Calcium-mediated rapid movements defend against herbivorous insects in Mimosa pudica

Animals possess specialized systems, e.g., neuromuscular systems, to sense the environment and then move their bodies quickly in response. Mimosa pudica, the sensitive plant, moves its leaves within seconds in response to external stimuli; e.g., touch or wounding. However, neither the plant-wide signaling network that triggers these rapid movements nor the physiological roles of the movements themselves have been determined. Here by simultaneous recording of cytosolic Ca2+ and electrical signals, we show that rapid changes in Ca2+ coupled with action and variation potentials trigger rapid movements in wounded M. pudica. Furthermore, pharmacological manipulation of cytosolic Ca2+ dynamics and CRISPR-Cas9 genome editing technology revealed that an immotile M. pudica is more vulnerable to attacks by herbivorous insects. Our findings provide evidence that rapid movements based on propagating Ca2+ and electrical signals protect this plant from insect attacks.

indicator GCaMP6f 16 and visualized the spatiotemporal dynamics of the cytosolic Ca 2+ concentration ([Ca 2+ ] cyt ) in real time. GCaMP6f showed that mechanical touch with forceps causes a rapid but spatially localized [Ca 2+ ] cyt increase in various organs, such as floral buds and roots ( Supplementary Fig. 1), without affecting the physiological response of leaf movement ( Supplementary Fig. 2). In contrast, touch and wounding a leaflet with scissors induced [Ca 2+ ] cyt increases at the base of leaflets (tertiary pulvinus) in parallel with leaflet movement, which was sequentially triggered along a rachilla (Fig. 1a, b; Supplementary Movies 1 and 2). High-speed Ca 2+ imaging revealed that wound-induced [Ca 2+ ] cyt increase preceded leaflet movement by up to 0.15 s (Fig. 1c, e and Supplementary Movie 3). Pretreating M. pudica leaves with a Ca 2+ channel blocker, La 3+ , prevented both the [Ca 2+ ] cyt increase and movement in response to wounding (Fig. 1d, f and Supplementary Movie 4). Therefore, a [Ca 2+ ] cyt increase in the pulvinus is correlated with rapid leaf movement.
We also analyzed the [Ca 2+ ] cyt propagation pathway from the wound site to pulvini on a rachilla by monitoring GCaMP6f fluorescence from the abaxial (lower) side of the leaves. Wounding by scissors immediately elicited a [Ca 2+ ] cyt increase at the wound site, which subsequently propagated in a leaflet vein at 1.31 ± 0.17 mm/s (n = 6;  2+ ] cyt signature on the rachilla was bimodal (Fig. 2d), with the first [Ca 2+ ] cyt peak propagating basipetally at 3.11 ± 0.41 and acropetally at 2.65 ± 0.38 mm/s (n = 10 each). Pretreatment of La 3+ and a Ca 2+ chelator, EGTA, retarded propagation of the [Ca 2+ ] cyt increase as well as the pulvinar [Ca 2+ ] cyt increase, whereas a slight [Ca 2+ ] cyt change was observed at the wound site ( Supplementary Fig. 3 and Supplementary Movies 6 and 7). These results suggest that [Ca 2+ ] cyt acts as a long-distance rapid signal triggering leaf movements in wounded M. pudica.

Ca 2+ and electrical signals are spatiotemporally coupled
Since non-wounding stimuli trigger only an AP 7 and mechanical wounding generates both an AP and VP propagating in a rachilla toward the pulvinus 5 , we investigated the spatiotemporal relationship between [Ca 2+ ] cyt transmission and the electrical signals. Touching a pinna tip evoked single-peak [Ca 2+ ] cyt signal and AP with leaflet movements in a rachilla ( Fig. 2e, g, i). Wounding a leaflet triggered propagation of bimodal [Ca 2+ ] cyt and electrical signals consisting of an AP and VP (the first and second peaks, respectively) in a rachilla (Fig. 2f Movies 11,12,15,16), suggesting that the longdistance transmission of Ca 2+ changes is spatiotemporally coupled with the AP and VP, triggering rapid leaf movements in touched or wounded M. pudica.

Rapid movements defend plants against insects
The physiological significance of the rapid movements of M. pudica is a long-standing mystery in plant science but have been speculated to relate to an antiherbivory defense response 2,[11][12][13] . To ask if these rapid movements do indeed serve as a defense response to wounding by insect herbivory, we used La 3+ -treated leaves that did not respond to wounding or touch (Fig. 3a, b; Supplementary Figs. 7 and 8a, b).
Grasshopper herbivores stayed and fed on the La 3+ -treated leaves more than on the control leaves (Fig. 3d). La 3+ -treated leaves lost 38.0% in weight after this feeding assay, which was~2-fold higher consumption than that of the control leaves [18.9% (n = 14 each); Fig. 3f]. Consistent with this result, the total residence time of grasshoppers was 75.3 ± 11.3 min on the control leaves and 161.2 ± 23.1 min on the La 3+treated leaves (n = 14 each; Fig. 3h). To exclude the possibility that La 3+ might change the palatability of leaves and influence the appetite of grasshoppers, we produced an immotile M. pudica in a different way, using CRISPR-Cas9 genome editing technology. ELONGATED PETIOLULE1 (ELP1), a putative transcription factor containing the ASYMMETRIC LEAVES2 (AS2)/LATERAL ORGAN BOUNDARIES (LOB) domain, is necessary for pulvinus development in Medicago truncatula 17 . Knocking out the homologous genes, ELP1B1/ELP1B2, rendered M. pudica immotile due to the lack of pulvini (Supplementary Figs. 9 and 10). Although elp1b1elp1b2 leaves propagated electrical signals upon wounding ( Supplementary Fig. 11), the leaves did not move ( Fig. 3c and Supplementary Figs. 7, 8c, d, 12a), indicating that the wound signaling pathway is unaffected by the loss of gene function and motility. As seen in the immobile La 3+ -treated leaves, the immotile elp1b1elp1b2 leaves lost 30.7% in weight by grasshopper attacks, which was approximately double that of wild-type (WT) leaves [15.3% (n = 12 each); Fig. 3e, g], and the total residence time of grasshoppers was 80.2 ± 12.4 min on the WT leaves and 142.7 ± 22.4 min on elp1b1elp1b2 leaves (n = 12 each; Fig. 3i). This effect was not limited to grasshoppers since we also used a generalist caterpillar and obtained similar results (Supplementary Fig. 12b-e). These observations indicate that rapid leaf movements defend against attack by a range of herbivorous insects reducing tissues lost and limiting the time the insects spend on the M. pudica leaves.

Feeding triggers Ca 2+ -mediated movement
To test if insect feeding induced Ca 2+ -based movement, we monitored [Ca 2+ ] cyt changes during grasshopper attack. A grasshopper chewing on leaflets caused a [Ca 2+ ] cyt increase in the tertiary pulvini and rapid leaflet movements ( Fig. 4 and Supplementary Movies 17 and 18). After the second feeding (Fig. 4, 60 s), the grasshopper's leg was pinched by folded leaflets, and the grasshopper repetitively shook its leg and moved away from this leaf (Supplementary Movie 18).

Discussion
Plants activate local and systemic defense responses within minutes to hours of insect contact, wounding, or herbivory, e.g., through production of the phytohormones ethylene and jasmonate, priming nondamaged regions to mount pre-emptive defenses [18][19][20][21] . Here, we demonstrated that M. pudica has a Ca 2+ /electrical signal-induced rapid defense response that actively repels predators. Compared to the hormone-based defenses, the motion-based defense is much faster, being activated and propagated throughout the plant body within seconds. This speed likely helps protect the plants from immediate insect attack. Touch triggered a single-peak [Ca 2+ ] cyt change and AP in the rachilla, sequentially inducing [Ca 2+ ] cyt increases at pulvini and subsequent leaflet movements (Figs. 1a, c, e, and 2e, i). Thus, M. pudica might sense herbivore contacts to activate motion-based defenses before leaves are damaged, but leaflet movements are restricted within the touched pinna because the AP cannot propagate over the secondary pulvinus toward distant pinnae 7 . Wounding elicits a VP that can pass through primary and secondary pulvini 5,7 , triggering defenses in both the local and systemic leaves. However, this model should be validated in the field.
Our model is based on pharmacological methods with possible side effects; e.g., La 3+ amplifies a secondary Ca 2+ increase induced by 0 mM K + medium in Arabidopsis thaliana roots 22 , causes plasma membrane depolarization that is reduced by raising the extracellular Ca 2+ concentration in Neurospora crassa 23 Fig. 13a-c). The dose-response curve indicated that the half maximal inhibitory concentration (IC 50 ) of La 3+ was 0.52 mM ( Supplementary Fig. 13d). These data suggest that the inhibitory effects on [Ca 2+ ] cyt dynamics and leaflet movements were due to La 3+ but not to the K + deficiency and membrane depolarization. To exclude the possibility that La 3+ -induced phosphate deficiency hindered [Ca 2+ ] cyt and electrical signals, we used verapamil, another Ca 2+ channel antagonist, which retarded Ca 2+ /electrical signal propagation in the rachilla ( Supplementary Fig. 14). We also reduced the period of EGTA treatment and mitigated the reduction in the Ca 2+ buffering capacity of EGTA with the pH buffer HEPES 26 . The [Ca 2+ ] cyt increase was limited within the wounded leaflets ( Supplementary  Fig. 15a, b), and the long-distance Ca 2+ /electrical signal was inhibited in the rachillae of leaves treated with EGTA and HEPES for a shorter period (Supplementary Fig. 15c-f). Furthermore, we treated membrane-permeable cytosolic Ca 2+ chelator (BAPTA tetraacetoxymethyl ester) together with EGTA to the rachillae. Despite the shortening of the incubation period to 0.5 h, wounding did not induce the long-distance propagation of the Ca 2+ and electrical signals in the rachillae ( Supplementary Fig. 16). Therefore, the Ca 2+ /electrical signal and leaflet movements are unlikely to be inhibited by the adverse effects of La 3+ and EGTA, supporting our model that the [Ca 2+ ] cyt changes coupled with the electrical signals act as the long-distance signal triggering leaf movements. Although now strongly linked to the leaflet movements, the molecular mechanisms driving the changes in [Ca 2+ ] cyt and their associated electrical signals remain elusive. In plants, anions are thought to play a role in AP 27 . Indeed, the amplitude of an AP is known to be dependent on extracellular Cl −28 and Ca 2+ (Supplementary Figs. 5 and 6) 15 in M. pudica. We speculate that the [Ca 2+ ] cyt increases might activate Ca 2+ -dependent Clchannels to transmit APs, as reported in Characeae algae 29 . Since VPs are related to the chemicals released upon wounding and transported through the xylem 1,30 , ligand-gated ion channels, e.g., GLUTAMATE RECEPTOR-LIKE family, might be involved in M. pudica's VP 31 . Leaf opening was not influenced by La 3+ treatment ( Supplementary Fig. 17), suggesting that the recovery process after rapid movement is regulated by signals other than [Ca 2+ ] cyt and AP/VP.

Plant material and growth conditions
Surface-sterilized seeds of M. pudica (an early-flowering WASE accession, Sakata Seed) were sown on germination medium [1/2× basal MS salts (Wako), 0.2% (w/v) gellan gum (Wako) and 0.05% (w/v) 2-morpholinoethanesulfonic acid (MES, Dojindo Laboratories); pH adjusted to 5.8 with KOH] in a square plate or in a mixture of culture soil (Nippi Engei Baido 1, Nihon Hiryo) and vermiculite (equal volumes) in a pot. The plates and pots of M. pudica were cultivated in a growth chamber (GB48, Conviron) under 12-h/12-h light/dark and 27°C/20°C cycles. One-week-old seedlings in the plates were used for Ca 2+ imaging of the roots and cotyledons ( Supplementary Fig. 1d, e), whereas plants more than 2 months old in the pots were used for the following experiments.
Isolation of the coding sequence of M. pudica ELP1B genes Taking advantage of the intronless structure of ELP1 coding sequences, we first isolated their partial sequences from M. pudica genomic DNA. The genomic DNA was obtained from M. pudica immature leaves using the CTAB extraction method 32 . Using KOD-Plus-Neo DNA polymerase (TOYOBO) and a pair of degenerate primers (5ʹ-ATGGCATCATCAA GCTCHTACAATTCNCCNTGYGC-3ʹ and 5ʹ-AGAAGATCAGCRTTAGC VGAATCRAGYTCYTTYTG-3ʹ), which were designed against wellconserved N-terminal and leucine zipper-like regions of leguminous ELP1 genes, we obtained PCR products of~330 bp in size. These fragments were cloned into a pCR-Blunt II-TOPO vector (Thermo Fisher Scientific) and sequenced using the ABI 3130xl sequencer (Applied Biosystems) and a BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific). Sequencing analysis revealed that M. pudica has at least 4 different sequences of ELP1 genes, named ELP1A1, ELP1A2, ELP1B1, and ELP1B2. The doubled chromosome number found in M. pudica 33 indicated a recent whole-genome duplication event in this species, and we consistently observed two gene pairs with highly similar sequences. Only 8 nucleotides differed in the isolated 262nucleotide region in ELP1A1 and ELP1A2, and 4 of 262 nucleotides differed in ELP1B1 and ELP1B2, indicating that these gene pairs originated from ancestral ELP1A and ELP1B by whole-genome duplication. Then, we isolated the full-length cDNA sequences of ELP1B1 and ELP1B2 genes by the rolling circle amplification-RACE method 34 . Briefly, total RNAs were extracted from 3-day-old seedlings or a shoot apex containing immature leaf primordia (~5 mm in length) using PureLink Plant  RNA Reagent (Thermo Fisher Scientific). The extracted RNAs were treated with DNase I and further purified using a silica membrane column (RNeasy Plant Mini Kit, Qiagen). Then, first-strand cDNA was synthesized with SuperScript III reverse transcriptase (Thermo Fisher Scientific) and phosphorylated oligo (dT) [12][13][14][15][16][17][18] primers (Thermo Fisher Scientific). After RNase H treatment and purification using a QIAquick PCR Purification Kit (Qiagen), the single-stranded cDNAs were circularized with CircLigase II ssDNA Ligase (Epicenter). The circularized cDNAs were purified with a QIAquick PCR Purification Kit and then subjected to rolling circle amplification with ϕ29 DNA polymerase (New England Biolabs) and phosphorothioate-modified random hexamers at 30°C for 24 h. Using these amplified cDNAs as the template, PCR amplification was performed with the primers 5ʹ-TATGGCTGTGTAGGAGCCATCTCC-3ʹ and 5ʹ-GAGTTGACGGCGTCTTC TCTCTGG-3ʹ. The PCR products were electrophoresed in 2% agarose gel, and several conspicuous DNA bands were excised from the gel. After purification using a FavorPrep GEL/PCR Purification Kit (FAVORGEN), the PCR fragments were phosphorylated with T4 polynucleotide kinase (New England Biolabs), purified again with a GEL/PCR Purification Kit, and then cloned into the SmaI site of the pGEM-3Zf (+) vector (Promega). We obtained the 5ʹ and 3ʹ sequences of ELP1B1 and ELP1B2 genes by Sanger sequencing, and the sequence information was used to design the genotyping primers described below.

Vector construction
For the LjUBQ1p (Lotus japonicas polyubiquitin promotor 35 )::GCaMP6f Ca 2+ biosensor construct, we first constructed a pSB11U2 vector containing LjUBQ1p, a multicloning site, and 35S terminator sequences in this order, together with a hygromycin resistance gene cassette. The SacI-StuI fragment of pIG121-Hm, which contained a NOS terminator and hygromycin resistance cassette, was inserted into the SacI-blunted EcoRI site of pSB11 (the resultant intermediate construct was named pSB11-NT-Hm, and we similarly indicate construct names hereafter). A 35S terminator sequence was PCR-amplified with primers (5ʹ-CCGA GCTCGGCCATGCTAGAGTCCGCAAAAATC-3ʹ and 5ʹ-AGGACGCGTAGG TCACTGGATTTTGGTTTTAGG-3ʹ) using pK7WGF2::hCas9 as the template, digested with SacI and MluI, and then inserted into the equivalent site of pSB11-NT-Hm to replace the NOS terminator with a 35S terminator (pSB11-ST-Hm). The LjUBQ1p fragment was PCR-amplified with primers (5ʹ-GCCTCGAGGAGAGAGGATTTTGAGGAAATAATTAAT TG-3ʹ and 5ʹ-TCTAGACTGTAATCACATCAACAACAGATAAAT-3ʹ) using pUB-Hyg as the template, digested with XhoI and XbaI, and then inserted into the equivalent site of pSB11-ST-Hm, generating the expression vector pSB11U2. The GCaMP6f fragment, gifted by Douglas Kim and the GENIE Project (Addgene plasmid #40755), was PCRamplified between the XbaI and BamHI sites and inserted into the equivalent sites of the pAN19 vector. The GCaMP6f fragment was isolated by XbaI and XmaI digestion and ligated into the equivalent sites of the pSB11U2 vector.
For CRISPR/Cas9 genome editing, we constructed a pSB11CAS11 vector, which possessed expression cassettes for hCas9, guide RNA, sGFP, and hygromycin resistance genes. The entire construction procedure of this vector reflects a series of modifications for functional improvement. A 35 S promoter fragment was excised from pIG121-Hm via digestion with HindIII and XbaI and then inserted into the equivalent site of pSB11-NT-Hm (pSB11S). A DNA fragment containing EGFP-hCas9 and the 35S terminator was PCR-amplified with primers (5ʹ-AC TCTAGACACCATGGTGAGCAAGGGCGAGGAG-3ʹ and 5ʹ-AGGACGCG TCAGGTACCAAGGTCACTGGATTTTGGTTTTAGG-3ʹ) using pK7WGF2:: hCas9 as the template, digested with XbaI and MluI, and then inserted into the equivalent site of pSB11S (pSB11CAS). An Arabidopsis U6 promoter fragment was PCR-amplified with primers (5ʹ-CAAGGTCTC GGTACCGGAGTGATCAAAAGTCCCAC-3ʹ and 5ʹ-AATGGTCTCCAATCG CTATGTCGACTCTATC-3ʹ) using pICH86966::AtU6p::sgRNA PDS as the template and then digested with BsaI. Similarly, a DNA fragment containing the guide RNA backbone and poly T terminator was PCRamplified with primers (5ʹ-TGAGGTCTCAGTTTTAGAGCTAGAAATAGC AAG-3ʹ and 5ʹ-GAAGGTCTCACGCGTAAAAAAAGCACCGACTCGGTG-3ʹ) using pICH86966::AtU6p::sgRNA PDS as the template and then digested with BsaI. Double-stranded DNA oligo containing a target sequence for M. pudica ELP1A genes was prepared by annealing two synthesized oligonucleotides (5ʹ-GATTGCAAACGTCCATAAAATATT-3ʹ and 5ʹ-AAACAATATTTTATGGACGTTTGC-3ʹ). These three fragments were introduced into the Acc65I-MluI site of pSB11CAS (pSB11CAS-ELP1A) through 4-fragment ligation to reconstruct a functional guide RNA expression cassette. This initial CRISPR/Cas9 construct was further subjected to two serial modifications. The first modification was intended to replace the 35 S promoter upstream of EGFP-hCas9 with LjUBQ1p and also to rearrange the position and orientation of the expression cassettes. To this end, a LjUBQ1p fragment was excised from pSB11U2 via digestion with XhoI and XbaI. A DNA fragment containing EGFP-hCas9 and the 35 S terminator was excised from pSB11CAS-ELP1A via digestion with XbaI and MluI. A DNA fragment containing the guide RNA cassette was PCR-amplified with primers (5ʹ-TCCTCGAGTTGCTAGCGGAGTGATCAAAAGTCCCAC-3ʹ and 5ʹ-GCAA GCTTGGACCGGTAAAAAAAGCACCGACTCGGTG-3ʹ) using pSB11CAS-ELP1A as the template and then digested with XhoI and HindIII. These three fragments were introduced into the HindIII-MluI site of pSB11-ST-Hm via 4-fragment ligation (pSB11CAS10-ELP1A). As the fluorescence intensity of EGFP-hCas9 fusion proteins was insufficient for use as a visual selection marker, we further intended to assign hCas9 and GFP to two separate expression cassettes. To this end, a DNA fragment containing a sGFP expression cassette was PCR-amplified with primers (5ʹ-TTACCGGTGCATGCCTGCAGGTCCCCAG-3ʹ and 5ʹ-AATAAGCTT CCCGATCTAGTAACATAGATGACACC-3ʹ) using pIF121-Hm 32 as the template and then digested with AgeI and HindIII. A DNA fragment containing the guide RNA cassette and LjUBQ1p was excised from pSB11CAS10-ELP1A via digestion with AgeI and NcoI. Similarly, a DNA fragment containing hCas9 and the 35 S terminator was excised from pSB11CAS10-ELP1A via digestion with NcoI and MluI. These three fragments were introduced into the HindIII-MluI site of pSB11-ST-Hm via 4-fragment ligation (pSB11CAS11-ELP1A). Finally, we replaced the pre-existing target sequence with that of ELP1B genes. To this end, double-stranded DNA oligo containing the ELP1B target sequence was prepared with oligonucleotides (5ʹ-GATTGGTGAGGAAGGAGCT CATTG-3ʹ and 5ʹ-AAACCAATGAGCTCCTTCCTCACC-3ʹ). This target sequence and adjacent PAM sequence were completely conserved in ELP1B1 and ELP1B2 genes, and thus, the single target sequence could be expected to recognize both genes simultaneously. A U6 promoter fragment was PCR-amplified with primers (5ʹ-CAAGGTCTCGCTAGCG GAGTGATCAAAAGTCCCAC-3ʹ and 5ʹ-AATGGTCTCCAATCGCTATGTC GACTCTATC-3ʹ) using pSB11CAS11-ELP1A as the template and then digested with BsaI. Similarly, a DNA fragment containing the guide RNA backbone and poly T terminator was PCR-amplified with primers (5ʹ-TGAGGTCTCAGTTTTAGAGCTAGAAATAGCAAG-3ʹ and 5ʹ-GAAGGT CTCACCGGTAAAAAAAGCACCGACTCGGTG-3ʹ) using pSB11CAS11-ELP1A as the template, and then digested with BsaI. These three fragments were incorporated into the AgeI-NheI site of pSB11CAS11-ELP1A to replace the guide RNA cassette. The resultant vector pSB11CAS11-ELP1B was used for the subsequent knockout experiment.

Mutation analysis of CRISPR/Cas9-mediated mutants
Eight independent transgenic T 0 plants were established for pSB111CAS11-ELP1B. After transfer into soil and subsequent plant growth, an immature leaf was excised and used for genomic DNA extraction using a DNeasy Plant Mini Kit (Qiagen). Genomic DNA fragments of ELP1B1 were PCR-amplified with a common ELP1B forward primer (5ʹ-TGCAAATTCCTGAGGAGGAAGTGCA-3ʹ) and a specific ELP1B1 reverse primer (5ʹ-TAGTCCCTGATCTTGCCCTCTG*A*T-3ʹ; asterisks indicate the two phosphorothioate modifications to prevent unwanted degradation of the primer by the proofreading activity of KOD-Plus-Neo DNA polymerase). Similarly, genomic DNA fragments of ELP1B2 were PCR-amplified with the common forward primer and a specific ELP1B2 reverse primer (5ʹ-TAGTCCCTGATCTTGCCCTCTG* A*G-3ʹ). After purification with a FavorPrep GEL/PCR Purification Kit, the DNA fragments were directly sequenced in both directions using an ABI 3130xl sequencer, a BigDye Terminator v3.1 Cycle Sequencing Kit, and the forward and reverse PCR primers. The electropherograms of the DNA sequencing, in which two different peaks were simultaneously detected because of the different sizes of indels in the two mutated alleles, were optically decomposed to each single sequence. Sequence analysis revealed various patterns of nucleotide insertions and deletions into the target genes between elp1b1elp1b2 line 1 and 8 T 0 plants ( Supplementary Figs. 9 and 10a).

High-speed and wide-field Ca 2+ imaging
For leaf imaging, the leaves were severed at the petiole bases, and the leaf petioles were inserted into water-filled 15-ml tubes through pores on the caps. The leaves were fixed with Parafilm and recovered at around 25°C under white LED light. When the leaflets were opened, the tube with the leaf specimen was carefully transferred onto the stage of a fluorescent microscope. For Ca 2+ imaging using a grasshopper, the leaf was floated onto the stage.
Upon fluorescence imaging, M. pudica leaves were stimulated with a touch by tweezers (or a micropipette tip), wounding by dissecting scissors, or grasshopper (Acrida cinerea) feeding. Several ROIs were analyzed over time using NIS-Elements and ZEN pro imaging software (for ROI positions, see Fig. 2b or Supplementary  Fig. 3c). Background noise was subtracted from the GCaMP6f signals on the leaves over time. The fractional fluorescence changes were calculated according to the equation ΔF/F 0 = (F − F 0 )/F 0 , where F denotes GCaMP6f fluorescence at a certain time and F 0 denotes the averaged baseline fluorescence defined as the average of F over~1 s without cytosolic Ca 2+ ([Ca 2+ ] cyt ) changes.

Simultaneous measurement of Ca 2+ and electrical signals
KCl, CaCl 2 (both from Wako), and MES were dissolved in water to prepare stock solutions (100× for MES, 1000× for KCl and CaCl 2 ). Extracellular fluid [ECF; 0.1 mM KCl, 1 mM CaCl 2 , and 1 mM MES; pH adjusted to 5.8 with Tris(hydroxymethyl)aminomethane (Rikaken)] was made by diluting the stock solutions. The leaf specimens were treated as described in the Ca 2+ imaging section using ECF instead of water.
Recording electrodes were prepared according to Sibaoka 37 with the following modifications: Ag/AgCl wires of 0.2 mm in diameter were inserted into truncated micropipette tips, and the tips were fixed to electrode holders and filled with 10 mM (or in some cases, 1 mM) KCl solution. A handmade electrode or, in some cases, an Ag/AgCl-type pellet electrode (1-HLA-003, Inter Medical), was used as a reference electrode. The handmade reference electrode was prepared by inserting an Ag/AgCl wire of 0.5 mm in diameter into a truncated micropipette tip that was filled with 10 mM KCl solidified by 0.5% (w/v) agarose (Rikaken). Two operational amplifiers (Axopatch 200 A, Axon Instruments), two headstage amplifiers (CV-201A headstage, Axon Instruments), a digitizer (Digidata 1322 A, Axon Instruments), and electrophysiology data acquisition software (Clampex 9.2, Axon Instruments) were used to detect surface potential changes. Simultaneous measurements were performed under the SMZ25 microscope in a Faraday cage. The leaf sample was placed upside down on the stage and exposed to excitation light. The reference electrode was inserted into the bathing fluid, and the two recording electrodes were placed on a rachilla (for recording electrode placements, see Fig. 2g, h and Supplementary Fig. 4a). The GCaMP6f fluorescence was acquired every 200 ms, and the surface potential changes were sampled at 5 kHz and low-pass filtered at 1 kHz. The fluorescence signals were analyzed as described previously (for ROI positions, see Fig. 2g, h and Supplementary Fig. 4a). The surface potential changes (ΔV) were calculated using the equation ΔV = V -V 0 , where V denotes the potential difference (PD) between the recording and the reference electrodes at a certain time and V 0 denotes the averaged baseline PD over 1 s with no electrical changes.  Fig. 15) to make a 50-mM solution, and the pH was adjusted to 7.0 with KOH and HCl. Verapamil hydrochloride (Sigma-Aldrich) was dissolved in ECF at a final concentration of 2 mM. The petioles of the leaf specimens were dipped into and treated with these solutions in 15-ml tubes for 2-3 h for La 3+ treatment, 3-5 h for EGTA treatment, or 5 h for verapamil treatment under white LED light. The leaves were then dipped in a control solution or 1 mM KCl for EGTA-treated leaves ( Supplementary Fig. 6) to stop the pharmacological treatments and were allowed to recover for at least 1 h under white LED light.

Pharmacological treatments
BAPTA tetraacetoxymethyl ester (BAPTA-AM, Tokyo Chemical Industry) was dissolved in DMSO to make a 100-mM stock solution. The stock solution was diluted 100 times using 10 mM HEPES solution with 50 mM EGTA, and the pH was adjusted to 7.0 using KOH and HCl. Pinnae were isolated from the leaves, and basal 7 leaflet pairs were removed from the base of the isolated pinnae to insert the pinnae into the solution. The isolated pinnae were inserted into 1.5-ml tubes filled with water through holes on the lids, fixed with Parafilm to the tubes, and placed in an incubator (LU-113, ESPEC) at 25°C overnight. The pinnae were transferred into PCR tubes with 0.3 ml of EGTA and BAPTA-AM, fixed to the tubes with Parafilm, and treated with the chelators under white LED light in the incubator at 25°C for 0.5 h. The treated pinnae were then put via holes in the lids into 1.5-ml tubes filled with 10 mM HEPES (pH was adjusted to 7.0 with KOH and HCl), fixed to the tubes with Parafilm, and maintained at~25°C for more than 1 h under white LED illumination. Subsequently, the Ca 2+ and electrical signals were measured (for positions of ROIs and recording electrodes, see Supplementary Fig. 16a).

Videography and photography
Movies of leaflet movements were acquired with a video camera (HERO5 BLACK, GoPro) or the aforementioned imaging systems, and the leaflet angle changes were estimated using NIS-Elements imaging software. Photographs of various tissues were collected by a stereomicroscope (Leica S9D, Leica Microsystems) equipped with a camera (Leica MC190 HD, Leica Microsystems) and processed using microscope software (Leica Application Suite, Leica Microsystems). The images acquired using a Leica S9D microscope (Supplementary Fig. 10d-f) were corrected by non-linear adjustment (gamma value = 0.60).

Feeding assay
WT and elp1b1elp1b2 leaves with 4 pinnae were cut at the petiole bases. A pair of control and 50 mM La 3+ -treated leaves or WT and elp1b1elp1b2 leaves was placed into a 50-ml glass bin filled with water. The leaves were fixed to the bins with a piece of Parafilm. Several bins containing leaves were placed in a plastic container (53 × 39 × 32 cm 3 ), and then the leaves were recovered at room temperature under white LED light. When the WT (control) leaves were expanded, 3-7 adult grasshoppers (Atractomorpha lata,~4 cm in length) or 2 caterpillars (Helicoverpa armigera,~2.5 cm in length) per bin, which had been fasted for~1 day, were put in the container. The assays were monitored with a HERO5 BLACK video camera for 3 h for the grasshoppers and 2 h for the caterpillars. The assays were repeated with off-food intervals (approximately 1 day). In every assay, the fresh leaves were weighed on an electronic balance (GR-202, A&D Company) before and after the experiments, and the ratio of these values was calculated. Thereafter, the pinnae were detached from the leaves using dissecting scissors and scanned by a photocopier (EP-M570T, Seiko Epson). The total residence time of the grasshoppers on a leaf was determined from the movie data. The transgenic leaves expressing LjUBQ1p::GCaMP6f were used as the WT leaves in these experiments.

Statistical analysis
The time point of the [Ca 2+ ] cyt increase was analyzed by t-test using the criterion of a rise to 3× SD above F 0 as an indicator of a detectable increase in the GCaMP6f signal, and the speed of the [Ca 2+ ] cyt signal was then calculated by dividing the distance between two ROIs by the interval between the calculated time points at the two ROIs using Excel (Microsoft) 19 . The timing of the detectable leaflet angle change was similarly determined. The inter-electrode distances were obtained from the imaging data, and the peak times of the surface potential changes were determined using electrophysiology analysis software (Clampfit, Axon Instruments). The action potential velocities were calculated using Excel by dividing the inter-electrode distances by the time intervals between the peak-to-peak time points.
We used GraphPad Prism (GraphPad Software) for the following statistical analyses. The velocities of the [Ca 2+ ] cyt increases and surface potential changes, the wet weight ratio of the leaves before and after the feeding assays, and the total residence time of insects on the leaves were analyzed by a two-tailed Wilcoxon matched-pairs signed rank test. IC 50 for the inhibitory effects of La 3+ was estimated by fitting a sigmoidal dose-response curve. Mean ± SEM values were represented in the main text and Supplementary Table 1 for the velocities of signal propagations and the total residence time of insects on the leaves.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request. Plasmid sequence data can be found in GenBank with the accession numbers pSB1 (AB027255) and pSB11 (AB027256). The GCaMP6f sequence is available at Addgene (#40755). ELP1B1 and ELP1B2 sequences are deposited in DNA Data Bank of Japan with the accession numbers ELP1B1 (LC731322) and ELP1B2 (LC731323). Source data are provided with this paper.