In order to establish whether the addition of BHT to the extraction solvents used in modification 5 favored carotenoid stability during analysis, transgenic maize TM2 was extracted with and without BHT. Samples were injected into a chromatograph 48 h after extraction. There was no statistically significant difference between the individual and total carotenoid content in samples in spite of using an anti-oxidant (Table 2). Indeed, the t calculated value was always lower than the t critical value for all cases. These results demonstrate that addition of BHT to the extraction solvents can be omitted provided that carotenoids are analyzed within 48 h of storage at −80 °C.

Most carotenoids are insoluble in water and soluble in organic solvents such as acetone, alcohol, THF, ethyl ether, chloroform and ethyl acetate [27]. Nevertheless, their solubility depends on the presence of different functional groups. To ensure the complete solubilization of these pigments and to avoid incompatibility of the injection solvent with the mobile phase, combinations of acetone and 2-propanol (iPrOH) with the mobile phase (only solvent A, ACN–MeOH 7:3, v/v) were tested as injection solvent. Acetone and iPrOH were selected because they are miscible with the mobile phase and less polar than ACN and MeOH. Therefore, it was considered that these solvents might contribute to increasing the miscibility of carotenes. Solvent A was included as one of the components of the injection solvent since it is advisable to prepare the sample in the operating mobile phase for the best peak shape and sensitivity [30]. Various aliquots of the same maize sample (TM1) were obtained. Each one was dissolved in the same volume of injection solvent. The injection volume was also identical in all cases. This approach facilitated comparison between the corresponding chromatographic peaks and hence evaluation of the effect of the various injection solvents.

Table 3 shows the injection solvents used to dissolve the carotenoids in TM1. No variations in retention time were observed when samples were dissolved in the different injection solvents (data not shown). The highest concentrations of pigments were obtained using a mixture of mobile phase–acetone 6.7:3.3 v/v rather than mobile phase alone as the injection solvent. For example, with this mixture the content obtained for zeaxanthin and lutein was 4.49 μg/g DW while with mobile phase alone this concentration dropped to 3.89 μg/g DW. A reduced percentage of acetone (sample dissolved in mobile phase–acetone 7.5:2.5 v/v) led to a decrease in the content of all carotenoids (Table 3). With this injection solvent, a concentration of 4.20 μg/g DW was obtained for zeaxanthin and lutein. The sample dissolved in ACN–MeOH–iPrOH 8.5:1:0.5 v/v/v did not show any increase in carotenoid contents (3.05 μg/g DW was obtained for zeaxanthin and lutein). Nevertheless, the lowest content of carotenoids were obtained dissolving the sample in 100% acetone. Acetone produced 2.20 μg/g DW of zeaxanthin and lutein content. Therefore, our results indicated that combinations of solvents were more advantageous to improve sample solubility than a single solvent such as acetone.

Nevertheless, acetone was required to increase the solubility of the sample under the specific chromatographic conditions used because of the polarity of carotenoids present in TM1 (antheraxanthin, adonixanthin, lutein, zeaxanthin and α-cryptoxanthin). Thus, the mixture of mobile phase–acetone 6.7:3.3, v/v was chosen as the injection solvent. Figure 1 shows the separation of a mixture of carotenoids in the transgenic maize line TM2 using the improved UHPLC system.

The solvents used to dissolve the carotenoid standards were chosen considering either the previously reported carotenoid solubility or the availability of the absorption coefficient of each pigment.

Stock carotenoid solutions were prepared in EtOH, acetone and hexane [31]. Carotenoid concentrations were determined spectrophotometrically. Table 4 shows the solvent and the value of A used to quantify each pigment. Working solutions were prepared from stock solutions by sampling an aliquot and diluting it with injection solvent. Solution concentrations were assessed by UHPLC analysis. For those carotenoids dissolved in hexane (canthaxanthin, β-cryptoxanthin, β-carotene, lycopene and phytoene), working solutions were prepared from stock solutions by evaporating an aliquot under nitrogen and diluting it with injection solvent. Calibration curves were obtained from peak area by injecting mixtures of standards. Table 5 shows the linear regression data for each carotenoid standard curve.

Solubilization problems were encountered for carotenes (when lycopene and β-carotene were dissolved in hexane) and for ketocarotenoids (when astaxanthin and canthaxanthin were dissolved in EtOH and hexane respectively). Chloroform, dichloromethane, hexane, ethyl acetate and THF [6,24,32,33] are known to dissolve lycopene and β-carotene. Thus, we chose hexane to solubilize these compounds. Initially, we attempted to prepare 100 μg/mL stock solutions of lycopene and β-carotene in hexane, but a precipitate was observed in the bottom of the vessels. Consequently, to ensure that carotenes were completely dissolved, stock solutions of carotenes were prepared again in hexane but in lower concentrations (Table 4). Thus, stock solutions of 24.85 μg/mL for β-carotene and 8.26 μg/mL for lycopene were prepared. We did not encounter solubilization problems with these concentrations. In addition, the calibration curves of these pigments (Table 5) indicated that the chromatographic peak areas of carotenes gave a linear plot throughout the concentration range studied.

Similarly, canthaxanthin and astaxanthin were not properly dissolved in hexane and EtOH respectively. Because of the problems of solubilization observed with these compounds, we determined and compared their concentrations using two different methods: dividing the mass of the carotenoid by the total volume of solution (theoretical concentration) and spectrophotometrically (experimental concentration).

The theoretical and experimental concentration obtained for canthaxanthin was 4.53 and 0.27 μg/mL, respectively whereas for astaxanthin it was 4.96 and 1.55 μg/mL, respectively. The lower concentrations of ketocarotenoids obtained experimentally indicated that hexane and EtOH were not appropriate solvents for canthaxanthin and astaxanthin, respectively. In addition, Table 5 shows that the calibration curves of these two pigments were characterized by a poor r-squared (R² < 0.94). Therefore, we prepared stock ketocarotenoid solutions in the injection solvent of 5.12 μg/mL for astaxanthin and 5.70 μg/mL for canthaxanthin (Table 4). Table 5 shows that the calibration curves of these pigments dissolved in the injection solvent gave a linear plot throughout the concentration range studied. The lack of information about carotenoid absorption coefficients in a variety of organic solvents hampers the use of more appropriate solvents for ketocarotenoids.

We did not encounter any solubilization problems with the concentration range used for: (a) violaxanthin, antheraxanthin, neoxanthin and lutein, dissolved in EtOH; or (b) zeaxathin, dissolved in acetone and (c) β-cryptoxanthin, dissolved in hexane. As reported previously [26,27,34,35], oxygen-functionalized carotenoids showed satisfactory solubility in MeOH, EtOH and acetone.

Given the concentrations of carotenoids expected in maize samples, we did not prepare standard concentrations above 40 μg/mL. However, in our experience, higher concentrations of oxygen-functionalized carotenoids can be prepared with the injection solvent used here when needed. For example, concentrations of 100 μg/mL can be prepared for violaxanthin and neoxanthin. If higher carotene concentrations were required, changes in the injection solvent should be made to increase their solubility. For example, when lycopene is the target analyte, the following injection solvents have been used: chloroform 100%, to analyze extracts of tomato fruit pericarps [36]; ethyl acetate 100% [37] and n-butanol–ACN–dichloromethane (3:7:0.1, v/v/v) [38] to analyze extracts of tomato fruit.

Konings et al. [35] prepared stock solutions of lutein, zeaxanthin, β-carotene and lycopene with the same solvents used in this study. However, they used a mixture of MeOH–THF (7.5:2.5, v/v) as injection solvent. Under the chromatographic conditions applied, they observed a higher linear range for lutein, zeaxanthin and β-carotene than for lycopene. The smaller linearity range of lycopene (from 0 to 3.5 μg/mL) was explained by the lower solubility of this compound in the injection solvent. Nevertheless, the choice of the injection solvent was a compromise between satisfactory solubility of carotenoids, compatibility with the mobile phase, and the absence of peak distortions.

When carotenoid standard solutions are used several times and stored under N₂ or Ar, their concentrations should be evaluated since the inert gas introduced various times into the vial evaporates the solvent, thereby changing the original carotenoid concentration. Thus, it is advisable to either divide the volume of carotenoid standard solutions into vials, putting only the volume required for each analysis into single vials, or to dry the standard solutions and redissolve these in each analysis. In addition, attention should be paid when many carotenoid standards at high concentrations are solubilized in the same solvent as some might precipitate. Thus, it is preferable to prepare various mixtures of carotenoids to ensure the complete solubilization of all analytes.

To protect carotenoids from degradation and oxidation, the extraction was conducted under limited light and THF was treated with sodium metal to remove peroxides. The procedure described by Naqvi [13] was used as reference method. Maize endosperm was excised by removing the seed coat and embryo. Only the maize endosperm was used for extraction because carotenoids were designed to accumulate only in this tissue by virtue of the genetic construct used to create the transgenic maize plants (endosperm-specific expression) [39]. Samples were freeze-dried and ground into a fine powder using a mortar and pestle. Maize seeds were lyophilized in order to prevent either microbial or chemical deterioration, which may be caused by the water content in the matrix. In addition, we chose to work with dried rather than fresh material because the former is more amenable to small-scale analysis and more easily homogenized. 50 or 100 [40] mg of sample was extracted with 15 mL of MeOH–THF (1:1, v/v) at 60 °C for 20 min and this mixture was continuously shaken. It was then put on ice until it reached room temperature and the liquid phase was filtered into a separatory funnel (if the residue exhibited color after extraction, it was re-extracted with 5 mL of the first extraction solvent at 60 °C for 5 min and the second extract was combined with the first one). 15 mL of hexane–diethyl ether (9:1, v/v) was added to the organic extract and the mixture was shaken vigorously. Then, 20 mL of saturated sodium chloride solution was added and the mixture was shaken again. The aqueous phase was removed and the organic phase was washed with water once again. The organic phase was concentrated under N₂ at 37 °C until the volume was adjusted to 5 mL. 1 mL of blank solution was transferred to a cuvette and this was used to set the baseline absorbance of the spectrophotometer at 450 nm. The absorbance of 1 mL of the organic phase was determined. This organic phase was returned to the tube and left under N₂ for further drying. When the sample was completely dry, Ar was flushed into the vial and carotenoids were stored at −80 °C until LC analysis.

No saponification step was included because carotenoids are generally not present in ester forms in maize [7]. This step also has the inherent disadvantage of causing carotenoid losses [12].

When carotenoids are the only pigments present in the samples, extraction is facilitated as the process can be monitored. Consequently, loss of color was used as an indication of complete or satisfactory carotenoid extraction from the matrix.

A blank solution was prepared by carrying out the same extraction process as described in Section 3.3.1, without sample.

BHT at a concentration of 0.1% was added to the extraction solvents used in modification 5: MeOH–ethyl acetate 6:4, v/v and hexane–diethyl ether 9:1, v/v. Each extraction was carried out in triplicate.

UHPLC-PDA analysis was carried out using an ACQUITY Ultra Performance LC^TM system linked to a photodiode array (PDA) 2996 detector (Waters, Milford, MA, USA). MassLynx^TM software version 4.1 (Waters) was used to control the instruments, and for data acquisition and processing. UHPLC chromatographic separation was performed on a reversed-phase column ACQUITY UPLC^® BEH 130Å C18, 1.7 μm, 2.1 × 100 mm (Waters) and a gradient system with the mobile phase consisting of solvent A, ACN–MeOH: (7:3, v/v) and solvent B, H₂O 100%. The gradient program used is shown in Table 6. Column and sample temperatures were set at 32 °C and 25 °C respectively. Before use, all solutions were passed through Millex 0.2-μm nylon membrane syringe filters (Millipore, Bedford, MA, USA). The injection volume was 5 μL.