Introduction

In the context of global change, priority needs to be given to biological nitrogen fixation to sustain both population growth and ensure a clean and healthy environment (Steffen et al. 2015). The Haber–Bosch process, which turns the nitrogen in the air into ammonia to make chemical fertilizers, involves the production of large amounts of CO2 which largely contribute to global warming. Furthermore, nitrogen pollution caused by chemical fertilizers in the runoff from agricultural fields is a major threat to human and animal health.

The ability to reduce atmospheric nitrogen to ammonia (nitrogen fixation) is restricted to prokaryotes, some of which form mutualistic symbioses with plants (Franche et al. 2009). Two kinds of associations lead to endosymbiotic plant-bacterial associations in which different nitrogen-fixing soil eubacteria are hosted in root nodules: filamentous Gram-positive Frankia bacteria associate with about 260 plant species belonging to eight different families called actinorhizal plants (Dawson 2007), whereas Gram-negative rhizobia associate only with legumes (Fabaceae) and Parasponia (Cannabaceae) (Trinick 1973, 1979).

Legumes, Parasponia and actinorhizal plants all belong to the same clade (Fabid), suggesting that a predisposition to evolve nitrogen-fixing root nodule symbioses appeared only once during evolution (Soltis et al. 1995; Doyle 1998, 2011). Recent work using evolutionary models supports the hypothesis of predisposition suggesting that a single evolutionary innovation occurred over 100 million years ago (MYA), and was followed by several evolutionary events leading to the emergence of the different kinds of symbiotic associations (Werner et al. 2014). In contrast to nitrogen fixing symbioses, the association between arbuscular mycorrhizal (AM) fungi involves about 80 % of land plants and originated more than 400 MYA (Remy et al. 1994). Several observations including the fact that some legume mutants defective in nodulation also lack the AM interaction, and the sharing of common signaling components, suggest that the evolutionary younger nitrogen-fixing symbiosis recruited functions from the older AM symbiosis (Parniske 2008).

Symbiotic associations with rhizobia and AM involve the exchange of signals enabling mutual recognition and activation of a signaling cascade. In the case of rhizobial associations, lipo-chitooligosaccharides (LCOs), called Nod factors, are synthesized in response to the flavonoids released in plant root exudates (Oldroyd et al. 2011). AM interactions begin when the plant hormones strigolactones stimulate the branching and growth of AM fungi (Buee et al. 2000), which respond by releasing chitin oligomers (COs) and LCOS, the latter being structurally similar to Nod factors (Maillet et al. 2011; Genre et al. 2013). Genetic analyses demonstrated that rhizobia-legumes and AM associations share a single signaling pathway, or «common symbiosis pathway» (CSP) (Kistner and Parniske 2002; Oldroyd et al. 2011). This pathway contains a receptor-like kinase SYMRK/DMI2, nuclear pores, and the potassium channel proteins required for the induction of calcium oscillations (Capoen et al. 2011; Gutjahr and Parniske 2013). A nuclear calcium- and calmodulin-dependent kinase (CCaMK/DMI3) interacting with transcription factors (Cyclops/IPD3) is also part of this common pathway. The nuclear Ca2+-spiking induced by AM fungi and rhizobia is likely decoded by CCaMK triggering infection and organogenesis programs (Levy et al. 2004; Mitra et al. 2004; Miwa et al. 2006). Among the infection related genes, vapyrin (VPY), a protein consisting of a major sperm protein (MSP) domain and an ankyrin domain, was found to be involved in AM and rhizobia infection (Murray et al. 2011; Gutjahr and Parniske 2013). The vpy-1 mutant displays normal Ca2+-spiking in root hairs in response to supplied Nod factors, and the elevation of VPY transcripts upon application of Nod factors depends on NFP, DMI1, and DMI3. These findings thus show that VPY acts downstream of the common signalling pathway (Murray et al. 2011). Likewise, a set of symbiotic genes are controlled by specific early transcription factors (TFs) such as CYCLOPS/IPD3 (Singh et al. 2014), NSP1 (Smit et al. 2005), NSP2 (Kalo et al. 2005), ERN1/ERN2 (Andriankaja et al. 2007) and NIN (Schauser et al. 1999).

The lack of a genetic system for Frankia together with the difficulty involved in obtaining actinorhizal plant mutants are two major obstacles to deciphering the molecular dialogue between the symbiotic partners. However, the use of RNA interference (RNAi) to downregulate candidate genes in plants and the growing number of sequenced Frankia genomes has enabled significant advances in our knowledge of actinorhizal symbioses in the last few years (Franche and Bogusz 2012; Bogusz and Franche 2015).

The aim of this review is to highlight the most recent findings in symbiotic signaling in actinorhizal symbiosis.

The infection process

Frankia can infect actinorhizal plants in two different ways, intracellular infection via root hair penetration (Fagales) or intercellular penetration (Rosales and Cucurbitales) (Berry and Sunell 1990). Intracellular infection occurs in several actinorhizal genera (Myrica, Comptonia, Alnus and Casuarina), where the infection process proceeds through root hair deformation and penetration. In addition to Frankia, other soil bacteria, including Pseudomonas spp., deform the root hairs of Alnus sp. (Knowlton et al. 1980). It has been hypothesized that these «helper» bacteria in the rhizosphere prepare the root hair surface for penetration of the root hair cell wall by Frankia (Knowlton et al. 1980). After invagination of growing filaments of Frankia into the curled root hairs, infection proceeds intracellularly in the root cortex (Berry et al. 1986). In the host cell, Frankia is embedded within a host derived growing tubular matrix enriched in polygalacturonans known as the infection thread (Berg 1999). Upon infection, cell divisions occur in the root cortex leading to a small protuberance called the prenodule, which consists of Frankia-infected and uninfected cells (Laplaze et al. 2000) (Fig. 1). While the prenodule develops, the nodule primordium results from mitotic activity in pericycle cells located opposite a protoxylem pole (Callaham and Torrey 1977). As the nodule primordium grows, the nodule cortex becomes infected intracellularly by Frankia hyphae coming from the prenodule (Fig. 1). No root hair deformation is observed during the intercellular root invasion process (Discaria, Eleagnus, Ceanothus and Cercocarpus) (Valverde and Wall 1999). Frankia hyphae penetrate the middle lamella between adjacent root epidermal cells and progress apoplastically between cortical cells, within an electron-dense matrix. Cell divisions are induced in the root pericycle opposite a protoxylem pole, leading to the formation of the nodule primordium. Frankia hyphae infect primordium cells in the nodule lobe where intracellular colonization occurs (Wall and Berry 2008). The prenodule step has not been reported in intercellular infection. In general, mature actinorhizal nodules are multilobed structures, each nodule lobe is a modified lateral root without a root cap, including a central vascular bundle, infected cells restricted to the cortex, and a meristem at the apex.

Fig. 1
figure 1

Schematic diagram of actinorhizal signaling in the intracellular infected C. glauca. Actinorhizal plant roots release flavonoids that trigger the production of Frankia signals. The nodulation signaling pathway is triggered when as yet unknown receptors sense Frankia signals. Activation of the receptors produces oscillations of calcium concentration (calcium spiking). Two members of the CSP, a leucine-rich repeat receptor kinase (SYMRK) and a calcium calmodulin-dependent protein kinase (CCaMK), are expressed following Frankia infection. Like in legumes, the NIN gene is expressed during preinfection stages in developing root hairs and during Frankia infection. Expression of Cg12 encoding a subtilisin-like serine protease and of CgAUX1 encoding an auxin influx carrier have been shown to be specifically linked to plant cell infection by Frankia (Svistoonoff et al. 2003; Péret et al. 2007). Expression in prenodules and in mature nodule lobes has also been demonstrated, suggesting that some of the genes involved in symbiotic signaling may have other roles in nodule function. Hemoglobin (CgHb) and metallothionein (CgMT1) genes have been shown to be highly expressed in cells filled with Frankia both in prenodules and in mature nodule lobes (Gherbi et al. 1997; Laplaze et al. 2000; Obertello et al. 2007)

Full size image

Plant symbiotic signaling molecules

By acting as auxin transport inhibitors, flavonoids play many different roles in signaling and in nodule development at different stages of rhizobium symbiosis (Wasson et al. 2006). It has long been postulated that flavonoids also play a significant role in actinorhizal symbioses including chemo-attraction, proliferation of Frankia, and plant host specificity (Smolander and Sarsa 1990; Benoit and Berry 1997; Van Ghelue et al. 1997; Hughes et al. 1999; Popovici et al. 2010). However, the mechanisms underlying the movement of the non-motile Frankia are not yet known. Recently, Casuarina cunninghamiana root extracts containing flavonoids were shown to be causing alterations to the surface components of the compatible Frankia strain CcI3 in relation to infectivity (Beauchemin et al. 2012). In C. glauca, flavonoids were shown to accumulate in nodules, and a possible role in nodule compartmentation between Frankia-infected and uninfected cortical cells was suggested (Laplaze et al. 1999). When an RNAi approach was used to silence the expression of the chalcone synthase gene from C. glauca, a major reduction in nodulation was observed (Abdel-Lateif et al. 2013). Since in CgCHS1-RNAi plants, the nodulation rate was affected whereas the nodule structure was not, it was suggested that flavonoids act during the early stages of symbiotic interactions. Future work should address the chemical nature of the specific flavonoids that act as early signals, and identify their specific targets, even if no homologs of canonical nod genes have been found in the Casuarina infective Frankia strain CcI3 (Normand et al. 2007). Previous studies in legumes that form indeterminate nodules have shown that certain flavonoids are required to control auxin transport during nodule initiation (Wasson et al. 2006; Zhang et al. 2009). To study this hypothesis in actinorhizal symbiosis, auxin transport was investigated in C. glauca as described in the following paragraph.

The role of auxin in actinorhizal symbiosis

Auxins play a crucial role in different aspects of plant growth and development (Ludwig-Muller 2011). They contribute to the development of lateral roots, shoots, and flowers, and are involved in defense mechanisms against pathogens. Auxins also play an important role in different growth processes associated with symbionts. The role of auxin has been described in ectomycorrhizal infection (Sukumar et al. 2013) and in the formation of arbuscules in AM symbiosis (Etemadi et al. 2014). In legumes, auxin is involved in indeterminate nodule formation (Mathesius 2008). More recently, auxin signaling was shown to contribute to rhizobial infection of M. truncatula root hairs (Breakspear et al. 2014).

Recent data suggest that auxin is a signal that regulates the Frankia infection process and nodule organogenesis. As treatment of C. glauca roots with the auxin influx inhibitor 1-naphtoxyacetic acid (1-NOA) perturbs the nodulation process, two genes encoding putative C. glauca auxin influx carriers were cloned and characterized (Péret et al. 2007). One of them, CgAUX1, was shown to be expressed in Frankia-infected cells during nodule development, suggesting that an influx of auxin could be involved in the infection process. It was also shown that auxins accumulate in Frankia-infected cells and that accumulation is driven by coordinated cell-specific expression of influx and efflux auxin transporters (Perrine-Walker et al. 2010). Frankia strains have the ability to produce auxins in pure culture (Wheeler et al. 1984; Hammad et al. 2003). However, since Frankia mutants defective in auxin biosynthesis are not available, the exact source of the auxin, i.e. plant or bacteria, is not known (Perrine-Walker et al. 2010). Transcriptional analyses in C. glauca expanded the range of transcription factors putatively involved in nodulation (Hocher et al. 2011; Diédhiou et al. 2014). CgIAA7, one of the auxin-response factor (AEF) genes in the AUX/IAA family, was identified in C. glauca (Champion et al. 2015). CgIAA7 is expressed both in plant cells infected by Frankia and in nodule primordia in C. glauca. This transcription factor, together with the auxin response factors (ARF) family of transcription factors, controls the expression of downstream auxin-responsive genes (Peer 2013). To interfere in the signaling pathway, dominant negative and positive versions of CgIAA7 were expressed during nodulation in C. glauca. Based on the impacts of the expression of different versions of IAA7 on the establishment of the symbiosis, these authors concluded that auxin could be a negative regulator of symbiosis. In the proposed model, which resembles that of the long-distance regulation mechanism described for AM symbiosis (Meixner et al. 2005), it is hypothesized that auxin accumulation in infected plant cells triggers the synthesis of an inhibitory diffusible factor, resulting in the inhibition of nodulation (Champion et al. 2015). This suggests that high auxin concentrations control nodulation to balance the nitrogen gains with energy demands. It is also possible that at low concentrations, auxin is involved in other processes such as the cell-wall remodelling that occurs during infection, limitation of plant defense mechanisms, or control of the division of pericycle cells leading to the initiation of nodule primordia.

In contrast to C. glauca, in the actinorhizal plant Discaria trinervis, nodulation is initiated by intercellular infection. Since treatment with 1-NOA also impaired nodulation efficiency in D. trinervis, the role of auxin during the nodulation process was investigated (Imanishi et al. 2014). Immunolocalization experiments indicated that, as previously reported in C. glauca, auxin accumulated in Frankia infected cells. However, in contrast to the findings concerning CgAUX1 in C. glauca, the activity of the promoter DtAUX1 from D. trinervis was restricted to the meristem region of the nodules and was not detected during infection by Frankia. This may be explained by the fact that (i) the role of auxin differs between Frankia intercellular penetration and intracellular hairy root infection mechanisms (ii) different auxin transporters may be involved in C. glauca and D. trinervis.

New insights into the actinorhizal signal transduction pathway

As mentioned above, phylogenetic studies have shown that all plants able to enter a root nodule symbiosis belong to the same clade, suggesting that they share a predisposition for symbiosis (Soltis et al. 1995; Doyle 1998). This finding prompted investigation of the possibility that the CSP is also shared by actinorhizal symbioses. The legume homolog of the receptor-like kinase gene SYMRK/DMI2 positioned upstream from the genes involved in calcium spiking, was isolated and its function in signaling was assessed in C. glauca (Gherbi et al. 2008) and Datisca glomera (Markmann et al. 2008). Following downregulation of SYMRK/DMI2 in both C. glauca and D. glomerata, a severe reduction in nodulation by Frankia and in colonization by AM fungi was shown in the transgenic plants. Furthermore C. glauca and D. glomerata SYMRK/DMI2 genes had the ability to complement knockout Lotus japonicus symrk mutants. These results point to the existence of the CSP in actinorhizal plants.

Transcriptional analysis of C. glauca and A. glutinosa expanded the range of CSP genes reported to be involved in AM and legume-rhizobium symbioses (Hocher et al. 2011; Demina et al. 2013; Diédhiou et al. 2014). Most of the CSP genes were identified in these two actinorhizal plants, including CCaMK, the calmodulin-dependent protein kinase (CCaMK), which may decode calcium spiking in response to a Frankia signal. In C. glauca, silencing of CgCCaMK resulted in lower nodulation and AM fungal colonization (Svistoonoff et al. 2013). The function of CgCCaMK in endosymbioses was supported by full complementation of the M. truncatula dmi3 mutant for nodulation and colonization by AM fungi (Svistoonoff et al. 2013). In legumes and Parasponia andersonii, a gain-of-function mutation of CCaMK has been shown to cause spontaneous nodulation without rhizobia (Gleason et al. 2006; Tirichine et al. 2006). Interestingly, when an auto-active form of CgCCaMK lacking the auto-inhibitory/CaM domain was expressed in C. glauca and D. trinervis, nodule organogenesis was also observed in absence of Frankia (Svistoonoff et al. 2013). Since D. trinervis is an actinorhizal member of the Rosales infected by Frankia via the primitive intercellular infection pathway, one can conclude that CCaMKs are central components of actinorhiza organogenesis, regardless of the mode of infection (Svistoonoff et al. 2013). Furthermore, since legume and actinorhial nodules differ anatomically, these results raise the question of the specific mechanisms that determine the organogenesis of the two different types of nodules. Additional evidence for the conservation of CSP in actinorhizal plants was provided by the observation of Ca2+ oscillations in A. glutinosa and C. glauca root hairs in response to exsudate to Frankia (Granqvist et al. 2015; Chabaud et al. 2016).

In legumes, the transcription factor NODULE INCEPTION, NIN, is linked to a «NOD»-specific pathway downstream from the CSP. The NIN gene is expressed early during the interaction with rhizobia, in both root epidermis and nodule promordia. NIN fulfills many functions by targeting genes involved in bacterial infection, nodule organogenesis, and autoregulation of nodulation (Marsh et al. 2007; Soyano et al. 2013; Yoro et al. 2014). CgNIN, a C. glauca sequence homolog to the Lotus japonicus LjNIN gene, was recently identified (Clavijo et al. 2015). This relationship and the finding that CgNIN belongs to the same cluster that includes legume NIN genes involved in symbiosis, led to its further characterization in the control of nodulation by Frankia. CgNIN-RNAi plants showed reduced nodulation, but AM colonization was not affected. Complementation of a M. truncatula nin mutant with a CgNIN under the control of MtNIN promoter revealed the absence of infection events, but small nodule-like structures were observed. These authors concluded that CgNIN can only partly complement the MtNIN mutant by activating early cell divisions, leading to nodule organogenesis. During the symbiotic process, the CgNIN promoter was shown to drive some expression in early infected cells including root hairs (Clavijo et al. 2015). Altogether, these data show that, like in legumes, NIN plays a central role in actinorhizal symbiosis suggesting, in addition to the CSP, conservation of a « NOD » specific pathway in the host plants nodulated by rhizobia and Frankia. In addition to NIN, several transcription factors of the GRAS and NF-Y families have been shown to play a critical role in rhizobia and AM symbioses (for a review, see Rípodas et al. 2014). Their role is to trigger the developmental program of nodule organogenesis and the infection process. Recently, transcriptional and phylogenetic analyses in C. glauca and A. glutinosa revealed several other potential transcription factors homologous to those involved in legume-rhizobia and AM symbioses (Diédhiou et al. 2014). Once again, these results suggest conservation of the mechanisms regulating gene expression in these three endosymbioses and extending far beyond the CSP.

The Frankia nodulation signal

In all rhizobia -except photosynthetic Bradyrhizobia- that elicit root-and stem nodules of legumes belonging to the Aeschynomene genus (Giraud et al. 2007), the synthesis of the N-acylated chitin oligomers of Nod factors is controlled by three specific enzymes, encoded by canonical nod genes nodABC (NodA—acyl transferase, NodB—chitin deacetylase, NodC—chitin synthase) (Oldroyd et al. 2011). The differences in the structure of Nod factors made by different rhizobia species are major determinants of host specificity) (Oldroyd et al. 2011).

Compared to rhizobia, one distinguishing feature of Frankia is that individual strains may have a gradient of specificity for different plant species, host origin is not always a determining characteristic for the classification of Frankia (Swensen and Benson 2008). Several phylogenetic chronometers including RAPD PCR, 16S and 23 s-rRNA, gene sequences (ITS region) between 16S and 23S-rRNA, and nif gene sequences have been used to identify the genetic variability and relationships among the Frankia strains isolated from different host plants (reviewed in Hahn 2008). Based on the 16S database, Normand et al. (1996) grouped the Frankia genus in four clusters: cluster 1, strains effective in Alnus (Betulacea), Casuarina, Allocasuarina (Casuarinaceae) and Comptonia, Myrica and Morella (Myricaceae); cluster 2, uncultured strains present in nodules of Dryas (Rosaceae), Coriaria (Coriariaceae) and Datisca (Dastscaceae); cluster 3, strains effective in Elaeagnaceae and Gymnostoma (Casuarinaceae); and cluster 4, non-infective or non-effective strains isolated from a range of host plants. The recent determination of several Frankia genomes made it possible to look for canonical nod homologs from rhizobia. In the Frankia genomes of symbionts belonging to cluster 1 (Frankia CcI3 and Frankia ACN14a) and cluster 3 (EAN1pec), several nodC and nodB distant homologs were observed spread throughout the actinobacterial genomes. Furthermore, they did not form a SYM-island and no nodA gene was detected (Normand et al. 2007). These findings are in accordance both with those of previous studies in which no molecules ressembling Nod factors were detected in Frankia supernatants (Cérémonie et al. 1999) and with the absence of rhizobial complementation of nod mutants with Frankia DNA (Cérémonie et al. 1998). Transcriptomics and proteomics approaches were used to identify Frankia genes that are essential for the production of symbiotic factor (Alloisio et al. 2007, 2010; Bickhart and Benson 2011; Udwary et al. 2011). However, only genes that fulfill metabolic functions in symbiosis have been identified so far. The difficulty involved in isolating Frankia symbiotic genes is amplified by the fact that, in the field of plant–microbe interaction, Gram-positive bacteria have received little attention compared with the intensive study of Gram-negative bacteria (Francis et al. 2010). Whether « omics » technologies will allow the identification of candidate genes involved in the synthesis of the symbiotic signals emitted by culturable Frankia strains remains to be seen. Recently, the genome sequence of the unculturable cluster 2 Frankia datiscae Dg1 revealed the presence of canonical nod genes nodABC (Persson et al. 2011). Although Nod factors were not purified, transcripts of these genes were detected, suggesting they play a role in the Frankia/Datisca symbiosis (Persson et al. 2015). Since actinorhizal symbioses involving cluster 2 Frankia strains are thought to be the oldest actinorhizal symbioses, it was hypothesized that canonical nod genes were probably lost during the evolution of Frankia symbionts (Persson et al. 2015). The BMG5.1 strain, which also belongs to the recalcitrant cluster 2, was isolated from Coriaria myrtifolia and reported to fulfill Koch’s postulates by reinfecting its host plants (Gtari et al. 2015). The genome of this isolate does not contain the canonical nodABC genes (M. Gtari, personal communication). This result contradicts the hypothesis that most Frankia symbionts lost their nod genes, and implies that Frankia dastiscae Dg1 may have recently acquired them.

The root hair deformations observed after contact with Frankia hyphae are similar to those observed following root incubation with a filtrate from a Frankia culture. Furthermore, a correlation was found between root hair deformation and the host range of the Frankia strains tested, suggesting that a Frankia diffusible factor involved in actinorhizal symbiosis is present in Frankia culture supernatant (Van Ghelue et al. 1997; Cérémonie et al. 1999). Initial attemps to isolate specific signal molecules with bioassays based on root hair deformation with Frankia culture supernatants were not sucessful (Van Ghelue et al. 1997; Bhuvaneswari and Solheim 2000). Of great interest is the demonstration that the CgNIN promoter was activated by diffusible factors synthetized by Frankia at a very early stage (14 h) after the incubation of the root system with a supernatant of Frankia CcI3. Consistent with the expression of CgNIN at the pre-infection stage, the inhibition of root hair deformation was observed in CgNIN-RNAi plants (Clavijo et al. 2015). This result paves the way for using transgenic plants ProCgNIN driving the expression of a reporter gene as a biological test to purify the diffusible Frankia symbiotic factors (Clavijo et al. 2015). Thus, transgenic C. glauca expressing the ProCgNIN:GFP fusion were used to identify the biologically active molecules in the Frankia supernatant (Chabaud et al. 2016). The candidate molecules for Frankia signals were found to be hydrophilic and chitinase-resistant (Chabaud et al. 2016), thereby confirming the results of previous studies (e.g. Cérémonie et al. 1999). Interestingly, these diffusible biologically active molecules were found to induce calcium spiking in C. glauca, strengthening the hypothesis of their role as signaling molecules (Chabaud et al. 2016). The purification and complete characterization of Frankia signaling molecules involved in symbiotic process will indeed be of great interest.

Conclusions

Recent research in actinorhizal signaling has highlighted the function of already known legume signaling components of the CSP, such as SYMRK/DMI2 and CCaMK/DMI3, or required for nodulation, such as NIN (see Fig. 1 for summary). These findings reinforce the hypothesis that these genes are part of the predisposition to evolve nodulation in Fabids. More research is needed to understand the mechanisms that determine actinorhizal specificity. First, the isolation and characterization of the Frankia symbiotic factor and its plant receptor complexes must be pursued. The identification of CgNIN as a potential marker for pre-infection events should help identify the active fractions obtained from Frankia culture supernatant. In addition to NIN, functional analyses of the transcription factors recently identified in C. glauca and A. glutinosa are essential to understand their specific role in actinorhizal symbiosis (Diédhiou et al. 2014). Since actinorhizal nodules are developmentally related to lateral roots, it would be interesting to focus on transcription factors, as these are also known to play a role in lateral root growth. In a recent work, a new GRAS-domain containing protein, named SIN1, involved in both lateral root and nodule formation, was characterized in common bean (Battaglia et al. 2014). How actinorhizal and legume plants recruited a pre-existing lateral root genetic program in response to Frankia and rhizobia, is a key question for the future transfer of nitrogen-fixing symbiosis to major non-legume crops.