If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
College of Food Science and Technology, Hunan Agricultural University, Changsha 410114, ChinaKey Laboratory of Functional Dairy, Co-constructed by Ministry of Education and Beijing Government, College of Food Science and Nutritional Engineering, China Agricultural University, Beijing 100083, China
Key Laboratory of Functional Dairy, Co-constructed by Ministry of Education and Beijing Government, College of Food Science and Nutritional Engineering, China Agricultural University, Beijing 100083, China
College of Food Science and Technology, Hunan Agricultural University, Changsha 410114, ChinaKey Laboratory of Functional Dairy, Co-constructed by Ministry of Education and Beijing Government, College of Food Science and Nutritional Engineering, China Agricultural University, Beijing 100083, China
Aroma is an important property of fermented milk, and it directly affects consumer acceptance. However, previous studies have mainly focused on analyzing the composition of aroma compounds in fermented milk in vitro, and the composition may be different from the real aroma composition that stimulates the sense of smell. Furthermore, the relationship between olfactory attributes and the release of aroma compounds was not fully understood. In this study, we selected 6 samples of fermented milk differing in aroma perception intensity based on our pretest. A descriptive sensory analysis focusing on orthonasal and retronasal olfaction of fermented milk was first conducted by semitrained panelists. Artificial saliva was mixed with the fermented milk samples and continuously stirred at 37°C for 15 s to simulate oral processing conditions. Headspace solid-phase microextraction–gas chromatography coupled with quadrupole time-of-flight mass spectrometry was applied to identify the head space composition of 6 kinds of fermented milk before and after the simulated oral processing. Twenty-five volatile compounds were identified in the fermented milks, 15 of which were predicted to have an influence on the olfactory perception of fermented milks during oral processing. Partial least squares regression analysis based on chemical and sensory data was then applied to explore the correlation between sensory perception and volatile aroma release. The results showed that oral processing greatly increased the perception of creamy aroma compounds, such as diacetyl and acetone, but did not increase the perception of dairy sour aroma compounds, such as butanoic acid and hexanoic acid. This study can help improve our understanding of the relationship between olfactory perceptions and the release of volatile aroma compounds under oral processing. It might also contribute to the design of palatable fermented milks catering to specific consumer preferences.
). Therefore, research focusing on aroma in fermented milk has received significant attention. More than 500 volatile compounds have been characterized in different types of fermented milk through many studies; however, not all these aroma compounds could be perceived, as some may not reach their threshold concentration for detection under daily consumption conditions.
Aroma compounds can be perceived by humans through the nasal cavity in 2 ways: orthonasal and retronasal olfaction. Breath plays an essential role in both perceptual approaches (
). Orthonasal olfaction occurs during sniffing, in which aroma compounds are inhaled directly from the headspace of food into the nasal cavity. Retronasal olfaction involves transportation of aroma compounds from the oral cavity to the nasal cavity. Aroma compounds within or around food first enter the mouth and are then transported to the nasal epithelium to be detected by receptors with breath during oral processing. (
). Therefore, oral processing is not only important for the ingestion and digestion of food, but also plays an important role in the retronasal olfactory perception of aromas (
Previous studies have shown that both the characteristics of fermented milk and oral conditions significantly influence the release of aroma compounds and olfactory perception. A lower release of volatile compounds has been observed in emulsion systems with higher viscosities (
). Oral conditions, such as saliva, pH, temperature, and chewing behavior, have been shown to influence the release and perception of volatile compounds (
). Many studies have focused on analyzing the composition of volatile compounds in fermented milk in vitro. However, the extraction conditions used by these studies to determine the accurate contents and types of aroma compounds were far from the actual conditions of orthonasal or retronasal olfaction (
). However, the influence of oral processing during the release of these compounds from fermented milk on their olfactory perception is not fully understood.
In this study, we selected 6 types of fermented milks with different aroma characteristics through a discrimination test and chemical analysis based on our preliminary experiments. Descriptive sensory analysis was applied by trained sensory panelists to explore the orthonasal and retronasal olfaction of fermented milk. Artificial saliva was mixed with the fermented milk samples and continuously stirred at 37°C for 15 s to simulate oral processing conditions. Headspace solid-phase microextraction–gas chromatography coupled with quadrupole time-of-flight mass spectrometry (HS-SPME–GC–Q-TOF-MS) was then applied to determine the composition of aroma compounds in the fermented milks and the release of these before and after simulated oral processing; this was followed by a principal components analysis (PCA), to identify the key aromas from each sample. Based on sensory and chemical data, partial least squares regression (PLSR) analysis was performed to explore the correlation between sensory perception and aroma release. Our findings might help understand the release of aroma compounds and olfactory perception under oral processing and contribute to the development of consumer-preferred fermented milks.
MATERIALS AND METHODS
Materials
Six kinds of fermented milk (samples A–F) were provided by the Key Laboratory of Functional Dairy (Beijing, China). The raw milk composition, manufacturing process, sugar content and pH of fermentation termination were the same among the 6 fermented milk samples [protein content, 3.1 ± 0.1% (wt/wt); fat content, 3.9 ± 0.2% (wt/wt); fermentation termination pH, 4.60 ± 0.01], but the starters were different (Streptococcus thermophilus and Lactobacillus bulgaricus for samples A, B, D, E, and F, with different ratios; S. thermophilus, L. bulgaricus, and Lactobacillus rhamnosus for sample C). The 6 samples were selected based on our previous hedonic sensory test, as they had different creamy odor preferences. Rheological tests were applied to determine the apparent viscosity of fermented milk (Supplemental Figure S1, https://doi.org/10.6084/m9.figshare.13271072.v1). Sugar, diacetyl, acetaldehyde, and δ-decalactone were either of natural or food grade, and other chemicals were of analytical grade. Diacetyl, acetaldehyde, δ-decalactone, and gastric mucin were purchased from Sigma-Aldrich (Temecula, CA). Sugar, sodium chloride, calcium chloride, potassium phosphate, potassium chloride, sodium azide, and sodium hydroxide were purchased from Adamas Reagents (Beijing, China).
Artificial saliva was prepared according to a previous report (
) and comprised 2.2 g/L gastric mucin (from porcine stomach), 0.381 g/L sodium chloride, 0.231 g/L calcium chloride, 0.738 g/L potassium phosphate, 1.114 g/L potassium chloride, 0.02% sodium azide, and trace sodium hydroxide (pH 7.0). Because the food matrix in the current study did not contain any starch, and α-amylase and mucins have no cumulative effect on aroma release (
), α-amylase was not included in the artificial saliva.
Orthonasal and Retronasal Olfaction of Fermented Milk
The orthonasal and retronasal olfaction of fermented milk was tested using descriptive sensory analysis. The descriptive sensory analysis was performed by 12 semitrained panelists (4 men and 8 women; mean age 23.3 ± 1.2 yr; all were recruited from China Agricultural University, Beijing) to evaluate the aroma perception of different fermented milk samples through orthonasal and retronasal olfaction. All participants were healthy, nonsmoking, not pregnant or lactating, not lactose-intolerant, and did not have any milk-related allergies. Subjects were instructed neither to eat nor drink anything other than water for 1 h before testing, nor to wear any scented product on the day of testing.
Descriptive analysis was conducted using an 11-point categorical scale. We chose this scale to train participants to link odor intensities with numbers. The lexicon and standard generated for the fermented milk samples included 4 aroma attributes. For all attributes except butter, 3 references were used, corresponding to points 3, 6, and 11 on the scale (Supplemental Materials, https://doi.org/10.6084/m9.figshare.13271072.v1). Four sessions of training (each lasting 1–2 h) were conducted to allow the panelists to consistently identify and score the attributes. An ANOVA of the data collected from the blind testing after training indicated that the panelists could consistently distinguish the different aroma compounds and their intensities.
Sensory assessments were conducted at the Sensory Laboratory in China Agricultural University, Beijing, within separate booths. Fermented milk (60 g of each) was served in 100-mL odorless transparent plastic cups coded with 3-digit random numbers. The samples were semisolid. The fluidity of the samples was not sufficient for drinking directly, and a spoon was needed to eat the samples. Each trial consisted of 1 sample, with 1-min intervals between trials; panelists were asked to take deep breaths or rinse mouth with water during the intervals. The fermented milk samples were stored in a freezer at 4°C. An ice box was placed in each booth, and the fermented milk samples were placed in the ice box during sensory evaluation to simulate the normal consumption temperature in day-to-day scenarios.
Sensory evaluation was divided into 2 sessions. In the first session, panelists sniffed each sample and scored the aroma based on orthonasal olfaction. In the second session, panelists were initially asked to first eat the sample, exhale through the nose after swallowing, and then finally score the aroma perception. An instruction on how the samples should be orally processed was provided before and during the test. Panelists were instructed to put a spoonful of fermented milk (approximately 15 g) into their mouths, keep their mouths closed, chew for 15 s, and then swallow the product. They could eat more if that helped them to evaluate the product. They were also instructed to breathe normally while chewing and exhale through their noses after swallowing. All evaluations were performed in triplicate for each sample, with 1 random 3-digit numbered fermented milk served per session. Six fermented milk samples were served during each evaluation. The order of sample presentation for each panelist was different and determined according to a Latin square design. Repeat measurements were performed at the same time on different days. Measurement repeats for the same sample were coded by different random 3-digit numbers, which means that each panelist evaluated 18 fermented milks with 3-digit numbers. Sensory data were collected using detailed questionnaires.
Simulated Oral Processing
Artificial saliva was added to fermented milk at a ratio of 1:10 (wt/wt) and stirred at a shear rate of 200 rpm in a water bath (JRA-2 magnetic stirring water bath, Tianlian Instrument, Beijing, China) at 37°C for 15 s (
) to simulate the residence time and mouth conditions for the oral processing of dairy drinks. Briefly, in our descriptive sensory evaluation, panelists were instructed to chew 15 g of sample for 15 s; therefore, the simulated oral processing time was set at 15 s. A relatively longer oral processing time was set to facilitate the release of flavor and aroma perception. According to a previous study, approximately 80 to 90% of the daily average salivary production is in the stimulated form, and the stimulated flow rate of saliva is, at maximum, 7 mL/min (
). Therefore, considering the salivary flow rate and oral residence time, the saliva was added at a ratio of 1:10 in this study. After simulated oral processing, the fermented milks were subjected to further characterization. It is worth noting that the mixing ratio varies depending upon the texture and oral residence time of the consumed food and might also differ during the course of oral processing, from intake to swallowing, in addition to being affected by other physiological and interpersonal factors (
). However, the dynamic profile of saliva incorporation in the food consumed is not within the scope of this study.
Headspace SPME
The aroma compounds in the 6 fermented milk samples were extracted using a manual headspace injection handle with an SPME fiber (50/30 μm DVB/CAR/PDMS, Bellefonte, PA). The fermented milk (20.0 ± 0.3 g) and sodium chloride (10.0 ± 0.2 g) were loaded into the SPME vial (40 mL, Sigma, Beijing, China). We dissolved 2-methyl-3-heptanone (internal standard) in an odorless and ultrapure solution to prepare the internal standard solution (81.1 μg/mL). This internal standard solution (80 μL) was then added to the sample vial. The vial was placed in the incubator for 15 min at 70°C. The SPME fiber was then inserted into the headspace of the extraction bottle to adsorb compounds for 30 min at 70°C. After extraction, the loaded SPME fiber was immediately injected into the injection port of the GC-Q-TOF-MS to desorb for 5 min at 230°C.
To measure the aroma compounds contributing to orthonasal and retronasal olfactory perceptions, the headspace composition of volatile compounds in fermented milk before and after the simulated oral processing were measured. For aroma compounds contributing to orthonasal perception, fermented milk (20.0 ± 0.3 g) was loaded into the SPME vial (40 mL). The vial was moved to the incubator for an incubation period of 15 min at 25°C. For aroma compounds contributing to retronasal perception, fermented milk (20.0 ± 0.3 g) was first processed by simulated oral processing and then loaded into the SPME vial (40 mL). The vial was placed in the incubator for 15 min at 37°C. Subsequently, the SPME fiber was inserted into the headspace of the extraction bottle to adsorb volatile compounds for 30 min at 37°C. After extraction, the loaded SPME fiber was immediately injected into the injection port of the GC–Q-TOF-MS instrument to desorb for 5 min at 230°C.
GC–Q-TOF-MS Analysis
A high-resolution Agilent 7250 GC-Q-TOF instrument (Agilent, Carpinteria, CA) equipped with a low-energy-capable electron ionization (EI) source was used in this study. Samples were analyzed on a DB-WAX column (30 m × 0.25 mm interior diameter, 0.25-μm film thickness; Supelco, Bellefonte, PA). Helium (99.999% purity) was used as the carrier gas, and the column flow rate was set at 1.20 mL/min (29.32 cm/s) in the splitless injection mode. The injector temperature was 230°C. The initial oven temperature was 40°C and was maintained for 2 min. The temperature was then increased to 180°C at 3°C/min, held for 2 min, increased to 220°C at 5°C/min, and finally steadied for 3 min. The mass spectrometry detection conditions were as follows: mass detector temperature, 150°C; electron impact mode, 70 eV; ion source temperature, 240°C; transmission line temperature, 250°C; and mass range m/z 40 to 450, in full scan mode.
For SPME–GC–Q-TOF-MS analysis, identification of the volatile compounds was based on matching library spectra and authentic standards. A semiquantitative method was used in this study (
). The concentrations of volatile compounds were determined using the ratio of the peak area of the odorant compound relative to that of the internal standard (2-methyl-3-heptanone) against the ratio of the concentrations of these compounds. All sample measurements were repeated in triplicate. The odor detection threshold was summarized according to other studies (
Re-investigation on odour thresholds of key food aroma compounds and development of an aroma language based on odour qualities of defined aqueous odorant solutions.
Identification and characterisation of headspace volatiles of fish miso, a Japanese fish meat based fermented paste, with special emphasis on effect of fish species and meat washing.
The SPSS 22.0 software (SPSS Inc., Chicago, IL) was used to analyze significant differences in results (P < 0.05). One-way ANOVA, followed by Duncan test, was used to analyze the differences in orthonasal and retronasal perception ratings and headspace aroma compositions before and after simulated oral processing. The analyses were performed separately (e.g., orthonasal creamy rating for sample A was compared with the orthonasal creamy rating for other samples; the headspace content of diacetyl for sample A was only compared with those of other samples before simulation). Independent t-test was used to analyze differences in certain attributes of the sample between orthonasal and retronasal perceptions as well as the difference in the headspace contents of certain aroma compounds before and after simulated oral processing. We performed PCA using XLSTAT 2019 (Addinsoft, New York, NY). Correlations between the descriptive sensory and chemical data (concentration changes of the volatile compounds) were determined by PLSR using the XLSTAT 2016 software (Addinsoft), where x variables were the chemical data and y variables were the sensory attributes.
RESULTS AND DISCUSSION
Orthonasal and Retronasal Olfaction of Fermented Milk
The comparative results of sensory evaluation among the 6 fermented milk samples are summarized in Figure 1. Scores from all the panelists were collected and analyzed. In the orthonasal olfactory perception analysis, samples A and F had the highest creamy and milk fat aroma perceptions, and sample E had the highest dairy sour perception. Samples B and C did not have a prominent orthonasal olfactory perception. In terms of retronasal olfactory perception, sample C had the strongest creamy perception, and sample F had the strongest dairy sour perception. The creamy perception of sample C was higher through orthonasal olfaction than through retronasal olfaction (P < 0.05). Dairy sour perception of sample E was higher through orthonasal olfaction than through retronasal olfaction (P < 0.05).
Figure 1Retronasal and orthonasal olfactory perception of fermented milks (A) radar icon for sensory evaluation results; (B) results from sensory evaluation. Results are mean ± SD (n = 12). a–fMean values in the same row with different letters differ in sensory evaluation score (P < 0.05). *Means of the orthonasal and retronasal perceptions of a specific attribute of the sample are significantly different (P < 0.05).
The reason behind the differences in orthonasal olfactory perceptions was that the ratios of starters applied in the processing of the different fermented milk samples were different. Retronasal olfactory perception was not consistent with orthonasal olfactory perception because the starters applied during oral processing, food matrix, and conditions of oral processing affect the release and perception of aroma compounds during mastication (
reported that the food matrix is a key factor affecting the release of volatile compounds due to the interaction between the compounds (based on their polarity) and the structure of the food matrix. In addition, different oral processing conditions, such as mastication, temperature, pH, and saliva, affect the release of volatile compounds to various degrees; mechanical movement of the oral cavity increases the release of volatile compounds, and the release of volatile compounds decreases with an increase in salivary content (
). However, the retronasal olfactory perception of volatile aroma compounds released from the fermented milk matrix under oral processing conditions is not fully understood. In this study, no significant difference was observed in apparent viscosity between fermented milk samples (Supplemental Materials, https://doi.org/10.6084/m9.figshare.13271072.v1), which indicated that the differences in their orthonasal perception were not caused by their viscosity. The orthonasal creamy perception of sample C was much higher than the retronasal creamy perception (P < 0.05), which indicated that oral processing can promote the release of aroma compounds and enhance perception. However, for sample E, the dairy sour orthonasal perception through was higher than the dairy sour retronasal perception (P < 0.05). This suggested that orthonasal olfaction might be important for the perception of acidic compounds. Furthermore, the creamy retronasal perception was higher in sample E than the creamy orthonasal perception (P < 0.05). Enhanced creamy retronasal perception may also mask dairy sour perception.
Semiqualitative Analysis of Dominant Volatile Compounds in Fermented Milk by SPME–GC–Q-TOF-MS
The compositions of volatile compounds in the 6 fermented milk samples are shown in Table 1. The contents of all aroma compounds exceeded their threshold of detection in water, which suggested that all of them had the ability to contribute to the aroma perception of fermented milk. We found that 4-penten-1-ol, diacetyl, 2-nonanone and 2-hydroxy-3-pentanone had odor perceptions of butter and hot milk, indicating that they were “creamy aroma compounds” and had creamy perception. Acetic acid, butanoic acid, hexanoic acid, heptanoic acid, and octanoic acid were identified to have sour and rancid perception. These compounds were identified as “sour aroma compounds,” contributing to the dairy sour perception of fermented milk. Differences were observed in the compositions of the 6 groups of samples. Sample C contained the highest amounts of aroma compounds, with “creamy aroma compounds” being the most abundant aroma compound type; sample B contained the lowest amounts of volatile compounds, but was rich in ketone compounds. Sample E contained the most types and highest contents of “sour aroma compounds.”
Results are expressed as mean ± SD (n = 3); ND = compound was not detected; NC = not clear; the odor detection threshold in water (μg/100 mL) and odor perception were obtained from references from VCF online (www.vcf-online.nl).
of total headspace aroma compounds in fermented milk samples
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds.
NC
NC
a–e Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds.
1 Results are expressed as mean ± SD (n = 3); ND = compound was not detected; NC = not clear; the odor detection threshold in water (μg/100 mL) and odor perception were obtained from references from VCF online (www.vcf-online.nl).
Aroma compounds determine the sensory perception of fermented milk. Acetoin and its reduction product, 2, 3-butanediol, are the main contributors to the perception of creamy aroma (
). In this study, the diacetyl content in sample C was significantly higher than that in the other samples, which could explain the strongest retronasal creamy perception of sample C. The composition of volatile compounds in fermented milk has been determined previously, with mainly nonvolatile acidic compounds, and volatile C2–C10 saturated fatty acids, such as lactic acid, pyruvic acid, and oxalic acid (
). Acids also contribute to the aroma perception of fermented milk. A small amount of acid enhances the sour perception of fermented milk, but high concentrations of some acids, such as butyric acid and valeric acid, are important contributors to poor flavor that can be perceived as “green” when fermented milk is corrupted (
). In this study, sample E had the highest content of butyric acid, which might have resulted in its weaker creamy perception, and had the strongest orthonasal dairy sour perception. Although the volatile compounds of fermented milk played a decisive role in aroma perception, the matrix of fermented milk and oral processing also greatly influenced the release and perception of volatile aroma. The effect of simulated oral processing on the release of volatile compounds in fermented milk will be discussed in the next section.
To investigate the similarities between the volatile compound compositions of the samples, we performed a PCA analysis of the results, as shown in Figure 2. The sample points A, D, and F were densely distributed, which indicated that the compositions of the volatile compounds in these 3 samples were similar. Sample C was characterized by very high contents of diacetyl, isopropanol, n-hexanol, and acetic acid (Table 1). Sample E was characterized by 2-nonanone, 4-hydroxy-3-hexanone, 2-undecyl ketone, and butyric acid, which had similar contributions to the first and the second principal components. Acidic compounds mask the perception of creamy aroma compounds, such as diacetyl and acetoin (
Effect of Oral Processing on the Release of Volatile Compounds
Oral processing affects the release and perception of aroma compounds and leads to a difference between the orthonasal and retronasal olfactory perceptions of food (
). To explore the compositions of aroma compounds that contribute to orthonasal and retronasal olfactory perceptions, we tested the headspace volatile composition of fermented milk before and after simulated oral processing. The results are shown in Table 2. The types and contents of volatile compounds measured before and after oral simulations were lower than those shown in Table 1. (total aroma compounds in fermented milk). The aroma compounds measured before and after oral processing comprised only acids, ketones, and alcohols.
Table 2Composition of headspace aroma compounds in fermented milk samples before and after simulated oral processing
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
a–e Mean values in the same row with different superscript letters differ (P < 0.05) in the contents of compounds. Results are expressed as mean ± SD (n = 3); ND = not detected
* and ** indicate that the contents of specific compounds in 1 sample were significantly different before and after simulated oral processing, *P < 0.05
The types and contents of acids and ketones were reduced to varying degrees when compared with the results in Table 1. . In sample B, only 5 and 10 types of headspace aroma compounds were detected before and after oral processing, respectively, indicating that sample B had the smallest variety of aroma compounds (12 types) among the samples. Contrarily, in sample C, 10 and 15 types of headspace volatile aroma compounds were detected before and after oral processing, respectively, indicating that sample C had the greatest variety of volatile aroma compounds (20 types) among the samples. These results indicated that not all of the volatile aroma compounds in fermented milk contributed to aroma perception, as some compounds were trapped in the food matrix or saliva. Simulated oral processing promoted the release of certain aroma compounds, as the types of aroma compounds found in retronasal olfaction (the aroma composition after simulated oral processing) were more than those found in orthonasal olfaction (the aroma composition before simulated oral processing).
Oral processing conditions, including mastication, temperature, and saliva, influence the release of aroma compounds and their retronasal olfactory perception. Mastication and oral temperature significantly promote the release of aroma compounds, while saliva may have different effects on the release of such compounds. The ions in saliva promote the release of volatiles due to the salt-out effect; however, a high salivary content may reduce the release of volatile compounds (
). However, the effect of oral processing on the release of different aroma compounds was not completely consistent due to differences in compound characteristics, such as polarity and volatility (
). We found that oral processing promoted the release of aroma compounds from the food matrix and increased the types and contents of aroma compounds released (Table 2). The most significant increase in release was that of ketone compounds. The diacetyl content in the 6 groups of samples doubled after oral processing, and 2-hydroxy-3-pentanone and acetone in sample A and that of 4-octanone in sample F, which were not found initially, were detected after simulated oral processing. This indicated that oral processing promoted the release of aroma compounds. However, the release effect of oral processing varied with the type of compound. In this study, the contents of acidic compounds (such as acetic acid, butyric acid, and caproic acid) in each sample increased after simulated oral processing, but not as much as those of ketones (i.e., diacetyl, acetone, and 4-octanone). The reason for this might be that these compounds have different volatilities and water solubilities. Due to the lack of intermolecular hydrogen bonds in ketones, they are more volatile than acids of the same molecular weight. In addition, acids are more soluble in water, and simulated oral processing may cause acids to dissolve in artificial saliva instead of volatilizing into the headspace.
Perceptual Pathway for the Key Volatile Aroma Compounds in Fermented Milk
To explore the perceptual approach for the key aroma compounds in fermented milk during consumption, we performed PLSR on the aroma compounds and sensory attributes; x variables were the category and content of aroma compounds (aroma categories were obtained from Table 2; aroma contents were obtained from Table 1), and y variables were the sensory data (Figure 3). Most x variables (aroma compound contents) and y variables (sensory attribute intensities) were loaded around the circle. The model quality (Q2 = 0.63) was higher than 0.50, which indicated that the PLSR model was well established. Dimension 1 explained 43.19% of the predictor variance (key volatile aroma compounds) and 37.24% of the response variance (aroma perception), and dimension 2 explained 29.65% of the predictor variance and 25.86% of the response variance. Overall, almost all the aroma compounds and their sensory attributes were located on the negative side of the axis for dimension 1 and the positive side of the axis for dimension 2, but most fermented milk samples were located on the positive side of the axis for dimension 1.
Figure 3Correlation matrix of instrument result and sensory perception of fermented milk samples. Blue circles represent sensory attributes, green plots represent different fermented milk samples (samples A to F), and red plots represent 15 kinds of aroma compounds. Red numbers represent different flavor components as follows: (1) cyclobutanol; (2) propanol; (3) methyl-2-buten-1-ol; (4) methyl-2-buten-1-ol; (5) 2,3-Butanedione; (6) 4-octanone; (7) 2-nonanone; (8) 4-hydroxy-3-hexanone; (9) 2-hydroxy-3-pentanone; (10) acetic acid; (11) butanoic acid; (12) pentanoic acid; (13) hexanoic acid; (14) heptanoic acid; and (15) octanoic acid. Dim = dimension.
). In this study, the correlation between sample C (which had a higher diacetyl content) and retronasal creamy perception was significantly higher than that of orthonasal creamy perception (P < 0.05). This result confirmed that oral processing affected the perception of aromas, especially for ketones, which contribute to cream perception. The correlation between sample E and orthonasal dairy sour was high, which was consistent with the highest contents and types of acids in sample E among all samples. This finding also indicated that people tend to perceive sourness through orthonasal olfaction. Studies have also reported that humans are more likely to distinguish acids through orthonasal olfaction (
). These results indicated that human orthonasal olfaction is more sensitive to acid perception.
As shown in Figure 3, orthonasal dairy sour and most acid aroma compounds, such as butanoic acid, pentanoic acid, hexanoic acid, heptanoic acid, and octanoic acid, were located on the negative axis side for dimension 1, and few of them were related to retronasal dairy sour. These results suggested that orthonasal olfaction is the primary pathway for the perception of acid compounds in fermented milk. As mentioned earlier, acids are water-soluble. When sour aroma was perceived through retronasal olfaction, sour aroma compounds could have been dissolved in saliva and perceived by taste. The distribution of retronasal milk fat and creamy perception was dense, indicating some similarity. Furthermore, diacetyl, which is a key aroma compound for creamy perception, had a higher correlation with retronasal creamy and milk fat than orthonasal perception, indicating that retronasal perception can be the primary pathway for perceiving creamy aroma. This phenomenon was due to the oral processing of ketones, such as diacetyl and 4-hydroxy-3-hexanone, which were released from the fermented milk (Table 2) and subsequently perceived. This is the first time we found that retronasal perception was the key perceptual pathway for compounds related to creamy perception.
). In this study, the variety and content of aroma were found to be different before and after simulated oral processing, which indicated that oral processing affected the release of aroma in fermented milk. We found that oral processing had a strong effect on the release of aroma compounds, such as diacetyl and acetone, which contributed to creamy perception, but its effect on the release of acids was not obvious. By combining sensory attributes and instrumentation data, we have shown that orthonasal creamy and retronasal dairy sour perceptions showed insignificant correlation with their respective aroma compounds, and the pathways for these 2 aroma perceptions in the fermented milk samples were different. The retronasal creamy perception and orthonasal sour aroma perception in fermented milk were high.
CONCLUSIONS
Oral processing affects the release of volatile aroma compounds in food, which results in differences between orthonasal and retronasal olfactory perceptions. However, the relationship between olfactory perception and the release of volatile aroma compounds under oral processing has not been fully elucidated. In this study, orthonasal and retronasal olfactory perceptions in 6 fermented milk samples were evaluated by trained panelists. Twenty-five volatile compounds were identified in fermented milk using HS-SPME-GC-Q-TOF-MS, 15 of which were predicted to have an influence on the olfactory perception of fermented milk during oral processing, as their contents varied significantly before and after simulated oral processing. Simulated oral processing greatly influenced the release of creamy aroma compounds, but had little effect on the release of dairy sour aroma compounds. By combining instrumentation data and sensory attributes, we have confirmed for the first time that creamy aroma perception in fermented milk had a high intensity through retronasal olfaction, and sour aroma perception had a high intensity through orthonasal olfaction. These results help in understanding aroma release and perception during the oral processing of fermented milk and how people perceive creamy aroma. This study was static and included 2 movements (before and after oral processing) during food consumption; further research on the dynamic release and perception of aroma compounds needs to be conducted in the future.
ACKNOWLEDGMENTS
Financial support was provided by the National Natural Science Foundation of China (Beijing, China; 31901717) and the Beijing Municipal Commission of Education Co-constructed Program (Beijing, China). The authors declare no conflict of interest.
REFERENCES
Aryana K.J.
Olson D.W.
A 100-year review: Yogurt and other cultured dairy products.
Re-investigation on odour thresholds of key food aroma compounds and development of an aroma language based on odour qualities of defined aqueous odorant solutions.
Identification and characterisation of headspace volatiles of fish miso, a Japanese fish meat based fermented paste, with special emphasis on effect of fish species and meat washing.
00022-9&id=gr1.jpg){kind=link}
00022-9&id=gr2.jpg){kind=link}
00022-9&id=gr3.jpg){kind=link}