Screening for Innovative Sources of Carotenoids and Phenolic Antioxidants among Flowers

Flowers have been used for centuries in decoration and traditional medicine, and as components of dishes. In this study, carotenoids and phenolics from 125 flowers were determined by liquid chromatography (RRLC and UHPLC). After comparing four different extractants, the carotenoids were extracted with acetone: methanol (2:1), which led to a recovery of 83%. The phenolic compounds were extracted with 0.1% acidified methanol. The petals of the edible flowers Renealmia alpinia and Lantana camara showed the highest values of theoretical vitamin A activity expressed as retinol activity equivalents (RAE), i.e., 19.1 and 4.1 RAE/g fresh weight, respectively. The sample with the highest total phenolic contents was Punica granatum orange (146.7 mg/g dry weight). It was concluded that in most cases, flowers with high carotenoid contents did not contain high phenolic content and vice versa. The results of this study can help to develop innovative concepts and products for the industry.


Introduction
Flowers have long held an important place in human societies. They have been used for ornamental purposes as well as in diverse dishes, mainly due to their appealing and diverse colors [1]. In addition, flowers have been used in traditional medicine [2]. More specifically, the use of flowers in the diet or as medicine dates back at least to 4000 BC, as documented in the Mesopotamic and Egyptian cultures [3]. Their traditional use in other cultures (Roman, Greek, Chinese, Indian, and European) is also well-known [4].
In recent years, there has been a growing interest in the study from different points of view of the health-promoting secondary metabolites present in flowers, including carotenoids and phenolics [5][6][7]. Indeed, the study of agronomic practices that can enhance the levels of these compounds in flowers or non-conventional technologies for their extraction are timely topics [8,9]. Carotenoids (carotenes and xanthophylls) are widespread and versatile compounds in nature, where they are important in processes including photosynthesis, the communication within and between species, the protection against oxidizing agents, and the modulation of membrane properties [10]. They are responsible for the red, yellow and orange colors of many flowers [11], which are important for pollination [12]. One of the main differences between carotenoids relative and other bioactive compounds is that some of them can be converted into vitamin A, which is an essential micronutrient.

Color Analysis
The colors were measured using a CM-700d colorimeter (Minollta, Japan). Illuminant D65 and 10 • observer were considered as references. The color parameters corresponding to the uniform color space CIELAB were obtained. The categorization of the samples by color (white, yellow, orange, red, pink, lilac and blue) was performed considering clusters of points in the a*b* plane. Thus, the samples were separated into three groups. Group A included white, yellow, and orange flowers, group B contained red and pink flowers, and group C included lilac and blue flowers. The color of some flowers could not be assessed instrumentally because of their small sizes.

Analysis of Carotenoids 2.4.1. Extraction and Saponification
The micro-extractions were performed under dim light and in triplicate. The best extraction mixture was selected after evaluating different extraction mixtures (hexane: acetone (v/v) (1:1), methanol: acetone: dichloromethane (v/v/v) (1:1:2), acetone: methanol (v/v) (2:1), and ethyl acetate: methanol: petroleum ether (v/v/v) (1:1:1)). For this purpose, the petals of Calendula × hybrid were used. Approximately 20 mg of homogenized freeze-dried powder was mixed with 1 mL of the appropriate solvent mixture and then vortexed, sonicated for 2 min and centrifuged at 14,000× g for 3 min. After recovering the colored fraction, the extraction was repeated with aliquots of 500 µL of the solvent mixture until color exhaustion. The organic colored fractions were combined and evaporated to dryness in a vacuum concentrator at a temperature below 30 • C. Calendula × hybrid is known to possess high amounts of esterified carotenoids, so the extracts were de-esterified by saponification [28]. For this purpose, the dry extracts were re-dissolved in 500 µL of methanolic potassium hydroxide (30%, w/v) and the mixtures were stirred for one hour in a nitrogen atmosphere at 25 • C. Next, 500 µL of dichloromethane and 800 µL of 5% aqueous NaCl (w/v) were added. The samples were vortexed and centrifuged at 14,000× g for 3 min and then the aqueous phase was removed. The carotenoid-containing phase was washed with water until neutrality of the wastewater. The colored phase was evaporated to dryness in a vacuum concentrator at a temperature below 30 • C and stored in a nitrogen atmosphere at −20 • C until the analysis.
The extraction mixture leading to the highest recovery of carotenoids was selected for the extraction of carotenoids from all the samples. All-trans-β-apo-8 -carotenal was used as an internal standard.

Spectrophotometric Analysis
The total carotenoid contents (TCC) of petroleum ether extracts of each flower were quantified by spectrophotometry by considering the absorbance reading at 450 nm and the molar absorptivity value of β-carotene in the solvent (εmol = 2592). The results were reported as µg/g dry weight (DW) [29].

Rapid Resolution Liquid Chromatography (RRLC) Analysis
The dry extracts were re-dissolved in 20 µL of ethyl acetate prior to their analysis by RRLC. The analysis was carried out using the method reported by [30] on an Agilent 1260 system equipped with a diode-array detector and a C18 Poroshell 120 column (2.7 µm, 5 cm × 4.6 mm) (Agilent, Palo Alto, CA, USA). The injection volume was 5 µL, the flow rate was 1 mL/min, and the temperature of the column was set at 30 • C. A mobile phase consisting of acetonitrile, methanol, and ethyl acetate was used with a linear gradient elution [30]. The chromatograms were monitored at 285, 350, and 450 nm for the quantification of phytoene, phytofluene, and the rest of the carotenoids (lutein epoxide, luteoxanthin, antheraxanthin, violaxanthin, lutein, cis-antheraxanthin, lycopene, zeinoxanthin, β-cryptoxanthin, β-carotene, and α-carotene), respectively. UV-Vis spectra were recorded from 250 to 750 nm. The individual carotenoids were identified with their corresponding standards and quantified using external calibration curves made with them whenever possible. The limits of detection (LOD) and quantification (LOQ) were calculated as three and ten times, respectively; the relative standard deviation of the analytical blank values were calculated from the calibration curve, using Microcal Origin ver. 3.5 software (OriginLab Corporation, Northampton, MA, USA). The LODs and LOQs ranged from 0.002 µg in phytoene to 0.070 µg in lycopene and from 0.007 µg in phytoene to 0.232 µg in lycopene, respectively. The LOD and LOQ were established on the basis of signal to noise (S/N) ratio of 3 and 10, respectively. The samples were analyzed in duplicate with double sample injection. The concentrations were expressed in µg/g DW and the TCC contents were calculated by adding up all the individual carotenoids.
2.5. Analysis of Phenolic Compounds 2.5.1. Extraction The protocol described by [31] was adapted for the extraction of smaller amounts of samples. Briefly, 1.5 mL of 0.1% acidified methanol was added to approximately 50 mg of freeze-dried petals, and the mixture was vortexed, sonicated for 2 min, and centrifuged at 4190× g for 7 min and at 4 • C; the supernatant was collected and the residue was submitted to the same extraction process twice with only 0.5 mL of the acidified methanol. The combined supernatant was stored at −20 • C until the analysis.

Spectrophotometric Analysis
The extract obtained was used for the determination of the total phenolic content (TPC) using the Folin-Ciocalteu assay, as described by [31], with slight modifications. Briefly, 50 µL of extract, 0.25 mL of Folin-Ciocalteu reagent, 0.75 mL of a solution of sodium carbonate (20%), and 3.95 mL of distilled water were mixed and left to stand for 2 h for the reaction to take place. Gallic acid was employed as a calibration standard and the absorbance was read at 765 nm with a Hewlett-Packard UV-vis HP8453 spectrophotometer (Palo Alto, CA, USA). The results were expressed as mg of equivalents of gallic acid per g of dry weight (mg GAE/ g DW) and allowed to define the injection volumes for the quantification by Ultra-High Performance Liquid Chromatography (UHPLC).

Ultra-High Performance Liquid Chromatography (UHPLC) Analysis
Prior to the injection, the extracts were concentrated to dryness, re-dissolved in 20 µL of 0.01% formic acid, and centrifuged at 4190× g for 7 min and at 4 • C. The UHPLC method was previously reported by [31]. An Agilent 1290 chromatograph equipped with a diode-array detector (Agilent Technologies, Palo Alto, CA, USA) set between 220 and 500 nm and an Eclipse Plus C18 column (1.8 um, 2.1 × 5 mm) were used. The column was kept at 30 • C, the injection volumes were in a range between 0.3 and 1.5 µL, the flow rate was 1 mL/min, and a linear gradient was used. Open lab ChemStation software was used for data acquisition and processing. The identification of the phenolic compounds was performed through a comparison of their retention times and UV-vis spectra, within the range 250-750 nm, with those of the available standards [31]. The chromatograms were monitored at 280 for the benzoic acids, hydroxycinnamic acids, flavones, and flavanones, and at 320 nm for the flavonols. Their quantification was carried out using external calibration curves of each of the compounds analyzed. The LODs and LOQs ranged from 0.006 µg in chlorogenic acid to 0.012 µg in p-hydroxybenzoic acid and 0.014 µg in chlorogenic acid to 0.041 µg in p-hydroxybenzoic acid, respectively. The LOD and LOQ were established on the basis of signal-to-noise (S/N) ratios of 3 and 10, respectively. The samples were analyzed in duplicate with double sample injection. The TPC was calculated by adding up all the individual phenolics.

Statistical Analysis
All the experiments were performed in triplicate with double injection, and the results were expressed as mean ± standard deviation (SD). The mean separation was made via Tukey's test. Differences were considered statistically significant for p values ≤0.01. The statistical analysis was performed using the STATGRAPHICS Centurion XVII software.

Color Parameters and Other Characteristics
The color parameters, humidity values, and culinary uses of the flowers are presented in Tables 1-3.

Selection of the Extraction Solvents
Four different extraction solvents were tested for the extraction of carotenoids in Calendula × hybrid (Figure 1). Acetone: methanol (v/v) (2:1) and ethyl acetate: methanol: petroleum ether (v/v/v) (1:1:1) showed the highest carotenoid extraction yields and there was no statistically significant difference between the two mixtures. The recovery of carotenoids obtained with this mixture, using all-trans-β-apo-8 -carotenal as internal standard, was 83%.

Selection of the Extraction Solvents
Four different extraction solvents were tested for the extraction of carotenoids in Calendula × hybrid (Figure 1). Acetone: methanol (v/v) (2:1) and ethyl acetate: methanol: petroleum ether (v/v/v) (1:1:1) showed the highest carotenoid extraction yields and there was no statistically significant difference between the two mixtures. The recovery of carotenoids obtained with this mixture, using all-trans-β-apo-8′-carotenal as internal standard, was 83%. In addition, quantitative data on individuals and TCC, assessed by liquid chromatography, are presented in Tables 4-6. An example of the resulting chromatogram is presented in Figure 2, and the frequency, mean contents, and standard deviations of carotenoids and major sources are presented in Figure 3, sections A, B, and C. In addition, quantitative data on individuals and TCC, assessed by liquid chromatography, are presented in Tables 4-6. An example of the resulting chromatogram is presented in Figure 2, and the frequency, mean contents, and standard deviations of carotenoids and major sources are presented in Figure 3, sections A, B, and C.

Phenolic Compounds
The quantitative data on individuals and TPC assessed by chromatographic analysis are presented in Tables 7-9. In addition, an example of the resulting chromatogram is presented in Figures 4 and 5 sections A, B, and C show the frequency, mean contents, and standard deviations of the phenolics and major sources.

Phenolic Compounds
The quantitative data on individuals and TPC assessed by chromatographic analysis are presented in Tables 7-9. In addition, an example of the resulting chromatogram is presented in Figures 4 and 5 sections A, B, and C show the frequency, mean contents, and standard deviations of the phenolics and major sources.

Color Parameters and Other Characteristics
The great majority of the flowers were edible (n = 111, i.e., 89%); 70% of the families studied (52 families) included edible flowers. For example, the families Asteraceae and Lamiaceae contained six and seven edible species, respectively [20]. Concerning their uses, the most frequent were in salads (31.3% of the total use of the flowers) and infusions (28.9%), followed by teas (15.7%), desserts (13.3%) and others, including as garnishes and colorants (Tables 1-3). The different culinary uses of flowers depend to some extent on their size, shape, and color, as suggested by other authors [4]. These characteristics varied considerably among the samples surveyed in the present study. Different shapes were found, such as tubular (e.g., Russelia equisetiformis Schltdl. Et Cham.), bilabial (e.g., Rosmarinus officinalis L.), flared (e.g., Punica granatum L.), and flowers that form part of a cluster (e.g., Plantago major L., Salvia splendens Sellow ex Schylt., Vitex agnus-castus L., Allium schoenoprasum L., and Lantana camara L.). On the other hand, the flowers showed a great variety of colors (Tables 1-3), such as white (e.g., Portulaca oleracea L.), yellow (e.g., Anthemis tinctoria L.), orange (e.g., Punica granatum L.), pink (e.g., Diantuhus caryophyllus L.), red (e.g., Pelargonium × hortorum), lilac (e.g., Petunia hybrid), and blue (e.g., Lavandula angustifolia Mill.). The color parameters ranged between 14.2 and 87.1, −9.2 and 57.2, −25.7 and 88.6, 2.9 and 89.9, and 3.4 and 359.6 for L* (lightness), a* (ranging from green to red), b* (ranging from blue to yellow), C* ab (chroma, the quantitative expression of color), and h ab (hue angle, the qualitative expression of color), respectively. The variety of colors found in the petals of flowers under study can be explained by the different contents of carotenoids and phenolics, which are usually the main contributors to the color of these structures [34,35].
The humidity of the petals ranged between 54.5 and 99.7%, a wider interval compared to that recently reported by other authors (70 and 95%) [4].

Selection of the Extraction Solvents
Regarding the quantification of carotenoids in flowers, there are several studies that use different extraction solvents; however, the mixtures acetone: methanol (2:1) and ethyl acetate: methanol: petroleum ether (1:1:1) in this study presented the highest extraction percentage. Acetone: methanol (2:1) was selected as the extraction solvent for the studied flowers due to its slightly higher yield and its simplicity of preparation.

Carotenoid Levels
At this point it is important to notice that saponification, which simplifies the identification of carotenoids, has the disadvantage that it leads to carotenoid losses [30], so the information provided must be interpreted with this in mind. This fact has been observed in the TCC levels of red and lilac flowers of Catharanthus roseus (3.7 µg/g DW and not detectable, respectively) and Pelargonium × hortorum (3.5 µg/g DW and not detectable, respectively). Although the TCC levels measured in non-saponified extracts by spectrophotometry showed values of 185, 132, and 100 µg/g DW, respectively, no individual carotenoids were detected by RRLC after the saponification of the extracts (data not shown).
On the other hand, flowers of the same family but different species presented different profiles in most cases. At this point it is important to notice that the profiles of the secondary metabolites of plants in general and carotenoids and phenolics in particular are dependent on different factors, including genotype as one of the most important, along with ambient/seasonal (light quality and quantity, temperature), and agronomic factors (irrigation, fertilization, etc.), among others [36][37][38].
Britton and Khachik proposed a criterion through which to classify food sources according to their carotenoid content expressed in mg/100 g fresh weight. According to this criterion, the contents of a specific carotenoid can be classified as low (0-0.1 mg/100 g), moderate (0.1-0.5 mg/100 g), high (0.5-2 mg/100 g), or very high (>2 mg/100 g).
The petals of the edible flowers Renealmia alpinia (15.0 mg/100 g FW of β-carotene and 15.8 mg/100 g FW of α-carotene) and Lantana camara (0.6 mg/100 g FW of β-carotene and 8.6 mg/100 g FW of α-carotene) showed the highest values of provitamin A carotenoids. The values of vitamin A activity of the samples can be expressed in terms of retinol activity equivalents (RAE), considering the equivalences 1 RAE = 12 µg of all-transβ-carotene = 24 µg of other provitamin A carotenoids [39]. Thus, the RAE per gram of fresh weight of Renealmia alpinia and Lantana camara are 19.1 and 4.1, respectively. Given that 1 RAE equals two retinol equivalent (RE), it can be estimated that 10 g of fresh flowers from Renealmia alpinia would provide 381.2 retinol equivalents (RE), which is approximately half the daily recommendation of vitamin A for adults (750 RE/day) by FAO and OMS [40].

Phenolic Compounds
The most frequent phenolic compounds in the set of flowers evaluated were mcoumaric acid (a phenolic acid), quercitrin, and quercetin (flavonoids) ( Figure 5, section A), which agrees well with the information reported by other authors indicating that phenolic acids and flavonoids are the predominant phenolic compounds in flowers [41]. On the other hand, values between 4.83 and 222.00 mg GAE/g DW of total phenolics have been described in 23 edible flowers elsewhere [4].
Flowers of the same species with different colors showed different profiles of phenolics, as opposed to what was observed in the case of carotenoids. Flowers of different species also exhibited different phenolic patterns. This may have been due to the fact that, as already mentioned, the contents of phenolic compounds and other secondary metabolites in plants are dependent on genetic factors, as well as climatic and agronomic conditions, among others [37,38].
In addition, the influence of different methods on the extraction efficiency of different compounds (in this case not only phenolics but also carotenoids), and therefore on their, quantification must be taken into account.

Benzoic Acids
Gallic acid displayed ranges between 0.1 and 38.1 mg/g DW. Pelargonium × hortorum red (38.1 mg/g DW), pink (24.5 mg/g DW), lilac (27.7 mg/g DW), and Pelargonium domesticum lilac (18.6 mg/g DW) were the samples with the highest contents. The content of p-Hydroxybenzoic acid ranged from 0.1 to 21.1 mg/g DW. The highest values were found in Plumbago auriculata white (21.1 mg/g DW), Chlorophytum comosum white (8.4 mg/g DW), Dahlia coccinea yellow (6.9 mg/g DW), and Vitex agnus castus lilac (6.5 mg/g DW). The m-coumaric acid values showed ranges between 0.04 and 19.5 mg/g DW. Verbena × hybrid pink (19.5 mg/g DW), Dianthus caryophyllus red (15.5 mg/g DW), Vitex agnus-castus lilac (15.5 mg/g DW), and Hydrangea petiolaris pink (12.3 mg/g DW) exhibited the most significant m-coumaric acid concentrations. The p-coumaric acid values oscillated between 0.1 and 17.6 mg/g DW. Catharanthus roseus red (17.6 mg/g DW) and Punica granatum orange (10.1 mg/g DW) were the samples surveyed with the highest values of p-coumaric acid. The levels of vanillic acid fell in an interval of 0.1-1.6 mg/g DW. This compound was detected only in a few species, such as Dianthus caryophyllus red (1.6 mg/g DW), Vitex agnus castus lilac (0.6 mg/g DW), Nerium oleander pink (0.5 mg/g DW), Celosia argentea red (0.1 mg/g DW), and Aglaonema commutatum yellow (0.1 mg/g DW). The Syringic acid totals 19.06 mg/100g DW, and 190.8 mg GAE/100 g DW for gallic acid, quercetin, and total phenolic, respectively, in Gardenia jasminoides [46]. Furthermore, other authors reported that Mirabilis jalapa is a good source of flavonoids and phenolic acids (ferulic acid and caffeic acid as major compounds) [47]. The aforementioned flowers, despite being used as coloring agents, did not stand out for their TPC in the present study.
Some petals are indeed highly concentrated sources of carotenoids, including provitamin A carotenoids. As an example, it has been estimated that 10 g fresh weight of the petals of Renealmia alpinia can provide 381.2 ER, which is approximately half the daily recommendation of vitamin A for adults (750 ER/day).
The samples with the highest TPC (assessed by liquid chromatography) were Punica granatum orange, Pelargonium × hortorum red and pink, and Hydrangea petiolaris pink and Plumbago auriculata white). The TPC in Punica granatum was approximately 15 times higher than the mean.
In summary, several petal matrices with interesting carotenoid or phenolic profiles (either by their total content or their levels of specific carotenoids or phenolics) were pinpointed. The information provided can help to design breeding programs aimed at producing flowers with increased carotenoid and/or phenolic levels and can be useful for the provision of natural colors for the agro-food or textile industries, as well as for the provision of beneficial compounds for the functional foods, nutricosmetics, and pharmaceutical industries.

Data Availability Statement:
The datasets generated for this study are available on request to the corresponding author.