Next Article in Journal
The Responses of the Ovary and Eyestalk in Exopalaemon carinicauda under Low Salinity Stress
Previous Article in Journal
Validation of a Portable eDNA Detection Kit for Invasive Carps
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Non-destructive Approach for the Prediction of pH in Frozen Fish Meat Using Fluorescence Fingerprints in Tandem with Chemometrics

1
Department of Food Science and Technology, Tokyo University of Marine Science and Technology, 4-5-7 Konan, Minato 108-8477, Tokyo, Japan
2
Department of Fisheries Technology, Patuakhali Science and Technology University, Dumki, Patuakhali 8602, Bangladesh
3
Department of Computer Science and Engineering, Toyohashi University of Technology, 1-1 Hibariga-oka, Tempaku-cho, Toyohashi 441-8580, Aichi, Japan
4
Department of Ocean Sciences, Tokyo University of Marine Science and Technology, 4-5-7 Konan, Minato 108-8477, Tokyo, Japan
*
Author to whom correspondence should be addressed.
Fishes 2022, 7(6), 364; https://doi.org/10.3390/fishes7060364
Submission received: 18 October 2022 / Revised: 28 November 2022 / Accepted: 28 November 2022 / Published: 30 November 2022

Abstract

:
The pH of fish muscle is an important index for quality assessment, but the traditional methods using a pH meter and probe/electrode are destructive, time-consuming, and laborious, making them unsuitable for on-line meat-quality monitoring. Hence, an approach of using fluorescence fingerprints (FFs) for the non-destructive prediction of pH in frozen fish fillets was trialled. Sixty-three live horse mackerel (Trachurus japonicus) and spotted mackerel (Scomber australasicus) were freshly harvested, sacrificed instantly, then preserved in ice, filleted, vacuum-packed, and frozen. Subsequently, the FFs of all frozen fillets were recorded using a fibreoptic-equipped fluorescence spectrometer, and the corresponding pH values of the samples were measured. After pre-processing, the masked FFs were modelled using partial least squares regression (PLSR) for the prediction of pH values. The results revealed that the developed method was accurate enough for predicting the pH changes in frozen horse mackerel and spotted mackerel fillets with R2 = 0.71 and R2 = 0.90, respectively. The proposed technique could be utilized as a rapid and non-contact alternative to traditional pH electrodes for the quality monitoring of fish products.

1. Introduction

Freshness is a significant quality parameter of fish, and there is a notable correlation between pH and fish freshness. The pH value of fish muscle is usually a complementary index for quality assessment that must be monitored because it is commonly related to the accumulation of lactic acid generated in the fish muscles using glycogen under anaerobic conditions. As the pH decreases due to the hydrolysis of adenosine triphosphate (ATP), a strong relationship between the drop in pH and the concentration of lactate in the muscle exists. Variations of pH changes occur depending on the fish species due to the effects of glycogen content and ATP reduction rate on pH decrease. The postmortem lowering of pH in fish significantly influences various quality aspects of the muscle, including texture, water-binding capacity, and resistance to microbial growth [1]. As an important quality parameter, pH is intimately related to protein precipitation and solubilisation, water-holding capacity, color, and drip loss of meat during processing [2]. Recent reports have demonstrated that high pH in fresh fish meat plays an important role in protein stabilization and in suppressing metmyoglobin formation during frozen storage [3]. The traditional methods of pH measurements in fish flesh using a pH meter and probe/electrode are destructive, time-consuming, and laborious, making them unsuitable for on-line meat-quality monitoring [4]. Therefore, there is a strong interest in developing a rapid, non-destructive, and non-contact sensing technique to predict the pH of fish meat in fisheries.
The measurement of the three-dimensional fluorescence fingerprints (3D-FFs) of frozen fish fillets has proven to be a very rapid, highly sensitive, selective, and non-destructive technique for monitoring freshness of the fish [5]. However, the pH of fish fillets changes with an increase in the storage period. The fluorescence spectra can be affected by various factors such as pH, temperature, color, and concentration quenching, and small pH changes can affect the intensity and spectral characteristics of fluorescence signals and even shift the maximum emission wavelength [6]. Considering all possible influences by other factors, the present study was conducted to predict, using 3D-FFs, the pH changes in frozen fish meat which were triggered during the postmortem ice-storage period (0-48 h). Fluorescence from ATP-related compounds has been detected for the quality assessment of both livestock [7] and frozen fish meat [5,8,9,10]. Additionally, the effects of pH on the fluorescence spectra of ATP were also investigated [11]. As the freshness of fish meat changes due to both pH decrease and the degradation of ATP-related compounds, non-destructive prediction of pH becomes necessary for the application of 3D-FFs in the evaluation of fish-meat freshness changes in early stage. In previous studies, changes in the pH of pork and chicken meat have been investigated using near-infrared and hyperspectral imaging techniques [2,12,13]. Also, pH-dependent fluorescence was observed by Wencel et al. (2014) [14]. Therefore, a novel approach, aimed at monitoring pH changes in frozen fish meat using the 3D-FFs is critically important. Accordingly, the main aims of this study were to determine the influence of pH on the 3D-FFs of frozen fish meat in tandem with the relevant chemometric approach and to confirm the pH sensitivity of fish fluorophores for the accurate prediction in pH values of fish meat during storage. To the best of our knowledge, this is the first study demonstrating an attempt to predict pH in frozen fish meat using the 3D-FFs.

2. Materials and Methods

2.1. Fish Sample Preparation

Horse mackerel (Trachurus japonicus) and spotted mackerel (Scomber australasicus) were freshly harvested and sacrificed instantly by the neck-breaking method to minimize the effect of slaughtering method on pH change [15]. High freshness can be maintained for a longer period by the neck-breaking method than by other methods, e.g., the struggled suffocating method of sacrifice [15], because a rapid decrease in pH occurs as a result of accelerated glycolytic reactions when fish die in agony due to struggling. To prepare fish samples with different postmortem pH, fishes were preserved in ice for different periods of 0, 0.5, 1, 1.5, 2, 3, 4, 5, 6, 7, 8, 12, 24 and 48 h (42 horse mackerel) and 0, 2, 6, 8, 12, 24 and 40 h (21 spotted mackerel) (n = 3; samples per storage condition). The fishes were then beheaded, eviscerated, filleted, individually packed in vacuum packs, and then frozen quickly in an air-blast freezer at −60 °C. The fish fillets were stored for one month at the same temperature (−60 °C) to avoid any possible change in their qualities until the time of acquiring their florescence fingerprints (FFs), because all enzymatic reactions and oxidations are suppressed in fish meat below −40 °C [3]. It was also reported that biochemical reactions such as protein denaturation, oxidation of lipid and protein, and discoloration due to accumulation of metmyoglobin could still take place in meat frozen and stored at temperatures higher than −20 °C, since sufficient unfrozen water remains available at these temperatures for such reactions to occur, and the optimum temperature for the frozen storage of meat is −40 °C given that a very small percentage of water is unfrozen at this point [16]. Cooling temperatures and rates which do not alter the pH of mackerel meat [17] were used in this study. Additionally, vacuum packaging of fish fillets before freezing was effective in preventing oxidative spoilage and in maintaining pH during frozen storage [18]. Thus, the postmortem color and pH variations of fish meat triggered by the ice-storage period (0–48 h) was maintained and there was no possibility of discoloration and changing of pH in the fish meat during quick freezing, frozen storage at −60 °C, or measurement of FFs at −30 °C (within 2 days).

2.2. Preparation of Buffer and Pure Solutions of Fluorophores

To determine the sensitivity of fluorescent compounds (fluorophores) to pH changes, standard reagent solutions of ATP-related compounds and their derivative nicotinamide adenine dinucleotides (adenosine 5′-triphosphate, ATP; adenosine 5′-diphosphate, ADP; adenosine 5′-monophosphate, AMP; oxidized nicotinamide adenine dinucleotide, NAD; and reduced nicotinamide adenine dinucleotide, NADH purchased from Oriental Yeast Company Limited, Tokyo, Japan) were prepared for fluorescence measurement. A 0.1 M Sorensen buffer solution (KH2PO4 and Na2HPO4) of four different pH values (5.0, 6.0, 7.0, and 8.0) was prepared to adjust pH with the aid of a pH meter (LAQUA F-72 pH meter, HORIBA Ltd., Kyoto, Japan). ATP (5 µmol/mL), ADP and AMP (1 µmol/mL), NAD (0.5 µmol/mL), and NADH (0.1 µmol/mL) solutions were prepared, and the pH values of all solutions were adjusted (5.0–8.0) using buffer solutions of different pH values. Pure buffer solutions with pH 5.0–8.0 were used as control samples.

2.3. Settings of Fluorescence Spectrometer for 3D-FFs Measurement

The three-dimensional fluorescence fingerprints (3D-FFs), also called the excitation–emission matrix (EEM), is a three-dimensional spectrum obtained by using a fluorescence spectrometer (F-7000; Hitachi High-Tech Science Corp., Tokyo, Japan). Although all fish fillets were frozen and stored in freezers at −60 °C after the postmortem processing, the samples were transferred to another small freezer at −30 °C one day prior to the 3D-FFs experiment to stabilize the fish-meat temperature and the same temperature was maintained during the fluorescence measurement. The 3D-FFs of frozen fish fillet samples, kept inside a small freezer (SC-DF25; Twinbird Corp., Niigata, Japan) at −30 °C, were acquired directly using an external Y-type fibre-optic probe in a dark room without thawing the samples. The fibre-optic probe was set 2 mm above the frozen fillet for remotely collecting fluorescence signals from every individual sample using a built-in FL Solutions software version 4.2 (Hitachi High-Tech Science Corp., Tokyo, Japan). The 3D-FFs were acquired from the ordinary flesh of fish meat only to avoid possible concentration quenching by the myoglobin of dark meat [19]. The excitation (Ex.) and emission (Em.) wavelengths were set in the same range of 250–800 nm at 10 nm intervals, 20 nm slit width, and 30,000 nm/min scanning speed. A light source was required to illuminate the fluorescence intensity, and the photomultiplier (PMT) voltage was adjusted to 400 V for all measurements. Only 150 s was required to obtain a single 3D-FFs spectrum of the fish fillet and all the spectra acquisition of a single fish species was completed within one day. After acquisition of the fluorescence spectra, the samples were repacked and kept frozen at −60 °C for one week to maintain their original quality until pH was measured by the traditional destructive method.
The 3D-FFs of the standard solutions of different fluorophores (ATP, ADP, AMP, NAD, and NADH) were obtained by placing them in the quartz cuvette in a cell holder inside the scanning compartment of the device. The measurement was conducted by setting the Em. wavelength at 200–600 nm at 10 nm intervals, with Ex. wavelengths at every 10 nm. For both excitation and emission, the slit width was adjusted to 10 nm, and the scanning speed was 60,000 nm/min. Triplicate measurements were conducted for each sample. First, the standard ATP solution with different pH values (5.0, 6.0, 7.0, and 8.0) was used as a model experiment, and then the other pure ADP, AMP, NAD, and NADH solutions with different pH values (5.0–8.0) were measured. The fluorescence intensity changed with the change in the intensity of the excitation light, and the detection voltage was adjusted to 400 V.

2.4. pH Measurement of Frozen Fish Meat

The pH of fish meat is an important indicator of freshness. After acquiring fluorescence spectra of all frozen fillet samples, the same muscle portion (ordinary meat) was dissected from each frozen fillet for pH measurement. Muscle was extracted, according to the method described by Rahman et al. (2019) in which each frozen sample was removed from its pack, and cylindrical subsamples (1 cm in diameter) were cut using a rotary saw from the same locations at which the 3D-FFs spectra were acquired [5]. After removing the peripheral and dark muscles, the excised subsamples were crushed into tiny pieces using a knife, a chisel, and a hammer without thawing. To maintain the original quality of the frozen fish fillets, the entire process of cutting, weighing, and homogenizing the muscle was performed inside a cold room (4.5 °C), and the cutting tools were maintained on dry ice at a low temperature during the whole process.
After removing the skin and dark muscle, frozen fish meat (approximately 3 g of ordinary muscle) was collected and homogenised with a cold solution of 0.02 M sodium iodoacetate (15 mL) using a polytron homogeniser (Model PT 10-35 GT; Kinematica AG, Lucerne, Switzerland) at 20,000 rpm to prevent enzymatic reactions [20]. The pH value was then measured using a digital pH meter (LAQUA F-72, HORIBA Ltd., Kyoto, Japan) calibrated with standard pH buffer solutions (pH 6.86 and pH 4.01 at 25 °C).

2.5. Processing of 3D-FFs Data

The excitation (λex) and emission (λem) wavelengths were plotted on the Y-axis and X-axis, respectively, and the fluorescence intensities were indicated using the color scale (from blue to red). As the 3D-FFs data contained scattered light that adversely affected the process of deduction of the highest peak intensity from the spectrum [21], the scattered light and non-fluorescence data were removed from the original fluorescence landscape while drawing contour plots using MATLAB 2016a (The Mathworks Inc., Natick, MA, USA) (Figure 1), and the highest peak intensities of pure fluorescent compounds were used for data analysis.

2.6. Multivariate Analysis of Fish 3D-FFs Data

For the multivariate analysis, fish 3D-FFs spectra and biochemical pH data were divided into two groups, the calibration and validation data-sets. As three fish samples were used for each ice-storage period (n = 3), first two sets of data were marked as the calibration data-set and last one was marked as the validation data-set. Then, partial least squares regression (PLSR) was applied for data analysis using JMP software (JMP pro. 12, SAS Institute, Inc. Cary, NC, USA), and the data were modelled by setting the wavelength pairs as predictors (X-variables) and the measured values of pH as the response variable (Y-vector). The coefficient of determination (R2), root mean square error estimated under calibration (RMSEC) and validation (RMSEV), and the optimum number of latent factors (LFs) of each PLSR model were determined using the leave-one-out method. Moreover, the primarily masked EEM data (1054 wavelength pairs) were also subjected to secondary masking (403 wavelength pairs), and then used to build the PLSR calibration and validation models (Figure 1).

3. Results and Discussion

3.1. pH Changes in Frozen Horse and Spotted Mackerel Fish Meat

pH has often been used as a complementary index to detect changes in fish freshness. Figure 2 shows the changes in pH of frozen horse mackerel fillets previously stored for 0–48 h. The initial pH of the fillets (0 h) was approximately 6.8, similar to that (6.85 ± 0.00) observed by Adeyemi et al. (2013) for the same fish species [22]. After 24 h of ice storage, the pH dropped to approximately 6.1, which was similar to that observed in the horse mackerel study of Maeda et al. (2007) [23]. Due to the sudden biochemical changes after death, the pH values of fish meat dropped rapidly within the first 12 h of ice storage followed by a gradual decrease reaching a value of approximately 6.0 after 48 h of storage. This could be attributed to the degradation of ATP and the production and accumulation of lactic acid in the fish muscle [1].
The meat pH of migratory fish such as tuna and mackerel may drop to around 5.5 as they lose freshness after death [3]. Although there is no pH threshold for high-quality fish meat, it has been shown that frozen mackerel meat treated at a specific subzero temperature range for a certain period of time before thawing can suppress the pH drop, and it was shown that the mackerel meat that maintained a pH around 6.2 retained more water-holding capacity and softening of the meat was suppressed [3]. Since the isoelectric point of myofibrillar proteins is around pH 5.5 and denaturation of myofibrillar proteins is accelerated below pH 6.0, it is believed that fish meat above pH 6.0 has high water-holding capacity, which contributes to a good texture of the thawed meat.
Figure 3 depicts the pH changes in frozen spotted mackerel, previously stored for 0, 2, 6, 8, 12, 24, and 40 h. The initial pH of spotted mackerel was approximately 6.65. It dropped noticeably to 6.03 within the first 6 h and then decreased gradually with the increasing ice-storage period. It is noteworthy that the decreasing trend of pH was faster in spotted mackerel than in the horse mackerel. At the end of ice storage (40 h), the average value of pH reached 5.84. The initial pH of our study was approximately 6.65, which was slightly lower than that reported by Ogata et al. (2016) for the same fish species [15]. This might be due to the antemortem and postmortem activities in the samples and to the freezing process. Since the muscle properties (e.g., the amount of glycogen, glycolytic-related enzymes, and their activity) of spotted mackerel were different from those of horse mackerel, their trends of pH change were also different.

3.2. 3D-FFs of Frozen Fish Meat

Figure 4 shows FFs of frozen horse mackerel fillet samples previously stored for different periods (0–48 h). The first intense fluorescent peak was located in the area of Ex. wavelength 250–320 nm and Em. wavelength 290–400 nm. This fluorescent peak might have been generated from a combination of ATP-related compounds and aromatic amino acid (tryptophan and tyrosine) residues of protein. The same peak was observed in the previous studies [5,8,9,24]. Additionally, a relatively smaller peak, which was assigned to NADH by Rahman et al. (2019), Chang et al. (2013), and Sádecká and Tóthová (2007) [5,25,26], was also observed at Ex. wavelength 330–420 nm and Em. wavelength 400–650 nm. Therefore, the 3D-FFs of fish meat were revealed at Ex. wavelength 250–420 nm and Em. wavelength 290–650 nm (403 wavelength pairs), containing most of the fluorophores related to fish quality.
As shown in Figure 4, the area and structure of fluorescent peaks of samples (0 h ice-storage period) slightly changed after 48 h of storage, and the pH dropped from 6.76 to 6.08. This might have been caused by the change in the concentration of fluorophores and affected by the acidity or alkalinity of the samples [6]. Since the pH is correlated with the concentration of different fluorophores, such as ATP-related compounds, and influenced the fluorescence of ATP [11], the pH changes during ice storage could be reflected in the 3D-FFs of fish meat. In other words, the fluorescence signals comprised a mixture of different fluorophores that may be affected by pH; thus, multivariate analysis was critically required to extract the information related to pH change in fish meat.

3.3. Prediction of pH in Fish Meat by FFs Coupled with PLSR Validation Models

During chemometric treatment, the entire 3D-FFs spectrum (3136 wavelength combinations = 56 Ex. × 56 Em.) was pre-processed and masked to remove both scattered light and non-fluorescent areas. After primary masking, 1054 wavelength pairs were obtained comprising 1054 different emissions at different excitation wavelengths. Then, secondary masking was also performed to remove irrelevant data (Figure 1) yielding 403 wavelength pairs. Rahman et al. (2019) proved that the 1054 wavelength pairs of the 3D-FFs were not as accurate as the 403 wavelength pairs in predicting the postmortem changes in fish meat, and sometimes over-fitting occurred during the chemometric modelling [5]. Therefore, the 403 wavelength pairs (Ex. 250–420 nm and Em. 290–650 nm) were used as predictor variables (X-matrix), and the values of pH were used as the response variable (Y-vector). Moreover, prior to the development of PLSR models, some spectral pre-processing methods, such as mean-centring and auto-scaling, were also used on the raw 3D-FFs to examine its feasibility in improving the prediction models. Auto-scaling measured the unit variance, whereas mean-centring scaled the fluorescence intensity at each wavelength to zero mean. In general, the performance of the PLSR models was evaluated with respect to R2, RMSECV, and LF. High R2, low RMSEC and RMSEV, and least number of LFs are the indicators of robust prediction models.
Table 1 presents the results of the PLSR models developed using masked Ex.–Em. wavelength combinations (403 variables) to predict pH changes in frozen fish fillets. The correlation between the fluorescence spectra (403 wavelengths) in the masked spectral range and the pH of horse mackerel samples (R2 = 0.88 and RMSE = 0.08), was obtained from the calibration data-set. This relationship was weaker in the validation model (R2 = 0.71 and RMSE = 0.16). However, the calibration and validation R2 (0.95 and 0.90, respectively) and RMSE values (0.07 and 0.11, respectively) for the PLSR models of pH prediction were higher in spotted mackerel fish meat. Four LFs were used in the spotted mackerel fish model, revealing a best-fit model for pH prediction in fish meat. The measured vs. predicted pH values using the above PLSR models (pH models shown in Table 1) are shown in Figure 5 (horse mackerel) and Figure 6 (spotted mackerel).
The PLSR validation model of pH prediction in frozen horse mackerel fish meat appeared to be a reasonable-fit model due to the lower coefficient of determination (R2 = 0.71); however, six LFs were used in this model, indicating the acceptance of the model (Figure 5). This phenomenon could be attributed to the narrow range of pH of the tested samples (pH = 6.80–6.01) involved in building such models [4]. As the pH changes in frozen horse mackerel, which was previously stored for a short period (0–48 h only), were monitored, the postmortem storage conditions were found to be unfavourable to triggering a large variation in the pH of fish meat. In contrast, the pH prediction by the PLSR validation model in frozen spotted mackerel (Figure 6) showed significantly high accuracy (R2 = 0.90). This could be because the metabolic rate of spotted mackerel meat was faster than that of horse mackerel. Additionally, the wider range of pH (6.77–5.75) in spotted mackerel meat during the postmortem ice-storage period would be one of the factors contributing to the high precision of PLSR model. Therefore, widening the pH range in horse mackerel meat by increasing the postmortem storage time and temperature and including more samples from different locations would play a vital role in enhancing the predictability of PLSR models for pH prediction in horse mackerel meat.
It is noteworthy that the fluorescence system did not measure the pH directly, as fluorescence spectroscopy does not employ electrodes for sensing the H+ ion. In essence, the recorded fluorescence signals of fish samples highlight the differences in fluorescence patterns occurring due to changes in the concentration of fluorophores, temperature, and inter-molecular forces in the fish meat at different pH levels [6]. Since fish meat is a multi-constituent medium, the proposed method must be applied to a large number of target components, resulting in a wide variation in pH, to determine the meat quality. Additionally, the temperature and time of post-harvest handling, the processing of the sample, and the fluorescence measurement temperature are critical factors that should be taken into consideration during the calibration [27]. Since pH is a reasonably good indicator of the final meat quality, accurate and non-destructive prediction of pH is extremely important for the fish industry. For instance, if pH of fish meat drops to a very low level at the time of fishing or remains low during frozen storage after death, protein denaturation may commence, and the commercial value will be decreased. The performance of pH prediction of the developed PLSR models (Table 1) in our study was much better than that of NIR spectroscopy, hyperspectral imaging, or fluorescence imaging reported by EIMasry et al. (2012) in beef meat (R2 = 0.73), Jia et al. (2017) in chicken breast fillets (R2 = 0.73), Liu et al. (2014) in salted pork meat (R2 = 0.79), and Rahman et al. (2021) in frozen shrimp meat (R2 = 0.53) [2,4,13,28].
In addition, it is important to describe why pH prediction by FFs would be a preferable freshness index over K-value prediction [9,10] during the early stage of postmortem in fish. In the previous studies [9,10], the postmortem ice-storage periods of fish for the K-value prediction were longer (0–12 days for EIMasry et al., 2015 and 0–9 days for Bui et al., 2018) than our study (0–48 h or 2 days only). During the short postmortem ice-storage period in our study, the K-values did not increase as high in horse mackerel fish (maximum 5.3%) (Figure S1). K-value is calculated as the ratio of non- phosphorylated ATP metabolites (i.e., HxR and Hx) to the total ATP breakdown products (ATP, ADP, AMP, IMP, HxR, and Hx). Basically, HxR and Hx are produced at later postmortem stages and their concentrations were much lower (maximum 0.7 μmol/g) in our study (data not shown). Thus, K-value seems to be an insensitive indicator for freshness of fish muscle as the rate of change in K-values during this stage was very slow and our finding agrees with the results of Mishima et al. (2005), who reported that K-values during the first 48 h reached around 5% [29]. Further PLSR validation models developed for predicting K-values using the same (pH prediction) wavelengths of FFs was rather poorly fitted (Figure S2), which confirms that the K-value was not a reliable index to predict the freshness changes by FFs at early postmortem stage. On the contrary, the initial pH of both horse mackerel and spotted mackerel meat (6.80 and 6.77) dropped rapidly within the first 12 h of ice storage due to the sudden biochemical changes after death, followed by a gradual decrease reaching a value of 6.01 and 5.75 after 48 and 40 h of storage, respectively (Figure 2 and Figure 3). This postmortem lowering of pH in both species of fish influences various quality aspects of the muscle, including texture, water-binding capacity, and resistance to microbial growth [1]. As pH is intimately related to the fish-meat quality, low pH may affect the color and drip loss of meat during processing [2]. Moreover, fish postmortem metabolites (e.g., ATP) are sensitive to pH (explained in the next section) and pH-dependent fluorescence was also observed [14]. That is why pH would be a preferable freshness index over K-value during the early stage of postmortem in fish and we succeeded to predict pH in fish meat by FFs.

3.4. pH Sensitivity of Pure Fluorophores and Their Relationship to Fish Meat Quality

The recorded fluorescence spectra are highly sensitive to pH value [7], and this pH-dependent fluorescence relationship could be easily monitored even in the flesh of fish meat. Fluorophores, which are associated with postmortem anaerobic glycolysis in fish meat, are mainly ATP-related compounds, oxidized and reduced forms of nicotinamide adenine dinucleotides (NAD and NADH, respectively). Since lactic acid and phosphate ions are generated during anaerobic glycolysis, which is involved in the hydrolysis of ATP, pH changes in fish meat are highly correlated with these fluorophores. Accordingly, pH sensitivity of different fluorophores confirmed the non-destructive pH prediction in frozen fish meat.
Figure 7 shows the 3D-FFs contour maps of different fluorophores (ATP-related compounds, NAD, and NADH standard solutions) at various pH. Considering the concentrations of these compounds in the fish body, pure ATP (5 µmol/mL), ADP, AMP (1 µmol/mL), NAD (0.5 µmol/mL), and NADH (0.1 µmol/mL) solutions were measured at different pH values (5.0–8.0). The 3D-FFs of the solutions appeared differently when the pH levels differed. There were significant relationships depicted between the fluorescence intensities of fluorophores (ATP, ADP, AMP, and NAD solutions) and pH levels.
As shown in Figure 7, The maximum fluorescence intensity of ATP solution was observed at 290 nm and 380 nm (λex and λem, respectively). The highest intensities were observed at pH 5.0, and the intensities declined drastically with increasing pH (Figure 8). At pH 5.0–6.0, the fluorescence intensities were highly affected by pH, whereas at pH 7.0–8.0, the intensities did not change drastically. Moreover, the highest intensity peaks at pH 5.0–6.0 were observed at specific Em. and Ex. wavelengths (290 nm and 380 nm, respectively). At pH 7.0–8.0, the Em. and Ex. wavelengths shifted to 300 nm and 410 nm, respectively, and lower intensities were observed (Figure 7). Therefore, the 3D-FFs contour plots of ATP solutions revealed that the strength of the fluorescence signal was influenced by the pH degree of the sample.
Similarly, the 3D-FFs contour plots of the standard solutions of ADP, AMP, and NAD showed a similar trend of fluorescence intensity with pH changes (Figure 7). At pH 5.0, the fluorescence intensities were the highest for ADP, AMP, and NAD solutions. These intensities declined with increasing pH and were the lowest at pH 8.0-like ATP solutions. However, the fluorescence of NADH solutions showed significantly higher intensities than those of other fluorophores, although there were no significant differences (ANOVA; p > 0.05) in the fluorescence intensities of NADH at different pH values (Figure 8). Thus, it was suggested that standard ATP, ADP, AMP, and NAD solutions were similarly influenced by pH changes, but no such trend was observed for the NADH solution (Figure 8) which may be attributed to the use of a higher concentration of NADH standard (0.1 μmol/mL) in this study.
The above-reported results revealed that pH had a potential effect on the overall features of 3D-FFs of ATP, ADP, AMP, and NAD solutions in terms of fluorescence intensity as well as on the locations of the remarkable peaks. In the case of the NADH solution, although the trend of intensity change was unclear, fluorescence was influenced by pH change (Figure 8). There are some reports of the effects of pH on the fluorescence emission of some indole compounds [30,31], whereas there are no reports on EEM of different food constituents. The results of the present study were supported by the observation that most aromatic molecules were fluorescent in neutral or acidic media, but the presence of a base led to the formation of non-fluorescent compounds [6]. Serotonin showed fluorescence emission when it was shifted from neutral pH to strongly acidic pH [30]. The fluorescence of amino-substituted proteins was notably influenced by pH, and other possible reasons were the intra-molecular charge migration and change in the molecular structure of tryptophan analogs [31], thus supporting our study. Moreover, not only the solution state of a standard fluorophore (ATP) but also the frozen state of such a fluorophore was influenced by changing pH [11], which is corroborative with the results of our novel pH-prediction technique in frozen fish meat.
The above statements indicate that most of the fluorophores (ATP-related compounds and nucleotides), associated with the postmortem changes in fish-meat freshness [5], were sensitive to pH. With the aging of fish, as the pH changes occurred parallel to the changes of different fluorophores (e.g., ATP), a strong relationship between the change in pH and the concentrations of such fluorophores exists. Therefore, pH-dependent fluorescence could be demonstrated in fish meat, even in a frozen state. Hence, it was concluded that pH prediction in frozen fish meat using FFs was reasonable, and the FFs could serve as a potential tool for the non-invasive and on-line monitoring of fish quality in the fish industry. However, the possibility of the application of the PLSR model for pH prediction by FFs is limited depending on fish species at present. Further studies including a large number of fish samples with different physiological conditions prior to fishing and in variations of cooling or freezing rate will be required to make this method applicable to a variety of fish species.

4. Conclusions

The present study was conducted to determine the feasibility of using three-dimensional FFs for the non-destructive prediction of pH in intact frozen fish meat. The masked FFs (Ex. 250–420 nm and Em. 290–650 nm) were utilized to build PLSR validation models to predict pH in frozen horse mackerel and spotted mackerel fish meat. pH changes in frozen horse mackerel fillets were predicted by the reasonable-fit FF model. The PLSR validation model of frozen spotted mackerel meat showed the best prediction (R2 = 0.90) of pH changes during the ice-storage period. In addition, the pH sensitivity of several standard solution of fluorophores which are included in fish meat and related to fish freshness was monitored to confirm the pH prediction by FFs. Small pH changes affected the intensity and spectral characteristics of FFs and even shifted the peak wavelength conditions. Significant effects of pH on the FFs of different fluorophores were observed, and PLSR models of FF spectra revealed accurate prediction of pH changes in frozen fish meat. This result indicated that the proposed technique could be utilized as a rapid and non-destructive alternative to traditional pH electrodes. Thus, this promising method of pH monitoring can facilitate the rapid prediction of quality changes, including sensory characteristics, in fish food.

Supplementary Materials

The following are available online at https://www.mdpi.com/article/10.3390/fishes7060364/s1, Figure S1: Changes in K-values of frozen horse mackerel fillets previously stored for different ice-storage periods; Figure S2: PLSR model for K-value prediction in frozen horse mackerel fillets previously stored in ice for 0–48 h.

Author Contributions

Conceptualization, M.M.R., M.S., N.N., S.N. and E.O.; methodology, M.M.R., M.S., N.N., M.N.A.R., S.N. and E.O.; software, M.M.R. and M.S.; validation, M.M.R. and M.S.; formal analysis: M.M.R. and M.S.; investigation, M.M.R. and M.N.A.R.; resources, S.N., T.H., K.O. and E.O.; data curation, M.M.R. and M.N.A.R.; writing—original draft preparation, M.M.R. and M.N.A.R.; writing—review and editing, M.M.R., M.S., N.N., M.N.A.R., S.N., T.H., K.O. and E.O.; visualization, M.M.R., M.S. and S.N.; supervision, S.N., T.H., K.O. and E.O.; project administration, S.N. and E.O.; funding acquisition, S.N. and E.O. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by grants from the Project of the NARO Bio-oriented Technology Research Advancement Institution (Research program on development of innovative technology) and the APC was funded by the Grant-in-Aid for Scientific Research provided by the Japan Society for the Promotion of Science, grant number JSPS 22F22088.

Institutional Review Board Statement

Ethical review and approval were waived for this study due to the following reasons: The fish study was conducted at the Tokyo University of Marine Science and Technology (TUMSAT) in 2017. However, the TUMSAT regulation on fish experiments was established in March 2020 and there was no particular requirement to apply for fish research approval prior to that date. According to the present regulations, the ethical rules for fish experiments are the same as those for animal experiments, but the application is not obligatory and is voluntary.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.

Acknowledgments

The authors would like to acknowledge the Ministry of Education, Culture, Sports, Science, and Technology (Monbukagakusho) of Japan for providing an FY2015 MEXT scholarship to the first author (Md. Mizanur Rahman) for pursuing his study in Japan. Additionally, we are grateful to Keisuke Moriya from the Tokyo University of Marine Science and Technology, Minh Vu Bui from the Toyohashi University of Technology, and the staff of the Iwate University Sanriku Reconstruction Regional Creation Promotion Organization Kamaishi Satellite for their support during the sample preparation.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Lougovois, V.P.; Kyranas, E.R.; Kyrana, V.R. Comparison of selected methods of assessing freshness quality and remaining storage life of iced gilthead sea bream (Sparus aurata). Food Res. Int. 2003, 36, 551–560. [Google Scholar] [CrossRef]
  2. Liu, D.; Pu, H.; Qu, J.; Sun, D.-W.; Wang, L.; Zeng, X.-A. Combination of spectra and texture data of hyperspectral imaging for prediction of pH in salted meat. Food Chem. 2014, 160, 330–337. [Google Scholar] [CrossRef] [PubMed]
  3. Nakazawa, N.; Okazaki, E. Recent research on factors influencing the quality of frozen seafood. Fish. Sci. 2020, 86, 231–244. [Google Scholar] [CrossRef] [Green Version]
  4. Elmasry, G.; Sun, D.-W.; Allen, P. Near-infrared hyperspectral imaging for predicting colour, pH and tenderness of fresh beef. J. Food Eng. 2012, 110, 127–140. [Google Scholar] [CrossRef]
  5. Rahman, M.M.; Shibata, M.; ElMasry, G.; Nakazawa, N.; Nakauichi, S.; Hagiwara, T.; Osako, K.; Okazaki, E. Expeditious prediction of post-mortem changes in frozen fish meat using three-dimensional fluorescence fingerprints. Biosci. Biotechnol. Biochem. 2019, 83, 901–913. [Google Scholar] [CrossRef] [PubMed]
  6. User Assistance PerkinElmer Ltd, An Introduction to Fluorescence Spectroscopy; PerkinElmer Inc.: Buckinghamshire, UK, 2000.
  7. Oto, N.; Oshita, S.; Makino, Y.; Kawagoe, Y.; Sugiyama, J.; Yoshimura, M. Non-destructive evaluation of ATP content and plate count on pork meat surface by fluorescence spectroscopy. Meat Sci. 2013, 93, 579–585. [Google Scholar] [CrossRef]
  8. Shibata, M.; Elmasry, G.; Moriya, K.; Rahman, M.M.; Miyamoto, Y.; Ito, K.; Nakazawa, N.; Nakauchi, S.; Okazaki, E. Smart technique for accurate monitoring of ATP content in frozen fish fillets using fluorescence fingerprint. LWT-Food Sci. Technol. 2018, 92, 258–264. [Google Scholar] [CrossRef]
  9. ElMasry, G.; Nagai, H.; Moria, K.; Nakazawa, N.; Tsuta, M.; Sugiyama, J.; Okazaki, E.; Nakauchi, S. Freshness estimation of intact frozen fish using fluorescence spectroscopy and chemometrics of excitation–emission matrix. Talanta 2015, 143, 145–156. [Google Scholar] [CrossRef]
  10. Bui, M.V.; Rahman, M.M.; Nakazawa, N.; Okazaki, E.; Nakauichi, S. Visualize the quality of frozen fish using fluorescence imaging aided with excitation-emission matrix. Opt. Express 2018, 26, 22954–22964. [Google Scholar] [CrossRef]
  11. Rahman, M.M.; Shibata, M.; Nakazawa, N.; Hagiwara, T.; Osako, K.; Okazaki, E. Effects of pH on the fluorescence fingerprint of ATP. Trans. JSRAE 2016, 33, 405–410. [Google Scholar]
  12. Barbin, D.F.; ElMasry, G.; Sun, D.-W.; Allen, P. Predicting quality and sensory attributes of pork using near-infrared hyperspectral imaging. Anal. Chim. Acta. 2012, 719, 30–42. [Google Scholar] [CrossRef] [PubMed]
  13. Jia, B.; Yoon, S.-C.; Zhuang, H.; Wang, W.; Li, C. Prediction of pH of fresh chicken breast fillets by VNIR hyperspectral imaging. J. Food Eng. 2017, 208, 57–65. [Google Scholar] [CrossRef]
  14. Wencel, D.; Abel, T.; McDonagh, C. Optical Chemical pH Sensors. Anal. Chem. 2014, 86, 15–29. [Google Scholar] [CrossRef] [PubMed]
  15. Ogata, Y.; Koike, H.; Kimura, I.; Yuan, C. Delaying Post-mortem Changes in the Muscle of Spotted Mackerel Killed by an Instantaneous Way of Neck-breaking and Bleeding. J. Fishsci. 2016, 10, 083–088. [Google Scholar]
  16. Leygonie, C.; Britz, T.J.; Hofman, L.C. Impact of freezing and thawing on the quality of meat: Review. Meat Sci. 2012, 91, 93–98. [Google Scholar] [CrossRef] [PubMed]
  17. Nakazawa, N.; Maeda, T.; Fukushima, H.; Wada, R.; Tanaka, R.; Okazaki, E. Effect of Cooling Conditions on the ATP Content and pH of Chub Mackerel (Scomber japonicus) Meat. Trans. Jpn. Soc. Refrig. Air Cond. Eng. 2019, 36, 49, (In Japanese with English abstract). [Google Scholar]
  18. Aberoumand, A.; Baesi, F. Effects of vacuum packaging in freezer on oxidative spoilage indexes of fish Lethrinus atkinsoni. Food Sci. Nutr. 2020, 8, 4145–4150. [Google Scholar] [CrossRef]
  19. Postnikova, G.B.; Shekhovtsova, E.A. Fluorescence Studies on the Interaction of Myoglobin with Mitochondria. Biochemistry 2012, 77, 280–287. [Google Scholar] [CrossRef]
  20. Bito, M. Changes in NAD and ATP Levels and pH in Frozen-Stored Skipjack Meat, in Relation to Amount of Drip. Bull. Japan Soc. Scient. Fish 1978, 44, 897–902. [Google Scholar] [CrossRef]
  21. Rinnan, A.; Andersen, C.M. Handling of first-order Rayleigh Scatter in PARAFAC Modelling of Fluorescence Excitation–Emission Data. Chemom. Intell. Lab. Syst. 2005, 76, 91–99. [Google Scholar] [CrossRef]
  22. Adeyemi, O.T.; Osilesi, O.O.; Onajobi, F.; Adebawo, O.; Afolayan, A.J. Stability study of smoked fish, horse mackerel (Trachurus trachurus) by different methods and storage at room temperature. Afr. Biochem. J. Res. 2013, 7, 98–106. [Google Scholar]
  23. Maeda, T.; Yuki, A.; Sakurai, H.; Watanabe, K.; Itoh, N.; Inui, E.; Seike, K.; Mizukami, Y.; Fukuda, Y.; Harada, K. Alcohol brine freezing of Japanese horse mackerel (Trachurus japonicus) for raw consumption. Trans. JSRAE 2007, 24, 323–330. [Google Scholar]
  24. Dufour, E.; Frencia, J.P.; Kane, E. Development of a rapid method based on frontface fluorescence spectroscopy for the monitoring of fish freshness. Food Res. Int. 2003, 36, 415–423. [Google Scholar] [CrossRef]
  25. Chang, C.Y.; Chang, C.C.; Hsiao, T.C. Fluorescence intrinsic characterization of excitation-emission matrix using multi-dimensional ensemble empirical mode decomposition. Int. Mol. J. Sci. 2013, 14, 22436–22448. [Google Scholar] [CrossRef] [PubMed]
  26. Sádecká, J.; Tóthová, J. Fluorescence spectroscopy and chemometrics in the food classification-a review. Czech J. Food Sci. 2007, 25, 159–173. [Google Scholar] [CrossRef]
  27. Andersen, C.M.; Mortensen, G. Fluorescence spectroscopy: A rapid tool for analyzing dairy products. J. Agric. Food Chem. 2008, 56, 720–729. [Google Scholar] [CrossRef]
  28. Rahman, M.M.; Bui, M.V.; Shibata, M.; Nakazawa, N.; Rithu, M.N.A.; Yamashita, H.; Sadayasu, K.; Tsuchiyama, K.; Nakauchi, S.; Hagiwara, T.; et al. Rapid noninvasive monitoring of freshness variation in frozen shrimp using multidimensional fluorescence imaging coupled with chemometrics. Talanta 2021, 224, 121871–121880. [Google Scholar] [CrossRef]
  29. Mishima, T.; Nonaka, T.; Okamoto, A.; Tsuchimoto, M.; Ishiya, T.; Tachibana, K.; Tsuchimoto, M. Influence of storage temperatures and killing procedures on post-mortem changes in the muscle of horse mackerel caught near Nagasaki Prefecture, Japan. Fish. Sci. 2005, 71, 187–194. [Google Scholar] [CrossRef]
  30. Leiva, A.D.; Schwartz, S. Relation of pH to Serotonin, Melatonin, and Other Indole Compounds Reacted with o-Phthaldialdehyde. Clin. Chem. 1976, 22, 1999–2005. [Google Scholar] [CrossRef]
  31. Lepthien, S.; Wiltschi, B.; Bolic, B.; Budisa, N. In Vivo Engineering of Proteins with Nitrogen-containing Tryptophan Analogs. Appl. Microbiol. Biotechnol. 2006, 73, 740–754. [Google Scholar] [CrossRef]
Figure 1. Scheme for preprocessing and masking of fluorescence spectra for multivariate analysis.
Figure 1. Scheme for preprocessing and masking of fluorescence spectra for multivariate analysis.
Fishes 07 00364 g001
Figure 2. Changes in pH of frozen horse mackerel fillets previously stored for different ice-storage periods.
Figure 2. Changes in pH of frozen horse mackerel fillets previously stored for different ice-storage periods.
Fishes 07 00364 g002
Figure 3. Changes in pH of frozen spotted mackerel fillets previously stored for different ice-storage periods.
Figure 3. Changes in pH of frozen spotted mackerel fillets previously stored for different ice-storage periods.
Fishes 07 00364 g003
Figure 4. Representative 3D-FFs of frozen horse mackerel fish meat previously ice-stored for 0–48 h.
Figure 4. Representative 3D-FFs of frozen horse mackerel fish meat previously ice-stored for 0–48 h.
Fishes 07 00364 g004
Figure 5. PLSR model for pH prediction in frozen horse mackerel fillets previously stored in ice for 0–48 h.
Figure 5. PLSR model for pH prediction in frozen horse mackerel fillets previously stored in ice for 0–48 h.
Fishes 07 00364 g005
Figure 6. PLSR model for pH prediction in frozen spotted mackerel fillets previously stored in ice for 0–40 h.
Figure 6. PLSR model for pH prediction in frozen spotted mackerel fillets previously stored in ice for 0–40 h.
Fishes 07 00364 g006
Figure 7. FF contour plots of ATP, ADP, AMP, and NAD solutions (X-axis = λex and Y-axis = λem in nm).
Figure 7. FF contour plots of ATP, ADP, AMP, and NAD solutions (X-axis = λex and Y-axis = λem in nm).
Fishes 07 00364 g007
Figure 8. Effect of pH on fluorescence intensities of different fluorophore solutions at highest peak condition.
Figure 8. Effect of pH on fluorescence intensities of different fluorophore solutions at highest peak condition.
Fishes 07 00364 g008
Table 1. Different PLSR models for the prediction of pH changes in frozen fish meat.
Table 1. Different PLSR models for the prediction of pH changes in frozen fish meat.
Frozen Fish Meat with Ice-Storage ConditionPartial Least Square Regression (PLSR) Models Using Wavelength Combinations of Ex. 250–420 nm/EM. 290–650 nm (403 Wavelength Pairs)
CalibrationValidationNumber of Latent Factors
R2RMSESample SizeR2RMSESample SizeLFs
Horse mackerel
(0–48 h)
0.880.08280.710.16146
Spotted mackerel
(0–40 h)
0.950.07140.900.1174
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Rahman, M.M.; Shibata, M.; Nakazawa, N.; Rithu, M.N.A.; Nakauchi, S.; Hagiwara, T.; Osako, K.; Okazaki, E. Non-destructive Approach for the Prediction of pH in Frozen Fish Meat Using Fluorescence Fingerprints in Tandem with Chemometrics. Fishes 2022, 7, 364. https://doi.org/10.3390/fishes7060364

AMA Style

Rahman MM, Shibata M, Nakazawa N, Rithu MNA, Nakauchi S, Hagiwara T, Osako K, Okazaki E. Non-destructive Approach for the Prediction of pH in Frozen Fish Meat Using Fluorescence Fingerprints in Tandem with Chemometrics. Fishes. 2022; 7(6):364. https://doi.org/10.3390/fishes7060364

Chicago/Turabian Style

Rahman, Md. Mizanur, Mario Shibata, Naho Nakazawa, Mst. Nazira Akhter Rithu, Shigeki Nakauchi, Tomoaki Hagiwara, Kazufumi Osako, and Emiko Okazaki. 2022. "Non-destructive Approach for the Prediction of pH in Frozen Fish Meat Using Fluorescence Fingerprints in Tandem with Chemometrics" Fishes 7, no. 6: 364. https://doi.org/10.3390/fishes7060364

Article Metrics

Back to TopTop