Modification of 13-hydroperoxide lyase expression in olive affects plant growth and results in altered volatile profile Plant Science

The C6 aldehydes, alcohols, and the corresponding esters are the most important compounds of virgin olive oil aroma. These C6 volatile compounds are synthesized via the 13-hydroperoxide lyase (13-HPL) branch of the lipoxygenase pathway. In this investigation, a functional analysis of the olive ( Olea europaea L.) 13-HPL gene by its overexpression and silencing in olive transgenic lines was carried out. With this aim, sense and RNAi constructs of the olive 13-HPL gene were generated and used for the transformation of embryogenic olive cultures. Leaves from overexpressing lines showed a slight increase in 13-HPL gene expression, whereas RNAi lines exhibited a strong decrease in their transcript levels. Quantification of 13-HPL activity in two overexpressing and two RNAi lines showed a positive correlation with levels of transcripts. Interestingly, RNAi lines showed a high decrease in the content of C6 volatiles linked to a strong increase of C5 volatile compounds, altering the volatile profile in the leaves. In addition, the silencing of the 13-HPL gene severely affected plant growth and development. This investigation demonstrates the role of the 13-HPL gene in the biogenesis of olive volatile com- pounds and constitutes a functional genomics study in olive related to virgin olive oil quality.


Introduction
Plants produce and emit a huge diversity of volatile organic compounds, which are released from different tissues such as leaves, fruits, flowers, and roots [1]. From a chemical point of view, these plant volatiles are organic lipophilic molecules characterized by low boiling points and high vapor pressures at ambient temperatures. They constitute a heterogeneous group of compounds that are synthesized from primary metabolites such as carbohydrates, amino acids, and fatty acids [2].
A group of fatty acid-derived plant volatiles of particular relevance are the green leaf volatiles (GLV), which consist of six straight-chain carbons (C6) compounds including aldehydes, alcohols, and esters. GLV are generated from linoleic and linolenic acids through the 13-hydroperoxide lyase (13-HPL) branch of the lipoxygenase (LOX) pathway by the successive action of several enzymes [3]. In a first step, different types of acyl-lipid hydrolases hydrolyze these polyunsaturated fatty acids from lipids. Next, LOX catalyzes their regiospecific oxidation, which can occur either at C9 (9-LOX) or at C13 (13-LOX) of the hydrocarbon backbone. The 13-hydroperoxides produced by 13-LOX are subsequently cleaved heterolytically by 13-HPL to C6 aldehydes. C6 aldehydes can then undergo reduction by alcohol dehydrogenases (ADH) to form C6 alcohols, which can finally be esterified with acyl-CoA derivatives in a reaction catalyzed by alcohol acyltransferases (AAT) [2].
Although un-stressed plants release only traces of GLV, they can be rapidly synthesized, within a few seconds after stress [4]. The increased emission of GLV can be caused by mechanical damage or herbivory, bacterial or fungal infection, and abiotic stresses including heat, high light, drought, and heavy metals [5]. GLV play key roles in plant defense responses and they have been implicated in numerous and important physiological functions such as to repel or attract herbivores and their natural enemies ("Plant indirect defenses"), to act as stress messengers to distal parts of the emitter plant or to neighbouring plants ("Plant communication"), to activate and prime plant defenses, to have antibacterial and antifungal properties, and to activate abiotic-stress related genes [4,5].
GLV are also important constituents of the flavor and fresh green aroma in plant foods such as fresh fruits and vegetables or juices such as virgin olive oil (VOO), and impart a characteristic odour referred to as the "green note" [6]. In fruits, volatile emission may have also evolved to facilitate seed dispersal by insects and animals. For humans, volatiles in fruits and vegetables have a considerable impact, as parameters of food quality and consumer preference [2]. For that reason, there is an increasing interest in the production of tailor-made aromas of diverse plant food products [1]. In this sense, the main efforts to improve aroma quality using a transgenic approach have been focused on tomato fruit, in particular by the modification of the LOX/HPL pathway. In an early attempt, the overexpression of an ADH resulted in the alteration of the aldehyde/alcohol ratio. Particularly, the increase in alcohols in the volatile fraction of transgenic tomato fruits was associated with the perception of a more intense ripe flavor at the early stages of ripeness [7]. In contrast, overexpression of 13-LOX or 9-HPL did not alter the volatile profile due to problems of cosuppression [8], or to the low content of the enzyme substrate [9]. In a more recent study, overexpression of ω-3 desaturases (FAD7 and FAD3) in tomato fruit increased the content of linolenic acid, the (Z)-hex-3-enal precursor, and reduced the hexanal/(Z)-hex-3-enal ratio resulting in an enhancement of the highly valued fresh notes [10].
VOO is one of the most appreciated edible plant oils, not only due to its potential health benefits but also to its excellent organoleptic properties [11]. In particular, VOO aroma is characterized by a unique balance of green and fruity attributes [12], and it plays a fundamental role in its sensory quality and consumer acceptance. Furthermore, VOO is the only food product requiring a sensory analysis to be classified in commercial categories. The absence of sensory defects in olive oil is necessary for the oil to be classified as "extra virgin''. This classification is ruled by the European Official Regulations for olive oil [13], and is performed by certified test panels, in which the aroma evaluation is an essential part.
The VOO volatile fraction consists of a complex mixture of more than 200 volatile compounds, among which, GLV are the most important either from a quantitative or qualitative point of view [14]. Especially prominent is the content of the aldehyde (E)-hex-2-enal, which was reported to account for more than 50 % of volatile fractions in European olive oils [15].
The participation of the 13-HPL branch of the LOX pathway in the biosynthesis of the C6 volatile compounds mainly responsible for VOO aroma has previously been demonstrated [16,17]. In particular, the branch with linoleic acid as precursor produces hexanal, hexan-1-ol, and hexyl acetate; the former and the latter being responsible for desirable perceptions. This branch can be considered as the green-sweet aspect of the global green flavor. A second branch is involved in the main green notes perception, producing (Z)-hex-3-enal, (Z)-hex-3-en-1-ol, and (Z)-hex-3-en-1-yl acetate. The third branch gives rise to (E)-hex-2-enal and (E)-hex-2-en-1-ol, and could be considered as the bitter-astringent aspect of the green sensory perceptions [18]. Additionally, Angerosa et al. [14] pointed out the relevance of the straight-chain five-carbons (C5) compounds also present in the volatile fraction of olive oil. C5 compounds would be generated through an additional branch of the LOX pathway that would involve the production of a 13-alkoxyl radical by 13-LOX in a homolytic way (Fig. 1).
De novo synthesis of the C6 and C5 volatile compounds occurs when enzymes and substrates meet as olive tissues containing chloroplasts such as leaves and fruits, are disrupted during homogenization [17]. They are qualitatively the same compounds found in VOO, which are responsible for its appreciated aroma. The VOO extraction process is carried out only by physical methods, involving olive fruit crushing, malaxation, and oil separation. Therefore, the synthesis of VOO aroma compounds seems to depend mainly on the enzymatic activities of the LOX/HPL pathway during the industrial oil extraction process and the availability of their substrates [17,[19][20][21][22].
In olive, different LOX isoforms have been purified and characterized from olive fruits [23,24] and olive callus [25], both soluble and membrane-bound. In addition, four LOX genes have been isolated and characterized to date: two LOX genes (Oe1LOX2 and Oe2LOX2), encoding isoforms that show strictly 13-LOX activity and possibly chloroplast localization [26], and two LOX genes (Oe1LOX1 and Oe2LOX1) encoding putative cytosolic isoforms that exhibit mainly 9-LOX activity [27,28]. All four olive LOX genes are expressed in mesocarp and leaf tissues, with major involvement of the Oe2LOX2 gene in the synthesis of VOO volatile compounds [26].
On the other hand, a native membrane-bound HPL enzyme has been purified and characterized from olive fruit [29]. Only one olive HPL gene (OeHPL) has been cloned and characterized so far [30,31]. The OeHPL gene codes for an HPL protein with a putative chloroplast localization and strict specificity for 13-hydroperoxides, explaining the absence of C9 volatile compounds in the aroma of olive oil [32]. The OeHPL gene is expressed in olive leaves and mesocarp, displaying a slight, though significant, maximum at the onset of fruit ripening [30]. Additionally, olive 13-HPL seems to be involved in the response to different abiotic stresses [33] and olive fly infestation [34].
Production of olive oils with improved aroma is considered a priority in olive breeding programs since new cultivars with the enhanced sensory quality might further stimulate VOO consumption. Because olive is a recalcitrant species for genetic transformation, only recently transgenic olive plants have been obtained to study processes such as flower induction [35], drought tolerance [36], or fungal resistance [37,38]. However, a functional genomic approach in olive with the aim of improving olive oil quality has not been addressed so far. The goal of this investigation was to assess the effect of overexpressing and silencing the 13-HPL gene, putatively involved in the generation of volatile compounds, on the volatile profile of the leaves of olive transgenic plants. In addition, the effect of 13-HPL gene manipulation on olive plant growth and development was also evaluated.

Generation of constructs and olive transformation
For 13-HPL overexpression, the binary vector p35SGUSINT [43], containing the GUS gene under the control of the CaMV35S promoter, was digested with EcoRI and SalI to eliminate the duplicated multicloning site, and then, with SmaI and SstI to remove the reporter gene. Subsequently, the complete coding sequence of the olive 13-HPL gene [30] was cloned into the BamHI site of the vector, yielding the plasmid p35S-OeHPL. For RNAi silencing, a 320 bp PCR fragment of the olive 13-HPL gene, located in the 5 ′ region of the open reading frame and outside of the cytochrome P450 domain, was cloned on both sides of the PDK intron in the pHANNIBAL vector [44]. The resulting plasmid was digested with NotI and the generated fragment was subcloned in pART27 [44], yielding the pRNAi-OeHPL plasmid. In both plasmids, p35S-OeHPL and pRNAi-OeHPL, the correct orientation of the gene was confirmed by restriction analysis, and the cloning junctions were checked by sequencing before transformation. Both binary vectors used for olive transformation contained the nptII gene for antibiotic selection. These plasmids were introduced into the disarmed A. tumefaciens strain AGL1 [45] by the freeze-thaw method [46]. Agrobacterium cultures were incubated at 28 • C in LB medium supplemented with 10 mg/L rifampicin and 50 mg/L kanamycin, at 250 rpm. Before embryogenic callus inoculation, the bacterial suspension was centrifuged at 5000 g, the pellet washed with 10 mM MgSO 4 , and diluted in the ECO medium at 0.5− 0.6 OD 600 .
Globular somatic embryos were transformed as described by Torreblanca et al. [47]. Briefly, the explants were incubated in a diluted Agrobacterium culture for 20 min under mild agitation, blotted dried on sterile filter paper, and cultured on ECO solid medium in Petri dishes at 25 • C in darkness for 2 days. Then, explants were washed with ECO liquid medium supplemented with 250 mg/L cefotaxime and timentin at 25 • C for 2 h, gently dried on sterile filter paper, and cultured on ECO solid medium supplemented with 250 mg/L cefotaxime, 250 mg/L timentin, and 200 mg/L paromomycin. Explants were incubated in darkness for at least 4 months with reculturing on fresh selection medium at weekly intervals during the first month and bi-weekly thereafter. To assure the recovery of non-chimeric embryogenic lines, explants that showed proliferation in the solid selection medium were grown individually in 250 mL culture flasks containing 40 mL of liquid ECO medium supplemented with 250 mg/L cefotaxime and 100 mg/L paromomycin, in an orbital shaker at 120 rpm for 3 weeks. Afterwards, suspensions were filtered through a 2-mm mesh and embryogenic calli cultured in solid selection medium. For plant regeneration, somatic embryos were cultured on ECO maturation medium, ECO medium without growth regulators and supplemented with 1 g/L activated charcoal for 8 weeks [48]. Then, mature embryos were transferred to germination medium, modified MS with 1/3 MS macroelements, MS microelements, 10 g/L sucrose [49], for 3 months under 40 μmol m − 2 s − 1 irradiance level. Excised shoots from germinated embryos were further multiplied in modified RP medium [50] as indicated by Vidoy-Mercado et al. [51]. For rooting, apical shoots, 2 cm long, were cultured on tubes containing 3 mL of basal RP medium supplemented with 10 mg/L indole-3-butyric acid (IBA) for 3 days and later on transferred to solid RP medium without hormones and supplemented with 1 g/L activated charcoal. Rooted shoots were acclimatized in jiffy trays with peat moss:perlite (1:1). After acclimatization, plants were potted in 12 cm pots containing the same mixture but supplemented with 2 g osmocote fertilizer per pot, and grown in a confined greenhouse for two years with a cooling system, 30 • C maximum temperature, and daylight conditions. Lowest temperature during winter time was 5 • C. From March to November, foliar fertilizer sprays were applied at 2-4 week intervals, alternating Stimufol, Isabión and Sequestrene 138 (Syngenta) as dosages indicated by the producers, and plants were watered when needed. For phenotypical characterization, shoot length and diameter of six plants from the different transgenic lines were recorded. Three biological replicates of fully developed young leaves (upper third of the plant) from olive 13-HPL transgenic and non-transgenic control lines were collected, frozen in liquid nitrogen, and kept at -80 • C for further molecular analysis, 13-HPL activity assay, and analysis of volatile compounds.

Molecular analysis of transgenic 13-HPL olive plants
Genomic DNA was extracted from leaves of in vitro plants using the protocol of Healey et al. [52]. To amplify the nptII gene the primers used were F: 5 ′ -GAGGCTATTCGGCTATGACTG-3 ′ and R: 5 ′ -ATCGGGAGCGGCGATACCGTA-3 ′ . All PCR reactions were prepared in a final volume of 20 μL containing 0.5-1 μl of genomic DNA and 0.5 μM of each primer. Amplification conditions consisted of 4 min at 95 • C, followed by 35 cycles of 1 min at 95 • C, 45 s at 59 • C, and 1 min at 72 • C, with a final extension step of 10 min at 72 • C.
Total RNA isolation was performed using 100 mg of frozen leaf tissue and the Spectrum™ Plant Total RNA kit (Sigma-Aldrich) following the manufacturer's instructions. RNA quality verification, removal of contaminating DNA, and cDNA synthesis were carried out according to Hernández et al. [53].
Gene expression analysis was performed by quantitative PCR (qPCR), using a CFX Connect real-time PCR System and iTaq Universal SYBR Green Supermix (BioRad, California, USA), as previously described [54]. Primers for gene-specific amplification of olive 13-HPL gene were previously designed [30]. qPCR data were calibrated relative to the 13-HPL gene expression level in control lines following the 2 − ΔΔCt method for relative quantification [55]. The data are presented as means ± standard deviation (SD) of three biological replicates, each having two technical replicates per 96-well plate.

13-HPL activity assay
For HPL activity assay, crude extracts were obtained by grinding 1 g of leaf frozen tissue in 20 mL 50 mM HEPES-NaOH (pH 7.5), 20 mM KCl, 2 mM MgCl 2 , 2 mM EDTA, 2 mM Na 2 S 2 O 5 , 7 mM DTT, 0.1 % ascorbate, 0.5 % Triton X-100, and 20 % PVPP. Grinding was carried out with an Ultraturrax homogenizer at 24,000 rpm, in three 1 min periods. The homogenates were filtered under vacuum through two layers of Miracloth and centrifuged at 27,000 g for 20 min at 4 • C. The supernatant was centrifuged again at 10,000 g for 10 min at 4 • C and used as a crude extract.
In vitro 13-HPL activity from crude extracts was measured by monitoring the decrease of A 234 due to the disruption of the conjugated diene chromophore of the substrate at 25 • C [33]. The 1.5 mL standard assay mixture consisted of 100 mM sodium phosphate buffer (pH 8.0), 8 μL of 10 mM substrate solution, and 10− 20 μl of enzyme solution. One unit (U) of 13-HPL activity is defined as the amount of enzyme catalyzing the formation of 1 μmol of product min − 1 . The 13-hydroperoxy isomer from the linolenic acid utilized as substrate was prepared using soybean LOX according to the method of Hamberg and Samuelsson [56].

Phenotypical analysis of selected transgenic 13-HPL olive plants
Transgenic lines S7 and i11 were selected for further characterization of the effect of 13-HPL manipulation on plant growth, in vitro and under greenhouse conditions. For in vitro characterization, control and transgenic shoot segments, 1.5− 2 cm long, with two nodes and deprived of the shoot apex were multiplied in modified RP medium as described in section 2.2 during four subcultures, 8 weeks per subculture. The average length of the main and first axillary shoots and the number of leaves per shoot were recorded at the end of each subculture. To evaluate rooting ability, apical shoots were subjected to a pulse of 10 mg/L IBA in liquid medium for three days, and then cultured in solid RP medium as described in section 2.2. The number of roots and the length of the main root were recorded after nine weeks of culture. Twenty nodal segments and 40 apical shoots per genotype were employed in the proliferation and rooting experiment, respectively.
Micropropagated control and transgenic selected lines were acclimated, transferred to the confined greenhouse, and grown for one year in 12 cm diameter pots with osmocote in the substrate and foliar fertilizer sprays as previously indicated, prior to their phenotypical and physiological characterization. Plants were watered when needed. Experiments were performed in June when the plants were in active growth. Shoot and root fresh weight, length of the main stem, number of lateral shoots, number and length of internodes in the main stem, number of leaves, and leaf area were recorded. Twelve plants and 25-90 leaves per genotype were measured.
A Pulse-Amplitude Modulated (PAM) chlorophyll fluorimeter (Junior-PAM, Walz) was used to determine the photosynthetic status of the plants growing in the greenhouse as described above. Leaves were covered with aluminum foil and maintained in darkness for at least 40 min and then, the parameter Fv/Fm was recorded. Afterwards, leaves were exposed to increasing intensities of actinic illumination to perform a rapid light curve (RLC). The relative electron transfer rate (ETR) was plotted against PAR to yield the following parameters: the initial slope of the RLC (α) which is related to the quantum efficiency of photosynthesis; maximum electron transport rate (ETR m ); minimum saturating irradiance (E k ). Four plants per genotype and four leaves per plant were measured.
Shoot water potential was measured at midday, between 12 and 13 h, with a Scholander pressure chamber (PMS Instruments). Plants were watered at field capacity the day before measurements. Four plants per genotype and three measurements per plant using the main stem and different lateral shoots were employed.

Analysis of volatile compounds
The analysis of volatile compounds was carried out through the leaf homogenate approach [10]. For this purpose, frozen leaves were homogenized in 2 mL 200 mM Na-phosphate buffer (pH 6.7), 250 mM sorbitol, 10 mM EDTA, 10 mM MgCl 2 , and 1% glycerol by means of an Ultraturrax at 24,000 rpm for 30 s. After an equilibrium period of 5 min at 25 • C, homogenate aliquots of 0.5 mL were taken into 12 mL vials containing 1.5 mL of a saturated CaCl 2 solution that halts enzymatic changes which might induce quantitative and qualitative alterations in the samples' volatile profile following the olive leaf homogenization. Then, vials were sealed and stored at − 18 • C until analysis. Homogenate samples were conditioned to room temperature and then placed in a vial heater at 40 • C. After 10 min of equilibrium time, volatile compounds from headspace were adsorbed on a solid-phase microextraction (SPME) fiber DVB/Carboxen/PDMS 50/30 μm (Supelco Co., Bellefonte, PA) according to Luaces et al. [57]. The sampling time was 30 min at 40 • C, and it was carried out in triplicate. Desorption of volatile compounds trapped in the SPME fiber was done directly into the GC injector. Volatiles were analyzed using an HP-6890 gas chromatograph equipped with a DB-Wax capillary column (60 m ×0.25 mm i.d., film thickness 0.25 μm; J&W, Scientific, Folsom, CA, USA). Operating conditions were as follows: N 2 as carrier, gas injector and detector at 250 • C, column was held for 6 min at 40 • C and then programmed at 2 • C min − 1 to 128 • C. Quantification was performed using individual calibration curves for each identified compound by adding known amounts of different compounds to re-deodorized high oleic sunflower oil. Compound identification was carried out on a 7820A/GC-5975/MSD system equipped with a similar stationary phase column and chromatographic conditions, matching against the Wiley/NBS Library, and by GC retention time against standards. Main volatile compounds were clustered into C5 and C6 groups according to the chain length [17,33].

Statistical analysis
Data were subjected to analysis of variance (ANOVA) using SPSS software version 23. The Levene test for homogeneity of variances was performed prior to ANOVA, and multiple mean comparisons were done by Tukey test. All tests were carried out at P = 0.05.

Generation of transgenic 13-HPL olive plants
Globular somatic embryos were inoculated with the A. tumefaciens strain AGL1 harbouring the binary vectors p35S-OeHPL or pRNAi-OeHPL for overexpression or RNAi silencing of the OeHPL gene, respectively. The transformation results obtained are summarized in Tab. S1. After the two steps selection procedure in solid and liquid medium containing paromomycin, nine overexpressing and 11 RNAi lines showing a good proliferation rate in the presence of the antibiotic were recovered. The final transformation rate based on paromomycin resistance was close to 0.5 %.
Five overexpressing (S lines) and five RNAi (i lines) independent transgenic lines were randomly chosen for further analysis. Somatic embryos from these lines were maturated following the procedure of Cerezo et al. [48] and later germinated in the medium of Clavero-Ramírez and Pliego-Alfaro [49]. The conversion rate, mature somatic embryos producing shoots and roots or only shoots, varied in the range 10-43 %, being the mean value slightly higher in the case of the RNAi lines (Tab. S1). Excised shoots from germinated embryos were multiplied in a modified RP medium as indicated by Vidoy-Mercado et al. [51] and later acclimatized to greenhouse conditions for further molecular and phenotypical characterization.

Molecular and biochemical characterization of transgenic 13-HPL olive plants
All transgenic lines analyzed showed the correct PCR amplification of a fragment from the kanamycin resistance nptII gene, using DNA isolated from leaves of the plants recovered from the putative transgenic somatic embryos and confirming their transgenic nature, while no amplification was detected using DNA from control non-transgenic plants (Fig. S1).
The transcript levels of the olive 13-HPL gene in leaves from overexpressing and RNAi transgenic lines were determined using qPCR. Four out of the five lines transformed with the p35S-OeHPL vector displayed 13-HPL mRNA levels at least 2-fold higher than control non-transformed plants, which suggests that the 13-HPL gene is effectively overexpressed in the transgenic overexpressing lines (Fig. 2a). The highest expression was detected in line S7. In the case of RNAi lines, 13-HPL expression decreased markedly in the five lines analyzed, ranging the level of gene silencing from between 85 % in line i3 to 35 % in line i20 (Fig. 2b).
In order to evaluate the 13-HPL activity, the two overexpressing lines with higher 13-HPL transcript levels, S5 and S7, as well as two RNAi lines with the lowest expression, i3 and i11, were chosen (Fig. 3). The two overexpressing lines analyzed showed a 3-fold increase of 13-HPL activity compared to the control line. In contrast, 13-HPL activity was reduced by 80 % in RNAi lines i3 and i11. Therefore, 13-HPL activity in transgenic olive leaves correlated well with the level of 13-HPL gene expression.  Crude extracts from leaves of transgenic 13-HPL olive lines were obtained and used to determine 13-HPL activity as described in Material and methods. Bars correspond to mean ± SD (n = 6). The means with different letters indicate significant differences by Tukey test at P = 0.05.

Effect of overexpression and silencing of 13-HPL on olive plants growth and development
After acclimatization, control and transgenic plants were maintained in a confined greenhouse under natural daylight and temperature conditions for over two years. During this time, three out of the five RNAi lines displayed a reduced growth, being their average values of shoot length and diameter of shoot lower than control or overexpressing lines, especially in the case of i3 and i11 plants (Fig. S2). Overexpressing lines showed a shoot length similar to control, except line S8 that was slightly smaller.
Based on these observations, transgenic lines S7 and i11, which showed the largest modifications on 13-HPL activity levels, were selected to further characterize the effect of this gene on plant growth, both in vitro and under greenhouse conditions. To determine the in vitro behavior, nodal segments were multiplied as described by Vidoy-Mercado et al. [51] during four subcultures. Shoot length was slightly reduced in both transgenic lines, as well as the number of leaves in S7 line (Table 1). Proliferated shoots were exposed to IBA for three days and transferred afterwards to basal RP medium for root emergence. The percentage of rooted shoots was similar in the control and i11 line, around 95 %, but slightly lower in the case of S7 plants, 80 %. The three genotypes showed a similar number of roots per shoot and root length (Table 1).
Afterwards, plants were successfully acclimated and maintained in a confined greenhouse for one year. As previously observed, i11 plants showed a reduced growth rate and vigor; some of the i11 plants died during the one year period of growth in the greenhouse, and most of them displayed a large number of senescent leaves (Fig. 4a). On the contrary, no evident differences were observed between S7 and control plants. After one year in the greenhouse, several growth variables were measured. No significant differences were detected in these variables between S7 and control plants except for shoot fresh weight and leaf area that were significantly higher in the 13-HPL overexpressing line ( Table 2). By contrast, RNAi line i11 showed a significant reduction in plant growth (Table 2). Root and especially shoot fresh weight were significantly lower in these plants than in control, resulting in a diminished shoot/root ratio (Fig. 4b). Shoot length, the number of lateral shoots, as well as the number of leaves, and the average leaf area were also diminished in i11 plants when compared with controls ( Table 2). The aspect of the plants after one year of growth is shown in Fig. 4c. Moreover, RNAi line i11 plants continued to show poor growth, leaves turned yellow, and many of these plants died.
To determine whether the reduced growth of 13-HPL silenced plants in the greenhouse was due to water restriction caused by a reduced root biomass or poor photosynthetic performance, shoot water potential and chlorophyll fluorescence kinetics were measured. The water potential of i11 shoots was higher than S7 but did not differ from control plants (Fig. S3). This variable correlated negatively with the shoot and root fresh weights (Pearson's coefficients of -0.64 and -0.61, both significant at P = 0.05, for shoot and root, respectively). On the other hand, transgenic and control plants displayed a similar Fv/Fm ratio, close to 0.78 (Tab. S2). This ratio is commonly used as an indicator of photoinhibition as a result of environmental stress [58]. Similarly, the three photosynthetic parameters obtained from the rapid light curve, quantum efficiency (α), maximum electron transport rate (ETR m ), and minimum saturating irradiance (E k ), were similar in the three genotypes (Tab. S2). Altogether, these results indicate that i11 plants growing in the greenhouse were neither water-stressed nor photoinhibited as a result of 13-HPL silencing.

Effect of overexpression and silencing of 13-HPL on olive plants volatile composition
The volatile compounds generated as a consequence of the homogenization of olive leaves are essentially the same as those obtained by homogenization of the mesocarp of the olive fruit [17] and, consequently, of those present in VOO. Keeping in mind that olive trees exhibit a long juvenile period and need about six years to produce enough fruits to obtain olive oil, the effect of the transformation of the olive tree with the 13-HPL gene on the volatilome of the leaf is a suitable approximation to know the effect that this transformation would have on the volatilome and aroma of VOO.
After confirming that 13-HPL activity was significantly increased or reduced in olive transgenic leaves, the volatile composition was analyzed in the same selected two overexpressing (S5, S7) and two RNAi (i3, i11) lines used for the 13-HPL activity assay. The results showed that in the case of the overexpressing lines a small increment in the content of the C6 volatiles was observed compared to the control lines (Fig. 5), mainly for (Z)-hex-3-en-1-ol (Tab. S3). In contrast, no significant changes were detected for C5 volatiles (Fig. 5; Tab. S3).

Discussion
Modification of volatile production via metabolic engineering has the potential to improve cultivated plant species. However, this requires both knowledge of the metabolic pathways responsible for the biosynthesis of the volatile compounds and identification of the genes that regulate their synthesis [1]. Notably, the information obtained from transgenic studies using model plants cannot be directly extrapolated to crop plants since the number of gene/isoforms involved in a metabolic step and their regulation mechanisms could be different, altering the expected result.
In this investigation, we have studied the effect of overexpressing and silencing the 13-HPL gene, which belongs to the LOX pathway, on the volatile composition in olive. Interestingly, since only one HPL gene is present in the olive genome [59], it could be more likely to detect altered properties in the transgenic olive plants.
Nine overexpressing and 11 RNAi independent transgenic olive lines were generated, and five of each class were randomly selected for further analysis. Observed transformation efficiency was 0.5 % (Tab. S1), lower than that obtained by Narváez et al. [37] when transforming olive globular embryos with the antimicrobial gene afp from Aspergillus giganteus (1.5 %), but higher than the value reported by Narváez et al. [38] following transformation with the AtNPR1 gene (0.28 %). These differences could be explained by the size of the transgene [38]. As usually noticed in most genetic transformation experiments, the transgene expression level varied significantly among the different lines. This is usually because of the effect of the copy number and other factors like the site of integration [60].
13-HPL activity augmented 3-fold in the leaves of the overexpressing Data correspond to mean ± SD (n = 20). Within each row, means with different letters indicate significant differences by Tukey test at P = 0.05.
lines S5 and S7, compared to the control line (Fig. 3). A similar increment was observed in transgenic Arabidopsis leaves overexpressing a bell pepper 13-HPL gene [61], although a much larger increase was detected in the case of a watermelon 13-HPL overexpressed in tobacco leaves [62]. On the contrary, 13-HPL activity highly decreased in the leaves of RNAi lines i3 and i11 (Fig. 3), as it has been reported for antisense-mediated depletion of the 13-HPL gene in transgenic potato [63] and Arabidopsis leaves [61]. The modified expression of the 13-HPL gene did not greatly affect the in vitro behavior of olive overexpressing S7 and RNAi i11 transgenic lines, which exhibited the biggest alterations on 13-HPL expression levels (Table 1). However, a differential behavior between both lines was observed when they were acclimated to ex vitro conditions in a confined greenhouse (Fig. 4, S2, Table 2). In the case of the overexpressing S7 transgenic line, no significant differences were detected in growth parameters in comparison to the control line, in agreement with the results obtained for tomato plants overexpressing a tea 13-HPL gene, whose growth and development was not affected during their whole life cycle [64]. Surprisingly, RNAi line i11 transgenic plants showed a significant reduction in growth rate and vigor, even reaching leaves senescence and death of the plant. This was a quite unexpected result since no effect on plant growth has been reported previously when the 13-HPL gene was silenced in annual plants such as potato [43], Nicotiana attenuata [65], or tomato [66]. In fact, in Arabidopsis thaliana Columbia-0 ecotype, a deletion in the coding region of the 13-HPL gene results in a truncated non-functional enzyme and no production of GLV [67], indicating that HPL activity is dispensable for normal growth and development [68]. The only exception reported so far is the rice hpl3 mutant, which was greatly affected in normal growth and development and displayed a severe phenotype, including necrotic lesions in the leaves, fewer tillers, decreased plant height, and partial sterility [69,70].
Although it has recently been reported that the 13-HPL branch of the oxylipin pathway is involved in tolerance to waterlogging [71] and protects against photoinhibition of photosynthesis [72], shoot water potential (Fig. S3) and chlorophyll fluorescence (Tab. S2) kinetics were  Data were obtained after one year of growth in the greenhouse and correspond to mean ± SD (n = 12). Within each row, means with different letters indicate significant differences by Tukey test at P = 0.05. On the other hand, when we analyzed the volatile compounds in the olive transgenic leaves of the overexpressing lines (Fig. 5, Tab. S3), only a slight increase in the C6 volatiles content was detected, mainly in (Z)hex-3-en-1-ol. Analogous results were obtained when a tea 13-HPL gene was overexpressed in tomato [64]. In general, when the 13-HPL gene has been overexpressed in leaves, no changes in the content of C6 volatile compounds have been observed [61,73,74]. All these data suggest that the flux through the HPL branch of the oxylipin pathway is not regulated by the level of HPL activity, but by the availability of the substrates precursors, linoleic and linolenic acids [75].
In the case of the RNAi lines (Fig. 5, Tab. S3), we detected a high decrease in the content of all C6 volatile compounds and their total content, as has previously been reported in the leaves of antisensemediated depleted 13-HPL in potato [63,76], N. attenuata [77], Arabidopsis [61], and tomato [66], as well as in the Arabidopsis 13-hpl mutants Columbia-0 ecotype [67], and T-DNA knockout line [78], and rice [69,70]. Not only that, concomitant with the strong reduction of C6 volatiles, a high increase of C5 volatile compounds was observed, including C5-aldehydes, C5-ketones, and C5-alcohols. The same effect has also been described in the antisense potato knockdown line [63,76] and the Arabidopsis T-DNA knockout line [78].
The generation of the C5 volatile compounds has been attributed to an additional branch of the LOX pathway. LOX would also catalyze, after the hydroperoxide formation, its cleavage via an alkoxy radical giving rise to the formation of stabilized 1,3-pentene radicals, as demonstrated in soybean seeds [79]. This radical would undergo subsequent non-enzymatic β-scission in a homolytic way to form a 1,3-pentene allylic radical that could be chemically dimerized leading to C10 hydrocarbons (known as pentene dimers) or react with a hydroxyl radical to produce C5 alcohols. The latter can be enzymatically oxidized by ADH to synthesize the corresponding C5 carbonyl compounds, as suggested to occur in soybean leaves [80]. In that way, because of the low HPL activity, fatty acid hydroperoxides are accumulated in the plant tissue and endogenous LOXs that compete for the substrate are recruited for its disposal [63,81]. Accordingly, Sánchez-Ortiz et al. [17] observed a decrease in the content of C5 compounds when HPL activity was increased during homogenization of olive fruit mesocarp. The essential involvement of a 13-LOX (TomloxC) in the synthesis of C5 volatiles was confirmed in tomato fruit [82]. The in vivo relevance, if any, of this alternative mechanism is not known. In fact, C5 volatiles have been shown not to have an important defensive function against the bacterial pathogen Xanthomonas campestris in tomato leaves [82].
The 13-HPL silenced olive transgenic lines produced volatiles with important differences concerning the sensory notes with respect to the control lines. The predominant green notes found in the aroma generated by the leaves of the control lines as a result of the higher presence of C6 alcohols and esters are in contrast with the sensory notes of the silenced lines due to the higher concentrations of C5 compounds. Therefore, given that the LOX pathway seems to be operative in both olive leaves and olive fruit according to the similar profiles of volatile compounds [17], this study represents a first step in the goal of improving the aroma of olive oil by modifying the activity level of the enzymes involved and constitutes an initial stage in the generation of olive oils with tailor-made aromas.
Olive oil quality may be defined in a number of ways, but the consumers' sensory perception of flavor is the ultimate determinant. Largely, the range of volatile compounds present in the oil determines its aroma. Hence, it is critical to understand the formation of these volatile compounds and promote certain favourable flavor attributes in olive oil [83]. In the future, it would be possible to generate new olive cultivars with modified expression/activity levels of specific genes/enzymes of the LOX pathway present in the olive fruit to improve some desired volatile compounds. For instance, enhancement of 13-HPL and reduction of ADH and AAT activities can be used to increase the "green" aroma. In the same way, increasing AAT activity can be applied to elevate the fruity aroma [84].

Conclusions
Modification of 13-HPL expression in olive results in an altered volatile profile in the leaves. In particular, silencing of the gene causes a high decrease of C6 volatiles as well as a strong increment of C5 volatiles. Furthermore, we have shown evidence that 13-HPL activity is essential for the normal growth and development of the olive tree, although the underlying mechanism is still unknown. The present work represents a functional genomics study, which confirms the physiological role of genes related to olive oil quality. This information will be useful in olive breeding programs to develop molecular markers for marker-assisted selection of new varieties with improved VOO aroma.

Author contributions
SC, EP-R, and NM were responsible for the obtainment, maintenance, and phenotypical characterization of transgenic plants. MLH, MDS, and JMM-R carried out the generation of constructs and the expression analysis.
LG-V and CS performed the activity assays and volatile analysis. CS, JAM, FP-A, and JMM-R designed the study. JMM-R prepared the manuscript and collected contributions from all authors.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.