Comparative Peptidomics Analysis of Fermented Milk by Lactobacillus delbrueckii ssp. bulgaricus and Lactobacillus delbrueckii ssp. lactis

Few studies have investigated the peptidomics of fermented milk by Lactobacillus delbrueckii. The aim of the present study was to interpret the peptidomic pattern of the fermented milk by five strains of L. delbrueckii ssp. bulgaricus and ssp. lactis prior to and after the simulated gastrointestinal digestion in vitro. The results indicated variations in the peptidomics among the samples, particularly between the samples of different subspecies. The peptides originating from β-casein were abundant in the samples of ssp. bulgaricus, whereas the peptides derived from αs1-casein and αs2-casein were more likely to dominate in those of ssp. lactis. For β-casein, the strains of ssp. bulgaricus displayed extensive hydrolysis in the regions of (73–97), (100–120), and (130–209), whereas ssp. lactis mainly focused on (160–209). The digestion appears to reduce the variations of the peptidomics profile in general. Among the five strains, L. delbrueckii ssp. bulgaricus DQHXNS8L6 was the most efficient in the generation of bioactive peptides prior to and after digestion. This research provided an approach for evaluating the peptide profile of the strains during fermentation and digestion.


Introduction
Lactobacillus delbrueckii, one of the thermophilic lactic acid bacteria (LABs), plays an important role in food fermentation, including ssp. bulgaricus and ssp. lactis, two subspecies isolated from milk niches. Previous studies showed that there were differences in carbohydrate metabolism between ssp. bulgaricus and ssp. lactis [1,2]. However, the characteristics of the peptidomic pattern and the profile of the bioactive peptides of the fermented milk from these two subspecies, as well as the influence of gastrointestinal digestion, have not been thoroughly investigated.
The proteolytic system provides peptides and free amino acids for the growth of strains during fermentation [3], and the cell-envelope proteinase (CEP) of LAB is essential for the hydrolysis of milk proteins [3][4][5]. The CEP of L. delbrueckii ssp. bulgaricus and lactis were classified as PrtB [6] and PrtL, respectively, with a unique amino acid substitution for substrate specificity at position 222 for PrtL and different hydrolysis patterns of the chromophoric peptides compared to PrtB [3]. How PrtB and PrtL from the two subspecies affect the peptide profile and generation of bioactive peptides of fermented milk has not been reported yet.
Milk fermentation by LAB is known for the generation of bioactive peptides [7,8] with different biological functions such as angiotensin-converting enzyme (ACE) inhibitors, antihypertensives, anti-inflammatories, antioxidants, and immunomodulators [9]. In particular, the lactotripeptides VPP and IPP were identified as ACE inhibitors and displayed the ability to reduce blood pressure in vivo; VPP and IPP have been detected in fermented milk by L. helveticus and could resist gastrointestinal digestion [10][11][12][13]. However, few studies have been conducted on the profile of peptides in fermented milk L. delbrueckii ssp. bulgaricus and lactis. Moreover, as gastrointestinal digestion affects the absorption and function of peptides, investigating the changes of the peptide profile during digestion are critical for evaluating the function of fermented milk.
Therefore, in the current study, the peptidomics of fermented milk by five strains of L. delbrueckii ssp. bulgaricus and ssp. lactis were investigated before and after the simulated gastrointestinal digestion in vitro in order to understand the pattern of peptide formation during fermentation and digestion.

Microorganisms and Culture Conditions
Three L. delbrueckii ssp. bulgaricus strains, DXJLHTS2M2, DQHXNS8L6, and 2038 and two L. delbrueckii ssp. lactis strains, D11M188 and ATCC12315, were used to ferment skim milk. All the strains were obtained from the Culture Collections of Food Microbiology, Jiangnan University (Wuxi, China).

Fermentation of Skim Milk
Fermented skim milks were manufactured using the previously referred to method with some modifications [14]. Skim milk was prepared by reconstituting 11% (w/w) skim milk powder. The bacterial strains were incubated into skim milk with an initial culture concentration of 1-5 × 10 6 CFU/mL and incubated at 37 • C for 12 h. The viable cell counts of L. delbrueckii strains were determined using the plating method. Then, 0.5 mL samples were added to 4.5 mL of 0.9% (w/v) sterile saline and diluted serially as appropriate. Plates were incubated at 37 • C for 48 h under anaerobic conditions. Colonies were counted on each plate, and live-cell counts were expressed as lg(cfu·mL −1 ). The pH was measured by a pH meter (Model 3100, Ohaus, Parsippany, NJ, USA). The experiments were conducted in three replications.

Preparation of Whey Fraction
Whey fraction of samples was prepared according to the previous method with moderate modifications [15]. The pH of 35 mL of the samples was adjusted to 4.6 and the supernatants were harvested after centrifugation at 10,000× g for 10 min at 4 • C and filtration with 0.45 µm nylon syringe filter. Then, ultrafiltration was applied using a membrane with a cut-off value of 10 kDa (EMD Millipore, Billerica, MA, USA), and the whey was stored at −20 • C prior to further analysis.

Simulated Gastrointestinal Digestion
The digestion process was conducted according to the previous study with slight modification [16]. The pH values of whey samples were adjusted to 2.0, and pepsin (Sigma Aldrich, St. Louis, MA, USA) was added at a ratio of 1:50 (w/w) enzyme to substrate. After digestion at 37 • C for 2 h, the pH was adjusted to 8 with 0.4 M NaOH, and then trypsin (Sigma Aldrich, St. Louis, MA, USA) was added to digest at 37 • C for 2 h. Then, samples were heated at 95 • C for 10 min to inactivate the enzyme activity. The supernatants of the digested samples were collected by centrifuge at 14,000× g for 5 min at 4 • C and filtered by a 0.45 µm nylon syringe filter. The digested samples were finally stored at −20 • C prior to further analysis.

Determination of Peptide Content
The peptide content of the samples was determined using Pierce TM Quantitative Colorimetric Peptide Assay (Thermo Fisher Scientific, Waltham, MA, USA). The absorbance was read at 480 nm and the results were expressed as mg·mL −1 .
The chromatography device was coupled to a Q-Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA), and the liquid chromatography (LC) eluent was electrosprayed from the end of the column at an applied voltage of 2.3 kV. MS analysis was performed in sensitivity, positive ions, and data-dependent analysis (DDA) modes, and MS data were collected in the 150-2000 m/z range with a scan time of 0.2 s. A maximum of 15 precursor ions with an intensity threshold of 10,000 counts was selected for ion trap collision-induced dissociation (CID) fragmentation and subjected to collision energy ramping from 8 V to 9 V for low mass and 40 V to 90 V for high mass. The MS/MS spectra were recorded on the 180 to 2000 m/z range with a scan time of 0.1 s.
Database searches via Maxquant (1.5.2.8; Available online: https://maxquant.net/ maxquant/ (accessed on 5 December 2014)) were performed using the UniProt databases (Available online: https://www.uniprot.org/ (accessed on 5 January 2021)) restricted to Bos taurus organism. A mass tolerance of 35 ppm and 3 missing cleavage sites as well as an MS/MS tolerance of 0.02 Da were allowed. Variable methionine oxidation was also considered. The relevance of protein and peptide identities was judged according to their identification generated by PEAKS Studio 7.0 (p < 0.05) and a false discovery rate < 1%.

Statistical Analysis
The mean ± standard deviation (SD) was used for the presentation of data, and multiple ANOVA was carried out with p < 0.05 considered a significant difference. Log10 scale transformation was used to carry out the statistical analysis of peptidomic, and principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were used, respectively, to visualize differences in peptidomic profiles between all samples and two subspecies groups in MetaboAnalyst (available online: https://www. metaboanalyst.ca (accessed on 30 September 2021)). Peptides detected in at least two of the three replicates were considered identified in samples, and the variable importance in projection (VIP) score was used to represent differences between two subspecies. The bioactivity of detected peptides was matched to the Milk Bioactive Peptide Database (available online: http://mbpdb.nws (accessed on 9 August 2021)). GraphPad Prism 8.0 (GraphPad Software, San Diego, CA, USA) was chosen to draw the figure.

Peptide Profile Analysis Revealed Variation among the Samples
The peptidomic structural differences of the skim milk, fermented milk samples, and digested samples are shown in Figure 1. For the samples of fermented milk, discrimination in the samples of two subspecies was observed, with skim milk clustered close to that of L. delbrueckii ssp. lactis, whereas the digestion appears to reduce the extent of discrimination of the fermented samples. The peptide content of the fermented milk of ssp. bulgaricus was generally higher than that of ssp. lactis, with the highest peptide content detected in the fermented milk of L. delbrueckii ssp. bulgaricus DXJLHTS2M2 and 2038 followed by DQHXNS8L6. (Figure S1), and variation in the pH value and the viable count was recorded among the samples (Table S1).

Comparative Peptidomic Analysis of Fermented Milk
The upset plot of the common and differential peptides of the fermented milk samples prior to and after digestion is shown in Figure 2. In total, 1266 peptides were detected in the samples, with 846 peptides found in the samples before digestion and 641 peptides found in the samples after digestion. The peptides ranged from 232 to 384 in the fermented milk samples, with 88 detected in skim milk. Among these peptides, 20-51% of the peptides originated from β-casein, followed by αs1-casein, αs2-casein, and κ-casein in general, as reported previously [17].  Figure 2a shows variation among the strains in the capacity of the generation of differential peptides, with the most abundant differential peptides (137) recorded in the fermented milk of ssp. lactis D11M188 and only 41 in that of ssp. bulgaricus DXJLHTS2M2.
After digestion, the number of peptides dramatically decreased in the samples of DXJLHTS2M2, 2038, and D11M188 ( Figure 2b). However, a 16% increase was observed in the digested sample of ATCC12315. Furthermore, the number of peptides derived from β-casein decreased with those from other proteins increased in the samples after digestion. The digested sample of ssp. lactis ATCC12315 possessed the most abundant differential peptides (87), while only 24 differential peptides were present in that of ssp. bulgaricus DXJLHTS2M2. In total, 28 peptides were shared by six digested samples, with 22 from the five digested samples of fermented milk. Comparison of the fermented samples was evaluated with PLS-DA, with R2X, R2Y, and Q2Y values of 0.979, 0.998, and 0.957, respectively ( Figure S2), confirming the accuracy and predictability of the model. A total of 369 differential peptides were identified with VIP scores > 1, and the top 20 in the fermented samples are shown in Figure 3b. The peptides originating from β-casein were frequently present in the group of ssp. bulgaricus, whereas peptides derived from αs1-casein and αs2-casein were more likely to dominate in the group of ssp. lactis as previously reported [4,18].
Variation was observed for the common peptides at the subspecies level, with 61% of peptides derived from β-casein for the three samples of ssp. bulgaricus and 70% from αs1-, αs2-, and κ-casein for two samples of ssp. lactis (Tables S2 and S3). The significant difference of Shannon index between the samples from two subspecies indicated that the strains of L. delbrueckii ssp. bulgaricus displayed more intensive proteolysis of milk proteins during fermentation than those of L. delbrueckii ssp. Lactis and digestion appears to significantly reduce the richness of the peptide profile (p < 0.05) (Figure 3c). Different LABs had different CEPs, which are responsible for the specificity for the substrate [19,20]. The CEP of L. delbrueckii ssp. bulgaricus was classified as PrtB [6], while PrtL of L. delbrueckii ssp. lactis has a unique amino acid substitution at position 222 for substrate specificity and different hydrolysis patterns of chromophoric peptides compared to PrtB [3]. A previous study indicated that both ssp. bulgaricus and ssp. lactis had similar hydrolysis activity against αs1-casein, αs2-casein, and β-casein based on the pattern of SDS-PAGE [3,18], which was inconsistent with the observation of the present study probably due to the variation in the experimental condition and analysis.
Comparison of the digested samples is shown in Figure 3d, in which R2X, R2Y, and Q2Y values (0.898, 0.991, and 0.688, respectively) are greater than 0.5 and could be used in recognition analysis ( Figure S3). Similarly, a total of 322 differential peptides (VIP > 1) were identified. The peptides of the top 20 VIP scores indicated variation between subspecies (Figure 3d), particularly for β-casein f (78-91), β-casein f (189-202), and β-casein f (60-68), with the first two peptides presented in digested samples of ssp. bulgaricus and the latter in ssp. lactis.

Caseins Cleavage Pattern Based on Peptidomic Analysis
Variation on the cleavage pattern of αs2and β-casein based on peptidomics was observed between the samples from the two subspecies in the heat map (Figure 4), in which ssp. bulgaricus displayed a stronger ability in hydrolyzing β-casein while ssp. lactis showed a preference to αs2-casein. For αs2-casein, regions of (95-128) and (142-207) were extensively hydrolyzed and lead to the generation of a large number of peptides (Figure 4a). Specifically, ssp. bulgaricus produced notable peptides from the region (95-115), whereas ssp. lactis generated more peptides from the region (142-207). Among these peptides from αs2-casein, 13 were unique in the samples of ssp. bulgaricus, while 12 were unique in ssp. lactis (Tables S2 and S3). Besides the different regions preferred by two subspecies, there were common cleavage regions such as (114-128) and (185-199) shared in two subspecies, which generated seven peptides. However, the number of peptides released from αs2-casein was smaller than that from β-casein during milk fermentation, probably due to the fact that αs2-casein possessed more α-helix and was located in the interior of the casein micelle, which limits the accessibility of the CEPs and peptidases [21,22].
It is worth noting that the C-terminal of the β-casein, a highly unstructured region, contributed a different pattern of hydrolysis between subspecies [22]. A previous study suggested that β-casein is an amphiphilic compound where the region (1-50) was hydrophilic and regions (153-175) and (187-209) were hydrophobic [23]. The results indicate that the CEPs of L. delbrueckii preferred hydrophobic regions of β-casein, similar to PrtS [20], with more extensive cleavage sites from ssp. bulgaricus than ssp. lactis during fermentation. No distinct profile between subspecies was observed for as1-casein ( Figure S4) and κ-casein ( Figure S5).

Pattern of Peptidomics of Fermented Milk after Digestion
As shown in Figure 4, the six digested samples had a very similar profile, which is consistent with our PCA analysis and the previous study [29]. As shown in Figure 4b, regions (95-115) and (142-207) of αs2-casein could be easily digested by pepsin and trypsin. Interestingly, after digestion, ssp. bulgaricus had some peptides remaining in the region (114-128) of αs2-casein, while ssp. lactis did not, which released a notable number of peptides through the fermentation of both subspecies and had a subspecies-based profile. In addition, the region (24-35) released several peptides in all fermented samples after digestion, such as αs2-casein f (25-33) and αs2-casein f (25-35). Alternatively, peptides from regions (73-97), (124-163), and (189-202) of β-casein still existed, while regions (100-120) and (164-188) diminished after the digestion of pepsin and trypsin (Figure 4b).
The long peptides, including the regions (202-209) of β-casein, were hydrolyzed into smaller peptides around arginine residues (position 202) at the C-terminal side (Figure 4b). This was probably due to the hydrolysis of trypsin as trypsin has been reported to generate peptides on the C-terminal side of lysine and arginine amino acid residues preferentially [30]. Otherwise, pepsin had an extensive cleavage specificity, which could cleave peptides with an aromatic acid on either side of the peptide bond, especially if the other residue is also an aromatic or a dicarboxylic amino acid, except for valine, alanine, or glycine linkages [31], but no distinct pattern could be found by the hydrolysis activity of pepsin.

Conclusions
In the present study, a difference was observed in the peptidomic feature of fermented milks before and after the simulated gastrointestinal digestion in vitro between subspecies, mainly originating from the hydrolysis of αs2and β-casein. Regarding αs2-casein, ssp. bulgaricus generated notable peptides in the region (95-115), whereas ssp. lactis formed more peptides and a more distributed pattern from the region (142-207). For β-casein, ssp. bulgaricus had more extensive cleavage distribution, including regions (73-97), (100-120), and (130-209), whereas ssp. lactis had a narrower distribution in the region (160-209) with fewer peptides. After digestion, L. delbrueckii ssp. bulgaricus DQHXNS8L6 had the most peptides, peptides resisting digestion, bioactive peptides, and short-sized peptides, together with robust growth and acidification capacity.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/foods10123028/s1, Figure S1: Peptide content of skim milk and fermented milks by Lactobacillus delbrueckii ssp. bulgaricus and Lactobacillus delbrueckii ssp. lactis after 12 h of fermentation at 37 • C before (black) and after (gray) simulated gastrointestinal digestion in vitro; Figure S2: PLS-DA score plot for fermented milks peptidomic of Lactobacillus delbrueckii ssp. bulgaricus and Lactobacillus delbrueckii ssp. Lactis; Figure S3: PLS-DA score plot for digested samples peptidomic of Lactobacillus delbrueckii ssp. bulgaricus and Lactobacillus delbrueckii ssp. Lactis; Figure S4: αs1-casein heat maps constructed by peptides present in all undigested and digested samples, respectively. Heat maps under the sequences indicated the frequency of the amino acids; Figure S5: κ-casein heat maps constructed by peptides present in all undigested and digested samples, respectively; Table S1: pH value and viable count of fermented milks by Lactobacillus delbrueckii ssp. bulgaricus and Lactobacillus delbrueckii ssp. lactis after 12 h fermentation under 37 • C.

Data Availability Statement:
The datasets generated for this study are available on request to the corresponding author.