Suppression of hypernodulation in soybean by a leaf-extracted, NARK- and Nod factor-dependent, low molecular mass fraction


  • Yu-Hsiang Lin,

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, The University of Queensland, St Lucia, QLD 4072, Australia
    Search for more papers by this author
  • Brett J. Ferguson,

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, The University of Queensland, St Lucia, QLD 4072, Australia
    Search for more papers by this author
  • Attila Kereszt,

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, The University of Queensland, St Lucia, QLD 4072, Australia
    Search for more papers by this author
  • Peter M. Gresshoff

    1. Australian Research Council Centre of Excellence for Integrative Legume Research, The University of Queensland, St Lucia, QLD 4072, Australia
    Search for more papers by this author

Author for correspondence:
Peter M. Gresshoff
Tel: +61733653550


  • Legumes regulate the number of nodules they form via a process called autoregulation of nodulation (AON). This involves a shoot-derived inhibitor (SDI) molecule that is synthesized in the shoots and is transported down to the roots where it inhibits further nodule development.
  • To characterize SDI, we developed a novel feeding bioassay. This involved feeding aqueous leaf extracts directly into the petiole of hypernodulating and supernodulating nark mutant plants of Glycine max (soybean). These mutants normally exhibit an increased nodulation phenotype because SDI is not produced and thus AON is nonfunctional.
  • Feeding wild-type leaf extracts presumed to contain SDI was successful in suppressing the increased nodulation phenotype, whereas feeding with Gmnark leaf extracts did not. Suppression activity was inoculation-dependent, Nod factor-dependent, required GmNARK activity, and was heat-, Proteinase K- and ribonuclease A-resistant. Wild-type extracts maintained suppressive activity even at a ninefold dilution. Sinorhizobium meliloti-inoculated Medicago truncatula leaf extracts from wild-type, but not from supernodulating mutant Mtsunn, suppressed hypernodulation in soybean.
  • Our results demonstrate that the petiole feeding bioassay is an efficient and effective technique to introduce aqueous extracts into plants. They also demonstrate that SDI is a small compound with an apparent molecular mass of < 1000 Da and is unlikely to be a protein or an RNA molecule.


Most leguminous plant species establish a symbiotic relationship with soil bacteria belonging to the Rhizobiaceae, leading to the formation of lateral root organs called nodules (Ferguson et al., 2010). Within the nodules the bacteria differentiate and convert atmospheric dinitrogen gas to ammonium, which is assimilated by, and transported within, the plant (Udvardi et al., 1988). Nitrogen fixation in legumes is a key component of global nitrogen cycles (Graham & Vance, 2003; Lau et al., 2008; Scott et al., 2008).

Nodule numbers are predominantly regulated by the plant via a process termed ‘autoregulation of nodulation (AON)’ or ‘feedback inhibition of nodulation’ (Delves et al., 1986; Gresshoff & Delves, 1986; Caetano-Anollés & Gresshoff, 1991; Ferguson et al., 2010). Hormonal and environmental factors can also influence nodule ontogenesis (Ferguson and Mathesius, 2003; Gresshoff et al., 2009).

Plant mutants defective in AON form an increased number of nodules resulting in a super-/hyper-nodulating phenotype (Carroll et al., 1985a; Duc & Messager, 1989; Pracht et al., 1994; Vuong et al., 1996; Wopereis et al., 2000; Penmetsa et al., 2003). Often, root growth abnormalities occur (Day et al., 1986; Wopereis et al., 2000; Buzas & Gresshoff, 2007), in part explaining the limited agronomic application of supernodulated legumes. Reciprocal grafting experiments between wild-type and supernodulating mutant shoots and rootstocks showed the nodulation phenotype of the root to be predominantly determined by the genotype of the shoot (Delves et al., 1986; Jiang & Gresshoff, 2002; Krusell et al., 2002). However, mutants such as nod3 in Pisum sativum (Postma et al., 1988; Li et al., 2009), Ljrdh1 (Ishikawa et al., 2008), and Ljtml (Magori et al., 2009) in Lotus japonicus show that components in the roots are also essential.

Positional cloning defined one AON component as a leucine-rich repeat (LRR) receptor kinase, designated GmNARK in soybean, LjHAR1 in L. japonicus, MtSUNN in Medicago truncatula and PsSYM29 in P. sativum (Krusell et al., 2002; Nishimura et al., 2002; Searle et al., 2003; Schnabel et al., 2005). This ‘autoregulation receptor kinase’ is structurally related to other plant genes encoding receptor kinases, including those involved in meristem maintenance (CLV1; Clark et al., 1997), perception of brassinosteroids (BRI1; Clouse et al., 1996), phytosulfokine responses (PSK; Matsubayashi et al., 2006), tracheogenesis (Motose et al., 2004), and flagellin-induced defense responses (Chinchilla et al., 2007). The LRR-RKs are also critical in mammalian defense (Kaparakis et al., 2007). GmNARK is expressed in phloem cells of all vascular bundles (Nontachaiyapoom et al., 2007), particularly in the leaf, but is generally not expressed in dividing tissues of the root and shoot.

According to the current model originally proposed by Gresshoff & Delves (1986), then expanded by Caetano-Anollés & Gresshoff (1991), AON (also termed AUT by some authors; Magori et al., 2009; Okamoto et al., 2009) involves three key components in a feedback regulatory loop acting via long distance signaling. This pathway commences with the production of ‘Q’ during the early stages of cortical cell division for lateral organ development in roots (Li et al., 2009). A putative CLE peptide or its precursor (Oelkers et al., 2008; Okamoto et al., 2009), functions as the root signal (Q) that travels to the phloem parenchyma of the leaf vein, most likely via xylem (Nontachaiyapoom et al., 2007). The perception of ‘Q’ by NARK/HAR1/SUNN/SYM28 is then presumed to initiate the production of a second signal termed shoot-derived inhibitor (SDI). This, in its function as a nodulation inhibitor, originates from leaves (Delves et al., 1986, 1992; Nishiwaki et al., 1997; Sheng & Harper, 1997) and is presumably transported from the shoot to the root via the phloem. Interestingly, nitrate, a known inhibitor of nodule formation, may be acting either systemically or locally depending on the species and experimental system. For example, transcriptome analysis of nitrate-treated M. truncatula split root systems showed downregulation of expected nodulation-related transcripts (Ruffel et al., 2008), but split root application of nitrate inhibited nodule number only on the exposed root portions of clover (Carroll & Gresshoff, 1983) and soybean (Hinson, 1975). The mechanism is still unknown but may involve possibly a CLE peptide and root-expressed NARK/HAR1 (Gresshoff et al., 2009; Okamoto et al., 2009), leading to the combined mutant loss-of-function phenotypes of: absence of AON (systemic); nitrate tolerance of nodulation (local); and diminished root development (Day et al., 1986; Wopereis et al., 2000; Buzas & Gresshoff, 2007).

Although the chemical nature of the long-distance signal SDI is currently unknown, studies have provided hints of its nature. For example, signals (presumably SDI) from one wild-type scion suppressed the supernodulation phenotype in the stock of another legume species (Delves et al., 1987; Harper et al., 1997). In addition, a transcriptional profiling study detected genes being downregulated in leaves, but not in roots of Bradyrhizobium japonicum-inoculated wild-type plants (Kinkema & Gresshoff, 2008). These included several genes involved in jasmonic acid (JA) biosynthesis or JA response, suggesting this classical plant hormone could have a role in AON downstream of GmNARK.

Some phytohormones clearly interact with AON suppression, though in many cases it is difficult to discern between cause and effect (Ferguson & Mathesius, 2003). For example, foliar application of methyl jasmonate (MeJA) to L. japonicus inhibited early stages of nodule development (Nakagawa & Kawaguchi, 2006). These inhibitory effects were observed in wild-type and hypernodulating har1 mutants, suggesting that MeJA acts after, or independent of, LjHAR1.

Foliar application of brassinosteroids (BRs) to the supernodulating soybean mutant, En6500 (mutated in GmNARK; Nishimura et al., 2002), also suppressed nodule development (Terakado et al., 2005). However, suppression was not observed in wild-type plants (Terakado et al., 2005). Moreover, BR biosynthesis and perception mutants of pea formed fewer nodules than wild type, indicating a role for BRs in nodule development, but not AON (Ferguson et al., 2005a).

Auxin also is postulated to play an important role in nodulation (Mathesius, 2008). The supernodulating M. truncatula mutant, sunn, showed higher levels of long-distance, shoot-to-root, auxin transport compared with its wild type (van Noorden et al., 2006). However, elevated levels of auxin transport in the nonAON ethylene-insensitive M. truncatula mutant, skl, may indicate a secondary role (Prayitno et al., 2006).

Nodule numbers are also negatively regulated by factors such as ethylene and nitrate (Carroll et al., 1985a,b; Ligero et al., 1987; Abeles et al., 1992; Lee & LaRue, 1992; Ferguson & Mathesius, 2003; Ferguson et al., 2005b). Application of ethylene to inoculated roots generally decreases nodule numbers (Ligero et al., 1987; Lee & LaRue, 1992; Ferguson et al., 2005b) as does nitrate fertilization (Carroll et al., 1985a).

In this study both wild-type Bragg and Williams and their respective hypernodulating/supernodulating mutants were used. Both hypernodulating and supernodulating genotypes (nts1116 and NOD4, respectively) were chosen to evaluate suppressive activity at both levels of the mutated phenotypic range. Hypernodulation and supernodulation were originally defined by Carroll et al. (1985b) as different degrees of increased nodulation with hypernodulation representing the more moderate phenotype as seen in nts1116 compared with nts1007 or NOD4. The hypernodulating nts1116 carries a missense mutation (V837A) in the GmNARK receptor kinase, leading to loss of AON (Searle et al., 2003) and in vitro kinase activity (Miyahara et al., 2008). By contrast, the supernodulating NOD4 has a missense mutation (V370D) in the GmNARK receptor LRR domain, presumed to be the ligand binding site (J. Batley et al., unpublished).

The chemical and structural definition of both Q and SDI signals is critical for the further understanding of this regulatory loop. Here we report on a critical step towards that goal as liquid leaf extracts suppressed GmNARK deficiency in soybean. The development of a petiole-feeding bioassay (allowing ‘developmental cross-feeding’) facilitated the testing of various plant extracts for SDI activity, and possibly other biological activities of regulatory molecules.

Materials and Methods

Plant material

Before germination, Glycine max (L.) Merr. genotype Bragg (wild-type), nts1116 mutant (Carroll et al., 1985a), Williams 82 (wild type) and NOD4 mutant (Pracht et al., 1993) soybean seeds were surface-sterilized by immersion in 70% (v : v) ethanol–hydrogen peroxide 3% (v : v) for 1 min and washed five times in sterile distilled water. The seeds were planted in sterilized vermiculite (grade 3) in glasshouses (28°C day, 25°C night, 18 h day-length). Bacterial inoculation was carried out 5 d after germination or 1 d after the start of feeding with 250 ml of Bradyrhizobium japonicum CB1809 (OD600 = 0.1) or the AN122 nodC mutant grown in yeast mannitol broth (YMB; Vincent, 1970). Plants were watered to saturation using a modified Herridge nutrients solution three times a week. Town water was used as required for the remaining days in the week. For nodule count studies, 3-wk-old plants were inoculated then uprooted and gently rinsed 8–11 d later to record their nodule number. For statistical analysis where comparisons were made between two treatments, means were discriminated using Student’s unpaired t-test.

Medicago truncatula genotype A17 (wild type) and Mtsunn mutant seeds (obtained from Dr U. Mathesius, CILR, ANU) were scarified with sandpaper. Seeds were surface-sterilized by immersion in 70% (v : v) ethanol : 3% (v : v) hydrogen peroxide for 5 min and washed five times in sterile distilled water. Seeds were germinated on 1.5% water agar plates (cell culture tested; Sigma) in the dark overnight at 29°C. Seedlings were transferred to 250 mm pots filled with sterilized vermiculite (grade 3). Bacterial inoculation was carried out 7 d after transfer to pots with 250 ml of Sinorhizobium meliloti 1021 (OD600 = 10−3) grown in Fahraeus medium (Vincent, 1970). As a minimum quantity of source material was required for the extraction process, source leaves were collected from 4- to 5-wk-old plants (after inoculation).

Sample preparation

For experiments using dye, 1% (v : v) blue dye FCF (133; Queen Fine Foods Pty. Ltd., 17-19 Wakefield St., Alderley, Australia) solutions were made using purified water. Sucrose solutions were made using 10% (w : v) sucrose powder (analytical grade, Univar, 17425 NE Union Hill Rd, Redmond, WA 98052, USA) dissolved in purified water. For leaf extracts, mature trifoliate leaves derived from source plants (4–5 wk old) were harvested and snap-frozen for storage at −80°C. The harvested leaves were ground to a powder in liquid nitrogen using mortar and pestle. 10% (w : v) leaf powder was added to a water-based buffer (10 mm 4-morpholinethanesulfonic acid (MES) at pH 6.2. The leaf mixture was stirred continuously for 30 min. The crude extracts were centrifuged (5000 g at 10°C for 30 min) to separate and sediment the cell debris. The crude extracts were then filtered through cellulose filter paper (Whatman Grade 1, 11 μm) to remove cell debris. The final pure extract was then freeze dried at −30°C. The dried samples were stored at −80°C.

Molecular sieving was conducted by passing water extracts of Bragg (inoculated) leaves through a Microsep Omega column (Pall Corporation, 25 Harbor Park Drive, Port Washington, NY 11050, USA) and collecting the volume with a presumed exclusion size of 10 kDa or 1 kDa, respectively. Molecular masses are indicative only and do not consider the Stokes radius of the molecule. For radiochemical feeding experiments, 1% (v : v) glucose, d-[3-3H] and 1% (v : v) mannitol, d-[1-3H(N)] (Perkin Elmer) were prepared using purified water.

Petiole feeding

A 20 mm silicon tubing (OD 4.8 mm; ID 2.6 mm) was connected to the base of a 5 ml syringe. An 80 mm (OD 2.4 mm; ID 1.6 mm) tube was then connected to the 20 mm tubing. The syringe assembly was attached to a 255 mm bamboo skewer with rubber bands. The bamboo skewer was then inserted into the soil near the recipient plant. This created a simple and adjustable stand for the feeding assembly. The target trifoliate petiole was cut perpendicularly using a sterile scalpel blade and was attached to well-fitting 80 mm silicon tubing. At least 10 mm of the petiole was inserted into the silicon tubing, which was then sealed airtight using Selley’s water-resistant sealant (nontoxic for plant tissue). The feeding solution was introduced via the syringe and air present in the silicon tubing was extracted by hypodermic syringe (1 cm3 ml−1 Terumo syringe and 30G needle). As a result, the feeding solution was able to enter the silicon tubing immediately (Fig. 1a). It was essential that air be removed from the tubing to prevent bubble formation. A 2 × 2 cm Parafilm square perforated a few times was used to seal the opening of the SFE to restrict contamination of the feeding solution.

Figure 1.

 Visualization of dye movement in petiole-fed soybean plants. (a,b) Cut petioles with the feeding apparatus. (c,d) Leaves and flowers before blue dye feeding FCF (133) feeding. (e,f) Blue dye staining in both leaves and flowers 10 h post-feeding. (g,h) Shoot cross-section below the treated petiole before and after dye feeding.

Test plants were grown for c. 4 wk to allow for the full maturation of three trifoliate leaves, then fed with different extracts for 9 d, and inoculated with B. japonicum CB1809 at the second day of feeding. For each condition, five test plants (grown in separate pots) were fed with the treatment extract (n = 5). Nodule numbers and lateral roots were counted and the lengths of the roots and shoots were measured to evaluate effects of feeding on general plant growth. Great care was taken to maintain the same conditions between feeding experiments, but variation of actual nodule numbers was still observed. This may be attributed to the complexity of the feeding bioassay, as external factors such as petiole health may influence the frequency of resetting of the bioassay. To compensate for these variations, results were presented according to the level of suppression (%). The percentage of nodulation suppression was calculated based on the number of nodules observed compared with the nodule number of nts1116 extract-fed control (nts1116, B. japonicum CB1809 inoculated) plants in the same experiment.

Enzyme treatment of extracts

Bragg and nts1116 water extracts were degraded using either proteinase K or RNase A enzymes. Freeze-dried extracts were resuspended in 9 ml of buffer (10 mm Tris (pH 8.0), 5 mM EDTA, 0.5% sodium dodecyl sulfate (SDS)). Proteinase K or RNAse A was added to a final concentration of 1 mg ml−1. The mixture was incubated overnight at 37°C. Enzymes are removed from mixture with 1 kDa filter via the Sigma 4K15 centrifuge (5000 g, 30 min). Protein samples (10 μl) were mixed with sample buffer (100 mm Tris, pH 6.8, 2% SDS, 5%ß-mercapto-ethanol, 15% glycerol) and incubated at 93°C for 10 min. Samples were cooled on ice then loaded onto 10% (w : v) acrylamide gel. The separated proteins were stained using Coomassie Brilliant Blue staining solution. RNA samples (15 μl) were mixed with loading buffer (95% formamide, 0.025% bromophenol blue, 0.025% xylene cyanol, 0.025% SDS, 0.5 mm EDTA) and incubated at 70°C for 10 min. Samples were cooled on ice then separated on 1.2% (w : v) agarose gel in 0.5× TBE buffer to verify RNA digestion.


Developing a bioassay for SDI

Petiole feeding is a reliable technique to introduce aqueous solutions into the plant  Blue dye FCF (133) was fed continuously for 10 h into the first trifoliate petiole of 4-wk-old nts1116 hypernodulation mutant plants (Fig. 1a,b). Dye staining was visually confirmed in leaves and flowers (Fig. 1c–f) and cross-sections of shoots (Fig. 1g,h). These findings verify our feeding method as a reliable technique to introduce aqueous substances into plants. The fact the dye was observed both above and below the feeding site demonstrated bidirectional translocation of the fed constituents, suggesting they likely travelled in the phloem or the apoplast. To further test this, radioactive [3H]-labeled glucose and mannitol were fed continuously to nts1116 plants for 0.5, 1.5 and 3 h (Fig. 2a) or 24 h (Fig. 2b). Radioactivity was detected at significant levels both above and below the petiole feeding site, including in the root, further demonstrating bidirectional translocation. This pattern of translocation and the speed (up to 5 cm h−1) suggested the radiolabel was transported in the phloem and apoplast (Wardlaw, 1974). More radioactivity was detected above the feeding site, and it travelled faster than that transported downwards, suggesting that some of the radiolabel was also transported in the xylem. Spikes detected in the flow of radioactivity in the shoot correlated with the presence of leaf attachment sites to the stem. The increased vasculature occurring at these locations is most likely the reason for the increased amount of radiolabel detected.

Figure 2.

 Translocation of radioactivity after petiole feeding in soybean plants. Fed at the petiole of the first trifoliate with either (a) [3H]-glucose at 0.5 (diamonds), 1.5 (triangles) and 3 h (circles) time-points or (b) [3H]-mannitol for 24 h. The arrow represents the location of the petiole feeding site. A stylized soybean plant illustrates the developmental positions of the sampled 1-cm plant segments.

A time-course for the uptake of a sucrose (10% w : v) solution was performed to ascertain the functionality of the petiole over the feeding period. After 24 h, 500 ± 110 μl of solution were taken up by the plant. Later time-points showed a gradual decrease in the rate of uptake until reaching a consistent level of 165 ± 25 μl absorbed per day. After 7 d of petiole feeding, c. 2 ml of the solution had been absorbed. Less concentrated solutions were taken up at an even greater rate (data not shown). Care was taken to work in sterile or semi-sterile conditions to prevent petiole closure (shown by browning of the cut site) or petiole abscission.

Establishing the optimal timing of AON assays in soybean  A nodulation time-course was performed to establish the shortest time point post-inoculation at which wild-type Bragg plants reliably exhibited significant AON suppression relative to its hypernodulating mutant nts1116. Significant difference in nodule number per plant was detected as early as 8 d after inoculating 4 wk-old wild-type and hypernodulating/supernodulating plants with B. japonicum (Fig. 3). This timing is consistent with previous reports using soybean (Carroll et al., 1985a; Delves et al., 1986). Based on these findings, 8 d post-inoculation was chosen as optimal to assess nodule numbers using petiole feeding.

Figure 3.

 Nodulation time-course of Bradyrhizobium japonicum-inoculated Bragg (A, WT), hypernodulating nts1116 (B, V837A) mutant, Williams 82 (C, WT) and supernodulating NOD4 (D, V370D) mutant soybean plants. To mimic the nodulation response in petiole-fed plants, inoculation occurred at 4 wk after germination. Values are means ± SE (n = 5); DAI, d after inoculation.

SDI is reliably detected in aqueous leaf extracts using petiole feeding  To test whether petiole feeding could be used as a reliable bioassay for detecting SDI activity, various aqueous leaf extracts were fed into 4-wk-old hypernodulating nts1116 plants. The fed plants were inoculated with B. japonicum 1 d after commencing petiole feeding (to avoid presumed ‘injury’ effects), and nodule numbers were determined 8 d after inoculation. No significant differences were detected in shoot or root growth after 9 d of feeding (Supporting Information Fig. S1). Feeding water-based leaf extracts from inoculated nts1116 plants did not suppress nodule numbers of nts1116 plants (Fig. S2). This demonstrated that the technique itself did not influence the bioassay and that leaf extracts from nts1116 plants did not suppress nodule numbers. By contrast, strong nodule suppression (up to 85% relative to nts1116 extract-fed control plants) was observed when nts1116 plants were fed with leaf extracts from B. japonicum-inoculated wild-type plants (Fig. 4a, Fig. S2). In addition, nodule numbers of supernodulating NOD4 plants were strongly suppressed (c. 50%) when they were fed with leaf extracts from the wild-type cultivars Bragg and Williams 82 (Fig. 4c, Fig. S3). No suppression of nodule numbers was observed when NOD4 plants were fed with leaf extracts from either NOD4 or nts1116 plants.

Figure 4.

 Differential nodulation suppression in hypernodulation and supernodulation genotypes of soybean by petiole-fed leaf extracts: 4-wk-old (a) hypernodulating nts1116, (b) Bragg (WT), (c) supernodulating NOD4 plants inoculated with Bradyrhizobium japonicum CB1809 at 24 h after commencement of petiole-feeding using extracts from (A) B. japonicum-inoculated nts1116, (B) Bragg, (C) NOD4, (D) Williams 82 leaves. The level of suppression was calculated as (a) the percentage of total nodules relative to those found on nts1116 extract-fed nts1116 plants (negative control), (b) percentage of total nodules relative to Bragg extract-fed Bragg plants, and (c) percentage of total nodules relative to NOD4 extract-fed NOD4 plants. Total nodule numbers per plant are shown in the Supporting Information Figs S2, S3 and S4. Values are means ± SE (n = 5).

The fact that nts1116 and NOD4 leaf extracts failed to suppress nodule numbers, whereas both Bragg and Williams 82 leaf extracts did, clearly demonstrates that the bioassay is SDI-specific. Moreover, these findings confirm that the inhibitory factor was absent in nts1116 and NOD4 mutants but present in the wild-type cultivars. Other aspects of plant growth, especially lateral root number per plant, were not visibly affected by the treatments (Fig. S1), further demonstrating the reliability of the bioassay.

To resolve whether supernodulation is caused by the absence of an inhibitor (the current model) or the increased presence of a presumed ‘nodulation stimulator’, leaf extracts were also fed into Bragg plants. Feeding Bragg plants with leaf extracts from inoculated nts1116 plants did not alter nodule numbers (Fig. 4b, Fig. S4). Similarly, feeding Bragg plants with leaf extracts from inoculated Bragg plants also did not further reduce nodule numbers (Fig. 4b, Fig. S4).

Partial characterization of the soybean shoot-derived inhibitory principle

While leaf extracts from inoculated WT plants significantly suppressed hypernodulation of nts1116 plants, this was not observed using leaf extracts from uninoculated wild-type plants (Fig. 5a, Fig. S5). This demonstrates that inoculating wild-type plants with B. japonicum was necessary for SDI production and the resulting suppressive property of the leaf extracts. To determine whether the suppressive activity was inoculation- or Nod factor-dependent, leaf extracts were made from Bragg plants inoculated with the B. japonicum AN122 nodC mutant. These extracts were unable to reduce the nodule numbers of nts1116 plants (Fig. 5b, Fig. S6). These finding are consistent with the current model of AON.

Figure 5.

 Regulation of nodulation suppression in soybean: 4-wk-old hypernodulating nts1116 plants inoculated with Bradyrhizobium japonicum 24 h following the commencement of feeding using various leaf extracts. (a) Leaf extracts from B. japonicum-inoculated (B) or -uninoculated Bragg (WT) plants (C). (b) Leaf extracts from B. japonicum AN122 nodC mutant-inoculated Bragg (WT) (D), or B. japonicum-inoculated Bragg (B) plants. (c) Leaf extracts from B. japonicum CB1809 inoculated plants. The leaf extracts were diluted at 3-fold (E), 9-fold (F), and 27-fold (G) or given at three times strength (H). (d) Leaf extracts from Sinorhizobium meliloti-inoculated Medicago truncatula (I, WT) and sunn mutant (J) plants. nts1116 plants fed with B. japonicum-inoculated nts1116 (A) leaf extracts were used as control for all experiments. Total nodule numbers per plant are shown in the Supporting Information Figs S5, S6, S8, S10. Values are mean ± SE (n = 5).

To test the heat stability of SDI, inoculated wild-type leaf extracts were heated to 80°C for 60 min, then fed into nts1116 plants. This treatment failed to destroy the suppressive activity of the leaf extract (Fig. 6a, Fig. S7) indicating that SDI is heat-stable. Molecular sieving through matrices excluding 10 kDa and 1 kDa-sized molecules were used to test the size of SDI suppressive activity. Feeding of 1 kDa wild-type leaf extracts to nts1116 plants showed a strong suppression (87%; Fig. 6a) indicating SDI is a small molecule.

Figure 6.

 Partial characterization of the ‘nodulation suppression activity’ with 4-wk-old hypernodulating nts1116 plants. (a) Leaf extracts from Bradyrhizobium japonicum-inoculated Bragg (B, WT) plants sieved through 1 kDa (C) and 10 kDa (D) filters or heat-treated (E). (b) Leaf extracts from B. japonicum-inoculated Bragg plants enzymatically treated with ribonuclease A (F, RNAse) or proteinase K (G) to degrade protein and RNA molecules, respectively. Mock-treated RNAse (H) and proteinase K (I) samples were also fed to nts1116 plants. nts1116 plants fed with B. japonicum-inoculated nts1116 (A) leaf extracts were used as control for both experiments. Values are means ± SE (n = 5). (c) RNA was observed in mock-treated (−) but not in RNAse-treated (+) samples. (d) Protein was detected in untreated (−) but not proteinase K treated (+) samples.

Biological generality of SDI was tested by feeding leaf extracts from inoculated M. truncatula wild-type and supernodulation sunn mutant plants (Schnabel et al., 2005) into nts1116 plants. Significant suppression was observed from wild-type leaf extracts. Interestingly partial suppression was also observed with sunn mutant extracts (Fig. 5d, Fig. S8).

Inoculated wild-type leaf extracts were treated with Proteinase K or RNase A, which destroy the majority of protein and RNA molecules, respectively. However, these treatments failed to lessen the suppression activity (Fig. 6b, Fig. S9). Gel electrophoresis was used to confirm that enzymatic treatments successfully degraded RNA and protein in treated aqueous leaf extracts (Fig. 6c,d).

The potency of SDI was tested by diluting inoculated wild-type leaf extracts 3-, 9- and 27-fold. When diluted to 3- and 9-fold, nodulation of nts1116 plants was significantly suppressed and similar to the degree of suppression achieved by undiluted wild-type extracts (Fig. 5b, Fig. S10). However, 27-fold diluted samples failed to suppress. It appears that 1/27 of the original strength was below the minimum concentration threshold required for effective SDI activity therefore suppression of nodulation was no longer observed.


Aqueous extracts from leaves derived from B. japonicum-inoculated wild-type soybean plants specifically suppressed supernodulation/hypernodulation in GmNARK mutant plants, opening the way for chemical fractionation and structural characterization based on activity-monitoring using the bioassay. The petiole feeding bioassay was shown to be an efficient and reliable technique for continually introducing solutions into the plant over long periods. Using various dyes and radiolabeled compounds, molecular transport above and below the petiole feeding site was observed within hours, indicating the introduced solutions were translocated bidirectionally, most likely via the phloem and apoplast (Wardlaw, 1974). Thus, the petiole-feeding technique could have wide-ranging applications in plant physiology for studying molecules involved in development, metabolism and growth.

Extracts, presumed to contain SDI, consistently and significantly reduced the nodule number of hypernodulating nts1116 and supernodulating NOD4 plants, often achieving nodulation phenotypes normally seen in wild-type plants. Although only 50% suppression was achieved in the NOD4 supernodulating mutant, this represents a large reduction of roughly 400 nodules. It is likely that better suppression was achieved with nts1116 because of its mutation (V837A) in GmNARK (Miyahara et al., 2008). This would mean that the mutant already has some endogenously produced SDI to go along with the SDI being exogenously fed into the plant. It would explain why the mutant exhibits a hypernodulation phenotype rather than the more severe supernodulation phenotype seen in nonsense mutants such as nts1007 and missense mutants in the putative ligand-binding site of the LRR-RK.

In contrast to the above findings, leaf extracts from plants where AON has not been activated (i.e. SDI is presumed to be absent) failed to alter the nodule numbers of nts1116 and NOD4 plants. This confirms that SDI is extractable in water from wild-type leaf tissue and can successfully be introduced into a soybean plant using the petiole feeding described here. Furthermore, our findings clearly demonstrate that the absence of B. japonicum-inoculation, and more specifically the absence of Nod factor (Fig. 5), or the presence of the Gmnark mutation (Fig. 4), remove the suppressive principle. Nod factor is an essential component for the induction of the nodule organogenesis signal cascade. The B. japonicum AN122 mutant is unable to produce Nod factor, therefore plants inoculated with the mutant fails to induce the NF-induced signal cascade. As a result, SDI is not activated because the AON pathway has not been triggered (Fig. 7). Similarly, the absence of Nod factor perception in nonnodulation mutant nod139, lacking a functional GmNFR5α and β-NF receptor component (Indrasumunar et al., 2010) fails to induce AON in approach grafts with wild-type Bragg (Caetano-Anollés & Gresshoff, 1990).

Figure 7.

 Model of the autoregulation of nodulation (AON) pathway in relation to nodule organogenesis. Nod factor (NF) from Bradyrhizobium is detected by the roots and induces a signal cascade resulting in cell divisions and the eventual infection of the nodule. Furthermore, the NF-induced signal cascade triggers the production of signal Q that interacts with NARK in the leaves. As a result, a shoot-derived inhibitor (SDI) is activated to suppress further nodule development in roots. Bradyrhizobium japonicum AN122 mutant is unable to produce NF, and soybean mutant nod139 (Mathews et al., 1989) known to be defective in the GmNFR5 Nod-factor receptor (Indrasumunar et al., 2010) fail to activate the AON pathway.

The fact that leaf extracts from inoculated nts1116 and NOD4 plants failed to suppress nodulation in our bioassay confirms that SDI is indeed GmNARK-dependent. This is fully consistent with proposed model of AON, where SDI is synthesized downstream of the activation of NARK (Delves et al., 1986). Furthermore, GmNARK is strongly expressed in the phloem, particularly in the leaf veins (Nontachaiyapoom et al., 2007), and SDI is presumed to be transported in the phloem (Delves et al., 1986); these findings fit well with the observations presented here and the existing AON model (Ferguson et al., 2010).

Also consistent with the current model of AON is the fact that SDI activity correlated strongly with B. japonicum inoculation and, more specifically, is Nod factor-dependent. The suppression was only observed in the bioassay when leaf extracts were taken from wild-type plants inoculated with Nod factor-producing rhizobia. This dependency on the presence of Nod factor is consistent with the proposed ‘activated state’ hypothesis, where the Rhizobium-produced Nod factor stimulates cell division in the root, resulting in an ‘activated state’ leading to the eventual production of the root-derived signal, Q (Caetano-Anollés & Gresshoff, 1991; Li et al., 2009). This signal, or a product of its action, is then transported and perceived by the NARK protein kinase in the leaf, which consequently leads to the production/activation of the inhibitory signal, SDI (Caetano-Anollés & Gresshoff, 1991). This explains why nodule suppression was only observed using inoculated wild-type extracts where the ‘activated state’ was triggered. In the absence of Nod factor, the ‘activated state’ of the roots would not be reached, hence AON would not be induced and nodulation suppression by SDI would not be observed.

Using heat treatment, we showed that SDI is a heat-stable molecule. In addition, using filtration approaches, we demonstrated that SDI is a small molecule of less than 1 kDa. Control exclusion experiments suggest that the stated size limit of the filter may not be perfectly accurate, depending on the chemical and structural features of the molecule (M. Djordjevic, ANU, pers. comm.). Our bioassay results used inoculated wild-type leaf extracts treated with either proteinase K or RNase A. Proteinase K is a broad-range protease which cleaves the peptide bonds next to carboxylic side chains of aliphatic, aromatic, hydrophobic amino acids (Ebeling et al., 1974). RNase A is a ribonuclease that degrades most RNA molecules, including siRNA and miRNA (Raines, 1998; Haupenthal et al., 2006). As extracts treated with Proteinase K or RNase A caused nodule suppression equivalent to that of nonenzymatically treated control extracts, it is likely that SDI is neither a protein nor an RNA.

Feeding the wild-type extracts at different strengths shows that there is a minimum required concentration threshold for effective SDI suppression activity. It is possible that suppression activity was no longer observed at 1/27 of the original strength because SDI concentration in the extracts has fallen below the minimum required concentration. Although SDI concentration is significantly diluted at 1/9 strength, sufficient quantity of SDI were still present to induce a suppression effect. Therefore, the minimum threshold would be situated between 1/9 and 1/27 of the original strength.

Feeding inoculated wild-type leaf extracts from M. truncatula plants into nts1116 soybean plants caused strong suppression of nodule development, suggesting that the suppressive principle of M. truncatula is somewhat compatible with the AON system of soybean. However, partial suppression was also observed in sunn mutant leaf extracts indicating incompatibilities between determinate and indeterminate legume systems. Alternatively, a nonspecific inhibitor may exist in the Medicago systems, but be absent in the soybean systems. Suggestions for broader cross-legume conservation of the AON process have come from grafting Glycine species, Phaseolus, M. truncatula and L. japonicus shoots and rootstocks (Delves et al., 1992; Jiang & Gresshoff, 2002; Krusell et al., 2002; Lohar & VandenBosch, 2005). The AON receptor kinases MtSUNN, GmNARK and LjHAR1 share strong sequence homology, orthologous protein structures and expression patterns (Nontachaiyapoom et al., 2007). Collectively, these findings suggests that SDI may be a universal signal across legumes species even between determinate and indeterminate nodule-forming species (Hirsch, 1992; Doyle, 1998).

Future studies using the bioassay will need to involve extracts fractionated by high-performance liquid chromatography to purify SDI, followed by mass spectrometry and nuclear magnetic resonance techniques to identify the signaling molecule. Knowing the identity of SDI will provide functional insights into the AON mechanism, which may lead to agricultural optimization of nodulation.


We thank the Australian Research Council for provision of a Centre of Excellence grant (CEO348212), the Queensland State Government Smart State Innovation Scheme as well as the University of Queensland Strategic Fund for support. We thank Dr Richard Webb for assistance with sample preparation. Meng-Han Lin and Dugald Reid are thanked for technical assistance and bioassay optimization. Dongxue Li and Mikiko Miyagi are thanked for general assistance. The B. japonicum strain AN122 was generously provided by Prof. Gary Stacey.