Influence of drought stress on increasing bioactive compounds of pomegranate (Punica granatum L.) juice. Exploratory study using LC–MS-based untargeted metabolomics approach

Water deficit is one of the most important abiotic factors that reduce the agricultural productivity. The impact of irrigation levels on bioactive compounds of one commercial pomegranate cultivar (‘Mollar de Elche’) was studied through LC–MS-based untargeted metabolomics approach. Two treatments were applied to explore crop responses to the timing and the regime of the irrigation applications. The main goal of the present study was to validate the untargeted metabolomics approach as analytical tool to find new markers related with the water deficit. Robust classification and prediction model was built with the use of supervised technique, partial least squares discriminant analysis (PLS-DA). The results revealed that pomegranate juices obtained from plants grown under DIr conditions experimented physiological stress, which significantly increased their bioactive compounds (polyphenols, phenylpropanoids, peptides, benzoic acid, tannins, and phospholipids) in comparison with the Control treatments. The results indicate that controlled irrigation stress would increase the production of bioactive compounds by increasing the phenylpropanoids metabolism. New metabolites were tentatively identified in pomegranate juice by the untargeted metabolomics techniques. This study showed, for the first time, the satisfactory application of the untargeted metabolomics approach using an UHPLC-QTOF system to identify bioactive secondary metabolites from pomegranate juices to discriminate among different irrigation conditions.


Introduction
During the last decade, the impact of climate change on water scarcity focused in the Mediterranean region has been highlighted [1].Thus, sectors such as agriculture might be negatively influenced, as it depends directly on water availability.Thus, to obtain positive results, the sensitivity of the crop to water deficit must be studied throughout the life cycle to determine the critical phases of the crop, to know in which periods this technique can be applied and in which periods full irrigation should be applied [2,3] Different studies have been published describing the physiological response, growth, yield, and fruit quality of pomegranate trees under different irrigation strategies, including sustained and regulated irrigation treatments.However, inconsistent results have been found, explained by differences in the amount and frequency of irrigation, cultural practices, differences in crop load, variety, age of the trees, climatic and soil characteristics, etc., which makes difficult to compare between similar field studies [4][5][6][7].
To both save water and maximize the fruit quality, different irrigation strategies, like the above described, have been positively applied to drought-tolerant species like pomegranate [3,8].Therefore, south-eastern of Spain has become the hub of a large number of pomegranate export companies [9].
Pomegranate (Punica granatum L.) is one of the oldest known edible fruit widely consumed as fresh fruit and juice.This fruit has generated great interest because its consumption has been associated with positive effects on human health [1,2].Thus, the presence of anthocyanins (monoglycosides and diglycosides of cyanidin, delphinidin, and pelargonidin), ellagic acid and ellagitannins (mainly punicalagins and punicalins), gallic acid and gallotannins, proanthocyanidins, flavanols, and lignans is responsible for health promotion in humans due to the biological activities exerted both directly or after an assimilation mediated through colonic biotransformation [10][11][12].Thus, it has been widely described that ellagitannins and ellagic acid are transformed by the gut microbiota to produce urolithins, bioavailable metabolites that can exert anti-inflammatory and anti-carcinogenic effects, and can reach high concentrations in both normal and tumor human colonic tissues [13].As antimicrobial, anti-inflammatory, astringent, antitussive, and antidiarrheal properties, pomegranate juice has gained a reputation as an easily accessible superfood and is being sold as a high-quality food item [14,15].
In recent years, a total analytical approach known as "untargeted metabolomics" has experienced a significant increase in food studies [16,17].Metabolomics focus on the study of low-molecular-weight molecules (metabolites) to explore unknown food constituents, generating a detailed and comprehensive metabolic chemical profile of them.The goal is the identification of metabolites (biomarkers) that can discriminate between sample populations and/or the generation of statistical models able to classify samples and predict class memberships [17].To enable the large-scale determination of unknown compounds, the use of high-throughput analytical techniques, such as high-resolution mass spectrometry (UHPLC-QTOF) is essential.Statistical treatment using statistical tools, such as principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA), is necessary for the discrimination/classification of the samples and the development of predictive models [18].
Due to the relevance of pomegranate as a potential source for the elaboration of promising functional foods, we addressed the effect of two different irrigation conditions of pomegranate cultures on the bioactive metabolite accumulation in their juices through LC-MS-based "untargeted metabolomics" approach.To the best of our knowledge, no information is available since this study is the first to describe an untargeted metabolomics approach using an UHPLC-QTOF system to identify bioactive secondary metabolites from pomegranate juices to discriminate among different irrigation conditions.

Experimental design and irrigation treatments
The experiment was carried out during the season of 2021 at the experimental farm of CEBAS-CSIC in Santomera (Murcia, Spain) (38°06′ N, 1°02′ W).The plant material was own rooted 15-year-old pomegranate trees (Punica granatum (L.) cv.Mollar de Elche) in a 3 m × 5 m spacing pattern and the total area cultivated was 0.8 ha.The sandy clay loam soil of the experimental site was characterized by a high stone content (39% by weight) and a bulk density of 1.37 g/cc.The volumetric soil water content at saturation, field capacity, and permanent wilting point was 49, 29, and 18%, respectively.Agro-meteorological data were recorded by an automated weather station located in the CEBAS-CSIC experimental field station, which read the values every 5 min and recorded the averages every 15 min (Fig. 1).Irrigation water had an electric conductivity of 0.8-1.0dS m −1 .
Pest control and fertilization practices were those usually used by local growers, and no weeds were allowed to develop within the orchard.The pomegranate trees had only one trunk.They are lightly pruned every year and sprouts and suckers are removed as they appear, to encourage fruit production.
Irrigation was conducted daily at sunrise and sunset, using a drip irrigation system, with a lateral irrigation line per tree row and six drippers per tree spaced 50 cm between drippers, set at a rate of 2.2 l h −1 .
During the season (May-October 2021) (day of the year, DOY, 121-280), two treatments were applied to explore crop responses to the timing and the regime of the irrigation applications: -Control, C: irrigation was scheduled to replace 120% of the estimated crop evapotranspiration (ET c ). Crop ET c was calculated as ET c = ETo × K c .ET o was calculated using the Penman-Monteith formula [19] and the K c values reported were based on results reported by Intrigliolo et al. [20] This was done to ensure that the potential crop water needs were replaced.-DI r : irrigation was applied at 25% of the water requirements of the crop during the ripening phase (3 weeks before harvest), while the rest of the season 120% ET c was applied (DOY 263-286).
The reductions in the quantity of water applied during the water deficit periods were achieved by reducing irrigation duration, while frequency of irrigation was always the same for all treatments.According to the theoretical irrigation, 5738.6 m 3 ha −1 and 4890.2 m 3 ha −1 of water were applied in the treatments Control and Dir, respectively, from 1 May 2021 to the harvest time, on 13 October 2021 (DOY 121-286).

Water relations determinations
The stem water potential (Ψ stem ), leaf osmotic potential (Ψ os ), and leaf osmotic potential at full turgor (Ψ 100s ) were determined at the end of the experiment.
Leaves were taken from the north facing side and were covered with aluminium foil for at least 2 h before measurements.Ψ stem was estimated, according to Scholander et al. [21], using a pressure chamber (Model 3000; Soil Moisture Equipment Co., Santa Barbara, CA, USA) in which leaves were placed in the chamber within 20 s of collection and pressurized at a rate of 0.02 MPa s −1 [22].Adjacent leaves were also collected, frozen immediately in liquid nitrogen (− 196 °C) and subsequently stored at − 30 °C.After thawing, the leaf osmotic potential (Ψ os ) was measured in the extracted sap using a WESCOR 5520 vapor pressure osmometer (Wescor Inc., Logan, UT, USA), according to Gucci et al. [23].The leaf osmotic potential at full turgor (Ψ 100s ) was estimated as indicated above for Ψ os , and then placed in distilled water overnight to reach full saturation.

Gas exchange measurements
Leaf stomatal conductance (g s ) and net photosynthesis (P n ) were determined at the same time and in the same leaves as stem water potential was measured, using a gas exchange system (LI-6400; LI-COR Inc., Lincoln, NE, USA), fitted with an infrared gas analyzer attached to a leaf chamber fluorimeter (LCF) (6400-40B, 2 cm 2 leaf area, Licor Bioscience, Inc., Lincoln, NE, USA).The reference CO 2 , photosynthetically active radiation (PAR), and speed of the circulating air flow inside the system were set at 400 ppm, at 1500 µmol m −2 s −1 , and at 500 µmol s −1 , respectively.

Yield and fruit quality
Pomegranate fruits were harvested at commercial maturity on 1-2 harvesting days starting at the beginning of October 2021 (DOY 286).The yield (expressed as marketable and total kg of fruits per tree) and marketable and total number of fruits per tree were determined in four trees per treatment (three replications).The mean fruit weight was calculated from total mass and number of fruits per tree.Four fruits per treatment were selected, which were peeled by hand and the arils were separated and squeezed to extract the juice.Pomegranate juice was centrifuged at 10,480 × g for 10 min and each supernatant was filtered through a 0.45 µm cellulose nitrate membrane filter.The total soluble solids (TSS) (expressed as ºBrix) and titratable acidity (TA) of the juice were measured by a refractometer ATAGO PAL-BX|ACID F5 Master Kit.

Pomegranate juice untargeted metabolomics analysis by UPLC-QTOF.
The analyses were carried out using an Agilent 1290 Infinity series LC system coupled to a 6550 I-Funnel Accurate-Mass QTOF (Agilent Technologies, Waldbronn, Germany) with a dual electrospray ionization interface (ESI-Jet Stream Technology) for simultaneous spraying of a mass reference solution that enabled continuous calibration of detected m/z ratios.
Non-diluted samples with hesperidin added as internal standard at a final concentration of 500 μM (to correct for injection variability between samples and minor changes in the instrument response) were injected (1 µL) into a reversed phase column, a Poroshell 120 EC column (3 × 100 mm, 2.7 µm) from Agilent Technologies (Waldbronn, Germany) operating at 30 °C and a flow rate of 0.5 mL/min.The mobile phases used were acidified water (0.1% formic acid) (phase A) and acidified ACN (0.1% formic acid) (phase B).Metabolites were separated using the following gradient conditions: 0-3 min, 5-18% phase B; 3-10 min, 18-50% phase B; 10-13 min, 50-90% phase B. Finally, the phase B content was returned to the initial conditions (5%) for 1 min and the column re-equilibrated for 2 more minutes.Data were acquired using the Mass Hunter Workstation software (version B.08.00, Service Pack 1, Agilent Technologies).The system was operated using both negative and positive ion polarity and data were acquired in centroid and profile mode, with a data storage threshold of 5000 absorbance for MS and 5000 absorbance for MS/MS.The operating conditions were as follows: gas temperature of 280 °C, drying nitrogen gas of 9 L/min, nebulizer pressure of 45 psi, sheath gas temperature of 400 °C, sheath gas flow of 12 L/min, capillary voltage of 3500 V, nozzle voltage of 500 V, fragmentor's voltage of 100 V, skimmer of 65 V and octopole radiofrequency voltage of 750 V. TOF spectra acquisition rate/time was 1.5 spectra/s and 666.7 ms/spectrum, respectively, and transients/spectrum was 5484.The mass range was between m/z 50 and 1100.To assure mass accuracy during the MS analyses, external calibration of the instrument was performed at the beginning of the batch, introducing a mixture of reference compounds (Tuning Mix).Besides, continuous internal calibration was performed during analyses with the use of signals m/z 112.9855 and m/z 1033.9881 in negative polarity and m/z 121.0509 and m/z 922.0098 in positive polarity.Auto-recalibration reference mass parameters were a detection window of 100 ppm and a minimum height of 1000 counts.MS/MS conditions were a collision energy of 20 eV and an acquisition time of 100 ms/spectrum.Data were processed using the Mass Hunter Qualitative Analysis software (version B.08.00, Service Pack 1, Agilent Technologies).All samples were injected in the same batch and the order of sample injection was randomized to avoid sample bias.A mixture with one replicate of each group of samples was used as 'quality control' (QC) and was injected at the beginning and at the end of the batch.Besides, methanol injections were included every three samples as a blank run to avoid the carry-over effect.

Data treatment
The raw data files were acquired in profile file mode and were exported to MZmine software (Version 2.53, Copyright (c) 2005-2015 MZmine Development Team) to create the data matrix.The raw data were pre-processed by a batch set of parameters including the mass detection, chromatogram builder, and deconvolution and alignment algorithm.The data matrix was exported to Mass Profiler professional (MPP, Agilent technologies, Waldbronn, Germany) and Metaboanalyst 5.0 online platform (https:// www.metab oanal yst.ca/) for parallel data management.Data matrices were processed including log transformation and auto scaling prior to univariate and multivariate analysis [24].The multivariate analysis PLS-DA (Partial least square discriminant analysis) was performed to study the classification of the data samples groups and evaluate the group differences.The univariate analysis was performed by MPP software after the multivariate analysis evaluation.Data treatment through MPP software included filters by frequency of the data matrix to reduce the sample variability within each study group.T test unpaired (corrected p value cutoff: 0.05; p value computation: asymptotic; multiple testing correction: Benjamini-Hochberg) statistics analysis was applied to the data matrix to filter significant entities along the different samples groups.The final list of features was used for metabolite identification with purchased METLIN databases and according to the exact mass.

Water relations and gas exchange
Stem water potential (Ψ stem ) was lower in trees irrigated with the DI r treatment than in trees irrigated with the Control treatment at the end of the experiment (Fig. 2A).The leaf osmotic potential (Ψ os ) did not show statistical differences between both treatments, while Ψ 100s was lower in trees irrigated with the DI r treatment compared to those irrigated with the Control treatment (Fig. 2B, C).
Although trees irrigated with the DI r treatment experienced a decrease numerically of g s compared to those irrigated with the Control treatment, there were no statistical differences between both treatments due to the high variability found in the measurements (Fig. 3A).Nevertheless, the irrigation with DI r treatment caused a significant reduction of P n values at the end of the experiment (Fig. 3B).

Yield and fruit quality
At harvest, yield parameters such as marketable fruit weight (FW), total FW, marketable fruit number (FN), total FN, and mean FW were similar between both treatments and did not show statistical differences (Table 1).Likewise, the total soluble solids (TSS) and titratable acidity (TA) values were similar between both treatments.

Multivariate model analysis
The data matrix was created using the most extreme of irrigation groups including Control and DIr to find the clearest differences of the irrigation variable.The pre-processing operations gave a data matrix based on 1470 entities from full data set.PLS-DA model of final data matrix was created by Metaboanalyst 5.0 online platform to evaluate the classification of the samples into groups (Fig. 4).The calculated PLS-DA model, based on eight samples and three components, described 99.3% of the variance (R 2 = 0.993) according to the cross-validation prediction of Q 2 = 0.503.The first two principal components PC1 and PC2 explained 33.3% and 18.1% of the total variability.The discrimination model PLS-DA showed large differences between the study groups.The irrigation variable was mainly explained by the component 1; therefore, the differences showed on the sample metabolomes might be affected by the irrigation.The VIP (variable importance in projection) score was used to measure importance of the entities if the PLS-DA model and the entities were filtered according to a VIP > 1.The filtered list result in 303 entities from the full data set.
The principal class of metabolites highly affected by the irrigation conditions was the polyphenols (mainly flavonoids) including 15 metabolites and especially under DIr condition, where 13 polyphenols were identified as upregulated metabolites.In the same way, metabolites belonging to the phenylpropanoids, benzoic acid, tannins, and phospholipids were identified as well as upregulated under DIr condition.Metabolites as flavonoids derivatives from catechins,  quercetins, delphinidin, kaempferol, luteonin and cyaniding [26,27], which are the most important constituents of the pomegranate and commonly present, but also benzoic acids [28], phenylpropanoids [29], and tannins derivatives [30] might be identified in pomegranate juice.However, some of these as 3′-methoxyfukiic acid, 3′-methoxytricetin 7-glucuronide, hesperetin-7-O-glucuronide, sanguiin H4 and hamamelitannin are the first time they have been tentatively identified in pomegranate juice.This study suggested that untargeted metabolomics tools could be used as new approach to identify secondary metabolites affected by the water irrigation stress.This study also tentatively identifies unusual metabolites in pomegranate juice as phospholipids or peptides; however, these metabolites are related with the abiotic stress acting as membrane signals [31,32].The abiotic stress would activate the membrane phospholipases releasing phospholipids and triggering a signal cascade to finally induce the phenylpropanoids metabolism via activation of the phenylalanine ammonia lyase enzyme (PAL) [33].Thereby, the presence of PI (21:0/0:0) and seryl-valine would increment the response to the abiotic stress under DIr condition and the induction of the phenylpropanoids 3′-methoxyfukiic acid and phenethyl 6-galloylglucoside.Additionally, the flavonoids described also belong to the super class of phenylpropanoids, so the results suggested that a 25% reduction of the irrigation water might increase the production of these metabolites via phenylpropanoids metabolism.Exceptionally only the metabolite 6-hydroxydelphinidin 3-glucoside has been tentatively identified as completely absent in the Control group and present in the Control group.This metabolite may be considered as exclusive biomarkers of the DIr irrigation condition.If confirmed, this finding would be of great relevance since the delphinidin is the metabolite responsible for the magenta ~ purple color of the fruit [34].A greater presence of these compounds implies an important added value to the product since many of these metabolites are bioactive and have proven effects on human health [35][36][37].Therefore, the results showed that only a reduction until 25% of the irrigation provoke the stress to induce the increment of the secondary metabolites described above.
To evaluate the suitability of a specific irrigation strategy in pomegranate, knowledge is required regarding the effect of water deficits on crop growth during different phenological stages [38].There are numerous studies focused on the application of deficit irrigation during several phenological stages such as flowering, set fruit or linear growth of the fruit [20,[39][40][41], as well as the application of sustained deficit irrigation throughout the entire season [7,20,[40][41][42].These studies showed that pomegranate is capable of developing mechanisms of tolerance to water stress, although with different results in terms of fruit production and quality.On the other hand, the suppression or reduction of irrigation in the ripening phase of the fruit has been less studied [43][44][45].In this experiment, the reduction of Ψ 100s in trees irrigated with DIr treatment indicated that the plant was able to actively accumulate solutes in the leaf tissues in response to a decrease in water availability.By lowering leaf osmotic potential and leaf water potential, plants were able to take up water from the soil and maintain turgor pressure and the physiological activity in tissues.This behavior was also observed by other authors [46,47] confirming the resilience of this species to water stress.The results of gas exchange also indicated a regulation of stomatal aperture in these trees to avoid water losses, which caused a reduction in the production of photoassimilates during this stage.However, contrary to what might be expected, the yield was practically not negatively affected.In fact, Galindo et al. [44] found that the suppression of irrigation during 15 or more days before harvest, coinciding with the ripening phase of pomegranate trees, reduced the total and marketable fruit yield.In this experiment, applying irrigation at 25% of ET c during 3 weeks could maintain was able to keep the plant minimally hydrated, and thus this could prevent production from declining.In addition, results of TSS and TA in fruit juice are also not in concordance with Galindo et al. [45] or Laribi et al. [40] who reported that these parameters increased in pomegranate fruits when deficit irrigation (also 25% of ET c ) was applied, although this period was longer than in our case (approximately 45 days).
The present LC-MS untargeted metabolomics study aimed to identify significantly different metabolites found in each irrigation treatment that differed from those found in the Control treatment, and therefore might be considered potential biomarkers of the fruit.To the best of our knowledge, there is no data published so far to be compared with.Thus, we have found an extra difficulty to discuss our results with previous bibliography.There are two previous studies based on an "untargeted metabolomics" in pomegranate juice but they are not related to irrigation treatments.Thus, a previous study by Tang and Hatzakis [48] based on an "untargeted metabolomics" NMR platform described several biomarkers related with pomegranate juice cultivar and geographical location.Furthermore, Dasenaki et al. [49] described the LC-MS "untargeted metabolomics" as an excellent tool for the detection of pomegranate juice adulteration.In summary, clear advantages have been shown in using an untargeted metabolomics approach whatever the objective and treatment being pursued.
The metabolome of pomegranate fruits cultivated under extreme different irrigation treatments revealed clear divergences.According to our results using LC-MS-based untargeted metabolomics approach, the DI r treatment in pomegranate plants produces an enhancement in the functional properties of the corresponding fruits.Pomegranate juices from DI r significantly increased the concentration of bioactive compounds, such as, polyphenols, phenylpropanoids, benzoic acid, peptides, and phospholipids.These metabolite families represent great relevance when establishing irrigation conditions given their close relationship with the quality of the fruit and the benefits on human health, as already mentioned.Despite this, these metabolites only represent the tip of the iceberg of the complete metabolome.The preliminary list of discriminant entities consisted in 303 and finally only 21 were tentatively identified in this study due the exploratory nature.This result presents a huge source of metabolites susceptible to be confirmed as biomarkers by MS/MS or authentic standard.Because of this, the results of the present study highlight the great importance of further untargeted metabolomics studies given the large amount of new knowledge that can be generated.

Conclusion
The physiological response of pomegranate trees after applying DI r treatment confirms the resilience of this species to water stress, by developing adaptation mechanisms such as osmotic adjustment and stomata regulation, without compromising yield.As previously mentioned, this study pursues the validation of the untargeted metabolomics approaches as confinable tool to find new biomarkers of water irrigation stress in the fruit and the results have been shown promises.Despite this, here a tentative identification was developed, due to the exploratory nature of the tool, which is a limitation that has to update to confirmation level in further studies.Therefore, although additional assays will be necessary to corroborate this outcome, DI r irrigation might be a promising way for increasing bioactive compounds content compared to those grown under conventional irrigation conditions.

Fig. 1
Fig. 1 Relative humidity (RH, dotted line), daily mean air temperature (T m , solid line) and daily precipitation (P, vertical bars) during the experimental period

Fig. 2 Fig. 3
Fig.2Stem water potential (Ψ stem ) (A), osmotic leaf potential (Ψ os ) (B), and leaf potential at full turgor (Ψ 100s ), (C) of pomegranate trees irrigated by different irrigation treatments.Different lowercase letters indicate significant differences between treatments according to T student 0.05 test

Fig. 4
Fig. 4 PLS-DA model of full data set.i Red dot: samples under irrigation type 1 condition (Control); ii green dot: sampler under irrigation type 2 condition (DI r )

Fig. 5 Fig. 6
Fig. 5 Heatmap of the metabolites up and down regulates in Control and DI r irrigation group