A Putative Ca2+ and Calmodulin-Dependent Protein Kinase Required for Bacterial and Fungal Symbioses

See allHide authors and affiliations

Science  27 Feb 2004:
Vol. 303, Issue 5662, pp. 1361-1364
DOI: 10.1126/science.1093038


Legumes can enter into symbiotic relationships with both nitrogen-fixing bacteria (rhizobia) and mycorrhizal fungi. Nodulation by rhizobia results from a signal transduction pathway induced in legume roots by rhizobial Nod factors. DMI3, a Medicago truncatula gene that acts immediately downstream of calcium spiking in this signaling pathway and is required for both nodulation and mycorrhizal infection, has high sequence similarity to genes encoding calcium and calmodulin-dependent protein kinases (CCaMKs). This indicates that calcium spiking is likely an essential component of the signaling cascade leading to nodule development and mycorrhizal infection, and sheds light on the biological role of plant CCaMKs.

The legume-rhizobia symbiosis fixes as much nitrogen worldwide as the chemical fertilizer industry, owing to the ability of rhizobial bacteria to induce the morphogenesis of a new plant organ, the legume root nodule, in which they fix nitrogen. Rhizobial signals, the lipochito-oligosaccharidic Nod factors, initiate symbiotic responses on the roots of legume hosts and are required for recognition, controlled infection, and nodule formation (14). These molecules induce a variety of responses, including rapid calcium influx and calcium spiking in root hair cells, specific gene induction, alterations in epidermal cell morphology, and cortical cell mitosis (4).

Genetic studies in the model legumes Medicago truncatula and Lotus japonicus have led to the identification of genes involved in the perception and transduction of Nod factors. Two types of transmembrane receptor–like serine/threonine kinases with putative extracellular regions containing LysM domains (LysM-RLKs), required for Nod factor responses and rhizobial infection, are hypothesized to form heterodimeric Nod factor receptors (57). In M. truncatula, three downstream genes, DMI1, DMI2, and DMI3, are required for both nodulation and the formation of arbuscular mycorrhizae (the AM symbiosis). The AM symbiosis occurs in most land plants (a notable exception is the Brassicaceae family, which includes Arabidopsis thaliana) and involves fungi of the order Glomales (810). Both the fungal and bacterial symbiosis partners trigger plant host genetic programs that permit controlled and localized infection. DMI2 encodes a leucinerich repeat receptor–like protein kinase (NORK) (11). Mutants of DMI1, DMI2, and DMI3 are blocked for most responses to Nod factors: They do not exhibit induction of root hair branching, early nodulin gene expression, and cortical cell division (3). However, they differ in their ability to respond to Nod factors by the induction in root hair cells of sharp oscillations in the concentration of cytoplasmic calcium (calcium spiking), which is lost in dmi1 and dmi2 but not in dmi3 mutants (12, 13). DMI3 thus seems to act immediately downstream of calcium spiking.

DMI3 maps to the south arm of chromosome 8 (14), between the two markers SDP1 and PU01, located 1 and 2 cM, respectively, from DMI3. These two markers were used to isolate primary bacterial artificial chromosomes (BACs), and an 800-kb contig of BACs that spanned the DMI3 region was assembled using a combination of chromosome walking from BAC end sequences and restriction endonuclease fingerprinting of the clones (fig. S1). Genetic markers originating from the BAC contig were developed and recombination events were used to position the DMI3 locus to a 190-kb interval. Two BACs encompassing this region were sequenced (GenBank accession numbers AY508218 and AY508219), which allowed the identification of candidate genes on the basis of sequence homologies. Among them we noted several genes coding for transcription factors (fig. S1). A putative calcium and calmodulin-dependent protein kinase (CCaMK) detected in the sequenced region appeared as a promising candidate for DMI3 (Fig. 1A). Using specific primers, we amplified this gene from genomic DNA of two dmi3 mutant alleles, TRV25 and T1-5, and sequenced it. The wild-type and TRV25 allele differed by a 14 – base pair (bp) deletion in the predicted kinase domain of the mutant, leading to a premature stop codon, whereas the T1-5 allele exhibited a point mutation in the kinase domain, leading to a stop codon. Identification of mutations in two independent mutant alleles together with the physical and genetic location of DMI3 provides strong evidence that this candidate gene is DMI3.

Fig. 1.

Structure of the M. truncatula DMI3 gene. (A) Intron-exon structure of DMI3 and position of the predicted protein domains and motifs. (B) Comparison of the structural organization of two types of calcium sensor kinases: a putative CCaMK (M. truncatula DMI3) and a CDPK (A. thaliana CPK1). The kinase domain (vertical bars), calmodulin binding domain (black box), and calcium-binding EF hands (hatched boxes) are indicated.

To test this hypothesis, we attempted to complement the Nod mutation in TRV25 using the wild-type genomic sequence of the candidate gene. Because the nodulation phenotype of TRV25 depends on the root genotype (14), we used Agrobacterium rhizogenes for root transformation. Roots appearing on TRV25 seedlings after A. rhizogenes infection were inoculated with the rhizobial symbiont Sinorhizobium meliloti. Eighty percent of plants transformed by A. rhizogenes carrying the candidate gene developed nodules (an average of five nodules per plant) (Fig. 2). Microscopic examination of a representative sample of mature nodules revealed the presence of bacteria within the central tissue. No nodules could be detected on roots of plants transformed by A. rhizogenes carrying the vector (15). This complementation, together with the above described genetic evidence, demonstrates that the candidate gene homologous to CCaMK is DMI3.

Fig. 2.

Functional complementation of the dmi3 mutation by the wild-type DMI3 genomic sequence. Roots of seedlings of the TRV25 dmi3 mutant, defective for nodulation, were transformed with an Agrobacterium rhizogenes strain, Arqua1, into which we introduced the pF5 plasmid that carries the wild-type DMI3 gene. Ninety percent of plants exhibited transformed roots, as detected by GUS activity. A total of 45 plants transformed by pF5 were tested, and 36 of them nodulated with an average of five nodules per plant. X-Gal treatment of a representative sample of nodules revealed the presence of bacteria inside the nodules. No nodules were detected on 26 control plants transformed by the pCAMBIA2201 vector. Bar, 1 mm.

To further substantiate that this gene is required for nodulation and mycorrhizal infection phenotypes, we sequenced the orthologous gene in pea, a legume phylogenetically related to Medicago for which mutants with phenotypes and syntenic map positions similar to that of TRV25 have been described (14, 1618) (table S1). As dmi3 mutants, pea mutants at the sym9 locus are altered downstream of the calcium spiking response induced by treatment of root hairs by Nod factors and are defective for root hair curling, infection thread formation, nodulation, and mycorrhizal infection (17, 18). Using a polymerase chain reaction (PCR) approach, we identified a pea gene that exhibited high overall sequence similarity (90% identity) to M. truncatula DMI3 (GenBank accession numbers AY502067, AY502068, and AY502069). Each sequenced allele from eight sym9 mutants of P. sativum, obtained in three different genetic backgrounds, exhibited single base-pair alterations in the predicted kinase domain that were likely to alter the protein function (Table 1 and fig. S2). The identification of mutations in eight alleles obtained from independent sym9 mutants further supports a role of CCaMKs in nodulation and mycorrhizal infection. In addition, it allows the molecular characterization of a major gene controlling both nodulation and mycorrhizal infection in pea.

Table 1.

Summary of M. truncatula dmi3 and P. sativum sym9 alleles.

Genetic background Allele Mutation
Mt cv. TRV25 14-bp deletion:
Jemalong 198-203/stop
Mt cv. T1-5 CGA/TGA : R97/stop
    A17 ENOD11-GUS
Ps cv. Frisson P1 CAA/TAA : Q230/stop
P2 CAA/TAA : Q230/stop
P3 CAA/TAA : Q230/stop
P53 TGG/TGA : W240/stop
Ps cv. Sparkle R72 TGT/TGA : S224/stop
Ps cv. Finale DK6 CTT/C-T : L188/stop
DK9 GGG/AGG : G202/R
DK22 TCT/TTT : S24/F

From the M. truncatula expressed sequence tag (EST) collection, we identified eight cDNAs from various libraries with sequences matching the DMI3 genomic sequence. Alignment of these sequences defined a gene structure with seven exons. The full-length open reading frame of 1569 nucleotides encodes a protein of 523 amino acids with a predicted molecular mass of 58,600 daltons. Comparative sequence analysis indicated that DMI3 belongs to the CCaMK group of serine-threonine protein kinases, together with proteins encoded by genes previously characterized from tobacco, lily, rice, and the moss Physcomitrella (Figs. 1 and 3; fig. S2) (19, 20). Like these proteins, DMI3 is predicted to share with the more common plant calcium-dependent protein kinases (CDPKs) an N-terminal kinase domain. However, it differs from CDPKs in the structure of its C-terminal calcium-binding regulatory domain, which has higher similarity to a mammalian visinin-like domain (with three calcium-binding EF hands) than to the calmodulin-like domain typical of CDPKs (with four EF hands) (Fig. 1B). Between the kinase and calcium-binding domains lies a calmodulin-binding domain that overlaps an autoregulatory domain. This structure allows regulation of the kinase activity by both calcium and calmodulin (21, 22).

Fig. 3.

Comparison of M. truncatula DMI3 (GenBank accession number AY502066) to the previously characterized Lilium longiflorum CCaMK (GenBank accession number U24188) (19). Conserved amino acids are shaded black. The serine/threonine kinase domain (vertical bars), calmodulin-binding site (black box), calcium-binding EF hand motifs (hatched bars), and autophosphorylation site (star) of CCaMK are indicated. DMI3 also has high sequence similarity to rice (GenBank accession number AK070533), tobacco (GenBank accession number AF087813), and Physcomitrella (GenBank accession number AY155462) putative CCaMKs (fig. S2).

Electronic Northern analysis of the M. truncatula EST collection indicated that DMI3 is preferentially expressed in root tissues (seven out of eight ESTs). This was confirmed by quantitative reverse transcription–polymerase chain reaction (RT-PCR), which showed that DMI3 expression in roots was 10-fold as high as that in flowers, whereas no expression was detected in leaves or stems (Fig. 4). A minor up-regulation of DMI3 was observed in nodules (Fig. 4). The expression of DMI3 was similar to, although slightly lower than, that of DMI2 (NORK), another M. truncatula gene required for nodulation and mycorrhizal infection.

Fig. 4.

Quantification of DMI3 mRNA levels in leaves, stems, flowers, roots, roots 48 hours after inoculation with S. meliloti, and root nodules. Relative transcript abundance was determined by quantitative RT-PCR and normalized against MtACTIN2 (MtACT2), which is constitutively expressed in all tissues tested.

On the basis of sequence homologies, we have identified DMI3 as a member of the small plant family of CCaMKs. DMI3 has high sequence similarity to rice, tobacco, and lily CCaMKs (73.5% identity between Medicago and lily CCaMKs). The latter protein has been studied in great detail at the biochemical level (19, 21, 2325) and shown to undergo two steps of calcium regulation. First, interaction of Ca2+ with the C-terminal EF hands results in autophosphorylation of CCaMK, which leads to increased affinity for calmodulin. In a second step, Ca2+/CaM binds to CCaMK, which allows substrate phosphorylation. In addition, autophosphorylation of CCaMK results in a time-dependent loss of enzyme activity (25). The structural features of DMI3 suggest that, like CCaMK, it might sense calcium directly through the three calcium-binding EF hands, and indirectly by binding a calcium-activated calmodulin. DMI3 might thus recognize complex calcium “signatures” such as sharp oscillations of cytoplasmic Ca2+ concentration. Proteins able to decode calcium oscillations have not yet been described in plants, but in animal cells Ca2+/calmodulin-dependent protein kinases II can be activated in a Ca2+-spike frequency-dependent manner (26). This property depends on their multimeric structure and ability to autophosphorylate (26).

Modulation of intracellular calcium concentration is associated with Nod factor signaling. Nod factors elicit two separable calcium responses in M. truncatula root hair cells: a rapid calcium influx that occurs within minutes, followed by the induction of calcium spiking about 10 min after application (27). Because these calcium responses are absent in different mutants unable to associate with rhizobia, they have been proposed to be part of the symbiotic signaling pathway. However, until now it has not been possible to rule out the possibility that calcium responses are a side branch of the main Nod factor signaling pathway (12). The identification of DMI3 as a putative calcium-sensitive effector protein now confirms the role of calcium as an integral part of the Nod factor transduction pathway. Furthermore, mutants of DMI1 and DMI2 (NORK) genes, which are unable to form nodules and mycorrhizae (3), are defective for most responses to Nod factors, including the calcium spiking response, but are still able to generate the rapid calcium influx response (12, 27). These results suggest that the calcium spiking, and not the rapid calcium influx, is the calcium signature recognized by DMI3, the role of which would be to translate this signal to various cellular components controlling nodulation and mycorrhizal infection responses. Future characterization of a calmodulin isoform that interacts with DMI3, together with the identification of targets of DMI3 kinase activity, should help to decipher the precise role of DMI3 in the Nod factor signaling pathway. In particular, although it is required for Nod factor activity, DMI3 is also proposed to be involved in negative regulation of this pathway (28). The requirement of DMI3 for mycorrhizal infection suggests that a signal transduction pathway with calcium as a second messenger controls the establishment of this fungal symbiosis. In this case, however, the nature of the signal(s) involved—putative Myc factors—is unknown, as is the type of calcium response(s) induced by this signal.

Little is known about the biological role of CCaMKs in plants. The preferential expression of the lily, tobacco, and rice CCaMKs in developing anthers and root tips (24, 29) has led to the suggestion that they could play a role in mitosis and meiosis (23). Here, using mutants in both M. truncatula and P. sativum, we provide evidence for the role of a CCaMK in the signaling pathway leading to mycorrhizal infection and in the developmental process of nodulation.

Members of the CCaMK group have been described in a number of plants, ranging from the moss Physcomitrella to higher plants, both monocotyledonous and dicotyledonous. This type of calcium-sensitive effector protein, which seems to be restricted to plants, is thus likely to be of ancient origin. This conclusion fits with the hypothesis of an ancient origin for the widespread arbuscular mycorrhizal symbiosis, which could have accompanied colonization of the land by plants more than 400 million years ago (810). We hypothesize that a CCaMK was used by mycorrhized plants to interpret a complex calcium signature elicited in response to a mycorrhizal signal. Interestingly, the CCaMK family has no known member in the sequenced genome of A. thaliana, a plant that, like most members of the Brassicaceae family, is unable to establish symbiosis with mycorrhizal fungi (20, 23).

In the legume family, genes involved in the signaling pathway of the mycorrhizal symbiosis that are homologous to DMI1, DMI2, and DMI3 could have been recruited later in evolution (∼70 million years ago), for establishing the signaling machinery of the rhizobial symbiosis. To determine whether in legumes DMI3 is interpreting the same calcium signature (a calcium spiking with defined frequency and amplitude) for both types of endosymbioses, or is recognizing two different calcium messages and informing the host cell of the presence of either the rhizobial or the mycorrhizal symbiont, is an exciting challenge for the future.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article