co-substrate for the oxidation of mono-lignols 6,7 . In the case of lignification, however, a 55 molecularly-defined signaling pathway that induces ROS production during lignification has not been 56 defined. A few years ago, our group identified a specific NADPH oxidase, RBOHF, to be required for 57 the localized formation of lignin in the root endodermis 8 . Lignin is a poly-phenolic polymer that is 58 generated by the radical-coupling of mono-lignols, oxidized through the action of ROS-dependent 59 peroxidases, as well as laccases 7 . The hydrophobic lignin polymer impregnates the cellulosic cell wall  More recently, we identified a pair of peptide ligands, a Leucine-rich repeat receptor-like 72 kinase (LRR-RLK) and a cytoplasmic kinase, whose phenotypes, genetic interaction and specific 73 subcellular localizations led us to propose that they combine into a barrier surveillance pathway.

74
Previous reports had shown that the CIF1/2 (CASPARIAN STRIP INTEGRITY FACTORs 1/2) 75 peptides are SCHENGEN3 (SGN3) (also called GASSHO1 (GSO1)) ligands and that the SGN1 and 76 SGN3 kinases govern Casparian strip integrity 15-18 . cif1 cif2 and sgn1, sgn3 mutants have similar, 77 3 discontinuous CS, caused by a discontinuous CSD, as well as a conspicuous absence, or strong 78 attenuation of, compensatory lignification and suberization observed in other CS mutants 13,15,[18][19][20][21] . 79 Their phenotypic similarities suggested that these factors act in one pathway. CIF1 and 2 peptides do 80 not express in the endodermis, where the CS is formed, but in the stele. In contrast, both SGN1 and 81 SGN3 present specific localization on the endodermal plasma membrane; SGN3 receptor-like kinase 82 resides along both sides of the CS, while palmitoylated SGN1 polarly localizes on cortex facing (outer) 83 plasma membranes 15 . Remarkably, their localization overlaps only in a small region next to the cortex-84 facing side of the CS 16 . This would require peptides from the stele to diffuse across the CS in order to 85 access the signaling complex. This is only possible while the CS is still permeable 15 (Fig. 1E). 86 This pathway would therefore allow the endodermal cell layer to probe the tissue-wide integrity 87 of its extracellular diffusion barrier and to respond to barrier defects through compensatory over-88 lignification 15 . Here we demonstrate that the receptor, cytoplasmic kinase and NADPH oxidase 89 molecularly connect into one pathway, with great resemblance to plant immune signaling pathways, 90 but whose direct action is to locally produce ROS for localized lignification. We establish the crucial 91 importance of the restricted subcellular localization of its components and demonstrate that stimulation 92 of this signaling pathway additionally leads to strong transcriptional activation of target genes that 93 further drive and sustain endodermal lignification, as well as suberization and endodermal sub-domain 94 formation and differentiation. We thus provide a full molecular circuitry in which an endogenous 95 peptide from the stele stimulates localized signaling kinases and NADPH oxidases in the endodermis, 96 causing extracellular ROS production at micrometer-scale precision and a precisely localized 97 lignification of the plant cell wall.

99
We generated a new cif1-2 cif2-2 double mutant allele by CRISPR-Cas9 in a pure Col  Central to the barrier surveillance model is the polar localization of SGN1, which is thought to 107 limit the potential for signal activation to the cortex side of the endodermis, requiring passage of CIF 108 peptides across the CS region (Fig. 1A,E). Consistently, application of peptide ligand to the media, 109 leading to stimulation from the outside, causes overlignification at the cortex-facing endodermal edges 110 (Fig. 1B). In an attempt to falsify the model we had proposed and to interrogate the importance of 111 4 SGN1 polar localization, we generated a SGN1 variant that localized in an apolar fashion, by adding a 112 myristoylation (myr) and palmitoylation (palm) motifs on the N terminus 22 . This myrpalm SGN1-113 mCitrine (Cit) variant was expressed under the control of the endodermis-specific CASP1 promoter, 114 which is strongly active during Casparian strip formation and complemented the sgn1 barrier 115 phenotype (Fig. S1A). In planta, the wild-type SGN1-Cit variant resides polarly on the cortex-facing 116 side of endodermal cells, while myrpalm SGN1-Cit was observed at both sides of the endodermal 117 plasma membranes, even though preferentially accumulation at the cortex side could still be observed 118 (Fig. 1C, 1D). Both variants were excluded from the central position where the Casparian strip domain 119 is formed (Fig. S1B) and the localization patterns of the two variants did not change when introgressed 120 into sgn1, sgn3 or cif1 cif2 mutants (Fig. 1C, 1D). An apolar SGN1 localization would allow SGN3 to 121 encounter SGN1 also on the stele-facing side, not only on the cortex side (Fig. 1E). This should lead 122 to constitutive signal activation in the absence of barrier defects, because the CIF peptides would now 123 be able to access a SGN3/SGN1 signaling module on the stele-facing side without crossing the barrier.

124
Indeed, we found that the apolar SGN1 variant caused both, ectopic lignin deposition and precocious 125 suberization in endodermal cells (Fig. 1D, S1F, G), as previously described for endodermal barrier 126 mutants. Yet, no barrier defect was observed in the lines complemented with apolar SGN1 (Fig. S1A) 127 and, consistently, we found CASP1-mCherry distribution to be normal in these lines, forming a Previous data showed that sgn1 is less sensitive to high doses of externally applied CIF peptide,   connection between the two events that would allow to maintain spatial information. In a cell primed 176 for lignification, i.e. with mono-lignol substrates available and polymerizing enzymes expressed, 177 lignification could be simply "switched on" by activating ROS production through NADPH oxidases.

178
Previously, we had found one of the NOXs, RBOHF, to be crucial for lignification at the CS. 179 Intriguingly, RBOHF is the only transmembrane protein known to accumulate at the Casparian strip 180 domain, safe the CASPs themselves 8 (Fig. 3A). Moreover, homologous NADPH oxidases, such as 181 RBOHB or RBOHD, also present in the endodermal plasma membrane, are excluded from this domain 182 8 (Fig. 3A). We therefore asked whether CIF peptides would induce lignification through activation of 183 6 RBOHF as a downstream component. To our surprise, CIF treatment still led to induced lignification 184 in rbohf, despite the fact that RBOHF, is strictly required for CS lignification in untreated conditions. 185 Yet, when RBOHD was knocked-out in addition to RBOHF, a complete absence of lignification upon 186 CIF treatment was observed (Fig. 3B). The fact that RBOHD single mutant neither showed defects in 187 developmental CS lignification, nor in CIF-induced lignification, indicates that RBOHF is required for 188 both processes, but that upon strong stimulation with exogenous CIF peptide application, RBOHD can 189 be additionally used. Indeed, using the general NADPH oxidase inhibitor diphenyleneiodonium (DPI) 190 in short-term co-treatment with CIF was also able to fully block CIF-induced ectopic lignification in 191 wild-type with intact Casparian strips, further supporting that NADPH oxidases act downstream in the 192 CIF/SGN3 pathway (Fig. S3A). The direct activity of NOX enzymes is not lignification, but production of superoxide (O2 -) that 197 becomes dismutated to hydrogen peroxide (H2O2) in the apoplast 1 . We had previously established that production, we developed a procedure to quantitatively assess cerium precipitates and checked whether 211 this localized ROS production is indeed dependent on the SGN3 pathway (Fig. 4D,E, S4A, see also 212 Experimental Procedures section). For SGN3, we found that already steady-state ROS production was 213 undetectable in the mutant and that there was no increase upon CIF2-treatment (Fig. 4D,E). The sgn1 214 mutant showed lower, but still detectable steady state ROS levels, but no significant increase upon 215 CIF2-treatment (Fig. 4D,E). Thus, the highly localized ROS accumulation induced by CIF2 is entirely 216 dependent on the localized SGN3/SGN1 receptor module. As in the case of lignification, we observed 217 that CIF2-induced ROS can be produced by either RBOHF or RBOHD, as only the double mutant 218 caused a complete absence of ROS after CIF-stimulation. As expected, but not previously demonstrated, 219 7 the steady-state ROS production at the CS observed before stimulation was exclusively dependent on 220 RBOHF, but not RBOHD (Fig. 4D,E).

222
In contrast to lignin polymerization, suberin might not require NADPH oxidase activity. It was 223 shown previously that CIF2 triggers excess suberization in WT (Fig. S3B) 15 . We found excess suberin 224 deposition in rbohD and rbohF upon peptide treatment, although a slight enhancement is already 225 observed in rbohF. In rbohDF, excess suberin deposition is very strong, even without treatment, likely 226 due to a strong activation of the surveillance system due to complete absence of a CS (Fig. S3B). These  The above data strongly suggest a direct connection between the SGN3/SGN1 kinase module 234 and the two NOX enzymes. We therefore conducted an in vitro kinase assay in order to ask whether 235 SGN1 can directly phosphorylate the N-terminal cytoplasmic region of RBOHF and RBOHD. We 236 found that recombinantly expressed TF-SGN1 could phosphorylate both the recombinant N-terminal 237 part of RBOHF and RBOHD (Fig. 5A). Plant NOX regulation has been intensively studied and shown 238 to be highly complex, requiring possibly interdependent activities of kinases, as well as small GTPases 239 26 . We therefore tested whether SGN1 might be sufficient for activation of RBOHF activity in a cellular 240 context. To do so, we made use of a heterologous reconstitution system in human HEK293T cells, 241 which show very low endogenous ROS production and for which it had been previously demonstrated 242 that plant NADPH oxidases can be expressed and their activation mechanism be studied 27 . As a 243 positive control the previously described calcium-dependent kinase complex of calcineurin B-like 244 (CBL) interacting protein kinases 26 (CIPK26) and CBL1 was used and shown to be active (Fig. 5B).

245
When expressing wild-type SGN1 in this cell line, we noticed that it did not activate RBOHF or 246 RBOHD, but that it also did not localize to the plasma membrane as in plant cells (Fig. S5A). However, 247 when we used the functional, constitutively plasma membrane-localized myrpalm-SGN1 version, a 248 significant induction of ROS production was observed for RBOHF and to a lesser extent for RBOHD 249 (Fig. 5B, S5B). The direct phospho-relay from SGN3 receptor, to SGN1 kinase to RBOHF and RBOHD 254 outlined above draws a direct molecular connection from perception of a peptide hormone stimulus to 255 8 cell wall lignification. Moreover, it accounts for both, the highly localized ROS production that we 256 observe upon CIF stimulation, as well as the observation that localization of lignification is correlated 257 with the site of active SGN3 signaling. Yet, the massive enhancement of lignification observed upon 258 CIF stimulation (Fig3B), the increase of CASP accumulation and ectopic patch formation (Fig. S2A), 259 as well as the non-localized formation of precocious and enhanced suberin (Fig. S3B,C), are additional  (Fig. 6A) 15,18 . Neither the single, nor the double NADPH oxidase mutants display 266 discontinuous CASP1-GFP signals, indicating that ROS production is not required for this aspect of 267 the CIF/SGN3 pathway (Fig. 6A). In order to directly demonstrate that formation of a continuous 268 CASP domain requires newly formed gene products, we treated our cif1 cif2 double mutant with CIF2 269 peptide in presence or absence of protein synthesis inhibitor cycloheximide (CHX). CASP1-GFP signal 270 strongly increased during 8 hours of CIF treatment, during which the discontinuous CASP1-GFP 271 domains became continuous. This effect was abrogated by CHX treatment (Fig. 6 B,C, Movie S1).

273
The NADPH oxidase-independent branch of the CIF/SGN3 pathway is associated with MAP kinase 274 stimulation and causes strong activation of gene expression.

276
In pattern-triggered immune receptor signaling, gene activation is thought to depend in large 277 parts on activation of Mitogen-activated protein (MAP) kinases 28 . We therefore tested whether CIF-278 treatment leads to MAP kinase phosphorylation and indeed found that CIFs can induce MAP kinase 279 phosphorylation in a SGN3-dependent manner, further extending the molecular parallels between 280 immune receptor signaling and the CIF/SGN3 pathway and suggesting that gene induction in the CIF 281 pathway might equally depend on MAP kinase signaling (Fig. S6B).

283
We then undertook an RNA profiling of seedling roots at different time points (30, 120 and 284 480 min) after CIF stimulation, using wild-type, cif1 cif2 and sgn3 mutants as genotypes. 930 genes 285 were found to be differentially expressed across any of the genotypes and time points combined, using 286 a stringent cut-off (adj.-pval. <= 0.05 ; logFC >= 1 or logFC<=-1 ) (Table S1). After normalization of 287 batch effects, the three replicates clustered closely with a large degree of variance explained by the 288 peptide treatment in wild-type and cif1 cif2 (Fig. S6A). Wild-type and cif1 cif2 samples showed nearly 289 identical responses (co-relation efficiency values were 0.90, 0.89, 0.94 at 30, 120, 480 min respectively), 290 with cif1 cif2 displaying a slightly stronger overall amplitude (Fig. 6D). Importantly, sgn3 had virtually 291 9 no differentially expressed genes across treatments, indicating the specificity of the CIF responses and 292 further corroborating that SGN3 is the single, relevant receptor for CIF responses in roots (Fig. 6D,

293
S6A, Table S1). Our data extends on the previous data of Nakayama et al., by showing that all 5 CASP 294 genes are differentially regulated upon CIF-treatment (Fig. S6C). Moreover, we observed upregulation 295 of MYB36, a central transcription factor for Casparian strip formation and CASP expression, thus 296 potentially accounting for the increases in CASP1-4 expression 20,29 (Fig. S6C). The fact that CIF 297 stimulates MYB36 expression also is consistent with the recent report that CIF treatment can enhance 298 ectopic endodermal differentiation, driven by overexpression of the SHR transcription factor 30 . indicates that many of the most significant, overrepresented terms in the "early response" (1) and 305 "strong and sustained" (2) gene clusters are related to immune and defense responses (response to 306 chitin, bacterium, callose, etc.), as well as responses to oxidative stress (Table S2). The top GO 307 categories of the late response cluster (3) were suberin/cutin biosynthesis, supporting our observation 308 that suberin accumulation is a late effect of CIF stimulation (Fig. S6D, S3B,C) 15 . Additional, enriched 309 terms were related to oxidative stress and cell wall remodeling, nicely fitting the CIF2-induced ROS 310 production we observed. Intriguingly, the cluster showing late downregulation of gene expression 311 contains many overrepresented GO terms related to a wide variety of transport processes, reaching 312 from water, nitrate and ammonium transport, to primary and secondary metabolite or hormone 313 transport (Fig. 6D, Table S2). We interpret this as a consequence of CIF-induced lignification and 314 suberization, which must profoundly impact endodermal ability for shuttling organic and inorganic 315 compounds, as well as signaling molecules between stele and cortical root cell layers 10 .

317
While GO terms related to lignification were also overrepresented, they were not as highly  (Fig. S6E,F). Moreover, MYB15, a transcription factor shown to be involved in stress and 327 MAMP(microbe-associated molecular pattern)-induced lignification 31 is about 6-fold induced after 30 328 min and about 17-fold after 120 min in both treated wild-type and cif1 cif2, nicely correlating with the 329 upregulation of peroxidases and laccases (Fig. S6E,F, Table S1). We then made use of two of the most 330 highly differentially expressed genes upon CIF2-treatment (PER15 and PER49) in order to establish 331 whether SGN1 and the NADPH oxidases are also required for gene regulation downstream of SGN3 332 activation. In qPCR analysis both peroxidase genes still showed a slight, but significant upregulation 333 upon CIF-treatment in the sgn1 mutant (compared to the complete absence of response in sgn3) (Fig.   334 6E-H), clearly indicating that, as for ROS production, SGN1 is an important, yet not absolutely NADPH oxidase mutants is in effect higher than in wild-type after CIF treatment, leading us to 343 conclude that ROS production and gene activation in response to CIF are two independent branches 344 of this signaling pathway, with the branching occurring downstream of the SGN1 kinase. We finally 345 wanted to know whether the mis-localization of the SGN1 kinase reported initially is only affecting 346 CIF/SGN3 signaling at the plasma membrane, or whether it also affects gene activation. We found that 347 both PER15 (Fig. 6I) and PER49 (Fig. 6J)  The data presented here delineate an entire signaling pathway. Previously, SGN3 had been 355 established as the receptor for CIF1 and 2, but its connection to SGN1 had exclusively been based on 356 genetic evidence. Here we show that SGN3, SGN1 and RBOHD/F are, biochemically and functionally, 357 part of a signal transduction chain, leading directly from localized peptide perception to localized ROS 358 production and lignification (Fig. 6K). Moreover, the pathway branches downstream of SGN1, leading 359 to MAP kinase activation and stimulation of gene expression. Some of the most strongly induced genes 360 being peroxidases and laccases that would further enhance and sustain lignification. The SCHENGEN 361 pathway therefore elegantly integrates fast, plasma membrane-based responses that maintain 362 positional information (ROS and lignin is produced close to where the ligand is perceived) and slower, 363 gene expression-based responses, which have lost positional information, but would allow to enhance 364 and maintain the ROS-burst-controlled activation of peroxidases (Fig. 6K).

366
The SCHENGEN pathway bears a striking overall resemblance to well-established signaling 367 pathways for perception of MAMPs. In MAMP perception, structurally similar receptor kinases, such 368 as FLS2 or EFR bind to microbial patterns and transduce this signal through kinases of the RLCKVII 369 family, such as BIK1 or PBLs, which are homologs of SGN1 32 . BIK1 in turn was shown to phosphorylate  The SCHENGEN pathway might not be limited to regulation of lignified diffusion barriers, 415 as it has been shown that it is also important in the formation of the embryonic cuticle. The peptide 416 ligand used in this context has not been identified and it remains unclear whether an equally precise 417 barrier surveillance mechanism is also acting to ensure separation of endosperm and embryo during 418 embryonic cuticle formation 36 . In this context, the SCHENGEN pathway might be used to drive 419 production and deposition of cutin instead of lignin and suberin as in the case of the endodermis. In 420 the future, it will be fascinating to investigate whether the SCHENGEN pathway is of even broader 421 developmental significance and to understand the molecular basis that enables its distinct, organ and 422 cell-type specific activities.