Main

As a ubiquitous second messenger, Ca2+ regulates many aspects of physiology and development in both animals and plants9,10,11, including reproduction. In animals, Ca2+ signals drive the motility of sperm12 and forecast successful fertilization13. In flowering plants, sperm are immobile and require a special delivery structure called the pollen tube, which navigates the female tissue and finds the ovule before releasing sperm1. From pollen germination to pollen tube guidance and pollen tube reception, each step requires intricate Ca2+ signalling14. However, the molecular mechanism underlying Ca2+ signalling in plant reproduction remains largely unknown. During pollen tube reception, interactions between the pollen tube and synergids in the ovule activates Ca2+ oscillations in both partners, which leads to the rupture of the pollen tube and synergid cell death and the initiation of fertilization2,3,15,16. On the female side, FER, LRE and NTA are three components from the same pathway required for synergid Ca2+ spiking in response to pollen tube arrival, but little is known regarding how they work together to mediate Ca2+ entry. We show here that pollen tube RALFs bind to FER–LRE co-receptors, which recruit NTA, a calcium channel, to form a receptor–channel assembly. This tri-molecular complex is regulated by Ca2+/calmodulin (CaM)-dependent feedback inhibition to drive Ca2+ oscillations in the synergid.

RALFs trigger synergid Ca2+ oscillation

FER and LRE are both required for pollen tube reception and may function as a co-receptor for unknown signals derived from pollen tubes4,5,17,18,19,20. We hypothesized that such signals may be mediated by RALF family peptides because some RALFs bind to either FER alone21,22 or to both FER and LRE-like proteins (LLGs) in other processes23,24,25. Among the 37 Arabidopsis RALFs, at least 8 (RALF4, RALF8, RALF9, RALF15, RALF19, RALF25, RALF26 and RALF30) are expressed in pollen tubes7. We expressed and purified the eight pollen-tube-derived RALF peptides and examined their interaction with FER23. We also included an ovule-derived RALF peptide (RALF34) that is closely related to pollen tube RALFs such as RALF4 and RALF19. The extracellular domain of FER (FERex) pulled down RALF4, RALF19 and RALF34 (Fig. 1a and Supplementary Fig. 1a). Consistent with this result, RALF4 and RALF19 acted antagonistically with RALF34 for pollen tube integrity through interactions with the same receptors of the FER family7.

Fig. 1: Pollen-derived RALFs bind to FER–LRE and trigger synergid [Ca2+]cyt changes in a FER–LRE–NTA-dependent manner.
figure 1

a, Pull-down assays showing the interaction of GST-tagged RALFs and MBP-tagged ectodomains of FER (FERex). Amylose resin was used to pull down MBP, followed by western blotting with antibodies against GST and MBP. n = 3 independent repeats. b, Interaction of GST-tagged LRE and MBP-tagged ectodomains of FER with or without RALFs (RALFx; each 100 nM) as indicated. Amylose resin pull-down and western blotting was performed as in a. n = 3 independent repeats. c, Co-IP of Myc-tagged LRE and Flag-tagged FERK565R expressed in N. benthamiana leaves with or without the addition of RALFs (each 5 μM) as indicated. Anti-Flag M2 affinity beads were used in co-IP, and western blots were probed with antibodies against Myc and Flag. n = 3 independent repeats. d–g, Representative Ca2+ spiking patterns in synergids in response to pollen tube (PT) arrival or to 0.5 µM or 2 µM RALF4 (R4) for WT (d), fer-4 (e), lre-5 (f) and nta-3 (g). W, water. h, Ca2+ oscillation periodicity of WT synergids in response to PT arrival or 0.5 µM RALF4 and RALF19 (R19). n = 8 ovules. i, The peak values of Ca2+ spiking as in d–g. n values are shown in Extended Data Fig. 3. Ovules were isolated from Col-0, fer-4, lre-5 and nta-3 flowers harbouring the synergid-specific GCaMP6, and fluorescence was recorded using an inverted microscope. Red triangles indicate time points at which the PT arrived or RALF4 was applied. Error bars depict the mean ± s.e.m. All P values were determined by two-tailed Student’s t-test. NS, not significant.

Source data

Full size image

As ligands for FER–LLG co-receptors, RALFs enhance interactions between FER and LLGs23. Consistent with this model, RALF4, RALF19 and RALF34 enhanced the interaction between GST-tagged LRE and MBP-tagged FERex, whereas several other RALFs did not (Fig. 1b, bottom, and Supplementary Fig. 1b). This observation was confirmed by co-immunoprecipitation (co-IP) assays with total protein samples from Nicotiana benthamiana leaf tissue expressing LRE–Myc and FERK565R–Flag (Fig. 1c and Supplementary Fig. 1c). RALF4 is secreted into the apoplast of the pollen tube26, and our promoter–β-glucuronidase (GUS) analysis confirmed the expression of RALF4 and RALF19 during pollen tube reception (Extended Data Fig. 1a). This result suggests that RALF4 and RALF19 interact with FER and LRE in the synergid. In summary, FER–LRE may function as co-receptors for pollen tube RALFs, including RALF4 and RALF19.

In response to pollen tube arrival, the synergids produce specific Ca2+ fluctuations required for pollen tube reception2,3. If RALFs signal the arrival of the pollen tube, they should produce a similar Ca2+ entry pattern in the synergids when applied to isolated ovules in vitro. We generated transgenic plants that express the Ca2+ indicator GCaMP6s27 driven by a synergid-specific promoter (pMYB98)28 and examined changes in cytosolic calcium concentration ([Ca2+]cyt) in the synergids in response to RALFs. In the wild-type (WT) synergids, RALF4 and RALF19, but not RALF34, induced increases in [Ca2+]cyt (Fig. 1d,i and Supplementary Videos 1, 2 and 3), which is consistent with the idea that RALF34 binds to the same receptors but functions differently7. Several other pollen tube RALFs (RALF8, RALF9, RALF15, RALF25, RALF26 and RALF30) that did not bind FER also failed to induce synergid [Ca2+]cyt changes (Extended Data Fig. 2). In the pollen tube reception assay, the receptive synergid and nonreceptive synergid in one ovule showed distinct Ca2+ spikes2. However, exogenous RALF4 and RALF19 induced similar Ca2+ transients in both synergid cells in one ovule, which suggests that RALFs in the solution may have diffused evenly towards the two synergids. By contrast, the pollen tube positions itself closer to one of the two synergids, which leads to asymmetrical signalling.

We then compared synergid Ca2+ changes triggered by the pollen tube2,3 with those induced by RALFs. As reported earlier3, synergid Ca2+ dynamics proceeded in three phases as the pollen tube progressed to reception: (1) [Ca2+]cyt oscillated at a regular pace when the pollen tube enters the micropyle and approaches the synergid; (2) [Ca2+]cyt was sustained at a higher level after the pollen tube penetrates the synergid; (3) [Ca2+]cyt was reduced when the synergid collapses. The amplitude and periodicity of Ca2+ oscillations triggered by RALF4 and RALF19 were similar to phase I of Ca2+ spiking induced by the pollen tube (Fig. 1d–i, Extended Data Fig. 3 and Supplementary Video 4). This result suggests that pollen-tube-derived RALF4 and RALF19 mimic the early phase of pollen tube arrival before mechanical penetration.

We then tested whether RALF4 and RALF19 induced such Ca2+ spiking in a FER–LRE-dependent manner. Synergids from fer-4 and lre-5 mutants failed to respond to RALF4, RALF19 or pollen tube arrival3 (Fig. 1e,f,i, Extended Data Fig. 3 and Supplementary Videos 5 and 6). In addition to the fer and lre mutants, a mutant lacking NTA was non-responsive to 0.5 µM RALF4/19, although a portion of the nta-3 synergid showed weaker responses to an increased level of RALF4/19 (2 µM) in Ca2+ imaging assays (Fig. 1g,i, Extended Data Fig. 3 and Supplementary Video 7). Compared to the fer and lre mutants, the weaker defect in nta-3 suggests that there may be partial functional redundancy with another NTA-like component in this system.

MLO proteins are Ca2+ channels

The NTA protein is a member of the MILDEW RESISTANCE LOCUS O (MLO) protein family6. Originally discovered as a genetic determinant for resistance against powdery mildew in barley29, the MLO proteins feature multi-transmembrane domains and a CaM-binding domain (CaMBD)30,31. The Arabidopsis genome encodes 15 MLO proteins (AtMLO1–AtMLO15), some of which are functionally linked to root thigmomorphogenesis32, powdery mildew susceptibility33 and pollen tube growth34. NTA (AtMLO7) is specifically expressed in synergids and appears to function downstream of the FER–LRE module in pollen tube reception6. The biochemical function of MLO proteins remains unknown, which represents a crucial gap in knowledge with respect to the signalling pathways in which they participate31. Genetic analyses of NTA indicates that it works together with FER–LRE co-receptors in the same pathway to induce Ca2+ influx. Because NTA and other MLO proteins are multi-transmembrane proteins, we hypothesized that NTA is one of the missing Ca2+-transporting proteins responsible for synergid Ca2+ entry.

To test whether MLO proteins transport Ca2+, we performed Ca2+ transport assays with all 15 MLO members from Arabidopsis (AtMLO1–AtMLO15), the barley MLO (HvMLO) and 2 MLO members from Physcomitrella patens (PpMLO2 and PpMLO3), which represent dicot, monocot and basal land plant MLO proteins, respectively. In single-cell Ca2+ imaging assays35, AtMLO2, AtMLO3, AtMLO4, AtMLO10, AtMLO12, HvMLO, PpMLO2 and PpMLO3 mediated Ca2+ entry when expressed in COS7 cells (Fig. 2a and Extended Data Fig. 4). To confirm these Ca2+ imaging results, we used patch-clamping to directly measure the transport activity of AtMLO2 expressed in HEK292T cells. We recorded large inward currents mediated by AtMLO2 that depended on external Ca2+ concentrations (Extended Data Fig. 5a,b). Moreover, AtMLO2 waspermeable to Ba2+ and Mg2+, but not to K+ or Na+ (Extended Data Fig. 5c–f,i–l). Furthermore, two typical Ca2+ channel blockers, lanthanum (La3+) and gadolinium (Gd3+), inhibited the AtMLO2-mediated inward currents (Extended Data Fig. 5g,h). Similar to AtMLO2, HvMLO also mediated Ca2+ influx (Extended Data Fig. 6). These results indicate that MLO proteins function as Ca2+-permeable channels.

Fig. 2: MLO family proteins, including NTA, are Ca2+-permeable channels.
figure 2

a, [Ca2+]cyt increases measured by single-cell fluorescence imaging in COS7 cells expressing various MLO proteins 1–15 (AtMLO1–AtMLO15). Hv, HvMLO; Pp2, PpMLO2; Pp3, PpMLO3. b, FER and LRE facilitated the PM localization of NTA–GFP. The white rectangle indicates the area magnified in the bottom panels. n = 3 independent repeats. Scale bars, 5 μm (bottom row) or 10 μm (top row). c, Co-IP of HA-tagged NTA, Myc-tagged LRE and Flag-tagged FER expressed in Xenopus oocytes with or without the addition of RALFs (each 5 μM) as indicated. Anti-Flag M2 affinity beads were used to co-IP, and western blots were probed with antibodies against Myc, HA and Flag. n = 3 independent repeats. d, [Ca2+]cyt increases measured by single-cell imaging of COS7 cells expressing NTA (N), FER (F), FERK565R (kinase-dead version) (KD) or LRE (L), or combinations thereof. e,f, Typical whole-cell recordings (e) and current–voltage curves (f) of inward currents in HEK293T cells expressing NTA, FER and LRE. g,h, Similar analyses were conducted for HEK293T cells expressing NTA, LRE and the kinase-dead version of FER. i, The C-terminal cytosolic tail of MLO1 facilitated the PM localization of NTA–GFP. NTA–MLO1 denotes the chimeric protein of NTA and MLO1 C-terminal tail. n = 3 independent repeats. Scale bars, 5 μm (right column) or 10 μm (left column). j,k, Representative cytosolic Ca2+ spiking curves (j) and statistical analysis of peak values (k) in COS7 cells expressing the NTA–MLO1 chimeric or original channels. l,m, Typical whole-cell recordings (l) and current–voltage curves (m) of inward currents in HEK293T cells expressing the NTA-MLO1 chimeric or original channels. For Ca2+ imaging in COS7 cells, n = 6 replicates, and about 60 cells were imaged in each duplicate. For patch-clamp, n = 8 cells. Error bars depict the mean ± s.e.m. All P values were determined by two-tailed Student’s t-test.

Source data

Full size image

The FER–LRE–NTA trio mediates Ca2+ entry

Many of the tested MLO proteins (including NTA) failed to mediate Ca2+ entry in HEK293T or COS7 cells (Fig. 2a, Extended Data Fig. 4 and Supplementary Videos 8 and 9). We speculated that they may require other components to be active or they may not be properly targeted to the plasma membrane (PM). Indeed, NTA primarily accumulates in a Golgi-associated compartment36 and relocates to the synergid filiform apparatus in a FER- and LRE-dependent manner6,37. In our Ca2+ transport assays, PM localization would be crucial for mediating Ca2+ entry if NTA is indeed a Ca2+ channel. NTA–GFP was largely localized to intracellular punctate structures in COS7 cells (Fig. 2b). When co-expressed with FER and LRE, however, NTA–GFP was targeted to the PM (Fig. 2b). Such PM targeting was not achieved by co-expressing NTA–GFP with either FER or LRE alone, which is consistent with the finding that LRE–LLG1 physically interacts with and chaperones FER to the PM19 and that FER is required for the redistribution of NTA to the PM6. We further showed that NTA directly interacted with FER and that LRE enhanced such an interaction (Fig. 2c and Supplementary Fig. 1d), which indicates that FER, LRE and NTA form a complex, which we refer here as the NTA trio.

As FER and LRE together target NTA to the PM, we tested whether the NTA trio produces a functional channel at the PM. We co-expressed NTA with FER and LRE in COS7 and HEK293T cells and then performed imaging assays and patch-clamp recordings, which showed that NTA mediated Ca2+ influx (Fig. 2d–f, Supplementary Video 10 and Extended Data Fig. 7). Similar to AtMLO2 and HvMLO, the NTA trio conducted currents carried by divalent cations (Ca2+, Ba2+ and Mg2+), but not monovalent cations (K+ and Na+) (Extended Data Fig. 8). The Ca2+ channel activity of the NTA trio was inhibited by La3+ and Gd3+, which also blocked synergid Ca2+ spiking (Extended Data Fig. 8). The kinase-dead version of FER also formed an active NTA trio (Fig. 2d,g,h), which is consistent with an earlier finding that the kinase activity of FER is not required for pollen tube reception38,39.

Our data suggest that NTA is an active Ca2+ channel but requires FER–LRE for targeting it to the PM. We tested this idea by constructing a PM-localized chimeric NTA–MLO1 protein36,37, which showed that NTA–MLO1 mediated Ca2+ influx independently of FER–LRE (Fig. 2j–m).

RALFs enhance FER–LRE–NTA activity

We then tested the effect of RALFs on the activity of the NTA trio. RALF4 and RALF19, but not RALF34, significantly enhanced the Ca2+ channel activity of the NTA trio (Fig. 3a–d), which is consistent with the finding that RALF4 and RALF19 strongly induce increases in synergid Ca2+ (Fig. 1d,i). We further confirmed this observation by reconstituting the RALF–FER–LRE–NTA pathway in Xenopus oocytes and monitoring channel activity by two-electrode voltage-clamp assays (Fig. 3e,f). The chimeric NTA–MLO1 co-expressed with FER and LRE was also enhanced by RALF4 and RALF19, but not by RALF34 (Fig. 3g–j). Regarding the mechanism underlying the RALF4- and RALF19-dependent activation of the channel, a previous study6 has shown that NTA is redistributed to the filiform apparatus of the synergid following the arrival of the pollen tube. We examined the PM localization of NTA in response to RALFs, but did not observe any discernible effect of RALF4 and RALF19 application (Extended Data Fig. 9a). In this mammalian cell system, FER and LRE clearly facilitated the PM localization of NTA (Fig. 2b), which implies that a portion of NTA can be localized in the PM of the synergid in a pollen-tube-independent manner. Following pollen tube arrival, RALF4 and RALF19, and possibly other pollen tube signals (for example, mechanical stimulus), may further activate the Ca2+ channel by recruiting the trio to a specific location (for example, the filiform apparatus).

Fig. 3: RALFs enhance the Ca2+ channel activity of the FER–LRE–NTA trio.
figure 3

a,b, Representative cytosolic Ca2+ spiking curves (a) and statistical analysis of peak values (b) in COS7 cells expressing the FER–LRE–NTA trio or mock cells treated with various RALFs. The arrowheads indicate the time points at which 10 mM Ca2+ was applied. c,d, Typical whole-cell recording traces using the ramping protocol (c) and amplitudes at −180 mV (d) of Ca2+-permeable inward currents in HEK293T cells expressing FER–LRE–NTA or mock cells treated with various RALFs. e,f, Typical two-electrode voltage-clamp recordings (e) and current amplitudes at −160 mV (f) of inward currents in Xenopus oocytes expressing FER–LRE–NTA or mock water-injected oocytes treated with various RALFs. g,h, Representative cytosolic Ca2+ spiking curves (g) and statistical analysis of peak values (h) in COS7 cells expressing the chimeric NTA–MLO1 or FER–LRE–NTA–MLO1 treated with various RALFs. i,j, Typical whole-cell recording traces using the ramping protocol (i) and amplitudes at −180 mV (j) of Ca2+-permeable inward currents in HEK293T cells expressing FER–LRE–NTA–MLO1 treated with various RALFs. For Ca2+ imaging in COS7 cells, n = 8 replicates, and about 60 cells were imaged in each duplicate. For HEK293T cell recordings, n = 8 cells. For oocyte recordings, n = 8 oocytes. Error bars depict the mean ± s.e.m. All P values were determined by two-tailed Student’s t-test.

Source data

Full size image

During the revision of this manuscript, five other pollen tube RALFs (RALF6, RALF7, RALF16, RALF36 and RALF37) were reported to bind FER, ANJEA (ANJ) and HERCULES RECEPTOR KINASE 1 (HERK1) and to function redundantly in polytubey block and pollen tube reception40. We analysed RALF37 in our assays and found that RALF37, similar to RALF4 and RALF19, also triggered synergid Ca2+ changes and activated the NTA trio (Extended Data Fig. 10). This result suggests that multiple RALFs derived from the pollen tube serve as signals to trigger synergid Ca2+ spiking, which in turn leads to pollen tube reception. Consistent with this observation, single mutants of ralf4 and ralf19 did not show any detectable phenotypic defects (Extended Data Fig. 1b,c).

NTA–CaM shapes synergid Ca2+ spiking

MLO proteins contain a CaMBD in the intracellular carboxy-terminal region30,31, which suggests that these proteins may be regulated by CaM binding, a typical autoregulatory mechanism for many Ca2+ channels in both animal and plant systems41,42. We examined how CaM affects the channel activity of MLO proteins by co-expressing CaM7 with the NTA trio or other MLO proteins, including AtMLO2, HvMLO and NTA–MLO1, in COS7 cells. Substantial inhibition of Ca2+ entry was observed in all cases (Fig. 4a,b), thereby revealing an inhibitory feedback mechanism of MLO channel activity by CaM. We confirmed this mechanism using a mutant NTA (named NTARR), in which Leu455 and Trp458 were mutated to Arg to abolish its CaM binding capacity37. NTARR failed to respond to CaM-mediated inhibition (Fig. 4a,b). Although these mutations in the CaMBD partially impaired the redistribution of NTA to the filiform apparatus in the synergid37, the NTARR trio was recruited to the PM in COS7 cells (Extended Data Fig. 9b), which is consistent with the finding that the NTARR trio still conducted Ca2+ entry.

Fig. 4: CaM inhibition of NTA Ca2+ channels is involved in modelling the Ca2+ spiking pattern in synergids.
figure 4

a,b, Typical Ca2+ spiking patterns (a) and peak values (b) in COS7 cells expressing MLO proteins and AtCaM7. The arrowheads indicate the time points at which 10 mM external Ca2+ was applied. n = 8 replicates, and about 60 cells were imaged in each replicate. For b, numbers are as indicated for a (red). c–f, Typical whole-cell recordings of inward currents in HEK293T cells expressing the NTA trio, AtCaM7 or mock cells when [Ca2+]cyt was 0 nM (c,d) or 1 μM (e,f). g, Current amplitudes at −180 mV of HEK293T cells expressing the NTA trio and AtCaM7 when [Ca2+]cyt was 0 nM or 1 μM. n = 8 cells. Numbers are as indicated in c,d (blue) and e,f (green). h–j, Representative Ca2+ spiking patterns in synergids in response to PT arrival or 0.5 µM RALF4 for WT (h), nta-3 (i) and NTARR (j). k, The peak values of Ca2+ spiking as in h–j. n values are shown in Extended Data Fig. 3. Error bars depict the mean ± s.e.m. All P values were determined by two-tailed Student’s t-test. l, Model of RALF–FER–LRE–NTA pathway leading to synergid Ca2+ changes. Following PT arrival, PT-derived RALF4 and RALF19 bind FER–LRE, and this complex recruits and activates NTA, a CaM-gated Ca2+ channel, to initiate Ca2+ spiking.

Source data

Full size image

As CaM binds to HvMLO in a Ca2+-dependent manner43, we proposed that CaM may inhibit NTA channel activity following increased [Ca2+]cyt as a negative feedback mechanism. We tested this hypothesis by titrating [Ca2+]cyt and expressing a CaM7 mutant lacking the Ca2+-binding EF motif44. The results showed that CaM7 required Ca2+ binding to inhibit the activity of the NTA trio (Fig. 4c,e,g). Similarly, the NTARR mutant (which is defective in CaM binding) became constitutively active (Fig. 4d,f,g). These results support a model in which RALFs activate the NTA channel to increase synergid [Ca2+]cyt to a threshold that in turn enables CaM binding and inhibition of NTA channel activity.

Specific Ca2+ spiking in synergids is essential for pollen tube reception2,3. We hypothesized that the Ca2+/CaM-dependent feedback inhibition of the NTA channel provides a mechanism for shaping such a Ca2+ signature. To test this idea in planta, we generated transgenic plants harbouring the NTARR mutant driven by the NTA promoter in the nta-3 mutant background and examined synergid [Ca2+]cyt spikes in response to RALF4. [Ca2+]cyt spiking in synergids was amplified in NTARR plants (Fig. 4h–k). We also observed higher levels of Ca2+ increase in NTARR synergids in response to pollen tube arrival and a disordered oscillation pattern compared with WT synergids (Fig. 4j,k). This result indicates that the NTARR mutant, which lacks CaM-dependent inhibition, produces a sustained increase in [Ca2+]cyt, which causes a defect in pollen tube reception37.

Conclusions

We identified pollen-tube-derived RALF peptides as ligands for the FER–LRE co-receptor complex that recruits NTA, a CaM-gated Ca2+ channel, to PM domains to initiate Ca2+ entry and pollen tube reception (Fig. 4l). This work demonstrated a mechanistic process that integrates the action of FER, LRE and NTA, three players genetically connected in synergid–pollen tube interaction. In addition, the identification of MLO proteins as Ca2+ channels uncovered the long sought-after common biochemical pathway (Ca2+ entry) that involves MLO functions in multiple physiological processes, including but may not be limited to, mildew resistance, root mechanosensing, pollen tube growth and fertilization in plants. Indeed, Ca2+ is a core component in all these processes11,14, and our finding here sets the stage for extensive future research to address mechanisms in various MLO-dependent processes. As FER–LLG co-receptors are often connected to Ca2+ spiking in other signalling processes beyond reproduction21, the identification of a MLO channel downstream of the FER–LRE co-receptors offers a possible mechanism for other RALF–FER–LLG-dependent pathways. In the context of Ca2+ signalling, which is a common theme in all eukaryotes, MLO proteins represent a family of Ca2+ channels specific to the plant kingdom, which suggests that instead of having fewer Ca2+ channels than animals as currently thought11, plants may feature channels distinct from animal counterparts and more of these channels await to be discovered.

In the context of reproduction, our study raises several important questions for future research into the mechanistic details of male–female interactions. For example, although RALF4 and RALF19 bind to FER–LRE and enhance the channel activity of NTA, the mechanism underlying this activation awaits resolution by structural analysis of the FER–LRE–NTA trio in the presence of the RALF ligands. Before pollen tube reception, pollen tube integrity and guidance also involve the function of several RALF peptides, FER family of receptor-like kinases and MLO proteins. Our study provides a strategy for further research to link these components in distinct Ca2+ signalling pathways. A previous report34 noted that MLO5 and MLO9 are trafficked together with cyclic nucleotide-gated channel 18 (CNGC18), another Ca2+ channel with an essential role in pollen tube growth and guidance. This raises an interesting question regarding the functional interplay of multiple Ca2+ channels in shaping specific Ca2+ signatures in pollen tubes, synergids and other cell types in plants42.

Methods

Plant material and growth conditions

Seeds were sterilized with 10% (v/v) bleach and sown on agar plates containing half-strength Murashige and Skoog (1/2 MS) medium (1/2 MS, 0.8% (w/v) Phyto agar and 1% (w/v) sucrose, pH adjusted to 5.8 with KOH). Plates were incubated at 4 °C for 3 days for stratification and then transferred to the soil pots in a 22 °C growth room with a 16-h light/8-h dark cycle (100 μmol m−2 s−1). The seeds for fer-4 (GABI_GK106A06), lre-5 (CS66102) and nta-3 (SALK_027128) were purchased from Arabidopsis Biological Resource Center. The ralf4 and ralf19 mutants were generated by CRISPR as previously reported7.

Transgenic plants

The coding DNA sequence (CDS) of GCaMP6s was PCR-amplified using HBT-GCaMP6-HA as the template27 and fused to the MYB98 promoter region28, amplified from Columbia-0 (Col-0) genomic DNA in the pCAMBIA 2300 vector. The binary construct was transformed into Arabidopsis thaliana (Col-0) plants through Agrobacterium (GV3101) using the floral dip method45. Transgenic plants were selected on 1/2 MS plates containing 50 mg l–1 kanamycin, and one homozygous transgenic pMYB98-GCaMP6s line was then crossed with fer-4, lre-5 and nta-3 and further brought to homozygosity with both the GCaMP6s and the fer-4, lre-5 and nta-3 genetic backgrounds. The NTARR mutant was produced by site-directed mutagenesis to replace Leu455 and Trp458 with Arg. The NTA promoter was PCR-amplified from Col-0 genomic DNA and fused with the NTARR CDS in the pCAMBIA 1305 vector and transformed into plants as described above.

β-Glucuronidase staining

The mature pistils of the transgenic plants carrying proRALF4/19:β-glucuronidase (GUS) were dissected to isolate intact ovules that were then fixed in 80% acetone overnight. Samples were then incubated with GUS staining buffer (50 mM sodium phosphate, pH 7.2, 2 mM potassium ferrocyanide, 2 mM potassium ferricyanide, 0.2% Triton X-100 and 2 mM X-Gluc). Images were taken with a Zeiss AxioObserver Z1 inverted microscope.

Aniline blue staining

Pollen grains of a freshly opened flower of WT or mutant lines were used to pollinate WT pistils that had been emasculated a day earlier. After 24 h, the pistils were fixed in acetic acid/ethanol (1:3) overnight. They were then washed stepwise in 70% ethanol, 50% ethanol, 20% ethanol and ddH2O. The pistils were treated with 8 M NaOH overnight to soften the tissues and then washed with ddH2O three times before staining with aniline blue solution (0.1% aniline blue, 50 mM K3PO4) for 2 h. The stained pistils were observed using a Zeiss AxioObserver Z1 inverted microscope.

Mammalian cell culture, vector construction and transfection

The CDS of GCaMP6s was amplified from HBT-GCaMP6-HA27 and cloned into a dual-promoter vector, pBudCE4.1 (Invitrogen), with each CDS for NTA, FER, LRE, NTA–MLO1 for co-expression in HEK293T or COS7 cells. The chimeric NTA–MLO1 CDS was generated as previously described36.

Mammalian cells were cultured in DMEM supplemented with 10% FBS in a 5% CO2 incubator at 37 °C with controlled humidity. HEK293T or COS7 cells were transfected using a Lipofectamine 3000 Transfection Reagent kit (Invitrogen). Plasmids for transfection were extracted from Escherichia coli (DH5α) using a Plasmid Mini kit (Qiagen), and 2 μg plasmid DNA was added into each well of 6-well plates (Nunc) containing the cells (70–80% confluent). To confirm that the cells were successfully transfected, green and/or red fluorescence signals were examined using an inverted fluorescence microscope (Zeiss AxioObserver Z1 inverted microscope) before patch-clamp and Ca2+ imaging experiments 48 h after transfection.

Whole-cell patch-clamp recording

The whole-cell patch-clamp experiments were performed using an Axopatch-200B patch-clamp setup (Axon Instruments) with a Digitata1550 digitizer (Axon Instruments) as previously described46. Clampex10.7 software (Axon Instruments) was used for data acquisition, and Clampfit 10.7 was used for data analysis.

To record Ca2+ currents across the PM of HEK293T cells, the standard bath solution contained 140 mM N-methyl-d-glucamine (NMDG)-Cl, 10 mM CaCl2, 10 mM glucose and 10 mM HEPES, adjusted to pH 7.2 with Ca(OH)2. The standard pipette solution contained 140 mM Cs-glutamate, 6.7 mM EGTA, 3.35 mM CaCl2 and 10 mM HEPES, adjusted to pH 7.2 with CsOH. Free [Ca2+] in the pipette solution was 175 nM, as calculated using the Webmaxc Standard (https://somapp.ucdmc.ucdavis.edu/pharmacology/bers/maxchelator/webmaxc/webmaxcS.htm). The 10 mM Ca2+ in the bath solution was removed to attain 0 mM Ca2+ or substituted with 10 mM Ba2+ or 10 mM Mg2+ as indicated. A ramp voltage protocol of 2-s duration from −160 mV to +30 mV (holding potential 0 mV) was applied 1 min after achieving a whole-cell configuration, and currents were recorded every 20 s, with 5 repeats in total for each cell. The five current traces were used for statistical analysis to produce average current–voltage curves.

For inward K+ current recordings in HEK293T cells, the bath solution contained 140 mM NMDG-Cl, 14.5 mM KCl, 10 mM glucose and 10 mM HEPES, adjusted to pH 7.2 with KOH. The pipette solution contained 145 mM K-glutamate, 3.35 mM EGTA, 1.675 mM CaCl2 and 10 mM HEPES, adjusted to pH 7.2 with KOH. The free [Ca2+] in the pipette solution was 100 nM, as calculated using the Webmaxc Standard.

For inward Na+ current recordings in HEK293T cells, the bath solution contained 140 mM NaCl, 10 mM glucose and 10 mM HEPES, adjusted to pH 7.2 with NaOH. The pipette solution contained 135 mM CsCl, 10 mM NaCl, 3.35 mM EGTA, 1.675 mM CaCl2 and 10 mM HEPES, adjusted to pH 7.2 with CsOH. The free [Ca2+] in the pipette solution was 100 nM, as calculated using the Webmaxc Standard.

A step voltage protocol of 4-s duration for each voltage from −160 mV to +60 mV with a +20 mV increment was used for K+ and Na+ current recordings in HEK293T cells 1 min after achieving a whole-cell configuration.

Two-electrode voltage-clamp recording from Xenopus oocytes

The CDS for NTA-3×HA, LRE-4×Myc and FER-3×Flag were cloned into the pGEMHE Xenopus oocyte expression vector. To construct LRE-4×Myc, the 4×Myc tag sequence was inserted after the first 60 bp of the LRE CDS encoding the signal peptide, followed by the downstream 438 bp of LRE as previously described23.

Two-electrode voltage-clamp assays were performed as previously reported35,44. The capped RNA (cRNA) was synthesized from 1 μg of a linearized plasmid DNA template using a mMESSAGE mMACHINE T7 kit (Ambion) and 10 ng of each cRNA, in a total volume of 46 nl, was injected into each oocyte. Injected oocytes were incubated in ND96 solution (96 mM NaCl, 2 mM KCl, 1 mM MgCl2, 1.8 mM CaCl2, 10 mM HEPES/NaOH, pH 7.4) at 18 °C for 2 days before electrophysiological recording. Oocytes were voltage-clamped using a TEV 200A amplifier (Dagan), a Digidata 1550 A/D converter, and recorded using CLAMPex 10.7 software (Axon Instruments). The pipette solution contained 3 M KCl. The standard bath solution contained 30 mM CaCl2, 1 mM KCl, 2 mM NaCl, 130 mM mannitol and 5 mM MES-Tris (pH 5.5). Voltage steps were applied from +40 mV to −160 mV in −20 mV decrements over 0.8 s.

Single-cell Ca2+ imaging in mammalian cells

HEK293T or COS7 cells expressing GCaMP6s and various combinations of candidate channel proteins were monitored using a Zeiss AxioObserver Z1 inverted microscope (Ivision 4.5 software) with a ×20 objective as previously reported35. The interval of data acquisition was 2 s. The standard solution for Ca2+ imaging contained 120 mM NaCl, 3 mM KCl, 1 mM MgCl2, 1.2 mM NaHCO3, 10 mM glucose, 10 mM HEPES, pH 7.5. About 60 s after initiation of the imaging procedure, the bath was perfused using a peristaltic pump with the standard solution supplemented with 10 mM Ca2+ and/or RALFs to elicit Ca2+ entry through active channels.

Synergid cell Ca2+ imaging

For the RALF-induced synergid [Ca2+]cyt increase experiment, unfertilized ovules were dissected from flowers as previously described47. The pistil was dissected to remove the ovules from the placenta using a surgical needle. The isolated ovules were placed in pollen germination medium (PGM), which contained 18% sucrose, 0.01% boric acid, 1 mM MgSO4, 1 mM CaCl2, 1 mM Ca(NO3)2 and 0.5% agarose (pH 7.0)48. After 2 h of incubation at 22 °C and 100% relative humidity, synergids expressing GCaMP6s were monitored using a Zeiss AxioObserver Z1 inverted microscope (Ivision 4.5 software) with a ×20 objective, and various RALFs were added to the ovules as indicated.

For the pollen-tube-induced synergid [Ca2+]cyt increase experiment, we followed a previously published protocol2,3. Dissected ovules of emasculated flowers expressing GCaMP6s were placed on PGM. Unpollinated pistils were cut with a razor blade (VWR International) at the junction between the style and ovary. The stigmas were placed on the PGM and manually pollinated with pollen grains expressing DsRed. Pollinated stigmas were positioned 150 µm away from the ovules, and pollen tube growth was monitored using a fluorescence microscope. Time-lapse Ca2+ imaging began after the pollen tube entered the ovule micropyle.

Protein localization

Transfected COS7 cells were washed with PBS and mounted onto slides for image acquisition with a Zeiss LSM 880 confocal microscope and ZEN2012 software.

Peptide purification

All tag-free RALF peptides used in this study were purified from insect cells (High 5). The pFastBac vector containing RALF4, RALF19 and LRX8 were gifts from J. Santiago (University of Lausanne), and RALF4 and RALF19 peptides were purified as previously reported49.

For RALF8, RALF9, RALF15, RALF25, RALF26, RALF30 and RALF34, the CDS encoding RALF mature peptides were cloned into a modified pACEBAC1 (Geneva Biotech) vector in which RALFs were amino-terminally fused to a 30K signal peptide, a 10×His tag, thioredoxin A and a tobacco etch virus (TEV) protease site.

High 5 cells were infected with virus with a multiplicity of infection of 3 and incubated for 1 day at 28 °C and 2 days at 22 °C at 110 r.p.m. on an orbital shaker. The secreted peptides were purified from the supernatant with a Ni2+ column (Ni-NTA, Qiagen), and incubated with TEV protease (NEB) to remove the tags. Peptides were further purified by size-exclusion chromatography on a Superdex 200 increase 10/300 GL column (GE Healthcare), equilibrated in 20 mM sodium citrate, pH 5.0, 150 mM NaCl. The peptides were diluted with sterile pure water before use.

Protein–protein interaction assays

For pull-down assays, MBP–FERex, GST–RALFs and GST–LRE were produced in E. coli Rosetta (DE3) by 0.1 mM IPTG induction overnight at 16 °C and bound to amylose or glutathione resins for purification as previously reported19,50. The pull-down buffer contained 40 mM Tris–HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 5% glycerol, 5 mM MgCl2, 1 mM PMSF, complete protease inhibitor cocktail (Roche) at 1:100 dilution, and 0.4% Triton X-100. Proteins were applied to amylose resin and incubated at 4 °C for 2 h with gentle mixing. The resin was washed three times in pull-down buffer. Proteins that remained bound to the resin were eluted by mixing with SDS–PAGE loading buffer, boiled for 5 min and subjected to 12% SDS–PAGE and western blotting.

For co-IP of tobacco leaves, 35S:FERK565R–3×Flag and 35S:LRE–4Myc constructs were co-transformed into Agrobacterium tumefaciens (strain GV3101) and infiltrated into N. benthamiana leaves23. Sixty hours after inoculation, leaves were detached and treated with 5 μM RALFs for 2 h before total protein was extracted and applied to anti-Flag M2 affinity agarose gel (Sigma-Aldrich). After incubation at 4 °C for 2 h with gentle mixing, the resin was washed three times in pull-down buffer, and the bound protein was eluted by mixing with SDS–PAGE loading buffer, boiled for 5 min and subjected to 10% SDS–PAGE and western blotting.

For co-IP of Xenopus oocytes, cRNAs of FER–3×Flag, NTA–3HA and LRE–4Myc were injected into oocytes, incubated for 3 days, followed by treatment with 5 μM RALFs for 2 h. Total protein was extracted in the pull-down buffer and then applied to anti-Flag M2 affinity agarose gel (Sigma-Aldrich). After incubation at 4 °C for 2 h with gentle mixing, the resin was washed three times in pull-down buffer, and the bound protein was eluted by mixing with SDS–PAGE loading buffer, boiled for 5 min and subjected to 10% SDS–PAGE and western blotting.

For chemiluminescence detection, the following antibodies were used: anti-GST–HRP (1: 2,000 dilution), anti-Myc–HRP (1: 2,000 dilution), anti-HA (1: 2,000 dilution), anti-MBP (1: 2,000 dilution) and anti-mouse secondary (1:20,000 dilution) antibodies from Santa Cruz Biotechnology; and anti-Flag antibody (1:4,000 dilution) from Sigma-Aldrich.

Image processing and data analysis

ImageJ (v.1.51j8) was used to analyse GCaMP6s signals over time at several regions of interest. To calculate the fractional fluorescence change (ΔF/F), the equation ΔF/F = (F − F0)/F0 was used, where F0 denotes the average baseline fluorescence determined by the average of F over the first 10 frames of the recording before the treatment.

Microsoft Excel in Office 365 and GraphPad Prism 7.0 were used for calculation and statistical analyses of the data. Adobe Illustrator CC 2019 was used for image assembly. Clampfit 10.7 was used to analyse and process data from the electrophysiological experiments. All experiments were independently reproduced in the laboratory.

Reporting summary

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