Introduction

Soil bacteria collectively called rhizobia are able to induce nitrogen-fixing symbiotic association with roots of leguminous plants. An effective symbiosis requires several bacterial genes, including nitrogen-fixation (nif) genes that encode nitrogenase responsible for N2 reduction and nodulation (nod) genes that encode Nod factors which trigger root hair deformations and cortical cell divisions (Debelle et al. 2001; Perret et al. 2000). Nod factors consist of β-1,4-linked N-acetyl glucosamine oligosaccharide backbone ranging in length from 3 to 5 residues that are substituted by an N-acyl chain on the non-reducing end and other chemical groups on the glucosamine residues. The backbone of the Nod factors is synthesized under control of the common nod genes, nodA (acyl transferase), nodB (deacetylase), and nodC (N-acetylglucosaminyl transferase), and next it is modified by different chemical groups (sulfate, fucosyl) that determine the specificity of rhizobium–legume interactions. The synthesis and transfer of chemical “decorations” to the backbone part of Nod factors is encoded by the host-specific nodulation (hsn) genes. In addition, the common nod genes contribute also to the bacterial host range. For example, NodA proteins recognize and transfer only specific fatty acids to the Nod core, the length of which determines NodC protein (Debelle et al. 2001; Downie 1998). Symbiotic nod and nif genes which are located on transmissible plasmids in Rhizobium, Sinorhizobium, and some Mesorhiozbium species or on genomic islands in Mesorhizobium loti and Bradyrhizobium japonicum may be transferred across divergent chromosomal lineages (González et al. 2003; Gottfert et al. 2001; Mierzwa et al. 2009; Nandasena et al. 2007; Sullivan and Ronson 1998; Xu and Murooka 1995; Zhang et al. 2008; Zou et al. 1997). It was shown for agricultural rhizobial populations (Laguerre et al. 1996; Sullivan et al. 1995) and concluded by discordance between the genealogies of nod and housekeeping loci (Laguerre et al. 2001; Suominen et al. 2001; Vinuesa et al. 2005a; Young and Haukka 1996). Such gene transfer may adopt rhizobial populations to new host plant and enable symbionts with the same genetic background but different nod genes to nodulate distinct legumes as it was found for R. leguminosarum forming symbiosis with Viciae, Trifolium, and Phaseoli species plants (Martínez et al. 1990). By contrast to nod genes, the phylogeny of nifH genes, which encode dinitrogenase reductase, was reported to be congruent with 16S rDNA phylogeny (Hennecke et al. 1985; Wdowiak-Wróbel and Małek 2010; Young 1992; Zhang et al. 2008); however, in some cases, it followed that of nodA genes (Haukka et al. 1998; Laguerre et al. 2001; Vinuesa et al. 2005a, b).

Symbiotic promiscuity of R. pseudoacacia was examined by Han et al. (Han et al. 2008), Ulrich and Zaspel (Ulrich and Zaspel 2000), Wei et al. (Wei et al. 2009) and Mierzwa et al. (Mierzwa et al. 2009). Those studies showed black locust to be nodulated by Rhizobium, Sinorhizobium, and Mesorhizobium genera bacteria, however, the latter ones clearly predominated in root nodules. Despite taxonomic diversity, R. pseudoacacia rhizobia from North America, Germany, and China shared very similar sequences of nodA, nodC, and nifH genes (Wei et al. 2009). According to Wei et al. (Wei et al. 2009), local rhizobia from China were adopted to symbiosis with black locust by lateral transfer of nod genes from R. pseudoacacia nodulators derived from North America where this plant is native.

In this paper, we present the phylogeny of symbiotic nodA, nodC, nodH, and nifH genes of R. pseudoacacia rhizobia derived from Poland and Japan and microscopic structure of black locust root nodules.

Materials and methods

Bacterial strains and growth conditions

For phylogenetic analysis of R. pseudoacacia rhizobium symbiotic genes, the following nodule isolates were chosen: RPP14, RPP20 (from Poland), RPJ3, RPJ5, RPJ6, and RPJ16 (from Japan; Mierzwa et al. 2009). Additionally, these strains and 16 other black locust rhizobia described in the paper of Mierzwa et al. (Mierzwa et al. 2009) were used in plant tests in order to determine their host range and the effectiveness of N2 fixation. Rhizobium strains were maintained and grown on yeast extract-mannitol medium (YEM) as described earlier (Mierzwa et al. 2009).

PCR amplification and sequencing of amplicons

For each R. pseudoacacia nodulator, total genomic DNA was isolated, as described previously (Mierzwa et al. 2009), and used as PCR template. PCRs were performed with the ReadyMixTMTaq kit according to the manufacturer’s recommendations (Sigma). The 660-bp fragment of nodA gene was amplified and sequenced using forward primer nodA-1 (5′-TGCRGTGGAARNTRNNCTGGGAAA-3′) and reverse one nodA-2 (5′-GGNCCGTCRTCRAAWGTCARGTA-3′) according to the procedure described by Haukka et al. (Haukka et al. 1998). The nodC gene sequences were amplified and sequenced with forward primer nodCFu (5′-AYGTHGTYGAYGACGGITC-3′) and reverse primer nodCI (5′-CGYGACAGCCANTCKCTATTG-3′) corresponding to the position 251–269 bp and 1160–1181 bp of nodC gene of R. leguminosarum bv. viciae, respectively. PCR was performed using the same cycling parameters as reported for this gene by Laguerre et al. (Laguerre et al. 2001). The 567-bp fragment of nodH gene was amplified and sequenced with forward primer TSnodH1 (5′-VTKGAGYAACGGTGARYTGCTCA-3′) and reverse one TSnodH2 (5′-GCGAAGTGAWSCCGCAACTC-3′) under the following conditions: preheating at 95°C for 2 min; 35 cycles of denaturing at 95°C for 45 s, annealing at 53°C for 30 s, and extension at 72°C for 2 min; and a final extension at 72°C for 7 min. A large ~780-bp long fragment of nifH gene was amplified and sequenced with primers NifH1 (5′-CGTTTTACGGCAAGGGCGG-3′) and NifH2 (5′-TCCTCCAGCTCCTCCATGGT-3′). nifH sequences were amplified following the procedure described by Perret and Broughton (1998). PCR products were purified by using Montage PCR Filter Units (Millipore, Massachusetts, USA) as recommended by the manufacturer. Sequencing reactions were performed with the ABI Prism BigDye Terminator Cycle sequence kit (Applied Biosystems model 310 DNA sequencer). Sequences were deposited in the GenBank database under accession numbers listed in the phylograms.

Sequence analyses

Multiple sequence alignments were generated using ClustalX (Thompson et al. 1997) and manually corrected using GeneDoc software (Nicholas and Nicholas 1997). Phylogenetic trees were generated by maximum likelihood (ML) approach using PhyML 3.0 and the nucleotide substitution model selected by the Akaike information criterion, as implemented in MODELTEST3.6 (Posada and Crondall 1998). Robustness of tree nods was estimated by bootstrap analysis with 100 replicates of each sequence using PhyML. The trees were displayed by using TreeView (Page 1996).

Plant tests

R. pseudoacacia microsymbionts were tested for symbiotic interactions with their original host plant (R. pseudoacacia), Trifolium repens, Lotus corniculatus, Medicago sativa, Ornithopus sativus, Phaseolus vulgaris, Glycine max, Astragalus glycyphyllos, Astragalus sinicus, Vicia sativa, Lupinus luteus, Amorpha fruticosa, and Amorpha californica. Seeds of legumes were surface-sterilized, germinated, and plants inoculated with bacterial suspension were grown in nitrogen-free Hoagland’s medium as described by Mierzwa et al. (Mierzwa et al. 2009). For each legume species–rhizobium interaction (ten strains from Poland and ten strains from Japan), three replicates were prepared. The symbiotic properties of R. pseudoacacia nodulators were assessed by the presence of nodules, shoot dry weight, and the acetylene-reducing activity (Turner and Gibson 1980). Non-inoculated plants were used as negative controls.

Light and electron microscopy

For the microscopy studies, a whole root nodules sampled 14, 28 days after inoculation (DAI), and nodule fragments sampled 91 DAI were fixed in a mixture of 2.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M Na-cacodylate buffer, pH 7.2, for 2 h under air pressure of −0.06 MPa at room temperature. Next, the samples were treated as described earlier (Kalita et al. 2006), embedded in glycid ether 100 epoxy resin (SERVA), and the resin was polymerized at 60°C for 24 h according to the manufacturer’s formula. Serial resin sections 2–4 μm thick (depending on the nodule size) and ultra-thin sections were cut and treated as described by Kalita et al. (Kalita et al. 2006). The anatomical observations and micrographs were done using light microscope Provis AX (Olympus) equipped with digital photo camera. The ultrastructure was observed under transmission electron microscope Morgagni 268 (FEI Company) operating at 80 kV. The images were adjusted using “Levels” or “Curves” tools in Photoshop 7.0 CE (Adobe) software.

Results

Phylogeny of symbiotic genes of R. pseudoacacia rhizobia

The R. pseudoacacia microsymbionts, two from Poland and four from Japan which represented three different phenogroups (Mierzwa et al. 2009), were studied for the phylogeny of their symbiotic loci by comparative analysis of nodulation nodA, nodC, nodH, and dinitrogenase reductase nifH genes. These genes have been widely used in studies of symbiosis genealogy in rhizobia. nodA PCR amplification with nodA-1, nodA-2 primers yielded a single band of about 600-bp. ML analysis of 437-bp nodA sequences of black locust rhizobia derived from Poland and Japan (this study) and China, Germany, and North America (present in the GenGank database), as well as other nodule bacteria, is shown in Fig. 1. The best-fit model selected by Akaike Information Criterion (AIC) was Tamura-Nei (TrN + I + G). All R. pseudoacacia rhizobia of nodA sequence identity 99–100% were grouped together with 100% bootstrap support (BS) and in bigger highly validated monophyletic cluster (98% BS) with R. giardini bv. giardini, R. gallicum bv. gallicum, R. tropici (symbionts of Phaseoli sp.), and M. amorphae (symbiont of Amorphae fruticosa) as a basal bifurcation. NodA sequence similarities of these bacteria and black locust rhizobia ranged from 73 to 79%. Other rhizobia, with 57–73% nodA sequence identities to R. pseudoacacia nodule isolates, formed four clearly separated lineages. On the outskirt of this tree, A. caulinodans (57% sequence similarity) was located.

Fig. 1
figure 1

Phylogenetic ML tree based on 437-bp nucleotide sequences of the nodA genes of R. pseudoacacia nodule isolates (shown in bold) and reference rhizobial strains. Only bootstrap probability values of ≥ 70% (for 100 pseudoreplicates) are indicated at the branching points. The scale bar indicates the number of substitutions per site. GenBank accession numbers are shown in parentheses

Full size image

Next, we concentrated on nodC gene evolutionary history. Phylogenetic analysis was carried out on 837-bp nodC fragment amplified and sequenced in all six tested R. pseudoacacia microsymbionts (data not presented) and on a shorter 419-bp segment when black locust rhizobia from Germany, China, and the USA (GenGank database) were included into studies (Fig. 2). ML approach to nodC gene phylogeny gave in both cases very similar tree topologies in TrN + I + G model selected by AIC. R. pseudoacacia rhizobia from Poland, Japan, China, Germany, and North America, with 97–98% nodC nucleotide identities, formed monophyletic branch together with M. amorphae, R. gallicum bv. gallicum, R. giardinii bv. giardinii, R. gallicum bv. phaseoli, and R. etli (76–82% sequence similarity) which splits in two well-supported major clusters. One of them contains nine black locust rhizobia and M. amorphae, the second one harbors bean endosymbionts (Fig. 2). At the basal position of this clade, Mesorhizobium sp. (Oxytropis) N33, M. septentrionale, and M. temperatum were situated. In respect to nodC sequences of the tested rhizobia, they showed differences in 17–21% nucleotides. Other rhizobia which displayed 55–77% nodC nucleotide sequence similarity to black locust isolates spread into five clearly resolved lineages with A. caulinodans as outlier (55% identical nucleotides).

Fig. 2
figure 2

ML phylogram based on 419-bp nucleotide sequences of the nodC genes of R. pseudoacacia nodule isolates (shown in bold) and reference rhizobial strains. Bootstrap values of ≥70% (for 100 pseudoreplicates) are indicated at the corresponding nodes. The scale bar indicates the number of substitutions per site. GenBank accession numbers are shown in parentheses

Full size image

To uncover the phylogeny of dinitrogen reductase of R. pseudoacacia microsymbionts, we amplified and sequenced the 689-bp fragments of nifH genes in all six tested rhizobia. Phylogenetic analysis was carried out on the whole this amplicon and its 469-bp segment when black locust rhizobia from China, Germany, and North America (GenGank database) were included into studies. Both nifH phylograms, obtained under the best-fit General Time Reversible model of sequence evolution (GTR + I + G) have a very similar topology. Only the tree for shorter nifH gene fragment is presented in Fig. 3. All R. pseudoacacia rhizobia from Europe, Asia, and North America with 97–100% nifH sequence identity to each other lied in the same clade in 98% of bootstrap replicates. They grouped with Mesorhizobium species (except M. loti) within well-resolved clade and shared with them 88–94% nifH sequence identity. Outside this clade, on the sister branch, M. loti along with R. tropici was positioned. nifH sequences from these bacteria and black locust endosymbionts revealed identity from 89 to 92%. Other rhizobia were placed in three clearly resolved clusters and one single strain lineage (R. etli). Bradyrhizobium species and A. caulinodans formed a highly validated, well-resolved cluster at the basal position of the corresponding tree. All these bacteria differed from the tested R. pseudoacacia microsymbionts at 29–10% nifH sequences.

Fig. 3
figure 3

Phylogenetic ML tree based on 469-bp nucleotide sequences of the nifH genes of R. pseudoacacia nodule isolates (shown in bold) and reference rhizobial strains. Only bootstrap probability values of ≥70% (for 100 pseudoreplicates) are indicated at the branching points. The scale bar indicates the number of substitutions per site. GenBank accession numbers are shown in parentheses

Full size image

In our studies on symbiotic genes of R. pseudoacacia microsymbionts, we also focused on the nodH host-specificity genes involved in transfer of activated sulfate from 3′-phosphoadenosine 5′-phosphosulfate (PAPS) to the Nod factor core structure (Debelle et al. 2001; Perret et al. 2000). Five hundred and sixty-five–base pair fragment of nodH gene was amplified and sequenced in all six tested strains. The nucleotide identity for nodH genes among black locust rhizobia was 97–100%. Phylogenetic analysis of nodH partial sequences with best-fit Hasegawa-Kishino-Yano model (HKY + G) of nucleotide substitution resulted in ML tree presented in Fig. 4. R. pseudoacacia nodule isolates from Poland and Japan clustered together at 100% bootstrap support and they grouped well with Mesorhizobium sp. (Oxytropis arctobia) N33 strain (80–81% sequence identity) and Sinorhizobium sp. BR816 (77% identity), suggesting that they might have been of monophyletic origin. The nucleotide sequence similarity of nodH R. pseudoacacia rhizobia, and Methylobacterium nodulans was 75–76%. Compared with the nodH sequences of Sinorhizobium meliloti strains and Rhizobium sp. LPU83, our isolates indicated a lower percentage of identical nucleotides, i.e., in the range from 73 to 74%. On the outskirts of nodH gene tree, R. tropici and Burkholderia phymatum were positioned with 68 and 66% sequence similarity, respectively.

Fig. 4
figure 4

ML phylogram based on 565-bp nucleotide sequences of the nodH genes of R. pseudoacacia nodule isolates (shown in bold) and reference rhizobial strains. Bootstrap values of ≥70% (for 100 pseudoreplicates) are indicated at the corresponding nodes. The scale bar indicates the number of substitutions per site. GenBank accession numbers are shown in parentheses

Full size image

Host plant associations

R. pseudoacacia microsymbionts derived from Poland and Japan were able to form nitrogen-fixing symbiosis not only with the host from which they were isolated but also with A. fruticosa and A. californica. None of our isolates nodulated: Trifolium repens, Lotus corniculatus, Medicago sativa, Ornithopus sativus, Phaseolus vulgaris, Glycine max, Vicia sativa, Lupinus luteus, Astragalus cicer, Astragalus glycyphyllos, and Astragalus sinicus (Mierzwa et al. 2009). After 6 weeks postinoculation, the dry matter of green parts of nodulated legumes was 2–3 times higher than that of uninoculated ones showing that the used bacteria fix N2 and supply it to the symbiotic partner (Table 1). Symbiotic interactions between the tested rhizobia and R. pseudoacacia, A. fruticosa, and A. californica were also evaluated for nitrogenase activity by the acetylene reduction test. The highest level of enzyme activity R. pseudoacacia rhizobia showed in association with their original host, and next in descending sequence with A. fruticosa and A. californica, although these bacteria were good diazotrophs in symbiosis with all three legumes (Table 1).

Table 1 Nitrogen-fixing activity of Robinia pseudoacacia rhizobia in symbiosis with different legumes
Full size table

Microscopic structure of R. pseudoacacia root nodules

The root nodules of R. pseudoacacia were initiated within deep layers of root primary cortex and became infected via an infection thread originating from the curled root hair. Up to 28–35 DAI, the root nodules were round with two strands of loose parenchyma on their surface (data not shown). Later, the nodules took an elongate shape (Fig. 5), and among 91 DAI nodules, bifurcated ones were found. Microscope observations proved the root nodules of R. pseudoacacia to be of indeterminate type, with persistent apical meristem, bacteroid tissue consisting of infected and uninfected cells and multilayered nodule cortex with a system of vascular bundles (Fig. 5). Within the bacteroid tissue, the following developmental zones were discernible: (1) infection thread penetration zone, (2) symbiosome multiplication zone, (3) differentiated zone, (4) senescent zone, and (5) saprophytic zone. The nodule meristem was narrow. Electron-dense precipitate or globules were observed in vacuoles. The organelles were relatively sparse and the plastids contained small starch grains. In 91 DAI nodules, the plastids within the nodule meristem, as well as the infection thread penetration zone, accumulated large amounts of ferritin (Fig. 6 and the insert). The infection threads, within the whole bacteroid tissue, contained a non-typical electron-empty layer between the infection thread wall and matrix. Starch accumulation was evident in the non-infected cells of differentiating bacteroid tissue, while in the infected cells, starch grains were small or not evident.

Fig. 5
figure 5

Anatomy of Robinia pseudoacacia root nodules, 91 DAI. The approximate boundaries of developmental zones are marked with thin arrows. NM nodule meristem, ITP infection thread penetration zone, DZ differentiating zone, DBT differentiated bacteroid tissue, SBT senescent bacteroid tissue, SAZ saprophytic (= saprotrophic) zone, OC outer cortex, IC inner cortex, white and black asterisks—infected and non-infected cells in the bacteroid tissue, respectively, arrowheads—cortical endodermis, double arrowheads—vascular bundle, arrow—cell of the meristemoid producing the “periderm”. Bar 600 μm

Full size image
Fig. 6
figure 6

Ultrastructure of infected cell from the infection thread penetration zone in Robinia pseudoacacia root nodules. N cell nucleus, V vacuole (note the electron-dense material), thin arrow—infection thread wall, double arrow—electron-empty layer between the infection thread wall and thread matrix, double arrowhead—rhizobium cell to be released from the infection thread, black asterisks—mitochondria, white asterisks—plastids, white arrowhead—a plastid with heavy accumulation of ferritin (shown enlarged in the insert). Bars 5 μm and 0.2 μm, respectively

Full size image

In the infected cells of differentiated bacteroid tissue, a large vacuole occurred at the centrally positioned nucleus (Fig. 7 and the insert). Plastids were devoid of starch grains. Together with mitochondria, they formed tight clusters at the intercellular spaces. Several bacteroids were observed in some symbiosomes.

Fig. 7
figure 7

Ultrastructure of differentiated bacteroid tissue in Robinia pseudoacacia root nodules. IS intercellular space, N cell nucleus, NIC non-infected cells of the bacteroid tissue, S symbiosomes, V vacuole with a fine-fibrillar material, black asterisks in insert—mitochondria, white asterisks in insert—plastids (note the close arrangement of plastids and mitochondria and their positioning at the IS), arrowheads—starch grains in NIC’s amyloplasts. Bars10 μm and 2 μm, respectively

Full size image

The senescence of bacteroid tissue-infected cells started with widening of peribacteroid space and fusion of symbiosomes (data not shown). In some symbiosomes, the peribacteroid membrane fused with the tonoplast. In such cases, the vacuolar lytic system could participate in the degradation of bacteroids, but most bacteroids were degraded within their symbiosomes. The degradation of the host organellae followed the degradation of symbiosomes. The infection thread wall became visibly loosened. In the non-infected cells, the size of starch grains increased, the vacuoles decreased, but the cytoplasm loose its density.

After the infected cell’s degradation was complete, some of the dead cells became populated by large number of rhizobia (Fig. 5), which released themselves from infection threads and never started the transformation into bacteroids. Such colonies of saprotrophic rhizobia were present also within numerous intercellular spaces.

Within the nodule cortex of young nodules, three typical layers were differentiated (Fig. 5): the multilayered inner cortex with the boundary layer, the monolayered cortical endodermis, and the outer cortex of loose parenchyma. In some parts of the older nodules, the inner cortex cells adjoining the cortical endodermis dedifferentiated and formed meristemoids that produced parenchymatous cells in regular files, resembling the arrangement of periderm cells (Fig. 5).

Discussion

R. pseudoacacia rhizobia originating from Poland and Japan, which were identified as Mesorhizobium genus bacteria by phenotypic, genomic, and 16S rDNA phylogenetic analyses (Mierzwa et al. 2009), were now studied for evolutionary history of nodulation (nodA, nodC,nodH) and nitrogen-fixation (nifH) genes. The products of all these symbiotic genes have well-defined functions. The NodA determines the type of N-acyl substitution of the Nod factor and therefore it plays a crucial role in host plant recognition; NodC controls the length of chitin oligosaccharide in Nod factor biosynthesis, NodH is the host range determinant and catalyzes the transfer of sulfate from PAPS to Nod factor, and NifH is the Fe–S protein subunit of nitrogenase (Debelle et al. 2001; Perret et al. 2000). Phylogenetic analysis of both nodA and nodC sequences of R. pseudoacacia rhizobia provided compelling evidence for their monophyletic origin and a strong support for the hypothesis that these bacteria form a sister lineage to Phaseolus sp. symbionts known to be taxonomically distant and to M. amorphae, symbiont of Amorpha fruticosa (Figs. 1, 2). We also suggest that these bacteria inherit nod genes from their common ancestor. Black locust rhizobia studied in this work as well as those from China, Germany, and North America have very similar or even identical nod sequences without mosaicism and without biogeographic patterns (Figs. 1, 2). The clustering of all R. pseudoacacia rhizobia in a single, tight clade of their own may suggest that dissemination of nod genes within R. pseudoacacia rhizobia lineage occurred mainly through vertical transmission and that lateral transfer of nod genes between these bacteria and other rhizobia is rather a rare event. We also suggest that an extremely low rate of nodA and nodC divergence noted for black locust symbionts is required to their proper function in symbiotic association with the host plant and that plant strongly affected evolution of common nod genes by selecting against changes preventing host nodulation. Several rhizobiologists discussed the role of leguminous plant as the key evolutionary force that provides cohesion to rhizobia by structuring of nodulation genes (Aguilar et al. 2004; Bena et al. 2005; Wernegreen and Riley 1999). Bipartition separating R. pseudoacacia symbionts and those of A. fruticosa (nodC gene tree) and Phaseolus sp. (nodA tree) may correspond to speciation events that guarantee optimal cooperation of both symbiotic partners. Extreme divergence of nodA and nodC sequences of Azorhizobium caulinodans in relation to those of the tested rhizobia that form stem and root symbiosis with plants of the same tribe Robinieae, i.e., Sesbania sp. and Robinia pseudoacacia, respectively (Perret et al. 2000), allows us to hypothesize that stem nodulation requires a specific adaptation of nod genes controlled by leguminous host.

R. pseudoacacia rhizobia were found to harbor nodH genes suggesting that these bacteria produce Nod molecules that bear a sulfate, although the functionality of these genes was not proved for them. The almost identical nodH sequences of black locust rhizobia are the most similar to the corresponding genes of Mesorhizobium sp. (Oxytropis arctobia) N33 and Sinorhizobium sp. BR816 strains. On the nodH gene tree, all these bacteria grouped together in a single clade (Fig. 4) suggesting that their nodH genes have a common evolutionary history. The inferred phylogeny seems to be not reliable due to a very low number of nodH sequences available and included into this study. Sulfation of Nod signal molecules was demonstrated in several rhizobium species. For the first time in S. meliloti, where it is required for alfalfa nodulation (Lerouge et al. 1990), and subsequently in Mesorhizobium sp. (Oxytropis arctobia) N33 which forms nodules on both arctic and temperate legumes such as Astragalus, Onobrychis, and Oxytropis species (Debelle et al. 2001), R. tropici, a broad host range bacterium that is able to nodulate Phaseolus vulgaris, Macroptilium atropurpureum, and Leucaena leucocephala (Poupot et al. 1993), M. huakuii, symbiont of Astragalus sinicus (Yang et al. 1999), Sinorhizobium sp. BR816, a heat-tolerant tropical strain isolated from Leucaena leucocephala that also nodulate common bean (Laeremans et al. 1997), and Methylobacter nodulans ORS 2060, symbiont of Crotalaria podocarpa (Renier et al. 2007).

In this work, we generated also nifH sequences of the tested R. pseudoacacia rhizobia and studied their evolutionary history. Analysis revealed that R. pseudoacacia is nodulated by rhizobia with very similar nifH genes despite the fact that some were from Poland and some from Japan. In the nifH tree (Fig. 3), black locust symbionts formed its own highly supported lineage associated with Mesorhizobium species in a robust, monophyletic cluster reflecting that nitrogenase Fe–S proteins of these bacteria have a common evolutionary history. The observed clustering of R. pseudoacacia rhizobia was congruent with that expected from the previous work on 16S rDNA phylogeny (Mierzwa et al. 2009) and compatible with the view of many authors that nifH gene phylogeny closely follows 16S rDNA evolutionary history (Hennecke et al. 1985; Wdowiak-Wróbel and Małek 2010; Zhang et al. 2008). The common evolutionary patterns of nifH and 16S rRNA genes and tight clustering of R. pseudoacacia rhizobia on its own branch may suggest that nifH genes of these bacteria require for optimal functioning a correct chromosomal background and that they evolved to a large degree parallelly with bacteria which carry them without a significant gene exchange with other bacteria.

In conclusion, the results from molecular analysis of symbiotic genes of R. pseudoacacia rhizobia suggest that: (1) common nod and nifH genes have independent evolutionary histories, i.e., nifH genes evolved to a large degree in a similar fashion as the bacteria that carry them, whereas the nodA and nodC gene phylogeny did not follow that of the core genes, (2) the studied symbiotic genes are well conserved, and (3) sym gene inheritance seems to be dominated by vertical descent.

Nod metabolites, encoded by common and host specificity nodulation genes of rhizobia, are key molecules in eliciting nodules in legume roots/or stems (Perret et al. 2000). Rhizobia vary significantly in their nodulation specificity. Some of them nodulate only a particular leguminous plants as for example R. leguminosarum bv. trifolii that forms symbiosis with Trifolium sp., and others are promiscuous and infect plants from various tribes and even various subfamilies as, for example, Rhizobium sp. NGR234 that infects 232 legume species of 112 genera tested and even forms nodules on non-legume Parasponia andersonii belonging to Ulmaceae family (Pueppke and Broughton 1999).

Microsymbionts of R. pseudoacacia represent a restricted range of host specificity. They were fastidious in the their choice of a symbiotic partner and formed effective associations with the host from which they were isolated but also with plants of Amorpha genus, i.e., A. fruticosa and A. californica (Table 1). It is worth noting that vice versa A. fruticosa also formed N2-fixing associations with R. pseudoacacia (data not presented).

Root nodules of the tribe Robinieae are generally of indeterminate type (Sprent 2001), although morphologically determinate nodules are also reported (Corby 1988). Here, it is confirmed in R. pseudoacacia. In this species, young nodules, if not examined in serial anatomical sections, may be mistaken for determinate ones due to their spherical shape together with the characteristic “lenticels” (data not shown). However, the older nodules developed a cylindrical shape typical for indeterminate nodules (Fig. 5). Also, as showed our observations under the light microscope (Fig. 5), the general anatomy of R. pseudoacacia nodules is similar to the well-studied indeterminate nodules of Medicago sp. (Timmers et al. 2000; Vasse et al. 1990) and e.g., Trifolium sp. (Łotocka et al. 1997). However, at the level of ultrastructure, some traits are observed that are specific for R. pseudoacacia nodules (data not shown). Among them, the most significant seems to be the absence of the II/III interzone at the proximal face of the differentiating bacteroid tissue. In the II/III interzone, as described by Vasse et al. (Vasse et al. 1990) in Medicago sp. and later observed in the other genera, the amyloplasts of the infected cells are positioned at the intercellular spaces and they rapidly accumulated especially large starch grains. In our work on R. pseudoacacia, very little or none starch was observed in the infected cell plastids (Figs. 6, 7), also at their late differentiation stage, and thus no II/III interzone can be discerned in the nodules of this species.