Evidence for structurally specific negative feedback in the Nod factor signal transduction pathway

Authors


For correspondence (fax: +1 650 725 8309; email: srl@leland.stanford.edu)

Summary

Nod factor is a critical signalling molecule in the establishment of the legume/rhizobial symbiosis. The Nod factor of Sinorhizobium meliloti carries O-sulphate, O-acetate and C16:2N-acyl attachments that define its activity and host specificity. Here we assess the relative importance of these modifications for the induction of calcium spiking in Medicago truncatula. We find that Nod factor structures lacking the O-sulphate, structures lacking the O-acetate and N-acyl groups, and structures lacking the O-acetate combined with a C18:1N-acyl group all show calcium spiking when applied at high concentrations. These calcium responses are blocked in dmi1 and dmi2 mutants, suggesting that they function through the Nod factor signal transduction pathway. The dmi3 mutant, which is proposed to function in the Nod factor signal transduction pathway downstream of calcium spiking, shows increased sensitivity to Nod factor. This increased sensitivity is only active with wild-type Nod factor and was not present when the plants were treated with mutant Nod factor structures. We propose that the Nod factor signal transduction pathway is under negative feedback regulation that is activated at or downstream of DMI3 and requires structural components of the Nod factor molecule for activity.

Introduction

Legumes have the capacity to form a symbiotic interaction with rhizobial bacteria. This interaction results in the formation of bacterially infected nodules on the roots of the host plant. Bacterially derived Nod factor is a critical signalling molecule for the establishment of the legume/rhizobial symbiosis (Dénariéet al., 1996; Long, 1996; Spaink, 2000). Isolated Nod factor applied to the appropriate plant host induces many of the responses elicited by the bacterial symbiont, including root hair deformation, cortical cell division and gene expression (Downie and Walker, 1999). Nod factors also activate a number of very early responses in the plant, notably the induction of ionic fluctuations across the plasma membrane and calcium spiking in the cytosol (Cardenas et al., 2000; Ehrhardt et al., 1992; Ehrhardt et al., 1996). Nod factor is necessary but insufficient for the induction of the infection thread formation involved in the invasion of rhizobia into the plant.

Specificity in the legume/rhizobial symbiosis is encoded in the structure of the Nod factor signal. All Nod factors consist of an oligomeric chitin backbone (3–5 N-acetylglucosamine residues) with an N-acyl group attached to the non-reducing terminal sugar (Figure 1a,c). A number of additional substituents are attached to the chitin backbone, and these substituents vary widely between Nod factors generated by different Rhizobium species (Dénariéet al., 1996; Downie and Walker, 1999; Long, 1996). The specificity of the plant/bacterial interaction is defined in a number of cases by the structure of the N-acyl group and the substituents attached to the Nod factor molecule. The Nod factor of Sinorhizobium meliloti, the symbiont of Medicago sativa and Medicago truncatula, carries an O-sulphate group on the reducing terminal sugar and an O-acetyl group and a specific C16:2N-acyl chain on the non-reducing terminal sugar (Figure 1c; Truchet et al., 1991).

Figure 1.

Nod factors and oligosaccharides assayed in this study.

(a) Tetra-N-acetylglucosamine (chitotetraose). (b) Sulphated chitotetraose. (c) S. meliloti Nod factor. Positions marked indicate functionally significant modifications to the molecule. Position (i) shows the O-acetyl group that is attached through the activity of NodL. Position (ii) indicates the O-sulphate group that is attached through the activity of NodH. Position (iii) indicates the N-acyl group. The structure of this group is defined by the activity of NodE and NodF.

Genetic studies in rhizobia have shown roles for the nod genes in the production of Nod factor. The nodABC genes are involved in the generation of the basic Nod factor structure, with NodC being responsible for the condensation of N-acetyl glucosamine residues to form the oligomeric chitin backbone, and NodB being responsible for the removal of a specific N-acetyl group which then allows NodA to attach the lipid moiety (Spaink, 2000). In S. meliloti, NodF and NodE are involved in the synthesis and transfer of the appropriate C16:2 lipid moiety (Ardourel et al., 1994; Demont et al., 1993; Spaink, 2000). S. meliloti NodH is a sulphotransferase and NodL is an acetyltransferase; these enzymes function in the attachment of the O-sulphate and O-acetyl groups, respectively (Ardourel et al., 1995; Bloemberg et al., 1994; Roche et al., 1991b; Spaink, 2000). Mutations in the nodABC genes block all responses of the host plant to the bacterial symbiont. Mutations in S. meliloti nodH block most host plant responses, indicating an absolute requirement for the sulphate group in the activity of this Nod factor (Faucher et al., 1988; Faucher et al., 1989; Roche et al., 1991a). S. meliloti strains carrying single mutations in either the nodF, nodE or nodL genes show slightly delayed nodulation, but few additional phenotypes (Ardourel et al., 1994). However, strains carrying a double mutation in the nodF and nodL genes are blocked for infection but still induce root hair deformation, cortical cell activation and gene expression in the host plant (Ardourel et al., 1994).

The model legume M. truncatula has been developed to aid in the genetic dissection of this symbiosis (Barker et al., 1990; Cook, 1999; Handberg and Stougaard, 1992). A number of nodulation-deficient (Nod) mutants of M. truncatula define specific steps in the Nod factor signal transduction pathway (Catoira et al., 2000; Oldroyd, 2001). Three Nod mutants of M. truncatula, altered in DMI1, DMI2 and DMI3 genes, show very little response to Nod factors, and these genes have been proposed to act early in the Nod factor signal transduction pathway (Catoira et al., 2000). These Nod mutants have been characterized for their calcium spiking response (Wais et al., 2000). dmi1 and dmi2 are blocked, while dmi3 is active for calcium spiking. Therefore, DMI1 and DMI2 were postulated to act upstream of calcium spiking in the Nod factor signal transduction pathway, while DMI3 is thought to act downstream of calcium spiking, but probably still early in the pathway. A fourth set of mutants, at the NSP locus, are Nod, but show a number of responses to Nod factor or S. meliloti and are active for calcium spiking (Catoira et al., 2000; Wais et al., 2000).

Genetic studies in M. truncatula have demonstrated a role for the plant hormone ethylene in the negative regulation of nodulation. skl, an ethylene-insensitive mutant of M. truncatula, shows a supernodulation phenotype that is associated with the defect in ethylene perception (Penmetsa and Cook, 1997). Further studies have shown that ethylene negatively regulates a component of the Nod factor signal transduction pathway at or upstream of calcium spiking (Oldroyd et al., 2001). This inhibition affects the number of cells activated for calcium spiking at set Nod factor concentrations. This measurement has been used to assay the responsiveness of the plant to Nod factor. The skl mutant was shown to have a 10-fold increase in sensitivity to Nod factor relative to wild-type plants (Oldroyd et al., 2001).

Here we assess the activity of different S. meliloti Nod factor and Nod factor-like structures for the induction of calcium spiking in M. truncatula and their activity in Nod mutants. We show that the most basic structure, the oligomeric chitin backbone, can activate calcium spiking when applied at high concentrations. Nod factors altered at either the reducing or non-reducing terminal sugar show dramatically reduced activity compared to wild-type Nod factor, but are more active than chitin alone. These Nod factors do not induce calcium spiking in the Nod mutants dmi1 and dmi2. The dmi3 mutant, however, shows increased responsiveness to Nod factor for the induction of calcium spiking relative to wild-type plants. This increased responsiveness was only seen with wild-type Nod factor and not with the additional structures tested. We propose that a component of the Nod factor signal transduction pathway at or upstream of calcium spiking is under negative feedback and that the negative feedback requires specific components on the Nod factor molecule for activation.

Results

The chitin oligomer is sufficient to induce a calcium spiking response when applied at high concentrations

It has been shown in Pisum sativum that, at high concentrations, the chitin oligomer alone can activate a calcium spiking response (Walker et al., 2000). We therefore tested the activity of tetra-N-acetyl-glucosamine (chitotetraose), which forms the backbone of S. meliloti Nod factor (Figure 1a), for the induction of calcium spiking in M. truncatula. We found that at high concentrations this structure could induce calcium spiking (Figure 2; Table 1). At 10−6 m chitotetraose, 14% of root hair cells showed calcium spiking; 38% of cells showed spiking at 10−5 m and 75% of cells showed spiking at 10−4 m. In most cases, the spiking was indistinguishable from Nod factor-induced calcium spiking, activated by 10−12−10−9 m Nod factor (Figure 2b). However, in a few cases, chitotetraose-induced calcium oscillations showed reduced periodicity and amplitude compared with Nod factor-induced calcium spiking (Figure 2a).

Figure 2.

Response of wild-type M. truncatula plants to chitotetraose.

The traces show the change in calcium levels in individual cells treated with different concentrations of chitotetraose and Nod factor. The time at which a specific concentration of chitotetraose or Nod factor was added is indicated on the trace. (a) An example in which the spiking response changes at differing chitotetraose concentrations and Nod factor application. (b) An example in which there is no difference in the spiking response with higher chitotetraose concentrations or Nod factor-induced spiking. The y axis represents the change in florescence after treatment of the raw fluorescence data with the equation Y = X(n + 1) − Xn. This transformation reduces the level of background fluctuations and amplifies rapid changes in the calcium levels, thus aiding in the identification of calcium spikes. CT4, chitotetraose; NF, Nod factor.

Table 1.  The number of cells and plants analysed in each treatment
GenotypeNumber of cells (number of plants) analysed
WT NFnodH NFLCO IV, C16:2, SLCO IV, C18:1, SCT4SCT4
  1. NF, Nod factor; CT4S, sulphated chitotetraose; CT4, chitotetraose.

WT56 (5)37 (5)34 (4)42 (5)41 (5)29 (6)
dmi349 (5)42 (5)43 (4)35 (4)41 (5)11 (3)
dmi3 (ACC)46 (5)     
nsp37 (4)     

As the O-sulphate group of the S. meliloti Nod factor is critical for induction of host plant responses, we tested the ability of sulphated chitotetraose (Figure 1b) to trigger calcium spiking. This molecule induced calcium spiking in 88% of root hair cells when applied at 10−7 m (Figure 3; Table 1), indicating that sulphated chitotetraose is approximately 1000-fold more active than chitotetraose for the induction of calcium spiking in M. truncatula.

Figure 3.

Dose response curves for different Nod factors and Nod factor-like structures on wild-type M. truncatula.

Multiple cells on wild-type plants were assessed for their calcium spiking response to various concentrations of Nod factors and chitin oligomers. The percentage of cells that show calcium spiking at each Nod factor concentration is indicated. CT4S, sulphated chitotetraose; NF, Nod factor.

The induction of calcium spiking by alternative Nod factor structures

Modifications of the terminal N-acetyl glucosamine residues of the S. meliloti Nod factor modulate its activity on host plants (Ardourel et al., 1994; Roche et al., 1991a). S. meliloti strains mutated in the nodH gene, which generate a Nod factor lacking the O-sulphate attachment on the reducing terminal sugar (Figure 1c), do not induce calcium spiking and are unable to activate a variety of additional host responses (Roche et al., 1991a; R.J.W. and S.R.L., unpublished data). S. meliloti strains carrying nodF/nodL mutations, the gene products of which modify the non-reducing terminal sugar (Figure 1c), activate calcium spiking as well as a number of additional responses in M. truncatula (Ardourel et al., 1994; R.J.W. and S.R.L., unpublished data). To further test the relative importance of these modifications on S. meliloti Nod factor, we assessed the activation of calcium spiking by Nod factors altered at both the reducing and non-reducing terminal sugars.

We found that nodH-derived Nod factor fails to activate calcium spiking in M. truncatula within the concentration range 10−12−10−9 m, a concentration range that activates calcium spiking with wild-type Nod factor. However, at higher concentrations, nodH Nod factor induces calcium spiking. To analyse the responsiveness of the plant to this unsulphated Nod factor, we assessed the number of root hairs that initiate calcium spiking at concentrations of nodH Nod factor ranging from 10−10−10−7 m (Figure 3; Table 1). The threshold concentration at which 50% of root hairs show calcium spiking is 5.6 × 10−8 m, compared to 2 × 10−12 m for wild-type Nod factor. This indicates that the loss of the sulphate group on the reducing terminal sugar of S. meliloti Nod factor leads to a reduction in activity of approximately 30 000-fold.

Modifications on the non-reducing terminal sugar of S. meliloti Nod factor are less critical for activity relative to the presence of the sulphate on the reducing terminal sugar (Figure 1c; Ardourel et al., 1994; Ardourel et al., 1995; Demont et al., 1993). However, Nod factors carrying a C18:1N-acyl group in place of the normal C16:2 structure, coupled with loss of the O-acetate, are incapable of activating infection structures, but can induce root hair deformation, cortical cell activation and gene expression (Ardourel et al., 1994). To assess the relative importance of modifications on the non-reducing terminal sugar for the induction of calcium spiking, we tested the synthetic Nod factor lipo-chito-oligosaccharide (LCO) IV, C18:1, S, which carries a C18:1N-acyl group and lacks the O-acetate (Demont-Caulet et al., 1999). This structure activates calcium spiking at concentrations of 10−11−10−10 m (Figure 3; Table 1). The threshold concentration at which 50% of cells show spiking is 3 × 10−10 m, indicating approximately 100-fold reduction in activity compared to wild-type Nod factor. To control for the use of synthetic Nod factors, we assessed the activity of LCO IV, C16:2, S, a synthetic equivalent of the Nod factor generated by an S. meliloti nodL mutant. NodLmutant strains act similarly to wild-type S. meliloti strains when applied to the plant. LCO IV, C16:2, S shows the same level of activity as wild-type Nod factor (Figure 3; Table 1).

M. truncatula dmi1 and dmi2 mutants do not show calcium spiking in response to any of the Nod factor or Nod factor-like structures

To assess whether the calcium spiking responses observed with high concentrations of altered Nod factors involve the same signal transduction pathway as wild-type Nod factor, we assessed these responses in M. truncatula dmi1-1 and dmi2-1 mutants. These mutants fail to induce calcium spiking in response to wild-type Nod factor (Wais et al., 2000). We found that the mutants were unable to elicit calcium spiking in response to 10−7 mnodH Nod factor or 10−7 m LCO IV, C18:1, S (Table 2). Furthermore, dmi1-1 and dmi2-1 mutants did not induce spiking with 10−7 m sulphated chitotetraose and dmi1-1 mutants did not show calcium oscillations with 10−4 m chitotetraose (Table 2). Hence, we conclude that all these responses are a function of the same genetic pathway that is involved in the induction of calcium spiking in response to wild-type Nod factor.

Table 2.  Calcium spiking with mutant Nod factors in dmi1 and dmi2: number of cells spiking/total cells (number of plants) analysed
 nodH NFLCO IV, C18:1, SCT4SCT4
  1. NF, Nod factor; CT4S, sulphated chitotetraose; CT4, chitotetraose.

dmi1–10/23 (4)0/16 (4)0/19 (4)0/12 (5)
dmi2–10/19 (4)0/13 (4)0/14 (4) 

dmi3, but not nsp, mutants show increased sensitivity to Nod factor

Previous examination of calcium behaviour in M. truncatula Nod mutants suggested that the dmi3 mutant may have a higher percentage of responsive root hair cells than wild-type plants presented with 10−9 m Nod factor (Wais et al., 2000). A possible explanation for this observation is that the dmi3 mutant has increased Nod factor sensitivity. To test this hypothesis, we analysed the dose response of the dmi3-1 mutant to wild-type Nod factor for the induction of calcium spiking. We found that this mutant shows increased sensitivity to Nod factor (Figure 4). The threshold concentration at which 50% of root hairs display calcium spiking in dmi3-1 is 3.6 × 10−13 m, compared with 2 × 10−12 m in wild-type plants, indicating an approximately 10-fold increase in sensitivity in the dmi3-1 mutant. As the DMI3 complementation group has only one allele, we were unable to test allelic variation for the increased sensitivity. We also examined the Nod factor sensitivity of nsp-1 mutants, which are blocked for nodulation but permissive for root hair deformation, gene expression and calcium spiking. nsp-1 mutants showed wild-type Nod factor sensitivity (Figure 4), suggesting that the increased sensitivity observed in dmi3-1 mutant is a function of a component of the Nod factor signal transduction pathway at or downstream of DMI3, but upstream of NSP.

Figure 4.

Dose response curves for dmi3-1 and nsp-1 mutants with S. meliloti Nod factor.

The percentage of cells that spike at various Nod factor concentrations was assessed for wild-type plants and the dmi3-1 and nsp-1 mutants. Note that the dmi3-1 mutant shows a 10-fold increase in sensitivity to Nod factor.

Ethylene inhibition of calcium spiking is functional in dmi3-1 mutant

Previous studies have indicated a 10-fold increased Nod factor sensitivity in the ethylene-insensitive skl mutant (Oldroyd et al., 2001). Due to the similarities in Nod factor sensitivities between the dmi3-1 and skl mutants, we hypothesized that ethylene inhibition of Nod factor signalling may be blocked in the dmi3-1 mutant. To test this hypothesis, we compared the Nod factor sensitivity of dmi3-1 plants grown on either 0.1 µm l-α-(2-aminoethoxyvinyl)-glycine (AVG), which inhibits ethylene biosynthesis, or 10 µm 1-amino-cyclopropane-1-carboxylic acid (ACC), which induces ethylene biosynthesis. We have previously shown that wild-type plants grown under these two conditions differ dramatically in their sensitivity to Nod factor (Oldroyd et al., 2001). We found that dmi3-1 plants grown on ACC showed reduced sensitivity to Nod factor for induction of calcium spiking compared to plants grown on AVG (Figure 5). We therefore conclude that ethylene inhibition is functional in the dmi3-1 mutant, and hence the increased Nod factor sensitivity in the dmi3-1 mutant must have an alternative explanation.

Figure 5.

Ethylene inhibition is functional in dmi3–1.

The effect of ethylene on the sensitivity of the dmi3-1 mutant to Nod factor was assessed by measuring the percentage of cells activated for calcium spiking at different Nod factor concentrations in plants treated with either 0.1 µm l-α-(2-aminoethoxyvinyl)-glycine (AVG) or 10 µm 1-amino-cyclopropane-1-carboxylic acid (ACC).

The increased sensitivity of dmi3 mutants is structurally specific for the Nod factor molecule

As the dmi3-1 mutant shows increased Nod factor sensitivity, we assayed the dmi3 response to mutant Nod factors and chitin oligomers. We performed dose response experiments in dmi3-1 using nodH Nod factor, LCO IV, C18:1, S and LCO IV, C16:2, S. We found that dmi3-1 mutants did not show an increased sensitivity to these structures compared to wild-type plants (Figure 6). There is a slight increase in the sensitivity of dmi3-1 to LCO IV, C18:1, S; however, this increase is small compared with that observed with wild-type Nod factor. Furthermore, dmi3-1 responded in a similar way to chitotetraose and sulphated chitotetraose at the equivalent concentrations as wild-type plants (Table 1). This suggests that the increased sensitivity in dmi3-1 is structurally specific for the wild-type Nod factor molecule.

Figure 6.

dmi3-1 does not show increased sensitivity to altered Nod factor structures.

Dose response curves for mutant Nod factors in wild-type and dmi3-1 plants, as assessed by the number of root hair cells spiking at different Nod factor concentrations. (a) LCO IV, C16:2, S. (b) LCO IV, C18:1, S. (c) nodH Nod factor.

Discussion

From studies in S. meliloti, it has been shown that components of the Nod factor molecule are critical for its activity on the host plant (Dénariéet al., 1996; Downie and Walker, 1999; Downie, 1998; Long, 1996; Spaink, 2000). Here we assess the relative importance of modifications of the reducing and non-reducing terminal sugars of the S. meliloti Nod factor for induction of calcium spiking in M. truncatula. We show that the presence of the O-sulphate group on the reducing terminal sugar is critical for Nod factor activation of calcium spiking. Alterations of the N-acyl group, coupled with loss of the O-acetyl group on the non-reducing terminal sugar modulate Nod factor induction of calcium spiking, but are less critical than the O-sulphate attachment. dmi1 and dmi2 mutants do not show calcium spiking in response to any of the mutant Nod factors, indicating that these two genes are part of a common or shared Nod factor signalling pathway. The dmi3 mutant shows increased sensitivity to wild-type Nod factor, but not mutant Nod factors or chitin oligomers. We have shown that the increased sensitivity in dmi3 cannot be explained by a block in ethylene inhibition of the Nod factor signalling pathway. We propose that DMI3 or a component downstream of DMI3 is involved in negative feedback regulation that is structurally specific for the Nod factor molecule.

The simplest structure tested that triggered calcium spiking in M. truncatula was chitotetraose, which forms the backbone of S. meliloti Nod factor. Chitotetraose also induces calcium oscillations in P. sativum, at a concentration of 10−8 m, compared with 10−6 m in M. truncatula. Similar to some of our observations in M. truncatula, the calcium spiking response in P. sativum showed increased frequency (shorter period between spikes) when the chitotetraose concentration was raised (Walker et al., 2000). M. truncatula, unlike P. sativum, requires an O-sulphate attachment on the reducing terminal sugar of the Nod factor molecule. Sulphated chitotetraose shows approximately 1000-fold increased activity in M. truncatula compared to chitotetraose, and has a comparable dose response to chitotetraose in P. sativum.

S. meliloti nodH mutants show very few responses in host plants (Roche et al., 1991a; Vernoud et al., 1999). In contrast, strains carrying double mutations in the nodFand nodL genes are unable to infect the plant, but can induce root hair deformation, cortical cell activation, early gene expression and calcium spiking (Ardourel et al., 1994; Vernoud et al., 1999; R.J.W. and S.R.L., unpublished data). In the dose response assays in this study, we found that the activity of LCO IV, C18:1, S, equivalent to the nodF/nodL Nod factor, was intermediate between that of wild-type Nod factor and nodH Nod factor. A limitation in this comparison, however, is that we analysed Nod factors generated using different approaches, and the resultant Nod factor isolations may differ with regard to purity and stability.

From initial studies in Medicago species with S. meliloti nodF/nodL, it has been proposed that there are two receptors for Nod factor: a stringent receptor involved in the initiation of infection thread formation that requires the O-acetyl and appropriate N-acyl attachments to the reducing terminal sugar, and a less stringent receptor that can tolerate alterations in the N-acyl group and loss of the O-acetyl group and is involved in root hair deformation, cortical cell activation and early gene expression (Ardourel et al., 1994). In this model, both receptors would require the O-sulphate attachment. An alternative explanation for the nodF/nodL studies states that the different Nod factor structures can differentially activate a single Nod factor receptor, with the outcome for the induction of plant responses being dependent upon the mode of pathway activation (Hirsch, 1992; Long, 1996; Oldroyd, 2001). We have analysed a number of molecules that could differentially activate the Nod factor pathway or alternative Nod factor receptors. Our data indicate that both stringent and non-stringent perception of Nod factor involves calcium spiking. Furthermore, we have shown that the induction of calcium spiking by the diverse structures tested is a function of the same pathway as dmi1 and dmi2 mutants are blocked for this induction. However, this does not rule out the possibility that two receptors exist, which converge into one pathway at or above DMI1 or DMI2. Our findings on the activity of LCO IV, C18:1, S, suggests that this molecule is less active than wild-type Nod factor for the induction of calcium spiking. This reduced activity could explain the nodF/nodL mutant phenotype, and hence provides support for the single receptor model.

Genetic analysis of Nod factor responses in Nod mutants of M. truncatula has revealed a number of genes that appear to function in Nod factor signal transduction (Catoira et al., 2000; Wais et al., 2000). Phenotypic analysis has shown that DMI1 and DMI2 act upstream of calcium spiking, while DMI3 acts downstream of calcium spiking in the Nod factor signal transduction pathway. All three genes, when mutated, block downstream responses to Nod factor. Here we show that dmi1 and dmi2 mutants are blocked for calcium spiking induction by a diversity of Nod factor structures, and that the dmi3 mutant has increased sensitivity to Nod factor. In this study, we are limited by the existence of a single dmi3 mutant allele, which means we are unable to differentiate allelic variability for the increased sensitivity to Nod factor. This increased sensitivity suggests that DMI3 is involved in negative regulation of the Nod factor signal transduction pathway upstream of calcium spiking.

Negative feedback regulation is a relatively common phenomenon in signal transduction pathways. Negative feedback can regulate the degree to which a pathway is activated or can completely inhibit pathway activation following an initial stimulation. The negative regulation by DMI3 may function as a negative feedback loop. This feedback must be activated by DMI3 or a component downstream of DMI3, but upstream of NSP, as nsp mutants do not show increased Nod factor sensitivity for the induction of calcium spiking. This negative feedback may involve direct interaction of Nod factor signal transduction proteins, or may require intermediate molecules. An alternative explanation for the negative regulation is that DMI3 acts in a different Nod factor-inducible pathway that does not involve calcium spiking, but regulates the Nod factor-dependent pathway that activates calcium spiking. However, the fact that dmi1, dmi2 and dmi3 have identical mutant phenotypes aside from calcium spiking suggests that these three proteins are components of the same signal transduction pathway.

The increased sensitivity of dmi3 to Nod factor-inducible calcium spiking is similar to the increased sensitivity observed in the ethylene-insensitive mutant skl (Oldroyd et al., 2001). However, the dmi3 mutant shows ethylene inhibition of Nod factor-inducible calcium spiking similar to wild-type plants, suggesting that the dmi3 mutation does not affect ethylene inhibition of Nod factor signal transduction. We are currently testing whether a skl/dmi3 double mutant will show an additive effect on the sensitivity of the plant to Nod factor. While we have not seen a link between DMI3 negative feedback and ethylene inhibition of the Nod factor signal transduction pathway, the similarity of the calcium spiking phenotypes in these two mutants may suggest a similar mechanism.

The DMI3 negative feedback regulation is structurally specific for the Nod factor molecule, as it is not active with nodH Nod factor, LCO IV, C16:2, S, LCO IV, C18:1, S, sulphated chitotetraose or chitotetraose. The structural specificity may be explained by two different models. The first model builds on the two-receptor model and assumes the presence of a stringent receptor that requires the complete Nod factor structure to be activated and a less stringent receptor that also perceives other Nod factors and Nod factor-like molecules. Both receptors converge at DMI1/DMI2 and can activate calcium spiking. In this model, DMI3 inhibits only the more stringent receptor and has no effect on the less stringent receptor. The negative feedback by DMI3 can be in place prior to Nod factor application, as there are no perception requirements for the induction of the inhibition. In the second model, a single Nod factor receptor exists; however, the different Nod factor structures differentially activate the pathway with differing outcomes for the induction of negative feedback. In this model, wild-type Nod factor, but not the mutant forms, induce the negative feedback regulation. Hence, Nod factor perception is a requirement for the induction of feedback regulation. Therefore, the negative feedback must be activated following Nod factor application rather than pre-existing prior to Nod factor stimulation, as in the first model. Currently no data exist to differentiate between these two possible models.

The perception of Nod factor appears to be complex, as the plant can differentially activate downstream events depending upon the structure of the Nod factor molecule. Studies in M. sativa have shown an absolute requirement for the O-sulphate group on the S. meliloti Nod factor (Roche et al., 1991a), and a requirement for the O-acetyl and appropriate N-acyl substitutions for the induction of bacterial infection (Ardourel et al., 1994). Here we show that nodH Nod factor, LCO IV, C18:1, S and sulphated chitotetraose can activate calcium spiking when applied at high enough concentrations. It appears that the perception of these structures in the plant is not regulated by or is unable to activate the negative feedback regulation of the Nod factor signal transduction pathway. This negative feedback regulation is dependent upon the presence of DMI3, suggesting that it is activated at or downstream of DMI3. Negative feedback regulation appears to define the responsiveness of the plant to Nod factor and may function in the regulation of the plant's response to S. meliloti.

Experimental procedures

Plant growth conditions

Seeds were scarified using 96% sulphuric acid for 5 min, followed by two rinses in sterile water. They were then sterilized using a commercial bleach solution for 3 min followed by five rinses in sterile water. The seeds were imbibed in sterile water for a minimum of 3 h at room temperature and then transferred to 4°C overnight. The seeds were then germinated overnight at room temperature. Seedlings were plated onto BNM medium (Ehrhardt et al., 1992) with 1.2% agar and 0.1 µm AVG and grown at 22°C in shaded conditions overnight. Root hairs on 1-day-old seedlings were assayed for calcium spiking. Seed of the cultivar Jemalong was used as the wild-type.

Nod factors and oligosaccharides

Wild-type Nod factor was isolated as described by Ehrhardt et al. (1996). nodH Nod factor was prepared as described by Roche et al. (1991a) and was provided by F. Maillet and J. Dénarié (INRA-CNRS, Toulouse, France). LCO IV, C18:1, S, LCO IV, C16:2, S and sulphated chitotetraose were generated as described by Demont-Caulet et al. (1999) and were provided by J.-M. Beau (Université Paris-Sud, Orsay, France). Tetra-N-acetyl chitotetraose was purchased from Seikagaku (Tokyo, Japan).

Calcium spiking analysis

Analysis of calcium spiking was performed as described by Ehrhardt et al. (1996), with slight modifications as described by Wais et al. (2000). For Nod factor dose response assays, the following modifications were made. The maximum number of root hairs that can be imaged (approximately 15 cells) was injected with Oregon green-dextran (Molecular Probes, Eugene, Oregon, USA). Cells were allowed to recover for approximately 30 min prior to imaging. The baseline fluorescence was established by imaging cells for 10 min in BNM. The first concentration of the appropriate Nod factor was then applied and every 30 min the concentration was increased 10-fold. The number of cells spiking at each concentration was counted. Tests of wild-type Nod factor were performed under perfusion using BNM medium. As the additional Nod factor structures required high concentrations to activate calcium spiking we were unable to use perfusion. Instead we applied the Nod factor by dropping 10 µl of the appropriate concentration into the 1 ml bath. To assess the effect of ethylene on the dmi3-1 mutant, 10 µm ACC was included in the overnight growth medium, the 1 ml BNM bath and the perfusion solution.

To assess the response of the dmi1-1 and dmi2-1 mutants to altered Nod factors and chitin structures, we performed the experiments as described above; however, two plants were analysed in each experiment. Approximately five root hair cells were injected with Oregon green-dextran per plant. Each plant was imaged separately. A baseline for calcium fluctuations was determined on the first plant and then either nodH Nod factor, LCO IV, C18−1, S or sulphated chitotetraose was added to a final concentration of 10−7 m, or chitotetraose to a final concentration of 10−4 m. The first plant was imaged for approximately 60 min and then the second plant in the bath was imaged for a further 20 min.

Acknowledgements

We thank Fabienne Maillet, Jean Dénarié and J.-M. Beau for generously providing the mutant and synthetic Nod factors. We also thank Jean Dénarié for critical reading of the manuscript.

Ancillary