Tomato (Solanum Lycopersicum L.) Response to Nickel Stress: Bioavailability, Accumulation and Allergenicity

Enrica Roccotiello (  enrica.roccotiello@unige.it ) University of Genoa Elena Nicosia Regione Liguria Lorenzo Pierdonà Czech University of Life Sciences Prague Pietro Marescotti University of Genoa Maria Antonietta Ciardiello Institute of Biosciences and Bioresources Ivana Giangrieco Institute of Biosciences and Bioresources Adriano Mari Associated Centers for Molecular Allergology Danila Zennaro Associated Centers for Molecular Allergology Denise Dozza IREN Laboratori s.p.a. Michele Brancucci Geospectra S.r.L. spin off Mauro Mariotti University of Genoa


Introduction
Food, speci cally vegetables, represents a major source of nickel (Ni) exposure 1 . Environmental contamination and cultural practices can increase Ni amount in vegetables posing signi cant risk for human health 2, 3 . Nickel is a ubiquitous trace element occurring in water, soil, air and in the biosphere.
This element is essential for several plants, microorganisms, and vertebrates 4 .
Tomato (Solanum lycopersicum L.) is a key vegetable worldwide, belonging to the Solanaceae family, the third most important commercial crop family from an economic point of view. This species has been used for a long time as a model plant in studies on disease response, genetics, and fruit ripening 5 . The fruit is produced for fresh consumption and processed products (e.g., tomato sauce, tomato paste, etc.) and is naturally rich of Ni, potentially affecting human health.

Nickel in soil and food
According to the European Directive 6 the limit values for Ni concentration in agricultural soils are 75 mg kg − 1 (Dry Weight) at pH 7 and 30 mg kg − 1 (Dry Weight) at pH 6, considering the higher bioavailability of most metals at lower pHs 6 .
The average Ni content in natural soils ranges between 13 and 37 mg kg − 1 , but signi cantly higher contents > 1000 mg kg − 1 occur in ultrama c soils 7 , which systematically exceed, up to one order of magnitude, the threshold values lid down by governments and environmental agencies (e.g., 8 ). Similarly, Ni content in agricultural soils generally does not exceed 100 mg kg − 1 9, 10 but it can reach thousands of mg kg − 1 in areas with ultrama c bedrocks 11,12,8 . Besides, Ni intakes of agricultural soils may derive from atmospheric fallout, super cial and underground waters, and direct anthropogenic inputs 13 . Among these, atmospheric fallout represents an important and widespread input of Ni to soils, also for remote areas, due to the worldwide increasing emission of Ni to the atmosphere 14 , mostly from coal and oil combustion 7 . Direct anthropogenic inputs to agricultural soils are primarily due to mineral fertilizers, pesticides, compost, sewage sludge, and manure 13 .
The mobility of Ni in soils is strictly controlled by organic matter, amorphous oxides (mainly Fe and Mn oxides) and clay minerals. Oxides and clay minerals mostly scavenge Ni through sorption mechanisms and can release this metal to soil solution depending on pH variations 7 . Moreover, although the solubility of Ni in water is generally low (< 2 µg/L), it can signi cantly increase in the presence of dissolved organic compounds because it can form soluble complexes with organic ligands 15 , becoming potentially bioavailable. For instance, Ni mobility is quite high in acidic organic-rich soils, where fulvic and humic acids are formed by the decomposition of organic material 7,14 .
Ni plays and important part in plant physiology, as a component of the enzyme urease 16 , which participated in urea hydrolysis in several plant species 17 . In some plants, Ni is an essential micronutrient, promoting growth and development 18 (Anjum et al. 2015). Ni toxicity levels are ∼10 µg [g dry weight (DW)] −1 in sensitive plant species 19 , and 50 µg g − 1 DW in moderately tolerant species 20 . Ni phytotoxicity varies with the bioavailability of the metal and with the plant species 18 Currently, a speci c legislation on Ni in food is missing. The only threshold values in the European legislation 21 is about drinking water with a maximum level of 20 µg/l, instead of 70 µg/l suggested by WHO 22 . Despite the lack of threshold values according to current laws regarding Ni in food, EFSA 23 set a tolerable daily intake for body weight equal to 13 µg Ni/kg. The EU commission raised awareness on Ni and adopted recommendation 24  Ni allergy is common worldwide and in EU, where it affects 10-15% of women 28 . The epidemiological data showed that 12.3-17.7% of the population is allergic to Ni 28 and must follow a Ni-avoidance diet (e.g., Italy, Spain and Poland which have the highest incidence of Ni allergies). Low-Ni tomato products would be of great importance for these patients.
To date, seven tomato allergens have been identi ed and registered by the WHO/IUIS Nomenclature 29 . They are: pro lin (Sola l 1), beta-fructofuranosidase (Sola l 2), fruit 9 kDa lipid transfer protein (9k-LTP, Sola l 3), Bet v 1-like protein (Sola l 4), cyclophilin (Sola l 5), seed 7 kDa lipid transfer protein (7k-LTP, Sola l 6) and a further 9 kDa lipid transfer protein from seeds (9k-LTP, Sola l 7). Additional allergens, or putative allergens, not yet included in the WHO-IUIS nomenclature, such as 11 S globulin, chitinase, glucanase, peroxidase, polygalacturonase, pectin methylesterase, thaumatin-like protein (TLP) and vicilin have been reported and entered in the Allergome database 30 . Pro lin, Bet v 1-like protein and LTP belong to allergen families that have been more widely studied compared to the others. Pro lins and Bet v 1-like proteins are classi ed as class 2 food allergens, that are heat-labile, easily degraded by the gastrointestinal proteases and responsible for localized oral allergy symptoms (OAS) 31 . In contrast, LTP belongs to class 1 food allergens which are heat and protease-stable. They are clinically very relevant allergens because their ingestion, inhalation and contact can cause symptoms that may include all the clinical severity levels of allergic reactions: OAS, gastrointestinal symptoms, urticaria-angioedema syndrome, food-dependent exercise-induced anaphylaxis and even anaphylactic shock 32,33 . The plant LTP family includes two subfamilies, 9k-LTP and 7k-LTP, according to their molecular masses corresponding to 9 kDa and 7 kDa, respectively 34 . However, the allergic sensitization to 9k-LTP is much more prevalent than that recorded for the smaller 7k-LTP. In tomato, both these LTP have been found in seeds and the 9 kDa one has been recorded in the fruit 34,35 .
In most cases, tomato genotypes have been analysed from agronomic and technological point of view without considering Ni content and allergenic protein production that could increase the risk of allergies. Key geochemical processes that lead to limited Ni plant uptake in plant tissues at various growth stages can then be induced in eld using different agricultural practices (irrigation, soil amendments, etc.).
This work aimed at assessing the S. lycopersicum response to Ni on the agronomic yield of tomatoes (i.e., plant biomass and fruit production) and the potential impact of Ni on the production of allergenic proteins (i.e., LTP, TLP, etc.).

Soil Ni concentrations
The peat-sand mix (1:2 v/v) used as growing substrate was analysed by XRF and ICP-MS to determine the Ni concentration at the starting condition. The results evidenced that the Ni mean content of the substrate was 32 mg kg -1 (range 31-32 mg kg -1 ). This mean Ni content was assumed as the background value for the whole experiment. The separate analyses of the two components (peat and sand) evidenced that only the sandy fraction of the mixture was characterized by Ni content above the instrumental detection limits (range 46-47 mg kg -1 ) thus representing the only component of the substrate mix to contribute signi cantly to the initial Ni content of the substrate mix.
Considering the relative nickel loss, at the end of the experiment a signi cant amount of Ni, added as NiSO 4 *6H 2 O, was leached from soil. Relative Ni output increases in a linear way (Figure 1), from Ni 30 to Ni 300. Nevertheless, the nal Ni concentration in soil resulted always higher than the starting value of the untreated substrate.
Plant biomass development and fruit production in response to Ni Considering plant productivity in terms of fruit produced (red, green, and total), biomass and fruit Ni accumulation, Spearman's correlations rank (Table 1) does not highlight signi cant differences between Ni concentrations in fruits and fruit biomass or number.
The Ni treatments do not correlate with tomato productivity while cultivars have negative signi cant correlations respect to the total number of fruits produced (ρ = -0.36 p<0.05; Table 2), but not respect to the fruit biomass. Green (G, unripe) and red (R, ripe) fruits do not show signi cant correlations respect to the other parameters.
Since cv 'Standard' is more productive respect to cv 'Ingrid' in terms of fruits produced, productivity data were analysed grouped per cultivars obtaining the same results previously cited.
The Kolmogorov-Smirnov two-sample test between controls and Ni treatments for Ni concentrations in fruit and fruit biomass (Table 3) showed that there is a signi cant difference between Ni in red, green and total fruit starting from Ni 60 only for cv 'Standard' and Ni 30 for cv 'Ingrid'. However, there are no signi cant differences for fruit biomass and productivity between controls and Ni treatments.
In addition, the Kolmogorov-Smirnov two-sample test to evaluate signi cant differences between controls and Ni treatments respect to plant biomass (root, stem, leaf, fruit DW and dry matter DM) ( Table 4) revealed signi cant difference for 'Standard' from Ni 60 for stem (DW and DM), and from Ni 300 for leaf (DM). For 'Ingrid' revealed signi cant difference from Ni 60 and Ni 120 for stem DM, and from Ni 300 for leaf (DW and DM).
Summarizing, plant biomasses show signi cant differences in response to increasing Ni levels with a clear evidence at Ni 300.
The same test between the two cultivars revealed signi cant differences between fruit and leaf DM, higher in 'Standard', supporting the evidence of a higher productivity of 'Standard' respect to 'Ingrid' ( Table 5).

Ni accumulation in tomato fruit
The one-way ANOVA performed on Ni treatments respect to Ni concentrations in soil and tomato at the end of the experiment ( Figure 2) highlighted signi cant differences from Ni 120 (P=0.0002) and Ni 300 (P=0.0001) respect to the control for soils and tomatoes with a marked signi cant difference also from Ni in tomatoes between Ni 120 and Ni 300 (P= 0.0001).
The Pearson's correlation between nal concentrations of Ni in soil respect to Ni in tomatoes is highly signi cant (r=0.83 P<0.01).

Tomato allergens expressed under Ni stress
Protein extracts preparation from tomato samples Figure 3 shows the protein concentration in the total extracts, ranging from 0.07 to 0.51 mg g -1 of the fruit. The sample showing the highest concentration is C-'Ingrid'-G, followed by C-'Standard'-G and Ni 60-'Ingrid'-R. The Figure shows that the protein concentration found in both the controls of green tomato (C 'Standard'-G and C 'Ingrid'-G) is higher than that observed in the same samples after Ni-treatments.

Analysis of LTP and TLP content in tomato protein fractions using biochemical methods
The analysis of RP-HPLC pro le of tomato total extracts showed that the LTP detection and the estimation of its amount in the samples was not easy. This was especially due to the low concentration of this allergen compared to the other protein components. To overcome this issue, considering that LTP proteins are characterized by a basic isoelectric point, a fraction enriched in basic proteins was obtained from total extracts by separations with an anion exchanger resin. The samples were then concentrated as reported in the Materials and Methods section and their protein pro le was obtained by RP-HPLC. As an example, Figure 4 shows the RP-HPLC pro le of the fraction containing the basic proteins obtained from the extract of C 'Standard'-R sample. The eluted peaks were collected and analysed by direct amino acid sequencing. Peaks eluted at 34.2 min and 34.8 min both provided the same N-terminal sequence, LSCGQVT. The similarity search against the UniProtKB database, with the BLASTP algorithm on the ExPASy server, allowed the identi cation of both the peaks as 9 kDa LTP, Sola l 3. At least two 9k-LTP found in the UniProtKB database had the experimentally obtained N-terminal sequence (accession numbers A0A3Q7HZ96 and K4D1U9). They have been labelled as Sola l 3a and Sola l 3b ( Figure 4). Therefore, the detection in the RP-HPLC elution pro le of more than one LTP peak indicates the presence of isoforms in the analysed samples. Figure 5 shows the amount of Sola l 3 estimated in the analysed tomato samples. It can be observed a certain variability of the Sola l 3 isoforms. However, it is not possible to observe any correlation between LTP concentration and the concentration of Ni applied in the treatments.
The component eluted at 49.2 min was identi ed as TLP (Sola l TLP) by N-terminal amino acid sequencing that provided the following sequence ATKEVRNNCP (Accession number in UniProtKb P12670). Figure 6 shows a decrease of TLP in the standard cultivar as a function of the increasing Ni concentration. The same effect is not observed in the 'Ingrid' cultivar.

Analysis of allergens content in the tomato samples by IgE inhibition tests
The allergens contained in the samples treated with Ni 300 were investigated by immunological tests and the results were compared with those of controls made of untreated tomato. Two samples of Ni 300treated red tomato and green tomato were analysed with the SPHIAa method 36 on the FABER system 37 . Figure 7 shows the IgE-binding inhibition results recorded on some allergens spotted on the FABER biochip, namely the tomato fruit extract, tomato seed extract, Bet v 1 and a Bet v 1-like protein, three pro lins and seven LTP (see Table 6 for details). In line with the observation that these tomato samples contained a very low number of seeds, results obtained show that the inhibition on the entire fruit extract is high, whereas lower values were recorded for the tomato seed extract.
The presence in the fractions of both, red and green fruits, of tomato 7k-LTP, Sola l 6, was indicated by the IgE-binding inhibition (100%) recorded on this allergen spotted on the FABER biochip. Inhibition with variable values was also observed on all the seven 9k-LTP contained in the biochip. Nevertheless, the inhibitions produced by red tomato on these LTP appear independent of Ni treatment. Differently, compared with the control, results obtained with Ni-treated green tomato show a higher inhibition on the peach LTP, Pru p 3, and to a lower extent on other LTP, such as the peanut Ara h 9, the kiwifruit Act d 10 and the maize Zea m 14. All together, these results suggest a higher concentration of LTP Sola l 3 in Ni 300-treated green tomato compared to the untreated samples.
Ni-treated tomato inhibited Bet v 1 with a lower e ciency, compared to the untreated samples. This result suggest that nickel could induce a decrease of a tomato Bet v 1-like allergen. However, the result obtained on the Bet v 1-like allergen Mal d 1 did not produce the same result. Compared to the untreated tomato, both red and green Ni 300-treated ones showed a signi cant lower IgE-binding inhibition on the three tested pro lins. This result is consistent with a reduction of pro lin concentration in Ni-treated tomato. In addition to LTP, pro lin and Bet v 1-like allergen, IgE-binding inhibition was detected also on the pomegranate GRP. Therefore, GRP may represent a new potential tomato allergen. The Ni-treatment seems to give only a weak effect on the GRP concentration in tomato.

Proteins response to Ni in tomato samples
The Pearson's correlation between fruit biomass and proteins expressed and nal concentrations of Ni in soil and in tomatoes (Table 7) highlighted signi cant correlation between Ni in tomato and thaumatin (r=0.40, P<0.01) and Ni in soil and thaumatin (r=0.51, P<0.01).

Discussion
The evaluation of Ni uptake and storage in tomato fruits and related, induced, allergens expression is a key point to set up agricultural practices to limit Ni mobilization from soil.
The sandy fraction of the starting peat-sand mix, i.e., the Ni-bearing fraction of the growing substrate, behaved as an inert medium throughout the 240 days of the experiment. The relative loss of Ni increased linearly with increasing nickel sulphate hexahydrate additions, from Ni 30 through Ni 300 mg kg − 1 . This evidence suggests that most of the Ni in the experimental medium might be leached out with the excess water or absorbed and translocated in the tomato plants. Considering the absence of signi cant precipitation of Ni-bearing salts within the substrate and in the saucer of the pots, it is plausible that most of the nickel has been effectively removed by the growing plants.
In addition, Ni is actively uptaken and stored in fruit when soil concentration exceeded 120 mg kg − 1 . At the end of the treatment each soil replicate had the same amount of Ni, whereas the concentration stored in fruit is as twice as much as in lower concentration (i.e., 30, 60, 120 mg kg − 1 respectively). Data from literature provided different evidence respect to this phenomenon sometimes highlighting a direct correlation of plant Ni as a function of the concentration of Ni in the soil 38, 39 , as in our study. However, growing condition and agricultural practices used are partly comparable.
Uptake from the cultivation soil is the main origin of metals and metalloids in edible parts of the plants.
However, recent reports indicated that tomato has the general ability to store metals in fruits [40][41][42][43][44] () even if those values were sometimes lower than the allowable concentrations by FAO/WHO 45,46 , sometimes with high level of hazard that could be related with environmental pollution (use of fertilizers/pesticides/industrial wastewater irrigation) and air pollution (due to emissions from industries and vehicles) 47 or local intensive agriculture practices, smelting, industries, and wastewater irrigation 42 .
Interestingly, a higher amount of leaf water is directly correlated with the highest Ni treatment, contrary to common response to heavy metal increase in tomato plants 48 . Plants that have speci c mechanisms to balance the impact of metals on plant physiology are able to increase water content at the leaf level to counteract the metal intake and also increase resistance to drought stress [49][50][51] .
Usually, controlled studies have shown that a high concentration of bioavailable Ni results in Ni toxicity, causing a measurable reduction in plant biomass 52,53 . The toxicity of metals, including Ni, on plant growth and water content 54-56, 53 is manifested as a decrease in transpiration through the decrease in water content and stomatal conductance together exert toxic effects on photosynthesis 4 , leading to a decrease in the photosynthetic rate. Interestingly, the highest level of water at the leaf level, when soil Ni reaches 300 mg kg − 1 , might suggest a speci c osmotic reaction to counteract the Ni intake. This is known in other plants, speci cally hyperaccumulators, where water intake represents a selected response to alleviate metal stress and Ni toxicity 57 .
The analysis of the total protein concentration in the different tomato samples revealed some high variations, even between the fruits that had received the same Ni treatment, thus suggesting that many factors can affect the protein content. However, the calculation of average amounts reduced the observed differences between the analysed fruits. In fact, aside from a few exceptions, most of the samples showed average values of about 0.3 mg of proteins per g of fruit. Nevertheless, no correlation between the protein concentration and Ni treatments was observed.
LTP is a relevant allergenic protein that can cause severe symptoms. It belongs to the group 14 of pathogenesis-related proteins . The estimation of the LTP content using biochemical methodologies did not allow the detection of any correlation between the Ni treatments and the concentration of this allergen in tomato. Nevertheless, RP-HPLC chromatographic separations show a clear variation of the concentration of two 9 kDa LTP (Sola l 3) isoforms contained in the tomato samples. However, the analysis of Sola l 3 isoforms as separate components, or as sum of the two detected isoforms added together, do not show correlation with the Ni treatments. The lack of this kind of correlation was con rmed, at least for red tomato, following the analysis of the LTP content performed with immunological inhibition tests consisting in the competition of extract components with the allergenic molecules spotted on the FABER biochip 37 . Conversely, IgE-binding inhibition results obtained with the green tomato suggest a possible Ni effect producing a slight increase in LTP content. In fact, we detected a much higher inhibition on the peach LTP, Pru p 3, produced by the Ni-treated green tomato compared with the control. The inhibition on the other tested LTP was not always in line with this outcome. However, the heterogeneity of the inhibition results could be due to the different IgE-binding epitope panels associated to the individual tested LTP and to the multiple LTP isoforms contained in tomato. Anyway, the result obtained for Pru p 3, that is an allergen well-known as the most powerful one of this protein family 58 , prompt further future investigations to better understand the Ni effects on the concentration of this relevant allergen.
In the standard tomato cultivar, a decrease of thaumatin as a function of Ni increase was observed by biochemical methods. Tomato thaumatin is a potential allergen although its clinical relevance is not clear yet. It is a pathogenesis-related protein belonging to the group 5 (PR-5). Similar to other components of PR-5, tomato thaumatin is involved in the plant response to biotic and abiotic stresses and many factors can promote induction and regulation of its expression 59 . Using immunological methods, a signi cant decrease of the allergenic pro lin Sola l 1 in the tomato samples treated with Ni 300 was observed. This effect is in line with the literature report describing the decrease of pro lin in the leaves of basil grown on soil containing 500 mg kg − 1 of Ni 60 . Therefore, the experimental results here described con rm the already reported Ni effect in decreasing the pro lin concentration in the plant tissues. Pro lins are actinbinding proteins, present in the cytoplasm of all eukaryotic cells where they play a key role in cell physiology. They are involved in processes such as organ development, wound healing, and defence from biotic attacks. Similarly, the Bet v 1-like proteins are pathogenesis-related molecules involved in host defence and their amount in Ni-treated tomato appears decreased. They belong to the group 10 (PR-10). Therefore, beyond the allergological implications, the observation that Ni can cause a decrease of the amount of thaumatin, Bet v 1-like protein and pro lin, which are proteins involved in relevant physiological processes and host defence, suggests a possible effect of this chemical element in weakening the plant, which can become more sensitive to environmental biotic and abiotic attacks.
The immunological experiments also allowed the detection of a new potential allergen, belonging to the family of GRP, not yet reported in tomato. In fact, this component contained in the tomato fraction competed with the pomegranate GRP 61 for the binding of speci c IgE. GRP are proteins involved in plant development and their expression is up-regulated by the plant hormone gibberellin. Results obtained in this study suggest a low effect of Ni treatment on the concentration of this protein in tomato, but a con rmation by additional investigations is desirable.
In our case, the ability of transferring Ni from soil to fruit without affecting plant viability in terms of biomass, number of fruits produced and fruit weight, poses signi cant risks to consumers. In fact, without the evidence of plant suffering, these Ni-rich fruit can potentially be eaten by consumers. The choice of low-Ni practices should be considered avoiding potentially contaminated matrices like wastewater or low-quality compost.

Substrate preparation
The peat-sand mix (1:2 v/v) was chosen as growing substrate; the substrate was autoclaved at 120°C for 20 min, and oven-dried at 60°C.
The nal pH of substrate was measured with a pHmeter (pH 210, Hanna Instruments) by mixing an aliquot of soil with deionized water (ratio 1:3).
To determine the Water Holding Capacity (WHC), soil was transferred into a 10 cm Ø pot, 100 ml of water were added to 100 ml of dry soil placed in a funnel on a graduated cylinder. After waiting at least 1 hour until the last drop, the WHC (%) was calculated based on the volume of water retained by the soil.

Experimental design
Three-months-old plants of Solanum lycopersicum L. 'Cuor di Bue Standard' (n = 25) and 'Cuor di Bue Ingrid' grafted Beaufort (n = 25) were purchased from a plant nursery specialized in tomato cultivars, in compliance with national and international regulations, then transferred to the experimental greenhouse of the University of Genoa. Plants were transplanted to 18 cm Ø pot containing 2 kg of substrate previously described (one plant per pot, 5 replicates each concentration) adjusted to pH 6.00. At the beginning of the experiment, soil was homogeneously hydrated up to 70% WHC with solutions of sterile deionized water and metallic salt (NiSO 4 *6H 2 O) at increasing concentrations of Ni (0, 30, 60, 120, 300 mg kg − 1 , respectively) to evaluate the plant response. The Ni concentrations were chosen based on the threshold values for environmental Ni, as mandated by European laws 6 .
Pots were transferred to greenhouse and the plants were grown under semi-natural conditions at controlled temperature (21-27°C) for 240 days and were irrigated two times a week with tap water to maintain 70%WHC. Each pot had a saucer to recover drainage water.
A water-soluble fertilizer Leader N-P-K (20-10-20 + MgO + Me) was dissolved in water at the concentration of 4 g l − 1 and supplied for each pot once a week.

Soil, plant, and fruit sampling
At the end of the experiment, the plants were subdivided into roots and shoots and were thoroughly rinsed, rst with tap water and then with deionized water. Soil was rstly removed manually from the root system and then thoroughly rinsed away with tap and deionized water. Soil was further sieved to collect the remaining thin roots. The fresh biomasses of the different plant organs were weighed separately. Mature fruits were collected during the whole fruiting stage (start on June until November 2017) and weighed to evaluate fresh biomass.
Soils, roots, shoots, and fruits were then oven-dried at 60°C for 48 hours (soils and plants) and for 96 hours (fruits). Dried samples were weighed to evaluate dry biomass and powdered using a ball mill (Retsch MM2000, Haan, Germany).
To evaluate plant water content, the Root/Shoot ratio for both fresh and dry biomass and the water content (100 * DW/FW) in root and shoot were calculated 57 .

XRF and ICP-MS analyses
The chemical analyses for nickel were carried out on dried and porphyrized soils and plant samples by using the X-MET7500 Field Portable X-Ray Fluorescence (FP-EDXRF) Spectrometers (Oxford Instruments) thanks to the collaboration with Geospectra S.r.l, Spin-Off of the University of Genoa. The FP-EDXRF instruments were calibrated using both fundamental parameters calibration determined by the manufacturer and site-speci c calibration standards (SSCS) representative of the matrix analysed by FP-EDXRF.

Estimation of the protein concentrations
The protein concentration of the total protein extract was determined by the Bradford method with the BIO-RAD Protein Assay (Biorad, Milan, Italy), using a calibration curve made with bovine serum albumin.
To estimate the amount of 9k-LTP (Sola l 3) and thaumatin-like protein (Sola l TLP) contained in each sample (expressed as µg of protein per gr of fresh tomato), the area of the RP-HPLC peaks was calculated and compared with that of peaks obtained with a known amount of tomato 9k-LTP and TLP, respectively.

Amino Acid Sequencing
Proteins collected from RP-HPLC were concentrated with a centrifugal vacuum concentrator (Savant Speedvac Plus SC110 A, Ramsey, Minnesota, USA). Next, a protein amount corresponding to about 200 mAU at 220 nm was subjected to automated direct sequencing of the N-terminal region with a Protein Sequencer PPSQ-33B (Shimadzu Corporation, Tokyo, Japan).
IgE inhibition experiments with the SPHIAa assay on the Patients' sera Sera used in this study were selected among those stored in the serum bank of ADL. These are residual sera deriving from venous blood sampling made for the routine allergy diagnosis by FABER test 31,68 . The features of each serum, in terms of content of IgE antibodies able to recognize and bind speci c individual allergens (speci c IgE) spotted on the FABER biochip, are recorded in the InterAll databank (version 5.0, Allergy Data Laboratories). Sera were selected based on the speci c IgE content. The chosen ones were free of IgE recognizing cross-reactive carbohydrate determinants (CCDs). In fact, they were tested negative against CCD-bearing proteins used as markers, namely bromelain from Ananas comosus and peroxidase from Armoracia rusticana.
In the SPHIAa experiments, IgE is used as a probe to detect the presence of structural determinants, that is the epitopes of the proteins (puri ed or in mixture) under investigation. Therefore, the selection of sera was independent of the clinical history and/or symptoms of patients. For the SPHIAa assay, a pool of four sera able to recognize relevant plant food allergens was prepared. They contained IgE recognizing allergens such as LTP, pro lin, Bet v 1-like proteins, GRP and thaumatin-like protein. The nal dilution of each individual serum co-incubated with the tomato extract sample was 1:8.
All patients gave their informed consent to the use of their clinical data for research purposes in an anonymous form. In view of the purely comparative nature of this study, along with the fact that all venous blood samplings were part of routine clinical practice and that a residual part of the routine sample was used for inhibition experiments, a formal approval by the Ethical Committee was not necessary.

Data analysis
The statistical analyses were performed with Statistica 8.0 (Statsoft Inc.) software. Results below the detection limits are presented as zero and were used as such in the calculations.
Nonparametric tests were used to avoid data transformation. Normality of parameters were evaluated with the Shapiro-Wilk test. Correlations between variables were analysed using Spearman's correlation coe cient (ρ) using P < 0.05 to indicate statistical signi cance, since most of data exhibit a non-normal statistical distribution.
The Kolmogorov-Smirnov two-sample test was used to evaluate differences between control and Ni treatments.
Parametric analyses were used for Ni in fruit and soil respect to allergenic proteins' production.
The dry matter (DM%), de ned as the fruit dry weight (DW)/fruit fresh weight (FW)*100 70-72 is also calculated for plant organs to evaluate the overall plant biomass production in response to an abiotic stress as in the case of soil nickel. Abbreviations DM, dry matter; DW, dry weight; FABER, P-Friendly Allergen Nano Bead ARray; FP-EDXRF, Field Portable X-Ray Fluorescence; FW, fresh weight; ICP-MS, inductively coupled plasma-mass spectrometry; 9k-LTP, Sola l 7, 9 kDa lipid transfer protein from seeds; RP-HPLC, reversed phase high-pressure liquid chromatography; Sola l 1, pro lin; Sola l 2, beta-fructofuranosidase; 9k-LTP, Sola l 3, fruit 9 kDa lipid transfer protein; Sola l 4, Bet v 1-like protein; Sola l 5, cyclophilin; 7k-LTP, Sola l 6, seed 7 kDa lipid transfer protein; SPHIAa, IgE Single Point Highest Inhibition Achievable assay; SSCS, site-speci c calibration standards; TLP, thaumatin-like protein; TFA, Tri uoroacetic Acid; WHC, Water Holding Capacity. Declarations ACKNOWLEDGMENTS Authors wish to thank Elena Mora, curator at Genoa Botanic Garden and Stefano Rosatto for the technical support in the greenhouse experiments. We also thank Chiara Rafaiani and Michela Ciancamerla for their skilful work in doing multiplex IgE inhibition experiments.