Acylated anthocyanins from organic purple-fleshed sweet potato (Ipomoea batatas (L.) Lam) produced in Brazil

Acylated anthocyanins from a purple-fleshed sweet potato (PFSP), obtained by organic cultivation in Brazil, were characterized after separation by a high performance liquid chromatography-diode array detector (HPLC-PDA). These anthocyanins were manually collected at the detector output, concentrated and injected into a high resolution mass spectrometer (ESIQTOF-MS2). Twenty-two acylated anthocyanins were detected. Among them, sixteen had been reported in the literature and six, derived from peonidin were reported for the first time in sweet potato roots in this study. These compounds showed molecular ions with accurate mass/charge ratios (m/z) of 909.2081, 961.3010, 961.2571, 963.3345, 1123.2932 and 1179.3862. Although anthocyanins in PFSP have already been extensively studied, the variety studied in this work is probably genetically different from all varieties and cultivars already researched, which would explain why these anthocyanins have not been observed in the previously studied varieties.


Introduction
Sweet potato, the tuberous root of Ipomoea batatas L., native to the Andes (Shan et al., 2012) is a nutritionally valuable food with high antioxidant activity (Kim et al., 2011). It is a high-yielding industrially important crop used in hunger relief in Asia and Africa, with a range of different colors and amounts of carotenoids, anthocyanins and phenolic acids (Zhao et al., 2014;Wang et al., 2016).
Consumption of anthocyanins is associated with reduced risk of cardiovascular and degenerative metabolic diseases, visual and brain function improvement, cancer chemoprevention, anti-inflammatory, hepatoprotective and antihyperglycemic activities even when bioavailability is low (Hou, 2003;Oliveira et al., 2019c;Smeriglio et al., 2016;Tsuda, 2012). Since acylated anthocyanins have higher processing stability, bioprotective capacity and resistance to overall simulated digestions than non-acylated anthocyanins they are desirable for their provision of color and health benefits Oliveira et al., 2019b;Yang et al., 2019).
The purple-fleshed sweet potato (PFSP) has a high content of acylated anthocyanins that provide greater color stability when compared to other vegetables as in the case of red fruits which have a high content of anthocyanins with low levels of acylation. Studies have shown a wide-ranging physiological functionality for these anthocyanins, as the majority have cyanidin or peonidin, or are mono-or di-acylated with at least one caffeic acid, which confers greater thermal and ultraviolet resistance, and antioxidant capacity. They degrade only partially when subjected to heating (Kim et al., 2015) and have been used in industrialized foodstuffs (Wang et al., 2012) such as juices, alcoholic beverages, pasta, flour, breads and others as natural dye and antioxidant. It may be recommended in healthy food for the prevention of chronic diseases related to certain lifestyles.
A PFSP has been cultivated for many years by family farmers in the organic production system in the state of Rio de Janeiro, Brazil. However, the anthocyanin profile of this plant has never been studied under these conditions. The objective of this study was to characterize the profile of the anthocyanins present in this PFSP variety and compare it with the profile of the varieties already studied in the works available in the literature. For this purpose the PFSP was cultivated in different places and seasons, for the same total period of cultivation, and the anthocyanins were separated by HPLC-DAD and characterized by ESI-QTOF-MS 2 .

Chemicals
The acetonitrile, methanol and formic acid used were HPLC grade, the water was ultrapure, and 0.5 g C 18 reverse phase cartridges and C 18 60A carbon 17 % particle size 40-63 µm for solid phase extraction (SPE) were used.

Plant material
The branches of a PFSP variety with purple peel, of unknown origin, commonly cultivated in Brazil by organic farming, were grown for the same total cultivation period of five months ( seasons and years (from 8 Nov 2017 to 3 Apr 2018) at an integrated agroecological production system, Seropédica-RJ (geographic coordinates: 22°45'13.18" S, 43°40'25.25" W, altitude of 36 m). The PFSP branches (approximately 0.3 m in length) were linearly distributed, separated by 0.3 m at the center of three parallel rows 1 m wide by 5 m long, separated by 1 m intervals. Only on the first day was fertilization implemented with 200 g m -2 fermented organic compound based of wheat bran (60 %) and castor bran (40 %) inoculated with compost accelerator, 22 g m -2 thermophosphate and 4 kg m -2 potassium sulfate. A drip irrigation system was also used. After five months the tuberous roots were harvested, peeled, cut, crushed, freeze dried, ground and frozen until use.

Sample preparation
The samples of PFSP tuberous root previously prepared and frozen until use as described above (30.0 g of each farm), after reaching room temperature (24 °C) were individually extracted for 30 min under stirring at 40 °C, with 300 mL of ultrapure water containing 0.068 mol L -1 formic acid, followed by filtration of the supernatant. The procedure was repeated three times until no more intense pink color was observed in the solvent. The total volume (900 mL) of each extract obtained was concentrated by SPE using manually prepared 10 g reverse phase C 18 60A cartridges.
The manually prepared cartridges were preconditioned with 100 mL of methanol, equilibrated with 100 mL of ultrapure water, loaded with the extract and washed with 100 mL of ultrapure water. The retained anthocyanins were eluted with 30 mL of acetonitrile with 5 % formic acid, collected and reserved for HPLC separation.

Separation and purification of anthocyanins
The solutions obtained in the previous step were injected into an HPLC, consisting of a separations module equipped with a photodiode array detector (PDA), acquiring chromatograms recorded from 200 to 600 nm. The column used was C18 (100 × 4.6 mm, particle size 2.4 µm). The column oven temperature was 35 °C, and the injection volume 30 µL with a mobile phase flow rate of 1.0 mL min -1 . The elution gradient was applied according to Lee et al. (2013) with certain modifications. It started with 90 % of ultrapure water with 5 % of formic acid (A) and 10 % of acetonitrile (B) decreasing to 86 % A in 12 min, reaching 80 % A in 3 min, and remaining at that level for 3 min, then rising further to 90 % A in 2 min, and remaining at that level for 2 min. Fourteen peaks with maximum absorption in the range of 518-532 nm and 308-337 nm were detected with the same chromatogram profile for both extracts. They were manually collected at the detector output. Twenty injections were required to ensure a sufficient amount was obtained for the characterization analyses.
The separated anthocyanins in the collected peaks were concentrated by solid phase extraction (SPE) in 0.5 g SEP-PAK C 18 reverse phase cartridges, conditioned with 1 mL of acetonitrile, equilibrated with 1 mL of ultrapure water, loaded with the collected peak diluted two times with ultrapure water, washed with 1 mL of ultrapure water and then eluted with 3 mL of acetonitrile containing 5 % formic acid. The samples were dried under air flow, solubilized in 500 µL of acetonitrile and analyzed by ESI-QTOF-MS 2 .

Identification of anthocyanins
The solutions obtained in the previous step were directly injected into the ESI-QTOF-MS 2 with ESI in positive mode and QTOF under the following conditions: capillary, sampling cone and extraction cone energies of 3.0 kV, 50.0 V and 3.0 V, respectively; source temperature of 80 °C, desolvation gas temperature and flow of 250 °C and 500 L h -1 and gas flow of 25 L h -1 (nitrogen was used as the nebulizing and drying gas). Depending on the analysis, collision energies used in the trap and transfer ranged from 25 to 35 V. The MS 2 spectral data, maximum absorption wavelength region, elution order and other studies that detailed the characteristic fragmentation of the PFSP anthocyanins were used to identify them (Goda et al., 1997;Grass et al., 2017;He et al., 2016;Hu et al., 2016;Islam et al., 2002;Jie et al., 2013;Lee et al., 2013;Montilla et al., 2010;Odake et al., 1992;Qiu et al., 2009;Terahara et al., 1999;Terahara et al., 2000;Terahara et al., 2004;Tian et al., 2005;Truong et al., 2010;Truong et al., 2012;Wang et al., 2017;Xu et al., 2015;Ying et al., 2011;Zhao et al., 2014;Zhu et al., 2017).

HPLC-DAD profile of PFSP
Using the described chromatographic conditions, fourteen peaks were separated ( Figure 1) which exhibited typical acylated anthocyanin spectral profiles showing maximum absorbance of approximately 525 nm, characteristic of anthocyanins, and approximately 320 nm, characteristic of acylation with hydroxycinnamic acid derivatives (Giusti and Wrolstad, 2003;Hang and Wrolstad, 1990), for every peak detected in both extracts (from different locations of farms and in a different season and year of cultivation) that have qualitatively identical chromatograms. Genetic, biochemical, physiological, ecological and evolutionary factors may both quantitatively and qualitatively influence the production of secondary metabolites such as anthocyanins, but the most important is the genetic factor (Endt et al., 2002;Figueiredo et al., 2008;Pichersky and Gang, 2000). As both crops used the same genetic material, the other different factors were not enough to alter the gene expression of anthocyanin biosynthesis. The band intensity ratio in the range of 308-337 nm with a band intensity of approximately 518-532 nm (E acyl.max / E vis.max ) (Giusti and Wrolstad, 2003;Hang and Wrolstad, 1990) at the spectrum of each peak showed, in a number of cases, higher ratios that characterize hydroxycinnamic derived di-acylated compounds and lower ratios that characterize hydroxycinnamic derived mono-acylation (Lee et al., 2013;Odake et al., 1992;Terahara et al., 2004). The E acyl.max /E vis.max ratios were low for peaks 1, 2, 3, 4, and high for peaks 5, 6, 7, 8, 9, 10, 11, 12, 13,14. The spectra had no interference from free phenolic acids since the procedure used in this work for anthocyanin extraction was more selective than those used in other studies. In this study, the solvent used to extract anthocyanins was 0.068 mol L -1 formic acid in water, which did not require phenolic acid elimination treatments and facilitated the extraction of anthocyanins compared to the solvents used in other studies, since the phenolic acids in PFSP roots are bound to other molecules and the acid concentration in the extraction solvent does not promote hydrolysis. Organic solvents (methanol, acetonitrile or ethanol) or their mixtures with acidic water, commonly used in other studies to extract anthocyanins from fruits, had difficulty in extracting anthocyanins from these PFSP samples. Due to their high affinity for starch, anthocyanins were preferably adsorbed on starch and did not migrate to the organic solvents tested. That is one reason for starch being a good wall material for microencapsulation of anthocyanins (Fang and Bhandari, 2010). The 0.068 mol L -1 formic acid in water solvent, more polar than organic solvents or mixtures of these solvents with acid water, was more efficient in extracting PFSP anthocyanins as it competed with starch for anthocyanin affinity. The difficulty in extracting anthocyanins from PFSP, which has high starch contents, through the use of organic solvents in general does not occur in the extraction of anthocyanins from fruits that usually have much lower starch contents. The interfering polar compounds that may have been present in the extract were eliminated by the washing step during SPE as Lee et al. (2013) in their anthocyanin purification procedure.

Identification of isolated anthocyanins by ESI-QTOF-MS 2
The results present three main possibilities: the loss of glycosyl bound to position 5 of anthocyanidin, the loss of sophorosyl in position 3 (higher probability) (Tian et al., 2005), and the loss of both, as in Figure 2. The fragmentation of glycosidic bond within the sophorose and between the aromatic acids and sophoroside was negligible (Tian et al., 2005) The loss of acylated sophorosyl, which is the preferred reaction, explains the higher intensity of the resulting fragment peak compared to the peak of glycosyl loss. The glycosyls and acyl moieties in anthocyanins were assumed to occur in positions already confirmed in other studies ( Figure 2) through mass spectrometry and nuclear magnetic resonance (NMR) (Goda et al., 1997;Montilla et al., 2010;Odake et al., 1992;Qiu et al., 2009;Terahara et al., 1999;Terahara et al., 2000;Terahara et al., 2004;Ying et al., 2011;Zhang et al., 2018). The data obtained for the anthocyanins detected are presented in an organized listing in Table 1.  An anthocyanin was observed at peak one, M + m/z 907.2524, which corresponds to M + =[C 41 H 47 O 23 ] + . Although that peak was not completely separate from peak 2 and it was not possible to differentiate them by their UV/Vis spectra only (Table 1), it was possible to tell them apart by selecting that precursor ion (m/z 907) through MS 2 . Its UV/Vis spectra was typical of mono-acylation with both maximum absorption bands of the anthocyanin and the acid derivative very intense. It was the most polar anthocyanin in the extract since it was the first in elution order and due to the presence of the fragment m/z 301. and the fragment m/z 463.1253 was more intense than the last, resulting from the loss of sophorosyl bonded to acyl moiety. When the acylation occurs in the carbon at the 6" position of the sophorosyl (Figure 2), as in this case, the intensity of the [M-acyl-sophorosyl] + peak is much lower than the M + and the [anthocyanidin] + peaks, although it is always more intense than the peak of the [M-glycosyl] + fragment, showing that the reaction of the loss of the sophorosyl occurs in a greater proportion than the loss of glycosyl as evidenced by Lee et al. (2013). When the acylation occurs in the carbon at the 6"' position of the sophorosyl (Figure 2 (He et al., 2016;Hu et al., 2016;Islam et al., 2002;Jie et al., 2013;Lee et al., 2013;Montilla et al., 2010;Tian et al., 2005;Truong et al., 2010;Truong et al., 2012;Wang et al., 2017;Xu et al., 2015;Zhu et al., 2017). At peak two, an unidentified anthocyanin was detected for the first time in PFSP, with M + m/z 1179.3862 (MS 2 precursor ion m/z 1179) and fragments m/z 1119. 3651, 973.3016, 841.2552, 737.2061, 605.1586, 505.1393 and 301.0712 [Peonidin] + . Two compounds were observed at peak three with typical UV/Vis spectrum of mono-acylates with hydroxycinnamic acid anthocyanins (Table 1) (He et al., 2016;Hu et al., 2016;Islam et al., 2002;Jie et al., 2013;Lee et al., 2013;Montilla et al., 2010;Tian et al., 2005;Truong et al., 2010;Truong et al., 2012;Wang et al., 2017;Xu et al., 2015;Zhu et al., 2017), with the ferulic acylation at the 6" position since the intensity of m/z 449 [M-feruloyl-sophorosyl] + is only higher than the m/z 787.2089 (Lee et al., 2013).
At peaks four and six an anthocyanin was detected for each peak, both with the similar M + and fragments (MS 2 Figure  3B). Both peonidin derivative isomers were detected for the first time in PFSP.
At peak eleven, an anthocyanin was observed with the coumaroyl and hydroxybenzoyl UV/Vis bands and a low intensity band of anthocyanin that characterized di-acylation (Table 1) Wang et al. (2017).
At peak twelve, an anthocyanin was observed with intense ferulic and p-hydroxybenzoyl acids UV/ Vis bands (  Truong et al. (2010), identified by Wang et al. (2017) and recently by Zhu et al. (2017).
In certain cases it was not possible to indicate the positions of a number of radicals on carbons 6" or 6"' in the sophorosyl of the anthocyanins (Figure 2) because no structural NMR analyses were performed and this information was not presented in the available literature.
It was possible to verify in the studied PFSP that the majority of the anthocyanins are the peonidin derivatives peonidin 3-O-(6"-O-feruloyl sophoroside)-O-5-glucoside followed peonidin 3-O-(6-O-feruloyl-6-O-caffeoyl sophoroside)-5-O-glucoside and peonidin 3-O-(6"-p-hydroxybenzoyl sophoroside)-5-O-glucoside. The anthocyanins first detected in this work should be investigated by other structural identification techniques that allow their molecular structures to be elucidated. The absence of non-acylated anthocyanins was noted, unlike any previously studied PFSP variety. The first reported anthocyanins in sweet potatoes in this study are