Fragaria × ananassa cv. Elsanta plants were hydroponically grown, as previously described (Valentinuzzi et al., 2015b), in either a full nutrient solution (control), composed as follows KH₂PO₄ 0.25 mM, Ca(NO₃)₂ 5 mM, MgSO₄ 1.25 mM, K₂SO₄ 1.75 mM, KCl 0.25 mM, Fe(III)Na-EDTA 20 μM, H₃BO₃ 25 μM, MnSO₄ 1.25 μM, ZnSO₄ 1.5 μM, CuSO₄ 0.5 μM, (NH₄)₆Mo₇O₂₄ 0.025 μM, or in a nutrient solution supplemented with 10 μM Se or 100 μM Se, supplied as Na₂SeO₄ (Sigma–Aldrich, reagent grade ≥ 98.0%). The nutrient solution was changed every 3–4 days. Plants were grown in a climate chamber with 14/10 h day/night, 24/19°C, 70% relative humidity (RH) and 250 μmol m^-2 s^-1 light intensity. Ten biological replicates were kept for each treatment.

Strawberry plants were hydroponically grown for 72 days. Strawberry fruits were harvested once at least 80% of the fruit surface showed a red coloration. Fruits skin color was evaluated determining the color parameters using a portable tristimulus colorimeter (Chroma Meter CR-400, Konica Minolta Corp., Osaka, Japan). Fresh weight (FW), yield per plant (g FW per plant), average fruit yield (g FW), average number of strawberry fruits per plant were assessed. At harvest, shoots and roots were separated assessing fresh weight (FW) and dry weight (DW) of the tissues together with the root to shoot ratios.

Titratable acidity, total soluble solid content and firmness of fresh strawberry fruits were determined as previously described (Valentinuzzi et al., 2015a).

Freeze-dried strawberry samples were homogenized and 100 mg of strawberry powder were extracted with 1 mL methanol (HPLC grade, Merck, Darmstadt, Germany). The mixture underwent sonication for 30 min at 4°C and the extracts were centrifuged at 14000 × g for 30 min at 0°C; afterward, the supernatant was collected and filtered through a 0.2 μm nylon filter (Valentinuzzi et al., 2015a). The content of total phenols of strawberry fruit extracts was determined following the Folin-Ciocalteau method (Folin and Ciocalteu, 1927; Atanassova et al., 2011), whilst the concentration of flavonoids and flavonols was determined by a pharmacopeia method, using rutin hydrate as reference compound (Miliauskas et al., 2004).

Organic acids and sugars were separated simultaneously by HPLC using a cation exchange column Aminex 87-H column (300 × 7.8 mm, 9 μm, Bio-Rad) using an isocratic elution with 10 mM H₂SO₄ as carrier solution at a flow rate of 0.6 mL min^-1. Organic acids were detected at 210 nm using a Waters 2998 photodiode array detector (Waters Spa, Italy), whilst sugars were detected by a refractive index detector (Waters Spa, Italy). Standard acids and sugars were prepared as individual stock solutions and then combined to give diluted reference standards. Organic anions and sugars were identified by comparing retention times of unknowns to pure compounds. Standards were purchased from Sigma–Aldrich (St. Louis, MO, United States).

Oven-dried shoot samples (60°C) and freeze-dried strawberry fruits were homogenized and approximately 0.5 g of each sample were acid digested with concentrated ultrapure HNO₃ (650 ml L^-1; Carlo Erba, Milano, Italy) using a single reaction chamber microwave digestion system (UltraWAVE, Milestone, Shelton, CT, United States). Selenium and macro- and micronutrient concentrations were then determined by ICP-OES (Arcos Ametek, Spectro, Germany), using tomato leaves (SRM 1573a) and spinach leaves (SRM 1547) as external certified reference material. The limits of detection for each element are reported as follow: Ca 1.80 μg L^-1, Fe 0.32 μg L^-1, K 1.10 μg L^-1, Mg 1.53 μg L^-1, Mn 0.37 μg L^-1, P 2.00 μg L^-1, S 9.00 μg L^-1, Se 11.10 μg L^-1.

The screening of plant metabolites was carried out on a hybrid quadrupole-time-of-flight mass spectrometer coupled to an UHPLC chromatographic system (UHPLC/Q-TOF) using a metabolomic analysis as previously set-up (Lucini et al., 2015). A 1290 liquid chromatograph, coupled to a G6550 Q-TOF mass spectrometer via a Dual Electrospray JetStream ionization system (all from Agilent technologies, Santa Clara, CA, United States), was used. Briefly, the instrument was run in the positive SCAN mode (MS-only) and operated to acquire spectra in the range of 100–1600 m/z in extended dynamic range mode. Reverse phase chromatographic separation was achieved using a Knauer BlueOrchid C18 column (100 × 2 mm i.d., 1.8 μm) and a gradient elution having as LC mobile phase a mixture consisted of (A) water (proteomic grade from VWR International, Radnor, PA, United States) and (B) methanol (LCMS grade from VWR). Formic acid 0.1% (v/v) and ammonium formate (5 mM) (both from Sigma–Aldrich, St. Louis, MO, United States) were added to both mobile phases. The gradient started with 5% B and increased to 90% B within 30 min, then was held for 5 min. The mobile phase temperature was set to 35°C, the injection volume was 3 μl and the flow rate was 220 μl min^-1.

Lock masses were continuously infused via a dedicated electrospray in the JetStream source to ensure mass accuracy; purine at m/z 121.0509 and HP-0921 at m/z 922.0098 were used with this purpose.

The raw data gained from the Q-TOF mass analyzer were processed by the Profinder B.06 software (from Agilent Technologies) using the “find-by-formula” algorithm. Compounds identification was based on both accurate mass and isotopic pattern (i.e., accurate spacing and isotopes ratio), and expressed as overall identification score. Compounds were then aligned (mass and retention time), annotated using the database exported from PlantCyc 9.5 (Plant Metabolic Network¹; released November 2014) and then subjected to a recursive analysis workflow having retention time as mandatory (with retention time tolerance of <0.1 min) in the second ID step. Once mass and retention time alignment was done, a filter-by-frequency post-processing filter was applied after deconvolution, retaining only those compounds that were present in at least 80% of replications within at least one treatment (Lucini et al., 2015). A second identification process, using the same procedure, was carried out to specifically profile phenolic compounds using the database Phenol-Explorer (Rothwell et al., 2013).

Therefore, based on the strategy applied, identification was carried out according to Level 2 (putatively annotated compounds) as set out by the COSMOS Metabolomics Standards Initiative².

The results are presented as means of at least three replicates ± standard error (SE). All the quantitative analyses have been carried out on at least three biological independent replicates. In the case of fruit quality analyses, each single independent replicate was obtained by pooling ten strawberry fruits at the same ripening stage. Statistical analysis was performed using Statgraphics (Statpoint technologies, Inc., Warrenton, VA, United States). Data were analyzed by analysis of variance (ANOVA), and means were compared using Student Newman Keul’s (SNK) test at p ≤ 0.05 to determine the significance of differences found.

Interpretation of metabolomic results was carried out using Mass Profiler Professional B.12.06 (from Agilent technologies). Compounds abundance was log2 normalized, abundances were normalized at 75th percentile and baselined versus the median of each compound in all samples. For statistical analysis from metabolomic data, unpaired t-test (P ≤ 0.001, Bonferroni multiple testing correction) and fold-change analysis (cut-off = 2) were combined into volcano plots. Furthermore, a multivariate partial least squares discriminant analysis (PLS-DA, N-fold validation with N = 4) was performed and variables loadings, which were used to build the class prediction model, were plotted according to their weight within the latent vectors. Compounds with the highest scores on the first and second latent vectors were exported from the covariance structures (loading plot) in the PLS-DA hyperspace.

Strawberry plants were hydroponically grown for approximately 10 weeks in either a full nutrient solution (control), or in a nutrient solution added with Se at a final concentration of 10 and 100 μM. Fruits were harvested once at least 80% of the fruit surface showed a red coloration, whilst shoots and roots were collected at the end of the growing period and subjected to subsequent analyses.

Shoots fresh weight increased by 20% in strawberries treated with 100 μM Se respect to the control plants and those treated with 10 μM Se (Table 1). Also the leaf area increased in strawberries treated with 100 μM Se by 17% in comparison with both control and 10 μM Se treated plants (Table 1). As a consequence and due to the fact that the root biomass did not increase with increasing Se concentration, the shoot to root ratio increased in plants supplied with the highest Se concentration (Table 1).

The chlorophyll content, evaluated as SPAD index values, was expressed as average between old and young leaves and it did not highlight any difference between the Se treated and control plants (Supplementary Figure S1).

The micro- and macronutrient content has also been assessed in shoots and roots of strawberry plants to further study the effect of Se on plant physiology and possible nutrient imbalances (Table 2). In shoots, the Se concentration reached 10.48 ± 1.20 μg g^-1 DW in plants supplied with 10 μM Se and 125.08 ± 13.89 μg g^-1 DW in those supplied with 100 μM Se. Beside Se, only the concentration of S in the shoots of plants treated with 100 μM Se was significantly higher than in control and in 10 μM Se-treated strawberry plants (Table 2).

In roots tissues, the Se concentration was about 19.02 ± 2.99 μg g^-1 DW when supplied with 10 μM Se and about 174.42 ± 14.35 μg g^-1 DW with 100 μM Se application (Table 2). Differently from shoots, the presence of increasing concentration of Se in the growth medium caused an alteration in the uptake of several macro- and micronutrients (Table 2). Calcium (Ca), magnesium (Mg) and manganese (Mn) showed higher concentration in roots in the case of 100 μM Se-treated plants, whilst potassium (K) concentration was increased in the roots of both 10 and 100 μM Se-treated strawberries (Table 2).

With the aim of determining whether the growth of plants in Se-fertilized hydroponic solution might result in the production of biofortified fruits, strawberries were analyzed for their macro- and micronutrient content (Table 2). In fruits, Se concentration showed increasing concentrations according with the increased Se supply to the nutrient solution. An addition of 10 and 100 μM Se led to an average Se content in fruits of 3.95 and 46.04 μg Se g^-1 DW, respectively. It is well established that Se and S use the same transporters being taken up by plants, therefore the high availability of one of the two element might decrease the absorption of the other. Nevertheless, in our experimental conditions, no differences in the S concentration have been highlighted between treatments (Table 2). Likewise, except for Ca, any other element was altered significantly in Se-supplemented strawberries (Table 2).

At commercial maturation stage (corresponding approximately to 80% of the total fruit surface colored) strawberry fruits were harvested and quality parameters were analyzed. The fruits color was recorded according to the CIELAB scale and no significant differences were highlighted depending on the Se fertilization as compared to control fruits (Supplementary Table S1). Total soluble solids (TSS), expressed as degrees Brix grade (°Bx), have been affected significantly by the treatment; in fact, a 30% increase in TSS has been observed in strawberries treated with 100 μM Se respect to the control and 10 μM Se-treated plants (Table 3). Titratable acidity showed no significant difference between treatments (Table 3). The sweetness index [expressed as the ratio of TSS and the acidity (Sturm et al., 2003)] of strawberry fruits harvested from control plants resulted 4.94, whereas plants supplied with 10 μM Se exhibited an index of 5.27 and those treated with 100 μM Se had an index of 6.46 (Table 3). On the other hand, firmness was not affected by Se supplementation (Table 3).

Fructose, sucrose and glucose were the predominant sugars detected in the extract of strawberry fruits (Figure 1A). The content of soluble sugars was significantly affected by the Se treatment imposed; in fact, the concentration of both fructose and sucrose resulted significantly higher in the plants supplied with 10 and 100 μM Se respect to the control plants; on the other hand, glucose showed no significant differences (Figure 1A).

Concerning organic acids, citric, malic, t-aconitic, shikimic and fumaric acid were the predominant compounds detected in the strawberry fruit extracts (Figures 1B,C). In particular, citric and malic acids were the two most abundant and they were typically present at concentrations 1000-fold higher than the other acids detected (Figures 1B,C). Total phenol content was not affected by Se treatment (Figure 2), however, the concentration of flavonoids and flavonols, which are subclasses of phenolic compounds, decreased significantly with increasing Se supplementation (Figure 2).

Metabolic profiling of strawberry fruits has been carried out by an untargeted analysis through a hybrid quadrupole-time-of-flight mass spectrometer coupled to an UHPLC chromatographic system. Metabolites were annotated by comparison with Plant Metabolic Network (PMN) database³. The Partial Least Square Discriminant Analysis (PLS-DA), carried out on the metabolites identified, showed a clear separation (overall validation accuracy after validation and model training = 100%) of the samples according to the Se concentration applied to the nutrient solution, thus demonstrating that the metabolome of fruits was differentially affected by the biofortification practice (Figure 3).

Differential compounds, as highlighted by the Volcano plot analysis, are given in Table 4. As reported, extreme variations (given as fold-change = 16 and p = 0) could be observed for a large number of metabolites; the alteration of the secondary metabolites occurred particularly in cytokinins and amino acids (Table 4). However, also other classes, such as alkaloids, aldoximes and brassinosteroids, have been influenced by the Se treatment. Among the cytokinins, dihydrozeatin was up accumulated in both 10 and 100 μM Se treated plants respect to the control fruits.

Within the amino acids metabolic pathways, Se treated strawberry fruits showed an over accumulation of phenylalanine, which belongs to the phenylalanine metabolism and it is an intermediate for the biosynthesis of N-acetyl-phenylalanine; furthermore, D-octopine, involved in the arginine and proline metabolism, was also up accumulated in Se biofortified fruits. On the other hand, the down accumulation of both indolylmethylthiohydroximate and N-formyl-D-kynurenine in Se treated plants suggests that Se enrichment affected the glucosinolate biosynthesis and the aromatic amino acid synthesis. Selenium treatments induced also an increase in the content of alkaloid such as gramine that might have a role in plant defense against insects (Wippich and Wink, 1985; Corcuera, 1993). Moreover, nitrogen-containing bisphosphonate, pamidronate has been upregulated in Se treated plants.

Afrormosin-7-O-glucoside, an isoflavone belonging to the family of flavonoids, was specifically down accumulated in 10 μM Se treated plants, whilst no significant alteration was detected in 100 μM Se treated fruits. Furthermore, the brassinosteroid typhasterol, which is an intermediate metabolite in the brassinosteroid and acts as plant hormone with plant growth-promoting activity (Takatsuto, 1986; Kauschmann et al., 1996), was down accumulated in 10 μM Se treated strawberries. The treatment with 10 μM Se induced also a down accumulation of 3,4-dihydroxymandelonitrile-β-glucoside, a member of the cyanogenic glycoside group, (-)-(R)-carvone, a monoterpenoid commonly found in essential oils, and (+)-copalyl diphosphate, involved in the biosynthesis of abietate, palustric acid or levopimaric acid. Interestingly, the (3E)-phytochromobilin, which is the chromophor group of the phytochrome involved in the perception of the red and far-red light, is up accumulated in strawberry fruits treated with 100 μM Se. Experimental evidence shows that the perception of light and the subsequent signaling are responsible for an increased biosynthesis and accumulation of flavonoids (Zoratti et al., 2014).

In order to gain further details on metabolites profile, particularly from a nutritional point of view, the Phenol-Explorer (Rothwell et al., 2013) database, encompassing only phenolic compounds occurring in plants, was also used. Flavonoids, in particular anthocyanins, and polyphenols were the predominant compound classes found in strawberry fruits (Table 5). Most of these metabolites have been up accumulated in both 10 and 100 μM Se treated plants compared to control plants. Only pelargonidin 3-O-sambubioside, an anthocyanin, has been down accumulated in strawberries supplied with 100 μM Se respect to control plants. On the other hand, delphinidin 3,5-O-diglucoside, another member of the anthocyanin class, has been found to be up accumulated in plants supplied with 10 μM Se. Among the polyphenols, 3,4-DHPEA-EDA, specifically a tyrosol, was the compound with the highest positive fold change of about 2000 in 100 μM Se treated plants compared to control plants. However, in strawberries supplied with only 10 μM Se the up-regulation occurred at 15 times fold-change. The second most abundant compound was coumestrol, which is an end product of the isoflavonoid biosynthesis or phenylpropanoid biosynthesis. It has been up accumulated approximately 1300 times in strawberries supplied with Se, independently from the concentration. The phenolics induced by supplementation were the same at either 10 or 100 μM Se, suggesting that the effect was specifically related to Se. Interestingly, when the highest and the lowest concentrations were compared, 100 μM Se treatment was slightly more effective in inducing the accumulation of anthocyanins, with delphinidin 3,5-O-diglucoside and pelargonidin 3-O-sambubioside having a negative fold-change of 6 and 2 respectively in 100 μM Se.

Naïve Bayesian biomarker discovery confirmed the involvement of phenolic compounds in response to Se supplementation, as coumestrol, conjugated ferulic acid and glycosylated anthocyanins [malvidin 3-O-(6″-p-coumaroyl)-glucoside and pelargonidin 3-O-sambubioside].