*Article* **Hydrolysate from Mussel** *Mytilus galloprovincialis* **Meat: Enzymatic Hydrolysis, Optimization and Bioactive Properties**

**Sara A. Cunha, Rita de Castro, Ezequiel R. Coscueta and Manuela Pintado \***

CBQF—Centro de Biotecnologia e Química Fina—Laboratório Associado, Escola Superior de Biotecnologia, Universidade Católica Portuguesa Rua Diogo Botelho 1327, 4169-005 Porto, Portugal; scunha@ucp.pt (S.A.C.); rita\_de\_castro@live.com.pt (R.d.C.); ecoscueta@ucp.pt (E.R.C.)

**\*** Correspondence: mpintado@ucp.pt

**Abstract:** Mussel production generates losses and waste since their commercialisation must be aligned with target market criteria. Since mussels are rich in proteins, their meat can be explored as a source of bioactive hydrolysates. Thus, the main objective of this study was to establish the optimal production conditions through two Box–Behnken designs to produce, by enzymatic hydrolysis (using subtilisin and corolase), hydrolysates rich in proteins and with bioactive properties. The factorial design allowed for the evaluation of the effects of three factors (hydrolysis temperature, enzyme ratio, and hydrolysis time) on protein/peptides release as well as antioxidant and anti-hypertensive properties of the hydrolysates. The hydrolysates produced using the optimised conditions using the subtilisin protease showed 45.0 ± 0.38% of protein, antioxidant activity via ORAC method of 485.63 ± 60.65 µmol TE/g of hydrolysate, and an IC<sup>50</sup> for the inhibition of ACE of 1.0 ± 0.56 mg of protein/mL. The hydrolysates produced using corolase showed 46.35 ± 1.12% of protein, antioxidant activity of 389.48 ± 0.21 µmol TE/g of hydrolysate, and an IC<sup>50</sup> for the inhibition of ACE of 3.7 ± 0.33 mg of protein/mL. Mussel meat losses and waste can be used as a source of hydrolysates rich in peptides with relevant bioactive properties, and showing potential for use as ingredients in different industries, such as food and cosmetics, contributing to a circular economy and reducing world waste.

**Keywords:** antioxidant; anti-hypertensive; proteins; sustainability; marine species; marine hydrolysates

#### **1. Introduction**

Mussels are highly consumed in several countries. Asia and Europe are considered the leading producers, estimated to produce about 1.05 and 0.5 million tonnes of mussel per year, respectively [1,2]. Mussel consumption has several advantages, for both the environment and consumers. Environmentally, mussel farming can be done with minimal greenhouse gas emissions, and thus low carbon footprint and few environmental impacts [3]. Mussels farming produces about 0.6 kg of CO<sup>2</sup> emission/kg edible product, while beef produces about 19.0–36.7 kg of CO<sup>2</sup> emission/kg edible product [3]. For consumers, mussel meat has low fat and low calories. Still, more importantly, mussels are a rich source of sodium, selenium, vitamin B twelve, zinc [1], and an interesting source of proteins since they are composed of about 58.7% of protein on a dry weight basis [4]. Due to their protein-rich meat, mussels have been described as a source of bioactive peptides with relevant biological properties. Bioactive peptides are fragments that are inert when inside proteins but show different properties when broken from the original protein [5]. Thus, enzymatic hydrolysis with proteases seems an interesting approach for obtaining bioactive extracts since these enzymes may break mussel proteins into smaller peptides, which may be associated with other biological and functional properties [6]. Different enzymes have been used to produce bioactive peptides from mussels, such as pepsin [7], flavourzyme [8], papain [8], and trypsin [9]. Marine species have often been described

**Citation:** Cunha, S.A.; de Castro, R.; Coscueta, E.R.; Pintado, M. Hydrolysate from Mussel *Mytilus galloprovincialis* Meat: Enzymatic Hydrolysis, Optimization and Bioactive Properties. *Molecules* **2021**, *26*, 5228. https://doi.org/10.3390/ molecules26175228

Academic Editor: Jesus Simal-Gandara

Received: 2 August 2021 Accepted: 25 August 2021 Published: 28 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

as a source of bioactive peptides, and some bioactivities have been associated to mussel peptides, such as antioxidant [10], anti-hypertensive [11], antimicrobial [12], anticancer [7], anti-inflammatory [13], anticoagulant [14], antidiabetic [15], and antiviral [16]. As far as we know, *Mytilus galloprovincialis* bioactive extracts are not so extensively explored, with the main studies being developed with *Mytilus coruscus* and *Mytilus edulis*. The farming of *M. galloprovincialis*, also known as the Mediterranean mussel, has mainly been developing along the Spanish Atlantic coast and in the Mediterranean area [2].

Mussel commercialisation generates losses and waste since they are submitted to a pre-selection before being delivered for sale, resulting in the rejection of broken mussels or those which fail to meet established criteria in the target market [17–20]. It is estimated that about 27% of produced mussels are discarded [4,20]. Thus, mussel meat waste can be used to produce bioactive hydrolysates with interesting properties for food, cosmetic, pharmaceutical, and nutraceutical industrial applications.

In this work, *Mytilus galloprovincialis* meat was submitted to different conditions according to two factorial designs to produce hydrolysates with a maximum level of soluble rich proteins/peptides and bioactive properties, more specifically antioxidant and anti-hypertensive. The hydrolysates were produced by enzymatic hydrolysis using two different enzymes, subtilisin and corolase. Therefore, this study aims to valorise mussel meat by creating bioactive hydrolysates with potential for various industries.

#### **2. Results**

#### *2.1. Mytilus Galloprovincialis Characterisation*

The mussel's meat was received under refrigeration on the day of capture. It was characterised before being minced according to a few nutritional characteristics, showing a protein content of 70.50 ± 13.44%, 90.30 ± 4.24% moisture, and 5.00 ± 0.00% lipids.

#### *2.2. Optimisation of the Production of Hydrolysates Rich in Proteins and Bioactive Properties*

Enzymatic hydrolysis is one of the main used methods to produce bioactive extracts, and is described for the mussel species *M. coruscus*, *M. edulis*, and *M. galloprovincialis*. Several enzymes have been used in the mussel species, such as papain [8], flavourzyme [21], and the digestive enzymes pepsin [7] and trypsin [9]. In this work, two different proteases, subtilisin and corolase, were used to produce hydrolysates rich in proteins/peptides and with antioxidant and anti-hypertensive properties. To understand the conditions that allow to achieve the production of hydrolysates with a better protein% and higher bioactive properties, two experimental designs were made, one for each protease. Different combinations of factors in an enzymatic hydrolysis may lead to differing effectiveness. Thus, we have used a factorial design with 15 combinations of the enzyme (%), temperature ( ◦C), and hydrolysis time (h) for each protease.

Mussel meat was initially minced until homogenised, thus creating uniformed biomass used for all the 60 hydrolysis reactions performed. Then, all the hydrolysis were performed using ultrapure water as the solvent, at 7.5 pH, with a ratio of 1:2 (*w*:*v*) (mussel biomass:water). The 60 hydrolysis reactions were performed using the factors combinations matrix generated by the experimental design, and protein/peptides and bioactivities were measured in the resulting supernatants.

In an enzymatic reaction, different factors combinations may lead to the production of extracts with different characteristics. Thus, the factorial designs allowed us to understand the best combination for optimising protein/peptide bioactive extract production. For both designs, the matrix and obtained results are presented as well as the Pareto charts obtained for each evaluated response, indicating the factors with the greatest influence for each variable studied. For each evaluated response, a multiple regression analysis of the experimental data allowed to obtain a model that can predict the responses and these are shown as Equations (1)–(6). An analysis of variance (ANOVA) was performed to evaluate the significance of each effect and to determine the factors that significantly affected protein % as well as antioxidant and anti-hypertensive properties.

**Run** 

#### 2.2.1. Experimental Design with Subtilisin Protease trates the response surface graphics obtained for the independent variables tested, showing their interactions when studying each dependent variable.

2.2.1. Experimental Design with Subtilisin Protease

*Molecules* **2021**, *26*, x FOR PEER REVIEW 3 of 18

tein % as well as antioxidant and anti-hypertensive properties.

Table 1 shows the experimental design matrix and the results obtained for the Box– Behnken factorial design performed with the subtilisin protease. Figure 1 shows the Pareto charts obtained for the design performed with the subtilisin protease. Figure 2 illustrates the response surface graphics obtained for the independent variables tested, showing their interactions when studying each dependent variable. **Table 1.** Box-Behnken factorial design matrix for three factors and three responses obtained for the subtilisin protease. **Factors Response 1 % Enzyme Hydrolysis Temperature Hydrolysis Protein Content ORAC (µmol ACE Inhibition** 

Table 1 shows the experimental design matrix and the results obtained for the Box– Behnken factorial design performed with the subtilisin protease. Figure 1 shows the Pareto charts obtained for the design performed with the subtilisin protease. Figure 2 illus-

shown as Equations (1)–(6). An analysis of variance (ANOVA) was performed to evaluate the significance of each effect and to determine the factors that significantly affected pro-

**Table 1.** Box-Behnken factorial design matrix for three factors and three responses obtained for the subtilisin protease. **(°C) Time (h) (%) TE/mg) (%)** 


<sup>1</sup> Values expressed as mean <sup>±</sup> SD of two replicates. 1 Values expressed as mean ± SD of two replicates.

**Figure 1.** Pareto charts with the effect of three experimental factors, in decreasing order, obtained for protein % (**A**), antioxidant (**B**) and anti-hypertensive (**C**) properties in the experimental design with subtilisin, showing the most influent factors. The vertical lines in the pareto charts represent the level of significance (*p* = 0.05). XA—Enzyme %; XB—Hydrolysis temperature; XC—Hydrolysis time. **Figure 1.** Pareto charts with the effect of three experimental factors, in decreasing order, obtained for protein % (**A**), antioxidant (**B**) and anti-hypertensive (**C**) properties in the experimental design with subtilisin, showing the most influent factors. The vertical lines in the pareto charts represent the level of significance (*p* = 0.05). XA—Enzyme %; XB—Hydrolysis temperature; XC—Hydrolysis time.

Protein content did not show significant variations among all the combinations tested since the range of variation for results was 45.05–49.12% of protein in the final hydrolysate. The Pareto chart (Figure 1A) and the ANOVA analysis (Table 2) show that protein release was not influenced by the linear factors, but only by the quadratic effect of the temperature. The temperature quadratic coefficient showed a negative effect, indicating an increase in protein % at intermediate values. Since none of the factors had a significant effect, the model was adjusted to best fit. Thus, only the linear effect of enzyme% and time, as well as the linear interaction between enzyme% and temperature and the quadratic effect of temperature were considered (Table 2). By multiple regression, the predicted response for the protein % could be expressed by the model in Equation (1).

Protein % = −13.1089 + 10.2225 × X<sup>A</sup> + 0.635 × X<sup>C</sup> − 0.22625 × XAX<sup>B</sup> − 0.0200714 × X<sup>B</sup> 2 (1)

**Figure 2.** Response surface graphs corresponding to the combined effect of enzyme % (XA), hydrolysis temperature (XB) and hydrolysis time (XC), on the protein (**A**–**C**), ORAC (**D**–**F**) and iACE (**G**–**I**) responses obtained for the mussel extracts produced with subtilisin. **Figure 2.** Response surface graphs corresponding to the combined effect of enzyme % (XA), hydrolysis temperature (XB) and hydrolysis time (XC), on the protein (**A**–**C**), ORAC (**D**–**F**) and iACE (**G**–**I**) responses obtained for the mussel extracts produced with subtilisin.



The final adjusted model showed a significant fit (*p* > 0.050). However, the R2 = 38.52%, indicating that only 38.52% of the variability observed may be explained by the model. This indicates that, for obtaining hydrolysates rich in proteins/peptides, the minimum studied variables could be used (0.5% enzyme, 40 °C, 1h) to achieve protein release superior to 40%. Although, since this work aims to produce hydrolysates with bioactive The final adjusted model showed a significant fit (*p* > 0.050). However, the R<sup>2</sup> = 38.52%, indicating that only 38.52% of the variability observed may be explained by the model. This indicates that, for obtaining hydrolysates rich in proteins/peptides, the minimum studied variables could be used (0.5% enzyme, 40 ◦C, 1 h) to achieve protein release superior to 40%. Although, since this work aims to produce hydrolysates with bioactive properties, the antioxidant and anti-hypertensive properties must be analysed.

properties, the antioxidant and anti-hypertensive properties must be analysed. **Table 2.** Analysis of variance (ANOVA) for protein % obtained for the subtilisin Box-Behken design. **Model Sum of Squares DF Mean Square F Value** *p***-Value**  XA (Enzyme %) 4.7524 1 4.7524 1.80 0.1955 XC (Time) 6.4516 1 6.4516 2.44 0.1345 XAXB 10.2378 1 10.2378 3.88 0.0637 XB2 30.0804 1 30.0804 11.39 0.0032 The antioxidant property, measured in the soluble hydrolysates by ORAC, was only significantly influenced by the linear effect of the hydrolysis temperature (XB) and enzyme % (XA) (*p* < 0.050), with hydrolysis time (XC) not showing a significant effect on this property (Figure 1B). Temperature and enzyme % positively affected antioxidant property, meaning that the response is directly proportional to the tested values. For ORAC, the interactions between variables did not show a significant effect, the same being verified for the quadratic effect of the three studied variables. By multiple regression, the predicted response for the ORAC could be obtained by the model in Equation (2).

R2 = 47.00, Adj − R2 = 38.52, CV = 1.62

$$\text{ORAC} = -0.0333894 - 0.0436106 \times \text{X}\_{\text{A}} + 0.0206284 \times \text{X}\_{\text{B}} \tag{2}$$

× XB <sup>2</sup> (1)

Total 109.624 29

Analysing the ANOVA results (Table 3), the final adjusted model showed a significant fit (*p* > 0.050). However, the R<sup>2</sup> = 45.65%, indicating that only about 45.65% of the variability in the antioxidant activity may be explained by the model.

**Table 3.** Analysis of variance (ANOVA) for ORAC results obtained for the subtilisin Box-Behken design.


The ACE inhibition % was evaluated using a concentration of 10 mg hydrolysate/mL. The linear effects of the three studied factors (XA, X<sup>B</sup> and XC) were significant for the variations observed on the ACE inhibition %, as well as the linear interaction between all the variables (XAXB, XAXC, and XBXC) and the quadratic effect of the hydrolysis time (X<sup>C</sup> 2 ) (Figure 1C). Thus, seven effects showed a *p* < 0.050, while the other two effects (X<sup>A</sup> 2 and X<sup>B</sup> 2 ) were not significant and consequently removed from the model. Temperature and enzyme % showed a high impact on the response. The linear effect of hydrolysis time (XC) and the linear interaction between temperature and enzyme % showed a negative contribution, which means that there is an increase of the ACE inhibition at intermediate values. Thus, the longer the reaction, the lower the inhibition of ACE will be, which can mean that the higher extension of the hydrolysis leads to the formation of peptides with less activity against ACE. The quadratic effect of hydrolysis time (X<sup>C</sup> 2 ) showed a positive effect on the response, which means that, considering the negative effect verified for XC, higher values of iACE were achieved near the minimum values studied. The other linear factors with a significant contribution positively affected the results, meaning that the anti-hypertensive potential is enhanced by the increase in enzyme % and by the higher temperatures.

By multiple regression, the predicted response for the iACE could be obtained by the model in Equation (3).

$$\begin{aligned} \text{iACE} &= -24.3485 + 38.3047 \times \text{X}\_{\text{A}} + 3.41606 \times \text{X}\_{\text{B}} - 45.3903 \times \text{X}\_{\text{C}} - 1.13715 \times \text{X}\_{\text{A}} \times \text{X}\_{\text{B}} + 6.3675 \\ &\times \text{X}\_{\text{A}} \times \text{X}\_{\text{C}} - 0.0200471 \times \text{X}\_{\text{B}} \,^2 + 0.2635 \times \text{X}\_{\text{B}} \times \text{X}\_{\text{C}} + 5.83804 \times \text{X}\_{\text{C}} \end{aligned} \tag{5}$$

Analysing the ANOVA results (Table 4), the final adjusted model showed a significant fit (*p* > 0.050) and an R<sup>2</sup> = 81.30%, indicating the variability observed in terms of ACE inhibition is highly explained by the model.



After analysing the results for each variable, we intended to maximise the hydrolysis in order to achieve higher protein content and antioxidant and anti-hypertensive properties. For that, a Derringer's desirability analysis was performed [22] (Table 5). The optimum conditions predicted were 52.5 ◦C, 1.5% of subtilisin, and an hydrolysis time of 3 h (Table 5).

**Table 5.** Optimal conditions predicted by the experimental design to maximise protein/peptide release and antioxidant and anti-hypertensive activities of the hydrolysates, and Derringer desirability to predict the optimal conditions for a multiple response.


The experiments were validated in triplicate, using the same biomass quantities and solvent volumes but adapting temperature to 52 ◦C, to work practically. The obtained protein content, ORAC, and ACE inhibition values were 46.70 ± 0.36%, 0.54 ± 0.029 µmol TE/mg hydrolysate and 70.21 ± 2.9%, respectively (Table 6). When comparing the results predicted by the factorial design, the obtained results were similar to those predicted by the design.

**Table 6.** Results predicted by the model, and results obtained in a validation and a scaled-up enzymatic hydrolysis, performed with the optimal conditions described in Table 5.


After optimising the enzymatic hydrolysis reaction, a scale-up was performed, in triplicate, increasing the amount of mussel biomass and solvent by 15 times and maintaining the tested ratio. The temperature was adjusted to 50 ◦C, to be easily adapted to an industrial scale. The obtained protein content, ORAC, and IC<sup>50</sup> for ACE inhibition values were 45.0 ± 0.38, 0.49 ± 0.061 µmol TE/mg hydrolysate, and 1.0 ± 0.56 mg protein/mL, respectively (Table 6). The obtained results indicated that the increase in the proportions seems to influence the evaluated responses negatively. Nevertheless, the obtained hydrolysates showed interesting protein/peptide values and antioxidant potential, making these extracts an interesting protein source and a potential ingredient for functional food or cosmetic formulations focused on anti-ageing properties.

#### 2.2.2. Experimental Design with the Corolase Protease

The experimental design matrix and the responses obtained for the Box–Behnken factorial design performed with the corolase protease are shown in Table 7. Figure 3 shows the Pareto charts, and Figure 4 shows the response surface graphics obtained for the independent variables tested, showing their interactions when studying each dependent variable.


**Table 7.** Box-Behnken factorial design matrix for three factors and three responses obtained for the corolase protease.

<sup>1</sup> Values expressed as mean <sup>±</sup> SD of two replicates.

**Figure 3.** Pareto charts with the effect of three experimental factors, in decreasing order, obtained for protein % (**A**), antioxidant (**B**) and anti-hypertensive (**C**) properties in the experimental design with corolase, showing the most influent factors. The vertical lines in the pareto charts represent the level of significance (*p* = 0.05). XA—Enzyme %; XB—Hydrolysis temperature; XC—Hydrolysis time. **Figure 3.** Pareto charts with the effect of three experimental factors, in decreasing order, obtained for protein % (**A**), antioxidant (**B**) and anti-hypertensive (**C**) properties in the experimental design with corolase, showing the most influent factors. The vertical lines in the pareto charts represent the level of significance (*p* = 0.05). XA—Enzyme %; XB—Hydrolysis temperature; XC—Hydrolysis time. oxidant (**B**) and anti-hypertensive (**C**) properties in the experimental design with corolase, showing the most influent factors. The vertical lines in the pareto charts represent the level of significance (*p* = 0.05). XA—Enzyme %; XB—Hydrolysis temperature; XC—Hydrolysis time.

and hydrolysis time (XC), on the protein (**A**–**C**), ORAC (**D**–**F**) and iACE (**G**–**I**) responses obtained for the mussel extracts produced with corolase. The protein content of the hydrolysates showed a variation in the range of 44.30– **Figure 4.** Response surface graphs corresponding to the combined effect of enzyme % (XA), hydrolysis temperature (XB) and hydrolysis time (XC), on the protein (**A**–**C**), ORAC (**D**–**F**) and iACE (**G**–**I**) responses obtained for the mussel extracts produced with corolase. **Figure 4.** Response surface graphs corresponding to the combined effect of enzyme % (XA), hydrolysis temperature (XB) and hydrolysis time (XC), on the protein (**A**–**C**), ORAC (**D**–**F**) and iACE (**G**–**I**) responses obtained for the mussel extracts produced with corolase.

49.77%. By analysing the Pareto chart (Figure 3A), protein% was positively influenced by the linear effect of enzyme% (XA) and temperature (XB), meaning that the increase in en-

The quadratic effect of the temperature was also positively significant (*p* < 0.050). By multiple regression, the predicted response for the protein % could be obtained by the model

zyme concentration and temperature leads to an increase of protein release from the mussel biomass. The time of the hydrolysis did not significantly affect the protein % (*p* > 0.050). The quadratic effect of the temperature was also positively significant (*p* < 0.050). By multiple regression, the predicted response for the protein % could be obtained by the model

Protein % = 61.4391 + 1.20687 × XA − 0.846161 × XB + 0.00951786 × XB2 (4)

The final adjusted model showed a significant fit (*p* > 0.050) (Table 8). Still, the R2 indicates that only about 60% of the variability observed in relation to the protein content is explained by the model, indicating that the increase of corolase %, temperature, and

Protein % = 61.4391 + 1.20687 × XA − 0.846161 × XB + 0.00951786 × XB2 (4)

The protein content of the hydrolysates showed a variation in the range of 44.30–

The final adjusted model showed a significant fit (*p* > 0.050) (Table 8). Still, the R2 indicates that only about 60% of the variability observed in relation to the protein content is explained by the model, indicating that the increase of corolase %, temperature, and

hydrolysis time is not beneficial to produce protein-rich hydrolysates.

hydrolysis time is not beneficial to produce protein-rich hydrolysates.

in Equation (4).

in Equation (4).

The protein content of the hydrolysates showed a variation in the range of 44.30–49.77%. By analysing the Pareto chart (Figure 3A), protein% was positively influenced by the linear effect of enzyme% (XA) and temperature (XB), meaning that the increase in enzyme concentration and temperature leads to an increase of protein release from the mussel biomass. The time of the hydrolysis did not significantly affect the protein % (*p* > 0.050). The quadratic effect of the temperature was also positively significant (*p* < 0.050). By multiple regression, the predicted response for the protein % could be obtained by the model in Equation (4).

$$\text{Protein } \%= 61.4391 + 1.20687 \times \text{X}\_{\text{A}} - 0.846161 \times \text{X}\_{\text{B}} + 0.00951786 \times \text{X}\_{\text{B}}^{\text{2}} \tag{4}$$

The final adjusted model showed a significant fit (*p* > 0.050) (Table 8). Still, the R<sup>2</sup> indicates that only about 60% of the variability observed in relation to the protein content is explained by the model, indicating that the increase of corolase %, temperature, and hydrolysis time is not beneficial to produce protein-rich hydrolysates.

**Table 8.** Analysis of variance (ANOVA) for protein % obtained for the corolase Box-Behken design.


The antioxidant activity was significantly influenced by the linear effect of enzyme% (XA) and temperature (XB) (*p* < 0.050), while the hydrolysis time (XC) did not show a significant effect (*p* > 0.050). Enzyme% showed a positive effect, meaning that the increase in enzyme concentration results in higher ORAC values. On the other hand, temperature had a negative effect. Thus, the higher the temperature the lower the ORAC values will be. Furthermore, the quadratic effect of both factors also showed a significant effect. The linear interaction between temperature and hydrolysis time had a significant negative effect (Figure 3B).

By multiple regression, the predicted response for the ORAC could be obtained by the model in Equation (5).

$$\text{ORAC} = 4.11383 + 0.158065 \times \text{X}\_{\text{A}} - 0.159981 \times \text{X}\_{\text{B}} - 0.0548188 \times \text{X}\_{\text{A}}^2 + 0.00154318 \times \text{X}\_{\text{B}}^2 - 0.0045 \times \text{X}\_{\text{B}} \times \text{X}\_{\text{C}} \tag{5}$$

The final adjusted model highly explains the antioxidant activity, showing a significant fit (*p* > 0.050) and R<sup>2</sup> = 87.09% (Table 9).



The ACE inhibition (iACE) % was evaluated in a concentration of 10 mg hydrolysate/mL. The only variable that showed a significant effect on iACE was the enzyme% (XA), with a positive effect (Figure 3C). Thus, the increase in enzyme concentration increases the anti-hypertensive potential of the hydrolysates, which may be explained by the formation of more peptides with the ability to inhibit the ACE.

By multiple regression, the predicted response for the iACE could be obtained by the model in Equation (6).

$$\text{iACE} = 4.11383 + 0.158065 \times \chi\_{\text{A}} \tag{6}$$

The ANOVA results for the adjusted model was verified to have a significant fit (*p* > 0.050) (Table 10). However, the model only explained 57.90% of the variability in the anti-hypertensive results.



The Box–Behnken design allowed to optimise the conditions that would enable higher results for the individual responses (Table 11). However, a Derringer's desirability analysis was performed to optimise multiple responses of the design (Table 11). Thus, the hydrolysis of the minced mussel meat with 3% of the enzyme, at 40 ◦C for 3 h, seems to represent the best conditions to obtain the higher results in terms of hydrolysate proteins/peptides content as well as antioxidant and anti-hypertensive properties.

**Table 11.** Optimal conditions predicted by the experimental design to maximise protein/peptide release and antioxidant and anti-hypertensive activities of the hydrolysates, and Derringer desirability to predict the optimal conditions for a multiple response.


An enzymatic hydrolysis was performed, in triplicate, using the optimised conditions according to the design for the purpose of validation. The experiment was performed using the exact quantities used in the design experiments. The temperature was adjusted to 40 ◦C to work practically. The obtained hydrolysates were freeze-dried and then evaluated regarding their protein content and antioxidant and anti-hypertensive potential. The hydrolysates showed a mean of 47.36 ± 1.02% of protein content, antioxidant activity of 0.65 ± 0.062 µmol TE/mg hydrolysate, and ability to inhibit the activity of ACE in 55.36 ± 2.12% (at 10 mg hydrolysate/mL). The obtained results, although slightly lower, were not so different from the predicted ones. Furthermore, a scale-up hydrolysis was performed, in triplicate, with an increase of 15 times the amount of mussel biomass and solvent, maintaining the ratio used in the experimental design. The final hydrolysates showed a mean of 46.35 ± 1.12% of protein content, antioxidant activity of 0.389 ± 0.021, and IC<sup>50</sup> for ACE inhibition of 3.7 ± 0.33 mg protein/mL (Table 12). The scaled-up results were verified to be slightly lower than the predicted ones and the validation

hydrolysates regarding protein content. However, the antioxidant activity showed a pronounced decrease.

**Table 12.** Results predicted by the model, and results obtained in a validation and scaled-up enzymatic hydrolysis, performed with the optimal conditions described in Table 11.


#### **3. Discussion**

Mussel meat has a high protein content, making it interesting to produce bioactive hydrolysates rich in proteins and bioactive peptides. However, the mussel *Mytilus galloprovincialis* is less exploited regarding its bioactive potential when compared to other mussel species, such as *M. coruscus* and *M. edulis*. Since we wanted to create a food-grade method, we chose two food-grade proteases to carry out enzymatic hydrolysis. Thus, to explore this mussel potential, we have performed two Box–Behken experimental designs, with two different proteases, aiming to obtain hydrolysates with interesting potential for industrial applications. Furthermore, we have not found studies with mussels from the genus *Mytilus* performing enzymatic hydrolysis with subtilisin or corolase. The most frequent enzymes found were mainly gastric enzymes, such as pepsin and trypsin, and non-gastric enzymes, such as papain and flavourzyme.

The mussel meat biomass used showed 70.50 ± 13.44% of protein on a dry weight (DW) basis and a moisture content of 90.30 ± 4.24%. These results show higher values of protein when compared to other studies with *Mytilus* sp. from Portugal and Spain that showed protein content variation from 39.17–42.94 (DW) and moisture % of 81.71– 87.59% [23]. However, these results are in line with the possible variations in protein content that can occur in different months, as shown by Çelik [24] in a study with *Mytilus galloprovincialis* indicating higher protein levels (74.64%) in February.

The protein % of the hydrolysates does not seem to be highly influenced by the determined models, indicating that enzymatic hydrolysis with both enzymes can produce hydrolysates with protein contents in the range of 40–48% (DW). So, to obtain mussel hydrolysates with a content of above 40%, the most economical and fastest conditions can probably be used.

The subtilisin protease optimised method was an enzymatic hydrolysis with 1.5% of enzyme with a duration of 3 h at 52 ◦C. In a scale-up test with these conditions, the final hydrolysates showed protein content, ORAC, and IC50 for ACE inhibition values of 45.0 ± 0.38, 0.49 ± 0.061 245 µmol TE/mg hydrolysate, and 1.0 ± 0.56 mg protein/mL, respectively. With the corolase, the optimised method was an enzymatic hydrolysis with 3.0% of enzyme with a duration of 3 h at 40 ◦C, obtaining scale-up hydrolysates with protein content, ORAC, and IC<sup>50</sup> for ACE inhibition values of 46.35 ± 1.12, 0.389 ± 0.021 µmol TE/mg hydrolysate, and 3.7 ± 0.33 mg protein/mL, respectively. The experimental design responses were not highly explained by the models, indicating that the system is highly variable, as necessary to enhance the process, or a plateau may have been quickly reached, which challenges the explanation of the variability in the models. However, the hydrolysates showed potential as proteins/peptides sources with antioxidant properties, bringing interest to the results. In both experiments, interesting protein values were obtained with a few hours of hydrolysis, which is in line with other studies showing that enzymatic hydrolysis with papain for 2 h was enough for achieving the maximum protein

extraction [25]. The obtained protein content (450 and 463 mg protein/g hydrolysate) was close to those obtained for *Mytilus edulis* by Vareltzis and Undeland (430 and 580 mg protein/g with acid and alkaline process, respectively) [26], but lower than those obtained by Neves et al. (735.45 ± 11.45 mg protein/g hydrolysate) [15]. The subtilisin method needs a lower enzyme% to obtain higher bioactive properties than the corolase, with the main difference being observed for the anti-hypertensive potential. Even though the corolase hydrolysate shows a higher protein %, this does not bring much potential for this hydrolysate due to the small difference compared to the subtilisin hydrolysate. So, mussel meat hydrolysate produced with the subtilisin protease appears to have more potential for further studies as an active ingredient, at least regarding the antioxidant and anti-hypertensive potential. However, it is important to highlight that the obtained values for the anti-hypertensive property are not very significant since IC<sup>50</sup> ≥ 1000 µg protein/mL [27]. The hydrolysate produced with corolase shows the lowest potential with an IC<sup>50</sup> = 3700 µg protein/mL. The subtilisin hydrolysate seems to be more promising, with an IC<sup>50</sup> = 1000 µg protein/mL. Bioactive peptides usually have a molecular weight (MW) less than 6 KDa [28], and the most efficient anti-hypertensive peptides are usually associated with MW lower than 3 KDa [29]. Several marine derived peptides with MW lower than 3 KDa have been described, such as the microalgae *Chlorella vulgaris* VECYGPNRPQF peptide (1.3 KDa; IC<sup>50</sup> of 29.6 µM) [30] and *C. ellipsoidea* VEGY peptide (467 Da; IC<sup>50</sup> of 128.4 µM) [31]; the macroalgae *Gracilariopsis lemaneiformis* TGAPCR peptide (604 Da; IC<sup>50</sup> of 23.94 µM) [32] and *Nannochloropsis oculata* LEQ peptide (369 Da; IC<sup>50</sup> of 173 µM) [33]. Thus, to increase the anti-hypertensive potential of the produced hydrolysates, a future approach may be to submit them to a ultrafiltration system using 3-KDa cut-offs, to concentrate peptides with lower MW [34]. Furthermore, the production of low MW peptides may also increase the antimicrobial potential of hydrolysates, thus presenting new possible pplications.

All the hydrolysis performed used the same mussel batch, initially minced and stored at −20 ◦C. The main goal was to assure that all the hydrolysis were performed with minimum mussel internal variations, since we wanted to compare a large number of extracts to optimize the hydrolysate production. The validation and scaled-up hydrolysis were also performed with the same batch, allowing us to precisely compare these extracts with those obtained using the experimental design, excluding possible mussel internal chemical variations. However, it is important to point out that mussel meat biochemical composition varies with the harvesting season, due to their reproductive cycle, environmental conditions, growth, and food availability [24]. Çelik et al. [24] showed that mussel protein content is highly related with the spawning seasons, with decreased protein levels being observed during this season, which increases after spawning time. So, different harvesting seasons lead to variations in the biochemical composition, which may be reflected in differences in mussel protein and amino acids, not only in terms of quantity, but also quality. Consequently, the enzyme action will produce different peptides over the seasons. Therefore, it would be expected that the ORAC and iACE results obtained for hydrolysates produced with the presented methods may differ between different mussel batches, depending on their harvesting season and other external factors. Furthermore, the mussel's digestive gland produces proteases, which also seems to be influenced by their diet [35], and mussels seem able to modulate their digestive enzyme activities in response to limited feeding and thermal stress [36]. Since endogenous proteases may also have either a proteolytic effect or serve as an enzymatic substrate in the hydrolysis, the amount of endogenous proteases may also contribute to the variability of results. Thus, in the future, it would be of great interest to perform the same hydrolysis in different mussel batches, harvested in different months, and perhaps from different locations, to examine the variability of the produced hydrolysates when influenced by the expected biochemical composition differences.

The production of multifunctional extracts from mussels may be an interesting approach for food applications since they are not only a source of proteins, but also present

bioactivities that can enhance consumer health, useful for the creation of functional food. Moreover, they may also be used as nutraceuticals or as cosmetic ingredients. Antioxidant food and nutraceuticals may help reduce levels of radical oxygen species that are constantly produced by the human organism, especially during high exposure to external factors, such as alcohol, tobacco smoke, and stress [37]. Hypertension has been associated as one of the main causes of cardiovascular diseases [38], with the angiotensin-converting enzyme (ACE) being one of the major enzymes involved in the process of blood pressure regulation [39]. Thus, multifunctional extracts may be incorporated in food matrices with health claims, to facilitate sale as functional food. However, for claiming health benefits, it is important to study the bioavailability of food matrices incorporating these hydrolysates [40], by analysing their resistance to the gastrointestinal (GI) tract enzymes and conditions, to verify if their properties are maintained throughout the GI tract passage [41]. In cosmetics, antioxidants are especially important for anti-ageing purposes since free radicals are highly associated with skin ageing. Thus, natural antioxidant hydrolysates used as active cosmetic ingredients may help decrease free radical damage and work as an alternative for synthetic antioxidant ingredients.

Furthermore, mussel protein and peptide hydrolysates are frequently associated with other properties, especially antimicrobial properties [42,43], but also anticancer [44], anti-inflammatory [45], anticoagulant [14], antidiabetic [15], and antiviral [46]. Thus, in the future, it would be interesting to study these hydrolysates for other bioactivities. Additionally, the water-soluble nature of these extracts makes it easy to incorporate them in several matrices. Although the freeze-drying process may lead to a loss of bioactivity, it is important for a better preservation of the hydrolysates, facilitating their incorporation in both solid and liquid matrices.

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

#### *4.1. Materials*

The enzymes used were subtilisin kindly supplied by Aquitex, and the commercial digestive-enzyme complex Corolase PP purchased from AB Enzymes GmbH (Darmstadt, Germany). The mussels were kindly supplied by Testa & Cunhas (Gafanha da Nazaré, Portugal).

#### *4.2. Mytilus Galloprovincialis Meat Characterisation*

The *Mytilus galloprovincialis* meat used was characterised, in triplicate, before being minced, according to a few nutritional characteristics. Total fat, protein, and moisture content were measured in accordance with the established standards PE.Q.AC.04 Ed.06, PE.Q.AC.03 Ed.07 (ISO 1871:2009), and PE.Q.AC.01 Ed.06 (NP 2282:2009), respectively.

#### *4.3. Enzymatic Hydrolysis Procedures*

When received, mussels were clean, and the meat was separated from the shell. Mussel meat was then minced until homogenised and stored at −20 ◦C for further analysis. A preliminary study was performed with different conditions, with variations on the enzymes concentration (0.5–4%), hydrolysis time (from 30 min to 4 h), and mussel/water ratio (*w*:*v*) (1:1, 1:2, 1:3), to understand the limits to be established for the experimental design. Concerning the experimental design, all the hydrolysis reactions for both enzymes were prepared using the previously stored mussel meat minced biomass. Briefly, mussel biomass was mixed with ultrapure water in a ratio of mussel:water of 1:2 (*w*:*v*) and pH was adjusted to 7.5. Then, the enzyme was added in the intended test concentration and the mixtures were incubated at the test temperature in an orbital shaker (Thermo Scientific™ MaxQ™ 6000) (conditions tested at Tables 1 and 7). The pH was verified and adjusted to 7.5 every 15 min. To stop the hydrolysis reaction, the samples were incubated at 90 ◦C for 10 min to inactivate the enzymes. Samples were centrifuged at 5000× *g* for 30 min, and the supernatant was collected and freeze-dried for further analysis.

#### *4.4. Experimental Design*

Two experimental designs, one with corolase and the other with the subtilisin protease, were implemented to establish the most influential factors that could produce a hydrolysate rich in proteins and bioactive properties. For that, a Box–Behnken design was selected. The factors evaluated were enzyme %, hydrolysis temperature (◦C), and hydrolysis time (hours). Enzyme % and hydrolysis time were chosen according to single-factor experiments (data not shown). The ORAC assay was performed for each hydrolysate. The temperature and the pH tested were selected according to the functioning range of the enzymes. The levels of the factors coded as −1, 0, and 1 were established and are shown in Table 13. The selected response variables were protein content as well as antioxidant and antihypertensive potential. Each design resulted in an arrangement of 15 treatments, executed in duplicate (a total of 30 runs). Each hydrolysis was performed as described before.

**Table 13.** Levels for 3 experimental factors for the two experimental designs.


*4.5. Statistical Analysis and Statistical Model*

The optimisation analysis was performed using Statgraphic Centurion software. All data were expressed as means ± standard deviation (S.D.). Means were considered statistically significant using a significance level of 0.05. Responses were adjusted to the second-order polynomial model (Equation (7)):

$$\mathbf{Y} = \mathfrak{B}\_{\mathsf{D}} + \mathfrak{B}\_{\mathsf{A}}\mathbf{X}\_{\mathsf{A}} + \mathfrak{B}\_{\mathsf{B}}\mathbf{X}\_{\mathsf{B}} + \mathfrak{B}\_{\mathsf{C}}\mathbf{X}\_{\mathsf{C}} + \mathfrak{B}\_{\mathsf{A}\mathsf{B}}\mathbf{X}\_{\mathsf{A}}\mathbf{X}\_{\mathsf{B}} + \mathfrak{B}\_{\mathsf{A},\mathsf{C}}\mathbf{X}\_{\mathsf{A}}\mathbf{X}\_{\mathsf{C}} + \mathfrak{B}\_{\mathsf{B},\mathsf{C}}\mathbf{X}\_{\mathsf{B}}\mathbf{X}\_{\mathsf{C}} + \mathfrak{B}\_{\mathsf{A},\mathsf{A}}\mathbf{X}\_{\mathsf{A}}{^2 + \mathfrak{B}\_{\mathsf{B}}\mathbf{X}\_{\mathsf{B}}{^2 + \mathfrak{B}\_{\mathsf{C}}} + \mathfrak{B}\_{\mathsf{C},\mathsf{C}}\mathbf{X}\_{\mathsf{C}}{^2 + \varepsilon} \tag{7}$$

where Y is the measured response; β0 is the constant; βA–βC are the coefficients associated with linear, quadratic, and interaction effects of the variables X<sup>A</sup> (enzyme %), X<sup>B</sup> (Temperature), and X<sup>C</sup> (Time), respectively, and ε is the residual error. In the final models for each variable, only the significant effects appear (*p* < 0.05). To optimise the multiple responses obtained, a Derringer's desirability function was applied to the results of each design [22].

#### *4.6. Protein Quantification*

Total nitrogen content was determined by the micro-Kjeldahl method. Briefly, 0.2 g of freeze-dried hydrolysate were digested with 1 g of Kjeldahl catalyst and 4 mL of H2SO<sup>4</sup> (ρ20 = 1.84 g/mL) at 400 ◦C for 2 h. The reaction was stopped with 20 mL of deionised water. The samples were distilled using 30 mL of NaOH 10 M. A boric acid solution with bromocresol and methyl red was used as indicator. The resulting solution was titrated with HCl 0.1 M. The total nitrogen and protein percentage were determined using the Equations (8) and (9), respectively, where f (HCl 0.1 m) = 0.0014 and Kjeldahl factor = 6.25.

$$\text{Total nitrogen (\%)} = \text{f} \times \text{(V}\_{\text{sample}} - \text{V}\_{\text{blank}}) \times \text{(100/sample weight)}\tag{8}$$

$$\text{Protein content (\%)} = \text{Total nitrogen (\%)} \times 6.25 \tag{9}$$

#### *4.7. Antioxidant Activity*

The antioxidant activity was measured by the Oxygen radical absorbance capacity (ORAC) assay, performed in a black 96-well microplate (Nunc, Denmark) according to the method described by Coscueta et al. (2020) [47]. Briefly, the reaction was carried out in 75 mM phosphate buffer (pH 7.4) at 40 ◦C. The final assay mixture was 200 µL, containing

fluorescein (70 nM, final concentration in well), 20 -Azobis (2-methylpropionamidine) dihydrochloride (AAPH) (12 mM, final concentration in well), and either Trolox (1–8 µM, final concentration in well), for the calibration curve, or sample. A control with PBS instead of the antioxidant solution was used. Before adding AAPH, the mixture was pre-incubated for 10 min at 37 ◦C. AAPH solution was added rapidly. The fluorescence was recorded at intervals of 1min for 80 min in a multidetection plate reader (Synergy H1; BioTek, Winooski VT, USA) with excitation and emission wavelengths of 485 nm and 528 nm, respectively. The equipment was controlled by the Gen5 BioTek software version 3.04. Antioxidant curves (fluorescence versus time) were first normalised to the curve of the blank corresponding to the same assay by multiplying original data by the factor fluorescence blank, t = 0/fluorescence control, t = 0. The area under the fluorescence decay curve (AUC) was calculated according to the trapezoidal method from the normalised curves. The final AUC values were calculated by subtracting the AUC of the blank from all the results. Regression equations between net AUC and antioxidant concentration were calculated.

#### *4.8. Anti-Hypertensive Activity*

The ACE-inhibitory activity was performed in a black 96-well microplate (Nunc, Denmark) according to the method described by Sentandreu & Toldrá (2006) [48] with some modifications [27]. This method is based on the ability of the angiotensin-I converting enzyme (ACE) to hydrolyse a specific substrate (o-aminobenzoylglycyl-p-nitrophenylalanylproline (Abz–Gly–Phe(NO2)–Pro)), generating the fluorescent product o-aminobenzoylglycine (Abz–Gly). A commercial Angiotensin-I converting enzyme (EC 3.4.15.1, 5.1 U/mg), purchased from Sigma Chemical (St. Louis, MO, USA), was previously diluted in 5 mL of a glycerol solution in 50% ultra-pure water. Then, ACE was diluted 1:24 with a 150 mM Tris buffer solution (pH 8.3), containing 1 µM of ZnCL2, reaching a final concentration of 42 mU/mL). A total of 40 µL of ultrapure water or ACE working solution was added to each microtiter-plate well, then adjusted to 80 µL by adding ultrapure water to blank, control, or samples. The reaction was initiated with the addition of 160 µL of the substrate solution (0.45 mM solution of ABz-Gly-Phe(NO2)-Pro (Bachem Feinchemikalien, Bubendorf, Switzerland) dissolved in 150 mM Tris buffer (pH 8.3) containing 1.125 M NaCl). The mixture was incubated at 37 ◦C for 30 min, and the fluorescence generated was measured using a multidetection plate reader (Synergy H1; BioTek, Winooski VT Vermont, USA) with excitation and emission wavelengths of 350 nm and 420 nm, respectively. To obtain the IC<sup>50</sup> of the inhibitory activity, which is the concentration of the sample that is required to inhibit the original ACE activity by 50%, serial dilutions of each sample were performed (1/1 to 1/32). A non-linear modelling of the obtained data was used to calculate the IC<sup>50</sup> values, using the 5 Parameter curve fit method and the Interpolate function from Gen5 software (BioTek Instruments).

#### **5. Conclusions**

Although marine species have often been described as a source of bioactive hydrolysates and bioactive peptides, the mussel *Mytilus galloprovincialis* has been less exploited. Due to its high protein level, this marine specie seems to be an interesting potential source of bioactive peptides. Thus, in this work, the factorial designs allowed to confirm the combination of experimental factors that leads to the production of the most efficient hydrolysate from the mussel *Mytilus galloprovincialis,* with the highest levels of proteins/peptides as well as antioxidant and anti-hypertensive activity. The use of enzymatic hydrolysis with food-grade enzymes presents the opportunity to create active ingredients that can be further explored to produce functional food, nutraceuticals, and cosmetics. Furthermore, the use of discarded mussels to produce functional ingredients for food, cosmetic, and pharmaceutic industries may contribute to the valorisation of world waste in a circular economy context.

**Author Contributions:** Conceptualisation, S.A.C. and M.P.; methodology, S.A.C., R.d.C. and M.P.; software, E.R.C.; validation, S.A.C. and R.d.C.; formal analysis, S.A.C., E.R.C. and M.P.; investigation, S.A.C. and M.P.; resources, M.P.; data curation, S.A.C. and E.R.C.; writing—original draft preparation, S.A.C.; writing—review and editing, S.A.C., E.R.C. and M.P.; visualisation, S.A.C. and E.R.C.; supervision, M.P.; project administration, M.P.; funding acquisition, S.A.C. and M.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research and the APC was funded by Fundo Europeu de Desenvolvimento Regional (FEDER), through the Programa Operacional Competitividade e Internacionalização (POCI) under the project VALORMAR: Valorização Integral dos recursos Marinhos: Potencial, Inovação Tecnológica e Novas Aplicações (POCI-01-0247-FEDER-024517); CBQF under the FCT project UIDB/50016/2020; and the individual FCT PhD research grant (ref. SFRH/BD/144155/2019) for the author Sara A. Cunha.

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

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The authors acknowledge Testa & Cunhas (Gafanha da Nazaré, Portugal) for the *Mytilus galloprovincialis* supply; and Aquitex S.A. for the subtilisin protease supply.

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

**Sample Availability:** Samples of the compounds are not available from the authors.

#### **References**


## *Review* **Phages and Enzybiotics in Food Biopreservation**

**José Ramos-Vivas 1,2 , María Elexpuru-Zabaleta <sup>1</sup> , María Luisa Samano 1,2 , Alina Pascual Barrera <sup>2</sup> , Tamara Y. Forbes-Hernández <sup>3</sup> , Francesca Giampieri 4,5,\* and Maurizio Battino 4,6,\***


**Abstract:** Presently, biopreservation through protective bacterial cultures and their antimicrobial products or using antibacterial compounds derived from plants are proposed as feasible strategies to maintain the long shelf-life of products. Another emerging category of food biopreservatives are bacteriophages or their antibacterial enzymes called "phage lysins" or "enzybiotics", which can be used directly as antibacterial agents due to their ability to act on the membranes of bacteria and destroy them. Bacteriophages are an alternative to antimicrobials in the fight against bacteria, mainly because they have a practically unique host range that gives them great specificity. In addition to their potential ability to specifically control strains of pathogenic bacteria, their use does not generate a negative environmental impact as in the case of antibiotics. Both phages and their enzymes can favor a reduction in antibiotic use, which is desirable given the alarming increase in resistance to antibiotics used not only in human medicine but also in veterinary medicine, agriculture, and in general all processes of manufacturing, preservation, and distribution of food. We present here an overview of the scientific background of phages and enzybiotics in the food industry, as well as food applications of these biopreservatives.

**Keywords:** bacteriophage; endolysin; enzybiotics; biopreservation

#### **1. Introduction**

Food preservation by suitable means is key in food safety and quality. There are several traditional and well-known physical preservation techniques such as refrigeration and pasteurization, but the modern industry is always looking for new procedures for food preservation to increase the product's shelf-life by minimizing the loss of nutritional quality and organoleptic properties. Presently, some modern biopreservation techniques rely on naturally occurring microorganisms (i.e., lactic acid bacteria) and their metabolites. These food preservatives are mainly used to produce safer food for the consumer, preventing the action of pernicious microbes which can cause food deterioration or even toxicity and therefore be dangerous to human health.

Moreover, bacteria -including multidrug-resistant bacteria- can reach food at different points in the food supply chain, from farm to postharvest, and processing such as slaughtering, fermentation, packaging and storage [1–5].

As most natural foods are highly perishable, by extending their half-life we can also control their native microbiota for proper preservation, maintaining their safety and quality.

**Citation:** Ramos-Vivas, J.; Elexpuru-Zabaleta, M.; Samano, M.L.; Barrera, A.P.; Forbes-Hernández, T.Y.; Giampieri, F.; Battino, M. Phages and Enzybiotics in Food Biopreservation. *Molecules* **2021**, *26*, 5138. https://doi.org/10.3390/ molecules26175138

Academic Editors: Manuela Pintado, Ezequiel Coscueta, María Emilia Brassesco and Ricardo Calhelha

Received: 13 July 2021 Accepted: 20 August 2021 Published: 25 August 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

As microorganisms produce a long list of molecules ranging from classic antibiotics to antibacterial enzymes, the control of indigenous populations in food can be achieved by adding these products directly. The paradigm of bacterial molecules used in the food industry as biopreservatives is Nisin, a bacteriocin produced by the Gram-positive bacterium *Lactococcus lactis*, one of the lactic acid bacteria most extensively used for the manufacture of dairy products [6]. Other well-known bacteriocins, such as Pediocin, Natamycin, Enterocin, and Leucocin [7], also have inhibitory properties against other microorganisms which makes them very interesting for use in the food industry. Some bacteria that produce these compounds have been used as probiotics. Current research on probiotics is quite promising and modern fashion trends push probiotics and bacteriocins from modulation of the gut microbiota toward a wide range of other health-promoting activities away from food, such as cancer treatment, skin health care, periodontal health, or allergies [8–11].

In addition, the use of bacteriocin producing strains or those that can compete against pathogens in the context of the food industry needs new approaches, mainly due to the increase in foodborne infections, the appearance of new production processes, the massive demand for food, and the changing consumer trends. Moreover, the extensive use of antibiotics against animal and human pathogens has also led to an increase in foodborne pathogens resistant to antibiotics, which makes the picture not reassuring at all [12–14].

Goodridge and Abedon published an article in 2003 where they proposed to use the terms "phage biocontrol" and "phage bioprocessing" to differentiate the application of bacteriophages in the farm or crops from their use in the food industry [15]. Several years later, Greer published a review of the control of foodborne bacteria using phages, including the effects of these microorganisms on food storage and preservation [16].

At that time, the excellent properties of endolysins to kill bacteria were already known, but their use to protect food from foodborne pathogens had not yet been effectively tested. One of the first murein hydrolases to be studied concerning food-related bacteria was that of the *Lactobacillus helveticus* bacteriophage 0303 [17]. This endolysin exhibited a broad spectrum of activity, killing different bacterial species such *Pediococcus acidilactici*, *Lactobacillus delbrueckii* subsp. *bulgaricus*, *Lactobacillus delbrueckii* subsp. *lactis*, *Lactobacillus acidophilus*, *Bacillus subtilis*, *Enterococcus faecium*, and several strains of *Lactobacillus helveticus*.

Problems of deterioration of the organoleptic properties have been described after physical treatments; also, consumers are increasingly demanding low-processed foods. One of the advantages of phages over the usual physical treatments is that phages do not modify any organoleptic properties of foods. Moreover, even with common treatments such as heat, team and UV light, a relatively high percentage of food products are lost due to subsequent microbial spoilage or microbial contamination; when food becomes contaminated, it will lead to food spoilage, and such food will no longer be fit for consumption.

Thanks to their ability to control or to inactivate spoilage and/or foodborne bacteria selectively, bacteriophages have great potential as food biopreservatives. Additionally, in terms of food biopreservation, enzybiotics are beginning to be increasingly studied in the field of food microbiology, taking advantage of the pull that in vitro successes have displayed against very important multidrug-resistant human and animal pathogens [18–20].

In this review, we discuss the use of phages and their lytic enzymes as a tool to eliminate or reduce spoilage bacteria and common foodborne bacterial pathogens.

#### **2. Why Bacteriophages?**

Bacteriophages are an alternative to antimicrobials in the fight against bacteria, mainly because they have a practically unique host range, which gives them great specificity. Apart from their selective activity, bacteriophages have been successfully tested to eliminate or weaken biofilms formed by different classes of both Gram-negative and Gram-positive pathogens in the food industry [21–24]. Biofilms are consortia of bacteria that persist on different surfaces or pipelines within the food industries that contaminate food at some point in the processing or packaging chain.

In addition to their potential ability to specifically control strains and biofilms of pathogenic bacteria, their use does not generate a negative environmental impact like in the case of antibiotics or disinfectants [25]. Other advantages of these viruses are: (i) safety—as they are not toxic to eukaryotic cells, (ii) the preservation of the organoleptic properties of food, and (iii) the control of multi-resistant bacteria since the tolerance of some strains to phages can often be overcome with the use of phage cocktails [14]. In addition, phages can be used in combination with antibiotics, bacteriocins, or even with probiotics.

The main limitations of bacteriophages as biopreservative tools in foods derive from the scarce knowledge of their genetics since the use of strains that may contain virulence factors, lysogeny, or antibiotic resistance genes is inadvisable. As an example, studies prior to this decade did not have the modern and inexpensive sequencing techniques that almost all laboratories can afford today. Furthermore, in some cases, it is necessary to use phage cocktails that are more difficult to characterize than individual strains. Additionally, we need to learn much more about their behavior within solid and liquid food matrices to optimize the amount of phage to be used in each case. The method of releasing phages on food is also important, since the phages must reach the largest number of bacteria possible so that they can effectively control them and reduce their number to safe values. In other words, phages and bacteria must be in contact with liquid but also with solid foods; moreover, as much bacterial contamination occurs initially at low numbers (a minimum bacterial density is a prerequisite) sometimes we must apply a large number of phages to those foods. Knowing the optimal number of viral particles (multiplicity of infection, MOI) to use for each food, as well as their infection kinetics in each food matrix, it is essential to understand how these phages are acting on their target pathogens [26–32]. Minimum host threshold requirement has been demonstrated for phages of different food pathogens [33,34]. As successful biopreservative agents, it is also important to consider phages' stability in food matrices under different environmental conditions such as water activity, salinity, temperature, pH, osmotic shock, and light (visible and UV). According to several authors, phages have a remarkable stability in foods [35–37]. Phage propagation on a susceptible host, purification, and phage or cocktail formulation are very relevant parameters too.

In some studies, in which a high number of phages are used, the bacterial lysis 'from without' can occur because many viral particles bind to the bacterial surface, leading to the production of numerous holes in the cell wall [38,39]. All these concepts must be better studied and understood in order to apply phages to food pathogens.

Although the application of phages will continue, there is a phenomenon that must always be kept in mind, the emergence of phage-resistant strains. When infecting bacterial cells, phages already face a range of antiviral mechanisms (i.e., restriction modification systems/enzymes), and they have evolved multiple tactics to avoid these mechanisms. In this co-evolution between bacteria and phages, most authors agree that phages can effectively raise a counter-resistance. Therefore, finding a new phage that can infect a bacterium will always be easier than finding an entirely novel family of antibiotics.

We do not know much about how often these resistant variants of phages used in the food industry appear, as few publications include assays to study this phenomenon. It is likely that researchers prioritize the study of efficacy over safety. Moreover, multidrug resistance, where a bacterium has obtained resistance mechanisms against several different families of antibiotics, is increasingly common, but this phenomenon does not occur when phages are used. Additionally, many studies suggest that phage combinations can be optimized to limit the emergence and persistence of resistance, therefore promoting the long-term usefulness of phage therapy. With regards to this issue, enzybiotics offer the advantage that they do not generate resistance because they act on essential targets for the bacteria's viability, so, it is difficult for bacteria to modify them.

The other most important issue in addition to the development of phage-resistant strains is phage spread. As bacteriophages applied to food can be easily transferred between facilities in the food industry, we must pay particular attention to the number of

phages used, and above all, to how they are applied to food. An undesirable effect would be the inactivation of starter cultures that initiate the fermentation processes. Despite the narrow spectrum of a specific phage, the problem of the phages spread within the food industries is real because it is not convenient; for example, to collaterally eliminate some species of lactic acid bacteria that confer characteristic properties to the products in which they are present [40].

As with isolated phages, phage cocktails can be used directly on food or surfaces and food handling tools in chain processing plants. Another advantage of phage cocktails is that they can be modified quickly and conveniently to deal with specific strains that may appear in a particular food manufacturing facility [41]. No articles were reviewed here where more than three bacteriophages or cocktails containing undefined strains were used because in the last few years there have been excellent reviews on that scope [26,41–43]. Moreover, Theuretzbacher's recent article in the currently available weaponry against superbugs indicates that more than 20 different bacteriophage-based products have been approved for the control of pathogenic bacteria related to the food industries or direct food contamination [44].

Our review of approximately 100 bacteriophages indicates that three families (*Myoviridae*, *Siphoviridae*, and *Podoviridae*) account for the majority of virulent phages for the most common food-borne pathogen species. Much work has focused on the biocontrol or biopreservation of foods with six of the most important food-borne pathogens: *E. coli* (mainly serotype 015:H7), *Listeria monocytogenes*, *Staphylococcus aureus*, *Clostridium* spp., *Campilobacter jejuni*, and *Salmonella* spp., (Table 1). In addition to those six important food pathogens, phages against many other bacteria capable of causing foodborne infections should begin to be studied. This would allow us to identify not only new phages but also interesting enzybiotics.



**Table 1.** Phages tested against food-borne pathogens and their proposed use as food biopreservatives.





According to the articles analyzed, the phages of the family *Myoviridae* were preferentially used to control *E. coli*. Other important food pathogens such as *C. jejuni*, *Salmonella*, *L. monocytogenes*, and *S. aureus* were controlled by *Siphoviridae* and *Myoviridae*. The analyzed studies showed that the *Podoviridae* family can infect all these species, but fewer phage strains of this family have been found to control bacteria in the different foods tested. Comparative genomics and morphological observation by transmission electron microscopy revealed that the phage LPSEYT, able to infect *Salmonella*, represents a new genus within the *Myoviridae* family [42]. This last example shows that if we go a little deeper into the genomic characterization of the isolated strains, we will be able to advance in the knowledge of the taxonomy of phages. Most of the phages used to control these pathogen species in food were isolated from wastewater, sewage, or other environmental samples; but many have also been isolated from different foods. One phage strain (EcpYZU01) of the *Corticoviridae* family was isolated from sewage samples and tested against *Enterobacter cloacae* in cucumber juice [43]. Finally, a phage (LPST94) from the *Ackermannviridae* family isolated from water was effective against *Salmonella* in foods [108,109]. This newly assigned family was recently added to the list of the International Committee on Taxonomy of Viruses ICTV catalog. The isolation of phages from sewage and water samples is common due to their abundance in these ecosystems. However, Scattolini et al., pointed out that the search and characterization of phages isolated in the same foods in which the pathogens can hide could be a good way "to integrate this control measure in an innovative, cost-effective, safe and environmentally friendly way" [86]. Therefore, it seems like a good idea to use phages in food safety which in turn come from food, especially for the consumer, who can identify fewer drawbacks than when consuming phages or their genetically manipulated enzybiotics.

Bacteriophages can also be used to prevent or to reduce colonization of domesticated livestock with bacterial pathogens before they enter the production chain [148]. After that, phages can be used to decontaminate inanimate surfaces made, for example, of stainless steel or to fight bacterial biofilms. Finally, phages can be used directly on food, both in unprocessed or ready-to-eat foods as well as processed foods, even stored at temperatures ranging from 4 ◦C to 20 ◦C.

Several cofactors tested with phages used in the control of *L. monocytogenes* in the food industry have been recently reviewed by Kawacka and coworkers [26,149]. Among those factors, we can find other bacterial cultures such as *Lactobacillus* spp., *Gluconocbacter assii*, the bacteriocins Nisin, Enterocin and Pediocin, and several compounds such as lauric arginate, potassium lactate, sodium diacetate, sucrose monolaurate.

#### **3. Spatial Distribution of Phages**

Bacteriophages' ubiquity is another advantage. It is estimated that there are 10 bacteriophages for every bacterium present on our planet, representing a virtually unlimited source, not only of virions but also of lytic enzymes. Phages are especially abundant in seawater and soil and have also been found in large quantities in wastewater. The potential use of bacteriophages as indicators of environmental contamination has also been investigated in the last few decades [150–155]. Perhaps the most impressive figures are that phages kill bacteria at rates of up to 40% of the total population of marine bacteria per day and that carbon flux through phage biomass is estimated at 145 gigatonnes per year, playing a crucial role in our planet's global carbon cycle [156,157]. They are also easily found on any animal or plant surfaces as they are part of the microbiota of most living things. Phages have also been isolated from a variety of foods, including ready-to-eat foods, fish and shellfish, milk products, meat, and vegetables [33,158–162]. Because of this, consumers are already in contact with food bacteriophages every day. Therefore, if researchers could offer an adequate explanation, it would help consumers to increase their acceptance of the use of food bacteriophages. In other words, they should accept their use as biopreservatives if we can explain well what this class of virus really is and how exactly they are used to fight "bad" bacteria in food.

[172].

#### **4. Morphology and Classification** of Ackermann or the criteria of the International Committee on Taxonomy of Viruses

*Molecules* **2021**, *26*, x FOR PEER REVIEW 11 of 25

**4. Morphology and Classification**

Initially, phages were characterized by transmission electron microscopy (TEM), followed by pulse-field gel electrophoresis and restriction endonuclease analysis. However, although TEM continues to be essential in publications on bacteriophage viruses, the quality of the images in many of the articles is questionable [163]. Most studies use the work of Ackermann or the criteria of the International Committee on Taxonomy of Viruses (ICTV) [164] to identify their phage isolates [165–167]. For further taxonomic classification and phage characterization, more detailed information, such as genomic data, has begun to be included in scientific publications [168–171]. (ICTV) [164] to identify their phage isolates [165–167]. For further taxonomic classification and phage characterization, more detailed information, such as genomic data, has begun to be included in scientific publications [168–171]. Most phages belong to the order Caudovirales. Based on the tail morphology, Cau‐ dovirales are divided into three families: *Myoviridae*, *Siphoviridae*, and *Podoviridae*. *Myovir‐ idae* phages are characterized by long straight contractile tails, *Siphoviridae* phages possess long flexible non‐contractile tails, and *Podoviridae* phages have short, non‐contractile tails

Initially, phages were characterized by transmission electron microscopy (TEM), fol‐

ity of the images in many of the articles is questionable [163]. Most studies use the work

Most phages belong to the order Caudovirales. Based on the tail morphology, Caudovirales are divided into three families: *Myoviridae*, *Siphoviridae*, and *Podoviridae*. *Myoviridae* phages are characterized by long straight contractile tails, *Siphoviridae* phages possess long flexible non-contractile tails, and *Podoviridae* phages have short, non-contractile tails [172]. Alternatively, we can also use the PCR technique and subsequent sequencing to par‐ tially characterize the isolated phages. For example, some authors used specific primers

Alternatively, we can also use the PCR technique and subsequent sequencing to partially characterize the isolated phages. For example, some authors used specific primers to detect the Major Capsid Protein (MCP) of reported *Salmonella* phages [158,159]. to detect the Major Capsid Protein (MCP) of reported *Salmonella* phages [158,159]. Augustine et al., also used PCR or multiplex PCR to perform a screening of virulence factors in DNA obtained from phages [35]. Tomat et al. used PCR to detect virulence factor

Augustine et al., also used PCR or multiplex PCR to perform a screening of virulence factors in DNA obtained from phages [35]. Tomat et al. used PCR to detect virulence factor genes (from diarrheagenic *E. coli* toxins) in two phages (DT1 and DT6) isolated from stool samples of patients with diarrhea [72]. genes (from diarrheagenic *E. coli* toxins) in two phages (DT1 and DT6) isolated from stool samples of patients with diarrhea [72]. Presently, full genome sequencing and analysis provide the key tool for taxonomic

Presently, full genome sequencing and analysis provide the key tool for taxonomic classification and for alerting the presence of "dangerous" genes that phage genomes may contain. We believe that it is necessary to sequence phage genomes to obtain information on the presence of antibiotic-resistant genes or virulence factors before determining their suitability for food applications. An outline with the steps followed for the isolation and characterization of phages for food biopreservation is shown in Figure 1. classification and for alerting the presence of "dangerous" genes that phage genomes may contain. We believe that it is necessary to sequence phage genomes to obtain information on the presence of antibiotic‐resistant genes or virulence factors before determining their suitability for food applications. An outline with the steps followed for the isolation and characterization of phages for food biopreservation is shown in Figure 1.

**Figure 1.** Steps followed for the isolation and characterization of phages*.* **Figure 1.** Steps followed for the isolation and characterization of phages.

DNA genomes of Caudovirales range in size from about 15 up to 500 kbp [173]. The study of the genome of phages is crucial today, but most investigations analyzed before to the last 10 years do not include the sequencing or annotation of these genomes. The complete genomes of phages are already included as a technique of characterization and phylogeny, but the in‐depth analysis of these genomes has only been carried out very recently; this even allows us to discover new subfamilies and new genera of phages in‐ DNA genomes of Caudovirales range in size from about 15 up to 500 kbp [173]. The study of the genome of phages is crucial today, but most investigations analyzed before to the last 10 years do not include the sequencing or annotation of these genomes. The complete genomes of phages are already included as a technique of characterization and phylogeny, but the in-depth analysis of these genomes has only been carried out very recently; this even allows us to discover new subfamilies and new genera of phages infecting food pathogens [43,125].

fecting food pathogens [43,125].

#### **5. Phage's Life Cycle**

*Molecules* **2021**, *26*, x FOR PEER REVIEW 12 of 25

To perpetuate themselves, phages must infect their host bacteria by binding to specific receptors on them. After injecting their nucleic acid into the bacterium's cytoplasm, phages can hijack the bacterium's cellular machinery to induce their own replication, through a process called the "lytic cycle", giving rise to hundreds or thousands of complete viral particles that will leave the cell after killing it (Figure 2). Alternatively, if the phage nucleic acid is inserted into the chromosome or within a plasmid of the bacterium, it can remain in a kind of dormant state known as the "lysogenic cycle," which will not produce new virus particles until conditions are favorable, or their genes are activated by some external stimulus. Lytic bacteriophages are the first choice to selectively kill bacteria in foods because lysogenic phages remain in the bacterial chromosome and will not multiply until the environment in which the bacterium is found allows for it, making lysogenic phages difficult to control. **5. Phage's Life Cycle** To perpetuate themselves, phages must infect their host bacteria by binding to spe‐ cific receptors on them. After injecting their nucleic acid into the bacterium's cytoplasm, phages can hijack the bacterium's cellular machinery to induce their own replication, through a process called the "lytic cycle", giving rise to hundreds or thousands of com‐ plete viral particles that will leave the cell after killing it (Figure 2). Alternatively, if the phage nucleic acid is inserted into the chromosome or within a plasmid of the bacterium, it can remain in a kind of dormant state known as the "lysogenic cycle," which will not produce new virus particles until conditions are favorable, or their genes are activated by some external stimulus. Lytic bacteriophages are the first choice to selectively kill bacteria in foods because lysogenic phages remain in the bacterial chromosome and will not mul‐ tiply until the environment in which the bacterium is found allows for it, making lyso‐ genic phages difficult to control.

**Figure 2.** Gram‐negative bacterium after lysis by phages. Numerous complete or incomplete phage heads and tails can be seen in the image. Inset: Detail of the boxed area showing two phages of the *Siphoviridae* family. Original magnification ×25,000. **Figure 2.** Gram-negative bacterium after lysis by phages. Numerous complete or incomplete phage heads and tails can be seen in the image. Inset: Detail of the boxed area showing two phages of the *Siphoviridae* family. Original magnification ×25,000.

#### *6.* **Enzybiotics 6. Enzybiotics**

There are three classes of bacterial cell wall hydrolases: animal lysozymes, bacterial autolysins, and phage lysins. All animal lysozymes share the ability to hydrolyze the β‐ (1,4)‐glycosidic bond between the alternating N‐acetylmuramic acid and N‐acetylglucosa‐ mine residues of the bacterial cell wall polymer called peptidoglycan. Their biological role is mainly antibacterial defense, but some lysozymes also work as food digestive enzymes in animal guts [174]. Bacterial cell wall hydrolases are involved in carefully remodeling the cell wall to maintain cell integrity but also participate actively in processes such as cell division, bacterial surface appendages' assembly, and the facilitation of bacterial secretion systems' stabilization [175,176]. Most of these autolysins are peptidoglycan hydrolases (PGHs) that can provoke bacterial autolysis, so their expression and activity need to be tightly regulated. There are three classes of bacterial cell wall hydrolases: animal lysozymes, bacterial autolysins, and phage lysins. All animal lysozymes share the ability to hydrolyze the β-(1,4) glycosidic bond between the alternating N-acetylmuramic acid and N-acetylglucosamine residues of the bacterial cell wall polymer called peptidoglycan. Their biological role is mainly antibacterial defense, but some lysozymes also work as food digestive enzymes in animal guts [174]. Bacterial cell wall hydrolases are involved in carefully remodeling the cell wall to maintain cell integrity but also participate actively in processes such as cell division, bacterial surface appendages' assembly, and the facilitation of bacterial secretion systems' stabilization [175,176]. Most of these autolysins are peptidoglycan hydrolases (PGHs) that can provoke bacterial autolysis, so their expression and activity need to be tightly regulated.

The third class of cell wall hydrolases are phage endolysins, enzymes that directly target bonds in the peptidoglycan of the bacterial cell wall. These so-called enzybiotics (for ENZYme antiBIOTICS) are synthesized at the end of the bacteriophage lytic cycle to lyse the bacterium they parasitize, producing a lysis "from within" in Gram-negative bacteria [177]. Most endolysins contain one or two enzymatically active domains (EAD) in the N-terminus (which cleave one of the bonds in the bacterial peptidoglycan) and one cell wall-binding domain (CBD) in the C-terminal region (which is involved in host bacterial recognition). Based on their EAD, enzybiotics can be broadly divided into three types: endopeptidases, amidases, and glycosidases.

On the other hand, in Gram-positive bacteria, endolysins are also able to lyse bacteria "from outside" during the phage adsorption at the bacterial surface [178,179].

Endolysins have an extensive structural variation and a diverse cleavage predilection for the molecules with glycosidic, amide, or peptide bonds present in the bacterial peptidoglycan [180,181]. The structure of endolysins can be either globular or modular. Globular endolysins are unique for phages infecting Gram-negative bacteria, whereas modular endolysins are found in phages with a Gram-positive host. Another class of phage enzymes is virion-associated peptidoglycan hydrolases which share a similar mode of action on the bacterial peptidoglycan [182–185]. A good example of these newly studied antibacterial molecules is the virion-associated peptidoglycan hydrolase HydH5 of *Staphylococcus aureus* bacteriophage vB\_SauS-phiIPLA88 [186]. Additionally, some phages can produce depolymerases to overcome bacterial protective layers such as proteinaceous S-layers [187] or polysaccharide capsules [188].

Among the advantages of enzybiotics, we include the possibility of totally or partially breaking the structure of bacterial biofilms. A biofilm can be defined as a structured community of bacterial cells enclosed in a self-produced polymeric matrix and adherent to an inert or living surface. Growth in biofilms enhances the survival of bacterial populations in the food industry environments, increasing the probability of causing food-borne infections. Due to the presence of extracellular material that protects biofilms, many phages have limited access to bacteria inside these structures. This can be solved using phages expressing exopolysaccharide depolymerases and endolysins. Endolysins can act effectively irrespective of the metabolic status of the cells (exponential and stationary phase cells) and are capable of killing planktonic cells as well as sessile cells. In this way, phage endolysins have been shown to be effective in eliminating biofilms formed by tenacious pathogens on different surfaces commonly used in the food industry [189–192]. Moreover, endolysins can be evaluated in combination with depolymerases or even with antibiotics to kill the underlying pathogen that formed the biofilm. On the other hand, as many pathogens build their biofilms based on different substances that form the biofilm matrix, it would be advisable to evaluate the activity of endolysins against biofilms that present a different proportion of proteins, nucleic acids, sugars or lipids.

Additionally, endolysins can kill "persister" bacteria that escape conventional antibiotics and even can kill the dreaded multi-resistant strains. Although there are not many studies in this regard, endolysins also offer the possibility of being used in combination with other molecules or with other solutions for the food industry, such as bacteriocins or probiotics. Furthermore, as gene-encoded proteins, enzybiotics are amenable to bioengineering strategies, both to optimize specificity and to increase yields [193,194]. An example is the construction of hybrid proteins consisting of LysSA11 -an endolysin of the *S. aureus* phage SA11 and the enzymatically active domain of LysB4- and endolysin from the *Bacillus cereus* phage B4 [195].

The search, characterization, and practical use of these phage-derived lysins have received less attention than phages, basically because they are more difficult for many laboratories to study. However, there is a growing body of work on these enzymes, particularly in the field of human and animal pathogens, which has encouraged researchers in other fields, including food safety, to begin promising work with enzybiotics. Not surprisingly, many enzybiotics have been successfully tested as biopreservatives or have been proposed by their discoverers as good candidates to be used in food against Gramnegative and Gram-positive bacteria (Table 2). The study of these enzymes in phages that

do not belong to the "selective group" of food pathogens could provide a wide range of new proteins with different properties and varied spectra.

**Table 2.** Enzybiotics tested against food-borne pathogens and their proposed use in foods.


Furthermore, enzybiotics can improve the narrow host spectrum of phages against both Gram-positive and Gram-negative bacteria. Therefore, the narrow host range of phages should be used to control specific spoilage or pathogenic bacteria, while the broadest spectrum of enzybiotics can be used to control different strains or species. Some of the newly isolated and characterized endolysins have a broad spectrum so they could be candidates for use in the food industry. An example is endolysin M4Lys, which has a peculiar mosaic structure [222].

The main limitation in the use of phage enzybiotics in food is their complicated production and purification, since relatively large amounts of proteins are needed even to be studied in in vitro assays. Another problem is their low resistance to high temperatures used in different processes in the food industry, such as disinfection. However, the search for new enzymes with new properties will make it possible to find thermostable and easy-toproduce forms in heterologous hosts such as *E. coli* and *Lactococcus lactis* [21,221,223–225].

#### **7. Concluding Remarks**

Many natural and eco-friendly methodologies for food preservation have been proposed in the last few years, but only limited data are available about the usefulness of most of them under industrial scale conditions, which needs proper attention to satisfy the requirements of the industry as well as the demand of the consumers [226–230]. Consequently, studies about the ability of the reported biopreservative agents to control the development of undesirable microorganisms when applied at the industrial scale are greatly required.

Studies on the biocontrol of food-borne pathogens in foods have generally produced very good results. However, not all are lights in the use of phages against pathogenic bacteria in food, there are also shadows. There are assays in which it was not possible to reduce the number of pathogenic bacteria in food using bacteriophages [136,231,232].

The use of phages in human and veterinary medicine has received much more attention than their use in the food industry; but the increasing appearance of antibiotic-resistant strains in the food industry has begun to make these viruses be seriously taken into account when seeking their (application for food safety), also in this context. Similarly, their lytic enzymes have not been sufficiently exploited in the food industry to date. However, this is beginning to change; indeed, after the successful use of lysozyme (animal) or Nisin (bacteria), enzymes are beginning to be seriously valued in the food industry. Phages offer new and interesting possibilities when planning the control of annoying microorganisms in food manufacturing, food biopreservation, or food processing. Additionally, their lytic enzymes, easily modifiable through molecular biology processes, offer a very wide range of possibilities both for direct application against bacteria, as well as for inclusion in food matrices or the preparation of antibacterial surfaces generated by biotechnology [233].

Virulent bacteriophages are naturally present in foods, therefore both phages and their enzybiotics would be exploited in different ways for food safety as the consumer demand for the use of ecofriendly biopreservatives is increasing. Contamination of ready-to-eat products with pathogenic bacteria is a more serious problem than the contamination of food that will then be cooked before being consumed since many of the cooking methods reduce the number of these bacteria. In this context, both phage and enzybiotics have been tested in ready-to-eat meals. However, not only is the use of phages and their enzymes in food is not only an area of incipient research, but the whole biology of phages is experiencing a new boom in all domains of research, mainly in human and veterinary health, where spectacular achievements have already been reached in some patients and farm animals.

Along with this increasing amount of isolation and characterization of phage strains capable of controlling important food-borne pathogens—it is always desirable to increase our armament against superbugs—we must make a parallel effort to understand more indepth their interaction with target pathogens, as well as their biology and ecology in food if we want to apply them in the different stages of the production chain, increasing their biopreservation capacity. At the molecular level, we must better characterize enzybiotics, study the possibility of applying them in different processes, and optimize their production so that their application is profitable for food producers and does not raise the price too much for consumers.

Furthermore, the safety and ubiquity of phages must be well explained to both food producers and consumers to avoid rejection of "the unknown" [234,235]. Bacteriophages are the most abundant microorganisms on the planet and even in our guts, with approximately 10<sup>14</sup> phage particles in our body [236]. As we have seen in this review, phages and their enzybiotics can be found in the environment, in animals, and in food we eat every day. Finally, some phage-based products for the control of pathogens in food are already being used in different countries after being approved by competent authorities, even in ready-to-eat products. Those products mainly include a cocktail of phages, for example against *E. coli* (EcoShield™), *L. monocytogenes* (ListShield™ and PhageGuard Listex™), and *Salmonella* spp. (SalmoFresh™) [237].

**Author Contributions:** Conceptualization, J.R.-V.; writing—original draft preparation, J.R.-V., M.E.-Z., M.L.S.; writing—review and editing, J.R.-V., M.E.-Z., M.L.S., A.P.B.; Visualization: T.Y.F.-H.; supervision F.G., M.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** Research in our group was supported by SODERCAN (Project RH20-XX-032, FAGOFOOD).

**Acknowledgments:** Tamara Y. Forbes-Hernández is supported by a "Juan de la Cierva-Formación" post-doctoral contract.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study, or in the writing of the manuscript.

#### **References**

