2.1. Analysis of foods

Protein-bound biotin was released by acid hydrolysis (Mock et al., 1992) with subsequent biotin quantitation by the HPLC/avidin-binding assay (Mock, 1997). This method for release of bound biotin has been optimized for human milk o and rat liver tissue to obtain maximum (>95%) release from protein while minimizing ( 10%) biotin oxidation to biotin sulfoxide (Mock et al., 1992). To evaluate whether this acid hydrolysis procedure would release biotin from foods without destroying it, we determined biotin release and recovery in three foods. These foods were chosen because they differed substantially in protein content. Beef liver was chosen for high protein content (267 mg/g), cooked sweet potato was chosen for the low amount of protein (16.5 mg/g), and whole wheat bread was chosen for the moderate amount of protein (97 mg/g). In order to assess destruction, biotin content was quantitated in the presence of enough supplemental protein (bovine serum albumin (BSA) 40 mg) to prevent biotin destruction even during hydrolysis of the lowest protein food samples (o40 mg protein/g); control contained no added protein. The effect of acid hydrolysis on free biotin was also determined. Total biotin content of each sample was determined using acid hydrolysis as previously described (Staggs, 2002). The biotin content of beef liver, sweet potato, and whole wheat bread did not differ significantly whether supplemental protein was added or not. Free biotin was not degraded during acid hydrolysis for any of the three foods. These data provide evidence that this acid hydrolysis method does release biotin from foods with a variety of protein levels, while minimizing degradation.

Sampling: Foods were selected for analysis based on three criteria: (1) substantial biotin content as judged by published values, (2) absence of published information for the biotin content, and (3) high frequency of appearance in the 24-h diet record of Arkansas children participating in a study of nutritional status in the Mississippi Delta (Staggs, 2002).

Food items were purchased from local restaurants and local chain grocery stores supplied by large-scale producers, local producers, and wholesalers. Food samples (Table 1) were obtained by a single collection of representative products available locally. Beef liver, chicken liver, pork chop, and eggs were purchased fresh and cooked by microwave prior to homogenization. All other meats, fish and poultry were purchased fully cooked from local restaurants. Vegetables (fresh, frozen, or canned) were analyzed without cooking with the exception of sweet potato, which was purchased fresh and cooked by microwave prior to analysis. Fruits were purchased as fresh or frozen depending on availability and peeled where applicable prior to homogenization. Breads and grains, sweets, entrees, condiments and sauces, beverages, nuts, and ‘miscellaneous’ food were purchased from commercial sources and homogenized without further preparation. All food samples (approximately one serving by weight per food) were weighed and homogenized in water maintaining a weight ratio of 1 part food to 4 parts water. Homogenates were stored at -20°C until acid hydrolysis.

Acid hydrolysis: Acid (1.5 mL of 3 mol/L HCl) was added to 0.5 mL of the aqueous homogenate. The mixture was vortexed and incubated at 100°C for 120 min; samples were vortexed again after the first 10 min and every 30 min thereafter. After incubation, samples were cooled to room temperature and centrifuged at 550g for 10 min. A 1-mL aliquot of the supernatant was adjusted to a pH 2.5 with 10 mol/L NaOH and filtered through a 0.45 mm filter.

Separation of biotin: Biotin was separated from its inactive metabolites by HPLC using a C18 reversed-phase column as previously described (Mock, 1997). A 0.5-mL aliquot of the filtrate was fractionated by reversed-phase HPLC. The HPLC fractions were collected at specific retention times determined by prior injection of radiolabeled biotin. To ensure maximum recovery of the biotin peak, four fractions (1.0 ml each) were collected. Injection and collection of radioactive standards utilizing this HPLC protocol produced recovery rates of approximately 95% (Mock, 1997). The HPLC eluate fractions containing biotin were dried to remove chromatography solvents and then rehydrated with 1 mL of distilled deionized water followed by an equal volume of 0.2 mol/L HEPES 2 mol/L NaCl, 0.05% Tween 20, pH=7 buffer. Rehydrated fractions were directly transferred into 96-well round-bottom plates and assayed as described below.

HRP-avidin assay for biotin: Previous studies have validated the HRP-avidin-binding assay for quantitating biotin (Mock, 1997). The HPLC/avidin-binding assay is very sensitive with detection limits of approximately 0.001 ng biotin and has the capability to accurately separate biotin from its inactive metabolites (Mock et al., 1992).

Initially, eluate fractions were incubated with avidin coupled to horseradish peroxidase and then transferred to a microtiter well coated with biotinylated BSA. Those HRP-avidin molecules with unoccupied biotin-binding sites bound to the biotinyl-BSA adhering to the well. The more biotin (or biotin metabolite) present in the sample (or standard), the fewer avidin molecules bound to the well. The amount of bound avidin is quantitated optically by bound HRP-avidin using a timed oxidation of O-phenylenediamine. Optical density was determined using a multiwell spectrophotometer at 490 nm with background correction at 630 nm. The concentration of biotin was determined by comparing the change in the optical densities to a standard curve constructed by diluting a known concentration of biotin.

Biotin (d-biotin) was obtained from Sigma Chemical Co. (St. Louis, MO, USA). Biotin stock was prepared as previously described by Mock (1997). Assay standards were prepared by diluting a 100 nmol/L biotin solution to a final concentration of 3000 pmol/L. Biotin standard concentrations were confirmed by comparison with radiolabeled biotin of a known specific activity.