Effects of cinnamaldehyde combined with ultrahigh pressure treatment on the flavor of refrigerated Paralichthys olivaceus fillets

The combined effects of cinnamaldehyde (CA) and ultrahigh pressure (UP) treatment on the flavor of olive flounder (Paralichthys olivaceus) fillets during storage at 4 °C for 20 days were investigated. Changes in total viable count, trimethylamine, ATP-related compounds, free amino acids, TCA-soluble peptides, electronic nose (E-nose) analysis and sensory quality were measured. The results indicated that CA and UP treatment, especially CA combined with UP, significantly reduced undesirable flavor compounds including inosine, hypoxanthine, TMA, and bitter amino acids, and accumulated pleasant flavor compounds such as inosine monophosphate and umami-related amino acids. In addition, the combination of CA and UP was shown to be more effective for retarding protein degradation and microbial growth than CA or UP treatment alone. In accordance with the results of E-nose analysis and sensory evaluation, CA combined with UP treatment had great potential for improving the flavor quality of refrigerated flounder fillets and extending their storage life.


Introduction
Olive ounder (Paralichthys olivaceus), a marine atsh species, is popular with consumers owing to its pleasant avor and abundant nutritional value. It is economically important for sheries and aquaculture, serving as a precious shing resource in Asia. 1 Recently, with the development of cold chain logistics, as well as changes in consumption concepts, sales of ounder llet are growing rapidly. Nevertheless, raw sh spoils easily and develops an unpleasant avor during post-mortem storage. The spoilage of raw sh is attributed to the actions of endogenous enzymes, microbial enzymes and lipid oxidation, leading to the deterioration of avor and texture to the point of a loss of edibility. Flavor variation is critical for determining consumers' preferences, oen foreshadowing changes in the quality of sh. 2,3 Therefore, it is essential to study how to prolong the shelf-life and maintain good avor quality of refrigerated sh llets.
Recently, cinnamon essential oil (CEO) has been considered a natural food preservative and is widely applied in aquatic products due to its excellent bacteriostatic and antioxidant properties. 4 Cinnamaldehyde (CA) is the main active component of CEO, accounting for 60-75% of the total oil. It has been identied as GRAS (Generally Recognized as Safe) and can be applied in food or antimicrobial food packaging according to the U.S. Food and Drug Administration. 5,6 Many researchers have explored the effect of cinnamon essential oil (CEO) on the quality and shelf-life of refrigerated aquatic products such as common carp (Cyprinus carpio), 7 Pacic white shrimp, 8 and rainbow trout. 9 These results suggested the positive effectiveness of CEO in prolonging the shelf-life of aquatic products during refrigeration. However, the use of essential oils in sh preservation is limited based on their peculiar avors and aromas, which affect sensory receptivity. 10 Lyu et al. 11 found that gamma radiation combined with CEO had a synergistic effect on maintaining sh quality, and additionally, the combination could reduce the radiation dose and concentration of CEO without diminishing the preservation effect. Thus, another preservation method in combination with CA is required to reduce its dosage and organoleptic impact on aquatic products.
Currently, the multi-hurdle technology has been widely used in food preservation. Ultrahigh pressure processing (UPP), as a non-thermal and promising technology, is commonly a feasible hurdle alternative. 12 UPP only acts on non-covalent bonded structures without damaging the protein's primary structure. It can deactivate spoilage microorganisms and enzymes, prolonging the shelf-life of raw sh and processed products during refrigerated storage. 13 In particular, ultrahigh pressure (UP) affects cell membranes' permeability via liquid medium and disturbs active transport mechanisms, resulting in an absence of nutrients, pH transformations and ultimately cell death. 14,15 In general, UPP can extend the shelf-life and improve the odor, taste, physicochemical properties as well as overall quality of sh muscles during chilled storage. 16 On the other hand, UPP beyond 150-200 MPa or higher can result in protein denaturation leading to undesired color changes and cookedlike appearance, and even accelerate lipid oxidation. 17 Therefore, the suitable selection of UPP parameters especially pressure or in combination with other preservation methods can abate the drawbacks and improve its effectiveness. There have been several prior studies on the use of CA or UP in food. 18,19 However, they are rarely combined for the preservation and avor retention of sh or other seafoods. Therefore, the present work is aimed at evaluating the effects of CA combined with UP treatment on the avor quality of Paralichthys olivaceus llets during refrigerated storage.

Sample preparation and treatment
Cinnamaldehyde was purchased from Shenzhen Guoxin Essence Perfume Co. Ltd. (Shenzhen, China). Hydroxypropyl-bcyclodextrin (HP-b-CD) was purchased by Henan Huarui Biotechnology Co. Ltd. (Henan, China). HP-b-CD solution was prepared by blending HP-b-CD with distilled water and stirring at 55 C until clear. Cinnamaldehyde was added into the prepared HP-b-CD solution and sonicated at 55 C with an ultrasonic cleaner (KQ-400KDB, Jiangsu, China) until the color of the mixture became turbid milky white. The nal preservative solution of CA consisted of 0.2% cinnamaldehyde (w/v) and 0.4% HP-b-CD (w/v). The concentration of the cinnamaldehyde was selected based on our preliminary study. 20 Fresh whole ounder (weight: 800 AE 100 g) were purchased from Lin Xi Street Aquaculture Market (Jinzhou, China) and instantly transported to the laboratory, where they were killed by percussive stunning. They were lleted by hand, followed by washing with cold sterile water. Two llets were obtained from each skin-off dorsal muscle of sh. Aerwards, every llet was cut into a sample with an average weight and length of 100 AE 8 g and 15 AE 0.2 cm. The llet samples were then randomly divided into four groups: (1) llets immersed in deionised water (control); (2) llets treated with deionised water prior to pressurized at 200 MPa for 10 min (UP); (3) llets immersed in a preservative solution of cinnamaldehyde (CA); (4) llets immersed in CA solution and then pressurized at 200 MPa for 10 min (UP + CA). The llets were dipped into the corresponding solution for 30 min. UP treatments were performed in ultrahigh pressure equipment (HPP.L2-600/0.6, Tianjin, China). All samples were separately packed in air-proof polyethylene bags and stored at 4 AE 1 C for subsequent quality analysis.

Total viable counts (TVC)
TVC of sh samples was determined using AOAC method. 21 TVC value was determined by the plate count method. The results were reported as lg CFU (colony forming units) g À1 .

ATP-related compounds
ATP-related compounds analysis was performed according to the method of Cai et al. 22 Determination of ATP-related compounds was performed using a reverse phase HPLC (Agi-lent1200; Agilent, CA, USA). Nucleotides, nucleosides, and bases were identied by comparing their retention times with those of commercially obtained standards. The content of each compound was calculated according to the peak areas.

Free amino acids (FAA)
Minced sh sample (2 g) was homogenated with 10 mL of 5% trichloracetic acid solution for 1 min. The homogenate was then centrifuged for 10 min at 7720 rpm. The above extraction process was repeated and the blended supernatants were diluted to 25 mL with distilled water. Then, the extract solution (1 mL) was ltered with a 0.22 mm membrane before being analyzed by an automatic amino acid analyzer (L-8900, Hitachi, Japan). The concentration of free amino acids (mg per 100 g sample) was determined by quantifying with standard amino acids.

TCA-soluble peptide
Three grams of chopped esh were homogenized with 27 mL trichloroacetic acid (5%, w/v). The samples were kept at 4 C for 1 h and centrifuged at 5460 rpm for 10 min. The content of TCA-soluble peptides in the supernatant was conducted using the method of Lowry 23 and expressed as mmol tyrosine per g muscle.

Trimethylamine (TMA)
TMA value determination was carried out by the AOAC method 21 with minor modication. The sh sample (2 g) was homogenized with 50 mL deionized water and 20 mL of 10% trichloroacetic acid. Aer ultrasonic treatment in an ice bath for 30 min, the sample was centrifuged at a speed of 10 000 rpm at 4 C for 6 min. The supernatant was neutralized to pH 4 with 1 M NaOH solution and diluted to 50 mL. Then, 4 mL of the above solution was mixed with 10% formaldehyde (1 mL), toluene (5 mL) and 25% KOH (3 mL) in a colorimetric tube, and heated at 30 C for 10 min. Then, 3 mL of the mixed solution was dried by 0.2 g anhydrous sodium sulfate and then blended with 3 mL picric acid solution (0.02%). The absorbance of the resulting reagent was recorded at 410 nm against the blank. A standard curve of trimethylamine hydrochloride was prepared and the concentration of TMA was calculated and expressed as mg per 100 g sample.

E-nose analysis
The aroma proles of sh samples treated by different methods were further determined using a PEN3 E-nose sensor system (Airsense Company, Germany). Two grams of minced muscle were placed into glass beaker and immediately sealed with plastic wrap. The beaker was rst incubated at 4 C for 20 min before injection. Then, the headspace gas was injected into the sensor chamber with a ow speed of 300 mL min À1 . The data collection time of E-nose detection lasted for 120 s.

Sensory evaluation
Sensory characteristics of llets were assessed according to the method of Zhou, Chong, Ding, Gu, and Liu 24 with some modications. Nine panelists graded for six odor attributes (pleasant odor, grassy odor, shy odor, amine odor, and rancid odor), using a nine-point hedonic scale (1-dislike extremely to 9-like extremely). A sensory score of 4 was deemed as the boundary of acceptability.

Statistical analysis
The mean data of three parallel experiments was the nal consequence. All data were performed by one-way-analysis of variance (ANOVA). Means separations were adopted by Duncan test at a signicance level of 5%. Principal component analysis (PCA) was applied to analyze E-nose data. Analyses were performed with the soware SPSS version 19.0.

Changes in TVC
The activity of microorganisms is the main factor responsible for sh spoilage and eventual change of avor. 25 Enzymes produced by microbial metabolism could cause protein and lipid degradation, resulting in the generation of volatile products. In general, sh muscles are rich in trimethylamine oxide (TMAO) and free amino acids that can easily form TMA and nitrogenous compounds due to microbial activity, leading to consumers' rejection. 16 As depicted in Fig. 1, the initial TVC value was 3.88 lg CFU g À1 at the rst day of storage, indicating that the ounder llets were of good quality. The TVC values of all the groups increased signicantly with the extension of storage time (P < 0.05). Additionally, during the entire storage period, the growth rate of treated samples was notably lower than that of the control group (P < 0.05), and there was no signicant difference in the aerobic bacterial count between the UP and CA groups (P > 0.05). Aer 16 days of storage, the TVC of control samples reached 7.42 lg CFU g À1 , exceeding the maximum limit (7 lg CFU g À1 ), as provided by ICMSF 26 for fresh llets. However, the TVC values of UP-and CA-treated samples reached 6.32 and 6.26 lg CFU g À1 on day 16, respectively, and the UP + CA samples showed the slowest growth rate of TVC, reaching 6.28 lg CFU g À1 on day 20. CA, as an electro-negative compound, could disturb biological processes involving electron transfer and protein synthesis, thus inhibiting microbial growth. 27 For ultrahigh pressure processing, it could effectively decrease the initial microbial load in sh muscles as well as the growth of spoilage microorganisms. 16 Similarly, Ojagh, Núñez-Flores, López-Caballero, Montero, and Gómez-Guillén 17 also found that the aerobic bacterial count of trout reduced by 5 lg CFU g À1 aer treatment with 300 MPa for 10 min. In sh, the development of spoilage bacteria such as pseudomonas and S. putrefaciens can lead to degradation and formation of foul odor during storage. 28 The results indicated that treating ounder llet samples with CA combined with UP retarded the microorganic growth synergistically, thus improving the avor quality and prolonging storage life.

ATP-related compounds analysis
The concentrations of ATP and its breakdown products are closely related to the avor and freshness of sh. Aer death, ATP in sh rapidly breaks down into ADP, AMP and IMP, due to endogenous enzymes. Subsequently, IMP degrades to inosine (HxR) and hypoxanthine (Hx). Among these compounds, IMP plays an important role in desirable avor, while HxR and Hx are responsible for off-avor and bitterness in sh muscle. 29 Hx can be generated by nucleotides' autolytic breakdown or/and bacteria such as Pseudomonas spp. and S. putrefaciens. 30 As can be seen in Fig. 2a, the concentration of ATP in the samples had an initial value of 1.19 mmol g À1 . With the extension of storage time, the concentrations of ATP were found to decrease signicantly in all groups (P < 0.05), especially during the rst 4 days of storage. In addition, the reduction in the ATP content of treated samples was prominently lower than that of the control group. Nevertheless, at the end of the storage period, the ATP content among different groups showed no signicant difference (P > 0.05), and the ATP content of UP + CA samples decreased by 1.10 mmol g À1 , which was slower than the other groups. The rapid degradation of ATP in ounder llets during storage might be caused by the activation of ATP enzymes. 31 The results of this study suggested that CA and/or UP treatment could affect the activity of ATP enzyme. Tariq et al. 32 reported that CA could inhibit ATPase enzymes and destroy the outer cell membrane. Truong et al. 16 indicated that the activity of Ca-ATPase and Mg-ATPase in sardine was declined with the increase of high pressure treatment from 100 to 500 MPa.
As one of the predominant umami nucleotides, IMP is mainly derived from the decomposition of AMP, due to the presence of AMP-deaminase and acid phosphatase. 29 The high content of IMP is a delicious avor enhancer of sh muscle. As shown in Fig. 2b, the IMP content of fresh ounder llets reached as much as 5.57 mmol g À1 at the beginning of storage. Signicant decreases in IMP content were observed in control This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 12573-12581 | 12575 Paper samples throughout the storage time (P < 0.05). In the treated samples, the IMP content showed a growing trend in the rst 4 days of storage, while signicant decreases in IMP content occurred during subsequent storage (P < 0.05). Additionally, the IMP concentrations in the treatment groups were prominently higher than those in the control group during the same storage period. Aer 20 days in storage, the IMP content of llets treated with UP, CA and UP + CA was 1.81, 1.98 and 2.15 mmol g À1 , respectively, whereas the control sample reached a concentration of 1.12 mmol g À1 , indicating that both UP and CA could substantially inhibit the interrelated enzyme activities and further suppress the IMP breakdown.
As presented in Fig. 2c and d, the initial content of Hx was obviously lower than that of HxR. This was in agreement with the results of previous reports on grass carp. 3 With the extension of storage time, the HxR content increased dramatically (P < 0.05) to the highest value on the 12th day and then declined signicantly with subsequent storage time (P < 0.05). Furthermore, the HxR content of treated samples was notably lower than that of control. The increase could be attributed to IMP consumption by 5 0 -nucleotidase and the decrease might be due to Hx formation decomposed by HxR. Meanwhile, microbial reproduction might also be a reason for the decline of HxR. Hx is a contributor to off-avor, and its accumulation is the predominant factor in sh decomposition and poor quality. 33 During the rst 4 days of storage, the Hx content among different groups showed no obvious difference (P > 0.05), possibly due to low levels of microorganisms and HxR in the initial storage. However, the Hx content increased sharply aer 8 days of storage (P < 0.05), reaching 6.34, 5.19, 5.05 and 4.54 mmol g À1 when stored for 20 d in control, UP, CA and UP + CA samples, respectively. The Hx content was dramatically lower (P < 0.05) in the UP + CA sample, suggesting that UP combined with CA was more effective in restraining microbial growth and protease activity, and thus the llets maintained better avor quality throughout the storage. Table 1 shows the contents of FAA in refrigerated llets on days 0, 4, 12 and 20. Sixteen amino acids were detected in fresh ounder llets, and the most abundant FAA was Lys, followed by Ala and Glu, adding up to 66.01% of total FAAs. The FAAs contribute to the taste of sh llets, specically umami, sweetness and bitterness. Glu, Asp, Ala, and Gly play an important role in the umami taste of food, 34 and their total content reached 21.66 mg/100 g in ounder llets. As shown in Table 1, the total FAA content, as well as that of Gly, Lys, Ser and Met, was found to signicantly decrease in all the groups rstly and then increase during the following storage period (P < 0.05). At the end of the storage period, the Lys, His, Leu and Phe content presenting for bitterness was lower in the UP + CA group than the other treated groups, while the umami FAAs such as Asp, Glu and Ala had accumulated. The FAAs can be generated by sh muscle proteolysis caused by endogenous and microbial enzymes. The changes in FAAs with storage time depend on the balance between their production and degradation into volatile and nonvolatile compounds. 35 The decrease  in the FAA content in sh muscle during storage might indicate their degradation and metabolism by bacteria, 36 while the increase might be owing to protein decomposition in sh muscle when subjected to higher proteolytic enzymatic activity aer 12 days of storage. 18 In addition, some FAA concentrations of the UP group were markedly higher than those of the control and CA treatment groups (P < 0.05). The results indicated that UP could promote amino acids accumulation, in accordance with the study by Yue et al. 36 who found that high pressure at 200 MPa could enhance the levels of taste FAAs in squid muscles during storage. By contrast with other treated samples, the CA samples had the lowest FAA concentration of 102.62 mg/ 100 g aer 20 days of storage (P < 0.05). The CA could effectively inhibit microbial growth and correlative enzymes, 37 thereby deferring the proteolysis caused by microbial enzymes in the later storage period. Therefore, UP combined with CA treatment might promote the accumulation of umami amino acids and delay the release of bitter amino acids, thus better maintaining the avor quality of llets. Similar results were reported by Yu et al. 3 who also found chitosan coating combined with essential oil contributed to the signicant accumulation of partial umami-associated FAA and the reduction of off-tasting histidine in refrigerated llets.

TCA-soluble peptides
The proteins in sh esh are susceptible to decomposition by microorganisms and enzyme, which affects the taste and avor, as well as the general freshness of sh. 25 Soluble peptides will be decomposed into amino acids and be further degraded to produce such volatile products as ammonia and amines, aldehydes, thiols, H 2 S, and indole, which nally causes some unpleasant odors. 3 As shown in Fig. 3, the level of TCA-soluble peptide in the fresh sample was approximately 0.22 mmol g À1 . The initial TCA-soluble peptides in post-slaughter llets might be produced by endogenous peptides and the accumulation of their degradation products. 38 The TCA-soluble peptide content of all the groups increased gradually throughout the storage period (P < 0.05), mainly correlating with the activity of autolysis and exogenous proteases. 39 Besides, the TCA-soluble peptide content in the control group was signicantly higher than those in the UP, CA and UP + CA groups at the same storage time (P < 0.05), demonstrating the effective inhibition of proteolysis by UP and/or CA treatments. This was consistent with the higher TVC and TMA values of the control sample compared with other treated samples. This result indicated that the control sample might have higher protease activity, leading to an increase in nitrogenous degradation products. In addition, protein catabolism and its nitrogenous degradation products were benecial for bacteria proliferation, which could further accelerate the decomposition of sh muscle. Therefore, the TCA-soluble peptides still showed a remarkable increase during the later storage period (P < 0.05). Aer 20 days of storage, the TCAsoluble peptide concentration of the control group reached 1.86 mmol g À1 , while that of the UP, CA and UP + CA treated groups were 1.41, 1.08 and 0.85 mmol g À1 , respectively. UP could deactivate the autolytic enzymes and microorganisms due to denaturation and/or modication of proteins, resulting in the inhibition of proteolytic degradation in sh muscle. The lower TCA-soluble peptide content of the CA group by comparison with the UP group could be due to superior antioxidative and antibacterial activities of CA. Apparently, the UP + CA group had the lowest TCA-soluble peptide content during storage, indicating that UP combined with CA treatment caused an intensively synergistic effect and better restrained the proteolytic degradation than UP or CA treatment alone.

Changes in TMA content
TMA is one of the main substances accountable for unpleasant shy odor in aquatic products, which is the major decomposition product of trimethylamine oxide aer microbial  metabolism. 40 As presented in Fig. 4, the initial TMA content was 4.01 mg/100 g in fresh sample. The TMA values in the control sample increased signicantly as the storage time increased (P < 0.05), while the treated samples showed a rapid increase aer 8 days of storage (P < 0.05). Although the TMA values increased as the storage period progressed in all samples, the values in the treated samples were prominently lower than that of the control, and the differences became remarkable aer the eighth day (P < 0.05). Besides, there was no obvious difference between the TMA values of the UP-and CAtreated samples at the corresponding storage time (P > 0.05).
In particular, the control sample increased to the value of 11.64 mg/100 g on the eighth day of storage. From the viewpoint of Ozogul et al., 41 a level of around 10-15 mg TMA/100 g in sh muscle means it is spoiled and unt for human consumption. On day 12 of storage, the TMA values of UP and CA samples reached 11.83 and 11.07 mg/100 g, respectively. However, the value in the UP + CA samples was still below 10 mg/100 g. Apparently, the accumulation of TMA was inhibited by CA and UP treatments, and the inhibiting effect was effectively promoted by the combination of CA and UP. The lower TMA values in the samples treated with UP may be attributed to the inhibition of proteolytic activity. 42 In early work, Bindu et al. 43 found that the TMA values of Indian white prawn aer high pressure treatment were signicantly reduced during chill storage, in accordance with the present results. Additionally, the TMA values measured in this study were observably low in the samples with CA treatments, suggesting that the growth of TMA-producing bacteria such as Pseudomonas spp. and S. putrefaciens was substantially restrained. 37 3.6. E-nose analysis by PCA E-nose technology, as a mimic sense of smell to detect and distinguish odors, is widely applied to assess food quality. 44 Enose analysis was carried out to distinguish the aroma proles of llets on the 0th, 4th, 12th and 20th days; a PCA loading plot of the different variables of ounder llet is presented in Fig. 5 though their dots showed a wide range, because the variance (0.12%) explained by the PC2 axis was extremely minor.

Sensory evaluation
Cinnamaldehyde itself has some odors and it can alter the avor of sh when added in large amounts. Although the llets had a very slight cinnamon odor aer being treated with CA, the concentration of 0.2% (w/v) did not cause unfavorable impact on the original avor of llets in our preliminary experiment, which was similar to the ndings reported by Zhang et al. 7 and Hu et al. 45 Sensory results of ounder llets were tested at 4 day intervals during the entire storage period. As described in Fig. 6, as the storage time increased, a continuous decrease in the sensory scores was observed in all samples, indicating that the odor of ounder llets gradually changed from pleasant to rancid with the extension of storage time. Aer 12 days of storage, the score of the control sample approached 3.9 exceeding the threshold of sensory rejection. 46 However, the sensory evaluations of treated samples were notably better than those of the control sample, approaching the scores of 6.1, 6.2 and 6.9 for the UP, CA, and UP + CA groups, respectively. As storage time went on, UP and/or CA treatment gradually had a positive impact on the sensory quality of llets, especially at the end of storage. This was in accordance with the results of TMA and ATP-related compounds. Sensory deterioration was attributed to bacteria spoilage and oxidation reactions, eventually resulting in the production of off-odor components, including volatile aldehydes and TMA. 47 Compared with the control, higher scores were obtained in the treated samples, suggesting that UP and/or CA treatment was effective in slowing ounder llets' sensory deterioration and preserving their avor quality, particularly with UP + CA treatment. High pressure treatment could inhibit the formation of biogenic amines and improve the odor and taste of sh esh, leading to better sensory quality compared with the untreated sample during storage. 16 Moreover, Zhang et al. 7 reported that vacuumpackaged common carp treated with CEO had better sensory quality than the control samples during refrigerated storage.

Conclusions
In summary, the present study indicated that a combination of CA and UP treatment had more positive effects on increasing IMP and umami-related amino acids, and reducing off-avor nucleotides and bitter amino acids in refrigerated ounder llets than did CA or UP treatments by themselves. In addition, the combination of CA and UP was more effective in reducing TVC, TCA-soluble peptides and putrid compound TMA.
Combining the results of E-nose analysis and sensory evaluation, it can be concluded that CA combined with UP treatment might be a promising method to retain avor quality of sh llet and improve its edible quality during refrigerated storage.

Conflicts of interest
The authors declared that there are no conicts of interest associated with this work.