Determination of free soluble phenolic compounds in grains of ancient wheat varieties ( Triticum sp.pl.) by liquid chromatography-tandem mass spectrometry

19 A liquid chromatography coupled-tandem mass spectrometry method was developed and validated 20 for the determination of free soluble phenolic compounds in eight ancient varieties of wheat 21 ( Triticum sp. pl.), notably Autonomia, Gentil rosso, Inallettabile, Leone aristato, Mentana, Poulard 22 di Ciano, Risciola and Terminillo. Trace compounds such as two conjugated flavones, vitexin 23 (17.13-34.32 µg/kg) and isovitexin (9.76-30.01 µg/kg), were also determined. 24 Poulard di Ciano, presumably an autochthonous wheat of the Reggio Emilia province (Northern 25 Italy), showed a peculiar quali-quantitative phenolic profile (7097.03 µg/kg total phenolic sum and 26 1.97 hydroxycinnamic acids sum to hydroxybenzoic acids sum ratio) along with a tetraploid 27 genome. Terminillo, Risciola, Gentil rosso, Mentana, and Leone aristato showed a hexaploid 28 genome and high concentrations of phenolic compounds (ranged from 6796.12 to 7605.78 µg/kg), 29 also in comparison with two modern varieties of bread wheat, Bolero and Blasco. 30 The targeted metabolomic approach proved to be effective to determine some secondary 31 metabolites of wheat. The richness in phenolic compounds combined with high rusticity and


Introduction
The human selection of cereal seeds throughout history was aimed at obtaining more productive and resistant plants that better responded to the environmental biotic and abiotic stresses (pests, weeds, drought, and other adverse climatic factors).The ancient varieties and landraces of wheats were characterized by the good adaptation to the local environment and climate of cultivation. 1 Moreover, these crops responded to the nutritional needs of the local populations.Some cultivars originated from local landraces due to the work of Italian geneticists, particularly F. Todaro and N.
Strampelli, during the first decades of the twentieth century. 2 After the Second World War, local varieties were replaced by "modern" cultivars and high energy input agriculture. 2,3These cultivars were mainly selected for cultivation and/or production related traits and characterized by a relative genetic uniformity. 1 The modern cultivars improved the yields and better complied to the requirements of the baking industry.On the other hand, this process inevitably led to the dramatic loss of agro-biodiversity observed in recent years.A part of the genetic heritage of the ancient cultivars came to us owing to the tenacity of "custodian farmers", who conserved and made these plant genetic resources available by cultivation over time, along with the higher rusticity and adaptability to marginal. 4,5ecent years, a new interest on ancient cereals has been expressed by consumers, manufacturers, farmers, and researchers.The main reasons are related to sensory and nutritional properties of the derived products, 6 as well as to the good adaptation and rusticity of these plants, which can make their cultivation very attractive in the light of the threat of climate change.In addition, the demand of ancient cereals was fostered by the constant increase of "green-minded" consumers, who are keen on recovering cultural values, traditional, safer, traceable, transparent, whole-wheat, and organic foods, and agriculture of low environmental impact. 7t and its derivatives (bread and other baked goods) are considered functional foods, based on the correlation between the consumption of products made from whole-wheat and the reduction of chronic diseases (cardiovascular diseases, diabetes, and cancer). 2,8These beneficial effects to health were put in relation to the presence of substances with antioxidant action and the fibrous component. 9,10ancient grains have a rich heritage of secondary metabolites, so-called "functional" or "bioactive" compounds, which are most lacking in modern cultivars, especially in terms of quality. 11These substances, with antioxidant properties, mainly include phenolic compounds (phenolic acids: hydroxycinnamic acids and hydroxybenzoic acids; flavonoids: flavonols, flavones, flavone-C-glycosides, flavone-O-glycosides, isoflavones, flavonones, anthocyanidins; other classes: coumarins, stilbenes, proanthocyanidins, lignans), aside from tocopherols and tocotrienols, and carotenoids (typical of the "red" varieties). 12In addition to performing an antiradical activity, freeradical scavenging, and metal chelating, these compounds are involved in many other metabolic functions and exert important effects on the immune and cardiovascular systems. 13,14olic identification and determination via liquid chromatographic techniques in common wheat (Triticum aestivum L.) have been reported in recent years.The identification was performed using LC-ESI-TOF-MS 8,12 and LC-ESI-MS/MS, 15 while the quantification was carried out using HPLC-UV-Vis, 16 HPLC-DAD, 10,[17][18][19][20] and HPLC-UV-MS. 21Other studies focused on quantification of phenolics in durum wheat (T.durum Desf) and other Triticum species using HPLC-DAD. 22,23ough detectors based on UV absorption are still commonly used for the determination of phenolic compounds, they lack in selectivity and sensitivity. 24In particular, for trace components in complex matrices, the impact of impurities on the noise is high, and it results in scarce precision and reliability.
In liquid chromatography, MS detectors equipped with a triple quadrupole analyzer are set to measure only selected reaction monitoring (SRM) transitions for each compound, thus highly improving sensitivity, selectivity, and precision in phenolic identification and quantification.In addition, the chromatographic run can be segmented into different time windows in order to search for each compound solely within its specific window and, in turn, to optimize each individual dwell time. 25 targeted metabolomics approach represents a modern and attractive strategy for crop and food characterization.Indeed, it allows the quantification of a predefined set of metabolites, typically a relatively small number of chemically characterized metabolites.Furthermore, compounds are kept in their native form, thus avoiding any chemical modification of their structure.
The present study aims to identify and quantify free soluble phenolic compounds in ancient varieties of wheat using a sensitive and reliable targeted metabolomics method based on LC-ESI-MS/MS.In this context, quali-quantitative analyses are carried out in eight ancient and two modern varieties, cultivated in the same conditions.
The sowing was carried out in autumn, for the vernalization requirement of some no-alternative varieties.The harvest was carried out manually over the course of two days to ensure a good level of maturity of all the cultivated varieties.Spikes were immediately bagged in suitably-labeled paper envelopes and stored at room temperature for a few days in a local laboratory suitable to shelter them from light and moisture.

Plant growth conditions and DNA extraction
The seeds of all the varieties were germinated in Petri dishes at 25 °C.Once the coleoptiles emerged (growth stage 0.7, according to Zadoks et al. 27 ) seedlings were transferred to polystyrene box cells filled with peat (70% organic matter, 0.6% nitrogen; pH = 6.0 and electric conductivity = 1.33 dS/m).Seedlings were grown for seven days in a growth chamber (Binder KBW 720, Tuttlingen, Germany) under a 16-h photoperiod with an irradiance of 180 µmol/m 2 /s (white fluorescent tubes Fluora 18W/77, Osram, Munich, Germany), relative humidity of 60% and day/night temperature of 25/15 °C.Young leaves (300 mg) of each cultivar were used for DNA isolation according to Stein et al. 28 Concentration and purity of extracted DNAs were measured through a Nanodrop 1000 spectrophotometer (Thermo Scientific, Wilmington, DE, U.S.A.).

Determination of ploidy level -genotyping
Diploid species belonging to Triticum genus have an A genome only, tetraploid species have A and B genomes, while the hexaploid species Triticum aestivum L. harbors three genomes: A, B, and D. 29 A, B, D genomes specific markers were thus used in the present study to discover the ploidy level of the tested genotypes.While amplification products were expected for all markers for Triticum aestivum samples, only A and B specific, and only A specific markers were expected for tetraploid and diploid genotypes, respectively.Following this strategy, six KASP SNP markers (two for each A, B, and D genomes) were selected from CerealsDB 3.0 (http://www.cerealsdb.uk.net) 30 (Table 1).
SNPs detection was performed using the Fluidigm EP1™ system (Fluidigm Corp., San Francisco, CA, U.S.A.).Contig sequences flanking the six SNPs (at least 60 bp before and after the polymorphism) were used by Fluidigm to design and synthetize the primers for the assay.Allelespecific forward primers (ASP1, ASP2) were labelled with the fluorochromes FAM and HEX.
Amplifications and data analysis were carried out according to Fluidigm EP1™ system protocols.

Milling
A gross sample of 10 g of intact and healthy caryopses of each variety was selected and considered as a single batch of raw material.Caryopses were ground using a small electric grinder (MC300, Moulinex), taking small amounts each time to avoid the heating of the sample.Finally, the flour was made finer and more uniform with mortar and pestle.

Optimization of the LC-ESI-MS/MS conditions
Standard stock solutions (50-100 mg/L) of each analyte were prepared using an aqueous solution (hereafter called EAT solution) containing 5% absolute ethanol, 3% acetonitrile, and 0.1% triethylamine, as a solvent.
The MS/MS parameters (parent ion, product ions, fragmentor, collision energy, and polarity of ion source) were optimized using standard solutions of each analyte (125 µg/L), introduced individually into the detector.For each analyte, selected reaction monitoring (SRM) transitions of the two most abundant fragments (product ions), were used.The experiment was carried out in a positive-ion mode and fell in the 50-500 m/z range.Source parameters were optimized as follows: gas (N 2 ) temperature, 350 °C; N 2 flow, 11 L/min; nebulizer pressure, 35 psi; capillary, positive 3500 V and negative 4000 V.
Peak identification included the comparison of the peak retention times to those obtained with chemical standards and the evaluation of the tandem mass spectroscopy experiments.For each compound, the most intense SRM transition was used as the quantifier ion (QI) and the least intense one as the qualifier ion (qi) to confirm the peak identification.The relative intensity of each peak (qualifier ion SRM to quantifier ion SRM percentage ratio) was also evaluated and verified to be within an uncertainty range of ±10%.

LC-ESI-MS/MS optimized analytical conditions
Secondary metabolites determination was carried out by liquid chromatography coupled with mass spectrometry using an Agilent Technologies RP-LC-ESI-MS/MS system (Santa Clara, CA, U.S.A.), consisting of an HPLC (Agilent Technologies 1200 Series) equipped with a degasser, a binary pump, an autosampler, and a 6410B triple quadrupole mass spectrometer.
The solvent system was composed by formic acid 0.1% in water (solvent A) and acetonitrile

Method of extraction, optimization and recovery tests (trueness)
The extraction protocol relied on the method described by Adom et al., 31 then modified by Dinelli and et al., 12 with some adjustments to optimize the yields of extraction.
Each sample (2.00 g of finely-ground flour) was introduced into a 50-mL plastic tube together with 10 mL of chilled ethyl alcohol (80%).After homogenization through the use of a vortex (150 sec), the tubes were centrifuged (4237R, A.L.C.International Ltd, Cologno Monzese, MI, Italy) at 4000 rpm for 5 min at 4 °C.The supernatant was transferred in another tube and stored at -20 °C.Ten mL of chilled ethyl alcohol (80%) were added to the pellet and the entire extraction cycle was repeated three times in total.Finally, all three aliquots were pooled to obtain about 30 mL of extract that was poured into a 100-mL round bottom flask and evaporated to dryness with a rotary evaporator (VV 2000, Heidolph Instruments, Kelheim, Germany), using a temperature of the water bath below 30 °C.The residue was re-dissolved with EAT solution and made up to 10 mL in a graduated flask with the same solution and, finally, transferred into a test tube.After insufflation of N 2 to prevent oxidation, each tube was closed by a screw cap with Teflon septum and stored at -20 °C until LC-MS/MS determination.The dilution factor of the method was 5.For each real sample, replications (n = 3) were carried out using different batches of raw materials.
To establish the optimal number of cycles of extraction, a recovery test was carried out using a homogenous gross sample of Mentana flour (20 g).After each extraction cycle (five cycles in total), carried out on 2.00 g of flour, the extract obtained was checked through the acquisition of the UV-Vis spectrum (using the V-730 UV/VIS spectrophotometer, Jasco Europe Ltd, Cremella, LC, Italy), from 235 to 800 nm.The same samples (n = 5) were also run using the LC-MS/MS method described above to calculate the individual analyte yield.The test was performed in triplicate.
The method trueness was also evaluated.Samples of Mentana finely-ground flour (2.00 g) were spiked with a standard mixture of all the analytes.Two fortification levels (three repetitions for each) were tested and the results were compared to those obtained with not-spiked samples.The mixture of pure standards was allowed to stand for 30 min to enhance the interaction with the matrix before the extraction.After each chromatographic run, the column was washed by injecting 5 µL of EAT solution.These samples were subjected to extraction and LC-MS/MS determination as described above.

Evaluation of linearity, limit of detection and lower limit of quantification (detectability), intra-day and inter-day repeatability (precision).
Solutions of increasing concentration of diluted pure standards were prepared to evaluate the linearity of the instrumental response.For each substance, the range of concentrations was established on the basis of the data present in the literature and by exploring the chromatographic traces of some real samples.
In the aqueous reference solution, the instrumental limit of detection (LOD) and the lower limit of quantification (LLOQ) were obtained by applying the equation: 32 LOD or LLOQ = (K sy/x)/b [Eq.1] where sy/x and b are the estimated regression standard deviation and the slope of the relevant analytical calibration function, respectively.K = 3 and K = 10 were chosen in order to obtain the LOD and LLOQ, respectively. 33 Precision was evaluated with an intra-day repeatability test on standard mixture solutions containing all the analytes (125 µg/L), injected five times, and an inter-day repeatability test carried out on samples of Mentana finely-ground flour (2.00 g) injected on five consecutive days in the same conditions.The relative standard deviations were calculated for each substance.

Total phenolic content
The total phenolic content (TPC) was analyzed using the Folin-Ciocâlteu method. 34In a 10-mL flask, 6 mL of water, 0.3 mL of the extract obtained through the method described above, and 0.5 mL of the Folin-Ciocâlteu reagent were added.After 1 min of shaking with a vortex, 2 mL of 15% aqueous sodium carbonate were added and the solution was made up to 10 mL with water.
Finally, this solution was mixed and left to stand at ambient temperature for 120 min.Absorbance was read at 750 nm (V-730 UV/Vis spectrophotometer, Jasco Europe Ltd, Cremella, LC, Italy) against a blank represented by the reagents only and compared with a standard gallic acid calibration curve.Results of triplicate analyses on different sample batches are given as mg/g of gallic acid equivalents (GAE).

Statistical analysis
Univariate and multivariate analyses were carried out on the data set.Differences amongst varieties were assessed by analysis of variance (one-way ANOVA) based on three replicates for each sample.When a significant effect (at least p ≤ 0.05) was showed, comparative analyses were carried out using the post hoc Tukey's multiple comparison test.Correlation matrix analysis and principal component analysis (PCA) were also applied to the whole data set.All tests were performed using Statistica version 8.0 software (Stat 180 Soft Inc., Tulsa, OK, USA).

Determination of ploidy level -genotyping
For nine of tested genotypes, amplification products were obtained for markers designed for A, B, and D genomes and thus proved their hexaploid nature (Table 1).Only one genotype -"Poulard di Ciano" -for which no amplification product for markers situated on D genomes was obtained, was identified as tetraploid species.

Determination of the optimal parameters for MS/MS detection
The optimal parameters (fragmentor voltage, V, and collision energy, V) for MS/MS detection, determined through the direct infusion in the spectrometer of a 125 µg/L solution of each pure standard compound, are reported in Table 2.For most of the phenolic acids (gallic, caffeic, p-coumaric, trans-ferulic, sinapic, and salicylic), the QI corresponded to a loss of water by the parent ion.The dwell-times for each transition were also optimized.High sensitivity was obtained in ESI positive mode for all compounds and the same type of parent ion (protonated molecular ion [M+H] + ) of high sensitivity was obtained.

Setup of chromatographic conditions and segmentation of the chromatogram
A satisfactory separation of the analytes was obtained with a double endcapped phenyl-hexyl column.This column was designed for the separation of substances with aromatic rings in their structure, through selective π-π interactions with the aromatic rings of the stationary phase.
The retention times of the analytes are showed in Table 2.The chromatographic method was optimized by means of successive tests, carried out by changing the composition of the mobile phase in order to obtain the best profile for peak separation and to reduce the total run time.The separation of the two critical couples of peaks (isovanillin-vanillin and isovitexin-vitexin) was considered.Moreover, inside each couple, the compounds showed the same transitions.The resolution (R) of both the couples of peaks was higher than 1.5, even if the performance of the column decreased during the time, in particular for the couple isovanillin-vanillin.However, isovanillin was never detected in any real sample.In addition, the chromatogram was divided into six time windows to optimize the dwell times of each compound (Table 2).

Optimization of the extraction protocol and trueness evaluation
The dissolution of the samples after the evaporation to dryness was a critical step of the method.
The chemical nature of the residue needed an alkaline solution to dissolve the polyphenols that contain weak-acids hydroxyls (pK a ≈ 10) as substituents, completely.This problem was overcome using triethylamine (0.1%) as an MS-compatible organic base. 35Moreover, the presence of a strong organic base, such as trimethylamine, was able to suppress the formation of sodium and potassium adducts, thus resulting in an improved MS sensitivity. 36epeated extraction cycles were performed on Mentana finely-ground flour to verify the reduction of the absorption at different wavelengths (280, 290, 326, 446, 474 nm), corresponding to the maximum absorptions in the UV-Vis spectrum.Each sample was also analyzed with the optimized LC-MS/MS method.
The sum of the first three cycles of extraction accounted for more than 90% of the absorption for all the monitored wavelengths (Table 3).In addition, the recovery of the single compounds carried out with LC-MS/MS after three cycles of extraction was considered satisfactory for most of the analytes (Table 3).Vanillin was the only compound that showed a recovery lower than 90% in these conditions.The reason is likely related to the chemical nature of this substance.In fact, it is not a phenolic acid, but it contains a less polar aldehyde function (pK a ≈ 17).However, it was decided to limit the number of cycles of extraction to three in order to reduce the total time for sample preparation, also.
The determination of the recoveries obtained by spiking a known amount of pure compounds into the sample (net of the concentrations found for the same not spiked sample) does not allow the estimation of the extraction yield of the analytes.In fact, the compartmentalization of the compounds inside cell structures needs a number of extraction cycles up to allow the analytes to be quantitatively extracted and dissolved into the solvent.As a matter of fact, the present method does not include the use of tools, such as SPE cartridges and filters that can give a significant reduction of the analytes yield.Nevertheless, a recovery test with two different fortification levels of standard concentration was carried out on the same sample (Mentana), also used for the recovery tests previously described.

Method validation (linearity, detectability, and precision determination)
The ranges of concentration used for calibration are showed in Table 4.For all the substances, the coefficient of determination (R 2 ) was higher than 0.995, except formononetin (0.991), thus showing a satisfactory linear correlation between concentration and response in the range studied. 32 and LLOQs are showed in Table 4.The results of the detectability evaluation provided LOD values in the standard solutions that were lower than 1 µg/kg for most of the analytes, mainly the trace compounds, and were lower than 10 µg/kg for all the main phenolic acids.However, considering the dilution factor of the method, the detectability showed values ranging from 0.62 and 31.8 µg/kg.
A review of the literature showed that the detectability of the proposed method is very satisfactory when compared with similar studies on the same topic.The ranges of LOD reported are: from 100 and 500 μg/kg; 17 from 80 and 400 μg/kg; 20 from 60 and 700 μg/L; 22 from 220 and 5360, μg/kg. 23intra-day precision was performed on a standard mix extracted and injected five times in the same day.The results were satisfactory (Table 4) and all relative standard deviations were contained within 10%, except daidzein (12.5%).The inter-day precision was evaluated for each analyte on a sample of Mentana.RSDs of all the analytes exceeding the LLOQ were below ±10% (Table 4).For a confidence level of 95% and coverage factor (k) = 2, the method has an uncertainty that is lower than ±20%. 32

Phenolic determination on real samples
The quantification of the phenolic substances in the real samples is reported in Table 5. ANOVA and Tukey's test were applied on the whole data set.Isovanillin was never detected, while (+)catechin, caffeic acid, and orientin did not show values above the detectability limits in any sample.
Formononetin was present with a concentration higher than LLOQ solely in Autonomia.All the other substances showed significantly different concentrations (p ≤ 0.001) among the varieties.
A concentration of gallic acid above 300 µg/kg was highlighted only in Mentana.Vanillic acid was present in significantly higher concentrations (around 2000 µg/kg) in Gentil rosso, Terminillo, Mentana, and Leone aristato, while syringic acid showed the significantly highest concentrations in Mentana, Risciola and Leone Aristato.
Risciola showed a high content of vanillin, followed by Inallettabile, even though the significant highest concentration was found in the modern cultivar Bolero.Terminillo showed the statistically highest concentration of trans-ferulic acid, followed by Risciola, Mentana, Gentil Rosso, Poulard di Ciano, and Leone Aristato, while the other varieties remained under 2000 µg/kg.Poulard di Ciano showed an extremely higher concentration of sinapic acid (above 1300 µg/kg) in comparison with all the other varieties.
Salicylic acid, isovitexin, vitexin, and daidzein were not found in concentrations above the LLOQ in some samples.In general, the concentrations of vitexin were higher than isovitexin, while daidzein was present in very tiny concentrations.Poulard di Ciano had the highest concentrations of both isovitexin and vitexin, closely followed by Gentil rosso.Terminillo, Leone aristato, Risciola and Gentil rosso showed the highest concentrations of daidzein, as well as of trans-cinnamic acid.
Mentana and Terminillo showed the significant highest hydroxybenzoic acids sum, followed by Leone aristato, Risciola and Gentil rosso.Poulard di Ciano showed the significant highest values for hydroxycinnamic acids sum along with Terminillo.Risciola and Gentil rosso, followed by Terminillo and Poulard di Ciano, highlighted the highest concentrations for flavones and isoflavones sum.
Poulard di Ciano showed a ratio between hydroxycinnamic acids sum and hydroxybenzoic acids sum (HC/HB) as high as 1.97, while all the other varieties had values that ranged from 1.07 to 1.32, except Blasco, which was the only variety with an amount hydroxybenzoic acids higher than hydroxycinnamic ones.
In recent studies aimed at differentiating certain modern and ancient cultivars of common wheat, Gentil rosso showed the highest total amount of phenolic substances, a significantly higher concentration of trans-ferulic acid than Inallettabile and higher concentrations of sinapic, syringic, and vanillic acids in comparison with Inallettabile and Bolero. 37Autonomia and Bolero varieties showed a similar concentration of trans-ferulic acid in a study carried out on wheat flour. 4ver, the comparison of data obtained in the present study with those presented in other papers is quite complex, due to the different extractive-quantitative approach herein applied.

Total phenolic content
The TPC of every sample is showed in Table 6.The one-way ANOVA showed statisticallysignificant differences (p ≤ 0.001) among the varieties.The highest figure was found in Poulard di Ciano, followed by Risciola and then Leone aristato, Terminillo and Mentana, while the lowest values were found in Inallettabile, Autonomia, as well as in the two modern varieties, Bolero and Blasco.
Apart from Bolero, data on TPC were found in literature solely for Autonomia, Inallettabile and Gentil rosso. 12,38,39Data on Gentil rosso, Bolero and Blasco were also available for wheat bread. 40greement with the results of the present work, the TPC relative amount found in Gentil rosso was higher than in Autonomia and Inallettabile.12,38,39 The TPC values obtained in this study for the modern cultivars were relatively lower, in accordance with those reported by several authors.12,38,39,40

Correlation matrix analysis
The relationships between the main parameters were studied by the linear correlation matrix (Table 7).Concentrations lower than the LLOQs of some parameters did not permit their inclusion in this study.TPC was mainly correlated (p ≤ 0.001) with the individual hydroxycinnamic acids (sinapic, p-coumaric, trans-ferulic, and trans-cinnamic) and their sum.It also showed a linear correlation (p ≤ 0.05) with gallic acid, hydroxybenzoic acids sum, and flavones and isoflavones sum.The hydroxybenzoic acids (gallic acid, vanillic acid, and syringic acid) showed a significant linear correlation amongst them, as well as with their sum, but also with some hydroxycinnamic acids and their sum.Vanillin did not show any significant linear correlation, neither with trans-ferulic acid (its precursor), nor with vanillic acid (its derivative). 41p-Coumaric acid was significantly correlated only with hydroxycinnamic acids, as well as with flavones and isoflavones sum and total sum.
Conversely, trans-ferulic acid, sinapic acid, and trans-cinnamic acid, aside from the significant correlations with hydrocinnamic-like compounds, also showed a correlation with hydroxybenzoiclike compounds, as well as the flavones and isoflavones sum.Apigenin showed a significant correlation only with flavones and isoflavones sum (p ≤ 0.001).The flavones and isoflavones sum was significantly correlated with hydrocinnamic-like compounds, due to their biochemical origin from 4-hydroxycinnamoyl-CoA. 41

Principal component analysis
A principal component analysis (PCA) was carried out on the autoscaled values to explore part of data set (only the parameters that did not show concentrations lower than the LLOQs) and to achieve information about the relationship amongst the variables and the overall distribution of the samples on the score plot.
The first four principal components (PCs), all with eigenvalues > 1.0, explained 88.57% of the total variance.All factors with eigenvalues < 1.0 were discarded, according to the Kaiser's criterion. 42the considered parameters (except vanillin) weighed on PC1 (52.94% of the total variance) with a negative sign (Fig. 1A).The hydroxybenzoic-like compounds were grouped together and weighed with a positive sign on the PC2 (18.44% of the total variance), while the hydroxycinnamic-like compounds were not grouped in a similar way.Total phenolic content weighed on PC2 with a negative sign and showed a linear correlation with sinapic acid.Apigenin weighed on PC3 (10.08% of the total variance) with a positive sign, as did flavones and isoflavones sum and vanillin (Fig. 1B).Vanillin weighed also on PC4 (7.12% of the total variance) with a negative sign.
Samples of Poulard di Ciano were clearly separated on the PC2 in the lower part of the plot (Fig. 2A), due to the high concentration of sinapic acid, p-coumaric acid, total phenolic content, and HC/HB.All the other samples were located on the PC1 according to the different amounts of the total sum, as well as the concentrations of trans-ferulic acid and trans-cinnamic acid, and hydroxybenzoic-like compounds.Inallettabile and Autonomia were located in the positive quadrant of the PC1, while Risciola and Terminillo, Gentil rosso and Leone aristato, due to their higher concentrations, were set in the negative quadrant of the PC1.Mentana samples were rather separated on the PC2, due to the high concentration of hydroxybenzoic-like compounds.The two modern varieties were located in the positive quadrant of PC1 and PC2 and were mainly characterized by the high concentration of vanillin (Bolero) and the low concentration of HC (Blasco).
Inallettabile was discriminated on PC3 for its high concentrations of vanillin and apigenin (Fig. 2B), while the PC4 permitted the separation of Bolero, due to its high concentration of vanillin.
In conclusion, this study permitted the determination of free soluble phenolic compounds in ancient varieties of wheat using a sensitive and reliable method based on LC-ESI-MS/MS.For the first time, applying this new metabolomic quantitative approach, polyphenolic compounds, such as vitexin and isovitexin, were determined in wheat.
Interesting phenolic profiles, in terms of amount and peculiarity, emerged.Poulard di Ciano, probably an autochthonous wheat of the Reggio Emilia province, showed a tetraploid genome.It furthermore displayed a characteristic high hydroxycinnamic acids/hydroxybenzoic acids ratio, which could be proposed as its peculiar biomarker.Its interesting profile certainly deserves a more thorough investigation.
Terminillo, Risciola, Gentil rosso, Mentana and Leone aristato showed high concentrations of phenolic compounds, with a balanced amount of hydroxycinnamic acids and hydroxybenzoic acids.
Although a more extensive comparison is needed to formulate wider conclusions, these varieties highlighted higher concentrations of secondary metabolites in comparison with the two modern varieties Bolero and Blasco.
The results of this study pick out the importance to work in the direction of recovery and reuse of ancient varieties, not only to safeguard the agro-biodiversity, but also for their advantageous characteristics.In particular, the great rusticity and adaptability to marginal soils that may therefore ensure stability of both yield and quality, the suitability for organic cultivation, along with their richness in secondary metabolites.
It is thus important to start up new genetic improvement programs aimed at obtaining a new wheat ideotype, harboring a right combination of technological and nutritional characteristics.This is particularly important for wheat, which is at the base of the diet of a large part of the world population, both in view of a further notable demographic growth and in light of the much-feared climatic changes, capable of affecting the agricultural production.Finally, the reintroduction of ancient grains could contribute to the diffusion of a more sustainable agricultural model, able to provide farmers with a greater added value to their production.

Table 1
Six Single nucleotide polymorphisms (SNP) markers used for genotyping assays and their properties.Experiment results are reported for each genotype.

Table 2 683
List of the retention times (t R ), time windows, molecular weights (MW), parent and product ions obtained after MS parameters optimization and a QI: quantifier ion.b qi: qualifier ion.FV c : Fragmentor voltage.CE d : Collision energy.qi/QI e : qualifier ion SRM to quantifier ion SRM percentage ratio.

Table 3 692
Recovery tests.Sum of the first three cycles of extraction (%) carried out through spectrophotometric reads at different wavelengths and through Comparison of recoveries (%) of each analyte obtained from samples spiked with two fortification levels of pure standards.All the tests were repeated three times using samples of Mentana (2.00 g) in the same conditions.

Table 4
Main parameters of method validation for standard compounds.Evaluation of linearity (R 2 ) for each range of concentration, straight line equations,

Table 5
Concentrations of the phenolic substances (µg/kg FW) in the real samples (modern varieties are reported in italics).

Table 6
Total polyphenols content determined by Folin-Ciocâlteu assay (three replicates) and expressed as mg of gallic acid equivalents (GAE)/g FW.Modern varieties are reported in italics.Results of the one-way ANOVA and the Tukey's test are reported as F values and lower case letters, respectively.Different letters identify samples significantly different (p ≤ 0.05). a

Table 7
Significant correlation matrix of the main parameters data set a .TPC c GA d VA e SA f V g pCA h tFA i SinA j tCA k A l HB m HC n FI b a p ≤ 0.05; bold p ≤ 0.01; bold and underlined p ≤ 0.001.b TPC: total polyphenols content.c GA: gallic acid.d VA: vanillic acid.e SA: syringic acid.