DELLA proteins are common components of symbiotic rhizobial and mycorrhizal signalling pathways

Abstract

Legumes form symbiotic associations with either nitrogen-fixing bacteria or arbuscular mycorrhizal fungi. Formation of these two symbioses is regulated by a common set of signalling components that act downstream of recognition of rhizobia or mycorrhizae by host plants. Central to these pathways is the calcium and calmodulin-dependent protein kinase (CCaMK)–IPD3 complex which initiates nodule organogenesis following calcium oscillations in the host nucleus. However, downstream signalling events are not fully understood. Here we show that Medicago truncatula DELLA proteins, which are the central regulators of gibberellic acid signalling, positively regulate rhizobial symbiosis. Rhizobia colonization is impaired in della mutants and we provide evidence that DELLAs can promote CCaMK–IPD3 complex formation and increase the phosphorylation state of IPD3. DELLAs can also interact with NSP2–NSP1 and enhance the expression of Nod-factor-inducible genes in protoplasts. We show that DELLA is able to bridge a protein complex containing IPD3 and NSP2. Our results suggest a transcriptional framework for regulation of root nodule symbiosis.

Introduction

The availability of nitrogen is important for plant growth and stress resistance. Legumes are able to form beneficial symbiotic associations with nitrogen-fixing rhizobial bacteria that provide nitrogen to the host plant1,2. Legumes develop a specific organ, called a nodule, which provides the proper microenvironment for ammonia generation through the reduction of molecular dinitrogen by rhizobia and nutrient exchange between both symbionts. Nitrogen fixation is energetically expensive and this symbiotic interaction is precisely regulated by the plant to adapt to environment conditions.

There are two major developmental programs in root nodule symbiosis: nodule primordia originate from root cortical cells which re-activate cell divisions, and simultaneously, root epidermal responses prepare the cell for bacterial infection2. The formation of nitrogen-fixing nodules is controlled by a host genetic programme that synchronizes these two processes3,4. The early development of rhizobial symbiosis is achieved through a chemical communication between plant and rhizobium, present in the rhizosphere. Flavonoids are released by the plant roots as a signal to rhizobia. In turn, rhizobia produce nodulation factors (Nod factors) that are recognized by the host plant to activate a common symbiotic signalling pathway required for mycorrhizal and rhizobial symbioses2,4,5. Over the past 20 years, progress has been made in understanding the genetic programme in Medicago truncatula and Lotus japonicus. Two LysM receptor kinases NFR1/LYK3, NFR5/NFP and a leucine-rich repeat receptor kinase DMI2/SYMRK are necessary for Nod factor perception6,7,8,9,10,11,12. Interestingly, it has been recently shown that NFR1/LYK3 is also involved in mycorrhizal symbiosis and is a new common symbiotic gene13. After perception at the plasma membrane, the remaining components are associated with nucleus, which are involved in both mycorrhizal and rhizobial symbioses. Two cation channels (Castor and Pollux) and three components at the core of the nuclear pore (NENA, NUP85 and NUP133) are located on the nuclear membrane14,15,16,17,18,19. These components are required for Nod-factor-induced calcium oscillations4. Perception of the calcium oscillations requires a nuclear-localized calcium and calmodulin-dependent protein kinase (CCaMK, known as DMI3 in M. truncatula)20,21. The activation of CCaMK is sufficient to induce symbiotic processes, as gain-of-function mutations in CCaMK induce nodulation in the absence of rhizobia22,23,24. CCaMK associates with and phosphorylates CYCLOPS (known as IPD3 in M. truncatula)25,26, which is essential for rhizobial and mycorrhizal colonization. Very recently, phosphorylated CYCLOPS has been shown to bind the promoter of NINand induce nodulation in the absence of rhizobia27.

Several nuclear-associated transcriptional regulators, including nodule inception (NIN)28,29, an ERF family protein (ERN)30,31, and two GRAS family proteins, Nodulation signaling pathway 1 (NSP1) and NSP2, are required for the expression of Nod-factor-induced genes and initiation of nodulation32,33. NSP1 and NSP2 form a hetero-complex that associates with promoters of Nod-factor-inducible genes, such asENOD11 and ERN1 (ref. 34). NSP1 and NSP2 were originally thought to function specifically in nodulation. However, it was recently shown thatNSP1 and NSP2 are also required for arbuscular mycorrhizal fungi (AMF)-associated lipochitooligosaccharide (LCO) signalling or mycorrhizal infection35,36,37,38. NSP1 contains a DNA-binding domain, whereas there is none in NSP2, which requires an interaction with NSP1 to associate with Nod-factor-inducible promoters34. The NSP1–NSP2 protein complex promotes the expression of NIN and ERN, two transcription factors with nodulation-specific functions. Interestingly, it has recently been shown that NSPs are not required for CYCLOPS-induced NIN expression, but they are required for CYCLOPLS-induced nodule organogenesis27. So it remains unknown how NSP1–NSP2 activity is associated with the action of CCaMK-CYCLOPS or whether the two protein complexes function independently in root nodule symbiosis.

The phytohormone gibberellic acid (GA) has been implicated in many aspects of plant biology, including seed germination, stem elongation, leaf expansion, pollen maturation and induction of flowering39,40. GA is perceived by GID1 (GA insensitive dwarf 1) and promotes the destruction of DELLA (SLR1, slender rice 1, one copy in rice genome) which is a nuclear protein that restrains the cell proliferation and expansion that drive plant growth39,40. DELLA is well known as a transcriptional suppressor that acts by directly binding DNA recognition domain of other transcript factors such as PIF3, 4 and inhibiting their transcriptional activities41,42. DELLA is an integrator of plant responses to hormones and environmental stresses, and DELLA-dependent growth restraints are advantageous in adverse environment stresses43.

DELLA family proteins are required for mycorrhizal symbiosis44,45. The mycorrhizal phenotype of della double mutant in M. truncatula shows DELLAs function at the arbuscule development stage44. Interestingly we recently showed that AMF infection and arbuscule development are both defective in the rice della mutant, slr1, suggesting an early signalling function of DELLAs in symbiosis45. This different mycorrhizal defect of mutants in rice and Medicago could be caused by genetic redundancy of DELLAs in Medicago. The role of GAs in the control of root nodule symbiosis has been investigated for over 60 years46,47. It has been shown that exogenous application of GAs inhibits the formation of infection threads and nodules, but other results show GAs can promote root nodule symbiosis in legumes48,49,50,51.

Here we show that DELLA proteins promote nodule development and infection thread formation during root nodule symbiosis. We provide evidence that DELLAs can promote CCaMK–IPD3/CYCLOPS complex formation and increase the phosphorylation of IPD3/CYCLOPS. We show that DELLAs can form a protein complex with NSP2–NSP1 and are able to bridge a protein complex containing IPD3/CYCLOPS and NSP2. We propose that DELLA proteins represent a missing link in the common symbiotic signalling pathway required for both rhizobial and mycorrhizal symbioses.

Results

GA inhibits root nodule symbiosis

It has previously been shown that gibberellins play an important role in root nodule symbiosis48,49,50. To better understand the nature of this regulation, we initiated studies to assess the effects of GA3 on nodulation efficiency in M. truncatula under our growth conditions. We grew M. truncatula cv. Jemalong line A17 plants in the soil pretreated with 0, 10−7, 10−6, 10−5 and 10−4 M GA3, concentrations of GA3 that are higher than endogenous GAs levels in plants, and quantified nodulation events at 7 and 14 days post inoculation with Sinorhizobium melilotistrain 1021 (Sm1021). As expected the shoots of GA-treated plants displayed increased shoot length and shoot fresh weight (Supplementary Fig. 1a,b). However, root length, lateral root number and root fresh weight of the plants treated with GA3 at concentrations from 10−7 to 10−5 M were not significantly changed compared with control plants (Supplementary Fig. 1c–e). Interestingly, we found that nodulation was severely impaired at 10−6–10−4 M GA3 treatment (Fig. 1a), while the shoot length increased after treated with GA3 (Fig. 1c). The generation of nodules involves a number of processes that may be regulated by GA3, including the re-initiation of cortical cell division to form a nodule primordium and the facilitation of bacterial infection through epidermal and cortical cells. We therefore assessed whether GA3 could regulate bacterial infection of the epidermis by quantifying the number of infection events and nodule development in plants grown at different concentrations of GA3. The infection frequency was also decreased with increasing concentrations of GA3 (Fig. 1b,d), suggesting that GAs regulates both nodule formation and bacterial infection independently of the shoot and root growth phenotype.

Figure 1: GA3 represses root nodulation development and rhizobial infection in M. truncatula.

Wild-type A17 plants were grown with different concentrations of GA3 and upon inoculation with S. meliloti strain 1021 harbouring a LacZ reporter and assayed for nodule development (a), rhizobial infection events (b), and plant development (c). Increasing GA3 concentration promotes plant growth but inhibits nodulation and rhizobial infection. Infection threads were counted following LacZ staining. Twelve plants were analysed for each GA3concentration. 1:mock, 2:10−7, 3:10−6, 4:10−5, 5:10−4 M GA3. (d) Representative images of infection threads visualized by staining of S. meliloti (strain 1021) expressing the LacZreporter. The left panel shows infection thread progression in to a nodule primordium. The middle panel shows that a representative image of infection thread progression was inhibited by treatment of 10−4 M GA3 and the right panel shows that a representative image of no infection thread progression in the most plant roots treated with 10−4 M GA3. Scale bars correspond to 2 cm (c) and 100 μm (d). These experiments were repeated three times with similar results. Error bars are standard error. The asterisk indicates a significant decrease to the control with Student’s t-test (**P<0.01).

Full size image

DELLAs are required for root nodule symbiosis

Gibberellin is perceived by GID1 and promotes the interaction of GID1 and DELLA, which in turn leads to the polyubiquitination and subsequent degradation of DELLA by the 26s proteasome in the nucleus39,40. Three DELLA genes and two highly homologous genes were found in the M. truncatula genome database Mt3.5 and Mt4.0 (Supplementary Fig. 2)44. When we transformed functional MtDELLA-GFP (Supplementary Fig. 3) driven by 35S promoter into M. truncatulausing hairy root transformation, no green fluorescent protein (GFP) signal was detected in the root, but occasionally very weak GFP signals were detected only in root tip cells, suggesting that the amounts of DELLA proteins are tightly regulated in M. truncatula. Consistent with the degradation of DELLAs by GA treatment in other plant species, GA3treatment induced the degradation of MtDELLA2 and MtDELLA3 in the nucleus (Supplementary Fig. 3a). Under non-symbiotic conditions, the expression of the three DELLAs was detected in epidermal cells, cortical cells and stele in the rhizobial infection zone (closed to root apical meristem) and elongation zone (Fig. 2a,b). Consistent with our results, expression of DELLA1 and DELLA2 in the epidermal cells (root hair cells) was also found in transcriptional data (http://mtgea.noble.org/v3/blast_result.php?s=226645). Further analysis showed that MtDELLA3 was upregulated 6 h post Nod factor treatment and that MtDELLA1 and MtDELLA2 were induced 7 day post inoculation (d.p.i.) with Sm1021, suggesting a role of DELLAs in root nodule symbiosis (Supplementary Fig. 4a,b). To assess whether DELLAsfunction in root nodule symbiosis, we generated transformed hairy roots expressing an MtDELLA hairpin allowing RNA-mediated interference of MtDELLAs. We used three different combination of RNA interference (RNAi) targeted three MtDELLAs and observed that expression levels of MtDELLAs were reduced in these RNAi plants (Supplementary Fig. 4e). The total nodule number was significantly decreased in MtDELLAs RNAi plants compared with the empty vector control (Fig. 2c), indicating that DELLA proteins are positive regulators in root nodule symbiosis.

Figure 2: MtDELLAs promote root nodule symbiosis.

(a) Whole-root GUS activity of pDELLA1:GUS, pDELLA2:GUS, pDELLA3:GUS, pNSP2:GUS, pIPD3:GUS are shown. Scale bars correspond to 2 mm. The black bars indicate that the region is selected for the transversal section in b. (b) Thin transversal sections (70 μm) of pDELLA1:GUS, pDELLA2:GUS, pDELLA3:GUS, pNSP2:GUS, pIPD3:GUS roots are shown. Scale bars correspond to 100 μm. (c) Numbers of nodules on MtDELLAs RNAi plant roots 4 weeks post inoculation with S. meliloti. n⩾8, where n denotes the number of plants. (d) qRT-PCR analysis of MtDELLAs transcript levels. The expression levels of MtDELLAs were detected in wild-type (R108) and della1, della2 and della3 mutants. Expression levels were normalized against the reference gene Elongation factor 1-alpha (EF1-α). The RNA was extracted from six individual plants of R108 or mutants. (e,f) Scores of nodule number in della2/della3 double mutants (e) and della1/della2/della3 triple mutants (f). (g) Quantification of infection threads in della double and triple mutants. The infection thread (IT) was visualized by staining of S. meliloti expressing the LacZ reporter and was counted at 7 d.p.i. n⩾13, where ndenotes the number of plants. Relative ITs density indicates the number of infection threads per root length per plant, and they were normalized against the wild-type (R108). (h) Nodule development is impaired in della1/della2/della3 triple mutant. There is no nodule formed in 12 plants out of 21 della1-1/della2/della3-1 triple mutants. Occasionally pink nodules were formed in della triple mutants. Scale bars correspond to 500 μm. (i) Nodule sections of the pink nodule and the white nodule in the della triple mutant, nodules were stained with toluidine blue. Scale bars correspond to 200 μm. This is a representative experiment repeated three times in c and d, and twice in e, f and g. Error bars represent s.d. The asterisk indicates a significant decrease relative to wild type or vector control with Student’s t-test (*P<0.05; **P<0.01).

Full size image

To further confirm the RNAi results, M. truncatula lines containing Tnt1insertions in MtDELLA1, MtDELLA2 and MtDELLA3 were obtained from a mutant population generated at the Samuel Roberts Noble Foundation (Supplementary Fig. 4) (http://medicago-mutant.noble.org/mutant/database.php) and the transcription levels ofMtDELLA1, MtDELLA2 and MtDELLA3 were greatly reduced in the dellamutants (Fig. 2d). Given genetic redundancy of DELLAs in Medicago,della2/della3 double mutants were generated and showed significantly decreased nodule number at 21 d.p.i. compared with its wild type (R108) (Fig. 2e). However, the root length of double mutant was not different from wild type (Supplementary Fig. 5). While few matured nodules were formed in the della2/della3 double mutants (Fig. 2e). We further generated the della triple mutants and observed very few mature pink nodules in della1/della2/della3 plants (Fig. 2f–i). Intriguingly, no nodules were formed in 12 out of 21 della1-1/della2/della3-1 triple mutants. Further analysis revealed the number of infection threads was greatly reduced in della triple mutants and the pink nodule formed on dellatriple mutants appeared to be aberrant (Fig. 2f–i). These results indicated that MtDELLAs are essential for root nodule symbiosis.

Gene induction by Nod factors is dependent on DELLAs

ENOD11 represents one of the earliest marker genes for response to Nod factor and rhizobia. The induction of ENOD11 requires the Nod factor signal transduction pathway and as such provides markers for the regulation of this signalling pathway52. To assess the effects of GA3on Nod-factor-induced gene expression, we used M. truncatula plants stably transformed with β-glucuronidase (GUS) driven by the ENOD11promoter. Seven-day-old plants were transferred into liquid BNM medium containing 10−7, 10−6, 10−5 and 10−4 M GA3. Following 6 h of GA3treatment, 1 nM Nod factor was added to the medium, and after 6 h of Nod factor treatment, the roots were used for GUS staining. Plants without Nod factor treatment only showed GUS activity in root caps, while Nod factor treatment activates ENOD11 in a specific region of the root (Fig. 3a). Pre-treatment with 10−4 M GA3 led to reduced induction of ENOD11 (Fig. 3a,b), but co-treatment with GA3 did not reduce induction of ENOD11 (Fig. 3a,b), suggesting that DELLAs may act at a very early stage in the signal transduction pathway. Consistent with the GA treatment data, we found that the induction of Nod-factor-induced genes was impaired in della2/della3 double mutants (Fig. 3c–g), suggesting that DELLAs may directly regulate the expression of early nodulin genes.

Figure 3: Nod-factor-induced gene expression is promoted byMtDELLAs.

(a) The effect of GA3 on Nod-factor-induced gene expression was assayed using transgenic plants expressing pENOD11:GUS. Note: 6 h pre-treatment of GA3 suppresses expression ofENOD11 (right panel), but GA3 co-treatment with Nod factor does not suppress expression ofENOD11 (left panel). 1:mock, 2:10−4, 3:10−5, 4:10−6, 5:10−7 M GA3. Scale bars correspond to 1 cm. (b) Real-time PCR results revealed that GA3 suppresses expression of Nod-factor-induced genes. Treatment with 1 nM Nod factor for 6 h induced expression of ENOD11,VAPYRIN, NIN, ERN and RIP10 and the induction was reduced upon pre-treatment with 10−5 M GA3 for 6 h. (c–g) Real-time PCR results revealed that MtDELLAs required for induction of nodulin genes. Seven-day-old della2/della3 double mutants and R108 were treated with 1 nM Nod factor or mock (BNM medium) for 6 h. In wild type the nodulin genes were induced, however these genes could not be induced in della2/della3 double mutants after Nod factor treatment. These experiments were repeated three times with similar results, one representative experiment was shown. Error bars are standard error (n=3). The asterisk indicates a significant induction relative to control with Student’s t-test (*P<0.01).

Full size image

A gain-of-function (GOF) mutation in a CCaMK or a gain-of-function of a cytokinin receptor (CRE1/SNF2) induces spontaneous cortical cells division and leading to spontaneous nodule formation in the absence of rhizobia2. The gain-of-function of CCaMK (1–311aa) or GOF-CRE1 was introduced into hairy roots to determine the effect of GA3-coupled DELLA degradation on the cortical cell division. The transgenic roots treated with GA3 showed a significant reduction in the number of spontaneous nodules in GOF-CRE1 lines (Supplementary Fig. 6a), suggesting that DELLAs are required for spontaneous nodule formation induced by cytokinin and DELLAs may function at the nodule development stage. Spontaneous nodule formation was also inhibited by GA3 treatment in overexpression of GOF-CCaMK plants (Supplementary Fig. 6b). Intriguingly, the spontaneous nodule formation induced by CRE1-GOF or CCaMK1-311-GOF was fully blocked indella2/della3-1 double mutants (Fig. 4), suggesting that DELLA proteins act downstream of CCaMK.

Figure 4: DELLAs promote spontaneous nodule formation induced by CRE1 and CCaMK.

Spontaneous nodule formation induced by gain-of-function CRE1 and CCaMK1-311 was blocked in della double mutants. The nodule number was counted 5 weeks after hairy root transformed plants being transferred to sterile vermiculite with perlite (n⩾6, where ndenotes the number of plants).

Full size image

DELLAs are able to interact with IPD3

Since DELLA proteins are involved in both root nodule and mycorrhizal symbioses, we propose that DELLAs may interact with the key components of the common signalling pathway. We observed that DELLA proteins MtDELLA1, MtDELLA2 and MtDELLA3 interacted with IPD3, but not CCaMK and NSP1 in a yeast two-hybrid (Y2H) assay, although CCaMK and NSP1 proteins were expressed in yeast (Fig. 5a;Supplementary Fig. 7c,d). MtDELLAs and IPD3 are also able to interact in in vitro pull-down assays (Fig. 5b) and bimolecular fluorescence complementation (BiFC) assays in Arabidopsis protoplasts (Fig. 5c). A strong fluorescence signal was observed in the nucleus of Arabidopsismesophyll protoplasts co-transformed with IPD3-nYFP and MtDELLAs-cYFP. Protoplasts transformed with MtDELLA-nYFP and the control vector or with IPD3-cYFP and the control vector were used as controls, which showed no positive signal (Fig. 5c). Taken together, these results show that MtDELLAs are able to interact with IPD3.

Figure 5: Interactions between MtDELLAs and MtCYCLOPS/IPD3.

(a) Yeast two-hybrid assays between IPD3 and MtDELLA1, MtDELLA2 or MtDELLA3. Yeast cells carrying different combinatory constructs are listed on the left. Serial dilutions (10 times) of yeast cells expressing the indicated proteins from the pGADT7 (AD) and pGBKT7 (BD) vectors were plated onto SD/-Leu-Trp (-LT) medium or SD/-Leu-Trp-Ade-His (-LTAH) medium with 30 or 60 mM 3-amino-1,2,4-triazole (3AT). (b) Pull-down assays between IPD3 and MtDELLA1, MtDELLA2 or MtDELLA3. MBP-IPD3 fusion protein but not MBP alone bound with HIS-tagged MtDELLA1, MtDELLA2 or MtDELLA3. (c) Detection of protein–protein interactions in Arabidopsis protoplast by BiFC. YFP fluorescence of leaves co-transformed with MtDELLAs-nYFP (N-terminal half of YFP) and IPD3-cYFP (C-terminal half of YFP). No protein interactions were detected in the other three combinations: MtDELLAs-nYFP--cYFP, IPD3-cYFP--nYFP and cYFP--nYFP. Scale bars, 20 μm.

Full size image

DELLAs can form a protein complex with IPD3 and CCaMK

It has been shown that CCaMK interacts with and phosphorylates IPD3/CYCLOPS26. To assess the relationship between CCaMK, IPD3 and MtDELLAs, we performed a yeast three-hybrid (Y3H) assay with the fusion proteins CCaMK-GAL4-DNA-binding domain (GAL4-DBD) and/or IPD3-GAL4 activation domain (GAL4-AD). Additive activation of the reporter was observed when MtDELLA proteins were present (Fig. 6a), suggesting that MtDELLA may form a complex with CCaMK and IPD3. We did not observe the interaction of CCaMK and MtDELLAs in the yeast two-hybrid assay (Supplementary Fig. 7). To further assess whether MtDELLA, CCaMK and IPD3 can form a protein complex, we co-expressed HIS-DELLA, CCaMK-S-tag and HIS-IPD3 in Escherichia coli and found that CCaMK could co-immunoprecipitate both IPD3 and DELLA2 (Fig. 6b), suggesting that MtDELLA, CCaMK and IPD3 are able to form a complex.

Figure 6: CCaMK–IPD3–DELLA protein complex formation.

(a) MtDELLAs increased the interaction between CCaMK and IPD3 in yeast three-hybrid assays. Data are presented as β-galactosidase activity. (b) CCaMK co-immunoprecipitates with IPD3 and DELLA2 in E. coli. CCaMK, IPD3 and MtDELLA2 are co-expressed in E. coli. (c) The effect of MtDELLAs on phosphorylation of IPD3 by CCaMK in vitro phosphorylation assays. The left histogram shows quantitative intensities of IPD3 phosphorylation calculated by ImageJ. The relative intensities were normalized against the control (bar1). (d) MtDELLA2 increased the phosphorylation of IPD3 by CCaMK. An increasing amount of MtDELLA2 was added to the reaction mix. The left histogram shows quantitative intensities of IPD3 phosphorylation calculated by ImageJ. The relative intensities were normalized against the bar1. Autoradiographs show the corresponding phosphorylated IPD3. Autoradiographs of kinase assays (32P) (upper images) and Coomassie staining of the gels (lower image) in c and d. This is a representative experiment that was repeated twice in cand d. (e) CCaMK increased interaction of MtDELLA and IPD3 in yeast three-hybrid assay. Data are presented as β-galactosidase activity. (f) M. truncatula ipd3-2 roots were transformed with IPD3-S50A-S155A, IPD3-S50D-S155D and IPD3. The pink nodules were formed on ipd3-2 mutants contained IPD3 and IPD3-S50D-S155D upon 35 day post inoculation with Sm1021. Scale bars correspond to 500 μm. (g) Yeast three-hybrid assay showed sites S50 and S155 of IPD3 are critical for interaction of IPD3 and MtDELLA. Alanine replacement of S50 and S155 of IPD3 abolished the interaction with MtDELLA. Data are presented as β-galactosidase activity. Results represent the means of three experiments ina, e and g. Error bars represent standard error. The asterisk indicates a significant increase relative to the control with Student’s t-test in a, e and g (*P<0.01).