Anti-Salmonella Activity and Peptidomic Profiling of Peptide Fractions Produced from Sturgeon Fish Skin Collagen (Huso huso) Using Commercial Enzymes

This study investigated peptide fractions from fish skin collagen for antibacterial activity against Escherichia coli and Salmonella strains. The collagen was hydrolyzed with six commercial proteases, including trypsin, Alcalase, Neutrase, Flavourzyme, pepsin and papain. Hydrolyzed samples obtained with trypsin and Alcalase had the largest number of small peptides (molecular weight <10 kDa), while the hydrolysate produced with papain showed the lowest degree of hydrolysis and highest number of large peptides. Four hydrolysates were found to inhibit the growth of the Gram-negative bacteria, with papain hydrolysate showing the best activity against E. coli, and Neutrase and papain hydrolysates showing the best activity against S. abony; hydrolysates produced with trypsin and pepsin did not show detectable antibacterial activity. After acetone fractionation of the latter hydrolysates, the peptide fractions demonstrated enhanced dose-dependent inhibition of the growth (colony-forming units) of four Salmonella strains, including S. abony (NCTC 6017), S. typhimurium (ATCC 13311), S. typhimurium (ATCC 14028) and S. chol (ATCC 10708). Shotgun peptidomics analysis of the acetone fractions of Neutrase and papain hydrolysates resulted in the identification of 71 and 103 peptides, respectively, with chain lengths of 6–22 and 6–24, respectively. This work provided an array of peptide sequences from fish skin collagen for pharmacophore identification, structure–activity relationship studies, and further investigation as food-based antibacterial agents against pathogenic microorganisms.


Introduction
The increasing prevalence of bacterial resistance to commercial antibiotics leading to failed treatments and increased expenses has become a public health concern [1]. Moreover, the increasing constraint on the use of chemical preservatives and the growing demand for natural antimicrobial agents in foods or food supplements has spurred emerging research towards the discovery of novel natural antimicrobial agents with a broad spectrum of antibiotic activity [2]. Consequently, the development of sustainable alternatives to synthetic antibiotics has become a research priority. Antimicrobial peptides (AMPs) are essential components of the innate immune response and a diverse group of individually unique

Enzymatic Hydrolysis of Fish Skin Collagen
Collagen was isolated from sturgeon fish skin as previously reported [26]. Lyophilized pepsin-solubilized collagen (1 g) was suspended in 200 mL of deionized water and hydrolyzed separately with six commercial enzymes at their optimal conditions (Table 1) with continuous stirring. During enzymatic hydrolysis, the pH of the mixtures was adjusted by adding 1 M NaOH or 1 M HCl. After 3 h of hydrolysis, the enzymatic reaction was terminated by changing pH as shown in Table 1 to inactivate each of the enzymes. The reaction mixtures were then centrifuged at 10,000× g for 30 min at 4 • C and the supernatants lyophilized using a freeze-drier to yield the collagen hydrolysates and stored at −20 • C until use. DH of the collagen hydrolysates was determined in a 96-well plate using the Ophthaldialdehyde (OPA) method described by Nielsen et al. [27], with some modifications. Briefly, 30 µL sample solution, standard (serine) or blank control (deionized water) was mixed with 225 µL OPA reagent, and after 2 min shaking at room temperature, absorbance the mixtures were measured at 340 nm using a microplate reader. Thereafter, DH was calculated as previously reported [28].

SDS Polyacrylamide Gel Electrophoresis (SDS-PAGE)
The molecular weight profile of the collagen and collagen hydrolysates was determined by SDS-PAGE according to the method of Laemmli [29], with a slight modification, using gradient resolving gel (6%, 9%, 12%, 15%, and 18%) with 3% (w/v) stacking gel. To prepare the samples, collagen and collagen hydrolysate powders (1 mg/mL) were dissolved in 0.02 M sodium phosphate buffer (pH 7.2) containing 1% (w/v) SDS and 3.5 M urea, and the solutions were gently stirred at 4 • C for 12 h. The mixtures were centrifuged at 5000× g for 5 min to remove undissolved debris. Thereafter, the samples were mixed with the sample buffer (2% SDS, 20% glycerol, and 0.5% bromophenol blue in 62.5 mM Tris-HCl buffer, pH 6.8, containing 1 M β-mercaptoethanol), followed by heat denaturation at boiling temperature for 5 min. Electrophoresis was conducted at a constant current at 120 V for 2 h, and the gel was stained overnight with Coomassie Brilliant Blue R-250 solution and distained with Milli-Q water by shaking for 4 h. A standard molecular weight protein marker ranging from 10 to 250 kDa was used to estimate the molecular weight of the collagen peptides. Image scanning of gels was done using the ChemiDoc Imaging System (BioRad Inc., Mississauga, ON, Canada).

Surface Hydrophobicity (H o )
Surface hydrophobicity (H o ) of the collagen hydrolysates was determined by the fluorescence method using ANS as the hydrophobic probe. Samples at concentrations ranging from 0.005 to 0.025% were mixed with 20 µL ANS in a Grenier UV-Star (96-well) microplate. Fluorescence of the mixture was then measured at excitation and emission wavelengths of 390 and 470 nm, respectively, using a Spark multimode microplate reader (Tecan, Switzerland). The slope of the fluorescence intensity vs. concentration plot was used to represent the surface hydrophobicity (H o ), as previously reported [30].

Dynamic Light Scattering (DLS) Analysis
Zeta (ζ)-potential and mean particle size of the collagen hydrolysates were evaluated by DLS using the Zetasizer Nano Series Nano-ZS (Malvern Instruments Ltd., Malvern, UK) at 25 • C. The hydrolyzed samples (0.05 mg/mL) were mixed in water and used to determine the zeta potential and particle size of the samples. All measurements were taken in triplicates at 25 • C with the Smoluchowski model at F (ka) 1.50 and backscattered angle of 173 • .

Solvent Fractionation of the Hydrolysates
Selected collagen hydrolysates were further subjected to solvent fractionation to separate the bioactive peptides as previously reported [12]. High-molecular-weight peptides in the samples were precipitated using ice-cold acetone (50%, v/v) followed by centrifugation at 8000× g for 20 min at 4 • C. Thereafter, acetone in the supernatants containing low-molecular-weight peptides was evaporated using a centrifugal vacuum evaporator, and the samples were freeze-dried to obtain the peptide fraction powders.

Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) and Peptidomics Data Analysis
Shotgun proteomics was used to identify peptides in the collagen peptide fractions. Lyophilized samples were separated by a 60-min gradient elution at a flow rate of 250 nL/min using an EASY-nLC integrated nano-HPLC system (Thermo Fisher, San Jose, CA, USA), which was directly interfaced with a quadrupole Orbitrap (Q-Exactive) mass spectrometer (Thermo Fisher, San Jose, CA, USA). The analytical column used was a PepMap RSLC EASY-Spray column (75 µm × 50 cm) packed with C18 resin (2 µm). Eluted peptides were then introduced into the Orbitrap Q-Exactive mass spectrometer, operated in the data-dependent acquisition mode using the MaxQuant software with a single full-scan spectrum (400-1500 m/z, 70,000 resolution) followed by 10 data-dependent MS/MS scans in the Orbitrap mass analyzer. For peptide identification, the spectral data were processed with MaxQuant version 1.6.10.43 software (Max Planck Institute of Biochemistry, Planegg, Germany) using the Andromeda peptide search engine, as reported by Tyanova et al. [31].

Antibacterial Activity of Collagen Peptides
Antibacterial activity of the collagen hydrolysates and the acetone fractions was performed using a microtest polystyrene 96-well microplate (VWR Tissue Culture Plates, Randor, PA, USA) as previously described [32]. Briefly, the microplate wells were filled by distributing 100 µL of BHI medium. Medium alone and medium with each strain inoculum were used as blank and control, respectively, on the same microplate. Then, 100 µL of each collagen hydrolysates and acetone fractions prepared in the medium were added to each well (Column 3 to Column 12) and 2-fold diluted with the medium starting from 100 mg/mL for collagen peptides and from 10 mg/mL for acetone fractions. The wells, containing 100 µL of media or sample solution, were inoculated with 100 µL overnight culture of the target strains diluted to a final concentration of 10 5 CFU/mL. The microplates were subsequently incubated at 37 • C, and absorbance at 650 nm was measured at 20 min intervals for 24 h, using the Spark multimode microplate. Viable bacterial strains were quantified after 6 h incubation with the acetone fractions or BHI medium as control using the standard plate counting method on BHI agar and expressed as CFU/mL.

Statistical Analysis
Data were collected in triplicate and analyzed using SPSS (Chicago, IL, USA). Results were expressed as mean ± standard deviation, and the differences between mean values of different samples were determined by the Least Significant Difference (LSD) test (p < 0.05).

Degree of Hydrolysis and Surface Hydrophobicity of the Hydrolysates
DH of the collagen hydrolysates produced with different enzymes (trypsin, Alcalase, Neutrase, Flavourzyme, pepsin, and papain) at their optimal temperature and pH conditions are shown in Table 1. Enzymes were used at the same enzyme-substrate ratio to compare their hydrolytic efficiencies. As shown in Table 1, trypsin and papain hydrolysate demonstrated the highest and lowest DH (p < 0.05), respectively, in producing the collagen hydrolysates. Interestingly, there was no significant difference between the DH of hydrolysates obtained with the single enzyme, pepsin and the multi-enzyme, Flavourzme. In contrast, the highest surface hydrophobicity was observed for collagen hydrolysate produced with papain, whereas hydrolysates obtained with Alcalase and trypsin had similar and the lowest H 0 values ( Figure 1A). There was no defined pattern in the DH or H 0 with respect to the origin of the enzymes (plant, animal or microbe) used in collagen hydrolysis.

Molecular Weight Distribution of the Collagen Hydrolysates
The molecular weight profiles of the fish skin collagen and its hydrolysates, determined by SDS-PAGE, are shown in Figure 1B. After treating collagen with the six commercial enzymes, the samples obtained with trypsin (lane C) and Alcalase (lane D) displayed peptide bands with the lowest molecular weights (<10 kDa). In contrast, hydrolysates obtained with the pepsin (lane G) and papain (lane H) showed bands with high molecular weights ranging from~100 to <10 kDa. The molecular weight profile follows a similar pattern as the DH (Table 1); hydrolysates with higher DH showed lower molecular weight bands on the SDS-PAGE gels, indicating extensive protein hydrolysis by the enzymes, and vice versa. Furthermore, SDS-PAGE showed the disappearance of the two major bands present in the original collagen samples (lanes A and B) in all the hydrolysates (lanes C-H), indicating that all six enzymes hydrolyzed the parent protein, although to different extents.

Surface Charge and Particle Size of the Collagen Hydrolysates
As shown in Figure 1C,D, the collagen hydrolysate particles had a net negative surface charge when dispersed in water. The magnitude of zeta potential for the hydrolysates produced with Alcalase and trypsin was significantly higher than the rest, and papain had the lowest surface charge. Conversely, similar particle sizes were observed for the hydrolysates produced with Alcalase, Flavourzyme, and pepsin, while that produce with papain had the highest particle size, and those produced with trypsin and Neutrase showed the lowest values. The particle size results are inversely correlated to the DH (Table 1) and surface charge results ( Figure 2C).

Antibacterial Activity of the Collagen Hydrolysates and Peptide Fractions
The collagen hydrolysates and their acetone fractions were evaluated for inhibitory activity against E. coli and four Salmonella strains using microdilution assay. As shown in Figures 2 and 3, some collagen hydrolysate at a concentration of 100 mg/mL moderately inhibited the growth of E. coli and S. abony strains, with crude hydrolysates obtained with papain, Alcalase, and Neutrase being the most potent. Specifically, hydrolysate produced with papain inhibited the growth of both strains, while Neutrase and Alcalase produced hydrolysates with selective antibacterial activity against E. coli and S. abony, respectively. To a lesser extent, collagen hydrolysate obtained with Flavourzyme exhibited a weak growth inhibitory effect against S. abony. Collagen hydrolysates produced with pepsin and trypsin did not show detectable antibacterial activity in both strains under our experimental conditions. Fractionation using acetone increased the specific activity of inhibitory substances in most samples. After 6-h incubation, the acetone fractions of hydrolysates produced with Neutrase and papain exhibited the highest and dose-dependent anti-Salmonella growth inhibitory activities (Figure 4). In contrast, fractions from hydrolysates produced with Alcalase and Flavourzyme showed less potency against the four Salmonella strains (data not shown). Neutrase produced the hydrolysate with the most potent activity against the four Salmonella strains, with the highest growth inhibition of 90% observed at 10 mg/mL against S. typhimurium ATCC 13311. The anti-Salmonella activity results were confirmed using the viable bacterial cell counts (CFU) in the presence of acetone peptide fractions of collagen hydrolysates produced with Neutrase and papain (10 mg/mL) after 6 h incubation (Table 2). When treated with Neutrase peptide fraction at 10 mg/mL, the CFU counts of the Salmonella strains were significantly (p < 0.05) reduced by more than two logs, with S. typhimurium (ATCC 13311) being the most sensitive strain (2.28 log reduction). The Salmonella strains were less, but significantly, affected by the papain peptide fractions with a log reduction range of 0.32-1.26. Control, bacteria in BHI medium without the peptide treatment; Neutrase and Papain, bacterial treated with 10 mg/mL of acetone fraction of collagen hydrolysates produced with Neutrase and papain, respectively. Different letters in each row represent significantly different mean values (p < 0.05).

Prole of Peptides in the Acetone Fractions
Peptide profiles of the collagen hydrolysate fractions determined by shogun peptidomics are shown in Tables 3 and 4 for samples produced with Neutrase and papain, respectively. A total of 71 and 103 peptides were identified in the fractions of collagen hydrolysates produced using Neutrase and papain, respectively, which displayed the most potent antibacterial activities against the Gram-negative bacteria. The length of identified peptides ranged from 6 to 22 and 6 to 24 amino acid residues, and the molecular mass ranged from 612.3 to 2266 Da and 540.2 to 2586 Da for the Neutrase and papain fractions, respectively. The vast majority of peptides in both samples were derived from Type 1 collagen alpha 1 chain, the main collagen protein component. In general, lower molecular weight peptides with 2-5 amino acid residues are less accurately identified by the shotgun peptidomics approach and, if present in the fractions, were not detected in this study.

Discussion
Biological activities of protein hydrolysates are dependent on the protein substrate, enzyme used for proteolysis, and hydrolysis conditions, such as the enzyme-substrate ratio (E/S), incubation time, temperature, and pH [33]. The DH of the proteins influences the size and structure of the released peptides [34]. Moreover, functional properties of food protein hydrolysates, such as protein solubility, surface hydrophobicity, emulsification, and foaming capacity, depend on the degree of hydrolysis. For instance, the solubility of protein hydrolysate increases with increased DH as observed in samples from yellow stripe trevally [35], Alaska pollack [36], barbel [13,14], anchovy [37], and Atlantic mackerel [12]. High DH would ensure that the peptides become soluble and accessible to interact with their targets in aqueous physiological environments. However, extensive hydrolysis may lead to the loss of the bioactive motif in a peptide. As reported previously, the fish skin collagens in this study consist of two different subunits with an average molecular weight of 110-150 kDa (α1 and α2 at band intensity ratio of 2:1), β (dimers) and small amounts of γ (trimers) [38]. These protein bands were hydrolyzed differently depending on the enzyme used. Interestingly, trypsin gave the highest DH, even higher than multi-enzyme Alcalase and Flavourzyme, resulting in peptides with the lowest MW profiles. Conversely, papain and pepsin gave the lowest DH resulting in several high MW bands. Notably, the original collagen protein bands were not detected in all the samples, signifying their hydrolysis by the enzymes. Similar results have been reported by Suárez-Jiménez et al. [39], Chi et al. [40] and Felician et al. [10], who observed the presence of low molecular weight polypeptide bands for collagens isolated from squid by-products, fish cartilage fish, and jellyfish, respectively.
Surface hydrophobicity of the protein hydrolysates in this study was inversely related to DH. Whereas the smaller sized peptides would be highly soluble, it is expected that the high molecular weight peptides from limited hydrolysis would interact more easily to form hydrophobic pockets or aggregates that bind the fluorescent molecular probe [41,42]. Furthermore, surface charge distribution is an important parameter for the determination of biomolecular structure and interactions in aqueous environments [30]. Results from this study indicated weak electrostatic stabilization (except for hydrolysates produced with Alcalase and trypsin) and inclination of the hydrolysate particles to aggregate in aqueous solution. The net surface charges of the hydrolysates in aqueous solution also decreased in magnitude with an increase in surface hydrophobicity, which is an outcome of peptide aggregation. The molecular profile, surface properties and intermolecular interactions of peptides are important determinants of their bioaccessibility and binding to molecular targets, which would influence their bioactivities.
Antimicrobial peptides have been derived from various food sources by enzymatic hydrolysis, fermentation or gastrointestinal digestion for combating food-borne pathogens [43]. Results in this study revealed that the extent of the antibacterial activity of collagen hydrolysates against the S. abony and E. coli varies with different enzymes used for hydrolysis. Sila et al. [14] demonstrated that hydrolysates with higher DH (16.2% and 14.53%) or lower DH (2.8%) did not inhibit the growth of Gram-positive and Gram-negative bacteria, compared to hydrolysates with DH 6.6%, which showed antibacterial effect against a broad spectrum of Gram-positive (Listeria monocytogenes, Staphylococcus aureus, Enterococcus faecalis, Micrococcus luteus and Bacillus cereus) and Gram-negative bacteria (Escherichia coli, Salmonella enterica, Pseudomonas aeruginosa, Klebsiella pneumonia, and Enterobacter sp.). In our study, the DH had no significant effect (p > 0.05) on antibacterial activity. [15] and [44] observed similar results in their studies on trout protein hydrolysate and yak kappa-casein hydrolysate, respectively. The biological attributes of peptides are largely influenced by their molecular structural properties, including amino acid composition, sequence, net charge, and chain length [43]. These properties may have influenced the antibacterial activities of the collagen hydrolysates. Notably, hydrophobicity is thought to be important because it facilitates the interaction of the peptides with bacterial cytoplasmic membranes [44]. This feature relates more to the molecular hydrophobicity of individual peptides instead of surface hydrophobicity.
The bioactivity mechanisms of antibacterial peptides can be by interaction with negatively charged cell surface components, such as lipoteichoic acid in the outer membranes of Gram-positive bacteria and lipopolysaccharides in the cell wall of Gram-negative bacteria. Antibacterial peptides produce pores, channels or peptide-lipid complexes in the outer membrane or the cell wall of bacteria, and disruption of cytoplasmic membranes occurs followed by cell lysis, which results in inhibition of cell function or death [43,45]. Because of the increasing bacterial resistance against many antibiotics and advances in peptide design and synthesis, there has been a heightened interest in the discovery of new effective antibiotics [46]. Moreover, food-based sources provide potentially cheaper and safer alternatives to chemical antibacterial agents [43]. In our study, the collagen hydrolysates contained several peptides, some of which may be inactive. Acetone fractionation of the most potent hydrolysates (produced with Neutrase and papain) resulted in significantly increased antibacterial activity against four S. species, even at lower concentrations. This indicates that the downstream processing led to the concentration of active peptides in the resulting fraction. Ennaas et al. [12] reported a similar pattern with acetone fractionation of mackerel by-product hydrolysates with antibacterial activity against Gram-positive (L. innocua) and Gram-negative bacteria (E. coli). The increase in antibacterial activity after fractionation with acetone was attributed to the hydrophobic nature of the fractionated peptides. Similarly, Ruangsri et al. [47] reported that acetonitrile fractions extracted from different Atlantic cod (Gadus morhua) tissues had higher antibacterial activity than the aqueous fractions.
Peptides present in the acetone fractions are important candidates for further evaluation as antibacterial agents or for identification of important bioactive structural motifs. Several of the fish collagen peptides obtained with Neutrase (71 peptides) and papain (103 peptides) had arginine (R) and/or histidine (H), and/or lysine (K) residues in their sequences. These cationic amino acid residues, in addition to the presence of hydrophobic domains in sequences of antibacterial peptides (AMPs), contribute to the formation of amphiphilic topology for adherence of the peptides to bacterial membrane [43,48]. Furthermore, identified peptides from both samples revealed the high occurrence of glycine (G) and proline (P) residues, which are common structural characteristics of several AMPs [11,12,49,50]. Previously, a 12-mer collagencin (GLPGPLGPAGPK; f291-302 of Petromyzon marinus collagen, S4R4C5_PETMA) identified in a fraction from Scomber scombrus (Atlantic mackerel) byproduct hydrolysate was reported to have antimicrobial activity against six Gram-positive and Gram-negative bacteria, including L. innocua, Lactococcus lactis, Carnobacterium divergens, Staphylococcus aureus, Streptococcus pyogenes, and E. coli [25]. Notably, the C-terminal hexamer fragment of collagencin (GPAGPK) was found in seven peptides in our study, including peptides 13, 14, 18 and 19 released by Neutrase (Table 3) and peptides 19 and 73 released by papain (Table 4). It is likely that these motifs played a role in the antibacterial activity of their respective fractions.

Conclusions
In this study, fish skin collagen was hydrolyzed with different commercial enzymes to release peptides possessing potent antibacterial activity. The results demonstrated that the degree of hydrolysis had no significant effect on antibacterial activity. Collagen hydrolysate produced with Neutrase and papain showed the most potent inhibitory activity against Gram-negative bacteria (Salmonella strains), and viable count confirmed the decrease of the cell population. Acetone fractionation significantly enhanced the growth inhibitory activity against four Salmonella strains, providing products with stronger potential for use against food-borne pathogens. Antibacterial peptides not only alter the cytoplasmic membrane but also inhibit intracellular targets such as nucleic acid synthesis, protein synthesis or enzymatic activity. Therefore, the large number of peptide sequences identified in the fractions by shotgun peptidomics suggest further investigation of a potentially multi-targeted approach to the antimicrobial effects. Taken together, this study provided an array of peptide sequences from the fish skin by-products for elucidating molecular mechanisms and further exploration as value-added antimicrobial nutraceutical products against foodborne pathogens.