A point mutation in the kinase domain of CRK10 leads to xylem vessel collapse and activation of defence responses in Arabidopsis

Abstract Cysteine-rich receptor-like kinases (CRKs) are a large family of plasma membrane-bound receptors ubiquitous in higher plants. However, despite their prominence, their biological roles have remained largely elusive so far. In this study we report the characterization of an Arabidopsis mutant named crk10-A397T in which alanine 397 has been replaced by a threonine in the αC helix of the kinase domain of CRK10, known to be a crucial regulatory module in mammalian kinases. The crk10-A397T mutant is a dwarf that displays collapsed xylem vessels in the root and hypocotyl, whereas the vasculature of the inflorescence develops normally. In situ phosphorylation assays with His-tagged wild type and crk10-A397T versions of the CRK10 kinase domain revealed that both alleles are active kinases capable of autophosphorylation, with the newly introduced threonine acting as an additional phosphorylation site in crk10-A397T. Transcriptomic analysis of wild type and crk10-A397T mutant hypocotyls revealed that biotic and abiotic stress-responsive genes are constitutively up-regulated in the mutant, and a root-infection assay with the vascular pathogen Fusarium oxysporum demonstrated that the mutant has enhanced resistance to this pathogen compared with wild type plants. Taken together our results suggest that crk10-A397T is a gain-of-function allele of CRK10, the first such mutant to have been identified for a CRK in Arabidopsis.


Introduction
Plant growth and development are modulated by a multitude of intrinsic growth regulators and environmental cues. Factors regulating development as well as environmental and pathogenic signals are mostly recognized by receptor-like kinases (RLKs), membrane-localized receptors that perceive and transduce these signals to the intracellular environment. Similar to animal receptor tyrosine kinases (RTKs), these receptors consist of an extracellular domain that perceives specific ligands, a single-pass transmembrane domain, and a cytoplasmic kinase domain that transduces the signal via phosphorylation of downstream target proteins in the cytoplasm in order to tailor a cellular response (Shiu and Bleecker, 2001;De Smet et al., 2009). Their highly variable extracellular domains are used for the classification of RLKs into subgroups, the largest of which (~200 genes in Arabidopsis) is characterized by leucine-rich repeats (LRR-RLKs) (Diévart and Clark, 2004). The well-studied brassinosteroid receptor BRASSINOSTEROID INSENSITIVE 1 (BRI1; AT4G39400) and the microbial pattern recognition receptors (PRR) FLAGELLIN SENSING 2 (FLS2; AT5G46330) and EF-TU RECEPTOR (EFR; AT5G20480), for example, are well-characterized members of this subgroup (Friedrichsen et al., 2000;Chinchilla et al., 2006;Zipfel et al., 2006).
Among the multiple subfamilies of RLKs found in plant genomes, the one containing the cysteine-rich receptor-like kinases (CRKs) is one of the largest with over 40 members in Arabidopsis. The signature motif for CRKs is the presence of, in most cases, two repeats of the DOMAIN OF UN- KNOWN FUNCTION 26 (DUF26) in their extracellular domain, which contains three cysteine residues in the conserved configuration C-X8-C-X2-C (Chen, 2001). Although the functional significance of the DUF26 domain remains to be elucidated, it was originally suggested to participate in redox sensing (Wrzaczek et al., 2010;Bourdais et al., 2015). More recent data obtained from the crystallographic analysis of the DUF26-containing ectodomain of the plasmodesmatalocalized proteins PDLP5 and PDLP8, however, point towards the involvement of the cysteine residues in forming disulfide bonds for the structural stabilization of the protein rather than redox regulation (Vaattovaara et al., 2019). The same study also revealed that the DUF26 domain shows strong structural similarity to fungal carbohydrate-binding lectins, which suggests that DUF26-containing proteins might constitute a group of carbohydrate-binding proteins in plants (Vaattovaara et al., 2019). Corroborating this hypothesis, the ability to interact with carbohydrates was demonstrated for the secreted DUF26-containing antifungal proteins Ginkbilobin2 (Gnk2) from Gingko biloba (Miyakawa et al., 2009(Miyakawa et al., , 2014 and Anti-Fungal Protein 1 (AFP1) from maize (Ma et al., 2018), both of which bind to the monosaccharide mannose, a component of fungal cell walls. Bona fide ligands for CRKs, however, still remain to be identified.
Despite the large number of CRKs among the RLK superfamily, very little is known about their specific biological roles and the regulation of downstream signalling events. Efforts to assign functions to members of this family have involved a comprehensive analysis of a collection of T-DNA knockout lines for 41 CRKs of Arabidopsis, which suggested a role for several members in the fine-tuning of stress adaptation and plant development (Bourdais et al., 2015). Most knockout lines, however, did not display obvious phenotypes, as is expected for a large gene family due to redundancy amongst its members. Studies in Arabidopsis also revealed that several CRKs are transcriptionally regulated by a wide variety of biotic and abiotic factors such as ozone, ultraviolet light, reactive oxygen species, the hormone salicylic acid (SA), and elicitation with pathogenderived molecules (Czernic et al., 1999;Du and Chen, 2000;Ohtake et al., 2000;Wrzaczek et al., 2010).
Functionally, CRKs belong to the RD subclass of Ser/Thr kinases (Vaattovaara et al., 2019), which typically carry a conserved arginine (R) immediately preceding the invariant aspartate (D) in subdomain VI required for catalytic activity, and are, in most cases, activated through autophosphorylation of the activation loop (Nolen et al., 2004). Although the ability to autophosphorylate as well as to phosphorylate substrates in vitro has been demonstrated for CRK2 and CRK7 (Idänheimo et al., 2014;Kimura et al., 2020), for example, detailed studies investigating the role of specific phosphorylation sites for the regulation of kinase activity are still outstanding for CRKs.
Establishing ligand-receptor pairs is not a trivial task and exploring regulation and function of RLKs without the knowledge of their activating ligands poses a challenge. In these circumstances, mutants harbouring gain-of-function alleles of RLKs, where kinase activity occurs in the absence of the ligand, can be a useful resource to study receptor regulation and function. In this study we describe the characterization of one such gain-of-function allele, crk10-A397T, obtained for CYS-TEINE-RICH RECEPTOR-LIKE KINASE 10 (CRK10; AT4G23180) of Arabidopsis. This mutation leads to the replacement of alanine 397 with threonine in the αC helix of the kinase domain of the protein, with the newly introduced Thr397 acting as an additional phosphorylation site in situ. We show that the crk10-A397T allele causes a dwarf phenotype in Arabidopsis, which is associated with the collapse of xylem vessels in roots and hypocotyls. Analysis of the transcriptome shows the occurrence of extensive transcriptional re-programming of immune-and cell wall-related genes in the hypocotyls of the mutant plants. Biochemical analyses reveal that these transcriptional changes are accompanied by alterations in cell wall composition. A pathogen assay indicates enhanced resistance to the soil-borne vascular pathogen Fusarium oxysporum. Taken together our results suggest that crk10-A397T is a gainof-function allele of CRK10 that leads to xylem vessel collapse and the activation of defence responses to pathogens.

Plant materials, growth conditions, and micrografting
Arabidopsis ecotype Col-0 plants were grown in Grobanks cabinets (CLF 2006, Plant Climatics, Germany) in Levington F2 + Sand compost in long day conditions (16 h/8 h), 23/18 ºC day/night temperature, and 200 µmol m −2 s −1 light intensity. For in vitro experiments, surface-sterilized seeds were cultivated on ½ Murashige and Skoog (MS) plates (Duchefa Biochemie). The T-DNA lines SAIL_427_E09 and SALK_116653 were obtained from the Nottingham Arabidopsis Stock Centre. Micrografting was performed according to the procedure described by Turnbull et al. (2002). Successful grafts were transferred to soil 7-10 d post-grafting.

RNA isolation, RNA-Seq library construction, and sequencing
Total RNA was extracted from hypocotyls of Arabidopsis plants using the RNeasy Mini Kit (Qiagen). Four biological replicates were isolated per genotype per time point with each biological replicate consisting of a pool of 50-60 hypocotyls. Samples were treated with DNase Turbo DNA-free kit (Thermo Fisher Scientific). Prior to library preparation RNA quality was assessed on a 2100 Bioanalyzer (Agilent Technologies). Library preparation and paired-end sequencing was performed by Exeter Sequencing Service (University of Exeter, UK) using the Illumina HiSeq 125 PE sequencing platform.

Quantification of transcript abundance by quantitative PCR
RNA was isolated using TRI Reagent (Merck) and treated with DNase I, Amplification Grade (Thermo Fisher Scientific) prior to cDNA synthesis with SuperScript III Reverse Transcriptase (Thermo Fisher Scientific).
All qPCR reactions were performed in a LightCycler 96 Real-Time PCR System (Roche Diagnostics) using the FastStart Essential DNA Green Master (Roche Diagnostics). Primers AtCRK10 forward (For)/ reverse (Rev) were used for quantification of CRK10 expression, and primers AtACT2 For/Rev (ACTIN2; AT3G18780) and AtUBC21 For/ Rev (UBC21; AT5G25760) were used as internal controls (see Supplementary Table S1). The 2 −∆∆Ct method (Livak and Schmittgen, 2001) was used to calculate relative expression.

Quantification of plant hormones
Three replicates, each containing between 75 and 100 mg (fresh weight) of hypocotyls isolated from 3-week-old wild type (WT) and mutant plants were prepared. Hormone analysis was performed using the platform provided by The Plant Hormone Quantification Service from Universitat Politècnica de València according to their protocol (https:// ibmcp.upv.es/services/plant-hormone-quantification/).

Generation of genetic constructs
The cDNA clone (U60398) containing full length CRK10 was obtained from the Arabidopsis Biological Resource Center (https://abrc.osu.edu/). Vector pJD330 was kindly provided by Dr D. R. Gallie, and vector RS 3GSeedDSRed MCS by Edgar B. Cahoon, University of Nebraska-Lincoln. A list of primers used in this study can be found in Supplementary  Table S1. The 35S:CRK10-NOSt construct was generated by amplification of the full-length cDNA of CRK10 from clone U60398 with primers CRK10 SalI For and CRK10 SacI Rev. The CRK10 cDNA sequence was subcloned into pJD330 between the 35S promoter and NOS terminator after removal of the β-glucuronidase (GUS)-containing fragment by restriction digestion (pJD330 35S:CRK10-NOSt). The 35S:CRK10-NOSt fragment was then amplified with 35S AscI For and NOSt AscI Rev primers and inserted into the binary vector RS 3GSeedDSRed MCS. The CRK10 Pro :crk10-A397T-NOSt construct was generated by replacing the 35S promoter in pJD330 35S:CRK10-NOSt with 1 kb of the native promoter of CRK10 (CRK10 Pro ) obtained by amplification of genomic DNA with primers CRK10 Pro SphI For and CRK10 Pro SalI Rev. In vitro mutagenesis (GeneArt; Thermo Fisher Scientific) with primers CRK10 A397T For and CRK10 A397T Rev was used to introduce the G>A mutation responsible for the replacement of A397 by T. CRK10 Pro AscI For and NOSt AscI Rev primers were used to amplify CRK10 Pro :crk10-A397T-NOSt for cloning into the RS 3GSeedDSRed MCS binary vector. The CRK10 Pro :GUS-NOSt construct was generated by replacing 35S in pJD330 with the CRK10 Pro sequence obtained by amplification with primers CRK10 Pro SphI For and CRK10 Pro NcoI Rev through restriction digestion. The CRK10 Pro :GUS:NOSt fusion was amplified with primers CRK10 Pro AscI For and NOSt AscI Rev for transfer into RS 3GSeedDSRed MCS. The translational fusion of CRK10 with mCherry was obtained by replacing the stop codon of CRK10 in pJD330 35S:CRK10-NOSt with a SacI restriction site (primers CRK10 wsc SacI For and CRK10 wsc SacI Rev) by in vitro mutagenesis. This restriction site was used to insert in frame the mCherry sequence that had been amplified with compatible primers (mCherry SacI For and mCherry SacI Rev). The CRK10-mCherry-NOSt fragment was then amplified with primers CRK10 SalI For and NOSt NotI Rev and cloned into the pEN-TR1A Dual Selection Vector (Thermo Fisher Scientific). Gateway cloning (Thermo Fisher Scientific) into destination vector pB2GW7 (Karimi et al., 2002) generated the final construct 35S:CRK10-mCherry-NOSt.

Transformation of plants
Detection of transient expression of fluorescent fusions was performed by infiltration of Nicotiana benthamiana leaves according to Sparkes et al. (2006). Expression of fusion proteins was observed 72 h post-infiltration.
In order to generate stable transformed lines Arabidopsis Col-0 was transformed by floral dip (Clough and Bent, 1998) with Agrobacterium tumefaciens GV3101 containing the respective constructs.

Microscopy
Light microscopy Thin sections were prepared by fixing plant tissue in 4% paraformaldehyde-2.5% glutaraldehyde followed by gradual dehydration with ethanol and infiltration with LRWhite resin (Agar Scientific). Sections were prepared using a Reichert ultramicrotome (section thickness: 1-2 µm) and stained with 0.5% potassium permanganate. Thin sections were observed with a Zeiss Axiophot microscope equipped with a Q-Imaging Retiga EXi CCD camera (QImaging, Canada).

Confocal microscopy
Confocal microscopy was performed using the Zeiss 780 LSM system. For detection of the autofluorescence of lignin, non-stained resin-embedded thin sections were imaged with an excitation wavelength of 405 nm and emission was collected at 451-480 nm and 560-612 nm. mCherry fluorescence from transiently transformed N. benthamiana leaves or stably transformed Arabidopsis hypocotyls was detected with a laser excitation wavelength of 561 nm and collection of emission at 578-639 nm. Plasmolysis was performed using a 0.8 M mannitol solution for 40 min.
Transmission electron microscopy Samples were prepared by fixing plant material by high-pressure freezing using a Leica HPM100, followed by freeze substitution with 100% ethanol (Leica EM Auto Freeze substitution) and infiltration with LRWhite resin (Agar Scientific). Ultra-thin sections were prepared with a Leica EM UCT ultramicrotome (section thickness: 90 nm) and were collected on pioloform/carbon-coated nickel grids (Agar Scientific) and stained with 2.5% uranyl acetate and Reynolds lead citrate (Reynolds, 1963). Ultrathin sections were imaged using a JEOL-2100Plus transmission electron microscope (JEOL, Japan) equipped with a Gatan OneView IS camera (Gatan, USA).

Fourier-transform infrared spectroscopy
Transverse cross sections of hypocotyls of 3-week-old WT and crk10-A397T mutant plants for Fourier-transform infrared spectroscopy (FTIR) analysis were prepared using a cryostat (CM1850 Cryostat, Leica Microsystems; section thickness: 20 µm) with three replicates per sample. Cross sections were washed with 70% ethanol and air-dried on barium fluoride (BaF 2 ) discs prior to analysis using a Nicolet iN10MX infrared microscope (Thermo Scientific) equipped with a ×15 infrared (IR) objective. FTIR maps were obtained by transmission aperture mapping with a mercury-cadmium-telluride detector and an x-y step size of 10 μm (20 × 20 μm 2 aperture). A total of 128 scans were averaged at 8 cm −1 for each image pixel. An empty spot on the BaF 2 disk was used as background. The maps were exported in ENVI format and processed in MATLAB (The MathWorks, Inc.). The spectra were truncated to 1800-700 cm −1 and the density map was calculated by averaging the IR absorption from 1800 to 800 cm −1 . Characteristic band intensities for the following components were mapped to visualize their distribution: 1018 cm −1 for pectin, 1033 cm −1 and 1050 cm −1 for hemicelluloses and cellulose, 1511 cm −1 for lignin, 1650 cm −1 for protein, and 1735 cm −1 for ester groups.

Quantification of total monosaccharides
Thirty hypocotyls of 3-week-old WT and crk10-A397T mutant plants were collected per replicate with three replicates per sample. Alcohol in-soluble residue (AIR) preparation was performed according to Goubet et al. (2009). A total of 200-600 µg of AIR per sample was hydrolysed in 2 M trifluoroacetic acid (Sigma-Aldrich) prior to quantification of acidic and neutral monosaccharides by high performance anion-exchange chromatography with pulsed amperometric detection. Acidic monosaccharides (glucuronic and galacturonic acid) were quantified using a Dionex ICS-3000 ion chromatography system (Thermo Scientific) and CarboPac PA-200 columns (Thermo Scientific). For the quantification of neutral monosaccharides, a Dionex ICS-5000+ equipped with eluent generator (Thermo Scientific) and CarboPac PA-20 columns (Thermo Scientific) were used. Chromeleon analytical software (version 7.2SR5; Thermo Scientific) was used for peak marking and quantification.

Histochemical analysis of β-glucuronidase activity
Plant tissue was incubated overnight in X-gluc (Melford) solution at 37 °C. Chlorophyll was removed with 80% ethanol prior to imaging using a Leica M205 FA stereomicroscope (Leica Microsystems).

Recombinant protein expression, purification, and analysis
Primers used to generate the 6× His-tagged CRK10 kinase domain (KD) constructs are listed in Supplementary Table S1. The KD of CRK10 was amplified with primers CRK10 KD SalI For and CRK10 KD NotI Rev and cloned into pENTR1A Dual Selection Vector (Thermo Fisher Scientific). In vitro mutagenesis was used to generate the gain-of-function mutation of CRK10 (primers CRK10 A397T For and CRK10 A397T Rev) and the dead kinase variant (primers CRK10 D473N For and CRK10 D473N Rev) prior to Gateway cloning with pDEST17 (Thermo Fisher Scientific). The 6× His tagged proteins were expressed in BL21 AI One-Shot Escherichia coli cells (Thermo Fisher Scientific). Bacterial cultures were grown to an OD 600 of 0.4-0.5 after which protein expression was induced by adding l-arabinose to a final concentration of 0.2%. After 3 h bacterial cells were lysed by sonication, His-tagged proteins were purified with the HIS-Select Nickel Affinity Gel (Sigma-Aldrich) and protein concentration determined by the Bradford method (Protein Assay Dye Reagent, Bio-Rad). For phosphatase treatment, 5 µg protein extract was treated with Lambda Protein Phosphatase (Lambda PP, New England BioLabs) for 1 h 30 min at 30 ºC. Samples were resolved by SDS-PAGE (NuPAGE 4-12% Bis-Tris Protein Gels, Thermo Fisher Scientific) and gels were stained with Quick Coomassie Stain (Generon). For western blotting, proteins were transferred to a polyvinylidene difluoride membrane (iBlot Transfer Stack, PVDF, Thermo Fisher Scientific) and hybridized with a His-probe (H-3) horseradish peroxidase monoclonal antibody (Santa Cruz Biotechnology). The membrane was washed and incubated with Amersham ECL Western Blotting Detection Reagent (GE Healthcare) according to manufacturer's instructions.

Determination of phosphorylation sites
Protein bands corresponding to His-CRK10kd WT and His-CRK-10kd A397T were excised from an acrylamide gel and sent to the Cambridge Centre for Proteomics, (https://proteomics.bio.cam.ac.uk/core-facility) for analysis by LC-MS/MS. Data were submitted to the Mascot search algorithm (Matrix Science, London UK, version 2.6.0) against a custom database consisting of the CRK10 WT and CRK10 A397T sequences and the UniProt Arabidopsis database (41552 sequences; 17578843 residues). A significance threshold value of P<0.05 and a peptide cut-off score of 20 were applied.

Protein modelling
A structural model of the kinase domain of CRK10 was generated by homology modelling using PyMOD 3.0 with default parameters (Janson et al., 2017). The kinase domain of BRI1 (PDB code 5LPV) (Bojar et al., 2014) was used as template retaining the ATP analogue (phosphoaminophosphonic acid adenylate ester) but no other heteroatoms.

Fusarium oxysporum infection assay
Susceptibility to infection by Fusarium oxysporum f. sp. conglutinans 699 (kindly provided by Prof. Antonio Di Pietro) was assessed by a root infection assay according to Masachis et al. (2016) with minor modifications. Arabidopsis seedlings (12 d old) grown in vitro were inoculated by immersing their roots for 20 min in a suspension of 1 × 10 6 microconidia ml −1 of F. oxysporum f. sp. conglutinans 699. Subsequently seedlings were transferred to soil and cultivated in a growth chamber under long day conditions and a temperature set at 28 ºC during the day and 25 ºC at night. For each repetition of the experiment, 80 plants per genotype were arranged in a randomized blocked tray design, and mortality was assessed daily between 7 and 20 d post-inoculation. The experiment was performed twice. To determine fungal burden, seedlings were inoculated as described above and sampled at 2 and 7 d post-inoculation. At these time points, total DNA was extracted (protocol adapted from Yu et al., 2019) from a pool of eight seedlings per genotype and the relative amount of fungal DNA was quantified by qPCR using the primers ACTIN1 For/Rev (F. oxysporum) and normalized to the Arabidopsis ACTIN2 gene (primers AtACT2 For/Rev) (see Supplementary Table S1). The experiment was repeated three times. Results were expressed relative to WT at 2 d post-inoculation.

Statistical analysis
Statistical tests were performed using Genstat software (Genstat for Windows 21st Edition; VSN International, Hemel Hempstead, UK). Student's t-test was used to assess statistical differences between two variants. To assess whether the pattern of segregation of the dwarf phenotype followed the expected 1:2:1 ratio, the chi-square statistic was used, where O i is the observed count for group i and E i is the expected count for group i. Under the null hypothesis of 1:2:1 segregation, this test statistic should follow a chi-square distribution with 2 degrees of freedom. The probability of survival of each genotype in the bioassay with F. oxysporum was assessed with a generalized linear model (Bernoulli distribution; logit link function fitted to the final mortality outcome of each plant); statistical significance of the genotypic effect was tested after removing variation associated with plant position within rows of different trays and quantified through a chi-square statistic of the difference in deviance. Statistical significance of the differences in fungal burden between genotypes was tested by ANOVA (expression levels were log-transformed to meet the ANOVA requirements, and each individual experiment was considered as a block).

crk10-A397T is a semi-dominant mutant allele of CRK10
The Arabidopsis mutant characterized in this report was isolated in a forward genetic ethyl methanesulfonate screen performed for an unrelated study. In brief, six rounds of backcrosses to the WT Col-0 parent were performed in order to clean the genetic background of the mutant before the in-depth characterization. The homozygous mutant has a strong dwarf phenotype and observation of the segregating F 2 progeny of the sixth backcross revealed the semi-dominant nature of the mutation, as WT, intermediate and dwarf phenotypes segregated according to a 1:2:1 ratio with heterozygous plants being clearly discernible (Supplementary Fig. S1A; χ 2 2 =2.36, P=0.308). To determine the underlying mutation responsible for the dwarf phenotype, whole genome sequencing was performed on bulk segregants derived from the sixth backcross. This returned a list of 15 candidate genes containing point mutations in coding regions. We noticed that a point mutation (G>A) in the fourth exon of CYSTEINE-RICH RECEPTOR-LIKE KINASE 10 (CRK10; AT4G23180) causes the replacement of alanine 397 by a threonine in the kinase domain of the protein (Supplementary Fig. S1B). As we considered this receptor-like kinase to be the most likely candidate among the 15 identified genes, we tested whether the dwarf phenotype could be rescued by constitutive expression of the WT cDNA sequence of CRK10 under the control of the 35S promoter. All T 1 transformants showed a WT phenotype ( Supplementary Fig. S1C), suggesting that the correct gene had been identified. To further confirm that the mutation in CRK10 causes the dwarf morphology, we recreated the G>A substitution by in vitro mutagenesis in the cDNA sequence of CRK10 and introduced this open reading frame into a crk10 knockout (KO) background (crk10-2, SAIL_427_E09, characterization of KO lines to follow) under the control of the 1 kb genomic region containing the putative native promoter of CRK10. A total of 25% of the recovered transformants were dwarfs, establishing a direct link between the dwarf phenotype and the mutant allele ( Supplementary Fig. S1D). Subsequently, we will refer to this mutant as crk10-A397T.
The crk10-A397T mutant is a dwarf WT Col-0 and crk10-A397T plants were phenotypically characterized for the duration of one entire life cycle. Although germination rate and establishment of seedlings were accelerated in the mutant ( Supplementary Fig. S2A, B) these differences were no longer apparent 1 week after sowing. No other obvious differences in growth were observed between WT and crk10-A397T seedlings until week 2, after which leaf expansion became restricted in the crk10-A397T mutant and small, dark green leaves were formed causing a reduction of more than 70% in rosette size at 4 weeks after sowing ( The crk10-A397T mutant has collapsed xylem vessels in roots and hypocotyls Dwarfism in plants is often caused by defects in the vascular system. To investigate whether the vasculature of the crk10-A397T mutant develops normally, we prepared transverse cross sections of resin-embedded hypocotyl, root, and stem samples of 5-week-old plants. The sections were stained with potassium permanganate, a lignin-specific dye that allows the observation of lignified xylem vessels and fibres. Imaging of the cross sections revealed that xylem vessels in the root and hypocotyl of the mutant plants were severely collapsed (Fig. 2C-F), whereas vessels in the stem remained largely unaffected ( Fig. 2A, B), as confirmed additionally by Maeule staining ( Supplementary Fig. S4A, B). To understand the progression of the phenotype, a developmental time series of hypocotyl cross sections spanning weeks 1-5 after sowing was analysed ( Fig. 2K-V). Cross sections of 1-and 2-week-old hypocotyls revealed disorganization of xylem vessels in crk10-A397T plants at an early developmental stage as they did not proceed to form the typical radial patterning observed in the hypocotyl vasculature of WT plants (Fig. 2K-N). At 3 weeks of age, the first deformed and collapsed xylem vessels became apparent in the mutant hypocotyls, a phenotype that is even more severe in 4-week-old plants (Fig 2O-R). At 5 weeks of age, following the onset of flowering, cross sections revealed the absence of fully differentiated xylem fibres in the hypocotyl of the crk10-A397T plants, in contrast to the WT (Fig. 2S-V). Differentiation of xylary fibres in Arabidopsis hypocotyls is associated with the switch to growth phase II of xylem development, which is triggered by the transition to flowering. We conclude that this switch is delayed in the crk10-A397T plants, despite flowering occurring simultaneously to WT plants.
Interestingly, the collapsed xylem vessels displayed darker brown staining in response to the dye compared with their WT counterparts, which suggests that their secondary cell walls were more heavily lignified (Fig. 2G, H). This hypothesis was reinforced by detecting the auto-fluorescence of lignin of these cells using confocal microscopy, as the autofluorescence of xylem vessels in the mutant hypocotyl was consistently more intense than the signal obtained from WT (Fig. 2I, J).

CRK10 is expressed in close association with vascular tissues and the protein localizes to the plasma membrane
Tissue-specific expression of CRK10 was determined by placing the reporter β-glucuronidase under the control of the 1 kb genomic sequence containing the putative promoter of CRK10 (CRK10 Pro :GUS). GUS expression was detected in the vasculature of the roots, cotyledons, petioles, leaves, hypocotyls, and inflorescence stem (  of the CRK10 transcript in hypocotyls and inflorescence stems of 3-and 6-week-old WT and crk10-A397T mutant plants, respectively, was confirmed by qPCR ( Supplementary  Fig. S5E). Subcellular localization of CRK10 was determined by analysing lines expressing a construct carrying the C-terminal translational fusion of CRK10 with the fluorescent protein mCherry under the control of the constitutive 35S promoter (35S:CRK10-mCherry). Both transient expression of the construct in N. benthamiana leaves and stable expression in transgenic Arabidopsis plants indicated that the fusion protein localized to the plasma membrane ( Supplementary Fig. S5F;  Fig. 3C). The presence of Hechtian strands, characteristic of the retracting plasma membrane from the cell wall following plasmolysis (Oparka, 1994), further confirmed this subcellular localization of the protein (Fig. 3D). In order to exclude a possible effect of the point mutation on the subcellular localization of the protein, we also expressed the 35S:CRK10 A397T -mCherry variant transiently in tobacco leaves. Similar to the native protein, CRK10 A397T -mCherry was found to localize to the plasma membrane, which suggests that the point mutation does not alter the subcellular localization of the protein and does not seem to affect its endocytosis ( Supplementary  Fig. S5G). Therefore, we conclude that CRK10 is expressed in close association with vascular tissues of below-and above -ground organs, and that the protein localizes to the plasma membrane of plant cells.
Collapsed xylem vessels in the root and hypocotyl are responsible for the dwarf phenotype of the crk10-A397T mutant Although CRK10 is expressed in tissues associated with vasculature in the stem, hypocotyl, and roots, as demonstrated by CRK10 Pro :GUS analysis, it is intriguing that in the crk10-A397T plants xylem vessel collapse occurs only in roots and hypocotyls. To investigate if the dwarf phenotype is solely due to the defects of the belowground tissues, or whether it is a 'whole-plant' response, we performed a micrografting experiment (Turnbull et al., 2002). In vitro-grown 4-day-old seedlings were used to generate combinations of WT rootstocks and crk10-A397T scions (WT/crk10-A397T) and vice versa (crk10-A397T/WT; Fig. 4C), as well as self-grafted plants as controls (WT/WT and crk10-A397T/crk10-A397T; Fig.  4B). Phenotypic assessment of successful grafts revealed that a WT scion grafted onto a crk10-A397T rootstock developed the characteristic dwarf phenotype of the mutant, whereas a mutant scion developed into a WT-like plant when grafted onto a WT rootstock (Fig. 4C). Our observations show that the root and hypocotyl system of the crk10-A397T plants are responsible for their dwarf phenotype, which is likely due to the presence of collapsed xylem vessels in these tissues.

Loss of function or overexpression of CRK10 does not have a phenotypic effect in Arabidopsis
In order to investigate whether increased levels of CRK10 expression have phenotypic effects we introduced a construct carrying the WT cDNA of CRK10 under the control of the constitutive 35S promoter in WT Arabidopsis plants. Two independent homozygous lines were generated and selected for further analysis (CRK10 OE-1 and OE-2). Compared with WT, qPCR performed on 4-week-old leaves detected a CRK10 transcript increase of 15 and 6 times for CRK10 OE-1 and OE-2, respectively ( Supplementary Fig. S6B), although growth and development were not altered (Supplementary Fig. S6A). In order to investigate whether the absence of the CRK10 transcript affects the phenotype of Arabidopsis plants, two homozygous T-DNA knockout lines for the CRK10 gene were isolated, crk10-2 (SAIL_427_E09) and crk10-4 (SALK_116653). Quantification of CRK10 transcript levels from leaves of 4-week-old plants by qPCR confirmed that crk10-2 and crk10-4 are a knockout and knockdown line of CRK10, respectively ( Supplementary Fig. S7B), but growth and development of both lines were indistinguishable from WT plants (Supplementary Fig. S7A). Cross sections of hypocotyls of 4-week-old crk10-2 and CRK10 OE-1 plants were imaged and showed that xylem vessels develop normally in both lines ( Supplementary Fig. S8A-D).

The A397T substitution is localized in the αC helix of the kinase domain of CRK10
According to the subdivision of eukaryotic kinase domains into 12 conserved subdomains (Hanks and Hunter, 1995), the A397T substitution is localized in subdomain III of the kinase domain of CRK10 (Fig. 5A), which corresponds to the αC helix motif in the three-dimensional structure of the protein. Homology modelling to the active kinase domain of the Arabidopsis BRASSINOSTEROID INSENSITIVE 1 (BRI1) positions Thr397 at the C-terminal end of the helix, with its side chain likely to be exposed on the surface of the protein (Fig. 5B).

The cytoplasmic kinase domain of CRK10 and crk10-A397T are active kinases
With CRKs being classified as Ser/Thr kinases, the replacement of Ala397 by a threonine (A397T) in the CRK10 kinase domain could have introduced a potential additional phosphorylation site. We therefore wanted to determine whether WT and mutant CRK10 are enzymatically active kinases and if differences in their autophosphorylation pattern could be detected. We addressed this question by investigating the autophosphorylation activity of the cytoplasmic kinase domain of CRK10 in situ when expressed in E. coli cells (Taylor et al., 2013). We purified the CRK10 WT cytoplasmic kinase domain as an N-terminal 6× His-tag fusion protein (His-CRK10kd WT ) from E. coli cells, as well as its 'dead' kinase counterpart that harboured the substitution of the essential aspartic acid 473 with an asparagine residue (His-CRK10kd WT-D473N ). Following separation of the recombinant proteins by SDS-PAGE and detection by anti-His immunoblotting, the dead kinase version His-CRK10kd WT-D473N migrated at the predicted molecular mass of 40 kDa, while the WT kinase version His-CRK10kd WT showed an electrophoretic mobility shift to a larger molecular mass, known to occur for phosphorylated proteins (Wegener and Jones, 1984;Fig. 5C). In order to determine whether the A397T substitution in the CRK10 kinase domain affects its autophosphorylation activity, we generated constructs in which Ala397 was replaced by threonine through in vitro mutagenesis (His-CRK-10kd A397T and His-CRK10kd A397T-D473N ). Although the dead kinase version His-CRK10kd A397T-D473N migrated at the same molecular mass as His-CRK10kd WT-D473N on SDS-PAGE gels, the mobility shift of His-CRK10kd A397T was increased when compared with the one observed for His-CRK10kd WT Fig. 4. The root-hypocotyl system is responsible for the dwarf phenotype of crk10-A397T mutant plants. Images of non-grafted plants (A), self-graft controls (B), and graft combinations (C) of WT and crk10-A397T mutant. Plants were imaged 3 weeks after micrografting was performed. The phenotype observed for the reciprocal grafting combinations was consistently observed in two independent repetitions of the experiment. An average number of 10 grafts per combination was recovered each time. Annotation: scion/rootstock. (Fig. 5C), suggesting additional sites were phosphorylated in His-CRK10kd A397T . In order to confirm that the electrophoretic mobility shift was due to the presence of phosphorylated residues, the purified recombinant proteins were treated with λ-phosphatase prior to separation by SDS-PAGE. Irrespective of the treatments, the dead kinase versions CRK10kd WT-D473N and His-CRK10kd A397T-D473N migrated at the predicted molecular mass as confirmed by SDS-PAGE and anti-His immu-noblotting (Fig. 5D). However, λ-phosphatase treatment of His-CRK10kd WT and His-CRK10kd A397T resulted in a clearly detectable shift to a lower molecular mass, consistent with the autophosphorylation of recombinant His-CRK10kd WT and His-CRK10kd A397T as being responsible for their electrophoretic mobility shift. Taken together, these results confirm that both His-CRK10kd WT and His-CRK10kd A397T are active kinases capable of autophosphorylation. Furthermore, the increased mobility shift of His-CRK10kd A397T compared with His-CRK10kd WT suggested the presence of additional phosphorylation sites in His-CRK10kd A397T .
The kinase domain of CRK10 autophosphorylates highly conserved residues in the activation loop and Thr397 is an additional phosphorylation site in His-CRK10kd A397T We next proceeded to identify which sites in the kinase domain of CRK10 were being phosphorylated by subjecting tryptic peptides of His-CRK10kd WT and His-CRK10kd A397T to analysis by LC-MS/MS. The Mascot probability-based algorithm was used to confirm the peptides match to the CRK10 kinase domain sequence. Individual MS/MS spectra were inspected for confirmation of phosphorylation sites, which led to the unambiguous identification of Thr340, Tyr363, Thr507, Ser508, Tyr514, Thr625, Ser662, and Thr664 as phosphosites in both His-CRK10kd WT and His-CRK10kd A397T proteins (Fig.  5E). Interestingly, Thr507, Ser508, and Tyr514 align to conserved phosphorylation sites in the activation loop of several RLKs, known to be essential for the activation of RD kinases ( Supplementary Fig. S9). Phosphorylated residues were also detected in the juxta-membrane region (Thr340) as well as in the C-terminal tail of CRK10 (Thr625, Ser662, and Thr664), which are predicted to act as regulatory sites for interaction with binding partners. In addition, the identification of two phosphorylated tyrosine residues (Tyr363 and Tyr514) classifies CRK10 as a dual specificity kinase and constitutes the first instance in which such activity has been reported for a CRK. Interestingly, Thr397 itself was identified as a phosphorylation site in the His-CRK10kd A397T kinase domain in situ (Fig. 5F), but whether this residue also acts as a phosphorylation site in vivo remains to be determined.

The hypocotyl transcriptome of crk10-A397T carries the signature of a plant responding to stress
As the dwarf phenotype of the crk10-A397T mutant is associated with the collapse of xylem vessels in the belowground organs, we chose to investigate the effect of this mutation on the transcriptome of hypocotyls isolated from 2-, 3-and 5-week-old WT and mutant plants. Principal component analysis showed good clustering of replicates according to genotypes and developmental time points ( Supplementary  Fig. S10A). Following normalization and statistical analysis of the sequencing results (q≤0.05; log 2 fold change threshold of ±1), we obtained 523 (2 weeks), 1836 (3 weeks), and 913 (5 weeks) DEGs, of which 274 were common to all time points. These DEGs were selected as the core set and taken forward for analysis ( Fig. 6A; Supplementary Fig. S10B; Supplementary Tables S2-S5). Comparison with public datasets using the GENEVESTIGATOR Signature tool (Hruz et al., 2008) revealed that the transcriptome signature of the crk10-A397T mutant was most similar to Arabidopsis plants challenged by fungal and bacterial pathogens (Sclerotinia sclerotorum, Plectosphaerella cucumerina, and Pseudomonas syringae) and exposed to abiotic stresses (treatment with fenclorim and sulfometuron methyl) (Fig. 6C). Equally, Gene Ontology (GO) enrichment analysis of the up-regulated genes within the core set (246 genes) with AgriGO v.20 (Tian et al., 2017; false discovery rate <0.05) revealed that terms associated with the biological functions 'Defence response' (GO:0006952; P=2.30 × 10 −26 ), 'Response to stimulus' (GO:0050896; P=1.40 × 10 −24 ) and 'Response to stress' (GO:0006950; P=2.00 × 10 −24 ) are significantly over-represented ( Fig. 6B; Supplementary Tables S6, S7). In accordance with the whole transcriptome data, marker genes indicative for the activation of the SA-and jasmonic acid (JA)-regulated defence pathways, such as pathogenesis-related and camalexin biosynthesic genes (Uknes et al., 1992;Thomma et al., 1998;Ahuja et al., 2012) are significantly up-regulated in the crk10-A397T mutant (Supplementary Table S8). Transcription factors belonging to the WRKY, MYB, and NAC-domain containing families are prominent among the regulatory genes induced by crk10-A397T, many of which have been associated with the modulation of stress responses (Supplementary Table S9). The analysis of DEGs of individual time points, especially at 3 weeks after sowing, revealed the up-regulation of genes involved in the biosynthesis, signalling, and homeostasis of the hormone abscisic acid (ABA) (Supplementary Table S10). Also at this time point numerous cell wall-related genes were found to be differentially expressed in the mutant transcriptome, implying that widespread changes in cell wall composition and assembly could be responsible for the xylem vessel collapse observed in mutant hypocotyls (Supplemental Table S11). In summary, biotic and abiotic stress-responsive pathways are constitutively up-regulated in the crk10-A397T mutant, suggesting that crk10-A397T is a gain-of-function allele of CRK10.

Differences in cell wall composition between wild type and mutant hypocotyls are manifold and complex
Xylem vessel collapse has been shown to be a consequence of altered cell wall composition leading to defective cell walls that are no longer able to withstand the negative pressure generated by transpiration, as for example described by Turner and Somerville (1997) for the irregular xylem (irx) mutants. As our transcriptomic dataset revealed the reprogramming of numerous cell wall-related genes, we proceeded with the characterization of the cell wall composition of the intact and collapsed xylem vessels in the hypocotyls of 3-week-old WT and crk10-A397T mutant plants by using FTIR spectroscopy, a powerful and rapid technique for analysing cell wall components and putative cross-links. The difference spectrum between the cell wall of an intact and collapsed xylem vessel revealed that complex changes had occurred in the collapsed vessel cell walls, with changes in hemicellulose composition and a reduced amount of ester cross-links among the most noticeable ( Supplementary Fig. S11A). We further performed FTIR microscopy on transverse cross sections of 3-week-old hypocotyls of WT and mutant in order to map the differences in chemical composition onto the anatomical structure ( Supplementary Fig. S11B-F). Heatmaps showed that the polysaccharide content was reduced in the cross sections in areas containing xylem elements in the mutant, and that the hemicellulose/cellulose ratio was altered ( Supplementary  Fig. S11C, D) as was the content of ester groups (Supplementary Fig. S11F). The complexity of the cell wall changes in the mutant, hinted at by FTIR, was further confirmed by quantification of total monosaccharides in the hypocotyls of 3-week-old WT and crk10-mutant plants (Supplementary Fig.  S12). The content of multiple monosaccharides was changed in the mutant, with fucose, arabinose, and galactose showing a highly significant reduction, and xylose a highly significant increase.
In numerous mutants with collapsed xylem vessels, alteration of the cell wall composition can frequently be visualized as changes in the ultrastructure of the cell wall when analysed by transmission electron microscopy, as reported for the irx1, irx3 irx8, irx7, and irx9 mutants (Taylor et al., 2000;Persson et al., 2007;Brown et al., 2011), for example. However, contrary to our expectations, transmission electron microscopy analysis performed on the secondary cell walls of intact WT vessels and collapsed xylem vessels of the crk10-A397T mutant showed no obvious differences in the ultrastructure of the secondary cell walls (Supplementary Fig. S13) suggesting that the biochemical alterations in cell wall composition are not reflected by obvious structural defects that could explain the collapse of these cells. In summary, our cell wall analyses showed that many components of the vessel cell walls in the mutant are altered, in accordance with the transcriptional reprogramming suggested by our RNA-Seq experiment.

crk10-A397T mutant hypocotyls contain increased levels of the stress hormones salicylic acid and abscisic acid
In order to corroborate the transcriptomic data, we quantified the levels of the stress hormones SA, ABA, and JA from the hypocotyls of 3-week-old WT and crk10-A397T plants (Supplementary Table S12). In agreement with the transcriptional induction of stress-responsive pathways, the levels of free SA and ABA were increased approximately 3 and 1.5 times, respectively, in the mutant hypocotyls. In contrast, JA levels were not significantly different to the WT. With defence responses constitutively up-regulated in the mutant, we next wanted to investigate whether this is reflected in an enhanced resistance to pathogens. Given that CRK10 expression is detected mainly in the vasculature, we chose the vascular pathogen Fusarium oxysporum f. sp. conglutinans 699, an isolate known to infect Arabidopsis, for the assay (Masachis et al., 2016). CRK10 OE-1 and crk10-2 lines were also included in the experiment as overexpression and knockout of other CRKs often showed enhanced/decreased resistance phenotype to pathogens (Acharya et al., 2007;Yeh et al., 2015;Yadeta et al., 2017). The progression of the infection was observed (Fig. 7A-D) and a time-mortality curve was recorded from 7 to 20 d post-inoculation (Fig. 7E). Our results showed that the susceptibility of WT and crk10-2 plants to the pathogens was very similar, with both genotypes reaching over 65% mortality at the end of the experiment. CRK10 OE-1 and crk10-A397T plants exhibit a similarly low mortality rate until 10 d postinoculation, although CRK10 OE-1 plants reach a final death toll of 47.5%, in contrast to the lowest overall death count displayed by the crk10-A397T mutant of around 18%. Statistical analysis (deviance test, χ 3 2 =19.68, P<0.001) confirmed the differences in the probability of survival between genotypes, with the crk10-A397T mutant having the highest chance of survival of 81.25%, followed by 52.5% for the CRK10 OE-1 plants, and just over 30% for both crk10-2 and WT (Fig. 7F). Fungal burden quantification by qPCR showed increased F. oxysporum biomass in WT and crk10-2 plants compared with CRK10 OE-1 and crk10-A397T mutant at 7 d post-inoculation, in agreement with the mortality trend results (Supplementary Fig. S14). The experiment was performed twice with similar results. Therefore, our results strongly suggest that crk10-A397T mutant plants are more resistant to infection with F. oxysporum, reinforcing our hypothesis that the transcriptional responses induced by the crk10-A397T mutant allele are effective at reducing the spread of root-infecting vascular pathogens.

Discussion
Previous efforts to assign functions to specific CRKs mainly focused on the characterization of T-DNA knock-out and overexpression lines. However, with a large multigene family, the effects of knocking out one specific member are often masked by redundancy and, in the absence of known stimulants, lines constitutively overexpressing CRKs are frequently phenotypically indistinguishable from WT. Here we report the characterization of the crk10-A397T mutant that harbours a gain-of-function allele of CRK10, to our knowledge the first such mutant obtained for this class of receptors in Arabidopsis. The point mutation responsible for the conversion of CRK10 into a gain-of-function allele (alanine 397 to threonine) lies in subdomain III of the kinase domain, at the C-terminus of the αC helix and at the start of the short αC-β4 loop that links the αC helix to the β strand 4 (Fig. 5A). The importance of this region for kinase regulation has been studied in numerous mammalian kinases, with mutations residing in this area often leading to kinase deregulation and disease. Equivalent studies in plant kinases are, however, absent. The combined αC helix and αC-β4 loop have been shown to be critical allosteric docking sites that play a crucial part in kinase regulation (Yeung et al., 2020). As the 3D model of the CRK10 kinase domain suggests, the substituted amino acid lies on that surface of the helix, which faces away from the active site of the kinase and that is known to provide an interface for interactions with regulatory domains and proteins (Fig. 5B). For human CDK2, for example, it was shown that residue Lys56, equivalent to the position of Thr397 in crk10-A397T, is involved in its interaction with cyclin necessary for the stabilization of the active form of the kinase (Jeffrey et al., 1995), whereas in the case of the Ser/Thr-protein kinase B-Raf (BRaf), where dimerization is thought to be an important part of the activation mechanism, the αC helix/αC-β4 loop provides the interface for dimer formation (Rajakulendran et al., 2009). This region can also function as a cis-regulatory site as shown for the human leucine-rich repeat kinase 2 (LRRK2), where it provides a firm docking site for the C-terminal residues of the kinase that keeps it in an inactive conformation (Deniston et al., 2020;Taylor et al., 2020). As the αC helix/αC-β4 loop is a feature common to all eukaryotic kinases, it seems reasonable to speculate that the amino acid substitution in crk10-A397T could perturb the interaction of CRK10 with a regulatory partner. It is noteworthy that the αC helix and αC-β4 loop is a highly conserved region within the family of CRKs, and only three residues are found to occupy the position equivalent to Ala397 of CRK10 among all the members: alanine, threonine, or serine (Supplementary Figs S15, S16). It now remains to be seen whether other members of the CRK family are similarly affected by an analogous mutation in their αC helix/ αC-β4 loops. It also is notable that the only other semi-dominant mutant reported for a CRK in rice, als1, which develops spontaneous lesions on leaf blades and sheaths, harbours a point mutation that also localizes in the vicinity of the αC helix (Du et al., 2019).
In situ phosphorylation analysis of the cytoplasmic kinase domains of WT and mutant CRK10 by LC-MS/MS determined that CRK10, being an RD kinase, shows the typical autophosphorylation pattern of conserved phosphorylation sites in the activation loop, on which the activity of this class of kinases depends ( Fig. 5E; Nolen et al., 2004;Beenstock et al., 2016). Thr507 and Ser508 were identified as unambiguous phosphorylation sites, with Ser508 likely to be functionally equivalent to Thr450 of BAK1, which plays a key role in its activation by maintaining the active conformation of the activation loop (Yan et al., 2012). However, we also detected phosphorylation of tyrosine residues (Tyr363 and Tyr514), with Tyr514 residing in the activation loop, which establishes CRK10 as a dual specificity kinase. Additional phosphorylated residues were detected in the juxtamembrane domain (Thr340) and at the C-terminus (Thr625, Ser662, and Thr664). Phosphorylation of residues in these non-catalytic regions has been shown to play an important role in the recognition and/or phosphorylation of downstream substrates, although these are unknown for CRKs Oh et al., 2012;Zhou et al., 2020). The phosphorylation pattern of His-CRK10kd WT and His-CRK10kd A397T was identical, with the notable exception of the substituted Thr397 functioning as an additional phosphorylation site in His-CRK10kd A397T (Fig. 5E). Whether these phosphorylation sites, which were identified in vitro, do play a role in vivo and the determination of their biological significance are questions which need to be addressed in future work. For example, introducing a phosphomimic variant of threonine 397 in planta could ascertain that there is a link between the phosphorylation of threonine 397 and the crk10-A397T mutant phenotype. It will also be important to determine if/ how the mutation affects kinase activity. As substrates phosphorylated by CRK10 remain unknown, artificial substrates such as myelin basic protein could be used in in vitro kinase assays in order to unravel whether the crk10-A397T phenotype can be linked to, for example, enhanced kinase activity. In addition, interacting partners of the WT and mutant variants of CRK10 in vivo could be identified, which will help to establish whether the A397T mutation perturbs and/or promotes protein-protein interactions, thus affecting the dynamics of the receptor complexes in which CRK10 resides.
Phenotypically the crk10-A397T mutant is a dwarf with severely collapsed xylem elements in the roots and hypocotyl, whereas xylem elements in the inflorescence stem remain intact (Figs 1, 2). A grafting experiment suggested that it is this defective vasculature in the belowground organs that causes the dwarf phenotype, as WT scions grafted onto crk10-A397T rootstocks become similarly stunted, whereas crk10-A397T scions grafted on WT rootstocks develop normally (Fig. 4). As CRK10 is expressed in vascular-associated tissues in both stem and hypocotyl, as shown by reporter lines (Fig. 3A, B) and qPCR, a tissue-specific expression of CRK10 cannot, therefore, explain why in the crk10-A397T mutant the xylem elements are defective in one organ but not the other. In the absence of detailed knowledge on how the mutation alters CRK10 function, an explanation for this observed discrepancy remains highly speculative. For example, one straightforward hypothesis could envisage that the alanine 397 substitution by threonine alters the substrate recognition domain and confers to it the ability to phosphorylate substrates that are present in the hypocotyl but absent in the stem. Alternatively, stem and hypocotyl might differ in the composition of the receptor complexes within which CRK10, in analogy to other receptor-like kinases, is anticipated to reside. As a result, the interaction with regulatory partners within the complex might be differently affected by the crk10-A397T mutation leading, for example, to kinase activation in one tissue but not the other.
Collapsed xylem vessels are thought to be a consequence of alterations in cell wall composition, leading to cell wall weakening and an inability to withstand the negative pressure generated by transpiration, as reported, for example, in the eskimo and irx mutants (Turner and Somerville, 1997;Lefebvre et al., 2011). The collapse of xylem vessels in the hypocotyl of the crk10-A397T mutant plants also seems to be accompanied by changes in cell wall composition, as suggested by the transcriptional reprogramming of multiple cell wall-related genes revealed in our transcriptomic dataset. Corroborating this finding, vessel cell wall analysis with FTIR and monosaccharide analysis of the hypocotyl revealed that changes occurring in the cell wall composition of the mutant were complex with several compounds being affected, amongst which hemicellulose and ester cross-links were the most prominent. The alteration in chemical composition, however, was not reflected in alterations of the ultrastructure of the vessel cell walls, as transmission electron microscopy analysis performed on cross sections of 3-week-old hypocotyls of the crk10-A397T mutant and wild-type was unable to detect any obvious differences. Preliminary analysis for the determination of the underlying cause of the vessel collapse remains therefore inconclusive and warrants a separate line of investigation that is beyond the scope of this work.
Collapsed vessel elements very likely impede water transport, which could be perceived as drought stress leading to an increase of ABA. This could explain the elevated levels of ABA that we detected in the hypocotyl of crk10-A397T and the up-regulation of genes involved in ABA synthesis, perception, and response (Supplementary Table S10). A similar increase in ABA levels has been observed in other cell wall mutants such as esk1 and irx1-6 whereas irx3, irx5, and irx9 contained numerous constitutively up-regulated ABA-responsive genes (Chen et al., 2005;Hernández-Blanco et al., 2007;Lefebvre et al., 2011;Faria-Blanc et al., 2018;Xu et al., 2020). The stunting of these mutants has been suggested to be the consequence of the response to drought signalling hormones resulting in the suppression of growth, which could also explain the dwarfism of crk10-A397T. It will be interesting to investigate whether dwarfing of crk10-A397T can be alleviated by an ABA-insensitive mutant such as abi1 (Koornneef et al., 1984).
Alteration of cell wall composition can lead not only to loss of cell wall rigidity, as it is increasingly being reported that modification of cell wall composition by genetic or chemical means leads to the constitutive activation of defence pathways and an altered resistance to pathogens. The primary cell wall mutant ixr1/cev1, for example, with reduced crystalline cellulose content due to the defective cellulose synthase CESA3, displays constitutive activation of JA and ethylene signalling (Ellis and Turner, 2001;Ellis et al., 2002). Transcriptomic data obtained for the secondary cell wall mutants irx1-6, irx5-5, and irx9 showed that defence-related genes are constitutively expressed in these mutants (Hernández-Blanco et al., 2007;Faria-Blanc et al., 2018). In line with these reports, our data showed that the signature of the core set of DEGs of crk10-A397T was most similar to the transcriptome of plants responding to biotic and abiotic stress (Fig.  6). Canonical SA-dependent marker genes (PR1, PR2, and PR5) and genes involved in the synthesis of the tryptophanderived antimicrobial compounds (camalexin, glucosinolates) are significantly up-regulated in the mutant, as are numerous transcription factors usually associated with co-ordinating stress responses (Supplementary Tables S8, S9). Concomitant with gene induction, SA levels in the crk10-A397T mutant hypocotyls are increased, whereas changes in JA levels were not significant. Induction of defence pathways due to cell wall impairment manifests itself frequently in alteration of disease resistance to a wide range of pathogens (Houston et al., 2016;Bacete et al., 2018). Molina et al. (2021), for example, showed that from a panel of 34 cell wall mutants affecting a wide range of different cell wall compounds, 29 had an altered, mainly enhanced, resistance response to pathogens comprising different parasitic lifestyles. In order to assess disease resistance of crk10-A397T we chose the root-infecting, hemi-biotrophic vascular wilt pathogen F. oxysporum to perform a pathogen assay, bearing in mind the vasculature-associated expression of CRK10. In agreement with the studies linking cell wall modification to altered disease resistance, we found that the crk10-A397T mutant was significantly more resistant to the pathogen, part of which could be due to the fact that collapsed xylem vessels act as physical barriers slowing pathogen progression in the roots (Fig. 7).
Taken together, the experimental evidence provided in this report suggests that crk10-A397T is a gain-of-function allele of CRK10 as activation of defence pathways occurs in the absence of a stimulus in the mutant. However, our observations do not currently allow us to infer the biological function of CRK10 as, according to Muller (1932), gain-of-function phenotypes can be the consequence of different types of mutations, namely neomorphic, hypermorphic, or antimorphic. As long as mode of action and regulation remain uncharacterized for CRKs, it is not possible to ascertain which type of gain-offunction mutation the crk10-A397T allele represents, which makes it difficult to determine the role of CRK10 in planta.
Another outstanding question in order to uncover the role of CRK10 remains the identification of its ligand. Intriguingly, plant DUF26 domains share strong structural similarity with fungal carbohydrate-binding modules, which led Vaattovaara et al. (2019) to propose that DUF26 proteins could be carbohydrate-binding domains, and cell-wall-derived carbohydrates or small extracellular molecules could represent candidate ligands. However, efforts to prove carbohydrate binding to the DUF26 domain have so far not been successful and bona fide ligands are still awaiting discovery.

Supplementary data
The following supplementary data are available at JXB online. Fig. S1. crk10-A397T is a semi-dominant allele of CRK10. Fig. S2. crk10-A397T mutant seeds germinate earlier than the WT and exhibit increased seedling size. Fig. S3. Reduction in rosette size and aborted apical meristem of crk10-A397T plants. Fig. S4. Xylem vessels in the stem of the crk10-A397T mutant do not collapse and lignification in the stem of WT and the crk10-A397T mutant is restricted to xylem vessels and fibres. Fig. S5. CRK10 is expressed in vascular tissues and the protein localizes to the plasma membrane. Fig. S6. Transgenic plants overexpressing the CRK10 transcript develop normally and resemble the WT. Fig. S7. T-DNA knockout mutants of CRK10 develop normally and resemble the WT. Fig. S8. The hypocotyls of crk10-2 and CRK10 OE-1 plants do not contain collapsed xylem vessels. Fig. S9. Alignment of the activation segment of eukaryotic kinase domains shows highly conserved residues and phosphorylation sites.   S11. The FTIR spectrum profile of collapsed xylem vessels in the crk10-A397T mutant hypocotyls shows marked differences from the spectrum of intact vessels in the WT. Fig. S12. The total monosaccharide content of hypocotyls of crk10-A397T mutant plants shows marked differences from that of WT plants. Fig. S13. The ultrastructure of the secondary cell wall of collapsed xylem vessels resembles that of intact vessels in the WT. Fig. S14. Fungal burden quantification at 2 and 7 d postinoculation with F. oxysporum. Fig. S15. Alignment of the αC helix segment of the kinase domain of the CRK family from Arabidopsis shows members of the family that contain alanine, threonine, or serine residues on position equivalent to Ala397 in CRK10. Fig. S16. Phylogenetic tree of the CRK family from Arabidopsis and their respective residue on position equivalent to Ala397 in CRK10. Table S1. Table of primers.  Tables S2-S5. Differentially expressed genes in the crk10-A397T mutant hypocotyls.
Tables S6, S7. Gene Ontology analysis. Tables S8-S11. Tables of differentially expressed genes per functional category. Table S12. Quantification of hormones in the hypocotyl of WT and crk10-A397T mutant plants.