Different Saccharomyces cerevisiae strains exhibit varying abilities to express heterologous genes. We aimed to screen for an S. cerevisiae strain with the strongest LL-37 protein expression capability. Plasmids containing the gene encoding LL-37 were constructed and transformed into S. cerevisiae cells. Fermentation was conducted at 30°C, 220 rpm, and LL-37 detection was performed through HPLC. The yeast strain that achieved the highest biomass and highest titer yield was selected as the chassis strain for subsequent experiments.
Objective: Select an S. cerevisiae chassis cell with the strongest LL-37 expression capability.
Method: Using pESC as the vector, we constructed the pESC-URA3-PGAL1-LL37 plasmid (Figure 1) through one-step cloning. Once the plasmid sequence was confirmed, it was transformed into four different S. cerevisiae chassis cells for expression, with yields detected by HPLC.
Figure 1: LL-37 expression plasmid
Results: Laboratory strains QY1.1 and QY1.2 showed significantly higher OD₆₀₀ values in fermentation medium compared to environmental isolates QY1.3 and QY1.4, reaching a maximum of 18.11 (±0.74), while QY1.3 peaked at only 12.08 (±0.63). Additionally, QY1.1 achieved maximum yield at 48 hours, reaching 386.81 (±16.07) μg/L, which was 7.67% higher (p=0.093) than the second-highest yield of 359.25 (±14.62) μg/L achieved by QY1.3 at 72 hours, and significantly higher than all other strains at all time points (Figure 2). Therefore, we selected QY1.1 as the chassis cell for further modifications.
Figure 2: Chassis cell screening results
Based on the initial chassis cell screening titer measurements and Western blotting (WB) detection results, target bands could only be detected from the cell wall pellet, and none from the supernatant. There was an issue where LL-37 was difficult to elute from the cell lysate. Combined with low titer yields, we decided to explore the intracellular localization of LL-37 to address the inability to detect target bands in the supernatant and the initially low production issues.
Objective: Investigate LL-37's intracellular secretion pathway to resolve low yield problems.
Method: We designed a pESC-PGAL1-LL37-eGFP fusion expression plasmid (Figure 3) and successfully transformed it into the CEN.PK113-7D strain for fermentation-induced expression, observing intracellular location via fluorescence microscopy.
Figure 3: Fluorescence localization plasmid construction
Results: Fluorescence microscopy observations showed that the LL37-eGFP fusion protein clearly localized to the cell membrane region (Figure 4). We hypothesize that this phenomenon may result from electrostatic adsorption between LL-37's positive charges and the negative charges on the S. cerevisiae cell membrane surface.
Figure 4: Fluorescence localization results
Based on fluorescence localization results, LL-37 binds to the S. cerevisiae cell membrane through charge attraction. High-salt, low-pH extraction solutions can release LL-37 from the cell membrane through electrostatic shielding, while surfactants dissolve the cell membrane, thereby releasing LL-37. Therefore, we decided to use these two methods to extract LL-37.
Objective: Optimize extraction conditions to successfully extract LL-37 from the cell membrane.
Method: We used high-salt, low-pH extraction solutions to extract LL-37 from the cell membrane. To improve extraction efficiency, we also employed surfactants (SDS, LDS, CHAPS, Tween-20, Triton X-100) for extraction.
Results: High-salt, low-pH extraction solutions successfully released LL-37 from membrane adsorption through electrostatic shielding, allowing detection of WB bands in cell disruption supernatant, though the cell disruption pellet still contained much unextracted LL-37 (Figure 5).
Figure 5: High-salt, low-pH extraction effect
Surfactant extraction efficiency was higher than high-salt, low-pH extraction solutions, extracting most of the LL-37 (Figure 6). However, surfactants clogged C18 columns, preventing HPLC titer detection, so we continued using high-salt, low-pH extraction solutions for subsequent extraction work.
Figure 6: Surfactant extraction effect
In S. cerevisiae, certain endogenous proteases degrade the heterologous gene encoding LL-37, significantly reducing yields. Multiple protease gene knockouts can reduce LL-37 degradation, while multi-copy integration can repeatedly insert LL-37 expression cassettes at specific genomic loci, increasing overall expression levels. Combining both approaches theoretically should achieve synergistic effects for higher yields.
Objective: Construct strains with both multi-gene knockouts and multi-copy integration, test their LL-37 expression levels at 5-L fermenter scale, and identify optimal high-yield engineered strains.
Method: A literature review was conducted to identify genes potentially capable of degrading LL-37. Single-gene knockout strains were then constructed in the pre-selected chassis strain. These single-knockout strains were fermented and screened to identify the genes with the most significant impact. Based on the screening results, strains with combinations of multiple gene knockouts were constructed. Subsequently, a GAL80 gene knockout was introduced into these strains, and the LL-37 expression cassette was integrated via multi-copy integration mediated by Ty1 sequences. These engineered strains were then cultured in fermenters alongside control strains for systematic yield comparison.
Results: Through literature review, 17 protease genes and 6 protein-sorting genes that potentially affect LL-37 yield were selected. Single knockouts of these genes were constructed in the pre-screened chassis strain, from which 10 genes demonstrating significant effects were identified, as shown in Figure 7.
Figure 7: Single gene knockout strain fermentation results. Black indicates that the gene is retained; white indicates that the gene is knocked out.
Based on the top four genes found to substantially impact LL-37 yield, we created combinations of double, triple, and quadruple knockouts, totaling 11 distinct strains. We first transformed these 11 multi-knockout strains with a free plasmid, confirmed gene expression by Western blot, and then measured yield after 48 hours of fermentation using HPLC. The results indicated that several combinations enhanced LL-37 production, with the quadruple knockout strain showing a particularly dramatic increase in yield (Figure 8).
Figure 8: Multiple gene knockout strain fermentation results. Black indicates that the gene is retained; white indicates that the gene is knocked out.
The five-knockout strain yps1Δ;prb1Δ;pep4Δ;ysp3Δ;gal80Δ::LoxP-URA3;ty1::LoxP-KanMX-GAL1-LL37H-GAL10-LL37H was then fermented in a 5-L bioreactor. Fermentation results are shown in Figure 9. The maximum yield reached 68.7 mg/L.
Figure 9: Time-course monitoring of yeast biomass, residual sugar, and LL-37 production.
Small peptide accumulation in cells is limited not only by protease degradation but also by overall cellular homeostasis regulation. The YKO (Yeast Gene Knockout) strain library is a systematic collection of Saccharomyces cerevisiae S288C mutants, each containing a precise deletion of a single non-essential gene. To identify gene knockouts that enhance LL-37 production, we randomly selected strains from this library and transformed them with an LL-37 expression plasmid. Through subsequent screening, we successfully identified several knockout strains that significantly increased LL-37 yield. These candidate strains were then subjected to sequencing analysis to determine the specific gene deletions responsible for the enhanced production phenotype.
Objective: Confirm genotypes of high-yield strains from semi-rational screening, analyze correlations between gene deletions and high LL-37 production, and provide new candidate gene targets for subsequent rational design.
Method: Perform gene sequencing and genotype confirmation on 24 candidate strains from semi-rational screening. Focus on analyzing the top 10 high-yield strains (Z≥3, equation: Zi = (Pi − Pwt)/σ), comparing their genetic differences with wild-type, particularly focusing on genes related to protein homeostasis regulation and stress response pathways.
Results: Sequencing results showed multiple genes can significantly improve LL-37 yields (Figure 10). The top 10 high-yield strains commonly lack genes related to the ubiquitin-proteasome system and endoplasmic reticulum stress response. These deletions significantly enhance small peptide accumulation in cells, breaking through previous limitations of relying solely on protease knockouts. This discovery not only explains the success mechanism of semi-rational screening but also provides new target references for subsequent rational optimization. Detailed annotations are shown in Table 1.
Figure 10: Semi-rational screening results
Table 1: Gene annotations for top high-yield strains
| Number | Z-value | Annotation |
|---|---|---|
| P2 B4 | 4.98084 | Putative ubiquitin-specific protease that cleaves ubiquitin-protein fusions; UBP9 has a paralog, UBP13, that arose from the whole genome duplication |
| P4 A2 | 4.86405 | Dubious open reading frame; unlikely to encode a functional protein, based on available experimental and comparative sequence data; overlaps with verified gene BUD7/YOR299W; mutation affects bipolar budding and bud site selection, though phenotype could be due to the mutation's effects on BUD7 |
| P3 C4 | 3.25414 | Proteasome-binding protein; interacts physically with multiple subunits of the 20S proteasome and genetically with genes encoding 20S core particle and 19S regulatory particle subunits; specifically blocks all three proteasome active sites; exhibits boundary activity which inhibits the propagation of heterochromatic silencing; contains a PI31 proteasome regulator domain and sequence similarity with human PSMF1, a proteasome inhibitor; not an essential gene |
| P4 B6 | 3.17161 | Ubiquitin-specific protease; specifically disassembles unanchored ubiquitin chains; involved in fructose-1,6-bisphosphatase (Fbp1p) degradation; similar to human isopeptidase T |
| P3 G3 | 3.16879 | Serine-threonine kinase and endoribonuclease; transmembrane protein that mediates the unfolded protein response (UPR) by regulating Hac1p synthesis through HAC1 mRNA splicing; role in homeostatic adaptation to ER stress; Kar2p binds inactive Ire1p and releases from it upon ER stress; involved in UPR-mediated suppression of aneuploidy |
| P2 H10 | 3.14384 | ER quality-control lectin; integral subunit of the HRD ligase; participates in efficient ER retention of misfolded proteins by recognizing them and delivering them to Hrd1p; binds to glycans with terminal alpha-1,6 linked mannose on misfolded N-glycosylated proteins and participates in targeting proteins to ERAD; member of the OS-9 protein family |
| P2 A5 | 2.90616 | ER-resident protein involved in peroxisomal biogenesis; ER-localized protein that associates with peroxisomes; interacts with Pex29p and reticulons Rtn1p and Yop1p to regulate peroxisome biogenesis from the ER; role in peroxisomal-destined vesicular flow from the ER; partially redundant with Pex31p; may function at a step downstream of steps mediated by Pex28p and Pex29p; PEX30 has a paralog, PEX31, that arose from the whole genome duplication |
| P3 F6 | 2.47209 | Protein of unknown function; subunit of evolutionarily conserved EMC (Endoplasmic Reticulum Membrane Complex) implicated in ERAD (ER-associated degradation) and proper assembly of multi-pass transmembrane (TM) proteins; EMC acts in yeast as an ER-mitochondria tether that interacts with outer membrane protein Tom5p of TOM (Translocase of the Mitochondrial Outer Membrane) complex; YDR056C is not an essential protein |
LL-37 production is regulated by GAL promoter expression. Our goal is to modify the original promoter to obtain stronger promoters, thereby increasing target product synthesis yields.
Objective: Design and modify the original inducible GAL promoter to control expression levels and substantially increase LL-37 expression.
Method: Construct reporter plasmids and perform multi-dimensional promoter engineering:
Results: Replacing the GAL promoter's GAL UAS with CLB2 UAS had minimal impact on cell growth, and the modified promoter strength significantly increased with increasing CLB2 UAS tandem numbers.
Due to initially low LL-37 synthesis yields and high time costs for WB detection, we attempted to construct reporter plasmids by connecting blue-green fluorescent protein genes at both ends of the GAL promoter, using fluorescence intensity to characterize promoter strength.
First, we used homologous recombination to construct two plasmids: connecting green eGFP downstream of the GAL1 promoter and blue mTagBFP2 downstream of the GAL10 promoter. After construction and bacterial verification, they were transformed into yeast for induction. Fluorescence microscopy observations confirmed successful construction of both plasmids (Figure 11).
Figure 11: Successful construction of reporter plasmids
After successful individual constructions, both fluorescent protein genes were connected to both ends of the GAL promoter in one plasmid, with fluorescence microscopy used to observe promoter expression. Although plasmids still showed uneven expression with obvious intercellular differences, both fluorescent proteins were clearly simultaneously expressed in the same cell, allowing analysis of overall fluorescence expression intensity (Figure 12).
Figure 12: Synergistic expression of blue and green fluorescent proteins driven by galactose-inducible promoter in *S. cerevisiae*
Using error-prone PCR for single-point random mutations of the GAL promoter sequence, we transformed the constructs into the same yeast strain CEN.PK 113-7D, using 96-well plates for high-throughput screening of the four strongest blue and green fluorescence intensities, then sequencing and comparing with original sequences. We conducted two rounds of experiments, but sequencing results showed no mutations occurred in these wells.
The first round of screening identified 4 wells with the highest relative fluorescence intensity: A11, B10, H1, H12 (Figure 13).
Figure 13: First round screening: Bidirectional GAL promoter engineering and dual fluorescence kinetic detection
The second round of screening identified 4 wells with the highest relative fluorescence intensity: A6, D8, E5, H1 (Figure 14).
Figure 14: Second round screening: Bidirectional GAL promoter engineering, dual fluorescence kinetic detection, and OD₆₀₀
For combinatorial promoter construction, the original GAL promoter UAS sequences remained unchanged, with strong core promoters TEF1 and GAP replacing the original GAL1. If effective, both GAL1 and GAL10 would be replaced simultaneously in one plasmid to improve construction efficiency (Figure 15).
Figure 15: Combinatorial promoter construction strategy
TEF1 and GAP sequences were directly synthesized by our sponsor Qingke Company. We first constructed two plasmids connecting single core promoters, amplified fragments by PCR, connected with backbone, and after normal transformation into yeast and plating, found fluorescence in colonies before induction due to promoter leakage (Figure 16).
Figure 16: Plate colonies before induction. (a) TEF1-connected plasmid (b) GAP-connected TDH3 plasmid
Subsequently, we constructed promoter plasmids with TEF1 and GAP connected on both sides of UAS, transformed into E. coli for plating, with most colonies verified correct by colony PCR (Figure 17).
Figure 17: Colony PCR verification of hybrid promoter plasmid construction
Correct bacterial plasmids were then transformed into yeast for induction, with fluorescence kinetics detection showing enhanced expression from combinatorial promoters but affecting cell growth (Figure 18).
Figure 18: Real-time fluorescence kinetic monitoring of hybrid promoter expression levels
We replaced the original UAS GAL with UAS CLB2, creating tandem arrays of one, two, and three UAS_CLB2 units (Figure 19).
Figure 19: Promoter engineering strategy for UAS sequence replacement
Three plasmids were constructed through homologous recombination, transformed into E. coli for bacterial PCR, with experimental results showing successful construction (Figures 20, 21).
Figure 20: Nucleic acid gel of UAS_CLB2×1 and UAS_CLB2×2 promoter construction
Figure 21: Nucleic acid gel of UAS_CLB2×3 promoter construction
The three plasmids were then separately transformed into yeast, inoculated into 96-well plates for induced fermentation, with 30 samples per plasmid plus controls for high-throughput screening, measuring fluorescence intensity at 0h, 24h, 36h, and 48h, with statistical data analysis (Figure 22).
Figure 22: Cell fluorescence intensity and growth OD₆₀₀ for 1×, 2×, 3× tandem arrays measured at 0h, 24h, 36h, 48h. a-d: Green and blue fluorescence time-course measurements; e-h: Real-time OD₆₀₀ measurements
In previous experiments, LL-37 was successfully synthesized intracellularly but consistently failed to be secreted extracellularly. This significantly complicates efforts to increase LL-37 yield and subsequent industrial-scale purification. Finding a method to promote its extracellular secretion is therefore essential.
Objective: Screen for effective signal peptides and fusion tags that facilitate the extracellular secretion of LL-37.
Method: We tested the impact of five different leader signal peptides, the α-factor signal peptide, and a SUMOSTAR fusion tag on LL-37 secretion.
Results: Initially, we constructed plasmids containing genes for five leader signal peptides (SUC2 sp, INU1 sp, PHO5 sp, SCW11 sp, HECH sp). After induction, both the supernatant and pellet of the fermentation broth were analyzed. The results were consistent: no LL-37 production was detected in the supernatant (Figure 23).
Figure 23: WB detection of LL-37 in supernatant and pellet of 6 expression combinations in three chassis cells
Subsequently, since the α-factor is a typical secretory signal peptide, we fused it to the N-terminus of LL-37. After induction, the supernatant and pellet were analyzed by Western blot. Although no corresponding band was detected in the supernatant, two distinct bands were observed in the whole-cell lysate (Figure 24). The 14.6 kDa band corresponds to the unprocessed precursor peptide, while the significantly darker 5.4 kDa band corresponds to the mature, cleaved peptide. This indicates that the α-factor signal peptide successfully guided LL-37 to the Golgi apparatus. During the late stages of the transport pathway, it was localized to the Golgi membrane and cleaved by the serine protease Kex2, resulting in the formation of mature LL-37.
Figure 24: WB detection of LL-37 synthesis with α-factor connection. S: supernatant, W: whole cell lysate, fermentation time: 36h
Despite these results, we continued efforts to achieve extracellular secretion of LL-37. We constructed a rational mutant of the SUMO tag, SUMOSTAR (R64T, R71E). Expression in the initial chassis strain was successful, and the expected bands were detected: the full-length precursor and the processed form of the signal peptide (Figure 25). However, LL-37 was still not detected extracellularly, potentially due to insufficient signal peptide efficiency. Subsequent work will focus on signal peptide engineering.
Figure 25: Western blot detection results of SUMO tag fusion protein
As the signal peptides did not effectively facilitate extracellular secretion of LL-37, we decided to employ fluorescent co-localization to investigate its intracellular transport pathway.
Objective: Investigate the intracellular transport pathway of LL-37.
Method: Use fluorescent co-localization techniques to determine the subcellular localization of LL-37 and identify potential issues in its intracellular transport process.
Results: We constructed pESC-PGAL1-LL37-eGFP and transformed it into the CEN.PK113-7D strain. After inducing expression, clear targeting to the cell membrane was observed via fluorescence microscopy (Figure 26).
Figure 26: Transport tracing of LL-37 using fluorescence microscopy
We also performed fluorescent localization for α-factor-LL37-eGFP, making observations after 12h, 24h, and 36h of induction. The results showed that LL-37 was indeed transported to vesicular structures suspected to be the endoplasmic reticulum and Golgi apparatus. However, it appeared to be stalled at this stage. Furthermore, due to the forced guidance by the signal peptide, there were very few instances of free LL-37 adsorbing to the cell membrane and lighting up the entire membrane (Figure 27). Additionally, due to the properties of LL-37, even if secreted, it may directly adsorb back to the cell membrane, resulting in little presence in the supernatant.
Figure 27: Fluorescence microscopy observation of *LL37-eGFP* localization every 12 hours
To gain more precise insight into the specific transport route of LL-37, we performed subcellular localization via fluorescent co-localization. We separately constructed strains expressing mCherry fused to marker proteins for the endoplasmic reticulum (CNE1), nucleus (NYV1), mitochondria (COX4), peroxisome matrix (PEX8), vacuole (PRC1), and cell membrane (SNC1). LL-37 was C-terminally fused to eGFP. The experimental results indicated that LL-37 accumulates extensively on the membranes of organelles (Figure 28).
Figure 28: Fluorescence co-localization of organelle guide proteins, LL-37, and merged images
To achieve large-scale LL-37 production, we aim to construct a low-cost, stable expression, and easy-to-operate fermentation system. We plan to remove the GAL system's galactose dependence through genetic engineering and alleviate glucose repression of expression, thereby simplifying fermentation processes and reducing costs.
Objective: Construct a yeast expression system independent of galactose induction and not repressed by glucose, achieving efficient LL-37 expression under single carbon source conditions, and screen for the most suitable low-cost, high-efficiency fermentation carbon source.
Method: We used the Cre-loxP system to knock out the chassis strain's GAL80 gene, removing its repression of Gal4 (Figure 29), allowing GAL system activation without galactose. We tested the effects of 8 different carbon sources on cell growth and LL-37 expression in gal80Δ strains.
During fermentation, samples were taken regularly to detect OD₆₀₀ and residual sugar to record chassis cell growth status. Western blotting (WB) and HPLC were used to analyze LL-37 expression levels under different carbon sources, combined with cost assessment to determine the optimal carbon source.
Figure 29: Principle of *GAL80* repression of *GAL4*
Results: We successfully constructed the gal80Δ chassis. WB results (Figure 30) showed this strain could initiate LL-37 expression without galactose and at low glucose concentrations, with expression levels increasing over fermentation time.
Figure 30: Detection results of OD and Western blot (WB) using different carbon sources
Among 8 carbon sources, glucose, fructose, sucrose, and maltose all ensured normal cell growth while achieving high LL-37 expression (Figure 31). After comprehensive consideration of cost and LL-37 expression efficiency, glucose was selected as the optimal fermentation carbon source.
We also preliminarily designed and constructed MIG1 knockout components, which will be used to further relieve glucose repression to achieve continuous LL-37 expression. Please refer to the Engineering page for more specific details.
Figure 31: Fermentation carbon source test results
Fermentation media contains large amounts of yeast endogenous fragments including proteins, metabolites, as well as medium components that may interfere with LL-37 purification. As LL-37 is a small antimicrobial peptide with characteristics significantly different from common proteins, we need to find a purification process suitable for LL-37.
Objective: Establish a purification method suitable for LL-37 to provide reliable samples for subsequent LL-37 performance evaluation.
Method: We tested Ni-NTA column purification, utilizing affinity between 6×His tag and Ni²⁺ for gradient elution separation of LL-37. We also tested magnetic bead purification, utilizing magnetic beads to improve LL-37 binding efficiency. We used Western blotting and Coomassie blue staining to evaluate LL-37 content at each step to verify purification effects of different methods.
Results: Ni-NTA column purification failed: Flow-through contained large amounts of LL-37, indicating almost no LL-37 bound to column material, possibly related to its small molecular weight and poor tag exposure (Figure 32).
Figure 32: Ni-NTA column purification results
Magnetic bead purification succeeded. Western blotting results shown in Figure 33: almost no LL-37 remained in flow-through, and clear LL-37 bands were detected in eluate, indicating magnetic bead purification is suitable for LL-37 purification.
Figure 33: Magnetic bead purification results
We also performed mass spectrometry analysis on desalted LL-37 obtained from magnetic bead purification, with results shown in Figure 34. Q2 represents LL-37 after magnetic bead purification and desalting, showing highly similar charge state distribution to His-tagged LL-37 standard, confirming successful His-LL-37 purification through magnetic beads and desalting, further demonstrating excellent magnetic bead purification effects.
Figure 34: Mass spectrometry analysis
To enhance the antimicrobial activity of LL-37, we plan to utilize model predictions to identify LL-37 variants that can significantly improve its antibacterial efficacy.
Objective: Use computational models to generate LL-37 variants to identify variants with superior antibacterial effects.
Method: The dry lab predicted LL-37 variant sequences with better antibacterial effects and greater stability. The wet lab synthesized variants using cell-free systems and determined the best antibacterial effects through plate counting.
Hypothesis: Predicted LL-37 variants include some with antibacterial effects superior to the original LL-37.
Results: N-terminal amidation significantly improved LL-37's antibacterial effect.
To improve LL-37's antibacterial efficiency, we constructed antimicrobial peptide evolution models and antimicrobial peptide activity evaluation models to generate variants and assess their antibacterial effects. Based on model predictions, we screened four best-performing candidate variants.
Figure 35: Structural diagrams of selected LL-37 variants based on modeling results
First, to determine D2P method synthesis effects, we successfully used homologous recombination to insert eGFP and LL-37 sequences into pD2P plasmids, verified by sequencing, then synthesized eGFP and LL-37, determining yields through fluorescence intensity measurement and WB detection.
Figure 36: Fluorescence kinetics of the D2P reaction
Results showed the D2P method could synthesize LL-37 and eGFP, reaching peaks at 1.5h and 4h respectively. Next, we used the D2P method to synthesize LL-37 variants, finding that eGFP control reaction solution addition clearly aided E. coli growth. Through company consultation, we learned that ProteinFactory's reaction system contains abundant energy and substrates needed for transcription and translation, being nutritionally rich, requiring subsequent centrifugation to remove the reaction solution. Meanwhile, LL-37 measurement results were unsatisfactory due to small round granular flocculent material appearing in 96-well plates (Figures 37, 38).
Figure 37: Antibacterial testing of LB medium and mixed reaction solution
Figure 38: Photograph of a 96-well microtiter plate containing antimicrobial assay solutions
We hoped to determine the cause of the flocculent material through inverted fluorescence microscopy observation (Figure 39). A is LB medium, B is LB+eGFP, C is LB+LL-37 and variants, D is LB+LL-37+no bacteria. B showed obviously higher E. coli density than A, C showed small amounts of E. coli, and D and C showed small round granular flocculent material. Multiple experiments found it slowly aggregated over time, with aggregation speeds varying greatly at different concentrations, and contamination causes were excluded.
Figure 39: Microscopic observation of different culture medium samples
Unable to determine the cause of the flocculent material, we ultimately chose the dilution plate spreading method, calculating bacterial survival rates to judge antibacterial effects. The results showed that LL-37 variant 1 had the best effect (Figures 40, 41).
Figure 40: Survival rates of bacterial cells after treatment with LL-37 and its variants
Figure 41: Representative images of bacterial colonies on agar plates at different time points
