Putative Porin of Bradyrhizobium sp. (Lupinus) Bacteroids Induced by Glyphosate

Abstract

Application of glyphosate (N-[phosphonomethyl] glycine) to Bradyrhizobium sp. (Lupinus)-nodulated lupin plants caused modifications in the protein pattern of bacteroids. The most significant change was the presence of a 44-kDa polypeptide in bacteroids from plants treated with the higher doses of glyphosate employed (5 and 10 mM). The polypeptide has been characterized by the amino acid sequencing of its N terminus and the isolation and nucleic acid sequencing of its encoding gene. It is putatively encoded by a single gene, and the protein has been identified as a putative porin. Protein modeling revealed the existence of several domains sharing similarity to different porins, such as a transmembrane beta-barrel. The protein has been designated BLpp, for Bradyrhizobium sp. (Lupinus) putative porin, and would be the first porin described in Bradyrhizobium sp. (Lupinus). In addition, a putative conserved domain of porins has been identified which consists of 87 amino acids, located in the BLpp sequence 30 amino acids downstream of the N-terminal region. In bacteroids, mRNA of the BLpp gene shows a basal constitutive expression that increases under glyphosate treatment, and the expression of the gene is seemingly regulated at the transcriptional level. By contrast, in free-living bacteria glyphosate treatment leads to an inhibition of BLpp mRNA accumulation, indicating a different effect of glyphosate on BLpp gene expression in bacteroids and free-living bacteria. The possible role of BLpp in a metabolite interchange between Bradyrhizobium and lupin is discussed.

The legume-Rhizobium symbiosis is a beneficial plant-microorganism interaction with high agronomical importance due to its ability to fix atmospheric nitrogen. It is characterized by the formation of root nodules on the host plant, a complex process regulated by both plant and bacteria (15, 24). Once bacteria are inside the root cells, they are surrounded by a peribacteroidal membrane (PBM) which initially originates from the plant plasma membrane (40, 43). Rhizobia and PBM form an organelle called the symbiosome, the nitrogen fixing unit. In the symbiosome, rhizobia are transformed into bacteroids (differentiated bacteria), which perform nitrogen fixation, thereby providing ammonium to the plant and obtaining sources of carbon and energy from the host.

The efficiency of nitrogen fixation depends upon the plant-bacteria combination. It has been previously shown that the symbiosis established between Lupinus albus plants and the slow-growing soil bacteria Bradyrhizobium sp. (Lupinus) is very effective (4). Lupin plants have a high seed protein value, a positive effect on soil fertility (39), and a tolerance to different abiotic stresses (13, 18).

The rhizobia-legume symbiosis is very sensitive to herbicides, since these compounds interfere directly with plant metabolism, and, indirectly, they exert a negative effect on the microsymbionts. In addition, they have adverse effects on rhizobia in the rhizosphere, thus disturbing the formation of nodules (44). Among herbicides, glyphosate (N-[phosphonomethyl] glycine) is one of the most frequently employed in agricultural practices, due to its immobility, quick inactivation, and degradation in soil (27). Also, several species of soil bacteria can metabolize glyphosate, including members of the Rhizobiaceae family (25, 31).

Glyphosate disrupts the biosynthesis of aromatic amino acids in the plant (1), inhibiting the shikimate pathway by competing with the substrate phosphoenolpyruvate for a binding site on the 5-enolpyruvylshikimato-3 phosphate synthase enzyme (EC 2.5.1.19). This enzyme is present in all plants, bacteria, and fungi but not in animals. In addition to causing a deficit in aromatic amino acids (phenylalanine, tyrosine, and tryptophan), glyphosate also interferes with the biosynthesis of secondary products derived from these aromatic amino acids. This affects the biosynthesis of proteins, phytoalexins, folic acid, cinnamic acids, lignin precursors, flavonoids, plastoquinone, and other phenolic compounds (11).

Previous studies of the effect of glyphosate treatment on the lupin-Bradyrhizobium sp. (Lupinus) system demonstrated early senescence and leaf necrosis after 24 h of herbicide application at concentrations as low as 2.5 mM (9). Moreover, glyphosate altered carbon and phenolic metabolism and caused an adverse effect on symbiosome ultrastructure, depending upon the time of exposure and concentration applied (9). In addition, it was observed that not only was nitrogenase activity decreased rapidly in Bradyrhizobium sp. (Lupinus) bacteroids but also the amount of the two nitrogenase components (Mo-Fe protein and Fe protein) diminished after glyphosate application (10).

Following on from the observation that protein content in lupin leaves and either nodules or bacteroids decreased as the glyphosate concentration increased (9), in this work we focused our attention on the study and characterization of a 44-kDa polypeptide which accumulated in the soluble fraction of Bradyrhizobium sp. (Lupinus) bacteroids at the highest doses of herbicide. Sequence analysis of this polypeptide showed significant homology to bacterial porins.

These proteins are located in the outer membrane (OM) of gram-negative bacteria and a group of gram-positive bacteria (6, 32), as well as in the membranes of different organelles such as mitochondria, peroxisome, and chloroplasts. They form long watery channels, allowing the diffusion of specific compounds such as mono- and disaccharides (LamB and ScrY), nucleosides (TsX), and ions (21) (specific porins) or hydrophilic molecules (<600 Da) (general porins) (30).

The expression and/or activity of these OM proteins is affected by external environmental conditions such as water availability, pH, pCa (− log [Ca²⁺]), osmotic pressure, oxidative stress, heavy metals, temperature, shortage of nutrients, anoxia, and salinity (3, 17, 30, 37). This suggests a possible role for the Bradyrhizobium sp. (Lupinus) 44-kDa polypeptide in the interaction between the bacteroids and the surrounding medium.

MATERIALS AND METHODS

Bacterial and plant culture.

Bradyrhizobium sp. (Lupinus) strain ISLU16 was grown at 28°C with orbital shaking in either Vincent (41) or tryptone-yeast extract medium at pH 6.8. Escherichia coli strains DH5α and SURE were grown in Luria-Bertani culture medium.

Lupinus albus L. cv. Multolupa seeds were surface sterilized in 0.1% HgCl₂ for 5 min and then washed several times with sterile distilled water. Seeds were sown in pots containing vermiculite and watered with nitrogen-free nutrient solution (23). Plants were grown in a growth chamber at 25/15°C (day/night) temperature and 60 to 70% relative humidity, with a 16-h photoperiod and a quantum irradiance of 190 μE m⁻² s⁻². The plants were inoculated twice (first at sowing and then 1 week later) with a 1 ml of a suspension of ISLU16 (10⁸ CFU) grown in Vincent medium (41).

Glyphosate treatment.

For free-living bacteria, glyphosate (Roundup; Monsanto) was added to cultures of Bradyrhizobium sp. (Lupinus) ISLU16 at concentrations of 31.25 and 62.5 μM.

For plants, the aerial part was sprayed with 2 ml of glyphosate in water using half of the field-recommended dose (10 mM), as well as lower doses (5, 2.5, and 1.25 mM), 5 weeks after sowing. Nodules were detached 3, 5, and 7 days after herbicide application, immediately frozen in liquid nitrogen, and stored at −80°C.

Isolation of bacteroids and protein extraction.

Bacteroids were isolated by centrifugation from 1 g of nodules as previously described (8), with some modifications (10). The resuspended bacteroid pellet was sonicated in an ice-salt bath with 25 pulses of 40 s at 50% intensity using a Vibrate-Cell model VC-375. Bacteroid breaking was verified by optic microscopy, after which the sample was centrifuged at 15,000 × g for 15 min at 4°C. Protein content in the supernatant was estimated by the Bradford assay (Bio-Rad), using bovine serum albumin as a standard. Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (22) using denaturing 12.5% polyacrylamide gels. Electroblotting of proteins to polyvinylidene difluoride membranes was carried out for 2 h at 20 V in a semidry system (Bio-Rad). The polyvinylidene difluoride membranes were washed in methanol for 5 min, followed by staining with Ponceau red for 3 min, which was halted with 50% methanol. After the stained protein band was cut out, the sequencing of its N termini was performed directly onto the membranes using the standard Edman degradation techniques in an Applied Biosystems Procise-494 protein sequencer.

Genomic DNA isolation and Southern analysis.

Bacterial DNA was isolated by phenol:chloroform:isoamyl alcohol extraction as described previously (2). The genomic DNA (15 μg) was digested with restriction enzymes BamHI, EcoRI, and HindIII (Amersham Pharmacia Biotech) and separated on 1.2% agarose gels. After electrophoresis, the DNA was transferred onto Hybond-N⁺ nylon membranes (Amersham Pharmacia Biotech) and hybridized with digoxigenin (DIG)-labeled RNA probe specific for the Bradyrhizobium sp. (Lupinus) putative porin (BLpp) (35). RNA probes were prepared by transcribing with SP6 RNA polymerase SacII-linearized plasmid Pbnm, which contained a 523-nucleotide (nt) DNA fragment corresponding to nt 193 to 716 of the BLpp gene.

Blots were washed and incubated with anti-DIG antibody, and immunodetections were performed according to the DIG Luminescent Detection kit instructions (Roche Diagnostics).

PCR amplification of genomic DNA.

To clone the gene encoding the BLpp protein, a Universal Genome Walker kit (Clontech) was used. Genomic DNA aliquots (0.1 μg) were each digested with restriction endonucleases (DraI, EcoRV, PvuII, ScaI, and StuI), DNA fragments were ligated to the Genome Walker Adaptor, according to the manufacturer's instructions. The nucleotide sequence of the specific primer n1 (nucleotide positions 120 to 145, 5′-TACGGCGCCGGCTTCTACTACATCC-3′) was designed based upon the N-terminal 20-amino-acid sequence of the 44-kDa polypeptide. Gene-specific primers n2 (nucleotide positions 273 to 298, 5′-TAAGGTCGACACGGGATTGAGCAGT- 3′), n3 (nucleotide positions 533 to 560, 5′-ATGGGCAAACTATCCGGCGAACAGCTT-3′), and n4 (nucleotide positions 1499 to 1523, 5′-TACGTAATGCTCGACCAGAAGTATG-3′) were designed after nucleotide sequence analysis of the different DNA amplified fragments. Primers (n1 to n4) were used for PCR-based DNA walking in the five libraries following the protocols supplied by the manufacturer. The resulting PCR products were electrophoresed in 0.8% agarose gels and stained with ethidium bromide. Gel-eluted DNA bands were cloned using the pGEM-T Easy vector (Promega) for E. coli DH5α and SURE competent cell transformation.

RNA isolation and Northern analysis.

Total RNA from nodules (100 to 300 mg) was isolated by phenol extraction and LiCl precipitation as described previously (16), and extraction of RNA from free-living bacteria was performed by using an RNeasy Kit (QIAGEN).

Electrophoresis of total RNA from nodules and bacteria was performed under denaturing conditions (35). Samples (5 μg) were denatured and separated on 1.5% agarose-formaldehyde gels at 7 V/cm for 2 h. RNA was transferred to nylon membranes (Hybond-N⁺; Amersham Pharmacia Biotech) and hybridized with the specific DIG-labeled BLpp RNA probe as described previously (35). Immunodetection of the bound probe was performed according to instructions for the DIG Luminescent Detection kit (Roche Diagnostics).

Sequencing analysis and model building procedure.

DNA sequencing was performed in an ABI PRISM 377 (Applied Biosystems/Perkin-Elmer) sequencer. Nucleotide and amino acid sequences were analyzed using bioinformatics services and the available sequence databases publicly provided by NCBI (National Center for Biotechnology Information) and European Bioinformatics Institute. Homology sequence analysis of BLpp conserved domains was carried out by the BLAST program at the NCBI server (http://www.ncbi.nih.gov/BLAST). The recognition of an overall fold for the BLpp protein was based on the resultant threading predictions from the 3D-PSSM (20) and Libellula (19) programs.

The multiple sequence alignment (Clustal W, version 1.82) of the BLpp conserved domain with the sequences from published porin proteins was used for phylogenetic tree reconstruction. The phylogenic analysis was performed according to the neighbor-joining method (34).

RESULTS

Effects of glyphosate on the protein pattern of lupin nodule bacteroids and free-living Bradyrhizobium sp. (Lupinus) ISLU16 bacteria.

To gain insight into the changes that glyphosate treatment could induce in the protein metabolism of bacteroids isolated from glyphosate-treated lupin nodules, the bacteroid-soluble protein pattern was analyzed by SDS-PAGE electrophoresis after 3, 5, and 7 days of treatment. As shown in Fig. 1, the most significant change was the detection, from 5 days onward, of a protein band of ∼44 kDa in bacteroidal extracts from plants treated with the herbicide at doses of 2.5, 5, and 10 mM, with a remarkable increase between 2.5 and 5 mM. By contrast, the band could not be visualized in bacteroid extracts from control, untreated plants or from those with a 1.25 mM dose of herbicide. To determine whether this change was also observed in free-living bacteria, the soluble protein extracts from bacteria grown in the presence and absence of glyphosate were analyzed. For this assay, the doses of herbicide used were 31.25 and 62.5 μM. These levels were chosen as the growth of bacteria in culture medium is adversely affected in the presence of glyphosate, such that doses higher than 62.5 μM fully inhibited bacterial growth (9). In this analysis, the ∼44-kDa band was undetected in the protein pattern from free-living bacteria grown in the presence or absence of glyphosate, and there were no appreciable changes between the extracts (Fig. 1).

FIG. 1.

SDS-PAGE of soluble proteins from bacteroids of Bradyrhizobium sp. (Lupinus) ISLU16 (B) and free-living bacteria (FLB) of control (lanes 1 and 6, respectively) and after 7 days of glyphosate treatment at concentrations of 1.25 mM (lane 2), 2.5 mM (lane 3), 5 mM (lane 4), 10 mM (lane 5), and 31.25 μM (lane 7). A total of 35 μg of protein was loaded per lane. Numbers on the left indicate molecular size markers in kilodaltons.

Characterization of the ∼44-kDa polypeptide (BLpp).

In order to characterize the ∼44-kDa polypeptide, the sequence of 20 amino acids at its NH₂ terminus was determined by the Edman degradation method. The established amino acid sequence was identical to sequences from two proteins from Bradyrhizobium japonicum (USDA110 strain), corresponding to a putative porin (Bjpp) (NP_773528) and to a hypothetical protein of unknown function (NP_771716). It also shared sequence similarity with other bacterial proteins, also described as porins, such as those from Rhodopseudomonas palustris and Mesorhizobium loti. Based upon these analyses, the protein was named BLpp.

To isolate the gene(s) encoding the protein, the genomic DNA libraries constructed with the Universal Genome Walker kit were amplified by PCR, using adaptor (outer or nested) primers and gene-specific primers. By using BLpp-specific primer n1, a DNA fragment of 523 nt was amplified from the StuI library, corresponding to nt 193 to 716 of the BLpp gene. A 5′ overlapping 468-nt fragment and a 3′ overlapping 981-nt fragment were isolated from each of the PvuII and EcoRV libraries, using BLpp-specific primers n2 and n3, respectively. Another DNA fragment of 969 nt was obtained from the StuI library by using BLpp-specific primer n4. The nucleotide sequence analysis of the 2,467-nt fragment contained in the four overlapping cloned DNA fragments showed that it harbored an open reading frame of 1,521 bp encoding a protein of 506 amino acids, with a calculated molecular mass of 52,922 Da. A possible ribosome binding site (GGAGG) was identified 6 nt upstream of the predicted translation start codon. Analysis of the nucleotide sequence showed that the BLpp gene shared a 66% identity with the gene bll5076 (gene identifier 1051796) encoding a hypothetical protein (NP_771716) of B. japonicum USDA110 (BA000040). The BLpp gene also shared identities of 60% with R. palustris (CGA009) locus tag RPA3423 (gene identifier 2691513), and 59% with segment 1/4 of M. loti R7A symbiosis island (MLO672112) and the gene mll4029 (BA000012.4), encoding Omp2b porin from M. loti MAFF303099.

The analysis of the deduced amino acid sequence of the encoded polypeptide predicted the existence of a signal peptide with the most likely cleavage site between positions 22 and 23 at the 3-amino-acid consensus peptidase cleavage site (AQA-AD). However, the experimental data obtained from direct sequencing of the protein showed that the N-terminal residue corresponded to amino acid position 34, thus indicating that processing of the polypeptide takes place between amino acid positions 33 and 34, where a putative glutamyl endopeptidase cleavage site was predicted by the Expasy Peptide Cutter server. Cleavage of the putative signal peptide and processing of the protein would result in a mature polypeptide of 473 amino acids with a predicted molecular mass of 49,638 Da. This value is in good agreement with the 44 kDa, the molecular mass of the BLpp polypeptide, as empirically determined by SDS-PAGE analysis. The processed polypeptide would have a theoretical pI of 6.56, and its most likely subcellular localization would be the bacterial OM, according to Psort prediction.

Analysis of the BLpp deduced amino acid sequence showed a high degree of identity with three hypothetical proteins of B. japonicum: 61% with the Bll4983 protein, called Bjpp (NP_771623); 60% with Bll5076 protein, Bjhp (NP_771716); and 59% with Bll6888, a protein described as putative porin, Bjpp2 (NP_773528) (Fig. 2). Moreover, BLpp also shared a high degree of identity with either putative porins or OM proteins from other bacterial species, showing identity values of 46% with the putative Omp2b porin (NP_945891) from R. palustris, 43% with an Omp2b porin from Brucella melitensis bv. abortus (NP_AAF80104), 37% with another porin (NP_107994), 29% with an Omp2b porin (BAB50784) from M. loti, and 30% with Omp2a and Omp2b porins (AAG15256 and AF300817, respectively) from Brucella cetaceae. It also shared 30% identity with a putative OM protein (NP_385162) from Sinorhizobium meliloti and 42% identity with an OM protein (NP_531717) from Agrobacterium tumefaciens.

FIG. 2.

Alignment of the predicted amino acid sequence of BLpp with three proteins with a high degree of identity. Conserved residues in proteins are shadowed in black, conservation of strong groups is shown in dark gray, and conservation of weak groups is shown in light gray. Amino acid numbers are shown at left. Underlining indicates the N-terminal amino acid sequence of BLpp.

From the multiple alignments of the predicted amino acid sequences from these proteins, a putative conserved domain of porins was identified. It consisted of 87 amino acids, located in the BLpp sequence 30 amino acids downstream from the N-terminal region. Comparisons of this domain among the amino acid sequences of BLpp and other putative porins indicated that BLpp shared 70% of identity with Bjpp, 71% with Bjpp2, and 68% with Bjhp. This reveals a high degree of amino acid sequence conservation of this domain in the porin family.

Phylogenetic analysis with the neighbor-joining method of this putative conserved domain among different porins indicated the existence of several clusters (Fig. 3). BLpp (AY452712) is grouped in a higher-level cluster with B. japonicum and R. palustris porins next to rhizobia and far away from porins of animal pathogens such as Brucella or Bordetella.

FIG. 3.

A neighbor-joining phylogenetic tree reconstructed from the multiple sequence analysis of a putative conserved domain of porins in different bacteria. Agroba-t, A. tumefaciens; Barton-h, Bartonella henselae; Bradyr-j, B. japonicum; Brucel-B14/94, Brucella sp. strain B14/94; Brucel-m, Brucella melitensis; Mesorh-l, M. loti; Rhodo-p, R. palustris; Sinorh-m, S. meliloti. Values for nodes with bootstrap support above 70% are indicated.

The overall fold of the protein could not be established since its overall sequence identity with any protein of experimentally determined three-dimensional structure is lower than 30%. However, the most reliable fold yielded by the threading methods used was a transmembrane beta-barrel corresponding to chain P of sucrose-specific porin (Protein Data Bank code 1A0T) from Salmonella enterica serovar Typhimurium, an OM protein from enterobacteria which belongs to the porin superfamily. According to the predicted secondary structure, BLpp protein would be basically a β-strand with α-helices encompassing residues 3 to 17 at the initiation of the signal peptide and at residues 392 to 418 (Fig. 4).

FIG. 4.

Predicted secondary structure of BLpp aligned to 1A0T. H, α-helix structure (arrow); E, β-strand (cylinder); C, coil; BLpp_PS, predicted secondary structure of BLpp; 1A0T, sequence of chain P sucrose-specific porin; 1A0T_SS, known secondary structure of chain P sucrose-specific porin. The panel at bottom shows top (left) and side (right) views of the 1A0T protein. The protein backbone is represented by ribbons. N and C indicate the amino- and carboxy-terminal ends.

Southern blot analysis of Bradyrhizobium sp. (Lupinus) DNA was performed by using the plasmid Pbnm as a probe for the bacterial DNA digested with each of the restriction enzymes BamHI, EcoRI, and HindIII. No restriction sites for these enzymes were present in the probe. In each BamHI-, EcoRI-, and HindIII-digested DNA, only one hybridization band of approximately 6,108 bp, 22,740 bp, and 21,226 bp, respectively, was observed, indicating that BLpp is encoded by a single gene in the bacterial genome (Fig. 5).

FIG. 5.

Southern blot analysis of Bradyrhizobium sp. (Lupinus) with a Pbnm RNA probe. Fifteen micrograms of genomic DNA was digested with the restriction enzymes (BamHI, EcoRI, and HindIII) and loaded on each lane.

Expression patterns of BLpp in Bradyrhizobium sp. (Lupinus) free-living bacteria and bacteroids under glyphosate treatment.

In order to determine whether the expression of BLpp was transcriptionally regulated and to analyze the effect of the glyphosate treatment, Northern blot analysis was performed on total RNA extracted from both Bradyrhizobium sp. (Lupinus) free-living bacteria and bacteroids. The analysis revealed that untreated free-living bacteria and bacteroids showed basal constitutive expression of BLpp RNA, but different BLpp expression patterns were observed for free-living bacteria and bacteroids after glyphosate treatment (Fig. 6A and B). Thus, in free-living bacteria BLpp RNA was not detected after herbicide treatment (Fig. 6B). By contrast, in bacteroids, BLpp RNA increased with increasing concentrations of herbicide (Fig. 6A). The quantification of the relative BLpp RNA transcript accumulation level indicated a dose effect in BLpp RNA accumulation, increasing when plants were treated with increasing doses of herbicide, from 2.5 mM to 10 mM.

FIG. 6.

Expression of BLpp mRNA in bacteroids from nodules of Lupinus albus-Bradyrhizobium sp. ISLU16 (A) and free-living bacteria (B), grown in the presence of different doses of glyphosate. The lanes in panel A are as follows: lane 1, nodules from control plants; lanes 2, 3, and 4, nodules from plants treated with 2.5, 5 and 10 mM glyphosate, respectively. The lanes in panel B are as follows: lane 1, free-living ISLU16 grown without glyphosate; lanes 2 and 3, ISLU16 grown with 31.25 and 62.5 μM glyphosate, respectively. Total RNA (5 μg) was loaded on agarose-formaldehyde gels, blotted onto nylon membrane, and hybridized to BLpp-specific DIG-labeled RNA probes. The expected size for the BLpp mRNA was 1.5 kb. Ethidium bromide staining of RNA was used as a loading control.

DISCUSSION

Biological nitrogen fixation is a complex system that is strongly related to the physiological state of the host plant. Therefore, environmental stresses not only have a detrimental effect on metabolism, growth, and plant development but also affect the Rhizobium-legume symbiosis.

As assayed in the present work, glyphosate treatment of nodulated lupin plants produces some changes in the protein pattern of Bradyrhizobium sp. (Lupinus) ISLU16 bacteroids, in particular, the accumulation of a 44-kDa bacteroidal polypeptide that was not detected in bacteroids from nodules of untreated plants. By amino acid sequence analysis of the N-terminal region and its coding region, this polypeptide has been identified as a putative porin (BLpp) encoded by the bacterial DNA. Thus, the data revealed that the increase of the 44-kDa polypeptide was not due to either plant or PBM protein contamination of the bacteroid protein extracts. Therefore, this represents the first time that a change in polypeptide pattern has been detected in bacteroids from nodules of plants treated with herbicides.

Based upon the analysis of the nucleotide sequence of the BLpp gene and of its deduced amino acid sequence, BLpp shares a high degree of identity with a putative porin of B. japonicum (Bjpp) and with other proteins also described as porins from R. palustris and M. loti. Both species have a close phylogenetic relationship to Bradyrhizobium spp. according to their 16S rRNA gene sequences (7). BLpp also shares identity with porin proteins from other less related phylogenetic bacteria such as Brucella. Such homology is not surprising because rRNA and DNA hybridizations have shown the close phylogenetic origin of Brucella and bacteria of the Agrobacterium-Rhizobium complex (7), and all of them are included in the alpha-2 subdivision of the class Proteobacteria (5, 42). These bacteria are characterized by living in close association with animals or plants, either as pathogens or as symbionts (29).

The identification of a putative conserved domain of the porin superfamily near the N-terminal amino acid sequence of BLpp and the similarity of the predicted secondary structure of BLpp to the sucrose-specific porin ScrY from S. enterica serovar Typhimurium (12, 14) further support the hypothesis that this protein could belong to the porin superfamily. Several clusters have been found in the phylogenetic analysis of this conserved motif, with BLpp forming a cluster at the bottom of the tree, where putative porins from B. japonicum and Omp2b porins from R. palustris are the closest relatives (Fig. 3). This is in agreement with the close phylogenetic relationship between these bacterial species (36).

Computational analysis of the BLpp protein indicates that its likely subcellular localization is the bacteria OM. In spite of this, electrophoretic analysis showed that the protein was extracted in the soluble fraction of Bradyrhizobium sp. (Lupinus) bacteroids from nodulated lupin plants treated with the highest glyphosate doses utilized. This anomaly could be due to the major changes in bacteroidal membranes that take place as a consequence of herbicide treatment (9).

The analysis of BLpp gene expression shows basal constitutive expression in both free-living bacteria and bacteroids of Bradyrhizobium sp. (Lupinus) from lupin nodules grown without glyphosate treatment. However, herbicide treatment has a dual effect depending on the bacteria state, with glyphosate treatment inhibiting BLpp gene expression in free-living bacteria while up-regulating its transcription in bacteroids (Fig. 6). This differentiation suggests that the induction of BLpp gene expression was not directly regulated by glyphosate in Bradyrhizobium sp. (Lupinus) bacteroids but in response to herbicide-induced changes in plant symbiosome metabolism. Such a response would explain why BLpp gene expression is increased at the higher glyphosate doses, when cellular damage and adverse effects are also higher.

Regulation of BLpp porin expression could occur as a response to the presence of chemical compounds into the medium. High concentrations of glyphosate have a toxic effect over free-living Bradyrhizobium sp. (Lupinus) bacteria (9), and an increase in the accumulation of hydroxybenzoic acids has been described in B. japonicum bacteria when grown in glyphosate medium (28). In this sense, it is plausible to consider that its transcriptional regulation is similar to that of OmpF porin synthesis in E. coli and Serratia marcescens, which is diminished when salicylate is present in the medium (3, 33).

Previous work has shown that integrity of the PBM plays an important role in the maintenance of turgor in the nonvacuolated infected cells of lupin nodules (9). Glyphosate treatment is detrimental to the symbiosome membrane, and as a consequence of its disintegration or rupture, osmotic adjustment and transport inhibition between bacteroids and plant cytosol would take place, limiting availability of solutes. This rupture of metabolic exchange could be the cause of the increase in BLpp gene expression as an attempt to preserve nutrient transport toward the interior of the bacteroids, such as has been described for E. coli OmpF porin expression under glucose limitation (26). Therefore, if the peribacteroid membrane plays a strategic role in a symbiotic exchange through the bacteroidal membrane (38), BLpp protein could play an important role in controlling metabolite exchange between the plant and bacteroid in lupine nodules, which would indicate that BLpp gene differential expression is an adaptative response to stress.

This works represents the first description of a porin from Bradyrhizobium sp. (Lupinus), but several aspects of the structure and function of BLpp protein are still to be determined. Future investigations will be required to unravel these questions.