Extraction and isolation of bixin from annatto seeds

Annatto (Bixa orellana L.) seeds were extracted with acetone, yielding a mixture containing 1.42% total carotenoids of seeds, and 66.81% natural bixin (by HPLC Area, Fig.1 ) of the total carotenoids extract. The quantifications were performed according to FAO/WHO Joint Expert Committee on Food Additives ⁵ detailed in the experimental section (Sample quantification) where the percentage was determined by HPLC-SPD at 450 nm using the peak area in the chromatogram. Data from literature ⁴ indicates that approximately 80% of the naturally occurring carotenoids present in the annatto seeds is bixin (1), (Fig.1) In this sense, the lower percentage obtained could be due to the variety of annatto acquired, or the post-harvest seed treatment ^19),(20. The isolation of natural 9’Z-bixin (1), (Fig.1) from annatto seeds was achieved with a peak in the HPLC-SPD chromatogram at 450 nm (100% of the total area), a fact indicating a high degree of purity.

Hemisynthesis of bixin derivatives

Hemisynthetic 9´Z-norbixin (2) (Fig.1) was obtained by the saponification of annatto seeds extract with 5% potassium hydroxide in methanol ²¹, with a yield of 85% of the total carotenoids extract, determined by HPLC-SPD area at 450 nm.

HPLC run with the products of saponification of the crude extract of total carotenoids of the seeds for the obtaining of (2) (Fig.1) also allowed the obtaining of 9’Z-methylbixin (3) (Fig.1) with a high degree of purity, HPLC-SPD chromatogram at 450 nm (91 % of the total area).

Hemisynthetic 9´Z-methylbixin (3) was obtained from natural 9’Z-bixin (1) through Steglich esterification of the later ²², with the use of N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC) and dimethylaminopyridine (DMAP). The yield was 74% of (1). This methodology is new for the obtention of 9´Z-methylbixin. Previously, hemisynthetic methylbixin was obtained via reaction of naturally occurring 9´Z-bixin with diazomethane or dimethylsulfate ²³. Both methodologies required very expensive and specialized equipment ²⁴, and the use of these reagents can cause genetic damage to the users that perform the reaction ²⁵. The use of EDC/DMAP allowed for hemisynthetic 9´Z-methylbixin (3) (Fig.1) to be obtained in a simple, economic way and with good yields. The isolation of (3) occurred with a high degree of purity according to HPLC-SPD chromatogram at 450 nm with 100 % of the total area.

Hemisynthetic E-methylbixin (4) was obtained by esterification in acid conditions (HCl + CH₃OH), of compound (1). The compound was obtained with a yield of 34% of (1) determined by HPLC-SPD area at 450 nm. The low yield may be due to the high concentrations of acid used or the prolonged reaction time. It is reported in the literature that high concentrations of acid produce carotenoid bleaching, meaning that the carotenoid is degraded in lower molecular weight components that do not absorb in the visible region ²⁶. The isolation of (4) presented a peak in the HPLC-SPD chromatogram at 450 nm of 97 % of the total area, which indicated a high degree of purity.

The ¹H- and ¹³C-NMR, UV/VIS, MS and MS² data of bixin (1) and bixin derivatives (2), (3) and (4) are in full agreement with the data reported in the literature ^{7), (23), (27), (28}.

Evaluation of the antioxidant capacity with the radical ABTS^•+

Table 1 shows the TEAC and α-TEAC results. Compounds (1) and (2) showed the highest TEAC and α-TEAC values. They did not present significant differences between values obtained in their respective assays. The same can be said for compounds (2), (3) and (4). This reveals that there are no significant differences between the antioxidant capacity of α-tocopherol and Trolox ²⁹.

The results showed that the esterification [nucleophilic substitution that gave rise to a methoxy group out of a hydroxy group that was replaced in the carboxylic acid residue (C-6) of carotenoid (1), to afford (3) or (4), Fig.1] decreased the TEAC and α-TEAC values. Evidence indicates that the aforementioned substitution almost completely decreases antioxidant capacity. Moreover, there are no differences in the antioxidant activity between compounds (1) and (2) revealed by both assays (TEAC and α-TEAC).

On the other hand, no differences between compounds (3) and (4) were observed regarding their antioxidant capacity, according to the currently employed antioxidant assays. In spite of the opposed stereochemistry at the C^9’=C^10’ bond, Z [(3)] versus E [(4)], the difference in the ability to quench ABTS radicals between these compounds, was irrelevant.

In Fig. 1, compound (1) with one terminal acid function (C-7,) and (2) with two (C-7 and C-7’), showed more antioxidant capacity, than (3) and (4), with ester functions at such positions . The derivatization seems to be the cause for the difference in the ABTS indexes (Table 1).

Is very important to recognize that (1) and (2) are the major compounds in the commercial annatto extract [30] and have a high antioxidant capacity compared with the bixin derivatives.

Table 1 TEAC and α-TEAC of bixin and bixin derivatives in scavenging the radical ABTS^•+ 

In the scavenging of the radical ABTS^•+ in the presence of such hydrogen-donating antioxidants ¹⁵, it seems that the acid proton of the acid function, could be responsible for the antioxidant capacity. It is possible that esterification (no acid proton) limits bixin-derivatives’ antioxidant capacity. The terminal acid function shows the possibility for decarboxylation (-CO₂) with formation of ·H, this is not possible for derivatives.

General Experimental Procedures

Bixin extract and fractions were isolated using a semipreparative HPLC system (HP 1050 Series, Hewlett-Packard/Agilent Technologies Co., Palo Alto, CA, USA), equipped with a ProntoSil C30 of 200 x 3,0 mm column (Bischoff, Leonberg, Germany). A separation was performed by gradient elution (2.5 ml/min) started with 100% of solvent A until reached 100% of solvent B in 50 min. Solvent B was a mixture of 90% methyl t-butyl ether: 6% methanol: 4% water. Solvent A was a mixture of 15% methyl t-butyl ether: 81% methanol: 4% water. Data analysis was performed using the OpenLab CDS data system.

The UV-vis spectra were recorded with a Genesys 10S UV-Vis spectrometer (Thermo Scientific, Waltham, MA, USA). The MS and MS² experiments were performed using a Waters UPLC ACQUITY class H coupled with XEVO TQD spectrometer (Waters Corporations, Milford, MA, USA). The samples were dissolved in MeOH (HPLC-grade) and measured in positive APCI mode. The samples were filtered in a 0.22 µm filters and analyzed with the following conditions: a cone voltage of 30 V, corona voltage of 5 V, probe temperature of 550 °C, carrier gas desolvation flow of 500 L/h, cone flow of 50 L/h and collision energy of 3 V. For the MS² spectra, the collision energy was increased between 7 and 20 V. The ¹H NMR (600 MHz), ¹³CNMR (151 MHz), and 2D NMR experiments ¹H-¹H COSY, HMBC and HSQC were run in a Bruker Avance 600 MHz NMR spectrometer using Bruker microprograms. The samples were dissolved in DMSO-d⁶ for (1) and (2) or CDCl₃ for (3) and (4). Chemical shifts were measured from internal TMS or residual solvent signals. UV, MS and NMR spectra available as supplementary material.

Plant material

Annatto (Bixa orellana L.) seeds were bought in a local market in Panama City (MERCA PANAMA, Panama).

Bixin Isolation

200 g of seeds were extracted with acetone (2000 mL). The acetone extract was dried on a rotary evaporator. 1 g of dried annatto seed extract was chromatographed on a 60 g silica gel column with a gradient of diethyl ether in hexane. The isolated bixin fraction was purified by semipreparative HPLC-SPD on a ProntoSil C30 250 x 10,0 mm (Bischoff, Leonberg, Germany) column.

Sample Quantification

The bixin and bixin derivatives content was determined by spectrophotometric methodology using to specific absorption coefficient and according to the FAO/WHO Joint Expert Committee on Food Additives ⁵. The sample’s absorbance was measured in a 1 cm quartz cuvette at 487 nm using a UV-visible spectrophotometer and acetone as a blank. The contents of bixin in the samples were calculated with the following equation ⁵:

in which:

A_s= the absorbance of the samples;

Fd = the dilution factor;

A^100% _{1 cm}3090 (1g/100 mL)^-1∗1 cm^-1 for bixin in acetone (specific absorption coefficient).

The concentration was expressed as total carotenoids in bixin form, and the purity was determined by HPLC-SPD at 450 nm using the peak area in the chromatogram.

To correct the quantification of compounds (2), (3) and (4) that were structurally different from bixin, but they have very similar UV-Vis spectra, a correction factor calculated as follows was used:

in which:

M.M. = The molecular mass of the compound;

M.M. = The molecular mass of bixin (394.5 g/mol).

This methodology was used in the quantification of all the samples in this work.

Hemisynthesis of Bixin Esters

Isolated natural bixin was submitted to esterification methods.

9´ Z-Norbixin (2)

Bixin (0.13 mmol, 50 mg) in 50 mL of dry diethyl ether, was dissolved in a 5% KOH methanolic solution. The sample was stirred in a Schlenk flask for five hours at 30 °C. At the beginning, 20 mL of chlorohydric acid (37 % w/w) was added slowly, and the diluted sample was transferred to a separation funnel and washed with diethyl ether (four times with 250 ml each), and 1 L of deionized water until it reached a neutral pH. The organic phase was evaporated in vacuo to obtain the products. The residue was rediluted in a mixture of 90 methyl t-butyl ether: 6 methanol: 4 water, for isolation. The desired compound was isolated by semipreparative HPLC-SPD, with a ProntoSil C30 250 x 10,0 mm column (Bischoff, Leonberg, Germany). Chromatographic separation was performed by using an isocratic mixture (90 methyl t-butyl ether: 6 methanol: 4 water) and flow of 2 mL/min.

9' Z-Methylbixin (3)

A total of 98 mg (0.51 mmol) of N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC) (2 equivalent), 0.3 mL of pyridine (1.19 mmol) (4.7 equivalent) and 8 mg (0.065 mmol) of dimethylaminopyridine (DMAP) were added to bixin (0.25 mmol, 100 mg) (1 equivalent) in dry dichloromethane (DCM). The sample was stirred for 24 hours at 30 °C. The sample was washed with HCl 2% (three times with 20 ml each), 5% NaHCO₃ (Three times with 20 ml each) and 50 mL of deionized water. The combined organic phase was evaporated in vacuo to obtain the products. The residue was adsorbed in three grams of Silica gel 60 and eluted in a column packed with Silica gel 60, with a gradual increment of diethyl ether in a hexane solution. The fraction with the highest content of the product was isolated by semipreparative HPLC-SPD with a ProntoSil C30 250 x 10,0 mm column (Bischoff, Leonberg, Germany). Chromatographic separation was performed using an isocratic mixture (90 methyl t-butyl ether: 6 methanol: 4 water) and flow of 2.5 mL/min.

All E-Methylbixin (4)

A total of 30 mL of a solution of acidified methanol (5 M) was added with bixin (0.78 mmol, 300 mg) in dry methanol, which was prepared by bubbling dry hydrogen chloride in methanol. The sample was stirred for 24 hours at 60 °C. A total of 10 g of sodium bicarbonate solid was slowly added to the mixture, evaporated in vacuo, dissolved with 200 mL of DCM, and washed with 400 mL of deionized water. The organic phase was evaporated in vacuo to obtain the products. The residue was adsorbed in 3 g of silica gel 60 and eluted in a column packed with silica gel 60 with a gradient of diethyl ether in a hexane solution. The fraction with the major content of the product was isolated by semipreparative HPLC-SPD with a ProntoSil C30 250 x 10,0 mm column (Bischoff, Leonberg, Germany). Chromatographic separation was performed by using an isocratic mixture (90 methyl t-butyl ether: 6 methanol: 4 water) and flow of 2 mL/min.

Evaluation of the antioxidant capacity with the radical ABTS^•+

The following experimental procedures were applied to evaluate the antioxidant capacity of 9´Z-bixin and bixin derivatives and, when possible, structure-activity relationships. Before analysis, Stock solutions of bixin and bixin derivatives (50 µmol/L) were prepared in diethyl ether and 2 mL were transferred to an amber vial. All were evaporated under a stream of nitrogen at 30 ± 1 °C. The samples were stored dry at -20° C. Exact concentrations of the stock solutions were determined spectrophotometrically using the methodology previously described. For the analysis, the samples were redissolved in 1 mL of methanol (100 μM), successive dilutions were prepared at 100, 70, 30 and 10 μM. All the carotenoid samples were analyzed in triplicate at four different concentrations (100, 70, 30 and 10 μM).

The Trolox equivalent antioxidant capacity (TEAC) and α-tocopherol equivalent antioxidant capacity (α-TEAC) procedure used an automated plate reader (KC4, Bio Tek, USA) with 96-well plates. Procedures for measuring the TEAC and α-TEAC values were carried out using Mϋller et al.’s (2010) method [29] with a few modifications. The radical solution of ABTS^•+ was prepared by passing an ABTS solution in 5 mM phosphate-buffered saline (PBS) solution, with pH 7.4, through a filter paper coated with manganese dioxide. This was followed by filtration with a syringe filter with a pore size of 0.22 μm. Before analysis, the stock solution of ABTS was diluted with PBS to obtain an absorbance of 0.75 ± 0.05 at 734 nm and stabilized for 2 hours. In the assay, 40 μL of carotenoid solution, Trolox (ca. 4.7-117.5 μM), α-tocopherol (ca. 4.4-111.6 μM) or blank (methanol) and 240 μL of ABTS were stirred for 30 s. Thereafter, the absorbance was determined at 734 nm for 20 min. For the TEAC, Trolox was used as a standard, and α-tocopherol was used as a standard for α-TEAC.

Statistics

Each sample analysis was performed in triplicate. All the presented results are the means (±SD) of at least three independent experiments. Statistical analysis (ANOVA with a statistical significance level set at p < 0.05, with a post hoc Student-Newman-Keuls (S-N-K) procedure) was carried out with SPSS 17.0 for Windows (SPSS Inc., Chicago, IL, USA).