H₂O₂ Is Required for Optimal Establishment of the Medicago sativa/Sinorhizobium meliloti Symbiosis

Leguminous plants can engage in a symbiotic interaction with Rhizobia and form new root organs, the nodules. The nodulation process is initiated by a complex signal exchange between both partners (5). Rhizobial invasion in the symbiotic model Medicago sativa/Sinorhizobium meliloti occurs via root hairs. The perception of bacterial nodulation factors by the host plant leads to cell division in the root pericycle and in the root cortex, where the nodule primordium forms. Simultaneously, root hairs deform and curl. The bacteria are entrapped in this curl, local cell wall is hydrolyzed, and a plasma membrane invagination occurs, leading to the formation of an infection thread (IT) (3). This plant-derived tubule filled with dividing and growing bacteria first progresses through the infected root hair and then traverses several cell layers toward the nodule primordium. Bacterial cells at the tip of the IT are released and enclosed in peribacteroid membranes in the host cell cytoplasm (10). Once inside the plant cells, the bacteria differentiate into nondividing nitrogen-fixing endosymbionts called bacteroids, which are able to fix atmospheric dinitrogen (22).

Important progress has recently been made in analyzing the mechanisms that initiate the establishment of the symbiosis and especially the mechanism of the perception of bacterial nodulation factors by the host plants (12, 18). The development of an IT inside root hair and the architecture of IT networks in nodules have been recently studied (9, 21). In fact, ITs can be considered as tubular ingrowths of the plant cell wall. In pea/Rhizobium leguminosarum symbiosis, a family of extensin-like glycoproteins associated with the lumen of these ITs has been identified (25). However, more data are necessary to clearly identify the biochemical mechanisms by which ITs grow. Reactive oxygen species like O₂ ⁻ and H₂O₂ could be implicated, since they have been implicated during the infection process and especially in the ITs (28). The observed distribution of H₂O₂ in ITs and the plant cell wall strongly suggests that H₂O₂ is of plant origin (30). To overcome this reactive oxygen species production, S. meliloti possesses at least two superoxide dismutases (SodA and SodC) and three catalases (KatA, KatB, and KatC). It has been shown that the corresponding genes are differentially expressed not only during free-living growth but also during nodule establishment. Indeed, both sod genes are expressed in ITs (27; D. Touati, personal communication). In contrast, only katB and katC genes, and not the uniquely H₂O₂-inducible katA gene (15), are strongly expressed in the ITs despite the detection of H₂O₂ all around the bacteria (16). Moreover, a nonefficient release into plant cells and a perturbed differentiation of bacteria into bacteroids have been observed for a katB katC double mutant (16). In free-living conditions, the coexpression of katB and katC has been observed only during the stationary phase, suggesting that most of bacteria within the ITs are potentially in a nonexponential growth phase. This has recently been corroborated by experiments showing that bacterial growth occurs only in a very small region on the tip of ITs (9). Moreover, on the basis of observations of root tips, the potential implication of H₂O₂ in an insolubilization of the glycoprotein matrix within IT lumen has been suggested (30).

To further comprehend the mechanisms of IT growth, we tried to modify the H₂O₂ content within IT by using an S. meliloti strain overexpressing one catalase. Among the three catalases, KatB was selected because it plays a housekeeping role, as it is expressed throughout all the growth phases of the free-living bacterium and also during symbiosis. Moreover, KatB shows a higher affinity for H₂O₂ than the two other catalases of S. meliloti (1). In order to overexpress KatB, the Salmonella enterica serovar Typhimurium trp promoter was fused upstream of the complete coding region of katB by four cloning steps. The first of these consisted of amplification of the katB open reading frame by PCR using the following primers: AJKATB1 (5′-CCATCGATCGCCCAATCGAGGAAGAGG-3′), creating a new ClaI restriction site, and AJKATB2 (5′-ACATGGGCTCGAGCCACGT-3′), creating a new XhoI restriction site (boldface in sequences indicates restriction sites). Subsequent steps included subcloning in pGEM-T (Promega corporation), double digestion ClaI and XhoI, and insertion into pTB93F (11) to create the pTBkatB ⁺⁺ plasmid. The efficiency of the trp promoter to overproduce KatB has been tested by transferring this recombinant plasmid into the recipient katB-null mutant GKBZ01 expressing a pkatB-lacZ fusion by triparental mating. Total catalase activities were 36 ± 6.7, 17 ± 7.2, and 380 ± 102 U/μg of protein in crude extracts of GMI211, GKBZ01, and GKBZ01/pTBkatB ⁺⁺, respectively, indicating that the trp promoter allows a 10-fold increase in the catalase activity in S. meliloti. However, the recombinant plasmid was unstable in S. meliloti during the plant infection process without antibiotic pressure (data not shown).

To avoid this problem, the chimeric construction allowing the overexpression of katB was integrated into the pSymA megaplasmid by recombination. Four more cloning steps were necessary before the transfer by triparental mating. The BamHI/BglII fragment of the pTBkatB ⁺⁺ plasmid, containing the ptrp promoter and 80% of the KatB open reading frame, was subcloned into the BamHI site of the pSUP202 plasmid, resulting in the pSUPkatB ⁺⁺ plasmid. A lacZ/Nm^r cassette extracted from the pKOK5 plasmid was inserted into the SalI restriction site of pSUPkatB ⁺⁺, resulting in the pSUPkatB ⁺⁺ lacZ plasmid. The recombinant RmkatB ⁺⁺ strain expressing both pkatB-lacZ and pTrp-katB fusions was obtained by triparental mating using the Rm1021 wild-type strain as that recipient strain. Expression levels of trp and katB promoters in the ITs are not known, and it was impossible to accurately measure the catalase activity from IT bacteria. However, the catalase capacities of free-living RmkatB ⁺⁺ were determined during growth in Luria-Bertani medium containing 2.5 mM MgSO₄ and 2.5 mM CaCl₂ (LB-MC). During the exponential phase, the total catalase activity in RmkatB ⁺⁺ was three times higher than that in Rm1021, whereas a 10-fold increase was observed during the stationary phase (Table 1). Analysis of the catalase activity pattern on native polyacrylamide gel by use of inhibition of diaminobenzidine oxidation by H₂O₂ (14) showed that the increase of total catalase activity in the stationary phase was largely due to an increase of KatB (Fig. 1A and B). To evaluate the resistance of RmkatB ⁺⁺ to exogenous H₂O₂, the survival of bacteria was checked by incubating for 30 min in LB-MC containing an increasing concentration of H₂O₂. The treatment was stopped by adding exogenous catalase, and bacteria were counted on the LB-MC plate. In the presence of 10 mM H₂O₂, wild-type Rm1021 bacteria were rapidly killed, whereas the survival of recombinant RmkatB ⁺⁺ cells was only slightly reduced (data not shown). The intracellular concentration of H₂O₂ was evaluated in both strains by use of an Amplex Red kit (Molecular Probes, Montluçon, France) as described previously for Agrobacterium tumefaciens (31). H₂O₂ concentrations of 124 nM and 52 nM in Rm1021 and RmkatB ⁺⁺, respectively, were estimated, indicating that the mutant strain has a lower internal H₂O₂ level.

To test the effect of katB overexpression on the symbiotic process, the symbiotic performance of recombinant RmkatB ⁺⁺ strain to its parental wild type was assessed by three independent experiments. The number of nodules per 72 plants was determined 7, 14, and 21 days after inoculation with the mutant RmkatB ⁺⁺ strain and with Rm1021 as the control. The values obtained for 7 and 14 days postinoculation indicate that the RmkatB ⁺⁺ strain presents a delayed nodulation phenotype following the infection (Table 1). Moreover, this delay does not appear to be due to a modification of the early steps of the infection process, since the expression levels of early plant nodulin genes ENOD12 (24) and MtN6 (20), determined by reverse transcription-PCR using total RNA from M. truncatula root tips, were found to be similar for infection with Rm1021 and RmkatB ⁺⁺ (data not shown). However, 3 weeks after infection, no significant difference between plants inoculated with Rm1021 or RmkatB ⁺⁺ was observed in terms of the number of nodules and the capacity of bacteroids to reduce the dinitrogen within nodules (data not shown). To determine more precisely the point at which the RmkatB ⁺⁺ strain is affected during the infection process, the plasmid pXLGD4 (17), which contains a hemA-lacZ fusion allowing a constitutive expression of lacZ, was transfected into the Rm1021 and RmkatB ⁺⁺ strains. A set of 40 plants were infected for each strain, and the efficiency of infection was estimated 5 days after infection using the following criteria: colonized curled root hairs, initiated ITs, extended ITs, and infected young nodules (Fig. 2). The global numbers of infection events were similar for both strains. In contrast, the number of extended ITs and the ratio of initiated ITs to colonized curled root hairs were significantly higher for Rm1021/pXLGD4 than for RmkatB ⁺⁺/pXLGD4. Moreover, the ITs showed an irregular aspect and a larger average diameter for RmkatB ⁺⁺ (7.34 ± 2.25 μm) than for Rm1021 (3 ± 1.07 μm). The diameters of 40 ITs from each strain were measured using images from bright-field microscopy (Fig. 3). H₂O₂ accumulation was analyzed in ultrathin sections of ITs of very young nodules as an electron-dense deposit stained with cerium chloride as describe previously (2, 28). When plants were infected with Rm1021, H₂O₂ was detected in most of the ITs, principally between the cell wall and the matrix (Fig. 4A and B). In contrast, no H₂O₂ could be observed in enlarged ITs obtained with RmkatB ⁺⁺, whereas the peroxide is still detectable in the cell wall (Fig. 4C and D). This indicates that the mutant has an enhanced ability to inactivate H₂O₂ associated with IT walls. As a control, the detection of H₂O₂ in ITs was completely abolished by treating the sample with 25 μg ml⁻¹ of catalase before adding cerium chloride, indicating that it reflects H₂O₂ accumulation (data not shown).

Other bacterial mutant strains have been shown also to exhibit a nodulation-delayed phenotype similar to that of RmkatB ⁺⁺, particularly those affected in their production of exopolysaccharides (4, 6, 8, 23). To determine if the exopolysaccharide content was affected in RmkatB ⁺⁺, two distinct tests were performed: one with the UV-fluorescent calcofluor, which binds a portion of succinoglycan (7), and the other with Sudan black B, which detects the galactoglucan and some succinoglycan, which are not involved in calcofluor fluorescence (19). No difference between Rm1021 and RmkatB ⁺⁺ was observed, indicating that the phenotype observed for RmkatB ⁺⁺ could not be explained simply by a modification of exopolysaccharide production (data not shown).

In conclusion, our results indicate that the H₂O₂ production observed within some ITs (28), which did not appear to induce an oxidative stress for bacteria (16), has a crucial role in optimizing IT development. This is in line with the hypothesis of an important H₂O₂ involvement in the biomechanics of IT growth via the cross-linking of glycosylated plant glycoproteins, namely, the root nodule extensins, which are localized in the extracellular matrix of legume tissues and in the lumen of ITs (13, 30). The reduced levels of H₂O₂ in the ITs obtained with RmkatB ⁺⁺ could influence the rate of cross-linking of these extensins, compromising a polarized growth of ITs and leading to their enlargement. The results presented here are in agreement with an early report by Salzwedel and Dazzo (26), where peroxidase activities were localized during root hair infection and were supposed to contribute to the infection process.

• Open in new tab
• Download powerpoint

FIG. 1.

Catalase activity of Sinorhizobium meliloti strains: Rm1021 (lane 1) and RmkatB ⁺⁺ (lane 2) in exponential (A) and stationary (B) phases. The locations of KatA, KatB, and KatC are indicated according to the work of Sigaud et al. (29). Twenty-five micrograms of total protein was loaded on each lane.

• Open in new tab
• Download powerpoint

FIG. 2.

Number of infection events 5 days after infection with Rm1021 and RmkatB ⁺⁺ strains. Experimental data were assessed for statistical significance by means of the Student t test; * indicates significant difference (P < 0.05). Abbreviations: CCRH, colonized curled root hair; IIT, initiated IT; EIT, extended ITs; YN, young nodule; WT, wild type.

• Open in new tab
• Download powerpoint

FIG. 3.

Structure of ITs in 5-day-old nodules. Nodules were induced by the Rm1021 (A and C) and RmkatB ⁺⁺ (B and D) strains. Black arrows show ITs, and open arrowheads show nuclei. Scale bars = 10 μm.

• Open in new tab
• Download powerpoint

FIG. 4.

Detection of H₂O₂ accumulation as electron-dense precipitate formed in the presence of cerium chloride (arrows) in ITs 5 days after infection with Rm1021 (A and B) and RmkatB ⁺⁺ (C and D). (A, C, and D) Transversal cross section of ITs. (B) Longitudinal cross-section showing the release of bacteria inside the plant cell. Magnitudes of pictures are indicated with scale bars.

• Copyright © 2007 American Society for Microbiology