**Untargeted Lipidomics Method for the Discrimination of Five Crab Species by Ultra-High-Performance Liquid Chromatography High-Resolution Mass Spectrometry Combined with Chemometrics**

**Jiaxu Yao 1, Jinrui Zhu 1, Minjie Zhao 2, Li Zhou 1,\* and Eric Marchioni <sup>2</sup>**


**Abstract:** In this study, ultra-high-performance liquid chromatography high-resolution accurate mass-mass spectrometry (UHPLC-HRAM/MS) was applied to characterize the lipid profiles of five crab species. A total of 203 lipid molecular species in muscle tissue and 176 in edible viscera were quantified. The results indicate that *Cancer pagurus* contained high levels of lipids with a docosahexaenoic acid (DHA) and eicosapntemacnioc acid (EPA) structure in the muscle tissue and edible viscera. A partial least squares discriminant analysis (PLS-DA) showed that PE 16:0/22:6, PE P-18:0/20:5, PA 16:0/22:6 and PC 16:0/16:1 could be used as potential biomarkers to discriminate the five kinds of crabs. In addition, some lipids, such as PE 18:0/20:5, PC 16:0/16:1, PE P-18:0/22:6 and SM 12:1;2O/20:0, could be used as characteristic molecules to distinguish between *Cancer magister* and *Cancer pagurus*, which are similar in appearance. This study provides a new perspective on discriminating crab species from MS-based lipidomics.

**Keywords:** seafood; molecular species; mass spectrum; quantification; identification; lipid

#### **1. Introduction**

Crab, as one of the most famous delicacies in the world, is greatly adored by people for its unique taste and distinct flavor [1]. Due to the different consumption habits, crab muscle and edible viscera are individually appreciated. The former is mainly from the claws, legs and abdomen and the latter from the hepatopancreas and gonads [2]. Crab is an excellent source of numerous nutrients essential for human health, including highly unsaturated fatty acids, proteins, minerals and vitamins [3], which bring benefits to humans, such as anti-inflammatory and immunity and cognitive enhancing properties [4].

The 2021 Power of Seafood report published by the Food Industry Association demonstrated that the overall seafood category was up 28.4% in retail around the world. In China, the demand for crab is also constantly increasing, as evidenced by the growth of crab farming from 821,000 tons in 2010 to 1,063,000 tons in 2020. Crab ranks third in global seafood production after shrimp and lobsters and has significant commercial value [5,6]. Mainly, crab is one of the most diverse of the decapod crustaceans, which can be found in both fresh and ocean waters [7]. The nutritional value of different crabs might be different due to the fact of their different nutritional components.

It has been documented that approximately 7000 species of crab survive throughout the world, either in freshwater lakes or in the ocean [8]. However, the majority of crabs are used for food, whether they are raised or wild. Commercially valuable crabs can be easily purchased for consumption, such as red king crab (*Paralithodes camtschaticus*), swimming

**Citation:** Yao, J.; Zhu, J.; Zhao, M.; Zhou, L.; Marchioni, E. Untargeted Lipidomics Method for the Discrimination of Five Crab Species by Ultra-High-Performance Liquid Chromatography High-Resolution Mass Spectrometry Combined with Chemometrics. *Molecules* **2023**, *28*, 3653. https://doi.org/10.3390/ molecules28093653

Academic Editor: Susy Piovesana

Received: 22 February 2023 Revised: 14 April 2023 Accepted: 19 April 2023 Published: 22 April 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

crab (*Portunus trituberculatus*), Chinese mitten crab (*Eriocheir sinensis*), Dungeness crab (*Cancer magister*) and brown edible crab (*Cancer pagurus*) [9].

Lipids, a major constituent of various foods, play vital roles in many cellular processes, such as energy storage, signal-mediated processes and tumor suppression [10–12]. Lipids contribute to the quality features of food products, including flavor and nutritional value. Numerous studies have focused on lipids, resulting in the emergence of lipidomics. Lipidomics, a branch of metabolomics, has superiority in providing lipid profiles of biological samples, especially those rich in lipids. In nutritional research, lipidomics has been routinely employed to elucidate the interactions among diet, nutrients and human metabolism to optimize food processing and to evaluate the nutrition of foods [13]. Consumers can establish guidelines for personalized nutrition based on lipid information [14].

Statistically, the number of publications on the lipidomics of marine animals showed an explosive increase over the past few years [15], and numerous studies have been conducted concerning the lipids of crabs. The research objectives are also focused on improving the level of beneficial lipids or survival rates by changing the dietary structure of crabs [16,17] and analyzing the lipid profile of crabs to determine their nutritional information [18]. Nevertheless, few studies have been performed comparing the lipids of multiple crab species. Thus, it is meaningful to comprehensively characterize and compare the lipid compositions among crab species from a lipidomics perspective.

In recent years, one of the most valuable technologies for lipid identification has been ultra-high-performance liquid chromatography coupled to high-resolution mass spectrometry [19], which possesses the advantages of accurate quality and excellent sensitivity. Quadrupole-Exactive high-resolution accurate mass spectrometry (HRAM) is a developed technique that can provide complete lipid molecular information without derivatization [20]. Thus, this study aimed to identify the lipid classes and lipid molecular species of the muscles and edible viscera tissue from five kinds of edible crabs by a UHPLC-HRAM/MS approach. Subsequently, the different compositions of the lipid profiles were analyzed by lipidomics combined with chemometrics. This is the first investigation involving an in-depth and comprehensive identification and comparison of the lipid profiles of different crabs. This study may help us better understand the nutritional values of edible crabs.

#### **2. Results and Discussion**

#### *2.1. Total Lipid Content of Crab Muscle and Edible Viscera*

The total lipid content of crab muscle and edible viscera are shown in Table S2. No significant difference (*p* > 0.05) was exhibited in the content of the crab muscle lipids. As for edible viscera, the five crabs exhibited a significant difference (*p* < 0.05). The total lipid content of the edible viscera from *P. camtschaticus* (766.3 mg/g) and *E. sinensis* (761.0 mg/g) showed a higher level than the other three crab lipids. Overall, the total lipid content of the edible viscera was higher than that in the muscles, which is consistent with the results of previous reports [2,21]. The reason for this is probably that crab muscles mainly store protein, while the viscera stores fat and cholesterol. The high level of total lipids in the crab samples results in biological weight gain, which is associated with consuming crab-containing diets [22]. In addition, many factors might affect the lipid content of different crab species, such as temperature [23], maturity [24] and diets [25]. The variation of the lipid content in crabs under different physiological conditions requires further experimental analysis.

#### *2.2. Validation of the UHPLC-HRAM/MS Method*

As described in this method, all lipids were quantified by the internal standard method. The internal standards were mixed with each test sample to quantify the lipid molecules, so the relative standard deviation (RSD) values of the peak areas of the internal standards could be used as a measure of the stability of the UHPLC-HRAM/MS instrument during the measurement of the sample [26]. The RSD values of the peak area of the internal standard

are displayed in Figure S1. It was found that the RSD values of 84.3% of features were less than 20%, which demonstrated that the signal was stable during the sample detection.

Further validation was carried out using seven internal standards. The linear regression equations derived from the different concentrations of the seven lipid standards are shown in Table S3, which reveals an excellent linearity (R2 ≥ 0.990). The limit of detection (LOD) value was less than 2.28 ng/mL, and the limit of quantification (LOQ) values were between 0.99 ng/mL and 7.52 ng/mL.

#### *2.3. Lipid Identification*

The muscle lipids and edible viscera lipids of the five crabs were identified by UHPLC-HRAM/MS, with the electrospray ionization (ESI) source in positive and negative ion modes. Different lipid classes were detected in specific ESI source patterns due to the fact of their polarities and electric charges [27]. The PC, LPC, PE, LPE, TAG and DAG were analyzed under the positive ion mode ([M + H]<sup>+</sup> and [M + NH4] +), while the remaining lipids were examined under the negative ion mode ([M − H]−, [M + HCOO−] − and [M − 2H]2−). The lipid profiles of the muscle (Figure 1a) and edible viscera (Figure 1b) from five crab species could be detected by ESI sources in the positive ion mode within 20 min. From the total ion chromatograms plot in the positive ion mode, the weakly polar lipid such as TAG were retained for a short time, and the retention times of the molecular species in the same lipid classes were similar. However, the different peak shapes indicate that that the contents of common lipid classes could be different in these crab samples. Figure 1c presents an example of extracted ion chromatograms (EICs) of *m/z* 876.8015, 750.5432, 832.5851 and 522.3554 from *C. magister* muscle lipids.

**Figure 1.** Total ion chromatograms of lipids in the positive ion mode from the (**a**) muscle and (**b**) edible viscera of five crabs. (**c**) Extracted individual ion chromatograms of the muscle lipids of *C. magister*, showing the species TAGs of 16:0/18:1/18:1, PE P-18:0/20:5, PC 22:2/18:5 and LPC 18:1.

The lipid molecular species were identified by MS/MS spectrometry on the basis of their characteristic mass values and fragment ions. Lipid classes can be classified by the types of polar headgroups, which exhibit characteristic fragments in the MS/MS spectrum. For example, *m/z* 184.0730 [C5H15NO4P]<sup>+</sup> is the typical fragment of the choline headgroup of PC, and *m/z* 241.0130 [C6H10O8P]− is the polar head of PI. In addition, other MS/MS fragments can provide information to infer the fatty acyl group of the lipids and the total relative molecular masses.

A total of 14 lipid subclasses and 203 molecular species were determined in the muscle lipids (Table S4) and 13 lipid subclasses and 176 molecular species in the edible viscera lipids (Table S5). Seventy-one and fifty-five common molecular species were detected, respectively, in the muscles and edible viscera of the five crabs (Figure 2a,b). The concentrations of each lipid subclass in the crab muscles and edible viscera are displayed in Tables S6 and S7, respectively. The proportion of each lipid subclass is calculated and shown in Figure S2. PE, PC and TAG were the main lipid subclasses in the crab muscle samples, while TAG was the predominant one in the edible viscera of *E. sinensis*, which accounted for more than 89%. These results are similar to that published by Wang et al. [28], who illustrated that crabs stored great numbers of TAGs in visceral organs for energy expenditure during starvation, molting or reproduction [29,30]. Especially, a relatively high level of PA was found in the crab muscles, which was not encountered in previous studies. According to research, more than 70% of PA is consumed at the time crabs mature from larvae [31]. The absence of PA might be one of the causes of the high mortality of crab larvae, so diets adding relevant lipids would be used to increase the survival rate of crabs. In comparison, phospholipids have a crucial effect on cell membranes, mainly by maintaining endogenous systems, serving as binding sites for proteins and participating in signaling [32], which play a vital role in the growth and metabolism of crabs [33,34]. Moreover, they are precursors of essential steroids [35].

**Figure 2.** Distribution of lipid molecular species in the (**a**) muscle and (**b**) edible viscera of five crabs.

It is well known that the lipids containing long-chain polyunsaturated fatty acyl chains (LC-PUFAs), such as EPA and DHA, play an irreplaceable role in brain development [36] and have properties such as antitumor [37] and anti-inflammatory [5]. In addition, LC-PUFAs exhibit better bio-efficacy and bioavailability than other lipids [38]. In this study, some DHA/EPA-TAG and DHA/EPA-PL molecules were detected in the muscle and edible viscera of the five crab species. As shown in Figure S3, the contents of DHA/EPA-PLs and DHA/EPA-TAGs (DHA/EPA-PLs + DHA/EPA-TAGs) were the highest (72.26%) in *C. pagurus* muscle, while they appeared to have the lowest levels in *E. sinensis* edible visceral tissue. The total contents of DHA/EPA-PLs and DHA/EPA-TAGs were the highest

in *C. pagurus*, with significant differences (*p* < 0.05) compared to the other groups. The DHA/EPA distribution in PLs and TAGs was consistent with the overall fatty acid composition of the plentiful PUFAs in marine animals and plant plankton [39]. Herein, our results indicate that crab muscles and edible viscera are an excellent source of DHA/EPA.

Furthermore, ether-linked phospholipids were found in the lipids of the muscle and edible viscera of the crabs, which are essential signaling molecules needed for the execution of complex life activity [40,41]. A representative MS/MS fragmentation spectrum in the positive ion mode of PE P-20:1/20:5 (*m/z* 776.5596) from *P. camtschaticus* muscle is shown in Figure 3a. This molecular species only existed in the *P. camtschaticus* muscle and was most abundant (3183 nmol/g) in plasmalogen PE. The product ion at *m/z* 141.0698 (C2H8NO4P) was the characteristic fragment of the PE molecule. The product ion at *m/z* 635.5086 corresponded to the loss of the ethanolamine headgroup. The oxygen at the sn-1 position attacked the phosphorus atom, which led to the formation of a new oxygen–phosphorous bond. At the same time, hydrogen was extracted from C-2 of the glycerol backbone, forming a double bond between C-1 and C-2 of the glycerol backbone [42,43]. Therefore, the ion fragment at *m/z* 418.3072 had the structure of 20:1 ether and C2H8NO3P, and the ion fragment at *m/z* 320.3316 was a neutral loss corresponding to H3PO4 on this basis.

**Figure 3.** MS/MS fragmentation pathway of (**a**) PE P-20:1/20:5 (*m/z* 776.5596) under a positive ion mode and (**b**) PI O-16:1/20:5 (*m/z* 867.5270), (**c**) PG 16:0/18:1 (*m/z* 747.5028) and (**d**) PA 18:1/20:5 (*m/z* 719.4657) under a negative ion mode.

The representative MS/MS fragmentation spectrum in the negative ion mode of PI O-16:1/20:5 (*m/z* 841.5156) from *C. magister* edible viscera is provided in Figure 3b. The product ion at *m/z* 301.2182 corresponds to the ion of EPA, and *m/z* 539.2991 corresponds to the loss of EPA [44]. The ion fragment at *m/z* 377.2467 was identified as the residue fragment, which was generated by the loss of the inositol headgroup of the ion fragment at *m/z* 539.2991. The ion fragment at *m/z* 557.3096 corresponded to the loss of C20:5 acyl group from the precursor ion. The ion fragment at *m/z* 241.0117 and 152.9948 were characteristic signals of the PI molecules, which were generated from the inositol headgroup. The former was a structure formed by dehydrating the head bond, while the latter was a five-membered ring structure formed by disconnecting two fatty acid chains and a six-membered ring.

PG 16:0/18:1 was detected in the muscles of all five kinds of crab, and the fragmentation pathways of PG 16:0/18:1 are displayed in Figure 3c. The ion fragment at *m/z* 747.5028 corresponds to the empirical formula C40H76O10P. The ion fragment at *m/z* 171.0054 was the C3H7O6P structure, which was the head group of the PG molecule, and the ion fragment at *m/z* 152.9949 was a five-membered ring structure formed by dehydration on this basis. As seen in Figure 3d, the PA 18:1/20:5 from *P. camtschaticus*, which had the highest level (750.6 ± 41.4 nmol/g) among all species, displayed signals at *m/z* 719.4567 in the MS<sup>−</sup> spectrum, which corresponded to the empirical formula C41H69O8P−. The ion fragments at *m/z* 281.2490 and *m/z* 301.2176 were oleic acid and EPA, respectively. The ion fragments at *m/z* 417.2411 and 4435.2516 were derived from the loss of the EPA and C20:5 acyl groups, respectively, from the precursor molecule.

#### *2.4. Multivariate Analysis*

As MS-based lipidomics generates a large amount of data in the analysis process [45], principal component analysis (PCA), a multivariate statistical method, was used to examine the correlations among the lipidomics data with multiple variables and to compare the lipid molecules and contents of the different crabs to facilitate the analysis and visualization. Figure 4a presents the PCA score plots (R2X = 0.994, Q2X = 0.982) of the crab lipids in the different muscles, displaying the clustering of each sample in the first two principal component (t1 and t2) score plots, accounting for 0.367 and 0.251 of the total variance, respectively. The PCA score plots (R2X = 0.997, Q2X = 0.985) of the edible visceral lipids in the different crabs indicate that the first two principal components (t1 and t2) accounted for 0.356 and 0.242 of the total variance. From the PCA score plot (Figure 4a), the five crab types were nonoverlapping with each other, which indicates that the metabolites of each crab were discrepant. Therefore, the PCA model had good separation and clustering results and can be used as a method to effectively distinguish different samples.

Furthermore, the separations were tested in the PLS-DA score plot. R2 represents the degree of fitting between the model and crab lipid data, and Q<sup>2</sup> (cumulative) represents the prediction ability of the model for new data [46]. For the crab muscle samples, the PLS-DA score plot is displayed in Figure 4b (R2X = 0.998, R2Y = 0.999, Q2 = 0.997), which demonstrates a good separation effect and prediction ability. After 999 permutations (Figure 4c), the values of R2 = (0.0, 0.062) and Q2 = (0.0, −0.823) of this model were more prominent than the random values of all samples, which is similar to the PLS-DA score plot of the edible visceral samples. All these data underlie the classification of the five crab species and show a good predictive ability for a new data set.

Subsequently, the value of VIP (variable importance in projection) was applied to screen for significant differences in the lipids of the five crab species, which was available from the PLS-DA model. In the crab muscle model, a total of 57 lipid molecules with VIP > 1 were analyzed by cluster analysis to assess the similarity of the crab samples (Figure 4d). According to the dendritic diagram, *C. magister* and *C. pagurus* was in one group. At the same time, 51 lipid molecules met VIP > 1, and *C. pagurus* and *P. camtschaticus* belonged to one group based on the similarities in the composition of the edible viscera. The results were also well reflected in the PCA diagram. Thus, studying the pedigree relationship of lipidomics characteristics can provide a reference for species classification.

Additionally, one-way analysis of variance (ANOVA) was conducted to determine the final statistically significant (*p* < 0.05) lipid species. In general, markers need to meet *p* < 0.05 and VIP > 1 [47], which could be identified as potentially characteristic metabolites, thus achieving the ability to identify species accurately. In the crab muscle model, 26 lipid molecules were selected as markers (Table S8), such as PE 16:0/22:6, PE P-18:0/20:5, PA 16:0/22:6 and PC 16:0/16:1. The distribution of 26 lipids in the muscle content of the five species of crab is shown in Figure S4a. For example, when the detected range of PE 16:0/22:6 was 8000 nmol/g, the crab species can be judged as *E. sinensis*. In the samples of edible viscera, 17 lipid molecules were selected as markers (Table S8), including TAG 16:0/20:1/18:2, SM 14:1;2O/22:0, PE 18:1/20:5 and TAG 16:0/18:1/20:1, whose distributions are shown in Figure S4b.

**Figure 4.** *Cont*.

**Figure 4.** (**a**) PCA score plot of identified lipids in muscle with R2X = 0.994 and Q2X = 0.982 and edible viscera with R2X = 0.997 and Q2X = 0.985. (**b**) PLS-DA score plot of identified lipids in muscle with R2X = 0.998, R2Y = 0.999 and Q2 = 0.997 and edible viscera with R2X = 0.998, R2Y = 0.999 and Q2 = 0.998. (**c**) Permutation constructed on the basis of the PLS-DA of the contribution molecular species in muscle and edible viscera, and (**d**) Hierarchical cluster dendritic diagram of muscle and edible viscera and hierarchical cluster analysis based on molecular species in muscle and edible viscera with VIP > 1. Colors represent different concentrations indicated by the color bar.

#### *2.5. Orthogonal Partial Least Squares Discrimination Analysis*

Due to the highly similar appearance of *C. magister* and *C. pagurus*, the two species were often confused without relevant books and professionals. Here, the differences between the two crabs were analyzed from a lipidomics perspective combined with a metrological approach.

In this study, the principle for the selection of significantly different lipid species between *C. magister* and *C. pagurus* is a sufficiently high variable importance for the projection (VIP > 1.3) and the univariate statistical analysis, including a fold change (log2(FC) ≥ 1 or ≤−1) and T-test criteria (*p* < 0.05). Overall, 174 lipids from muscles were calculated and are

shown in volcano plots in Figure 5a, among which 78 lipids were upregulated and 68 were downregulated in *C. magister* compared with *C. pagurus*. Analogously, Figure 5b demonstrates that 76 lipids from edible viscera were upregulated and 43 were downregulated. In addition, the VIP values of each lipid molecule were obtained from orthogonal partial least squares discrimination analysis (OPLS-DA). Figure 5c,d represent the VIP value map of some quantified metabolites from the muscles and edible viscera, respectively. Finally, 20 lipid molecules from muscles and 17 lipid molecules from edible viscera were selected as the characteristic molecules to distinguish *C. magister* and *C. pagurus* (Table 1).

**Figure 5.** An orthogonal partial least squares discrimination analysis was performed for *C. magister* and *C. pagurus*. Volcano plots of lipids from (**a**) muscle and (**b**) edible viscera. VIP maps of lipid molecules in (**c**) crab muscle and (**d**) edible viscera. Dark blue indicates VIP > 1.


**Table 1.** The significantly different lipids in the muscles and edible viscera of *C. magister* and *C. pagurus*.

VIP: variable importance in projection; FC: fold change.

#### **3. Materials and Methods**

#### *3.1. Ethical Statement*

In this study, all experimental animals were adequately cared for, their pain minimized, and killed painlessly. In addition, the number of all animals was controlled to the minimum required to obtain scientific results.

#### *3.2. Sample Preparation*

Five living female farmed crab species (Figure S5), including *P. camtschaticus* (*Paralithodes camtschaticus*), *E. sinensis* (*Eriocheir sinensis*), *C. magister* (*Cancer magister*)*, P. trituberculatus* (*Portunus trituberculatus*) and *C. pagurus* (*Cancer pagurus*), were purchased from a local seafood market in Wuhan, China, in October 2020, with the help of professionals (12 fresh crabs of each species, individually weighed, *P. camtschaticus*: 1500–1700 g; *E. sinensis*: 150–200 g; *C. magister*: 750–800 g; *P. trituberculatus*: 250–300 g; and *C. pagurus*: 900–1000 g). The ages of each species of crab shell were similar, and all crab limbs were free of damage. According to a previous study [48], all crabs were electrocuted through 110 volts and 2 amps of current for 10 s with a Crustastun machine (Studham Technologies, Scotland, UK) and killed immediately. The edible viscera and muscle were manually separated from the crabs, immediately preserved in liquid nitrogen and then lyophilized using a vacuum freeze-dryer (Labconco Corporation, Kansas, MO, USA) with a vacuum of 0.06 MPa. The freeze-dried crab tissues were stored at −80 ◦C until analysis.

#### *3.3. Standards and Reagents*

The PL standards were purchased from Avanti Polar Lipids, including phosphatidylcholine (PC 17:0/17:0), phosphatidylethanolamine (PE 17:0/17:0), phosphatidylglycerol (PG 17:0/17:0), phosphatidic acid (PA 17:0/17:0), phosphatidylinositol (PI 8:0/8:0), sphingomyelin (SM d18:1/12:0) and SPLASH LIPIDOMIX (isotopic internal standard mixture). Glyceryl trilinoleate (TAG) was purchased from Sigma Aldrich. Other chemicals, including methanol, chloroform, acetonitrile and ammonium formate, were purchased from Macklin Biochemical Technology Co., Ltd. (Shanghai, China) (high-performance liquid chromatography or analytical reagent grade).

#### *3.4. Lipid Extraction*

All biological replicates were sourced from 12 crabs of the counterpart species, and a triplicate analysis was performed for each crab species. Total lipids were extracted using the methods described by Yang et al. with minor modifications [49]. Briefly, the freeze-dried edible viscera and muscle tissues were weighed and homogenized in 10 mL of methanol in a homogenizer (Bead Mill 4, Fisherbrand, Tokyo, Japan) for 30 s (×2 cycles). Subsequently, 10 mL of the homogenate (1 g) was transferred to a beaker, and 20 mL of chloroform was added. The mixture was placed in an ultrasonic bath and processed for 30 min at 20 ◦C. Then, the mixture was centrifuged at 4000× *g* for 10 min. The supernatant was transferred to a clean bottle, and the lower phase was re-extracted twice with chloroform:methanol (2:1, *v/v*). Finally, the organic solvents were combined and evaporated under a vacuum rotary evaporator (40 ◦C, 350 kPa). The residue was further dried using a gentle stream of nitrogen. Dried lipid extracts were weighted and stored at −20 ◦C for further analysis by UHPLC-HRAM/MS. All analyses were completed within two weeks after extraction.

#### *3.5. Lipid Identification and Semi-Quantification*

Lipid molecular species were identified with MS-DIAL software (version 4.48) and Xcalibur 4.0 software (ThermoFisher, Waltham, MA, USA), as well as the LipidMaps and LipidBank websites. False positive data were manually excluded based on database matching and fragmentation information.

Seven PL standards, including PC 17:0/17:0 (standard for PC and lysophophatidylcholine (LPC)), PE 17:0/17:0 (standard for PE and lysophosphatidylethanolamine (LPE)), PG 17:0/17:0 (standard for PG, lysophosphatidylglycerol (LPG) and cardiolipin (CL)), PA 17:0/17:0 (standard for PA and lysophosphatidic acid (LPA)), SM d18:1/12:0 (standard for SM), PI 8:0/8:0 (standard for PI and lysophosphatidylinositol (LPI)) and TAG 18:2/18:2/18:2 (standard for TAG and diacylglycerol (DAG)), were used for the construction of the calibration curves. These standards were quantitatively dissolved in a chloroform:methanol (2:1, *v/v*) mixture as stock solutions and then diluted into a series of concentrations, with a constant concentration of internal standards from SPLASH LIPIDOMIX (as displayed in Table S1 of the Supplementary Materials), including PC 15:0/18:1(d7), PG 15:0/18:1(d7), PE 15:0/18:1(d7), TAG 15:0/18:1(d7)/15:0, PI 15:0/18:1(d7), SM d18:1/18:1(d9) and PA 15:0/18:1(d7). The internal standard method was applied to quantify the lipid concentrations of crab samples [50]. Briefly, the same concentration of internal standards was added to each crab lipid, and the samples were injected for detection. The crab lipids were quantified according to the calibration curves of the peak area ratio of the standard and internal standard within the same class so the concentration (mg/mL) of lipids could be obtained. Subsequently, the concentration (nmol/g) of each lipid molecule in a dried biological sample was available by calculation. Furthermore, all molecular quantifications were based on the EICs of individual lipids, with *m/z* expansion at ±5 ppm, and all crab samples were analyzed with three duplicate samples.

#### *3.6. UHPLC-HRAM/MS Analysis*

#### 3.6.1. UHPLC Conditions

The total lipids of the different crabs were identified using a UHPLC system (Dionex, UltiMate, 3000 RSLC) coupled with QE Orbitrap/MS (ThermoFisher Scientific, Bremen, Germany). A BEH-HILIC column (100 × 1.0 mm, 1.7 μm, Sigma–Aldrich/Supelco, Bellefonte, PA, USA) was used for chromatographic separation. The flow rate was 0.1 mL /min, and the column temperature was 40 ◦C. Eluent A and eluent B were 5 mM ammonium formate in water and in acetonitrile, respectively. The chromatographic gradient elution mode was as follows: 0 min, 5% B; 4 min, 5% B; 10 min, 40% B; 15 min, 40% B; 16 min, 5% B; and 20 min, 5% B. The injection volume was 2 μL.

#### 3.6.2. Quadrupole-Exactive High-Resolution Accurate Mass Spectrometry

The desolvation ESI source parameters were set as follows—electrospray voltage: 3.2 kV; a sheath gas flow rate: 35 arbitrary units (arb unit); auxiliary gas flow rate: 10 arbs; capillary temperature: 325 ◦C; heater temperature: 350 ◦C; collision energy: 35 eV; dynamic exclusion: 10 s; and isolation window: 3.0 *m/z*. The data were acquired in the positive ionization mode (*m/z* 120–1800) and negative ionization mode (*m/z* 120–1800) with dependent MS/MS acquisition. The resolutions of the full-scan spectra and the fragment spectra were 140,000 and 70,000, respectively. All lipid identifications were determined by MS and MS/MS, with an MS mass error of <5 ppm and MS/MS mass error of <8 ppm.

To investigate the background contamination and the method's validation, quality control procedures were applied to this experiment, which contained solvent blanks, solvent spiked with internal standards and matrix blanks without spiking the internal standards [51]. To test the stability, the RSD values of the peak areas of the internal standards in three parallel samples were calculated. The LOD and LOQ were the concentrations of the standards corresponding to the signal-to-noise (S/N) values of 3 and 10 [52].

#### *3.7. Statistical Analysis*

The data are expressed as the mean ± SD. SPSS software (version 24.0) was used for the one-way analysis of variance, and *p* < 0.05 was considered statistically significant. Lipid molecule species were identified and quantified with MS-DIAL software (version 4.48) and Xcalibur 4.0 software (ThermoFisher, Waltham, MA, USA), as well as the LipidMaps and LipidBank websites. MetaboAnalyst 5.0 was used for the cluster analysis, and SIMCA 14.1 was used for the PCA, PLS-DA and OPLS-DA.

#### **4. Conclusions**

This study applied the UHPLC-HRAM/MS approach to discriminate the lipid molecules in the muscles and edible viscera of five kinds of crabs, including *P. camtschaticus*, *E. sinensis*, *C. magister*, *P. trituberculatus* and *C. pagurus*. Combined with chemometric analysis, 14 lipid subclasses in muscle tissue and 13 lipid subclasses in edible viscera were detected and quantified. *C. pagurus* had the highest DHA/EPA-PL molecule content in the muscle and edible viscera tissue of five crab species. Our work can help researchers better understand the lipid composition of crab muscle and edible visceral tissues and better select or assess their nutritional values.

In addition, we developed a new way to distinguish two kinds of crabs with similar appearances (*C. magister* and *C. pagurus*) based on lipidomics. In the muscle samples, twenty characteristic lipids were screened, such as PE 18:0/20:5, PC 16:0/16:1, PE P-18:0/22:6 and SM 12:1; 2O/20:0, as the basis for distinguishing *C. magister* and *C. pagurus*. Seventeen characteristic lipids, such as PE P-18:0/22:6, PC 16:1/18:1, PC 18:1/18:1 and PE P-18:0/20:5, were screened from the edible visceral samples. However, our work focused on the differences in the crab varieties, while some factors that might affect the lipid composition, such as maturity and living environment, were not considered. Therefore, the effects of various factors on the lipid composition of crabs will be investigated in further study.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules28093653/s1, Figure S1: RSD distribution of the peak area of seven internal standards in 10 groups of different samples; Figure S2: Lipid composition (%) in the muscles and edible viscera of five crab species; Figure S3: The percentage contents of DHA/EPA-PLs and DHA/EPA-TAGs in the total lipids extracted from crab muscle and edible viscera. The data are expressed as the mean ± SD. Different letters indicate a significant difference (*p* < 0.05) among the different samples; Figure S4: Boxplots of the significantly different lipids in the muscles (A) and edible viscera (B) of five crab species; Figure S5: Optical images of the crab species; Table S1: Concentration (ng/mL) of 7 lipid standards and corresponding internal standard (IS) for the construction of calibration curves; Table S2: Total lipids content in five kinds of crabs; Table S3: The linear regression equations, linear regression coefficients (R2), LOD and LOQ of lipid standards; Table S4: Molecular species of lipids extracted from the muscles of five kinds of crabs; Table S5: Molecular species of lipids extracted from the edible viscera of five kinds of crabs; Table S6: Concentrations of the different lipid subclasses in the five kinds of crab muscles (nmol/g); Table S7: The concentrations of different lipid subclasses of the five kinds of edible viscera (nmol/g); Table S8: The significantly different lipids in the muscle and edible viscera of five crab species.

**Author Contributions:** J.Y. and J.Z. were responsible for the experimental study, data analysis and manuscript writing; M.Z. was responsible for the revision of the manuscript; L.Z. and E.M. were responsible for the overall design, planning of the project and revision of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants from the National Natural Science Foundation of China (31501521) and the Natural Science Foundation of Hubei Province (2022CFB258). The authors thank the Analytical and Measuring Center, School of Pharmaceutical Sciences, South-Central Minzu University, for the spectral analyses.

**Institutional Review Board Statement:** Animal use and care were confirmed by the IAEC of South-Central Minzu University (SYXK (Wuhan) 2016-0089, no: 2021-SCUEC-AEC-033), Wuhan, China.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data that support the findings of this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

**Sample Availability:** Not applicable.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

## *Article* **Histamine and Tyramine in Chihuahua Cheeses during Shelf Life: Association with the Presence of tdc and hdc Genes**

**Eduardo Campos-Góngora \*, María Teresa González-Martínez †, Abad Arturo López-Hernández , Gerardo Ismael Arredondo-Mendoza, Ana Sofía Ortega-Villarreal and Blanca Edelia González-Martínez \***

† This paper is written in loving memory of Ma. Teresa González Martínez, a great human being and skillful researcher.

**Abstract:** Cheese is a product of animal origin with a high nutritional value, and it is one of the most consumed dairy foods in Mexico. In addition, Chihuahua cheese is the most consumed matured cheese in Mexico. In the production process of Chihuahua cheese, maturation is carried out by adding acid lactic microorganisms, mainly of the *Lactococcus* genus and, in some cases, also the *Streptococcus* and *Lactobacillus* genus. As part of the metabolism of fermenting microorganisms, biogenic amines can develop in matured foods, which result from the activity of amino decarboxylase enzymes. In cheeses, histamine and tyramine are the main amines that are formed, and the consumption of these represents a great risk to the health of consumers. In this work, the presence of biogenic amines (histamine and tyramine) was determined by HPLC at different times of the shelf life of Chihuahua cheeses. In addition, the presence of genes hdc and tdc that code for the enzymes responsible for the synthesis of these compounds (histidine and tyrosine decarboxylase, or HDC and TDC) was determined by molecular techniques. A significant correlation was observed between the presence of both histamine and tyramine at the end of shelf life with the presence of genes that code for the enzymes responsible for their synthesis.

**Keywords:** Chihuahua cheese; histamine; tyramine; hdc and tdc genes; HDC and TDC proteins

#### **1. Introduction**

Cheese is a dairy product with a high nutritional value. Since it is one of the most consumed foods of animal origin, it is part of the basic food basket in several Latin American countries. In Mexico, a wide variety of cheeses from different regions of the country (Panela, Oaxaca, Cotija, Asadero, Chihuahua, Sierra, etc.) are consumed, with fresh cheeses being the most consumed. Within the group of matured cheeses, the most consumed is Chihuahua or Mennonite cheese, so-called because it was originally made by this resident community of the state of Chihuahua. Although this type of cheese originated in Chihuahua, it is currently prepared by different companies throughout the country, and it is also exported to the United States to meet the demands of the Hispanic community [1,2].

In the dairy industry, the starter cultures used to produce fermented products (cheese, cottage cheese, butter, kefir, yogurt, etc.) are microorganisms that are used to ferment glucose, transforming it into lactic acid. The microorganisms that are most used in these processes are lactic acid bacteria (LAB), whose main characteristic is that they cause a decrease in pH, which inhibits the growth of other microorganisms in fermented food, also promoting the coagulation of casein [3]. Starter cultures grow from the beginning of inoculation until reaching very high cell densities (108–109 CFU/mL) in the first hours of fermentation. Subsequently, there is a gradual decrease throughout the cheese maturation process. Although starter cultures can be made up of different types of microorganisms, mesophilic bacteria cultures of the genera *Lactococcus, Lactobacillus,* and *Leuconostoc*, or

**Citation:** Campos-Góngora, E.; González-Martínez, M.T.; López-Hernández, A.A.; Arredondo-Mendoza, G.I.; Ortega-Villarreal, A.S.; González-Martínez, B.E. Histamine and Tyramine in Chihuahua Cheeses during Shelf Life: Association with the Presence of tdc and hdc Genes. *Molecules* **2023**, *28*, 3007. https:// doi.org/10.3390/molecules28073007

Academic Editor: Andrea Salvo

Received: 25 February 2023 Revised: 14 March 2023 Accepted: 22 March 2023 Published: 28 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Universidad Autónoma de Nuevo León, Centro de Investigación en Nutrición y Salud Pública, Facultad de Salud Pública y Nutrición, Monterrey 64460, Nuevo León, México

**<sup>\*</sup>** Correspondence: eduardo.camposg@uanl.mx (E.C.-G.); blanca.gonzalezma@uanl.mx (B.E.G.-M.)

thermophilic species of *Lactobacillus,* such as *L. delbrueckii, L. helveticus*, or *Streptococcus thermophilus*, are generally used for the maturation of cheeses [4,5].

During cheese maturation, LABs are not only responsible for the production of lactic acid, but they also participate in proteolysis by hydrolyzing polypeptide chains and releasing amino acids that serve as precursors for the formation of volatile compounds largely responsible for the organoleptic properties of cheese [6]. The amino acids produced are catabolized by an initial pathway in which an aminotransferase enzyme acts, followed by two catabolic pathways where deamination or decarboxylation occurs. While deamination generates aromatic products due to the action of dehydrogenases (α-ketoacid and ammonia) or oxidases (aldehyde and ammonia), by-products of decarboxylation of biogenic amines can be generated (histamine, tyramine, putrescine, cadaverine, etc.). Some of these compounds can cause adverse physiological effects in susceptible consumers [7–9].

The activity of decarboxylase enzymes depends on the genetic potential, as well as other factors such as the pH of food (slightly acidic), availability of free amino acids, presence of fermentable carbohydrates (glucose), presence of pyridoxal phosphate (cofactor of the decarboxylation reaction), maturation temperatures, humidity, and a low concentration of salts, which favors the action of bacteria with amino decarboxylase enzymes [10–12]. Although temperatures above 15 ◦C favor the development of microorganisms and, therefore, of biogenic amines, the activity of the amino decarboxylase enzyme has been reported at temperatures of 4 ◦C [13–16].

The formation of histidine and tyramine (the biogenic amines that present a greater health risk) during the metabolism of LABs used in the elaboration of cheese is due to the presence of histidine decarboxylase and tyrosine decarboxylase, the enzymes responsible for their synthesis encoded by the hdc and tdc genes, respectively. Under normal conditions, the concentrations of biogenic amines in food do not present a health risk because the digestive system of the human organism has an efficient mechanism to eliminate these molecules through the enzymes monoamine oxidase (MAO) and diamine oxidase (DAO). However, under specific conditions, such as high concentrations of amines or in individuals that present deficiency or inhibition of these enzymes, there may be some risk of adverse reactions [17]. Symptoms of histamine poisoning occur in manifestations of the digestive system, such as nausea, vomiting, and diarrhea, in addition to hypotension, edema, tachycardia, headaches, and migraines [18,19].

Fermented cheeses have also been linked to severe hypertensive crises in patients who take medications that inhibit the enzyme monoamine oxidase (MAO). The amine related these events is tyramine, therefore, these clinical manifestations are called "cheese reactions" [20,21]. Although the symptoms appear from within a few minutes to three hours of ingestion of the contaminated product, it has been described that its duration depends, in large part on: (1) The physiological condition of the patient, reestablishing in hours or up to several days; and (2) The concentrations of histamine and other amines ingested with food [22]. On the other hand, it has been reported that an intake of 5–10 mg of histamine can be considered a risk factor for sensitive people, from 10 to 100 mg is considered a tolerable limit, from 100 to 1000 mg produces a medium intoxication, and amounts greater than 1000 mg produce severe intoxication [23]. Despite the toxic effects related to the biogenic amines, histidine and tyramine, in Mexico, there is no regulation of the allowable limits of these amines in cheeses. It is noteworthy that the production of biogenic amines is carried out mainly during the maturation process of Chihuahua cheese, which covers an approximate period of two months at temperatures between 4 and 27 ◦C. The production of these compounds continues during the storage of the cheese, usually at refrigeration, between 10 and 20 ◦C [24,25].

The aim of this study was to evaluate the concentration of histamine and tyramine using High-Performance Liquid Chromatography (HPLC) in Chihuahua cheeses at different times of their shelf life, and to determine the presence of the genes that code for the enzymes responsible for the synthesis of these compounds in cheese samples using molecular techniques (Polymerase Chain Reaction: PCR).

#### **2. Results**

#### *2.1. Microbiological Quality of the Cheese Samples*

Results corresponding to the microbiological quality of cheeses are presented in Table 1. As expected in matured products, all the cheeses have a high LAB content (ranging from 7.6 to 9 Log CFU/g). The total count of aerobic mesophilic bacteria was found between 3.1 and 3.8 Log CFU/g, and the total coliform bacteria was between 2.6 and 3.4 Log CFU/g. Statistical analysis showed a significant difference between the different brands.


Capital letters represent each one of included cheeses. Values correspond to average ± standard deviation of Log CFU/g in three independent experiments. Different superscripts a–e indicate significant differences between cheese brands (*p* < 0.05). Log CFU/g: Logarithm of Colony Forming Unit per gram. LAB: Lactic Acid Bacteria.

At the end of the shelf life, the microorganisms recognized as histamine and tyramine producers were identified by API biochemical system. The identification of bacteria was performed up to the species level; the results are presented in Table 2. *L. pentosus* was found in three different brands, while *L. rhamnosus* was found in two brands.

**Table 2.** Biochemical identification of strains recognized as histamine and tyramine producers in the Chihuahua cheeses.


Identification of bacteria based on API biochemical tests at the end of the shelf life. Capital letters represent each different included cheese. HDC: histidine decarboxylase; TDC: tyrosine decarboxylase; ND: bacteria were not detected.

#### *2.2. Histamine and Tyramine Quantification*

The biogenic amine quantification was carried out by the HPLC method with a fluorescence detector (FLD). The retention times for histamine and tyramine were 15 and 30 min, respectively. Calibration curves carried out with the standards showed linearity with different concentrations (25, 50, 100, 250, and 400 mg/L) of compounds. A correlation coefficient of 0.993 was determined for histamine and 0.987 for tyramine. Based on these values, a limit of detection (LOD) of 25 mg/L was determined for both amines under the experimental conditions of this study.

HPLC analysis was performed at three moments in the shelf life of cheeses (Table 3). In the first week of its acquisition (beginning of the shelf's life), the presence of his-

tamine was not detected in any of the cheeses, while tyramine was present in 37.5% of the cheeses (Cheeses B, C, and F). Tyramine presence reached concentrations ranging from 34–122 mg/kg. At half of shelf life, the presence of histamine was detected only in one of the cheeses (12.5% of cheeses), while tyramine was detected in 62.5% of the cheeses (five out of eight). Interestingly, it was observed that the concentration of tyramine in Cheeses B, C, and F (in which tyramine was detected from the first analysis) increased significantly to reach values of 160, 160, and 139 mg/kg, respectively. At this stage, in Cheese A, both histamine and tyramine were detected at concentrations between 45 mg/kg and 59 mg/kg, respectively.


**Table 3.** Detection of biogenic amines in Chihuahua cheeses.

Biogenic amine detection was performed by HPLC at three moments in the shelf life of cheeses. Capital letters represent each one of included cheese. Values correspond to average ± standard deviation of amines concentration (mg/kg) determined in three independent experiments. mg/kg: correspond to amines quantity by kilogram of cheese.

At the end of the shelf life (approximately four months after its acquisition) histamine was detected in three of the eight cheeses (A, D, and G), which corresponds to 37.5% of them. In Cheese A, histamine concentration increased by more than 400% (from 45 to 192 mg/kg). On the other hand, tyramine was detected in 75% of cheeses (six out of eight) in concentrations from 115 mg/kg to 209 mg/kg.

It is noteworthy that while the concentration of tyramine increased in Samples B, C, D, and F over time, in Cheeses E and H, the presence of histamine and tyramine was not detected at the different storage times analyzed.

#### *2.3. Detection of hdc and tdc Genes in Cheese Samples*

For the determination of bacterial DNA in the cheese samples, the first-stage PCR amplification was performed using the specific oligonucleotides PO (GAGAGTTTGATC-CTGGCTCAG) and 338-F (GCTGCCTCCCGTAGGAGT), which were reported by Ventura, et al., (2001) [26]. These oligonucleotides allow for the amplification of a 332 base-pair fragment corresponding to the gene that encodes the 16S ribosomal subunit of eubacteria. In all the cheeses analyzed, the expected PCR products were obtained (data not shown), indicating the presence of bacterial DNA in all cheese samples.

For the specific analysis of hdc gene presence, specific oligonucleotides (HDC1: TTGACCGTATCTCAGTGAGTCCAT and HDC2: ACGGTCATACGAAACAATACCATC), which were designed and reported by Fernandez et al., (2006) [27], were used. In addition, the obtained PCR products (bands of 174 bp) were found only in three of the samples (Cheeses A, D, and G), showing that 37.5% of cheeses analyzed contain bacteria able to produce HDC (Figure 1A).

**Figure 1.** PCR amplification of histidine decarboxylase (**A**) and tyrosine decarboxylase (**B**) genes in LAB present in Chihuahua-type cheeses. (**A**) Amplification of histidine decarboxylase gene. Oligonucleotides HDC1 and HDC2 were used to amplify a 174-bp fragment of hdc gene in the bacterial DNA present in the different cheese samples (line A-H); genomic DNA corresponding to *Lactobacillus* sp. 30a strain was used as positive PCR-control (lane CTL+). (**B**) Amplification of tyrosine decarboxylase gene. Degenerated oligonucleotides TER-F and TER-R were used to amplify a 252-bp fragment of tdc gene from bacterial DNA present in the different cheese samples (line A-H) genomic DNA corresponding to *L. brevis* strain was used as positive PCR-control (lane CTL+). DNA molecular weight marker was included in the electrophoresis analyses (line MW).

On the other hand, to test the usefulness of degenerated primers designed to specifically amplify the tdc gene in LAB, DNA obtained from the cells of *L. brevis* (reference strain) was used as a template (Figure 1B, Line CTL+). PCR reactions using the degenerate oligonucleotides TER-F (GCWAAYYTDGARGGDYTHTGGTATGC) and TER-R (CCAWGAATART-GYTTHGTTTGTGG) were performed at different alignment temperatures (50–60 ◦C). As PCR products, DNA bands of the expected size (252 bp) were obtained. In six of eight cheeses (A, B, C, D, F, and G), the corresponding band to the tdc gene was observed, suggesting the presence of bacteria carrying the tdc gene, in 75% of the cheeses analyzed (Figure 1B).

#### **3. Discussion**

Chihuahua cheese is the matured cheese more consumed in Mexico and, in general, in several Latin American countries. Lactic acid bacteria are used in the Chihuahua cheese elaboration. However, LAB can produce biogenic amines during the process, causing adverse physiological effects in cheese consumers [9]. In Mexico, there is no regulation about the permitted levels of these compounds. In addition, there is no standardized process allowing the biogenic amines identification or species capable of producing them in this type of food.

Biogenic amine concentration can vary between cheeses since it depends on different factors during its elaboration: the quality of the raw material, the presence of strains used as starters, or the presence of contaminating microorganisms with amino decarboxylase activity. It has been reported that certain bacteria produce biogenic amines: *L. buchneri, L. fermentum, L. helveticus, L. rhamnosus L. brevis* [28], *L. fermentum, L. plantarum, L. helveticus* [29], *L. sakei,* and *L. pentosus* [30]. These species are commonly found in cheeses and are used as starter lactic acid bacteria cultures. On the other hand, both histamine and tyramine

formations in cheeses have also been related to the presence of different bacterial species known as non-initiating lactic acid bacteria, mainly *Lactobacillus curvatus* and *L. lactis lactis*, respectively [28,30,31].

In this work, it has been demonstrated that the presence of biogenic amines in samples (histamine and tyramine) of Chihuahua cheese at different storage times. During the first week of shelf life, the histamine concentration was low or undetectable by the HPLC methodology. On the other hand, tyramine was found in concentrations ranging from 34–122 mg/kg in three of the eight samples. During the mid-shelf-life analysis, histamine and tyramine were detected in 12.5% and 62.5% of the samples, respectively. In the final tests (end shelf life), 37.5% of the cheeses were positive for the presence of histamine, and 75% for tyramine.

These results suggest that biogenic amine production takes place not only during the maturation process of the cheeses, but that such compound production continues during their shelf life. This phenomenon has already been reported by Diaz–Cinco et al., (1992) [32], who stated that cheese samples in their study were stored between 5–25 ◦C for 12 days, and the concentrations of histamine and tyramine increased both in higher temperatures and longer storage times. It is noteworthy that samples in our study were always kept at 1 ◦C during the storage period. This could indicate that even when cheeses are stored at low temperatures, the production of amines continues. The effect of low temperatures is well-known in delaying the growth rate of several bacterial microorganisms. However, such a condition is probably not adequate for the inactivation of enzymes responsible for amine production.

Histamine and tyramine concentrations in the different cheese samples showed marked variations despite belonging to the same type of cheese. These variations could be because cheeses were made by different manufacturers who probably used different starter cultures (or different strains), variations in the production process, and conservation of raw material.

In previous studies on matured cheeses, tyramine has been highlighted as the most frequent of the biogenic amines [33–35]. Similarly, in this study, tyramine was detected in six of the eight analyzed samples at the end of shelf life, while histamine was detected only in three of eight cheese samples. This biogenic amine is considered of particular interest due to the vasoactive effect it produces on susceptible consumers who have inhibition of MAO (monoamine oxidase enzyme). It has been described that the intake of 6 mg of tyramine showed mild effects, while the intake of 10–25 mg increased the risk of hypertension when ingested in combination with MAO inhibitors [36].

Despite the well-documented toxic effects of biogenic amines, current regulations in Mexico related to the maximum concentration of histamine only exist for fresh and processed fish, and there are no regulations for tyramine. According to the Norma Oficial Mexicana (NOM-242-SSA1-2009) [37], the limit of histamine is 100 mg/kg, which is similar to the value established by the European Union for fermented foods [38]. Within this margin, only a sample of Chihuahua cheeses analyzed in this study did not comply with this regulation. The Food and Drug Administration (FDA, Silver Spring, MD, USA) sets a maximum limit of 50 mg/kg for histamine in foods and considers of risk the concentrations between 50–200 mg/kg [39]. According to this, it could be stated that at the end of shelf life, 37% of Chihuahua cheeses may be a health risk (since three cheese samples presented higher concentration levels).

Using degenerated primers allowed for the detection of bacterial species capable of producing biogenic amines along the shelf life of cheeses. The presence of these LAB species at early shelf life could not be detected by routine microbiological methods, probably because the number of bacteria was not enough for its detection by these methods. However, such bacteria quantity could be detected by PCR due to the detection limit and sensibility this technique possesses. In our results, the use of degenerated primers agrees with the detection of tyramine by the HPLC technique (six of eight cheese samples), but it is

interesting to note that the HPLC detection in six samples was performed only at the end of the shelf life.

#### **4. Materials and Methods**

#### *4.1. Sample Origin*

Eight Chihuahua kinds of cheese of different brands marketed in the Monterrey, N.L., Mexico metropolitan area were selected and analyzed. Cheese samples were kept at refrigeration temperature (2–8 ◦C) during transportation, and they were kept refrigerated at 1 ◦C throughout the study. A one-letter code was assigned to each sample. Relevant information was recorded (the elaboration and expiration date and the production batch). Three samples of each cheese were selected, and the analyses were carried out in triplicate.

#### *4.2. Microbiological Analyses*

Microbiological analyses were carried out during the first week of the acquisition of the cheeses. The analyses were performed according to the criteria of the Norma Oficial Mexicana (NOM-121-SSA1-1994) [40], which establishes sanitary specifications for fresh, matured, and processed cheeses, and the Norma Mexicana (NMX-F-209-1985) [41], which establishes specifications of the product named "Chihuahua-type cheese".

The aerobic mesophilic bacteria count was performed by seeding in trypticase soybean agar, and samples were incubated at 37 ◦C for 24 h. For the coliform bacteria cultures, the specific culture medium Rida® COUNT coliform (R-Biopharm AG Darmstadt, Germany) was used; the cultures were incubated for 24 h at 35 ◦C, and the presence and count of *Escherichia coli* were determined using Petrifilm *E. coli*/Coliform Count Plate (3M, Minnesota, USA) following the procedure recommended by the manufacturer. The count of LAB was made using cheese samples of 10 g and were homogenized with 90 mL of saline solution (0.85%) in sterile jars. Next, each bacterial culture was serially diluted (10−1–10−10) and pour-plated onto MRS agar and incubated at 37 ◦C for 48 h under anaerobic conditions using Gas-Pack-System (BD, Ontario, Canada), as recommended by [42]. The determination of LAB was performed in triplicate for each sample, and the identification was performed using the API® (BioMérieux, Marcy-l'Étoile, France) biochemical system.

#### *4.3. Histamine and Tyramine Analysis*

The determination and quantification of the biogenic amines (histamine and tyramine) were carried out in three stages of the cheeses' shelf life: start (during the first three days after the acquisition); half (in the middle of its shelf life, considering the product label statement); and end (at the end of the shelf life, regarding the expiration date declared on the label). For the amine extraction from cheese samples, the methodology described by Elsanhoty, Mahrous, and Ghanaimy (2009) [43] was followed. Briefly, samples (10 g of cheese) were mixed with 10 mL of 10% trichloroacetic acid solution and homogenized for 15 min (Ultra-turrax homogenizer®, Daigger, IL, USA); the products obtained were centrifuged 10 min at 3000 rpm at 4 ◦C (Eppendorf, model 5804 R, Hamburg, Germany). The supernatant was filtered (Whatman Paper No. 1), transferred to 15 mL polypropylene tubes (Corning, NY, USA), and kept at −20 ◦C until use.

The quantification of histamine and tyramine was performed following the official method 977.13 of AOAC (Association of Official Analytical Chemists). This method is sensitive for the identification and quantification of histamine in seafood and consists of the extraction of the compound, the formation of a derivative with *o*-phthaldialdehyde (OPA), and the detection of the fluorescence developed [44]. For the formation of the fluorescent derivatives, 300 μL of cheese sample diluted in 120 μL of 0.4 M borate buffer solution were added 5 min before to the HPLC analysis with 120 μL of o-phthaldialdehyde reagent (200 mg of OPA mixed in 9 mL of methanol, 1 mL of 0.4 M sodium borate, pH 10, and 160 μL of 2-mercaptoethanol) (Sigma–Aldrich, Saint Louis, MO, USA).

The fluorescent derivatives were separated by HPLC (high-performance liquid chromatography) for amine quantification (Thermo Scientific, Spectra System, Waltham, MA,

USA). The HPLC System was equipped with a fluorescence detector fixed to 338 nm and 430 nm (absorption and emission wavelength, respectively). A Waters Ultrasphere ODS 5 μm (4.6 × 250 mm) C18 column was set, and a circumvention gradient formed by 12.5 mM of phosphate buffer, pH 6.5 (Eluent A), and acetonitrile (Eluent B) was used. The gradient started with an 85:15% ratio of Solvents A and B, respectively, and a flow of 0.9 mL/min during 3 min; subsequently, Solvent B was increased to 60% with a final flow rate of 1.3 mL/min (3–24 min) and then returned to 15% of Solvent B (24–50 min) and remained isocratic for five more minutes.

Histamine dihydrochloride and tyramine hydrochloride (Sigma–Aldrich, Saint Louis, MO, USA) were used as standards for the identification and quantification of these compounds in the cheese samples. Standard solutions were prepared at different concentrations (25, 50, 100, 250, and 400 mg/L), and a standard calibration curve was constructed. For the HPLC analysis, 10 μL of standards and samples were injected.

#### *4.4. Molecular Biology Techniques Standardization*

Standard techniques were used for the molecular analysis (DNA extraction and PCR analysis). To establish the molecular protocols that would allow the identification of bacteria-producing biogenic amines, the reference strains ATCC 33222 of *Lactobacillus* 30a and ATCC 367 of *Lactobacillus brevis*, were used as controls. For preservation and experimental conditions, both strains were grown in MRS broth for 48 h and were kept stirring at 200 rpm. The *Lactobacillus* 30a culture was performed at 37 ◦C under anaerobic conditions, while the *L. brevis* culture was performed under aerobic conditions at 30 ◦C.

#### 4.4.1. DNA Extraction

DNA extraction from reference strain cultures was carried out following the technique described by Hoffman and Winston, (1999) [45], in which the cell wall is broken both by mechanical fractionation with glass beads and by chemical lysis using a buffer solution (TSNT buffer: 2% Triton, 1% SDS, 100 mM NaCl, 10 mM Tris-HCl, pH 8.0) followed by a DNA purification step with a phenol–chloroform mixture (50:50) and subsequent precipitation of DNA with 3 mM of sodium acetate and ethanol (96%).

For the direct extraction of bacterial DNA present in cheeses, an adaptation of the technique described by Randazzo et al., (2002) [46] was used. DNA purification was performed using the SV Total RNA Isolation System kit (Promega, Madison, WI, USA). Briefly, cheese samples (5 g) were placed in sterile dilution jars (130 mL with screw cap) containing 40 mL of a sterile solution of sodium citrate (2%) and were incubated at 45 ◦C for 30 min. Subsequently, 200 μL of sterile glass beads (0.1 mm diameter) were added, and the mixture was stirred manually for 5 min and left to stand for 10 min at room temperature. Next, 1 mL of the supernatant was transferred to a sterile 1.5 mL polypropylene tube and centrifuged for 3 min at 12,000 rpm (Eppendorf mini spin centrifuge; Eppendorf; Hamburg, Germany); the supernatant was eliminated. This process was repeated until the pellet product of all volumes (10 mL) of the homogenized cheese was obtained. From this product, the DNA was obtained using the SV Total RNA Isolation System kit, following the instructions from the manufacturer; the obtained DNA was diluted with sterile water (50 μL) and stored at −20 ◦C until use.

#### 4.4.2. Amplification of Genes Encoding hdc and tdc

The presence of the genes encoding decarboxylase enzymes (histidine and tyrosine) on DNA obtained from cheese was determined by PCR amplification. PCR results obtained from DNA amplification from different cheese samples were considered positive when PCR reactions were positive in the three representative cheese samples. For the identification of hdc and tdc genes in cheese samples, the conditions described by Fernandez et al., (2006) [27] were used. For the hdc gene amplification, the specific oligonucleotides HDC1 and HDC described by Ventura et al., (2001) [26] were used. For the tdc gene identification, a pair of degenerate oligonucleotides (TER-F and TER-R) was designed on the sequences from the tdc gene corresponding to different bacterial species.

#### 4.4.3. Design of Degenerated Oligonucleotides for the tdc Gene

Nucleotide sequences corresponding to genes that code for the enzyme tyrosine decarboxylase from different species of bacteria (*Enterococcus hirae*, *E. durans*, *E. faecium*, *Lactobacillus curvatus*, *L. brevis and Streptococcus* sp.) were obtained from databases. Alignments of these sequences (which allowed identifying the conserved regions) were made (both the amino acids and nucleotides levels) using the BLAST (http://blast.ncbi.nlm.nih.gov/Blast. cgi, accessed on 10 July 2020) and CLUSTAL-W (http://www.genome.jp/tools/clustalw/, accessed on 10 July 2020) programs. The regions that presented higher conservation were identified and used to design the degenerate oligonucleotides TER-F (forward primer) and TER-R (reverse primer).

The standardization of the experimental conditions of PCR amplification and determination of the specificity of both the specific oligonucleotides for hdc and the degenerate ones for the tdc genes was carried out using DNA obtained from the reference strains ATCC 33222 of *Lactobacillus* 30a and ATCC 367 of *L. brevis*, respectively. PCR reactions were performed in a Thermocycler (Axygen, Therm-1000, Maxygene) using a temperature gradient (50–60 ◦C) as an alignment/hybridization temperature. Final PCR conditions were 95 ◦C for 5 min (a cycle), followed by 95 ◦C for 45 s, 50–60 ◦C for 45 s, and 72 ◦C for 2 min (35 cycles), with a final extension cycle (72 ◦C for 5 min). The products obtained from PCR reactions were analyzed by electrophoresis in 1.5% agarose gels, stained with ethidium bromide [47], and were visualized under UV light with a Gel-Doc Photo Documentation System (Vision Works, UVP, Upland, CA, USA).

#### *4.5. Statistical Analysis*

Statistical analyses were performed using the statistical software package IBM SPSS Statistics V21.0. Differences between treatments were analyzed using the analysis of variance (ANOVA) test. The statistical significance was determined at *p* < 0.05. Data corresponding to each variable are expressed as means ± SD, unless otherwise indicated. All determinations were performed by triplicate.

#### **5. Conclusions**

In this study, the presence of biogenic amines, histamine and tyramine, was detected in 37% and 75%, respectively, at the end of the shelf life of the Chihuahua cheese tested. It was observed that at this point, the number of samples positive for the presence of histamine and tyramine increased, suggesting that even at cooling temperature, the production of these compounds continues. The presence of genes encoding for hdc and tdc was detected by molecular techniques using degenerated primers. The detected tyramine levels (<600 mg/kg) do not present a risk to healthy consumers (without MAO inhibitors prescription); in the other hand, the histamine content at the end of the shelf of cheeses (>50 mg/kg) represent a health risk, according to EFSA and FDA regulations. The presence of biogenic amines in foods, such as Chihuahua cheese, could be controlled by the strict use of good manufacturing hygiene practices and raw material selection.

In Mexico, there is no regulation about the maximum concentration of biogenic amines (histamine or tyramine) in cheeses. Although there are standardized protocols for amine detection, these processes are time-consuming and labor-intensive. In this work, we presented evidence of the usefulness of degenerated primers to be employed as tools for amplification by PCR techniques to achieve a direct detection of bacterial species that produce biogenic amines.

Moreover, we suggest the appliance of molecular techniques as a detection tool for bacteria that produce biogenic amines in food as it can prevent the early development of these bacteria and, therefore, biogenic amines in food. In addition, maximum regulatory limits on biogenic amines must be declared to protect consumers and avoid possible public health problems.

**Author Contributions:** Conceptualization, B.E.G.-M. and E.C.-G.; investigation and data curation M.T.G.-M., A.A.L.-H., A.S.O.-V. and G.I.A.-M.; methodology, M.T.G.-M., A.A.L.-H., A.S.O.-V. and G.I.A.-M.; validation, formal analysis, and resources, M.T.G.-M., A.A.L.-H., A.S.O.-V. and G.I.A.-M.; writing—original draft preparation, A.A.L.-H., A.S.O.-V., G.I.A.-M. and E.C.-G.; writing—review and editing, E.C.-G., A.A.L.-H., A.S.O.-V. and G.I.A.-M.; supervision, B.E.G.-M. and E.C.-G.; project administration and funding acquisition, B.E.G.-M. and E.C.-G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported in part by the PROMEP-SEP Program with a fellowship to M.T.G.-M. and by the UANL-PAICYT Program (No. CE 1025-11 and CN 354-14). The APC was funded by UANL-Facultad de Salud Pública y Nutrición.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
