New Phenolic Compounds in Posidonia oceanica Seagrass: A Comprehensive Array Using High Resolution Mass Spectrometry

The studies on the Posidonia oceanica Delile (P. oceanica) phenolic composition have been focused on the foliar tissues and have often neglected the phenolic compounds in rhizomes or roots alike. With the current improvements in high resolution mass spectrometry (HRMS) analyzers, such as the Orbitrap MS, there is a new opportunity to more deeply study P. oceanica. One of the benefits is the possibility of conducting an exhaustive phenolic monitoring, which is crucial in the search for new stressor-specific biomarkers of coastal deterioration. For this purpose, the different tissues (leaf, rhizome, and root) of P. oceanica seagrass from several marine sampling areas were analyzed through target, suspected, and non-target screenings. This paper brings a fast and tissues-specific extraction, as well as a detection method of phenolic compounds applying for the first time the potential of HRMS (Exactive Orbitrap) in P. oceanica samples. As a result, 42 phenolic compounds were satisfactorily detected, of which, to our knowledge, 24 were not previously reported in P. oceanica, such as naringenin, naringenin chalcone and pinocembrin, among others. Information here reported could be used for the evaluation of new stressor-specific biomarkers of coastal deterioration in the Mediterranean waters. Furthermore, the followed extraction and analytical method could be considered as a reference protocol in other studies on marine seagrasses due to the exhaustive search and satisfactory results.


Introduction
Posidonia oceanica Delile (P. oceanica) is a long-living, slow-growing marine angiosperm (seagrass) endemic to the Mediterranean Sea. The P. oceanica meadows have been identified as priority habitats by European conservation legislation [1]. These meadows supply essential ecosystem services for human well-being [2]. However, P. oceanica has been observed to be especially susceptible to anthropogenic disturbances [3][4][5]. Studies have detected the presence and bioaccumulation of pollutants in this seagrass, such as heavy metals. Ref. [6] brought insight into their application as contamination bioindicators. It has also been reported that P. oceanica responds to contamination-driven stress as well as to natural-driven stress (grazing, competition, microbial settlement, and eutrophication among others) by altering the amount of phenolic compounds [7,8]. Due to their specific and rapid response when changes in the environment occur, phenolic compounds have been proved to be suitable biomarkers or early warnings [9][10][11][12]. These compounds are specialized metabolites that benefit the bearing organism through several biological properties (e.g., antioxidant capacity) [13]. Interest in this field has been growing, given the potential medical contributions of such phenolic compounds to human health and personal care: immunostimulant and antitumoral drugs or products for skin aging [14,15]. P. oceanica phenolic content has been previously reported [14,16]. Of the researched compounds, ferulic acid, caffeic acid, chicoric acid, caftaric acid, and cinnamic acid have been identified as major phenolic compounds [17,18]. Studies commonly focused on the determination of phenolic compounds in foliar tissues (temporal structures marked by seasonal variations), and little attention has been paid to the tissues with a longer lifespan less marked by seasonal variations, i.e., rhizomes and roots. To the best of our knowledge, only one study reported the phenolic fingerprint in P. oceanica roots [19], and available studies based on rhizomes have only reported total phenols (following mainly the Folin-Ciocalteu assay) or lignin content [18,[20][21][22]. However, several studies on terrestrial plants have observed specific phenols in these tissues reacting to specific changes in the environment or taking part in relevant activities, such as the reduction of contaminants in soil or their role in plant resistance to deficit conditions [23][24][25].
Additionally, studies based their identification/detection of the phenolic compounds on liquid (LC), gas chromatography (GC), or pyrolysis GC (Py-GC) coupled to low resolution mass spectrometry (LRMS) [15,18,26]. Nowadays, the improvements in high resolution mass spectrometry (HRMS) and its ability to detect a higher number of compounds, bring a new opportunity to obtain a comprehensive array of phenolic compounds in P. oceanica. The implementation of HRMS analyzers could also lead to (1) a better differentiation between genuine signals and artifacts driven signals, (2) accurate detection of target and non-target compounds, and (3) retrospective analysis of samples.
Hence, the objective of this study is double fold. Firstly, the determination of phenolic compounds in P. oceanica for the first time, to our knowledge, by ultra-high performance liquid chromatography combined with HRMS, concretely with an Orbitrap analyzer (UHPLC-Orbitrap MS), using target, suspected, and non-target analysis. Secondly, the evaluation of the phenolic content in the different P. oceanica tissues (leaf, rhizome, and root) from different sampling points. The findings of this study provide new reference information in the field of marine seagrass chemistry that could be useful in future investigations of P. oceanica phenolic compounds' suitability as stress-specific biomarkers.

Extraction Procedure
Two different procedures were compared in the tissues of one of the collected P. oceanica samples (FAN7 leaves, rhizome, and roots) for extractant composition evaluation [17,27]. Procedures varied between a mixture of methanol/water 8:2 v/v and 5:5 v/v (pH 4 in both cases). After ca. 15 min extraction, samples were subject to a target screening using an in-house database. This database used 93 commercially available standards of the most common phenolic compounds detected in agri-food products [28,29]. In this preliminary screening, up to 20 compounds were identified and considered for the extractant composition evaluation. The results showed that the optimal composition was tissue-dependent (Table S1). For instance, for rhizomes and roots, an extractant mixture of 8:2 v/v methanol/water was proved to be the most adequate in terms of phenolic profile sensitivity in all cases, i.e., number of compounds, higher peak intensities (described as NL, Normalization Level), better peak shape, or a greater presence of confirmation fragments. In contrast, leaves showed overall relatively better results when using the 5:5 v/v mixture (~72% of the cases). An example of how the intensity varied according to the employed extractant composition and the studied tissue is shown in Figure 1. Observed differences in extraction efficiencies could be attributed to the type of plant material (e.g., ligneous and rigid in the case of the rhizome and roots) and the compounds affinity to the different tissues. The results obtained for leaves confirmed that the extraction followed in previous studies (5:5 v/v mixture) yield relatively better results than the 8:2 v/v mixture [15][16][17]30]. Nonetheless, the scarcity of data on roots and rhizomes hindered the comparison between our results and other published procedures. No available information on rhizomes was found and, after a thorough search of the relevant literature, only one article was found regarding the extraction of the roots, in which non-diluted acetone was used in a three day extraction procedure [19]. In all observed articles, extraction times ranged from several hours to days or consumed elevated solvent-to-plant. In contrast, the extraction method employed in this study proved to be less time consuming and employed a reduced solvent volume, following the green chemistry approach.
Therefore, based on the assessment of the results, the extractant composition selected was a mixture of methanol/water 5:5 v/v for leaves, and 8:2 v/v in the case of rhizome and roots which allowed for the detection of a greater number of compounds (up to four) for each tissue.

Chromatographic Conditions
Two sub-2-μm columns, widely employed in the phenolic compound's studies, were also tested in the analysis of FAN7 leaves. A target screening was performed using the inhouse database to evaluate the chromatographic separation performance. Such columns were Acquity C18 column (2.1 mm × 100 mm, 1.7 μm particle size; Waters, Milford, MA, USA) and Hypersil GOLD TM (2.1 mm × 100 mm, 1.9 μm particle size; Thermo Fisher, San Jose, CA, USA).
Besides, two common aqueous mobile phases were also evaluated, 30 mM ammonium acetate aqueous solution pH 5 [27] and 4 mM formic acid aqueous solution pH 3 [17]. Methanol was employed as the organic eluent along the process.
Regardless of the employed columns, compound separation remained the same. On the other hand, elution times differed, showing an increase in RT when using the Acquity C18 column, which could be attributed to the more reduced particle size. However, overall, relatively higher intensities or NL were reached when using Hypersil GOLD TM , as shown in Figure S1.
As for the mobile phases and their respective pH, they have been seen to affect the peaks' intensity and the presence or absence of compounds and their confirmation fragments. In general, intensities were lower when ammonium acetate was employed, which in some cases hindered the apparition of confirmation fragments. Average better signals were observed in the formic acid mobile phase with pH 3, confirming effective binding of the phenolic compounds to the stationary phase. Up to six compounds more were observed when using formic acid. Examples are caffeic acid, p-coumaric acid or ferulic acid, phenolic compounds that were only confirmed when using formic acid. As an example, the case of ferulic acid is shown in Figure S2, which was absent when using ammonium. The results obtained for leaves confirmed that the extraction followed in previous studies (5:5 v/v mixture) yield relatively better results than the 8:2 v/v mixture [15][16][17]30]. Nonetheless, the scarcity of data on roots and rhizomes hindered the comparison between our results and other published procedures. No available information on rhizomes was found and, after a thorough search of the relevant literature, only one article was found regarding the extraction of the roots, in which non-diluted acetone was used in a three day extraction procedure [19]. In all observed articles, extraction times ranged from several hours to days or consumed elevated solvent-to-plant. In contrast, the extraction method employed in this study proved to be less time consuming and employed a reduced solvent volume, following the green chemistry approach.
Therefore, based on the assessment of the results, the extractant composition selected was a mixture of methanol/water 5:5 v/v for leaves, and 8:2 v/v in the case of rhizome and roots which allowed for the detection of a greater number of compounds (up to four) for each tissue.

Chromatographic Conditions
Two sub-2-µm columns, widely employed in the phenolic compound's studies, were also tested in the analysis of FAN7 leaves. A target screening was performed using the in-house database to evaluate the chromatographic separation performance. Such columns were Acquity C18 column (2.1 mm × 100 mm, 1.7 µm particle size; Waters, Milford, MA, USA) and Hypersil GOLD TM (2.1 mm × 100 mm, 1.9 µm particle size; Thermo Fisher, San Jose, CA, USA).
Besides, two common aqueous mobile phases were also evaluated, 30 mM ammonium acetate aqueous solution pH 5 [27] and 4 mM formic acid aqueous solution pH 3 [17]. Methanol was employed as the organic eluent along the process.
Regardless of the employed columns, compound separation remained the same. On the other hand, elution times differed, showing an increase in R T when using the Acquity C18 column, which could be attributed to the more reduced particle size. However, overall, relatively higher intensities or NL were reached when using Hypersil GOLD TM , as shown in Figure S1.
As for the mobile phases and their respective pH, they have been seen to affect the peaks' intensity and the presence or absence of compounds and their confirmation fragments. In general, intensities were lower when ammonium acetate was employed, which in some cases hindered the apparition of confirmation fragments. Average better signals were observed in the formic acid mobile phase with pH 3, confirming effective binding of the phenolic compounds to the stationary phase. Up to six compounds more were observed when using formic acid. Examples are caffeic acid, p-coumaric acid or ferulic acid, phenolic compounds that were only confirmed when using formic acid. As an example, the case of ferulic acid is shown in Figure S2, which was absent when using ammonium. Such absence might be due to the charge of ferulic acid in increased pH, hampering the binding to the column.
As a result, the selected conditions for this analysis were: Hypersil GOLD TM column, and formic acid as the aqueous mobile phase.

Phenolic Identification
After evaluating the extraction conditions, target, suspected, and non-target screenings of all tissues (leaves, rhizome, and roots) of five P. oceanica seagrasses were conducted as indicated in Figure S3. During the target screening, compound peaks of each molecule were located using information from the in-house database, such as the parent theoretical mass and confirmation/characteristic fragments, and performing pseudo MS/MS experiments. The risk of false-positive was also reduced by monitoring each parent's peak mass spectrum and comparing it to the theoretical molecule spectrum (or simulation) to confirm the ion ratios for the isotopic pattern [28].
Results of the target screening revealed 22 phenolic compounds from which 13 were detected for the first time to our knowledge in P. oceanica (Table 1). These newly reported specialized metabolites consisted of flavones (apigenin, baicalein, and luteolin), flavonols (galangin), isoflavones (biochanin A, genistein, and glycitein), flavonons/flavanones (eriodictyol, naringenin, pinocembrin, and sakuranetin/isosakuranetin), chalcones (naringenin chalcone), and kaempferol-3-O-glucoside/luteolin-4'-O-glucoside (IUPAC names can be found in Table S2). These phenolic compounds have already been proved to play essential roles inhibiting cancer or offering antioxidant properties [15,31,32]. More biological properties are summarized in Table S3, Supplementary Materials. Table 1. Retention time (R T ) and m/z ions for the identification and confirmation of the detected target compounds in P. oceanica roots (•), rhizomes ( ), and/or leaf (♦). Mass error corresponds to the smallest average mass error found among the three tissues. Note that some of the detected phenolic compounds were isomers, with the same exact mass, R T , and fragments. Since they could not be separated by chromatographic or mass resolution and the presence of both isomers could not be confirmed, these were recorded and counted as only one phenolic compound [28]. For the suspected screening, a second list with 38 compounds was developed by gathering published information on the P. oceanica phenolic fingerprint and other Mediterranean seagrasses (namely Cymodocea nodosa, Zostera marina, and Zostera noltii). Out of the 30 considered suspected compounds previously detected in P. oceanica, nine were tentatively identified in the samples and included in Table 2. On the other hand, the not detected suspected compounds are shown in Table S4. Note that the tentative identification of a compound was reached when (1) the difference between the exact mass of the candidate (calculated from the elemental formula) and the exact mass of the target compound felt within ±5 ppm of mass error, (2) confirmation fragments were present and, (3) spectrum matched among experimental and theoretical peaks in terms of ion ratios for the isotopic pattern ( Table 2). Figure 2 shows a chromatogram and mass spectrum of isorhamnetin-3-Omalonylglucoside as an example. Suspected compounds identified in other Mediterranean seagrasses, such as apigenin-7-O-glucoside and rosmarinic acid or the sulfated flavonoids apigenin-7-sulfate, diosmetin-7-sulfate, and luteolin 7-sulfate, were not detected in this study in agreement with the literature [32][33][34].  During the non-target screening, the Compound Discoverer Software (Thermo Fisher Scientific, Les Ulis, France) was employed to identify potential candidates. In this search, the software subtracts all exact masses and the respective molecular formulas. Finally, these formulas were compared to information on phenolic compounds gathered in open databases, searching for matches. Only those highlighted as a full match and meeting the criteria settled for a tentative identification were considered. The non-target analysis revealed 11 compounds not previously identified in P. oceanica, such as sophoraflavanone B and two curcuminoids (Tables 3 and S2). In addition, through the non-target screening several compounds detected during the target and suspected analysis were further confirmed: chicoric acid, p-coumaric acid, ferulic acid, fertaric acid, bioachin A, genistein, naringenin, isorhamnetin, isorhamnetin-3-glucoside, and quercetin-3-O-glucoside. The fact that certain compounds were only observed in the non-target screening whilst others were only detectable during the target and suspected mode highlights the need to combine the three searching modes and the benefits of the retrospective analyses for environmental samples, as recently suggested [36]. During the non-target screening, the Compound Discoverer Software (Thermo Fisher Scientific, Les Ulis, France) was employed to identify potential candidates. In this search, the software subtracts all exact masses and the respective molecular formulas. Finally, these formulas were compared to information on phenolic compounds gathered in open databases, searching for matches. Only those highlighted as a full match and meeting the criteria settled for a tentative identification were considered. The non-target analysis revealed 11 compounds not previously identified in P. oceanica, such as sophoraflavanone B and two curcuminoids ( Table 3 and Table S2). In addition, through the non-target screening several compounds detected during the target and suspected analysis were further confirmed: chicoric acid, p-coumaric acid, ferulic acid, fertaric acid, bioachin A, genistein, naringenin, isorhamnetin, isorhamnetin-3-glucoside, and quercetin-3-O-glucoside. The fact that certain compounds were only observed in the non-target screening whilst others were only detectable during the target and suspected mode highlights the need to combine the three searching modes and the benefits of the retrospective analyses for environmental samples, as recently suggested [36]. Table 3. Average retention times (R T ) and m/z ions of the non-target compounds in the Orbitrap system for P. oceanica. Compounds detected in P. oceanica roots (•) rhizomes ( ), and/or leaves (♦). Interestingly, among the first-time detected compounds, biochanin A, glycitein, gambiriin A1, and tracheloside were only observed in the underground tissues ( Figure 3). To our knowledge it is also the first time that these compounds have been reported in any marine seagrass, it being more common to find them in other samples such as legumes and algae for biochanin A and glycitein, respectively [37,38]. Interestingly, among the first-time detected compounds, biochanin A, glycitein, gambiriin A1, and tracheloside were only observed in the underground tissues ( Figure 3). To our knowledge it is also the first time that these compounds have been reported in any marine seagrass, it being more common to find them in other samples such as legumes and algae for biochanin A and glycitein, respectively [37,38]. Considering all P. oceanica samples and tissues, a total of 42 phenolic compounds were detected (Table S5). Among them, flavonoids and phenolic acids represented the main groups, as shown in Table 4. These two families were further investigated, revealing the cinnamic acid subclass responsible for the uneven distribution among tissues. Information on the average number of compounds per tissue is shown in Table S6. The rest of the phenolic subclasses show a relatively similar number of phenolic compounds in leaf, rhizome, and root. This difference in P. oceanica leaves could be ascribed to a greater exposition to UV radiation, current motion, shifts in water temperature, epiphytes load, or turbidity, among others [5,39]. For instance, marine water is relatively more unstable than sediment in terms of environmental conditions.

Sampling
In this study, five samples of P. oceanica (EE3, AL2x, AL3, FAN7, and CG4) were taken from different sites along the coast of Almeria (Alboran Sea, Spain), as can be observed in Figure 4. Samples were collected manually by scuba divers in July, October, November, December 2019, and January 2020. The sampling sites were distributed at different water depths, from 1.5 m at CG4 to 9 m at EE3. More relevant information concerning the characteristics of the sampling points can be observed in Table 5.

Sample Pre-Treatment
After collection, samples were immediately stored in portable fridges at a low temperature until arrival at the laboratory. Subsequently, sand and salt were removed by

Sample Pre-Treatment
After collection, samples were immediately stored in portable fridges at a low temperature until arrival at the laboratory. Subsequently, sand and salt were removed by rinsing with distilled water and samples were divided into three parts: leaf, rhizome, and root. All parts were kept in petri plates and weighted. Note that young leaves and basal sheathes were not considered in this study.
Afterward, samples were stored at −20 • C (48 h) and freeze-dried at −50 • C (48 h) in a Thermo Electron Corporation Heto PowerDry LL3000 freeze-dryer (Thermo Fisher Scientific, Bremen, Germany). In the case of the leaves, this part was conducted in two separate steps of 24 h. In between steps, leaves were gently cleaned from epiphytes (dried crust) using a brush. Subsequently, all samples were homogenized by powdering in a Mixer Mill MM 200 (2 min at 25 r/s) and stored in desiccators until extraction.

Chemical and Reagents
LC/MS-grade water and methanol were purchased from Merck KGaA (Darmstadt, DE) and Riedel-de-Haën™ (Seelze, Germany), respectively. Formic acid was purchased from Fisher Scientific (Waltham, MA, USA). Galangin standard was obtained from Extrasynthese (Genay, France). P-coumaric acid, kaempferol-3-O-glucoside, quercetin standards and ammonium formate were purchased from Sigma-Aldrich (St Louis, MA, USA). All reagents were of analytical grade. Employed standards had a purity of >99%. Individual standard solutions of 300 mg l −1 were prepared in methanol. Resulting stock standards solutions were kept in amber bottles. Subsequently, a multi-compound working solution (50 mg l −1 ) was prepared by diluting each stock solution aliquot with methanol. All solutions were stored at 4 • C in an amber bottle until analysis.

Extraction Procedure
Specific extractions were developed regarding the investigated tissues from previous procedures with minor modifications [17,27]. Briefly, 3 mL of methanol/water 5:5 v/v solution (pH 4, acidified with formic acid) were added to 15 mL falcon tubes with 150 mg dry weight of leaf powder, whereas 3 mL of methanol/water 8:2 v/v solution was used in the case of rhizome and roots. All different mixtures were then sonicated for 4 min at room temperature (~25 • C) and centrifuged (5000 rpm, 10 min). One single extraction cycle was carried out since preliminary studies proved it to be adequate to monitor the phenolic compounds [27]. After ca. 15 min extraction the resulting supernatant was filtered (45 µm, Fisher Scientific, Madrid, Spain) into 2 mL LC vials and analyzed.

Chromatographic Conditions
Chromatographic analyses of P. oceanica tissues were conducted on a Thermo Scientific Transcend TM 600 liquid chromatography (Thermo Fisher Scientific, San Jose, CA, USA). The employed column was Hypersil GOLD TM . The chromatographic separation was performed using a mobile phase that comprises water (1% formic acid and 4 mM ammonium formate) as eluent A and methanol as eluent B.
The column temperature during analysis was maintained at room temperature (25 • C), the flow rate was settled at 0.2 mL min −1 and the injection volume at 10 µL along the process.

Orbitrap-MS Analysis
In this study, a single Orbitrap mass spectrometer (Exactive™, Thermo Fisher Scientific, Bremen, Germany) was used for MS analyses. The mass spectra were acquired employing four alternating acquisition functions: full MS, without fragmentation (higher collisional dissociation, HCD, collision cell was switched off), mass resolving power 25,000 FWHM (full width at half maximum); scan time 0.25 s, ESI+, and ESI−; all-ion fragmentation (AIF), with fragmentation (HCD on, collision energy 30 eV), mass resolving power 10,000 FWHM, ESI+, and ESI−. Mass range in full scan mode was set at m/z 100-1000, whereas, for MS/MS monitoring, it was set at m/z 70-700.
Data acquisition and processing were carried out using Trace Finder Version TM 4.0 and Xcalibur TM Version 2.2.1 (Thermo Fisher Scientific, Les Ulis, France) in Qual browser mode. Software searching criteria was set on 5 ppm mass tolerance error.

Phenolic Identification
As mentioned above, the identification of compounds was performed in three steps. Firstly, a target screening was carried out, submitting an in-house database containing 93 phenolic compounds to the software to elucidate the presence or absence of listed compounds. Further important information for the compound identification, such as molecular formula, retention time (R T ), exact theoretical mass of the molecular ion and characteristic fragments, and ionization mode, were also included in the in-house database.
Secondly, a suspected screening was conducted. In this case, previously detected phenolic compounds in P. oceanica, as well as in other Mediterranean seagrasses, were considered. This list was developed based on available data (literature and open databases such as PubChem, ChemSpider, and Human Metabolomics), adding to the study 30 suspected compounds detected in P. oceanica and eight only detected in the other Mediterranean seagrasses (Table 3 and Table S4). The exact mass of these suspected compounds was calculated using their chemical formula and the corresponding ionization mode (ESI−/ESI+). Confirmation fragments were also retrieved from the available literature and open databases. Note that several of the found suspected compounds were already present in our in-house database and were therefore treated as target compounds.
In third place, a non-target analysis was performed using Compound Discoverer Software (Thermo Fisher Scientific, Les Ulis, France). For that purpose, the raw data obtained from the sample and blank injections were submitted to the software. The software carried out a structural analysis, performing a metabolite profiling base on several selected databases (CheBI, KEGG, MolBank, Nature Chemistry, Sigma Aldrich, and Phenol Explor-

Conclusions
The analysis of the three P. oceanica tissues using, for the first time, UHPLC-Orbitrap MS in target analysis mode in combination with retrospective analyses (suspected and non-target modes) contributed to the detection of 42 phenolic compounds. Out of these, 24 have not been previously reported in this marine angiosperm. The vast majority of the detected compounds belong to the flavonoid family, highlighting the presence of flavonols and flavanones, although some phenolic acids such as cinnamic and benzoic acids were also found.
Additionally, the distribution, in terms of number of total detected phenolic compounds, was relatively higher in leaves followed by rhizome and in last place roots. It seems that the cinnamic acids subclass is responsible for such a difference (caffeic acid, caftaric acid, chicoric acid, p-coumaric acid, fertaric acid, ferulic acid, and zosteric acid). However, only through the detailed research of the three tissues the complete P. oceanica phenolic fingerprint was unveiled, since several compounds were merely observed in underground tissues (e.g., bioachin A, glycitein, gambrin A1, and tacharoside).
This study brings new reference information in the field of bioactive compounds in marine seagrasses, namely a comprehensive array of phenolic compounds in P. oceanica tissues. In addition, the number of flavonoids, compounds widely employed as biomarkers, present in P. oceanica has been increased. Perhaps these first-time observed flavonoids could bring lacking information in the field of seagrasses chemical reactions.
The 42 detected compounds that make up the phenolic fingerprint of P. oceanica should be considered in future research to observe their behavior under different conditions and further evaluate their potential as specific-biochemical markers.