Introduction

Selenium (Se) is an essential micronutrient required in humans and animals for the function of a number of selenium-dependent enzymes, such as glutathione peroxidase (GPx) and thioredoxin reductase [1]. It has been also demonstrated that certain Se-containing compounds (e.g., methylselenocysteine, γ-glutamyl-methyl-selenocysteine) are effective chemoprotective agents, reducing the incidence of certain types of cancer [2, 3]. However, at high dose, Se may be toxic to mammals [4, 5]. The concentration range between physiological requirement and lethality is very narrow, as consequences both Se deficiency and toxicity are common worldwide problems. Plants can play a pivotal role in this respect: For example, the abilities of plants to absorb, sequester, and volatilize selenium can be exploited to manage environmental Se contamination by phytoremediation strategies [6, 7]. On the other hand, Se-enriched plants may be useful as “Se delivery system” (in forage and crops) to supplement the diets of livestock and humans in Se-deficient regions [8, 9]. Moreover, the ability of some plants to transform selenium into bioactive compounds (e.g., anticarcinogenic compounds) has also important implications for human health [10–12].

The correlation of bioavailability and toxicity of Se with its chemical form triggered interest in Se speciation in plants. Actually, in plants, Se supplied as inorganic form (low bioavailability, potentially toxic) could be converted to safer highly bioactive species, improving nutritional properties [4]. Nowadays, to increase Se concentration in crop plants on Se-deficient soils, it has become a widespread agricultural practice in many countries to use Se-enriched fertilizers. This form of agronomic biofortification is for example successfully practiced over 30 years in Finland, New Zealand, and Denmark [8, 12, 13]. Some chemical forms of selenium are more readily available to plant uptake than others, for example in soil, selenite, being more retained on solid phases, is less bioavailable than selenate [14]. Apart from the total Se uptake, different enrichment treatments normally undergo certain metabolic changes, leading to several final products showing diverse translocation and accumulation within plant compartments [15–17]. For this reason, it is important to evaluate which form of Se should be used for plant supplementation to obtain high content of this element in final plant and to know whether inorganic Se is efficiently assimilated into organic forms. Therefore, its distribution in different parts of the plant, as well as characterization and quantification of individual chemical species, including bioactive compounds, become an issue. Very few papers have been published so far on speciation analysis of Se in plants, and most of the reported works have been performed with Se hyper-accumulator plants (i.e., Brassica, Allium) which, being able to accumulate several thousand mg kg−1 of Se, were shown to be good candidate to investigate the formation of the different Se compounds after plant exposure [18, 19]. However, information available on accumulation and speciation of Se in non-accumulator plants (i.e., forages and grasses), which could serve as model for general environmental conditions, is still scarce, probably due to the lack of analytical approaches, allowing the detection of Se species at very low concentrations. Liquid chromatography coupled to inductively coupled plasma mass spectrometry (HPLC/ICP-MS) is a widely accepted tool for screening and quantification of elemental species in plant tissue [20–22], but the detection and the accurate quantification of the separated Se species are not always possible, owing to the fact that species concentrations are below the detection limit. The low sensitivity for Se in ICP-MS is partly due to several interferences on major Se isotopes and the presence of many stable isotopes (74Se (0.89 %), 76Se (9.36 %), 77Se (7.63 %), 78Se (23.4 %), 80Se (49.6 %), and 82Se (8.73 %)). To overcome this issue, an attractive way to enhance the sensitivity of HPLC/ICP-MS measurement for Se and to get reliable species determination consists in the application of highly enriched stable isotopes as tracers in metabolic studies. Metabolites of a precursor labelled with an enriched stable isotope can be much more easily detected than Se with natural abundance. In addition, several nutritional forms of Se labelled with different isotopes can be simultaneously supplied into the growth media; consequentially, metabolites of different precursors can be traced in a single experiment. Various factors such as absorption rate, translocation and speciation as well as potential competition mechanisms between the added species can be thus compared using identical host plant under the same growing and detection conditions. However, if this isotopic tracer approach has been successfully applied in rats [23–25], and in environmental samples such as soils and sediments [26, 27], it has, to the best of our knowledge, never been performed in order to monitor Se biotransformations in plants.

In this study, a non-hyperaccumulator plant, ryegrass (Lolium perenne L.), was chosen as model system for being an easy and fast to grow plant. Moreover, ryegrass is one of the most important forage varieties in terms of contribution to pastures composition and an increase of Se accumulation has been previously evidenced in greenhouse experiments, as well as non negative effect on plant yield, when Se was supplied as selenite or selenate forms at low concentration [28]. Although there are few studies on the determination of total Se in ryegrass [29, 30], Se speciation in this plant has not been reported to our knowledge. We used two stable isotopes, 77Se and 82Se, to label respectively selenite and selenate sources added to hydroponic growth media (double spike) in order to monitor respective bio-incorporation of these two species in roots and leaves of ryegrass. HPLC/ICP-MS was used for Se species determination after water extraction and enzymatic hydrolysis of freeze-dried samples. The objectives were to determine both the capacity of ryegrass to take up and accumulate the Se inorganic forms and the corresponding metabolic products in different plant tissues.

Materials and methods

Instrumentation

Total Se concentration was determined with an Agilent 7500ce ICP-MS instrument (Agilent Technologies, Santa Clara, CA) equipped with an octopole collision/reaction cell (C/RC). The setting and acquisition parameters were previously optimized in our laboratory [27] (Table S1 in the Electronic Supplementary Material, ESM). An Agilent 1100 series HPLC pump was coupled to ICP-MS for speciation analyses. Chromatographic separation was carried out on a porous graphitic carbon stationary phase (PGC, Thermo Hypercarb column 10 cm × 4.6 mm i.d) with a formic acid mobile phase (240 mmol L−1, pH 2.4 adjusted with ammonia), delivered at 1 mL min−1 flow rate and a gradient program with increasing amount of methanol (from 1 to 35 %, as detailed in Table S1 in ESM). The column used for size exclusion chromatography (SEC) was a Superdex Peptide 10/300 GL (Tricorn, Amercham Biosciences) with a fractionation range of 0.1–7 kDa and a total volume of 24 mL. The elution phase was a 30-mM Tris–HCl buffer (pH 7.5) at a flow rate of 0.6 mL min−1. The calibration of the column was accomplished with a standard mixture of myoglobin (17 kDa), insulin (5.8 kDa), and vitamin B12 (1.35 kDa), with UV detection at 254 and 280 nm, giving a calibration curve log(MW) = 6.291—0.155(tr), where tr is the retention time in minutes and MW the molecular weight in Da.

SEC fractions were collected and analyzed off-line for Se total content and its speciation by PGC-HPLC/ICP-MS. For water extracts (roots and leaves), the eluate was collected every 3 min between 11 and 32 min of elution, leading to 7 fractions for each sample type. For driselase extract of roots, the eluate was collected every 3 min between 15 and 36 min (7 fractions), and for driselase extract of leaves between 21 and 32 min (4 fractions). The collection of fractions was repeated 6 times, and corresponding fractions were pooled (final volume of each fraction about 11 mL). The injection volumes were of 100 and 200 μL for RP and SEC chromatography, respectively.

Reagents and standards

Selenium stock standard solutions of 1000 mg L−1 were prepared in water and stored in the dark at 4 °C. D,L-Selenomethionine (SeMet), L-selenocystine (SeCys2), methylselenic acid (MeSeOOH), sodium selenite (Se(IV)), sodium selenate (Se(VI)) and Se-methylselenocysteine (MeSeCys) were obtained from Sigma-Aldrich (SigmaUltra, ≥99 %). Working standard solutions were prepared freshly prior to use by diluting the respective stock solutions in ultrapure water. All reagents used for mobile phases (formic acid, methanol, and ammonia) were of analytical grade and purchased from Sigma-Aldrich (St. Louis, MO, USA). Nitric acid (Ultrex, J.T. Baker, Deventer, The Netherlands) and hydrogen peroxide 30 % (Fisher Scientific) were used for mineralization procedure. Tris(Hydroxymethyl)aminomethane, protease type XIV from Streptomyces griseus and driselase from Basidiomycetes used in extraction procedures were from Sigma-Aldrich. Highly enriched elemental 77Se(0) and 82Se(0) powders were purchased from Isoflex (Moscow, Russia).

Preparation of isotopically enriched Se tracers (77Se(IV) and 82Se(VI))

Elemental Se was converted into Se(IV) and Se(VI), respectively, by nitric acid and a combined nitric acid/hydrogen peroxide oxidation reaction, according to Van Dael et al. [31]. Both standard solutions were characterized in terms of isotopic composition and of concentration by isotope dilution. Speciation analysis of the isotopically enriched Se standards was also performed to check complete conversion of elemental selenium into selenite and selenate. The results concerning the enriched solution are as follow: [77Se]: 1015 ± 3 mg kg−1, [77Se(IV)]: 1010 ± 6 mg kg−1, 98.3 % 77Se and 1.7 % 78Se and [82Se] 947 ± 4 mg kg−1, [82Se(VI)]: 950 ± 5 mg kg−1, 99.2 % 82Se and 0.8 % 80Se.

Plant growth and samples preparation

Ryegrass (Lolium perenne L.) plants (two replicates per treatment) were cultivated in a controlled atmosphere growth chamber (1 m2 surface area and 1.5 m height) with a 10-h per day period (two 400 W metal halide discharge lamps, Growth spectra, MH400W, E40), a temperature of 25:23 °C (day/night), and relative humidity of 60–70 %. Dried seeds of English ryegrass (30 mg cm−2) were planted on mesh screen (mesh diameter of 1 mm) stretched out at the top of white plastic container. The plants were grown in duplicate in two plastic tanks filled with 16 L of hydroponic solution (a modified Hoagland nutrient solution [32]).

After 2 weeks of plant growth, 77Se(IV) and 82Se(VI) tracers were simultaneously supplied in the medium at an initial concentration of approximately 100 μg L−1 of Se for each species. This concentration was chosen based on previous work to optimize growth conditions (unpublished data) showing that ryegrass growth is threatened by stunting, chlorosis, withering, and necrosis of leaves above 200 μg L−1. Hydroponic solutions were replaced every 2 weeks with a fresh nutritive solution containing Se tracers at the same initial concentrations (about 100 μg L−1). As the ryegrass plants grew up, the evapotranspiration gradually increased (up to 140 ml per hour), and the addition of deionized water was required to ensure stable nutrients concentrations. The water condensates drained out of the growth chamber were measured each day, and an equivalent volume of deionized water was therefore added in the plastic tanks.

As consequence of plant uptake, the tracer concentrations decreased during plant growth, aliquot of nutrient solutions were therefore analyzed before and after replacement (every 2 weeks) in terms of total Se content and speciation. After 6 weeks, the plants were harvested and the roots were separated from the leaves. Roots were then washed with ultra-pure water to remove all the traces of nutrient solution. Both tissues, from each culture repetition, were freeze-dried, ground with an automatic agate mortar obtaining four powdered homogenized samples (2 samples/plant compartment) and weighted (dry weight, DW).

Procedures

Total digestion

For the determination of total Se in L. perenne L., 0.250 g of powdered homogenized samples was digested with 2 mL of HNO3 and 0.5 mL of H2O2 in a closed vessel microwave digestion system (Ethos Touch Control, Milestone) assisted by a two-step temperature program (5 min up to 180 °C and held for 10 min, working with a maximum power of 1000 W). Finally, the digested samples were made up to 30 mL with ultra-pure water and filtered to 0.45 μm and analysed by ICP-MS. The four samples were digested and analysed in triplicate.

Validation of procedure was performed by digesting a certified reference material (Ryegrass ERM-CD281) measured total Se (0.022 ± 0.001 mg kg−1 DW) being in complete agreement with theoretical value (0.023 ± 0.004 mg kg−1 DW).

Sequential extraction protocol

Selenium species were extracted from the plant tissues using a sequential extraction protocol including the following steps: (1) ultra-pure water, (2) driselase (mixture of cell wall digesting enzymes), and (3) protease type XIV. For each step, different extraction conditions were tested and the optimized protocol was defined based on a compromise between extraction yield, species stability and time consuming.

According to the results of the optimization experiments (described below), the retained sequential extraction protocol for further analysis was as follow: about 0.1 g plant material was subjected to:

  1. 1.

    Extraction with 5 mL of ultra-pure water assisted by ultrasonic probe (SONICS VibraCellTM) (25 % of energy, 20 s pulse,) for 4 min, repeated three times, for extraction of soluble selenium compounds. The supernatants collected at each extraction step were then pooled for analysis.

  2. 2.

    Enzymatic hydrolysis with driselase: the residue from water extraction was extracted twice with 5 mL of Tris–HCl buffer solution (8 mM, pH 7.5) containing 10 mg of driselase. The solution was kept at a constant temperature of 37 °C in a thermostat bath and constantly stirred for 24 h. The supernatants were then pooled for analysis.

  3. 3.

    Enzymatic hydrolysis with protease: 5 mL of a Tris–HCl buffer solution (8 mM; pH 7.5) containing 1 mM CaCl2 and 20 mg of protease type XIV was added to the pellet from driselase extraction. The solution was incubated in a digitally controlled immersion thermostat bath for 20 h at 37 °C. After centrifugation, the supernatant was digested once again with 1 mL of Tris buffer solution containing 10 mg of protease XIV for 20 h at 37 °C.

For extracts obtained from enzymatic hydrolyses (driselase and protease), an aliquot of the solution was discarded for total analysis and the remaining was cut off filtered through a 100-kDa pore size membrane to eliminate high weight molecular compounds that might disturb chromatographic separation. Total Se content was determined before (noted as extracted 77Se or 82Se) and after cutoff filtration (noted as extracted 77Secut-off and 82Secut-off).

Samples were sequentially extracted in triplicate, and a blank was subjected to the same procedure. Protease residues were mineralized for the determination of non-extracted Se allowing mass balance calculation. These precautions allowed to control that no contamination or loss occurred during extractions.

Extraction efficiencies are calculated as the percentage of total Se content in the freeze-dried plant tissues (DW).

Determination of Se in the digested samples and extracts

Endogenous and tracers (77Se and 82Se) Se concentrations, for total and speciation analyses, were determined by reverse isotope dilution (RID), with natural abundance Se standard used as a spike and, 78Se/77Se and 78Se/82Se ratios selected as measured isotopes pairs by using Eqs. 1 to 3 according to Hintelmann and Evans [33].

$$ C=\frac{C^{\hbox{'}}\times {w}^{\hbox{'}}\times {W}_{\mathrm{at}}\times \left(R\kern0.5em \times \kern0.5em {}^{77-82}A_{\mathrm{natural}}{-}^{78}{A}_{\mathrm{natural}}\right)}{w\times {W}_{\mathrm{at}}^{\hbox{'}}\times \left({}^{78}A_{\mathrm{sample}}-R\times \kern0.5em {}^{77-82}A_{\mathrm{sample}}\right)} $$
(1)

With:

C: Se concentration in the sample

C′: Se concentration of standard (natural abundance) spiked to the sample

w, w′: masses of sample and spike

W at ,W at′: atomic masses of sample and spike

R: 78Se/77Se or 78Se/82Se ratio measured after spike addition

XAsample : abundance of X isotope in the sample before spike addition

XAnatural: abundance of X isotope in the natural standard used as spike

$$ \left[S{e}_{\mathrm{ambient}}\right]=\frac{C\times {\kern0.5em }^{78}{A}_{\mathrm{sample}}}{{}^{78}{A}_{\mathrm{natural}}} $$
(2)
$$ \left[S{e}_{\mathrm{tracer}}\right]=\frac{C\kern0.5em \times {}^{78}A_{\mathrm{sample}}-\left[\left(\frac{{}^{78}{A}_{\mathrm{natural}}}{{}^{77-82}A_{\mathrm{natural}}}\right)\times \kern0.5em C\kern0.5em \times {}^{77}A_{\mathrm{sample}}\right]\kern0.5em }{\left(\frac{{}^{78}A_{\mathrm{tracer}}}{{}^{77-82}A_{\mathrm{tracer}}}\right)-\left(\frac{{}^{78}A_{\mathrm{natural}}}{{}^{77-82}A_{\mathrm{natural}}}\right)} $$
(3)

With:

XAtracer: abundance of isotopes 78 and 77 or 82 of tracers (i.e., 77SeIV or 82SeVI) and natural abundance Se

Optimum spike-to-sample ratio was calculated for each sample to obtain minor error propagation during ID calculation [34].

All isotopes intensities were mathematically corrected from interferences due to the formation of SeH+ and BrH+ following the determination of Se and Br hydridation factors (fSe and fBr) in Se and Br standards solutions measured each three samples (bracketing method; [27, 35]).

All isotope ratios obtained were further corrected for mass bias using an exponential model after determination of the mass bias factor with natural abundance Se standards [36]. C/RC conditions and correction method defined in Tolu et al. [27] were evaluated for reagents used in the present work showing a good accuracy and precision of measured isotopic ratios (error < 4 %; RSD < 5 %) [27].

Se concentrations were then expressed in micrograms of Se per kilogram based on precise weights of vegetal material (DW) and reagents used for extraction or mineralization.

Total uncertainties associated to concentrations of total Se and Se species were defined by combining analytical uncertainties to those related to the mean and standard deviation of values of extractions replicates (n = 3). Analytical uncertainties were calculated according to random error propagation method along the calculation steps [27].

Values referring to the “whole plant” were calculated from Se tracer concentrations measured in each compartment (μg g−1 of Se) and corresponding compartment biomass (g).

Results and discussion

Nutritional solutions

Total concentrations of the two Se tracers were determined in aliquots of freshly prepared nutritional solution and solution after two weeks of growth. At the beginning, Se concentrations of both tracers amounted approximately to 100 μg L−1 (Table S2 in the ESM). After 2 weeks of plant growth, initial concentrations of both tracers decreased of about 60–66 % as result of plant uptake. Speciation analyses of nutritional solutions were performed as well in order to check eventual transformations of supplied Se species occurred during the considered period. Results resumed in Table S2 in the ESM showed that the tracers were stable in terms of speciation in the growth media remaining after 2 weeks of culture in the chemical form initially used for Se enrichment.

Comparative efficiencies of different extraction procedures

In order to monitor respective bio-incorporation of inorganic Se and elucidate possible further metabolism in root and leaf compartments, samples were sequentially extracted by reagent solutions designed to selectively leach different classes of selenium species. In this study, a sequential approach was designed consisting of extraction of (1) water soluble Se species, (2) Se bound to plant cell wall and (3) Se-amino acids incorporated into protein structure. Besides the maximum efficiency, the stability of Se species during the procedure should be taken into account when choosing extraction conditions. A preliminary study was thus performed to select appropriate extraction procedures in order to obtain maximum extraction efficiency as well as to ensure integrity of species. The first step of sequential extraction aimed to recover water-soluble Se species. Three different water extraction protocols were tested as follow: 0.1 g sample was extracted either once, twice or three times with 5 mL of ultra-pure water in an ice bath using an ultrasonic probe. In the case of repeated extractions, the supernatants were collected after centrifugation before further extraction(s). All supernatants were pooled together before analysis.

Figure 1a shows the results obtained considering total Se extractabilities (expressed as % Seextracted/Setotal). For both samples and both tracers, Se recovery after three successive water extractions was almost twice higher compared to single extraction; it was thus selected as first step of sequential extraction protocol.

Fig. 1
figure 1

Proportion of extracted Se (% Seextracted/Setot) under tested conditions used for: a water extraction and b driselase extraction. Error bars represent SD for triplicate extractions of two samples/plant compartment

Full size image

Comparison of total Se measured in extracts from once and twice repeated digestions with driselase evidenced a significant increase of extractability after two successive extractions with driselase (Fig. 1b) which were therefore retained for the sequential extraction protocol. A proteolytic digestion was performed on driselase residue as last step. Most widely reported conditions used amount of 30 mg protease type XIV for digestion of 0.1 g of plant material performed in 5 mL of Tris–HCl buffer solution (8 mM; pH 7.5) [37, 38]. We compared three different protocols to bring that amount; each protease addition was incubated 20 h at 37 °C: (1) single step digestion with 30 mg of enzyme, (2) first digestion step with 20 mg of enzyme followed by further digestion of supernatant adding 1 mL of buffer solution containing 10 mg of enzyme, and (3) first digestion step with 10 mg of enzyme followed by two successive digestions of supernatants obtained by adding 1 mL of buffer solution containing 10 mg of enzyme.

Differences in extracted selenium tracer recoveries were not statistically significant among the three procedures (RSD: 1–5 %). However, when the extraction procedure involved successive protease hydrolysis of supernatant, an increase in SeMet concentration was observed (20–60 % higher than one step digestion). A procedure including at least one further proteolytic digestion of supernatant seems thus to provide complete digestion of proteins and consequently release of seleno-amino acids.

For each step of extraction procedure, integrity of species during treatment (especially when applying successive ultrasonic rounds) was also verified. HPLC-ICPMS analyses of water and enzymatic extracts showed identical Se-containing peaks after successive extraction steps, indicating that no species degradation occurred during the treatments. Relative standard deviations obtained for total Se determinations were lower than 4 % for the selected procedures. Moreover, HPLC-ICPMS analyses of extracts showed superimposable chromatograms for triplicate treated samples, confirming extractions procedures repeatability (RSD < 5 %).

Allocation of 77Se and 82Se tracers in plant tissues

Total concentrations of both tracers and endogenous Se were measured in freeze-dried samples after digestion for each tissue type (roots and leaves). Total Se accumulation in the whole plant is the same for both tracers, accounting for 7.6 ± 0.8 and 7.3 ± 1.2 mg kg−1 for 77Se and 82Se, respectively. These results indicate that, in used experimental conditions, the total amount of Se taken up by the plant is independent of the chemical species supplied in the grown media in agreement with previous studies performed in hydroponic solutions, both oxyanions being equally available for plant uptake [39]. However, different uptake could be observed in the case of plant cultivation on soil, where Se(IV) resulted to be less bioavailable being more strongly sorbed on soil solid phases compared to Se(VI) [14].

Although total Se uptake was similar, Se allocations and translocations in plant tissues (leaves or roots) strongly depend on supplied Se form used for the enrichment treatment (Tables 1 and 2). Namely, total 82Se in leaves (8.1 ± 1.8 mg kg−1) was found higher than that found in roots (5.1 ± 0.2 mg kg−1). On the contrary for 77Se tracer, highest concentrations were observed in roots (23 ± 1 and 3.0 ± 0.2 mg kg−1 for the roots and the leaves, respectively). The calculated translocation factor defined as the ratio between Se concentrations in leaves and roots is strongly higher for 82Se (1.6 ± 0.4) than that of 77Se (0.13 ± 0.01), suggesting an increased rate of Se translocation from roots to leaves in ryegrass after supplementation with Se(VI) versus Se(IV). These results are in agreement with previous studies performed with different plant species (soybean, wheat, broccoli, rice, Indian mustard, sugar beet, and maize) clearly demonstrating that Se translocation from roots to leaves is dependent on the form of Se supplied, selenate being much more easily transported than selenite or any organic species, like SeMet [6, 16, 17, 32, 40].

Table 1 Total and identified species concentrations of 77Se tracer in leaves and roots. Corresponding proportions in extracts (expressed as % of total 77Se in plant compartment) are given between brackets. Values are mean ± SD from triplicate digestions or sequential extractions of two samples/plant compartment
Full size table
Table 2 Total and identified species concentrations of 82Se tracer in leaves and roots. Corresponding proportions in extracts (expressed as % of total 82Se in plant compartment) are given between brackets. Values are mean ± SD from triplicate digestions or sequential extractions of two samples/plant compartment
Full size table

Regarding sequential extraction procedure, total Se concentrations measured in the three selective extracts and in digested residual pellet as well as the corresponding extraction efficiencies for leaves and roots samples are presented in Tables 1 and 2 for 77Se and 82Se tracers, respectively.

The first extraction step carried out on the plant tissue by mean of ultrapure water assisted by sonication extracted most of 82Se present in roots and leaves samples, accounting for almost 80 % of total 82Se present in the compartment. Driselase and protease digestions extracted almost the same amount of 82Se, about 10 % in both samples. No significant decrease in 82Se concentrations measured after 100 kDa cut-off filtration of enzymatic extracts was observed.

In roots, water-soluble 77Se accounted for 25 ± 1 % of total 77Se. The digestion with driselase liberated an additional 20 ± 1 % (bound to cell walls) and the proteolytic lysis allowed recovery of an additional 12 ± 1 %. Indeed, the remaining 46 ± 2 % (residue) might correspond to 77Se incorporated in more complex structures which could not be released with the above reagents. Different proportions of 77Se were observed in leaves extracts compared to root extracts, accounting for 51 ± 5, 31 ± 4 and 16 ± 2 % for water, driselase and protease extracts, respectively.

Comparison of total content before and after cut-off filter evidenced 77Se losses of about 5 and 15 % related to total 77Se in roots and leaves, respectively. It means that the tracer was incorporated into HMW macromolecules (>100 kDa) released after cell wall digestion. On the contrary, no significant decrease of 77Se concentration after cut-off was observed in protease extracts.

Although these results display a near-complete extraction (up to 90 %) for both tracers in leaves and for 82Se in roots, a large amount of 77Se remained in the unextracted fraction of roots.

Quantitative speciation of Se

Species determination by PGC-HPLC/ICP-MS

PGC separation was firstly performed to obtain information on the occurrence of Se metabolites coming from the two precursors (77Se(IV) and 82Se(VI)) and their distribution in the plants. Identification of Se containing chromatographic peaks in extracts was performed by spiking commercially available Se standards. Complete separation and elution of available standards can be achieved within the first 10 min of the separation run. Quantification of identified Se species was then performed by standards addition method based on peak areas.

Quantitative results (concentrations and proportions) for identified Se species determined in the different extracts and plant compartments are presented in Tables 1 and 2 for 77Se and 82Se tracers, respectively. Chromatograms of roots water extracts revealed the presence of tracers as inorganic Se forms (77Se(IV) and 82Se(VI)) used for Se enrichment in the growing media. 82Se(VI) amounts to 100 % of total water extracted 82Se and consequently 81 % of total 82Se in roots. On the contrary, 77Se(IV) only accounts for 29 % of water extracted 77Se and 8 % of total 77Se measured in roots, although no additional chromatographic peaks containing 77Se were detected. Regarding leaves samples, for 82Se, similar to that observed in roots water extract, only 82Se(VI) could be detected corresponding to almost 100 % of total 82Se extracted with water. For 77Se, two major peaks corresponding to 77Se(IV) and 77SeMet were observed. A smaller peak was also detected with a retention time corresponding to that of 77MeSeCys. Calculated extraction efficiencies for 82Se(VI) are in agreement with results obtained for others plants (chicory, dandelion, lamb’s lettuce, and parsley) that showed water extraction recovery of selenate in leaves of different plants ranging from 45 to 85 % [41].

Example chromatograms corresponding to driselase extracts of roots and leaves are shown in Fig. 2. For both samples, two major Se peaks matching the retention times of MeSeCys and SeMet and a minor selenocystine peak were detected. Some Se-containing peaks whose retention times do not match the ones of any Se available standard were also detected. In roots extracts, these peaks were more intense for 77Se isotope than for 82Se one. The opposite was observed for leaves. An increase of baseline signal was observed between 15 and 30 min (methanol gradient) mostly for 77Se tracer, suggesting some elution of Se compounds or Se peptides.

Fig. 2
figure 2

PGC–HPLC ICP-MS example chromatograms of driselase extracts (diluted 2.5 times) of a roots and b leaves. The y-scale of both figures is expanded as the high amount of SeMet would not allow showing the smaller peaks

Full size image

Considering quantitative data obtained in the Se fraction released by driselase hydrolysis, it appears that both inorganic Se tracers were metabolized by the plant into species associated with cell wall, actually no more 82Se(VI) could be detected and 77Se(IV) only accounted for between 1.2 and 3.6 % of total extracted 77Se measured after cutoff (77Secut-off). SeMet and MeSeCys are the most abundant detected known Se species; their concentrations varying as a function of plant compartment and Se tracer. SeMet was present at a higher concentration in leaves compared to roots. As already observed in water extracts, most of 77Se were present as unknown compounds, quantified species accounting for 14 and 42 % of total extracted 77Secut-off in roots and leaves, respectively. In roots extract, a large amount of 82Se was also undetected, accounting for almost 64 % of total extracted 82Secut-off.

The chromatographic profiles of protease XIV extracts of roots and leaves are shown in Fig. 3. Two main peaks were detected, corresponding to SeMet and MeSeCys by standards matching retention times. MeSeOOH, Se(IV) and SeCys2 could be detected at μg kg−1 level for 77Se in roots and for 82Se in leaves. Additional peaks corresponding to unknown selenium-containing species were also observed, some of them eluting similarly than those observed in the driselase extract. For the two tracers, similar profiles in terms of detected species were observed, however with different intensities.

Fig. 3
figure 3

PGC–HPLC-ICP MS example chromatograms of protease extracts a roots and b leaves. The y-scale of both figures is expanded as the high amount of SeMet would not allow showing the smaller peaks

Full size image

According to the quantitative speciation data for both tracers in roots and leaves after proteolytic digestion, SeMet represents the main extracted species accounting for 80 and 90 % of total extracted Secut-off in leaves and for 52 and 90 % in roots for 77Se and 82Se, respectively.

Little amount of SeCys2 could be observed in protease extract which could reflect either the unspecific (SeCys2 detection) or the specific (SeCys detection) selenium incorporation into the plant proteins. In fact, in these extraction conditions, the detection of SeCys is not possible as it is rapidly oxidized to SeCys2.

Selenium distribution as a function of molecular weight (SEC /ICP-MS)

Considering the large proportion of unknown species in aqueous and driselase extracts, an insight into speciation of selenium in these extracts was attempted by analytical scale size-exclusion chromatography that allows obtaining information about the molecular weight of extracted seleno-molecules.

For 77Se tracer, water extracts of leaves and roots show similar elution profiles with a peak eluting in the exclusion volume and two peaks in the <1.3 kDa range, but with important discrepancies in peaks intensity, higher in roots than in leaves, as shown in Fig. 4a, b. The first peak (RT 12 min) corresponding to the exclusion volume probably corresponds to selenium containing proteins. A baseline increase was observed between 14 and 23 min, probably corresponding to elution of Se-containing peptides. In the low molecular weight (LMW) region, two major peaks were also observed which may correspond to the free seleno-amino acids, inorganic selenium and small selenium metabolites. In the case of 82Se, the SEC/ICP-MS profiles for leaves and roots water extracts showed only one peak eluting in LMW region corresponding thus to selenate, as previous PGC chromatographic analysis showed that selenate was the only species detected in water extract (100 % of extracted 82Se). Chromatograms of driselase extracts of leaves and roots samples indicate, for both tracers, the presence of selenium species eluting after the 1.3 kDa marker region, corresponding to LMW species. SEC profiles, showing two major peaks at higher elution times, were quite similar for leaves and roots extracts, but differed among the two tracers. In root extracts, peak intensity at m/z 77 was higher than the one at m/z 82, whereas on the contrary was observed in leaves extracts. Moreover, some minor Se species eluting between 14 and 23 min could be observed only in SEC chromatograms of 77Se tracer in roots extracts.

Fig. 4
figure 4

SEC-ICP MS profile of 77Se tracer in water extracts of: a roots and (b) leaves (arrows indicate the retention time of calibration molecules (myoglobine 17 kDa, insulin 5.8 kDa and vitamin B12 1.35 kDa). c Proportion of 77Se in defined molecular weight fractions

Full size image

Subsequently, in order to analyze Se distribution as a function of the molecular weight, SEC fractions were collected, and their selenium content was determined off-line. The sum of Se content measured in all collected fractions was equal to the total selenium amount in extract, evidencing no loss during the chromatographic separation. The proportions of 77Se in each collected fraction of water extracts (% of total extracted 77Se) as a function of calibrated molecular weight fractions are summarized in histogram depicted in Fig. 4c.

In water extracts of roots and leaves, respectively, 64 and 50 % of total extracted 77Se are associated to the fraction lower than 1.3 kDa, corresponding mainly to inorganic selenium, free amino acids and small peptides. For leaves and root driselase extracts, about 90 % of total extracted 77Se appears in this low molecular weight fraction. These results indicate that 77Se associated with cell wall is mostly in the form of low molecular weight metabolites (less than 1.3 kDa).

Multidimensional chromatography was further applied by PGC–HPLC/ICP-MS analysis of SEC collected fractions. Namely, fractions eluting in the LMW region (<1.3 kDa) were directly analyzed, while first fractions collected in the HMW area were subjected to enzymatic hydrolysis with protease type XIV in order to release selenium species bound to proteins. For 77-Se tracer, chromatographic profiles of water extract HMW fractions (W1, W2, and W3) of roots and leaves after proteolytic digestion reveal the presence of two major peaks with retention times matching those corresponding to SeMet and SeCys2 standards. A little amount of 77-selenite was detected in fraction W1 of root sample. No additional peak corresponding to unknown species could be observed. In each collected fraction above 1.3 kDa, quantification of seleno-amino acids do not account for the total 77Se measured in the fraction (26–83 % of total 77Se in each fraction). This result indicates that all the soluble macromolecules could not be hydrolyzed by the proteolytic digestion, suggesting the non-protein nature of these compounds and/or an incomplete digestion.

With regard to root driselase extracts, no Se-containing peaks could be detected after proteolytic digestion of fractions, suggesting a non-protein nature of Se species eluting in HMW region and indicating that dissolution of the cell walls using a mixture of pectinolytic enzymes does not release any protein. However, a significant amount of low molecular weight compounds could be released only after the destruction of the wall structure.

In the case of LMW fractions obtained from water and driselase extracts of both samples, PGC–HPLC/ICP-MS analysis confirmed the presence and amount of Se compounds previously determined by direct analysis of extracts, but no others species could be detected, probably due to their relatively low concentration in the analyzed samples.

Metabolism of 77Se(IV) and 82Se(VI) in ryegrass

The comparison of the proportions of Se species versus total Se uptaken by the whole plant indicates strong differences between both tracers, suggesting different metabolisms (Table 3). These results demonstrate that applied Se(VI) is less efficiently converted into organo-selenium compounds than Se(IV). Precisely, supplementation with 82Se(VI) leads to an accumulation of Se essentially as selenate, accounting for almost 75 % of total Se in whole plant, the rest being converted into organic compounds such as SeMet (9 %) and unknown metabolites (14 %). This low conversion yield might be due to the energy-demanding reduction step of Se(VI) into Se(IV) which is required for selenate transformation into seleno-amino acids in the metabolic pathway of selenium in plants [6, 42].

Table 3 Proportion of Se species related to total Se tracer content in whole plant and plant compartments
Full size table

Opposite to 82Se(VI), added 77Se(IV) is readily metabolized by the plant; actually, the remaining amount of inorganic Se(IV) corresponds to only 6 % of total Se content in plant. SeMet coming from the 77Se tracer is the most abundant identified organic species, accounting for 19 % of total amount and, mostly incorporated into protein structures and to less extent as free soluble amino acid. However, the major pool of 77Se (about 37 %) is represented by unknown compounds, and a large amount of the tracer remains un-extracted (32 % of total 77Se in the whole plant).

Comparing the occurrence of Se metabolites and their distribution in the plant obtained by different selective extraction procedures, a possible metabolic pathway as function of the chemical species could be proposed as depicted in Fig. 5a, b for 82Se and 77Se, respectively. After 82Se(VI) absorption by root via sulfate transporter [15, 17], 82Se appears to be mainly translocated to the leaves with little amount remaining in the roots. 82Se tracer is weakly metabolized in roots remaining essentially in form of highly mobile inorganic Se(VI), leading to significant rate of translocation. The little amount of 82Se(IV) observed in root extracts (1 % of total 82Se in roots) suggests small transformation of 82Se(VI) into organic compounds through a reduction pathway via selenite. Actually, Se(VI) can be reduced to Se(IV) that can be converted downstream to selenide [43], a part of which is metabolized into organo compounds such as SeMet, SeCys MeSeCys and others LMW metabolites (Fig. 5a).

Fig. 5
figure 5

Schematic overview of proposed metabolism pathway following uptake of a 82Se(VI) and b 77Se(IV). Font size is proportional to the relative concentration of the species and thickness of arrows is proportional to transport rate

Full size image

On the basis of chromatographic profiles and decrease in 77Se abundance from root to leaf, it is suggested that Se(IV), after absorption through plant roots, is reduced into selenide that in turn is quickly transformed into organic Se before transport to leaves. SEC/ICP-MS profiles of water extracts of roots and leaves suggested incorporation of 77Se tracer supplied in form of selenite to both HMW and LMW compounds. HPLC/ICP-MS analysis of fractions eluting in HMW region after proteolytic digestion indicated the incorporation of 77Se tracer into soluble proteins and peptides mostly as SeMet, and the occurrence of non-protein macromolecules containing Se. The high metabolism rate into high molecular weight compounds probably restricts further translocation to the leaves of the plant which supports the total analysis results. The highest recovery of 77Se tracer obtained after enzymatic hydrolysis of roots compared to water extraction indicates that exposure of the ryegrass to Se(IV) promotes its transformation into organic compounds and incorporation into proteins or molecules associated with cell wall.

Our findings are consistent with results from previous speciation studies carried out on different accumulator and not accumulators plants species supplied with Se(VI) or Se(IV). These works report that in selenate-treated plants, Se is translocated to leaves predominantly as selenate, whereas in selenite-treated plants Se is quickly converted into organic Se before transport, only a small amount of selenite being directly transported from root to leaf [12, 15, 17, 20, 44–47]. However, some differences are observed if we compare the proportion of selenite-derived compounds obtained in this study with those obtained with accumulator plants. Namely, Se-hyperaccumulator species (Brassica, Astragalus) and secondary accumulators (Allium) metabolize primarily Se into various non-proteinogenic Se amino-acids such as γ-glutamyl-selenocysteine, selenocystathionine or methylselenocysteine, this latter being the most abundant organo compound (up to 80 % of total Se) [18–21]. Selenium tolerance of these plants seems to be related to the synthesis of these compounds, allowing plants to accumulate high amounts of selenium without toxicity symptoms [48]. Kotrebai et al. reported that proportions of selenium compounds in selenium-enriched samples are dependent on the total selenium concentrations, being present mainly as SeMet and γ-glutamyl-selenocysteine in samples with Se contents below 333 mg kg−1 and as MeSeCys in those with higher concentration [49]. Similarly, Cubadda et al. found an increasing amount of MeSeCys (until 21–68 %) with increasing Se supply in wheat [50]. In the present study, the low presence of MeSeCys (<2 %) could, thus, indicate that applied concentrations did not require a response from the ryegrass to minimize Se-induced toxicity.

Conclusion

This work presents for the first time the application of double Se-stable isotopes tracing in plant (exposed to 77Se(IV) and 82Se(VI)), including determination of tracers distribution and speciation analysis, in order to simultaneously monitor the bio-incorporation of two inorganic species commonly used as plant enrichment sources.

A sequential extraction protocol, targeting different Se fractions in cell plant, and based on combination of various state-of-the-art extraction procedures in plant was optimized in order to reach maximum Se recoveries and to ensure stability of species. Through the use of isotopically enriched tracers, it was shown that inorganic selenium can be taken up and bio-transformed, i.e., converted into organic compounds and/or incorporated into Se-containing proteins. In general, total selenium measurement showed that supplementation with Se(VI) resulted in similar total Se level in whole plant compared to that enriched with Se(IV). However, allocation in plant compartments and metabolism of Se depends on the nature of initial inorganic species. Selenite was found to be readily metabolized in roots into both HMW and LMW compounds and showed low mobility to leaves, while supplementation with selenate resulted in its accumulation in leaves essentially as original inorganic form. Therefore, at equal bioavailability, the degree of metabolic conversion of Se(IV) into organic species is largely higher than the one of selenate, leading to lower accumulation in leaves. To conclude, the present labeling methodology demonstrates several merits, such as an increase of detection sensitivity allowing speciation study in not accumulator plants, and the possibility of simultaneous monitoring of different species, allowing comparing biological responses to various compounds of the same element. This opens new approaches for its application to other types of plant for a better understanding of their metabolic pathways. Further investigations regarding identification by molecular mass spectrometry of unknown Se-containing peaks detected in the different extracts are required to better substantiate possible Se metabolic pathway.