Reactive oxygen species are produced as an early event in plant defense response against avirulent pathogens. We show here that alfalfa responds to infection with Sinorhizobium meliloti by production of superoxide and hydrogen peroxide. This similarity in the early response to infection by pathogenic and symbiotic bacteria addresses the question of which mechanism rhizobia use to counteract the plant defense response.

In nitrogen‐poor soils, rhizobia elicit nodule formation on legume roots, within which they differentiate into bacteroids that fix atmospheric nitrogen. Protection against reactive oxygen species (ROS) was anticipated to play an important role in Rhizobium–legume symbiosis because nitrogenase is extremely oxygen sensitive. We deleted the sodA gene encoding the sole cytoplasmic superoxide dismutase (SOD) of Sinorhizobium meliloti. The resulting mutant, deficient in superoxide dismutase, grew almost normally and was only moderately sensitive to oxidative stress when free living. In contrast, its symbiotic properties in alfalfa were drastically affected. Nitrogen‐fixing ability was severely impaired. More strikingly, most SOD‐deficient bacteria did not reach the differentiation stage of nitrogen‐fixing bacteroids. The SOD‐deficient mutant nodulated poorly and displayed abnormal infection. After release into plant cells, a large number of bacteria failed to differentiate into bacteroids and rapidly underwent senescence. Thus, bacterial SOD plays a key protective role in the symbiotic process.

In Escherichia coli , the expression of sodB , which encodes iron superoxide dismutase, has been suggested to be activated by Fur, the iron-responsive global regulator initially characterized as a transcriptional repressor. We investigated sodB regulation by functional analysis of the sodB promoter using sodB-lac fusions with various truncated and mutated promoters. Several cis - and trans -acting elements involved in sodB regulation have been identified. The β-galactosidase activity of sodB-lacZ reporter fusions and RNA analysis showed sevenfold iron-dependent, Fur-mediated activation of expression. A region just downstream from −10, including a large palindromic sequence encompassing the +1 position followed by a 14-bp AT-rich motif, is the site of Fur positive regulation, and the integrity of both sequences was required for full Fur-mediated activation. The life span of sodB mRNA was three times longer in a fur ⁺ strain, indicating that Fur-mediated activation proceeds, at least in part, at the posttranscriptional level. The H-NS and IHF histone-like factors also affected sodB expression. IHF slightly repressed sodB expression independently of Fur regulation. In contrast, H-NS negative regulation operated only in the absence of Fur. Remarkably, p sodB behaved like a “pure extended -10” promoter. Deletion of the −35 region did not affect expression, whereas expression was totally abolished by a TG-to-CC mutation in the extended −10 sequence TGcTACCCT.

Fur (ferric uptake regulation protein) activates sodB expression, increasing expression levels by a factor of seven and sodB transcript stability by a factor of three. Post-transcriptional regulation of sodB was investigated by searching for endoribonucleases that might be involved in sodB mRNA degradation. The activation of sodB expression was significantly reduced if both the RNaseE and RNaseIII genes were mutated. This correlated with cleavage at a palindromic sequence located in the 5′ untranslated region of the sodB transcript. An RNA-binding assay showed that Fur did not directly protect the sodB transcript. It was hypothesized that the persistence of Fur-mediated activation of sodB expression in the RNase double mutant was probably due to an effect at the transcriptional level. Therefore, it was investigated whether Fur had a direct transcriptional effect in vitro. Fur bound the sodB promoter region with low affinity, but it was not able to increase sodB transcription. H-NS-mediated repression of sodB expression, which has been shown to be Fur-dependent, was characterized. No DNA-bending region was identified in the sodB promoter region. H-NS did not interfere with the post-transcriptional effect of Fur. Fur-dependent H-NS and the Fur post-transcriptional effect were not additive. This suggests that Fur and H-NS effects are indirect and may be mediated by a common intermediate.

Sinorhizobium meliloti Rm5000 is an aerobic bacterium that can live free in the soil or in symbiosis with the roots of leguminous plants. A single detectable superoxide dismutase (SOD) was found in free-living growth conditions. The corresponding gene was isolated from a genomic library by using a sod fragment amplified by PCR from degenerate primers as a probe. The sodA gene was located in the chromosome. It is transcribed monocistronically and encodes a 200-amino-acid protein with a theoretical M _r of 22,430 and pI of 5.8. S. meliloti SOD complemented a deficient E. coli mutant, restoring aerobic growth of a sodA sodB recA strain, when the gene was expressed from the synthetic tac promoter but not from its own promoter. Amino acid sequence alignment showed great similarity with Fe-containing SODs (FeSODs), but the enzyme was not inactivated by H ₂ O ₂ . The native enzyme was purified and found to be a dimeric protein, with a specific activity of 4,000 U/mg. Despite its Fe-type sequence, atomic absorption spectroscopy showed manganese to be the cofactor (0.75 mol of manganese and 0.24 mol of iron per mol of monomer). The apoenzyme was prepared from crude extracts of S. meliloti . Activity was restored by dialysis against either MnCl ₂ or Fe(NH ₄ ) ₂ (SO ₄ ) ₂ , demonstrating the cambialistic nature of the S. meliloti SOD. The recovered activity with manganese was sevenfold higher than with iron. Both reconstituted enzymes were resistant to H ₂ O ₂ . Sequence comparison with 70 FeSODs and MnSODs indicates that S. meliloti SOD contains several atypical residues at specific sites that might account for the activation by manganese and resistance to H ₂ O ₂ of this unusual Fe-type SOD.