Effect of Lactobacteria on Bioactive Peptides and Their Sequence Identification in Mature Cheese

An in silico study that featured the effect of starter cultures on the bioactivity and other health benefits of peptides in semi-hard cheese is presented in this contribution. Model Caciotta-type cheese samples were obtained in laboratory conditions in two variations. Sample A included starter cultures of Lactococcus lactis subsp. lactis and Lactococcus lactis subsp. cremoris. Sample B included starter cultures of Lactococcus lactis subsp. lactis, Lactococcus lactis subsp. cremoris, and a culture of lactobacilli Lacticaseibacillus casei. The in silico method showed that the peptides inhibited angiotensin-converting enzymes (ACE) and ipeptidyl peptidase IV (DPP-4), as well as possessed antioxidant properties. Lactococcus lactis subsp. lactis and Lactococcus lactis subsp. cremoris had a greater effect on the formation of bioactive peptides.

Dairy products are consumed in all parts of the world, and the dairy sector is the most prominent market of functional food with global prospects. Safe and live microbes can be fermented with specially cultivated strains and obtain health-promoting properties that reduce the risk of certain diseases if these functional products are part of one's daily diet. Cheese is one of such dairy products [21][22][23][24][25]. Cheese is a fresh or mature dairy product that is obtained by coagulating milk with enzymes, microorganisms, acids, etc. This product is an integral part of almost all traditional cuisines, and its historical role in human diet can hardly be overestimated. Cheese is easy to digest and rich in nutrients, which makes it an important and versatile source of proteins, short-chain fatty acids, vitamins, and minerals, depending on the region [26][27][28][29][30].
Biochemical reactions that occur in semi-hard and hard cheeses during ripening shape their sensory profile. These reactions result from the metabolism of lactic acid bacteria (LAB) introduced as starter cultures during the production process, as well as bacteria that are initially present in raw milk or may enter it from the environment. The nutrient Table 1. Main technological parameters for the production of laboratory samples of cheese Caciotta. ).

Starting Components Basic Cheese-Making Parameters
The standard Caciotta cheese has a protein/fat ratio of 0.92. Table 1 presents the main technological parameters applied in the laboratory production of of cheese Caciotta.  . Mature Caciotta-type cheese. Figure 1. Mature Caciotta-type cheese.

Micrographs of the Lactobacilli
Micrographs of the lactobacilli were obtained using a scanning electron microscope with systems for an energy-dispersive microanalysis, a wave dispersion microanalysis, and a Nova NanoSEM 450 backscattered electron diffraction analysis system (Czech Republic) [51].
The test sample with a volume of about 1-5 g was placed in a liquid dispersion module (volume 500 mL). The measurement started automatically as soon as the absorbance value reached the specified value.

Protein Analyses
The protein mass fraction was determined using a Rapid N Cube total nitrogen (protein) analyzer by the Dumas method: after burning the sample, the total nitrogen was registered by a thermal conductivity detector [52].

The Molecular Weight Distribution
The molecular weight distribution was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-Na) [51]. Proteins were fractioned in denaturing polyacrylamide gel (separating 12% and focusing 4%) with 0.1% of SDSNa. The electrophoresis procedure was performed at a single buffer with the addition of 0.1% SDS-Na at 15 mA. The gel was dyed with 0.2% of Kumassi R-250 dye (prepared in glacial acetic acid) and then rinsed three times with distilled water. Gel visualization and analysis were performed using the Gel Doc XR+ Gel Documentation System. The molecular weight was calculated using the Peptide Mass Calculator (http://rna.rega.kuleuven.be/masspec/pepcalc.htm (accessed on 1 June 2022)).

Amino Acid Analyses
The amino acid sequence of the hydrolysate peptides was determined by the method of matrix-activated laser desorption/ionization on a MALDI Biotyper (Bruker), equipped with a UV laser (Nd) in the positive ion mode using a reflectron; the accuracy of the measured monoisotopic masses after the additional calibration by trypsin autolysis peaks was 0.005% (50 ppm). The spectra were obtained in the mass range of 600-5000 m/z, choosing the laser power that was optimal for reaching the best resolution. To obtain the fragmentation spectra, the tandem device mode was used; the measurement accuracy of fragment ions was no lower than 1 Da. [53].

Biological Activity of Peptides In Silico
The peptide bioactivity was assessed in silico using the PeptideRanker online server (http://distilldeep.ucd.ie/PeptideRanker/ (accessed on 5 June 2022)), which ranks peptides according to their potential biological activity [54].

Visualization of Dihedral Amino-Acid Angles
The visualization of the dihedral angles ϕ against ψ of amino acid residues in the protein structure was performed using the SWISS-MODEL resource (https://swissmodel. expasy.org/interactive (accessed on 8 June 2022).) according to the simulated Ramachandran Maps [55].

D protein Structure Modeling
The structure modeling stage employed the SWISS-MODEL service (https://swissmodel. expasy.org/interactive (accessed on 10 June 2022)), which includes the SWISS-MODEL repository and the SWISS-MODEL interactive workspace. This automated protein structure homology modeling platform generates 3D protein models using a comparative approach and a database of existing models for key reference proteomes based on UniProtKB [55].

Modeling the Structure of Peptides
The structure of the peptides was modeled using the PepDraw online tool (http:// www2.tulane.edu/~biochem/WW/PepDraw/ (accessed on 15 June 2022)). The MBPDB database of bioactive milk peptides made it possible to identify and determine the protein and the properties of the peptides [56].

Results
The model cheese samples were synthesized according to the technology described in Section 2.1. The model cheeses (250 ± 20 g) ripened at 12-14 • C and relative humidity of 80-85%. Their chemical composition and active acidity were determined on day 90 ( Table 2). It is shown in Table 1 that the chemical composition of the model cheeses was almost the same. However, the active acidity was lower in the experimental samples with Lac. casei (0.5% + 0.5%). The biochemical properties of LAB include acid formation energy, limiting acidity, ability to ferment citric acid salts, curd quality, proteolytic activity, etc. [24]. Lactic acid streptococci have different activities. L. lactis was the first microorganism isolated in pure culture (in 1873 by Lister). L. lactis subsp. lactis is homofermentative bacteria. During ripening, these bacteria ferment glucose via the fructose bisphosphate pathway also known as the Embden-Meyerhof-Parnas (E.M.P.) pathway, which is similar to that of alcohol. Pyruvate, however, does not decarboxylate to acetaldehyde, like in alcoholic fermentation: it is used directly as an electron (hydrogen) acceptor. D-lactate dehydrogenase in LAB marks the formation of D(-)-lactic acid, while L-lactate dehydrogenase marks L(+)-lactic acid. DL-lactic acid is determined by the synthesis of two lactate dehydrogenases of different stereospecificity accompanied by the L(+)-lactic acid formation. They are strong acid formers and exhibit proteolytic activity during cheese ripening [58][59][60][61].
Unlike L. cremoris neither ferment maltose and dextrin nor de-aminates arginine. At low cultivation temperatures (15-20 • C), some strains form a significant number of volatile acids. The energy of acid formation in L. cremoris is weaker than in L. lactis. Lac. casei is homofermentative and ferments lactose, releasing mostly lactic acid; however, these bacteria develop slowly in milk. Lac. casei has prominent saccharolytic properties. It ferments fructose, galactose, mannitol, mannose, raffinose, ribose, salicin, sorbitol, trehalose, esculin, etc. Glucose is fermented without gas formation. Lactobacilli produce a number of hydrolytic enzymes, e.g., lactase, which breaks down lactose (milk sugar) and prevents lactase deficiency [62,63]. Lac. casei is found in various cheeses, especially at the late ripening stages. Lac. casei can form chains with different numbers of cells and produce gas from sodium citrate [64]. LAB is shown in Figure 2.
Unlike L. cremoris neither ferment maltose and dextrin nor de-aminates arginine. At low cultivation temperatures (15-20 °C), some strains form a significant number of volatile acids. The energy of acid formation in L. cremoris is weaker than in L. lactis. Lac. casei is homofermentative and ferments lactose, releasing mostly lactic acid; however, these bacteria develop slowly in milk. Lac. casei has prominent saccharolytic properties. It ferments fructose, galactose, mannitol, mannose, raffinose, ribose, salicin, sorbitol, trehalose, esculin, etc. Glucose is fermented without gas formation. Lactobacilli produce a number of hydrolytic enzymes, e.g., lactase, which breaks down lactose (milk sugar) and prevents lactase deficiency [62,63]. Lac. casei is found in various cheeses, especially at the late ripening stages. Lac. casei can form chains with different numbers of cells and produce gas from sodium citrate [64]. LAB is shown in Figure 2.

Lc. lactis subsp. lactis
Lc. lactis subsp. cremoris Lac. casei The change in casein fractions during the proteolysis of milk proteins was determined in the model cheese samples by the electrophoresis in polyacrylamide gel ( Figure  3).  The proteolytic activity during cheese ripening depends on several factors, such as type of coagulant, native milk microbiota and starter cultures, residual effect of the coagulant and native milk proteases, which can be affected by the moisture content in the cheese, its temperature, and relative humidity, ripening conditions, etc. Electrophoresis is one of the most common observation methods for cheese ripening. It can detect various casein fractions and protein breakdown products throughout the whole ripening process [65]. Fractions of α-casein are more susceptible to proteolysis, while degradation of β-casein occurs much less frequently [66]. Electrophoresis methods separate proteins by molecular weight and compare the staining intensity of polypeptide chains in the polyacrylamide gel.
Clear casein bands with different levels of hydrolysis throughout the ripening period, which indicates that the process involved several factors are shown in Figure 3. Although in Figure 3 are clear bands of various casein fractions, the fractions with the lowest molecular weight prevailed both on day 60 and day 90 of ripening. All samples demonstrated peptides with a molecular weight of 1.1-14.5 KDa. As for the sensory evaluation, cheese B had a more pronounced taste and aroma on day 90 due to Lac. casei The proteolytic activity during cheese ripening depends on several factors, such as type of coagulant, native milk microbiota and starter cultures, residual effect of the coagulant and native milk proteases, which can be affected by the moisture content in the cheese, its temperature, and relative humidity, ripening conditions, etc. Electrophoresis is one of the most common observation methods for cheese ripening. It can detect various casein fractions and protein breakdown products throughout the whole ripening process [65]. Fractions of α-casein are more susceptible to proteolysis, while degradation of β-casein occurs much less frequently [66]. Electrophoresis methods separate proteins by molecular weight and compare the staining intensity of polypeptide chains in the polyacrylamide gel.
Clear casein bands with different levels of hydrolysis throughout the ripening period, which indicates that the process involved several factors are shown in with a molecular weight of 1.1-14.5 KDa. As for the sensory evaluation, cheese B had a more pronounced taste and aroma on day 90 due to Lac. casei in its starter culture. After 90 days of ripening, both cheeses were tested for peptide sequences and their bioactivity (Table 3).    Eighteen potentially bioactive peptides out of 115 peptide sequences found in cheeses A and B are shown in Table 3. Their bioactivity ranged from 0.547239 to 0.870583 units. The results of the assessment of bioactive properties using the database of bioactive milk peptides MBPDB confirmed the biologically active properties of the identified peptide sequences. Further studies were carried out using the in silico method using the online resources presented in Sections 2.6-2.10. Further research featured the effect of the cultures on the protein structure. In Figures 4 and 5 are the Ramachandran plots that visualize the dihedral angles of the polypeptide backbone (ψ and ϕ) in proteins.
Each point on the Ramachandran plots represents one amino acid. In a polypeptide, the backbone bonds rotate relatively freely. These rotations are represented by torsion angles Phi (ϕ) and Psi (ψ), respectively. The white areas correspond to conformations where the polypeptide atoms come closer than the sum of their van der Waals radii. These regions are sterically forbidden for all amino acids except glycine, which is unique in that it lacks a side chain. Glycine molecules were observed in both cheese samples. Dark green areas correspond to conformations without steric collisions, i.e., they are the regions allowed for the arrangement of amino acids, namely the α-helix and β-helix conformations. Light green areas show the regions that are allowed in case the van der Waals radii are slightly shorter, i.e., the atoms can move a little closer together. This additional region corresponds to the left α-helix.
H-histidine; I-isoleucine; K-lysine; L-leucine; M-methionine; N-asparagine; P-proline; Q-glutamine; R-arginine; S-serine; T-threonine; V-valine; W-tryptophan; Y-tyrosine; Bold font indicates potential bioactive peptides Eighteen potentially bioactive peptides out of 115 peptide sequences found in cheeses A and B are shown in Table 3. Their bioactivity ranged from 0.547239 to 0.870583 units. The results of the assessment of bioactive properties using the database of bioactive milk peptides MBPDB confirmed the biologically active properties of the identified peptide sequences. Further studies were carried out using the in silico method using the online resources presented in Sections 2.6-2.10 Further research featured the effect of the cultures on the protein structure. In Figures 4 and 5 are the Ramachandran plots that visualize the dihedral angles of the polypeptide backbone (ψ and φ) in proteins.   L-amino acids cannot form extended sections of the left helix, but individual residues sometimes adopt this conformation. As a rule, they are represented by glycine, but they can also be asparagine or aspartate if the side chain forms a hydrogen bond with the main chain and, therefore, stabilizes this otherwise unfavorable conformation. The forbidden regions for amino acids in the model space are usually associated with the steric hindrance between the C-β-methylene group of the side chain and the atoms of the main chain. Glycine has no side chain and can take Phi (ϕ) and Psi (ψ) angles in all four quadrants of the Ramachandran plot. As a result, glycine is often found in the turning regions of proteins where any other residue would be sterically hindered.
The main areas are represented by the two dark green areas, while the three allowed areas are light green. The nuclear regions (dark green in Figures 4a and 5a) represent the most favorable combinations of ϕ and ψ and contain the highest number of points. The allowed areas (light green) either cluster around the main areas or detach from the main area. Nevertheless, they contain fewer data points than the main areas (dark green). The white areas are prohibited for amino acids.   Each point on the Ramachandran plots represents one amino acid. In a polypeptide, the backbone bonds rotate relatively freely. These rotations are represented by torsion angles Phi (φ) and Psi (ψ), respectively. The white areas correspond to conformations where the polypeptide atoms come closer than the sum of their van der Waals radii. These regions are sterically forbidden for all amino acids except glycine, which is unique in that it lacks a side chain. Glycine molecules were observed in both cheese samples. For a more detailed analysis, we modeled Ramachandran plots for glycine (B), preproline (C), and proline (D). These amino acids have different local steric hindrance properties and can take into account the effect of neighboring sequences. The proline side chain is covalently linked to the preceding N backbone. Thus, proline is more strictly prohibited than conventional residues. The residues, immediately before Pro (prePro), are limited as a result of the steric interaction with the proline ring. The other 16 types of amino acids prefer different regions, but their outer contours that separate the allowed regions from rogue ones match very closely. That is why they are all grouped together in the general case of distribution (Figures 4a and 5a). The general Ramachandran plot shows that the main amino acids in cheese sample B are within the right α-helix, since the points representing the location of the amino acids in the figure are presented at the lower left border of the plot. The amino acids of cheese A are concentrated in the upper left border, which means that this sample has a left-handed β-helix.
With the help of the SWISS-MODEL online service (https://swissmodel.expasy.org/ interactive (accessed on 10 June 2022)), they developed the protein structure models of the studied cheeses ( Figure 6).
Based on the obtained amino acid sequences, we determined the degree of hydrophilicity and hydrophobicity of proteins. Their graphs were built using the ProtScale online service based on the Kite and Doolittle scale (Figure 7).
In Figure 7a, it is illustrated that the peptides possess hydrophilic properties since most graph peaks on axis Y range from -2.8 to 0, which hints at the hydrophobic properties of the peptides on axis Y. Most hydrophobic regions have the following peptide sequences: TVDDKHYQKA and TESQSLTL. Most peaks are amino acids: aspartic acid, threonine, and serine. These amino acids have hydrophobic properties and 1-4 uncharged side radicals. Protein regions with such amino acid residues can hydrate and interact with other similar residues by hydrogen bonds. These peaks characterize the order of amino acids with hydrophobic properties. As for the control sample, the maximal value on the hydrophobicity scale belongs to threonine located in the 11-T region: it is 2.875 units. The highest hydrophilic properties belong to valine located in the 127-V region: it is 2.798.
pre-proline (C), and proline (D). These amino acids have different local steric hindrance properties and can take into account the effect of neighboring sequences. The proline side chain is covalently linked to the preceding N backbone. Thus, proline is more strictly prohibited than conventional residues. The residues, immediately before Pro (prePro), are limited as a result of the steric interaction with the proline ring. The other 16 types of amino acids prefer different regions, but their outer contours that separate the allowed regions from rogue ones match very closely. That is why they are all grouped together in the general case of distribution (Figures 4a and 5a). The general Ramachandran plot shows that the main amino acids in cheese sample B are within the right α-helix, since the points representing the location of the amino acids in the figure are presented at the lower left border of the plot. The amino acids of cheese A are concentrated in the upper left border, which means that this sample has a left-handed β-helix.
With the help of the SWISS-MODEL online service (https://swissmodel.expasy.org/interactive (accessed on 10 June 2022)), they developed the protein structure models of the studied cheeses ( Figure 6).
(a) Protein structure model of cheese A (b) Protein structure model of cheese B Figure 6. Protein structure models of cheese A and cheese B. Figure 6. Protein structure models of cheese A and cheese B. In Figure 7a, it is illustrated that the peptides possess hydrophilic properties since most graph peaks on axis Y range from -2.8 to 0, which hints at the hydrophobic properties of the peptides on axis Y. Most hydrophobic regions have the following peptide sequences: TVDDKHYQKA and TESQSLTL. Most peaks are amino acids: aspartic acid, threonine, and serine. These amino acids have hydrophobic properties and 1-4 uncharged side radicals. Protein regions with such amino acid residues can hydrate and interact with other similar residues by hydrogen bonds. These peaks characterize the order of amino acids with hydrophobic properties. As for the control sample, the maximal value on the hydrophobicity scale belongs to threonine located in the 11-T region: it is 2.875 units. The highest hydrophilic properties belong to valine located in the 127-V region: it is 2.798.
Cheese B (Figure 7b) has the following characteristic trait. The main part of the peaks in the hydrophobic region is in regions 131-154 and 198-202 with peptides CSEKLDQW, LCEKL, HAQQKEPIM, and IPNPI Q, respectively. Most peaks above 0 are represented Cheese B (Figure 7b) has the following characteristic trait. The main part of the peaks in the hydrophobic region is in regions 131-154 and 198-202 with peptides CSEKLDQW, LCEKL, HAQQKEPIM, and IPNPI Q, respectively. Most peaks above 0 are represented by amino acids Q, P, L, K, I, and E, which possess hydrophobic properties and 1-3 uncharged side radicals. Isoleucine (150-I) has a maximal hydrophobicity value of 2.00 units. Like in the control sample, the peptide sequences of the test sample have a predominantly hydrophilic nature.
The obtained data confirm the results published in [67,68] on the hydrophobic and hydrophilic properties of these amino acids. The next research stage featured the characteristics of potential bioactive peptides. Their structure was compiled using online servers. The characteristics of bioactive peptides are presented in Table 4.         The assessment in Table 4  The assessment in Table 4 shows that the peptide samples are ACE inhibitors, DPP-4 inhibitors, and antioxidants. Based on the in silico studies, Lc. lactis subsp. lactis and Lc. lactis subsp. cremoris in the control sample had a greater effect on the development of bioactive peptides. Thus, the control sample contained thirteen bioactive peptide sequences, while the experimental sample with Lac. casei had only seven.
The sections of the bioactive peptides in cheeses A and B are shown in Figures 8 and  9.
* The in silico study was based on the PeptideRanker online service; ** according to the MBPDB database of bioactive milk peptides; *** the modeling was based on the PepDraw online service.
The assessment in Table 4 shows that the peptide samples are ACE inhibitors, DPP-4 inhibitors, and antioxidants. Based on the in silico studies, Lc. lactis subsp. lactis and Lc. lactis subsp. cremoris in the control sample had a greater effect on the development of bioactive peptides. Thus, the control sample contained thirteen bioactive peptide sequences, while the experimental sample with Lac. casei had only seven.
The sections of the bioactive peptides in cheeses A and B are shown in Figures 8 and 9.
* The in silico study was based on the PeptideRanker online service; ** according to the MBPDB database of bioactive milk peptides; *** the modeling was based on the PepDraw online service.
The assessment in Table 4 shows that the peptide samples are ACE inhibitors, DPP-4 inhibitors, and antioxidants. Based on the in silico studies, Lc. lactis subsp. lactis and Lc. lactis subsp. cremoris in the control sample had a greater effect on the development of bioactive peptides. Thus, the control sample contained thirteen bioactive peptide sequences, while the experimental sample with Lac. casei had only seven.
The sections of the bioactive peptides in cheeses A and B are shown in Figures 8 and  9.   The identified bioactive peptides had aromatic rings that were represented mainly by tryptophan, tyrosine, and phenylalanine. As for the peptide bioactivity in Table 3, peptide sequence SCDKFL with high peptide bioactivity 0.8444337 was found only in cheese B with Lac. casei. Peptide sequences MKPWI (0.853622) and PSGAW (0.870583) were found both in the control cheese A and the experimental cheese B. These sequences also demonstrated a high bioactivity of peptides. All these peptide sequences included such essential amino acids as lysine, isoleucine, leucine, phenylalanine, methionine, and tryptophan, which are beneficial for human health and must be included in the diet. All these sequences inhibit the angiotensin-converting enzyme (ACE). ACE inhibitors are The identified bioactive peptides had aromatic rings that were represented mainly by tryptophan, tyrosine, and phenylalanine. As for the peptide bioactivity in Table 3, peptide sequence SCDKFL with high peptide bioactivity 0.8444337 was found only in cheese B with Lac. casei. Peptide sequences MKPWI (0.853622) and PSGAW (0.870583) were found both in the control cheese A and the experimental cheese B. These sequences also demonstrated a high bioactivity of peptides. All these peptide sequences included such essential amino acids as lysine, isoleucine, leucine, phenylalanine, methionine, and tryptophan, which are beneficial for human health and must be included in the diet. All these sequences inhibit the angiotensin-converting enzyme (ACE). ACE inhibitors are responsible for multifactorial actions in the human body, e.g., they relax blood vessels, thus, reducing blood pressure. In medical practice, ACE inhibitors are known to reduce the hospitalization incidence for heart failure as they increase life expectancy, exercise tolerance, and life quality [1,24,69].

Conclusions
Bioactive peptides appear as a result of biochemical and microbiological reactions under the action of proteolytic enzymes and microorganisms in starter cultures during the cheese ripening. They can act as ACE inhibitors. Regular consumption of mature cheeses not only satisfies the need for protein and essential amino acids but also, under certain circumstances, makes it possible to reduce or avoid taking pharmacological drugs. This issue, however, requires further research in cooperation with medical scientists. The authors believe that cheese can be regarded as a functional product. Cheese has a long shelf life: the longer the period of its ripening, the more bioactive peptides and amino acids they accumulate. This research proves that the accumulation of bioactive peptides in different cheeses can be predicted depending on the strain of microorganisms in their starter cultures. Biopeptide studies open certain prospects because various strains of microorganisms used in the food industry, including cheese-making, are so beneficial that they can potentially replace some pharmacological preparations, thus realizing the health-via-food concept. Further research will feature different parameters and raw materials in cheese formulations.