Roselle calyx, decoction, and by-product

Dried roselle (Hibiscus sabdariffa L.) calyxes were obtained from local producers from Guerrero, Mexico. Calyxes were disinfected with 1 % Nobac citrus 373 solution for 10 min by immersion, drained, and then dried at 45 ºC for 48 h in a forced circulation oven (BF 400, Binder, Tuttlingen, Germany). Subsequently, 100 g of dried and disinfected roselle calyxes were subjected to a decoction process with 1 L of boiling water for 15 min as reported by ^{Amaya-Cruz et al. (2019)}. The used calyces (by-product) were separated from the liquid (decoction) by decantation. The roselle by-product was dried as previously described, then ground to a size particle < 420 µm. The roselle calyx, decoction, and by-product were stored at -20 ºC until they were further analyzed. Roselle decoction was thawed at 4 ºC for 24 h and subsequently brought to room temperature before undergoing chemical analyses, while roselle calyx and by-product were directly analyzed. The decoction process was performed in triplicate. Within each decoction replicate, the by-product and the calyx were both analyzed in triplicate for each assay.

Polyphenols extraction from roselle calyx and by-product

The roselle decoction was directly utilized to determine the polyphenol composition. The roselle calyx and by-product were subjected to a sequential polyphenol extraction with organic solvents as reported by ^{Hassan et al. (2011)}. Briefly, 500 mg of each sample were extracted with 5 mL of 50:50 (v/v) methanol:water, adjusted to pH 2 using hydrochloric acid. The extraction was performed at room temperature for 60 min under continuous stirring. Then, samples were centrifuged (1500 x g for 10 min). Supernatants were recovered and the residue was further extracted with 5 mL of 70:30 (v/v) acetone:water as previously described. The combination of these supernatants formed the extractable polyphenol (EPP) fraction, utilized to quantify extractable polyphenols, flavonoids, anthocyanins, and proanthocyanidins. The residues were dried at 45 º C for 25 h, forming the non-extractable polyphenol (NEPP) fraction utilized to quantify hydrolysable polyphenols (HPP) and non-extractable proanthocyanidins (NEPA).

Total and extractable phenolic compounds

The polyphenol content was determined as reported by ^{Singleton et al. (1999)} with minor modifications. Briefly, 10 µL of each sample were mixed with 25 µL of 1 N Folin Ciocalteu, 125 µL of a 20 % (w/v) aqueous solution of sodium carbonate, and 40 µL of distilled water. The reactions were incubated for 30 min under darkness. Then, absorbances were measured at 765 nm. The results were expressed as mg of gallic acid equivalents/mL of roselle decoction or mg of gallic acid equivalents/g of roselle calyx or by-product.

Total and extractable flavonoid content

The flavonoid content was determined as reported by ^{Oomah et al. (2005)} with minor modifications. Briefly, 100 µL of each sample were mixed with 20 µL of a 1 mg/mL methanolic solution of 2-aminoehtyldiphenylborate, and 130 µL of distilled water. Then, absorbances were immediately measured at 404 nm. The results were expressed as mg of rutin equivalents/mL of roselle decoction or mg of rutin equivalents/g of roselle calyx or by-product.

Total and extractable anthocyanin content

The anthocyanin content was determined as described by ^{Giusti and Wrolstad (2001)} with minor modifications. Briefly, 50 µL of each sample were mixed with 175 µL of each buffer: 0.25 M potassium chloride at pH 1 or 0.40 M sodium acetate at pH 4.5. Then, absorbances were immediately measured at 510 and 700 nm. The results were expressed as mg of cyanidin 3-O-glycoside equivalents/mL of roselle decoction or mg of cyanidin 3-O-glycoside equivalents/g of roselle calyx or by-product.

Extractable proanthocyanidins

The extractable proanthocyanidins (EPA) content was determined as reported by ^{Zurita et al. (2012)}. Briefly, 500 µL of each sample were mixed with 4.5 mL of a 95:5 (v/v) buthanol:hydrochloric acid solution containing 0.07 g/L of iron chloride. The reactions were incubated in a boiling water bath for 1 h. Then, samples were centrifuged at 1500 x g for 10 min and supernatants were recovered. Absorbances were measured at 450 and 550 nm. The results were expressed as mg of proanthocyanidin equivalents/mL of roselle decoction or mg of proanthocyanidin equivalents/g of roselle calyx or by-product.

Hydrolysable polyphenols content

A sequential alkaline and acid hydrolysis was carried out as reported by ^{Quatrin et al. (2019)}. The NEPP residues were incubated with 5 mL of 10 M sodium hydroxide and 12 mL of distilled water under continuous stirring for 16 h. Then, their pH was adjusted to 2 - 3 with 6 M hydrochloric acid. Samples were centrifuged at 2000 x g for 10 min and supernatants were recovered. Afterwards, the residues were re-extracted with 5 mL of distilled water as previously described. The supernatants were mixed to quantify polyphenols as described. Results were expressed as mg of gallic acid equivalents/mL of roselle decoction or mg of gallic acid equivalents/g of roselle calyx or by-product, reflecting the alkaline HPP content. Then, the residue obtained during the alkaline hydrolyses was subjected to acid hydrolysis with 2.5 mL of hydrochloric acid. Samples were incubated at 85 ºC for 30 min. Then, their pH was adjusted to 2 - 3 with 10 M sodium hydroxide. Afterwards, the residues were re-extracted with 5 mL of distilled water as previously described. The supernatants were mixed to quantify polyphenols as described. The results were expressed as mg of gallic acid equivalents/mL of roselle decoction or mg of gallic acid equivalents/g of roselle calyx or by-product, reflecting the acid HPP content.

Non-extractable proanthocyanidins content

The NEPA content was determined as reported by ^{Zurita et al. (2012)} in the NEPP residue as previously described. The results were expressed as mg of proanthocyanidin equivalents/mL of roselle decoction or mg of proanthocyanidin equivalents/g of roselle calyx or by-product.

Polyphenol profile by UPLC-ESI-QToF MS

The roselle decoction was filtered using a syringe filter with PTFE membrane (13 mm, 0.2 µm). The EPP and HPP extracts obtained from the roselle calyx and by-product were vacuum-dried using a Speedvac system (Savant, Thermo Fisher Scientific, MA, USA), re-suspended in 200 µL of mobile phase, and subsequently filtered. The polyphenol profile was assessed in an Ultra-Performance Liquid Chromatograph coupled to a high-resolution Quadrupole/Time-of-Flight Mass Spectrometer (UPLC-QToF MSE, Vion, Waters Co., MA, USA). The autosampler was maintained at 4 ºC. The chromatographic separation was carried out with 1 µL of each sample injected into a C18 column (2.1 x 100 mm, 1.7 µm, BEH Acquity, Waters Co.) at 35 ºC. The mobile phase was constituted by (a) water with 1 % formic acid and (B) acetonitrile with 1 % formic acid using the gradient conditions throughout 17 min as reported by ^{Reynoso-Camacho et al. (2021)}. An atmospheric pressure electrospray ionization (ESI) source was used at negative and positive mode at 120 ºC using nitrogen as desolvation gas (800 L/h at 450 ºC) and as cone gas (50 L/h). The mass spectra were acquired at 100-1200 m/z at low (6 V) and high (15-45 V) collision energy. Data were mass corrected during acquisition using a 50 pg/mL leucine enkephalin solution at 10 µL/min.

Polyphenol identification was carried out by comparison of the experimental mass with the theorical mass according to their molecular formula, with a mass error <10 ppm cut-off and by the analysis of the fragmentation pattern. For compounds for which a standard was available, quantification was carried out using its corresponding calibration curve. In cases where a commercial standard was unavailable, quantification was conducted using the most similar standard available. The results were expressed as mg/mL of roselle decoction or mg/g of roselle calyx or by-product.

Antioxidant capacity

The antioxidant capacity was determined in the roselle decoction, as well as in the EPP and HPP fractions of the roselle calyx and by-product, using three widely recognized methods. The 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging assay was performed following the procedure outlined by ^{Fukumoto and Mazza (2000)}. The 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) radical scavenging assay was conducted according to the method described by ^{Re et al. (1999)}. The ferric reducing antioxidant power (FRAP) assay was carried out as reported by ^{Firuzi et al. (2005)}. The results were expressed as mmol of Trolox equivalents/mL of roselle decoction or mmol of Trolox equivalents/g of roselle calyx or by-product.

Statistical analysis

Results are presented as mean values ± standard deviation. Statistical analysis was performed using one-way analysis of variance (ANOVA), followed by means comparison by Tukey’s test (p < 0.05), using the JMP software (v14.0).

Total polyphenol content of roselle calyx decoction

As shown in Table 1, the roselle decoction exhibits a mean polyphenol content of 111 mg/100 mL, significantly higher than other herb-based beverages (^{Rothwell et al., 2013}). It is worth noting that the proanthocyanidin content in roselle decoction is 4.4 times higher than its anthocyanin content, despite the latter being traditionally associated with its health benefits. Commonly consumed polyphenol-rich beverages, such as Camellia sinensis (tea) infusion and coffee, typically contain a mean total polyphenol content ranging from 62 to 105 mg/100 mL (^{Rothwell et al., 2013}). This positions roselle decoction as a competitive antioxidant beverage.

Total extractable and non-extractable polyphenol content of roselle calyx and its decoction by-product

As the global demand for flavorful herb-based beverages grows, there is a concurrent increase in waste generated from the decoction and infusion processes (^{Debnath et al., 2021}). As shown in Table 1, used roselle calyxes (decoction by-product) displayed a lower content of extractable polyphenols (1.7-fold) compared to roselle calyxes before the decoction process, attributed to the release of these hydrophilic constituents into the decoction (beverage). This trend was observed in most of the extractable polyphenols identified in used roselle calyxes (Table 2). Notably, the antioxidant capacity assessed by ABTS and DPPH scavenging assays remained similar between roselle calyxes and their by-product, while the FRAP capacity decreased significantly (1.4 - fold; Table 5).

The release of hydrophilic constituents into the decoction led to the concentration of non-extractable constituents in the roselle decoction by-product, resulting in a higher content of both acid and alkali hydrolysable polyphenols (2.3- and 1.3-fold; Table 1) as comparison to the calyx before the decoction process. These hydrolysable polyphenols include low-molecular-weight polyphenols covalently bound to the polysaccharide constituents of dietary fiber through covalent bonds, hydrogen bonds, and/or hydrophobic interactions (^{Báez-García et al., 2023}).

Identification and quantification of polyphenols and organic acids in roselle calyx, decoction, and its by-product

Delphinidin sambubioside emerged as the major anthocyanin identified in roselle calyxes, decoction, and its by-product. The MS^E spectra obtained for this component are depicted in Figure 1, revealing a cation radical ([M]⁺) with a m/z of 597 and a fragment at m/z 303 corresponding to the delphinidin aglycone after the disaccharide moiety is lost. Although delphinidin hexoside, a minor anthocyanin was also identified, our study did not detect cyanidin derivatives, possibly due to the deep red-purple color of the calyxes used, which is attributed to delphinidin derivatives. For instance, ^{Hinojosa-Gómez et al. (2020)} reported both delphinidin- and cyanidin- sambubioside as major anthocyanins in roselle calyxes from red cultivars. It is noteworthy that delphinidin 3-sambubioside extracted from roselle calyxes decreases hepatic steatosis in both in vivo and in vitro studies by decreasing fatty acid lipogenesis and augmenting fatty acid beta-oxidation (^{Long et al., 2021}).

Figure 1 High-resolution mass spectra of delphinidin sambubioside. Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 1. Espectro de masas de alta resolución de delfinidina sambubiosida. El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

Regarding non-pigmented flavonoids, previous research have identified quercetin and kaempferol aglycones and their derivatives in various roselle varieties and plant parts (^{Da-Costa-Rocha et al., 2014}). In our study, these flavonoids were identified as minor components. Some flavones like gossypetin, hibiscitrin, sabdaretin, and hibiscetin, previously reported in roselle flowers (^{Da-Costa-Rocha et al., 2014}), were not detected.

In the context of phenolic acids, the major hydroxybenzoic acid was dihydrobenzoic acid hexoside, appearing at 315 m/z as a deprotonated molecular ion. A fragment at 153 m/z resulted from the loss of the hexoside moiety, while another at 109 m/z stemmed from CO₂ detachment (Figure 2). Accordingly, protocatechuic acid has been identified as major component in roselle calyx, with a higher content as compared to other Mediterranean plants (^{Rababah et al., 2011}) and edible flowers (^{Trinh et al., 2018}). Protocatechuic acid (3,4-dihydroxybenzoic acid) and its glucoside derivative are well-known for their antioxidant and anti-inflammatory properties (^{Kakkar and Bais, 2014}); moreover, in silico studies suggest that protocatechuic acid shows a better inhibitory potential of phosphoenolpyruvate carboxikinase, a key enzyme involved in hepatic insulin resistance, than the antidiabetic drug metformin (Mohanty and Bhadra, 2020).

Figure 2 High-resolution mass spectra of dihydroxybenzoic acid hexoside. Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 2. Espectro de masas de alta resolución de ácido dihidroxibenzoico hexósido. El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

The major hydroxycinnamic acids were caffeoylquinic acids, including chlorogenic acid, observed at 353 m/z as a deprotonated molecular ion. Fragments corresponding to quinic acid (191 m/z), caffeic acid (179 m/z), and decarboxylated caffeic acid (135 m/z) were also noted (Figures 3 and 4). Interestingly, these compounds have been reported to contribute to the flavor of roselle (^{Selli et al., 2021}) and are recognized as primary precursors for developing coffee flavor and aroma (^{Lin et al., 2022}). Coumaroulquinic acid was also found as a major phenolic acid, also identified as a deprotonated ion at 337 m/z, with fragments at 191 m/z (quinic acid), 163 m/z (coumaric acid), and 119 m/z (decarboxylated coumaric acid; Figure 5). Accordingly, chlorogenic acids have been identified as the major phenolic acid in roselle extracts, contributing with almost 18 % (^{Yang et al., 2023}), which shows high inhibitory activity against the angiotensin converting enzyme, suggesting an anti-hipertensive potential (^{Salem et al., 2020}). Conversely to our study, coumaroylquinic acids have been identified as minor components in roselle extracts (^{Yang et al., 2023}).

Figure 3 High-resolution mass spectra of caffeoylquinic acid isomer I. Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 3. Espectro de masas de alta resolución de ácido cafeoilquínico isómero I. El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

Figure 4 High-resolution mass spectra of caffeoylquinic acid isomer III (chlorogenic acid). Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 4. Espectro de masas de alta resolución de ácido cafeoilquínico isómero III (ácido clorogénico). El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

Figure 5 High-resolution mass spectra of coumaroylquinic acid isomer II. Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 5. Espectro de masas de alta resolución de ácido cumaroliquínico isómero II. El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

Among the organic acids, hibiscus acid highlighted as the major compound, consistent with previous studies (^{Sapian et al., 2023}). This organic acid appeared in its deprotonated form ([M-H]-) at 207 m/z, with fragments at 189 and 127 m/z (Figure 6). Additionally, citric and quinic acids were identified as deprotonated ions at 191 m/z with their representative characteristic fragment at 111 and 127 m/z, respectively (Figures 7 and 8). Importantly, the hibiscus acid and citric acid content, the major organic acids identified in this study, remained unaffected by the decoction process (Table 2). It is noteworthy that most studies carried out with roselle have associated its beneficial health effects to their high content of anthocyanins, phenolic acids, and non-pigmented flavonoids, whereas only few studies have focused on the bioactive potential of organic acids (^{Izquierdo-Vega et al., 2020}). In this regard, Morales-Luna et al. (2018) demonstrated that white roselle calyx extract, with a high organic acid content and no anthocyanins, exert a similar anti-obesogenic effect than red roselle calyx extract with a high anthocyanin content but low organic acid content. Among the composition of organic acids found in roselle, citric acid has been attributed with high antioxidant, anti-inflammatory and anti-coagulant activities, whereas hibiscus acid shows anti-diabetic and anti-hypertensive potential. On the other hand, hydroxycitric acid has been recently proposed as an anti-obesity nutraceutical; however, most studies have been assessed in hydroxycitric acid isolated from Garcinia cambogia (Yang et al., 2020).

The roselle decoction analyzed in this study was found to have hibiscus acid as the major component, accounting for 79.9 % of the identified organic acids, followed by delphinidin sambubioside (Table 2). Moreover, dihydroxybenzoic acid hexoside represented 73.7 % of all hydroxybenzoic acids, while the three caffeoylquinic acid isomers collectively made up to 80.5 % of the hydroxycinnamic acids. These results are consistent with previous studies demonstrating that roselle calyx decoctions are rich in bioactive compounds, known for their numerous health benefits (^{Da-Costa-Rocha et al., 2014}; ^{Ríaz and Chopra, 2018}; ^{Bedi et al., 2020}; ^{Sapian et al., 2023}).

Figure 6 High-resolution mass spectra of citric acid. Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 6. Espectro de masas de alta resolución de ácido cítrico. El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

Figure 7 High-resolution mass spectra of hibiscus acid. Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 7. Espectro de masas de alta resolución de ácido hibiscus. El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

Figure 8 High-resolution mass spectra of quinic acid. Upper mass spectra acquired at 6 V (low collision energy) and lower mass spectra acquired at 15-45 V (high collision energy).  

Figura 8. Espectro de masas de alta resolución de ácido quínico. El espectro de masas superior fue adquirido a 6 V (baja energía de colisión) y el espectro de masas inferior fue adquirido a 15-45 V (alta energía de colisión). 

Interestingly, organic acids were identified in both acid and alkali hydrolysable fractions, with higher amounts in the acid hydrolysis fraction. It is noteworthy that quinic acid has been reported to be obtained after the hydrolysis of caffeoylquinic acids and other phenolic acids esterified with quinic acid (^{Clifford et al., 2017}). Nevertheless, hibiscus acid and quinic acid were found to be increased by 1.3-fold in the roselle decoction by-product compared to the roselle calyxes. The non-extractable fractions of the roselle decoction by-product primarily consisted of hydroxybenzoic acid derivatives, which were enriched by 1.9 to 2.2-fold after the decoction process (Table 4).

Antioxidant capacity of roselle calyx, decoction, and its by-product

The results indicate that the non-extractable fraction of the roselle decoction by-product exhibits significant antioxidant capacity when assessed using the DPPH method, showing a 1.1- to 3.4-fold increase as compared to roselle calyxes before the decoction process (Table 5). However, the antioxidant capacity, as determined by the ABTS and FRAP assays, is notably lower, with reductions of 1.6- to 9.9-fold and 7.0- to 2.4-fold, respectively. These findings suggest that the non-extractable fraction of the roselle decoction by-product contains antioxidant components, such as polyphenols and organic acids, that are effective in donating protons and scavenging the DPPH radical. On the other hand, it appears to have a limited ability to donate electrons, as indicated by the reduced capacity in scavenging the ABTS radical and reducing Fe³⁺. This variation in antioxidant activity across different assays highlights the diverse mechanisms by which antioxidants can operate and underscores the importance of considering multiple methods for a comprehensive evaluation of antioxidant potential (Danet, 2021).

Notably, this study presents the first quantitative profile of the non-extractable fraction of roselle calyxes and their decoction by-product. Moreover, we demonstrate that hibiscus acid, the major roselle organic acid, is not only found as a free component but can also be found bound to the matrix by glycosidic linkage, as observed in the composition of the acid hydrolysable fraction. Few studies have determined non-polyphenol constituents in the non-extractable fraction. In this regard, ^{Dong et al. (2021)} identified eleven organic acids, along with several polyphenols, bound to carrot dietary fiber which were released by alkaline hydrolysis, contributing to its antioxidant capacity. Therefore, further studies must be undertaken to fully characterize the non-extractable fraction of plant sources and wastes.