Crosstalk between the nodulation signaling pathway and the autoregulation of nodulation in Medicago truncatula

Authors

  • Isabel M. L. Saur,

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, Plant Science Division, Research School of Biology, The Australian National University, Canberra 0200, Australia
    Search for more papers by this author
  • Marie Oakes,

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, Plant Science Division, Research School of Biology, The Australian National University, Canberra 0200, Australia
    Search for more papers by this author
  • Michael A. Djordjevic,

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, Plant Science Division, Research School of Biology, The Australian National University, Canberra 0200, Australia
    Search for more papers by this author
  • Nijat Imin

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, Plant Science Division, Research School of Biology, The Australian National University, Canberra 0200, Australia
    Search for more papers by this author

Author for correspondence:
Nijat Imin
Tel: +61 2 6125 5099
Email: nijat.imin@anu.edu.au

Summary

  • A subset of CLAVATA3/endosperm-surrounding region-related (CLE) peptides are involved in autoregulation of nodulation (AON) in Medicago truncatula (e.g. MtCLE12 and MtCLE13). However, their linkage to other components of the AON pathways downstream of the shoot-derived inhibitor (SDI) is not understood.
  • We have ectopically expressed the putative peptide ligand encoding genes MtCLE12 and MtCLE13 in M. truncatula which abolished nodulation completely in wild-type roots but not in the supernodulating null mutant sunn-4. Further, root growth inhibition was detected when MtCLE12 was ectopically expressed in wild-type roots or synthetic CLE12 peptide was applied exogenously.
  • To identify downstream genes, roots of wild-type and sunn-4 mutant overexpressing MtCLE12 were used for quantitative gene expression analysis. We found that, in 35S:MtCLE12 roots, NODULE INCEPTION (NIN, a central regulator of nodulation) was down-regulated, whereas MtEFD (ethylene response factor required for nodule differentiation) and MtRR8 (a type-A response regulator thought to be involved in the negative regulation of cytokinin signaling), were up-regulated. Moreover, we found that the up-regulation of MtEFD and MtRR8 caused by overexpressing MtCLE12 is SUNN-dependent.
  • Hence, our data link for the first time the pathways for Nod factor signaling, cytokinin perception and AON.

Introduction

Legumes can establish symbioses with nitrogen-fixing soil bacteria (generically termed rhizobia) resulting in lateral organs called root nodules. Root nodules enable legumes to grow in nitrogen-poor soils and this symbiosis imparts an important role for legumes in sustainable agricultural production worldwide (Ferguson et al., 2010). To trigger root nodule formation, host roots secrete flavonoids and other compounds that attract rhizobia to colonize the rhizosphere and root surfaces, inducing nodulation (nod) genes (Redmond et al., 1986). The induction of nod genes results in the synthesis and secretion of lipochitin oligosaccharides called Nod factors (Denarie et al., 1996). In many legumes, Nod factors trigger one systemic and two root-specific pathways which control nodule initiation and development as well as nodule numbers.

One of the root-specific pathways controlling nodulation is mediated by Nod factors which interact with LysM kinase complexes (Stougaard, 2000). Nod factor signal transduction is mediated by the symbiosis receptor kinase to trigger calcium spiking, calcium/calmodulin-dependent protein kinase activity and the activation of the transcription factors NODULATION SIGNALING PATHWAY (NSP1, NSP2) and NODULE INCEPTION (NIN) (Schauser et al., 1999; Catoira et al., 2000; Oldroyd & Long, 2003; Gleason et al., 2006; Oldroyd & Downie, 2008). Recently, NSP1 and NSP2 were shown to form homo- and hetero-complexes that bind to the promoters of the NIN and ENOD11 genes (Hirsch et al., 2009). NSPs and NIN are necessary for the dedifferentiation of the cortical cells required for nodule initiation and formation (Crespi & Frugier, 2008).

A second endogenous root-specific pathway, also essential for cortical cell dedifferentiation, is controlled by the plant hormone cytokinin (Murray et al., 2007; Tirichine et al., 2007). This pathway also leads to activation of NSP1/2 and NIN expression and involves the CRE1 histidine kinase receptor and primary type A response regulators, which have been implicated in the negative feedback regulation of cytokinin signaling (Lohar et al., 2004; Ferreira & Kieber, 2005; Gleason et al., 2006; Gonzalez-Rizzo et al., 2006; Doerner, 2007; Murray et al., 2007; Tirichine et al., 2007; Frugier et al., 2008; Vernie et al., 2008).

Systemic feedback regulation known as autoregulation of nodulation (AON) also controls nodule number. Grafting experiments show that, during AON, the earliest formed nodules suppress further nodulation through root-to-shoot and shoot-to-root communication (Delves et al., 1986; Caetano-Anolles & Gresshoff, 1991; Searle et al., 2003). Recent studies show the involvement of CLAVATA3/endosperm-surrounding region-related (CLE) peptides in AON in Lotus japonicus (LjCLE-RS1 and -RS2; Okamoto et al., 2009), Medicago truncatula (MtCLE12 and MtCLE13; Mortier et al., 2010) and soybean (GmNIC1; Reid et al., 2011). Overexpressing these specific CLE peptides suppresses nodule formation in wild-type plants (Okamoto et al., 2009; Mortier et al., 2010; Reid et al., 2011), whereas in L. japonicus this effect is abolished when overexpressing LjCLE-RS1 and -RS2 in a mutant background (har1-4) bearing a missense mutation in the exodomain of the XI subgroup of leucine-rich repeat receptor-like kinase (LRR-RLK) (Nishimura et al., 2002; Okamoto et al., 2009). However, the overexpression of MtCLE13 in the M. truncatula sunn-1 line, bearing a missense mutation in the endodomain of this gene, results in reduced nodulation (Mortier et al., 2010). HAR-1 and SUNN are orthologous to Arabidopsis CLAVATA1 (Clark et al., 1997; Krusell et al., 2002; Schnabel et al., 2005) and it would be expected that they play similar roles in AON, so this potential discrepancy in the behavior of sunn-1 and har1-4 has not been explained experimentally. It has been shown that the activation of MtCLE12, MtCLE13, LjCLE-RS1 and LjCLE-RS2 occurs downstream of NIN shortly after nodule primordium initiation (Okamoto et al., 2009; Mortier et al., 2010). In Arabidopsis, similar CLE peptides have been found to be involved in root elongation, vascular differentiation and other developmental regulation especially in the shoot apical meristem (Strabala et al., 2006).

Using reverse genetic approaches and gene expression analysis, we have linked, for the first time, the Nod factor and cytokinin signaling pathways with AON. We also show that overexpressing MtCLE12 inhibits root growth and alters the expression of the auxin-responsive reporter GH3:GUS in the root.

Materials and Methods

Plant materials

Seeds of Medicago truncatula Gaertn. cv Jemalong genotype A17 wild-type, sunn-4 mutant and M. truncatula 2HA line carrying a GH3 promoter-GUS reporter fusion gene (GH3:GUS) were germinated and inoculated as previously described (Holmes et al., 2008).

Exogenous application of synthetic MtCLE12

The CLE12 and CLE13 peptides were synthesized at the Biomolecular Resource Facility, The Australian National University. The 12-amino-acid (aa) peptides corresponding to the conserved domain of MtCLE12 (RLSPGGPNHIHN) and MtCLE13 (RLSPAGPDPQHN) were synthesized using FMOC (fluorenylmethyloxycarbonyl) chemistry and solid-phase peptide synthesis with a Symphony/Multiplex multiple peptide synthesizer (Protein Technologies, Inc. Tucson, AZ, USA). The peptides were purified via reverse-phase high-performance liquid chromatograhy and the quality checked by MALDI-TOF/TOF mass spectrometry (ABI 4800; Applied Biosystems, Foster City, CA, USA). CLE12 and CLE13 were used at a final concentration of 10 μM in Fåhraeus-medium (Kusumawati et al., 2008). The root length of wild-type plants was measured 4 d after transfer to Fåhraeus-medium containing the synthetic peptide. For nodulation assays, plants were inoculated with Sinorhizobium meliloti strain 1021 (S. meliloti) or S. medicae 1022 (Terpolilli et al., 2008) 4 d post-germination and nodule numbers were counted at 21 d postinoculation (dpi).

Agrobacterium rhizogenes-mediated hairy root transformation

Polymerase chain reaction fragments corresponding to the full-length open reading frames of MtCLE26 (a novel CLE of unknown function that is similar to Arabidopsis CLE20), MtCLE12 and MtCLE13 were amplified from M. truncatula cDNA and cloned into the pK7WG2D vector (Karimi et al., 2002). The respective constructs were transformed into A. rhizogenes strain Arqua1 grown on Luria Bertani medium with the appropriate antibiotics (Bertani, 1951). After germination, the seed coat was peeled off and the root radicle tip removed 3 mm from the tip. The cut roots were coated with A. rhizogenes containing the respective vectors. Seedlings were grown on 15-cm-diameter plates with sloped Fåhraeus-medium containing 25 mg l−1 kanamycin. The plates were incubated at 20°C for 1 wk and transferred to 25°C (16 h photoperiod and photon flux density of 150 μmol m−2 s−1). Transgenic roots were identified by the presence of green fluorescent protein (GFP) with an Olympus SZX16 stereomicroscope equipped with a GFP filter unit (Model SZX2-FGFPA; Shinjuku-ku, Tokyo, Japan).

Assessment of nodule numbers and harvesting of plant material for quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis

Hairy roots derived from A17 or sunn-4 plants developed nodules within 10 dpi with S. meliloti. To ensure that the overexpression of MtCLE12 or MtCLE13 suppresses nodule formation on A17 plants, nodule numbers were assessed at 21 dpi. sunn-4 has a short root phenotype (Schnabel et al., 2005). Hence, to minimize possible confounding growth inhibitory affects, in sunn-4 the effect of MtCLE12 overexpression on the nodule numbers was assessed at 10 dpi. To match the time for the assessment of nodule numbers with the gene expression analysis, roots were harvested at the same time as their nodulation was assessed.

RNA extraction, cDNA synthesis and qRT-PCR analysis

RNA extraction, cDNA synthesis and qRT-PCR analysis were performed as previously described (Kusumawati et al., 2008). Normalization was conducted by calculating the differences between the CT of the target gene and the CT of MtUBQ10 (MtGI accession number TC161574). Relative gene expression levels were calculated by plotting against the lowest expressed sample after the normalization. For M. truncatula A17, two biological (independent root samples), two experimental (independent cDNA synthesis) and three technical repeats (independent real-time PCR) were done. For sunn-4, two biological and three technical repeats were done. Table S1 (Supporting Information) lists the primers used.

β-Glucuronidase (GUS) staining and sectioning

β-Glucuronidase (GUS) activity was localized in transgenic hairy roots carrying GH3:GUS (Vitha et al., 1995). Staining and sectioning were performed three times, each time taking the roots of four plants, and similar results were obtained each time. Staining was examined with a Nikon SMZ1500 stereomicroscope and photographed with a mounted Digital Sight DS-Ri1 camera (Nikon Inc., Melville, NY, USA). For sectioning, roots were embedded in 3% DNA grade agarose (Progen Biosciences, Brisbane, Australia) and sectioned at 120 μm thickness (1000 Plus; Vibratome Company, St Louis, MO, USA). The sections were mounted on glass slides in water, examined with a Leica DMBL microscope (Leica Microsystems Inc., Deerfield, IL, USA) and photographed using a mounted SPOT RT slider CCD camera (Diagnostic Instruments, Sterling Heights, MI, USA).

Results

Suppression of nodulation by ectopic expression of MtCLE12 and MtCLE13 requires SUNN

MtCLE12 and MtCLE13 expression is limited to roots forming nodules, whereas MtCLE26 is expressed in various tissues of M. truncatula (Benedito et al., 2008, Figs 1a,d, S1). The microarray data by Benedito et al. (2008) show different expression patterns for MtCLE12 compared with those of Mortier et al. (2010) detected by qRT-PCR. Mortier et al. (2010) also showed a SUNN-dependent negative effect of MtCLE13 overexpression on nodulation. We therefore overexpressed MtCLE12, MtCLE13 and MtCLE26 in the roots of wild-type (A17) and the sunn-4 mutant, a functional ortholog of har1-4. qRT-PCR analyses confirmed the overexpression of the respective genes (> 45-fold, n > = 6, < 0.001, Student’s t-test). Consistent with the results of Mortier et al. (2010), no nodules were detected on A17 roots carrying 35S:MtCLE12 or 35S:MtCLE13. Roots transformed with the empty vector showed 1.7 ± 0.8 nodules per plant and roots carrying 35S:MtCLE26 showed 2.7 ± 1.2 nodules (21 dpi,  11, < 0.001, Fig. 1b,e). In contrast to results overexpressing MtCLE13 in a sunn-1 background (Mortier et al., 2010), nodule numbers were not significantly different when MtCLE12 or MtCLE13 overexpressed in the sunn-4 background (10 dpi,  10, < 0.05, Fig. 1c,f). These results demonstrate that the inhibition of nodule formation caused by MtCLE12 and MtCLE13 overexpression is SUNN-dependent and that MtCLE26 overexpression causes no significant effect on nodule numbers.

Figure 1.

Expression level of MtCLE12 in different tissues and the effect of the ectopic expression of the MtCLE12 and MtCLE13 in Medicago truncatula wild-type and sunn-4 mutant plants. (a, d) Expression level of MtCLE12 in various tissues of M. truncatula A17. MtCLE12 can only be detected in the roots that form nodules. dpi, days postinoculation with Sinorhizobium meliloti. Root-0 dpi, nitrogen starved for 4 d before inoculation. (b, c) Comparison of M. truncatula A17 and sunn-4 roots overexpressing MtCLE26, MtCLE12 or MtCLE13 together with the green fluorescent protein as a visible marker. (e) Wild-type plants transformed with the empty vector (pK7WG2D) show, on average, 1.7 ± 0.8 nodules per plant; 35S:MtCLE26 roots show 2.7 ± 1.2 nodules per plant. No nodules were detected on 35S:MtCLE12 and 35S:MtCLE13 transgenic roots (= 12). (f) The overexpression of MtCLE12, MtCLE13 and MtCLE26 did not have any significant effect on nodule numbers in sunn-4 mutant plants. The control showed 4 ± 0.5 nodules per root; 35:MtCLE26 showed 4.6 ± 1.6 nodules per root; 35S:MtCLE12 showed 4.3 ± 0.9 nodules per root; and 35S:MtCLE13 showed 3.8 ± 0.5 nodules per root ( 10). Nodule numbers were assessed at 21 dpi for A17 and at 10 dpi for sunn-4 (refer to the Materials and Methods section for the rationale). Error bars, ± SD. Value denoted with *** are significantly different from the control (pK7WG2D) at the < 0.001 level (Student’s t-test). (a, d) Values are levels of Affymetrix probe signal based on microarray data from the M. truncatula Gene Expression Atlas (Benedito et al., 2008). Bar, 1 mm (b, c).

MtCLE12 overexpression or the exogenous application of its synthetic peptide inhibits root elongation

An inhibition of root elongation was observed when MtCLE12 was overexpressed in A17 roots (recorded at 28 d post-transformation; Fig. 2b,f). Roots overexpressing MtCLE12 were 0.7 times shorter ( 49, < 0.01) than control hairy roots (Fig. S2). The supplementation of synthetic 12-aa CLE12 peptide (corresponding to the CLE domain of MtCLE12; Fig. S3) also inhibited root growth. The root length was 0.4 times shorter (= 18, < 0.001) than control roots (Figs 2a,e, S2). Root growth inhibition was not detected by Mortier et al. (2010) when MtCLE12 was overexpressed or when synthetic 13 aa CLE12 was applied exogenously. In contrast to the results obtained for MtCLE12, exogenous application of synthetic 12-aa CLE13 peptide or the ectopic expression of MtCLE13 did not affect root growth (= 17, > 0.05 data not shown).

Figure 2.

The effect of exogenous application of CLE12 peptides and oxerexpression of MtCLE12 on root growth and GH3 expression. (a, e) Medicago truncatula A17 28 d post-transformation: (a) transformed with empty vector pK7WG2D as a control; (e) transformed with 35S:MtCLE12. Overexpressing MtCLE12 significantly reduces root length compared with the control. (b, f) M. trunactula A17 wild-type grown on Fåhraeus-medium: (b) without supplement; (f) with addition of the 12-amino-acid CLE12 at 10 μM. The addition of CLE12 peptide significantly reduced primary root growth. (c, d, g, h). Roots and root sections (120 μm thick) of M. truncatula GH3:GUS, 28 d post-transformation; (c, d) transformed with pK7WG2D; (g, h) transformed with 35S:MtCLE12. Bars: 10 mm (a, b, e, f); 250μm (c, g); 100 μm (d, h).

Ectopic expression of MtCLE12 alters the pattern of GH3:GUS expression in roots

To assess whether the root growth inhibition induced by MtCLE12 overexpression results from disturbing auxin distribution, 35S:MtCLE12, 35S:MtCLE26 and the empty vector were transformed into M. truncatula 2HA carrying the GH3 promoter-GUS reporter fusion. GH3 is an early auxin-responsive gene and its expression is thought to reflect auxin distribution (Mathesius et al., 1998). Transgenic roots overexpressing MtCLE12 showed increased GH3:GUS activity in the phloem, endodermis and cortex (Fig. 2g,h). In roots transformed with the empty vector, GH3:GUS is detected in the vascular bundles only (Fig. 2c,d). These results were confirmed by qRT-PCR; M. truncatula 35S:MtCLE12 roots showed an 11-fold increase in GH3 expression compared with roots transformed with the empty vector (= 6, < 0.01, Fig. S4). Transgenic roots carrying 35S:MtCLE26 showed GH3:GUS expression similar to the control (data not shown).

Crosstalk between the nodulation signaling pathway and the AON

Mortier et al. (2010) demonstrated that MtCLE12 expression is dependent on NIN. However, using qRT-PCR analysis, we found that the NIN expression was 9.6-fold down-regulated (21 dpi; = 12, < 0.01) in inoculated 35S:MtCLE12 roots compared with roots transformed with the empty vector (Fig. S5), suggesting a negative feedback regulation of NIN via MtCLE12 induction. Type A response regulators have been implicated in the negative feedback regulation of cytokinin signaling and define a common pathway with NIN (Lohar et al., 2004; Ferreira & Kieber, 2005; Gleason et al., 2006; Gonzalez-Rizzo et al., 2006; Doerner, 2007; Murray et al., 2007; Tirichine et al., 2007; Frugier et al., 2008; Vernie et al., 2008). Since MtRR8 shows a similar expression pattern to MtCLE12 in root samples (Benedito et al., 2008; Figs 1, S6) we hypothesized that MtRR8 acts downstream of MtCLE12. Indeed, qRT-PCR analysis showed a sixfold increase (21 dpi; = 12, < 0.001) in the relative expression of MtRR8 in A17 roots carrying 35S:MtCLE12 compared with the empty vector control (Fig. 3a). Vernie et al. (2008) showed that the type A primary response regulator MtRR4 is controlled by MtEFD, a transcription factor, belonging to the ethylene response factor V group. The non-sense mutant efd-1 produces more nodules than A17 plants, whereas MtEFD overexpression negatively regulates nodulation (Vernie et al., 2008). We therefore also analyzed the expression of MtEFD and found a 2.3-fold increase (21 dpi; = 12, < 0.01) of MtEFD in A17 roots carrying 35S:MtCLE12 compared with the roots transformed with the empty vector (Fig. 3b). Collectively, these results suggest that MtEFD and type A response regulators such as MtRR4 and MtRR8 function downstream of MtCLE12 peptide signaling during AON.

Figure 3.

MtRR8 and MtEFD expression levels in the transgenic roots of Medicago truncatula A17 and sunn-4 overexpressing MtCLE12. (a, b) Expression of MtRR8 and MtEFD in the wild-type A17 transgenic roots at 21 dpi (d postinoculation with Sinorhizobium meliloti). Compared with the empty vector controls, transgenic roots overexpressing MtCLE12 induced sixfold (= 12, < 0.001) and 2.3-fold up-regulation (= 12, < 0.01) of MtRR8 and MtEFD, respectively (light gray bars, wild-type). (c, d) Expression of MtRR8 and MtEFD in sunn-4 background at 10 dpi. The expression of MtRR8 in transgenic roots overexpressing MtCLE12 was not significantly different (= 6, P > 0.05) from control roots, whereas the expression of MtEFD in roots overexpressing MtCLE12 was down-regulated 2.3-fold (= 6, < 0.01) (dark gray bars, sunn-4). Error bars, ± SD of biological and technical repeats combined. Values denoted with ** or *** differ significantly from the empty vector control at the P < 0.01 and < 0.001 levels, respectively (Mann–Whitney test and Student’s t-test).

In order to investigate the involvement of type A response regulators and MtEFD in AON, we determined the expression of these genes in the sunn-4 background overexpressing MtCLE12. By day 10, sunn-4 plants clearly developed nodules and were therefore harvested for expression analysis. In contrast to the A17 roots carrying 35S:MtCLE12 (21 dpi), the relative expression of MtRR8 was not significantly different when MtCLE12 was overexpressed in a sunn-4 background (10 dpi, = 6, > 0.05, Mann–Whitney test and Student’s t-test) whereas the expression of MtEFD was reduced 2.3-fold (10 dpi, = 6, < 0.01, Mann–Whitney test and Student’s t-test) compared with the empty vector control (Fig 3c,d). These results do not show whether MtEFD is actually required for the up-regulation of type A response regulators by overexpressing MtCLE12. To assess this, we overexpressed MtCLE12 in efd-1. Unfortunately, we found that plants transformed with 35S:MtCLE12 did not grow hairy roots, while 90% of the plants transformed with the empty vector did (= 50). We conclude that this is most likely the result of the combined effect of the mutation in MtEFD and the negative effect of MtCLE12 on root growth.

Discussion

The lack of nodulation in A17 roots carrying 35S:MtCLE12 and 35S:MtCLE13 confirms the results found by Mortier et al. (2010). We found that this effect was completely abolished in the sunn-4 background. The low nodule numbers in our experimental system are most likely the result of the Sinorhizobium strain used. S. meliloti 1021 inefficiently nodulates M. truncatula roots grown on agar-based medium compared with other more highly efficient strains such as S. medicae 1022 (Fig. S7; van Noorden et al., 2006; Terpolilli et al., 2008). Allelic differences between sunn-4 and sunn-1 most likely explain the differing results obtained from overexpressing MtCLE12 and MtCLE13. The sunn-4 mutation produces a truncated protein and is more functionally equivalent to har1-4 than to sunn-1 (Krusell et al., 2002; Schnabel et al., 2005). By contrast, we found that overexpressing MtCLE26 had no significant effect on nodulation in either the wild-type or sunn-4 mutant, indicating that functional differences exist between the nodule-specific CLE peptides (derived from MtCLE12, MtCLE13, LjCLE-RS1/RS2 or GmNICI; Fig. S3) and other CLE peptides. The ectopic expression of MtCLE12 or the addition of synthetic 12-aa CLE12 peptide inhibited root elongation, indicating that this 12-aa CLE12 peptide is biologically active. By contrast, nodulation was not affected by the exogenous application of the synthetic CLE12 peptide (data not shown). This may be explained by the inability of the exogenously applied CLE12 to reach its target site(s) where AON receptors are located. Alternatively, the 12-aa synthetic CLE12 might not bind to the AON receptor(s), whereas it may interact nonspecifically with receptors in the root apical meristem to inhibit root growth. The final forms of the active peptides and their putative receptors remain to be determined.

Roots transformed with 35S:MtCLE12 showed increased GH3 expression compared with controls, which has not been shown before. This suggests a role for MtCLE12 in modulating auxin signaling or sensitivity. The local increase of GH3 expression may reflect an auxin increase which could inhibit root growth and be involved in a process independent of AON, because high auxin concentrations inhibit root elongation (Swarup et al., 2007). Therefore, the shortened root phenotype of sunn mutants might be the result of the demonstrated increased amount of auxin transported from the shoot to the root (Schnabel et al., 2005; van Noorden et al., 2006). Because a relatively high ratio of cytokinin to auxin is thought to favor nodule initiation (Souleimanov et al., 2002), it cannot be excluded that the inhibition of nodule formation in 35S:MtCLE12 roots is also the result of unbalanced auxin concentrations.

Quantitative analysis showed a strong reduction of NIN expression in 35S:MtCLE12 roots, although Mortier et al. (2010) showed that NIN is necessary for MtCLE12 expression. Therefore, we propose a negative feedback regulation that implicates cytokinin signaling and NIN in AON. Our data indicate that type A response regulators and MtEFD act downstream of both MtCLE12/MtCLE13 and SUNN. Thus, we propose a model that links Nod factor signaling, cytokinin perception and AON (Fig. 4). In brief, cytokinin signaling and Nod factor perception are initiated by rhizobial epidermal infection and lead to the activation of the nodulation signaling pathway involving NSP1, NSP2 and NIN, which elicit the formation of nodules as well as the induction (directly/indirectly) of MtCLE12 and MtCLE13; in turn this up-regulates type A response regulators via SUNN and the shoot-derived inhibitor (SDI, Ferguson et al., 2010). MtEFD might be one of the factors involved in the up-regulation of type A response regulators, which in turn negatively regulate cytokinin signaling. There may be other unknown factors independent of MtEFD that modulate type A response regulators and cytokinin signaling, since a mutation in MtEFD does not show the same super-nodulating phenotype as sunn mutants. The down-regulation of cytokinin signaling modulates Nod factor signaling downstream genes (e.g. NIN) and therefore negatively regulates cell divisions and nodule formation. This provides the plant with a tight regulatory control mechanism to control nodule numbers in tune with the physiological status of the plant.

Figure 4.

A simple model for the crosstalk between the nodulation signaling pathway, cytokinin signaling and the autoregulation of nodulation (AON) during indeterminate nodule formation. Nod factor perception and cytokinin signaling are initiated by rhizobial epidermal infection and lead to the activation of the nodulation signaling pathway involving NODULATION SIGNALING PATHWAY 1 and 2 (NSP1, NSP2) and NODULE INCEPTION (NIN) transcription factors, which stimulate the cortical and pericycle cell divisions that give rise to the founder cells of the nodule organ (for a review, see Crespi & Frugier, 2008b). NIN also induces MtCLE12 expression which, via SUNN, up-regulates the unknown shoot-derived inhibitor(s) (SDI) and subsequently induces the type A response regulators and possibly other components such as MtEFD. Type A response regulators lead to the down-regulation of cytokinin signaling, which modulates downstream genes of Nod factor signaling (e.g. NIN) and therefore negatively regulates cell divisions and nodule formation (Ferreira & Kieber, 2005; Doerner, 2007). Red arrows, cytokinin signaling pathway; blue arrows, Nod factor signaling pathway; green arrows, autoregulation of nodulation pathway; black arrow, nodule formation. CCaMK, calcium/calmodulin-dependent protein kinase; CK, cytokinin; CRE1, Cytokinin Response1; EFD, ethylene response factor required for nodule differentiation; NF, Nod factor; NFP/LYK, nod factor perception/LysM domain-containing receptor-like kinase.

The different time points used to assess MtRR8 and MtEFD expression levels in A17 and sunn-4 backgrounds do not allow direct comparison. However, these time points were chosen to match the gene expression analysis with the assessment of the nodule numbers on the same roots. Since 35S:MtCLE12 roots do not develop nodules 10 or 21 dpi, we conclude that it is unlikely that major changes in MtRR8 and MtEFD expression occur between 10 and 21 dpi. The Medicago gene expression atlas shows lower expression levels for MtEFD and MtRR8 in A17 control roots at 10 compared with 21 dpi (Fig. S6), which would only positively affect the significance of the increased MtEFD and MtRR8 expression in 35S:MtCLE12 roots at 10 dpi.

Mortier et al. (2010) showed that NIN is necessary for MtCLE12 expression. Our analyses showed a strong reduction of NIN expression in 35S:MtCLE12 roots. It cannot be completely excluded that this is the result of the lack of nodule formation in 35S:MtCLE12 roots because NIN is strongly induced in, and required for, nodule formation (Schauser et al., 1999). However, the severe reduction of NIN expression in 35S:MtCLE12 roots correlates with the lack of nodules formed, indicating MtCLE12 function upstream of NIN.

It remains to be shown how other putative components of AON, such as KLAVIER, ROOT-DETERMINED NODULATION and TOO MUCH LOVE, can be integrated into this model. How auxin is linked to this process also remains to be determined.

The proposed function of MtEFD and type A response regulators downstream of SUNN mirrors the results found in Arabidopsis shoot apical meristem regulation. Here, the CLAVATA3-dependent activation of CLAVATA1 (orthologous to SUNN) down-regulates the transcription factor WUSCHEL, resulting in suppression of cytokinin biosynthesis by the transcription of several Arabidopsis response regulator genes (Leibfried et al., 2005). These data already demonstrate the negative influence of type A response regulators on meristem size and that their repression by WUSCHEL is necessary for proper meristem function. Moreover, WUSCHEL is thought to create a negative feedback loop via the CLV3/CLV1 cascade (Brand et al., 2000; Schoof et al., 2000), resulting in the down-regulation of cytokinin signaling (Leibfried et al., 2005). Recently, Kondo et al. (2011) showed that CLE10, which is preferentially expressed in the root vascular system, inhibits protoxylem vessel formation in Arabidopsis by repressing specifically the expression of two type A response regulators, ARR5 and ARR6, revealing that crosstalk between CLE signaling and cytokinin signaling occurs. Although the factors downstream of SDI are not known, it is likely that WUS-like homeodomain transcription factors (WOX TFs) might be good candidates for this, since several CLE peptides are known to act through WOX TFs (Jun et al., 2008; Hirakawa et al., 2010). The silencing of a M. truncatula gene encoding a WOX TF (MtGI accession number TC104580) results in a Nod phenotype (mtrnai.msi.umn.edu/), indicating a role for the WOX TFs during nodule development. Nevertheless, further study is required to verify this interaction. Taken together, it is likely that cytokinin signaling functions downstream of the CLE/WUS signaling pathway and the crosstalk between CLE and cytokinin signaling pathways may be a common feature of CLE peptide-dependent cell-to-cell communication, further supporting our model that regulatory CLE peptides function in AON through cytokinin signaling. Our model is consistent with: (a) unaltered EFD expression after Nod factor or ethylene treatments; (b) the down-regulation of MtENOD11 in 35S:MtEFD roots (Vernie et al., 2008); and (c) the lack of MtENOD11 expression in roots overexpressing MtCLE12 (Mortier et al., 2010). Our model is also consistent with spontaneous nodule formation (independent of rhizobia) in the gain of function histidine kinase mutants cre1 in M. truncatula and snf2 in L. japonicus. Here nodule numbers are autoregulated in a NIN-dependent manner (Gonzalez-Rizzo et al., 2006; Murray et al., 2007; Tirichine et al., 2007). This model is also consistent with NIN-dependent autoregulation in the calcium/calmodulin-dependent protein kinase mutant dmi3 in M. truncatula and snf1 in L. japonicus (Levy et al., 2004; Tirichine et al., 2006; Marsh et al., 2007).

It is possible that CLE12 and CLE13 ligands directly interact with SUNN in the shoot, but this remains to be determined experimentally. Although many studies show AON is primarily dictated by long-range mechanisms involving shoot-located SUNN (Schnabel et al., 2005), it is also possible that a component of AON is mediated by a local suppression of MtEFD that is independent of SUNN. This may enable some degree of local control of nodule formation. Similarly, in Arabidopsis, WUSCHEL expression occurs through both CLAVATA-dependent and -independent pathways by changing cytokinin signaling (Gordon et al., 2009). Therefore, the SUNN-independent regulation of MtEFD might also involve type A response regulators and therefore cytokinin signaling. This hypothesis is consistent with the unaltered expression of type A response regulators in 35S:MtCLE12 root of sunn-4.

Acknowledgements

This work was supported by the Australian Research Council through the Australian Research Council Centre of Excellence for Integrative Legume Research (grant CE0348212). We thank Huiming Yang, Mahmut Kare and Nadia Radzman for their technical assistance, and Charles Hocart and Jeremy Weinman for critical reading of the manuscript. We also thank Douglas Cook, Ulrike Mathesius, Julia Frugoli, Sandra Moreau and Pascal Gamas, and John Howieson for kindly supplying the A17, GH3:GUS, sunn and efd-1 seeds and S. medicae strain 1022, respectively.

Ancillary