Nitrogen-fixing root nodulation is confined to four plant orders, including > 14 000 Leguminosae, one nonlegume genus Parasponia and c. 200 actinorhizal species that form symbioses with rhizobia and Frankia bacterial species, respectively. Flavonoids have been identified as plant signals and developmental regulators for nodulation in legumes and have long been hypothesized to play a critical role during actinorhizal nodulation. However, direct evidence of their involvement in actinorhizal symbiosis is lacking.
Here, we used RNA interference to silence chalcone synthase, which is involved in the first committed step of the flavonoid biosynthetic pathway, in the actinorhizal tropical tree Casuarina glauca. Transformed flavonoid-deficient hairy roots were generated and used to study flavonoid accumulation and further nodulation.
Knockdown of chalcone synthase expression reduced the level of specific flavonoids and resulted in severely impaired nodulation. Nodule formation was rescued by supplementing the plants with naringenin, which is an upstream intermediate in flavonoid biosynthesis.
Our results provide, for the first time, direct evidence of an important role for flavonoids during the early stages of actinorhizal nodulation.
Actinorhizal plant symbioses with actinobacteria from the genus Frankia are widespread, are ecologically significant and, in a variety of ecosystems, serve as a pioneer system in early successional plant community development. Unlike the host plants involved in the rhizobia–legume (Fabaceae) symbiosis, which, with the exception of Parasponia, belong to a single family, actinorhizal plants constitute a diverse group of > 200 species from eight different families. Legume and actinorhizal nodules differ in their ontogeny and structure (Pawlowski & Bisseling, 1996). However, phylogenetic studies have shown that all plants able to enter a root nodule symbiosis belong to the same clade (Eurosid I), thus suggesting that they share a predisposition for symbiosis (Soltis et al., 1995; Doyle, 1998, 2011; Bell et al., 2010). The molecular basis of this predisposition is not yet known. Nitrogen-fixing root nodule symbioses also share several key features with arbuscular mycorrhizas (AMs), which are formed by the majority of land plants with fungi belonging to the phylum Glomeromycota (Schüssler et al., 2001). This led to the hypothesis that pre-existing AM genes were recruited during the evolution of root nodule symbiosis (Parniske, 2008). Recent work by Gherbi et al. (2008b) and Markmann et al. (2008) demonstrated that the symbiosis receptor-like kinase gene SymRK is a linchpin in the symbioses of Casuarina glauca and Datisca glomerata, playing a role similar to that played in legume nodulation and mycorrhiza formation. Moreover, Hocher et al. (2011) suggested that the nodulation predisposition is related to the presence of a similar nodule organogenesis pathway in all root nodule symbioses.
The establishment of both root nodule and AM symbioses starts with the exchange of symbiotic signals in the soil between the plant and the symbionts. This molecular dialog involves key signaling molecules that determine the specific recognition of the symbiotic partners (reviewed in Gough & Cullimore, 2011). Key factors have been characterized regarding the symbiotic signals produced by bacteria and fungi. In the legume–rhizobia interaction, rhizobial genomes contain common nodulation nodABC genes encoding products that direct the synthesis of substituted lipo-chito-oligosaccharides (LCOs), called Nod factors, that interact with dedicated receptors to trigger the symbiotic program in Fabaceae (Lerouge et al., 1990). Like rhizobia, the fungus Rhizophagus irregularis secretes LCOs that stimulate the formation of AM in plants (Maillet et al., 2011). In the actinorhizal symbioses involving Alnus glutinosa, it is known that Frankia enters the root using a deformed root hair mechanism, possibly mediated by a diffusible factor (Van Ghelue et al., 1997). A chemical approach (Cérémonie et al., 1999) enabled partial characterization of this factor, but its complete structure and function are still not known. Indeed, neither Van Ghelue et al. (1997) nor Cérémonie et al. (1999) linked root hair deformation to nodule induction. To date, complete genome sequencing of Frankia strains infective on Alnus (ACN14) and Casuarina (CcI3) has confirmed the absence of the canonical rhizobia nodABC genes (Normand et al., 2007a) in these strains. We can therefore hypothesize that either a different mechanism is required by Frankia strains to initiate nitrogen-fixing symbioses or actinobacteria use different enzymes to synthesize LCOs.
On the plant side, in the legume–rhizobia symbioses, specific flavonoids secreted into the rhizosphere have been shown to attract compatible rhizobia to plant roots and also to interact with the NodD protein of rhizobia to activate transcription of other nod genes responsible for the synthesis of Nod factors (Hassan & Mathesius, 2012). Moreover, recent studies using RNA interference (RNAi)-mediated silencing of key flavonoid biosynthetic genes revealed that flavonoids also regulate auxin transport and initiate nodule primordial cell division (Wasson et al., 2006; Zhang et al., 2009). Although the exact role of flavonoids during the establishment of AM remains unclear, data suggest that they play a regulatory role in the first stages of AM colonization of plant roots and in a later stage of the AM association (Steinkellner et al., 2007; Hassan & Mathesius, 2012).
Little is known about plant signaling molecules in actinorhizal symbioses. Some evidence of chemo-attraction and proliferation of Frankia bacteria has been reported in the rhizosphere of several actinorhizal species (Vessey et al., 2005) and flavonoids have also been suggested to act as plant signals that activate the production of the Frankia root hair deforming factor (Prin & Rougier, 1987; Van Ghelue et al., 1997). However, it is unclear whether flavonoids act as early signals exchanged between the plant and Frankia. Benoit & Berry (1997) showed that flavonoid-containing preparations from seed washes of red alder (Alnus rubra) enhanced nodulation by Frankia in this species. These results were reinforced by the observation that flavonols (quercetin and kaempferol) contained in black alder (A. glutinosa) root exudates are able to enhance the level of nodulation (Hughes et al., 1999) and that root hair curling is increased by exposure of Frankia to A. glutinosa root filtrate (Prin & Rougier, 1987; Van Ghelue et al., 1997). Recent work by Popovici et al. (2010) on the Myrica gale–Frankia actinorhizal symbiosis showed that specific flavonoids isolated from M. gale fruit induce phenotypic (growth and nitrogen fixation) and transcriptional modifications in Frankia strains, in congruence with the symbiotic compatibility of the strains. These authors have thus provided evidence for the involvement of specific flavonoids in determining symbiotic specificity in the M. gale–Frankia symbiosis. These data are consistent with the findings of a recent study showing that Casuarina root exudates containing flavonoids alter the physiology, surface properties, and plant infectivity of Frankia strain CcI3 (Beauchemin et al., 2012). Moreover, changes in the surface of Frankia were correlated with effects on the plant–microbe interaction. The authors showed that pre-exposure to plant root exudates allowed Frankia CcI3 to nodulate the host plant earlier than untreated cells, suggesting that treatment is beneficial for the infection and nodulation process (Beauchemin et al., 2012).
In addition to their potential role in signaling, flavonoids are also thought to be involved in actinorhiza development and functioning (Laplaze et al., 1999). Histological analysis revealed a cell-specific accumulation of phenolics in C. glauca nodule lobes, creating a compartmentation in the nodule cortex. Biochemical analyses showed that these phenolic compounds belong to the flavan class of flavonoids. The function of this compartmentation is not well understood, but clearly some signal exchange with the endophyte is needed for its development. Laplaze et al. (1999) hypothesized that cell-specific flavan biosynthesis and accumulation delimit cortical compartments containing Frankia-infected cells and might restrict endophyte invasion. In parallel, Kim et al. (2003, 2007) found that the gene coding for chalcone isomerase (CHI) in Elaeagnus umbellatae was highly expressed in root nodules, with levels increasing during nodule development. A study based on a C. glauca expressed sequence tag (EST) database and microarrays enabled identification of different genes coding for enzymes involved in flavonoid biosynthesis (Hocher et al., 2006). The results of a kinetic study of the effect of inoculation with Frankia to characterize temporal expression patterns of these genes are consistent with the involvement of flavonoids in actinorhizal symbiosis (Auguy et al., 2011).
Taken together, the data collected on actinorhizal symbioses provide indirect evidence that flavonoids are involved in nodulation. In order to gather genetic evidence of an important role of flavonoids in actinorhizal symbioses, we used an RNAi strategy to silence expression of the chalcone synthase (CHS) gene, which encodes the enzyme responsible for the first committed step in flavonoid biosynthesis, in C. glauca hairy roots and studied the subsequent effects of gene silencing on C. glauca nodulation.
Materials and Methods
Casuarina glauca Sieb. ex Spreng. seeds were provided by the Australian Tree Seed Center (CSIRO, Acton, Australia) and grown as described in Franche et al. (1997). Agrobacterium rhizogenes strain ARqua1 (Boisson-Dernier et al., 2001) was used for hairy-root transformation of C. glauca. Frankia strain CcI3 (Normand et al., 2007a) was used to nodulate C. glauca as described in Svistoonoff et al. (2003).
Construction of the plant transformation vector
A Casuarina glauca chalcone synthase cDNA clone (CgCHS1) was isolated by Laplaze et al. (1999). The RNAi vector used to silence CgCHS1 transcripts was constructed as follows: a 522-bp coding region towards the 5′ end of CgCHS1 showing 89% similarity with the CHS sequence of Medicago truncatula (MtCHS) (Wasson et al., 2006) was amplified by PCR using forward (5′-ggggacaagtttgtacaaaaaagcaggctCAAGGCTCAAAGGGCTGAG-3′) and reverse (5′-ggggaccactttgtacaagaaagctgggtCTGTTGTTCTCGGCAAGGTC-3′) primers (Supporting Information Fig. S1). The PCR product was cloned using Gateway recombination technology (Invitrogen, Life Technologies, St Aubin, France) in pKellsgate vector. This vector is a modified version of the pHKN29 binary vector (Kumagai & Kouchi, 2003) and was obtained in our laboratory (H. Gherbi & S. Svistoonoff, unpublished) by introducing the Gateway RNAi cassette from the pHellsgate12 vector at SpeI/SacI sites (Helliwell & Waterhouse, 2003). This construct was verified by restriction digests and sequenced before being introduced into the A. rhizogenes ARqua1 strain by electroporation.
Generation of composite C. glauca
Alnus rhizogenes-mediated transformation of 45-d-old C. glauca seedlings was performed following previously described methods (Gherbi et al., 2008a,b). After 1 month, transgenic roots expressing GFP fluorescence were positively selected using a GFP filter under UV light on a stereomicroscope (MZFLIII; Leica, Wetzlar, Germany) and nontransgenic roots were removed. No interference with flavonoid fluorescence was observed, regardless of the transgenic line.
Real-time RT-PCR analysis of CgCHS1 expression
As described in Gherbi et al. (2008a,b), we selected six RNAi and three transgenic control plants showing a high level of GFP fluorescence in the root system. Three nontransgenic plants were also selected. Total RNA was extracted from c. 100 mg of root tissue from each selected plant using the RNEasy Plant MiniKit (Qiagen, Courtaboeuf, France) and quantified using the NanoNrop ND-1000 spectrophotometer (ThermoScientific, Wilmington, DE, USA). RNA was treated with DNAse I using the turbo DNA free kit (Ambion, Life Technologies) to remove contaminating genomic DNA. Using SuperScriptIII H− reverse transcriptase (Invitrogen), 100 ng of RNA per sample was reverse-transcribed. Quantitative RT-PCR (qRT-PCR) was then performed as described in Hocher et al. (2006) with the following primers: qCgCHS1-F: GAACTGCCACTCCTCCCAACTG and qCgCHS1-R: GCTCTGTCTTGTGCTCACTGTTTG. The transcript abundance of the C. glauca ubiquitin gene in each RNA preparation was used as an internal standard (Hocher et al., 2006).
For nodulation studies, composite plants were transferred into tubes (Gibson, 1963) containing a modified Hoagland solution (Santi et al., 2003). After 1 month, the plants were inoculated with a 2-ml aliquot of a dense suspension (25 μg protein ml−1) of Frankia CcI3. Tubes were kept lying horizontally for 1 h before being filled with nitrogen-free Hoagland's solution at pH 6.8. Plants were then cultivated in a growth chamber under normal conditions (25°C, an average 45% humidity and a 16-h photoperiod, Photosynthetically Active Radiation (PAR) of 150 μmol m−2 s−1).
To test the effect of flavonoid supplementation on nodulation, each composite plant was grown in hydroponic conditions in 500-ml pots containing a modified Broughton and Dillworth (BD) medium supplemented with nitrogen for 3 wk as described by Svistoonoff et al. (2010); inoculation with Frankia CcI3 was performed in BD without nitrogen. At the time of inoculation, 10 μM naringenin (Sigma-Aldrich) was added.
Ten C. glauca nodules sampled from five different composite control and CgCHS1-RNAi plants were fixed and dehydrated as described in Svistoonoff et al. (2003). Samples were embedded in Technovit 7100 resin (Heraeus Kulzer, Hanau, Germany) as recommended by the manufacturer. Thin 4-μm sections were cut with an HM355S microtome (MicroM; Thermo Fisher Scientific, Walldorf, Germany). Sections were then colored with 0.01% Toluidine Blue and mounted in Clearium Mounting (Surgipath; Leica). Samples were observed under a DMRB microscope (Leica).
To study flavonoid localization, 5-mm root segments and nodules showing a high level of GFP fluorescence were selected and embedded in 3% DNA grade agarose (Sigma-Aldrich). The roots and nodules were then cross-sectioned to 170 μm thickness on a HM650V vibratome (MicroM) and flavonoids were stained by incubating fresh sections for 5 min in diphenylboric acid-2-aminoethyl ester (DPBA; Sigma-Aldrich) (0.25% w/v in water supplemented with 0.1% triton v/v) according to the method of Sheahan & Rechnitz (1992), modified by Peer et al. (2001). The sections were then transferred to glass slides, covered and immediately observed under a DM6000B microscope (Leica). Images were taken using UV illumination and a 4′,6′-diamidino-2-phenylindole (DAPI) filter (excitation at 380 nm maximum, 340 nm minimum, and emission at 800 nm maximum, 425 nm minimum). Control sections of nontransgenic roots were colored and observed under the same settings to exclude the possibility that the fluorescence detected in transgenic sections was based on GFP.
Analysis of flavonols, anthocyanins and proanthocyanidins
For use as reagents and solvents, high-performance liquid chromatography (HPLC) grade acetonitrile and methanol were purchased from Merck (Darmstadt, Germany); acetone, formic acid, trifluoroacetic acid and hydrochloric acid (37%) were purchased from Prolabo (Normapur™ grade) (VWR, Fontenay-sous-Bois, France). All flavan-3-ol monomers ((+)-catechin, (−)-epicatechin, (−)-epicatechin-3-O-gallate, and (−)-epigallocatechin), phloroglucinol and l-ascorbic acid were purchased from Sigma-Aldrich. Each phloroglucinol derivative was obtained and characterized as described in Souquet et al. (2004) and Mané et al. (2007). Malvidin-3-O-glucoside and quercetin-3-O-glucoside were purchased from Extrasynthèse (Genay, France).
Forty-two-day-old hairy roots of control and CgCHS1-RNAi C. glauca plants showing GFP fluorescence were collected, rapidly frozen in liquid nitrogen and stored at −80°C. For flavonoid extraction, root samples were ground to a fine powder using a Freezer/Mill 6770 (SPEX Forensics, Edison, NJ, USA). Sample powder (250 mg) was mixed for 1 h with 800 μl of the extraction solution (acetone/water/trifluoroacetic acid (70 : 30 : 0.05, v/v/v)) according to the protocol described by Olle et al. (2011). Fifty microliters of a p-hydroxy methyl ester solution (3 g l−1 in methanol) was added to each sample as an internal standard. Immediately after extraction and centrifugation, each supernatant was divided into two parts. The first part was used for the analysis of flavonols and anthocyanins, and the second part for the analysis of proanthocyanidins (condensed tannins). Phloroglucinolysis was performed on the second part according to the protocol described by Verries et al. (2008). We made a five-fold concentrate to ensure sufficient flavonoids for further Ultra Performance Liquid Chromatography (UPLC) quantification. Both fractions were finally centrifuged and filtered through 0.45 μm mesh before injection into the UPLC system.
Chromatographic runs were performed on a Waters Acquity UPLC-DAD system (Waters, Milford, MA, USA), on an Acquity BEH C18 column (10 mm length × 1 mm internal diameter, 1.7 μm particle size; Waters) at 35°C. The mobile phase consisted of water/formic acid (99/1, v/v) (solvent A) and methanol/formic acid (99/1, v/v) (solvent B). The flow rate was 0.08 ml min−1. The elution program was as follows: isocratic with 2% B (1 min), 2–30% B (1–10 min), isocratic with 30% B (10–12 min), 30–75% B (12–25 min), 75–90% B (25–30 min), and isocratic with 90% B (30–35 min).
Electrospray Ionization Tandem Mass Spectrometry (ESI-MS/MS) analyses were performed with a Bruker Daltonics Amazon (Bremen, Germany) mass spectrometer equipped with an electrospray source and an ion trap mass analyzer. The spectrometer was operated in the positive ion mode (capillary voltage, 2.5 kV; end plate offset, −500 V; temperature, 200°C; nebulizer gas, 10 psi and dry gas, 5 l min−1). Collision energy for fragmentation used for MS2 experiments was set at 1. Identifications were achieved on the basis of the molecular ion mass, fragmentation, UV-visible spectra and relative retention times compared to those of standards. Concentrations were determined from external standard curves calculated from standards. (+)-catechin, (–)-epicatechin, (–)-epicatechin-3-O-gallate and (–)-epigallocatechin (280 nm) were used for quantification of these 4 monomers and their phloroglucinol derivatives; quercetin 3-O-glucuronide (360 nm) was used for both quercetin-3-O-glucoside and quercetin 3-O-glururonide quantification; malvidin 3-O-glucoside (520 nm) was used for cyanidin 3-O-glucoside quantification. Proanthocyanidin polymer length, estimated using the average degree of polymerization (DP), was calculated as the molar ratio of the sum of all proanthocyanidin units to the sum of terminal units.
Acetylene reduction activity (ARA) assay
Nitrogenase activity was determined using the acetylene reduction activity (ARA) assay on CgCHS1-RNAi composite plants, and plants transformed with the empty transformation vector were used as negative controls. For each construct, five test tubes containing one composite plant were tested for ARA according to Meilhoc et al. (2010).
Unless otherwise specified, data are expressed as the arithmetic average ± standard deviation of three biological replicates. ANOVA (one and two factors), Newman–Keul, Student, Kuskal–Wallis and Conover–Inman tests were carried out with a significance threshold set at 0.05.
Sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: CgCHS1, AJ132323; MtCHS, AJ277211.1; CgUBI (Ubiquitin), CO037049.
Silencing of the CHS gene altered flavonoid profiles in transgenic C. glauca roots
We used RNAi to silence the expression of chalcone synthase (CgCHS1) in hairy roots of C. glauca. In C. glauca, chalcone synthase is encoded by at least two genes, as shown by Southern blot (Laplaze et al., 1999). To silence the entire CHS gene family, we designed one RNAi construct targeting the most highly conserved region of the CHS genes from C. glauca and several additional plant chalcone synthase sequences (Wasson et al., 2006). Transgenic controls consisting of hairy-root plants transformed with the pHKN29 vector containing the GFP reporter gene but lacking the RNAi cassette were also generated. A total of 184 RNAi composite plants and 54 transgenic control composite plants were obtained in three independent experiments. Consistent with results obtained in our previous studies (Gherbi et al., 2008a,b), the fluorescence was obtained in an average of 20% of the hairy roots. To test the efficiency of CHS gene knockdown in RNAi roots, as described by Gherbi et al. (2008b), CgCHS1 expression was measured by qRT-PCR in six CgCHS1-RNAi roots sampled on six different chimeric plants showing a high level of GFP fluorescence. Primers designed in the conserved region of the CHS gene were used. An 80–97% reduction in CgCHS1 mRNA levels was observed in RNAi roots compared with transgenic control pHKN29 roots (Fig. S2).
To confirm that the silencing of CgCHS1 gene expression led to a reduction of flavonoid metabolites inside the root tissues, we used two assays: fluorescence microscopy and UPLC coupled to ESI-MS/MS. We carried out fluorescence microscopy using DPBA, which is used to enhance the fluorescence of flavonoids in plant tissue (Sheahan & Rechnitz, 1992; Peer et al., 2001). Analyses were conducted on roots sampled on 10 control pHKN29 and 10 CHS-silenced hairy-root plants. Intense fluorescence was observed across the root section, with the strongest signal in the pericycle and protoxylem in transgenic control plants (Fig. 1b), whereas CgCHS1-silenced hairy roots showed a low level of fluorescence in these cells and almost no fluorescence in the cortical cells compared with controls (Fig. 1d). Control conditions without DPBA staining are shown in Fig. 1(a,c).
For root flavonoid detection and quantification, three sets of 2-month-old nonnodulated hairy roots from the C. glauca transgenic control and CHS-silenced hairy roots with positive GFP fluorescence were analyzed with UPLC/ESI-MS/MS. As shown in Fig. 2(a,b), the presence of compounds belonging to two flavonoid families, flavonols (360 nm) and anthocyanins (520 nm), were detected in transgenic hairy-root controls. Peaks eluted at 15.9 and 16.1 min (Fig. 2a) were identified on the basis of their UV-visible spectra (λmax 256 and 354 nm, respectively) and MS spectra [M+H]+ = 479 and [M+H]+ = 465, with a fragment ion at [M+H]+ = 303 mAU characteristic of two flavonols, quercetin 3-O-glucuronide and quercetin 3-O-glucoside, respectively. The peak eluting at 11.2 min (Fig. 2b) was identified as cyanidin 3-O-glucoside on the basis of its UV spectrum (λmax 517 nm) and its parent and fragment ions at [M+H]+ = 449 and 287, respectively. The same profiles were observed in CgCHS1-RNAi roots as in controls (Fig. 2a–c). Table 1 shows that a significant reduction in flavonol levels (three-fold) in CgCHS1-RNAi roots compared with transgenic control pHKN29 roots was found. Levels of cyanidin 3-O-glucoside also decreased (three-fold) in CgCHS1-RNAi roots, although they did not differ significantly from those in the control (possibly because levels were close to the detection limit). The most abundant group of flavonoids detected in C. glauca roots corresponded to proanthocyanidins (condensed tannins), which were analyzed after phloroglucinolysis (Fig. 2c; Table 1). Analysis of retention time, UV-visible absorption spectra and mass spectra enabled identification of different constitutive units of proanthocyanidins. Peaks at 8.4, 10.7 and 12.4 min were identified as catechin, epicatechin ([M+H]+ = 291) and epicatechin 3-O-gallate ([M+H]+ = 443). Mass fragmentation enabled identification of the four peaks eluted at 3.5, 5.8, 6.4 and 8.6 min as phloroglucinol derivatives, with the typical loss of 126 atomic units of mass (uma) (loss of C6O3H6 of phloroglucinol). The peak at 3.5 min, showing a typical UV spectrum of epigallocatechin, a molecular ion at [M+H]+ = 431 and a fragment ion at m/z = 305, was identified as an epigallocatechin phloroglucinol derivative. The peaks eluted at 5.8 and 6.4 min, showing parent and fragment ions at [M+H]+ = 291 and 287, respectively, and UV spectra of catechin or epicatechin, were identified as catechin and epicatechin phloroglucinol derivatives. Lastly, the peak eluted at 8.6 min was identified as the phloroglucinol derivative of epicatechin-3-O-gallate on the basis of its mass signals at [M+H]+ = 567 and 441, and its UV spectrum (Fig. 2c). The total proanthocyanidin level was significantly reduced by more than four-fold in CgCHS1 silenced plants compared with transgenic control roots (Table 1). However, the DP of proanthocyanidins and the proportions of the various constitutive units were similar in transgenic controls and CgCHS1-silenced plants, indicating that the knockdown of the chalcone synthase gene affects the level of condensed tannins, but not their composition.
Table 1. Levels of flavonoid compounds measured and identified by Ultra Performance Liquid Chromatography (UPLC) coupled to Electrospray Ionization Tandem Mass Spectrometry (ESI-MS/MS) in transgenic control pHKN29 and chalcone synthase (CgCHS1) RNA interference (RNAi) roots of Casuarina glauca
Concentration (mg g−1 FW)
Control pHKN29 plants
Values ± SE are the mean of three replicates. The Kruskal–Wallis test is significant for P =0.0001 at the 5% level. A Conover–Iman test by pair was then applied to compare values for transgenic control versus RNAi roots. *, values are significantly different at the 5% level.
As expected, our results confirmed that a reduction in CgCHS1 expression led to a marked decrease in all flavonoid levels in CgCHS1-RNAi roots of C. glauca.
CHS-silenced hairy roots led to reduced nodulation
To investigate the ability of CgCHS1-silenced hairy roots to form nodules, selected transgenic control and CgCHS1-silenced hairy-root plants were inoculated with the Frankia CcI3 strain and examined three times a week for 3 months. A total of 126 RNAi composite plants and 54 transgenic control composite plants showing high GFP fluorescence in hairy roots were analyzed in three independent experiments. Five weeks after inoculation, plants transformed with the control vector began to develop nodules similar in size and shape to those produced on nontransgenic roots (Fig. 3a). Silencing of CgCHS1 resulted in a 1-month delay in nodulation in RNAi plants (Fig. 4). Remarkably, 84 d after inoculation, 56% of control hairy roots were nodulated, while only 30% of CgCHS1-silenced hairy roots had nodules (Fig. 4). In three independent experiments each performed on at least 15 plants, transgenic control roots produced an average of 10 nodules per plant, whereas CgCHS1-RNAi roots only produced an average of three nodules per plant (Table S1). Nodulated CgCHS1-RNAi roots developed a gradient of nodule phenotypes ranging from small uni-lobed to large multi-lobed. However, no major alteration in shape or size was found in mature nodule lobes. CgCHS1-silenced nodules were mostly white, while control hairy-root nodules were yellowish to reddish in color (Fig. 3a,b). Histological analysis of 10 white nodules of CgCHS1-RNAi roots revealed the absence of phenolic compounds, contrasting with cell-specific accumulation of these products in transgenic control nodules (Fig. 3c–h). These phenolic compounds previously described as flavonoids accumulated mainly in endodermal cells and in a few layers of cortical cells below the periderm (Laplaze et al., 1999). Furthermore, compared with yellowish nonnodulated control plants, CgCHS1-RNAi nodulated plants displayed green leaves, suggesting that effective symbiosis with Frankia occurred (data not shown). This was confirmed by testing the ability of CgCHS1-RNAi nodules to fix nitrogen via ARA assays. Indeed, CgCHS1-RNAi nodules exhibited an ARA rate comparable to that observed for transgenic control nodules (Fig. S3).
Naringenin supplementation can restore nodulation in CgCHS1-RNAi roots
To confirm the link between alteration of nodulation and flavonoid reduction in CgCHS1-RNAi roots, we attempted to restore nodulation by supplementation of the culture medium with naringenin, the central precursor of most flavonoids. Preliminary experiments conducted on nontransgenic plants revealed that addition of 10 μM naringenin at the time of Frankia inoculation improved nodulation. In that case, we observed more rapid appearance of nodules (c. 1 wk earlier) and an increase in nodule number (10%) (data not shown). Thus, we used these experimental conditions to complement transgenic CgCHS1-RNAi plants. Fig. 5(a) shows the effect of application of naringenin on transgenic control and CgCHS1-RNAi root nodulation. We observed that, in all cases, nodulation started 21 d after inoculation with Frankia. Thus, it seems that experimental plant growth conditions used for this experiment shortened the time required for nodulation even in untreated plants. However, flavonoid-deficient roots treated with naringenin increased their nodulation levels up to that of transgenic controls. Addition of naringenin increased significantly the average number of nodules on CgCHS1-RNAi roots (up to 10) compared with naringenin-untreated RNAi roots (average of three nodules per plant) (Table S2). The number of nodules on CgCHS1-RNAi roots treated with naringenin was thus similar to that obtained on vector control roots, which showed a rate of 11 nodules per plant on average. CgCHS1-silenced nodules treated with naringenin were yellowish in color (data not shown). In addition, longitudinal sections of mature nodules developed on CgCHS1-RNAi roots supplemented with naringenin showed the presence of larger amounts of phenolic compounds (Fig. 5c,d) than in untreated roots (Fig. 5b).
Taken together, our results indicate that a reduction in CgCHS1 expression results in impairment of actinorhiza formation related to a decrease in the level of flavonoid.
Numerous reports have suggested that flavonoids are involved in actinorhizal symbioses but, to date, there has been no genetic evidence to support this hypothesis (Abdel-Lateif et al., 2012). In this work, by silencing C. glauca CgCHS1, we provide direct genetic evidence that these compounds play a role in actinorhizal nodulation of C. glauca.
The (80–97%) reduction in CgCHS1 mRNA level measured by qRT-PCR compares well with the previously reported efficient reduction of CgSymRK (52–76%) and GUS (46–94%) observed in C. glauca hairy roots through the use of RNAi (Gherbi et al., 2008a,b). Biochemical analyses conducted with C. glauca hairy root material are in accordance with previous reports showing that the major phenolic components of roots belong to the flavan class of flavonoids (Laplaze et al., 1999). As expected, the levels of the identified proanthocyanidin were stongly reduced in CgCHS1-RNAi roots. Two flavonols and one anthocyanin were also identified in hairy roots as quercetin 3-O-glucuronide, quercetin 3-O-glucoside and cyanidin 3-O-glucoside, respectively. These compounds were identified previously in several Casuarina species and suggested as taxonomic markers (Saleh & El-Lakany, 1979). The levels of both flavonols were considerably reduced in CgCHS1-RNAi roots and nodules, which was confirmed by histological analysis.
The reduction of nodulation using CgCHS1 RNAi and the successful complementation of the root flavonoid-deficient phenotypes with naringenin to produce a nodulation rate comparable to that of the transgenic controls clearly showed that flavonoids are involved in C. glauca nodulation. In CgCHS1-RNAi plants, the nodulation rate was affected while the nodule structure was not; this suggests a role of flavonoids in the early stages of Frankia infection. It is worth noting that complementation was more efficient when naringenin was added at the moment of the inoculation of the plant with Frankia (data not shown). This observation suggests that contact with the root is required for naringenin to act upon flavonoid-deficient roots to restore nodulation. It is tempting to hypothesize that specific flavonoids released from C. glauca roots act as signaling molecules as is the case in rhizobia–legume symbioses. In legumes, specific flavonoids released from the roots interact with the NodD protein of rhizobia to activate transcription of other nod genes responsible for the synthesis of LCOs, called Nod factors (Dénarié et al., 1996). However, no homologs of common nod genes from rhizobia were found in the genome of the Frankia strain CcI3 nodulating C. glauca. If a Nod factor-like process is involved in actinorhizal symbioses, it is not based on a standard set of nod genes (Normand et al., 2007b). Therefore, the potential target of flavonoids for induction of the still unknown Frankia nodulation genes remains unidentified. However, the recent study of Popovici et al. (2010) identified candidate genes in Frankia strain ACN14a that respond to Myrica gale seed exudate-specific flavonoids. These would be prime targets for functional analysis.
Recently, Casuarina cunninghamiana root exudates were shown to alter surface properties and plant infectivity of Frankia strain CcI3 (Beauchemin et al., 2012). Yet, flavonoids present in the crude root exudates might be responsible for the higher infection rate in control hairy roots than in CgCHS1-RNAi roots. Characterizing the effect of various flavonoid components isolated in our work on nodulation should help to identify the active compounds.
Downstream from this infection, no significant differences were observed in CgCHS1-RNAi nodule morphology and tissue organization relative to the control pHKN29 nodules. CgCHS1-RNAi roots developed white nodules and light microscopy revealed a gradient of infection and differentiation in the cortex that is present in transgenic and nontransgenic controls. It therefore appears that, despite the decrease in the amount of flavonoids in CgCHS1-RNAi nodules, the late symbiotic events including nodule cell infection and nodule development are fully supported. Laplaze et al. (1999) hypothesized that layers of phenolic-containing noninfected cells in the C. glauca nodule cortex are a constitutive plant defense response in roots and contribute to protection against secondary infection. The distribution of infected and uninfected cells within the cortex of CgCHS1-RNAi nodules did not seem to differ from that of wild type. Thus, further work is needed to fully understand the role of the accumulation of flavonoids leading to compartmentalization of the cortex containing Frankia-infected cells. In legumes, it has been suggested that Nod factor perception could induce certain flavonoids that inhibit auxin transport, causing accumulation of auxin at the nodule initiation site leading to the initiation of nodule primordia (Mathesius et al., 1998; Boot et al., 1999; Wasson et al., 2006). As endogenous regulators of auxin transport, flavonoids have also been suggested to control lateral root initiation in Arabidopsis thaliana (Brown et al., 2001). Because actinorhizal nodule lobes have the same origin and structure as lateral roots, they are considered as modified lateral roots. However, it is worth noting that Péret et al. (2007) observed that the C. glauca auxin influx carrier gene CgAUX1 is expressed in lateral root primordia, but not in nodule primordia. Thus, the absence of phenotypic alterations in CgCHS1-RNAi mature C. glauca nodules might result from a unique developmental program for actinorhizal nodules and a different mechanism of manipulation of root phytohormone balance in actinorhizal versus legume nodule induction. Further studies are now needed to determine whether or not there is a link between flavonoid and auxin during actinorhizal nodulation.
In summary, our data point to a role for flavonoids in the early stages of the process of infection of the roots of C. glauca with Frankia. Additional studies are still needed to establish what are the flavonoid changes in the plant that enable efficient Frankia nodulation. However, in view of the recent finding of common symbiotic events for actinorhizal, legume and arbuscular mycorrhiza formation (Gherbi et al., 2008b; Hocher et al., 2011), our data support the conclusion that flavonoids are among the common signaling elements that are essential for root endosymbioses.
Financial support was provided by IRD and the Agence Nationale de la Recherche (Project SESAM 2010 BLAN 1708 01). K.A-L. was supported by a fellowship from the Egyptian government. We thank S. Guyomarc'h for helpful discussion concerning the histological results and S. Dussert and A. Barnaud for their help with statistical analysis.