**The Role of** *Enterococcus faecium* **as a Key Producer and Fermentation Condition as an Influencing Factor in Tyramine Accumulation in** *Cheonggukjang*

**Young Kyoung Park 1, Young Hun Jin 1, Jun-Hee Lee 1, Bo Young Byun 1, Junsu Lee 1, KwangCheol Casey Jeong 2,3 and Jae-Hyung Mah 1,\***


Received: 10 June 2020; Accepted: 8 July 2020; Published: 11 July 2020

**Abstract:** The study evaluated the role of *Enterococcus faecium* in tyramine production and its response to fermentation temperature in a traditional Korean fermented soybean paste, *Cheonggukjang*. Tyramine content was detected in retail *Cheonggukjang* products at high concentrations exceeding the recommended limit up to a factor of 14. All retail *Cheonggukjang* products contained *Enterococcus* spp. at concentrations of at least 6 Log CFU/g. Upon isolation of *Enterococcus* strains, approximately 93% (157 strains) produced tyramine at over 100 μg/mL. The strains that produced the highest concentrations of tyramine (301.14–315.29 μg/mL) were identified as *E*. *faecium* through 16S rRNA sequencing. The results indicate that *E. faecium* is one of the major contributing factors to high tyramine content in *Cheonggukjang*. During fermentation, tyramine content in *Cheonggukjang* groups co-inoculated with *E. faecium* strains was highest at 45 ◦C, followed by 37 ◦C and 25 ◦C. The tyramine content of most *Cheonggukjang* groups continually increased as fermentation progressed, except groups fermented at 25 ◦C. At 45 ◦C, the tyramine content occasionally exceeded the recommended limit within 3 days of fermentation. The results suggest that lowering fermentation temperature and shortening duration may reduce the tyramine content of *Cheonggukjang*, thereby reducing the safety risks that may arise when consuming food with high tyramine concentrations.

**Keywords:** *Cheonggukjang*; *Enterococcus faecium*; tyramine; biogenic amines; fermentation temperature; fermentation duration; tyrosine decarboxylase gene (*tdc*)

#### **1. Introduction**

*Cheonggukjang* is a traditional Korean soybean paste produced by fermenting soybeans with *Bacillus subtilis*. Traditional methods of *Cheonggukjang* production utilize rice straw added to steamed soybeans for a short fermentation period of approximately 2–3 days, while starter cultures are used instead of rice straw for modern methods of production [1,2]. Fermentation of *Cheonggukjang* is a process involving microbial enzymatic proteolysis resulting in uniquely characteristic savory aromatic and flavor properties [3]. Consumption of *Cheonggukjang* has been reported to be associated with numerous benefits such as antioxidative, antihypertensive, thrombolytic, and antimicrobial properties [4,5]. However, despite the beneficial properties of *Cheonggukjang*, potentially hazardous biogenic amines (BAs) may be produced during fermentation of the proteinous food rich in precursor amino acids.

The majority of BAs are formed through the reductive amination of ketones and aldehydes, as well as the decarboxylation of amino acids by microbially produced enzymes [6]. Though BAs are

essential for the regulation of protein synthesis, nucleic acid functions, and membrane stabilization in living cells, consumption of food products with high concentrations of BAs may result in toxicological effects [7–10]. The excessive intake of food products such as mackerel, pacific saury, sardines, and tuna may result in "scombroid poisoning" owing to potentially high concentrations of toxic histamine that may cause symptoms similar to an allergic reaction including diarrhea, dyspnea, headache, hives, and hypotension [10–13]. Overconsumption of foods with high concentrations of tyramine may potentially result in a "cheese crisis" with various symptoms including heart failure, hemorrhages, hypertensive crisis, high blood pressure, and severe headaches [9,10,14,15]. Such a high content of tyramine produced by microbial tyrosine decarboxylase activity has occasionally been found in tyrosine-rich foods such as cheese [16,17] and soybean-based fermented products [18–20]. Therefore, Ten Brink, et al. [21] suggested BA toxicity limits of 30 mg/kg for β-phenylethylamine, 100 mg/kg for histamine, and 100–800 mg/kg for tyramine in foods.

Previous studies by Ko, et al. [18], Jeon, et al. [19], and Seo, et al. [20] on the BA content of *Cheonggukjang* have shown that tyramine in particular has been detected in high concentrations up to 1913.51, 251.66, and 905.0 mg/kg, respectively. Ibe, et al. [22] suggested that *Enterococcus faecium* may be largely responsible for the BA content of *Miso* (a Japanese fermented soybean paste). Notably, numerous studies have reported that *Enterococcus* spp. possess the tyrosine decarboxylase gene (*tdc*) [23,24]. Moreover, in particular Kang and Park [25] and Kang, et al. [26] confirmed the presence of *E. faecium* in *Cheonggukjang*, while a previous study by Jeon, et al. [19] showed that *Enterococcus* spp. isolated from *Cheonggukjang* exhibited tyramine production at concentrations of at least 351.59 μg/mL. Taken together, the previous reports imply that *E. faecium* may also be responsible for the BA content of *Cheonggukjang*. Meanwhile, the growth of *Enterococcus* spp. has been reported to occur at temperatures ranging from 10 ◦C up to 45 ◦C that overlap with *Cheonggukjang* fermentation temperatures ranging from 25 to 50 ◦C [27–30]. The corresponding range in temperature may be beneficial for *E. faecium* growth and tyramine production during the fermentation of contaminated *Cheonggukjang* products. Furthermore, a previous study reported that tyramine content increases in fermented soybeans as fermentation progresses [19]. According to Bhardwaj, et al. [31], the production of tyramine by *E. faecium* strains may be affected by incubation conditions such as temperature and time. Therefore, the current study assessed the safety risk of BAs (particularly tyramine) in *Cheonggukjang*, clarified the microorganism responsible for tyramine accumulation, and evaluated the effect of fermentation temperature/duration on *E. faecium* growth and subsequent tyramine production in the food.

#### **2. Materials and Methods**

#### *2.1. Cheonggukjang Products*

Six representative, but different *Cheonggukjang* products were purchased from various retail markets in the Republic of Korea and stored at 4 ◦C until further experimentation. Within a day of storage, the BA content of *Cheonggukjang* products was measured, followed by physicochemical and microbial analyses.

#### *2.2. Physicochemical Analyses*

To investigate the influencing factors such as pH, salinity, and water activity on BA content in *Cheonggukjang*, the physicochemical properties of *Cheonggukjang* samples (retail *Cheonggukjang* products purchased and *Cheonggukjang* groups fermented in this study) were measured as described below. Samples weighing 10 g using an analytical balance (Ohaus Adventurer™, Ohaus Corporation, Parsippany, NJ, USA) were homogenized with 90 mL of distilled water using a stomacher (Laboratory Blender Stomacher 400, Seward, Ltd., Worthing, UK). The pH of the homogenates was measured using a pH meter (Orion 3-star pH Benchtop Thermo Scientific, Waltham, MA, USA), while salinity was measured using the procedure described by the Association of Official Analytical Chemists

(AOAC; Official Method 960.29) [32]. The water activity of the samples was measured using an electric hygrometer (AquaLab Pre, Meter Group, Inc., Pullman, WA, USA).

#### *2.3. Microbial Analyses*

The analysis of the microbial community in *Cheonggukjang* samples was conducted using Plate Count Agar (PCA; Difco, Becton Dickinson, Sparks, MD, USA); de Man, Rogosa, and Sharpe (MRS; Conda, Madrid, Spain) agar; and m-Enterococcus Agar (m-EA; MB Cell, Seoul, Korea) for the enumeration of total mesophilic viable bacteria, lactic acid bacteria, and *Enterococcus* spp., respectively. Samples weighing 10 g were homogenized with 90 mL of sterile 0.1% peptone saline using a stomacher. The homogenates were 10-fold serially diluted with 0.1% peptone saline up to 10−5, and 100 μL of each dilution was spread on PCA, MRS agar, and m-EA in duplicates. Incubation conditions were set according to the manufacturer's instructions: PCA at 37 ◦C for 24 h and m-EA at 37 ◦C for 48 h under aerobic condition; MRS agar at 37 ◦C for 48 h under anaerobic condition. Anaerobic condition was achieved using an anaerobic chamber (Coy Lab. Products, Inc., Grass Lake, MI, USA) containing an atmosphere of 95% nitrogen and 5% hydrogen. After incubation, the bacterial concentrations of the *Cheonggukjang* samples were calculated by enumerating the colony-forming units (CFU) on the plates of respective media with approximately 10 to 300 colonies [33] and adjusting for the dilution.

#### *2.4. Isolation and Identification of Enterococcus Strains from Retail Cheonggukjang Products*

A total of 169 *Enterococcus* strains were isolated from retail *Cheonggukjang* products according to the method described by Mareková, et al. [34], with minor modifications. Upon enumeration of colonies on m-EA, individual colonies were streaked on MRS agar and incubated at 37 ◦C for 48 h under anaerobic condition. Single colonies were streaked again on MRS agar and incubated under the same conditions. The pure single colonies were inoculated in MRS broth, incubated at 37 ◦C for 48 h, and stored at −70 ◦C using glycerol (20%, *v*/*v*).

The identities (at species level) of the individual *Enterococcus* strains that displayed the highest tyramine production were further investigated through sequence analysis of 16S rRNA gene amplified with the universal bacterial primer pair 518F (5 -CCAGCAGCCGCGGTAATACG-3 ) and 805R (5 -GACTACCAGGGTATCTAAT-3 ) (Solgent Co., Daejeon, Korea). The identities of sequences were determined using the basic local alignment search tool (BLAST) of the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov/BLAST/).

#### *2.5. Preparation of Cheonggukjang*

To investigate the effect of fermentation temperature on tyramine production by *E. faecium*, several temperatures were set for in situ *Cheonggukjang* fermentation experiments. The temperature for *Cheonggukjang* fermentation (intermediate-temperature group) was determined based upon previous studies in which 37 ◦C was reported as the temperature commonly used for *Cheonggukjang* production [19,35,36]. In addition, the temperatures of 25 ◦C and 45 ◦C used by other studies for *Cheonggukjang* fermentation were utilized for the low and high temperature groups, respectively [29,30].

White soybeans (*Glycine* max Merrill) were purchased from a retail market in the Republic of Korea. The soybeans were soaked in distilled water at 4 ◦C for 12 h, and subsequently drained for 1 h. Approximately 200 g of soybeans were adjusted to a final salinity of 2.40% according to the salinity of *Cheonggukjang* outlined in the 9th revision of the Korean food composition table [37] and subsequently steamed at 125 ◦C for 30 min using an autoclave. The steamed soybeans were cooled to 50 ◦C and inoculated with bacterial inocula in M/15 Sörensen's phosphate buffer (pH 7) to final concentrations of approximately 6 Log CFU/g of *B. subtilis* KCTC 3135 (also designated as ATCC 6051; type strain) and 4 Log CFU/g of *E. faecium* KCCM 12118 (ATCC 19434; type strain) or *E. faecium* CJE 216 (strain isolated from *Cheonggukjang* and selected owing to both strong tyramine production and *tdc* gene expression). The control group (without any *E. faecium* strains) was inoculated with only *B. subtilis* KCTC 3135 to a final concentration of 6 Log CFU/g. The sizes of inocula were selected with

consideration of the cell count of each microorganism in *Cheonggukjang* products determined in our previous study [19]. The inoculated steamed soybeans were then fermented at 25 ◦C, 37 ◦C, or 45 ◦C for 3 days. Approximately 20 g of the fermented soybeans were collected daily during fermentation to measure the BA content as well as physicochemical and microbial properties. Fermented soybeans sampled during fermentation were stored at −70 ◦C for further testing, as required.

#### *2.6. BA Analyses in Cheonggukjang Samples and Bacterial Cultures*

#### 2.6.1. BA Extraction from *Cheonggukjang* Samples and Bacterial Cultures

Quantification of the BA content of *Cheonggukjang* was conducted as previously described by Ben-Gigirey, et al. [38]. Five grams of *Cheonggukjang* with 20 mL of 0.4 M perchloric acid (Sigma-Aldrich, St. Louis, MO, USA) were homogenized by vortex (Vortex-Genie, Scientific industries, Bohemia, NY, USA) and stored at 4 ◦C for 2 h. The mixture was then centrifuged at 3000× *g* for 10 min at 4 ◦C (1736R, Labogene, Seoul, Korea), and the supernatant was collected. Upon resuspension of the pellet with 20 mL of 0.4 M perchloric acid, the mixture was stored at 4 ◦C for 2 h and centrifuged again at 3000× *g* at 4 ◦C for 10 min. The supernatant was combined with the previously collected supernatant and adjusted to a final volume of 50 mL with 0.4 M perchloric acid. Then, the extract was filtered through Whatman paper No. 1 (Whatman International Ltd., Maidstone, UK).

The bacterial production of BAs was measured using the procedures described by Eerola, et al. [39], modified by Ben-Gigirey, et al. [38,40], and further modified in the present study to culture *Enterococcus* spp. based on Marcobal, et al. [41]. A loopful (10 μL) of glycerol stock of each enterococcal strain was inoculated in 5 mL of MRS broth supplemented with 0.5% of each amino acid, including L-histidine monohydrochloride monohydrate, L-tyrosine disodium salt hydrate, L-ornithine monohydrochloride, L-lysine monohydrochloride (pH 5.8), and 0.0005% pyridoxal-HCl (all from Sigma-Aldrich) and incubated at 37 ◦C for 48 h. Approximately 100 μL of the broth culture was then transferred to another tube containing 5 mL of the same medium. Upon incubation at 37 ◦C for 48 h, the broth culture was filtered using a sterile syringe with a 0.2 μm membrane (Millipore Co., Bedford, MA, USA). Then, 9 mL of 0.4 M perchloric acid were added to 1 mL of the filtered broth culture and mixed by a vortex mixer. The mixture was reacted in a cold chamber at 4 ◦C for 2 h and centrifuged at 3000× *g* at 4 ◦C for 10 min. The extract was filtered through Whatman paper No. 1.

#### 2.6.2. Preparation of Standard Solutions for High Performance Liquid Chromatography (HPLC) Analysis

Standard solutions with concentrations of 0, 10, 50, 100, and 1000 ppm were prepared for tryptamine, β-phenylethylamine hydrochloride, putrescine dihydrochloride, cadaverine dihydrochloride, histamine dihydrochloride, tyramine hydrochloride, spermidine trihydrochloride, and spermine tetrahydrochloride (all from Sigma-Aldrich). Internal standard solution with the same concentrations was prepared using 1,7-diaminoheptane (Sigma-Aldrich).

#### 2.6.3. Derivatization of Extracts and Standards

Derivatization of BAs was conducted according to the method described by Eerola, et al. [39]. One milliliter of extract or standard solution prepared as aforementioned was mixed with 200 μL of 2 M sodium hydroxide and 300 μL of saturated sodium bicarbonate (all from Sigma-Aldrich). Two milliliters of dansyl chloride (Sigma-Aldrich) solution (10 mg/mL) in acetone were added to the mixture and incubated at 40 ◦C for 45 min. The residual dansyl chloride was removed by adding 100 μL of 25% ammonium hydroxide and incubating for 30 min at 25 ◦C. Using acetonitrile, the mixture was adjusted to a final volume of 5 mL and centrifuged at 3000× *g* for 5 min. After filtration using 0.2 μm pore-size filters (Millipore), the filtered supernatant was kept at 4 ◦C until further analysis using HPLC.

#### 2.6.4. Chromatographic Separations

Chromatographic separation of BAs was conducted according to the method previously developed by Eerola, et al. [39] and modified by Ben-Gigirey, et al. [40]. An HPLC unit (YL9100, YL Instruments Co., Ltd., Anyang, Korea) equipped with a UV/vis detector (YL Instruments) and Autochro-3000 data system (YL Instruments) was used. For chromatographic separation, a Nova-Pak C18 4 μm column (150 mm × 4.6 mm, Waters, Milford, MA, USA) held at 40 ◦C was utilized. The mobile phases were 0.1 M ammonium acetate dissolved in deionized water (solvent A; Sigma-Aldrich) and acetonitrile (solvent B; SK chemicals, Ulsan, Korea) adjusted to a flow rate of 1 mL/min with a linear gradient starting from 50% of solvent B reaching 90% by 19 min. A 10 μL sample was injected and monitored at 254 nm. The limits of detection were approximately 0.1 μg/mL for all BAs in standard solutions and bacterial cultures, and about 0.1 mg/kg for all BAs in food matrices [42].

#### *2.7. Gene Expression Analyses in Bacterial Cultures and Cheonggukjang*

#### 2.7.1. RNA Extraction and Reverse Transcription

Expression analysis of tyrosine decarboxylase gene (*tdc*) involved RNA extraction from bacterial cultures (for in vitro experiments) and *Cheonggukjang* samples (viz., *Cheonggukjang* groups prepared through fermentation; for in situ fermentation experiments) with a Ribo-Ex Total RNA isolation solution (Geneall, Seoul, Korea). The extraction was conducted according to the manufacturer's instructions with minor modifications as follows. To prepare bacterial culture for in vitro gene expression analysis, a loopful (10 μL) of glycerol stock of each enterococcal strain was inoculated in 5 mL of MRS broth supplemented with 0.5% L-histidine monohydrochloride monohydrate, L-tyrosine disodium salt hydrate, L-ornithine monohydrochloride, L-lysine monohydrochloride (pH 5.8), and 0.0005% pyridoxal-HCl (all from Sigma-Aldrich) and incubated at 37 ◦C for 48 h. Approximately 100 μL of the broth culture was then transferred to another tube containing 5 mL of the same medium and incubated under the same conditions. As for *Cheonggukjang* samples, 10 g of *Cheonggukjang* were gently mixed with 40 mL of phosphate buffer in a sterile bag, and the liquid part was collected. Subsequently, 3 mL of the bacterial culture or all liquid part of the mixture were immediately transferred into a 50 mL conical tube and centrifuged at 10,000× *g* at 4 ◦C for 5 min. After removing the supernatant, the pellet was suspended with 7 mL of phosphate buffer and centrifuged under the same conditions. Then, the pellet was homogenized with 800 μL of Ribo-Ex reagent in a bacterial lysing tube (Lysing Matrix B; MP Biomedicals, Santa Ana, CA, USA) using a Precellys 24 homogenizer (Bertin Technologies, Montigny, France) with two cycles for 30 s at 6800 rpm, pausing for 90 s between cycles. Approximately 200 μL of chloroform were added to the lysate, vortexed, and centrifuged at 10,000× *g* for 1 min. Approximately 400 μL of the supernatant were mixed with 600 μL of chilled absolute ethanol. The mixture was reacted at −70 ◦C for 15 min and purified with a Nucleospin RNA kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's instructions. The quality of the extracted RNA was evaluated using a NanoDrop 1000 spectrophotometer (Thermo Fisher, Waltham, MA, USA).

ReverTra Ace qPCR RT Master Mix with gDNA Remover kit (Toyobo, Osaka, Japan) containing reverse transcriptase, RNase inhibitor, oligo (dT) primers, random primers, and deoxynucleoside triphosphates (dNTPs) was used to synthesize cDNA from 1 μL of extracted RNA according to the manufacturer's instructions. Reverse transcription was conducted under the following conditions: 37 ◦C for 15 min, 50 ◦C for 5 min, and 98 ◦C for 5 min. After the reaction, the resulting cDNA was stored at −70 ◦C until quantitative PCR analysis.

#### 2.7.2. Quantitative PCR Analysis

As designed by Kang, et al. [43], *q-tdc* F (5 -AGACCAAGTAATTCCAGTGCC-3 ) and *q-tdc* R (5 -CACCGACTACACCTAAGATTGG-3 ) primers were used for the quantitation of *tdc* gene expression by *E. faecium*. The primers for reference genes including *q-gap* F (5 -ATACGACACAACTCAAGGACG-3 ) and *q-gap* R (5 -GATATCTACGCCTAGTTCGCC-3 ) [34], along with *tufA*-RT F (5 -TACACGCCACTAC

GCTCAC-3 ) and *tufA*-RT R (5 -AGCTCCGTCCATTTGAGCAG-3 ) [44] were used for the normalization of *tdc* gene expression. The efficiency of each set of primers for reverse transcription quantitative polymerase chain reaction (RT-qPCR) was determined by the following equation: *<sup>E</sup>* <sup>=</sup> <sup>10</sup> (−1/*S*) <sup>−</sup> 1, where *E* is the amplification efficiency and *S* is the slope of standard curves generated through threshold cycle (Ct) values of serial dilutions of cDNA obtained from reverse-transcription of RNA from *E. faecium* KCCM 12118.

For the RT-qPCR analysis of *tdc* gene expression in bacterial cultures and *Cheonggukjang* samples, 5 μL of a 10-fold diluted cDNA were added to 15 μL of a master mix containing 10 μL of Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA), 3 μL of RNase free water, and 1 μL of each primer (forward and reverse; 500 nM). Subsequently, thermal cycling was conducted using an Applied Biosystems 7500 Real-Time PCR system (Applied Biosystems) with the thermal cycling conditions programmed as follows: initial denaturation at 95 ◦C for 10 min; 40 cycles at 95 ◦C for 15 s (denaturation step), and 60 ◦C for 60 s (annealing and elongation steps, unless otherwise mentioned). Annealing and elongation conditions for primer *tufA*-RT were set at 55 ◦C for 60 s. Melting curve analysis was conducted using the RT-PCR system to confirm the specificity and to analyze the amplified products. Ct values were detected when the emissions from fluorescence exceeded the fixed threshold automatically determined by thermocycler software. Relative expression of *tdc* genes was further calculated by the 2−(ΔΔct) method, normalized to the expression levels detected in *E. faecium* KCCM 12118 (refer to Figure 2) or *Cheonggukjang* groups fermented at 37 ◦C (refer to Figure 6), and expressed as n-fold differences to compare gene expression levels in different bacterial cultures and *Cheonggukjang* samples.

#### *2.8. Statistical Analyses*

Data were presented as means and standard deviations of duplicates or triplicates. All measurements on retail products were performed in triplicates, while the other experiments were conducted in duplicate. The significance of differences was determined by one-way analysis of variance (ANOVA) with Fisher's pairwise comparison module of the Minitab statistical software, version 17 (Minitab Inc., State College, PA, USA), and differences with probability (*p*) value of <0.05 were considered statistically significant.

#### **3. Results and Discussion**

#### *3.1. Physicochemical Properties of Retail Cheonggukjang Products*

Physicochemical and microbial properties as well as BA content in retail *Cheonggukjang* products were analyzed to estimate the contributing factors to BA content (particularly tyramine) in*Cheonggukjang* (Sections 3.1–3.3). Table 1 displays the physicochemical properties of *Cheonggukjang* products purchased from retail markets in the Republic of Korea. The pH ranged from 6.39 to 7.05, with an average pH of 6.84 ± 0.23 (mean ± standard deviation). The results were similar to the study conducted by Lee, et al. [45], which reported the average pH of *Cheonggukjang* to be 7.0 ± 0.8. Jeon, et al. [19] and Yoo, et al. [46] also reported the average pH of *Cheonggukjang* to be pH 6.07 ± 0.72 (range of pH 4.62–8.14) and pH 7.21 ± 0.59 (range of pH 5.89–7.95), respectively. Such differences in the pH of the *Cheonggukjang* products may be owing to different fermentation conditions [47] and/or fermentation metabolites [48]. The salinity of retail *Cheonggukjang* products ranged from 1.95 to 9.36% with an average salinity of 5.16 ± 2.78%. In comparison, Ko, et al. [18], Jeon, et al. [19], and Kang, et al. [49] reported the average salinity of *Cheonggukjang* to be 2.12 ± 1.66% (0.12–11.51%), 1.56 ± 1.19% (0.10–5.33%), and 3.51 ± 2.45 (1.64–8.39%), respectively. Though the salinity of the *Cheonggukjang* products was found to vary substantially, Ko, et al. [18] suggested that the large differences in *Cheonggukjang* salinities may be traced to the production process, as some methods utilize the addition of different amounts of salt to preserve the fermented soybean product. The water activity of retail *Cheonggukjang* products ranged from 0.919 to 0.973 with an average of 0.951 ± 0.019. In a previous study by Kim, et al. [47], the average

water activity was found to be 0.962 ± 0.028 (0.857–0.991). Overall, the physicochemical properties of retail *Cheonggukjang* products analyzed in the current study were mostly similar to the values reported in previous studies. Although the results of the current study did not show any correlation between physicochemical properties and BA content (especially tyramine) based on linear regression analyses (data not shown), it is noteworthy that the ranges of the physicochemical parameters were within the specific conditions for the growth of *E. faecium*, which are as follows: pH, from 4 to 10 [50]; salinity, up to 7% [50]; water activity, above 0.940 [51].


**Table 1.** Physicochemical properties of retail *Cheonggukjang* products.

<sup>1</sup> CJ: *Cheonggukjang*; <sup>2</sup> Mean ± standard deviation were calculated from triplicate experiments.

#### *3.2. Microbial Properties of Retail Cheonggukjang Products*

Table 2 shows the microbial properties of retail *Cheonggukjang* products. The number of total mesophilic viable bacteria ranged from 8.54 to 9.81 Log CFU/g, with an average of 9.27 ± 0.45 Log CFU/g. Comparatively, Ko, et al. [18] and Jeon, et al. [19] reported the total counts of viable mesophilic bacteria of *Cheonggukjang* products to be 7.50 ± 1.01 Log CFU/g (5.30–9.98 Log CFU/g) and 9.65 ± 0.77 Log CFU/g (8.23–11.66 Log CFU/g), respectively. The wide range of total mesophilic viable bacteria may result from an insufficient standardization of *Cheonggukjang* manufacturing processes such as different fermentation materials and conditions [18,47]. *Enterococcus* spp. were detected at concentrations of 6.64–7.99 Log CFU/g, with an average of 7.17 ± 0.49 Log CFU/g. The number of lactic acid bacteria was found to be approximately 6.66–8.12 Log CFU/g, with an average of 7.09 ± 0.58 Log CFU/g (Table 2). For comparison, a study by Kang and Park [25] showed that *Enterococcus* spp. were detected in all 31 *Cheonggukjang* products at concentrations of 3.51–8.46 Log CFU/g, with an average of 5.95 ± 1.60 Log CFU/g. In the report, approximately 58% and 16.8% of the isolated *Enterococcus* strains were identified as *E. faecium* and *E. faecalis*, respectively. The presence of *E. faecium* in *Cheonggukjang* was also reported by Kang, et al. [26]. The reported results on the presence of *Enterococcus* spp. at high concentrations in *Cheonggukjang* concurred with the findings of the current study. As *E. faecium* has been reported as a pathogenic and/or tyramine-producing bacterium detected in some foods including Chinese and Japanese fermented soybean products, previous studies have mentioned that preventative measures are necessary to avoid contamination during the manufacturing of fermented foods [52–55]. The traditional *Cheonggukjang* production process may also be susceptible to contamination by harmful microbes owing to the reliance on rice straw containing *B. subtilis* for fermentation [56]. In fact, according to Heu, et al. [57], rice straw contains a variety of bacteria, including mesophiles, thermophiles, coliforms, and actinomycetes, as well as fungi. Moreover, as sterilization processes are not utilized in the manufacturing of *Cheonggukjang*, occasional contamination by tyramine-producing bacteria such as *E. faecium* may be present in the final product. The results of the current and previous studies suggest that further research appears to be necessary for the development of methods to inhibit *E. faecium* growth during the manufacturing of *Cheonggukjang* as well as other fermented soybean products described above.


**Table 2.** Microbial properties of retail *Cheonggukjang* products.

<sup>1</sup> CJ: *Cheonggukjang*; <sup>2</sup> Mean ± standard deviation were calculated from triplicate experiments; CFU: colonyforming units.

#### *3.3. BA Content of Retail Cheonggukjang Products*

*Cheonggukjang* contains abundant BA precursor amino acids such as lysine, histidine, tyrosine, and phenylalanine [58]. The high amino acid content may pose a risk for conversion into BAs during *Cheonggukjang* fermentation. In the present study, tryptamine, β-phenylethylamine, putrescine, cadaverine, histamine, tyramine, spermidine, and spermine contents in retail *Cheonggukjang* products were detected at concentrations of 70.63 ± 44.74 mg/kg, 36.22 ± 29.55 mg/kg, 10.80 ± 5.07 mg/kg, 18.57 ± 9.08 mg/kg, 8.37 ± 2.40 mg/kg, 457.42 ± 573.15 mg/kg, 121.92 ± 19.69 mg/kg, and 187.20 ± 110.27 mg/kg, respectively (Table 3). A previous study suggested toxicity limits of 30 mg/kg for β-phenylethylamine, 100 mg/kg for histamine, and 100–800 mg/kg for tyramine in foods [21]. Therefore, the evaluation of the BA content of the *Cheonggukjang* products was continued with regard to the aforementioned BA intake limits with the exception of tyramine (at 100 mg/kg). Evaluation of β-phenylethylamine content in two *Cheonggukjang* products revealed that concentrations exceeded the recommended limit of 30 mg/kg, with one product exceeding the limit by a factor of approximately 3. Though the histamine content of all *Cheonggukjang* products was found to be below the recommended limit of 100 mg/kg, the tyramine content of three products exceeded the 100 mg/kg limit by factors of 2, 9, and 14, respectively. Other studies have reported similarly high concentrations of tyramine in *Cheonggukjang*. Ko, et al. [18] and Seo, et al. [20] reported the highest concentrations of tyramine in *Cheonggukjang* products, exceeding the recommended limit by factors of 19 and 9, respectively. Furthermore, Cho, et al. [59], Han, et al. [60], and Jeon, et al. [19] reported that *Cheonggukjang* products contained high concentrations of tyramine, which exceeded the recommended limit by up to factors of 5, 5, and 3, respectively. Altogether, as β-phenylethylamine and tyramine content of several *Cheonggukjang* products exceeded the recommended limits, overconsumption of such fermented soybean products may occasionally result in adverse effects on the body. Moreover, *Cheonggukjang* was found to contain other BAs enhancing the toxicity of β-phenylethylamine and tyramine. Therefore, further research remains necessary for precautionary measures and remedial methods to reduce the BA content of *Cheonggukjang* to ensure the safety of the fermented soybean food.


**Table 3.** Biogenic amine (BA) content of retail *Cheonggukjang* products.

<sup>1</sup> CJ: *Cheonggukjang*; <sup>2</sup> TRP: tryptamine, PHE: β-phenylethylamine, PUT: putrescine, CAD: cadaverine, HIS: histamine, TYR: tyramine, SPD: spermidine, SPM: spermine; <sup>3</sup> Mean ± standard deviation were calculated from triplicate experiments. Mean values in the same column followed by different letters (A–F) are significantly different (*p* < 0.05).

#### *3.4. In Vitro BA Production by Enterococcus Strains Isolated from Retail Cheonggukjang Products*

Microbial decarboxylation of free amino acids is one of the main causing factors in the production of BAs, and various microorganisms, including *Bacillus*, *Clostridium*, Enterobacteriaceae, enterococci, *Lactobacillus*, and *Pseudomonas*, are capable of producing the decarboxylases responsible for the conversion of amino acids into BAs [21,61,62]. Considering previous studies in which *Enterococcus* spp. have been suggested to be responsible for tyramine accumulation in Chinese and Japanese fermented soybean products [52,53], the current study analyzed the BA production capabilities of 169 enterococcal strains isolated from retail *Cheonggukjang* products using an MRS broth-based assay medium. As shown in Figure 1, the production of tryptamine, β-phenylethylamine, putrescine, cadaverine, spermidine, and spermine was observed at concentrations lower than 10 μg/mL. Histamine production by 168 of the 169 strains was detected at quantities lower than 2 μg/mL; however, only one strain was capable of producing histamine at 96.06 μg/mL. Though tyramine production ranged from ND (not detected) to 315.29 μg/mL, 157 strains (about 93%) produced over 100 μg/mL. Through 16S rRNA sequencing, the seven strains (CJE 101, CJE 115, CJE 119, CJE 128, CJE 130, CJE 210, and CJE 216) that produced the highest levels of tyramine (301.14–315.29 μg/mL; refer to Figure 2) among the enterococcal strains were all identified as *E. faecium*. Novella-Rodríguez, et al. [63] suggested that the presence of Enterobacteriaceae or enterococci may result in the production of BAs in contaminated food products. Marcobal, et al. [64] demonstrated that *E. faecium* possesses a gene that codes an enzyme capable of L-tyrosine decarboxylation. According to Ibe, et al. [22], high levels of tyramine in *Miso* (a Japanese fermented soybean paste) products may partially result from tyramine production by *E. faecium*. Jeon, et al. [19] reported that *Enterococcus* spp. exhibited strong production of tyramine ranging from 0.41 μg/mL to 351.59 μg/mL in assay media. The author also found tyramine-producing *Bacillus* spp. (up to 123.08 μg/mL) and suggested that the species is one of the major tyramine producers in *Cheonggukjang* along with *Enterococcus* species based on the in situ fermentation experiment. Consequently, the present results suggest that *Enterococcus* spp. (particularly *E. faecium*) may be largely responsible for high tyramine concentrations in *Cheonggukjang*.

**Figure 1.** Biogenic amine (BA) production by *Enterococcus* strains (n = 169) isolated from retail *Cheonggukjang* products. Error bars indicate standard deviations calculated from duplicate experiments. <sup>1</sup> TRP: tryptamine, PHE: β-phenylethylamine, PUT: putrescine, CAD: cadaverine, HIS: histamine, TYR: tyramine, SPD: spermidine, SPM: spermine.

**Figure 2.** Comparison of tyramine production and *tdc* expression by *E. faecium* strains. -: tyramine production, : *tdc* gene expression. The expression levels observed in *E. faecium* strains isolated from retail *Cheonggukjang* products were normalized to that detected in *E. faecium* KCCM 12118 (type strain). The *tdc* gene expression was not detected in *E. faecium* strain CJE 210. Mean values followed by different letters are significantly different (*p* < 0.05). Error bars indicate standard deviations calculated from duplicate experiments.

#### *3.5. Selection of Tyramine-Producing E. faecium Strain for Cheonggukjang Fermentation Based on Tyrosine Decarboxylase Gene Expression In Vitro*

In the current study, the efficiency of primer sets *q-tdc* (for the quantitation of *tdc* gene expression) along with *q-gap* and *tufA*-RT (for the normalization of *tdc* gene expression) was calculated to be 100.71%, 94.84%, and 95.03%, respectively. An amplification efficiency between 90 and 110% indicates that the results of gene expression obtained using RT-qPCR are reproducible [65].

The aforementioned primer sets were used to detect *tdc* gene expression by the seven *E. faecium* strains (CJE 101, CJE 115, CJE 119, CJE 128, CJE 130, CJE 210, and CJE 216) with the highest tyramine production in vitro as described in the previous section (note that the primer sets were also used for in situ gene expression analysis). Among the strains, *E. faecium* strain CJE 216 showed the highest expression level of *tdc* gene (Figure 2). Considering the highest *tdc* gene expression as well as tyramine production in vitro, the CJE 216 strain was selected as an inoculant for fermentation experiments in the next section.

*3.6. Tyramine Production by E. faecium during Cheonggukjang Fermentation at Various Temperatures*

3.6.1. Changes in Physicochemical and Microbial Properties during *Cheonggukjang* Fermentation at Various Temperatures

As the results of the previous sections indicated that *E. faecium* was most likely one of the major contributing factors to high levels of tyramine in *Cheonggukjang*, in situ fermentation experiments were performed to empirically investigate the influence of *E. faecium* on tyramine content in *Cheonggukjang*. For the in situ fermentation experiments, three experimental groups of *Cheonggukjang* were used: control group inoculated with only *B. subtilis* KCTC 3135, and other two groups co-inoculated with the *B. subtilis* strain and each *E. faecium* strain (*E. faecium* KCCM 12118 or *E. faecium* CJE 216). Each group was further divided into three groups based on fermentation temperatures of 25 ◦C, 37 ◦C, and 45 ◦C (low-, intermediate-, and high-temperature groups, respectively). As shown in Figure 3, the changes in the physicochemical and microbial properties of *Cheonggukjang* were measured at 24-hour intervals for 3 days of fermentation. The pH of all *Cheonggukjang* groups was lowest on day 2, with progressively lower pH depending on the fermentation temperature, independent of which inoculum was used. The pH on day 2 of *Cheonggukjang* fermentation at 25 ◦C, 37 ◦C, and 45 ◦C ranged from pH 6.34 to 6.36, pH 5.90 to 6.09, and pH 5.49 to 5.88, respectively (Figure 3a). Loizzo, et al. [66] suggested that decarboxylases are produced by bacteria owing to a mechanism to neutralize acidic environments that restrict the growth of the bacteria. A previous study reported that low pH between 4.0 and 5.5 may result in the production of BAs [67]. Therefore, in this study, as the *Cheonggukjang* groups fermented at 45 ◦C (high-temperature group) resulted in a lower pH than other groups fermented at 37 ◦C and 25 ◦C (intermediate- and low-temperature groups, respectively), regardless of inoculum, the BA content was expected to be detected at the highest concentration among all *Cheonggukjang* groups. As for water activity, all *Cheonggukjang* groups remained within 0.95–0.97 during fermentation (Figure 3b).

**Figure 3.** Physicochemical and microbial properties during *Cheonggukjang* fermentation at various temperatures. (**a**) pH, (**b**) water activity, (**c**) total mesophilic viable bacterial counts. •: 25 ◦C, -: 37 ◦C, : 45 ◦C (inoculated with only *B. subtilis* KCTC 3135); •: 25 ◦C, -: 37 ◦C, : 45 ◦C (inoculated with *B. subtilis* KCTC 3135 and *E. faecium* KCCM 12118); ◦: 25 ◦C, : 37 ◦C, : 45 ◦C (inoculated with *B. subtilis* KCTC 3135 and *E. faecium* CJE 216). Error bars indicate standard deviations calculated from duplicate experiments.

The counts of total mesophilic viable bacteria, most probably attributed to *B. subtilis* inoculated, showed that microbial concentrations started from approximately 6 Log CFU/g on day 0 and remained at approximately 8–9 Log CFU/g throughout *Cheonggukjang* fermentation at all three temperatures, regardless of the presence or absence of *E. faecium* inoculum (Figure 3c). The total mesophilic viable bacteria in *Cheonggukjang* increased as fermentation temperature decreased; however, on day 1, those in the groups fermented at 25 ◦C and 45 ◦C showed growth up to 8 Log CFU/g, while those in the groups fermented at 37 ◦C exhibited the highest counts at 9 Log CFU/g. The results concurred with a previous finding that 37 ◦C is the optimal in situ growth temperature for *B. subtilis* during *Cheonggukjang* fermentation [29]. Similarly, Mann, et al. [68] reported the optimal in vitro growth temperature for *B. subtilis* strains isolated from *Cheonggukjang* to be 37 ◦C.

The enterococcal count in *Cheonggukjang* co-inoculated with *E. faecium* KCCM 12118 at approximately 4 Log CFU/g (and *B. subtilis* KCTC 3135 at 6 Log CFU/g as well) increased by 1.63 Log CFU/g, 3.52 Log CFU/g, and 4.06 Log CFU/g after 3 days of fermentation at 25 ◦C, 37 ◦C, and 45 ◦C, respectively (Figure 4). In *Cheonggukjang* co-inoculated with *E. faecium* CJE 216 at 4 Log CFU/g (and *B. subtilis* KCTC 3135), enterococcal count increased at all fermentation temperatures of 25 ◦C, 37 ◦C, and 45 ◦C by 3.48 Log CFU/g, 4.78 Log CFU/g, and 4.80 Log CFU/g, respectively, by day 3. *Enterococcus* spp. were not detected in the control group for the duration of the fermentation period. The results displayed progressively higher enterococcal counts that increased alongside rising fermentation temperatures with the highest enterococcal counts in *Cheonggukjang* fermented at 45 ◦C (high-temperature group). The findings were comparable to a previous study by Morandi, et al. [69], which showed that lower fermentation temperatures weakened *E. faecium* growth as the reported generation time at 25 ◦C was nearly two times longer than at 37 ◦C. *E. faecium* has been reported to display active growth within the temperature range of 37–53 ◦C, with an optimal growth temperature of 42.7 ◦C [50,70]. The current and previous studies, therefore, indicate that the use of high fermentation temperatures such as 45 ◦C may enhance *E. faecium* growth, thereby increasing the potential for high tyramine production during *Cheonggukjang* fermentation.

**Figure 4.** Effect of fermentation temperature on enterococcal counts in *Cheonggukjang* inoculated with (**a**) *B. subtilis* KCTC 3135, (**b**) *B. subtilis* KCTC 3135 and *E. faecium* KCCM 12118, and (**c**) *B. subtilis* KCTC 3135 and *E. faecium* CJE 216. •: 25 ◦C, -: 37 ◦C, : 45 ◦C. Error bars indicate standard deviations calculated from duplicate experiments.

#### 3.6.2. Effect of Fermentation Temperature on Tyramine Content in *Cheonggukjang*

The tyramine content of *Cheonggukjang* co-inoculated with either *E. faecium* KCCM 12118 or *E. faecium* CJE 216, along with *B. subtilis* KCTC 3135, was measured during fermentation, as seen in Figure 5. The tyramine content of the control group without *E. faecium* was detected at concentrations that did not exceed 10 mg/kg in all fermentation conditions (Figure 5a). In contrast, other groups with *E. faecium* contained higher levels of tyramine, which indicated that *E. faecium* was capable of and responsible for producing tyramine in *Cheonggukjang*. In *Cheonggukjang* groups co-inoculated with *E. faecium* KCCM 12118, initial tyramine content increased by 0.78 mg/kg, 33.36 mg/kg, and 101.17 mg/kg at 3 days of fermentation at 25 ◦C, 37 ◦C, and 45 ◦C, respectively (Figure 5b). As for *Cheonggukjang* groups co-inoculated with *E. faecium* CJE 216, initial tyramine content increased by 1.59 mg/kg, 74.11 mg/kg, and 85.14 mg/kg at 25 ◦C, 37 ◦C, and 45 ◦C, respectively, by day 3 of fermentation (Figure 5c). All *Cheonggukjang* groups fermented at 25 ◦C contained the lowest tyramine concentrations at less than 10 mg/kg during the entire fermentation duration. However, at 45 ◦C, the *Cheonggukjang* group co-inoculated with *E. faecium* KCCM 12118 displayed an exceptionally high tyramine content detected at 105.13 ± 5.68 mg/kg, exceeding the recommended limit, as expected owing to the acidic pH described in Section 3.6.1. Both *E. faecium* strains appeared to continuously produce tyramine during *Cheonggukjang* fermentation at 37 ◦C and 45 ◦C (Figure 5b,c). The results of the current study are in agreement with findings reported by Kalhotka, et al. [71], which showed a stronger in vitro tyramine production by *E. faecium* incubated at 37 ◦C than at 25 ◦C. According to Morandi, et al. [69], *E. faecium* metabolic activity was detected to be higher at 37 ◦C than at 25 ◦C during milk fermentation. The previous studies have indicated that lower temperatures may reduce both metabolic activity and tyramine production of *E. faecium*. BA content during fermentation at higher temperatures may even reach dangerously high levels as reported by Kang, et al. [43]. In the same report, tyramine concentrations in *E. faecium*-inoculated *Cheonggukjang* fermented at 45 ◦C for

48 h increased (up to 698.67 mg/kg) during the fermentation period, and consequently exceeded the recommended limit for consumption. Jeon, et al. [19] also reported strong tyramine production by *Enterococcus* spp. during soybean fermentation at 37 ◦C. The report demonstrated that tyramine concentrations continued to increase as fermentation progressed. Given the results, safety precautions regarding the limitation of fermentation duration and temperature appear to be necessary as extended periods of fermentation as well as high fermentation temperatures may increase tyramine content in *Cheonggukjang* beyond the recommended safe limit for consumption. Besides, the results showing a lower tyramine content in *Cheonggukjang* during fermentation at lower temperatures coincide with the results in the previous section that displayed a reduction in enterococcal count alongside a decrease in fermentation temperature. Taken together, the present study indicates that lower fermentation temperatures inhibit enterococcal growth, thereby limiting acid production and maintaining low levels of tyramine in *Cheonggukjang*. Therefore, utilizing lower temperatures for *Cheonggukjang* fermentation may reduce the risks associated with *Enterococcus* growth and tyramine accumulation.

**Figure 5.** Effect of fermentation temperature on tyramine content in *Cheonggukjang* inoculated with (**a**) *B. subtilis* KCTC 3135, (**b**) *B. subtilis* KCTC 3135 and *E. faecium* KCCM 12118, and (**c**) *B. subtilis* KCTC 3135 and *E. faecium* CJE 216. •: 25 ◦C, -: 37 ◦C, : 45 ◦C. Error bars indicate standard deviations calculated from duplicate experiments.

3.6.3. Effect of Fermentation Temperature on *tdc* Gene Expression by *E. faecium* Strains in *Cheonggukjang*

The changes in *tdc* gene expression by tyramine-producing *E. faecium* strains were detected and quantified during fermentation of *Cheonggukjang* at 25 ◦C, 37 ◦C, and 45 ◦C. As *Cheonggukjang* is mostly fermented at 37 ◦C, the *tdc* gene expression detected in *Cheonggukjang* groups fermented at 45 ◦C was normalized to that detected in the corresponding groups fermented at 37 ◦C according to the *E. faecium* strains used as inoculants. In *Cheonggukjang* fermented at 25 ◦C, tyramine content continuously remained at concentrations lower than 10 mg/kg, and *tdc* gene expression by *E. faecium* KCCM 12118 and *E. faecium* CJE 216 was not detected in all *Cheonggukjang* groups. In contrast, the highest *tdc* gene expression by *E. faecium* KCCM 12118 was detected in *Cheonggukjang* fermented at 45 ◦C and was upregulated in the range of 1.90- to 7.15-fold throughout *Cheonggukjang* fermentation, compared with that in *Cheonggukjang* fermented at 37 ◦C (Figure 6a–c, left). As for *Cheonggukjang* fermented at 45 ◦C with *E. faecium* CJE 216, downregulation of *tdc* gene expression was observed at 0.82-fold on day 1, and the expression was then upregulated in the range of 1.90- to 3.39-fold thereafter (Figure 6a–c, right). Consequently, both tyramine content and *tdc* gene expression were highest in *Cheonggukjang* groups fermented at 45 ◦C (viz., high-temperature group). Nonetheless, the variation in detected *tdc* gene expression levels during fermentation did not necessarily reflect the tyramine content observed for *Cheonggukjang*. After one day of fermentation, the *Cheonggukjang* group with *E. faecium* CJE 216 fermented at 37 ◦C contained a lower tyramine content than at 45 ◦C; however, *tdc* gene expression was slightly higher at 37 ◦C as described right above. The results showed that there may be differences between gene expression level and enzyme activity (and products thereof). Glanemann, et al. [72] reported that, in vitro, the mRNA response levels do not necessarily reflect the protein response levels

or enzyme activity. As previously suggested by Ladero, et al. [73], while the correlation between BA content and gene expression is not always linear, RT-qPCR remains a reliable method to detect and quantify BA-producing bacteria in food products. Similarly, in our preliminary tests conducted under different incubation conditions, tyramine production by *E. faecium* strains in an assay medium appeared to be insignificantly related to *tdc* gene expression level (data not shown). Therefore, utilizing HPLC analysis appears to be essential for the quantification of BA content and/or bacterial BA production in food samples [31,73,74]. When utilized in conjunction, the complementary methods, that is, HPLC and RT-qPCR, sufficiently allow for the quantitative analysis of both the BA content and tyramine-producing bacteria (including enterococci) in food products [24,31]. In the present study, the results derived from both techniques indicated that the fermentation of *Cheonggukjang* at high temperatures results in increased *tdc* gene expression and tyramine production. Therefore, low-temperature fermentation appears to be necessary to minimize both *tdc* gene expression and tyramine production by *Enterococcus* spp. and thereby ensure the safety of fermented soybean products.

**Figure 6.** Effect of fermentation temperature on *tdc* expression by *E. faecium* strains in *Cheonggukjang* on (**a**) day 1, (**b**) day 2, and (**c**) day 3 of fermentation. -: 37 ◦C, : 45 ◦C. <sup>1</sup> *Cheonggukjang* groups were co-inoculated with *B. subtilis* (KCTC 3135) and *E. faecium* (KCCM 12118 or CJE 216) strains. The expression levels observed in groups fermented at 45 ◦C were normalized to those detected in the corresponding groups fermented at 37 ◦C. Expression of *tdc* gene was not detected in *Cheonggukjang* fermented at 25 ◦C. Error bars indicate standard deviations calculated from duplicate experiments.

#### **4. Conclusions**

The current study assessed the safety risk of tyramine in *Cheonggukjang*, diagnosed the microbial causative agent (i.e., *E. faecium*) responsible for high tyramine levels, and evaluated the impact of fermentation temperature on enterococcal growth (as well as acid production and *tdc* gene expression) and tyramine production. Of the retail *Cheonggukjang* examined, half of the products contained tyramine content that exceeded the recommended limit for safe consumption by up to a factor of approximately 14. *E. faecium* strains isolated from the retail *Cheonggukjang* products were highly capable of producing tyramine in assay media, which indicated that the species is principally, or at least partly, responsible for tyramine accumulation in the food.

During in situ fermentation at different temperatures, the tyramine content of *Cheonggukjang* groups co-inoculated with *B. subtilis* (used as an inoculant to ferment soybeans) and *E. faecium* (either isolated in this study or designated previously as the type strain) strains was highest at 45 ◦C, followed by 37 ◦C and 25 ◦C. On the other hand, the control group inoculated with only *B. subtilis* strain (without any *E. faecium* inoculants) had the lowest tyramine content at all fermentation temperatures, which supported the notion that *E. faecium* may be a key producer of tyramine in *Cheonggukjang*. Another implication of the results was that a lower fermentation temperature leads to a lower tyramine content below the recommended limit in *Cheonggukjang*, even though the tyramine content continually increases during fermentation. Therefore, low temperatures and a short fermentation duration may reduce the accumulation of tyramine caused by *E. faecium* growth in *Cheonggukjang*, thereby reducing the safety risks associated with consuming food with high BA concentrations.

**Author Contributions:** Conceptualization, Y.K.P. and J.-H.M.; Investigation, Y.K.P., Y.H.J., J.-H.L., B.Y.B., and J.L.; Formal analysis, J.L.; Writing—original draft, Y.K.P.; Writing—review and editing, Y.K.P., Y.H.J., K.C.J., and J.-H.M.; Supervision: J.-H.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (no. 2020R1I1A3052118).

**Acknowledgments:** The authors thank Jae Hoan Lee and Alixander Mattay Pawluk of Department of Food and Biotechnology at Korea University for technical assistance and English editing, respectively.

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

#### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

## **Biogenic Amines, Phenolic, and Aroma-Related Compounds of Unroasted and Roasted Cocoa Beans with Di**ff**erent Origin**

**Umile Gianfranco Spizzirri 1, Francesca Ieri 2,\*, Margherita Campo 2, Donatella Paolino 3, Donatella Restuccia <sup>1</sup> and Annalisa Romani <sup>2</sup>**


Received: 27 June 2019; Accepted: 30 July 2019; Published: 1 August 2019

**Abstract:** Biogenic amines (BAs), polyphenols, and aroma compounds were determined by chromatographic techniques in cocoa beans of different geographical origin, also considering the effect of roasting (95, 110, and 125 ◦C). In all samples, methylxantines (2.22–12.3 mg kg<sup>−</sup>1) were the most abundant followed by procyanidins (0.69–9.39 mg kg<sup>−</sup>1) and epicatechin (0.16–3.12 mg kg−1), all reduced by heat treatments. Volatile organic compounds and BAs showed variable levels and distributions. Although showing the highest BAs total content (28.8 mg kg−1), Criollo variety presented a good aroma profile, suggesting a possible processing without roasting. Heat treatments influenced the aroma compounds especially for Nicaragua sample, increasing more than two-fold desirable aldehydes and pyrazines formed during the Maillard cascade and the Strecker degradation. As the temperature increased, the concentration of BAs already present in raw samples increased as well, although never reaching hazardous levels.

**Keywords:** cocoa nibs; roasting; bioactive amines; polyphenols; volatile organic compounds; geographical areas

#### **1. Introduction**

Cocoa beans represent the basic raw material in the production of chocolate and cocoa-based products. During processing, cocoa beans undergo important manipulations, including fermentation and roasting, which drastically influence the quality of the final product. During fermentation, cocoa beans are exposed to the action of various microorganisms and enzymes, while in the roasting process, high temperatures determine important modifications on the cocoa bean's composition [1].

Cocoa is a food rich in polyphenols, mainly flavonoids, procyanidins, and flavan-3-ols. The preservation or enhancement of cocoa procyanidins is of great importance since these compounds, despite their poor bioavailability, have been related to the health beneficial effects of cocoa, particularly in cardiovascular diseases [2]. Polyphenol and xanthine content in cocoa seeds changes during ripening and during the processing phases [3]. Microbial activity during fermentation and the drying process contribute to settling the amounts of theobromine and caffeine and their relative abundances, polyphenol amounts, in particular of catechin and epicatechin, and the amounts of organic acids, sugars, mannitol, ethanol, and alkaloids, thus influencing the quality and the biological properties of the finished product [4]. At the beginning of fermentation, during the first three days, the highest contents of total phenolic compounds and total anthocyanins prevailed in cocoa beans. Finally, at the end of

cocoa beans fermentation, the lowest contents of total phenolic compounds, and total anthocyanins, were observed [5].

In addition, roasting temperature has been seen to have effects on the flavanols amounts causing losses and structural modifications, particularly epimerization of both monomers and polymers. At high roasting temperatures, a progressive loss of (−)-epicatechin and (+)-catechin and an increase in (−)-catechin were observed as a result of heat-related epimerization from (−)-epicatechin; additionally, a temperature-related epimerization of procyanidin dimers has been reported [6,7]. These structural modifications could have negative effects on the biological properties of the product, being (−)-epicatechin the most bioavailable isomer and (−)-catechin the one with the lowest bioavailability. Roasting processes may also cause reduction of the content of hydroxycynnamic compounds, clovamide in particular [8], with a possible further reduction of the antioxidant activity.

The secret of the flavor of chocolate, so highly appreciated worldwide, resides mainly in its volatile aromatic fraction. Its complex composition depends on the cocoa bean genotype and is the consequence of several processes [9,10]. To date, descriptive studies have identified >600 volatile compounds in cocoa and chocolate products [11,12], mainly pyrazines, esters, amines and amides, acids, and hydrocarbons. Besides cocoa flavor precursors, also toxic molecules, such as biogenic amines (BAs), can be produced during processing. BAs are a class of organic, basic, nitrogenous compounds with low molecular weight, are usually part of bioactive molecules of cocoa beans and derivatives. They mainly derived by decarboxylation of corresponding amino acids, due to the action of suitable enzymes widely distributed in spoilage bacteria and other microorganisms, as well as in naturally occurring and/or artificially added bacteria involved in food fermentation [5,13]. Although several amines (i.e., natural polyamines) are present in living cells and contribute to promoting many human physiological functions, these compounds can represent a serious health hazard for humans, when present in food in significant amounts or ingested in the presence of potentiating factors, such as amine oxidase-inhibiting drugs, alcohol, and gastrointestinal diseases. Then, their attendance in foodstuffs is often undesirable, because often associated with several of pathological syndromes, such as headaches, respiratory distress, heart palpitations, hypo- or hypertension and several allergenic disorders [14]. In particular, tyramine, β-phenylethylamine, and histamine have been considered as the initiators of hypertension and dietary-induced migraines, while the neurotransmitter serotonin is essential in the regulation of appetite, body temperature, and sleep [15].

Generally speaking, a variety of factors determining cocoa and cocoa derivatives quality are strongly related to the cocoa beans processing, from the opening of the fruit until the end of industrial processes [16]. In addition, qualitative characteristics of the cocoa beans are a consequence of the differences in the farming practices regarding growing, fermenting, and drying the cocoa beans, with significant differences sometimes found in samples from the same country [16,17].

Recently, it has been reported that in cocoa beans the amino acid oxidative decarboxylation can be also obtained during food processing suggesting a new chemical, heat-induced formation of BAs [18]. It follows that, in addition to the amino acid catabolism produced by microorganisms, amino acids can also be degraded chemically as a consequence of thermal treatment of foods [19]. These reactions are responsible for the formation of taste and flavor compounds, reducing, at the same time, the concentration of essential amino acids and contributing to the accumulation of compounds that may be dangerous for consumers, such as BAs [20]. It follows that two main reasons can be underlined accounting for the analysis of BAs in foods: first their potential toxicity; second the possibility of using them as food quality markers as their concentration can be related with the hygienic-sanitary quality of the process and with the freshness of the raw materials and the processed products. As BAs have been widely exploited as important indicators of safety and quality in a variety of foods, many papers appeared in recent years reporting their quantitative determination in many fermented and non-fermented foods, including cocoa and its derivatives [13].

In this work, the quantitative determination of biogenic amines, xanthine, and polyphenol molecules was performed by liquid chromatography (LC) techniques on cocoa beans from different origin. The aroma profile of cocoa nibs was investigated by headspace solid-phase micro-extraction (HS-SPME) combined with gas chromatography-mass spectrometry (GC-MS). Additionally, the same compounds were monitored after roasting process at different temperatures to establish a possible correlation between heating and different biomolecules profiles.

#### **2. Materials and Methods**

#### *2.1. Samples*

Seven cocoa beans samples from different world areas and years were considered as reported in Table 1. All fruits studied in this work were considered as well fermented. Approximately 500 g portions of Coviriali and O'Payo cocoa beans of uniform size were random selected from 50 kg of each nib variety peeled and roasted in a forced airflow-drying oven ROASTER CENTOVENTI vertiflow® system (Selmi Chocoloate Machinery, Cuneo, Italy) at Meraviglie S.r.l. (Verona, Italy). Applied in these studies, heat treatment parameters were chosen to obtain a range of roasted beans with acceptable physico-chemical and sensory properties. The parameters of thermal processing were optimized to avoid over-roasting, varying time of roasting as soon as bean cracking occurs, as follows: temperatures of 95, 110, and 125 ◦C for 60, 30, and 20 min respectively. Samples from all bean types were prepared using differentiated methods according to the analyses carried out as reported in the subsections of the experimental section.


**Table 1.** List of samples. nrnot roasted; r195 ◦C; r2110 ◦C; r3125 ◦C.

#### *2.2. Chemicals*

BAs spermine (SPM, tetrahydrochloride), spermidine (SPD, trihydrochloride), putrescine (PUT, dihydrochloride), histamine (HIM, dihydrochloride), tyramine (TYR, hydrochloride), β-phenylethylamine (PHE, hydrochloride), cadaverine (CAD, hydrochloride), as well as dansyl chloride, ammonia (30%), perchloric acid, and LC solvents (acetonitrile and methanol LC grade) were purchased from Sigma-Aldrich (Milford, MA, USA). Ultrapure water was obtained from Milli-Q System (Millipore Corp., Milford, MA, USA). Filters (0.45 and 0.20 μm) were purchased from Sigma-Aldrich. SPE C18 cartridges (0.5 g) were obtained from Supelco Inc. (Bellefonte, PA, USA). All GC chemicals were from Sigma-Aldrich (Milford, MA, USA). The HPLC grade standards (±)-catechin hydrate, theobromine, caffeic acid, quercetin-3-glucoside were purchased from Sigma-Aldrich (Milford, MA, USA).

#### *2.3. Samples Preparation for the Analysis of Polyphenols and Xanthines*

The peeled cocoa beans were crushed in a mortar, then 1.0 g accurately weighed of crushed material was extracted in 10.0 mL of a solution EtOH:H2O 70:30 at pH 3.2 by addition of HCOOH, at room temperature, for 24 h under stirring. The solid material was removed by filtration under vacuum and the extracts analyzed by high performance liquid chromatography coupled with diode array detection and electrospray ionization mass spectrometer (HPLC-DAD-ESI-MS) and by high performance liquid chromatography coupled with diode array detection and fluorescence detector (HPLC-DAD-FLD).

#### *2.4. Chromatographic Conditions Xanthine and Polyphenol Determination*

The method used for the quali-quantitative analysis, and described below, was optimized according to literature data and previous studies of this research group about polyphenolic compounds and xanthines, and modifying them based on the specific results [8,21–23]. The analyses were performed using a HP-1200 Liquid Chromatograph with a DAD and a fluorescence detector and a HP-1100 MSD API Electrospray (Agilent Technologies, Palo Alto, CA, USA) operating in negative and positive ionization mode. Gas temperature was 350 ◦C, flow rate 10.0 L/min, nebulizer pressure 30 psi, quadrupole temperature 300 ◦C, capillary voltage 3500 V, and fragmentor 120 eV.

For the chromatographic separation a Luna C18 250 × 4.60 mm, 5 μm (Phenomenex, Torrance, CA, USA) column was used operating at 26 ◦C. A multistep linear solvent gradient starting from 95% H2O at pH 3.2 by addition of HCOOH (A), up to 100% CH3CN (B) was performed with a flow rate of 0.8 mL min−<sup>1</sup> over a 63 min period. In detail, the applied gradient started with 95% A; from 95% A to 85% A in 20 min; isocratic elution 85% A until 30 min; from 85% A to 75% A in 9 min; isocratic elution 75% A until 47 min; from 75% A to 15% A in 2 min; isocratic elution 15% A until 53 min; from 15% A to 0% A in 2 min; 0% A until 60 min; from 0% A to the initial conditions in 3 min. Figure 1 reports on the chromatographic profile of raw O'Payo bean extract (2016) by DAD at 280 nm (A) and 330 nm (B) and FLD ex. 280 nm; em. 315 nm (C).

**Figure 1.** Chromatographic profile of raw O'Payo bean extract (2016): diode array detection (DAD) 280 nm (**A**) and 330 nm (**B**); fluorescence detector (FLD) ex. 280 nm; em. 315 nm (**C**). TBR—Theobromine; CAF—Caffeine; PRO—Procyanidins; CAA—Caffeoyl aspartic acid; QHE—Quercetin hexoside; QAR—Quercetin arabinoside; CAT—Catechin; EPI—Epicatechin.

#### *2.5. Calibration*

Quantitation of xanthines, flavonols, hydroxycinnamic derivatives, and procyaninides was performed by HPLC-DAD using five-point regression curves built with the available standards. Curves with an r2 > 0.9998 were considered. Calibration was performed at the wavelength of the maximum UV-Vis absorbance, applying the correction of molecular weight. In particular, the extinction coefficient of each quantified compound being comparable to that of the specific standard used for its calibration, the weights in mg were calculated by multiplying the weight obtained from the calibration by a correction factor given by the ratio between the molecular weight of the compound and the molecular weight of the standard used for its calibration. In particular, xanthines were calibrated at 280 nm using theobromine as reference; procyanidins were calibrated at 280 nm using catechin hydrate as reference; hydroxycinnamic derivatives were calibrated at 330 nm using caffeic acid as reference; flavanols were calibrated at 350 nm using quercetin as reference. The quantitation of catechin and epicatechin was performed by HPLC-FLD, using a five point calibration curve (r2 = 0.9999) built with standard solutions of catechin hydrate. The fluorescence detector was set as follows: excitation wavelength 280 nm; emission wavelength 315 nm [24].

#### *2.6. HS-SPME-GC-MS Analyses*

Headspace solid-phase micro-extraction combined with gas chromatography–mass spectrometry (HS-SPME-GC-MS) was selected as the most suitable technique to recover and analyze the Volatile Organic Compounds (VOCs) in peeled cocoa beans samples. Samples were ground and homogenous powders were obtained. A total of 1 g of the powdered sample, was placed into a 20-mL screw cap vial fitted with PTFE/silicone septa. An Internal Standard (IS) in suitable amount was added to each sample (IS: ethylacetate-D8; 1-Butanol-D10; ethyl hexanoate-D11; 5-methyl-hexanol; acetic acid-D3; Hexanoic acid-D11; 3,4-Dimethylphenol;). The Internal Standard was used for normalizing the analyte responses over the area of the IS, to minimize the instrumental error during the time of analysis.

After some trials aimed at optimizing amounts of sample, exposure time, and temperature, SPME conditions were set as follows: after 5 min of equilibration at 60 ◦C, VOCs were absorbed exposing a 1-cm divinilbenzene/carboxen/polydimethylsiloxane SPME fiber (DVB/CAR/PDMS by Supelco) for 15 min into the vial headspace under orbital shaking (500 rpm) and then immediately desorbed at 280 ◦C in a gas chromatograph injection port operating in split less mode. The chromatographic analysis was performed in a GC system coupled to quadrupole mass spectrometry using an Agilent 7890a GC equipped with a 5975C MSD (Agilent Technologies, Palo Alto, CA, USA). The separation of analytes was achieved by an Agilent DB InnoWAX column (length 50 m, id 0.20 μm, df 0.40 μm). Chromatographic conditions were: initial temperature 40 ◦C, then 10 ◦C min−<sup>1</sup> up to 260 ◦C, hold for 6.6 min. Compounds were tentatively identified by comparing calculated Kovats retention index and mass spectra of each peak with those reported in mass spectral databases, namely the standard NIST08/Wiley98 libraries. Quadrupole MS operated in full-scan mode from which the specific ions of the analyte were extracted. Only the compounds with higher intensity were identified in order to select major compounds over a complex mixture of VOCs. Each sample was analyzed in triplicate.

#### *2.7. Amine Standard Solutions and Calibration*

A calibration curve was built starting from 1.0 mg mL−<sup>1</sup> standard solution of each amine in purified water and preparing 12 BAs standard mixtures to a final volume of 25 mL employing HClO4 0.6 mol L−1. Amine final concentrations were 0.1, 0.5, 0.8, 2.0, 4.0, 5.0, 10.0, 16.0, 25.0, 50.0, 75.0, and 100.0 μg mL<sup>−</sup>1. The comparison between the retention times of peaks of samples and standard solutions allowed the identification of each BA. Standard concentration against peak area allowed to build a calibration plot, and six independent replicates for each concentration level were performed. Moreover, the matrix effect was evaluated by comparison of external calibration plots, depicting concentration of standard solutions versus peak area, with standard addition method plots, depicting peak area versus concentration of standard solutions added to the sample. No significant matrix effect was recorded because of the slopes of the two plots were not significantly different. Quantitative determination was then accomplished by direct interpolation in the external calibration plot of each BA. Chromatogram of a standard mixture of BAs is displayed in Figure 2A.

**Figure 2.** LC-UV chromatogram of biogenic amines (BAs) standard mixture at concentration of 100 μg mL−<sup>1</sup> (**A**) and sample 1 (**B**). The chromatogram was obtained employing gradient conditions as specified in the Materials and Methods section.

#### *2.8. BAs Extraction and Purification*

The extraction of BAs from peeled cocoa beans samples was performed by adding 20 mL of HClO4 0.6 mol L−<sup>1</sup> to about 5.0 g of grounded sample, in a 50.0 mL test tube. The mixture was homogenized (vortex at 40 Hz for 40 min), centrifuged (10,000× *g* for 20 min), filtered (syringe filter 0.20 μm), collected in a plastic vial and purified by SPE on a C18 sorbent (conditioning: 2.0 mL of H2O and 2.0 mL (two times) of CH3OH; loading: 5.0 mL of the basified sample; washing: 2.0 mL of NH4OH at pH 11.0; eluting: 2.0 mL (two times) of CH3OH). Nitrogen gas was employed to dry eluting solution providing a solid residue that was re-dissolved in a plastic test tube with 1.3 mL of extraction solvent.

To perform recovery experiments sample 1 was spiked, before the extraction procedure, with an aliquot of standard mixture of BAs. Specifically, 5.0 g of peeled cocoa beans were spiked with 1.0 mL of 25.0 mg L−<sup>1</sup> BAs standard solution. Method validation was obtained in terms of recovery percentages, linearity, intra- and inter-day repeatability, limits of quantification and limits of detection (LOQs and LODs), to confirm analytical suitability [25].

Dansylation reaction was performed by adding at 1.0 mL of standard solution (or acid sample extract spiked with BAs or acid sample extract) 200 μL of NaOH 2.0 mol L−1, 300 μL of saturated NaHCO3 solution and 2.0 mL of dansyl-chloride solution (10.0 mg mL−<sup>1</sup> in acetone prepared just before use). After 30 min, dansyl-chloride in excess was removed with 100 μL of NH4OH 25% (v/v) and the suspension filtered by a 0.45 μm syringe filters. Finally, 20 μL was injected for LC-UV analysis. Figure 2B shows a chromatogram of sample 1.

#### *2.9. Chromatographic Conditions for BAs Quantification*

Jasco PU-2080 instrument equipped with a Rheodyne 7725 injector with a 20 mL sample loop and a gradient pump (PU-2089 plus, Jasco Inc., Easton, MD, USA) was employed to obtain the chromatograms. The system was interfaced with an UV detector operating at λ = 254 nm (UV-2075, Jasco Inc., Easton, MD, USA). Data were collected and analyzed with an integrator Jasco-Borwin1. A reverse-phase C18 column (250 mm × 4.6 I.D., 5 mm) (Supelco Inc., Bellefonte, PA, USA) equipped with a C18 guard-pak (10 mm × 4.6 I.D., 5 mm) were used (Supelco Inc., Bellefonte, PA, USA) for separation of BAs. Two solvent reservoirs containing (A) purified water and (B) acetonitrile were used to separate all the BAs with a gradient elution which began with 3 min of isocratic program A-B 50:50 (v/v) reaching after 20 min A-B 10:90 (v/v). Then 3 min of isocratic elution was carried out and 4 min further where necessary to restore again the starting conditions (A-B 50:50, v/v). A constant flow at 1.2 mL min−<sup>1</sup> was employed.

#### *2.10. Statistical Analyses*

All analyses were performed in triplicate and data were expressed as mean ± relative standard deviations (RSD). Studies of the correlation coefficient and linear regression, calculation of average, assessment of repeatability, standard deviation, and RSD were performed using Microsoft Excel 2010 software. Significance was performed using a one-way analysis of variance (ANOVA) test, employing Duncan's multiple range test at significance level *p* < 0.05.

#### **3. Results and Discussion**

#### *3.1. Polyphenol Content in Cocoa Beans*

To identify the phenolic compounds and xanthines, UV-Vis absorption spectra, mass spectra and literature data were used and combined for tentative identification of the analytes. In the present study, catechin and epicatechin were quantified through calibration by using a FLD detector, whereas oligomeric and polymeric procyanidins were quantified through DAD calibration. FLD calibration was used for catechin and epicatechin because of its higher sensibility and specificity, needed for catechin in particular, that often partially coelutes with caffeine, and is always present in low amounts with respect to this latter (average amount of catechin with respect to caffeine: 5.2% in the analyzed samples). In these conditions, the fact that both catechin and caffeine have also very similar wavelengths of maximum UV-Vis absorption hinders a correct quantification of the flavanol by using a DAD detector. Conversely, unlike caffeine, catechin and epicatechin emit a very intense fluorescence signal in the experimental conditions (excitation wavelength 280 nm; emission wavelength 315 nm), easily detectable and measurable also in presence of methylxanthines [24,26]. On the other hand, for the calibration of the other procyanidins a DAD detector is needed, because fluorescence detection is insensitive to procyanidins containing a gallic acid ester and/or gallocatechins as monomeric units [23]. In this case, the use of a fluorescence detector could lead to an underestimation of the total procyanidin content.

In Table 2 polyphenol distributions and total amounts in fermented, not-roasted cocoa beans of different origin are reported. Xanthines, theobromine in particular, are the most abundant compounds and their amounts decrease with increasing roasting temperature, after an initial slight increase by roasting at 95 ◦C, probably due to a further loss of water after the drying process.

Among the raw samples analyzed, the highest in polyphenols were 1nr (Camino verde 2015), 3nr (Coviriali 2015) and 6nr (O'Payo 2016), respectively with 10.67, 10.53, and 12.96 mg g−<sup>1</sup> total polyphenols. In all samples, it is possible to observe a clear predominance of epicatechin with respect to catechin; the highest [epicatechin]:[catechin] ratio (27.9) was found for the raw sample from Madagascar (2015), but its low content in polyphenols (1.44 mg g−1) suggests a lower quality with respect to the other samples under study. The other two samples harvested in 2015 (Camino verde and Coviriali) appear not to contain catechin, so it was impossible to evaluate the ratio even though the epicatechin amount appears to be high in particular for the Camino verde sample (2.84 mg g−1). The lowest [epicatechin]/[catechin] ratio was found for the raw samples Cheni and Criollo (5.0 for both the samples).



For Coviriali 2016 and O'Payo 2016 beans it was possible to compare the polyphenols contents after roasting at 110 ◦C, confirming that the total amounts of polyphenols significantly decrease with respect to the raw samples; Coviriali beans were roasted also at a higher temperature (125 ◦C) with a further decreasing of total polyphenols. According to previously reported data [27,28], also [epicatechin]/[catechin] ratios follow the same trend depending on the roasting process at high temperature. Procyanidins, quantified as catechin equivalents, are the most abundant polyphenols in all samples and their amounts also appear to be negatively influenced by the roasting process [28]. Two flavonolic compounds were found, quercetin hexoside and quercetin arabinoside, in low amounts and apparently not depending on the roasting temperature. The only hydroxycynnamic derivative present in the extracts was caffeoyl aspartic acid [3].

The characteristics of cocoa beans and derived food products, such as hydrophilic and volatile secondary metabolites profile, organoleptic properties, antioxidant activity etc., depend not only on fermentation and processing methods, but also on several variables related to genetics, geographical regions of cultivation, agronomical practices, and climatic conditions [29,30]. In particular, concerning phenolic compounds, the low amount of total polyphenols (1.28 mg g−1) found for Criollo variety is reliable based on literature data that identify this variety as the one with the lowest content of polyphenols [31,32]. The Forastero variety includes also Nacional Forastero that is the Forastero variety cultivated in northern Ecuador [31]. The analyzed Forastero samples were harvested in 2015 (Nacional Forastero from Ecuador, sample 1nr, and Forastero from Junìn region, Peru, sample 3nr) and 2016 (Forastero from Junìn region, sample 4nr, and Forastero from Satipo region, Peru, sample 5nr). For Forastero beans harvested in 2015, no significant difference was found between their contents of total polyphenols; interestingly, catechin was not detected in either of the two samples. Conversely, statistically significant differences were found between epicatechin and procyanidins contents. For Forastero beans harvested in 2016, total polyphenols are higher in the sample from Satipo region than in the one from Junìn region, but it must be taken into consideration that the total polyphenols content varies also according to climate variations, thus possibly from one year to another, as it can be seen for the two Junìn region samples of 2015 and 2016 (10.55 and 5.16 mg/g total polyphenols respectively). Moreover, a small difference was found between their contents of catechin, whereas the difference between epicatechin amounts is more evident. The differences between polyphenolic compositions are evident for the two samples of Trinitario variety from Sambirano region, Madagascar (sample 2nr) and Bali, Indonesia (sample 6nr). Sample 6nr, not roasted, is the highest in polyphenols among all the samples analyzed in the present study (12.96 mg/g total polyphenols), and its polyphenols content consists mainly of procyanidins (9.17 mg/g). Sample 2nr shows a consistently lower total polyphenols content (1.44 mg/g), of which 1.00 mg/g procyanidins remaining the most represented polyphenolic subclass. Monomeric flavanols are also higher in the sample from Indonesia, but the epicatechin/catechin ratio is better for the other sample (12.9 vs. 27.9). Again, it must be noted that the samples 2nr and 6nr differ not only concerning their geographical origins but also regarding the years of production (2015 and 2016).

#### *3.2. VOCs in Cocoa Beans*

After a successful fermentation process, it is necessary to reduce the water content of the cocoa seeds to between 5% and 8% and this is achieved by drying [33]. The drying process is not only important in preserving the cocoa seeds but also plays a very crucial role in the development of cocoa flavor and the overall quality of the raw cocoa seeds. Cocoa is dried to minimize the formation of molds and to reduce the acid level and astringency of the beans. Then the roasting of the cocoa seeds takes place in the consumer countries. Among the cacao beans analyzed, the 7nr sample (Criollo) beans will be used by a chocolate maker that produces raw chocolate, so beans will be not roasted and never reach temperatures of more than 42 ◦C. Especially for this product, the quality of cocoa beans is a very big determinant of the final taste of the chocolate. Cocoa flavor resides in volatile fraction, which is composed of a complex mixture of up to 600 compounds with new research continuously increasing this number [12].

HS-SPME coupled to GC-MS has proven a valuable tool for analysis of volatile and semi-volatile compounds from cocoa and chocolate products. The technique is very sensitive to experimental conditions and in this study, the DVB/CAR-PDMS fiber was found to afford the most efficient extraction of both volatile and semi-volatile compounds from the analyte's headspace according to literature [34].

The HS-SPME-GC-MS analysis of raw cocoa seeds allowed the extraction of a complex mixture of VOCs and the compounds with higher intensity were selected and reported in Table 3. The key VOCs considered in this work belonged to the class of alcohols, aldehydes, esters, acids, ketones, pyrazines, and terpenes and they have all been previously reported in other works [12,34].

The main VOCs were associated with the aroma of vinegar (acetic acid) and with the aroma of roasted and nutty (tetramethyl-pyrazine). High level of acetic acid could influence in a negative way the final aroma of chocolate, moreover pyrazines were considered important contributors to the desirable chocolate aroma [12] and changed during roasting [35]. Additionally, the aldehydes 2-methylbutanal and 3-methylbutanal are reported to have a strong influence on the chocolate flavor [12]. Phenylethyl alcohol and 3-Methylbutyl acetate were considered key aroma compounds and were associated to the odor of floral, rose and sweet, fruity, banana, respectively. There were significant differences among the types of beans, in particular for the abovementioned VOCs (Table 3). The 2nr (Madagascar) sample showed highest level of acetic acid and tetramethyl-pyrazine, instead of 1nr (Camino Verde) showing the lowest values. O'Payo beans (sample 6nr) showed highest levels of the aldehydes with chocolate and almond aroma, 2-methylbutanal, 3-methylbutanal and benzaldehyde. Cheni beans (sample 5nr) showed highest level of banana flavor (3-Methylbutyl acetate) and Coviriali beans (sample 4nr) the highest value of rose aroma (Phenylethyl alcohol).

The 7nr sample (Criollo) showed a large number of VOCs, not reported in Table 3, as 2-Heptanol for the class of alcohols, as 2-Pentanol acetate for esters and as 2-Heptanone for ketones. The high variety of VOCs, the low level of acetic acid and the good quantity of pyrazines and aldehydes confirmed the high quality of this variety and the possible use without roasting to produce raw chocolate.

Forastero, Criollo, Trinitario, and Nacional, the variety grown in Ecuador, exhibit differences in flavor characteristics that can be attributed to original variety but also growing conditions and geographical origin [36]. The extent to which other factors such as climate and soil chemical compositions influence the formation of flavor precursors and their relationships with final flavor quality remains unclear [36]. Some authors have studied the influence of cocoa's origin on the composition in volatile compounds and profile comparison allowed beans, liquor, and chocolate from various geographical origins to be distinguished [29,37].

Trinitario samples from Madagascar (2nr) and Nicaragua (6nr) exhibited high differences in many of key VOCs considered and also Forastero samples from different areas of Peru, 4nr and 5nr, showed significant differences in volatile composition.

Cocoa beans underwent the roasting process by means of a dry heat treatment, changing chocolate flavor. Flavor precursors developed during fermentation interact in the roasting process. Aldehydes and pyrazines are among the major compounds formed during roasting. They are formed through the heat induced Maillard reaction and Strecker degradation of amino acids and sugars [35]. The roasting process not only generated and increased the concentration of some flavor compounds through pyrolysis of sugars, but also reduced the amount of minor compounds affecting the final quality of chocolate [38]. The degree of chemical changes depends on the temperature applied during the process [38]. The HS-SPME-GC-MS analysis of roasted beans confirmed changes in VOCs during roasting as the decrease of acetic acid, especially in samples 4 where a higher level was present, and the increase of pyrazines in both samples (Table 3). Additionally, 3-methylbutanal increased with roasting, especially at 95 and 110 ◦C, while at 125 ◦C there was a return to initial values, due to the prevalence of the volatilization phenomenon compared to the production one. The loss of minor compounds that influence the chocolate's aroma as phenylethyl alcohol and benzaldehyde was observed by increasing the roasting temperature, in particular at 125 ◦C.




Differentlettersexpresssignificantdifferences (*<sup>p</sup>* < 0.05).ndmeansnotdetected.nrnotroasted;r195◦C;

 ◦C;

 ◦C.

#### *3.3. BAs in Cocoa Beans*

It can be assumed that, prior to roasting, the bacterial decarboxilation of the amino acids plays the main role in the BAs production in fresh cocoa beans. In fact, during fermentation, cocoa proteins can be hydrolyzed by microorganisms to release free amino acids, although their total amount can considerably vary [39]. Usually, low amounts of total free amino acids, mostly acidic, were detected in the unfermented seeds. It has been shown that after fermentation, acidic free amino acids decreased, while total free amino acids, as well as hydrophobic free amino acids, increased [40]. The latter aspect seems to be related to the characteristics of the aspartic endoprotease and the carboxypeptidase present in cocoa beans, as the first preferentially attacks hydrophobic amino acids of the storage proteins and the second releases single hydrophobic amino acids [40]. Considering the different optimal temperature and pH of these enzymes, proteolysis primarily depends on the fermentation conditions: duration and intensity of acidification, temperature, and aeration [39]. Once free amino acids are released, they can undergo decarboxylase activity by some bacterial enzymes to form amines [41]. Microbiota evolution during cocoa bean fermentation has been studied extensively, also owing to its importance in the formation of the precursor compounds of the cocoa flavor [42]. It was found that yeasts, filamentous fungi, lactic and acetic acid bacteria as well as members of the genus Bacillus, are typically present, all of them being able to produce BAs [43]. As a consequence of the protection mechanism of bacteria against the acid medium, decarboxylase activity is favored by low pH values during fermentation [44]. Moreover, contaminating bacteria can also decarboxylate amino acids to support a further accumulation of BAs.

In Table 4 BA distributions and total amounts in fermented, not-roasted cocoa beans of different origin are reported. Quantities of total BAs ranged from 13 mg kg−<sup>1</sup> in sample 3a to 28.8 mg kg−<sup>1</sup> in sample 7nr, never reaching hazardous concentrations. Total BAs concentrations collected in Table 4 are in agreement with a recent study, who recorded the evolution of BAs in fresh cocoa beans over a fermentation period of seven days, reaching a maximum level of 39.6 mg kg−<sup>1</sup> at the fourth day of fermentation [5]. On the contrary, considering samples of the same geographical origin, Oracz and Nebesny (2014) reported for raw cocoa beans from Ecuador and Indonesia, much lower total BAs content, not exceeding 5.0 and 6.0 mg kg<sup>−</sup>1, respectively. However, only five BAs were considered in this study, neglecting natural polyamines PUT, SPM, and SPD, as well as, HIS and CAD, representing in our study the most abundant compounds [44].

As can be seen from data in Table 4 some variations are present, depending on the sample. This is not surprising, as it was already underlined that, when analyzing samples of cocoa beans from different countries, several attributes can be very different [45]. In fact, wide variations have been obtained considering cocoa beans coming from big producing countries, much more emphasized in samples obtained from smaller producing countries. Moreover, differences were not only country-dependent, but also farmer-dependent, as significant discrepancies were found in quality attributes of cocoa beans from the same country [39]. To this regard, it is noteworthy that samples 3nr and 4nr were collected from the same farm but harvested respectively in 2015 (sample 3nr) and 2016 (sample 4nr). As it can be seen, BAs profiles and concentrations did not significantly differ, implying a high degree of farming standardization.

*Foods* **2019**, *8*, 306


**Table 4.** Biogenic amines (BAs) in fermented and roasted cocoa beans samples. Results in mg kg−1 vegetal material.

125

nrnot roasted; r195 ◦C; r2110 ◦C; r3125 ◦C.

PHE—β-phenylethylamine,

PUT—putrescine,

CAD—cadaverine,

HIS—histamine,

TYR—tyramine,

SPD—spermidine,

SPM—spermine.

Considering BAs profiles, the data obtained in this study clearly showed that BAs present in all samples at higher concentrations were SPM (3.5–7.2 mg kg−1), HIS (3.1–5.3 mg kg−1), PUT (1.5–3.4 mg kg<sup>−</sup>1), and CAD (1.1–2.1 mg kg<sup>−</sup>1), while TYR (not detected (nd)–4.9 mg kg−1), SPD (nd–7.3 mg kg<sup>−</sup>1) and PHE (nd–1.5 mg kg−1) were present more rarely and at variable concentrations. The presence of natural polyamines in the cocoa beans is expected since they are ubiquitous in plants and all living organisms. It is also known the ability of the bacteria to produce some amines, e.g., TYR, as a protection against the acidic environment, while low levels of PHE in cocoa and derivatives seem to be associated with their aphrodisiac effects and mood lifting [46]. Guillen-Casla et al. (2012) [20] reported that TYR, PHE, serotonin, and HIS were the main amines in cocoa beans, although also PUT, dopamine, and ethanolamine have also been determined. Comparison among samples of same geographical origin (samples 1 and 7), displayed comparable (Ecuador) or higher (Indonesia) amounts of PHE, while our samples always showed much higher concentration of TYR for both raw cocoa beans [44]. In addition, do Carmo Brito et al. (2017) [5] found different results. Only tryptamine, TYR, SPD, and SPM were present during fermentation of fresh cocoa beans, while CAD and PUT where always undetectable in all the analyzed samples. SPD and SPM concentrations increased from the beginning to the end of fermentation, while TYR reached its maximum level at the fourth day of fermentation, decreasing afterward to initial contents [5].

It can be concluded that the differences already recorded for total BAs concentrations are much more evident when considering BAs profiles. This is a very common situation already underlined for many other foods supporting BAs accumulation. Considering that the aminogenesis takes origin from multiple and complex variables, all of which interact, a direct overlapping of the data arising from different studies (or from different samples of the same study, if they are not produced in the same way) is generally difficult to accomplish. Many parameters concerning either the hygienic conditions of the raw materials or the production process, as well as the preservation techniques, influence BAs levels and distributions [5,16,44].

In Table 4 the evolution of BAs concentration evaluated for sample 4nr and 6nr at different roasting temperatures (95, 110 and 125 ◦C samples 4r1, 4r2, 4r3 and 95 and 110 ◦C for sample 6r1, 6r2) is reported. As can be seen, the roasting temperature is strictly related to the amine total amount, reaching for sample 6<sup>b</sup> the maximum level of 58.3 mg kg−1, in agreement with Oracz and Nebesny (2014) [44]. They underlined that temperature and relative humidity of air during roasting influenced the BAs concentrations and profiles a lot. As can be noted from Table 4, sample 6nr contained all the considered amine before roasting. Each amine concentration raised after processing, although to a different extent. In particular, PUT concentration showed the highest increasing factor (7.3) followed by TYR, CAD, HIS, PHE amounts with increasing factors between 2.1 and 2.6. SPD and SPM contents recorded the lowest enhancing, both with an increasing factor of 1.5. Although, with different profiles and distributions in comparison with Oracz and Nebesny (2014) [44], data obtained in our study confirmed the influence of the thermal processing on the increase of BAs concentrations in cocoa beans. This effect has been related to the transformation of free amino acids caused by the treatment at high temperature. In fact, it is now well established that during the Strecker degradation, the thermal decarboxylation of amino acids can occur in the presence of α-dicarbonyl compounds formed during the Maillard reaction or lipid peroxidation [19]. To this regard, literature data confirmed that asparagine, phenylalanine, and histidine changed in the corresponding amines 3-aminopropionamide, PHE, and HIS either in cocoa or in model systems [47].

As far as samples 4nrr3 are concerned, the impact of temperature on BAs concentrations during roasting showed a different behavior. Once again, all the amines exhibited higher quantities at the end of the thermal treatment mainly HIS (increasing factor 3.3) and PUT (increasing factor 2.8), followed by CAD (increasing factor 1.6) and SPM (increasing factor 1.3). To this regard, Hidalgo et al. (2013) [47] reported that the thermal degradation of histidine was more easily produced in comparison with that of phenylalanine. This effect could explain the higher increasing factors of HIS set against with those of PHE, for both samples series 6nrr2 and 4nrr3.

As can be seen in Table 4, before roasting, sample 4nr did not contain TYR and PHE, both appearing respectively only after a thermal treatment at 110 and 125 ◦C, supporting the idea that PHE is generated mainly by thermal decarboxylation of phenylalanine and not by biochemical reactions [44]. Additionally, in the case of TYR, traces of this compound, absent in green coffee (Rio quality), were detected in roasted samples after 16 min at 220 ◦C [48], thus demonstrating its "thermogenic" formation.

The different situations underlined by data in Table 4 probably depend on the complexity of the heat-induced formation of BAs. In fact, amines and amino acid-derived Strecker aldehydes, are simultaneously produced in food products during roasting, due to parallel pathways through the same key intermediates. Reactive carbonyl compounds started these degradations and the ratio between both aldehydes and amines generated is related to the carbonyl compound involved in the reaction and the experimental conditions, including amount of oxygen, pH, temperature, time, as well as the presence of other compounds such as antioxidants or amino acids [19]. In particular, additional amino acids were shown to play an important role in the preferential formation of either Strecker aldehydes or amino acid-derived amines by amino acid degradation in the presence of reactive carbonyl compounds. In this sense, the formation of PHE and phenylacetaldehyde in mixtures of phenylalanine, a lipid oxidation product, and a second amino acid was studied to determine the role of the second amino acid in the degradation of phenylalanine produced by lipid-derived reactive carbonyls. The presence of the second amino acid usually increased the formation of the amine and reduced the formation of the Strecker aldehyde to a differ extent depending on the considered amino acid [19]. The reasons for this behavior are not fully understood, although the obtained results suggested that they seem to be related to the other functional groups (mainly amino or similar groups) present in the side chains of the amino acid. To this regard, the limited aldehydes concentrations, especially at 125 ◦C (Table 3), could support this hypothesis.

Finally, the effect of antioxidants on BAs formation during roasting should be also considered. To this regard, the effect of the presence of phenolic compounds [49] on the degradation of phenylalanine, initiated by lipid-derived carbonyls was studied, to determine the structure-activity relationship of phenolics on the protection of amino compounds against modifications produced by carbonyl compounds. The obtained results showed that, among the different phenolic compounds assayed, the most efficient phenolic compounds were flavan-3-ols followed by single m-diphenols. The efficiency of these molecules was dependent on their ability to rapidly trap the carbonyl compounds. In this way the reaction of the carbonyl compound with the amino acid was avoided. This implies that the carbonyl-phenol reactions involving lipid-derived reactive carbonyls can be produced more rapidly than carbonyl-amine reactions, supporting the idea that antioxidants can provide a protection of amino compounds during thermal treatments of cocoa beans. In this sense, the loss of flavan-3-ols as the roasting temperature increased (Table 2) might be responsible of the limited BAs accumulation in the roasted cocoa beans.

#### **4. Conclusions**

Many classes of compounds present in cocoa nibs can be evaluated as indicators of quality and safety of raw materials and consequently of the final products.

In particular, along with a high [epicatechin]/[catechin] ratio, indicating a better bioavailability of flavanols, a high content on polyphenols could be considered as a favorable attribute of cocoa beans. This is related either to the health qualities of these compounds or to their capacity of preserving other compounds from chemical oxidation or enzymatic degradation, thus increasing stability and general characteristics of the product. According with the obtained results and taking into consideration that cocoa beans used for producing chocolate are usually roasted, the sample 6nr (O'Payo 2016) appears to be the one with the best quality, showing a good content in polyphenols also after roasting at 110 ◦C (12.96 mg g−<sup>1</sup> raw sample vs. 3.04 mg g−<sup>1</sup> roasted sample. On the contrary, although showing the highest [epicatechin]:[catechin] ratio (27.9), the sample 2nr seems to possess a lower quality among considered samples, in relation to its low polyphenols content (1.44 mg g<sup>−</sup>1).

The monitoring of the volatile aromatic fraction, as reported for the not volatile one, suggested the same conclusions. Sample 2nr, showed lower quality having high levels of acids that influence, in a negative way, the final aroma of chocolate. Besides, among the analyzed raw samples, low level of acetic acid and the highest levels of the aldehydes with chocolate and almond aroma confirmed the high quality of sample 6nr (O'Payo 2016), as described from polyphenols analysis. The analysis of roasted beans confirmed changes in VOCs during roasting, as the decrease of acetic acid, especially in sample 6, and the increase of pyrazines associated with the nutty, cocoa, peanut-like aroma. The roasting temperature at 125 ◦C seemed to cause a loss of some minor compounds involved in the aroma of chocolate such as alcohols, aldehydes, and esters, resulting therefore excessive for the tested variety.

Considering BAs as cocoa quality markers as well, their total levels seem to indicate an opposite trend in comparison to that underlined by polyphenols and aroma compounds. In fact, among raw cocoa nibs, sample 6nr showed the second higher BAs total concentration, indicating a medium quality among considered samples. However, it should be underlined that, after roasting at 110 ◦C, amine total amounts showed an increasing factor of 2.22 (6nr vs. 6r2) and of 2.37 (4nr vs. 4r2) implying, among the analyzed samples, a lower attitude of sample 6nr to form amines during heat treatment. Anyway, from the food safety point of view, not alarming BAs amounts were found in all samples, both raw and roasted. All BAs concentrations increased after roasting, although to a different extent depending on the sample and on the considered amine. The latter aspect supports the idea that heat induced amines formation/accumulation probably during the Strecker degradation where aldehydes and amines compete to be formed, and at the same time BAs accumulation was lowered by the polyphenols intervention.

**Author Contributions:** Conceptualization, A.R. and D.R.; Methodology, U.G.S., F.I., M.C., D.R. and A.R.; Software, U.G.S., F.I. and M.C.; Validation, U.G.S., F.I. and M.C.; Formal Analysis, U.G.S., F.I., M.C. and D.P.; Investigation, U.G.S., F.I., M.C. and D.P.; Resources, U.G.S., F.I., M.C., D.R. and A.R.; Data Curation, U.G.S., F.I., M.C. and D.P.; Writing—Original Draft Preparation, U.G.S., F.I., M.C., D.R. and A.R.; Writing—Review and Editing, U.G.S., F.I., M.C., D.R. and A.R.; Visualization, U.G.S., F.I. and M.C.; Supervision, A.R., D.R.; Project Administration, A.R.; Funding Acquisition, A.R.

**Funding:** This research received no external funding

**Acknowledgments:** We thank Giorgio Sergio of Meraviglie S.r.l., Sommacampagna (VR), Italy for supplying us with cocoa beans and technical support and we thank Sixtus (BANDO A—POR CREO FESR 2014–2020. Linea 1.1.2) and project BIOSINOL—PSR 2014–2200 for financial support.

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

#### **References**


© 2019 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 (http://creativecommons.org/licenses/by/4.0/).
