Effects of Cerium and Titanium Oxide Nanoparticles in Soil on the Nutrient Composition of Barley (Hordeum vulgare L.) Kernels

The implications of metal nanoparticles (MeNPs) are still unknown for many food crops. The purpose of this study was to evaluate the effects of cerium oxide (nCeO2) and titanium oxide (nTiO2) nanoparticles in soil at 0, 500 and 1000 mg·kg−1 on the nutritional parameters of barley (Hordeum vulgare L.) kernels. Mineral nutrients, amylose, β-glucans, amino acid and crude protein (CP) concentrations were measured in kernels. Whole flour samples were analyzed by ICP-AES/MS, HPLC and Elemental CHNS Analyzer. Results showed that Ce and Ti accumulation under MeNPs treatments did not differ from the control treatment. However, nCeO2 and nTiO2 had an impact on composition and nutritional quality of barley kernels in contrasting ways. Both MeNPs left β-glucans unaffected but reduced amylose content by approximately 21%. Most amino acids and CP increased. Among amino acids, lysine followed by proline saw the largest increase (51% and 37%, respectively). Potassium and S were both negatively impacted by MeNPs, while B was only affected by 500 mg nCeO2·kg−1. On the contrary Zn and Mn concentrations were improved by 500 mg nTiO2·kg−1, and Ca by both nTiO2 treatments. Generally, our findings demonstrated that kernels are negatively affected by nCeO2 while nTiO2 can potentially have beneficial effects. However, both MeNPs have the potential to negatively impact malt and feed production.


Introduction
Nanotechnology is expected to be in widespread use by 2020, promoting a market of three trillion dollars' worth of nanotechnology-based products with six million workers [1]. The applications of nanotechnologies in medicine, consumer goods, heavy industry, information and communication technologies, electronic devices, and environmentally friendly energy systems are developing at a much faster pace than our knowledge of their impact [2]. Several questions have been raised about the fate of nanomaterials (defined as materials with at least one dimension between 1 and 100 nm) [3] used in the agro-environment and those resulting from uncontrolled or accidental flows such as metal nanoparticles (MeNPs). For this reason, the Food and Agriculture Organization of the United Nations (FAO), as well as many countries adhering to the Organization for Economic Co-operation and Development (OECD), recognized the need for early research to fill the existing gaps in knowledge on the toxicity of nanoparticles, their bioaccumulation, oral exposure and risks of ingestion by target organisms, which are the key factors needed for risk assessment of nanomaterials [4,5].
Human diets obtain minerals primarily from grains. However, uptake of nutrients by plant roots is affected by abiotic and biotic stressors and so is the mineral storage in plant organs [6]. It is therefore appropriate to examine whether MeNPs in soil are able to influence mineral accumulation in the grains of food crops. Vascular plants and especially crops are of concern as they could be exposed to risks from MeNPs bioaccumulation and their subsequent entry into the food chain [7,8]. The effects of MeNPs on plants will depend, among other things, on the soil type and plant species [9], and indeed contrasting results are reported among dicot and monocot plants [10][11][12][13][14][15][16]. It is therefore crucial to investigate the impact of MeNPs on crops.
This paper reports the observations of barley (Horedum vulgare L.) kernels produced by plants grown in nCeO 2 or nTiO 2 amended soil. Barley is among the world's most important crops, being the fourth cereal after maize, rice and wheat for global grain production in 2013 [37]. Due to its adaptation, N uptake and utilization efficiency (N UtE ), barley is cultivated on more than 49 million hectares [37,38]. It has relevant economic importance for animal feed, malting and brewing and is an important food in some parts of the world [37].
The main aims of our experiment were to: (i) evaluate the bioaccumulation of Ce and Ti in kernels; (ii) verify the mineral nutrients content in kernels compared to that in control plants; and (iii) monitor the changes, if any, in seed quality parameters.

Soil Characterization
The soil used in our pot experiment was collected from the top 40 cm of an agricultural field at the University of Udine, Italy (46˝04 1 53 11 N, 13˝12 1 34 11 E). The soil was air dried and sieved through a 2-mm sieve. For the soil characterization, samples (5 replicates) were oven-dried at 40˝C for 48 h. The samples were then analyzed according to Pansu and Gautheyrou [39] for particle size distribution (Bouyoucos hydrometer method), pH (potentiometric measurements in a 1 to 2.5 of soil and Milli-Q ® water suspension), electrical conductivity (EC, conductometric measurements in a 1 to 5 of soil and Milli-Q ® water suspension), cation exchange capacity (CEC), available P (sodium bicarbonate extractable P at pH 8.5, Olsen method) and equivalent carbonate (calcimeter method). Total organic carbon (TOC) and total nitrogen (TN) contents were determined through an Elemental CHNS Analyzer (Vario Micro Cube, Elementar Analysensysteme GmbH, Hanau, Germany) using up to 10 mg of finely ground soil. Carbonates from the soil were previously removed by adding drops of hydrochloric acid (18%). The soil was classified as sandy clay loam and its characteristics are reported in Table 1.

Nanoparticles Characteristics
Cerium(IV) oxide (nCeO 2 ) and titanium(IV) oxide anatase (nTiO 2 ) nanoparticles were purchased from Sigma-Aldrich (Milwaukee, WI, USA) (ID product 544841 and 637254, respectively), which described them as having mean diameters of <25 nm (BET). Previous characterization showed that nCeO 2 and nTiO 2 are of different shapes, mainly rhombus and disks, with an average size˘SE of 22.7˘1.3 nm and 24.1˘0.7 nm (TEM). The specific surface area of nCeO 2 and nTiO 2 resulted as being 46.1 m 2¨g´1 and 61.6 m 2¨g´1 (BET), respectively. The size distribution of the nanoparticle powders by AFM and in aqueous metrics by DCS, zeta potential and size in aqueous metrics by DLS have been reported previously [40].

Plant Growth and Yield Parameters
Nanoparticles were added to the soil before sowing [40]. Four mixtures of soil were prepared with nCeO 2 or nTiO 2 at 500 and 1000 mg MeNPs¨kg´1. The control treatment received no MeNPs. Five polyethylene pots (4 L each) were prepared per treatment (for a total of 25 pots). The range of concentrations (0, 500, and 1000 mg MeNPs¨kg´1) was chosen considering that nanomaterial content in soil is expected to grow in the future due to anthropic activities [41]. Similar concentrations are also reported in numerous articles dealing with nanoparticles interaction with higher plants [42].
A greenhouse experiment was initiated at the experimental farm of the University of Udine (Italy) on 9 April 2014. Seeds of a two-row spring barley (Hordeum vulgare L., var. Tunika) obtained from S.I.S. S.p.A. (San Lazzaro di Savena, Bologna, Italy) were sown in pots [40]. More details regarding the experimental design, soil, and greenhouse conditions were reported in a previous paper [40].
All seeds germinated and pots were watered to maintain the soil at 60% of water holding capacity (WHC). During growth pots were individually weighed and watered on a weekly basis to compensate for evapotranspiration. At Zadoks growth stage 92 (Ripening, kernel is hard and can no longer be dented by thumb-nail) [43], kernels were harvested, counted and weighed for 100-kernels weight and grain yield estimation. Flag leaf area was measured using a LI-3100C Area Meter (Li-Cor Corporation, Lincoln, NE, USA). Kernels from the main shoot were then separated for the analyses described below. A subsample of kernels was oven-dried at 105˝C for 48 h for ICP-AES analysis while for the other analyses kernels were oven-dried at 60˝C for 48 h and ground to fine powder.

Amino Acid and Crude Protein Analysis
Acid hydrolysis with HCl at 110˝C for 22-24 h was used for total amino acids except for sulfur amino acids and tryptophan. For sulfur amino acids, performic acid oxidation for 16 h followed by acid hydrolysis with HCl was used. For tryptophan, alkaline hydrolysis with sodium hydroxide was performed at 100˝C for 4 h. After extraction, samples were derivatized at 55˝C for 10 min with 20 µL of AccQ-Fluor reagent (6-aminoquinolyl-N-hydroxysuccinimidyl carbamate) and injected in HPLC [44]. All reagents were of HPLC grade. Amino acids analysis was performed using a LC 200 Perkin Elmer pump fitted with an ISS-100 auto sampler (20 µL loop) and a fluorimetric detector (Perkin Elmer, Norwalk, CT, USA), λ excitation of 250 nm and λ emission of 395 nm. Separation was achieved by using one AccQ-Tag amino acid analysis column (Waters Corporation, Milford, MA, USA) and one Waters pre-column filter. The column was thermostatted at 37˝C and the flow rate was 800 µL¨min´1 [45]. Mobile phase A consisted of acetate-phosphate aqueous buffer, mobile phase B was acetonitrile 100% and eluent phase C was Milli-Q ® water. L-α-amino-n-butyric acid was used as internal standard. For the analysis of cysteine, methionine and tryptophan it was necessary on some occasions to mix two or three samples to obtain sufficient material for analysis.
Total N content was determined through an Elemental CHNS Analyzer (Vario Micro Cube, Elementar Analysensysteme GmbH, Hanau, Germany) using up to 2.5 mg of finely ground samples. Lastly, crude protein (CP) was estimated by multiplying the nitrogen content by 5.45 [46].

Element Concentrations Analysis
About 300 mg of material were digested in 10 mL of a 1 to 4 (v/v) mixture of 37% (v/v) HCl (99.999% trace metals basis, Sigma-Aldrich) and 67% (v/v) HNO 3 (RS superpure for trace analysis, Carlo Erba Reagents, Cornaredo, Italy) in Teflon cylinders for 10 min at 175˝C in a microwave oven (CEM, MARS Xpress, Matthews, NC, USA). After mineralization plant extracts were filtered (0.45 µm PTFE) and diluted. Total B, Ca, Cu, Fe, K, Mg, Mn, Na, Ni, P, and Zn contents were determined by an ICP-AES (Vista MPX, Varian Inc., Palo Alto, CA, USA) with an internal standard solution of Y. Total Ce and Ti contents were determined by an ICP-MS (Aurora M90, Bruker, Bremen, Germany) with an internal standard solution of 72 Ge and 89 Y. The accuracy of the analytical procedure adopted for ICP-AES analysis was checked by running standard solutions every 20 samples. Quality control for both ICP-AES and ICP-MS was performed using reagent blank samples, and triplicate readings for each sample. Certified standard reference material (tomato leaves 1573a from the National Institute of Standards and Technology, Gaithersburg, MD, USA) was treated as the samples (n = 3). The recovery of the elements in the standard material was on average 97% of the certified values with an RSD average of 1.3%. Method detection limits in ICP-AES were: 160 µg¨L´1, 56 µg¨L´1, 4 µg¨L´1, 10 µg¨L´1, 21 µg¨L´1, 77 µg¨L´1, 110 µg¨L´1, 3 µg¨L´1, 54 µg¨L´1, 8 µg¨L´1, 31 µg¨L´1, 5 µg¨L´1 and 6 µg¨L´1 for Ca, Ce, Cu, B, Fe, K, Mg, Mn, Na, Ni, P, Ti and Zn respectively. Method detection limit in ICP-MS was 1 µg¨L´1 for both Ce and Ti.
Total S content was determined through an Elemental CHNS Analyzer (Vario Micro Cube, Elementar Analysensysteme GmbH, Hanau, Germany) using up to 2.5 mg of finely ground samples.

Statistical Analysis
The experiment was set up as a completely randomized design and pots were randomly reallocated each week. Each analysis consisted of five replicates per treatment, unless otherwise stated. Two-way mixed effects analysis of variance (ANOVA) [47] was performed for all the data after verification of normality (Kolmogorov-Smirnov test) [47] and homogeneity of the variance (Hartley's F max -test) [47]. The multiple comparisons of means were based on the minimum significant difference (MSD) method obtained from the T statistic for equal replicates or from the T' statistic for not equal replicates in the case of cysteine, methionine and tryptophan analysis [47]. Pearson's product-moment correlation coefficients and significance were calculated between Ce or Ti content in kernels and all the other measured parameters [47].

Barley Biometric and Yield Parameters
Flag leaf area and 100-kernels weight were unaffected by the presence of cerium oxide (nCeO 2 ) and titanium oxide (nTiO 2 ) nanoparticles in the soil (Table 2). However, a reduction in flag leaf area and an increase in 100-kernels weight can be discerned under nCeO 2 treatments but this is hidden by a high SE. Instead, nanoparticles treatments affected plants in terms of kernel quantity and grain yield, but the dose-responses are unclear for these traits (Table 2). In both nCeO 2 and nTiO 2 treatments the 500 mg¨kg´1 level impaired kernel quantity and plant grain yield compared to the control; conversely, the 1000 mg¨kg´1 treatments did not cause any significant limitation and, although not significant, there was a beneficial effect from the 1000 mg nTiO 2¨k g´1 treatment (Table 2).

Amylose and β-Glucans Concentrations in Kernels
The effects of nCeO 2 and nTiO 2 on amylose and β-glucans concentrations in barley kernels are displayed in Table 3. On the whole, metal nanoparticles (MeNPs) treatments had a marked effect on starch composition, as the amylose concentration significantly decreased (till 21% on average) compared to the control. However, there were no specific responses across the different MeNPs treatments. No significant differences were found in β-glucans concentration between control and MeNPs-treated plants. Table 3. β-Glucans and amylose concentration in barley kernels at ripening from main shoot grown in soil treated with none (Control), 500 mg nCeO 2¨k g´1, 1000 mg nCeO 2¨k g´1, 500 mg nTiO 2¨k g´1 and 1000 mg nTiO 2¨k g´1.

Amino Acids Concentrations in Kernels
The effects of nCeO 2 and nTiO 2 in the soil on amino acid concentration and crude protein (CP) in kernels are displayed in Table 4. Overall, glutamic acid, proline and leucine were the most abundant amino acids in kernels with concentration ranges of 32-43, 15-21 and 9-12 mg¨g´1 respectively (Table 4).
More in detail, nCeO 2 and nTiO 2 did not significantly modify concentrations of arginine and tryptophan but significantly and similarly increased cysteine, glutamic acid lysine and proline concentrations in all treatments with respect to the control. Aspartic acid concentration was unaffected at 500 mg¨kg´1 of both nCeO 2 and nTiO 2 while significantly and similarly increased at 1000 mg¨kg´1 of both nCeO 2 and nTiO 2 .
The other amino acid responses varied depending on which MeNPs were added to the soil. In particular, nCeO 2 treatments did not modify concentrations of alanine, glycine, histidine, isoleucine, leucine, phenylalanine, serine, tyrosine and valine. On the contrary, methionine and threonine concentrations significantly increased only at the highest nCeO 2 treatment. Both nTiO 2 treatments significantly increased alanine, glycine, isoleucine, leucine, methionine, threonine, tyrosine and valine concentrations. Conversely, only the highest nTiO 2 treatment significantly increased histidine, phenylalanine and serine concentrations. However, in most amino acids the highest concentration was consistently measured at 1000 mg nTiO 2¨k g´1 treatment.
Finally, CP in kernels was significantly increased by approximately 19% in MeNPs-treated plants compared to the control (Table 4). Table 4. Amino acid (mg¨g´1) and crude protein (CP, %) concentration in barley kernels at ripening from main shoot grown in soil treated with none (Control), 500 mg nCeO 2¨k g´1, 1000 mg nCeO 2¨k g´1, 500 mg nTiO 2¨k g´1 and 1000 mg nTiO 2¨k g´1.

Ce and Ti Concentrations in Kernels
The effects of nCeO 2 and nTiO 2 in the soil on Ce and Ti concentrations in barley kernels are displayed in Table 5. We did not observe significant differences among the treatments for Ce and Ti concentrations in kernels as the SE of means were high (on average 45% of the mean). Nevertheless, a similar mean and a nearly homogeneous SE among treatments for Ce concentrations indicate that there were no differences among treatments. With regard to Ti, similar observations can be made except for the 1000 mg nTiO 2¨k g´1 treatment that had the highest mean but also a relatively high SE. Thus, although not significantly different, at the 1000 mg nTiO 2¨k g´1 treatment an enhancement of Ti can be envisaged with respect to the other treatments. Table 5. Cerium and titanium concentration in barley kernels at ripening from main shoot grown in soil treated with none (Control), 500 mg nCeO 2¨k g´1, 1000 mg nCeO 2¨k g´1, 500 mg nTiO 2¨k g´1 and 1000 mg nTiO 2¨k g´1.

Nutrient Elements Concentrations in Kernels
Cerium oxide nanoparticles (nCeO 2 ) treatments did not significantly modify several macro-nutrient concentrations (Ca, Mg, and P), Na (a beneficial nutrient) ( Table 6) and all micronutrients except B (Table 7). Of the affected macro-nutrients, K was significantly lowered by both nCeO 2 treatments, while S only by 1000 mg nCeO 2¨k g´1 treatment. Under 1000 mg nCeO 2 treatment, B concentration was significantly lowered by approximately 62% compared to the control. Table 6. Concentration of macro-nutrients, Na, and Na/K and Na/Ca ratios in barley kernels at ripening from main shoot grown in soil treated with none (Control), 500 mg nCeO 2¨k g´1, 1000 mg nCeO 2¨k g´1, 500 mg nTiO 2¨k g´1 and 1000 mg nTiO 2¨k g´1.

Treatment
Ca (mg¨kg´1 dw) K (mg¨kg´1 dw) Mg (mg¨kg´1 dw) Na (mg¨kg´1 dw) P (mg¨kg´1 dw) S (mg¨kg´1 dw) Na/K Na/Ca  Regarding nTiO 2 treatments, Ca was significantly increased and S significantly lowered by both nTiO 2 treatments, while K was lowered only by 1000 mg nTiO 2¨k g´1 (Table 6). Among micronutrients, only Mn and Zn reached significantly higher concentrations at 500 mg nTiO 2¨k g´1 but not at 1000 mg nTiO 2¨k g´1 treatment (Table 7).
In all the treatments the Na/K ratio was unaffected while the Na/Ca ratio was significantly affected by both nTiO 2 treatments (Table 6).

Pearson's Product-Moment Correlation
Pearson's product-moment correlation coefficients showed that only Mn (r = 0.859, df = 13, p < 0.001) concentration was positively and significantly correlated with Ce concentration in kernels from different concentrations of nCeO 2 in soil. On the contrary, there were no significant correlations between Ti concentration in kernels and all the other measured parameters under nTiO 2 treatments.

Barley Biometric and Yield Parameters
There could be a parallel effect on source and sink organs in nCeO 2 treated plants: the apparent reduction in flag leaf area is tentatively mirrored by a severe drop in sinks volume (kernel quantity). The severe reduction in kernel quantity under the 500 mg nCeO 2¨k g´1 treatment (and to a lesser extent under the 1000 mg nCeO 2¨k g´1 treatment), is associated with an increase in 100-kernel weight, but the compensation is not sufficient and the resulting plant grain yield was negatively affected. Our results are partially in agreement with those obtained by Rico et al. [16], who observed a reduced spike production in barley. Remarkably, these authors observed that plants exposed to 500 mg nCeO 2¨k g´1 did not form grains [16].

Amylose and β-Glucans in Kernels
Starch is the main carbohydrate in barley, accounting for approximately 65% to 75% of the grain dry weight. Amylose and amylopectin are the two components of starch. Barley types can be classified as normal (25%-27% amylose), waxy (below 5% amylose) and high-amylose (>35% amylose) [48]. The barley material used (cv. Tunika) was released as a two-row spring barley, and marketed in Italy for malt or feed production, as its high-amylose and moderate β-glucans traits are functional for both purposes. In general, the high-amylose trait in cereals is connected to resistant starch, which in turn has a positive role in human nutrition [49]. Amylose content has been tagged as the most sensitive parameter to heat stress in Japonica rice, maize and wheat [50], while there was no conclusive evidence for consistent changes in amylose content in barley grains exposed to high temperatures [51]. However, there was a decrease of amylose as a response to abiotic stress in our barley experiment.
Consistently with our results, Rico et al. [10] observed a lowered starch content in grain from rice treated with 500 mg nCeO 2¨k g´1 although significantly only in two of the three analyzed varieties. Conversely, Zhao et al. [14] did not observe significant differences in starch content in cucumber fruit. However, as well as analyzing a different species, they determined starch content in the whole cucumber fruit and not separated for skin, pulp and seeds [14].
Finally, as there were no significant differences in β-glucans concentration between control and MeNPs-treated plants, we hypothesize that MeNPs uptake has a limited impact on cell wall deposition in barley endosperm.

Amino Acids in Kernels
Our data on amino acid composition are consistent with other barley kernel composition studies [52,53]. In general, a total of 18 amino acids have been identified in barley proteins: with respect to animal growth requirements, lysine and threonine are the first and second most limiting amino acids, with methionine and tryptophan in third and fourth positions, respectively [53]. Since lysine is the first limiting amino acid in cereal grain protein, an increase in its level results in improved nutritional quality. However, high lysine strains and mutants (as Hiproly barley, found in the world barley collection in 1969) have been found to be associated with lower grain weight, and reduced yields.
A positive correlation between proline in leaves and plant stress is supported by a large body of data. Besides acting as an excellent osmolyte, proline plays different roles during stress, as a metal chelator, an antioxidative defense molecule and a signaling molecule [54]. An overproduction of proline, which in turn imparts stress tolerance by maintaining cell turgor or osmotic balance, is a common observation in stressful environments. We did not follow proline in leaf tissues, but MeNP-treated plants added on average 37% more proline in kernels compared to controls, which is the second largest increase, after the 51% increase in lysine. Metal nanoparticles (MeNP) treatments therefore probably operate as abiotic stressors on barley plants, and the observed proline increase in kernels could be related to the in-season proline evolution in green tissues. Contrasting results were reported by Rico et al. [13,16] for different amino acids in both barley and wheat exposed to nCeO 2¨k g´1.
Taking into account that MeNPs caused a decrease in grain yield per plant (´25%) comparable to the increase in CP (19%) compared to the control, the increased CP concentration could be due to less CP dilution by a reduced carbohydrate accumulation in MeNPs-treated plants.
From a different standpoint, the increased CP is in line with the increased total amino acids concentrations under MeNP treatments, on average 23% higher than in controls. Apart from tryptophan, similar effects were observed for each amino acid separately. Indeed, the magnitude of CP, total amino acids and single amino acid concentrations in MeNP treatments were comparable.

Ce and Ti in Kernels
As reported in our previous studies in vitro [21] and in kernels from shoots other than the main one [40], we did not observe significant differences among the treatments for Ce and Ti concentrations. This is in agreement with authors who studied different species under nCeO 2 [13,15,19], or under nTiO 2 exposure in soil [55]. It could be due to high data variation among replicates within treatments, but as the data overlapped a great deal among treatments, we are confident there is no enhanced Ce translocation into kernels under nanoparticle treatments compared to the control. An enhancement of Ti translocation under 1000 mg nTiO 2¨k g´1 treatment is instead notable. However, our results suggest that translocation of manufactured nCeO 2 and nTiO 2 (at least of nominal sizes of 25 nm) into barley kernels is hardly possible. Unfortunately, we were unable to test this hypothesis under TEM-EDAX analysis. In fact, due to the very hard seed coat, we were unable to cut kernels into slices, as required for TEM-EDAX analysis. Thus, additional efforts are needed to determine the content and distribution of Ce and Ti in the kernels. However, other authors observed that nCeO 2 treatments largely increased Ce concentration in fruits and seeds of different species [10,36,56] including barley [16]. The contrasting data reported in barley by Rico et al. [16] can be explained by the different physical-chemical properties of the nanoparticles used in the two studies as discussed in more detail below in Section 4.5. A root to fruit translocation of nTiO 2 was also observed in cucumber after plant exposure to nTiO 2 in soil [57].

Nutrient Elements in Kernels
Potassium is the most abundant cation in cytoplasm, actively maintaining osmotic potential and stabilizing pH between 7 and 8, the optimum range for most enzyme activities. As it is also necessary for protein synthesis and other metabolic processes [6], its reduction at both nCeO 2 levels and at 1000 mg nTiO 2¨k g´1 treatment could have a negative impact on enzyme activities (and kernel quality).
Sulfur is a structural constituent of coenzymes and secondary plant products containing amino acids cysteine and methionine and can also act as a functional group directly involved in metabolic reactions [6]. Therefore the reduction in S that occurred at both nTiO 2 treatments could theoretically affect glutathione synthesis and the antioxidant capacity of kernels. However, the expected reduction in cysteine and methionine was not observed; in fact, they were enhanced (Table 4). We speculate that even if S translocation to the kernels was lowered, the small amount of available S was nonetheless effectively directed to protein synthesis, avoiding its use for other purposes.
Boron is involved in a number of metabolic pathways and the severe reduction in B concentration could indicate a restricted permeation of B due to nCeO 2 presence.
Cerium oxide nanoparticles interfered with Mn, demonstrating a tendency for Mn accumulation at increasing levels of Ce in kernels as demonstrated by a significant Pearson's product-moment correlation. With respect to the few published articles about Ce correlation with other elements in grains of barley and other species, we did not find any evidence of correlation between other elements and Ce [10,13,16].
Other authors reported contrasting data and data only partially in agreement with our results. For example, similarly to our results, Rico et al. [10] observed no changes in concentrations of most of the nutrients in grain from three rice varieties grown in soil with 500 mg nCeO 2¨k g´1 compared to the control. Notable exceptions regarded Al, Fe, K, Na and S, which responded differently in the three analyzed varieties, indicating not only a dose effect but also a variety response. Also, wheat grown in soil exposed to nCeO 2 concentrations lower than 500 mg¨kg´1 showed no change in kernel nutrients except for S and Mn which had an inverse hormetic response [13].
Rico et al. [16] observed significant increases in both macro-and micro-nutrients in grains from barley plants exposed to nCeO 2 at different levels, which is in disagreement with our observations. Peralta-Videa et al. [12] reported no effects on S and K content in pods from soybean plants grown at nCeO 2 treatments similar to ours. Aluminum, Na and Ca concentrations in soybean pods were negatively affected, while Cu and P were increased at 1000 mg nCeO 2¨k g´1 treatment [12]. Different results were also obtained by Zhao et al. [14], who analyzed mineral content of cucumber fruit from plants grown at comparable nCeO 2 concentrations to ours. These authors did not observe significant differences in any element concentration between treated and control plants; the exceptions were Mo that decreased at both treatments and Mg that had a hormetic response [14]. Finally, under nCeO 2 treatments comparable to ours, significant increases of K and Mn were observed in undeveloped corn cobs but not in developed ones [15].
Contrary to nCeO 2 , scant results are available about the effects of plants grown in soil containing nTiO 2 . Only Servin et al. [57] reported data from a full mineral analysis of cucumber fruit. They added nTiO 2 to soil at concentrations from 250 to 750 mg nTiO 2¨k g´1 [57]. There was no substantial effect on macro-and micro-nutrients in fruits, apart from P and K, in which a hormetic effect was detected [57]. Instead, and in agreement with our results, Klingenfuss [55] reported no significant differences in P concentrations in wheat grains after exposure of plants in soil with a range of nTiO 2 varying from 1 to 1000 mg nTiO 2¨k g´1. Unfortunately, Klingenfuss did not report the concentrations of other nutrients [55]. However, these results can indicate a species-specific response and possibly diverse effects on mineral concentrations in eudicots and monocots, with the latter being more affected.
The high levels of Ca in both kernels and vegetative tissues such as roots and shoots [40] could be linked to the higher yield grain under nTiO 2 treatments. Calcium is a messenger between environmental factors and plant responses in terms of growth and development [6]. nTiO 2 (or the released ionic form) could potentially have a beneficial effect enhancing absorption and translocation of Ca [58]. Moreover, other authors showed that seeds immersed in nTiO 2 solution and sprayed with nTiO 2 over shoots could enhance plant growth through several mechanisms including the enhancement of photosynthesis, enzyme activity and even by a supposed new N 2 fixation mechanism in the air [32][33][34][35]59]. However, in our experiment we included nTiO 2 in the soil and the observed benefits could be explained by translocation of at least minimum amounts of nTiO 2 in shoots.
The Na/K and Na/Ca ratios are useful indicators of plant response to stress and can also indicate kernels quality. The unaffected Na/K ratio indicates no major metabolic disorders [60]. On the contrary the Na/Ca ratio was reduced by 2.4-fold on average in both nTiO 2 treatments, which suggests an increased competitive inhibition between absorption of Na and Ca that can possibly mitigate eventually harmful effects of Ti-similarly to what was observed for rice under nCeO 2 exposure [10]. Alternatively, as Na concentrations did not change significantly from the control whereas Ca concentrations increased significantly under nTiO 2 treatments, this could be due to a beneficial effect of nTiO 2 that can enhance Ca absorption and translocation as observed above. Curiously, this is not the case for nCeO 2, although other authors reported a reduction in the Na/Ca ratio for rice grains after exposure of plants to nCeO 2 in soil [10]. In kernels under nTiO 2 treatments, the reduction in Na/Ca ratio and increased concentration of Ca can be beneficial for human nutrition on the one hand, but can negatively affect the eating quality of kernels on the other.
A separate comment needs to be made about the contrasting data reported in the only barley full life cycle study performed [16]. Besides the fact that the authors cultivated another variety of barley in a different soil, the contrasting results obtained from our experiments and also supported by the low Ce quantity found in kernels from shoots other than the main one [40] can tentatively be explained by different nCeO 2 sizes (8 nm vs. 22.7 nm) and the methodology used to apply MeNPs to the soil. In fact, Rico et al. [16] sonicated a nCeO 2 solution prior to diluting it in soil and mixing, thus possibly making nCeO 2 more bioavailable. It is also possible that a larger quantity of ionic Ce, toxic to plants [28,30], was released in soil by nCeO 2 in the experiment by Rico et al. [16] which is supported by their larger BET surface area (93.8 vs. 46.1 m 2¨g´1 ).

Conclusions
Among the metal nanoparticle (MeNP) treatments, some significant effects (on kernel numbers, 100-kernel weight, and Zn and Mn concentrations) were visible only at lower concentrations and not at higher concentrations (hormetic response). Different nanoparticles can therefore positively or negatively influence different nutritional values. Although the observed differences between 500 and 1000 mg¨kg´1 MeNP treatments might be due to MeNP interactions (agglomeration and/or association) with soil constituents, this hypothesis remains to be tested at higher concentrations. Indeed, toxicity of MeNPs could be closely related to their chemical composition, structure, particle size and surface area. Substantial research is needed on MeNP size, their physico-chemical characteristics and their interactions with soil components.
Both MeNPs had a negative impact on the amylose content of kernels, with a reduction in grain yield. This was associated with an increase in crude protein (CP) and most amino acids. Interestingly, lysine, the essential and largely deficient amino acid in cereal grains, showed the largest percentage increase.
In terms of MeNPs impact on barley food production, our data are far from being exhaustive; however, a possible negative impact on kernel energy content, K and S concentrations might be counterbalanced by CP and lysine increase. Moreover, nTiO 2 treatments also increased Ca, Mg and Zn kernel concentrations.
Our findings demonstrate that cerium oxide nanoparticles (nCeO 2 ) and nTiO 2 act differently on the nutritional quality of barley kernels. Generally, barley kernels were found to be more negatively affected by nCeO 2 , while nTiO 2 can potentially have a beneficial effect. This study provides the first proof that nCeO 2 and nTiO 2 nanoparticles can have significant and contrasting impacts on the composition and nutritional value of barley kernels.

Conflicts of Interest:
The authors declare no conflict of interest.