The need of RMs for the development of analytical methods has long been discussed, because spike and recovery tests do not reflect the natural binding of matrices or major elements in living organisms (protein, carbohydrates, and fats) and trace elements or contaminants at ppm level. In spite of the potential risk of fumonisin contamination in rice, no RM of rice containing fumonisins is commercially available. In this study, the applicability of the ShakeMaster grinding machine, recently used for the preparation of in-house RM of rice containing sterigmatocystin at 3 ppm level [16], was examined. About 40 g of Thai rice culture was diluted with blank Thai rice in a ShakeMaster machine until 2,000 g of mixed material of Thai rice containing FB1 at 0.1-1.0 ppm level was obtained. The values for FB1 concentration of randomly picked up from this mixed material were sufficient to pass the homogeneity test (Table 1). Therefore, the method of mixing and grinding in a ShakeMaster machine was proved to be applicable for the preparation of 2,000 g of an in-house RM containing fumonisins at lower level (<1 ppm) in the matrix of Thai rice. In case of preparation for larger scale, more efficient system which enables the grinding of 1 kg of material at once might be required.

Values of FB1 and FB2 in the in-house RM were compared for different extraction volume and submergence periods, as shown in Table 2 and Table 3. For the extraction of fumonisins from food samples, two-fold of sample volume of methanol-water (3 + 1, v/v) (for example, 100 mL of methanol-water (3 + 1, v/v) to 50 g of corn) has often been used as extraction procedure in the past since the ratio was adopted by AOAC Official Method 995.15, a validated method for the analysis of fumonisins in corn [17]. However, in our previous study this extraction ratio (two-fold of sample volume) was not sufficient for fumonisins in Japanese rice and a five-fold increase of further 2.5-fold increase of solvent volume (for example, 50 mL of methanol-water (3 + 1, v/v) to 10 g of Japanese rice) was required [18]. In the Thai rice matrix, we compared six-fold, eight-fold and ten-fold solvent for extraction. As shown in Table 2, significant higher values of FB1 and FB2 were obtained in eight-fold solvent volume (80 mL) than in six-fold solvent volume (60 mL), while no significant difference was observed between eight-fold solvent volume (80 mL) and ten-fold solvent volume (100 mL). We adopted another extraction technique using submergence in water prior to the addition of methanol, which has been effective in the extraction of fumonisins in Japanese rice [18]. The technique was also proved to be effective and showed dramatic increase in the values of FB1 and FB2 in the matrix of Thai rice (Table 3).

The recoveries of FB1 and FB2 spiked to blank Thai rice are shown in Table 4. Spike and recovery test was done in triplicate at 0.2 ppm-spiked level and the average value was shown with standard deviation (S.D.). Considering the result of Table 2, the eight-fold volume of extraction solvent was adopted and the effect of submergence was evaluated. As shown, submergence was also worked and resulted in low S.D. for repeatability (1-2%) and proper recovery (62-67%) (Table 4: 30 min submergence) (Figure 1D). On the other hand, without submergence, the value of FB1 was under 0.06 ppm which means the recovery was below 30% (Table 4: 0 min submergence) (Figure 1C). For the clean-up step for fumonisins, a kind of solid phase extraction (SPE) cartridge of strong anion exchange (SAX), Bond Elut SAX, was used followed by AOAC Official Method 995.15 [17], and it is proved to work well.

Thai rice harvested in the 2006 was obtained from a retail shop and stored at 4 °C until used as a sample. Standard FB1 and FB2 were purchased from Wako Chemicals (Osaka, Japan). All other reagents were of HPLC grade. Stock solutions (100 μg/mL each) were prepared in acetonitrile-H₂O solution (1 + 1, v/v) as recommended by AOAC Official Method 995.15 [17], and were diluted to the desired concentration with acetonitrile-H₂O solution. Fumonisin stocks and standard working solutions were kept at 4 °C up to 6 months.

Wild type Fusarium verticillioides MAFF239106 (Genebank, National Institute of Agrobiological Resources, Ministry of Agriculture, Forestry and Fisheries, Japan) was used for the preparation of artificially molded Thai rice contaminated with fumonisins. The fungus was pre-inoculated and grown on a potato dextrose agar plate for 3 days. Thirty grams of Thai rice wetted with 15 mL of water for three hours was autoclaved and served as a fumonisin producing culture medium. The fungus grown on the plate was sliced as three debris of agar and inoculated on the above medium, and kept for two weeks at 25 °C until to accumulate ca. 50 ppm of fumonisins in the Thai rice culture medium. The culture was ground with blank Thai rice at the ratio of 1:10 in a ShakeMaster (Biomedical Sciences, Tokyo, Japan) equipped with a container of 400 g capacity to make a fine powder. This procedure was repeated to obtain 2,000 g of powdered Thai rice containing low-level of fumonisins (0.1-1.0 ppm). Randomly picked up in-house RM samples of 5.0 g each were extracted in a 50 mL tube with a screw cap using 40 mL methanol-H₂O solution (3 + 1, v/v) by vigorous vortexing for 1 min with a Vortex Genie2 (Scientific Industries, NY, USA), and the supernatant was purified and analyzed in the procedure described in 3.3. Homogeneity of artificially prepared fumonisins contaminated Thai rice powder was confirmed by one-way analysis of variance (ANOVA). The powder which passed the homogeneity test as shown in Table 1 was served as an in-house RM.

Commercial Thai rice (ca. 60 g) was milled for 3 min with a Waring Laboratory Blender (Model 7012S, Waring Commercial, CT, USA) to yield a fine powder which passes through 0.6 mm screen. An accurately weighed 10.0 g of Thai rice powder was extracted in a 300 mL Erlenmeyer flask with a ground-in stopper using 80 mL methanol-H₂O solution (3+1, v/v) by vigorous shaking on a shaker (Laboratory shaker, TAITEC, Japan) for 60 min. As for extraction solvent, the portion of water (20 mL) and that of methanol (60 mL) was used separately as follows. First, the sample was mixed and submerged with 20 mL of water in a flask with a stopper for 30 min with vortex on a vortex machine (Vortex Genie2) for 30 sec at 10 min. Second, extraction was followed by the addition of 60 mL of methanol. The extract was purified by a SPE cartridge (Bond Elut SAX, Varian, CA, USA). An accurate volume of filtrate (8.0 mL of Thai rice extract) was loaded onto the SAX cartridge and washed with 3 mL of methanol-water (3 + 1, v/v) followed by 3 mL of methanol. Fumonisins were eluted with 7 mL 1% methanolic acetic acid at a flow rate less than 1 mL/min. The eluate was collected in a 10 mL amber glass tube and evaporated at 60 °C under a gentle flow of nitrogen. The tube was rinsed with 1 mL methanol and evaporate additional methanol to dryness to ensure that remain of acetic acid has evaporated. The residue was redissolved in 1.0 mL acetonitrile-water (1 + 1, v/v) and served as the sample solution for automatic injection to HPLC-FL with pre-column derivatization as previously reported [18].

For the recovery test of spiked sample, standard fumonisins solution (0.8 mL of 2.5 μg/mL FB1 and FB2 mixture to make 0.2 ppm (=mg/kg) spiked samples) were added 5 minutes prior to the addition of extraction solvent, and followed by the extraction and cleanup procedure as described above. Recovery was calculated by the equation as follows:

Recovery (%) = (A - B)/A × D × 100

where A = signal of spiked sample, B = signal of blank sample, D = dilution factor (=5.0 or 10.0 (g)/40 or 80 (mL) × 8.0 (mL)/1.0 (mL) = 1.0). As mentioned in 3.3, we added water and methanol for extraction separately. The sum of 20 mL water and 60 mL methanol becomes a little less than 80 mL (ca. 78 mL), which substantially treated as 80 mL.

HPLC-FL analysis was performed using an LC-10A series HPLC system (Shimadzu, Kyoto, Japan) equipped with a C18 L-column (250 × 4.6 mm i.d., 5 μm spherical particle size; CERI, Tokyo, Japan), an auto-injector SIL-10A (Shimadzu) and a fluorescence detector RF-10AXL (Shimadzu). The mobile phase was isocratic and composed of methanol-0.1 M sodium phosphate monobasic (77 + 23, v/v), adjusted to pH 3.3 with o-phosphoric acid. The mobile phase flow-rate was 1.0 mL/min and column oven temperature was 40 °C. Injection was made by programming with precolumn derivatization so that 10 μL of fluorescent reagent mixture (0.3 M o-phthalaldehyde in methanol - 0.1 M disodium tetraborate - 2-mercaptoethanol (1 + 5 + 0.05, v/v) and 10 μL of sample solution was mixed followed by injection of 5 μL. Fluorescence was measured with excitation at 335 nm and emission at 440 nm with a slit width of 12 nm. Chromatographic data was analyzed using Chromatopac system (Shimadzu). Calibration curves were based on the analysis of working standard solution in the ranges of 0.05-1.0 μg/mL (0.05, 0.1, 0.2, 0.5 and 1.0 μg/mL) for FB1 and FB2. LOD was calculated from the standard curve (r² > 0.99) and the LOQ was determined as twice the LOQ.