Lotus corniculatus-rhizobia symbiosis under Ni, Co and Cr stress on ultramafic soil

Ultramafic/serpentine soils constitute a stressful environment with many plant growth constrains such as a lack of macronutrients and high levels of potentially toxic metals. We considered the adaptive strategy of Lotus corniculatus L.-rhizobia symbiosis to Ni, Co and Cr stress conditions. L. corniculatus nodulating rhizobia from ultramafic soil were isolated, identified and tested for nitrogen fixation, metal tolerance and plant growth promoting abilities. The structural and immunocytochemical analyses of root nodules were also performed. The isolates effective in nitrogen fixation were identified as Rhizobium and Mesorhizobium tolerant to Ni, Co, and Cr. Some strains directly promoted root growth of L. corniculatus and non-legume Arabidopsis thaliana under metal stress. The metal treated nodules showed structural alternations, i.e. enhanced accumulation of phenols and wall thickening with higher cellulose, hemicellulose, pectins, glycoproteins and callose content. Our results revealed that metal tolerant, growth promoting rhizobacteria inhabiting L. corniculatus root nodules may improve plant growth in the ultramafic environment. Accumulation of phenols and reorganization of nodule apoplast can counteract harmful effects of Ni, Co and Cr on the symbiosis. These findings imply that L. corniculatus-rhizobia symbiosis is an important element of plant adaptation to metal stress occurring on the ultramafic soils.


Introduction
Ultramafic soils originate from weathered ultramafic rocks such as peridotite and/or serpentinite. These kinds of rocks have similar chemical composition but they differ in their mineral content. Peridotite consists mostly of olivine and pyroxene, whereas serpentinite, formed from peridotite as a result of low-temperature hydratation, is composed mostly of serpentine group minerals (Coleman 1971;Evans et al. 2013). This various mineralogy of the parent rocks is responsible for differentiation of certain ultramafic soil features such as their morphology and mobility of some elements (Alexander 2004;Kierczak et al. 2016). Despite the differences in their origin, ultramafic soils show specific, common physical and chemical features. As habitats for plant life, these soils offer a combination of multiple limitations known as the "serpentine syndrome". Among them, chemical properties such as low availability of calcium in relation to magnesium, deficiency of essential macronutrients (N, P and K) and high levels of potentially toxic metals, mostly Ni, Cr and Co, are especially stressful for plants and most markedly affect the serpentine flora (Brady et al. 2005;Kazakou et al. 2008). These typical of serpentine substrates metals are microelements metabolically important and essential for plant development but toxic when present at excessive concentrations. For instance, Ni is required for urea metabolism (Gerendás et al. 1999), but at elevated levels (> 10 mg kg -1 ) it causes growth inhibition, chlorosis, necrosis and wilting in non-tolerant plants by impairing a large variety of physiological processes or conditions (e.g., photosynthesis, respiration, water relations, chlorophyll and protein content, metal homeostasis as well as activity of H+-ATPase, and nitrate and glutathione reductase) (Ahmad and Ashraf 2011;Yusuf et al. 2011;Bhalerao et al. 2015).
The ultramafic soils with their unusual chemical composition often support distinctive floras and unique plant communities, which have been of interest to many researchers (e.g. Brooks 1987;Kruckeberg 2002;Proctor 2003;Reeves et al. 2007). In dry grasslands on ultramafic soils in Poland, representatives of Fabaceae (legumes) are abundant and frequent floristic components both on natural ultramafic outcrops (Żołnierz 2007) and on ultramafic mine soils (Kasowska and Koszelnik-Leszek 2014). Their ability to grow in metal polluted and low-fertile environments has been explained by their capacity to form symbiotic interactions with nitrogen-fixing rhizobia and, in majority of these species, with arbuscular mycorrhizal fungi (Pajuelo et al. 2011;Hao et al. 2014;Smith and Read 1997). Rhizobia housed in root nodules provide the plant with reduced nitrogen in exchange for photosynthates (Vincent 1970). Despite comprehensive research on deleterious effects of metals on the rhizobia, nodulation and nitrogen fixation ability of non-metal tolerant legumes (Chaudri et al. 1993;Zahran 1999;Gruber and Galloway 2008;Giller et al. 2009;Franzini et al. 2010;Lafuente et al. 2010), little is known about the effect of metals on the nodule structure and functioning (Ibekwe et al. 1995;Wani and Khan 2013;Sujkowska-Rybkowska and Ważny 2018). Nodules represent a unique environmental niche for symbiont accommodation and proliferation, by providing microsites in metalcontaminated soils where the toxic influence of ions may be limited. On the other hand, in metal polluted soils nodules are the main sites of metal accumulation (Gopalakrishnan et al. 2015). To understand the possible adaptive mechanisms of legume-rhizobia symbiosis to metal stress conditions it is necessary to determine cellular organization of nodules. The most important alteration under toxic metals observed in nodules is thickening of their cell walls that may prevent symbiosis impairment by metal stress Borucki 2012, 2015;Sujkowska-Rybkowska and Ważny 2018). Plant cell wall is the first barrier against toxic metals and wall thickening is one of metal exclusion strategy in plants ( Le Gall et al. 2015). The cell wall structure comprising cellulose microfibrils and noncellulosic neutral polysaccharides embedded in pectin matrix with proteins and phenolic compounds, confers metal binding ability (Probst et al. 2009;Krzesłowska 2011). Excess accumulation of toxic metals in the wall often leads to its stiffening and thickening (Probst et al. 2009;Krzesłowska 2011).
Lotus-rhizobia symbiosis is highly specific and Lotus species are nodulated by Mesorhizobium loti, Bradyrhizobium sp., Rhizobium sp. and Ensifer sp. with different levels of effectiveness (see review Lorite et al. 2018). This effectiveness depends on the habitat conditions and little is known about natural rhizobia population associated with native legumes growing on ultramafic soils (Mengoni et al. 2010). Some rhizobia may improve plant growth (plant growth promoting bacteria -PGPB) on contaminated soils through synthesis of phytohormones, siderophores, phosphate solubilization, and reduction of toxic effects of metals (Wani et al. 2007;Karthik et al. 2017). Phytohormones (e.g. auxins) produced by rhizosphere microbe enhance plant growth by directly promoting cell division, cell elongation, root initiation, and/or altering expression of specific genes (Taghavi et al. 2009;Davies 2010). Phosphate solubilizing bacteria can strongly increase phosphorus uptake by plants (Rodríguez and Fraga 1999), and siderophore increases the amount of bioavailable iron at metal contaminated sites . The siderophore can also form stable complexes with metals, thus decreasing their mobility (Neubauer et al. 2000). Apart from the ability to establish nitrogen-fixing symbiosis, many rhizobia are known to form growth promoting associations with non-legumes, such as maize, rice (Gutiérrez-Zamora and Martínez-Romero 2001; Ladha and Reddy,2003) and Arabidopsis (Zhao et al. 2017). One of the main challenges in PGPB research is to understand how the taxonomic diversity of PGPB is related to their efficiency in plant growth promotion under different stresses.
Lotus corniculatus L. (bird's foot trefoil) is one of the most scientifically and ecologically important Lotus species (Escaray et al. 2012). This legume is a pioneer plant colonizing marginal and polluted soils and it is the most commonly used Lotus species for restoration of contaminated sites (de los Santos et al. 2001;Escaray et al. 2012). It can grow on alkaline or acidic soils and shows high tolerance to aluminum, manganese and salinity (Blumenthal and McGraw 1999). It naturally occurs on different metal-rich soils including ultramafic ones (Banuelos et al. 1992;Hao et al. 2014;Kasowska and Koszelnik-Leszek 2014;Mohamad et al. 2016;Fagorzi et al. 2018), which indicates its high adaptability. However, the knowledge about metal tolerance ability of ultramafic legumes and their associated rhizobia, including their diversity, metal-tolerance, plant growth promoting abilities and symbiotic relations is limited (Abou-Shanab et al. 2007;Pal et al. 2007;Rajkumar et al. 2009;Mengoni et al. 2010;Lorite et al. 2018). These natural metal-tolerant, nitrogen fixing, growth promoting rhizobia are of a great importance for their potential use in agriculture and phytoremediation of metal contaminated areas (Hao et al. 2014).
Considering the above, in the present study we analysed some mechanisms of L. corniculatus-rhizobia symbiosis on the ultramafic soil and investigated the plant and rhizobial adaptations to these soil constrains. Our aim was to work out the answers to the following research questions: (i) What is a taxonomic position of the nodulated rhizobia and do they tolerate Ni, Cr and Co? (ii) Are the strains effective in nitrogen fixation and do they exhibit plant growth promoting abilities in the presence of toxic metals? (iii) Does the nodule structure change under metal stress? and (iv) Are the observed alterations a manifestation of adaptation strategy to the presence of toxic metals?

Study area and samples collection
Ultramafic rocks and their soils in Poland occur as a few outcrops only within the Sudetes (Lower Silesia region, SW Poland). They form here a few small bodies and larger massifs with one of the largestthe Szklary Massive. The Szklary Massive outcrop is located within the Fore Sudetic Block and is composed of partially serpentinized peridotites cut by subordinate granitoids and lamprophyre veins (Kierczak et al. 2007). Geographically, the Szklary Massive lays in the Sudetic Foothills within the Niemczańsko-Strzelińskie Hills (Supplement, Fig. S1).
The partially serpentinized peridotite from the Szklary Massif consists mainly of serpentine, olivine, and amphiboles. It contains relatively high amounts of Ni (up to 2341 mg kg -1 ), Cr (up to 2915 mg kg -1 ) and Co (up to 164 mg kg -1 ). The most important source of Ni are there olivine and serpentine, while Cr and Co are mostly bounded in spinel group minerals The soil type formed on this bedrock was classified as Eutric Leptosol (Humic, Magnesic; Kierczak et al. 2016). On the area of the Szklary Massive, exploitation of nickel ores took place from the end of 19th century. The excavation has been done by underground techniques initially and then by surface mining. Mining and Smelting Plant in Szklary operated until 1983, and after the mine has been closed their area was reclaimed and forested in the overwhelming. Some slopes of excavations and the mine pit terrain were revegetated in the spontaneous succession process. Their flora and its ecological characteristic were described by Kasowska (2005) and Kasowska and Koszelnik-Leszek (2014).
L. corniculatus plants and ultramafic soil were taken in June 2016 from the mine pit surface (N 50°38'81.4"-50°38'83.2"; E 16°50'21.2"-16°50'20.9") at sites where this species dominated in the plant cover from three sample points (SP). At least three specimens were collected from each SP for obtaining a representative sample. These samples consisted of both the plants, being at the flowering phase, and the soil bulk, containing the roots and the soil of the rooting zone (10-20 cm depth). The soil bulks were moisturized with tap water and the whole samples were packed into plastic bags and taken to a laboratory as soon as possible to preserve rhizobia vitality. Seeds of L. corniculatus were further taken in the same sample points in August 2016.

Content of Ni, Cr and Co in ultramafic soil and plant parts
Soil samples, received from the soil bulks, were airdried and sieved through a 2-mm sieve. The total concentration of Ni, Cr and Co was determined after soil digestion in aqua regia according to ISO 11466.3. Soil samples weighed to 0.250 g were placed in a 250 ml Pyrex Erlenmeyer flasks. The pre-digestion step was done at room temperature (RT) for 24 h with a mixture of 12 M HCl and 17 M HNO 3 in the 3:1 (v/v) ratio. The obtained suspensions were digested on a hot plate at 130°C for 15 min. Then, they were cooled until RT, filtered through an ashless Whatman 41 filter and, finally, diluted to 25 ml with 0.17 M HNO 3 .
The geochemically reactive forms of the studied elements were analyzed after extraction in 0.43 M HNO 3 (Groenenberg et al. 2017). The extraction solution was prepared by dilution of 30 mL concentrated HNO 3 (65%, Merck, analytical grade) in 1000 mL ultrapure water. The samples were shaken during 2h with soil: solution ratio 1:10 (weight/volume). The concentrations of both total and reactive forms of metals were determined by atomic absorption spectrometry with a Varian SpectrAA 220FS apparatus. Samples of birdsfoot trefoil shoots and roots, separated from the soil bulks, were thoroughly washed with tap and distilled water, and dried at 40°C. Ground plant material was oven ashed at 475°C overnight. Concentration of metals, after ash dissolution in 6M HCl (Sillanpää 1982), were determined with a Varian SpectrAA 220FS AAS apparatus. The samples from each SP were analysed separately and all analysis and measurements were done in two analytical replicates. To determine L. corniculatus ability to transferring metals from roots to aboveground parts the translocation factor (TF) was evaluated as a ratio of the metal content (mg kg -1 ) in the shoots to this metal content (mg kg -1 ) in the roots.

Strain isolation and molecular identification
Because of the very small amount of nodules occurred at the ultramafic L. corniculatus roots, the plant specimens collected at all SPs were pooled to obtain enough material for all microbiological tests. For rhizobia isolation, about twenty nodules, at least two effective ones per ultramafic plant, were used. Rhizobia strains were isolated according to Vincent (1970) method and grown on YEM medium. The nodulation ability of isolated strains were performed in the authentication test on serpentine bird's foot plants. The seeds of L. corniculatus were scarified with 96% sulfuric acid (H 2 SO 4 ) for 6 minutes and washed in sterile deionized water five times and after germination for 3 days on Petri dishes containing 1% water agar were transferred to pots (5 seedlings per pot) filled with sterile ultramafic soil and singly inoculated. Soil was sterilized for 3 days at 100°C for 1h each day and then allowed to cool for 72 h (Ogar et al. 2015). The effectiveness of soil sterilization was confirmed by the lack of nodules on non-inoculated plants and their rapid death.
Thirty-three isolates formed nodules on bird's foot roots during the authentication test and were initially characterized by rep-PCR (BOXA1R) analysis. The primer BOXA1R (5' CTACGGCAAGGCGA CGCTGACG 3' (Versalovic et al. 1994;Koeuth et al. 1995) was used to generate BOX-PCR profiles of isolates. Reactions were denatured at 95°C for 2 min, subjected to 35 cycles of 95°C for 30 s, 53°C for 30 s and 72°C for 8 min. A final extension at 72°C for 7 min was added. PCR mixtures were set up as per manufacturers protocol with DreamTaqTM DNA Polymerase (Thermo Fisher Scientific, USA). BOX-PCR profiles were visualized by separation on 1% agarose gel electrophoresis using a The PowerPac™ Universal power supply (Bio-Rad Laboratories, USA), using Tris-acetate-EDTA buffer. A molecular weight marker (GeneRuler 1 kb DNA Ladder, Thermo Fisher Scientific, USA) was included in all gels. Gels run at 26°C at 100 V constant voltage, until the TriTrack DNA Loading Dye (6X) (Thermo Fisher Scientific, USA) reached the bottom of the gel. The gels were stained with SimplySafe™ (EURX, Poland) (Supplement, Fig. S2). Twenty three strains representing different types of BOX were selected for further analysis with two housekeeping gene markers (glnII and dnaK) to assess the level of diversity of the selected isolates. Primers used for PCR amplification are listed in Table 1.
PCR amplification reactions were performed following the procedures described elsewhere (Stępkowski et al. 2005;Beukes et al. 2016). The reaction mixture was first denatured at 95°C for 2 min and then subjected to 35 cycles of denaturation at 95°C for 30 s, annealing at 60°C (glnII) or 55°C (dnaK) for 30 s, elongated at 72°C for 1.5 min and a final elongation step was conducted at 72°C for 7 min. In all PCR reactions DreamTaq TM DNA Polymerase (Thermo Fisher Scientific, USA) was used. PCR products were purified with an ExtractMe DNA-gel-out kit (BLIRT, Poland) and sequenced on an ABI3100 Automated Capillary DNA sequencer (Applied Biosystems, USA). Sequence chromatograms were visualized and manually edited using BioEdit and the generated sequences subjected to BLASTn searches (Altschul et al. 1990;Benson et al. 2005).

Estimation of bacterial metals tolerance
All strains were checked for Ni, Co and Cr tolerance on SLP (Jiang et al. 2008) agar plates (not shown). 1mM stock solutions of metals salts sterilized by filtration (0.20 μm) was added to sterile agar SLP medium as follow: Ni 0.2mM, 2mM and 5mM as NiSO 4 ; Co 2mM, 5mM and 10mM as CoCl 2; and Cr 2mM, 5mM and 10mM as K 2 CrO 4. Selection of the chemical compounds and concentrations used in the experiment was made on the basis of previous studies described in the literature (Rooney et al. 2007;Li et al. 2009;Wani and Khan 2013) and these metal doses were comparable to the total and/or the reactive forms in the examined ultramafic soil from which the bacteria strains were isolated. Additionally, for strains with the highest metal tolerance, the analysis of bacterial growth in liquid SLP medium supplemented with increasing concentrations of metals was performed. The strains were grown in liquid medium for 72 h at 28°C and 100 μl of 10 8 cell/ml bacterial culture were deposited onto SLP agar plates and incubated at 28°C for 24-72 h. The test in liquid SLP medium was carried out with the same metals concentrations and bacterial growth has been measured at based on the absorbance at 600 nm using a Biospectrometer (Eppendorf). Three replicates were made for each bacterial strain and each test was repeated three times.

Determination of Plant Growth Promotion (PGP) traits
All isolated strains were checked for indole acetic acid (IAA) biosynthesis, siderophore production and phosphate solubilization abilities. In the case of nitrogenase activity four strains (L9, L19, L42, L50), most effective in nodulation in the authentication test, were examined for the efficiency in nitrogen fixation. Each test was repeated twice.
IAA estimation was determined in Luria Bertani medium (LB) supplemented with 0.2% tryptophan and/or metals, Ni (0mM, 0.2mM, and 2mM, as NiSO 4 ,), Co (0mM, 2mM, and 5mM as CoCl 2 ) and Cr (0mM, 2mM, and 5mM as K 2 CrO 4 ) and bacterial isolates. After 24h of growth in 28 0 C, the cultures were centrifuged and two milliliters of supernatant of the bacteria culture (10 8 cell/ml bacterial culture) was mixed with 4mL of the Salkowski reagent and after 1h of incubation at 28 0 C the absorbance at 530 nm was measured using Biospectrometer (Eppendorf) (Star et al. 2012). The concentration of IAA was calculated using a calibration curve of pure IAA as a standard (Sigma-Aldrich).
Siderophore production was detect using chromeazurol S (CAS) agar medium (Alexander and Zuberer 1991). Bacterial cultures were grown for 24h in 28 0 C in liquid SLP medium without metals or supplemented with metals similar as for metal tolerance tests and then  Stępkowski et al. (2005) 20μl of each cultures (10 8 cell/ml bacterial culture) were spotted on CAS agar plates and grow at 28 0 C for week. The orange haloes around colonies indicates production of siderophores. Phosphate solubilization ability of bacterial strains was carried out on Pikovskaya (PVK) agar medium (Pikovskaya 1948;Zaidi et al. 2006). The plates were inoculated with 20μl of bacteria culture (10 8 cell/ml) growing in SLP medium with or without metals in doses similar as for metal tolerance tests. After one week of incubation at 28 0 C plates were checked for haloes presence. Solubilization potency was measured based on Phosphate Solubilization Index (PSI) as Total diameter of halo zone/Colony diameter (Morales et al. 2011).
Nitrogen fixation capacity was quantified indirectly by acetylene reduction assay (Hardy et al. 1968). After three months of growth in laboratory conditions cut nodules were quickly placed in 5ml glass bottles filled with 10% acetylene, incubated for 1h at 28 0 C and ethylene content was measured using a gas chromatograph with a mass spectrometer (ATI Unicam). Nitrogenase activity was expressed in C 2 H 4 nanomoles per hour for 1 gram of fresh weight of nodules. The measurement was performed in three technical repetitions.

Roots growth promotion by selected rhizobia
Roots growth promotion ability of most PGP active Rhizobium (L9) and Mesorhizobium (L42) strains were checked for host birdsfoot trefoil and Arabidopsis thaliana L. (Col-0) plants. Seeds of ultramafic L. corniculatus were prepared as described above. A. thaliana seeds were surface sterilized with 0.1% sodium hypochlorite for 5min and washed 20 minutes with distilled water. Subsequently, the both types of seeds were put on vertical placed plates with Gamborg medium (with 1% Bacto Agar) and left to grow for 5 days in constant conditions (16h photoperiod, 22 0 C temperature and photon flux density of 170 μmol . m -2. s -1 ). Next the seedlings were transferred to fresh Gamborg vertical placed plates (6 plants per plate) supplemented with metals (0; Ni 0.2mM; Co and Cr 0.5mM; doses based on Aka and Babalola 2017) and were or not inoculated with 100μl of freshly prepared L42 or L9 bacterial cultures (optical density of 0.2 at 600 nm). Growth of seedlings was assessed by measuring length of primary root and lateral roots formation after 8 days of growth. Each plate was in triplicate.

Microscopic analysis of nodules
Ultramafic L. corniculatus seeds were used to obtain control plants, which were grown for three months in laboratory conditions (14h photoperiod, photon flux density of 300 μmol/m 2 /s, 22°C/17°C day/night regime, 70% relative humidity) in pots filled with sterile keramzyt and inoculated with consortium of isolated rhizobia (L9 and L42) and watered once a week with nitrogen-free Fahreus medium (Vincent 1970).
For microscopic analysis, ten effective nodules, both from the ultramafic soil and the control plants, were used. A part of nodule hand sections were stained with dimethylglyoxime for Ni localization (Seregin and Kozhevnikova 2011) or 1% (v/v) calcofluor white to visualize cellulose under UV (Olympus-Provis, Japan). For (ultra)structural studies collected nodules were fixed in Karnovsky medium for 2h in RT, washed with 0.1M cacodylate buffer, and subjected to a routine preparation procedure for epoxy resin, described previously (Sujkowska-Rybkowska and Borucki 2012). The semithin sections stained with 1% toluidine blue or 1% alcoholic solution of safranin-O (Broda 1971) for phenolic inclusions demonstration were analysed under light microscopy (Olympus-Provis, Japan). Ultrathin sections collected on formvar-coated grids, short stained with uranyl acetate and lead citrate were examined under a transmission electron microscope (Morgagni). Digital images were saved as jpg files and if necessary were adjusted using Photoshop CS 8.0 (Adobe Systems, USA) software by non-destructive tools (levels and/or contrast) throughout the whole area of an image. The figures were prepared using the Corel DRAW 11 software (Corel Corporation, Canada).

Immunolocalization of cell wall components in nodules
Ten nodules from five plants, both from the ultramafic soil and the control were sampled, then fixed in 4% paraformaldehyde in MSB buffer for 2h in RT and subjected to a routine procedure for BMM resin embedding, described previously (Sujkowska-Rybkowska and Borucki 2015). Semi-thin sections of nodules embedded in BMM resin were transferred to silane-prep slides (Sigma-Aldrich). Some sections were stained with 1% toluidine blue for light microscopy and other were blocked with 3% bovine serum albumin (BSA) (Sigma-Aldrich) in 0.01M PBS buffer for 1h at RT. After three washing with 0.01M PBS, several primary rat monoclonal antibodies: LM19 (anti-unesterified homogalacturonan HG); LM20 (anti-partially methylesterified HG); LM25 (anti-xyloglucan); LM1, JIM11 and JIM12 (anti-extensin EXT); LM2 and JIM13 (antiarabinogalactan-proteins AGP)(Plant Probes) diluted 1:3 were used, and anti-β-1,3-glucan (anti-callose) mouse antibody (Biosupplies) diluted 1: 40 in PBS containing 0.5% (w/v) BSA was applied overnight at RT. The control consisted of nodules sections nontreated with primary antibody. Following rinsing with PBS the sections were incubated for 1,5 h at RT in PBS containing 0.5% (w/v) BSA and FITC goat anti-rat IgG antibody (Sigma-Aldrich) or Alexa Fluor 488 goat antimouse antibody (Invitrogen) diluted 1: 200. After washing with PBS the sections were closed in a drop of PBS and observed under fluorescent microscopy with used of triple DFTR filter (Olympus-Provis, Japan). The obtained digital images were processed as described above. Labeling for each antibody was carried out twice on three different nodules.

Data analysis
Analysis of significant differences among means of the metal contents in L. corniculatus organs was performed using the Student's dependent t-test. The two-way ANOVA and the post-hoc Tukey HSD test were used for analysis of significant differences among means in the trial of bacterial strains tolerance to metals. In the case of nitrogenase activity, nodule number, root length and number of lateral roots the one-way ANOVA and the post-hoc Tukey HSD test were performed. All analyses have been carried out at the level p=0.05 with Statistica software version 13 (TIBCO Software Inc. 2017). Table 2 shows the content of total and geochemically reactive Ni, Cr and C. Both forms of nickel occurred in significantly higher amounts than other metals, and generally the levels of all elements significantly varied at individual sampling points. Total content of all metals was the highest at SP1 and the lowest at SP3, and total content of Cr and Co was more diverse than that of Ni. The reactive content (0.43 M HNO 3 extractable forms)

Ni, Cr and Co content in ultramafic soil
can be considered potentially available for uptake by biota. The share of reactive Ni in its total pool was about 50% in all samples and was generally the highest among all metals. The content of reactive Cr and its share in the total pool was the lowest. The reactive Co level was higher than that of Cr but its share in the total pool varied greatly between the samples.
Ni, Cr and Co content in L. corniculatus organs and metal translocation Table 3 presents mean content of Ni, Cr and Co in shoots and roots of L. corniculatus from the ultramafic soil. Nickel was present at significantly higher levels than Cr and Co both in the roots and shoots, and the content of Ni in the roots exceeded 100 mg kg -1 . Accumulation of all metals was significantly higher in the roots than in the shoots but in the case of Co the difference was not statistically significant (t-test, p≥0.05). Translocation factor (TF) indicates a plant species tolerance strategy to high levels of metals. In L. corniculatus, its value fell below 1 for all studied elements.

Comparative analysis of rhizobial sequences
GlnII gene was amplified for 18 strains and dnaK for 21 strains. Preliminary results of sequence analysis of dnaK and glnII genes and alignment with databases deposited in NCBI GenBank revealed that the isolates represented the genera Rhizobium and Mesorhizobium (see Table 4). Blast searches of partial glnII and dnaK sequences revealed that the 18 isolates belonged to either Rhizobium or Mesorhizobium species (Table 4) Bacterial tolerance to metals The experiment on solid SLP medium demonstrated the ability of all tested rhizobia to grow in the presence of low doses of Ni (data not shown). The isolates Rhizobium (L9, L19) and Mesorhizobium (L42, L50) were most effectively growing on Ni-enriched plates and also in the presence of Co and Cr. These strains were further analyzed in liquid medium, where the influence of metals on the bacterial growth was estimated by measuring optical density (OD) at 600 nm ( Fig. 1). In Fig. 1a it can be observed significant higher metal tolerance of Rhizobium strains (L9, L19) than Mesorhizobium ones (L42, L50) to Ni at its lowest dose (0.2mM). With the intermediate Ni doses (2.0 and 5.0mM) there were no statistically significant difference among Rhizobium (L9, L19) and Mesorhizobium (L42, L50) strains. The tolerance of Rhizobium strains (L9, L19) to Co were significantly higher compare to Mesorhizobium ones (L42, L50) from low to highest Co doses (Fig. 1b). The observed significant increase in Rhizobium (L9, L19) growth at the intermediate Co doses compare to the lowest Co level (0.2mM) would suggest the higher tolerance of these strains to Co than in the case of Mesorhizobium ones. However, the general dose response of Mesorhizobium (L42, L50) strains to Co were similar to Rhizobium ones but the effects were not statistically significant. In the case of Cr, statistically significant decrease of all strains growth, both Rhizobium and Mesorhizobium, occurred compare to control. However, like after applying Co, Rhizobium (L9, L19) strains were more tolerant but the differences in relation to Mesorhizobium were not statistically significant (Fig. 1c). In both genera, the highest concentrations of Co and Cr negatively affected bacterial growth while medium doses of these metals stimulated this process after a significant growth drop at the lowest Co and Cr levels ( Fig. 1b and c).

Bacterial PGP features
In control (non-metal conditions) all strains were capable of solubilizing phosphate, fourteen strains produced siderophore and six synthesized IAA (Table 5). Four metal-tolerant strains of Rhizobium (L9 and L19) and Mesorhizobium (L42 and L50) exhibited higher PGP abilities in the presence of Ni, Co and Cr (Table 5). Under metal stress, both genera increased their production of IAA. A maximum amount of IAA (9.6μg/ml) was produced by L9 strain growing in LB broth supplemented with 5mM Cr. These strains intensified also siderophore production to make iron more bioavailable under presence of any dose of Ni. Application of 2mM Co and Cr curbed siderophore synthesis in L9 and L42 strains. Furthermore, all strains were capable of phosphate solubilization on agar medium in presence of nickel. Ni increased phosphorus solubilization potential

Nitrogenase activity and nodulation
Four strains (L9, L19, L42, L50) most effective in nodulation in the authentication test were examined for their efficiency of nitrogen fixation. Mesorhizobium strain L42 was the most efficient of all examined strains, and showed the significantly highest nitrogenase activity and nodulation per plant compare to other strains. Rhizobium strains L19 alike L9 was the lowest in nitrogenase activity as well in nodulation (Fig. 2).
Direct root growth promotion and lateral root formation of Arabidopsis thaliana and L. corniculatus colonized by Rhizobium and Mesorhizobium Two isolates, Rhizobium L9 and Mesorhizobium L42 which showed high nitrogen fixation ability, highest PGP traits and Ni tolerance were evaluated for in vivo PGP potential with their original host L. corniculatus and non-legume A. thaliana (Fig. 3). Root growth and lateral roots formation of A. thaliana and L. corniculatus were significantly inhibited by metals (Ni 0.2mM, Co and Cr 0.5mM). Inoculation by Rhizobium L9 and Mesorhizobium L42 alleviated to some extent these negative effects, and in the case of L. corniculatus, the statistically significant increase in the number of lateral roots was found for both rhizobia strains. However, it seems that Rhizobium L9 was more effective in the root growth promotion and the roots formation under metal stress. The inoculated and unexposed to toxic metals plants had longer the main root and formed more lateral roots compare to control, and these effects were statistically significant for Rhizobium L9 with only one exception of the number of roots in L. corniculatus. Mesorhizobium L42 favored the root growth and the lateral roots formation but the differences between means, in relation to control, were not statistically significant.
Structural analysis of L. corniculatus nodules The ultramafic L. corniculatus involved small plants with a thick main root and pink and apparently functional nodules formed on thin lateral roots (Fig. 4a). The nodules were determinate, more-or-less spherical with many striped lenticels on their surface (Fig. 4b). They were active in nitrogen fixation, as indicated by red color (leghemoglobin) of their interior (Fig. 4c). The plants and nodules showed no visual toxicity symptoms when they grew on ultramafic soil, and nodules accumulated Ni in nodule cortex tissues (Fig. 4d,e). Light microscope analyses of ultramafic nodules revealed typical features of determinate nodules, i.e. lack of a meristem and a homogenous population of symbiotic cells in central N 2 -fixing zone (Fig. 4f,g). Uninfected cells were devoid of starch granules (Fig. 4g).
The mixture of infected and uninfected cells in the central tissue of the ultramafic and control nodules was surrounded with non-infected cortical cells with vascular bundles (Fig. 5a,m) and numerous dark vacuolar inclusions (phenolic-like substances) were observed in metal-treated nodules (Fig. 5a). The phenolic nature of vacuolar inclusions was confirmed by safranin stain (Fig. 5b, c). The nodule cortex of both nodules contains three to five layers of cortex cells, one or two layers of dark staining nodule endodermis and five to six layers of nodule parenchyma cells (Fig. 5a,m). In ultramafic nodules, parenchyma cells Phosphate solubilizing potential (PSI*) 0mM Four level [no production (-), low (+), medium (++) and high (+++) production] were established *PSI -Phosphate solubilization index The values are presented as means ± standard deviation Plant Soil (2020) 451:  were devoid of starch granules in contrast to control ones (Fig. 5a,m,n). Ultrastructural studies confirmed that vacuoles of the cortex tissues were filled with dark phenolic-like substances (Fig. 5d-f). Intercellular spaces of the cortex tissues were large with dark inclusions (Fig. 5g-i). Some dark precipitates were also visible inside the cell wall of the nodule cortex (Fig. 5h), in intercellular spaces (Fig. 5j, k), and vacuoles of uninfected cells (Fig. 5l). In control nodules, phenolic-like substances in nodule cortex tissue occurred sporadically (Fig. 5m-o) and no dark inclusions were present in parenchyma cells or intercellular spaces (Fig. 5n). The infected cells of ultramafic nodules had one or two large central vacuoles instead of a few small ones observed in control nodules and were densely packed with multi-bacteroid symbiosomes (Fig. 6a and e). The host cell cytosol of metal-treated nodules was dense with large mitochondria in aggregates and plastids lacking starch grain. Poly-3hydroxybutyrate (PHB) granules were absent in bacteroids of metal treated nodules (Fig. 6b-Dd) in contrast to the control ones ( Fig. 6f-h). The most visible changes in nodules exposed to metals concerned cell walls. We observed in metal-treated nodules intense wall thickening of infected cells, especially in the cell corners (Fig. 6a-d). Such thick walls were not detected in the control nodules ( Fig.  6e-h). The thick walls sometimes contained electron-dense precipitates (metal ions?) (Fig. 6b,d).
Apoplast changes in L. corniculatus nodules under metal stress To reveal changes in the cell walls of L. corniculatus nodules under metal stress, epifluorescent microscopy was combined with cytochemical staining and immunocytochemical analysis with specific monoclonal antibodies against cell wall components ( Fig. 7 and 8).
In metal treated nodules we saw thickened walls (Fig. 7) that contained more cellulose than control, as confirmed by intense staining with calcofluor white (Figs. 7b and 8b). The immunofluorescence localization of pectins (LM19 and LM20 antibodies), extensin (EXT) (LM1, JIM11 and JIM12), arabinogalactan protein (AGP) (LM2 and JIM13), xyloglucan (LM25) and callose revealed that these cell wall components were strongly induced during metal stress as compared with nonmetal treated nodules ( Fig. 7 and 8). A comparative assessment of the degree of pectin methylation in nodule apoplast was carried out in this study using antibodies specific to high-methyl-esterified pectin (LM20 antibody) or low-methyl-esterified pectin (LM19). In metal-treated nodules intensely green fluorescent signal against both pectins (Fig. 7c,d) and EXT ( Fig. 7e-g) was observed mainly in thickened walls. JIM11, JIM12 and LM1 EXT antigens were present in higher amounts in the cell walls and intercellular spaces of bacteroidal tissue and intercellular spaces of nodule parenchyma in metal treated nodules compare to control ones (Figs. 7e-g and 8e-g). Regarding AGP, we studied distribution of epitopes in the nodules recognized by JIM13 and LM2 antibodies (Figs.7h, i and 8h, i). Interestingly, in metal treated nodules the AGP epitope recognized by LM2 antibody appeared not only in the thick cell wall but also in the cytoplasmic compartments of the infected cells (Fig.   7I). In control nodules (Fig. 8i), LM2 epitope labeled only cell walls of uninfected cells. In control nodules JIM13 strongly labeled only nodule endodermis (Fig.  8h), whereas in metal treated nodules JIM13 fluorescent signal was detected in cell wall and also in the cytoplasmic compartments of infected cells (Fig. 7h). LM25 strongly labeled thick walls of metal-treated nodules (Fig. 7j), whereas in control nodules the staining was less intense (Fig. 8j). In metal-treated nodules, a strong green fluorescence signal of callose was observed predominantly in plasmodesmata between the uninfected and infected cells (Fig. 7k), whereas a weaker signal was seen for these cells in control nodules (Fig. 8k). No green fluorescence signal was detected in any types of nodules without primary antibodies (Figs. 7l and 8l).

Discussion
As showed in this study, the ultramafic soil contained high amounts of Ni, Cr and Co, and Ni content was particularly high. Mean total content of Ni was over 48 times higher, and in the case of Cr and Co several times higher than in the soils of Europe (Salminen et al. 2005). The results of the soil extraction with 0.43 moldm -3 HNO 3 also revealed potentially high availability of Ni for ultramafic biota that can adversely affect both the   plants and the microorganisms. The importance of Cr and Co as potentially harmful elements seems to be much smaller than Ni because of their lower contents and less mobility. This obtained sequence of the metal extractability (Ni>Co>Cr) usually occurs in the ultramafic soils (Kazakou et al. 2008;Kumar and Maiti  2013; Kierczak et al. 2016). In comparison with other studies of ultramafic soils from the Szklary Massif, the total form levels of Ni and especially Cr were lower than obtained by other researches (Żołnierz 2007;Kierczak et al. 2016). These differences, as well as high variability of the elements content among samples, were fluorescence of cellulose is shown in blue. c-k Weak immunostaining of non-cellulosic wall components in control nodules. l Visible lack of green fluorescence signal when primary antibodies were omitted. En -nodule endoderma; IC -infected cell; Ucuninfected cell; vb-vascular bundle. Scale bar= 20μm probably mostly due to the specificity of the sampling substrate which, as the mine or smelt waste material, did not derive from the natural ultramafic outcrops with an undisturbed soil profile but were anthropogenically transformed.
Our analysis of metal content in plant organs and TF values confirmed a tendency to Ni, Cr and Co accumulation in the roots and their low translocation ability to the shoots, which suggest L. corniculatus exclusion strategy of dealing with these metals. The shoots accumulated 59.5 mg kg -1 d.w. of Ni, 5.4 mg kg -1 d.w. of Cr and 2.4 mg kg -1 d.w. of Co, and these values are much higher than the normal concentrations in plants (Markert 1994). According to Kabata-Pendias (2010), toxic levels of these metals in shoots reach 10-100 mg kg -1 d.w. for Ni, 1-2 mg kg -1 d.w. for Cr and 30-40 mg kg -1 d.w. for Co. The roots accumulated Ni, Cr and Co at 167.1, 20.5 and 8.5 mg kg -1 d.w., respectively. Despite the fact that Ni and Cr content in the plant shoots exceeded their toxic values, no toxicity symptoms were observed. The translocation factor revealed L. corniculatus tolerance strategy to high levels of toxic metals as its values fell below 1 for all studied elements. Such a defense strategy is common for metal-tolerant plants (Baker and Brooks 1989). Our results confirmed earlier findings concerning this species from ultramafic soil reported by Pędziwiatr (2018). The plants restraining metals in roots accumulate them in root endodermis and cortical tissues (Parrota et al. 2015). In our study, the presence of Ni in L. corniculatus nodules was confirmed by histochemical staining with dimethylglyoxime and revealed the nodule ability to Ni accumulation in their cortex tissues. Gopalakrishnan et al. (2015) reported that at metal contaminated sites nodules may be the main spots of metal accumulation.
Although L. corniculatus is native to several countries in Europe and is a pioneer plant on metal polluted sites, very few studies have explored diversity of its symbiotic partner on ultramafic soils (Mengoni et al. 2010;Escaray et al. 2012;Lorite et al. 2018). In this work, the preliminary characterization of the L. corniculatus nodulated rhizobia was conducted using partial sequences of two housekeeping genes (glnII and dnaK). BLASTn and Blast Microbes searches revealed a phylogenetic affinity within the genera Rhizobium and M e s o r h i z o b i u m . F o u r s t r a i n s r e p r e s e n t e d M. septentrionalea species reported in the nodules of L. corniculatus plants growing in non-metal contaminated Norwegian soils (Gossmann et al. 2012). There is also a report concerning R. leguminosarum isolated from bird's foot trefoil nodules from Swedish soils (Apomah and Huss-Danell 2011). Most importantly, this work uncovered a novel lineage within the genus Rhizobium, and presumably also a new metal-resistant species in the genus Mesorhizobium. These strains will be subjected to further, more detailed phylogenetic and genomic characterization.
Studies on ultramafic soils indicated that only a small group of rhizobia inhabiting nickel-contaminated soils developed resistance mechanisms against this toxic metal (Pereira et al. 2006;Abou-Shanab et al. 2007;Pal et al. 2007;Chaintreuil et al. 2007;Fagorzi et al. 2018). In our study, Rhizobium (L9, L19) and Mesorhizobium (L42 and L50) strains most effective in nodulation on ultramafic soils showed multi metal tolerance. They were tolerant not only to Ni (up to 0.2 mM), but also to Co and Cr (up to 5 mM). Schmidt and Schlegel (1994) showed that regarding the Co and Ni resistances, it is frequent the appearance of multiresistance since the nickel-cobalt resistance genes are carried by the same plasmid. A recent Porter et al. (2017) genome-wide association study identified candidate genes for adaptation of Mesorhizobium strains to nickel in ultramafic soils. Delorme et al. (2003), Pereira et al. (2006) and R u b i o -S a n z e t a l . ( 2 0 1 8 ) r e p o r t e d t h a t R. leguminosarum bv. viciae isolated from metal polluted sites exhibited abroad spectrum of metal resistance. In our study, we observed an increased growth of R. leguminosarum and M. septentrionale at the intermediate Co and Cr doses. It would suggest the high tolerance of these strains to these metals. Similar stimulation effect of medium metal doses on rhizobia growth was observed in our previous studies (Sujkowska-Rybkowska and Ważny 2018;Sujkowska-Rybkowska et al. 2020). Delorme et al. (2003) indicated that in polluted soils, the tolerance of bacteria to one or more metals may be due to the existence of distinct, independent metal tolerance mechanisms in their organisms. Other reports show a presence of metal tolerating Bradyrhizobium (Wani et al. 2007), R. leguminosarum (Rubio-Sanz et al. 2018 and Mesorhizobium strains isolated from nodules of legume plants growing on nickel contaminated soils (Porter and Rice 2013). It has been suggested that rhizobia possess different Nitolerance systems (Chaintreuil et al. 2007;Fagorzi et al. 2018). Several studies show that bacteria inhabiting ultramafic soils developed different metal tolerance mechanisms to maintain low intracellular metal concentrations (Chaintreuil et al. 2007;Haferburg and Kothe 2007;Zielazinski et al. 2013). Rhizobial exo-and lipopolysaccharides influence metal resistance by forming complexes with metal ions through electrostatic interactions (Liu et al. 2001). Ensifer, Mesorhizobium and Bradyrhizobium strains have active Ni efflux systems (Chaintreuil et al. 2007;Zielazinski et al. 2013). Some highly tolerant R. leguminosarum strains accumulate metals in intracellular spaces (Purchase et al. 1997;Purchase and Miles 2001;Pereira et al. 2006). Mesorhizobium bacteria have active Ni efflux systems to prevent toxic intracellular nickel concentrations (Maynaud et al. 2013). On the other hand, rhizobia survival in metal contaminated soil can be related to physical protection of organic matter and clay minerals (Ibekwe et al. 1997;Giller et al. 1998) or to the existence of specific "niches" in the soil such as root nodules, where metal contamination may be minimal.
Our study indicated that apart from metal resistance isolated strains of Rhizobium (L9, L19) and Mesorhizobium (L42 and L50) exhibited high PGP abilities under metal (Ni, Co or Cr) presence (Table 5). These strains were capable of producing high amounts of auxin and siderophore and showed phosphate solubilizing ability under metal stress. Many reports have provided evidence that nodulating rhizobia may induce plant growth directly by nitrogen fixation, phosphate solubilization, siderophore or phytohormone synthesis and ACC deaminase production (Stearns et al. 2005;Ma et al. 2011;Gopalakrishnan et al. 2015;Karthik et al. 2017). Rhizobial PGP traits improving growth of legumes in metal contaminated soils were observed for Bradyrhizobium under Cd stress (Guo and Chi 2014), and Rhizobium nodulating Prosopis juliflora on metal-contaminated soil (Rai et al. 2004). Wani et al. (2008) showed similar effects of Mesorhizobium strain on chickpea grown in Cr contaminated soils. Phosphorus solubilization was reported for Rhizobium sp., and Mesorhizobium sp. (Vessey 2003;Gopalakrishnan et al. 2015). Rhizobium, Sinorhizobium and Bradyrhizobium sp. are known to produce siderophores (Carson et al. 2000;Arora et al. 2001). It therefore seems that Rhizobium and Mesorhizobium strains may improve L. corniculatus growth and increase plant stress tolerance on ultramafic soils.
In this study, we also showed a direct promotional effect of R. leguminosarum bv. viciae (L9) and M. serpentionale (L42) on L. corniculatus and nonlegume Arabidopsis thaliana (Fig. 3). Seedling root growth and lateral root formation in A. thaliana and host plants was promoted by inoculation with both strains, especially under metal presence. Our findings confirmed the tests for PGP abilities of isolated rhizobia and indicated the existence of beneficial interactions of rhizobia not only with legume plants. Apart from the ability to form nitrogen-fixing symbiosis with legumes, many rhizobia are known to develop growth promoting associations with non-legumes, such as maize, rice (Gutiérrez-Zamora and Martínez-Romero 2001; Ladha and Reddy 2003), and Arabidopsis (Zhao et al. 2017). In metal contaminated soil rhizobia may be an important microflora not only for their host but also for neighboring plants growing in these difficult conditions.
The ultramafic legumes should have specific mechanisms of tolerance and adaptation to these soil conditions, the more so because there is a strong evidence from pot experiments that some metals, especially Ni, affect nitrogen fixation in legume nodules and significantly limit their yield (Athar and Ahmad 2002). To understand possible adaptive mechanisms of Lotus-rhizobia symbiosis under metal stress it is necessary to determine cellular organization of the symbiotic nodules. However, there are few studies that discuss the effects of toxic metals on nodule structure and functioning at their natural sites (Ibekwe et al. 1995;Wani and Khan 2013;Sujkowska-Rybkowska and Ważny 2018). Nodules represent a unique environmental niche for symbionts accommodation and proliferation by providing microsites in metal-contaminated soils where toxic influence of metal ions is limited. In this study, we show that Ni is detained in ultramafic nodule cortex tissue, which was shown with dimethylglyoxime dyeing. Interestingly, in this study electron microscope analysis revealed structural alterations (wall thickening and phenols accumulation) in the ultramafic L. corniculatus nodules that may prevent negative effects of metals on the symbiosis. We found clear differences in the structure between control and metal-treated nodules. L. corniculatus under control and metal stress conditions formed determinate nodules typical of the tribe Loteae (Brewin 1991). The central part of nodules accounted bacteroidal tissue with uninfected and infected cells with bacteroids. We noted that the bacterial conversion to bacteroids in metal-treated L. corniculatus nodules was associated with a lower PHB accumulation compare to control nodules. A similar PHB decrease was observed in L. corniculatus nodules after forage harvest (Vance et al. 1982). In metal-treated nodules infected and uninfected cells were devoid of starch granules. Lower starch accumulation in nodules exposed to metals could be a result of low C supply from the plant to the nodule caused by metal stress since photosynthesis is impaired by metals (Paunov et al. 2018). The ultramafic and control nodules were surrounded by a multilayer cortex consisting of parenchyma for gas diffusion, endodermis for ion transport and outer cortex serving as a barrier against pathogens and toxic ions (Brewin 1991). The vacuoles of the nodule endodermis and some cortical and parenchymal cells were filled with dark phenolic substances, the nature of which was confirmed by safranin staining (Broda 1971). These phenolic vacuolar inclusions could be flavolans (condensed tannins) that are naturally accumulated in L. corniculatus nodules (Monza et al. 1992). In metal treated nodules we observed increased number of cortex cells filled with tannins. The adaptation of L. corniculatus plants to metal stress could be attributed to enhanced synthesis of phenolic compounds. The protective role of phenols against toxic metals consisting in metal chelation and scavenging harmful reactive oxygen species generated in the presence of toxic ions is well known (Sharma et al. 2012). Lafuente et al. (2015) revealed the induction of the synthesis of phenolic compounds (phenylpropanoids, isoflavonoids) in nodules of Medicago plants in the presence of arsenic. Recent studies of Raklami et al. (2019) showed that in nodules of M. sativa plants inoculated with metal resistant strains and cultivated in the presence of metals the gene coding phenylalanine ammonia lyase is expressed. This enzyme is the start point of two important pathways: secondary metabolism of phenolic compounds and lignin synthesis.
In ultramafic nodules we noted significant wall thickening of the infected cells, especially in their corners. Plant cell wall is the first barrier against toxic metals and cell wall construction is very important in plant resistance to abiotic stress (Le Gall et al. 2015). The cell wall structure comprising cellulose microfibrils and non-cellulosic neutral polysaccharides embedded in pectin matrix with proteins and phenolic compounds, confers metal binding ability (Probst et al. 2009;Krzesłowska 2011). Cell wall thickening is one of metal exclusion strategy in plants, and apoplast (cell walls and intercellular spaces) is the main site of toxic ion sequestration (Zornoza et al. 2002;Krzesłowska 2011). Thickening of the cell wall helps to increase mechanical strength and raises its binding capacity of metal ions thus preventing their entry into the protoplast (Krzesłowska 2011;Le Gall et al. 2015). This leads to modifications in cell wall structural properties (Krzesłowska 2011;Le Gall et al. 2015). In our experiment, some precipitates visible in metal-treated nodule apoplast could be metal ions sequestrated in the walls and intercellular spaces. However, such metal distribution should be confirmed by more advanced techniques, such as Energy Dispersive X-ray (EDX) microanalysis or X-ray fluorescence (XRF) imaging, which are planned at further stages of our research.
We used immunohistochemical method with different antibodies directed against wall components to analyze changes in the nodules cell wall composition under metal stress. We observed apoplast remodeling in metal-treated nodules, which may be considered as indicators of adaptation strategy of L. corniculatus-rhizobia symbiosis to metal stress conditions. Cell walls of metal-treated nodules were thickened and contains more cellulose and hemicellulose compare to control nodules, as indicated by dyeing with calcofluor and LM25 antibody labelling ( Figs.6-8). The higher deposition of these wall components is commonly observed in thick secondary walls (Le Gall et al. 2015). Recent studies of Tsyganova et al. (2019) revealed that highly methylesterified pectins (detected by JIM7 and LM20 antibodies) showed a uniform localisation in the cell walls of non-metal treated M. truncatula and Pisum sativum nodules, whereas low methylesterified pectins recognised by JIM5 and LM19 localized mainly in the walls of infection threads. In this study, in metal-treated nodules thickened walls of infected cell, especially their corners, were intensively labelled with pectins antibodies (LM19 and LM20). The increase of pectins, especially low-methylesterified in walls of metal treated roots (Krzesłowska 2011) and nodules (Sujkowska-Rybkowska and Borucki 2015) was often observed. Pectins together with hemicellulose play important role in metals binding (Krzesłowska 2011;Le Gall et al. 2015). Wall thickenings might be a result from stiffening of the polysaccharide network. De-esterified pectins can be cross-linked by calcium ions or metal ions what impacting on cell wall mechanical properties and metal immobilization (Krzesłowska 2011). In our studies, in addition to pectins the thick walls of infected cells of metal-treated nodules contains also higher amount of EXT (recognized by LM1, JIM11 or JIM12 antibodies) compare to control. Additionally, intercellular spaces of uninfected cell and cortex tissue of metal-treated nodules were significantly filled with these structural glycoproteins. EXTs are responsible for cell wall stiffening (Kieliszewski and Lamport 1994). In root nodules, this glycoproteins intercellular localization in nodule parenchyma play role as diffusion barrier that helps protects nitrogenase from inactivation (Vanden Bosch et al. 1989;Rathbun et al. 2002). Similar occlusions in the intercellular spaces have been reported in nodules under different stresses (de Lorenzo et al. 1993;Iannetta et al. 1995;Sujkowska-Rybkowska and Borucki 2014). Such modifications might render cell walls less permeable to metals thus limiting its entry in the cell interior and bacteroidal tissue. In present study thick walls and cytoplasmic compartments of infected cells of metal treated nodules show also presence of AGP (recognized by LM2 antibody). Such AGP cytoplasmic localization may indicate an extensive internal membrane system (Šamaj et al. 2000). LM2 epitope is closely associated with endomembranes of the endoplasmic reticulum, Golgi apparatus and tonoplast of different plant cells (Šamaj et al. 2000). These glycoproteins are involved in different cellular and developmental processes and response to different biotic and abiotic stresses (Lamport et al. 2006). AGPs may be transported as a compartments of the endomembrane flow participating in AGP exocytosis to apoplast, where AGP play role as pectin plasticizers (Lamport et al. 2006). AGP may reducing the formation of pectate gels and affecting cell wall extension (Serpe and Nothnagel 1994). AGPs are also localized in the infected cells of legume nodules and play an essential role in symbiotic tissue (Cassab 1986;Fruhling et al. 2000). In pea nodules, the only other nodule examined to date, JIM13 epitope recognized AGP was associated with nodule lignified endoderma (Rae et al. 1991). Surprisingly, metal treated L. corniculatus nodules also showed increased amount of AGP recognized by JIM13 antibody not only in nodule endodermis but also in cytoplasmic compartments of infected cells. This may indicate the presence of new bacteroidal tissues-specific AGP epitopes under metal stress. Metal treatment also induced in nodules accumulation of callose in plasmodesmata between infected and uninfected cells. Callose regulates the plasmodesmata permeability and limits metal ions movement into the cytoplasm (O'Lexy et al. 2018). Taking together, the results suggest that L. corniculatus nodules respond to Ni, Cr and Co with changes in the cell wall components (celullose, hemicellulose, pectins, EXT, AGP and callose), which may limit the influx of harmful ions to the cytoplasm of infected cells with symbiosomes and act protectively on L. corniculatus-rhizobia symbiosis in the presence of these metals.

Conclusions
Our results indicate for the first time that L. corniculatus nodules are inhabiting by beneficial metal-tolerant, growth promoting rhizobacteria, which have the potential to improve plant growth and alleviate Ni, Cr and Co stress on ultramafic soils. In addition, in these nodules, the accumulation of phenols and remodeling of apoplast can counteract the harmful effects of these metals on both bacterial symbiont and host. These findings imply that L. corniculatus-rhizobia symbiosis is an important element of plant adaptation to the metal stress occurring on ultramafic soils.