The symbiotic interaction of legumes and rhizobia results in the formation of nitrogen-fixing nodules. Nodulation depends on the finely coordinated expression of a battery of genes involved in the infection and the organogenesis processes. After Nod factor perception, symbiosis receptor kinase (SymRK) receptor triggers a signal transduction cascade essential for nodulation leading to cortical cell divisions, infection thread (IT) formation and final release of rhizobia to the intracellular space, forming the symbiosome. Herein, the participation of SymRK receptor during the nodule organogenesis in Phaseolus vulgaris is addressed. Our findings indicate that besides its expression in the nodule epidermis, in IT, and in uninfected cells of the infection zone, PvSymRK immunolocalizes in the root and nodule vascular system. On the other hand, knockdown expression of PvSymRK led to the formation of scarce and defective nodules, which presented alterations in both IT/symbiosome formation and vascular system.
Successful formation of nitrogen-fixing nodules, as a result of the symbiotic interaction between legumes and rhizobia, is a tightly regulated process. Nodulation is initiated when bacteria belonging to the Rhizobiaceae family and plant roots establish a specific molecular dialogue that triggers a series of coordinated, although complex, molecular, cellular and physiological responses in the root hair cells. This leads to the bacterial invasion of root hairs and the further development of a new root-derived organ, the nodule [reviewed by Crespi & Frugier (2008)]. Genetic and molecular analysis of legume–rhizobia nodulation models have provided insights into the spatial and temporal recruitment of pre-existing molecular mechanisms involved in root development (Szczyglowski & Amyot 2003), and revealed the functional nature of some of the players [reviewed by Oldroyd & Downie (2008); Den Herder & Parniske (2009); Høgslund et al. (2009); Ferguson et al. (2010)].
Activation of the nodule development program is induced when the plant root hairs perceive the rhizobial-produced signals (lipochitooligosaccharides) known as Nod factors (Geurts, Fedorova & Bisseling 2005; Radutoiu et al. 2007). Consecutively, the tip of responsive root hairs displays some physiological changes, that is, an increase in Ca+2 influx, Ca+2 oscillations/spiking and cytoskeletal rearrangements, among others [reviewed by Cárdenas et al. (2000)], as well as some morphological deformations such as swelling and curling [reviewed by Oldroyd & Downie (2008)]. These responses promote rhizobial attachment to the root hair and favour the entrapping of bacteria into an infection pocket resulting in the formation of an elongated tubular structure, the infection thread or IT [reviewed by Den Herder & Parniske (2009)]. In parallel, the nodule organogenesis programme is switched on, prompting the meristematic growth of recently dedifferentiated cortical cells to form the nodule primordium. As the young nodule emerges from the root, some steps of differentiation and cell organization occur, which result in the formation of the central and the peripheral tissues. Finally, rhizobia invade the nodule primordia cells and differentiate into nitrogen-fixing bacteroids [reviewed by Patriarca et al. (2004)]. Depending on the legume species, two types of nodules are formed: indeterminate nodules (i.e. Medicago nodules) which are oval-shaped and characterized by a continuously active meristem at the apical zone, and determinate nodules, globular in shape, which do not present a persistent meristem (i.e. Lotus and Phaseolus). Both have a peripheral vascular tissue. Current knowledge on Phaseolus vulgaris nodule development is mainly limited to the work by Tatéet al. (1994) and Papadopoulou, Roussis & Katinakis (1996); however, in some extent, it has been largely assumed to be equivalent to Lotus japonicus nodulation.
At the molecular level, it has been demonstrated that the perception of the Nod factors turns on a signal transduction pathway, where several key players have been identified, for instance, LysM kinase-like and leucine-rich repeat receptor-like kinase (LRR-RLK) receptors, calcium calmodulin-dependent protein kinase (CCamK/DMI3), CYCLOPS, the nucleoporins Nup85, Nup133 and NENA (a member of the Sec13/Seh1 protein family), as well as some transcription regulators (NSP1, NSP2 and ERN), among others [reviewed by Crespi & Frugier (2008); Oldroyd & Downie (2008); Den Herder & Parniske (2009)].
LRR-RLK receptors belong to a family of proteins involved in a diversity of molecular mechanisms (Diévart & Clark 2004; Tör, Lotze & Holton 2009). Although a large number of genes encoding LRR-RLK receptors are present in the plant genome, the biological function of the vast majority remains to be deciphered (Diévart & Clark 2004). The primary structure of LRR-RLK receptors is characterized by an amino-terminal containing a signal peptide sequence, an ectodomain containing several well-conserved LRR motifs in tandem, a transmembrane segment and a cytoplasmic domain with typical serine/threonine protein kinase signature (Afzal, Wood & Lightfoot 2008). The LRR-RLK receptor component of the signalling cascade involved in the nodulation process [hereafter named SymRK, which stands for symbiosis receptor kinase-like receptor; Stracke et al. (2002)] contains three LRR motifs denoting an extracellular ligand-induced activation (Bella et al. 2008; Holsters 2008); however, no LRR-binding candidates have been identified. Orthologs of SymRK have been described in several legumes and non-leguminous plants; they were named in accordance to the nomenclature assigned to the corresponding mutant or gene. For example, in L. japonicus, Sesbania rostrata, Datisca glomerata and Casuarina glauca, it is reported as SymRK, whereas in Medicago truncatula, Medicago sativa and Pisum sativum are named DMI2, NORK and Sym19, respectively. The functional roles of the SymRK in rhizobial nodule development have been revealed by the analysis of legume mutants impaired for nodulation and the use of molecular approaches such as gene cloning, knockdown and overexpression experiments (Catoira et al. 2000; Endre et al. 2002; Stracke et al. 2002; Capoen et al. 2005; Gherbi et al. 2008; Markmann, Giczey & Parniske 2008). When inoculated with a compatible endosymbiont, symrk and dmi2 mutants display swelling and branching of the root hairs, but further steps are blocked (Endre et al. 2002; Stracke et al. 2002). Therefore, it is clear that mutations in SYMRK gene interrupt the program leading to nodule development soon after Nod factor perception. Neither epidermal responses (i.e. root hair curling and entrapment of bacteria) nor IT formation, nor cortical cell division are achieved (Stracke et al. 2002). Ectopic expression of a full-length DMI2 cDNA under control of 35S promoter in a M. truncatula dmi2 mutant favoured the formation of ITs, although with an increased diameter and extensive branching, but no symbiosomes were observed, indicating that regulated spatio-temporal expression is required to fully rescue the wild-type phenotype (Limpens et al. 2005). Transgenic roots expressing a chimeric protein DMI2-GFP depicted the localization of DMI2 to the IT membrane, as well as to the plasma membrane of cells in the distal part of the infection zone (Limpens et al. 2005). The essential role of this receptor in symbiosome formation has also been established in SymRK knockdown lines in S. rostrata (Capoen et al. 2005). Beyond its key functions in nodulation, SymRK is also necessary in symbiotic interactions between vascular plants and Glomus intraradices to form arbuscular mycorrhiza (AM) association and non-leguminous angiosperms and actinobacteria to form actinorhizal nodules (Gherbi et al. 2008; Markmann et al. 2008). Moreover, the ortholog of SymRK from either actinorhizal or from non-nodulating but AM-forming plants restores both rhizobia and AM endosymbiosis in legume mutants deficient in the expression of this receptor. However, expression of either the tomato or the rice version of SymRK carrying two, instead of three, LRR domains under control of LjSymRK promoter in the L. japonicus mutant symrk-10 selectively restored AM symbiosis, but nodulation remained impaired. Neither IT formation nor nodule development functions were complemented by Lycopersicon SymRK, whereas some nodule primordia, generally devoid of bacteria, and, in rare cases, small nodules were observed in roots of symrk-10 expressing the rice version of the receptor (Markmann et al. 2008).
Herein, insights into an unexplored function of SymRK related to the development/function of the vascular tissue in P. vulgaris nodules are provided. PvSymRK immunolocalizes in the nodule vascular bundles, parenchyma as well as in the root central cylinder. We have also found that different down-regulated levels of PvSymRK transcript result in an impairment in nodulation, as well as a deficiency in both IT/symbiosome formation, as previously described, and development of the nodule vascular bundles.
MATERIALS AND METHODS
Germination of common beans (P. vulgaris L. cv. Negro Jamapa)
Bean seeds were surface sterilized with hypochlorite solution at 20% for 5 min and then with absolute ethanol for 1 min, followed by five washes with sterile water. Surface-sterilized seeds were germinated for 2 d (Cárdenas et al. 1995), planted in pots containing sterilized vermiculite and inoculated with Rhizobium tropici (strain CIAT 899) or Rhizobium etli (strain CE3), when indicated. Roots and nodules were harvested at different post-inoculation times (hpi, hours post-inoculation; dpi, days post-inoculation) and stored at −80 °C until use, or alternatively, fixed and processed as described.
Accession numbers of sequences used in this work are as follows: Alnus glutinosa, AY935263; Astragalus sinicus, AY946203; C. glauca, EU273286; D. glomerata, AM271000; Glycine max, ADH94611; Lupinus albus, AY935267; Lycopersicon esculentum, AY935266; L. japonicus, AF492655; Lathyrus sativus, DQ403853; M. sativa, AJ418368; M. truncatula, AF491998; Muscodor albus, AJ428991; Oryza sativa, XM478749; Papaver rhoeas, AM270999; P. sativum, Q8LKZ1; Populus trichocarpa, AM851092; P. vulgaris, GI:312434884 (this work); S. rostrata, AY751547; Tropaeolum majus, AY935265; Vicia hirsuta, AJ428990; Zea mays, DQ403195.
Total RNA was extracted from frozen roots (infected and non infected), leaves, shoots and nodules, using the guanidine isothiocyanate method (Chomczynski & Sacchi 1987), and treated with one unit of RNAse-free DNAse I (Roche, Indianapolis, IN, USA) at 37 °C for 15 min, followed by inactivation at 65 °C. For RT-PCR, first-strand cDNA was synthesized from 2 µg of total RNA using M-MLV RT (Invitrogen, Carlsbad, CA, USA) and poly-dT oligonucleotide, according to the manufacturer's instructions. Real-time RT-PCR (qRT-PCR) reactions (25 µL) containing 200 ng of total RNA, gene-specific primers and iScript™ One-Step RT-PCR Kit with SYBR®Green (Bio-Rad, Hercules, CA, USA) were performed in an iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad), according to the manufacturer's instructions. All RT-PCR experiments were performed in triplicate. Specific oligonucleotides used in this analysis are as follows: for RT-PCR, PvSymRK-Up, 5′-GAAGATTTATGGTACTAGGT-3; PvSymRK-Lw, 5′-TGTCAAGGCTACTCTGGA-3′; PvAquaporin-1-Up, 5′-CGCCGCTGTTTGAGCCCTCG-3′; PvAquaporin-1-Lw, 5′-TTGCGCATCGTTTGGCATCG-3′; for real-time RT-PCR, PvSymRK-RTUp, 5′-GAATTCTATGATGGAATTACCAGAAATTTGGG-3′; PvSymRK-RTLw: 5′-CTGGCTTTGCAACTGAAGGG-3′; PvEF1a-RTUp, 5′-GGTCATTGGTCATGTCGACTCTGG-3′; PvEF1a-RTLw 5′-GCACCCAGGCATACTTGAATGACC-3′.
Plasmid construction and generation of transgenic hairy roots
Construction of pTdT-PvSymRK-RNAi, driving the expression of PvSymRKi under control of the ectopic 35S promoter, was created as follows: a 452 bp PCR fragment unique to the non-conserved region and not overlapping the LRR motifs of the PvSymRK coding sequence was amplified using gene-specific forward primer PvSymRKi-Up (5′-CACCTGAAGGGTTTGAAAACATAGC-3′) and reverse primer PvSymRKi-Lw (5′-TGGCAGGCAAATCCTGTAAGAGG-3′). PCR product was further cloned into the destination vector pTdT-DC-RNAi (Valdés-López et al. 2008) using Gateway technology (Invitrogen). The resulting construct was used for Agrobacterium rhizogenes strain K599-mediated transformation, which allowed the generation of composite plants, mainly as described by Estrada-Navarrete et al. (2007). Putative transgenic hairy roots expressing red fluorescence tdTomato reporter, encoded in the vector, were screened by using a fluorescence stereomicroscope (SteREO Discovery V8, Zeiss, Le Pecq, France). No fluorescent emergent root primordia were dissected off to favour the growth of transgenic fluorescent roots. Composite plants were inoculated with R. tropici (strain CIAT 899). Roots and nodule structures were collected as indicated.
Construction driving the overexpression of PvSymRK-cMyc was generated by a PCR amplification strategy using PvSymRK cDNA and PvSymRK-specific primers: forward 5′-CACCATGGGGTTTGAAAACATAGCATG-3′ and reverse 5′-TCACAGATCTTCTTCAGAGATCAGTTTCTGTTCTCTCGGCTGTGGATGGGA-3′ (which includes sequence coding of cMyc epitope for protein tagging purposes). PCR product was cloned in the plant expression vector pEarleyGate 203 (Early et al. 2006).
Antibodies and Western blot analysis
Polyclonal antibodies against PvSymRK were raised in rabbits immunized with a PvSymRK-His6X recombinant protein spanning 449 amino acids of the extracellular region of PvSymRK (from residue 33 to 482). Western blots were performed according to standard procedures. Briefly, frozen plant tissues were homogenized in 20 mM Tris–HCl buffer, pH 7.4 containing 150 mM NaCl, 1 mM ethylenediaminetetraacetic acid, 1% Triton X-100, 0.1% dithiothreitol (DTT), 1 mM phenyl methyl sulfonyl fluoride (PMSF), 10 µg mL−1 leupeptin, 5 µg mL−1 chymostatin, and 10 µg mL−1 aprotinin. Proteins from crude extracts were precipitated with 20% trichloroacetic acid. The protein pellet was washed several times with cold acetone and further resuspended in reducing and denaturating loading buffer containing 2 mM 2-mercaptoethanol and 2% sodium dodecyl sulphate. Proteins (30 µg) were loaded onto a 10% sodium dodecyl sulphate–polyacrylamide gel electrophoresis, separated under denaturating conditions and electro-transferred to a nitrocellulose membrane (Immobilon NC, Millipore, Billerica, MA, USA). The membranes were blocked in phosphate buffered saline (PBS) containing 5% non-fat milk and 0.1% Tween 20, incubated with either anti-PvSymRK serum, monoclonal anti-beta tubulin (kindly provided by Juan Olivares; Amersham Life Science, Piscataway, NJ, USA) or anti-cMyc E910 monoclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). Antibodies were diluted 1:1000 in blocking solution. The signal was developed with corresponding goat anti-rabbit or anti-mouse IgGs conjugated to horse radish peroxidase (Millipore) and Luminata Forte Western HRP substrate (Millipore). PageRuler Pre-Stained Protein Ladder (Fermentas, International Inc, Canada) was used as reference. Protein quantification was carried out with the Bio-Rad Protein Assay (Bio-Rad).
Specimens were fixed in a mixture of 3% paraformaldehyde and 0.5% glutaraldehyde in 50 mM Na-phosphate buffer, pH 7.2, for 1 h at room temperature. Subsequently, the samples were washed with the same buffer and embedded in paraffin (Paraplast Tissue Embedding Medium, SPI Supplies, West Chester, PA, USA), according to standard procedures. Sections (8–10 µm) obtained in a Jung Biocut microtome (Leica, Wetzlar, Germany) were used for immunolocalization purposes according to standard protocols. Briefly, sections were incubated with 5% not-fat milk in Tris buffered saline and tween 20 or TBST (10 mM Tris–HCl, pH 7.5, 150 mM NaCl and 0.05% Tween 20) for 2 h at room temperature to further incubate in the presence of anti-PvSymRK serum (diluted 1:50 in TBST). Immunolocalization signal was revealed by Alexa Fluor 488 (diluted 1:100 in TBST). Sections were further mounted in PBS–glycerol and visualized by laser scanning confocal microscopy (LSM 510 Meta, Zeiss).
Histology and light microscopy
Specimens were fixed in a mixture of 2.5% glutaraldehyde and 4% paraformaldehyde in 0.1 M Na-cacodylate buffer, pH 7.2 at 4 °C for 16 h, post-fixed with 1% osmium tetroxide for 2 h at 4 °C and dehydrated through a series of ethanol (from 10 to 100%). Samples were then embedded in LR-White resin (London Resin Company, Ltd, England) and polymerized under ultraviolet light at −20 °C for 48 h. Semi-thin sectioning (0.5–1.0 µm) was performed using an ultramicrotome (Ultracut R, Leica) and stained with 0.1% toluidine blue. Sections were analysed under a DMLB microscope (Leica). Digital images were taken using an AxioCam MRc CCD camera (Zeiss Microscopy&Images) controlled by the AxioVision image acquisition software (Zeiss Microscopy&Images). Thereafter, image processing was performed using Adobe Photoshop, version 7.0 software (Adobe Systems, Inc., San Jose, CA, USA).
Isolation of P. vulgaris SymRK cDNA
In order to provide insights into the development of the nitrogen-fixing nodules in P. vulgaris, we have searched for signalling components known to play a key role, not only during early stages of the legume–bacteria interaction, but along nodule organogenesis. Using a NORK probe (kindly provided by Dr G. Kiss), we have isolated a full-length cDNA coding for a P. vulgaris SymRK, a putative ortholog of L. japonicus SymRK (Stracke et al. 2002), DMI2 and NORK (Endre et al. 2002), and S. rostrata SrSymRK (Capoen et al. 2005). PvSymRK cDNA sequence includes both 5′ and 3′ untranslated regions (322 and 126 bp, respectively) and 2757 bp of coding sequence. The open reading frame encodes a polypeptide of 919 amino acid residues with a calculated molecular mass of 103.2 kDa. Protein sequence alignment confirmed that the primary structure of PvSymRK conserves the characteristic features of other SymRK (Supporting Information Fig. S1); it shares 85, 82, 78 and 78% amino acid identities with SrSymRK, LjSymRK, MtDMI2 and MsNORK, respectively. Regarding the phylogenetic position of PvSYMRK gene, we have performed a standard bioinformatics analysis of several SymRK sequences from different angiosperms. A phylogenetic tree with three main subgroups was obtained (Fig. 1). One subgroup clusters sequences from plants of the family Fabaceae (legumes); the second subgroup contains a cluster of SymRK sequences from actinorhizal plants, which is separated from the subgroup of sequences from non-legume plants forming AM, but unable to establish symbiosis with bacteria (Fig. 1). By Southern blot analysis of P. vulgaris genomic DNA using sequences coding for the non-conserved extracellular moiety of PvSymRK as probe, we have found that PvSYMRK is a single-copy gene (Supporting Information Fig. S2).
PvSYMRK gene is expressed only in roots and nodules
The expression of the PvSYMRK gene in different common bean tissues was addressed by using a RT-PCR approach. The results shown in Fig. 2 indicate a tissue-specific accumulation of PvSymRK transcripts in roots, both inoculated and not inoculated with rhizobia, as well as in nodules (Fig. 2b); however, no detectable expression was found in shoots and leaves (Fig. 2a). Our data also indicate that PvSymRK transcripts are present along the nodulation process with an apparently increased accumulation at late developmental steps (10–15 dpi, Fig. 2b). Similar results were observed in M. truncatula nodules (Bersoult et al. 2005).
Immunolocalization of PvSymRK in the vascular system of both roots and nodules
To analyse the cellular localization of this receptor in common bean roots and nodules, we have raised rabbit polyclonal antibodies against a recombinant version of the extracellular region of PvSymRK expressed in Escherichia coli. Antibody specificity was assessed by Western blot analysis of whole extracts from either nodules (15 dpi) or wild-type roots or hairy roots expressing PvSymRK tagged with a cMyc epitope at the C-terminal. In all cases, anti-PvSymRK antibodies recognize a double band of approximately 100 and 110 kDa, respectively, which is in agreement with the expected size of PvSymRK (Supporting Information Fig. S3). Although occasionally it was not possible to resolve them, detection of two bands was highly reproducible and specific as demonstrated by running in parallel a Western blot analysis using anti-cMyc monoclonal antibodies (Supporting Information Fig. S3a). A similar pattern was obtained when whole nodule (15 dpi) extracts were analysed (Supporting Information Fig. S3b). Detection of a double band may reflect a post-translational modification of this receptor; this is supported by in vitro phosphorylation of the LjSymRK intracellular domain (Yoshida & Parniske 2005) and putative sites for N-glycosylation which are predicted by sequence analysis (data not shown). No cross-reaction with other proteins was observed (Supporting Information Fig. S3).
Immunolocalization of PvSymRK in nodules at several developmental stages was addressed, as illustrated in Fig. 3. Representative images from three independent nodules are shown. In longitudinal sections from roots at 3 dpi, we have found PvSymRK associated to the central cylinder and epidermis of the root. A faint signal was also detected in the intercellular space of the nodule primordium, as well as in the nascent nodule provascular trace (Fig. 3a). At 6 dpi, PvSymRK was observed at the base of the young nodule, in the zone where the nodule vascular bundle is expected to initiate. PvSymRK signal was also interspersed in a not yet well-organized central tissue (Fig. 3b). According to Tatéet al. (1994), in an emergent 1 week post-inoculated nodule in P. vulgaris, the central region is a cluster of cytoplasm-rich cells in which few vacuolated cells are present. At later stages (12 and 15 dpi), PvSymRK was detected in the central tissue, associated to uninfected cells, whereas no signal was observed in infected cells (Fig. 3c–e). PvSymRK was also found in the inner cortex of the nodule (pre-infected zone, usually containing abundant ITs), which includes the parenchyma and the vascular bundles (Fig. 3c,d). Distribution of the signal in the proximal zone of the nodule allowed us to propose that PvSymRK is also expressed in the vascular bundles connecting the root central cylinder with the nodule vasculature (Fig. 3c, see arrowhead). PvSymRK signal was observed around the root central cylinder (Fig. 3c, see arrow labelled as rcc). It is worth mentioning that no signal has been detected in negative control experiments using pre-immune serum (Supporting Information Fig. S4).
The nodulation process in P. vulgaris is impaired by the down-regulated expression of PvSymRK
To gain further insights into the role of the PvSYMRK gene in nodule organogenesis in P. vulgaris, we have knocked-down its expression by RNAi-mediated gene silencing in composite plants. PvSymRK RNAi construct was based on a non-conserved region of the extracellular moiety of PvSymRK (excluding the LRR motifs), unique to this receptor. The design of the PvSymRK RNAi construct was supported by a BLAST analysis in which the PvSymRK RNAi sequence (452 bp) was used as query to search against NCBI non-redundant nucleotide collection or legume sequence databases, including those available for P. vulgaris. This analysis did not hit sequences other than SYMRK orthologous genes; therefore, it can be considered unambiguously that small RNAs (21–23 bases) derived from PvSymRK RNAi would target specifically PvSymRK transcripts. PvSymRK RNAi was expressed under control of the ectopic promoter 35S (henceforth called PvSymRKi hairy roots). Control hairy roots were generated by transformation with A. rhizogenes carrying either no vector or an empty binary vector (pTdT-DC-RNAi, with no RNAi coding sequences).
As it is generally well accepted that RNAi-mediated strategies lead to a wide range of gene-silencing efficiencies, we have monitored the accumulation of PvSymRK transcripts in representative rhizobia-inoculated single hairy roots (21 dpi) from independent events of transformation. As illustrated in Fig. 4a, in all RNAi-expressing transgenic roots, except PvSymRKi 1, the level of PvSymRK mRNA was significantly reduced in comparison to the levels detected in control transgenic roots (K599 and pTdT). A decrease in the accumulation of this transcript correlates with a low PvSymRK protein expression in hairy roots harvested at 5 dpi, as confirmed by Western blot analysis (Fig. 4b,c). PvSymRK RNAi does not target other LRR-RLK receptor genes, as assessed by qRT-PCR analysis of KLAVIER (KLV) and NARK (Searle et al. 2003; Miyazawa et al. 2010). Transcript accumulation of neither PvKLV nor PvNARK was decreased in PvSymRKi hairy roots (Supporting Information Fig. S5).
RNAi-mediated silencing of PvSYMRK gene impaired the nodulation process, as revealed by direct observation under stereoscopic microscope and counting nodule structures generated in silenced hairy roots at 21 dpi. As summarized in Table 1, from 14 control hairy roots (pTdT), we have calculated an average of 22.9 ± 3.15 (±SE, n = 320) wild-type nodules per single hairy root. In contrast, PvSymRKi hairy roots (77 in total) presented decreased nodulation ability, in accordance to the non-nodulating phenotype of symrk mutants (Endre et al. 2002; Stracke et al. 2002). We have found 4 ± 0.65 (average ±SE, n = 306) nodule structures per single PvSymRKi hairy root. At first sight, it was noted that most nodule structures produced in PvSymRKi hairy roots were not fully developed; they were white (meaning no nitrogen fixation activity), smaller than wild-type nodules and did not present the typical prominent ridges on the surface of the nodule owing of lenticels [usually overlying the vascular traces, as described by Guinel (2009)]. Having as a reference the nodules generated in control hairy roots (mainly mature, wild-type nodules) and in accordance to their size and appearance, nodule structures produced in PvSymRKi hairy roots and harvested at 21 dpi were classified as type I (whitish, poorly developed, tiny nodule-like structure having ∼0.4–0.5 mm in diameter), and type II or III when nodule-like structures looked like nodules at different stages of development, heterogeneous in their size but considerably smaller than those developed in control hairy roots; as a general approach, we considered small or medium size nodule-like structures as type II or III, respectively. As observed in representative samples harvested later after inoculation (28 dpi), neither type I nor types II and III nodule-like structures showed signs of further development (Fig. 5a). Finally, a group of PvSymRKi hairy roots generated nodule structures that resembled wild-type nodules. A graphical analysis of their distribution is presented in Fig. 5b,c. It is noteworthy to mention that 32.5% of the PvSymRKi hairy roots (25 out of 77) did not generate nodule structures but were included in the statistical analysis (Fig. 5b). A group of 25 out of 77 PvSymRKi hairy roots (32.5%) presented few type I nodule-like structures (2.8 ± 0.21 per hairy root; average ±SE, n = 71), but neither type II nor type III nodule-like structures were found. Whereas 13 silenced hairy roots presented a mixture of types II and III nodule-like structures, from these, 7.9 ± 0.68 (average ±SE, n = 71) were type II nodule-like structures per hairy root and were found in 11.7% of PvSymRKi hairy roots (nine out of 77). An average of 21.3 ± 4.91 (±SE, n = 85) type III nodule-like structures were counted in 5.2% of the PvSymRKi hairy roots (four out of 77). The last 14 (18.2%) PvSymRKi hairy roots appeared to produce structures similar to nodules produced in wild-type roots (5.7 ± 1.6 per hairy root; n = 79). Figure 5c summarizes these results.
Table 1. Inhibition of nodulation in transgenic hairy roots expressing PvSymRK-RNAi
Number of single hairy roots studied
Nodule structures per hairy root (average ± SE)
Control hairy roots generated by transformation with Agrobacterium rhizogenes K599 carrying either no vector (a) or an empty binary vector pTdT (b).
Hairy roots generated by transformation with A. rhizogenes K599 carrying the construction pTdT-PvSymRK-RNAi.
Total number of nodule structures per group of composite plants.
25 out of 77 hairy roots expressing PvSymRK-RNAi did not generate nodule structures.
Down-regulation of PvSymRK interferes with the development of the nodule vascular system
In order to analyse the cell organization in types I, II and III nodule-like structures, resin-embedded samples were fully sectioned and observed under the microscope. A comparative histological analysis of wild-type and these nodule-like structures was performed. This allowed us to describe some effects of a down-regulated expression of PvSYMRK gene on nodule development. Figures 6 and 7 show representative images corresponding to either longitudinal or transversal middle sections of nodule-like structures or a wild-type nodule. As previously described by Tatéet al. (1994) and Guinel (2009), two major tissue types can be distinguished in the wild-type nodule produced in P. vulgaris roots: the peripheral tissue, which contains the nodule vascular bundles, and the central tissue (Fig. 6a). The latter consists of ITs and large infected cells with interspersed smaller, vacuolated, uninfected cells containing amyloplasts (Fig. 6b,c). Typically, the nodule vasculature originates from the root central cylinder as a bifurcated structure connecting the nodule to the root (Fig. 6a). It further branches around the circumference of the nodule and differentiates acropetally, giving rise to several vascular bundles containing phloem surrounding xylem (Fig. 6a,d,e). In contrast, the cell organization observed in a tiny and irregular in shape type I nodule-like structure reflects the phenotypic severity and penetrance of the misregulated expression of PvSYMRK gene (Fig. 6f–k). This type I nodule-like structure presented a prominent but disorganized central tissue containing uninfected, compact, cytoplasm-rich cells with a condensed nucleus, suggesting that they maintained a meristematic activity (Fig. 6g,h,j,k). Two to three layers of vacuolated cells were surrounding the central tissue (Fig. 6f). Few ITs were observed (Fig. 6j); however, no rhizobia-infected cells were detected. Analysis of serial sections revealed the lack of vascular bundles in this tiny type I nodule-like structure. Nonetheless, it seems that the irregular tissue forming the proximal zone of this nodule-like structure contains a hint of a provascular trace, defective in its development (Fig. 6f,i). Amyloplasts were hardly observed (Fig. 6g,h,j,k), which may reflect a starvation status as a consequence of the lack of vascular bundles; therefore, no nutrients would be supplied to this nodule-like structure. According to Tatéet al. (1994), in P. vulgaris nodules, the provascular traces are originated by cell division activity of the pericycle and inner cortex cells facing the infected root hair. At 6 dpi, procambial strands between the nodule primordium and the root central cylinder are already present (Guinel 2009). In 1-week-old nodules, the vascular bundles and the nodule endodermis (an extension of the endodermis of the root central cylinder) are forming (Tatéet al. 1994). Supporting Information Fig. S6 presents a histological section of a P. vulgaris root at 5 dpi, forming two nodule primordia and a lateral root primordium. Nodule primordia appeared to be at slightly different developmental stages that, considering the level of cortical cell division and extent of meristematic activity, may correspond to 3 and 5 dpi, respectively. In both, the process leading to the formation of the provascular traces had already initiated. Two provascular traces became defined in the primordium at 5 dpi (as indicated by arrowheads). Such a tissue organization is not distinguished in the proximal zone of type I nodule-like structure illustrated in Fig. 6f,i, suggesting an arrest/blockage early in vascular bundle development.
Sections of two nodule-like structures (type II and III, respectively) are illustrated in Fig. 7. Analysis of a transverse section of the type II nodule-like structure showed that its nodule vascular bundles were ectopically distributed at the middle of a presumably disorganized central tissue (Fig. 7a,b). The central tissue of the type III nodule-like structure contained abundant, wide ITs, some of them crossed the intracellular spaces (Fig. 7d–f). Infected cells were also found (Fig. 7f), although they were considerably less numerous in comparison to infected wild-type nodules. In this type III nodule-like structure, the nodule vasculature was not fully mature, as indicated by the analysis of two provascular traces. The uncompleted maturity of these traces was observed through consecutive transverse sections collected from the apical to the middle zone (Fig. 7g,h). Abundant cytoplasm-rich, nucleated cells present in the provascular bundle suggest an active meristem. Tracheary elements were not observed in the provascular traces (Fig. 7g). As described by Guinel (2009), vascular bundles in determinate nodules usually contain four to six tracheary elements, which are all small in diameter and their number does not change with the branch order. As illustrated in Fig. 6d, these elements are easily identified in a longitudinal section of a vascular bundle from a wild-type nodule and are indicative of xylem differentiation (Evert & Eichhorn 2006; Guinel 2009). Regarding vascular branching, apparently, it is not affected in this type III nodule-like structure, since two to three independent traces were observed in middle zone sections (Fig. 7c). Altogether, these suggest that a tightly regulated spatio-temporal expression of PvSymRK is determinant in more than one step in the nodule developmental programme. A dose-dependent effect may directly or indirectly switch on/off a mechanism controlling cell differentiation events, including pre-vascular and mature vascular bundle formation.
In order to improve our understanding of the function of SymRK in the organogenesis of the nodule, we considered imperative to investigate the relationship between the spatio-temporal distribution of this receptor and the development of P. vulgaris nodules. Cell/tissue organization of nodules generated when PvSymRK expression was down-regulated was also investigated. In addition to the previously described role of SymRK in IT growth and rhizobial delivery (Mirabella 2004; Bersoult et al. 2005; Capoen et al. 2005; Limpens et al. 2005), our findings suggest a correlation between SymRK expression and the development of the nodule vascular system.
Immunolocalization studies at several stages of P. vulgaris nodule development showed that PvSymRK is expressed in the root epidermis and cortical cells forming the nodule primordium, in accordance to the transcriptional activity of DMI2 promoter previously reported (Bersoult et al. 2005; Holsters 2008). PvSymRK was not detected in infected cells, it was rather found in uninfected cells of the central tissue (infection zone), as well as in cells of the inner cortex and, presumably, in ITs. A main finding was the clear association of PvSymRK to the vascular tissue of both roots and nodules. In young nodules, PvSymRK was detected at the nodule proximal zone, where the nodule vascular elements arise from the root vasculature. Therefore, PvSymRK expression is apparently a continuous from the root vascular cylinder to the fully differentiated vascular bundle in the nodule, suggesting a role of PvSymRK in the development/function of the nodule vasculature. It is noteworthy to mention that the spatio-temporal distribution of PvSymRK resembles the pattern of ENOD40 expression in the dividing root cortical and pericycle cells, as well as in the nodule vascular bundles (Yang et al. 1993; Papadopoulou et al. 1996; Charon et al. 1997; Imaizumi-Anraku et al. 2000; Varkonyi-Gasic & White 2002). Favery et al. (2002) proposed a role of ENOD40 in cell-to-cell communication processes between vascular and cortical root tissues. ENOD40 transcription requires the expression of SymRK, as revealed by the analysis of a dmi2 mutant (a SymRK deficient M. truncatula mutant; Catoira et al. 2000). Beyond the induction of ENOD40 transcription, it would be thus interesting to explore a possible functional relationship between PvSymRK and ENOD40.
Whether there is a correlation between the expression of PvSymRK and the biology of the nodule vasculature was reinforced by results obtained from the histological analysis of several nodule-like structures produced in nodulation-deficient PvSymRKi hairy roots. We focused our attention on rhizobial invasiveness and tissue organization of these impaired nodules, with special emphasis on the vascular system. As in other legumes, regardless of whether they form indeterminate or determinate nodules, a significant decrease in the expression of SymRK led to an inefficiency or even a blockage in the delivery of rhizobial from ITs to cortical cells, in agreement with previously reported data (Endre et al. 2002; Stracke et al. 2002; Capoen et al. 2005; Limpens et al. 2005). More interesting, down-regulated expression of PvSymRK was correlated with a defective vascular system phenotype in nodule-like structures. This phenotype ranged from the presence of immature vascular traces to the ectopic localization of the vascular bundles or even the absence of vasculature. Whereas the control nodule presented three to four differentiated vascular bundles. Together, our findings strongly suggest that a finely regulated, spatio-temporal expression of PvSymRK is required for a well-coordinated differentiation and/or development of the nodule vascular bundles in P. vulgaris. It is reasonable to postulate a similar SymRK regulatory activity in any other legume nodules. Although it has been reported that dmi2 mutants are not affected in lateral root development (Limpens et al. 2005), a rigorous analysis of root architecture in dmi2 roots will provide insights into the role of SymRK in root vascular system.
Establishing the connection between the expression of SymRK and nodule vascular system is not an easy task. The literature reports some mutants deficient in their nodule vascular system, but a rigorous analysis, indicating which developmental step is affected, has not been performed [as reviewed by Guinel (2009)]. The irregular cell organization at the proximal zone of the type I nodule-like structure, devoid of well-defined vascular traces (Fig. 6f,i), reminds the ineffective nodule-like structures formed by L. japonicus cerberus and alb1 mutants (Imaizumi-Anraku et al. 2000; Yano et al. 2006, 2009). In the small bumps formed in cerberus mutants, bifurcated but not fully developed vascular bundles were occasionally initiated (Yano et al. 2009); whereas in most of the nodule-like structures produced in alb1, only a single provascular trace was observed (Imaizumi-Anraku et al. 2000). An ectopically located vascular system, similar to what is found in type II nodule-like structure (Fig. 7), has been observed in nodule-like structures developed on the P. vulgaris mutant R69 (Shirtliffe & Vessey 1996). The genetic pathway that regulates the development/function of the nodule vascular system in legumes may also include NIN (Schauser et al. 1999; Marsh et al. 2007) and M. truncatula microRNAs, miR166 and miR167, targeting HD-ZIP III transcription factors and auxin response factors, respectively (Boualem et al. 2008; Hirakawa, Kondo & Fukuda 2010).
Another possible interpretation of the altered vasculature system observed in nodule structures generated in PvSymRKi hairy roots is the blockage of the nodule organogenesis as a consequence of an impaired infection process. Although we do not exclude that possibility, it may not be the case. In the absence of rhizobia, snf1 and snf2 mutants produce spontaneous, genuine, empty nodules, with an ontogeny similar to that of wild-type nodules (Gleason et al. 2006; Tirichine et al. 2006); double mutants symrk snf1 and symrk snf2 conserve the ability to spontaneously nodulate (Hayashi et al. 2010; Madsen et al. 2010). In other words, infection deficiency of PvSymRKi hairy roots does not necessarily imply a blockage in nodule organogenesis. It would be interesting to explore the effect of misregulated expression of other genes (p.e. CYCLOPS and CERBERUS; Yano et al. 2008, 2009) involved in the coordination of progressive IT growth and nodule organogenesis.
It is well known that signal transduction mediated by SymRK is linked to the synchronous participation of two genetic pathways responsible for an effective nodulation. Namely, SymRK expression is required for IT/symbiosome formation (Capoen et al. 2005; Limpens et al. 2005) and for root cortical divisions (Limpens 2004; Kosuta et al. 2011). Division of pericycle cell (mediated by the induction of ENOD40) is also part of SymRK signalling cascade leading to the formation of the nodule primordium (Capoira et al. 2000). It is worth noting that pericycle cell division is the first step in the nodule vascular development programme (Guinel 2009). Single mutation in the highly conserved GDPC motif located upstream the first LRR motif in LjSymRK uncouples SymRK-dependent initiation of IT formation in root hairs from the cortical cell division response, therefore suggesting that different inputs (p.e. local signalling molecules; Hirakawa et al. 2010) may lead to non-equivalent outputs, namely activating a diversity of SymRK-mediated signal transduction pathways (Kosuta et al. 2011). To improve our knowledge on how SymRK participates in diverse fundamental processes in nodulation, it becomes mandatory to identify and localize SymRK-signalling molecules, a challenging task. Although not investigated yet, a possibility to be taken in account as potential secondary signals are sensing hormones (p.e. cytokinins and auxins) involved in nodulation (Oldroyd & Downie 2008; Ding & Oldroyd 2009; Takanashi, Sugiyama & Yazaki 2011). It is not excluded that SymRK may be sensing both local and systemic signals, as infection initiation, nodule primordium formation and division of pericycle cells take place in different tissues (Limpens 2004). Potential intracellular proteins recruited as part of the machinery that transduces SymRK signal activation have been identified (remorins, MtHMGR, an ARID DNA-binding protein or LjSIP1; Kevei et al. 2007; Zhu et al. 2008; Lefebvre et al. 2010). However, understanding their functional relationship with SymRK interactors is a complicated work, but it may provide some insights.
Altogether, these observations and our data lead us to propose that SymRK is a multifunctional regulator associated to an extensive signalling network, even more diverse than previously described, linking the infection and nodule organogenesis pathways. Our data suggest that SymRK-mediated regulation is also related to the nodule vascular bundle. Certainly, future lines of research might extend our understanding of SymRK functions in nodule development.
We thank Dr Joseph Dubrovsky and Dr Luis Cárdenas for critical discussion and review of the manuscript. We also thank David Sardineta for the cloning of PvSymRK cDNA, Nayeli Sánchez for the paraffin inclusions and sectioning and Juan Olivares for providing anti-beta tubulin antibodies and Unidad de Síntesis y Secuenciación de ADN from Instituto de Biotecnología, Universidad Nacional Autónoma de México (IBT-UNAM) (Dr Paul Gaytan and M.C. Eugenio Becerra). We also thank Dr György B. Kiss from the Institute of Genetics, Szeged, and Agricultural Biotechnology Center, Gödöllő, Hungary for providing the NORK probe, Drs Miguel Lara and Georgina Hernández from CCG-UNAM for the fluorescence stereomicroscope and Dr Hilda Lomelí from IBT-UNAM for DMLB microscope facilities. This work was supported in part by DGAPA-UNAM IN204907 (CQ), DGAPA-UNAM IN221209 (RSL) and CONACyT 56631 (CQ) grants.