The bacterial strains and plasmids used are listed in Table 1. R. etli CE3 (a spontaneous Sm^r mutant of R. etli CFN 42^T) 31 was used as the wild type strain. R etli strains were routinely grown in complex TY medium 32 . E. coli strains were grown aerobically in complex LB medium 33 . B^-medium 34 , which contains 10 g l^-1mannitol as the sole carbon source, was used as minimal medium for R. etli. When appropriate, trehalose and glucose were also used as carbon source at a final concentration of 20 mM. Osmotic strength of this medium was increased by the addition of 0.1 to 0.2 M final concentration of NaCl. pH was adjusted to 7.2 (for TY) or 5 (for B^-). Solid media contained 2% of Bacto agar (Difco). E. coli cultures were incubated at 37°C. R. etli cultures were incubated at 28°C or 35°C 29 . When used, filter sterilized antibiotics were added at the following final concentrations (μg ml^-1): ampicillin (ap), 150 for E. coli; chloramphenicol, 30 for E. coli; gentamicin (Gm), 20 for E. coli, 25 for R. etli; streptomycin (Sm) 20 for E. coli, 40 for R. etli; spectinomycin (Spc) 80–100 for R. etli and nalidixic acid 20 for R. etli. When appropriate, the following compounds were added to the media (final concentration): X-Gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside, Sigma, 40 μg/ml), IPTG (isopropyl-β-D-1-thiogalactopyranoside, Sigma, 25 μg/ml). Growth was monitored as the optical density of the culture at 600 nm (OD₆₀₀) with a Perkin-Elmer Lambda 25 UV/Vis spectrophotometer.

Aliquot volumes (1 ml) of B^- medium cultures in early stationary phase were harvested by centrifugation. Cell pellets were washed with the same medium without any carbon source, centrifuged for 5 min at 13000 rpm and, after removing the supernatant, vacuum dried. Two variations of the protocol described by Manzanera et al. 39 were used. In a first step, two replicates of all samples were dried by vacuum in a Memmert V0200 vacuum oven at 20°C and 313 mbar for 20 h. After that, for each sample, one replica was taken out from the oven, sealed and stored at 28°C, and the other was subjected to a further step under vacuum consisting on a temperature ramping of 2°C/min with a 15-min pause after every increase of 2°C, up to a maximum temperature of 30°C, followed by storage at 28°C. For assessment of viability, after variable time periods, dried samples were resuspended in 1 ml of TY complex medium, and serial dilutions were plated on TY plates, incubated at 28°C, and counted to determine CFU. Viability was measured before (taken as 100% survival) and just after drying, and at 4 days, 1, 2, and 3 weeks storage, and expressed as percentage of viable cells.

R. etli wild-type and otsA mutant strain (CMS310) were grown in B^-medium with 0.2 M NaCl at 28°C until early-stationary phase. Cells were collected by centrifugation and washed with the same medium without any carbon source. Cell pellet was resuspended in 10 ml of extraction mixture (methanol:chloroform:water; 10:5:4) and extracted by gently shaking for 30 min at 37°C. Cell debris was removed by centrifugation, and supernatants were extracted once with chloroform:water (1:1) and freeze-dried. The solids were dissolved in D₂O (0.6 ml). ¹³C-NMR spectra were recorded at 25°C on a Brucker AV500 spectrometer at 125 MHz. The chemical shifts are reported in ppm on the δ scale relative to tetramethylsilane. Signals were assigned by comparison with previously published chemical shift values 6 and compared with ¹³C-NMR of pure compounds.

Trehalose determination was performed basically as described by Blazquez et al. 40 by the following procedure. For free-living cells, pellets from 15 ml of early stationary phase cultures in B^-medium were washed with isotonic carbon-free medium and resuspended in 1 ml of the same medium. Cells were lysed by 30 min of incubation at 95°C and, after centrifugation, the supernatant was used to determine the trehalose content in a total volume reaction of 200 μl containing 100 μl of the supernatant, 90 μl of 25 mM sodium acetate buffer (pH 5.6) and 0.02 U of trehalase (Sigma). For each sample, endogenous glucose was monitored by performing a parallel reaction in which trehalase was substituted by water. After overnight incubation at 37°C, the glucose released by trehalose hydrolysis was determined by adding 150 μl of the previous reaction to 150 μl of a mixture of 0.66 mg ml^-1 Aspergillus niger glucose oxidase (Sigma), 0.25 mg ml^-1 horseradish peroxidase in 0.5 M phosphate buffer, pH 6.0 (Sigma), and 50 μl of 2.33 mg ml^-1 o-toluidine (Panreac). After 30 min of incubation at 37°C, 1.5 ml of water was added to the samples and absorption was measured at 420 nm in a Perkin Elmer Lambda 25 UV/Vis spectrophotometer. Values were compared to those obtained from stock solutions of glucose standards in a concentration range of 0 to 1000 μgml^-1. Finally, trehalose content was calculated from the glucose content by performing a standard curve with commercial trehalose (Sigma) ranging from 1 to 5 mM. Trehalose concentration was expressed as μmol mg protein^-1. Nodules were fractionated into bacteroids and nodule cytosol as described by Delgado et al. 41 . Trehalose content was determined colorimetrically as described above.

The same cultures were used for determination of both trehalose and protein content. 1 ml aliquots were taken at early stationary phase and cell protein content was determined in triplicate by using a bicinchoninic acid (BCA) proteinassay kit (Pierce) as described by García-Estepa et al. 42 .

Plasmid DNA was isolated from E. coli with a Wizard Plus SV miniprep kit (Promega), and genomic DNA was isolated with a SpinClean Genomic DNA Purification kit (Mbiotech). Restriction enzyme digestion and ligation were performed as recommended by the manufacturers (Amersham-Pharmacia Biotech and Fermentas). DNA sequencing was performed by Newbiotechnic (Seville, Spain). To generate the R. etli CE3 otsAch mutant CMS310 (otsAch::Ω), a 4.119-bp fragment from the R. etli genome containing 394-bp of the adjacent gene frk, otsAch and 1.488-bp of the pgi gene, was amplified with Pfu Turbo DNA polymerase (Stratagene) by using two synthetic oligonucleotides (otsA ^R-FW: 5’-AAGACGGCTGTGAACGACGAG-3’ and otsA ^R-RV: 5’-CAAATCCGACATCGTCAAATTCTC-3’). The resulting PCR fragment was cloned into pUC19-301 digested with EcoRV to obtain the plasmid pMOtsA1. This plasmid was digested with BglII-XbaI, to remove the 4.2-kb fragment containing the otsA region which was cloned in pSKbluescript previously digested with BamHI-XbaI to obtain the plasmid pMotsA4. Subsequently, a BglII recognition site was generated in otsAch gene sequence, using the PCR-based QuickChange Site Directed Mutagenesis Kit (Stratagene) and the primers: otsA ^R BglII FW (5’-GAAGAGAGGGCATTGGCGA

A

Plasmids were transferred from E. coli to R. etli by triparental mating on TY medium, using pRK600 as a helper plasmid 37 , as described by Vargas et al. 43 but with a 1:2:1:(donor:receptor:helper) ratio.

The sequence of the R. etli CFN 42 genome is available at NCBI microbial genome database ( http://www.ncbi.nlm.nih.gov/genomes/lproks.cgi; Ac N°: NC_007761), and at http://www.ccg.unam.mx/retlidb/. Sequence data were analyzed using BLAST (NCBI http://ncbi.nlm.nih.gov/BLAST). ORF assignments of the metabolic pathways more relevant for this work was performed by comparing the information available at the Kyoto Encyclopedia of Genes and Genomes (KEGG) 44 and MetaCyc 45 . Codon preference was analysed at the Kasuza Codon Use Database ( http://www.kazusa.or.jp/codon/). Phylogenetic and molecular evolutionary analyses were conducted using MEGA version 5 46 . Sequences were aligned with ClustalW (1.6) using a BLOSUM62 matrix, and manually edited. The phylogenetic tree was inferred using the Neighbor-Joining method 47 , and the evolutionary distances were computed using the Poisson correction method. The rate variation among sites was modelled with a gamma distribution (shape parameter = 1) and all positions containing gaps and missing data were eliminated only in pairwise sequence comparisons. The robustness of the tree branches was assessed by performing bootstrap analysis of the Neighbour-Joining data based on 1000 resamplings 48 .

Seeds of Phaseolus vulgaris, var. Negro jamapa were surface sterilized in 96% (v/v) ethanol for 30 s, followed by 5% sodium hypochlorite for 5 min and then thoroughly washed five times with sterilized water 41 . Sterilized seeds were incubated two hours in sterile distilled water in darkness. The imbibed seeds were deposited on plates containing 1% water-agar, and let them to germinate at 30°C during 60 h. Three day after sowing, selected uniform seedlings were planted in sterile 2-kg pots (4/pot) containing a mixture of vermiculite/sand (1:1 v/v) as substrate. Each seedling was inoculated with 1 ml of the strains culture of R. etli wild-type and otsA mutant strain (CMS310) containing 10⁸ cells/ml at the log-phase of growth. Plants were grown in controlled environmental chambers (night/day temperature 19/25°C, photoperiod 16/8 h, PPF 400 μmol m^-2 s^-1 and relative humidity 60 to 70%) 49 . Plants were watered with N-free mineral solution 50 and water alternately. When plants were three weeks old, they were randomly separated into two sets: control and drought stress. Drought stress was imposed by withholding water for 5 days (moderate drought) or for 10 days (severe drought). Control plants were supplied daily with nutrients solution to field capacity. Leaf water potential (Ψ_w) was measured in the first fully expanded leaf of common bean plants with C52 sample chambers connected to a HR-33 T psychrometer (Wescor. Inc., Logan UT, USA).

Plant and nodule dry weight were determined after drying fresh plant material that was heated at 60°C for 48 h. Total nitrogen content per plant was determined by the Kjeldahl method 51 . Nitrogenase activity was analyzed by using the acetylene reduction activity (ARA) assay. A Hewlett-Packard model 5890 gas chromatograph (Agilent Technologies, S.L., Madrid) equipped with a flame ionization detector was operated with a molecular sieve 5A (60 to 80 mesh) column (180 × 0.32 cm). N₂ at 60 ml min^–1 served as a carrier gas. Oven, injector, and detector temperatures were 60, 90, and 110°C, respectively. Nodules (0.3 g) were placed in 17-ml tubes that were filled with 10% acetylene. Gas samples (0.5 ml) were taken from the tubes for ethylene analyses after incubation for 10 and 20 min. Concentration of ethylene in each sample was calculated from standards of pure ethylene.

Leghemoglobin content was measured by fluorimetry as previously described by La Rue and Child 52 . Nodules (0.3 g) were ground with 4 ml Lb extraction buffer (Na₂HPO₄·2H₂O 40 mM (pH 7.4); NaH₂PO₄·H₂O 10 mM (pH 7.4); K₃Fe (CN)₆ 0.02%; NaHCO₃ 0.1%) supplemented with 0.1 g polyvinylpolypirrolidone (PVPP). The homogenate was centrifuged at 12 000 rpm at 4°C for 20 min, to retain the supernatant. Clear supernatant (50 μl) and saturated oxalic acid (3 ml) were mixed in screwcapped tubes, which were sealed and autoclaved for 30 min at 120°C and then let to cool to room temperature. The fluorescence of the solutions was measured with a Shimadzu (Shimadzu Scientific Instruments, Kyoto, Japan) spectrophotofluorometer equipped with a mercury-xenon lamp and a RF-549 red-sensitive photomultiplier. The excitation wavelength was 405 nm and the emission monochromator setting was 650 nm. The difference in fluorescence between heated and unheated samples was proportional to haem protein concentration.

Heat stress induces accumulation of trehalose in yeasts 25 and bacteria such as E. coli 26 or Salmonella typhi serovar Typhimurium 27 . In rhizobia, including R. etli 10 , trehalose synthesis has been shown to be stimulated by salinity, but its role against heat stress has not been yet tested. In this study, we compared the influence of salinity and high temperature on growth and trehalose accumulation in R. etli. For this purpose, R. etli wild-type strain was grown up to early stationary phase in B^-minimal medium alone or with 0.2 M NaCl, at 28°C and 35°C, and trehalose content was determined colorimetrically as described in Materials and Methods. As shown in Figure 1, osmotic stress alone caused a delayed growth, but high temperature alone did not influence growth of R. etli. However, growth of cells subjected to both stresses was more impaired than that of cells grown under osmotic stress alone, showing an attenuated exponential phase, and reaching final O.D₆₀₀ values below 0.9. As shown in Figure 1, under non stress conditions, trehalose levels in R. etli were below 0.025 μmol/mg protein. To determine trehalose content in response to high temperature stress, we compared the accumulation of trehalose at 28°C and 35°C in cells grown without NaCl added. Under these conditions, trehalose accumulation by R. etli cells increased by 2.2-fold, but trehalose levels remained very low. However, a pronounced response in trehalose accumulation was observed due to salinity stress at both temperatures. Thus, trehalose levels in cells grown in minimal medium with 0.2 M NaCl at 28°C and 35°C were 13.5- and 5.04- higher, respectively, than trehalose levels in cells grown in minimal medium without NaCl added. These data suggest that, although temperature stress alone induces some trehalose synthesis by R. etli, trehalose biosynthesis in this microorganism is mainly triggered by osmotic stress.

Trehalose synthesis and catabolism have proven to be relevant for the symbiotic performance of rhizobia 5 10 21 22 . To get an overview of the metabolism of trehalose in R. etli, we inspected its genome for genes involved in trehalose synthesis, transport and degradation. Genes for trehalose metabolism were scattered among the chromosome and plasmids a, c, e, and f (see Additional file 1: Table S1 and Additional file 2: Figure S1A, for a complete description of gene annotation and gene clustering). As suggested by Suarez et al. 10 putative genes encoding the three so far known trehalose synthesis pathways in rhizobia (TreYZ, TreS and OtsAB) are present in R. etli. First, genes encoding trehalose synthesis from glucose polymers were found in plasmid p42e (treY), and the chromosome and plasmid p42f (two copies of treZ). Second, two genes encoding a putative trehalose synthase (TreS) were found in the chromosome and plasmid p42f. The product of the chromosomal putative treS gene presented similar length and significant sequence identity to known trehalose synthases from bacteria, but the product of the plasmid f-borne treS-like gene carried an additional domain of unknown function (DUF3459).Third, two genes were annotated as otsA, one located in the chromosome (otsAch) and one in plasmid p42a (otsAa). Both products showed homology to trehalose 6-phosphate synthases from other rhizobia, but the identity was much higher for OtsAch. In addition, a gene annotated as otsB was located in plasmid c. As trehalose is synthesized by R. etli from mannitol (see Figure 1), we searched for genes involved in mannitol transport and conversion into glucose. The genome of R. etli does not encode a specific mannitol phosphotransferase, suggesting that mannitol does not use this system to enter the cell. Instead, we found smoEFGK (encoding a sorbitol/mannitol ABC transporter), mtlK (encoding a mannitol 2-dehydrogenase that converts mannitol to fructose), and xylA (encoding a xylose isomerase that converts fructose to glucose. All these findings suggest that R. etli can convert mannitol into glucose via fructose. R. etli CE3 grown in minimal medium B^- also accumulates glutamate (see below). Since B^- does not contain ammonium, the most plausible route for glutamate biosynthesis from mannitol is through the enzyme glucosamine-6-phosphate synthase, which converts D-fructose-6-phosphate and L-glutamine into D-glucosamine-6-phosphate and L-glutamate. Two copies of the encoding gene (glmS) were found in R. etli chromosome (Additional file 1: Table S1, Figure 2). A previous study suggested that R. etli can degrade trehalose 53 . Therefore, we also looked for genes involved in uptake and degradation of trehalose. We did not find a trehalose phosphotransferase system (PTS) for the uptake of trehalose, but two putative ABC uptake systems homologous to those operating in S. meliloti 22 23 were found that might be involved in the uptake of trehalose, sucrose, and/or maltose. These were encoded in plasmid p42f (ThuEFGK), and the chromosome (AglEFGK). Regarding trehalose degradation, neither E. coli treA- or treF- like genes for periplasmic or cytoplasmic trehalases, respectively, nor genes belonging to glycoside hydrolase family 15 trehalases 16 17 , were found in the R. etli genome. However, orthologs to the thuAB genes, which encode the major pathway for trehalose catabolism in S. meliloti 21 , were found in the chromosome and plasmid p42f. In addition, three copies of treC, encoding putative trehalose-6-phosphate hydrolases, were identified in the chromosome. All three TreC proteins belonged to the family 13 of glycoside hydrolases 16 , but they did not cluster together (see the phylogenetic tree in Additional file 2: Figure S1B). The metabolism of trehalose in R. etli inferred from its genome sequence is summarized in Figure 2.

As two copies of OtsA (OtsAch and OtsAa, Figure 3A) were encoded by the R. etli genome, we investigated their phylogenetic relationship. First we aligned the amino acid sequences of both R. etli OtsA proteins with the sequences of characterized trehalose-6-P- synthases, and compared motifs involved in enzyme activity. All residues corresponding to the active site determined in the best studied E. coli trehalose-6-P synthase 54 were conserved in R. etli OtsAch and OtsAa (data not shown). However, the identity between both proteins was only of 48%, and the gene otsAa was flanked by putative insertion sequences in the R. etli genome. In addition, the otsAch copy and R. etli genome had a similar codon use, whereas the otsAa copy showed a different preference for Stop codon, and codons for amino acids as Ala, Arg, Gln, Ile,Leu, Phe, Ser, Thr, and Val. These findings suggested that otsAa might have been acquired by horizontal transfer. To check this hypothesis, we constructed a phylogenetic tree with sequences of proteobacterial OtsA proteins available in the data bases, including OtsAs from R. etli and other rhizobia like R. leguminosarum bv. trifolii 7 , S.meliloti 5 , and B. japonicum 2 . As shown in Figure 3B, the resulting phylogenetic tree showed four separated branches, with a generally homogeneous distribution of phylogenetic groups. The first branch was formed by OtsA proteins from β- and γ -proteobacteria, including OtsA from E. coli and Salmonella enterica. The second cluster was mainly composed of OtsA proteins from γ-proteobacteria, including some halophilic representatives such as C. salexigens and Halorhodospira halophila. The third branch grouped OtsAs from α-proteobacteria, including the two R. etli OtsA proteins. Whereas R. etli OtsAch grouped with OtsAs from R. leguminosarum, S. meliloti, Rhizobium sp. NGR234 and Agrobacterium vitis, R. etli OtsAa constituted a separated sub-group within the α-proteobacterial branch. The fourth branch was composed by OtsA proteins from δ-protobacteria. Some incongruences were found, as B. japonicum and Mesorhizobium proteins did not clustered with OtsA proteins from other rhizobia. In summary, this phylogenetic analysis supports the hypothesis that otsAa was transferred to R. etli or its ancestor from a related α-proteobacteria, which did not belong to the Rhizobium/Agrobacterium group.

From the above phylogenetic analysis, otsAch was chosen as the most promising candidate to encode a functional trehalose-6-P-synthase. To check this, the corresponding mutant (strain CMS310) was constructed by insertion of an omega cassette within otsAch (Figure 3A), followed by double recombination in the chromosome of the wild-type strain. Then, we used natural abundance ¹³C-NMR to determine the compatible solute pool of wild type and otsAch cells grown at 28°C in 0.2 M NaCl B^- medium up to early-stationary phase. As shown in Figure 4A, the ¹³C-NMR spectrum of R. etli wild-type strain contained three sets of chemical shifts, which were assigned to trehalose (61.2, 70.4, 71.7, 72.8, 73.2 and 93.9 ppm), mannitol (63.9, 70.0 and 71.6 ppm) and glutamate (27.6, 34.2, 55.4, 175.2 and 181.9 ppm). Although¹³C-NMR is only a semi-quantitative technique, it was evident that trehalose levels were much higher than those of mannitol and glutamate, suggesting that trehalose is the major compatible solute of R. etli under these conditions. Mannitol was absent when glucose was used as a sole carbon source (data not shown), indicating that it was accumulated by R. etli after its uptake from the external medium. Chemical shifts corresponding to trehalose were not present in the spectrum of the R. etli otsAch strain, where only signals corresponding to mannitol were detected (Figure 4B). From these results, we conclude that the product encoded by otsAch is involved in trehalose synthesis in R. etli. Moreover, at least under the conditions tested, the otsAa copy does not seem to be functional.

We investigated the effect of a mutation in otsAch on R. etli heat tolerance. For this purpose, we compared the growth of wild-type and otsAch strains in minimal medium B^- under different combinations of osmotic (0.0 M to 0.2 M) and heat (28°C or 35°C) stresses. As previously described (see Figure 1), at optimal temperature (28°C), the wild-type strain grew optimally without NaCl added. At higher salinities (0.1 to 0.2 M NaCl), wild-type cells showed a delayed growth, but eventually they reached a stationary phase with absorbance values comparable to those of cultures without NaCl (Figure 5A).

Interestingly, otsAch cells grown at 28°C with mannitol showed a biphasic pattern, which was independent of the NaCl concentration, although cells exposed to osmotic stress were more affected than cells grown without NaCl. The first stage consisted of an attenuated exponential phase of 20 h (if compared with that of the wild type) followed by 30 h of arrested growth with OD₆₀₀ values of around 0.5 units. In the second one, growth was restarted, showing a second exponential phase during 40 h, followed by a second stationary phase with absorbance values comparable to those of the wild type strain (Figure 5A). As in otsAch cells collected at the beginning of the first stationary phase (see Figure 4B), trehalose was absent from extracts prepared from samples harvested at the entrance of this second stationary phase. Instead, they contained large amounts of glutamate (Figure 4C). However, when glucose and trehalose were used as the sole carbon source, this biphasic pattern of growth was not observed. Growth of the otsAch strain with both carbon sources was delayed with respect to the wild-type strain, even in the absence of osmotic stress (see Additional file 3: Figure S2).

At 35°C, R. etli wild-type strain was able to grow well in B^- medium with NaCl concentrations ranging from 0 to 0.15 M. As described above (see Figure 1), growth of the wild type was impaired at 35°C with 0.2 M NaCl, showing absorbance values not exceeding 1.0 unit of OD₆₀₀(Figure 5B). At this temperature, growth of the otsA mutant was severely affected, regardless of the salinity of the culture medium, with cultures showing OD₆₀₀ around 0.5 OD units. The above data suggest that trehalose is essential for growth of R. etli at high temperature.

Involvement of trehalose in desiccation tolerance in rhizobia has been firmly established in R. leguminosarum bv. trifolii 7 . On the other hand, in S. meliloti 55 or rhizobia nodulating Acacia 56 , desiccation tolerance was stimulated by osmotic and/or temperature pre-treatment. To check the influence of trehalose on desiccation tolerance of R. etli, wild type and otsAch strains were grown at 28°C in minimal medium B^-alone or additioned with 0.2 M NaCl, and harvested at early stationary phase. For cell drying, we used two variants of the protocol described by Manzanera et al. for E. coli 39 , a drying process (induced by vacuum at 30°C) or a drying + high temperature process (including a second step with a controlled increase of temperature from 20 to 30°C under vacuum).

In the absence of osmotic stress, both wild type and otsAch strains showed survival levels under 0.01%, regardless of the drying protocol (data not shown). In contrast, wild type cells osmotically pre-conditioned by the presence of 0.2 M NaCl showed ca. 35% survival levels after drying, although viability after 4 days storage dropped down to 1.4% (Figure 6). Compared to the drying treatment, the drying + high temperature protocol did not enhance wild type cell survival (Figure 6). The same drying experiments were performed with trehalose-deficient (otsAch) cells harvested at early stationary phase. Regardless of the protocol used, the otsAch strain showed ca 3-fold lower survival levels than the wild type strain after the drying process, and a null viability after 4 days storage. These findings suggested (i) a beneficial effect of osmotic stress in R. etli tolerance to desiccation, and (ii) a role of trehalose on desiccation tolerance in R. etli.

To analyze if the otsAch mutation modifies the capacity of R. etli to fix nitrogen in symbiosis, common bean plants were inoculated with R. etli wild-type and the otsAch strain. After inoculation, plants were grown under optimal (control plants) or water deficit conditions and were evaluated for nodulation, plant dry weight, total nitrogen content, nitrogenase activity, and leghaemoglobin content of the nodules. Plant water status during the different treatments was monitored by measuring water potential (Ψ_w) of the first fully expanded leaf. Water potential in plants subjected to drought stress by holding irrigation for 5 days reached values of about −1 ± 0.25 MPa (moderate drought). When irrigation was stop for 10 days, leaf Ψ_w reached values of about −2 ± 0.3 MPa (severe drought). The control plants maintained a leaf Ψw of −1 ± 0.4 MPa. The effect of either moderate or severe drought stress in leaf Ψw of plants inoculated with the otsAch mutant was similar to that of plants inoculated with the wild type (data not shown).

Independently of the plant treatment, no significant differences were observed in nodulation, plant growth parameters, and nitrogen fixation parameters among plants inoculated with any of the strains (Table 2). A moderate drought did not affect nodules number (NN), nodule dry weight (NDW), plant dry weight (PDW), and total nitrogen content (TN) of plants inoculated with either the wild-type or the otsAch strain (Table 2). Specific nitrogenase activity expressed as acetylene reduction activity (ARA) and leghaemoglobin (Lb) content of the nodules as an estimation of nodule functionality were also measured. Regardless of the plant treatment, inoculation of plants with the otsAch mutant did not affect significantly ARA or Lb content compared to those plants inoculated with the wild-type strain (Table 2).

Nodules number (NN), nodule dry weight [NDW, (mg plant^-1)], plant dry weight [PDW, (g plant^-1)], total nitrogen content [TN, (mg plant^-1)], acetylene reduction activity [ARA, (μmol C₂H₄ h^-1 g^-1 NDW)],leghaemoglobin [Lb, (mg Lb g^-1 NDW)], and trehalose (Tre) in bacteroids (B) and nodule cytosol (C) [μmol gDW^-1] content in nodules and plants subjected or not (control) to moderate or severe drought conditions. Values in a column followed by the same lower-case letter are not significantly different as determined by the Tukey HSD test at P ≤ 0.05 (n = 9).

ND. Not detected.

As shown in Table 2, NN and NDW per plant was negatively affected by a severe drought since a decrease of about 45% and 53% in those parameters was observed in plants inoculated with the wild-type strain compared to control plants. A similar decrease of NN (43%) and NDW (49%) was observed in plants subjected to a severe stress and inoculated with the otsAch mutant compared to control plants (Table 2). After a severe drought, a 53% and 49% reduction of PDW was observed in plants inoculated with the wild-type or the otsAch mutant, respectively. Plants inoculated with any of the strains and subjected to severe drought showed a similar reduction of about 30% in TN compared to control plants (Table 2). Plants inoculated with the wild-type strain and subjected to severe drought showed an inhibition of ARA of about 36% compared to control plants. This activity was similarly dropped in nodules produced by the otsAch mutant under severe drought (41% compared to control plants) (Table 2). A severe drought provoked a significant decline in Lb content of about 35% in plants inoculated with the wild-type strain compared to control plants Likewise, this parameter was also reduced of about 39% in plants inoculated with the otsAch mutant and subjected to a severe drought (Table 2). Finally, trehalose content in bacteroids of the wild type and otsAch strains was similar, regardless of the treatment, suggesting that under symbiotic conditions (i.e. with other trehalose precursors available) other trehalose synthesis genes (i.e. TreS or TreYZ) may be operating. Trehalose was not detected in the cytosol of nodules induced by either the wild type or the otsAch strain under any condition tested, suggesting that trehalose in the R. etli-induced nodules is synthesized only by the microsymbiont.