Calcium signals the making of symbiosis
Plant cell nuclei respond to signals from symbiotic nitrogenfixing rhizobial bacteria or arbuscular mycorrhizal fungi with oscillating Ca2+ release. Charpentier et al. identified a trio of responsible Ca2+ channels in a legume. These channels contain nuclear localization signals and are expressed in root cell nuclear envelopes. The channels function early in the establishment of symbiosis to produce oscillations in Ca2+ release from nuclear stores.
Science, this issue p. 1102
Abstract
Nuclear-associated Ca2+ oscillations mediate plant responses to beneficial microbial partners—namely, nitrogen-fixing rhizobial bacteria that colonize roots of legumes and arbuscular mycorrhizal fungi that colonize roots of the majority of plant species. A potassium-permeable channel is known to be required for symbiotic Ca2+ oscillations, but the calcium channels themselves have been unknown until now. We show that three cyclic nucleotide–gated channels in Medicago truncatula are required for nuclear Ca2+ oscillations and subsequent symbiotic responses. These cyclic nucleotide–gated channels are located at the nuclear envelope and are permeable to Ca2+. We demonstrate that the cyclic nucleotide–gated channels form a complex with the postassium-permeable channel, which modulates nuclear Ca2+ release. These channels, like their counterparts in animal cells, might regulate multiple nuclear Ca2+ responses to developmental and environmental conditions.
Nuclear calcium (Ca2+) signals transduce a variety of stimuli in animal (1) and plant cells (2). Plant genomes lack genes similar to those that encode mammalian nuclear-localized Ca2+ channels (3), implying that plants have alternative mechanisms for nuclear Ca2+ release. Legume root cells show nuclear Ca2+ oscillations in response to signaling molecules from nitrogen-fixing rhizobial bacteria and arbuscular mycorrhizal (AM) fungi (4). In Medicago truncatula, the potassium (K+)–permeable channel DMI1 (does not make infections 1) and the Ca2+-dependent adenosine triphosphatase (Ca2+-ATPase) MCA8 are necessary for symbiotic Ca2+ oscillations; both are located on nuclear membranes, suggesting that the nuclear envelope [contiguous with the endoplasmic reticulum (ER)] acts as the Ca2+ store (5). Here we identify the Ca2+-permeable channels responsible for Ca2+ release from the nuclear envelope–ER stores.
We screened the M. truncatula protein database (6) to identify transmembrane proteins containing motifs present in Ca2+ channels (PF00622, PF02815, PF08763, PF00036, and PF06459), motifs related to ion channels (PF00520, PF07885, PF02386, PF02026, and PF08709), and nuclear localization signal (NLS) motifs (fig. S1A). We assessed their impact on symbioses by silencing the genes that encode these candidates in M. truncatula roots. We found that members of the cyclic nucleotide–gated channel (CNGC) family are required for symbiotic associations (fig. S1, B to E). Among the 21 CNGCs of M. truncatula, 14 are predicted to contain NLS motifs such as those in DMI1 (5) and CASTOR (7) (fig. S2), and these fall into CNGC groups II, III, IVA, and IVB (8) (fig. S2A).
To differentiate which NLS-containing CNGCs contribute to symbiotic signaling, we used RNA interference to reduce expression of one or several genes from each CNGC group (figs. S3 and S4). Silencing of CNGC15a, CNGC15b, and/or CNGC15c—all members of group III—correlated with defects in symbiotic associations (Fig. 1A; fig. S3; and fig. S5, A, C, and D). Of the other group III members, CNGC16 is not expressed in roots (fig. S5B), and silencing CNGC17 and CNGC18 did not reduce nodulation (fig. S5, E to J).
(A) CNGC15a, CNGC15b, and CNGC15c belong to subclass C of the group III CNGCs. The phylogenetic tree includes CNGC group III members from Amborella trichopoda (Am, yellow squares), Oryza sativa (Os, green triangles), Arabidopsis thaliana (At, blue polygons), and M. truncatula (Mt, red dots). Numbers at the branch points indicate the percentage bootstrap values (for 1000 iterations) of the consensus tree. (B) The cngc15a, -b, and -c mutants were assessed at 25 days after inoculation with S. meliloti strain 2011 and (C) 5 weeks after inoculation with R. irregularis (WT, wild-type M. truncatula). (D) Complementation of the nodulation and (E) mycorrhization phenotypes of cngc15a, -b, and -c by expressing CNGC15a, -b, and -c fused to GFP, driven by the Lotus japonicus Ubiquitin promoter (EV, empty vector). (F) Infection pockets (IP) and infection threads (IT) were quantified from two biological replicates, 7 days after inoculation with S. meliloti. (G) Nod factor–induced ENOD11 expression and (H) NS-LCO–induced NADP dpt Ox (nicotinamide adenine dinucleotide phosphate–dependent oxidoreductase) expression in WT and cngc15 mutants after 2, 6, and 24 hours of treatment. Values in (G) and (H) are from three biological replicates each of ten roots, and expression is normalized to that of EF-1α (elongation factor 1α). In (B), (D), (F), (G), and (H), error bars show standard deviation. In (C) and (E), error bars show standard error. The number of plants analyzed is given at the bottom of each column. *P < 0.05; **P < 0.01; ***P < 0.001 (t test).
We identified mutants in CNGC14, CNGC15a, CNGC15b, and CNGC15c, as well as in a CNGC gene that falls outside group III, CNGC4a, as a control. One or two insertion alleles were identified for each gene (fig. S6), and these were confirmed to reduce mRNA levels (fig. S6). Nodulation and fungal colonization were assessed in all mutant lines 25 days after inoculation with Sinorhizobium meliloti or 5 weeks after inoculation with Rhizophagus irregularis. The mutants cngc15a, cngc15b, and cngc15c all showed reduced nodulation and reduced R. irregularis colonization (Fig. 1, B and C, and fig. S7, A and B), whereas cngc4a and cngc14 did not. The symbiotic defects of cngc15 mutants were complemented with their respective genes (Fig. 1, D and E). The mutant lines showed defects in rhizobial infection (Fig. 1F) and induction of early gene expression in response to lipochitooligosaccharide (LCO) signals produced by S. meliloti (Nod factor) or R. irregularis [nonsulfated (NS)–LCO] (Fig. 1, G and H). We conclude that CNGC15a, CNGC15b, and CNGC15c function early in the establishment of both rhizobial and mycorrhizal associations.
Oscillations in nuclear Ca2+ levels are a feature of the common symbiotic signaling pathway (9). The mutants cngc15a, -b, and -c, but not cngc4a or cngc14, reduced nuclear-associated Ca2+ oscillations in response to Nod factor treatment (Fig. 2A and fig. S7C). Fewer cells responded and many cells showed defective Ca2+ oscillations in cngc15 mutants (Fig. 2B), as measured by interspike intervals and probability density functions. Defects included poor maintenance of Ca2+ oscillations and irregular oscillation frequencies (Fig. 2, A and B, and fig. S7C). Similar effects were observed after treatments with a mycorrhizal-produced signaling molecule (fig. S8). Nod factor also activates an influx of Ca2+ across the plasma membrane, but considering the location of the CNGC15 proteins (discussed below), we do not believe that these proteins will be involved in this Ca2+ response. Our data demonstrate that the CNCG15 proteins are required for full activation of nuclear-localized Ca2+ oscillations.
(A) Nod factor (10−9 M)–induced Ca2+ spiking, assayed in WT and cngc mutants. The number of cells analyzed is shown in fig. S7C. **P < 0.01; ***P < 0.001 (χ2 test). Below, black traces show examples of calcium traces that were scored as WT or abnormal. (B) Representative Ca2+ traces of WT roots and mutants, showing detrended Oregon green/Texas red ratios in arbitrary units. The panels on the right show density plots of average interspike intervals (ISI) during 60 min, starting 7 min after the first spike (n = 26 WT, 26 cngc15b-2, 24 cngc15a, and 17 cngc15c-1; PDF, probability density function). (C) CNGC15 proteins complement cch1/mid1. Discs containing α factor inhibit growth in the mutant but not the WT or CNGC15-complemented strains. CNGC15a and CNGC15b with their C termini truncated (ΔCterm) and CNGC4 do not complement cch1/mid1. (D and E) Average current-voltage (I-U) relationship in oocytes expressing CNGC15a (D) or injected with water (E) in presence of CaCl2 or BaCl2. Error bars represent standard deviation (n = 4).
We generated double mutants to further assess the function of these channels, but we found equivalent defects to those observed in the single mutants (fig. S7, C to E). Our attempts to generate a triple mutant by crossing cngc15b-1 cngc15c with cngc15a-1 cngc15b-2 or cngc15a with cngc15b-1 cngc15c were unsuccessful. The efficiency of the crosses, as demonstrated by pod formation, dropped from ~90% (for instance, 9 in 10 crosses of cngc15a-1 with cngc15b-2 were successful) to 0% for the generation of the triple mutant (no successful crosses in 103 attempts) (fig. S7F). Fertilization requires coordinated Ca2+ signals from the female gametophyte in response to the Ca2+ dynamics of the pollen tube (10, 11). CNGC15a, CNGC15b, and CNGC15c are all expressed in flowers and pods (fig. S5B), and cngc15a and cngc15b had decreased fertilization rates, similar to the double mutants cngc15b-1 cngc15c and cngc15a-1 cngc15b-1 (fig. S9). This suggests a possible function for these channels during fertilization in both female and male gametophytic Ca2+ signaling.
Our results demonstrate that CNGC15 proteins support Nod factor– and Myc factor–induced nuclear-localized Ca2+ oscillations. We assessed the location of these channels by using functional green fluorescent protein (GFP) fusions (Fig. 1, D and E), analyzed by confocal microscopy and transmission electron microscopy of GFP immunogold–labeled sections (Fig. 3 and figs. S10 and S11A). We found that CNGC15a, CNGC15b, and CNGC15c all localized to the nuclear envelope. We also generated antibodies specific to CNGC15a (fig. S11B), with which we were able to confirm the presence of CNCG15a specifically on nuclear membrane fractions (Fig. 3K). These CNGC proteins and DMI1 have predicted NLS motifs in the N terminus or between the first few transmembrane domains. This is similar to many nuclear membrane–targeted proteins in yeast and animals (12); however, additional work is necessary to test the relevance of this positioning of the NLS in plant proteins.
Confocal microscopy of roots expressing the plant marker DsRed alone (A and B) and with CNGC15b:GFP (C and D), CNGC15a:GFP (E and F), CNGC15c:GFP (G and H), and DMI1:GFP (I and J) (where the colon denotes fusion). In (A), (C), (E), (G), and (I), the overlay of the green and red channels is shown. In (B), (D), (F), (H), and (J), the green channel is shown. Scale bars, 20 μm. (K) The microsomal fraction depleted of nuclei (–Nuc) and the membrane fraction of purified nuclei (+Nuc) from M. truncatula roots, assessed with indicated antibodies. Antibodies to H+ATPase and MCA8 were used as plasma membrane and ER–nuclear membrane markers, respectively. The presence of CNGC15a was detected with its native antibody.
To determine whether CNGC15 subgroup members are permeable to Ca2+, we expressed them in the cch1/mid1 yeast mutant. In response to mating pheromone (α factor), cch1/mid1 fails to generate cytosolic Ca2+ signals, resulting in growth arrest (13). This phenotype can be complemented by the release of Ca2+ into the cytosol from any Ca2+ store. Each of the three CNGC15 proteins was individually able to complement the cch1/mid1 growth-arrest phenotype (Fig. 2C), suggesting that they are Ca2+-permeable. Additionally, expression of CNGC15a in Xenopus laevis oocytes triggered an inward Ca2+ current (Fig. 2D) that was not observed in control oocytes (Fig. 2E), further demonstrating the Ca2+ permeability of this group of channels.
Mathematical modeling has demonstrated that the interplay between the K+-permeable channel DMI1, a Ca2+ channel, and a Ca2+-ATPase can give rise to sustained Ca2+ oscillations (14). Based on our findings regarding the involvement of CNGC, we adapted this model to introduce a cyclic nucleotide–gating mechanism for the Ca2+ channel (15) and demonstrated that this system can recapitulate the observed Ca2+ oscillations (fig. S12). This model suggests that the activation of DMI1 and CNGC15 must occur simultaneously; introducing lag times in their activation stopped the Ca2+ oscillations (fig. S13). One possible mechanism for such simultaneous activation would be the presence of both channels in a single complex. We found that the soluble C terminus of DMI1 and the N termini of CNGC15a, CNGC15b, and CNGC15c interact in yeast (Fig. 4A and fig. S11C). Using bimolecular fluorescence complementation (16) of the full-length proteins, we observed interactions at the nuclear envelope between DMI1 and the CNGC15 proteins in both Nicotiana benthamiana leaves and M. truncatula roots (Fig. 4, B and C); we did not observe interactions in control iterations (fig. S14). Nod factor had no effect on these interactions (fig. S15), implying that the complex is maintained after activation.
(A) Yeast two-hybrid assays between the C-terminal domain of DMI1 as bait (BD) and the N-terminal domains of CNGC4a, the N-terminal domains of CNGC15 proteins, and the C-terminal domain of DMI1 as prey (AD) (details are given in the supplementary materials). Murine p53 and its interacting partner SV40 large T antigen (simian virus large tumor antigen) were used as a control. SD-LW, synthetic dropout medium lacking Leu and Trp; SD-AHLW, synthetic dropout medium lacking adenine, His, Leu, and Trp. (B) N. benthamiana leaves transiently expressing N-terminal Venus (NE) fused to the N termini of each CNGC15 and C-terminal Venus (CE) fused to the C terminus of DMI1. DsRed was present in all vectors. Scale bars, 16 μm. (C) M. truncatula root cells expressing the same fusions. Scale bars, 10 μm. (D) A model of CNGC15 and DMI1 dynamics at the inner nuclear envelope (INE). DMI1 interacts physically with CNGC15 to control the simultaneous opening of both channels through direct binding of a secondary messenger to DMI1 or CNGC15 after stimulation with Nod factor (+NF).
We have shown that CNGC15 proteins are responsible for nuclear Ca2+ oscillations in the symbiotic signaling pathway of M. truncatula. CNGCs have been demonstrated to locate at vacuoles (17) or the plasma membrane (18). In contrast, CNGC15 proteins are located at the nuclear envelope and are permeable to Ca2+. We propose that the location of these Ca2+ channels allows a targeted nuclear release of the ER Ca2+ store. Physical interactions between the CNGC15 proteins and DMI1 may support synchronous activation and modulate the Ca2+ signal (Fig. 4D). Further work on this nuclear channel complex should clarify how it is regulated and its implication in developmental processes such as fertility.
Supplementary Materials
www.sciencemag.org/content/352/6289/1102/suppl/DC1
Materials and Methods
Figs. S1 to S15
Tables S1 to S2
References and Notes
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- Acknowledgments: We thank J.-M. Ané for the construct 35S:DMI1:GFP, E. Peiter for the cch1/mid1 mutant, G. Calder for confocal help, E. Barclay for immunogold labeling, P. Bailey for bioinformatics, K. Mysore and J. Wen for Tnt1 mutant screening, and M. Hopkins for models of CNGC action. The work was supported by the Biotechnology and Biological Sciences Research Council (grants BB/J004553/1 and BB/J018627/1) and by the European Research Council (SYMBIOSIS project). M.C. and G.E.D.O. directed the research. M.C. performed the experiments with contributions by J.S. (calcium spiking), T.V.M. and R.J.M. (mathematical modeling), G.V.R. (phylogeny), K.F. (immunogold labeling and electron microscopy), E.S. (RNA extraction and quantitative reverse transcriptase polymerase chain reaction), and J.T. and A.-A.V. (electrophysiology). M.C., G.E.D.O., and D.S. wrote the manuscript. The supplementary materials contain additional data.
