Constitutive activation of leucine-rich repeat receptor kinase signaling pathways by BAK1-interacting receptor-like kinase 3 chimera

Receptor kinases with extracellular leucine-rich repeat domains (LRR-RKs) form the largest group of membrane signaling proteins in plants. LRR-RKs can sense small molecule, peptide or protein ligands, and may be activated by ligand-induced interaction with a shape complementary SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE (SERK) co-receptor kinase. We have previously shown that SERKs can also form constitutive, ligand-independent complexes with the LRR ectodomains of BAK1-interacting receptor-like kinase 3 (BIR3) receptor pseudokinases, negative regulators of LRR-RK signaling. Here we report that receptor chimaera in which the extracellular LRR domain of BIR3 is fused to the cytoplasmic kinase domains of the SERK-dependent LRR-RKs BRASSINOSTEROID INSENSITIVE1, HAESA and ERECTA form tight complexes with endogenous SERK co-receptors in the absence of ligand stimulus. Expression of these chimaera under the control of the endogenous promoter of the respective LRR-RK leads to strong gain-of-function brassinosteroid, floral abscission and stomatal patterning phenotypes, respectively. Importantly, a BIR3-GSO1/SGN3 chimera can partially complement sgn3 Casparian strip formation phenotypes, suggesting that GSO1/SGN3 receptor activation is also mediated by SERK proteins. Collectively, our protein engineering approach may be used to elucidate the physiological functions of orphan LRR-RKs and to identify their receptor activation mechanism in single transgenic lines.


Introduction
Plant-unique membrane receptor kinases characterized by an extracellular domain, a single membrane spanning helix and a cytoplasmic dual-specificity kinase domain control many aspects of plant growth and development, form the first layer of the plant immune system and mediate symbiotic interactions (Hohmann et al., 2017). LRR-RKs form the largest class of receptor kinases known in plants (Shiu and Bleecker, 2001). Members of the family have been shown to sense small molecule (Wang et al., 2001), peptide (Gómez-Gómez and Boller, 2000;Matsubayashi, 2014;Santiago et al., 2016) and protein ligands (Huang et al., 2016;Lin et al., 2017;Zhang et al., 2017).
Complex structures and quantitative biochemical comparisons of different ligand-activated LRR-RK -SERK complexes have revealed a structurally and functionally conserved activation mechanism, relying on the interaction of the ligand bound receptor LRR ectodomain with the shape-complementary ectodomain of the SERK co-receptor (Santiago et al., 2013;Wang et al., 2015;Santiago et al., 2016;Hohmann et al., 2017;Lin et al., 2017;Hohmann et al., 2018b). The ligand binding specificity of plant LRR-RKs is encoded in their LRR ectodomains (Hohmann et al., 2017;Okuda et al., 2020). The kinase domain of the receptor, not of the SERK co-receptor, confers cytoplasmic signaling specificity (Hohmann et al., 2018b;Chen et al., 2019;Zheng et al., 2019).
Protein engineering approaches have been previously employed to dissect the LRR-RK receptor activation in planta: A fusion protein combining the extracellular and trans-membrane domains of BRI1 (outerBRI1, oBRI1) with the cytoplasmic kinase domain of the rice immune receptor XA21 (innerXA21, iXA21) could trigger immune signaling in rice cells upon stimulation with brassinisteroids (He et al., 2000). It is now known that both BRI1 and XA21 rely on SERK coreceptor kinases for receptor activation Nam and Li, 2002;Santiago et al., 2013;Hohmann et al., 2018b;Chen et al., 2014). The heteromeric nature of LRR-RK -SERK complexes has been validated in planta using similar protein engineering approaches. Co-expression of a protein chimera of the immune receptor FLAGELLIN SENSITIVE 2 (FLS2) and its co-receptor SERK3 (oFLS2-iSERK3) with an oSERK3-iFLS2 construct led to immune signaling after stimulation with the FLS2 ligand flg22 in a transient expression system (Albert et al., 2013). Stable transgenic lines co-expressing oBRI1-iSERK3 and oSERK3-iBRI1 construct could partially rescue the BRI1 loss-of-function mutant bri1-301 (Hohmann et al., 2018b).
The signaling specificity of the cytoplasmic kinase domain of LRR-RKs has been dissected using an oBRI-iHAESA chimera, which rescued the floral abscission phenotypes when expressed under the control of the HAESA promoter in the haesa hsl2 double mutant (Hohmann et al., 2018b).
A similar approach has been recently used to demonstrate that the LRR-RKs BRI1 and EMS1 share a common cytoplasmic signaling cascade (Zheng et al., 2019). However, all these approaches rely on ligand stimulus.
Expressing these chimera under the control of endogenous/context-specific promoters, we obtain strong gain-of-function phenotypes for different LRR-RK triggered developmental signaling pathways. In addition, an oBIR3-iGSO1/SGN3 chimera supports a SERK-dependent activation mechanism for the LRR-RK GASSHO1/SCHENGEN3 in Casparian strip formation (Pfister et al., 2014;Okuda et al., 2020). Our strategy allows for the identification of gain-of-function phenotypes of orphan LRR-RKs whose ligands are unknown, and enables the elucidation of their receptor activation mechanism.

Results
We compared the structure of a previously reported BRI1 -brassinolide (BL) -SERK1 complex (Protein Data Bank ID 4SLX, http://rcsb.org) with the recently reported complex structure of a BIR3 -SERK1 complex (PDB-ID 6FG8) (Santiago et al., 2013;Hohmann et al., 2018a). The 5 121 122 123 124 125 BRI1 and BIR3 ectodomains bind SERK1 using non-identical but overlapping binding surfaces ( Figure 1A). As in the BRI1 -SERK1 complex, the C-termini of BIR3 and SERK1 are in close proximity in the complex structure ( Figure 1B,C). Based on the structural similarities, we generated an oBIR3 -iBRI1 chimera, in which the BIR3 ectodomain and trans-membrane helix are connected to the cytoplasmic domain of BRI1 ( Figure 1D) (see Methods).
We expressed the oBIR3 -iBRI1 fused to a C-terminal mCitrine (mCit) fluorescent protein tag under the control of the pBRI1 promoter in a previously characterized bri1 null mutant (Jaillais et al., 2011). oBIR3 F146A/R170A -iBRI1 and oBIR3 -iBRI1 D1027N , which block BIR -SERK complex formation (Hohmann et al., 2018a) and BRI1 kinase activity (Bojar et al., 2014;Hohmann et al., 2018b), respectively, were used as controls. Independent oBIR3 -iBRI1 transgenic lines, but none of the control lines displayed the wavy hypocotyl phenotype characteristic of gain-of-function brassinosteroid mutants (Figure 2 A). Importantly, the wavy hypocotyl phenotype observed in oBIR3 -iBRI1 lines was also visible when plants were grown in the presence of the brassinosteroid biosynthesis inhibitor brassinazole (BZR) (Asami et al., 2000) ( Figure 2A). This suggests that oBIR3 -iBRI1 triggered brassinosteroid signaling does not depend on endogenous brassinosteroids ( Figure 2A). Consequently, we found all oBIR3-iBRI1 but none of the control lines to be constitutively active when expressed in the det2 background (Chory et al., 1991), in which brassinosteroid levels are reduced (Fujioka et al., 1997) (Figure 2A). Quantification of three independent oBIR3-iBRI1 T3 lines revealed strong gain of function phenotypes, which are even more pronounced than the previously reported phenotype of the constitutively active bes1-1D mutant ( Figure 2A) (Yin et al., 2002). Consistent with a constitutive activation of the brassinosteroid signaling, we found BES1 to be dephosphorylated in oBIR3 -iBRI1 but not in the control lines ( Figure 2B), and BES1 dephosphorylation to also take place in the det2 background ( Figure 2C). We next performed co-immunoprecipitation (co-IP) experiments in our stable lines and found oBIR3 -iBRI1 and oBIR3 -iBRI1 D1027N to efficiently interact with the endogenous SERK3 co-receptor in vivo, while the oBIR3 F146A/R170A -iBRI1 control, which disrupts the interaction of the isolated BIR3 and SERK1/3 ectodomain in vitro (Hohmann et al., 2018a), could no longer bind SERK3 in vivo ( Figure 2D). Taken together, the BIR3 ectodomain can promote brassinosteroid independent interaction with SERK3, and possibly other SERKs in vivo, resulting in a constitutive activation of the brassinosteroid signaling pathway. The control lines further suggest that this signaling complex is formed and stabilized by the ectodomains of BIR3 and SERK3, and requires the catalytic activity of the BRI1 kinase domain for signaling ( Figure 2A).
We next tested if BIR3-based protein chimera can be used to activate a functionally distinct LRR-RK signaling pathway. The LRR-RK HAESA (HAE) shares the overall structure and 6 155 156 activation mechanism with BRI1 (Santiago et al., 2013(Santiago et al., , 2016Hohmann et al., 2018b), but the two receptors control very different developmental processes (Li and Chory, 1997;Jinn et al., 2000). We expressed an oBIR3-iHAE fusion protein ( Figure 3A) fused to a C-terminal mCit tag under the control of the pHAE promoter in the hae hsl2 mutant, in which floral organ abscission is delayed (Stenvik et al., 2008). We observed that expression of oBIR3-iHAE but none of the control lines rescued the floral abscission phenotype of the hae hsl2 mutant ( Figure 3B, C). Consistently, we found oBIR3 -iHAE and oBIR3 -iHAE D1027N but not oBIR3 F146A/R170A -iHAE to interact with SERK3 in co-IP assays ( Figure 3D). SERK proteins have been previously shown to allow for receptor activation of ERECTA family receptor kinases in protoderm formation and stomatal patterning (Meng et al., 2015). ER forms constitutive complexes with the LRR-RLP TMM to sense EPF peptides in stomatal patterning (Yang and Sack, 1995;Nadeau and Sack, 2002;Lee et al., 2012Lee et al., , 2015Lin et al., 2017), but it is not understood at the mechanistic level how SERK co-receptor kinases allow for receptor activation of this LRR-RK/LRR-RLP signaling complex (Lin et al., 2017). To test if the receptor activation mechanism is conserved among BRI1, HAESA and ER, we expressed a chimeric oBIR3-iER construct fused to a C-terminal yellow fluorescent protein YPET specifically in the stomata lineage under control of the meristemoid-specific MUTE promoter ( Figure 4A) (Pillitteri et al., 2007). Previous experiments demonstrated that constitutive activation of the ERECTA pathway in differentiating meristemoids leads to developmental arrest of guard mother cells (Lampard et al., 2009). To test the signaling specificity of our oBIR3-iER chimera, we also expressed a chimeric fusion of the innate immunity receptor FLS2 (Gómez-Gómez and Boller, 2000) under the control of the MUTE promoter (oBIR3-iFLS2-YPET) ( Figure 4A).
For each construct, we selected three representative lines according to YPET expression and measured density of mature stomata on cotyledons. The oBIR3-iER lines showed a drastic reduction of mature stomata and an increase in meristemoid-like cells on the leave surface ( Figure   4B, C). In contrast, none of the oBIR3-iFLS2 lines showed any significant deviation from the wildtype phenotype, despite being expressed at a similar or higher level than the BIR3-ER chimeras ( Figure 4D).
To analyze the observed phenotype on a molecular level, we tested expression of the guard mother cell (GMC)-specific transcription factor FAMA (Ohashi-Ito and Bergmann, 2006) and guard cell-specific Dof-type transcription factor SCAP1 (Negi et al., 2013). The three independent BIR3-ER lines displayed a strong reduction of FAMA and SCAP1 expression ( Figure 4E Finally, we tested if fusion of the BIR3 ectdomain to the LRR-RK GSO1/SGN3 (Tsuwamoto et al., 2008;Pfister et al., 2014) could restore the apoplastic barrier defects in the sgn3-3 mutant (Pfister et al., 2014). GSO1/SGN3 directly senses the peptide ligands CASPARIAN STRIP INTEGRETY FACTORS 1 and 2 (CIF1, CIF2) to ensure proper formation of the Casparian strip, an endodermal diffusion barrier enabling selective nutrient uptake in the root (Pfister et al., 2014;Nakayama et al., 2017;Doblas et al., 2017;Okuda et al., 2020). A biochemical interaction screen has recently identified SERK proteins as putative co-receptor kinases for GSO1/SGN3 (Okuda et al., 2020), but it is presently unclear if SERKs mediate GSO1/SGN3 receptor activation in vivo ( Figure 5A). We generated oBIR3-iSGN3, oBIR3-iSGN3 F146A,R170A and oBIR3-iSGN3 D1102N protein chimera expressed under the control of the pSGN3 promoter in the sgn3-3 mutant background. As previously described, the sgn3-3 mutant has a non-functional apoplastic barrier that can be visualized and quantified by visualizing the uptake of the apoplastic tracer propidium iodide (PI) along the root and its access to the central vasculature ( Figure 5B,C). We found that oBIR3-iSGN3 but none of the control lines restored sgn3-3 apoplastic defects ( Figure 5C, D), indicating a SERK-mediated GSO1/SGN3 receptor activation mechanism in the Casparian strip formation.

Discussion
The identification of a constitutive, ligand-independent interaction between the LRR ectodomains of two plant membrane signaling proteins prompted us to investigate if protein chimera between the BIR3 ectodomain and the cytoplasmic domain of various receptor kinases would lead to constitutively active signaling complexes. Despite the significant structural differences between LRR-RK -SERK and BIR -SERK complexes, our data demonstrates that a wide range of oBIR3 -iLRR-RK chimera are functional in planta.
Expression of the oBIR3 -iBRI1 chimera resulted in a strong, constitutive activation of the brassinosteroid signaling pathway. The gain-of-function effect is much stronger than previously described for the BRI1 sud1 and SERK3 elongated alleles, respectively (Belkhadir et al., 2012;Jaillais et al., 2011;Hohmann et al., 2018a), and comparable to constitutive activation of BES1 ( Figure 2A) (Yin et al., 2002). The constitutive signaling activity of the oBIR3 -iBRI1 chimera depends on the ability of the BIR3 ectodomain to bind SERK ectodomains and on the kinase activity of the BRI1 cytosolic segment (Figure 2). This reinforces the notion that formation of the heterodimeric extracellular signaling complex drives LRR-RK receptor activation, and that signaling specificity is encoded in the kinase domain of the receptor, not the co-receptor (Bojar et al., 2014;Hohmann et al., 2018b;Zheng et al., 2019). The phenotypes of oBIR3 -iBRI1 and bes1-1D plants in addition suggest that little signal amplification appears to occur throughout the brassinosteroid signaling pathway (Figure 2).
Analysis of the oBIR3 -iHAE chimera revealed a strongly conserved activation mechanism between different, SERK-dependent LRR-RK signaling pathways, as previously suggested (Hohmann et al., 2018b) (Figure 3). In addition, our experiments suggest that BIR ectodomains are able to interact with SERK proteins in the abscission zone, and thus BIR proteins may act as negative regulators of HAESA / HSL2 mediated signaling cascades in wild-type plants (Figure 3).
In this respect, it is noteworthy that the BIR suppressor SoBIR1/EVERSHED has been previously characterized as a genetic component of the floral abscission signaling pathway (Leslie et al., 2010).
ERECTA family kinases have been previously shown to require SERK co-receptor kinases to control stomatal patterning and immune responses (Meng et al., 2015;Jordá et al., 2016). Our functional oBIR3 -iER chimera now suggests, that, despite the requirement for TMM, EPF bound ER signaling complexes are activated by SERK proteins in very similar ways as previously reported other LRR-RKs (Hohmann et al., 2017) (Figure 4). Expression of the oBIR3 -iER chimera in meristemoid cells lead to a similar phenotype as described for the expression of constitutively active versions of YODA, MKK4, and MKK5 (Lampard et al., 2009). This strongly indicates that the oBIR3 -iER chimera displays constitutive, ligand-independent signaling activity.
The specificity of signal transduction seems to be largely maintained, as expression of oBIR3 -iFLS2 led to wild-type like stomatal development. At the molecular level, however, we observed a significant increase of FAMA expression in all tested oBIR3 -iFLS2 lines. This is consistent with an antagonistic regulation of these two pathways (Sun et al., 2018). The observed upregulation of FAMA expression, however, did not significantly alter stomata density. This is likely because the transcriptional activation of the oBIR3-iFLS2 construct in this experiment happens only in meristemoid cells and might be compensated by post-transcriptional regulation.
Taken together, the simple, lego-style assembly of BIR3 chimera (Figure 7) and the availability of suitable control lines now enables the genetic characterization of orphan LRR-RKs with unknown/unclear loss-of-function phenotypes and dissection of their potential activation mechanism. BIR3 protein chimera may also be of use for biochemical or genetic interaction screens, in which a constitutively active form of the receptor is desirable.

Plant material, growth conditions and generation of transgenic lines
To design chimeric receptor kinases, the transmembrane helix of all LRR-RKs was first predicted using TMHMM (version 2.0, https://services.healthtech.dtu.dk/service.php?TMHMM-2.0) (Krogh et al., 2001). The native signal peptide, extracellular domain and the transmembrane helix from AtBIR3 (residues 1-246) were fused to the juxtamembrane and kinase domains of the respective receptor (BRI1 residues 815-1196, HAE 649-999, SGN3 899-1249). No additional linker sequences were added (Figure 7). Fragments were amplified from Arabidopsis thaliana (ecotype Col-0) genomic or cDNA and cloned into pDONR221 (ThermoFisher Scientific) using Gibson-cloning technology; mutations were introduced through site directed mutagenesis (Supplemental Table 1). Binary vectors were assembled using the multi-site Gateway technology into the binary vector pB7m34GW, harboring a Basta resistance gene (ThermoFisher Scientific). All constructs were introduced into Agrobacterium tumefaciens strain pGV2260, and Arabidopsis plants were transformed using the floral dip method (Clough and Bent, 1998).

Hypocotyl growth assay
Seeds were surface sterilized, stratified at 4 °C for 2 d, and plated on ½ MS, 0.8 % agar plates supplemented with 1 VM brassinazole (BRZ, from a 10 mM stock solution in 100 % DMSO, Tokyo Chemical Industry Co. LTD) or, for the controls, with 0.1 % (v/v) DMSO. Following a 1 h light exposure to induce germination, the plates were wrapped in aluminium foil and incubated in the dark at 22 °C for 5 d. The plates were then scanned at 600 dpi on a regular flatbed scanner (CanoScan 9000F, Canon), hypocotyl lengths measured using Fiji (Schindelin et al., 2012) and analyzed in R (R Core Team, 2014) (version 3.6.1) using the packages mratios (Kitsche and Hothorn, 2014) and multcomp (Hothorn et al., 2008). Rather than p-values, we report unadjusted 95% confidence limits for fold-changes. A mixed effects model for the ratio of of a given line to the wild-type Col-0, allowing for heterogeneous variances, was used to analyze log-transformed endpoint hypocotyl lengths. To evaluate the treatment-by-mutant interaction, the 95 % two-sided confidence intervals for the relative inhibition (Col-0: untreated vs. BRZ-treated hypocotyl length)/(any genotype: untreated vs. BRZ-treated hypocotyl length) was calculated for the logtransformed length.

Plant protein extraction and immunoprecipitation
Seeds were plated on ½ MS, 0.8 % agar plates and grown for ~ 14 d after surface sterilization and stratification. Seedlings were harvested, padded dry carefully, snap-frozen in liquid N 2 , and ground to fine powder using pre-cooled mortar and pestel. 1 g of powder per sample was resuspended in 3 ml of ice cold extraction buffer (50 mM Bis Tris pH 7.0, 150mM NaCl, 10 % (v/v) glycerol, 1 % Triton X-100, 5 mM DTT, protease inhibitor cocktail (P9599, Sigma)) and agitated gently at 4 °C for 1 h. Subsequently, samples were centrifuged (30 min, 16,000 g, 4 °C), the supernatant then transferred to a fresh tube and the protein concentration estimated through a Bradford assay with a BSA standard curve.

Stomata density measurements and microscopy
Seven-day old T2 seedlings were used to determine stomata density. For confocal imaging, seedlings were incubated in 10mg/L propidium iodide (PI) solution for 30 min, and then washed with water. Abaxial epidermal regions of cotyledons were imaged using a Zeiss LSM 780 NLO microscope equipped with a Plan-Apochromat 25x/0.8 Imm Korr DIC objective. PI staining was visualized with an excitation wave length of 514 nm and emission was recorded between 566 nm and 643 nm. Mature stomata were counted in a 0.25 mm by 0.25 mm epidermal area for three seedlings of each line.

Expression analysis
Seven-day old T2 seedlings were used to analyze expression levels of the transgene as well as endogenous genes. For each independent line, RNA was extracted from 24 pooled T2 seedlings using the RNase® Plant Mini Kit (QiaGen). cDNA was synthesized using the RecertAid First Strand cDNA Synthesis Kit (Thermo Scientific). Relative abundance of the endogenous FAMA and SCAP1 transcripts as well as chimeric YPET-containing BIR3 transcripts were measured by quantitative RT-PCR (program: 1. 50 for 10 min, 2. 95 for 5 min, 3. 95 for 10 s, 4. 60 for ℃ for 10 ℃ for 10 ℃ for 10 ℃ for 10 30 s; repeat step 3 -4 40 times; 5. 95 for 10 s, 6. ramp 65 to 95 and increase 0.5 every 5s, ℃ for 10 ℃ for 10 ℃ for 10 Propidium iodide permeability assay and confocal microscopy of wild-type and complemented sgn3-3 plants. Propidium Iodide (PI) permeability assay were performed on 5 d old seedlings. In brief, the seedlings were stained in dark for 10 mins in 10µg/ml PI, rinsed twice in water and quantified as previously described (Naseer et al., 2012). Endodermal cell numbers were quantified using a Leica Epifluorescence microscope. Representative confocal images were acquired with a Leica SP8, with excitation and detection windows set as follows for PI: excitation -488 nm, emission -500-550 nm.
Confocal images were processed and analyzed using ImageJ. For samples treated with the CIF2 peptide, the seedlings were grown on ½ MS for 3 days followed by transfer to ½ MS + 10µM CIF2 peptide for 2 d and subsequently analyzed for PI permeability. Statistical analyses were done in the R environment (R Core Team, 2014). For multiple comparisons between genotypes, Kruskal-Wallis' test was performed and nonparametric Tukey's test was subsequently used as a multiple comparison procedure. Different letters indicates significant difference (P<0.05). Data are presented as box plots overlaid with dot plots.

Protein expression, purification and size exclusion chromatography
The coding sequence of AtNIK1 32-248 was amplified from Arabidopsis thaliana cDNA,     deviations. Individual data points are shown as dots. Expression in wild type was arbitrarily set to 1.
Significant differences to wild-type levels are indicated by an asterisk (t-test; p<0.05).
(F) Relative normalized expression of SCAP1. Normalized expression values determined by quantitative RT-PCR are shown as average of three replicates. Error bars display standard deviations. Individual data points are shown as dots. Expression in wildtype was arbitrarily set to 1.
Significant differences to wild-type levels are indicated by an asterisk (t-test; p<0.05).      Depicted are box plots (center line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range; points, outliers) together with the raw data (depicted as individual dots, grouped per plate) and mean ± standard deviation alongside. The raw data for oBIR3-iBRI1 chimera in the bri1-null background is shown in (A) and in the det2 -background in (B). Untreated: white, BRZ treated: blue. For each sample n=50 biologically independent hypocotyls, coming from 5 different ½MS plates, have been measured.