Andean Blueberry of the Genus Disterigma: A High-Resolution Mass Spectrometric Approach for the Comprehensive Characterization of Phenolic Compounds

Wild neotropical blueberries, endemic of Central and South American areas, are promising yet still undisclosed sources of bioactive compounds. Most research studies have addressed wild and cultivated blueberries from Europe and North America, despite the extremely wide variety of wild neotropical species. In the present paper, for the first time, the phenolic composition of Disterigma alaternoides was investigated through ultra-high-performance liquid chromatography coupled to high-resolution mass-spectrometric analysis followed by accurate data analysis and compound validation with a dedicated structure-based workflow. D. alaternoides, which belongs to a closely related genus to that of the common blueberry, grows exclusively in the Andean regions over 2000 above sea level. Thanks to the dedicated analytical platform, 249 phenolic compounds were tentatively identified, including several anthocyanins, flavonoids, phenolic acids, and proanthocyanidins. Thenature and heterogeneity of identified phenolic compounds demonstrate once more the need for a more profound knowledge of such still uncharted matrices.


Introduction
The vast majority of the research activity on the berries of the plant family Ericaceae has addressed temperate species of Vaccinium [1][2][3], which is only 1 of the 32 genera of the tribe Vacciniae of the family Ericaceae. Among these species, the most known are blueberry (Vaccinium Corymbosum) [4], bilberry (Vaccinium myrtillus) [5], cranberry (Vaccinium macrocarpon) [6], and lingonberry (Vaccinium vitis-idaea) [7]. Nevertheless, more than 600 species of berry-producing Ericaceae are native to the Neotropical realm, including South America, Central America, and the Caribbean islands [8]. Several neotropical blueberries in the Andean region of South America are widely consumed raw or in different preparations [9].
Berries of the genus Vaccinium have been raising interest for their extremely high content in flavonoids, anthocyanins, phenolic acids, and tannins, which have been demonstrated to exert a wide range of biological activities [10,11]. In a recent paper by Rutledge et al. [12], blueberry phenolics were associated with a cognitive enhancement in healthy elder adults. Likewise, Stull et al. [13] reported that consumption of the whole blueberry reduces the blood glucose level in vivo. For these reasons, blueberries are often referred to as "super-fruits". At present, the composition of blueberries from North American and European regions has been widely investigated [4,14,15]. In a recent study by Ancillotti et al. [2], the polyphenol composition of cultivated V. corymbosum and wild V. myrtillus and V. uliginosum were evaluated by liquid chromatography coupled to high-resolution mass

Phenolic Compound Extraction
Freeze-dried berries were extracted as previously reported with slight modifications [22]. Briefly, 0.2 g of freeze-dried berry samples were extracted with 10 mL CH 3 COCH 3 /H 2 O/CH 3 COOH (70:29.5:0.5, v/v/v). The extract was sonicated for 15 min in an ice bath and then centrifuged for 10 min at 2000× g. The supernatant was collected, and the procedure was repeated once. The supernatants were mixed and concentrated to 4.5 mL using a Speed-Vac SC 250 Express (Thermo 164 Avant, Holbrook, NY, USA). Then, 500 µL of MeOH was added to the sample, and the final extract solution (H 2 O/MeOH, 90:10 v/v) was filtered through a 13-mm Acrodisc Syringe filter with a 0.2 µm GH Polypro membrane (Pall, Ann Arbor, MI, USA). Finally, the extract was aliquoted and stored at −20 • C for further analysis.

UHPLC-HRMS Analysis
Phenolic compound chromatographic separation was carried out by a Vanquish binary pump H (Thermo Fisher Scientific, Bremen, Germany), equipped with a thermostated autosampler and column compartment, on a Kinetex core-shell C18 column (100 mm × 2.1 mm i.d.) with a particle size of 2.6 µm (Phenomenex, Torrance, CA, USA) at 40 • C and with a flow-rate of 600 µL min −1 . The injection volume was 10 µL. The mobile phases consisted of H 2 O/HCOOH (99.9:0.1, v/v; phase A) and ACN/HCOOH (99.9:0.1, v/v; phase B). The elution gradient was optimized in a previous study [22]. The chromatographic system was coupled to a Q Exactive hybrid quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific) with a heated ESI source. The ESI source parameters were set as reported in our previous work [23]. The detection was conducted in TOP 5 data-dependent acquisition (DDA) mode for low-and high-molecular-weight phenolic compounds. An exclusion list containing the most intense ions detected in the blank sample, consisting of H 2 O/MeOH (90:10, v/v), was added to the mass-spectrometric method. For low-molecularweight phenolic compound analysis (flavonoids, anthocyanins, and phenolic acids) and high-molecular-weight polyphenol analysis (tannins), MS data were acquired in the range 150-1000 m/z and 300-2000 m/z, respectively, with a resolution (full width at half maximum, FWHM, at m/z 200) of 70,000. In full scan mode, the automatic gain control (AGC) target value was 200,000 and the maximum ion injection time was 100 ms. The isolation window width was 2 m/z. MS/MS fragmentation was performed with a resolution (FWHM, at m/z 200) of 35,000 with AGC target value set at 100,000 and dynamic exclusion set to 3 s. Fragmentation was achieved in the higher-collision dissociation (HCD) cell at three values of normalized collision energy (NCE), namely, 20-50-80 NCE in the positive ion mode and 20-40-60 NCE in the negative ion mode, based on the results of a previous study [24]. All samples were run in triplicate.

Phenolic Compound Identification
Raw data obtained from three consecutive injections and the blank sample were processed by Compound Discoverer 3.1 (Thermo Fisher Scientific) using a customized method specifically dedicated to phenolic compound analysis [24,25]. Customized databases were generated by combining free phenolic compounds (aglycones) with a series of sugars, aliphatic, and aromatic acids, and complete IDs, accurate masses, and molecular formulas were implemented in the mass list feature for the automatic matching of extracted m/z ratios (45,567 combinations). Moreover, detailed HCD fragmentation spectra for flavonoids and phenolic acids were implemented in the compound class scoring section for automatic MS/MS spectra matching. The parameters for the predict composition tool were adapted to phenolic compounds. Extracted m/z from the raw chromatograms were grouped, aligned, and filtered to remove background compounds found in the blank sample, m/z values not associated with compounds present in the databases, and the features lacking MS/MS spectra. Filtered compounds were manually validated by matching fragmentation spectra to those of available standards or spectra reported in the literature. When data were lacking, phenolic compounds were tentatively identified according to the characteristic fragmentation spectra. The identification data for the tentatively identified compounds are discussed in the following sections and summarized in Tables 1-4 and  Tables S1-S4 with the related confidence level according to Schymanski et al. [26].

Results and Discussion
The identification of phenolic compounds is a critical issue in the phytochemical analysis research field. Phenolic compounds are, in fact, a structurally diverse class of compounds, encompassing a wide range of molecular weights, acid-base properties, and structure complexity [27]. Flavonoids, anthocyanins, and phenolic acids are often present as glycoconjugates to sugars or sugar derivatives and/or acylated to aliphatic and aromatic acids. Historically, the analysis of phenolic compounds has been based on UV-Vis spectroscopy due to the extensively aromatic systems in their structures [28]. However, in recent years, untargeted HRMS has become the foremost technique for phenolic compound identification as it allows extending the characterization to a wide range of compounds, also without the need for analytical standards [29]. Because of their extreme complexity, HRMS raw data cannot be handled without using software programs that render accessible the large datasets by m/z extraction, adduct grouping, and feature alignment. Moreover, when highly composite phytocomplexes are analyzed by untargeted HRMS, there is a need for tools that simplify the datasets for a more accessible manual validation of the compounds.
For the characterization of D. alaternoides, untargeted HRMS followed by a suspect screening data processing was employed, based on a methodology that was previously implemented on Compound Discoverer 3.1 by our research group [24]. Because of the wide range of bond energies in phenolic compound structures (from the weak acetal to the strong aromatic bonds), the acquisition was performed with a three-stepped NCE of 20-50-80 and 20-40-60 for the positive and negative ion mode, respectively. To obtain a larger number of chromatographic points per peak, separate chromatographic runs for each polarity were preferred to polarity switching mode. Top 5 DDA mode is widely used for untargeted analysis with orbitrap-based instrumentation as it allows high-quality MS/MS spectra for the five most intense ions for each scan in full-scan mode [30]. Compound Discoverer is based on a system of blocks and nodes that can be customized by the user for the development of specific data-processing methods. For this purpose, an extensive database of 45,567 phenolic compound derivatives was generated by considering flavonoids, phenolic acids, and tannins in their free and conjugated to sugars and acids. The database, which was implemented in the mass list tool, was employed to filter the extracted and aligned features to remove calculated masses not included in the database. Manual validation was also aided by the compound class scoring tool, which matches the experimental MS/MS to theoretical fragmentation of the flavonoid or phenolic acid core. According to this approach, 16 and 233 phenolic compounds were identified and tentatively identified, respectively. The use of DDA mode for data acquisition allowed the annotation of several compounds even though some of them coeluted. In fact, as precursor ions are sequentially isolated and fragmented with an isolation window of 2 m/z, whenever compounds differing from more than 2 Da coeluted, they could still be tentatively identified. In Figure 1, the classes of identified phenolics are reported in terms of the number of identifications and the total peak areas per class. The largest class of compounds was phenolic acids with 108 compounds, followed by flavonoids (87 compounds); proanthocyanidins (36 compounds); and, finally, anthocyanins (18 compounds). Despite being the less numerous, anthocyanins were the most abundant class in terms of total peak area with almost 45% of the total. Flavonoids and phenolic acids were equally distributed (with 28.2% and 25.5%, respectively). Finally, proanthocyanidins comprised only the 1.7% of the total peak area. Identification for each class is discussed in the following sections.
Manual validation was also aided by the compound class scoring tool, which matches the experimental MS/MS to theoretical fragmentation of the flavonoid or phenolic acid core. According to this approach, 16 and 233 phenolic compounds were identified and tentatively identified, respectively. The use of DDA mode for data acquisition allowed the annotation of several compounds even though some of them coeluted. In fact, as precursor ions are sequentially isolated and fragmented with an isolation window of 2 m/z, whenever compounds differing from more than 2 Da coeluted, they could still be tentatively identified. In Figure 1, the classes of identified phenolics are reported in terms of the number of identifications and the total peak areas per class. The largest class of compounds was phenolic acids with 108 compounds, followed by flavonoids (87 compounds); proanthocyanidins (36 compounds); and, finally, anthocyanins (18 compounds). Despite being the less numerous, anthocyanins were the most abundant class in terms of total peak area with almost 45% of the total. Flavonoids and phenolic acids were equally distributed (with 28.2% and 25.5%, respectively). Finally, proanthocyanidins comprised only the 1.7% of the total peak area. Identification for each class is discussed in the following sections.

Anthocyanin and Flavonoid Composition
Anthocyanins are a peculiar class of phenolic compounds synthesized via the phenylpropanoid pathways but differently from the other flavonoids, characterized by a positive charge on the oxygen of the C-ring of the basic flavonoid structure [31].

Anthocyanin and Flavonoid Composition
Anthocyanins are a peculiar class of phenolic compounds synthesized via the phenylpropanoid pathways but differently from the other flavonoids, characterized by a positive charge on the oxygen of the C-ring of the basic flavonoid structure [31]. Because of this positive charge, anthocyanins are commonly determined in positive ion mode in the form of molecular ions [M] + since the corresponding adduct [M-2H] − in negative mode is generated with a noticeably lower sensitivity. In Table 1, the 18 tentatively identified anthocyanins are listed, while in Table S1 further details on the identification are reported (adducts, molecular weights, confirming peaks, and peak areas).
Despite their generally high peak areas (with compound 5 comprising alone almost 40% of the total peak area), the tentatively identified anthocyanins are numerically inferior compared to European and North American blueberries, for which more than 50 anthocyanins were previously reported [2]. Regarding the aglycones, methylated compounds were significantly under-expressed compared to V. myrtillus and V. corymbosum, with only two minor peonidin derivatives identified (compounds 12-13). Malvidin and petunidin derivatives, which are major constituents of blueberries, were not found. It is worth mentioning that malvidin derivatives have not been identified in V. floribundum, an Andean blueberry that grows in the same regions as D. alaternoides [19,20]. In a previous paper by Ma et al. [9], phenolic markers for discriminate North American and Neotropical blueberries were studied, comprising another member of the genus Disterigma (D. rimbachii). Malvidin derivatives were effectively under-expressed in Neotropical blueberries compared to North American ones. The absence of malvidin derivatives was also apparent from the color of the extract, which is significantly more reddish (and less purplish) than those of blueberry and bilberry.
Besides anthocyanins, 87 other flavonoids were tentatively identified in D. alaternoides, mostly flavanol derivatives. Among the several aglycones belonging to this class, quercetin derivatives were the most abundant with more than 97% of the total flavonol peak area, followed by minor amounts of kaempferol, myricetin, and isorhamnetin. The flavonol composition of Andean blueberry was noticeably similar to that of other genus Vaccinium species, except for laricitrin and syringetin derivatives, which were not identified in the D. alaternoides extract [2,32]. Similar to malvidin, laricitrin and syringetin are O-methylated compounds. Their simultaneous absence could indicate a lower degree of methylation in the flavonoid constituents of Andean blueberries compared to European and North American blueberries.
Flavonoids were analyzed in both positive and negative ion modes, with the latter providing generally higher ionization efficiencies. In Table 2, the annotated flavonoids were reported alongside some details, i.e., retention time, proposed formula, experimental m/z, accuracy, main diagnostic product ions, and confidence level in ESI(−), except for compounds 79, 80, and 88, which were uniquely identified in ESI(+). In Table S2, further details were provided, including complete MS/MS spectra in both ion modes. The determination of the position of the glycoconjugation on the flavonol structure is a great analytical challenge when authentic standards are not available. The sugar-aglycone bond can undergo both heterolytic and homolytic cleavage in the negative ion mode, producing an aglycone ion [Y 0 ] − and a radical aglycone ion [Y 0 -H] − , respectively. Differently from the hydroxyl position on the aromatic rings (e.g., position 7 on the A-ring or position 4 on the B-ring), when a sugar is bound to position 3 (on the non-aromatic C-ring of the flavanol structure), the homolytic cleavage is favored [33]. Based on these pieces of evidence, whenever the radical aglycone ion had a higher abundance than the aglycone ion, the compounds were ascribed to 3-O-monosaccharide derivatives. Whenever the aglycone ion was more abundant or in the case of more than one glycosylation, the position was not indicated, as positions 7 and 4 are not distinguishable by HRMS [34]. In agreement with previous findings on other species of the Vaccinium genus, the majority of flavonols were 3-O-glycosylated [2].

Phenolic Acid Composition
To date, phenolic acids in berries from the Ericaceae family have been largely neglected to date compared to flavonoids and flavonoid derivatives, despite their interesting biological activities and high abundance in species of the Vaccinium genus [1]. Anthocyanin-rich matrices are often only analyzed in the positive ion mode [19], which is unsuitable for analyzing strong acid compounds. Moreover, whereas flavonoid structures are somehow consistent in different matrices, phenolic acids encompass a more comprehensive range of compounds, resulting in the need for several analytical standards for targeted analyses. In the case of V. floribundum, which is the most similar berry to D. alaternoides in terms of anthocyanin and flavonoid composition, no more than 7 phenolic acids have been reported so far by previous liquid chromatography coupled to MS analyses [19,35,36]. The reported phenolic acids comprised mainly hydroxycinnamic acids conjugated to quinic and shikimic acid, with the most abundant being chlorogenic acid. In the present paper, a total of 108 phenolic acids and phenolic acid derivatives have been tentatively identified in the Andean blueberry extract by HRMS analysis in the negative ion mode, a number that was significantly higher than reported for blueberries of the genus Vaccinium [2,7,19,32,37]. In Table 3 and Table S3, details of the annotated phenolic acids were reported.
Unlike previous studies on blueberries, the annotated phenolic acids presented a more significant structural variability and could be grouped into six main categories, i.e., arbutin conjugates, quinic and shikimic acid conjugates, hydroxycinnamic acid glycosides, hydroxybenzoic acid glucosides, coumaroyl iridoids, and free phenolic acids. The large number of tentative identifications, with tremendous structural heterogeneity and a high total peak area (more than 25%), implied phenolic acids have a role more important than expected in the composition of D. alaternoides. Figure 2 shows the total peak area for each of the six classes of compounds. Arbutin derivatives were the most abundant compounds (40% of the total peak area), followed by quinic and shikimic acid conjugates (38%). Hydroxycinnamic and hydroxybenzoic glycosides contributed to the total peak area with 10.6% and 7.2%, while coumaroyl iridoids and free phenolic acids were present in minor amounts (ca. 2%). than expected in the composition of D. alaternoides. Figure 2 shows the total peak area for each of the six classes of compounds. Arbutin derivatives were the most abundant compounds (40% of the total peak area), followed by quinic and shikimic acid conjugates (38%). Hydroxycinnamic and hydroxybenzoic glycosides contributed to the total peak area with 10.6% and 7.2%, while coumaroyl iridoids and free phenolic acids were present in minor amounts (ca. 2%).  Arbutin conjugates presented extremely high concentrations despite being numerically a minor class (10 out of 108 compounds). As a matter of fact, caffeoyl arbutin (compound 165) was the single most abundant phenolic acid in terms of peak area. Arbutin is a glucoside of hydroquinone primarily found in blueberry leaves [38] and is known for its skin-whitening properties [39] as well as its efficacy in the treatment of various urinary tract infections [40]. Caffeoyl arbutin was identified by a prior loss of 110 Da (hydroquinone). This cleavage generates a dehydration on the sugar moiety due to the extremely strong C-O phenolic bond of hydroquinone. Therefore, the subsequent sugar cleavage led to a loss of 144 Da (glucose-2H 2 O) rather than the usual neutral loss of 162 Da (glucose-H 2 O). In a previous paper by Ieri et al. [41], several arbutin derivatives were identified in a bilberry leaf extract, including caffeoyl and coumaroyl arbutin, as well as their acetyl derivatives, in good agreement with our findings. Other than caffeoyl arbutin (compounds 163 and 165), other derivatives were identified, i.e., caffeoyl hexosyl arbutin (three isomers, compounds 153, 159, and 160), caffeoyl acetyl arbutin (two isomers, compounds 182 and 183), caffeoyl methoxyarbutin (compound 169), and hydroxybenzoyl arbutin (two isomers, compounds 161 and 180), which were identified with the same logic as described for caffeoyl arbutin. Considering how neglected phenolic acids generally are, arbutin derivatives are likely to be present in all blueberry fruits rather than be solely present in D. alaternoides.
The identified phenolic acid conjugates were mainly hydrophobic hydroxycinnamic derivatives (caffeoyl; coumaroyl; and, to a lesser extent, feruloyl, and sinapoyl conjugates) rather than hydrophile hydroxybenzoic derivatives. Gallic acid and its polymeric derivatives (gallotannins and ellagitannins) were scarcely represented, while several minor glycoconjugates of benzoic, hydroxybenzoic, and dihydroxybenzoic acid have been tentatively identified. Despite a large number of identified compounds (60 compounds), phenolic acid glycoconjugates represented just 18% of the total peak area, likely due to a large number of minor positional isomers. Among the other minor constituents, two coumaroyl iridoids (compounds 166-167) were tentatively identified; these compounds, which are characteristic of cranberry (V. macrocarpon), are of great interest for their possible role in healing urinary tract infections [42].

Proanthocyanidin Composition
Proanthocyanidins are non-hydrolyzable oligomers of flavanols, mainly (epi)catechin and (epi)gallocatechin, and are distinguished into two subclasses according to their linkage. A-type proanthocyanidins present a double linkage from positions 7 and 8 on the ring A of the terminal unit to positions 2 and 4 on the ring C of the extension unit (2β→O→7; 4β→8), while B-type proanthocyanidins present a single interflavanoid bond (4β→8). For simplicity, species with one or more A-type interflavanoid bonds are commonly defined as A-type proanthocyanidins [43]. The standard nomenclature is perfectly suitable for dimers, as there is either one A-type bond or one B-type bond. Nevertheless, it is worth specifying that, concerning oligomers with more than two units, there are more than two possibilities, e.g., in the case of trimers, two B-type bonds, two A-type bonds, and one for each kind, with the latter two both falling under the definition of A-type proanthocyanidins. For sake of clarity, whenever more than one A-type bond was present in the oligomer, a prefix was added to the name (compound 246). In Table 4 and Table S4, detailed data of the 36 tentatively identified proanthocyanidins were reported. Despite being efficiently ionized in both positive and negative ion mode, proanthocyanidins have been only analyzed in negative polarity for the generally higher ionization efficiency, the higher clarity of the MS/MS spectra, and the minor interference of contaminants and noise. Proanthocyanidin fragmentation pathways involve quinone methide (QM) fissions, retro-Diels-Alder (RDA) ring openings, and heterocyclic ring fissions (HRF). QM fissions generate two distinctive sections of the oligomer, named terminal and extension unit (or β unit). Compared to other proanthocyanidin-rich matrices, such as tea [44], strawberry [22], and even bilberry [2], the identified compounds were primarily A-and B-type procyanidins, oligomers of the sole catechin and epicatechin. A-type procyanidins were more abundant than B-type ones both in terms of the number of identifications (20 vs. 16) and the total peak area (74% vs. 26%), in good agreement with previously found for bilberry [2].

Conclusions
Wild neotropical berries are still an undisclosed rich source of bioactive compounds. In the present paper, almost 250 phenolic compound derivatives were tentatively identified, including anthocyanins, flavonoids, phenolic acids, and proanthocyanidins, in Disterigma alaternoides, an Andean blueberry of the unfamiliar genus Disterigma of the family Ericaceae. These results indicated, once more, the need for more profound and capillary knowledge on these exotic berry species. Many of the identified compounds are indeed known to exert important biological activities. The high number of tentative identifications was achieved thanks to a dedicated analytical platform based on HRMS and data analysis that allowed a comprehensive yet accessible phenol characterization. The annotated compounds were in general agreement with the composition of other blueberry, with high anthocyanin and flavanol glycoconjugated. The anthocyanin and flavonoid pattern, however, was more similar to that of other Andean blueberries, such as V. floribundum, with the absence of highly methylated malvidin, petunidin, and syringetin derivatives. Phenolic acids, which are a generally less investigated class of phenolic compounds, were instead the most numerous and heterogenous class. Several phenolic acids that conjugated to arbutin, which is present in the leaves of blueberry plants, were reported for the first time in blueberry fruit extracts. The extremely rich composition of D. alaternoides represents a pivotal result in the field of neotropical berries, which are emerging as possible "superfruits" for the biological activities of their compounds.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/separations8050058/s1, Table S1: detailed identification data for the annotated anthocyanins in Disterigma alaternoides; Table S2: detailed identification data for the annotated flavonoids in Disterigma alaternoides; Table S3: detailed identification data for the annotated phenolic acids in Disterigma alaternoides; Table S4: detailed identification data for the annotated proanthocyanidins in Disterigma alaternoides.