Phylogeny of Symbiotic Genes and the Symbiotic Properties of Rhizobia Specific to Astragalus glycyphyllos L.

The phylogeny of symbiotic genes of Astragalus glycyphyllos L. (liquorice milkvetch) nodule isolates was studied by comparative sequence analysis of nodA, nodC, nodH and nifH loci. In all these genes phylograms, liquorice milkvetch rhizobia (closely related to bacteria of three species, i.e. Mesorhizobium amorphae, Mesorhizobium septentrionale and Mesorhizobium ciceri) formed one clearly separate cluster suggesting the horizontal transfer of symbiotic genes from a single ancestor to the bacteria being studied. The high sequence similarity of the symbiotic genes of A. glycyphyllos rhizobia (99–100% in the case of nodAC and nifH genes, and 98–99% in the case of nodH one) points to the relatively recent (in evolutionary scale) lateral transfer of these genes. In the nodACH and nifH phylograms, A. glycyphyllos nodule isolates were grouped together with the genus Mesorhizobium species in one monophyletic clade, close to M. ciceri, Mesorhizobium opportunistum and Mesorhizobium australicum symbiovar biserrulae bacteria, which correlates with the close relationship of these rhizobia host plants. Plant tests revealed the narrow host range of A. glycyphyllos rhizobia. They formed effective symbiotic interactions with their native host (A. glycyphyllos) and Amorpha fruticosa but not with 11 other fabacean species. The nodules induced on A. glycyphyllos roots were indeterminate with apical, persistent meristem, an age gradient of nodule tissues and cortical vascular bundles. To reflect the symbiosis-adaptive phenotype of rhizobia, specific for A. glycyphyllos, we propose for these bacteria the new symbiovar “glycyphyllae”, based on nodA and nodC genes sequences.

The taxonomy of rhizobia does not globally reflect the symbiotic features of bacteria, particularly their host plant range. It is well established that rhizobia with a broad host range are able to nodulate a high number of fabacean genera, as for example Ensifer fredii NGR234, which forms symbiotic interactions with 232 fabacean species, belonging to 112 genera and with nonfabacean plants of the genus Parasponia (the Cannabaceae family) whereas, rhizobia exhibiting a narrow host range can nodulate only a few hosts, e.g. Rhizoium leguminosarum bv. trifolii form nodules only on plants of the genus Trifolium [4,5].
The effective rhizobium-fabacean symbiosis requires several classes of bacterial symbiotic genes, including nif ones encoding enzymes responsible for atmospheric nitrogen fixation and nod genes necessary for the synthesis of Nod factors, which act as nodule morphogenesis signals and trigger the plants towards nodule formation [6][7][8]. All Nod factors are β-1,4 linked Nacetyl-glucosamine oligomers ranging in length from 3 to 5 residues which are N-acylated at the non-reducing end and substituted by different chemical groups at both ends. The synthesis of the N-acylated chitin oligomers of the Nod factors is controlled by common nodA, nodB and nodC genes, that occur in all symbiotic rhizobia except for photosynthetic Bradyrhizobium genus strains BTAi1 and ORS278, which elicit root-and stem-nodules on Aeschynomene sensitive and Aeschynomene indica [9]. The common nodABC genes are spatially and functionally linked and they encode an acyltransferase, a chitin oligomer deacetylase and a chitin oligomer synthase, respectively [10][11][12][13]. The NodA and NodC proteins also determine the host range of the rhizobia. Various nodA gene products recognize and transfer specific fatty acids to the chitin oligomers whereas NodC proteins determine the length of the chitooligosaccharide chain. The other nod genes, e.g. nodH, nodZ and nodS, present in different combinations in rhizobial species, encode characteristic decorations of the Nod factors backbones and are primary determinants of host nodulation specificity [14][15][16][17].
The development of root nodules is a multistep-process, which drives rhizospheric rhizobia into an endocellular plant niche [18,19]. In most cases, rhizobia are internalized in plant cells, via an endocytosis-like process, surrounded by a plant plasma membrane forming symbiosomes with bacteria (bacteroids) inside them and next, the fixation of atmospheric nitrogen can occur. Depending on the host genus, nodules induced by rhizobia can be "indeterminate", with persistent meristematic growth or "determinate" characterized by an early cessation of meristematic activity and growth through the expansion of infected cells in the central part of nodule [19][20][21].
Presently, rhizobia within species are grouped into biovars (symbiovars) according to their symbiotic host-specificity and mainly on the basis of nodC gene phylogenetic analysis [3]. Jordan [22] was the first to use the term "biovar" to design three groups of R. leguminosarum bacteria, which are able to nodulate Trifolium sp. (R. leguminosarum bv. trifolii), Vicia sp. (R. leguminosarum bv. viciae) or Phaseolus sp. (R. leguminosarum bv. phaseoli). Later, these three symbiovars of R. leguminosarum were supported by the sequence analysis of nodC genes [23]. The concept of symbiovars seems to be correct in the case of plants restrictive for nodulation (e.g. Trifolium sp., Vicia sp., Cicer sp.) but this issue is still being discussed in the case of fabacean promiscuous for nodulation, that form symbiosis with different rhizobia biovars harboring different nodulation genes (e.g. Phaseolus sp.) [24,25]. Currently, most symbiovars were determined using phylogenetic criteria based on the nodC [23,[26][27][28][29], nodA [29][30][31] and even nifH gene sequences [32][33][34] with a clear trend of using nodC gene to define symbiovars within rhizobium species. It seems, that in order to better understand rhizobium-fabacean symbiosis, the characterization and phylogenetic analysis of symbiotic genes should be included in minimal standards, when new fabacean symbionts are being described.
The aim of our study was determination of the phylogeny of the nodACH and nifH genes of A. glycyphyllos symbionts, affiliation of these rhizobia into symbiovar based on nodA and nodC gene sequence analyses, determination of the host plant range of studied rhizobia and description of the microscopic structure of A. glycyphyllos root nodules.

Bacterial strains and growth conditions
For the phylogenetic analysis of nodACH and nifH genes of A. glycyphyllos symbionts, the following six nodule isolates were used; AG1, AG7, AG15, AG27 (closely related to M. ciceri bacteria) and AG17 and AG22 (closely related to the bacteria of species M. amorphae and M. septentrionale) [35]. Furthermore, these strains and 22 other A. glycyphyllos nodulators described by Gnat et al. [35] were used in laboratory plant tests, in order to determine their host range and N 2 fixation effectiveness. Additionally, in plant test Mesorhizobium albiziae CCBAU61158, Mesorhizobium caraganae CCBAU11299, M. amorphae ICMP15022, Mesorhizobium chacoense USDA4963, M. ciceri USDA3383, Mesorhizobium gobiense CCBAU83330, Mesorhizobium huakuii USDA4779, Mesorhizobium loti USDA3471, Mesorhizobium plurifarium USDA3707, M. septentrionale SDW018, Mesorhizobium temperatum LMG23931 and Mesorhizobium tianshanense USDA3592 were used in order to determine their ability to form symbiosis with A. glycyphyllos. The nodule bacteria were maintained at 4°C and cultured in a yeast extract-mannitol medium (YEM) at 28°C, as described by Vincent [36].

Total DNA isolation
For the PCR-sequencing experiments, genomic DNAs were isolated from 30 ml of a 3-day-old bacterial culture in a YEM medium, according to the method of Pitcher et al. [37]. The purity and concentration of the DNA were determined with a spectrophotometer (Bio-Rad, Smart-Spec TM 3000).

PCR amplification of symbiotic genes and amplicons sequencing
The amplification reactions were performed with a ReadyMix TM Taq kit, following manufacturer's specifications (Sigma). The 660-bp fragment of the nodA gene was amplified and sequenced using a forward primer, nodA-1 (5'-TGCRGTGGAARNTRNNCTGGGAAA-3') and a reverse one, nodA-2 (5'-GGNCCGTCRTCRAAWGTCARGTA-3') according to the procedure described by Haukka et al. [38]. The 890-bp long nodC gene fragment was amplified and sequenced with a forward primer, NodCFu (5'-AYGTHGTYGAYGACGGITC-3') and a reverse primer, NodCI (5'-CGYGACAGCCANTCKCTATTG-3') using the same cycling parameters as reported for this gene by Laguerre et al. [26]. The 567-bp fragment of the nodH gene was amplified and sequenced with forward, TSnodH1 (5'-VTKGAGYAACGGTGAR YTGCTCA-3') and reverse, TSnodH2 (5'-GCGAAGTGAWSCCGCAACTC-3') primers, using 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 780-bp long fragment of the nifH gene was amplified and sequenced with NifH1 (5'-CGTTTTACGGCAAGGGCGG-3') and NifH2 (5'-TCCTCCAGCTCCTCCATGGT-3') primers according to the protocol of Perret and Broughton [39]. PCR products were purified using Montage PCR Filter Units (Millipore, Massachusetts, USA) as recommended by the manufacturer. The purified amplicons were electrophoresed, in 1% agarose gel, to estimate the amount of DNA and then both strands of each amplified DNA were sequenced, with an ABI Prism BigDye Terminator Cycle sequence kit (Applied Biosystems, Foster City, CA, USA) and analyzed on a 3500 Genetic Analyzer (Life Technologies) sequencer, following manufacturer's instructions. Sequences were deposited in the GenBank database under the accession numbers listed in the phylograms.

Phylogenetic analysis
The symbiotic genes sequences of A. glycyphyllos nodule isolates and those of related bacteria from GenBank were aligned using ClustalW [40]. Next, they were manually corrected using Gen-eDoc software [41]. Phylograms were generated in MEGA program version 6.0 [42] using the neighbor-joining method (NJ) [43] based on Kimura's two-parameter model (K2P) [44]. The stability of the strains groupings in the trees was estimated by bootstrap analysis (1,000 replicates) using the same MEGA software. The resulting trees were displayed using Tree View [45].

Plant tests
All 28 A. glycyphyllos nodule isolates [35] were studied for their ability to form a symbiotic interaction with their native host (A. glycyphyllos), Amorpha fruticosa, Astragalus cicer, Astragalus sinicus, Robinia pseudoacacia, Trifolium pratense, Lotus corniculatus, Medicago sativa, Ornithopus sativus, Phaseolus vulgaris, Glycine max, Vicia sativa and Lupinus luteus. For plant tests, seeds were surface-sterilized [46], germinated on 0.8% agar-water plates and next, the obtained seedlings were placed in glass tubes filled with a N-free Hoagland's nutrient solidified with agar [47]. Three days later, the seedlings' rootlets were inoculated with a rhizobial suspension (10 8 cells per tube) and grown for six weeks at 21°C, with 14 h of light, per 24 h photoperiod. Non-inoculated plants, grown in a N-free Hoagland's medium [47], were included as negative controls. For each rhizobial strain-fabacean species, six replicates were prepared. The symbiotic properties of A. glycyphyllos rhizobia were evaluated by the presence and color of the nodules on inoculated plants, shoot dry weight and the acetylene reducing test, as an indicator of nitrogenase activity [48]. Additionally, the symbiotic interaction of A.

Statistical analysis of plant-test data
Date from plant tests, concerning symbiotic interaction of A. glycyphyllos and A fruticosa with rhizobium strains studied, were expressed as mean ± SD values. Differences between the studied groups of plants were analyzed, by a two-way analysis of variance (ANOVA) followed by Duncan's multiple-range test (IBM SPSS Statistics 21, IBM Corporation, Somers, NY, USA). Differences at p < 0.05 were considered statistically significant.

Light and electron microscopy
For microscopy studies, inoculated A. glycyphyllos plants (prepared as for the plant test) with the first nodules visible, were carefully pulled out of the glass tubes and transferred into pots, filled with a sterile perlite supplemented with a N-free Hoagland's medium [47]. Plants were grown under 14 h light/10 h darkness photoperiod and whole root nodules were sampled 14 weeks after inoculation. The nodules were surface-cut for better penetration of reagents and fixed in a mixture of 5% glutaraldehyde and 4% paraformaldehyde in a 0.1 M Na-cacodylate buffer, pH 7.2, for 12 h under an air pressure of -0.06 MPa at room temperature. Samples were rinsed, contrasted and dehydrated, as described earlier [15]. Next, samples were embedded in a glycid ether 100 epoxy resin grade hard, which was prepared (and polymerized) according to the manufacturer's formula (SERVA). Central longitudinal sections of 2 μm thick were cut in a sagittal plane and prepared for light microscopy, as described by Kalita et al. [15]. Ultra-thin sections, for ultrastructure observations, were placed on slot-grids coated with Formvar (SERVA), and treated as described by the same authors. Anatomical observations and micrographs were done using a light microscope Provis AX (Olympus). The ultrastructure was observed under a transmission electron microscope, Morgagni 268 (FEI Company) operating at 80 kV.

Results and Discussion
Effective symbiosis between rhizobia and fabaceans, initiated by nitrogen starvation of the host plant, requires several bacterial symbiotic genes, including nitrogen-fixation (nif) ones, determining the reduction of N 2 into ammonium and nodulation (nod) genes that encode Nod factors triggering nodule formation [7,8,[49][50][51]. The basic structure of the Nod factors produced by different rhizobium species is very similar. It consists of a β-1,4-linked N-acetyl-D-glucosamine backbone, substituted at the non-reducing end by an acyl chain [13,52]. Depending on the bacterial species, Nod factors can differ in the number of glucosamine residues present, the structure of the acyl chain and substitutions at the reducing and non-reducing terminal gluscosamine [11,52,53].
In this study, we have focused on the phylogeny of A. glycyphyllos symbionts' common nodA and nodC genes, the nodH host specificity gene, as well as the nifH gene encoding dinitrogenase reductase. Phylograms of these four loci were compared, in order to determine whether their evolutionary history was similar and to explain whether nod and nifH loci were transferred across chromosomal lineages. The A. glycyphyllos nodulators studied represent different phenons and genomotypes of liquorice milkvetch symbionts [35,54] and they, based on the 16S rRNA gene sequence analysis, were determined to be the close phylogenetic neighbors of M. amorphae and M. septentrionale (AG17 and AG22) and M. ciceri (AG1, AG7, AG15 and AG27) [35].
The nodA gene phylogeny The nodulation gene nodA determines the type of N-acyl substitution on the nonreducing end of the Nod factor core and thus plays a significant role in determining the symbiotic specificity of nodule bacteria [8,38,53]. The 549 bp long, intragenic fragments of the nodA gene were sequenced for six A. glycyphyllos symbionts studied and aligned to nodA sequences of 24 reference rhizobia by CLUSTALW. The alignment showed 79 constant positions, 337 variable parsimony informative and 133 variable, but parsimony uninformative ones. The ratio of transitional to transversional pairs (ti/tv) was 0.94. In the phylogram of nodA sequences, liquorice milkvetch nodulators, with 99-100% nodA sequence identity to each other, were in the Mesorhizobium species cluster and formed a common group with them, at 86% bootstrap support (BS) (Fig 1). They were phylogenetically divergent from all other rhizobia included in the analysis and in the phylogram formed an independent branch with 100% BS. At the base of this lineage (as a separate cluster) M. ciceri, M. opportunistum, and M. australicum strains representing symbiovar biserrulae, with a 91-92% nodA sequence similarity to A. glycyphyllos rhizobia genes, were located. The nodA gene sequences similarities of A. glycyphyllos symbionts and other reference Mesorhizobium species, were in the range of 77-80%, and much lower to nodA sequences of the Bradyrhizobium sp., Ensifer sp., Rhizobium sp. and Neorhizobium galegae; i.e. 63-66, 66-69, 64-67 and 61%, respectively.
The most striking aspect of nodA gene evolution revealed by the nodA gene tree is the fact that all isolates from the A. glycyphyllos root nodules, regardless of their species designation, were recovered in one strongly supported clade, which indicates the possibility of lateral transfer of the nodA gene from a common ancestor to bacteria being studied. The 99-100% nodA sequence similarity of rhizobia specific for A. glycyphyllos points also to a relatively recent (in evolutionary scale) horizontal transfer of this gene to the bacteria being studied. The occurrence of horizontal transfer of the symbiosis genes was reported earlier by Sulivan et al. [55]. Due to this process, the mesorhizobia native to New Zealand, obtained nod genes (present in symbiosis islands) from an inoculant strain M. loti ICMP3153 and became able to N 2 fixing association with the introduced plant, L. corniculatus [55][56][57]. A similar event took place in Australia, where local soil bacteria have acquired the ability to nodulate the pasture legume, Biserrula pelecinus, via lateral transfer to them of a symbiosis island from the commercial inoculant, M. ciceri bv. biserrulae strain WSM1497 [31]. Symbiosis genes of the genus Mesorhizobium bacteria are mobile, usually being found in plasmids or transmissible, chromosomal islands [16,[57][58][59][60][61].
The nodC gene phylogeny The nodC gene encoding N-acetylglucosaminyltransferase, responsible for the first step in Nod factor assembly, is related to the host range of the nodule bacteria [7,12]. The 614 bp long, nodC genes alignment of milkvetch symbionts, studied here and 24 reference rhizobia, yielded information about nodC sequence similarities of these bacteria and showed 84 constant characters, 337 parsimony informative and 133 variable, but parsimony uninformative ones as well as revealed moderate bias towards transitions (ti/tv = 1.18).
The analysis of the nodC sequences of A. glycyphyllos symbionts provided results congruent to those of the nodA genes. In the nodC gene phylogram (Fig 2), A. glycyphyllos nodule isolates, with a 99-100% sequence identity to one another, formed an independent branch close to a clearly separate cluster comprising of bacteria classified as symbiovar biserrulae i.e. M. ciceri, M. opportunistum, and M. australicum with a 92-93% sequence similarity to the rhizobia tested. The high similarity between the nodC sequences of liquorice milkvetch nodulators suggests that nodC genes of the bacteria studied have a monophyletic origin and derive from the same ancestor. As in the nodA gene dendrogram (Fig 1), the Mesorhizobium species and A. glycyphyllos symbionts (an 80-82% sequence similarity, not taking into account symbiovar biserrulae strains) formed a separate, well supported clade (78% BS). The nodC genes of the Rhizobium etli bv. phaseoli+R. giardinii bv. giardinii+ R. gallicum, Ensifer fredii, R. tropici, Bradyrhizobium sp., E. meliloti+E. medicae and N. galegae, with sequence similarities to nodC liquorice nodulators genes of 76-82, 77, 70-76, 74 and 68%, respectively, formed separate phylogenetic lineages on the nodC gene dendrogram. The grouping of A. glycyphyllos nodulators into one, clearly separate monophyletic cluster, close to the symbiovar biserrulae mesorhizobia and together with other Mesorhizobium species, shown in the nodC gene phylogram (Fig 2), was consistent with that presented in the nodA gene tree (Fig 1). Thus, both of these genes, of rhizobia specific for A. glycyphyllos nodulators, appear to share a common genealogy and probably derive from a single ancestor. It is highly probable that evolution of the host and changes in the host's environment have exerted selection pressure on liquorice milkvetch symbionts and shaped the structure of the bacterial nod genes. It is also worth emphasizing, that nodC and nodA alleles of symbiovar biserrulae rhizobia split off at a basal position of the liquorice milkvetch symbionts, which correlates with the close relationship of these bacteria host plants; i.e. Biserrula pelecinus and A. glycyphyllos belonging to the same subtribe Astragalinae [62]. To reflect the genetically based symbiosis adaptive phenotype (ecotype), displayed by liquorice milkvetch nodulators, we propose the new symbiovar "glycyphyllae" for bacteria forming nitrogen-fixing associations with A. glycyphyllos. The nodC gene, encoding a chitin synthase and related to host specificity [7,12,26], is commonly used for symbiovar determination in rhizobia [28,29,31,[63][64][65][66]. Currently, within the genus Mesorhizobum three following symbiovars are described, i. The nodH gene phylogeny The nodH gene, which encodes sulfotransferase involved in the transfer of a sulfate group to the reducing end of the Nod factors [7,14,[67][68][69], was identified in the A. glycyphyllos symbionts genomes by the PCR technique. The 491 bp fragment of the nodH gene was amplified in all six studied strains. Sequencing of the PCR products confirmed their similarity to the nodH genes of rhizobia, in which sulfated Nod factors were chemically supported [67][68][69]. The alignment of nodH gene sequences of liquorice milkvetch nodulators and reference rhizobia exhibited 180 constant positions, 80 variable but parsimony uninformative and 231 parsimony informative ones. The ti/tv ratio (1.11) points to a little bias towards transitions in the analyzed alignment. The obtained in this work nodH gene sequences (sharing 98-99% sequence identity with one another) were the most closely related to those of the M. ciceri, M. opportunistum and M. australicum sb. biserrulae strains (94% sequence similarity), and next, to the sequences of M. huakuii-symbiont of A. sinicus and Mesorhizobium sp. N33-symbiont of Oxytropis arctobia (subtribe Astragalinae) (83-84% and 76-77% sequence identity, respectively). Nucleotide identity of the nodH genes of A. glycyphyllos symbionts and other nodule bacteria, included in the analysis, was in the range from 67 to 74%. The phylogenetic analysis of the nodH sequences resulted in the NJ tree, presented in Fig 3. All six liquorice milkvetch nodulators were grouped within one, tight, monophyletic cluster (97% BS). In a sister, clearly independent clade, symbiovar biserrulae strains were placed (100% BS) and, at the base of this cluster, M. huakuii was located in a separate, highly supported branch. All these bacteria formed a strongly supported clade (100%) which indicates its robustness. Outside of this cluster, the other rhizobial strains harboring nodH genes were positioned. The presence of nodH genes in the genome of rhizobia specific for A. glycyphyllos suggests that these bacteria produce a 6-O sulfated Nod factor. Sulfation of Nod factors is an important determinant of host plant specificity and was documented in; E. meliloti (a symbiont of Medicago sp., Melilotus sp. and Trigonella sp.) [70], R. tropici (a symbiont of P. vulgaris, Macroptilium atropurpureum and Leucaena leucocephala [67], Ensifer sp. BR816 (a symbiont of L. leucocephala and P. vulgaris) [71], M. huakuii (a symbiont of A. sinicus [72], Mesorhizobium sp. N33 (a symbiont of Astragalus sp., Onobrychis sp. and Oxytropis sp.) [68] and Methylobacter nodulans ORS2060 (a symbiont of Crotalaria podocarpa) [73].

The nifH gene phylogeny
In our present study, on the phylogeny of symbiotic genes of A. glycyphyllos nodule isolates, the nifH gene encoding the Fe protein of the nitrogenase complex, was also sequenced and analyzed. The nifH gene sequences of strains studied and the reference rhizobia were aligned for a stretch of 602 nucleotides. This alignment exhibited 334 constant characters, 26 variable but parsimony uninformative and 242 parsimony informative ones. Additionally, a moderate transversion bias (ti/tv = 0.94) was also found. Bacterial strains isolated from A. glycyphyllos root nodules, with little variation across their nifH gene sequences (1%), were clustered together with the Mesorhizobium species (77-80% sequence similarity omitting symbiovar biserrulae strains) and formed a group clearly separate from all of them (Fig 4). Their closest phylogenetic neighbors were the M. ciceri, M. opportunistum and M. australicum sb. biserrulae strains which showed 91-92% nifH sequence identity with the studied milkvetch nodulators. The A. glycyphyllos symbionts formed two robust, closely related subclusters in the upper part of the nifH phylogram (AG17+AG22 and AG1+AG7+AG15+AG27) as in the 16S rDNA dendrogram [35]. In the branches, sister to those comprising of the mesorhizobia, the genus Rhizobium and Ensifer strains (sharing 64-69% identical sequences with bacteria tested) were placed. On the outskirts of the nifH gene tree, the Bradyrhizobium species, with 63-66% nifH sequence identity to A. glycyphyllos nodulators, were located (Fig 4). The nifH gene-based tree demonstrated a phylogenetic relationship of A. glycyphyllos symbionts, similar to that based on the common nod genes. Similar paths of the rhizobia nifH and common nod genes phylogenies may be explained by the close proximity of these genes to one another on symbiotic plasmids or chromosomal symbiotic islands, and by the lateral transfer of these mobile elements, mainly across species within genus and in some instances, between bacteria belonging to different genera [57][58][59][60][61]. The lateral transfer of symbiotic genes from a common ancestor, that lived in soil, to AG17 and AG22 strains (related to M. amorphae and M. septentrionale) as well as AG1, AG7, AG15 and AG27 isolates (related to M. ciceri) could enable these bacteria an effective

Host range and symbiotic efficiency
Rhizobia forming N 2 fixing interactions with fabaceans exhibit various degree of symbiotic specificity, from very narrow as in the case of bacteria associated with plants of tribes; Trifolieae, Cicereae or Viceae [7] to broad host range (promiscuous) as for example strains isolated from nodules of L. leucocephala, Gliricidia sepium, Prosopsis cineraria, P. vulgaris or Sesbania rostrata [5,7,74,75]. The host range and symbiotic effectiveness of 28 A. glycyphyllos root nodule isolates, classified as a moderately-slow growing mesorhizobia [35], were analyzed in laboratory plant tests on 13 fabacean species. All studied A. glycyphyllos mesorhizobia displayed narrow symbiotic efficiency in terms of the host range and N 2 -fixation capacity determined by shoot dry mass production and the acetylene reduction test ( Table 1).
The liquorice milkvetch symbionts studied here formed pink, cylindrical nodules with substantial nitrogenase activity when inoculated on their original fabacean host, i.e. A. glycyphyllos. 54% of these bacteria also developed nodules on A. fruticosa but they were substantially less efficient in N 2 fixation with this host than with A. glycyphyllos. Six weeks after inoculation of A. glycyphyllos and A. fruticosa with the test bacteria, the shoot dry weight of plants with root nodules was significantly different (p<0.05) from that of-N controls and ranged from 8.0 to 16.25 mg (uninoculated control from 5.96 to 8.21 mg) and from 7.0 to 12.0 mg (uninoculated control from 4.82 to 7.86 mg) per plant. Furthermore, the acetylene reduction activity of rhizobial   [76]. Fabaceans, similar as rhizobia, exhibit also a different range of symbiotic specificity and single plant species may associate with several rhizobial genomotypes as for example R. pseudoacacia [7] or only with unique root-nodule bacteria, e.g. B. pelecinus, a pasture species [77]. In laboratory plant test, A.  (Table 1). This conclusion was supported by acetylene reduction test (an index of nitrogenase activity) ( Table 1), which showed that nitrogenase activity of above-mentioned Mesorhizobium species in symbiosis with A. glycyphyllos is lower than that of studied liquorice milkvetch nodule isolates ( Table 1). The Astragalus species promiscuity was described earlier and it is known, that these fabaceans are nodulated by Mesorhizobium, Rhizobium, Sinorhizobium and Bradyrhizobium species, harboring diverse symbiotic genes, mainly by the genus Mesorhizobium strains [1,78]. The mechanisms of Astragalus sp.-rhizobium promiscuity is not explained. It can be associated with a variety of host-released flavonoids, i.e. rhizobial nod genes inducers and with variety of Nod factors released by microsymbionts [7,79].
The microscopic structure of A. glycyphyllos root nodules Symbiotic interaction between diazotrophic rhizobia and fabaceans leads to formation of root and occasionally stem nodules classified into two major types, i.e. determinate and indeterminate based on their morphology, development, and physiology. The nodule structure is largely determined by the plant [80]. The major regulators of nodule development are phytohormones. Some of them play distinct or even opposite roles in the organogenesis of determinate and indeterminate nodules [81]. Plant hormones are also produced by rhizobia and these may too affect nodulation process [82].
The rhizobia studied here, induced cylindrical nodules on the A. glycyphyllos roots, resembling the indeterminate nodules of such fabaceans as, e.g.; M. sativa [83], T. repens [84], C. arietinum [85], B. pelecinus [77]. Anatomically, the liquorice milkvetch nodules were differentiated into persistent apical meristem, bacteroid tissue consisting of infected and uninfected cells, as well as multi-layered nodule cortex, with a vascular system (Fig 5A). Within the bacteroid tissue, a developmental zonation was discernible (Fig 5A). Proximal to nodule meristem, cells were penetrated by infection threads (Fig 5B) and infected endocytotically by rhizobia, which was a symptom of the onset of bacteroid tissue differentiation. In the infected cells, differentiation involved considerable cell growth, symbiosome multiplication, proliferation of membranous organelles and the decrease of vacuolation. Gradually, the amyloplasts were formed and together with closely-associated mitochondria they were translocated into the vicinity of intercellular spaces. Dividing symbiosomes were observed, even in the cells adjoining the so-called interzone II/III (Fig 5C). In the final stage of infected cell differentiation, which took place in the interzone II/III, starch grains increased the size rapidly and acquired an elongated form (Fig 5C), the single vacuole, located centrally, was formed close to the cell nucleus. In the symbiosomes, peribacteroid space was narrow (Fig 5D). Bacteroids were rounded or slightly elongated in sections, with both outer and cytoplasmic membranes (the latter forming occasional invaginations) clearly discernible. In fully differentiated infected cells, starch grains were smaller than in the interzone II/III cells and the symbiosomes were similar to those in the interzone II/III. Gradually, the senescence zone was formed due to the degenerative changes in the ultrastructure of symbiosomes (Fig 5E). The peribacteroid space widened locally and bacteroids' cytoplasm became heterogeneous (Fig 5E). Invaginations of the cytoplasmic membrane gradually increased and took the shape of intricately twisted tubules. Concurrently, the degenerative changes occurred in the host organelles and tonoplast became fragmented (not shown). Infection threads with intact rhizobial cells were visible in the senescent zone. Within the completely degraded cells, only the so-called "ghost membranes" (remnants of bacteroids' cell wall) along with infection threads were present.
At the proximal end of the bacteroid tissue, closest to the point of attachment to the "parent" root, a large saprotrophic zone was formed (Fig 5A). After the infected cell's degradation was complete, some of the dead cells became populated by a large number of rod-shaped rhizobia. They were embedded in a fibrillar matrix, which had a characteristic pinwheel pattern (Fig 5F) resulting from the orderly arrangement of fibrils.
The bacteroid tissue of nodule was surrounded by lateral tissues: a) the multilayered inner cortex with vascular bundles, b) the monolayered cortical endodermis, built of tightly-arranged cells and c) an outer cortex of loose parenchyma (Fig 5A).
Generally, the anatomy of A. glycyphyllos nodules was similar to that of other indeterminate ones [46,77,[83][84][85], however, some traits specific to liquorice milkvetch nodules were observed at the ultrastructure level; e.g. the extensive invaginations of bacteroids cytoplasmic membrane, especially evident in the senescent bacteroids.
In summary, the six A. glycyphyllos nodulators studied, closely related to members of the M. amorphae and M. septentrionale species (AG17 and AG22) and M. ciceri bacteria (AG1, AG7, AG15 and AG27), harbor highly conserved nodACH and nifH genes and all these bacteria were clustered together in a clearly separate, strongly supported clade, in corresponding genes trees. The most plausible explanation for incongruence between phylogenies based on 16S rRNA and symbiosis genes, seems to be the lateral transfer of symbiotic information, from a common ancestor to the bacteria studied. Rhizobia specific to A. glycyphyllos have a narrow host range and they were classified to the new symbiovar "glycyphyllae", based on phylogenetic analysis of the nodA and nodC genes. Nodules induced on liquorice milkvetch roots are typically indeterminate, in terms of their histological structure.