Cryptic bioactivity capacitated by synthetic hybrid plant peptides

Evolution often diversifies a peptide hormone family into multiple subfamilies, which exert distinct activities by exclusive interaction with specific receptors. Here we show that systematic swapping of pre-existing variation in a subfamily of plant CLE peptide hormones leads to a synthetic bifunctional peptide that exerts activities beyond the original subfamily by interacting with multiple receptors. This approach provides new insights into the complexity and specificity of peptide signalling.

activity similar to CLE41 (Fig. 1f), even though KIN was created by the swap of amino-acid residues between the two CLV3-type peptides, CLV3 and CLE25. We further analysed dose-response relationships in this assay (Fig. 1g). CLV3 exerted a negative effect on the stele growth at as low as 30 nM, whereas CLE41 showed a positive effect in higher concentrations (41 mM). Interestingly, KIN exhibited negative effects in lower concentrations, whereas conversely it displayed positive effects in higher concentrations, demonstrating that KIN exerts both CLV3 and CLE41 types of activities by itself.
Genetic dissection of bifunctional CLE bioactivity. The dual activity of KIN could be attributed to its target receptors. CLV1 and CLV2 are receptor genes involved in CLV3 signalling and the SAMs of their loss-of-function mutants are resistant to CLV3 treatment 9,17,18 . Unlike the wild-type SAM (Fig. 1d), the mutant SAMs were maintained even after a 10-day treatment of 10 mM CLV3 and KIN (Fig. 2a), suggesting that KIN acts through intrinsic CLV1/CLV2 pathways. Furthermore, CLV2, but not CLV1, is responsible for the root-shortening activity of CLV3 peptide 18 . Indeed, the clv2-101 mutant was resistant to KIN and to CLV3 as well in terms of both the root length and RAM size ( Supplementary Fig. 3a,b), indicating that KIN exerts the root-shortening activity via CLV2.
We next examined responses of these mutants to the peptides in the stele-thickening assay. As described above, a lower concentration (100 nM) of CLV3 or KIN reduced the stele width of wild-type plants, whereas a higher concentration (10 mM) of CLE41 or KIN thickened the stele (Figs 1g and 2b). In contrast to our assays where plants were exposed to CLV3 soon after germination, the CLV3-type inhibitory activity was not observed in the previous report in which 3-day-old seedlings were treated with the peptide 8 , suggesting that sensitivity to CLV3 might differ by plant age. The response pattern of clv1-101 mutant was similar to that of the wild type, suggesting that CLV1 does not participate in CLE signalling in stele thickening. In contrast, clv2-101 mutant was insensitive to the inhibitory activity of 100 nM CLV3 and KIN (Fig. 2b). Strikingly, 10 mM KIN showed a stronger effect than CLE41 in the clv2-101 mutant (Fig. 2b). This phenomenon was similar to the previously reported synergistic effect of the simultaneous treatment of CLV3 and CLE41, which does not require functional CLV2 (ref. 8). Simultaneous treatment of 10 mM CLV3 and 10 mM CLE41 showed a strong activity similar to the KIN treatment in clv2-101, although these activities were not observed in wild type in our experimental conditions (Fig. 2b). As clv2-101 is insensitive to the inhibitory activity of CLV3 in stele thickening (Fig. 2b), this mutant serves as an ideal genetic background to detect the positive effect of CLV3-type peptides. Indeed, the dose-response assay in clv2-101 showed that, in the presence of 10 mM CLE41, both CLV3 and KIN increase the stele thickening at concentrations above 1 mM (Supplementary Fig. 4). These data further support the notion that KIN exerts both activities of CLV3 and CLE41 by itself.
CLE41 treatment causes discontinued xylem strands in leaf vein due to its inhibitory activity on differentiation of undifferentiated vascular cells into xylem cells 7 . To further confirm whether KIN behaves similar to CLE41, we examined xylem strands after the KIN treatment. In this analysis, we used clv2-101, because the mutant is resistant to growthinhibitory effects caused by CLV3 and KIN, and therefore we could obtain leaves at a comparable growth stage between different peptide treatments. We found that, similar to CLE41, the application of KIN caused inhibition of xylem differentiation, which was not observed in CLV3 treatment (Fig. 2c). In summary, KIN possesses both CLV3-and CLE41-like activities in all assays examined (Fig. 2d).
To address whether KIN exerts the CLE41-like activity through the interaction with TDR, the only known receptor for CLE41, we performed peptide treatment experiments using tdr-1 and cle41-1 mutants (Fig. 3a). Both of these mutants show reduction in stele width due to the loss of intrinsic CLE41-TDR signalling 8,19 . Application of CLE41 complemented the cle41-1 mutant phenotype. As the KIN application also rescued the cle41-1 mutant defect (Fig. 3a), KIN could function as CLE41. On the other hand, the receptor mutant tdr-1 was insensitive to ARTICLE exogenous KIN and to CLE41 as well (Fig. 3a), suggesting that KIN acts through TDR to promote stele growth.  Fig. 7), suggesting that the binding of KIN to each receptor is not likely to be the major determinant of the difference in effective concentration for CLV3-and CLE41-type bioactivities. The difference is likely to be caused by other factors such as locations of target tissues and downstream signal transduction pathways.
Role of specific residues for bioactivities of CLE peptides. To elucidate the structural basis of the dual activity exerted by the hybrid peptide KIN, we examined the function of specific residues of CLE peptides. The amino-terminal residue of CLE peptides, which is conserved as R in CLV3-type peptides or H in CLE41type peptides, has been recognized as an essential residue for their activities according to the previous Ala-scan assays 1,2,22 . Consistently, deletion of the N-terminal residue from KIN reduced its bioactivity at B100-fold (Supplementary Fig. 8a-d, peptide 18). However, KIN exerts both CLV3 and CLE41 activities even though its N terminus is R, raising a possibility that the N terminus may not be important for the specificity of CLE activities. Indeed, KIN-H 1st peptide also showed a dual activity similar to KIN, both in stele-thickening and root-shortening assays ( Supplementary Fig. 8a,b,e,f, peptide 19), indicating that the N-terminal residue is not responsible for the specificity. This was further supported by the fact that CLE41-R 1st retained the CLE41 activity with no CLV3 activity ( Supplementary Fig. 8a,b,e,f, peptide 20).
In addition to H 1st , CLE41 has the characteristic S 11th , which is conserved only among CLE41-type peptides in the CLE family 23 . We found that CLE41-H 11th exhibited a dual activity, whereas CLE41-H 12th showed only CLE41 activity ( Supplementary  Fig. 8a,b,e,f, peptides 21 and 22). CLE41-H 11th H 12th showed CLV3 activity but lost CLE41 activity ( Supplementary Fig. 8a,b,e,f, peptide 23). On the other hand, CLV3-S 11th exhibited neither CLV3 nor CLE41 activity ( Supplementary Fig. 8a,b,e,f, peptide 24). This H-to-S substitution also reduced the CLV3-type activity of KIN, although the effect was moderate ( Supplementary Fig. 8ad, peptide 25). Collectively, CLV3 requires H 11th for its activity and the S 11th of CLE41 prevents the peptide from displaying the CLV3 activity.
In the recently published crystal structures of the CLE41-TDR complex, the O g atom of S 11th forms a hydrogen bond with the e-amino group of K 397th of TDR 12,13 . We analysed the stability of the hydrogen bond at room temperature (300 K) in molecular dynamics (MD) simulations based on the atom coordinates of the CLE41-TDR complex 12 . We considered multiple alternative models for protonation states of titratable residues at 300 K, especially H 1st of CLE41 and D 303rd of TDR (Supplementary Information). The overall structure of CLE41 peptide was considerably more flexible at 300 K compared with the simulation at 77 K (mimicking the crystal), as shown by reduced fractions of native contacts (Supplementary Fig. 9a). The fraction of native contacts was lower in simulations with protonated D 303rd of TDR (D þ 303rd versus D 303rd ) at 300 K, while not influenced significantly by the protonation states of H 1st of CLE41 (H þ 1st versus H 1st , Supplementary Fig. 9a). The higher flexibility was observed especially around the carboxy terminus of CLE41 as shown in Supplementary Fig. 9b by the root mean squared fluctuation of each C a -atom. Consequently, the duration of the hydrogen-bond formation between S 11th and TDR was reduced at 300 K (45% and 16% of the entire simulation time with unprotonated and protonated D 303rd , respectively), compared with the stable hydrogen bond at 77 K (Supplementary Information). Collectively, it is likely that the interaction of S 11th with TDR is significantly reduced at room temperature compared with the X-ray structure, which may explain why the mutation on S 11th had little effect on the bioactivity of CLE41 in the previous report 1 . In contrast, N 12th , which is essential for the bioactivity 1 , interacted with TDR 495% of the time (Supplementary Information), in spite of the increased flexibility at 300 K, which is due to the formation of a flexible network of hydrogen bonds with several residues of TDR.
The MD simulations raised a possibility that the side chain of the 11th residue of CLE peptides might not contribute   significantly to their affinities with their intrinsic receptors. To unequivocally address this, we examined the interaction between mutated peptides and receptors by the competitive displacement assay with [ 125 I]ASA-KIN. As expected, CLV3-S 11th retained the interaction with CLV1 ( Supplementary Fig. 10, left), indicating that the H-to-S substitution does not compromise the binding of CLV3 with CLV1, even though it abolishes the bioactivity. Thus, the S 11th of CLE peptides hampers CLV1 misactivation at a level other than the direct ligand-receptor interaction. Conversely, CLE41-H 11th , which is a bifunctional CLE peptide (Supplementary Fig. 8), interacted with both TDR and CLV1, although the interaction with CLV1 was not strong (Supplementary Fig. 10), showing the contribution of H 11th to the interaction between CLE peptides and CLV1. The S 11th of CLE41 is highly conserved in a number of flowering plants 23 and even in gymnosperms and ferns 24 , although it is not required for CLE41 activity according to the previous Ala-scan assay 1 , implying that unwanted dual activity, which may be detrimental to organized growth, has been selectively avoided during the molecular evolution of CLE41 genes.

Discussion
Here we demonstrate that the bifunctional CLE peptides, which have not been identified in nature, can be artificially engineered by using genetic variation among natural CLE peptides. Plant peptide hormones are typically encoded in a gene family, which contains small variations in the mature ligand sequences and each variation can have a unique role in exerting specific bioactivities. Some variations, such as S 11th of CLE41, can be important to avoid unwanted cell signalling. In principle, sequence variations in natural peptide hormones are products under selection pressures in each evolutionary path. Importantly, peptides can take multiple mutational routes to reach or avoid specific bioactivities, as demonstrated in the engineering of bifunctional CLE peptides using different natural CLE peptides as starting materials. Thus, we propose that the hybrid synthesis of artificial peptides would provide a powerful methodology to use the natural genetic diversity as a source to mine cryptic bioactivities evolutionarily hidden in the genome and to engineer artificial cell signalling. For instance, given that genetic diversities in some peptide hormone families determine species-specific reproductive barriers 25,26 , our approach could be used as a means to overcome reproductive barriers for the production of new beneficial plant/ crop species.

Methods
Preparation of peptides. Peptides were synthesized by Fmoc chemistry with a peptide synthesizer (CS136XT, CSBio). Hydroxyprolines were not included in the peptides used in this study. [ 125 I]ASA-KIN was synthesized as described previously 10 . Fmoc-KIN (3.5 mg), 4-azidosalicylic acid succinimidyl ester (1.6 mg, Pearce) and NaHCO 3 (1.0 mg) were dissolved in 200 ml of 50% acetonitrile for 12 h in the dark with shaking at room temperature. Fmoc-ASA-KIN was purified by reverse-phase HPLC, lyophilized and deprotected in 25% piperidine in water for 1 h in the dark with gentle shaking at room temperature. The deprotected peptide was purified by reverse-phase HPLC to yield 1.8 mg of analytically pure ASA-KIN. ASA-KIN was further radioiodinated by the chloramine T method, as described previously 10 . The labelled peptide was purified by reverse-phase HPLC, to yield analytically pure [ 125 I]ASA-KIN with specific radioactivity of 93 Ci mmol À 1 .
Photo-affinity labelling. Aliquots (1,000 mg) of microsomal proteins for Halo-tagged receptors (CLV1-HT 10  with an ultraviolet lamp (model ENF-260C/J (365 nm), Spectronics Co. Ltd) at a distance of 1 cm. The cross-linked membrane proteins were solubilized, immunoprecipitated by using HaloTag antibody and separated by SDS-PAGE on a 7.5% acrylamide gel. The dried gels were exposed to the bio-imaging plate (MS 2,025, Fujifilm) for 2 days at room temperature and the plates were analysed using a bio-imaging analyser (Typhoon FLA 900, GE).
Bioassay. For root-length measurement, plants were germinated and grown vertically on half-strength Murashige and Skoog (MS) medium supplemented with 1% sucrose and peptide/control solution at 22°C under continuous light. To observe the RAM, 4-day-old roots were excised and mounted in clearing solution (chloral hydrate/glycerol/water ¼ 8:1:2) before imaging with light microscope (Axio Imager.A2, Zeiss).
To observe the SAM, plants were germinated and grown at 22°C under continuous light on half-strength MS medium supplemented with 1% sucrose and 10 mM peptide/control solution. To make sections, roots and leaves were cut off from 10-day-old seedlings, then fixed in FAA solution (50% ethanol:10% formalin:5% acetic acid in water) and embedded into Technovit 7,100 resin according to the manufacturer's instructions (Heraeus Kulzer). Four-micrometrethin sections were made using a microtome (RM2235, Leica), stained with 0.05% toluidine blue and mounted in Entellan New (Merck) before observation with a light microscope (Axio Imager.A2, Zeiss).
For the observation of stele and leaf vein, seeds were germinated and cultured with shaking at 110 r.p.m. at 22°C under continuous light in liquid half-strength MS medium supplemented with 1% sucrose and peptide/control solution. 10-day-old seedlings were fixed in a 1:3 mixture of acetic acid/ethanol, washed with water and mounted in clearing solution (chloral hydrate/glycerol/water ¼ 8:1:2) before imaging with a light microscope (Axio Imager.A2, Zeiss).
Statistical analysis. Statistical analysis was performed with Excel (Microsoft) or R (www.R-project.org). Two-sided Welch's t-test was performed with Excel. For the multiple comparison, analysis of variance and Tukey's honest significant difference test were performed with R-package 'agricolae'. The sample size was determined based on the previous studies 7, 13 . Exact sample size for each data is shown in Supplementary Data 1.
MD simulation. Detailed methods of structure preparation for simulation and calculation of pK a of titratable residues are provided in Supplementary Note 1.
Data availability. The authors declare that all data supporting the findings of this study are available within the manuscript and its Supplementary Information files or are available from the corresponding authors upon request.