Flavourzyme is sold as a peptidase preparation from Aspergillus oryzae. The enzyme preparation is widely and diversely used for protein hydrolysis in industrial and research applications. However, detailed information about the composition of this mixture is still missing due to the complexity. The present study identified eight key enzymes by mass spectrometry and partially by activity staining on native polyacrylamide gels or gel zymography. The eight enzymes identified were two aminopeptidases, two dipeptidyl peptidases, three endopeptidases, and one α-amylase from the A. oryzae strain ATCC 42149/RIB 40 (yellow koji mold). Various specific marker substrates for these Flavourzyme enzymes were ascertained. An automated, time-saving nine-step protocol for the purification of all eight enzymes within 7 h was designed. Finally, the purified Flavourzyme enzymes were biochemically characterized with regard to pH and temperature profiles and molecular sizes.

Properties of oil-in-water (O/W) emulsions affecting initial dynamic flavour release were studied in real time considering mouth conditions. Aroma molecules from different chemical classes at concentrations typically present in beverages were used. The emulsion droplet diameter showed no significant influence on the dynamic flavour release. No barrier properties of the emulsifier were found, as the flavour release from equilibrated emulsions flavoured via either the oil phase or the aqueous phase showed no significant difference. Emulsifier concentrations above the critical micelle concentration did not influence the release. Even though the chemical composition of the lipids had considerable influence on flavour release, phase transition during equilibration from the liquid to the solid state insignificantly affected the initial dynamic release process.

The initial dynamic flavor release from sucrose solutions was modeled. Modeling was based on the theoretical hydration behavior of sucrose, theoretical physicochemical data of flavor volatiles, and process parameters of a headspace apparatus used for model validation. The rate-limiting factor determining the initial flavor release was the hydration of sucrose, which in turn depends on the molarity of sucrose in the solution and, therefore, on the actual amount of nonbound water. Improved solubility of the more hydrophilic compounds due to their orientation toward the hydration shells of the sugar molecules was considered. The viscosity of nonassociated water forming the microregion for mass transfer of volatiles was considered instead of the bulk solution viscosity. Experimental validation of the model by real-time measurements of dynamic flavor release using foodlike flavor concentrations confirmed the above theory. Increasing sucrose concentrations resulted predominantly in increased flavor release, and bulk solution viscosity showed no effect.

A fully computer-controlled apparatus was designed. It combines a glass reactor with a temperature-controlled hood, in which headspace volatiles are captured. Flavored liquids can be introduced into the reactor and exposed to conditions of temperature, air flow, shear rate, and saliva flow as they occur in the mouth. As the reactor is completely filled before measurements are started, creation of headspace just before sampling start prevents untimely flavor release resulting in real time data. In the first 30 s of flavor release the concentrations of the volatiles can be measured up to four times by on-line sampling of the dynamic headspace, followed by off-line trapping of the samples on corresponding Tenax traps and analysis using GC-TDS-FID. Flavor compounds from different chemical classes were dissolved in water to achieve concentrations typically present in food (micrograms to milligrams per liter). Most of the compounds showed constant release rates, and the summed quantities of each volatile of three 10 s time intervals correlated linearly with time. The entire method of measurement including sample preparation, release, sampling, trapping, thermodesorption, and GC analysis showed good sensitivity [nanograms (10 s)(-1)] and reproducibility (mean coefficient of variation = 7.2%).

No abstract