Unravelling the relation between natural microbiota and biogenic amines in Brazilian dry-cured loin: a chemometric approach

Abstract

Exploratory data analysis was used to evaluate the influence of NaCl concentration on the relation between endogenous microbiota (mesophilic aerobic bacteria, psychrotrophic bacteria, lactic acid bacteria, Enterobacteriaceae and Staphylococcus spp.) and biogenic amines (BAs) (histamine, tyramine, putrescine, cadaverine and spermidine) content in Brazilian dry-cured loin (BDL). Using hierarchical cluster analysis (HCA), initial data analysis led to samples separation into a higher and a lower NaCl cluster. The analysis of variance showed that the microbial counts did not differ between the clusters (p > 0.05). However, the higher NaCl cluster showed a lower level of BAs. Further, principal component analysis (PCA) demonstrated a negative correlation between the microbial counts and BAs content in the lower NaCl cluster, thus indicating the possibility of the higher BA’s content in the lower NaCl cluster being a result of a stress–response mechanism. On the other hand, in the higher NaCl cluster, the salt concentration had an inhibitory property in BA’s formation except for histamine. The collective results point to a NaCl threshold to minimise the production of BAs in BDL. The findings of the present exploratory study highlight the possibility of intervention for BA minimization without the need for designed starter cultures or preservation treatments.

Graphical Abstract

Open in new tabDownload slide

Introduction

Dry-cured meat products are popular due to their attractive and distinct sensorial characteristics (Morales et al., 2013). Although considered stable foods due to the curing process, studies have shown that dry-cured meats can harbour microbiological hazards (Mutz, et al., 2019; Mutz, et al., 2019). Inherently to its microbial diversity, potentially toxic metabolic compounds can be present in the final product, among which biogenic amines (BAs) stand out (Jairath et al., 2015; Ruiz-Capillas & Herrero, 2019).

BAs are organic nitrogenous bases of low molecular weight and biological activity (Vasconcelos et al., 2021). BAs present in food are classified as exogenic amines, which are anti-nutritional factors formed due to the decarboxylation of free amino acids by positive-amino acid decarboxylase bacteria (Jairath et al., 2015). In addition, proteolytic activities by contaminating microorganisms increase the content of free amino acids in foodstuffs (Rodríguez et al., 1998; Sun et al., 2016). Thus, BAs are present in numerous dry-cured and fermented meats due to the growth of fermentative bacteria such as lactic acid bacteria and spoilage bacteria such as Enterobacteriaceae and Pseudomonas spp. (Landeta et al., 2013; Li et al., 2019; Sang et al., 2020; Rosario et al., 2021). Among BAs, tyramine, histamine, cadaverine, putrescine and spermidine can be highlighted as relevant for being commonly described in dry-cured meats (Ruiz-Capillas & Herrero, 2019; Vasconcelos et al., 2021). BAs are a subject of interest in food science due to their potential for food poisoning (Ucar et al., 2021; Vasconcelos et al., 2021).

Histamine and tyramine are highlighted as the most toxic BAs in foodstuff (Jairath et al., 2015). Common symptoms of intoxication by these amines include migraines, headaches, allergic reactions, skin irritation and increased blood pressure (Wójcik et al., 2021). Moreover, other BAs, such as spermidine, are also related to food allergies (Kalač, 2014; Ruiz-Capillas & Herrero, 2019). On the other hand, putrescine and cadaverine are both associated with aggravating the effects of more toxic amines, besides being proposed as a quality index for fresh meat products (Guerrero-Legarreta et al., 1991; Vinci & Antonelli, 2002). However, such an index is not applicable for fermented and cured meats as those are commonly present in high concentrations in these products (Stratton et al., 1991; Ruiz-Capillas & Jiménez-Colmenero, 2005).

As potentially toxic subproducts BAs pose a risk for consumers’ health, and their control has been studied under many approaches, such as the use of preservation treatments (Kim et al., 2005; Ruiz-Capillas et al., 2007; Naila et al., 2010), selection of starter cultures (Komprda et al., 2001; Sun et al., 2016; Van Ba et al., 2016) and use of preservatives (Ucar et al., 2021). Alternatively, the study of the physicochemical characteristics of cured meats as affected by its process conditions and formulation can give insights into possible interventions to prevent BAs formation. This aspect is highly relevant considering the trend for healthier food, including the protected design of origin (PDO) products, such as the Brazilian dry-cured loin (BDL, Socol). BDL is a cured loin that goes through a spontaneous fermentative process produced by small-scale and family-sized enterprises, without adding preservatives, such as nitrite and nitrate, and possessing low implementation potential for preservation technologies.

Dry-cured meats, as BDL, exhibit a distinct physicochemical profile when compared with fresh meats. More specifically, their a_w, pH, acidity and NaCl content are different due to the technological processing involved in dry-curing (Mutz, et al., 2019). Our previous study with this cured meat found correlations between its physicochemical characteristics and oxidative state, where a_w and NaCl content were highly important (Rosario et al., 2020). In addition, these characteristics are directly involved in the microbiota present in the meat product and their BAs production, as evidence suggests that they affect the activity of amino acid decarboxylases (Bozkurt et al., 2004; Bubelová et al., 2015). Therefore, the present study uses chemometrics to unravel the intrinsic correlations between the BDL microbiota and BAs content as influenced by the cured-meat physicochemical characteristics.

Material and Methods

Dry-cured loin samples

A total of 21 BDL (Socol) samples, comprising three distinct production batches, were acquired from the same producer in Venda Nova do Imigrante city, Espírito Santo state, Brazil (20°19’31.9 S 41°07’56.6 W). The samples were produced during August–October 2017 (three months). The pigs (F1) were a crossbreeding between the “Landrace” and “Large White” breeds, fed corn and soybean meal (main base), and slaughtered when the weight between 90 and 100 kg was reached. The pork loin (Longissimus dorsi) was dry-cured with NaCl and 3g kg⁻¹ of a mixture (1:1) of black pepper (Piper nigrum) and garlic (Allium sativum). The ripening period was conducted for three months at ambient temperature (between 10 °C and 25 °C) and relative humidity (82.5%–84%) (Rosario et al., 2020). After the maturation, samples were vacuum packed and transported in sanitized boxes to the laboratory (25 °C) for analysis. The final BDL was vacuum-packed in individual units of approximately 200 g and transported to the laboratory. The loins were then sliced into pieces of 1.0 mm thick and on average 23.5 ± 0.9 cm² superficial area.

Microbiological analysis

The procedures were performed according to the American Public Health Association methodology described in the Compendium of Methods for the Microbiological Examination of Foods (Downes & Ito, 2001). First, 10 g of BDL slices were homogenized in 90 mL of distilled water with 0.85% sodium chloride. Homogenization was performed in a stomacher (Yka Tecnologia, Rio Grande do Sul, Brazil) for 2 min. Cultivation was performed with a spiral plater (Eddy Jet 2, IUL Instruments, United States) using mode E50. A culture medium specific for each microbial group was used. Mesophilic aerobic bacteria were plated on plate count agar (HiMedia^®, Mumbai, India) and incubated at 35 °C for 48 h. Psychrotrophic bacteria were plated on plate count agar and incubated under aerobic conditions at 7 °C for 10 days. Molds and yeasts were plated on Potato Dextrose Agar (HiMedia^®, Mumbai, India) and incubated under aerobic conditions at 25 °C for 7 days. Lactic Acid Bacteria (LAB) were poured plated on De Man, Rogosa, and Sharpe Agar (Merck^®, Darmstadt, Germany), covered with an additional agar overlay and incubated at 35 °C for 72 h under anaerobiosis. Halophilic bacteria were plated on plate count agar with 5% of NaCl under aerobic conditions at 35 °C for 48 h (Dinamica^®, Rio de Janeiro, Brazil). Enterobacteriaceae were plated on Violet-Red-Bile-Glucose agar (VRBG) (Merck^®, Darmstadt, Germany) and incubated under aerobic conditions at 35 °C for 48h. Staphylococcus spp. were plated on Baird Parker Agar (HiMedia^®, Mumbai, India) containing egg yolk emulsion and potassium tellurite (Sigma-Aldrich^®, Switzerland) incubated under aerobic conditions at 35 °C for 48 h. Characteristic colonies (grey-black shiny with an opaque zone around the colony) were considered Staphylococcus spp. according to Baird-Parker (1962). The colonies were counted in the electronic counter (Flash & go, IUL Instruments, United States). All results were expressed in a logarithm of colony-form units per gram (log cfu/g).

Physicochemical analysis

Water activity and total chlorides

Water activity was measured by direct reading of the minced sample in Pawkit meter (Decagon Devices, Pullman, WA, USA) according to the manufacturer. Total chlorides were determined by the method of Mohr (AOAC, 2012). Briefly, 5 g of BDL sample were homogenized with 50 mL of distilled water using an Ultra Turrax 18 basic (IKA, Wilmington, NC, USA) at 11,000 RPM. The homogenate was then added of K₂CrO₄ agent and titrated with AgNO₃. The data were expressed in mg of NaCl per 100 g.

pH and total titratable acidity

pH was determined using a digital pH-meter (PHS3BW, Bel engineering, Piracicaba, SP, BRA) with 10 g of sample homogenized in 100 mL of distilled water (AOAC, 2012). Total Titratable Acidity (TTA) was determined by the titrimetric method using NaOH 1 N in 5 g of sample homogenized in 50 mL of distilled water (AOAC, 2012). The results were expressed in mg of lactic acid per 100 mg of BDL.

Biogenic amines

The method of extraction, derivatization and the chromatographic conditions used herein were previously described and validated by our research group (Lázaro et al., 2013; Vieira et al., 2020). The determination coefficients of the calibration curves were ≥0.992, the accuracy and the intermediate precision ranged from 0.09% to 2.43% and 1.15% to 7.24%, respectively. The recovery was between 64.40% and 112.22%. The detection limit ranged from 0.03 to 1.25 µg.L⁻¹, while the quantification limit was between 0.15 and 5 µg.L⁻¹.

Briefly, 5 g of the BDL samples were extracted with 10 mL of perchloric acid (5%) using an Ultra Turrax 18 basic (IKA, Wilmington, NC, USA) at 11,000 rpm for 1.5 min. The extracted sample was neutralized with 2 M sodium hydroxide (reaching pH >12) and derivatised by 50 μL of benzoyl chloride. Next, the mixture was maintained for 40 min in a water bath at 30 °C. After that, the extraction of BAs derivatives with diethyl ether was performed twice, followed by evaporation under N₂ and resuspension with 1.0 mL of the mobile phase.

Histamine, tyramine, putrescine, cadaverine and spermidine determination was carried out by chromatographic analysis using a Shimadzu HPLC system (Shimadzu, Kyoto, Japan) consisting of an LC-20AT pump, SPD-M20A diode array detector, CT20A oven, SIL-20AC autosampler, and CBM-20A controller. The separation of the amines was made using a spherisorb ODS2 C18 column (15 X 0.46 cm I.D, 5 μm particle size), as stationary phase and acetonitrile: water (42:58, v/v) as mobile phase under isocratic condition. The column temperature was 20 °C, and the flow rate was 1 mL min⁻¹ with a total run time of 15 min. The BAs were identified by retention times and spiking the samples with the suspected amine and quantified by interpolating peak area with external standard curves using LC Solution software. Typical chromatograms of the BAs derivatives in acid extract for standard mixture of individual compounds and extract from a commercial BDL sample is shown in supplementary material (Figure S1).

Statistical analysis

All microbiological, physicochemical and BAs analyses were done in three replicates for each BDL sample. First, data on 16 variables of the 21 BDL were standardized to eliminate the order of magnitude effect among the responses. Then, to investigate the 21 BDLs natural grouping, the hierarchical cluster analysis (HCA) was used following Ward's method and Euclidean distance (Granato et al., 2018; Rosario et al., 2020). The HCA technique works using an agglomerative approach where each sample starts as a unique cluster. Then, the Euclidean distance between the samples is calculated, and the linkage criterion between two different samples is set by Ward’s method, which seeks to minimize intra-cluster variance (Brereton et al., 2017). Furthermore, the significance of the variables for the cluster distinction was evaluated with multivariate analysis of variance (MANOVA) at a 0.05 significance level.

To evaluate the specific relationship between BAs, and the BDL physicochemical and microbiological variables in each formed cluster, principal component analysis (PCA) based on a correlation matrix was performed. PCA reduces the dataset dimensionality while retaining its variance in new unrelated principal components (PCs). Further, due to its graphical outputs, the technique provides a very intuitive and visual approach for data analysis, with the plots of scores and loadings readily showing the samples and variable importance to the plot, respectively (Brereton et al., 2017).

Pearson’s correlation (p < 0.05) was employed to assess the overall correlation between the variables. All statistical analyses were performed using Statistica^® 10 software (Statsoft Inc., OK, USA) was used.

Results and Discussion

Overall BDL characteristics

To obtain a clear insight into the underlying connections between the 16 studied variables, clustering analysis of the 21 BDL samples was performed by HCA. The defined threshold separated the samples into two clusters, cluster 1 with 12 samples and cluster 2 with 9 samples (Fig. 1). The clusters were formed according to the studied characteristics of the BDL, where BAs, a_w, and NaCl content were significant physicochemical variables for the cluster’s formation (p < 0.05) (Table 1). NaCl content can be highlighted as an important factor among the physicochemical variables. It is a fundamental ingredient in dry-cured meat production and acts as one of the meat preservers. NaCl acts as a hurdle for microorganism growth in cured products due to the chloride ion and the lowering the a_w of the product (Mutz, et al., 2019). Indeed, NaCl content explains 84.3% (r2 0.843, p < 0.05) of the behaviour of a_w in the product (Figure S2). Therefore, each cluster was named based on their NaCl level (Table 1), a higher NaCl cluster (cluster 1), and a lower NaCl cluster (cluster 2).

As shown in Table 1, the higher NaCl cluster samples presented lower values of a_w, spermidine, cadaverine, putrescine, tyramine and psychrotrophic count. On the other hand, the lower NaCl cluster showed higher values for the mentioned variables (Table 1). In this way, NaCl, as a mandatory ingredient of this product, acts as a controller for BAs formation. Indeed, the use of NaCl inhibits the growth of microorganisms with decarboxylase activity, which in meat are generally Enterobacteriaceae, LAB and Pseudomonas spp. (Suzzi & Gardini, 2003; Vasconcelos et al., 2021). However, in the present study, there was no significant difference in the counts of these microbial groups between the higher and lower NaCl clusters (Table 1). Therefore, the lower BA content in BDL may be related to the impairment of the decarboxylase enzyme activity. Corroborating our finding, Laranjo et al. (2017) also found an increase in the BA content for dry-cured sausages with distinct salt contents without difference in LAB and enterobacteria counts. Indeed, although many microbial groups can produce BAs, the decarboxylase activity depends on the microbial strain and the environment’s physicochemical factors (Bozkurt et al., 2004; Ruiz-Capillas & Herrero, 2019).

Looking individually at the BAs, the only amine that did not statically differ between the clusters was histamine. Histamine and tyramine are the two BAs directly involved in food poisoning cases (Ruiz-Capillas & Jiménez-Colmenero, 2005). Although these amines are known to be present in dry-cured products (Jairath et al., 2015; Ruiz-Capillas & Herrero, 2019), there is not a specific limit set by legislation for their presence in these meat products. Among BAs, histamine is the one with a legislation limit, with a 200 mg kg⁻¹ maximum for fish by the Australian and New Zealand codex and EFSA, 100 mg kg⁻¹ for fish set for the Brazilian Ministry of Agriculture, Livestock and Food Supply, and 50 mg kg⁻¹ for food in general set by the United States FDA (Ruiz-Capillas & Herrero, 2019). Cadaverine and putrescine were the most present amines in both clusters. These amines are not considered toxic, although they enhance the toxicity of histamine and tyramine (Ruiz-Capillas et al., 2007).

The average BAs values for both clusters are in the range described in the literature for cured meats (Ruiz-Capillas & Jiménez-Colmenero, 2005; Jairath et al., 2015). However, care should be taken when considering the food poisoning risks of individual products, as food products are not generally consumed isolated, but in the context of meals (Ruiz-Capillas & Herrero, 2019). Determining the toxicity threshold for BAs is difficult as the toxic effects depend on individuals' detoxifying capability besides the ingested concentration (Benkerroum, 2016). The concentrations of 50 mg kg⁻¹ of histamine and 600 mg kg⁻¹ of tyramine were considered the no-observable-adverse-effect-level (NOAEL) for healthy individuals and used in risk assessment models performed by the European Food Safety Authority (EFSA, 2011). However, individuals with hypertension, coronary and gastrointestinal problems are more sensitive to BAs due to lower activity of their mono, di and polyamine oxidases (the detoxifying enzymes that metabolize BAs) (Jairath et al., 2015). Furthermore, individuals under monoamine oxide inhibitors (MAOI) can be either sensitive or highly sensitive, with the tyramine toxicity threshold decreasing to values of 6–50 mg kg⁻¹ (EFSA, 2011). Therefore, regarding the potential food poisoning risks imposed by the BAs level found in BDL clusters, it is essential to understand the conditions in which they can be minimized.

Correlation between the microorganism and BAs production in different NaCl levels

Each cluster was considered a distinct BDL scenario due to the significant differences in their NaCl levels. Therefore, the higher and lower NaCl clusters were individually evaluated with PCA to understand the distinction in their intrinsic relationships between microorganisms and BAs.

The PCA model built with three components explained 73.1% of the data variation for the lower NaCl cluster. Analysis of the variable contribution to constructing the principal components (loadings) shows that all the microorganisms are represented by either the first or second component (see Table S1). Tyramine and putrescine are oppositely located to mesophilic bacteria, halophilic bacteria, Enterobacteriaceae and LAB in the first component (Fig. 2a). For the second component, cadaverine and spermidine are closely located to Staphylococcus and mould and yeasts counts, respectively (Fig. 2a). Lastly, in the third component, water activity is closely located to histamine (Fig. 2b). Overall, PCA of the low NaCl shows a negative correlation between the microorganisms' counts and BAs (Fig. 2).

Moreover, a Pearson correlation analysis between the microbial groups and BAs was conducted. The correlations were classified according to Evans (1996) in relation to their coefficients as: very weak (0.00–0.19); weak (0.20–0.39); moderate (0.40–0.59); strong (0.60–0.79) and very strong (0.80–1.00). It was found a significant (p < 0.05) negative correlation with the BAs content, where cadaverine strongly negatively correlated to psychrotrophic (−0.66), and moulds and yeasts (−0.66). Furthermore, tyramine was also strongly negatively correlated to psychrotrophic (−0.75) and moulds and yeasts (−0.77). This scenario where the microbial counts are inversely related to the BAs could imply the formation of BAs as a mechanism of stress–response related to the BDL NaCl levels (Gardini et al., 2016; Mutz, et al., 2019). Indeed, reports in the scientific literature describe the upregulation of the decarboxylase enzymes genes as part of the protection mechanisms against osmotic and acid stresses (Rossi et al., 2011; Alvarez-Ordóñez et al., 2015; Gardini et al., 2016).

The first three components of the higher NaCl cluster explained 66.5% of the total data variance. By the loadings, it can be seen that the first component is mainly composed of spermidine, tyramine, mesophilic, halophilic and Staphylococcus counts (Table S2). At the same time, the second component is composed of histamine, putrescine and Enterobacteriaceae. Furthermore, the third component is composed of cadaverine and psychrotrophic counts. By looking into the score plot made from the first two components (Fig. 3a), it is possible to observe that, in the higher NaCl cluster, in general, the microbial counts are closely located to the BAs.

In contrast to the lower NaCl environment, putrescine is closely located to psychrotrophics, and moulds and yeasts, both represented by the second component (Fig. 3a). Meanwhile, tyramine is close and positively related to halophilic counts represented by the first component (Fig. 3a). Further, spermidine is also closely located to Staphylococcus and represented by the first component (Fig. 3a). The Pearson correlation analysis corroborated the PCA findings. Strong significant (p < 0.05) correlations were found in this cluster for histamine and Staphylococcus (0.60); spermidine with mesophilic (0.77), halophilic (0.67) and Staphylococcus (0.74); putrescine with psychrotrophic (0.64) and tyramine with mesophilic (0.60) and halophilic (0.69). Besides presenting a positive correlation between the microorganism count and the BAs level, the higher NaCl cluster presented a lower BAs content. Therefore, this can indicate that the BAs concentration in this cluster of samples are related to the growth of microorganisms capable of surviving this level of NaCl.

Moreover, as the effect of NaCl in BAs production is highly dependent on the species and strain, it is possible that (i) the bacteria capable of growth in a higher NaCl environment are less able to produce BAs or (ii) in the higher NaCl cluster, the NaCl exerted an inhibitory effect on the BAs production, as opposed to the impact in the lower NaCl cluster. The effect of different NaCl levels varies and can either inhibit the decarboxylase enzyme or increase the formation of BAs (Gardini et al., 2016). In addition, it was shown that bacteria possess distinct decarboxylase enzymes, an inducible that increase its activity as the NaCl level increases, while a constitutive enzyme decreases its activity Morii & Kasama, 2004). However, in the literature, there is no defined threshold or range describing the impact of distinct NaCl levels in the BAs content of meat products. Therefore, regardless of the exact mechanism, it is plausible to infer that the study of NaCl concentration on dry-cured meats can lead to an efficient inhibition strategy for BAs formation.

Many studies focus on finding suitable starter cultures to produce cured meats, aiming inhibition of pathogens and decreasing BAs production (Lu et al., 2010; Sun et al., 2016; Van Ba et al., 2016). Alternatively, the use of preservatives or preservation technologies has been approached to increase cured meats' safety (Simon-Sarkadi et al., 2012; Mozuriene et al., 2016). Somehow, for products that rely on natural microbiota for fermentation and maturation, as BDL (Rosario et al., 2020), the use study of the NaCl concentration on the final product can lead to an option for a safer product without the need of adding preservatives or changing the formulation of the product. Therefore, a more comprehensive study on the threshold of NaCl used in the products to balance microorganism growth and toxic BAs formation is desirable to grant a safe and healthy meat product.

Conclusion

According to HCA, NaCl content plays a significant role in the microbiological and physicochemical characteristics of BDL. Samples from the lower NaCl cluster presented higher BAs content, characterized by a high level of cadaverine. Further investigation by PCA indicates the possibility of a stress–response mechanism being responsible for the elevated concentration of BAs in the lower NaCl samples. On the other hand, in the higher NaCl cluster, the lower formation of BAs has related to the endogenous microbial counts in the BDL and the likely inhibitory effect of NaCl on the decarboxylase enzyme. Therefore, enlightening the possibility for new studies on the NaCl threshold for improved health and safety of artisanal cured meats that do not use starter cultures.

Acknowledgments

The authors are thankful for the financial support provided by the Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) Brazil – grants [E-26/200.891/2021] and [E-26/200.691/2021], the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) – grant number [313119/2020-1], and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) Brazil – Finance Code001

Ethical statement

Ethics approval was not required for this research.

Author contribution

Yhan Mutz: Conceptualization (supporting); Data curation (lead); Formal analysis (lead); Investigation (equal); Methodology (equal); Writing – original draft (lead); Writing – review & editing (equal). Denes Kaic Alves Rosario: Conceptualization (lead); Data curation (equal); Formal analysis (equal); Methodology (lead); Supervision (lead); Validation (equal); Writing – original draft (equal); Writing – review & editing (equal). Yago Alves de Aguiar Bernardo: Data curation (equal); Formal analysis (equal); Writing – review & editing (equal). Carla Paulo Vieira: Data curation (equal); Formal analysis (equal); Writing – review & editing (equal). Rodrigo Vilela Pinto Moreira: Data curation (supporting); Formal analysis (supporting). Patricia Bernardes: Conceptualization (equal); Project administration (equal); Supervision (equal); Writing – review & editing (equal). Carlos Conte-Junior: Funding acquisition (equal); Project administration (equal); Resources (equal); Supervision (equal); Visualization (equal); Writing – review & editing (equal).

Data availability statment

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References