Strain YH4 was isolated from a mud sample collected from Yuncheng Salt Lake, Shanxi Province, China. Culturable bacterial strains were initially enriched using modified LB medium (10% NaCl, 1% tryptone, 0.5% yeast extract, pH 7.2) under aerated conditions at 30°C with 200 rpm for 24 h. Protease-production strains were subsequently screened on selective skim milk agar plates (10% NaCl, 0.5% peptone, 0.1% yeast extract, 1% skim milk, 1.5% agar) incubated at 37°C for 48 h. Strain YH4 was identified based on its proteolytic activity, characterized by distinct hydrolysis zones on the skim milk agar medium. Genomic DNA was extracted using a bacterial genomic DNA extraction kit, and the 16S rRNA gene was amplified via polymerase chain reaction (PCR) with universal primers 27F and 1492R (Wang et al., 2020). The amplified product was sequenced by BGI Co., Ltd. (China) for taxonomic identification.

To evaluate halotolerance, strain YH4 was inoculated into fresh LB medium containing NaCl concentrations ranging from 0 to 5 M at a 2% (v/v) inoculum density. Cultivation was conducted in 30°C with 200 rpm. Growth was monitored after 24 h of incubation via optical density measurements at 600 nm (OD₆₀₀) to assess salt-dependent viability.

The activated strain was inoculated into a fermentation medium for enzyme production, containing 2% corn flour, 1% wheat bran, 2% soybean meal, 0.1% Na₂HPO₄, 0.03% KH₂PO₄, 0.1% CaCl₂, 0.1% Na₂CO₃, and 100 mL artificial seawater (pH 7.2). The culture was prepared with a liquid volume of 25 mL per 250 mL flask and incubated at 30°C with 200 rpm for 1–5 days. Crude enzyme solutions were harvested at daily intervals to assess protease production over time.

Extracellular protease profile was detected using casein as the substrate via the substrate immersion enzyme assay (Liu et al., 2015). To classify protease types, inhibitors phenylmethylsulfonyl fluoride (PMSF, 2.5 mM) and 1,10-phenanthroline (OP, 2.5 mM) were incorporated into the substrate solution, and residual enzyme activity was measured according to the method described by Liu et al. (2019b).

Following fermentation at 30°C with 200 rpm for 96 h, the broth was centrifuged at 12,000 × g for 30 min to isolate the supernatant as the crude enzyme solution. Ammonium sulfate ((NH₄)₂SO₄) was added to the supernatant to achieve 50% (w/v) concentration, followed by overnight precipitation at 4°C. The precipitated protein was collected via centrifugation (12,000 × g, 30 min), redissolved in 20 mM Tris–HCl buffer (pH 8.0), and dialyzed (MWCO 10 kDa) to remove low-molecular-weight impurities. The dialyzed sample was loaded onto a HiTrap Capto DEAE column (5 mL; GE Healthcare, United States) pre-equilibrated with 20 mM Tris–HCl buffer (pH 8.0) using an NGC chromatography system (Bio-Rad, USA). Bound proteins were eluted with a linear NaCl gradient (0–1.0 M) in the same buffer at a flow rate of 1 mL/min. Fractions exhibiting protease activity were pooled and further purified by gel filtration chromatography on a Superdex 75 Increase 10/300 GL column (24 mL; GE Healthcare, United States) at a flow rate of 0.5 mL/min. High-activity fractions were collected, and the molecular weight and purity of the protease were verified via SDS-PAGE under reducing conditions. The purified enzyme, designated EYHIII, was quantified using a BCA protein assay kit (Beyotime, China) with bovine serum albumin (BSA) as the standard.

For structural identification, the excised SDS-PAGE band corresponding to EYHIII was submitted to Sangon Biotech (Shanghai, China) for LC–MS/MS analysis. The acquired mass spectrometry data were aligned against the NCBI non-redundant protein database to identify homologous sequences. Conserved domains were identified, and gene-specific primers (Table 1) were designed to amplify flanking regions via thermal asymmetric interlaced PCR (TAIL-PCR). The full-length EYHIII gene sequence was assembled from overlapping PCR products and validated by sequencing. The tertiary structure of EYHIII was predicted using the homology modeling tool Phyre2.¹

To determine the optimal temperature for EYHIII activity, the enzyme was diluted appropriately and mixed with 2% (w/v) casein in a 1:1 (v/v) ratio. The mixtures were incubated at temperatures ranging from 30 to 80°C for 10 min, and enzyme activity was quantified using the Folin phenol method (Liu et al., 2019a,b). All optimal condition tests (thermal/pH/salt) utilized the condition exhibiting maximum activity as the baseline control (100% activity), thereby enabling the calculation of relative activities at other conditions. For thermal stability assessment, EYHIII solutions were incubated at 50, 60, and 70°C for intervals of 10–60 min. In the experiment of optimum enzyme activity temperature and pH value, the enzyme activity of protease was high at 50°C and pH8.0. And pre-tests showed <5% activity deviation at 50°C/pH 8.0 across 3 biological replicates. Thus, all stability tests (thermal/pH/salt) used the same baseline control (untreated enzyme at 50°C/pH 8.0) to ensure cross-experiment comparability.

The pH optimum of EYHIII was evaluated by assaying activity at 50°C across a pH gradient (4.0–12.0) using the following 0.02 M buffers: Citrate-Na₂HPO₄ (pH 4.0–7.0), Tris–HCl (pH 7.0–9.0), and Gly-NaOH (pH 9.0–12.0). For pH stability analysis, the enzyme was pre-incubated in these buffers at 4°C for 24 h, and residual activity was measured under standard conditions (50°C, pH 8.0).

Halotolerance was assessed by incubating EYHIII in NaCl solutions (0–4 M) at 4°C for 24 h, followed by residual activity measurement at 50°C and pH 8.0. The influence of NaCl on instantaneous activity was similarly tested without pre-incubation.

To evaluate metal ion effects, EYHIII was incubated for 10 min with 2.5 mM or 10 mM solutions of K⁺, Ca²⁺, Na⁺, Ba²⁺, Co²⁺, Mn²⁺, Mg²⁺, Al³⁺, Cu²⁺, Zn²⁺, Fe³⁺, or Fe²⁺. Activity in the absence of metal ions was defined as 100%. All assays were conducted at 50°C and pH 8.0.

The substrate specificity of EYHIII was determined by measuring its hydrolytic activity toward the following substrates: gelatin, casein, azocasein, synthetic dipeptides, soluble collagen, and insoluble collagen. Gelatin, casein, and elastin were purchased from Sigma-Aldrich (United States). Soluble collagens were extracted from porcine and salmon skins as described by Wu et al. (2018). Insoluble collagen substrates—bovine collagen fiber BF-50, porcine collagen PP-100, and bovine collagen BP-SF—were provided by Shanghai Sidao Yilang Investment Management Co., Ltd. (China). Synthetic dipeptide substrates (FA-Gly-Leu-NH₂, FA-Gly-Phe-NH₂, FA-Gly-Val-NH₂, FA-Lys-Ala-OH, FA-Ala-Arg-OH, FA-Glu-Glu-OH) were synthesized by Suzhou QiangYao Biotechnology Co., Ltd. (China).

Casein hydrolysis: Proteolytic activity was quantified using the Folin–Ciocalteu method (Liu et al., 2019a).

Insoluble collagen and gelatin hydrolysis: Activities were measured according to Wu et al. (2017). One unit (U) of enzyme activity was defined as the amount of enzyme releasing 1 μmol leucine from collagen per hour or 1 μmol tyrosine equivalents from gelatin per minute.

Azocasein hydrolysis: A reaction mixture containing 40 μL of EYHIII and 40 μL of 2% (w/v) azocasein was incubated at 50°C for 10 min. The reaction was terminated with 80 μL of 10% (w/v) trichloroacetic acid, centrifuged at 12,000 × g for 1 min, and 100 μL of the supernatant was mixed with 100 μL of 0.5 M NaOH. Absorbance was measured at 400 nm, with one unit defined as the enzyme causing a ΔA₄₀₀ of 0.01 per minute (Coelho et al., 2016).

For soluble collagen substrates (salmon and porcine), 2–4 μL of EYHIII (0.2 mg/mL) was incubated with 20 μL of collagen (5 mg/mL) at 50°C for 10–30 min. Reactions were terminated by adding 5 μL denaturing buffer, boiling for 5 min, and analyzing degradation patterns via SDS-PAGE (12% gel, Coomassie staining) (Wu et al., 2018).

The synthetic dipeptide substrate was diluted to 1 mM in 20 mM Tris–HCl buffer (pH 8.0). EYHIII was similarly diluted in the same buffer to ensure consistent experimental conditions. A 100 μL of EYHIII was mixed with 100 μL of the dipeptide substrate, and the mixture was transferred to a cuvette in an Agilent Cary 60 UV–Vis spectrophotometer. After a 30 s equilibration period, absorbance changes were monitored in real time using the instrument’s Time Course Measurement software. The 200 μL reaction system maintained a final substrate concentration of 0.5 mM (1% DMSO) and was analyzed at 25°C for 600 s, with data points recorded at 1 s intervals (Xie et al., 2009).

The catalytic efficiency (k_cat/ K_m) was derived from the linear decrease in absorbance at 345 nm, which reflects the reaction velocity. Under the assumption [S]₀<<K_m, the following relationship was applied:

where:

K_cat/K_m:Apparent second-order rate constant, s⁻¹·M⁻¹.

v: Enzymatic reaction rate, M·s⁻¹.

k: Absolute slope of the linear absorbance decline at 345 nm, s⁻¹.

b: Cuvette path length, 1 cm.

∆ε345: Molar extinction coefficient, −317 M⁻¹ cm⁻¹.

[E]₀: Initial enzyme concentration, M.

[S]₀: Initial substrate concentration, M.

The antioxidant activity of collagen hydrolytic peptides was assessed using three established assays: DPPH (1,1-diphenyl-2-picrylhydrazyl) radical scavenging activity, hydroxyl radical scavenging activity, and oxygen radical absorbance capacity (ORAC). These assays were performed according to methodologies described by Liu et al. (2019a) and Xiao et al. (2024), with minor adaptations to optimize conditions for collagen-derived peptides.

Human umbilical vein endothelial cells (HUVECs) were seeded in 96-well plates at a density of 1 × 10⁵ cells/mL in RPMI-1640 medium supplemented with 10% (v/v) fetal bovine serum (FBS) and incubated for 12 h at 37°C under 5% CO₂. Cells were treated with collagen hydrolysate peptides (100–600 μg/mL) for 12 h. After treatment, 10 μL of MTT solution (5 mg/mL) was added to each well, followed by a 4 h incubation. The medium was removed, and 150 μL of DMSO was added to each well. Plates were agitated on an orbital shaker (10 min, low speed), and then absorbance was measured at 490 nm to evaluate cell viability (Liu et al., 2019a).

HUVECs (1 × 10⁵ cells/mL) were seeded in 24-well plates and cultured for 12 h. Cells were washed thrice with PBS and incubated for 12 h in RPMI 1640 containing 35 mM glucose and hydrolysate peptides (50–300 μg/mL). DCFH-DA (2,7-dichlorofluorescein diacetate, 1:1000 dilution in RPMI 1640) was added, and cells were incubated for 1 h. Excess probe was removed by washing with 10 mM PBS. Fluorescence intensity (indicative of ROS levels) and cell morphology were analyzed using a Nikon ECLIPSE TE2000-U fluorescence microscope (excitation/emission: 485/535 nm) (Wu et al., 2018; Liu et al., 2019a).

Treated HUVECs were harvested, washed twice with ice-cold PBS, and lysed via ultrasonic homogenization (200 W, 3 s pulse/10 s interval, 8 cycles, ice bath). Lysates were centrifuged (12,000 × g, 15 min, 4°C), and supernatants were collected. The activities of superoxide dismutase (SOD), catalase (CAT), and glutathione peroxidase (GSH-Px) were quantified using commercial assay kits (Beyotime, China), following manufacturer protocols.

All results are reported as the mean ± standard deviation (SD) of triplicate experiments, with error bars representing the SD. Statistical analysis was conducted using SPSS software (version 23, SPSS Inc., Chicago, IL, United States). Differences between groups were assessed using analysis of variance (ANOVA) followed by Duncan’s multiple comparison test, with a significance level set at p < 0.05. Asterisk notation: * (p < 0.05), ** (p < 0.01), *** (p < 0.001). Graphs were generated using Origin 9.1 (OriginLab, Northampton, MA).

Protease-producing strains were isolated from the hypersaline sediments of Yuncheng Salt Lake using casein-containing screening plates. Strain YH4 exhibited prominent proteolytic activity, as evidenced by clear hydrolysis zones on casein agar (Figure 1A). Scanning electron microscopy (SEM) revealed a vibrioid morphology characteristic of the genus Salinivibrio (Figure 1C). The 16S rRNA gene sequence of strain YH4 was analyzed using BLAST against the GenBank database, and homologous sequences from phylogenetically related species were retrieved to construct a maximum-likelihood phylogenetic tree (Figure 1B). The analysis demonstrated that the 16S rRNA sequence of YH4 (GenBank accession: KR870826.1) exhibited the highest similarity (99%) with Salinivibrio costicola Pb-WC11147T, confirming its classification within the Salinivibrio genus. Growth analysis under varying NaCl concentrations (0–5 M) revealed optimal growth at 1–3 M, with negligible growth outside this range (Figure 1D), thereby confirming Salinivibrio sp. YH4 as a moderate halophile.

Extracellular protease production by Salinivibrio sp. YH4 exhibited time-dependent dynamics during fermentation. Protease activity increased significantly after 24 h, reaching a plateau by 84 h (Figure 2A). Zymographic analysis revealed distinct substrate-specific protease profiles. Two dominant proteases, EYHI and EYHIII, were identified through caseinolytic activity (Figures 2A,B), while four proteases (EYHI, EYHII, EYHIII, EYHIV) were identified through gelatinolytic activity (Figures 2C,D). Notably, EYHI and EYHIII exhibited dual substrate specificity for both casein and gelatin, whereas EYHII and EYHIV were specific to gelatin. Inhibition assays further classified these proteases. EYHIII and EYHIV activities were inhibited by 2.5 mM 1,10-phenanthroline (OP), while EYHI and EYHII activities were inhibited by 2.5 mM phenylmethylsulfonyl fluoride (PMSF). These findings demonstrate that Salinivibrio sp. YH4 primarily secretes metalloproteases, with EYHIII identified as the predominant metalloprotease, alongside serine proteases.

The extracellular crude enzyme extract of Salinivibrio sp. YH4 was purified using ammonium sulfate precipitation (50% saturation), dialysis, and sequential chromatography. Initial DEAE anion-exchange chromatography (20 mM Tris–HCl, pH 8.0) with isocratic elution (25% 1 M NaCl) yielded a single active fraction (F2) that exhibited protease activity (Figure 3A). Subsequent purification by gel filtration chromatography (using a Superdex 75 column) resolved three protein peaks, of which only F2-2 displaying proteolytic activity (Figure 3B). SDS-PAGE analysis confirmed that the purified enzyme, designated EYHIII, had a molecular weight of approximately 35 kDa (Figure 3C).

LC–MS/MS analysis of EYHI identified five peptide sequences that were 100% identical to the zinc metalloprotease precursor from Salinivibrio proteolyticus (Table 2). Based on these peptide sequences, similar protease sequences were searched in the NCBI database, and the conserved regions were identified. Primers were then designed using the gene sequences from these conserved regions to amplify the corresponding EYHIII protease gene. Three sets of nested primers were then designed, and TAIL-PCR was employed to obtain the 3′ and 5′ untranslated regions of EYHIII. The resulting sequences were assembled and validated. The final full-length enzyme sequence of EYHIII comprised 611 amino acids (GenBank: QAB01367.1). Furthermore, the full-length EYHIII enzyme exhibited 99.51% similarity with the M4 family metallopeptidase from Salinivibrio proteolyticus and 79.05% similarity with the M4 family metallopeptidase from Salinivibrio socompensis, indicating that EYHIII belonged to the M4 family.

BLAST analysis of the conserved domains within the full-length EYHIII enzyme revealed the presence of an FTP domain, a PepSY domain, an M4 neutral protease domain, and a C-terminal PPC domain (Figure 4A). The predicted tertiary structure adopted a heart-shaped configuration, with the N-terminal domain primarily comprised of β-sheets and the C-terminal domain dominated by α-helices (Figure 4B). Within the catalytic domain, the zinc-binding residues were H345, H349, and E369, forming a proteolytic triad essential for enzyme activity. The Expasy database predicted the molecular weight of protease EYHIII to be 66.67 kDa. However, SDS-PAGE electrophoresis and mass spectrometry analysis of purified EYHIII revealed a molecular weight of 33.3 kDa.

As shown in Figure 5A, the optimal temperature for protease EYHIII activity ranged from 50 to 60°C, which was consistent with the characteristics of typical mesophilic enzymes. Although its peak activity occurred at 60°C, thermal stability tests revealed a significant gradient in heat tolerance (Figure 5B). After incubation at 50°C for 1 h, the enzyme activity remained largely unaffected, demonstrating both high catalytic efficiency and excellent short-term stability near its optimal temperature. However, stability declined sharply as the temperature exceeded 55°C. For instance, a 20-min treatment at 60°C resulted in approximately 40% activity loss, and prolonged exposure to higher temperatures led to an 80% reduction in activity after 30 min at 70°C.

Figure 5C demonstrated that EYHIII exhibited higher catalytic activity under neutral to slightly alkaline conditions. The enzyme showed optimal activity between pH 7.5 and 9.5 and maintained over 90% of its activity across a pH range of 7.0–10.0. The pH stability results further confirmed that EYHIII primarily exhibited catalytic activity under alkaline conditions, displayed enhanced stability in alkaline environments.

The effects of NaCl concentration (0–4 M) on the enzyme activity of EYHIII illustrated in Figure 5D. The presence of 0.1 M NaCl enhanced protease activity, and at 1 M NaCl, EYHIII retained over 80% of its activity. However, at 4 M NaCl, activity dropped to less than 20%. Notably, after 24 h of incubation in varying salt concentrations, residual activity measurements indicated that EYHIII was highly stable in salt solutions.

Figure 5E depicted the impact of various metal ions on the catalytic activity of EYHIII at different concentrations. While K⁺, Ca²⁺, Na⁺, and Ba²⁺ had minimal effects, the presence of 10 mM Mn²⁺ significantly enhanced the enzyme activity. Conversely, 10 mM Mg²⁺ caused a reduction in activity by over 50%. Co²⁺ inhibited the activity by approximately 35% at 2.5 mM and around 70% at 10 mM, while 2.5 mM Cu²⁺ nearly completely inhibited EYHIII activity. Furthermore, EYHIII demonstrated heightened sensitivity to 2.5 mM Al³⁺, Cu²⁺, Zn²⁺, Fe²⁺, and Fe³⁺.

We selected casein, gelatin, azocasein, synthetic dipeptides, and both soluble and insoluble collagen to evaluate EYHIII’s substrate specificity. Casein, a major milk protein with a complex structure, mimics natural protease challenges. Gelatin, derived from collagen, is stable, soluble, and demonstrates enzyme performance on modified proteins. Azocasein, a modified form of casein, enables sensitive and quantitative tracking of enzyme activity. Synthetic dipeptides, with their clearly defined structures, help elucidate the enzyme’s catalytic mechanism. Soluble collagen is easier to hydrolyze, while insoluble collagen requires stronger enzymatic activity. This diverse substrate selection highlighted the enzyme’s versatility and potential for various applications.

Oxidative damage to endothelial cells is a key factor in the development of various diseases, such as diabetes mellitus and atherosclerosis. Consequently, antioxidant protection of endothelial cells is critical for preventing and treating vascular dysfunction-related conditions. To evaluate the biocompatibility of fish collagen hydrolysates in HUVECs, the MTT assay was employed. In viable cells, intracellular succinate dehydrogenase reduces MTT to form insoluble blue-violet formazan crystals that accumulated within the cells. These crystals were subsequently dissolved in DMSO, and the absorbance was measured at 570 nm. Within a specific range, the absorbance was proportional to the number of viable cells. Treatments with 200 and 600 μg/mL of the hydrolyzed peptides were not only non-toxic but also resulted in a higher cell count compared to the control (Figure 9A), indicating that the peptides were biocompatible within the tested concentration range and suitable for subsequent ROS experiments.

High glucose levels are known to increase oxidative stress in peripheral tissues, leading to the accumulation of intracellular reactive oxygen species (ROS). DCFH-DA, a cell-permeable probe, is hydrolyzed by intracellular lipases to yield DCFH, which is retained within the cell; ROS then oxidize DCFH to generate fluorescent DCF. To evaluate the intracellular ROS scavenging effects of the hydrolyzed peptides, HUVECs were labeled with DCFH-DA. As shown in Figure 9B, cells exposed to 35 mM glucose exhibited a significantly higher DCF fluorescence intensity than the blank control, confirming that high glucose stimulation increases ROS levels and induced an oxidative stress state in HUVECs. This result validated the high-glucose oxidative stress model. Furthermore, treatment with different concentrations of mixed hydrolyzed peptides resulted in a dose-dependent reduction in intracellular fluorescence. Even at 50 μg/mL, the peptides decreased ROS levels; at 100 μg/mL, a greater number of cells exhibited diminished fluorescence; and at 200 μg/mL, the fluorescence intensity was comparable to that of the blank control. At 300 μg/mL, no detectable intracellular fluorescence was observed. These findings indicated that the unpurified mixed hydrolyzed peptides effectively mitigated oxidative damage in HUVECs in a concentration-dependent manner.

Cells possess a complex antioxidant defense system comprising both antioxidant molecules and enzymes, such as glutathione peroxidase (GSH-Px), superoxide dismutase (SOD), and catalase (CAT). GSH-Px catalyzes the reduction of H₂O₂ and various organic peroxides to H₂O or corresponding alcohols using reduced glutathione (GSH) as a substrate, while SOD converts superoxide anions to H₂O₂, and CAT decomposes H₂O₂ into H₂O. The effects of hydrolyzed peptides on the activities of these antioxidant enzymes in high glucose-damaged HUVECs were shown in Figures 10. In the high glucose-damaged group, the activities of GSH-Px, SOD, and CAT were reduced relative to the blank group. However, only the decrease in CAT activity was statistically significant (p < 0.05), whereas the reductions in GSH-Px and SOD activities were not (p > 0.05). Treatment with low concentrations of hydrolyzed peptides (50–100 μg/mL) did not result in significant changes in enzyme activities compared with the high glucose-damaged group (p > 0.05). In contrast, increasing the peptide concentration to 200 μg/mL led to significant increases (p < 0.05) in enzyme activities: CAT activity rose from 85.68 to 110.96%, GSH-Px from 73.57 to 135.79%, and SOD from 87.63 to 103.55%. These findings indicated that high-concentration fish collagen hydrolysate (200 μg/mL) significantly upregulated GSH-Px, SOD, and CAT activities in glucose-stressed HUVECs, counteracting oxidative damage by enhancing endogenous antioxidant defenses.