1 Introduction

Metabolite composition is an important parameter for assessing food quality and safety (Wishart, 2008). In this regard, analysis of food molecular nutritional composition (metabolome) and its variations associated with quality becomes a very useful foodomics approach for food quality assessment at a molecular level. This is because metabolomics analysis can provide more chemical details than traditional food component analysis (Gallo and Ferranti, 2016). Nuclear magnetic resonance (NMR)-based metabolomics technique is among the core foodomics technologies that holistically capture the metabolite composition of food (Wishart, 2008). It has been proven to be a valuable tool for assessing authenticity (Li et al., 2016; Petrakis et al., 2016), safety (Cappello et al., 2018), and quality (Harnly et al., 2018) of food. Such applications have been especially conducted to understand the quality change of aquatic food such as Atlantic salmon fillets (Shumilina et al., 2015), vinasse pike eel (Chen et al., 2017), and mud crab (Li et al., 2019).

Mud crab (Scylla paramamosain) is among the most consumed crabs in China due to its delicious flavor, taste, and nutritional composition (Ye et al., 2011). It is generally believed that dead mud crab is not edible, resulting in abandonment of dead mud crabs and huge economy loss. Actually, a rapid accumulation of harmful metabolites such as biogenic amines and sulfur compounds have been found in dead Chinese mitten crab (Eriocheir sinensis) (Yang et al., 2016; Zhu et al., 2017; Wang et al., 2018). Precisely, trimethylamine (TMA) is toxical to the human (Anthoni et al., 1991) and has been widely used as an important indicator of freshness of aquatic products (Zhao et al., 2002; Shumilina et al., 2015). Our previous study revealed a significantly increased TMA level with fermentation of crab paste, which may pose a health risk to humans (Chen et al., 2016). Furthermore, TMA and total volatile base-nitrogen (TVB-N) both increase with storage time in whole blue crabs (Callinectes sapidus) due to metabolism by spoilage bacteria (Parlapani et al., 2019). In addition, histamine is frequently implicated in fish poisoning (Mercogliano and Santonicola, 2019). Though the quality of mud crab changes after death, whether dead mud crab is still safe for consumption is yet to be determined. The postmortem metabolite changes can be an important indicator for assessing the quality and safety of mud crab; however, little information is available for such changes.

In this study, NMR-based metabolomics analysis and multivariate data analysis were applied to systematically analyze the postmortem changes in metabolite composition of edible tissues, including muscle and hepatopancreas, of mud crab at room temperature, on ice, or at −20℃. The present study aimed to: i) investigate the postmortem metabolite profile changes of mud crab with different storage methods; and ii) assess the safety and quality of dead mud crab.

2 Materials and Methods 2.1 Chemicals and Reagents

The following chemicals and reagents (all of analytical grade) were purchased from commercial suppliers: methanol, dipotassium hydrogen phosphate trihydrate and sodium dihydrogen phosphate dihydrate (Sinopharm Chemical Co., Ltd., Shanghai, China); sodium 3-trimethylsilyl [2, 2, 3, 3-²D₄] propionate (TSP) and deuterated water (D₂O, 99.9% in D) (Cambridge Isotope Laboratories, FL, Miami, USA). Phosphate buffer (K₂HPO₄/NaH₂PO₄, 0.1 mol L⁻¹, pH 7.48) was prepared in water (H₂O) containing 80% D₂O (to provide an NMR field lock) and 0.001% TSP (w/v) (to serve as an internal standard) (Xiao et al., 2008).

2.2 Sample Treatment

A total of 160 live male mud crabs (about 150 g) were purchased from a local Crab Breeding Company (Ningbo, China) and transported to the laboratory within approximately one hour. Eight live crabs were randomly collected as live control group (LCK group). Four experimental treatments were set up: dead crabs kept at room temperature (RT group); dead crabs kept on ice (OI group); dead crabs kept at −20℃ (DMT group); live crabs kept at −20℃ (LMT group). For LMT group, 40 live crabs were randomly chosen and anesthetized. Then they were frozen at −20℃ and kept in −20℃ for 1, 2, 4, 8, and 16 d. For RT group, the dead crabs were obtained immediately after natural death of crabs at room temperature. Eight recently dead crabs were immediately chosen as dead control group (DCK group). The remaining 32 dead crabs were collected after they were kept at room temperature for 2, 4, 8, and 12 h. For OI group, additional 32 recently dead crabs were kept on ice and collected after they were kept on ice for 2, 4, 8, and 12 h. For DMT group, additional 40 recently dead crabs were kept at −20℃ and collected at five time intervals: 1, 2, 4, 8, and 16 d. After the samples were collected at certain time, muscle and hepatopancreas tissues were collected from each sampled crab immediately and snap frozen in liquid nitrogen, followed by storage at −80℃ until metabolomics analysis. Eight replicates (one crab per replicate) were used for each sampling time point for each group.

2.3 Metabolite Extraction of Muscle and Hepatopancreas

Each tissue sample (300 mg) was homogenized with 1 mL ice-cold aqueous methanol (methanol/water = 2:1) using a tissuelyzer (QIAGEN, Hilden, Germany) at 20 Hz for 90 s. After 10 min of centrifugation at 12000 r min⁻¹ and 4℃, the resulting supernatant of the homogenate mixture was collected. The same extraction procedure was repeated once more for the remaining solid residues. The supernatant was combined and condensed in vacuo for removal of methanol. After lyophilization, each extract was dissolved in 600 μL phosphate buffer. Following a 10 min centrifugation at 12000 r min⁻¹ and 4℃, 550 μL of supernatant from each extract was pipetted into 5 mm NMR tube for NMR analysis.

2.4 NMR Analysis

NMR analysis of the tissue extracts was conducted on a Bruker Avance Ⅲ 600 MHz spectrometer (Bruker Biospin, Germany) operating at 600.13 and 150.90 MHz for ¹H and ¹³C, respectively. One-dimensional ¹H NMR spectra were acquired at 298 K using a NOESYGPPR1D pulse sequence with an inverse cryogenic probe. The water signal was suppressed by a weak continuous wave irradiation during a 2 s relaxation delay and a 100 milliseconds mixing time. Meanwhile, a 90˚ pulse length was adjusted to approximately 10 microseconds. Sixty-four transients were collected into 32768 data points for each ¹H NMR spectrum, with a spectral width of 20 ppm. For metabolite signal assignment purposes, five two-dimensional (2D) NMR spectra were acquired for selected samples and processed with similar parameters described previously (Aue et al., 1976a, 1976b; Braunschweiler and Ernst, 1983). These 2D spectra included ¹H J-resolved spectroscopy, ¹H-¹H correlation spectroscopy, ¹H-¹H total correlation spectroscopy, ¹H-¹³C heteronuclear single quantum coherence, and ¹H-¹³C heteronuclear multiple bond correlation spectra.

2.5 NMR Data Processing and Multivariate Data Analysis

All ¹H NMR free induction decays were multiplied by an exponential line broadening function of 0.5 Hz before Fourier transformation by TOPSPIN (V2.0, Bruker Biospin, Germany). Furthermore, these spectra were phase- and baseline-corrected manually and calibrated to the proton signal from TSP at δ 0.00. For multivariate data analysis, the region δ 0.8–9.2 of each ¹H NMR spectrum was uniformly bucketed into bins with a 2.4 Hz width. Regions δ 4.7–5.1 and δ 3.34–3.37 were removed to eliminate solvent signals. The integrated areas of all bins were normalized to the total spectral intensity for each spectrum to compensate for overall concentration differences between the samples.

The spectral datasets were imported into SIMCA-P⁺ software (V12.0, Umetrics, Umeå, Sweden) for multivariate data analysis. An unsupervised principal component analysis (PCA) was initially performed using the mean-centered NMR data to generate an overview of group clustering and search for possible outliers. A supervised orthogonal projection to latent structure discriminant analysis (OPLS-DA) was sequentially carried out using unit variance-scaled NMR data (Trygg and Wold, 2002; Berg et al., 2006). A seven-fold cross validation was used with NMR data (X-matrix) and group information (Y-matrix) (Trygg et al., 2007). The quality of all OPLS-DA models was evaluated with R²X, R²Y, and Q² values, which respectively indicate the explained variations and model predictability, and further validated for their robustness with a cross validation-analysis of variance (CV-ANOVA), with P < 0.05 set as the significance level (Eriksson et al., 2008). In order to facilitate the interpretation of results, a backscaled transformation of the loadings of each variable contributing to the model (Cloarec et al., 2005) was conducted and plotted with a color-coded coefficient that indicates the weights of discriminatory variables in MATLAB (V7.1, Mathworks Inc., Natick, MA, USA) using an inhouse developed script. A hot color (red) indicates that the metabolite is significantly different between groups, while a cool color (blue) indicates no significant differences. A cutoff value of 0.666 was utilized in this study and the metabolites with an absolute value of r > 0.666 were considered statistically significant (P < 0.05).

3 Results 3.1 NMR Spectra for Muscle and Hepatopancreas

Resonance peaks were assigned to specific metabolites based on data obtained from other researches (Fan and Lane, 2008; Chen et al., 2016) and were further confirmed according to the homonuclear or heteronuclear correlations provided by a series of 2D NMR spectra (Fig. 1). The detailed metabolite NMR data are listed in Table 1. A total of 32 metabolites were identified. The dominating metabolites identified from the muscle and hepatopancreas NMR spectra included amino acids (leucine, isoleucine, valine, threonine, alanine, methionine, glutamate, glutamine, lysine, glycine, arginine, tyrosine, phenylalanine, and tryptophan), a derivative of glycine (sarcosine), organic acids (lactate, acetate, succinate, formate, sarcosine, fumarate, and taurine), organic amines (trigonelline, histamine, and TMA), ATP and its breakdown products (adenosine diphosphate (ADP), adenosine monophosphate (AMP), and inosine), betaine, trimethylamine-N-oxide (TMAO), glucose, and 2-pyridinemethanol. Most of the low-molecular-weight metabolites found in the muscle and hepatopancreas spectra were similar. However, the different tissues contained their own dominant metabolites. For instance, acetate, fumarate, AMP, ADP, ATP, and trigonelline dominated in the muscle, whereas sarcosine and tryptophan dominated in the hepatopancreas. The differences in composition might be related to the structure and function of the tissue. In the same tissue, the composition of metabolites was similar between different groups; however, there were differences in peak intensity in the ¹H NMR spectra between live and dead crabs (Fig. 1). For instance, the dead crabs had an increased inosine level and decreased ATP level in the muscle, and an increased lactate level in the hepatopancreas compared with the live crabs.

3.2 Metabolite Changes Associated with Storage Method

To obtain more details of time-dependent changes of metabolites in the muscle and hepatopancreas of dead crabs due to the influence of different storage methods, multivariate data analysis was further performed on the NMR data. PCA analysis was first performed on the normalized ¹H NMR data of all crab muscle extracts. The mean values of PC1 and PC2 from each muscle group were calculated for the PCA model. The trajectory plot of PC1 versus PC2 illustrating the global metabolic changes during the whole storage time as well as the influence of storage method is shown in Fig. 2. Here, each coordinate represents an average of scores for eight muscle hepatopancreas extracts in a storage group at a particular storage time. For muscle, the trajectory illustrated that the greatest change of metabolic profiles among the four groups was found in RT crab muscle. The metabolic profiles obtained from OI crab muscle presented a relative stable change. Although the storage time was prolonged to 16 d, the metabolic profiles obtained from DMT and LMT crab muscles still presented a smaller change than those of RT crab muscle. Moreover, OI, DMT, and LMT crab muscles presented a different metabolic change. Among them, LMT crab muscle had the closest metabolic profile to fresh LCK crab muscle. For hepatopancreas, the metabolic trajectory also demonstrated the greatest change of metabolic profile in RT crabs. Storage OI obviously reduced the change of metabolic profile relative to storage at RT. Similarly, 16 days storage at −20℃ presented a smaller change in the metabolite profiles than 12 h storage at RT. These observations suggest that storage method had an obvious influence on metabolite composition of edible tissues in mud crab and the storage temperature plays an important role in it.

To obtain detailed information on the altered metabolites of dead crabs associated with different storage methods, OPLS-DA model was created on the NMR data of each crab group against the live crab group. Eight constructed OPLS-DA models had good qualities based on the high Q²Y values and low P values obtained from CVANOVA evaluation (Fig. 3 left and Fig. 4 left). These valid OPLS-DA models were constructed from the muscle NMR data of RT8, RT12, LMT8, and LMT16, as well as the hepatopancreas NMR data of RT2, RT12, DMT8, and DMT16. Corresponding coefficient plots of OPLS-DA (Fig. 3 right) display a significant intergroup difference of metabolites between live control crabs and dead crabs under different storage conditions. The corresponding correlation coefficients of these metabolites are listed in Table 2. Our results showed that 8 h storage at RT induced a significant elevation of muscle lactate, succinate, and inosine, accompanied with a decline of muscle taurine, TMAO, and ATP (Fig. 3A). Moreover, 12 h storage at RT caused a significant elevation of muscle TMA, lactate, and formate, and a significant decline of muscle TMAO and ATP (Fig. 3B). When the live crabs were stored at −20℃, the muscle formate and AMP were significantly accumulated, whereas the muscle ADP and ATP were significantly depleted at 8 d postmortem (Fig. 3C); the muscle lactate and AMP were significantly accumulated, whereas the muscle tyrosine and ADP were significantly depleted at 16 d postmortem (Fig. 3D).

For the hepatopancreas, 2 h storage at RT resulted in a significant decrease of leucine, isoleucine, valine, tyrosine, phenylalanine, and TMAO (Fig. 4A), whereas 12 h storage at RT resulted in a significant increase of lactate, alanine, succinate, and formate (Fig. 4B). After storing the dead crabs at −20℃, a significant elevation in the levels of isoleucine, leucine, valine, glycine, and inosine was observed at 8 d postmortem (Fig. 4C). A significant elevation in the level of inosine and a significant depletion in the level of sarcosine were further observed at 16 d postmortem (Fig. 4D).

The dynamic changes of some metabolites with different storage time were calculated by comparing the concentration ratios of dead crab groups with those of the live control group and are shown in Fig. 5. Overall, a persistent rise of the muscle TMA as well as of the muscle and hepatopancreas lactate, formate, and succinate occurred throughout 12 h storage at room temperature. The muscle lactate, formate, and TMA had the most substantial rise up to 3.7, 9.8, and 5.6-fold, respectively, relative to live control level at 12 h postmortem, whereas the hepatopancreas succinate had a 1.6-fold elevation relative to live control level. We also found a lasting decline in ATP level, a first increase and then decrease in the level of ADP and AMP, as well as a lasting elevation in the level of inosine in the muscle at room temperature. These changes were highlighted in a 91.0% depletion of ATP and 618.1% accumulation of inosine at 12 h postmortem. Such a rapid hydrolysis of ATP is in agreement with previous report, where ATP is degraded within the first 24 h postmortem (Ocaño-Higuera et al., 2011). However, storage on ice obviously reduced the changes of the above metabolites, except for ADP and AMP. Notably, only slight changes occurred in the levels of lactate, formate, succinate, and TMA in the muscle and/or hepatopancreas during 12 h of storage on ice. These observations suggest that storage on ice is an efficient method for reducing the metabolite profile changes in mud crabs.

Furthermore, instead of storage on ice, storage of dead crabs at −20℃ also mitigated the changes in metabolites (Fig. 6). For instance, the levels of muscle succinate, formate, and lactate were increased by 0.8, 2, or 3 folds relative to live control levels. Less change in the level of these two organic acids was observed in the hepatopancreas during 16 d of storage. Moreover, it is important to note that the increase in the muscle TMA level was below 30% during 16 d of storage at −20℃, with an exemption of a 76.1% increase at the last day of storage. In addition, a similar change in the level of ATP, ADP, AMP, and inosine was observed in the muscle of dead crabs at −20℃ compared with those stored on ice. When we kept the anesthetized live crabs at −20℃, the highest increase in the levels of lactate and formate in the two tissues was about two folds or less relative to live control levels during 16 d of storage (Fig. 6). Moreover, the increase in the TMA level was below 70%; an overall decline in the levels of ATP and ADP as well as an overall elevation in the levels of AMP and inosine were observed in the muscle of live crabs during 16 d of storage at −20℃. In addition, we also observed that other metabolites including certain amino acids, taurine, fumarate, histamine, betaine, and sarcosine in the muscle and/or hepatopancreas presented changes below 30% relative to those of live control crabs under four storage conditions (data not shown).

4 Discussion

Freshness is the key determinant to decide whether mud crab is edible or not. A body of evidence reveals that TMA is an important indicator of spoilage of aquatic products such as fish (Zhao et al., 2002; Kyrana and Lougovois, 2010; Heising et al., 2014), shrimp (Boziaris et al., 2011), and crab (Chen et al., 2016). This volatile base is largely responsible for a pungent odor and unpalatable taste (van Waarde, 1988). Importantly, TMA can lead to acute poisoning after consumption of the flesh of Greenland shark (Somniosus microcephalus) (Anthoni et al., 1991) and causes a teratogenic toxicity in mice (Guest and Varma, 1991). Imperatively, the rapid elevated TMA in the dead crab muscle from 8 hour postmortem at room temperature strongly indicates a big loss of freshness of mud crab. Although no legislation limit regarding TMA has been set for crustaceans, the rejection limit in fish is usually 5–10 mg per 100 g muscle (Marrakchi et al., 1990). According to the known concentration of a reference compound (TSP in this study), the semi-quantitative concentration of TMA in the muscle of RT12 crabs was (15.7 ± 3.6) mg per 100 g muscle. Although this value is lower than those of whole blue crab (Parlapani et al., 2019) and processed crab meat (Cancer pagurus) (Anacleto et al., 2011), it exceeds the rejection limit, suggesting that crab stored at room temperature exhibits an unacceptable quality at 12 h postmortem. TMA concentrations in the other crab groups were below the rejection limit, suggesting an edible quality. Generally, TMA is produced from TMAO by bacteria or enzymes in dead marine animals (Gram and Dalgaard, 2002). Our observation of the concomitantly decreased level of TMAO supports the activity of this metabolic pathway in the dead crab muscle at RT. Furthermore, TMA production has been used as an indicator of bacterial activity in crustaceans (Anacleto et al., 2011; Boziaris et al., 2011). Spoilage bacteria, such as Pseudomonas and Shewanella which are capable of producing TMA and other TVB-N, have been found in blue crab (Parlapani et al., 2019). Given this information, bacteria may play an important role in TMA production in this study. Another evidence to support the role of bacteria growth in this study is the accumulated lactate, formate, and succinate in the crab tissues. This is because these organic acids are usually produced by bacterial mix acid fermentation. Moreover, some species from Leptotrichiaceae and Natranaerovirga pectinivora, which can produce a range of organic acids, have been found in the midgut of mud crab (Zhang et al., 2020). Taken together, we speculate that bacteria grew rapidly in the crab tissues from 8 h postmortem at room temperature. As organic acids have been regarded as products of fish spoilage (Xu and Yang, 2010), the increased level of TMA and organic acids both indicate the serious spoilage of dead crab from 8 hours postmortem, which is unacceptable for human consumption.

K value is another broadly adopted freshness indicator of aquatic product which can be calculated as ratio of non-phosphorylated ATP breakdown products (inosine and hypoxanthine) to total concentration of ATP and its breakdown products (Ocaño-Higuera et al., 2011). In this study, only muscle ATP, ADP, AMP, and inosine can be detected by NMR method. The possible reason that inosine 5'-monophosphate and hypoxanthine were not detected is that the levels of these two metabolites were below the detection limit of NMR analytical method. Although the precise K value cannot be obtained based on the NMR results, a 90% decrease of ATP, 14.9% decrease of ADP, 50.7% increase of AMP, and 618.1% increase of inosine at 12 h postmortem strongly indicate an increased K value of dead mud crab at room temperature. On the other hand, inosine and hypoxanthine have been used as freshness indicators in the cold storage fish (Ocaño-Higuera et al., 2011). This is because inosine and hypoxanthine have a bitter taste (Schilichtherle-Cerny and Grosch, 1998). In contrast, AMP has a bouillon-like taste. Therefore, the decreased AMP level from 4 h postmortem and increased inosine level indicate a deteriorated flavor of mud crab at room temperature. In addition, inosine production is also related to both autolytic deterioration and bacterial spoilage (Ocaño-Higuera et al., 2011), which further supports the bacterial growth in dead mud crab at room temperature. Based on the above metabolite indices, we speculate that dead mud crab preserved an edible quality at RT for at least 4 h.

Previous study reported that TMA production is a temperature-dependent process (Shumilina et al., 2015). In this study, the decreased storage temperature on ice reduced the TMA production during 12 h of storage. However, storage at −20℃ can effectively inhibit its production until 16 d in both dead and live mud crabs. In this case, temperature seems to play a crucial role in TMA production. A low storage temperature (4℃) is also a key factor that determines the substantial elevation in TMA until day seven of the fermentation of crab paste (Chen et al., 2016). The low TMA production may be due to the inhibition of microbial growth. This is because the level of organic acids including lactate, formate, and succinate in the mud crab stored on ice were lower than those stored at room temperature. Storage at −20℃ resulted in a similar inhibitory effect on the production of organic acids as storage on ice during 16 d. On the other hand, storage OI also resulted in an inhibitory effect on ATP degradation. Furthermore, AMP degradation and inosine formation were simultaneously retarded. A similar effect on ATP degradation occurred in the muscle of live and dead crab at −20℃. However, the rate of ATP degradation was slower in live crab than those in dead crab, causing a more bouillon-like taste and less bitter taste in live crab than in dead crab during 16 d of storage. Based on these metabolite indices, we speculate that dead mud crab preserved an edible quality for at least 12 h on ice and at least 16 d at −20℃.

In addition, we also noted some changes in levels of other metabolites (such as amino acids, betaine, and taurine) in the edible tissues of mud crab associated with different storage methods. These metabolites are important nutrients for humans and probably present an attractive flavor. For instance, free amino acids usually contribute an umami, sweet, or sour taste (Park et al., 2002), whereas betaine contributes a sweet taste. However, only less than 30% change occurred in most of these metabolites, indicating a weak contribution to quality. Furthermore, histamine is frequently involved in fish poisoning (Kanki et al., 2004; Mercogliano and Santonicola, 2019). However, less than 32% increase of histamine occurred in the muscle and hepatopancreas of dead mud crab kept at room temperature, indicating a minor contribution to inedibility.

5 Conclusions

The postmortem metabolite changes in the muscle and hepatopancreas of mud crab indicate that freshness can be controlled by appropriate storage methods. NMR-based metabolomics approach coupled with multivariate data analysis is a powerful tool for assessing the quality of mud crab. Since TMA, organic acids, ATP, and its breakdown products are closely related to storage time, these metabolites can be jointly used as good indicators for freshness of crab edible tissues under different storage conditions. Our results show that dead mud crab preserves an edible quality at least for 4 h at room temperature, 12 h on ice, and 16 days at −20℃. Thus, the dead crabs should not always be discarded without considering the freshness, which are still edible if be stored in a correct manner.

Acknowledgements

Financial support from the 2025 Technological Innovation for Ningbo (No. 2019B10010), China Agriculture Research System-CARS48 and K. C. Wong Magna Fund in Ningbo University, and the Special research funding from the Marine Biotechnology and Marine Engineering Discipline Group in Ningbo University (No. 422004582) are greatly acknowledged.