Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Perception of Symbiotic Signals at the Plasma Membrane

Genetic and molecular studies of host–microbe interaction identified a two-way signal exchange as a central mechanism for partner recognition during the symbiotic establishment. Plant-produced strigolactones are recognized and perceived by the obligate arbuscular mycorrhiza (AM) fungi, while specific flavonoids or isoflavonoids secreted from the host roots (Peters et al. 1986; Spaink et al. 1989) are recognized by the symbiotic rhizobia. It has been well established that in rhizobia, this recognition has a direct consequence, as the flavonoid-activated rhizobial NodD proteins promote transcription of bacterial nod genes involved in synthesis and secretion of lipochitin oligosaccharides (Nod factors). These molecules serve as major bacterial signals detected by the legume host (Mulligan and Long 1985; Spaink et al. 1991; Truchet et al. 1991). Chitin-derived signals, both in the form of GlcNac tetra/pentamers and acylated GlcNac oligomers, are also produced by the AM fungi and are perceived by the plants as symbiotic signals (Maillet et al. 2011; Genre et al. 2013), but the AM genes involved in synthesis of these molecules await identification.

Rhizobial Nod factors trigger physiological responses, gene expression, and cell division in susceptible legume roots during establishment of root nodule symbiosis (RNS). As a consequence, a new plant organ, the root nodule, hosting the symbiont is formed (reviewed in Oldroyd et al. 2011). The rhizobial symbiont of Lotus japonicus, Mesorhizobium loti strain R7A, produces a pentameric Nod factor with a 4-O-fucose at the reducing end, which either has an acetyl group or a proton in position 3 or 4. The non-reducing end is N-methylated and N-acylated (cis-vaccenic acid or stearic acid) and has a carbamoyl group in position 3 (Lopez-Lara et al. 1995; Bek et al. 2010). The acetylated fucosyl group is important for effective Nod factor signalling and the absence of this decoration leads to host-dependent nodulation phenotypes among different Lotus species: L. japonicus, L. filicaulis, L. corniculatus and L. burttii (Rodpothong et al. 2009).

Perception of Nod factors in legumes is mediated by receptor kinases containing LysM modules in their extracellular domains. In Lotus, the two receptor kinases perceiving the Nod factor signal, NFR1 and NFR5, are predicted to have a topology where single-pass transmembrane domains anchor the proteins to the plasma membrane exposing the LysM domains to the extracytoplasmic space, and the serine/threonine kinase to the cytoplasm (Madsen et al. 2003; Radutoiu et al. 2003; Madsen et al. 2011). Based on their domain similarity, and the fact that mutations in both Nfr1 and Nfr5 genes equally abolish Nod factor or rhizobia recognition, a heteromeric receptor complex (NFR1–NFR5) was proposed to initiate signal transduction following perception of a correctly decorated Nod factor (Radutoiu et al. 2003). Follow-up studies using bimolecular fluorescence complementation (BiFC) in Nicotiana benthamiana and leek cells revealed NFR1 and NFR5 localization at the plasma membrane and their interaction upon co-expression, supporting the originally suggested model (Madsen et al. 2011). The intracellular region of the NFR1 protein has all the subdomains of a typical kinase, and based on in vitro analyses, it has been shown to have the capacity for autophosphorylation and for NFR5 phosphorylation. On the other hand, NFR5, which lacks important kinase subdomains, and has therefore been considered a pseudokinase, failed to display kinase activity in a similar in vitro assay with myelin basic protein as a catalytic substrate (Madsen et al. 2011). The observation that a deletion of nine amino acids in the NFR5 kinase domain of the nfr5-1 mutant abolishes the symbiotic interaction (Madsen et al. 2003) indicates that the intracellular region of NFR5 is also crucial, possibly serving as an interacting domain for downstream signalling components. A plasma membrane-associated remorin, SYMREM1, which is specifically induced during RNS, has been identified both in Lotus japonicus and in Medicago truncatula, and it has been shown to interact with Nod factor receptors in both model legumes (Levebvre et al. 2010; Tóth et al. 2012). Recently, using an Y2H approach, a ROP GTPase has been identified as an NFR5 kinase interactor, and transcriptional deregulation of this gene by RNAi led to a symbiotic defective phenotype (Ke et al. 2012). With the exception of LjSYMREM1, no other interaction partners of NFR1 have been identified so far. However, the fact that particular amino acids and defined regions of the NFR1 kinase domain play a major role in downstream symbiotic signalling has been clearly shown by using point mutation constructs (Madsen et al. 2011) and domain swaps between Lotus NFR1 and Arabidopsis CERK1 (Nakagawa et al. 2011). CERK1 recognizes chitin (Miya et al. 2007; Wan et al. 2008), a microbial signal that is biochemically similar to rhizobial Nod factor. The chitin octamer is a PAMP produced by fungal pathogens, and CERK1 provides resistance in plants by activating a specific signalling cascade, leading to activation of defence genes (reviewed by Gust et al. 2012). NFR1 and CERK1 proteins show high sequence and structural similarity, especially in the intracellular region (Radutoiu et al. 2003; Miya et al. 2007; Wan et al. 2008). It is therefore unclear how specificity is achieved in the downstream signalling upon binding of symbiotic Nod factor or pathogen-derived chitin to the ectodomain. It has been shown that a chimeric NFR1-CERK1 construct containing the YAQ amino acid sequence from NFR1 in the αEF helix of CERK1 kinase allows reconstitution of symbiotic signalling in the nfr1 mutant (Nakagawa et al. 2011). This poses an important question: Does the YAQ sequence allow for the recruitment of specific downstream symbiotic partners or prevent the interaction with CERK1 partners acting in the defence pathway?

The presence of GlcNac-binding LysM modules in the extracellular regions of NFR1 and NFR5 represents a strong indication for the receptors’ ability to bind rhizobial Nod factors. However, biochemical studies showing direct binding proved to be challenging due to the recalcitrant nature of NFR proteins and the amphiphilic properties of Nod factor ligands. Recently, a breakthrough in the biochemical analysis of NFRs-Nod factor binding ability has been achieved, and detailed studies using plant-produced proteins showed the ability of both Lotus receptors to directly bind, with high affinity, the Nod factors produced by M. loti (Broghammer et al. 2012). Using two different techniques, it has been shown that full length NFR1 and NFR5 proteins have K d values for Nod factor binding in the nanomolar range. These are comparable with the ligand concentrations inducing membrane depolarization and calcium spiking in legume roots (Radutoiu et al. 2003; Miwa et al. 2006). These recent biochemical evidence complemented previous molecular studies that demonstrated the ability of NFR1 and NFR5 to mediate Nod factor perception and to ensure the specificity in the legume–rhizobia interaction. By transforming M. truncatula with the L. japonicus Nfr1 and Nfr5 genes, it was shown that the two receptors act in concert as host determinants, allowing M. truncatula, the non-host, to recognize and be infected by M. loti, the symbiont of Lotus (Radutoiu et al. 2007). Recognition of M. loti triggers initiation of nodule organogenesis in the root cortex as well as infection thread formation in the root hairs. This extended NFR1- and NFR5-mediated signal cascade is dependent on both Nod factor synthesis and structure. By using domain swap experiments and amino acid substitutions between NFRs of related Lotus species, it has been shown that a single amino acid variation, L118 to F, in the LysM2 domain of NFR5 plays a major role in discriminating M. loti and R. leguminosarum DZL Nod factors in L. filicaulis (Radutoiu et al. 2007). Interestingly, the same domain has been found in CERK1 to bind chitin (Liu et al. 2012), and homology modelling of the NFR5 LysM2 domain onto the CERK1 structure identified a possible binding groove, indicating a direct interaction with the ligand (M. Blaise, pers. communication). However, the presence of three LysM domains in the NFR5 and NFR1 receptors (Madsen et al. 2003; Radutoiu et al. 2003) suggests the involvement of more than one LysM domain in Nod factor perception. Three lines of evidence support this notion: (i) the non-nodulation phenotype caused by an amino acid substitution in the LysM1 domain of the M. truncatula homologue of NFR5 called NFP (Arrighi et al. 2006), (ii) the involvement of LysM1 of the pea SYM37 NFR1-like receptor in distinguishing ‘European’ and ‘Middle East’ Rhizobium leguminosarum bv. viciae strains (Zhukov et al. 2008) and (iii) the reported binding affinity of the CERK1 LysM2 domain to chitin (Liu et al. 2012) is significantly lower (K d  = 44, 8 μM) compared to the one found for Nod factor (Broghammer et al. 2012) in the case of NFR1 (K d  = 4.9 nM) and NFR5 (K d  = 10 nM). Surprisingly, a very high-affinity chitin-binding site (K d  = 280 pM), formed intramolecularly by the LysM1 and LysM3 domains, has been identified in the crystal structure of the C. fulvum fungal effector Ecp6 (Sanchez-Vallet et al. 2013). By comparison with the CERK1 structure, the possibility of a similar LysM1–LysM3 arrangement in the ectodomain of the receptor has been excluded (Sanchez-Vallet et al. 2013). However, functionally CERK1 acts as a dimer, and NFR1–NFR5 forms a heterodimer, therefore the possibility of a corresponding high-affinity binding groove, formed in these cases intermolecularly, represents a very attractive hypothesis.

Protein–carbohydrate recognition events are central to cell–cell communication, cellular defence mechanisms, protein trafficking, and host–microbe recognition (Sacchettini et al. 2001). In the past couple of years, an important role for chitin oligosaccharides (COs) or their derivatives, as signal molecules in plant and animal developmental processes and defence mechanisms, has emerged. In addition, carbohydrate-based microbe-associated molecular patterns (MAMPs), which are partly unidentified, are thought to be produced by microbes or plants during microbial infection (Boller and Felix 2009; Gimenez-Ibanez et al. 2009). Compared to other plant species, such as Arabidopsis or rice, the family of LysM receptor kinase in legumes, has expanded, but so far only Nfr1, Nfr5 and Lys3 (Kawaharada et al. manuscript in preparation) have an assigned function in rhizobial infection. In total, 17 loci encoding LysM proteins were identified in Lotus as having a typical receptor kinase structure (Lohmann et al. 2010). Analysis of the Lys genes location in the L. japonicus genome revealed the contribution of both tandem and segmental duplications for the expansion of this gene family. Seven Lys genes are arranged in tandem repeats (Nfr1-Lys1-Lys2, Nfr5-Lys12, Lys13-Lys14), while the occurrence of segmental duplication is supported by the syntenic regions on: (i) chromosome 4 containing Lys11, the closest paralog of Nfr5, and chromosome 2 where Nfr5-Lys12 resides and (ii) chromosome 6 where Lys6 is located and the Nfr1-Lys1-Lys2 region. A tempting hypothesis for the evolutionary diversification of this receptor family in legumes is their unique capacity to decipher various structures of microbe-derived molecules produced by an extended spectrum of interacting organisms: associative, symbiotic, and parasitic bacteria or fungi. Gene expression studies performed on Lotus identified 13 Lys genes expressed in roots and nodules, which represent attractive candidates for the suggested cortical receptor of rhizobial Nod factors (Madsen et al. 2010) and for Myc signal(s) perception. However, the identification of receptors proves to be challenging due to the complex nature or the unknown structure of these microbial signals, and the possible redundancy among receptor genes. Rhizobial genes involved in Nod factor synthesis are transcriptionally active in the cortical infection threads (Timmers et al. 1998 ; Schlaman et al. 1991), but neither the structure, nor the composition of these signals is currently known. In the case of AM, both chitin oligomers and acylated forms of chitin with different decorations are able to induce calcium spiking or to activate nodulin genes (Maillet et al. 2011; Genre et al. 2013), and as a consequence of such complex signalling from the microbe, one could expect a complex deciphering mechanism in the host as well.

In both AM and RNS, the symbiotic signal perception at the plasma membrane leads to the activation of calcium spiking and ultimately to mycorrhiza- and nodulation-associated gene expression. Both symbioses share a common genetic programme, which was co-opted from a pre-existing AM pathway during the evolution of RNS (Kistner and Parniske 2002). In Lotus japonicus, at least eight genes of this common pathway have been identified so far: SymRK, Nup85, Nup133, Nena, Castor, Pollux, CCaMK and Cyclops (Stracke et al. 2002; Mitra et al. 2004; Kistner et al. 2005; Imaizumi-Anraku et al. 2005; Kanamori et al. 2006; Saito et al. 2007; Groth et al. 2010). This poses an interesting question: How are distinct rhizobial and AM fungal stimuli integrated by one common pathway to activate specific downstream signalling events for either symbiosis?

2 Signalling from Plasma Membrane to Nucleus

The receptor-like kinase (RLK) SYMRK (NORK/DMI2) mediates symbiotic signal transduction following Nod factor perception, and symrk mutants are deficient in both RNS and AM (Endre et al. 2002; Stracke et al. 2002). SYMRK is composed of an intracellular kinase, a transmembrane domain and an extracytoplasmic region consisting of leucine-rich repeats (LRRs) and a malectin-like domain (MLD) (Antolìn-Llovera et al. 2014). At least three distinct SYMRK variants with different sizes of the extracellular region exist in angiosperms. The longest version, containing three LRRs, is needed for RNS, while shorter versions are sufficient for AM (Markmann et al. 2008). SYMRK interacts with both SYMREM1 and NFR5, potentially acting as a coreceptor of NFR5 in symbiotic signalling (Tóth et al. 2012; Antolìn-Llovera et al. 2014). The alteration of a conserved extracellular ‘GDPC’ sequence in the symrk-14 mutant affects symbiotic development in the epidermis, but not in the cortex (Kosuta et al. 2011). Dependent on the ‘GDPC’ sequence, full length SYMRK is cleaved in planta, resulting in the release of the extracellular MLD. The cleavage product lacking the MLD (SYMRK-ΔMLD) outcompetes full length SYMRK for NFR5 interaction, suggesting that the MLD interferes with NFR5 binding (Antolìn-Llovera et al. 2014). Moreover, SYMRK-ΔMLD is rapidly degraded if the LRR region is present. Removal of the LRRs stabilizes the truncated SYMRK and results in an increased formation of infection threads (Antolìn-Llovera et al. 2014). The degradation of SYMRK-ΔMLD is potentially mediated by the E3 ligase SINA, an interactor of SYMRK, whose ectopic expression was correlated with reduced SYMRK protein levels and impaired infection thread development (Den Herder et al. 2012). Taken together, these results emphasize the importance of intricate SYMRK regulation in symbiotic signalling.

The components involved in signal transduction from the PM receptors to the nuclear calcium spiking machinery have not been determined; however, screens for interaction partners of known signalling components identified candidates for missing pieces. Among the interactors of SYMRK, a 3-hydroxy-3-methylglutaryl coenzyme A reductase1 (HMGR1) coenzyme (Kevei et al. 2007) as well as the MAP kinase kinase (MAPKK) SIP2 (Chen et al. 2012) has been identified, and silencing of either of these causes nodulation defects. MAP kinase or HMGR signalling might be involved in downstream signal transduction. HMGRs are the rate controlling enzymes of the mevalonate pathway that produces sterols, isoprenoids and in particular cytokinins, which are necessary for the induction of nodule morphogenesis (reviewed in Oldroyd et al. 2011). Metabolites of the HMGR pathway could act as symbiotic secondary messengers (Kevei et al. 2007). Although MtHMGR1 is upregulated during initial stages of mycorrhizal symbiosis (Liu et al. 2003), neither HMGR1 nor SIP2 was implicated with a function in mycorrhiza colonization, indicating that additional pathways downstream of SYMRK might be involved in AM signal transduction.

3 Generation of Nuclear Calcium Spiking

Nuclear and perinuclear calcium oscillations are initiated after contact with both rhizobia and AM fungi (Kosuta et al. 2008; Chabaud et al. 2011; Sieberer et al. 2012) and can be triggered directly by the addition of rhizobial lipo-chitooligosaccharide nodulation factors (NF) (Ehrhardt et al. 1996; Miwa et al. 2006; Sieberer et al. 2009) and short-chain chito-oligomers (COs), which are present in AM fungal exudates (Genre et al. 2013). Forward and reverse genetic screens in legumes have identified proteins involved in the generation of the calcium response. The closely related ion channels CASTOR and POLLUX (M. truncatula DMI1) are required for calcium spiking, and mutant alleles are deficient for RNS and AM colonization (Ané et al. 2004; Imaizumi-Anraku et al. 2005; Charpentier et al. 2008). The channels are located in the nuclear envelope (Riely et al. 2007; Charpentier et al. 2008), and in the case of DMI1, preferentially targeted to the inner side of the nuclear membrane (Capoen et al. 2011). Electrophysiological and functional analyses revealed that the proteins are cation channels. CASTOR showed a preference for K+, and POLLUX could complement a yeast K+ import and export mutant (Charpentier et al. 2008). Symbiotic signalling in L. japonicus requires both CASTOR and POLLUX. In M. truncatula, DMI1 alone is sufficient and expression of Dmi1 in Lotus was able to complement a castor/pollux double mutant (Venkateshwaran et al. 2012). This functional difference between POLLUX and DMI1 was pinpointed to a single amino acid exchange in the putative selectivity filter region of the channels (alanine in POLLUX, serine in DMI1), which resulted in an increased mean channel opening time of DMI1 compared with CASTOR. Exchanging the filter region of POLLUX to that of DMI1 allowed for the complementation of dmi1 and castor/pollux double mutants; however, the same was not true when the change was introduced into CASTOR, which failed to rescue either dmi1 or castor/pollux but was still able to rescue a castor single mutant (Venkateshwaran et al. 2012). This finding indicates a functional or regulatory difference between CASTOR and POLLUX/DMI1 that goes beyond their K+ conductivity.

A mathematical model predicted that calcium-dependent activation of DMI1 and voltage-dependent opening of calcium channels in addition to the presence of a calcium pump are sufficient for sustained calcium oscillations (Granqvist et al. 2012). A SERCA type calcium pump, M. truncatula Calcium ATPase8 (MCA8), was localized to both the inner and the outer nuclear membrane. Silencing of Mca8 perturbed spiking and resulted in reduced mycorrhization (Capoen et al. 2011).

Several hypotheses regarding the function of CASTOR and POLLUX/DMI1 in the calcium spiking machinery have been put forward. One model assumes that CASTOR and POLLUX/DMI1 are activated by secondary messengers, causing K+ to flow into the perinuclear space. This would cause hyperpolarization of the nuclear membranes and in turn could lead to the opening of voltage-gated calcium channels (Venkateshwaran et al. 2012). In a slightly different model, it was suggested that for continued calcium spiking both DMI1 and calcium channels would need to be simultaneously activated by the binding of second messenger molecules. In this case, DMI1 (as well as CASTOR and POLLUX) would predominantly act as a counterion channel, but also initially contribute to the activation of the calcium channels by hyperpolarization of the nuclear membrane (Charpentier et al. 2013).

The role of the nuclear pore complex in symbiotic signalling remains poorly understood. Mutations in three nucleoporin genes, Nup85, Nup133 and Nena (Seh1), abolish calcium spiking and cause defects in both RNS and AM symbiosis, similar to the phenotypes observed in castor and pollux mutants (Kistner et al. 2005; Kanamori et al. 2006; Saito et al. 2007; Groth et al. 2010). Yeast and vertebrate homologues of NUP85, NUP133 and NENA are part of the nuclear pore NUP107-160 subcomplex, which is an essential component of the NPC scaffold and required for NPC assembly (Walther et al. 2003; Doucet et al. 2010). Given the apparent lack of broad pleiotropic defects in the mutants, the subcomplex likely remains at least partially intact but can no longer fulfil certain functions required in symbiotic signal transduction. Aberrations in structure or distribution of the NPCs could prevent ongoing calcium oscillations by affecting the electrophysiological properties of the nucleus. Alternatively, changes in the NPC scaffold could also interfere with nucleo-cytoplasmatic transport of symbiotic proteins or messengers (reviewed in Binder and Parniske 2013). While import and export of macromolecules through the central NPC channel does not depend on the NUP107-160 subcomplex, larger membrane proteins (>~25 kDa), which are imported from the outer to the inner nuclear membrane, have to pass through both the central channel as well as the NPC scaffold (Meinema et al. 2011). As this implies a remodelling of nucleoporin connections in order to create on opening, it is conceivable that structural defects in the nup mutants can impair proper localization of nuclear envelope membrane proteins such as CASTOR, POLLUX, MCA8, or the calcium channels and thus affect calcium spiking.

4 Decoding the Calcium Signal

As the likely primary decoder of symbiotic calcium signatures, a nuclear calcium- and calmodulin-dependent kinase (CCaMK) plays a central role in symbiotic signal transduction (reviewed in Singh and Parniske 2012). ccamk mutants do not form infection threads, nodules and arbuscules when inoculated with rhizobia or AM fungi (Levy et al. 2004; Mitra et al. 2004). A calmodulin (CaM)-binding domain and three calcium-binding EF-hands mediate CCaMKs regulation during calcium spiking (Swainsbury et al. 2012). Calcium binding induces a conformational change in the protein (Swainsbury et al. 2012) and promotes its autophosphorylation (Takezawa et al. 1996; Sathyanarayanan et al. 2000). Constitutive activation of CCaMK, caused by mutations in an autophosphorylation site (T265D, T265I), leads to spontaneous nodule development in the absence of rhizobia (Gleason et al. 2006; Tirichine et al. 2006). Negative regulation of CCaMK as a result of autophosphorylation within the calcium/CaM-binding domain (Liao et al. 2012; Routray et al. 2013) is also required for normal cortical infection and AM development (Liao et al. 2012), demonstrating an intricate modulation of CCaMK activity during symbiotic signalling. Autoactive CCaMK mutants are able to restore nodulation, and AM symbiosis in the mutants symrk, castor, pollux, nup85 and nup133 (Hayashi et al. 2010; Madsen et al. 2010). This indicates that the primary function of these genes is the activation of calcium spiking and highlights the importance of calcium signalling in symbiotic signal transduction.

It emerges that the calcium oscillations themselves may carry cell type and stage-specific information. Live cell imaging demonstrated a transition from low- to high-frequency spiking during apoplastic cell entry that was very similar for both mycorrhizal and rhizobial symbionts (Sieberer et al. 2012). Previously, calcium oscillations induced by AM fungal hyphae were described to be less regular than Nod factor-induced spiking (Kosuta et al. 2008; Chabaud et al. 2011); however, so far there has been no proof of differential decoding of AM and Nod factor-induced calcium signals by CCaMK.

CCaMK forms a complex with the nuclear protein CYCLOPS (M. truncatula IPD3) (Singh et al. 2014), which is essential for microbial infection (Messinese et al. 2007; Yano et al. 2008). CYCLOPS was revealed to be a novel type of transcriptional activator, which upon phosphorylation by CCaMK binds to a CYCLOPS responsive cis element in the Nin promoter and actives Nin gene expression (Singh et al. 2014). Phosphorylation of CYCLOPS S50 and S154 is critical for promoter binding and symbiotic development. An autoactive phosphomimetic mutant version of CYCLOPS (S50D/S154D) triggers spontaneous nodule formation independent of CCaMK, indicating that CYCLOPS acts as a master regulator of root nodule organogenesis (Singh et al. 2014).

5 Common Symbiosis Genes Involved in Microbial Accommodation

Several AM genes with putative functions in membrane trafficking are also involved in RNS. M. truncatula Vapyrin (Petunia Pam1) is required for arbuscule formation and efficient fungal entry of the root (Reddy et al. 2007; Feddermann et al. 2010; Pumplin et al. 2010; Murray et al. 2011) and deletion of the gene prevents rhizobial infection threads from reaching the cortical cell layer, resulting in an increased number of uninfected nodule primordia (Murray et al. 2011). Vapyrin encodes a protein with an N-terminal VAMP-associated protein (VAP)/major sperm protein (MSP) domain and a C-terminal ankyrin-repeat domain. Based on the domain structure and observed localization in the nucleus, cytosol and in distinct puncta in colonized cells, VAPYRIN was proposed to be involved in membrane trafficking and cellular rearrangement during symbiotic accommodation; however, this has not been experimentally verified (Pumplin et al. 2010; Murray et al. 2011). Two closely related M. truncatula vesicle-associated membrane proteins (VAMP) are involved in rhizobium–legume symbiosis and AM (Ivanov et al. 2012). Silencing of both Vamp721d and Vamp721e inhibits arbuscule and symbiosome formation and blocks bacterial release from the infection thread. Both gene products localize to small vesicles, which accumulate at bacterial release sites near symbiosome membranes and VAMP721e also accumulates at the tips of arbuscule branches, potentially at the periarbuscular membrane (Ivanov et al. 2012). While other VAMP72 proteins in A. thaliana are recruited during the interaction with biotrophic fungi (Kwon et al. 2008), VAMP721d/e are not present in the Arabidopsis genome. Considering that Arabidopsis has lost several symbiosis genes (Zhu et al. 2006), it is possible that the exocytotic pathway involving VAMP721d/e is specific to perimicrobial membrane synthesis.

6 Transcription Factors Involved in Early Symbiotic Responses

Transcriptional regulation is important to integrate signalling pathways and to coordinately regulate molecular networks. The calcium signal that is decoded by CCaMK leads to the expression of subsets of genes. Autoactive CCaMK substitutes for Nod factors to activate expression of ENOD11, which is induced by both rhizobial and mycorrhizal infection (Gleason et al. 2006; Journet et al. 2001). CCaMK activation seems to be an intracellular switch that activates transcriptional networks.

Several transcription factors that regulate symbiotic processes have been identified so far. Among them, the GRAS family transcription factors, NSP1 and NSP2 (Nodulation Signalling Pathway 1 and 2), act furthest upstream of the CCaMK-mediated pathway. They were initially identified as factors specific to RNS (Catoira et al. 2000; Oldroyd and Long 2003; Kaló et al. 2005; Smit et al. 2005; Heckmann et al. 2006; Murakami et al. 2006), and then involvement in mycorrhizal colonization was recently identified (Maillet et al. 2011; Delaux et al. 2013). nsp1 and nsp2 mutations downregulate expression of Vapyrin and Enod11 (Oldroyd and Long 2003; Hirsch et al. 2009; Murray et al. 2011). This is consistent with the idea that Nsp1 and Nsp2 are common to both AM and RNS. In nodulation processes, nsp1 and nsp2 mutants exhibit phenotypes similar to loss-of-function ccamk mutants in regard to symbiotic root hair responses as well as nodule formation. Root hairs of these mutants are deformed in response to Nod factors, but do not display root hair curling caused by rhizobial infection (Catoira et al. 2000; Oldroyd and Long 2003; Heckmann et al. 2006; Murakami et al. 2006). Autoactive CCaMK mutants do not rescue phenotypes of nsp1 and nsp2, unlike those of common SYM mutants defective in the nuclear calcium spiking (Hayashi et al. 2010; Madsen et al. 2010). These results suggest that a site in the nodulation processes where NSP1 and NSP2 act is close to that of CCaMK and that the GRAS family proteins are required for the CCaMK-mediated pathway. How activities of the GRAS proteins are regulated is an important issue for understanding symbiotic signal transduction.

NSP1 and NSP2 form a heterodimer and bind to the promoters of the transcription factors M. truncatula Ern1 (ERF Required for Nodulation1) and Nin (NODULE INCEPTION) in vitro as well as that of Enod11 in vitro and in vivo (Hirsch et al. 2009). Expression of Ern1 and Nin is induced by rhizobial infection depending on NSP1 and NSP2 (Murakami et al. 2006; Marsh et al. 2007; Hirsch et al. 2009; Cerri et al. 2012). NSP2 also interacts with an AM-specific GRAS family protein, RAM1 (Required for Arbuscular Mycorrhization1) (Gobbato et al. 2012), which directly targets Ram2 expression. Both Ram genes were identified during a forward genetic screen that aimed to identify loci specifically involved in mycorrhizal signalling (Gobbato et al. 2012; Wang et al. 2012). Multiple dimerization of the GRAS family transcription factors is involved in the regulation of symbiotic processes. The heterodimerization between NSP1 and NSP2 seems to be important for NSP2 function, because an NSP2 derivative with an amino acid substitution in the domain responsible for binding with NSP1 resulted in the reduction of nodulation efficiency (Hirsch et al. 2009).

ERN1 coordinately regulates Enod11 expression with the NSP1–NSP2 complex by targeting a cis-acting element different from those for NSP1–NSP2 complex. ERN1 is required to activate Nod factor-elicited Enod11 expression during early pre-infection, while NSP1–NSP2 mediates Enod11 expression during subsequent rhizobial infection (Cerri et al. 2012). NIN is a RWP-RK domain-containing transcription factor specific to and essential for RNS (Schauser et al. 1999). Root hairs of nin mutants are deformed and excessively curled in response to rhizobial infection, and failed to initiate infection thread development. This indicates that NIN acts downstream of CCaMK and the GRAS transcription factors to regulate symbiotic root hair responses (Marsh et al. 2007). Nin is activated by cytokinin through a L. japonicus cytokinin receptor, LHK1 (Tirichine et al. 2007). Gain-of-function LHK1 spontaneously induces nodules without rhizobial infection (Tirichine et al. 2007). Ectopic expression of Nin also induces cortical cell divisions in the absence of rhizobia (Soyano et al. 2013). NIN regulates cortical cell divisions downstream of the cytokinin signalling.

LjNF-YA1 and LjNF-YB1 have been identified as direct targets of NIN (Soyano et al. 2013). They are involved in stimulation of cell division. They encode different subunits of a CCAAT-box binding heterotrimeric complex. Knock-down of LjNF-YA1 prevents the nodule formation. Co-overexpression of the two Lotus NF-Y genes stimulates cells division in lateral root primordia as well as cortical cell division. NF-Y regulates expression of its target genes by influencing histone modification and requires an additional transcriptional activator to efficiently activate transcription. NIN may also regulate other transcription factors that act together with the NF-Y to induce cortical cell division. Arabidopsis NIN-like proteins (NLPs) play a central role in the transcriptional regulation of nitrate-responsive genes and target nitrate-responsive elements (Konishi and Yanagisawa 2013), which are almost identical to NIN-binding nucleotide sequences. Nitrate is known as an inhibitor of nodulation. There may be a linkage between nodulation control and nitrate-response pathways.