• The variability of the developmental responses of two contrasting cultivars of pea (Pisum sativum) was studied in relation to the genetic diversity of their nitrogen-fixing symbiont Rhizobium leguminosarum bv. viciae.
• A sample of 42 strains of pea rhizobia was chosen to represent 17 genotypes predominating in indigenous rhizobial populations, the genotypes being defined by the combination of haplotypes characterized with rDNA intergenic spacer and nodD gene regions as markers.
• We found contrasting effects of the bacterial genotype, especially the nod gene type, on the development of nodules, roots and shoots. A bacterial nod gene type was identified that induced very large, branched nodules, smaller nodule numbers, high nodule biomass, but reduced root and aerial part development. The plants associated with this genotype accumulated less N in shoots, but N concentration in leaves was not affected.
• The results suggest that the plant could not control nodule development sustaining the energy demand for nodule functioning and its optimal growth. The molecular and physiological mechanisms that may be involved are discussed.
Pea is a grain legume of world economic importance for feeds and human consumption. As for other legumes able to establish an interaction with rhizobia, nitrogen nutrition of pea relies on both soil mineral N absorption and symbiotic nitrogen fixation (SNF). Its specific bacterial partner is Rhizobium leguminosarum bv. viciae (Rlv). Under field conditions and low N mineral availability, SNF is the main source of N acquisition in pea, which can provide as much as 80% of total N requirement (Voisin et al., 2002).
Symbiotic nitrogen fixation is performed by rhizobia inside the nodules, which are the symbiotic organs formed on the roots of the host legume. The bacteria supply ammonium to the plant, while the sources of carbon and energy necessary for SNF are obtained from the plant photosynthates. However, the recent findings of Lodwig et al. (2003) show that the metabolic dependence of the two symbiotic partners is more complex than a simple exchange of products of photosynthesis and ammonium, also involving amino-acid cycling.
Nodule formation and host specificity are triggered by mutual signalling. Expression of the rhizobial nodulation (nod) genes is induced by flavonoids secreted by plant roots in conjugation with the product of the rhizobial regulatory gene nodD. The Nod proteins catalyse the biosynthesis of nodulation factors (Nod factors), which trigger early symbiotic responses in plant roots (reviewed by Perret et al., 2000).
Variability of symbiotic traits, and of characters associated with plant growth and photosynthesis, has been found in the pea–rhizobia interaction (Hobbs & Mahon, 1983; Skøt, 1983; Fesenko et al., 1994, 1995; Mårtensson & Rydberg, 1996; Santalla et al., 2001; Labidi et al., 2003). All these studies detected additive effects of both pea cultivar and rhizobial strain, while cultivar × strain interactions were of minor quantitative importance. The general trend is that plant genotype contributes more to shoot and seed biomass, while rhizobial genotype is more responsible for traits associated with SNF, especially for N content in shoots and seeds. However, samples of rhizobial strains investigated so far were small (four to 18) compared with the great genetic polymorphism reported within Rlv indigenous populations (Palmer & Young, 2000; Laguerre et al., 2003; Depret et al., 2004). Therefore the variability of plant responses linked to the microbial partner might have been underestimated. The study of Fesenko et al. (1995), using 481 strains from different geographical areas in Russia, was an exception. Only 10% of the strains were found in this study to be effective in improving significantly N accumulation in shoot when inoculated to pea seeds grown in pots in soil conditions. However, the genetic diversity represented by all these strains was not reported.
In addition to genetic diversity among Rlv strains, great differences in the genetic structure of Rlv indigenous populations could be found among soils (Palmer & Young, 2000; Laguerre et al., 2003; Depret et al., 2004). An important issue is to investigate whether this genetic variability contributes to explaining variation in N uptake, and more generally in the field performance of pea lines. This question will be especially relevant if functional variability of the symbiotic interaction is linked to genetic variability of the rhizobial partner. Therefore, in this study, we addressed this question more specifically. Our strategy was to utilize a sample of strains representative of the diversity of predominant Rlv genotypes identified in pea-nodulating indigenous populations in France, which is the major producing area of field pea in Europe, and some additional strains from other origins described as efficient N-fixers with pea. Two contrasting pea cultivars were used to investigate the quantitative importance of the strain × cultivar interaction. We studied the correlations between Rlv genotypes and traits associated with plant development. Two components of the Rlv genome were analysed: the genomic background and the regulatory nodD gene region. Correlation analysis between the measured parameters of plant growth was also performed to improve our understanding of the impact of the symbiosis on plant development.
Materials and Methods
Rhizobium leguminosarum bv. viciae (Rlv) strains
A collection of 42 bacterial strains was used in this study (Table 1). Most of the strains were previously characterized by restriction fragment-length polymorphism analysis of PCR-amplified DNA fragments (PCR–RFLP) with restriction enzyme HaeIII (Laguerre et al., 1992, 2003; Rigottier-Gois et al., 1998; Depret et al., 2004; Bourion et al., 2007). The target DNAs were the intergenic spacer regions between 16S and 23S rDNA (IGS) and the nodulation gene region nodD. The strains were categorized into IGS and nod haplotypes based on their RFLP patterns. The same genotyping method was used to characterize the additional strains.
Table 1. Strains used, genotypic characteristics† and origin
Plant of origin
Strains were characterized by PCR–RFLP fingerprinting using the intergenic spacer regions between 16S and 23S rDNA (IGS types) and the nodulation gene region nodD (nod types) as target DNAs.
Used with cv. Austin in spring 2005 and with cv. Frisson in autumn 2005. For the other strains, inoculation of both cultivars was performed on the same sowing date.
Two distinct commercial varieties of pea (Pisum sativum L.) were used: cv. Frisson, which is a leafy winter cultivar registered in 1979 in the French variety catalogue, and cv. Austin, which is a recent spring afila (semileafless) cultivar widely cultivated in France. Both cultivars have short internode length and are day-neutral (Lejeune et al., 1999). When grown together in field conditions, they flowered at the same developmental stage, 14–14.5 nodes (Bourion et al., 2007), but length of the main stem was higher for cv. Frisson. Pea plants were grown in a glasshouse under natural light and watered automatically with an N-free nutrient solution as described previously (Laguerre et al., 1993). Temperatures were maintained between 20 and 23°C during the day (16 h) and 18°C during the night. Seeds were inoculated with cell suspensions of Rlv strains (c. 108 per seed). Each treatment (pea cultivar × strain) was carried out in five replicates in a randomized design. As all the treatments (strain–cultivar combinations) could not be performed simultaneously, peas were sown in April 2004, April 2005 and September 2005. Four treatments (two strains used with the two cultivars) were repeated at two different sowing dates, and inoculation with Rlv strain IAUb11 was included in the three experiments. For eight strains, inoculations of cvs Austin and Frisson were carried out on two different dates (Table 1). To compare treatments performed on different sowing dates, all the values measured were standardized by comparison with the IAUb11 treatments. The ratio between each value and the mean value for IAUb11 in the same experiment was multiplied by the overall mean for IAUb11 over all experiments. A treatment consisting of noninoculated plants supplied with mineral N (1.25 mm KNO3 and 1.875 mm Ca(NO3)2, i.e. 5 mm N added to the nutrient solution) was included in the September 2005 experiment.
Harvesting and measurements
The plants were harvested at the beginning of flowering (7–8 wk after sowing, depending on date of sowing). The four plants in each pot were separated into shoots and root systems. In the two 2005 experiments, one plant per pot from three replicate pots was randomly taken to separate roots and nodules. All the nodules were then collected and counted. Dry weight of shoots (pooling the four plants for each of the five replicate pots), roots and nodules was determined after oven-drying at 80°C for 48 h for shoots and roots, and at 105°C for 24 h for nodules. The total N content of aerial parts was measured by the Dumas dry-combustion method with a Carlo Erba automatic analyser (NCS1500).
To compare treatments performed in different experiments, all the values measured were standardized by comparison with the IAUb11 inoculation included in all experiments. Each value was multiplied by the ratio between the value and the corresponding mean value over all IAUb11 pots. anovas for all the plant parameters and correlation were performed using xlstat software ver. 7.5.2 (Addinsoft, http://www.xlstat.com). Means were classified using the LSD test at the 0.05 probability level.
The 42 Rlv strains studied were distributed into 17 genotypes defined by the combination of eight distinct genomic backgrounds and six symbiotic genotypes, based on the use of rDNA IGS and nodD gene regions as markers (Table 1). Shoot biomass production was analysed with the whole sample of strains, but measurements of below-ground parameters were performed only in the two latter experiments, in spring and autumn 2005, for a total number of 33 strains representing 14 genotypes. Missing data concerned strains with genotypes 6b, 17a and 17b (Table 1). The growth (shoot DW) of plants cultivated in autumn 2005 was reduced by a factor 1.7–2.2 by comparison with experiments carried out in spring 2004 and 2005 for the IAub11 treatment that was used to standardize the data. The data were reproducible among the two experiments conducted in spring. The same results were observed for the mean values of shoot DW over all the cultivar–strain treatments. This could be explained by the reduced photoperiod and light intensity in autumn, the experiments having been performed under natural light conditions. After standardization of the data, we confirmed the reproducibility of the data among sowing dates for one strain (IIFb1) inoculated in spring 2004 and 2005, and another one (P1NP2K) inoculated in spring and autumn 2005. Additionally, eight strains representing four distinct genotypes were inoculated to cv. Austin in spring 2005, but to cv. Frisson in autumn. Strain rankings were very similar in both experiments, as confirmed by statistical analysis which did not reveal a strain–cultivar interaction for this sample (P = 0.15).
Strain-dependent variability of plant responses
The results showed the great variability of plant developmental responses according to the rhizobial strain inoculated (Fig. 1). The strain effect was statistically highly significant for all parameters measured (Table 2). The largest variation (factor of 80) was observed for mean DW per nodule, which spanned from 0.14 to 11.1 mg (Fig. 1c). The strains formed two groups according to the plant phenotype induced. Seven strains belonging to genotypes 1d, 18d, 39d and 18b formed big nodules (BNO) with a considerably higher total biomass than did the others, which formed smaller nodules (Fig. 1a). The big nodules were pink and exhibited branched coralloid forms corresponding to multiple nodule meristems, which were concentrated on the oldest part of the root system. By contrast, the small nodules (SNO) had an elongated cylindrical shape, and were more widely spread on the roots. The colour of these nodules varied from pink to dark purple.
Table 2. anova of nodulation, root and shoot traits of two pea (Pisum sativum) cultivars inoculated with rhizobial strains representing diverse genotypes
The BNO phenotype was associated with nodule numbers that were mostly lower than the general mean value of 250 nodules per plant obtained with the other strains (Fig. 1b), and the DW per nodule was high (0.7–11 mg) compared with the mean value over all other treatments (0.32 mg). Root development was clearly depressed by most BNO-inducing strains (Fig. 2). The root DW of BNO plants ranged from 150 to 290 mg per plant, while the mean value for strains inducing an SNO phenotype was 326 mg per plant (Fig. 1d). As a consequence, the proportion of nodule biomass in the below-ground compartment (N/RN) reached 36–78% for BNO plants, the mean value for SNO plants being only 17% (Fig. 1e).
Shoot biomass was also generally reduced in the BNO plants up to a factor of two compared with the maximal values recorded in this work (Fig. 1f). Plants were shorter, while numbers of nodes were similar to those observed with the SNO plants in the autumn 2005 experiment. However, in the spring 2005 experiment, the node number was 10–15% lower for the BNO plants at harvest. Frisson was especially sensitive to BNO-inducing strains, but shoot development was also affected in cv. Austin with the strains inducing the biggest nodules.
A striking result was the high biomass of nodules in BNO plants, which represented up to 20% of total plant biomass, while the mean value of 2% obtained with SNO plants was in the expected order of magnitude based on previously reported results for different pea cultivar–rhizobial strain combinations (Mårtensson & Rydberg, 1996).
Variability of responses was also observed within BNO and SNO plants. Highly significant strain effects, or at least significant differences between some strains, were found within both groups for all parameters. Correlation among the different parameters was then analysed. The results from the analysis of linear relationships are given in Table 3. Nodule number was negatively related to nodule size for both BNO and SNO plants (only with cv. Austin for the BNO plants). However, more significant correlation between these two parameters was found by using power functions (r2 > 0.70). Significant linear relationships, which were the best models, were found between other parameters, but these relationships varied according to plant phenotype and genotype. Nodule number was positively related to root biomass in the SNO plants, while no significant correlation between root and nodule parameters was found with the BNO plants. Nodule biomass was negatively related to shoot biomass, and less significantly to root biomass, in the BNO plants of cv. Austin, and a strong linear positive relationship between root and shoot biomass was found for the BNO plants of both cultivars. By contrast, nodule biomass was positively related to root biomass in SNO plants of cv. Austin, and no correlation was observed between the biomass of the below-ground organs and of shoots. Notably, two contrasting phenotypes of the below-ground compartment were observed among the SNO plants of cv. Austin inoculated with the eight strains that induced the highest values of shoot biomass (DW ≥ 2.90 g per plant). The DW values of roots and nodules were among the highest recorded for four strains, but among the lowest for the four others.
Table 3. Matrix of correlations† among nodule, root and shoot parameters‡ for pea (Pisum sativum) plants forming big nodules (BNO) and small nodules (SNO)
Values above and below diagonal are Pearson correlation coefficients obtained for cvs Austin and Frisson, respectively.
*, **, ***
, correlations are significant at P≤ 0.05, P≤ 0.01, P≤ 0.001, respectively.
There was a significant main cultivar effect for all parameters except nodule number (Table 2). The mean values of the different parameters were higher with cv. Frisson, but this may be caused by differences in plant phenology between the two contrasting cultivars. In particular, cv. Frisson being an earlier cultivar than Austin, emergence of plants and appearance of the first flowers were 3–5 d later for cv. Austin. At harvest, cv. Austin plants had formed three to four fewer nodes than cv. Frisson plants, all plants having been harvested at the same time. We also found that the effect of the rhizobial partner was more pronounced with cv. Frisson, especially with respect to nodulation and, as mentioned above, shoot development, the range of variation being larger than for cv. Austin.
Strain × cultivar interaction was found to be significant for all parameters except for the ratio N/RN (Table 2). However, the interaction had a minor effect compared with the strain effect. The interaction was especially highly significant for nodule number, and for root and shoot DW. For instance, the highest value for root biomass (461 mg per plant) was obtained with cv. Austin plants inoculated with strain P221 (genotype 18a), while the mean value was only 268 mg per plant for the cv. Frisson–strain P221 association. Conversely, cv. Frisson plants showed a high production of shoot biomass in symbiosis with strain P115 (genotype 16c), while this strain was among the less efficient with cv. Austin plants (mean values of 3.9 and 2.3 g per plant, respectively).
Variability in plant response is linked to rhizobial genotype
Although variability linked to strain was observed within rhizobial genotype (e.g. within genotype 18a for root and shoot DW, Fig. 1d,f), the genotype effect was highly significant for all parameters (Table 2). The BNO phenotype associated with low root and shoot biomass was found with the four strains classified in genotype 1d (Fig. 1). For SNO phenotype, both nodule number and root and shoot biomass also showed variability linked to rhizobial genotype. Some rhizobial genotypes (e.g. 1g, 32g and 1j) were the most beneficial for nodule, root and shoot development with both cultivars. Conversely, genotypes such as 1a formed fewer nodules and induced less root development.
A statistically significant interaction between cultivar and rhizobial genotype was found only for nodule number and shoot DW. Strains showing genotype 1g induced significantly more nodules with cv. Frisson than with cv. Austin, while for the other genotypes the values were not significantly different among cultivars. Strains with genotype 1d had a particularly severe negative effect on shoot biomass of cv. Frisson plants.
The two DNA markers were then analysed separately. The plant responses were linked to nod types without statistically significant interaction with the cultivar except for DW per nodule and shoot DW (Table 2), but again this can be explained by the more pronounced response of cv. Frisson according to nod types. Clearly, the six strains showing nod type d induced the BNO phenotype independently of the IGS type it was associated with (types 1, 18 and 39, Fig. 1; Table 4). Within SNO plants, those associated with nod type a were generally the less productive in nodule, root and shoot development. However, cases of interaction between nod types and bacterial genomic background were observed. The effect of nod type g varied considerably according to the IGS type it was associated with, the 18g combination being among the less efficient for nodule and root development, in contrast to 1g and 32g (Fig. 1). Also, cv. Frisson plants inoculated with combinations 1a and 18a produced significantly more shoot biomass than plants inoculated with 17a.
Table 4. Nodulation, root and shoot traits and biomass partitioning in relation to rhizobial nod type
Mean values followed by the same letter within a column are not significantly different by LSD (P ≤ 0.05).
Numbers within parentheses separated by semicolons are the number of strains tested with pea (Pisum sativum) cvs Frisson and Austin, respectively, included in the statistical analyses, nod types represented by a single strain being excluded. For nod type a, fewer strains were analysed for nodulation and root traits than for shoot DW (nine and 13, respectively).
N/NR and N/SRN are proportions represented by nodules in the biomass of the below-ground compartment and of the entire plant, respectively.
nd, not determined.
g (6; 9)
j (5; 4)
c (3; 3)
a (9–13; 9–13)
b (6; 6)
d (4; 6)
More generally, lesser but significant effects of IGS type were observed on plant responses (Table 2), but it was not possible to analyse the IGS type × nod type interaction because all combinations of IGS and nod types were not available. Therefore the effect of IGS type was probably linked to the interaction with nod type in several cases. For instance, strains with IGS types 6 and 17 were only found associated with two nod types (a and b) that were less effective than other nod types such as g and j for shoot biomass production. However, in other cases, it was possible to detect a main effect of the IGS type, irrespective of the nod type associated by a three-factor anova (IGS type, nod types, and cultivar) when analysing shoot DW using a sample of 16 strains representing the six combinations of IGS types 6, 17 and 18 with the two nod types a and b. Higher shoot DW was found with IGS type 18, and there was no significant interaction between IGS and nod types.
Nitrogen accumulation in shoot
Nitrogen nutrition was then analysed for a sample of BNO and SNO plants with contrasting responses. No significant strain effect on shoot N concentration was found (Table 5). Therefore the values of accumulated shoot N varied according to shoot DW. A treatment consisting of plants fed with a nonlimiting quantity of mineral N was included. This treatment could be considered as a control for optimal N nutrition according to Ney et al. (1997), as it showed the lowest N concentration for maximal biomass N accumulation. N2 fixation by several strains was as efficient as mineral N adsorption for N uptake.
Table 5. Nitrogen accumulation in shoot of pea (Pisum sativum) cv. Frisson plants grown in autumn 2005 in relation to nodulation phenotype and shoot development
Nodule efficiency (specific activity) for N2 fixation was then compared among treatments by calculating the ratio between shoot N accumulated per unit of nodule DW. These ratios varied by a factor of 20 with very low nodule efficiency for the BNO plants inoculated with strains LRBA7 and P1NP2K.
The present study revealed contrasting effects of the bacterial partner in the pea–rhizobium symbiosis. By using a sample of genetically diverse strains of Rlv, significant variability of plant responses linked to the rhizobial genotype was found for all plant compartments. The rhizobial genotype strongly influences nodule morphogenesis and biomass partitioning. The pea genotype made a major contribution to the variability of the developmental plant responses, as expected by choosing two contrasting pea cultivars, but a remarkable result was that variation in nodule number was explained mainly by the effect of the bacterial partner. Strain × cultivar interaction was of minor quantitative importance, which fully confirms previous studies (Hobbs & Mahon, 1983; Skøt, 1983; Fesenko et al., 1994, 1995; Labidi et al., 2003). Two main phenotypes could be described, the BNO plants producing fewer nodules but much more nodule biomass, producing less developed root systems and aerial parts, and accumulating less N than the SNO plants.
The nodule proportion of total biomass in BNO plants (12–14%) was higher than those previously reported, which ranged from 0.5 to 8% (DeJong et al., 1981; Skøt, 1983; Vessey, 1992; Mårtensson & Rydberg, 1996; Santalla et al., 2001). However, cases of BNO phenotype could be suspected in previous studies. Skøt (1983) reported that pea plants inoculated with Rlv strain 128C53 formed larger nodules and less root and shoot biomass than the three other pea–strain associations analysed. Also, Sagan & Gresshoff (1996) described an Rlv strain, MSDJ1243, that formed nodules with cv. Frisson showing morphological characteristics similar to those of BNO plants. Strains characterized as inducing the BNO phenotype appear to be widespread. They represented 21% of our sample of strains. The nod type d was found to be linked to this phenotype, and we confirmed recently that strain MSDJ1243 shows this genotype (results not shown). So far, nod type d was detected in seven out of eight fields sampled in three French regions, in proportions varying from 1 to 63% of nodule isolates (Laguerre et al., 1996, 2003; Depret et al., 2004). This genotype was also identified in two strains originating from North America included in this study, and in field isolates from Germany and UK (Rigottier-Gois et al., 1998). In the sole French soil previously studied in which it was not detected, nod type b was among the predominant ones (50% of isolates) in pea nodules collected in the field (G.L. and G.D., unpublished data). The single representative of nod type b that was analysed for nodulation and root traits in the present work induced formation of big nodules, but roots and shoots were not affected. Six other strains with nod type b were analysed further, but they did not form big nodules (data not shown). However, this genotype, especially when associated with IGS type 6, was performing poorly in shoot biomass production based on the study of 12 strains including those included in the present work. Our results raise the question whether high frequency in pea-nodulating indigenous populations of rhizobial genotypes that have a negative impact on root and/or shoot development under the controlled conditions used in this study may alter the field performance of the crop. In particular, BNO-inducing rhizobia may be more beneficial for plant growth in environmental biotic and abiotic conditions that may limit nodule development. Field experiments are in progress to evaluate the impact of inoculation of BNO-inducing strains on plant growth and seed yield under agronomic conditions.
The finding that nodule biomass was negatively correlated to biomass of other organs in BNO plants contrasts with correlation analysis in SNO plants, and with previous studies reporting positive relationships between these parameters (DeJong et al., 1981; Mårtensson & Rydberg, 1996; Santalla et al., 2001). High nodule biomass was generally considered as a trait of efficient symbiosis. On the other hand, experiments based on comparison of nitrogen sources (N2 or mineral N) led to the conclusion that nodules grow and fix N2 at the expense of roots (Schulze et al., 1999; Voisin et al., 2003), and also at the expense of shoots during the vegetative stage (Voisin et al., 2003). This was explained by the higher C cost of N2 fixation as compared with root mineral N absorption. Studies of hypernodulating pea mutants, which formed many more small nodules and higher nodule biomass than the wild type, but produced less developed roots and shoots or lower seed yield, also agreed with the carbon-economy theory (Salon et al., 2001; Bourion et al., 2007). Similarly, the BNO plants are unable to limit nodule growth and to adjust energy supply with demands of both nodule development and functioning, and optimal growth of the other organs. Indeed, C rather N uptake was probably the limiting factor of plant growth and N accumulation in BNO plants, as they were fully functional for N2 fixation as shown by N concentrations in shoots. Therefore this type of symbiosis looks more like parasitism than mutualism, growth of the symbiotic organs being achieved at the expense of the others.
The range of variation (up to a factor of five) observed for biomass in the different SNO-plant compartments did not reveal any correlation between below-ground and shoot biomass. Therefore C partitioning cannot explain these variations. This may result from variation in C respiration and/or in root C exudation, but physiological mechanisms other than competition for C among the different plant organs may also be affected by the symbiosis.
The negative relationship between nodule biomass and nodule number obtained with both SNO and BNO plants is commonly observed, and should result from autoregulation of nodule number, which is inhibition of nodule formation by already existing ones, and is known to be controlled by the plant (Duc & Messager, 1989; Sagan & Duc, 1996; Oka-Kira & Kawaguchi, 2006). Our study shows that this phenomenon is expressed more intensively in the BNO plants and may be linked to the extensive development of individual nodules. Sagan & Gresshoff (1996) demonstrated that high nodule biomass accumulation and branching induced by strain MSDJ1243 was caused by prolonged activity of the nodule meristem. Persistence of meristem activity was maintained in two shoot-controlled hypernodulating pea mutants in which the inhibition of nodulation is impaired, which led the authors to conclude that the rhizobial partner is involved in the control of nodule meristem development. Further studies using pea mutants affected in autoregulation of nodule number are in progress to investigate whether the rhizobial partner may play a direct role in this regulation, or whether limitation of nodule extent is linked to the level of development of nodule meristem.
BNO-inducing strains appear able to overcome plant-defence mechanisms that have been suggested to control bacteroid differentiation in coordination with the host cell in galegoid legumes, including pea (Mergaert et al., 2006). We found that induction of the BNO phenotype was restricted to two nod types among the six analysed. This phenotype was obtained independently of the rhizobial genomic background associated with nod type d. Therefore variability in nod gene expression or in the structure of Nod protein and Nod factors might be responsible for variation in nodule development. Elevated concentrations of Nod factors or high expression of nod genes inhibit nodulation of pea cv. Afganistan by Rlv strain TOM (Hogg et al., 2002). However, little is known about variability in structure of Nod proteins and Nod factors within specific rhizobial groups.
Alternatively, nod genes might just be markers reflecting variability in other genes that have coevolved with them. Additionally, interactions between symbiotic genes and other bacterial genes only present in certain genotypes cannot be excluded. In Rlv, as in several other rhizobial species, the symbiotic genes are carried by a plasmid called the symbiotic plasmid. The symbiotic plasmids harbour many other genes in addition to the symbiotic genes (Young et al., 2006), and vary in size (Laguerre et al., 1992), indicating that genetic information is not necessarily conserved among Rlv strains. In addition to Nod factors, rhizobia produced metabolites that may interact with plant hormonal biosynthesis, signalization pathways and/or regulation. Certain rhizobial strains synthesize phytohormones such as auxin and cytokinin (Phillips & Torrey, 1972; Ernstsen et al., 1987). Rlv strains have been reported to modulate ethylene biosynthesis by the host plant (Ma et al., 2003a, 2003b). Ethylene is an important hormone involved in several steps of nodule organogenesis, and was shown to inhibit nodulation of pea and other legume species (reviewed by Guinel & Geil, 2002).
The rhizobial genotype had a strong influence on root and shoot development. Growing of nodules at the expense of the other plant organs may explain this result, as discussed above. However, a direct additional effect of the rhizobial partner on plant growth through production of hormones or interaction with phytohormone levels cannot be excluded. Also, the symbiosis-associated signal molecules, Nod factors, act as phytohormones by influencing nonnodulated root development in host plants when applied at submicromolar concentrations (Souleimanov et al., 2002; Olah et al., 2005). At similar concentrations, Nod factors also induce physiological changes in various nonlegumes, enhancing their early growth (Souleimanov et al., 2002; Prithiviraj et al., 2003). The rhizobia may play a more general role in plant growth in addition to the N2-fixing symbiosis. Plant growth-promoting effects have been reported for rhizobia (reviewed by Dakora, 2003), but more comprehensive studies are required. We are currently comparing the developmental responses of different host and nonhost plants to investigate whether or not the rhizobial effects described in the present study are specific to the pea–rhizobium symbiosis. The genome of one Rlv strain, 3841, has been completely sequenced (Young et al., 2006). This strain belongs to genotype 1g according to our classification, and it induced a typical SNO phenotype when inoculated to cv. Frisson in a recent additional 2007 experiment (G.L., unpublished data). However, root and shoot development were among the lowest recorded for SNO plants. A comparative genomic analysis of additional Rlv strains showing contrasting effects on plant growth would provide information on genetic similarities and differences among strains. This would be a starting point to predict functional bacterial elements playing a role in nodule development, and more generally in plant growth.
We thank Stéphanie Jaxel, Michèle Bours, Florent Robvieux, Céline Faivre, Colette Catroux, Daniel Pouhair and Jacques Sommer for their help in the glasshouse work. We are grateful to Guilhem Debrosses and Christophe Salon for scientific discussion. This work was funded by the Regional Council of Burgundy.