Because boron (B) and calcium (Ca2+) seem to have a strong effect on legume nodulation and nitrogen fixation, rhizobial symbiosis with leguminous plants, grown under varying concentrations of both nutrients, was investigated. The study of early pre-infection events included the capacity of root exudates to induce nod genes, and the degree of adsorption of bacteria to the root surface. Both phenomena were inhibited by B deficiency, and increased by addition of Ca2+, resulting in an increase of the number of nodules. The infection and invasion steps were investigated by fluorescence microscopy in pea nodules harbouring a Rhizobium leguminosarum strain that constitutively expresses green fluorescent protein. High Ca2+ enhanced cell and tissue invasion by Rhizobium, which was highly inhibited after B deficiency. This was combined with an increased B concentration in nodules of plants grown on B-free medium and supplemented with high Ca2+ concentrations, and that can be attributed to an increased B import to the nodules. Histological examination of indeterminate (pea) and determinate (bean) nodules showed an altered nodule anatomy at low B content of the tissue. The moderate increase in nodular B due to additional Ca2+ was not sufficient to prevent the abnormal cell wall structure and the aberrant distribution of pectin polysaccharides in B-deficient treatments. Overall results indicate that the development of the symbiosis depends of the concentration of B and Ca2+, and that both nutrients are essential for nodule structure and function.
Boron (B), a micronutrient required in micromolar concentrations by plants, has been implicated in several physiological processes (reviewed in Blevins & Lukaszewski 1998): cell wall structure and synthesis; membrane structure and membrane-associated reactions; reproduction; nitrogen fixation; and phenolic metabolism. Although its role in the cross-linking of the pectin component rhamnogalacturonan II (O’Neill et al. 2001) has clearly been established, the diversity of symptoms due to B deficiency in plants make it unclear as to whether the micronutrient plays several roles or they are secondary effects.
The requirement of B for symbiotic N2 fixation was first suggested by Brenchley & Thornton (1925). Based on that long-standing report, we have studied the involvement of B on rhizobial symbiosis and, during the last decade, the possible roles of B in each step of nodule development and nodule morphogenesis have been established. As in other plant tissues, B is needed for the maintenance of nodule cell wall and membrane structure both in indeterminate (pea) (Bolaños et al. 1994) and determinate (bean) nodules (Bonilla et al. 1997). Moreover, a requirement for B has been reported for rhizobial infection and the nodule invasion process, probably due to a B-mediated inhibition of the binding of infection thread matrix glycoprotein (MGP) to the cell surface of Rhizobium that results in bacteria endocytosis from infection droplets (Bolaños, Brewin & Bonilla 1996). More recent studies indicate that this micronutrient is also essential for symbiosome development and bacteroid maturation (Bolaños et al. 2001), and for early events in plant–bacteria signalling (Redondo-Nieto et al. 2001).
In plants, calcium (Ca2+) is also implicated in a large number of physiological processes (Leonard & Hepler 1990). In plant cells, most of the Ca2+ is involved in the structure and function of cell wall and membrane, and cytosolic free Ca2+ is important as a second messenger in the signalling mechanism of many plant responses (Bush 1995). Regarding rhizobia-legume symbiosis, an impaired N2-fixation due to Ca2+ deficiency was reported by Greenwood & Hallsworth (1960). Later, Lowter & Loneragan (1968) reported that a high Ca2+ supply increased the number of nodules and Munns (1970) reported that Ca2+ is especially required for early infection events. More recently, Richardson et al. (1988) demonstrated that high Ca2+ increased the amount of nod-gene inducing compounds in root exudates. Additionally, it has been reported that calcium acts as a second messenger in the Nod-factor signalling pathway (see Cárdenas et al. 2000 and refs. therein).
The study of the interaction between both B and Ca2+ is an important topic in plant mineral nutrition. The amount and availability of either nutrient influence its tissue distribution (Ramón, Carpena & Gárate 1990) and the level of requirement of the other nutrient for optimal plant growth (Teasdale & Richards 1990). Interaction of both nutrients has a role in the transport of IAA through the cell membrane (Tang & de la Fuente 1986). Nevertheless, most investigations regarding the B–Ca2+ relationship have been focused on the structure and function of the cell wall. The assembly of cell wall polysaccharides is influenced by both nutrients. Calcium cations produce coordination bonding with carboxyl groups along pectin polysaccharides (Demarty, Morvan & Thellier 1984), and boric acid binds to the apiosyl residues promoting dimerization of rahmnogalacturonan II (O’Neill et al. 1996, 2001). Fleischer, O’Neill & Ehwald (1999) demonstrated that the larger pore size of B-deprived cell walls is due to the presence of monomeric instead of dimeric rhamnogalacturonan II and that removal of Ca2+ by chelating agents enlarged pore size, suggesting a cooperative role of B and Ca2+ on the cell wall pore. In this regard, Van Duin et al. (1987) described that Ca2+ is able to form complexes with borate-polyhydroxy–carboxylates through direct interaction with the borate anion. Furthermore, Kobayashi et al. (1999) and Ehwald et al. (2002) demonstrated that Ca2+ promoted and stabilizes in vitro and in vivo formation of dimers of borate-rhamnogalacturan II and proposed that Ca2+ stabilizes pectin polysaccharides of the cell wall through ionic and coordinate bonding in the polygalacturonic acid region.
Carpena et al. (2000) suggested that a specific B–Ca2+ relationship mediated mobilization of B from old to new growing tissues in nodulated B-deficient pea plants and a higher requirement of B for nodules than for other plant tissues. Additionally, a balanced B–Ca2+ nutrition has been reported to increase salt-tolerance in the Rhizobium-legume symbiosis by ameliorating nodule development (El-Hamdaoui et al. 2003). In view of this previous knowledge, the aim of this study was to gain insight into the relationship between both plant nutrients in the establishment and development of the rhizobial symbiosis, as well as in the nodule structure and function. Early pre-infection events and cell invasion by rhizobia have been investigated in nodulated legumes growing with different B and Ca2+ concentrations. Moreover, due to the importance of the cell wall, especially in the nodule cortex, for protection of the infected tissues and for the maintenance of the oxygen barrier (for a review, Hunt & Layzell 1993), the effects of B and Ca2+ nutrition on the ultrastructure of nodule cell wall were also studied.
MATERIALS AND METHODS
Growth of plants and inoculation
Pea (Pisum sativum cv. Argona), bean (Phaseolus vulgaris cv. Delinel) and alfalfa (Medicago sativa cv. Moapa) seeds were surface-sterilized with 70% (v/v) ethanol for 1 min and 10% (v/v) sodium hypochlorite for 20 min, soaked for 4 h in sterile distilled water and then germinated on wet filter paper at 25 °C. After 4 d the seedlings were transferred to plastic growth pots and cultivated on perlite with FP medium for legumes (Fahraeus 1957), as the nutrient solution (0.68 mm CaCl2; 0.5 mm MgSO4.7H2O; 0.7 mm KH2PO4; 0.68 mm Na2HPO4.2H2O; 50 µm Fe-EDTA; 9.3 µm H3BO3; 10.6 µm MnSO4.H2O; 0.7 µm ZnSO4.7H2O; 3.2 µm CuSO4.5H2O; 1 µm Na2MoO4.2H2O).
For B-free (–B) cultures, B was removed from the micronutrients solution. For cultures with the normal content of B (+B), the micronutrient (as H3BO3) was added to a final concentration of 9.3 µm B, for high B levels (++B), it was added to a final concentration of 46.5 µm. All solutions were prepared and stored in polyethylene containers previously demonstrated not to release boron even under sterilizing conditions (Mateo et al. 1986). During preparation of nutrient solutions and growth media, and during growth of plants, Amberlite IRA-743 (Sigma Chemical Co., St. Louis, MO, USA) was added at a concentration of 1 g L−1. This boron-binding-specific resin, which strongly complexes H3BO3 on its N-methylglucamine functional groups, with an adsorption capacity of up to 5 mg B g−1 resin (Asad et al. 1997), allows the removal of possible B contamination. The solutions and media were tested for B prior to using them, and no B was detected (detection limit was 0.02 µg mL−1). Boron concentration was determined using Azomethine H at pH 5.1 (Wolf 1974) and a Technicon Automatic Analytical System (Tarry Town, NY, USA) (Martínez et al. 1986). Before inoculation, the pea seedlings were grown without added B for a period of 15 d (including germination time) to drain B stored in the seeds. Calcium was added as CaCl2 to a concentration of 0.68 mm (+Ca2+ treatments). Some plants were fed with no (–Ca2+), one-half (1/2 Ca2+) or a double concentration of Ca2+ (+2Ca2+).
The plants were inoculated with 1 mL per seedling of about 108 cells mL−1Rhizobium leguminosarum bv. viciae strain 3841 for pea, strain B625 for bean, and Shinorhizobium meliloti strain 2011 for alfalfa, from an exponential culture in tryptone–yeast extract (TY) medium (Beringer 1974). Inoculated plants were maintained in a growth cabinet at 22 °C day/18 °C night temperatures with a 16/8 h photoperiod and an irradiance of 190 µmol m−2 s−1. Relative humidity was kept between 60 and 70%.
Antibodies and antisera
The rat monoclonal antibodies JIM 5 and JIM 7 (kindly supplied by Dr N. J. Brewin) recognize polygalacturonic acid (pectin) and methyl-esterified polygalacturonic acid, respectively (Knox et al. 1990). Antiserum anti-RGII (kindly supplied by Dr T. Matoh) recognizes rhamnogalacturonan II polysaccharide that localized at the cell wall–membrane interface (Matoh et al. 1998).
Nitrogenase activity was measured in four plants per treatment from four independent experiments by chromatographic determination of reduced acetylene (ARA) (Postgate 1971).
For mineral analysis, 50 plants of each treatment were randomly selected. Shoots, roots and nodules were separated from the plant and dried at 80 °C for 24 h to determine dry weight. From each sample, 0.5 g were oven-ashed and acid-digested (1 m HCl) at 70 °C for 30 min. Boron was determined on the acid-digested samples as described above. Ca2+ was analysed by atomic absorption spectrophotometry in a Perkin Elmer Analyst 800 spectrophotometer (Perkin Elmer, Willesley, MA, USA). To avoid interference, a solution of 0.5% La(NO3)3, 0.02% CsCl, 5% HCl was added to samples and standards in a 1 : 10 proportion, and Ca2+ was measured at 422.7 nm with a PE6017 lamp.
For the in vitro measurement of nod gene activity, surface extracts were obtained from roots (5 g fresh weight) of uninoculated 15-day-old-pea plants through methanolic extraction, according to the method of Maxwell et al. (1989). The analysis of nod gene induction by methanolic extracts was made by determination of β-galactosidase activity, adapted from Miller (1972). The bioassay used a genetically modified strain of R. leguminosarum, strain D24, which contained a transcriptional fusion of the promoter of the nodABCIJ operon to the lacZ gene. The capacity of induction was expressed as a percentage of β-galactosidase activity induced by 0.5 mm hesperetin, a nod gene-inducing flavone (988.2 ± 72.7 β-galactosidase activity units).
Adsorption of rhizobia to pea roots was assayed by the method of Lodeiro et al. (1995). Two-week-old plants of each treatment were incubated with 50 mL of a suspension of 103 cells mL−1 of a late-exponential-phase R. leguminosarum 3841 liquid culture at 28 °C with rotary shaking at 50 r.p.m. After four washes, roots (equal weight of each treatment) were plated on yeast extract-mannitol agar. The relative degree of adsorption was expressed as percentage of inoculated rhizobia that developed microcolonies on roots (%A or adsorption index).
Each experiment and determinations were performed in quadruplicate.
Microscopy on nodule sections
Nodules were selected at a comparable stage of development, harvested, and fixed overnight at 4 °C in 2.5% (v/v) glutaraldehyde in 0.1 m sodium cacodylate buffer, pH 7.2. For structural purposes, nodules were dehydrated in an ethanol series and embedded in Araldite, or LR White resin, for immunocytochemistry studies. Semi-thin (0.5 µm) sections were stained with toluidine-blue for light microscopy observations. Ultra-thin (silver-gold in colour) sections of nodules were stained with uranyl acetate and lead citrate or processed for immunogold staining (as described by Rae et al. 1991), following transmission electron microscopy. A goat anti-rat IgG (to reveal monoclonal antibodies labelling) or anti-rabbit IgG (to reveal antisera labelling) conjugated to colloidal gold (10 nm gold particles) (Amersham, Little Chalfont, Bucks., UK) was used as a secondary antibody in immunochemistry analysis. In control experiments (not shown) the first antibody was omitted or replaced by an irrelevant antibody, and no staining was detected.
For the study of nodule development with epifluorescence microscopy, pea plants were inoculated with R. leguminosarum strain 3841 carrying pHC60, a rhizosphere-stable plasmid that constitutively expresses green fluorescent protein (GFP) (Cheng & Walker 1998). For the acquisition of that plasmid by strain 3841, a three parental conjugation with two E. coli strains portraying pHC60 and the helper plasmid pRK2013, respectively, was made. The nodules developed were analysed with a fluorescence microscope coupled with a reflected fluorescent light attachment. Images of GFP-labelled bacterial cells were obtained by using a filter set consisting of a 400–490 nm (BP490) bandpass exciter, a 505-nm dicroic filter and a 530-nm long-pass emitter (EO530).
All the experiments were repeated at least four times. All data were statistically analysed by the one-way anova test with a sample size n ≥ 15. Data in Figures and Tables are means ± standard deviation.
Inoculated determined, and undetermined nodule-forming legumes (pea, bean, and alfalfa) were grown in media with different B and Ca2+ concentrations. As shown in Fig. 1, both the level of nodulation and nitrogen fixation in peas were strongly modulated by both nutrients. The highest rate of nitrogenase activity (measured as ARA; Fig. 1A) was obtained in nodulating peas grown in 9.3 µm B and 0.68 mm Ca2+. Although, higher Ca2+ concentrations (up to 6.8 mm, data not shown) were tested, no significant increase of nitrogenase activity was reached (Fig. 1A). Conversely, a significant boron toxicity was attained at 46.5 µm B (5B, +Ca2+), and a boron deficiency (–B, +Ca2+) resulted in a strong inhibition of nitrogenase activity (about 80%), as previously reported (Bolaños et al. 1994). Figure 2 shows that B-deficient pea nodules presented alterations in the cell wall, appearing irregularly shaped, and contained aberrant symbiosomes (Fig. 2B & E). Moreover, cells from nodules developed under B toxicity were also irregularly shaped (Fig. 2C) and the peribacteroid membrane of the symbiosomes was wrinkled and disrupted (Fig. 2F) in comparison with control nodules (Fig. 2A & D).
Interestingly, different Ca2+ concentrations also affected nitrogen fixation. The effects on the nitrogenase activity depended on the levels of boron available in the medium (Fig. 1A). For example, in pea plants grown with optimal B-level in the medium (9.3 µm), the increase in Ca2+ concentration (up to 1.36 mm, +2Ca2+ treatments) or decrease (down to 0.34 mm, +1/2Ca2+ treatments) of Ca2+ concentration both caused a reduction in nitrogenase activity (Fig. 1A). By contrast, the treatment of +2Ca2+ produced a four-fold increase in the activity of -B plants (–B +2Ca2+). Alternatively, one half reduction of Ca2+ levels in plants grown with toxic levels of boron (5B +1/2 Ca2+) resulted in a modest recovery in nitrogen fixation. In treatments in which Ca2+ was not included in the nutrient solutions (–Ca2+), nitrogenase activity was highly inhibited, and the increase in boron concentrations (5B –Ca2+) resulted in only a small recovery. Levels of Ca2+ higher than 1.36 mm (+2Ca2+ treatments) did not increase nitrogen fixation of B-deficient treatments (data not shown).
Figure 1B shows the number of nodules developed in plants. Nodulation was reduced by B deficiency at normal calcium concentration (–B +Ca2+ treatments), however, this normal Ca2+ level, optimal for nitrogen fixation and plant growth, was suboptimal for nodule induction, since nodulation significantly increased in 2Ca2+ treatments. This was independent on the level of B added. Observations made on bean and alfalfa plants gave similar results to those described for pea (data not shown).
Figure 3 and Table 1 show pea (Fig. 3A) and bean (Fig. 3B) plants grown with different B and Ca2+ concentrations 3 weeks after inoculation with R. leguminosarum strain 3841 (pea) or strain B625 (bean), respectively. Results correlated with the effects of the different B and Ca2+ treatments on nodulation and nitrogen fixation showed in Fig. 1. Boron-deficient plants, which were poorly nodulated and had a highly inhibited nitrogen fixation, showed a reduction of both shoot and root fresh weight not only due to B deficiency but also to a lack of N (–B +Ca2+ treatments). Those effects were recovered by the addition of 1.36 mm Ca2+ (2Ca2+), which also recovered nodulation and nitrogen fixation.
Table 1. Shoot and root fresh weight of 3-week-old Pisum sativum and Phaseolus vulgaris plants
Shoot weight (g)
Root weight (g)
Shoot weight (g)
Root weight (g)
See ‘Materials and Methods’ for experimental conditions. Values followed by the same letter are not significantly different. P ≤ 0.05.
1.01 ± 0.23a
0.49 ± 0.14a
2.75 ± 0.25a
1.22 ± 0.37a
0.29 ± 0.13b
0.13 ± 0.09b
1.59 ± 0.22b
0.33 ± 0.15b
0.80 ± 0.27a
0.50 ± 0.13a
2.38 ± 0.18a
1.01 ± 0.20a
1.10 ± 0.42a
0.60 ± 0.19a
2.58 ± 0.21a
1.23 ± 0.29a
Table 2 shows B and Ca2+ contents in shoots, roots and nodules of nodulated pea and bean plants grown with different B and Ca2+ concentrations. The content of Ca2+ was similar in any tissue analysed and at any treatment in pea. In bean plants the content was higher in +2 Ca than in +Ca treatments. Boron content diminished in tissues derived from plants of –B treatments both in pea and bean. It should be emphasized that the content of B was lower in roots of –B +2Ca than in –B +Ca treatments, especially in pea plants. Conversely, the content of B was higher in nodules of –B +2Ca than in –B +Ca treatments, suggesting a remobilization of B from roots to growing nodules by influence of the supplement of Ca.
Table 2. Contents of B (µg g−1 dry weight) and Ca2+ (mg g−1 dry weight) in shoots, roots and nodules of Pisum sativum and Phaseolus vulgaris plants grown with different B and Ca2+ treatments
B (µg g−1 dry weight)
Ca2+ (mg g−1 dry weight)
See ‘Materials and Methods’ for experimental conditions. Values followed by the same letter are not significantly different. P ≤ 0.05.
43.1 ± 6.9a
23.3 ± 5.6a
53.5 ± 5.7a
11.1 ± 2.3a
6.2 ± 0.1a
5.6 ± 1.2a
23.6 ± 4.3b
8.4 ± 1.5b
3.9 ± 0.2b
12.2 ± 3.3a
5.7 ± 0.6a
5 ± 0.9a
19.5 ± 3.6b
5.2 ± 0.3c
14.3 ± 1.6c
13.9 ± 2.1a
6.1 ± 0.3a
5.4 ± 0.2a
37.4 ± 6.1a
28.3 ± 4.2a
56.1 ± 3.8a
16.8 ± 3.6a
7.1 ± 1.3a
6.1 ± 1.7a
38.1 ± 1.2a
17.2 ± 0.8a
73.5 ± 3a
18.1 ± 3.3a
12.3 ± 0.2a
8.4 ± 1.1a
11.1 ± 0.3b
10.3 ± 0.4b
6.7 ± 0.3b
14.6 ± 3.2a
11.3 ± 0.2a
6.7 ± 0.7a
12.4 ± 0.3b
8.9 ± 0.3c
30.3 ± 1.1c
17.9 ± 2.2a
15.2 ± 0.3b
13.6 ± 0.3b
31.0 ± 0.9a
14.1 ± 0.6a
60.3 ± 4.1d
25.2 ± 2.3b
15.7 ± 0.2b
15.6 ± 1.3b
Two main observations can be extracted from the above results. On one hand, nodulation of 2Ca2+ treated plants increased at any tested concentration of B. On the other hand, the 2Ca2+ treatments were able to induce nitrogenase activity, from 10% (to B +Ca2+) to 50% (–B +2Ca2+), of the activity from control nodules (+B +Ca2+). Given the significance of these observations, both legume–rhizobia early interactions and nodule anatomy were further investigated.
To investigate the causes of treatment effects on nodule number, nod gene activity and adsorption of rhizobia roots were analysed in pea (Table 3). The capacity of methanolic extracts from roots of –B +Ca2+ plants to induce nod gene activity was about six times lower than in the control (+B +Ca2+), as previously shown (Redondo-Nieto et al. 2001). The increase in Ca2+ concentration (+2Ca2+ treatments) during plant growth led to the increase of nod gene induction capacity of methanolic root extracts, independent of the level of B. Moreover, the absence of B resulted in a diminution of adsorption of rhizobia to roots, as shown by the adsorption index (%A) of –B +Ca2+ and +B +Ca2+ treatments. Supplemental Ca2+ increased %A, which was highest in the +B +2Ca2+ treatment.
Table 3. Measurement of early interactions between Rhizobium leguminosarum 3841 and pea roots
nod gene induction %βGal activity
Adsorption index (%A) Percentage of inoculated rhizobia
See ‘Materials and Methods’ for experimental conditions. Values followed by the same letter are not significantly different. P ≤ 0.05.
73.84 ± 12.42a
11.32 ± 2.0a
12.15 ± 1.38 b
1.52 ± 1.1b
318.50 ± 37.21c
5.08 ± 1.4c
250.80 ± 50.27 c
19.6 ± 5.6a
Previous work has shown that B deficiency resulted in abortion of infection threads and low levels of invasion in pea nodules (Bolaños et al. 1996). Therefore, we investigated whether supplemental calcium could enhance rhizobial invasion of nodule tissues in B-deficient plants. As expected, cell invasion was shown to be impaired in nodules of B-deficient plants, as seen by examining plants that had been inoculated with bacteria expressing a constitutive GFP construct (Fig. 4). Supplementation of B-deficient plants with additional calcium (Fig. 4C) enhanced invasion of nodule cells.
Light microscopy examination of pea and bean nodules (Fig. 5) reinforced fluorescent localization of GFP. Treatment of –B +Ca2+ led to indeterminate (Fig. 5B) and determinate (Fig. 5E) nodules that were almost empty of bacteria. The supplement of Ca2+ (Fig. 5C & F) not only enhanced rhizobial infection and spreading, but also preserved normal nodule structure. Boron deficiency in both indeterminate and determinate nodules showed abnormal nodule anatomy, without clear distinction between cortical, infected and uninfected regions (Fig. 5B & E). The differentiation of these regions and its boundaries were clearly recovered in –B +2Ca2+ nodules (Fig. 5C & F).
Despite the reported effects of Ca2+ on nodulation and nodule anatomy, the recovery of nitrogen fixation in B-deficient nodules treated with Ca2+ was only partial (50% of the control, Fig. 1A, +B +Ca2+ treatment). Given the important role of both nutrients in cell wall architecture, we investigated the ultrastructure of nodule cell walls (Fig. 6). Boron-deficient walls of both infected and cortical cells presented abnormalities, including thicker and thinner zones or even disrupted sections (Fig. 6B). Cell walls were irregularly shaped at the ultra-structural level in –B +2Ca2+ treatments (Fig. 6C), in comparison with control nodules (Fig. 6A). Immunocytochemical studies on cells from the nodule cortex showed that, in control nodules, un-esterified pectin (polygalacturonic acid) recognized by JIM 5 antibody can be mainly immunolocalized to the middle lamella and in the interface between cell wall and intercellular spaces (Fig. 6D) and rhamnogalacturonan II polysaccharide recognized by anti-RGII antiserum is localized in the cell wall–intercellular space interface (Fig. 6G). By contrast, in –B +Ca2+ (Fig. 6E & H) and –B +2Ca2+ (Fig. 6F & I) both polysaccharides appeared uniformly gold-labelled throughout the cell wall and even in the cytoplasm of cells (arrows). Immunolabelling with JIM 7 antibody that recognizes methyl-esterified pectin was uniform throughout the walls in all three treatments (data not shown). The altered distribution of pectin polysaccharides in B-deprived cell walls is indicative of an abnormal structure of the cell wall under B deficiency, and addition of Ca2+ did not completely recover it.
In this study, it has been confirmed that the development of nodulated legumes (Fig. 3) and the N2-fixing activity derived from the symbiosis between these plants and their host rhizobia are influenced by the levels of both B and Ca2+ (Fig. 1). Furthermore, a supplement of Ca2+ was able to partially prevent the negative effects of B deficiency on pea nodule development and function. The increase of B content in –B nodules due to additional Ca2+ (Table 2; –B +2Ca2+ treatments) preventing severe B deficiency could be on the basis of these effects.
Results in Fig. 1B indicated that a concentration of 1.36 mm Ca2+ (2Ca2+) can suppress the inhibitory effects of B deficiency in nodulation, because Ca2+ increased the number of nodules independently of the concentration of B. The first step in the symbiotic interaction is the exchange of diffusible signals between plant and bacteria, which leads to the production of Nod factors following nod gene activation (Spaink 2000). Modifications of the level of both nutrients during plant growth varied the capacity of root extracts to induce rhizobial nod genes (Table 3; nod gene induction column). Boron deficiency led to exudates with a lower induction capacity (as previously described by Redondo-Nieto et al. 2001). This effect could be due to the influence of B nutrition on the metabolism of phenolics. Boron-deficient plants typically enhance production of quinones instead of flavonoids (Marschner 1995), being consistent with a decrease of nod gene-inducing capacity of exudates derived from –B +Ca2+ pea roots. Conversely, the concentration of 2Ca2+ resulted in a marked increase of induction activity at any concentration of B. These results suggest that there might be an increase of nod gene-inducing flavonoids in pea roots treated with Ca2+ as reported for other legumes (Richardson et al. 1988).
Besides the exchange of diffusible signals, the adsorption of rhizobia to roots is needed to initiate nodule formation in pea (Kannenberg & Brewin 1994). Data from adsorption experiments (Table 3; %A column) indicate that the level of both B and Ca2+ had an influence on the degree of attachment of bacteria to pea roots. This interaction is supposed to be mediated, among other components by bacterial Ca2+-dependent ricadhesins and cellulose fibrils, and plant lectins (Kannenberg & Brewin 1994). The role of B in the maintenance of the integrity of the cell wall is well established (Blevins & Lukaszewski 1998), and an unstable wall structure due to B deficiency, similar to that shown in Fig. 6, will diminish the adsorption of rhizobia to the root surface.
The addition of 2Ca2+ during growth of –B plants increased the adsorption index, albeit only 50% of the control. Figure 6 shows that the structure of B-deficient cell walls was still abnormal in –B +2Ca2+ treatments. Therefore, the root surface must still be altered and with a low capacity to interact with bacteria. Nevertheless, 2Ca2+ treatments always increased %A, even in the presence of B. This could possibly be due to an increased secretion of mucilage in roots (Bennet, Breen & Bandu 1990) in which cellulose fibrils can entangle the bacteria, and to a higher adhering capacity of Ca2+-dependent ricadhesins (Smit, Kjine & Lugtenberg 1989).
This study also shows that, in addition to the special Ca2+ requirement for nodulation, both B and Ca2+ are important for nodule development and function, because the decrease in either nutrient also inhibited nitrogen fixation (Fig. 1A). Following nodule development, rhizobia invade the plant through a transcellular tunnel (the infection thread) sheathed with cell wall material and an endocytosis-like process from unwalled infection droplets (Brewin 1991). Within the threads, rhizobia are embedded in intercellular plant-derived matrix material, including a plant matrix glycoprotein (MGP), which is secreted by plant cells into the lumen of the infection thread as an early response to rhizobial infection (VandenBosch et al. 1989; Rae, Bonfante-Fasolo & Brewin 1992), and that has recently been characterized as an extensin-like protein (Rathbun, Naldrett & Brewin 2002). Boron plays an important role at these stages because it is able to modulate plant–bacteria cell surface interactions. In the absence of B, the plant-derived infection thread MGP can attach to the cell surface of rhizobia. Therefore, the bacterium cannot progress through the infection thread and becomes unable to reach the endophytic environment. The presence of B (but not of Ca2+, pH changes, salt or high ionic strength) specifically inhibits the in vitro attachment of MGP to bacteria cell surface attachment and promotes the rhizobial interaction with the plant cell membrane (Bolaños et al. 1996). Consequently, B deficiency interferes with infection thread development, arresting it at early stages prior to endocytosis, and therefore leading to poorly invaded nodules similar to those shown in Figs 4B and 5B (indeterminate pea nodules), and Fig. 5E (determinate bean nodule).
Interestingly, the presence of 2Ca2+ rescued the invasion phenotype of -B nodules (Figs 4C, 5C & F). Table 2 indicates an increase up to five times in the content of B in –B +2Ca2+ nodules compared with –B +Ca2+ treatments, albeit still lower than the control. Therefore, these nodules are not under severe B deficiency. Since only B and not Ca2+ was able to mediate the cell surface interactions previously described (Bolaños et al. 1996), leading to endocytosis of rhizobia, it could be postulated that the increase of B after Ca2+ addition promoted cell invasion by Rhizobium. Carpena et al. (2000) reported that addition of high Ca2+ during pea growth under B deficiency could mediate mobilization of B from old to young tissues. Table 2 suggests that it can also be true in nodulated plants, since B increase in nodular tissue was accompanied by a decrease of B content in roots of –B +2Ca2+ treatments. Given the adsorption capacity of the B-specific resin Amberlite IRA 743 (Asad et al. 1997) it is unlikely that this B came from impurities and/or contaminations, and therefore it might have come from old tissues.
The ultra-structural study of nodule cell walls (Fig. 6) shows that B is essentially required for the in vivo assembly of pectin as part of the cell wall architecture, and that the increase of B content in nodules due to addition of Ca2+ (Fig. 6C, F & I) was not sufficient to prevent the abnormal cell wall structure. The distribution of both homogalactur-onan and rhamnogalacturonan is altered in –B +Ca and –B +2Ca2+ nodules. These observations agree with studies by Kobayashi et al. (1999) and Ehwald et al. (2002) who postulate that B cross-links pectin polysaccharides through diester bonds with apiosyl residues from the rhamnogalacturan II region (O’Neill et al. 1996, 2001), and Ca2+ stabilizes pectin through bonding in the polygalacturonic acid region and through acidic sugars of rhamnogalacturonan II. Not only pectins are abnormally distributed but also hydroxyprolin-/proline-rich proteins are accumulated in the cytoplasm instead of being covalently bound to the cell wall of B-deficient nodules (Bonilla et al. 1997). Nodule cell walls contained more protein than root cell walls (Frueauf et al. 2000), therefore B-deficient nodule cell walls may be more disorganized than root cell walls. Since the formation of the complex di-rhamnogalacturonan II-B is determinant for cell wall density through allocation of Ca2+-pectate and hence other wall components (Matoh & Kobayashi 2002), the addition of extra Ca2+ to B-deficient plants did not result in recovery of wall packing (Fig. 6C, F & I). Moreover, cytoplasmic localization of wall pectin (Fig. 6E & F, arrows) and proteins (Bonilla et al. 1997) could indicate abnormal targeting of Golgi-derived vesicles in B-deficient tissues, as postulated by Goldbach (1997) and as occurring with targeting of other macromolecules in B-deficient nodules (Bolaños et al. 2001). Abnormalities of B-deficient cell wall structure lead to a decrease of cell wall elasticity (Findeklee & Goldbach 1996) accompanied by an unusual fragility. Therefore, B-deficient nodule cell walls, with disorganized pectin polysaccharides and without associated structural proteins, lose such properties and hence lose functionality that cannot be prevented by Ca addition.
The antibodies used in this study were kindly provided by Dr Nicholas J. Brewin and R. leguminosarum D24 by Dr J. Allan Downie (John Innes Centre, Norwich, UK). Antiserum to rhamnogalacturonan II was provided by Dr Toru Matoh (Kyoto University, Japan). This work was supported by MCYT no. BOS2002-04164-CO3-02.