The structure and activity of nodulation-suppressing CLE peptide hormones of legumes

Additional keywords: autoregulation of nodulation, legume nodulation, nitrate-regulation of nodulation, nodule, plant peptide signalling, symbiosis.

The common agricultural practice of using nitrogen-based fertilisers to increase crop yields has been highly successful in generating sufficient food for the world’s ever-growing population. It has been a major part of the ‘green revolution’ instigated more than 50 years ago. However, adverse economical and ecological consequences are beginning to outweigh the benefits of nitrogen fertiliser use (Erisman et al. 2008; Sutton et al. 2011; Jensen et al. 2012).

Symbiotic nitrogen fixation represents an alternative to chemical nitrogen fertiliser use. It involves a relationship mainly formed between plant species of the family Fabaceae, commonly known as legumes, and soil bacteria, collectively referred to as rhizobia. Major legume crop and pasture species include soybean, pea, common bean, clover, cowpea, medic, chickpea, lentil and peanut. Biological nitrogen fixation from this legume–rhizobia relationship currently results in ~50–70 Tg of nitrogen added into global agricultural systems each year (Herridge et al. 2008; Jensen et al. 2012).

The legume-rhizobia relationship is signified by the formation of a new plant organ, called the nodule. Nodule development is orchestrated by a complex signalling interaction (Ferguson and Mathesius 2003, 2014; Ferguson et al. 2010; Desbrosses and Stougaard 2011; Oldroyd 2013). Once formed, the nodule acts to house the rhizobia that provide the plant with a useable form of reduced nitrogen (namely ammonia) using a specialised enzyme complex to ‘fix’ un-reactive atmospheric di-nitrogen gas (N₂). In return, the rhizobia are provided with a carbon source derived from photosynthesis, predominately malate (Udvardi et al. 1988). In addition to increasing current crop yields, this process is exploited in agriculture to improve the nitrogen content and structure of soils by using legumes as rotation crops (Jensen et al. 2012).

Nodulation is costly to the host plant in terms of resources; as a result, the plant has developed both local and systemic mechanisms to control its nodule numbers (Delves et al. 1986; Gresshoff and Delves 1986; recently reviewed by Reid et al. 2011b). Local control mechanisms responding to high soil nitrate directly prevent or delay nodule development (Carroll et al. 1985a; Reid et al. 2011a). A systemic control mechanism, called the ‘autoregulation of nodulation’ (AON), is closely associated with the nitrate regulatory pathway, but is induced by rhizobia, not nitrate, and acts systemically through the shoot, rather than locally in the root (Kosslak and Bohlool 1984; Delves et al. 1986; Gresshoff and Delves 1986; Reid et al. 2011a, 2011b).

The AON process begins with the production of a root-derived signal (Gresshoff and Delves 1986), which is expressed in response to a transcription factor, called NIN, involved in cortical cell division during early nodulation events (Soyano et al. 2014). This signal, formerly called ‘Q’ (Gresshoff and Delves 1986), is now known to be a CLAVATA/Embryo surrounding region (ESR) related (CLE) peptide. To date, CLE peptide-encoding genes having a role in nodulation have been identified in Glycine max (L.) Merr. (soybean), Medicago truncatula Gaertn., Lotus japonicus (Regel) K.Larsen and Phaseolus vulgaris L. (common bean) (Fig. 1; Okamoto et al. 2009, 2013; Mortier et al. 2010, 2012; Lim et al. 2011; Saur et al. 2011; Reid et al. 2011a, 2013; Ferguson et al. 2014). Recent biochemical advances have enabled the isolation and identification of one of the nodulation CLE peptides of L. japonicus, called LjCLE-RS2. The mature signal of this CLE peptide is 13 amino acids in length, is derived from a much larger prepropeptide and is post-translationally modified with three β 1-2 linked arabinose moieties at Hyp7 (Okamoto et al. 2013).

Fig. 1.  Multiple sequence alignment and domain structure of the nodulation CLE prepropeptides. Shown are the amino acid sequences of the known nodulation-suppressing CLE peptides of Glycine max (soybean), Phaseolus vulgaris (common bean), Lotus japonicus and Medicago truncatula. The alignment was obtained using CLUSTALW multiple alignment (Larkin et al. 2007) in Geneious Pro 6.0. Shading of individual amino acids represents conservation amongst the prepropeptides, with the darker the shading the more highly conserved the residue. The CLE domain is highly conserved, with many other conserved residues found in the signal peptide and C-terminal extension domains. Conservation is particularly strong between orthologous genes of the different species. Not shown are the homeologous/duplicate copies of the soybean genes, which may have no-, reduced- or an alternative-function.

The nodulation-suppressing CLE peptide signal is exported from the root and transported via the xylem by an unknown mechanism (Okamoto et al. 2013) to the leaf phloem parenchyma (Nontachaiyapoom et al. 2007) where it is perceived by a leucine-rich repeat serine-threonine receptor kinase (LRR RK), called GmNARK in soybean, LjHAR1 in L. japonicus, MtSUNN in M. truncatula, PsSYM29 in Pisum sativum L. (pea), GsNARK in Glycine soja Siebold & Zucc., and PvNARK in common bean (Krusell et al. 2002; Nishimura et al. 2002; Searle et al. 2003; Schnabel et al. 2005; Ferguson et al. 2014). These LRR RKs may act in a complex with other receptors to perceive the CLE peptide ligand. This includes factors such as LjCLAVATA2/PsCLAVATA2 and LjKLAVIER (Miyazawa et al. 2010; Krusell et al. 2011). Additional research has identified other factors that may interact with the LRR RK directly, or function downstream of it, to relay the perception of the signal and trigger downstream signalling events. This includes the kinase-associated protein phosphatases, GmKAPP1 and GmKAPP2 (Miyahara et al. 2008), the putative ubiquitin fusion degradation protein, GmUFD1a (Reid et al. 2012) and the root-acting F-box protein, TOO MUCH LOVE, LjTML (Magori et al. 2009; Takahara et al. 2013). Following the perception of the nodulation CLE peptide signal, a shoot-derived inhibitor (SDI) signal is produced and transported to the roots, likely via the phloem, to inhibit further nodulation development (Delves et al. 1986; Lin et al. 2010, 2011; Reid et al. 2011b; Sasaki et al. 2014). Recent studies in L. japonicus have indicated a role for cytokinin as a potential SDI-candidate in AON (Sasaki et al. 2014).

The gene encoding for the LRR RK is expressed in both shoot and root tissues (Krusell et al. 2002; Nishimura et al. 2002; Searle et al. 2003; Schnabel et al. 2005; Nontachaiyapoom et al. 2007) and plants having mutations in it exhibit both supernodulation (due to a lack of AON control) and nitrate-tolerant nodulation phenotypes (e.g. Carroll et al. 1985a, 1985b). Grafting studies using soybean have demonstrated that the GmNARK LRR RK is required for both AON in the shoot (Delves et al. 1986; Reid et al. 2011a) and nitrate-regulation of nodulation in the root (Reid et al. 2011a). Similar to AON, the nitrate regulation of nodulation mechanism in soybean begins with the production of a CLE peptide that is predicted to be perceived by GmNARK. However, unlike AON, this CLE peptide, called GmNIC1, responds to nitrate, not rhizobia, and acts locally in the root, not systemically in the shoot. These findings helped to confirm that there are two independent pathways controlling nodulation: the systemic rhizobia-induced AON pathway and the local nitrate-induced regulation of nodulation pathway (reviewed by Reid et al. 2011b). No candidates for the root-derived inhibitor (RDI) have been identified to date, but as it is produced downstream of GmNARK, it may be similar to, or even the same as, the SDI signal in AON.

Unlike soybean, L. japonicus and M. truncatula appear to have overlapping local and systemic molecular mechanisms that act to regulate nodulation in response to both rhizobia and nitrate (Okamoto et al. 2009; Mortier et al. 2010). Although the reason for this difference amongst species is unknown, it is likely that it relates to genomic duplication events undergone in soybean that have enabled genetic divergence and the development of new molecular signals and mechanisms through the process of neofunctionalisation (Schmutz et al. 2010). This may also explain why soybean has three functional CLE peptides that are known to regulate nodule numbers, in addition to three homeologous (duplicate) copies that may have no-, reduced- or diverged-function (Reid et al. 2011a), whereas L. japonicus and M. truncatula appear to have only two such peptides (Table 1). Interestingly common bean, which shared a duplication event with soybean 53 million years ago, has orthologous copies of the three soybean CLE peptide genes, but lacks the duplicate copies of each of these genes as a result of not undergoing the more recent genome duplication event approximately 13 million years ago (Ferguson et al. 2014).

Table 1.  Known nodulation CLE peptides and their key reported features
ND, not determined

The CLE peptides that act as a trigger for AON and nitrate-regulation of nodulation belong to a large group of heavily processed, cysteine-poor secreted plant peptides related to AtCLV3 in Arabidopsis thaliana (L. Heynh.) (Matsubayashi 2014). AtCLV3 functions in the CLAVATA pathway to regulate the shoot apical meristem stem cell population. It acts as a ligand to a receptor-complex involving AtCLV1, AtCLV2 and AtCORYNE (Ogawa et al. 2008). The AtCLV1 receptor is a LRR RK that is highly similar in structure to the LRR RKs that are central to nodulation control. Other Arabidopsis CLE peptides of note that are similar to the nodulation CLE peptides of legumes include AtCLE1 to AtCLE7, which have roles in root architecture and development (Cock and McCormick 2001; Strabala et al. 2006; Oelkers et al. 2008; Araya et al. 2014).

The mechanisms controlling nodulation in legumes are highly conserved, as demonstrated by the interspecific function of AON CLE peptides from soybean in common bean and from M. truncatula in pea (Osipova et al. 2012; Ferguson et al. 2014). There are, however, many differences in the sequences, structures and inducing factors of the various nodulation CLE peptides that allow for specificity of function (Fig. 1; Tables 1, 2). These similarities and differences, and how they impact on nodule suppression, are reviewed here.

Table 2.  Amino acid sequence similarity (%) amongst the known nodulation CLE prepropeptides
Similarities are based on alignments obtained using CLUSTALW multiple alignment tool in Geneious Pro 6.0

Mature CLE peptide signals are derived from prepropeptides consisting of 3–4 domains: an N-terminal signal peptide, a variable region and a CLE domain, with some also having a C-terminal extension (Fig. 1). Sequence similarities amongst the nodulation CLE prepropeptides shows the orthologous copies are most similar (Fig. 2; Table 2); however, it is likely that similarities and differences in the individual domains are most critical for driving specificity. Here we discuss the function, conservation and importance of each domain, particularly in respect to their role in the suppression of nodulation.

Fig. 2.  Phylogenetic tree of nodulation CLE prepropeptides. The known nodulation-suppressing CLE peptides of Medicago truncatula, Lotus japonicus, Glycine max (soybean) and Phaseolus vulgaris (common bean) are shown, together with AtCLV2, the Arabidopsis CLE peptide most similar to the nodulation CLE peptides, and AtCLV3 as an outgroup. The tree was generated using PhyML 3.0 (Guindon and Gascuel 2003) in Geneious Pro 6.0 and constructed using the maximum likelihood approach. A branch was supported in 1000 bootstrap replications, with bootstrap confidence values expressed as a percentage of the 1000 bootstrap replications (Felsenstein 1985).

The N-terminal hydrophobic signal peptide (also referred to as a transit peptide) is widely thought to be responsible for exporting the prepropeptide out of the cell (la Cour et al. 2004; Lim et al. 2011). It is ~30 amino acids in length and is critical to the specificity of the peptide (Fletcher et al. 1999; Reid et al. 2013). This domain has a role in exporting the AtCLV3 propeptide into the extracellular space (Rojo et al. 2002). A similar role for this domain is predicted for the nodulation CLE peptides.

Amongst the known nodulation CLE prepropeptides, the signal peptide domain has a 37% pairwise identity and contains a leucine-rich motif (Fig. 1), commonly observed in exported proteins (la Cour et al. 2004). Also present within this domain is a conserved motif of five amino acids (TLQAR; Table 1), which is predicted to be a site of cleavage (Okamoto et al. 2009). It has been noted that GmNIC1, PvNIC1 and MtCLE12 show lower conservation of amino acid residues within this motif. They are also the only known nodulation CLE peptides to lack the C-terminal extension domain (Table 1; Fig. 1). Outside of this motif, conserved sequence residues within the signal peptide can be seen amongst predicted orthologues of the nodulation CLE peptides (Fig. 1; Reid et al. 2011a; Ferguson et al. 2014).

The functional importance of the variable domain, the least conserved of the four domains, remains unknown (Ni and Clark 2006; Meng et al. 2010; Reid et al. 2013). Indeed, AtCLV3 shows function without this domain (Fiers et al. 2006). The size of the domain is also highly variable (31–50 amino acids). However, recognition and cleavage immediately before the Arg1 residue of the CLE domain requires at least four to five residues of the variable domain to be present for correct processing of AtCLV3 (Kondo et al. 2008; Ni et al. 2011; Xu et al. 2013). An additional amino acid at residue 39 within the variable domain of AtCLV3 is also predicted to be a cleavage site (Xu et al. 2013).

There are no residues that are 100% conserved across the variable domain between the known nodulation CLE peptides (Fig. 1), although it shows a 19.4% pairwise identity and, as with other domains, residues are conserved between orthologues. This is particularly evident between the nodulation CLE peptides of the closely-related bean and soybean species (Fig. 1; Ferguson et al. 2014).

The CLE domain, from which the peptide is named, denotes the mature/active peptide sequence. It is located at the C-terminus and is the most conserved region (Cock and McCormick 2001; Oelkers et al. 2008). The consensus amino acid sequence of the nodulation-suppressing CLE peptides is RL(A/S)PGGPDPQHN(X) (Fig. 1). The domain is 12 or 13 amino acids in length and contains 50% identical sites, with 77.4% (12 amino acids) and 75.2% (13 amino acids) pairwise identity between the known nodulation CLE peptides. LjCLE-RS2 of L. japonicus is the only structurally-confirmed nodulation CLE peptide, and is 13 amino acids in length (Okamoto et al. 2013). However, the nitrate-induced GmNIC1 peptide of soybean and its orthologue in bean, PvNIC1, have a stop codon at position 13 and therefore can only be 12 amino acids in length (Fig. 1; Reid et al. 2011a; Ferguson et al. 2014). This may influence their functional properties, such as their apparent lack of long distance transport.

Notably, GmRIC1, PvRIC1, GmRIC2, and PvRIC2 are the only nodulation CLE peptides known to contain an Ala residue at position 3 of the CLE domain, presumably a result of polyploidisation and subsequent species divergence amongst the legumes (Figs 1, 2; Stefanović et al. 2009; Schmutz et al. 2010). There are four other residues within the CLE domain of the known nodulation CLE peptides that contain sequence divergence from the consensus sequence: Gly5 > Glu5 (GmRIC1 and PvRIC1) or Ala5 (MtCLE13); Asp8 > Asn8 (MtCLE12); Pro9 > His9 (MtCLE12) or Gln9 (GmNIC1); and Gln10 > His10 (GmRIC1 and PvRIC1) or Ile10 (MtCLE12) (Fig. 1). It is not yet known how the activity of the CLE peptide is affected by these sequence divergences. Only the one at position 8 in MtCLE12 is predicted to be critical for function (i.e. the suppression of nodulation) based on site-directed mutagenesis work using soybean (Reid et al. 2013); however, this nonsynonymous substitution from an uncharged asparagine to a negatively charged aspartic acid is conservative and may not affect activity.

Recent research has indicated that, despite sequence-redundancy of the CLE domain, there is likely some specificity between pathways and/or species that are dependent on sequence. Okamoto et al. (2013) were unable to elicit a plant response in L. japonicus from exogenous application of the mature AtCLV3 peptide, but saw a reduction in nodules when LjCLE-RS2 was applied with the correct post-translational modifications. Chimeric genes that swapped the CLE domains of GmNIC1 and GmRIC1 also impacted on the suppression of nodulation compared with their respective native genes (Reid et al. 2013). In contrast, GmRIC1 overexpression in common bean and MtCLE13 overexpression in pea strongly suppressed nodulation inter-specifically (Osipova et al. 2012; Ferguson et al. 2014), indicating that these CLE peptide encoding genes can function in the AON pathways of other legume species. However, overexpression results of any kind should always be interpreted with care.

The C-terminal domain of the known nodulation-suppressing CLE peptides is small, at ~6–9 residues in length, and is even completely absent from some (Fig. 1; Table 1). Indeed, of the known nodulation CLE peptides, GmNIC1, PvNIC1 and MtCLE12 all lack the C-terminal extension in its entirety (Mortier et al. 2010; Reid et al. 2011a; Ferguson et al. 2014). However, the remaining nodulation CLE peptides all contain the domain, as do AtCLV3 and GmCLV3 of the CLAVATA pathway (Fiers et al. 2006; Wong et al. 2013).

The C-terminal domain is thought to act as a protective mechanism from degradative protease enzymes in the xylem, which the peptides would encounter during systemic transport (Oelkers et al. 2008; Okamoto et al. 2009; Ni et al. 2011; Reid et al. 2011a). It is characteristic of the rhizobia-dependent, systemically acting, CLE peptides, and is not present in the nitrate-induced, locally-acting GmNIC1 of soybean and its orthologue in bean, PvNIC1 (Reid et al. 2011a; Ferguson et al. 2014). This would appear to further support a role for the domain in protection during long-distance xylem transport. Moreover, overexpressing a chimeric construct that added the C-terminal domain of GmRIC1 to GmNIC1 enhanced the suppression of nodulation compared with that of the native GmNIC1 (Reid et al. 2013). In contrast, the removal of the domain from GmRIC1 did not alter its ability to suppress when overexpressed (Reid et al. 2013), but this may be due to the overexpression technique masking or over-compensating for the true function of the modified construct. MtCLE12 also lacks the C-terminal domain and is both induced by rhizobia and predicted to be transported systemically (Mortier et al. 2010), so the exact need for the domain remains puzzling.

Two conserved proline residues are present within the C-terminal extension of all seven nodulation CLE peptides that contain the domain (Fig. 1; Okamoto et al. 2009; Mortier et al. 2010; Reid et al. 2011a; Ferguson et al. 2014). Site-directed mutagenesis and overexpression of GmRIC1 modified to encode two alanine residues in place of these two proline residues did not alter the suppressive activity of the peptide, consistent with the unclear role of this domain (Reid et al. 2013).

The mature nodulation-suppressive CLE peptide of L. japonicus, LjCLE-RS2, is 13 amino acids in length and is hydroxylated at Pro4 and Pro7, with Hyp7 further modified to contain three arabinose sugars connected via β-1-2-linkages. These modifications are predicted to be made in the extracellular fluids (Okamoto et al. 2013). This is consistent with mature AtCLV3, AtCLE2 and AtCLE9 glycopeptides, which also contain a Hyp7 having three linked L-arabinose sugars (Kondo et al. 2006; Ohyama et al. 2009; Okamoto et al. 2013; Shinohara and Matsubayashi 2013). All of the nodulation CLE peptides contain motifs associated with arabinose modifications that are present in other plant proteins/peptides (Matsubayashi 2014). The hydroxyproline O-arabinosyltransferase (HPAT) gene that controls CLE arabinosylation in Arabidopsis is called AtHPAT3 (Ogawa-Ohnishi et al. 2013). MtRDN1 and PsNOD3 are likely orthologues of AtHPAT3 and are thought to be responsible for the arabinosylation of the nodulation-suppressing CLE peptides (Ogawa-Ohnishi et al. 2013). Mutations in these genes result in a supernodulation phenotype (Jacobsen and Feenstra 1984; Postma et al. 1988; Sagan and Duc 1996; Li et al. 2009; Schnabel et al. 2011), indicating that the peptides require the arabinose sugars for their activity.

Application of synthesised arabinosylated-LjCLE-RS2 to leaves of L. japonicus plants caused a reduction in nodulation in an LjHAR1-dependent manner (Okamoto et al. 2013). However, root or shoot application of synthetic nodulation CLE peptides devoid of modifications did not affect nodulation, although altered root growth was observed (Okamoto et al. 2009; Saur et al. 2011). Moreover, application of AtCLV3 with the arabinose modifications also had no effect on nodulation (Okamoto et al. 2013). Shinohara and Matsubayashi (2013) demonstrated that the binding of the AtCLV3 CLE peptide to the AtCLV1 LRR receptor-kinase declined as the arabinose chain length decreased, whereas AtCLE9 showed no change in receptor binding efficacy to its receptor, BAM1, a CLV1/BAM-family LRR RK, in the absence of the arabinose chain (Shinohara et al. 2012). Further, tracheary element differentiation inhibitory factor (TDIF) peptides synthesised with or without hydroxyproline residues can mimic the function of the naturally occurring peptide, which contains Hyp4 and Hyp7 (Sawa et al. 2006).

In addition to post-translational modifications to critical residues, the structural configuration of the CLE peptide ligand is also likely to impact markedly on receptor interactions. Gly6 is proposed to allow for rotation, most likely because of its small size, complementing evidence for a boomerang curve in the peptide’s configuration, with both ends of the peptide bending away from the arabinosylation at Hyp7 (Okamoto et al. 2013; Shinohara and Matsubayashi 2013; Song et al. 2013). Notably, Gly6 is 100% conserved amongst the known nodulation CLE peptides (Fig. 1). Site-directed mutagenesis of Gly6 to Ala6 significantly reduced the nodule suppressive activity of GmRIC1 (Reid et al. 2013). Song et al. (2013) altered Gly6 of AtCLV3 into 18 other amino acids; no substitution was able to rescue the phenotype of Atclv3 mutant plants. Similar specificity is expected for the nodulation CLE peptides. In addition to Gly6 and Pro7, residues Arg1, Pro4, Asp8, His11 and Asn12 of GmRIC1 were required for full nodulation-suppression activity in soybean (Reid et al. 2013). Similarly, TDIF also lost activity when the CLE domain residues His1, Val3, Gly6, Asn8, Pro9 and Asn12 were changed into an alanine residue via site-directed mutagenesis (Ito et al. 2006; Sawa et al. 2006).

It has been noted that locally-acting CLE peptides, including GmNIC1 and PvNIC1 (Fig. 1), in addition to AtCLV3, GmCLV3 and LjCLV3, all contain His12 (Reid et al. 2011a; Okamoto et al. 2011; Wong et al. 2013; Ferguson et al. 2014). This may indicate a role for this residue in local, but not systemic, transport of the peptide. Constructs having swapped the CLE domain of the systemically-acting GmRIC1 and the locally-acting GmNIC1 showed an altered inhibition of nodulation when overexpressed compared with the native peptides (Reid et al. 2013). Whether residue 12 plays a specific role in the transport or recognition of the peptide is of interest to determine.

As noted above, Arg1 of the AtCLV3 CLE domain has been shown to be critical for binding and processing of the mature CLE peptide, with at least 4–5 residues upstream of Arg1 required for proper recognition of the signal (Kondo et al. 2008; Ni et al. 2011; Xu et al. 2013). It is hypothesised that a subtilisin with endoproteolytic activity cleaves the CLE peptide, with a carboxypeptidase processing the C-terminal extension where present (Ni et al. 2011; Djordjevic et al. 2011). However, to date, there is little known about the mechanisms and sites of proteolytic cleavage in the nodulation CLE peptides.

All of the known nodulation-suppressing CLE peptides are upregulated in expression by the presence of rhizobia and/or the available soil nitrogen content (Table 1). Phylogenetic analysis shows that they cluster according to their mode of induction (Fig. 2). Evidence for other environmental factors such as phosphate and soil acidity, inducing or influencing the expression of CLE peptide-encoding genes also exists.

The presence of compatible rhizobia, and possibly more specifically the rhizobia-produced Nod factor signal, elicits the expression of systemically-acting CLE peptide-encoding genes that function in AON. These CLE peptides include: LjCLE-RS1, LjCLE-RS2, MtCLE12, MtCLE13, GmRIC1, GmRIC2, PvRIC1, and PvRIC2 (Table 1; Okamoto et al. 2009, 2013; Mortier et al. 2010, 2012; Lim et al. 2011; Reid et al. 2011a, 2013; Saur et al. 2011; Hayashi et al. 2012; Ferguson et al. 2014). Overexpression of these peptides in wild-type legume plants results in a complete abolishment of nodulation, but does not alter the nodulation pattern in NARK mutants, demonstrating that they act in a NARK-dependent manner (Okamoto et al. 2009; Mortier et al. 2010; Reid et al. 2011a; Lim et al. 2011).

Laser microdissection of root sections indicate that LjCLE-RS1 and -RS2 are expressed in the stele and outside of the endodermis (cortex and epidermis) (Okamoto et al. 2009). Promoter:GUS reporter fusion studies have shown that MtCLE13 is expressed in the inner cortex during early nodulation and later in dividing cells of the cortex and pericycle. In contrast, MtCLE12 is not expressed early but instead is expressed throughout young nodules and in meristematic tissues of the elongating indeterminate nodule (Mortier et al. 2010). Finally, Lim et al. (2011) have shown that GmRIC2 is expressed in the pericycle and inner cortex during early nodule development, and later in the outer cortex of more developed nodules.

Time-course experiments have revealed different but overlapping expression patterns for these genes within a species (Table 1). Soybean GmRIC1 is induced early (within 12 h) after inoculation with infection-capable (Nod factor producing) Bradyrhizobium japonicum, whereas GmRIC2 expression is induced later (48–72 h) and remains elevated in expression for longer (Reid et al. 2011a; Hayashi et al. 2012). The rhizobia-induced peptide encoding genes of common bean, PvRIC1 and PvRIC2, exhibit a similar pattern of expression (Ferguson et al. 2014). Likewise, M. truncatula MtCLE13 is expressed earlier than MtCLE12, although both are also expressed in later stages of nodulation (Mortier et al. 2010, 2012). LjCLE-RS1 and LjCLE–RS2 are both upregulated within 3 h of inoculation (Okamoto et al. 2009). Similar to GmRIC2, MtCLE13 and LjCLE-RS1 transcript levels appear to remain elevated for longer compared with MtCLE12 and LjCLE–RS2, respectively (Okamoto et al. 2009; Mortier et al. 2010; Reid et al. 2011a).

When compared with wild-type plants, a significant increase in expression of both LjCLE-RS1 and LjCLE–RS2 was also observed in the hypernodulating mutant of L. japonicus, too much love, possibly indicating that their synthesis is directly linked to the number of nodules being formed (Magori and Kawaguchi 2010). Interestingly, the plant hormone cytokinin, which has a role in early nodule development (reviewed in Ferguson and Mathesius 2014), has also been shown to induce the expression of some nodulation-suppressing CLE peptide genes (Lim et al. 2011; Mortier et al. 2010, 2012), consistent with the idea that the initiation of the AON pathway is linked to early cell divisions. Additional studies are required to further understand the expression patterns of these nodulation-suppressing CLE peptides, both within and between species.

GmNIC1 and LjCLE-RS2 are the only nodulation-suppressing CLE peptide-encoding genes that are confirmed to respond to nitrate (Table 1; Okamoto et al. 2009; Reid et al. 2011a). GmNIC1 is specifically induced by nitrate and not co-induced by the rhizobial microsymbiont, whereas LjCLE-RS2 is reported to be induced by both. PvNIC1 is also likely to be induced by nitrate as the candidate orthologue of GmNIC1 (Ferguson et al. 2014). To date, no CLE peptide-encoding gene of M. truncatula has been reported to respond to nitrate, although evidence suggests the existence of a locally acting, nitrate-responsive mechanism that acts in a MtSUNN-dependent manner to regulate nodulation (Jeudy et al. 2010). We note that AtCLE2, the Arabidopsis gene most similar to the nodulation-suppressing CLE peptides, has a role in root development and is also induced by nitrate (Scheible et al. 2004; Araya et al. 2014).

Overexpression of the locally-acting GmNIC1 in wild-type soybean reduces nodule numbers by ~50% compared with empty vector controls (Reid et al. 2011a). Although significant, this suppressive ability is far from that of GmRIC1 and GmRIC2, as discussed above. Confirmation is required to determine whether this is unique to soybean or is shared with the closely related orthologues identified in common bean (Ferguson et al. 2014).

Numerous factors can influence the extent of nodulation and it is possible that some do so by inducing, or otherwise influencing, the production, transport, perception or response to a CLE peptide(s). Recently, split-root and grafting studies using soybean grown in low pH conditions revealed a novel systemic mechanism that acts via GmNARK in the shoot to inhibit nodulation of the root (Lin et al. 2012; Ferguson et al. 2013). This suggests that soil acidity may act via a CLE peptide to suppress nodulation. Two CLE peptide-encoding genes of L. japonicus, called LjCLE19 and LjCLE20, have been shown to be upregulated in the presence of phosphate (Funayama-Noguchi et al. 2011); however, a specific role for these peptides in plant development has not been reported. It has been noted that although CLE peptides are nearly-exclusive to plants, they also exist in plant-parasitic nematodes (e.g. Bakhetia et al. 2007), which appear to use the peptides to initiate the formation of feeding structures in host roots (reviewed by Mitchum et al. 2012). Also noted is that nematodes are easily genetically transformed through simple feeding, suggesting that perhaps nematode CLE genes were plant-derived. Whether nematodes, or any other pathogen, can also induce a plant-encoded CLE peptide(s) is of great interest to determine.

In addition to the nodulation-suppressing CLE peptides reported above, two further CLE peptide-encoding genes have been identified that are expressed in response to rhizobia inoculation, namely LjCLE3 (Okamoto et al. 2009) and MtCLE4 (Mortier et al. 2010). Overexpression of these genes does not alter the nodulation phenotype when compared with empty vector controls. It is now of interest to determine the role of these peptides in nodulation to determine why they are responsive to rhizobia inoculation and how they function in this symbiosis.

The role of CLE peptides functioning as hormone signals in plant development is only just beginning to emerge, with the activity of most remaining to be elucidated. How and where the CLE peptides are induced, whether they are transported and act locally or systemically, how they are perceived, the downstream signals they induce, and their precise role in various plant developmental pathways are all features of great interest to establish in this burgeoning research field. Indeed, establishing the function of CLE peptides acting in critical plant developmental processes will considerably help to advance the current molecular knowledgebase of a variety of plant signalling networks.