**2. Materials and Methods**

#### *2.1. Plant Material and Olive Sampling*

In total, nine olive cultivars were selected for this work; they were grown in groves located in three different Italian regions: Sardinia, Abruzzo and Apulia.

The Sardinian cultivars "Sivigliana da olio", Semidana and "Corsicana da olio" have been chosen for their suitability to be processed as table olives (Sivigliana da olio), for their resistance to some common diseases and high productivity (Corsicana da olio) and for the increasing interest of growers to introduce it in all Sardinian orchards (Semidana) [45,46]. Specialized nonirrigated orchards were located in the province of Sassari (Sardinia, Italy); olives of Sivigliana and Semidana cultivars were hand harvested in Ittiri (40◦37 15.7 NL 8◦32 20.7 EL), while samples of Corsicana were picked in Usini (40◦39 39.0 NL 8◦31 32.7 EL).

With regard to Abruzzo (Central Italy), the varieties Dritta, Tortiglione and Gentile dell'Aquila were selected as the most representative of the provinces of Pescara, Teramo and L'Aquila, respectively, also in consideration of their specific adaptation to the local pedoclimatic conditions correlated with peculiar agronomic and biochemical traits [47,48]. Dritta and Tortiglione were grown in orchards located in Notaresco, in the province of Teramo (42◦38 54.6 NL, 13◦52 57.4 EL), while Gentile dell'Aquila was grown in orchards from Vittorito, in the province of L'Aquila (42◦08 16.6 NL, 13◦48 37.3 EL).

In the Apulia region (Southern Italy), in addition to some well-known and notable cultivars for olive oil extraction, a great number of other varieties and/or landraces are present [3]. Among these, three have been considered in this study, specifically the varieties Bambina, Oliva Rossa and Cima di Melfi, for which few reports are currently available in the literature [49,50]. The Apulian genotype Bambina was cultivated in the orchard located in Gravina di Puglia (40◦49 0 NL, 16◦25 0 EL), while genotypes Cima di Melfi and Oliva Rossa were grown in the orchard of Putignano (40◦51 0 NL, 17◦7 0 EL), both in the province of Bari.

For the molecular characterization, young leaves of each cultivar were collected and stored at −20 ◦C before use. For gene expression analysis, drupes were harvested at different developmental stages of fruit, generally from mid-October to the beginning of November: T1, yellow-green olives; T2, turning olives; and T3, almost dark olives (Table 1). In order to minimize the effects of fruit asynchronous maturation within the same tree, drupes were harvested from the external parts of the canopy of trees. After harvesting, olives were immediately frozen in liquid nitrogen and stored at −80 ◦C until further processing. For two of the considered time-points (T2 and T3), one part of the collected drupes was immediately used for virgin olive oil extraction and biochemical characterization. An additional timepoint at 100% ripening (T4, fully dark olives, from the end of November to the beginning of December) was considered exclusively for oil production and biochemical characterization (Table 1). For each cultivar/genotype, three biological replicates were taken. Sampling was carried out for two consecutive years.



#### *2.2. Genetic Characterization*

Leaf samples were collected from three plants per cultivar, lyophilized and finely pulverized before use. Total genomic DNA was extracted from 200 mg of dry tissue according to [51], checked for quantity and quality and normalized to 50 ng μL−1. Olive genotypes were molecularly characterized by means of 12 nuclear SSRs (Table S1) [52–54], chosen on the basis of their suitability and reliability proven in several studies about olive

variety identification [55–57]. Primer pairs were synthesized by Thermo Fisher Scientific (Waltham, MA, USA) and all forward primers were labeled with one of the following dyes: 6FAM™, NED™, VIC® and PET™. The amplification reactions were carried out in a final volume of 12.5 μL, using a T100 thermal cycler (Bio-Rad Laboratories, Segrate, MI, Italy) according to [9]. Two microliters of each PCR product was added to 0.5 μL of GeneScan™ 600 LIZ® Size Standard (Applied Biosystem, Foster City, CA, USA) and 9.5 μL of Hi-Di Formamide (Applied Biosystem, Foster City, CA, USA), and successively separated by capillary electrophoresis using an automatic sequencer ABI PRISM 3100 Avant Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Detection, sizing and data collection were carried out by means of the GeneMapper® genotyping software v.5.0 (Applied Biosystems, Foster City, CA, USA) as in [58]. To estimate the genetic distances among the considered genotypes, cluster analysis based on the Unweighted Neighbor Joining method was performed using DARWIN software v. 6.0.010 (http://darwin.cirad.fr, accessed on 26 April 2019), with 1000 bootstrap values for tree construction.

#### *2.3. Oil Extraction and Characterization*

Around 10 kg of drupes were collected from three trees per each sampling point and variety. Fruits were then divided into three aliquots representing the biological replicates. Oil extraction was performed within 5 h after harvesting at laboratory scale. The extraction system was made up of a semi-industrial scale hammer crusher (RETSCH GmbH 5657, Haan, Germany) working at 2850 rpm and a basket centrifuge with a bowl of 19 cm working at 2700 rpm (Marelli Motori S.p.A., Arzignano, VI, Italy) [59]. Briefly, about 1 kg of olives per sample and replicate was crushed and then the olive paste was transferred into the basket centrifuge for the oil recovery. Once extracted, the virgin olive oils (*n* = 3) were stored at 20 ◦C in 100 mL dark glass bottles until the analyses.

Two phenolic alcohols (i.e., hydroxyl-tyrosol and tyrosol) and the oleacein (3,4- DHPEA-EDA) were chosen as target phenolic compounds for HPLC analyses, being amongst the most abundant ones in olive oils and because of their powerful antioxidant activity and impact on oil sensorial feature [60]. These phenolic compounds were extracted by liquid–liquid extraction using a mixture of methanol/water (70/30 *v*/*v*) according to previous papers [61]. The extraction procedure was similar for both the Folin–Ciocalteu assay and for HPLC analysis, with the only difference being that in the latter case 250 μL of a 100 mg kg−<sup>1</sup> solution of gallic acid as internal standard for quantification was added. Total phenolic compounds (TPC) were quantified spectrophotometrically [61] by means of a calibration curve of pure gallic acid and the results are expressed as gallic acid equivalent (GAE, mg kg−1). HPLC-DAD analysis was carried out as previously reported [61] using a UHPLC binary system (Dionex Ultimate 3000 RSLC, Waltham, MA, USA). The identification of hydroxy-tyrosol, tyrosol and 3,4-DHPEA-EDA was performed by comparing the peak retention times with those obtained by the injection of pure standards and/or with data in the literature [62]. The results are expressed as gallic acid equivalent (GAE, mg kg<sup>−</sup>1).

#### *2.4. Polyphenolic Compound Gene Expression*

The mesocarps of frozen olives were mechanically crushed by the use of a tissue lyser and total RNA was extracted from 100 mg of the obtained powder according to the manufacturer's instructions of Spectrum Plant Total RNA Kit (Sigma-Aldrich, St. Louis, MO, USA). An additional step for on column genomic DNA digestion was added. RNA quantity and quality were checked by spectrophotometric measurement using a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and by electrophoresis on 1.2% Certified Molecular Biology Agarose gel (Bio-Rad Laboratories, Segrate, MI, Italy) in 1X TBE buffer (1 M Trizma base, 1 M Boric Acid, 20 mM EDTA, pH 8.3).

Quantitative real-time polymerase chain reactions (qRT-PCRs) were carried out on three genes involved in the polyphenolic biosynthesis, named TYRD for tyrosine/dopa decarboxylase, CuAO for copper amine oxidase, and ALDH for alcohol dehydrogenase. The used primer pairs are listed in Supplementary Material (Table S2). The reverse transcription of 700 μg RNA samples was performed with the SuperScript™ VILO™ cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. qRT-PCR reactions were carried out using the SsoAdvanced Universal SYBER® Green Supermix (Bio-Rad Laboratories, Segrate, MI, Italy) and the CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories, Segrate, MI, Italy). Thermal cycling parameters were: initial denaturation at 95 ◦C for 3 min, followed by 40 cycles of 95 ◦C for 10 s and 60 ◦C for 30 s. The specificity of the amplification product per each primer pair was confirmed by evaluating the melting curve through an increase of 0.2 ◦C every 5 s from 65 to 95 ◦C. For qRT-PCR assay, each amplification reaction was run in triplicates of three biological replicates. The elongation factor 1α (EF1α, AM946404) was selected as a reference gene for normalization and the comparative Ct method (2−ΔΔCt method) was used to analyze the expression levels of the selected genes [63].

#### *2.5. Statistical Analysis*

All values of chemical and genetic analysis are means (*n* = 3) ± standard deviation (sd). The statistical analysis was performed, for samples of each region, by one-way analysis of variance (ANOVA) with sampling time as the group factor and post-hoc Fisher's LSD test, using the software Statistica 10.0 for Windows. Differences were considered to be significant when *p* < 0.05.
