Probiotic L. paracasei subsp. tolerans FX-6 was used in the present study, previously isolated from Tibetan kefir (traditional fermented milk from Tibet, China), and stored at −80°C in the College of Food Science, South China Agricultural University, Guangzhou, China. Both Escherichia coli ATCC 25922 and Staphylococcus aureus ATCC 63589 were stored at −80°C in the microbial culture laboratory. DPPG, DPPE, DPPC, and CHOL were acquired from Corden Pharma (Liestal, Switzerland). Cholesterol and calcein were purchased from Sangon Biothch (Shanghai, China). All the solvents and chemicals used in this study were of analytical grade.

The tested bacterial strains used in this study E. coli and S. aureus were liquid cultured to the late logarithmic growth stage. After reaching this stage, the antimicrobial peptide F1 was added at a final concentration of 5 × MIC, after which it was cultured for 2 h at 37°C and sampled at 0.5, 1, and 2 h. Afterwards, centrifugation (2,000 rpm/min) for 10 min was used to collect the cells, after which the following steps were taken; (1) three rinses (10 min/rinse) with 5 mmol/L PBS buffer; (2) fixation with osmium acid for 1 h; (3) three rinses (10 min/rinse) with 5 mmol/L PBS buffer; (4) dehydration using a series of ethanol concentrations (30, 50, 70, 80, 90, and 100%). Among these, dehydration steps using 30, 50, 70, and 80% ethanol lasted for 10 min; cells were dehydrated with 90 and 100% ethanol twice per concentration (10 and 15 min/step, respectively); (5) dehydration using 100% acetone (twice, 15 min/step) (6) soaking in acetone: EPON812 resin in a 3:1 ratio for 2 h; (7) soaking in acetone:EPON812 resin in a 1:1 ratio overnight; (8) soaking in acetone:EPON812 resin in a 1:3 ratio for 6 h; (9) soaking in pure EPON812 resin twice for 12 h each time; (10) polymerize at 45°C for 24 h; (11) polymerize at 60°C for 48 h; (12) ultrathin sectioning followed by staining with 3% uranyl acetate and imaging using TEM (18).

One milliliter of bacteria (E. coli and S. aureus) was removed and cultured to the late logarithmic growth stage. After reaching this stage, it was centrifuged, washed twice with sterile ultrapure water, and then 950 μL of sterile ultrapure water and 50 μL of antimicrobial peptide F1 were added. The final concentration was 1 × MIC. Ultrapure water was used as a negative control, and the antimicrobial peptide Triton X-100 was used as a positive control. Samples were incubated for different times (30, 60, 90, 120, 150 min) in a water bath (37°C) and then centrifuged (4,000 rpm/min, 10 min). After centrifugation, 0.1 mL of the supernatant was removed and 9.9 mL of ultrapure water was added to dilute it. An atomic absorption spectrometer was used to detect the concentration of potassium and magnesium ions.

Lipid extraction was based on the method used by Dennison et al. (19) with minor modifications. Briefly, the bacterial solution that had been cultured to the logarithmic growth phase was placed in a 50 mL centrifuge tube and centrifuged at 4,000 rpm/min for 15 min at 4°C. The supernatant was re-suspended by adding PBS buffer, and then centrifuged at 4,000 rpm/min for 10 min, after which the supernatant was discarded. A small amount of buffer was added to the pellet and cells were transferred to a 300 mL beaker. Then 60 mL of buffer and 180 mL of chloroform methanol (1:2, v/v) solution were added and the resulting solution was stirred for 90 min. The liquid was then transferred from the beaker to a 500 mL separatory funnel, after which it was allowed to stand for separation of the different layers. The lowest liquid layer was collected and placed in a 50 mL round bottom flask and the solvent was evaporated at 40°C. The remaining organic solvent was blown off using nitrogen, after which 5 mL of PBS buffer was added to the extracted phospholipids. The sample was then sonicated for 30 min in a water bath to disperse the adherent film in the buffer. Using a 300 W intermittent ultrasonication for 5 min, the liquid nitrogen was repeatedly frozen and thawed for 8 cycles, after which it was repeatedly extruded for 10 times with a 100 nm film of liposome extruder to evenly distribute the individual liposomes. From this liposome dispersion, 3 mL was taken and subjected to a particle size analyzer, three cycles at a time. After determining the initial particle size, the liposomes were separated into the following two groups: (1) control group (37°C for 10 min) and (2) experimental group (1 mg/mL F1 and placed at 37°C for 10 min). A particle size analyzer was then used to detect the resulting particle sizes; each group had three samples run in parallel and the experiment was repeated three times independently.

Briefly, 10 mg of phospholipid was weighed and the phospholipid composition of the four groups was as follows: (1) 100% DPPG, (2) 30% DPPG and 70% DPPE, (3) 100% DPPE, (4) 90% DPPC and 10% cholesterol. To each group, 5 mL of chloroform was added to fully dissolve the included phospholipids and a 50 mL round-bottom flask was used to evaporate to dryness at 40°C. The remaining organic solvent was blown off using nitrogen to form a film on the bottom of the flask. Five milliliter of calcein solution (60 mM) was added to the above-mentioned flask, after which the solution was fully hydrated at 60°C and probed using 300 W intermittent ultrasound for 5 min. The solution was then filtered 8–10 times using a 100 nm filter to obtain a mixture of liposomes with a uniform particle size. From this mixture, 1 mL was removed and passed through a 1.5 × 60 cm dextran gel G-50 column to remove any free calcein from the liposomes. The mobile phase uses PBS buffer, and each 4 mL sample is collected in a test tube separately. Then, a multi-functional microplate reader was used to separately detect the fluorescence value of the liquid in each tube. The excitation and emission wavelengths were 485 and 530 nm, respectively.

To this, 10% Triton X-100 was added and mixed thoroughly prior to re-measuring. The sample with the largest fluorescence change before and after adding Triton X-100 was selected for the following experiment. The calcein-embedded liposome fluid was added to a 96-wells plate (200 μL per well) and its fluorescence value was detected (excitation and emission wavelengths were 485 and 530 nm, respectively). Afterwards, 10 μL of different concentrations of F1 (0, 20, 40, 60, and 120 μM) solution was added to each group. Fluorescence was measured after every 2 min. Finally, 10% Triton X-100 was added and thoroughly mixed to determine the final fluorescence value to indicate complete leakage. The following formula was then used to calculate the different fluorescence leakage values: Leakage % = (F_t – F₀)/(F – F₀) × 100%, where F is the fully leaked fluorescence value, F₀ is the initial fluorescence value, and Ft is the measured fluorescence value (20, 21).

Liposome (70% is DPPG and 30% is DPPE) sample was weighed (1 mg/mL) and prepared into a liposome solution according to the previously described method in section Effect of Antimicrobial Peptide F1 on Artificial Liposomes Embedded With Calcein. The surface of the quartz wafer was cleaned with alcohol, then dried with nitrogen and loaded into the instrument. The samples were then sequentially introduced into the crystal chamber. The operating procedure was as follows: (1) 0–20 min (buffer used to make a stable baseline); (2) 20–30 min (rinse with 1 mg/mL liposome solution at a flow rate of 100 μL/min); (3) 30–40 min (pump stopped and solution was allowed to stand; the liposomes were fully ruptured and spread; (4) any liposomes that were not adsorbed onto the wafer were washed away with buffer at a flow rate of 100 μL/min for 40–50 min; (5) antibacterial peptide F1 (50 μL/min) was flowed over the wafer for a 50–60 min rinse; (6) 60–90 min (no solution flow and F1 was allowed to fully work on the membrane); (7) ~90–100 or 90–130 min (unadsorbed and detached components were washed away with buffer at a flow rate of 100 μL/min) (22).

High-purity antibacterial peptide F1 (1 mg) was added and mixed well into the liposome solution prepared according to the previously described method in section Detection of Bacterial Cell Ion Leakage. Samples were removed from this liposome solution at 1 and 10 min before F1 addition. TEM sample preparation was as follows: The treated liposome suspension was added dropwise to a 200-mesh carbon support membrane and dried overnight at 40°C. Then, 2% phosphotungstic acid was added dropwise to negatively stain the sample. After 1 min, the dye was sucked away and dried. Then, the prepared carbon support film was observed by TEM (23).

All assays were conducted in triplicate and the average values were taken for data analysis. The data was expressed as mean values ± standard deviation (SD). The statistical significance was measured using one-way analysis of variance (ANOVA) and a multiple comparison by Tukey's test. Moreover, statistically significant difference was determined at P < 0.05. Standard curves and other figures were created using GraphPad Prism 5.0 (GraphPad Software Inc., San Diego, CA, USA) and Origin 8.5 [OriginLab (Microcal), CA, USA].

The ultrastructural changes of S. aureus over time after addition of the antimicrobial peptide F1 are shown in Figure 1. Normal S. aureus at time 0 h and without the effects of antimicrobial peptide F1 on its cell wall and membrane are clearly visible. The cell morphology is the typical morphology of S. aureus with a uniform cytoplasm, clear and complete cell wall and cell membrane, and smooth surface. After 0.5 h of incubation with antimicrobial peptide F1, the surface of the S. aureus cell membrane became rougher, with blurred cell wall and cell membrane and an uneven cytoplasm. After 1 h incubation with the antimicrobial peptide F1, the S. aureus cell membrane was almost mixed with the entire cell and the surface contour became blurred. The cell morphology was notably deformed and normal cells were not present, indicating that the intracellular lysate had leaked due to rupturing of cells and a complete dried out phenomenon occurred. After 2 h incubation with the antimicrobial peptide F1, the morphology of S. aureus cells was severely deformed and the cells were severely lysed, making it impossible to identify the cellular morphology.

Antimicrobial peptide F1 was also applied and investigated against E. coli, and the results had been published in a previous paper (15). Similar to the results observed in present study against S. aureus, the E. coli membrane was also gradually damaged after addition of the antimicrobial peptide F1. These result indicate that the antimicrobial peptide F1 has the potential to simultaneously damage the membranes of Gram-positive and Gram-negative bacteria (24). The main components of Gram-positive bacterial membrane are peptidoglycan, periplasmic space, and plasma membrane. The cell membrane of Gram-negative bacteria is composed of an outer membrane (lipopolysaccharide and protein), periplasmic space, and plasma membrane (25). Gram-positive bacteria have a thick peptidoglycan layer and no outer lipid membrane; comparatively, Gram-negative bacteria have a thin peptidoglycan layer along with an outer lipid membrane (26). Given this, the antimicrobial peptide F1 is likely to destroy the membrane structure by destroying the plasma membrane components shared by both Gram-positive and Gram-negative bacteria.

The potassium ion leakage from E. coli showed an increasing trend from 0 to 90 min and at the initial concentration of ions in F1 group was found higher than other groups. These results showed that F1 can contact and destroy the bacterial membrane in a short time (<30 min). In all groups, the concentration of potassium ion reached the maximum in 30 min and remained relatively stable, indicating that potassium ion is very sensitive to bacterial membrane damage. The potassium ion concentration for S. aureus continued to rise slowly during the first 120 min, while the potassium ion leakage rate suddenly increased after 120 min. Our results also indicated that the antibacterial peptide F1 had a stronger effect on both bacteria in the first 150 min than the membrane breaker Triton X-100. After 150 min, the ion leakage concentrations of the antimicrobial peptide F1 and Triton X-100 treatment groups were similar (Figures 2A–D). Compared with the negative control, F1 caused a notably larger amount of magnesium ion leakage in the somatic cells of E. coli and S. aureus. Moreover, the effect was better than that of the positive control of Triton X-100.

Some studies showed that most of the antimicrobial peptides from lactic acid bacteria form holes in the bacterial cell membrane, causing intracellular ions to leak out and ultimately causing the dissipation of the proton-driven potential (27). Although different antimicrobial peptides achieve this ultimate end, their target mechanisms are different. Currently, the detergent damage model and pore channel model are the two models that illustrate this mechanism (28). Both models are related to membrane damage and proton dynamic potential dissipation. Studies have shown that some antimicrobial peptides cause damage to bacterial cell membranes, decrease the fluidity of the membrane, and/or cause loss of its semi-permeability. Ultimately, resulting in leakage of intracellular electrolytes and nucleic acids to the outside of the cell, simultaneously affecting the various metabolic pathways of the cell. Interference with the regulation of osmotic pressure inside and outside the cell eventually leads to cell death (29). Potassium and magnesium ions are two types of metal ions found in bacterial cells. Under normal circumstances, they exist stably inside the cells. When the cell membrane of the bacteria is either permeabilized or ruptured, a large amount of leakage will follow. Given this, most researchers use the detection of potassium ions or changes in magnesium ions concentration to assess cell membrane integrity (30).

To demonstrate that the antimicrobial peptide F1 directly reacts with bacterial liposomes, thus liposomes were extracted from both from E. coli and S. aureus. The liposomes made from the extracted lipids of E. coli and S. aureus have a degree of stability, and the particle size remained unchanged (~200 nm) after being left untouched for a period of time (Figure 3). After applying the antimicrobial peptide F1, the size of the liposomes significantly increased. After the antimicrobial peptide F1 application, the size of liposomes in the 0.1% antimicrobial peptide F1 group was significantly larger than that of the 0.05% antimicrobial peptide F1 liposomes. This difference indicated that high concentrations of antimicrobial peptide F1 interfered with liposomal stability. This instability was caused by interacting with liposomes to increase their size. The increase in particle size of S. aureus liposomes was larger than that of E. coli liposomes, indicating that antimicrobial peptide F1 had different degrees of damage to liposomes of different strains, indicating that F1 has a certain selectivity for liposome damage.

Calcein has self-quenching properties. When high concentrations of calcein are found inside liposomes, the fluorescence value is low. When antibacterial peptides damage liposomes, calcein leakage starts and resulting in an increased fluorescence. The internal strength increases with the degree of leakage; the molecular diameter is smaller than that of probes such as FITC, which is suitable for liposomes with small particle size (31).

According to the fluorescence working curve of calcein shown in Supplementary Figure 1, it is clear that the fluorescence value of calcein has a concentration range that showed a linear change. Taking the level of fluorescence leakage after adding 10% Triton X-100 as an indicator of the complete leakage of liposomes. More specifically, it has a good linear relationship across a fluorescence intensity range of 0–400,000. This quantitatively reflects the leakage level of the probe. When the concentration exceeds 3 μg/mL, the fluorescence intensity exceeds this range and the fluorescence value no longer significantly increases. As shown in Figures 4A,B, after treatment with F1, the above two liposomes had strong fluorescence leakage. Moreover, the leakage degree increased with increasing F1 concentration. The leakage rate of DPPG liposomes under the same F1 concentration was significantly greater than that of the DPPG + DPPE liposomes. When the liposomes contained DPPE molecules, the efficiency of F1 was reduced. This may be because the addition of DPPE molecules (without charge) reduces the overall negative charge. As shown in the aforementioned figures, the fluorescence leakage after F1 acted on the liposomes was very fast. More specifically, a leakage level of more than 80% was observed within the time span of 2 min; over the following 2–10 min, subsequent leakage rate increase was not very high. This indicated that the interaction between the antimicrobial peptide F1 and the phospholipid membrane occurred quickly and when F1 contacted other liposomes, the amount of leakage increased. DPPC and cholesterol are the main components of mammalian cell membranes (32). DPPC molecules are not charged, and cholesterol has a protective effect on the stability of phospholipid membranes. As shown in Figures 4C,D, the effect of F1 on DPPC liposomes was significantly smaller than that on DPPG and DPPG + DPPE. The effect on DPPC + CHOL was smaller, and the fluorescence leakage rate in these liposomes never exceeded 10%. This reduction in permeability was due to the cholesterol molecules introducing conformational ordering of the lipid chains and creating a denser and more rigid barrier for the calcein to cross (33).

TEM is commonly used to characterize the internal structure of nanoemulsions (34). Nano-sized liposomes are dispersed in the buffer and their integrity is observable using transmission electron microscope (TEM). As shown in Figure 5, before adding F1, liposomes were found in a regular and uniform spherical shape as liposomes primarily appeared as unilamellar spherical vesicles consisting of a lipid bilayer (higher density bands) surrounding an aqueous core (same density as exterior aqueous phase) (35). One min after adding the antimicrobial peptide F1 to the system, the liposomes were fused and presented as a large liquid phospholipid membrane. At this point, the phospholipid bilayer completely lost its structure. After the antimicrobial peptide F1 interacted with the phospholipid membrane for 10 min, the phospholipid membrane broke into an irregular dispersion. When examined closely, it was clear that the original liposome became a transparent phospholipid membrane structure. Although the liposomes were clearly fused, they maintained a membranous structure. When acting upon for a length of time, the phospholipid membrane ruptured, and the precipitate accumulated. This led to increased agglomeration between liposomes over time, the liposome membrane ruptured and the average particle size increased.

Conclusively, the study's results showed that the antimicrobial peptide F1 disturbed the structure of the phospholipid membrane, thereby destroying the cell membrane structure of the tested bacteria and causing the cell body contents to leak out ultimately resulting in cell death.

QCM-D is an emerging technology based on quartz crystal microbalance (QCM) technology (36). QCM-D monitors the frequency (F) and energy dissipation (D) of the quartz crystal surface in real time, thereby monitoring the dynamics of the surface of the membrane solid support membrane and measuring the quality and viscoelasticity of the phospholipid membrane (37). As shown in Figure 6A, before adding F1, the frequency and dissipation of the wafer surface were relatively stable, and the phospholipid membrane had reached a fairly stable state. Fifty minutes min after F1 had been added, the frequency and dissipation fluctuated greatly. The frequency significantly decreased, indicating that the quality of the membrane surface increased. Moreover, the membrane surface quality continued to increase slowly after stopping the flow of F1. After allowing the flow into the buffer to rest, the frequency and the dissipation value both recovered and quickly stabilized. The final frequency and dissipation values had a certain difference from before flowing into F1, indicating that both the quality of the membrane surface and its viscoelasticity had increased. Taken together, these results showed that this concentration of F1 was adsorbed on the surface of the phospholipid membrane, which changed the physical parameters of the membrane surface. As shown in Figure 6B, the change on the surface of the phospholipid membrane at 0.2 mg/mL F1 was almost the same as that observed at 0.1 mg/mL. However, the frequency of the membrane surface was more notably changed. This indicated that F1 adsorbed more on the membrane surface, but did not show any obvious destructive effects. As shown in Figure 6C, when 0.3 mg/mL F1 was added, the frequency of the membrane surface also decreased, and the dissipation phenomenon increased. When F1 was continuously in contact with the phospholipid membrane, the frequency of the membrane surface began to slowly rise, and dissipation was slow and had decreased. At this stage, this indicated that the membrane mass was reduced, the viscoelasticity was reduced, and/or the membrane was thinned. After flowing into the buffer, the frequency increased and was closer to the initial value than when the concentration was low. This indicated that more material was washed away on the surface of the wafer. It may be that the phospholipid membrane was damaged due to the action of F1, causing part of the phospholipid molecules to fall off. As shown in Figure 6D, after adding 0.5 mg/mL of F1, both the mass and viscoelasticity of the membrane increased rapidly during the adsorption period. During the standing process, the frequency significantly decreased, which indicated that the mass of F1 decreased after interacting with the phospholipid membrane. Moreover, the D value also slowly increased (4).

After adding the buffer, the frequency increased to a large degree, indicating a large amount of damage to the phospholipid membrane and the washing away of a large amount of material on the wafer. As shown in Figures 7A–D, when 0.1 and 0.2 mg/mL of F1 were individually applied, the dissipation at 3 octave and 11 octave increased with increasing frequency. Comparatively, at 0.3 mg/mL, the F and D values began to change. When the concentration reached 0.5 mg/mL, its change law became more complicated and the curve of the 3 octave frequency changed to the upper left direction. The slope of the curve was first larger and then smaller, and the curve of the 11 octave frequency changed to the lower left. Taken together, these results showed that the viscoelasticity of the phospholipid membrane on the wafer surface varied at different depths (14). As shown more intuitively in Figures 8A–D, the value of the membrane changed at different frequency doublings. At lower F1 concentrations, the surface quality of the film increased; notably, this increase became larger with increasing concentration. At the same concentration, the amount of mass increase decreased with an increase of the frequency doubling. As shown, the change on the surface of the film was greater, while the internal impact was less. With increasing F1 concentration, the mass change was not proportional to the concentration, but began to decrease. After applying 0.3 mg/mL F1, the frequency of each frequency doubling was reduced; that is, the mass increased, but the increase was <20 Hz. This was notably <0.2 mg/mL. When the concentration reached 0.5 mg/mL, the frequency still decreased at the three times the frequency, while the frequency of five to eleven-times was significantly increased. Taken together, this indicated that the internal quality of the membrane was decreased and the membrane was no longer intact after F1 application.

In summary, the results of the QCM-D experiment showed that different F1 concentrations had different effects on the surface of the phospholipid membrane. When either 0.1 or 0.2 mg/mL F1 was applied over the membrane surface, adsorption occurred. Moreover, the amount of adsorption increased with increasing F1 concentration. However, when F1 with a concentration higher than 0.3 mg/mL was applied onto the surface of the phospholipid membrane, it caused significant damage to the phospholipid membrane. At 0.3 mg/mL of F1, the membrane surface quality was slightly reduced. At 0.5 mg/mL of F1, the membrane quality was significantly reduced, but the viscoelasticity increased. The changes in the properties of the membrane surface and the interior of the membrane were not fully synchronized, but as the concentration increased, the amount of molecular aggregation also increased and affected the membrane integrity.