Characterization and Modulation of the Bitterness of

Mar 2, 2016 - Using in vitro receptor assays, hTAS2R14 was shown to be the main bitter receptor involved in their perception, with EC50 values of 14 a...
0 downloads 9 Views 3MB Size
Article pubs.acs.org/JAFC

Characterization and Modulation of the Bitterness of Polymethoxyflavones Using Sensory and Receptor-Based Methods A. Max Batenburg,* Teun de Joode, and Robin J. Gouka Unilever R&D Vlaardingen, P.O. Box 114, 3130 AC Vlaardingen, The Netherlands S Supporting Information *

ABSTRACT: An obstacle in the application of many “health ingredients” is their alleged off-flavor. We used a combination of chemical, sensory, and biological analyses to identify the bitter components in citrus peel-derived polymethoxyflavone preparations, claimed to be functional in the lowering of cholesterol. Nobiletin (56−81%) and tangeretin (10−33%) were found to be the main bitter components. Using in vitro receptor assays, hTAS2R14 was shown to be the main bitter receptor involved in their perception, with EC50 values of 14 and 63 μM, respectively. Our analysis provided several routes for off-flavor reduction. Purification is an option because a purified, single PMF species proved to be considerably less bitter upon application in emulsified foods, due to limited solubility in the aqueous phase. A second route, also demonstrated in vivo, is C5-specific demethoxylation, in line with the finding that 5-desmethylnobiletin does not activate hTAS2R14. A third route could be the use of TAS2R14 antagonists. As a proof of principle, several antagonists, with IC50 values ranging from 10 to 50 μM, were identified. KEYWORDS: polymethoxyflavone, polyphenol, nobiletin, tangeretin, bitter receptor, citrus



INTRODUCTION So-called “functional foods”, claiming health promotion or disease prevention from new ingredients or higher doses of existing ingredients, represent a rapidly expanding market segment. Plant-derived natural antioxidants in particular are very popular fortifications in this type of products. Citrus fruits are rich in flavonoids, a class of secondary metabolites commonly known for their antioxidant activity.1 Best known are the flavanone glycosides naringin and hesperidin, but especially mandarins, Citrus nobilis or Citrus aurantium, contain significant quantities of the nonglycosylated polymethoxyflavones (PMFs; Figure 1). PMFs are a class of components claimed to have biological activities, including anti-inflammatory, anticarcinogenic, and antiatherogenic properties,2−6 and even weight loss activity.7 The blood cholesterol lowering potential is of particular interest, because the mode of action is

presumed to be different from that of plant sterols, which are currently applied in margarines and dairy products and, hence, might create an additive effect if combined with the sterols. PMFs have been claimed to inhibit synthesis and secretion of cholesterol,8 whereas sterols inhibit intestinal uptake. However, commercial PMF preparations exhibit a bitter taste, which could hamper product acceptance. More knowledge about the bitterness of individual PMFs could help to develop functional food products with the lowest possible bitterness, for example, by removing the most bitter components from the raw materials and/or by selecting the raw materials with the lowest bitterness. Masking of the bitterness, for example, by blockers of the receptors on the tongue, or conversion of the most bitter into less bitter PMF components are alternative options to improve palatability. In this respect a patent describing the conversion of PMFs into desmethyl-PMFs has been filed by Firmenich,9 but does not describe the influence on the bitterness of the PMFs. Li et al.10 also describe demethylation, not in the context of sensory effects but showing a marked increase of (anticarcinogenic) bioactivity. The aim of this study was first to investigate if non-PMF components present in the commercial raw materials contribute significantly to the bitter taste, which would allow reduction of the bitterness via PMF purification. Subsequently, the relative bitterness of the individual components was compared. This was performed by preparative HPLC separation and sensory evaluation of the obtained fractions with a trained panel. Complementary information was obtained via taste receptor assays. Among the 25 human bitter receptors, Received: Revised: Accepted: Published:

Figure 1. Structures of some polymethoxylated flavones (PMFs). © 2016 American Chemical Society

2619

December 10, 2015 March 2, 2016 March 2, 2016 March 2, 2016 DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626

Article

Journal of Agricultural and Food Chemistry

Initially, isocratic water/ethanol 45:55 was used to obtain fractions (from sample A) for the bitterness evaluation of the PMFs in water. For the fractionation of sample B, the method was modified using water/methanol 25:75 instead of ethanol to allow higher sample load. A 0.9 mL sample was injected in a flow of 10 mL/min, using UV detection (210, 280, and 330 nm). Fractions were collected in 20 mL vials and recombined covering the UV peaks in the chromatogram; the solvent was evaporated to dryness with the rotavapor (R114, Büchi, Switzerland), operated at 40 °C. For sample B, the last traces of methanol were removed by purging with nitrogen gas overnight and repeated freeze-drying from an aqueous solution. Analysis of Isolated Single PMFs by Analytical HPLC-MS. The purity and recovery of the fractions from the preparative HPLC step were measured with the analytical HPLC-ion-trap-MS system: column, Phenomenex Fusion RP, 4.6 × 250 mm, particle size = 4 μm; oven, 35 °C; eluents, A = water/methanol 50:50 and B = methanol; flow, 1 mL/min; gradient, %B 1 min, 0%, 20 min, 100%, 30 min, 100%, 32 min, 0%, 42 min, 0%; detection, UV 210, 280, and 330 nm; injection volume, 10 μL. Ion-trap MS settings were as follows: ionization, positive electrospray (ES+); cap exit voltage, 100 V; acquisition range, 100−1000 Da; mode, ultrascan; number of averages, 40; post column split ratio, MS/waste = 1:5. Solubility of Nobiletin and Tangeretin in Water and MCT Oil. PMFs have limited solubility in water and also in oil. Values for triglyceride oil have been reported in the range of 2000−3000 mg/L.14 To obtain an estimation of the solubility, individual solutions of nobiletin, tangeretin, and commercial mixtures were made in MCT oil of 60 °C at a range of concentrations from 4000 to 20000 mg/L. The solutions were cooled and stored for 5 weeks at room temperature to observe the recrystallization over time. To test the solubility of the individual PMFs in water, saturated solutions were kept overnight at room temperature and 60 °C. The next day the dissolved PMF concentrations were measured with HPLC-UV 330 nm against a standard of sample A. The PMFs were found to be chemically stable for days in water and oil solutions up to a temperature of 100 °C. Demethylation of PMFs. A method of demethylation of PMFs at the 5-position by heating in an organic solvent/hydrochloric acid mixture is described in a patent9 and in a scientific publication.10 One gram of sample B was dissolved in 150 mL of food grade ethanol, 20 mL of concentrated hydrochloric acid was added, and the solution was boiled under reflux for 1 week, after which the ethanol and hydrochloric acid were gently evaporated. The removal of hydrochloric acid was checked with a piece of pH-indicative paper in the vapor above the flask. As long as it turned red, the evaporation was continued. To prevent excessive spattering, a few milliliters of ethanol was added a few times. Sensory Evaluations in Water. Panel Training. In water the solubility of sample A is about 50 mg/L. This appeared too low for sensory evaluation. At least 100 mg/L is needed to clearly perceive the bitterness and be able to score samples with reasonable accuracy on a scale. To reach the higher concentrations needed, the sample was dissolved in water at 80 °C overnight, after which the sample is stable for at least a day at room temperature without precipitation (metastable solution), under the condition that all of the glassware was cleaned thoroughly and sonicated for 15 min with ultrapure water to remove all particles before the sample was dissolved. A panel (seven males, five females, and aged 35−60 years) was selected on the basis of basic taste recognition and trained using ranking tests and scoring tests using solutions of sample A, at concentration levels ranging from 0 to 200 mg/L. To all samples was added one drop of yellow colorant E104 to give all samples the same color. The samples were evaluated using scoring against references with scores of 3 and 8 corresponding to 100 and 200 mg/L of sample A, using the metastable solutions mentioned earlier. Blind scoring was followed by discussion and retasting. The final, considerably improved, performance of the panel is illustrated in the dose−response study of Figure 3A. The preceding toxicological assessment indicated low risk;

the ones responding to PMFs were identified, and the efficacy of the various PMF species as receptor agonists was determined. The receptor assay was also used to identify antagonists in a screening of a library of 2500 chemical compounds. On the basis of the results, various routes to reduce the bitterness of PMFs in food applications are discussed and elaborated.



MATERIALS AND METHODS

Chemicals. Ultrapure water was obtained from a Barnstead Nanopure Diamond water purification system. Food grade ethanol (Ph Eur) and hydrochloric acid (Ph Eur) were obtained from Merck, Darmstadt, Germany; methanol (Ph Eur), dimethyl sulfoxide (D8418), probenecid (P8761), and doxycyclin (D9891) were from Sigma-Aldrich (Steinheim, Germany). Tween-60 was from Kerry Bioscience, Zwijndrecht, The Netherlands, and yellow colorant E104 from Wusitta Backzutaten, Germany. Fluo-4 AM (F14202), hygromycin B (10687-010), geneticin (10131-027), zeocin (R25001), and blasticidin (R210-01) were all from Life Technologies Invitrogen, Bleiswijk, The Netherlands (now Thermo Fisher Scientific). Materials. Two PMF preparations, obtained from Chengdu SMP Phyto Extracts Corp (Chengdu, China), were purified by stirring for 1 h at 20 °C with three parts (w/w) of ethanol, filtering over a 2 μM filter, washing with cold ethanol (0 °C), and overnight drying at 50 °C. The composition of these samples is given in Table 1. The materials

Table 1. Composition (Percent) of Raw Materials (Samples A and B) Purified with Ethanol Extraction PMF component

sample A

sample B

sinensetin nobiletin desmethyl-nobiletin tangeretin pentamethoxyflavones heptamethoxyflavones OH-tetramethoxyflavones OH-pentamethoxyflavones other flavones

0.2 81.3 nd 10.1 0.3 0.0 0.1 0.3 1.0

0.6 55.8 4.3 33.4 0.2 0.1 0.5 0.4 0.7

total flavones

96.3

96.0

contain mainly nobiletin and tangeretin, as identified via LC-MS and quantified on the basis of LC-UV 330 signals. Potassium caseinate was obtained from DMV International, Veghel, The Netherlands, Sisterna SP70 sucrose ester from Brenntag Specialties, Loosdrecht, The Netherlands, and guar gum from Danisco, Denmark. Medium-chain triglyceride (MCT) oil was from Chempri BV, The Netherlands, and was deodorized via steam-stripping. HEK293 T-Rex Flp-In cells and plasmids pcDNA4 and pcDNA5/FRT were all from Life Technologies Invitrogen. DMEM (BE12-604F) and FBS tetracyclin-free (DE14830F) were obtained from Lonza, Verviers, Belgium. For the identification of bitter receptor blockers, a chemical compound library from Specs (Zoetermeer, The Netherlands), referred to as the World Diversity Set, was used. Tyrode’s buffer (140 mM NaCl, 5 mM KCl, 10 mM glucose, 1 mM MgCl2, 1 mM CaCl2, and 20 mM Hepes, pH 7.4) with 0.5 mM probenecid was used for dilution of compound− DMSO stock solutions and for calcium imaging assays. Preparative HPLC(-MS). PMFs can be separated and analyzed by RP-HPLC-(MS).11−13 The HPLC-MS equipment used in this study consists of a combined Agilent 1100 preparative-scale HPLC and an Agilent 1200 analytical scale HPLC, coupled to an Agilent 6320 iontrap mass spectrometer. The Phenomenex Hydro-RP 20 × 250 mm column, with a particle size of 4 μm, was kept at 35 °C using a Polaratherm Semiprep 9020 oven. 2620

DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626

Article

Journal of Agricultural and Food Chemistry

In Vitro Assessment of hTAS2R Activation by Intracellular Calcium Release. Activation of bitter taste receptors expressed in HEK293 cells leads to release of intracellular Ca2+.16 This was measured using the fluorescent calcium dye Fluo-4 AM (2.5 μM) in a Flexstation II 384 or Flexstation III (Molecular Devices Corp., Sunnyvale, CA, USA) for 90 s (excitation, 485 nm; emission, 520 nm). The expression of hTAS2Rs in HEK293 cells, the maintenance of the cells, and the measuring procedure were performed as reported earlier.17 Stock solutions of test compounds were prepared in DMSO and diluted to the appropriate concentration in Tyrode’s buffer, not exceeding a DMSO concentration of 1% (v/v). The signals from noninduced cells, not expressing the bitter receptors, were always measured in parallel to verify the specificity of receptor activation. Compounds that gave high nonspecific signals with noninduced control cells were excluded from further testing. Data were analyzed with SoftMax Pro 5.4 software (Molecular Devices Corp.). The fluorescence values (ΔF), as measure of receptor activity, were calculated by subtracting the baseline fluorescence (F0) prior to compound addition from the maximum fluorescence (Fmax) after addition of the bitter compounds. Dose−response curves were generated with Graph Pad Prism software (version 4 for Windows, Graph Pad Software, San Diego, CA, USA). For the identification of TAS2R14 antagonists, a screening was performed with 2500 compounds from the Specs World Diversity Set. Compounds (100 or 200 μM) were tested in combination with 100 μM PMF sample B. Potential inhibitors were retested in more detail in a second screening. Specificity was assessed by comparing the effect of the antagonists on a set of control receptors, TAS2R39, and the endogenous β-adrenergic and purinergic receptors.18

studies were subject to ethical review by the Internal Ethical Committee of Wageningen University. Bitterness Scoring of HPLC Fractions. The evaporated fractions of sample A were recombined in water in such a way that the concentrations of the individual PMF components match the concentrations in a 200 mg/L (metastable) solution of sample A. This means that the nobiletin fraction contained 163 mg/L component, and the tangeretin fraction contained 20 mg/L component. The concentrations of the components in the other fractions were much lower. The HPLC fractions were evaluated using scoring against references with scores of 3 and 8 corresponding to 100 and 200 mg/L of sample A dissolved in water (metastable solution). Five milliliters of the fractions was offered to the panel members in three-digit coded glass tubes in random order, in three sessions on different days. The order of presentation of the samples was different for each panel member, and so was the coding in each session. Besides fractions containing individual PMF components, also a reconstituted sample was prepared in which all fractions were pooled together. This sample was used to check if all the bitterness of the raw material before HPLC is recovered from the column. Bitterness of Nobiletin Relative to Tangeretin. A range of increasing concentrations of nobiletin was evaluated against a fixed concentration of 36 mg/L of tangeretin (saturated) in water at 60 °C. The concentrations of nobiletin were 1.5, 2.0, 2.5, 3.0, 3.5, and 4.0 times the concentration of tangeretin, which correspond to 54, 72, 90, 108, 126, and 144 mg/L respectively. The panelists had to choose which nobiletin concentration had a bitterness equal to the 36 mg/L tangeretin solution. Thirty-six untrained assessors performed the test. Sensory Evaluations in Emulsion. Panel Training for Ranking and Scoring Test. Sample B was taken up in MCT oil at elevated temperature to increase the speed of dissolution. Emulsions were made using 2.5% of the MCT oil solution and 0.4% sucrose ester emulsifier, using a Silverson homogenizer. A range of PMF concentrations was made including 0, 8000, 16000, 24000, and 32000 mg/L sample B in the oil (later it appeared that these concentrations are metastable, i.e., higher than the equilibrium solubility). The concentrations in emulsion were 0, 200, 400, 600, and 800 mg/L, respectively. The samples were coded and colored as described. The panel members were asked to first rank the samples and then give a score between 0 and 12, using references 3 and 8, which corresponded to concentrations of 16000 and 32000 mg/L of sample B in the oil, respectively. The training was done twice on different days, the second time using concentrations between the earlier mentioned concentrations. Bitterness of Nobiletin and Tangeretin in Emulsion. Fifteen percent MCT oil-in-water emulsions were made with 2.5% potassium caseinate as emulsifier and 0.1% guar gum. Reference standards of 3 and 8 for scoring of the bitterness were made from sample B at concentrations of 5000 and 10000 mg/L in MCT oil. The samples were remixed by the individual panel members just before tasting because of creaming of the samples. Panel members were asked to determine the most bitter sample of each pair and to score both samples of each pair from 0 to 12 against the references 3 and 8. Two sessions were held on different days with different sample codes, resulting in six 2AFC (two-alternatives forced choice) tests15 per person. Bitterness of a Single PMF or Demethylated PMFs versus a PMF Mixture. Saturated solutions of pure nobiletin and PMF sample B were prepared at 80 °C. Emulsions were prepared using the MCT suspensions at a level of 10% in deionized water containing 0.1% Tween-60 and homogenized with the Silverson homogenizer for 4 min at 4000 rpm and 1 min at 8000 rpm. The sample with pure nobiletin was adapted with a few drops of colorant E104 to match the color of the other sample. Evaluation was as described above, against references of the same concentration. Panel members were asked to stir the samples well before tasting because they contained undissolved PMF material. Demethylated sample B was compared to the original sample B following the same procedure.



RESULTS AND DISCUSSION Activity-Guided Fractionation. PMFs have been reported to be bitter,19 but it was not clear to what extent the overall bitterness of the commercial preparations is caused by the PMFs only or also by impurities. The relative bitterness of the various PMF species was also unknown. To assess the contribution of the individual PMF components, PMF samples A and B were fractionated by preparative RP-HPLC, following an LC-Taste procedure as described before.20 For sensory evaluation of the fractions, eluents based on ethanol were preferred over those made with methanol or acetonitrile, which are better and more common HPLC solvents. Figure 2A shows that PMF preparations can be well separated into individual components via reversed phase HPLC using ethanol. The loading, however, was restricted to 10 mg of extract per run. Combining 30 injections allowed us to obtain sufficient material of the individual PMFs for receptor studies, also for desmethylnobiletin, which is not commercially available, and for the sensory evaluation of the fractions. In ranking and scoring tests, it appeared that about 30 evaluations per sample were necessary to get good scoring results with acceptable confidence intervals. A good correlation was obtained between PMF concentration and bitterness (see Figure 3A). Despite the thorough training, our panel could not distinguish, however, 0 and 50 mg/L PMF, possibly related to the common tendency of tasters to avoid end-of-scale scores. The bitterness of HPLC fractions of sample A was evaluated at the concentrations the components have in the original sample dosed at 200 mg/L (Figure 3B). To a large extent the bitterness is related to nobiletin, which makes up >80% of the mass. Its bitterness is slightly lower than, but not significantly different from, the bitterness of the reconstituted sample, indicating that the contribution of the other PMFs to the bitterness is low at the concentrations they are present in this sample. 2621

DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626

Article

Journal of Agricultural and Food Chemistry

injections yielded 2.3 g of nobiletin and 1.4 g of tangeretin from 4.1 g of starting material. The recovery and purity, as based on LC-MS analysis, were close to 100%. The bitterness of tangeretin at a fixed concentration was compared with a set of nobiletin solutions of various concentrations. The samples were presented at elevated temperature to allow a concentration of the poorly watersoluble tangeretin that is perceived as clearly bitter. Because preliminary data suggested a large variation between individuals, a large group of untrained tasters (n = 36) was asked to identify the equi-bitter nobiletin solution. It appears that tangeretin on a weight basis is perceived as 2−3 times more bitter than nobiletin on average (Figure 4) in this concentration range.

Figure 2. Preparative HPLC chromatogram of (A) sample A, using water/ethanol = 45:55 eluent, and (B) sample B, using water/ methanol = 25:75 eluent.

In fact, nobiletin is the only PMF showing significant bitterness at the actual concentration in this sample. The scores of the other fractions are not significantly different from those obtained from pure water. The concentrations in these fractions (e.g., tangeretin is only 19 mg/L, the others even lower) are apparently below the threshold level for these components, as suggested by Figure 3A. The correlation of the PMF concentration in the fractions and the bitterness indicates that non-PMF impurities play a minor role in the overall bitterness of the PMF extracts. Bitterness of Tangeretin and Nobiletin Compared. For further evaluation with sensory panels the PMF separation was adapted, changing from ethanol- to methanol-based eluents, combined with multiple freeze-drying steps of the fractions to fully remove the solvent before tasting. Due to the higher solubility of the PMFs in the water/methanol eluent compared to water/ethanol, better separation efficiency, and lower backpressures, the sample load can be much higher. The separation between desmethylnobiletin and tangeretin is sacrificed to get a much higher efficiency (see Figure 2B). For sample B only 23

Figure 4. Bitterness of tangeretin relative to that of nobiletin in water at 60 °C. Nobiletin was tasted at a range of concentrations and its bitterness compared to that of a fixed concentration (36 mg/L) of tangeretin. Panelists had to indicate the nobiletin concentration at which bitterness was equal.

Application of the PMF extracts is envisaged in margarines and other emulsified oil−water products. Bitterness assessment of the individual PMFs was therefore extended to emulsions. Several (oil in water) emulsion systems were evaluated, with a variety of oil and emulsifier types. Some of these ingredients introduced a distractive off-flavor or trigeminal effect, obscuring the bitterness; others, such as Tween-60, reduced the bitterness in a neutral way, leading to lower panel sensitivity. Eventually

Figure 3. (A) Bitterness of an aqueous solution of a PMF mixture (sample A) as a function of dosage. (B) Bitterness scores of fractions of sample A (see Figure 2A). The reconstituted sample is made of all fractions pooled together. The blind control is a 100 mg/L solution of sample A, which should score 3. The error bars represent a confidence interval of 95%. 2622

DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626

Article

Journal of Agricultural and Food Chemistry

Figure 5. (A) Bitterness of a PMF mixture (sample A) in an oil-in-water emulsion as a function of dosage. (B) Bitterness scores for nobiletin and tangeretin at 1200 mg/L in the emulsion: (left pair of bars) whole panel; (middle pair) panelists sensitive to tangeretin; (right pair) panelists insensitive to tangeretin. The error bars represent a confidence interval of 95%.

Figure 6. Bitter receptor activation by PMFs. Screening of HEK293 cells, each stably expressing one of the 25 hTAS2Rs, toward activation by (A) tangeretin (250 μM), (B) PMF mix (250 μM), and (C) nobiletin (250 μM). Numbers correspond with TAS2R genes; nomenclature according to the HUGO Gene Nomenclature Committee. Noninduced cells were used as negative controls (gray background). (D) Dose−response plot for the activation of TAS2R14 by nobiletin (Nob) and tangeretin (Tan). Data are from a representative experiment with averages of samples measured in duplicate. Error bars depict the SD.

water, proximal to the receptors, largely determines the perceived bitterness, the higher specific bitterness of tangeretin (Figure 4) is completely canceled out here by its lower partitioning into the water phase. Hence, in practice, there will be no benefit of using one over the other in food products. In the 2AFC test preceding the scoring, half of the panelists consistently indicated, in six replicates, either tangeretin or nobiletin as being highest in bitterness. In line with the tasting in water there appears to be a large variation between individuals in the perception of the relative bitterness of the two major PMFs. In fact, two groups can be discriminated (Figure 5B), giving a similar score to the nobiletin emulsion but a very different score to the tangeretin emulsion, either significantly lower or significantly higher than that of the nobiletin emulsion. According to the dose−response relationship of Figure 5A, the difference between the groups equals a 3fold difference in concentration. This mixed population of subjects may be related to genetic variation in the bitter receptor(s) involved.

15% MCT oil emulsified with 2.5% potassium caseinate and 0.1% guar gum was found to be optimal. A concentration of purified individual PMFs in the oil of around 8000 mg/L, resulting in a level of 1200 mg/L in the emulsion, was needed to perceive a clear bitterness. This 8000 mg/L level is very close to the maximal solubility of both tangeretin and nobiletin in MCT oil that we measured (data not shown). Figure 5A shows that the sensory panel is well able to discriminate different concentrations of PMF extract in a rankand-score test. Despite the lower number of replicates, the standard deviations are only slightly larger than for aqueous solutions (Figure 3A), and at low dosages discrimination seems to be even somewhat better. Tasting of pure nobiletin and tangeretin pointed out that these PMF species in an emulsion are perceived as equally bitter (Figure 5B). At first sight this seems to contradict the evaluations in water above, but it should be realized that the solubility of nobiletin in water, around 50 mg/L at ambient temperature, is 2.5 times higher than that of tangeretin, being 19 mg/L. As the concentration in 2623

DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626

Article

Journal of Agricultural and Food Chemistry Receptor Activation by the Individual PMFs. Of all five established taste qualities, the perception of bitterness is probably the most complex, involving no fewer than 25 different receptors of the T2R or TAS2R family. The large set of receptors is presumably needed to cover the wide variety of chemical structures evoking a bitter taste, including xanthenes such as caffeine, peptides, and terpenes as well as flavonoids and many more.21 The bitter receptor family, discovered in 2000,22,23 is still only partially deorphanized.24 Some bitter receptors are very broadly tuned, whereas others recognize only one or a few ligands of very similar structure.24 We used a cellbased assay to identify the bitter receptor(s) involved in the recognition of a PMF extract and purified tangeretin and nobiletin, screening 25 human bitter receptors expressed in HEK293 cells (only 24 are shown in Figure 6A−C). Robust increases in intracellular calcium levels were observed only with bitter receptor TAS2R14, especially when stimulated with tangeretin or with the PMF extract, as shown in Figure 6A−C. The TAS2R14 response to nobiletin was lower. Nobiletin and the PMF extract also slightly activated some other receptors, TAS2R7, TAS2R39, TAS2R40, and TAS2R46. To determine the efficacy and potency of the two pure PMF species for TAS2R14, which is expected to be related to their relative bitterness, a dose−response curve was generated. Due to the lower solubility at high concentrations, the dose− response curve was not fully complete and the EC50 values had to be partly based on extrapolation. As shown in Figure 6D, half-maximal activation (EC50) is reached at about 14 μM for tangeretin and 63 μM for nobiletin. Maximum efficacy was reached at approximately 250−500 μM for both compounds. As can be derived from Figure 6D, the activation of 90 μM (36 mg/L) tangeretin is matched by that of around 300 μM (120 mg/L) nobiletin. Hence, an approximately 3−4 times higher bitterness on weight basis would be expected, whereas in the previous paragraph a difference of a factor of 2−3 was observed in sensory evaluation in this PMF concentration (Figure 4). Our results indicate that measured receptor activation and sensory assessment of the relative bitterness are in quite good agreement. The two PMFs were also tested in a range of concentrations with four other, less activated, bitter receptors: TAS2R7, TAS2R39, TAS2R40, and TAS2R46. Apart from TASR14, these receptors were activated only by high concentrations of nobiletin and not by tangeretin, as illustrated in Figure 7, indicating that TAS2R14 is the main receptor for these PMFs. To obtain information about the relationship between chemical structure and bitter receptor activation, a series of other flavones and flavanones was evaluated for activation of TAS2R14. Some of these components, sinensetin and 5desmethylnobiletin, are present at low levels in the commercial PMF extracts investigated. Sinensetin activated the receptor only at the highest concentrations tested (up to 80 μM), indicating that sinensetin is not as potent as nobiletin or tangeretin and suggesting that it tastes bitter only at higher concentrations. Remarkably, after correction for control cells, 5desmethylnobiletin was found not to be an activator of TAS2R14 (Figure 8). Two other, commercially available, polymethoxyflavones and -flavanones were also tested, showing a variety of affinities to the receptor. The limited solubility of the compounds did not allow testing at higher concentrations. Therefore, a full dose−response curve could not be obtained, preventing the calculation of EC50 values.

Figure 7. Dose−response plot for the activation of hTAS2R7 (●), hTAS2R39 (◆), hTAS2R40 (▲), and hTAS2R46 (▼) by nobiletin. TAS2R14 (■) responses are included for reference. Solid symbols with solid lines correspond with TAS2R-receptor containing cells and open symbols with dotted lines with cells containing no receptor (NR). Error bars depict SD.

Bitterness Reduction of PMF-Fortified Foods. The chemical, sensory, and biological analysis indicated that the bitterness of commercial PMF ingredients resides in the active compounds. Nevertheless, the results provided several leads for bitterness reduction of foods, in particular, emulsified foods, fortified with PMFs. Purification to Single PMF Compounds. An approach for bitterness reduction could be based on the finding that PMF mixtures are perceived as considerably more bitter than the single molecular species, dosed at the same or even higher concentration in an emulsion system (compare Figure 5). This is explained by the low solubility of PMFs in water. If we assume that the individual compounds do not affect each other with respect to their water−oil partitioning, the total concentration of PMF in the water phase of an emulsified product will be much lower when a pure single PMF component is used instead of a mixture of PMFs at the same total PMF concentration on product. Hence, the water phase, which is in direct contact with the receptors, will taste less bitter. Especially in a margarine, where the amount of PMF will be high (0.3−1.0%) to be effective, in addition to nobiletin and tangeretin, even the minor PMFs will contribute considerably to the bitterness. The concentration of these minor PMFs in the water phase will increase to their saturation level as the dose of the mixture is increased. According to this hypothesis, separation into single, pure, PMF species could be a route to bitterness reduction. To assess this option pure nobiletin and sample B were directly compared in a 2AFC test followed by a scoring of the bitterness. To simulate spread conditions, the total PMF dosage was as high as 0.33% on emulsion, a large fraction of which remained undissolved. The results of Figure 9A show that the PMF mixture is indeed far more bitter than the pure nobiletin. Moreover, the bitterness of the nobiletin sample is the same as in Figure 5B, despite the almost 3 times higher dosage. Apparently, the bitterness does indeed not increase further when the saturation level is exceeded, and the undissolved fraction does not contribute to the bitterness. The data indicate that highly purified single PMFs have a distinct advantage over mixtures in terms of perceived bitterness in practical situations when the solubility will be exceeded. Such purified single PMFs can be very highly dosed without increasing the bitterness, which is not the case for the original mixture (compare panels A and B of Figure 9). It should be realized, however, that at 2624

DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626

Article

Journal of Agricultural and Food Chemistry

Figure 8. Activation of hTAS2R14 by various polymethoxyflavones and -flavanones. Solid symbols with solid lines correspond with hTAS2Rreceptor containing cells (A) and open symbols with dotted lines with cells containing no receptor (B): (▲) tangeretin; (■) nobiletin; (▼) 5,7dihydroxy-6,8,3′-trimethoxyflavone; (●) 3′-hydroxy-5,6,7-trimethoxyflavanone; (◆) 5-desmethylnobiletin; (×) sinensetin.

demethylation is an effective route to bitterness reduction of PMFs. It should be mentioned, however, that the difference in specific receptor activation is not necessarily the (only) explanation of the lower bitterness. It may be that differences in solubility between PMFs and their demethylated counterparts play a role. This could be sorted out by experiments at equal concentration in water. The use of hydrochloric acid and ethanol is allowed in foods, enabling the conversion in a foodgrade process, also at industrial scale. It should of course be validated that the conversion does not lead to loss of health effects, but the available data suggest a higher rather than a lower physiological effect.10 Bitterness Masking by PMF Antagonists. A third route to bitterness reduction could be via TAS2R14 receptor antagonists. From a screening with 2500 synthetic compounds, several TAS2R14 receptor antagonists were obtained that effectively inhibited TAS2R14 activity, with IC50 values ranging from 10 μM to about 50 μM. The set of compounds was too low and too diverse for an in silico screening, and their nature also did not allow validating their effect in vivo before extensive safety evaluation; therefore, further experiments were not carried out. In that respect, a more preferred solution would be to use natural compound libraries for screening, as natural antagonists will greatly enhance the chance of approval. The fact that 5desmethylnobiletin is structurally very similar to nobiletin, but does not activate TAS2R14, could make it a candidate antagonist. This proof of principle confirms, however, the possibility of identifying specific TAS2R14 antagonists in vitro, although their effect in vivo could still be influenced by other aspects. For example, whereas our heterologous receptor assay is similar to those routinely used for bitter receptor deorphanization studies by others with good correlations to in vivo effects (e.g., TAS2R38 activation by 6-n-propylthiouracil, PROP),25 it cannot be excluded that some receptors which are activated in vivo might not turn up in vitro. Furthermore, in vitro expression levels of bitter receptors might not always be correlated with their expression in vivo. Finally, Brockhoff et al.26 showed that bitterness perception is quite complex and, in fact, the result of a combination of activation and suppression by a compound on all 25 receptors. Our work shows that the combination of sensory and receptor-based studies allows us to define several routes for bitterness control, based on physical and chemical as well as biological principles. It should also be possible to combine the three methods to further minimize the off-taste. The approach is not specific for PMFs and can be applied to bitterness reduction of “neutraceuticals” in general or, even more broadly, to reduction of any possible type of off-flavor encountered. This

Figure 9. (A) Bitterness of nobiletin and PMF mixture (sample B) in emulsion at 3300 mg/L dosage. (B) Bitterness of native and demethylated PMF mixture (sample B) in emulsion at 1075 mg/L. The error bars represent a confidence interval of 95%. Ten percent MCT oil emulsions were used with 0.1% Tween-60 as emulsifier.

industrial scale execution of such purifications may not be trivial. PMF Demethylation. The recombinant receptor assays showed that 5-desmethylnobiletin, in contrast to the major PMF species nobiletin, is not effective in TAS2R14 activation. This suggests that bitterness reduction of commercial PMF mixtures might be achieved by demethylation at the C5 position, assuming that the conversion affects the bitterness of other PMF species similarly. Isolation of 5-desmethylnobiletin from the commercial mixtures is not a feasible option due to the low natural abundance of this PMF. PMFs can be converted into 5-desmethyl-PMFs (leaving a hydroxy group at the C5position) by boiling in ethanol in the presence of hydrochloric acid. The process is very simple, although quite slow,9,10 taking a full week for conversion. One gram of PMF sample B was subjected to this procedure, and analysis of the product with LC-UV/MSn showed a full conversion of the main PMF components to the corresponding desmethyl-PMFs, the conversion product of nobiletin with the correct molecular ion mass eluting at the position of the natural 5desmethylnobiletin in the original sample. The fragmentation results in MS3 and MS4 confirmed that the demethylation, both for nobiletin and for tangeretin, had indeed taken place in the A-ring, the corresponding fragment having lost 14 Da like the parent ion. The demethylated PMF mixture was compared with the starting material via sensory evaluation in the MCT oil/water emulsion system. Figure 9B shows that the bitterness perceived by the trained panel was indeed considerably lower after demethylation, in line with our intention and in line with the expectations based on the receptor studies. This shows that 2625

DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626

Article

Journal of Agricultural and Food Chemistry is essential for the commercial success of “functional foods”, because all consumer evidence indicates that food products that do not have an appealing taste will not be accepted, whatever the health claim made.



spectrometry and liquid chromatography/nuclear magnetic resonance as complementary analytical techniques for unambiguous identification of polymethoxylated flavones in residues from molecular distillation of orange peel oils (Citrus sinensis). J. Agric. Food Chem. 2006, 54, 274− 278. (14) Li, Y.; Zheng, J. K.; Xiao, H.; McClements, D. J. Nanoemulsionbased delivery systems for poorly water-soluble bioactive compounds: Influence of formulation parameters on polymethoxyflavone crystallization. Food Hydrocolloids 2012, 27, 517−528. (15) Meilgaard, M.; Civille, G. V.; Carr, M. S. Sensory Evaluation Techniques, 3rd ed.; CRC Press, 1999; pp 100−102. (16) Chandrashekar, J.; Mueller, K. L.; Hoon, M. A.; Adler, E.; Feng, L.; Guo, W.; Zuker, C. S.; Ryba, N. J. P. T2Rs function as bitter taste receptors. Cell 2000, 100 (6), 703−711. (17) Roland, W. S. U.; Vincken, J. P.; Gouka, R. J.; van Buren, L.; Gruppen, H.; Smit, G. Soy isoflavones and other isoflavonoids activate the human bitter taste receptors hTAS2R14 and hTAS2R39. J. Agric. Food Chem. 2011, 59, 11764−11771. (18) Atwood, B. K.; Lopez, J.; Wager-Miller, J.; Mackie, K.; Straiker, A. Expression of G protein-coupled receptors and related proteins in HEK293, AtT20, BV2, and N18 cell lines as revealed by microarray analysis. BMC Genomics 2011, 12, 14. (19) Drewnowski, A.; Gomez-Carneros, C. Bitter taste, phytonutrients, and the consumer: a review. Am. J. Clin. Nutr. 2000, 72, 1424− 1435. (20) Reichelt, K. V.; Peter, R.; Paetz, S.; Roloff, M.; Ley, J. P.; Krammer, G. E.; Engel, K.-H. Characterization of flavor modulating effects in complex mixtures via high temperature liquid chromatography. J. Agric. Food Chem. 2010, 58, 458−464. (21) Meyerhof, W.; Born, S.; Brockhoff, A.; Behrends, M. Molecular biology of mammalian bitter taste receptors. A review. Flavour Fragrance J. 2011, 26, 260−268. (22) Adler, E.; Hoon, M. A.; Mueller, K. L.; Chandrashekar, J.; Ryba, J. P.; Zuker, C. S. A novel family of mammalian taste receptors. Cell 2000, 100, 693−702. (23) Matsunami, H.; Montmayeur, J.-P.; Buck, L. B. A family of candidate taste receptors in human and mouse. Nature 2000, 404, 601−604. (24) Meyerhof, W.; Batram, C.; Kuhn, C.; Brockhoff, A.; Chudoba, E.; Bufe, B.; Appendino, G.; Behrens, M. The molecular receptive ranges of human TAS2R bitter taste receptors. Chem. Senses 2010, 35, 157−170. (25) Bufe, B.; Breslin, P. A.; Kuhn, C.; Reed, D. R.; Tharp, C. D.; Slack, J. P.; Kim, U. K.; Drayna, D.; Meyerhof, W. The molecular basis of individual differences in phenylthiocarbamide and propylthiouracil bitterness perception. Curr. Biol. 2005, 15, 322−327. (26) Brockhoff, A.; Behrens, M.; Roudnitzky, N.; Appendino, G.; Avonto, C.; Meyerhof, W. Receptor agonism and antagonism of dietary bitter compounds. J. Neurosci. 2011, 31, 14775−14782.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jafc.5b05833. Analysis of the effectiveness of demethoxylation of PMF sample B (PDF)



AUTHOR INFORMATION

Corresponding Author

*(A.M.B.) E-mail: [email protected]. Fax: +31 (0) 10 4605236. Phone: +31 (0)10 4605644. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS We thank Leo Brunia and Rob Diks for the ethanol extractions. REFERENCES

(1) Pietta, P.-G. Flavonoids as antioxidants. J. Nat. Prod. 2000, 63, 1035−1042. (2) Bok, S. H.; Lee, S. H.; Park, Y. B.; Bae, K. H.; Son, K. H.; Jeong, T. S.; Choi, M. S. Plasma and hepatic cholesterol and hepatic activities of 3-hydroxy-3-methyl-glutaryl-CoA reductase and acyl CoA-cholesterol transferase are lower in rats fed citrus peel extract or a mixture of citrus bioflavonoids. J. Nutr. 1999, 129, 1182−1185. (3) Kurowska, E. A.; Manthey, J. A. Hypolepidemic effects and absorption of citrus polymethoxylated flavones in hamsters with dietinduced hypercholesterolemia. J. Agric. Food Chem. 2004, 52, 2879− 2886. (4) Schmanke, H. Polymethoxyflavone in Zitrusfrüchten: blutlipidsenkende und antikanzerogene Eigenschaften. Ernaehrungs Umschau 2008, 55, 290−295. (5) Manthey, J. A.; Bendele, Ph Anti-inflammatory activity of an orange peel polymethoxylated flavone, 3′,4′,3,5,6,7,8-heptamethoxyflavone, in the rat carrageenan/paw edema and mouse lipopolysaccharide-challenge assays. J. Agric. Food Chem. 2008, 56, 9399−9403. (6) Li, S.; Lo, C.-Y.; Ho, C.-T. Hydroxylated polymethoxyflavones and methylated flavanoids in sweet orange (Citrus sinensis) peel. J. Agric. Food Chem. 2006, 54, 4176−4185. (7) Talbot, S. M. Weight loss with citrus flavonoids. Patent Appl. US 2007/0224299 A1. (8) Kurowska, E. A.; Manthey, J. A. Regulation of lipoprotein metabolism in HepG2 cells by citrus flavonoids. Adv. Exp. Med. Biol. 2002, 505, 173−179. (9) Frerot, E. (Firmenich SA) Process to prepare flavones. Patent Appl. WO/2007/083263. (10) Li, S.; Pan, M. H.; Lai, C.-S.; Lo, C.-Y.; Dushenkov, S.; Ho, C.-T. Isolation and syntheses of polymethoxyflavones and hydroxylated polymethoxyflavones as inhibitors of HL-60 cell lines. Bioorg. Med. Chem. 2007, 15, 3381−3389. (11) Wang, Z.; Li, S.; Ferguson, S.; Goodnow, R.; Ho, C.-T. Validated reversed phase LC method for quantitative analysis of polymethoxyflavones in citrus peel extracts. J. Sep. Sci. 2008, 31, 30− 37. (12) Justesen, U.; Knuthsen, P.; Leth, T. Quantitative analysis of flavonols, flavones and flavanones in fruits, vegetables and beverages by high-performance liquid chromatography with photo-diode array and mass spectrometric detection. J. Chromatogr. A 1998, 799, 101− 110. (13) Weber, B.; Hartmann, B.; Stöckigt, D.; Schreiber, K.; Roloff, M.; Bertram, H.-J.; Schmidt, C. O. Liquid chromatography/mass 2626

DOI: 10.1021/acs.jafc.5b05833 J. Agric. Food Chem. 2016, 64, 2619−2626