Knockout primers (mrdA-F, mrdA-R), identification primers (mrdA-R1), and gRNAs were designed using Snapgene, based on the mrdA gene sequence (GenBank: X04516.1) (Supplementary Table 1). Bacterial genomes were edited using the CRISPR-Cas system in conjunction with the Red recombination system. Knockout positive clones (∆mrdA) were successfully obtained.

Prior to experimentation, 96-well plate covers were treated with a solution of 9 mL absolute ethanol and 1 mL Triton X-100 for 15 s (Brewster, 2003), placed under a UV lamp in a laminar flow hood for 15 min for ventilation and drying. Overnight cultures of both the E. coli MG1655 mrdA mutant and wild-type strains were diluted to a final concentration of OD600 = 0.01 in Luria–Bertani (LB) broth and transferred into the prepared 96-well plates, with four replicates for each strain. The plates were then sealed with parafilm to prevent evaporation. Growth curves were measured using a microplate reader under the ABS kinetic method, with readings taken at 600 nm (OD600) every hour for 15 h at a constant temperature of 37°C. Each well was shaken circularly for 3 s before every reading.

Minimal inhibitory concentrations (MICs) of melittin were determined for both E. coli MG1655 mrdA knockout and wild-type strains using Mueller–Hinton (MH) and LB media (Wiegand et al., 2008). Wild-type and mutant strains were cultured in LB medium to the stationary phase, and the optical density (OD) was uniformly adjusted to 1.0. Melittin was added to a final concentration of 20 μg/mL and placed in a shaker at 37°C and 220 rpm for 2 h. Bacterial cultures were serially diluted across six gradients (10-fold dilutions), and 10 μL of each dilution was spotted onto LB agar plates. After 8 h of inverted incubation at 37°C, bacterial growth was observed to assess melittin resistance.

The pET-28a (+) plasmid was employed as the expression vector, using EcoRI and SalI as the restriction enzyme sites. Primers containing homologous recombination arms and CDS amplification primers were designed using Snapgene (Supplementary Table 2). The E. coli BL21 (DE3) strain was used as a host to successfully construct an overexpression strain of mrdA (designated as mrdA-OE). Additionally, a pET-28a (+) empty vector was transformed into E. coli BL21 (DE3) as a control strain (vector).

Cultures were grown in self-induction medium (Supplementary Tables 3, 4) (Studier, 2014), and target protein expression was assessed at OD600 values of 0.4, 0.6, 0.8, and 1.0. For each time point, 100 mL of bacterial culture was collected, centrifuged at 8,000 rpm for 5 min at 4°C. The supernatant was discarded, and the pellet was resuspended in 10 mL of ice-cold phosphate-buffered saline (PBS) buffer. After repeated centrifugation, the pellet was further resuspended in 1 mL of ice-cold ultrasonic buffer (Supplementary Table 5) and thoroughly mixed. The bacterial suspension was sonicated using an saline ice bath for ultrasonic rupture of bacteria, with ultrasonic parameters set to 400 W, working for 5 s and with 6-s intervals, for a total of 20 min. Subsequently, centrifugation was performed at 14,000 rpm for 40 min at 4°C to separate the supernatant and pellet. The pellet was suspended in 800 μL PBS, and total protein concentrations in both the supernatant and pellet were measured using the bicinchoninic acid assay (BCA) method. Protein loading was standardized to 40 μg, with a loading volume of 20 μL. Samples were incubated with 5× protein loading buffer in a 95°C-water bath for 5 min, followed by centrifugation at 12,000 rpm for 1 min. Protein expression was quantified using 10% sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) with Coomassie blue staining and western blotting (WB). WB was used to detect protein expression (Supplementary Table 5).

The MICs of melittin were determined for E. coli strains overexpressing mrdA. Spot plate experiments were performed by exposing the mrdA-overexpressing E. coli strain to melittin for 2 h. Simultaneously, 5 μL of the mrdA-overexpressing strain and the control strain (cryopreserved at −80°C) were inoculated into 5 mL of LB medium containing kanamycin (Kan) at a dilution of 1:1000 and cultured at 37°C for 12 h on a shaker at 200 rpm.

The bacterial suspensions were then diluted from the induction medium to a final concentration of OD600 = 0.01 and transferred into 96-well plates. A plate reader was used to measure OD600 at a constant temperature of 37°C, with readings taken every hour for a total of 6 h. Upon reaching an OD600 of 0.6, melittin at 4 times the MIC concentration was added to the wells, and measurements were continued for an additional 13 h using the plate reader.

Mutant and wild-type E. coli strains were cultured in LB medium for 12 h overnight, while bacterial cultures containing either the empty vector or mrdA-overexpressed strains were grown in autoinduction medium for the same duration. All bacterial cultures were adjusted to an OD of 1.0 in a uniform volume of 150 mL. Melittin was then added at a final concentration of 20 μg/mL, and the cultures were incubated at 37°C with shaking at 200 rpm on a shaker for 2 h. The OD was readjusted to 1.0, and the final volume of each culture was standardized to 100 mL.

The cells were harvested by centrifugation at 8,000 rpm for 5 min at 4°C, and the wet bacterial pellet was weighed. The pellet was washed four times with PBS and suspended in 5 mL of ice-cold PBS. The bacteria were ruptured using sonication at 4°C and the resulting suspension was centrifuged at 14,000 rpm for 40 min. The isolated supernatant was filtered through a 0.22-μm filter membrane and stored at −40°C for subsequent analysis. Parameters for high-performance liquid chromatography (HPLC) are shown in Supplementary Table 6.

Reagents and solutions used in this experiment were prepared as shown in Supplementary Table 7. Peptidoglycan sacculus were extracted from each strain following the classical method described by Schaub and Dillard (2017). Strains were cultured to stationary phase in 2,000 mL of LB medium, and the pellets were harvested by centrifugation at 6,000 × g for 10 min at 4°C. Each strain’s pellet was uniformly weighed to 4 g, resuspended in 10 mL of PB solution, and added dropwise to an equal volume of boiling PB buffer containing 8% SDS (w/v). This mixture was further boiled for 1 h.

After boiling, the suspension was cooled to room temperature and placed on a shaker at 220 rpm overnight at 37°C. The suspension was then centrifuged at 130,000 × g for 30 min at 15°C. The supernatant was discarded, and the pellet was washed multiple times with PB solution, followed by centrifugation until the SDS was fully removed. The resulting pellet was the extracted peptidoglycan “sacculus,” which was weighed and documented.

A 1 mg/mL solution of the peptidoglycan in PBS was prepared, and melittin was added to a final concentration of 20 μg/mL. This mixture was placed on a shaker at 37°C with a speed of 220 rpm for 3 h. After the reaction, the suspension was centrifuged at 130,000 × g for 30 min, and the supernatant was collected for HPLC assay analysis. The amount of melittin adsorption capacity of the extracted peptidoglycan from different strains (intracellular melittin, or GQ) was calculated using the following formula: Adsorption = (amount of GQ added − amount of GQ in supernatant)/weight of peptidoglycan (mg).

E. coli cultures were prepared by adding 5 mL of LB medium to the cells, followed by incubation at 37°C with shaking at 200 rpm until the OD reached 1.0. After routine culturing, the bacterial suspension was centrifuged and the supernatant was discarded. The pellet was then fixed with a Gluta fixative (2.5% glutaraldehyde, v/v) at a volume 40 times that of the bacterial pellet, mixed thoroughly, and stored at 4°C for 4 h. After fixation, the suspension was centrifuged at 6,000 rpm for 5 min, and the supernatant was discarded. The pellet was washed three times with distilled water, centrifuged, and the supernatant was discarded.

Dehydration was carried out using a gradient of ethanol concentrations of 30, 50, 70, 80, and 90%, with each step lasting 15 min, and the samples were centrifuged at 6,000 rpm for 5 min between each step. The samples were further dehydrated in 100% ethanol for 15 min twice. Following this, the samples were placed in a 1:1 mixture of ethanol and tert-butanol for 20 min and then replaced with 100% tert-butanol twice for 20 min each. After the final replacement, the sample containing a small amount of tert-butanol was pre-cooled and then dried under vacuum in a freeze-dryer precooled at −10°C for 12 h. Once dehydration was complete, the samples were sputter-coated with a gold layer to produce a 5-nm-thick conductive film and observed under a scanning electron microscope (SEM).

Both wild-type and mrdA mutant strains were routinely cultured, with the mrdA-overexpressing strains grown in a self-induction medium. After adjusting the OD of each bacterial solution to 1.0, 1 mL of the culture was centrifuged at room temperature for 5 min at 6,000 rpm. The bacterial pellets were suspended in 300 μL of culture medium. Melittin was added to a final concentration of 20 μg/mL, and the cultures were incubated at 37°C and 200 rpm for 2 h.

Following incubation, 20 μL of each bacterial solution was pipetted onto 1 cm × 1 cm glass coverslips and allowed to dry at room temperature. The dried samples were dehydrated and processed as described above and sputter-coated with gold. The samples were observed under a scanning electron microscope.

For morphological analysis, eight cells were randomly selected from each SEM field of view and measured using the SEM’s built-in software. The data were analyzed and plotted using GraphPad software. SEM images were randomly selected for further examination. The total number of cells and the number of intact cells were counted based on the criteria for cell rupture and loss of cell contents. The proportion of intact cells in each figure was calculated as: the number of intact cells/the total number of cells in the field of view.

Growth curve comparison: The differences in growth curves between different strains were analyzed using repeated measures two-way analysis of variance (ANOVA). A p-value of <0.05 was considered indicative of significant differences in growth between strains.

Antimicrobial peptide resistance comparison: To compare the antimicrobial peptide resistance between overexpressing strains (mrdA-OE) and unloaded strains (vector) at different time points, the following calculations were performed:

OD600 values of E. coli overexpression strains without antimicrobial peptide treatment (mrdA-OE) (a) minus OD600 values of E. coli unloaded transformed strains (vector) at corresponding time points (b), OD600 values of mrdA-OE treated with antimicrobial peptide (a′) minus vector OD600 values at corresponding time points (b′), and a − b values at different time points were normalized by (a′ − b′)/(a − b) to compare antimicrobial peptide resistance between mrdA-overexpression strains and unloaded strains. A ratio >1 was considered evidence of resistance to the antimicrobial peptide in the mrdA-overexpression strains.

Determination of statistically significant differences: ns (not significant): p > 0.05; *: p ≤ 0.05; **: p ≤ 0.01; ***: p ≤ 0.001; ****: p ≤ 0.0001.

The growth trend of mrdA knockout E. coli (∆mrdA) was not significantly different from that of wild-type E. coli (wt) in LB medium (p-value = 0.0812, Figure 1a). Furthermore, there was no significant difference in growth between the two strains at different time points (Figure 1a). The results confirmed that deletion of the mrdA gene did not impact the growth of the E. coli MG1655 strain.

The findings from our melittin resistance testing have provided intriguing insights into the role of the mrdA gene in E. coli’s defense against melittin, a potent antimicrobial peptide. Our data revealed a stark difference in the minimal inhibitory concentration (MIC) of melittin required to inhibit the growth of the wild-type and the mrdA knockout (∆mrdA) strains of E. coli. Specifically, the wild-type strain demonstrated a higher tolerance to melittin, with an MIC of 32 μg/mL, indicating that it could withstand relatively higher concentrations of melittin without its growth being significantly impeded. In contrast, the ∆mrdA strain exhibited a considerably lower MIC of 8 μg/mL, suggesting that it is more susceptible to the antimicrobial effects of melittin. This marked increase in sensitivity in the absence of the mrdA gene underscores the gene’s potential role in conferring resistance to melittin. The substantial reduction in the MIC for the ∆mrdA strain implies that the mrdA gene may play a crucial part in the bacterial cell’s defense mechanisms against melittin, possibly by modulating the cell’s outer structure to impede melittin’s access to its targets within the cell.

Spot plate experiments additionally confirmed that the wild-type strain grew over time despite exposure to melittin at a final concentration of 20 μg/mL. Colony numbers were 100-fold higher after 2 h of melittin treatment compared to 0 h. ∆mrdA, on the other hand, showed a significant numerical reduction, with colony numbers decreasing to 3 × 10⁻⁴ after melittin treatment for 2 h compared to 0 h (Figures 1b–d).

Before induction, the MIC values for melittin were the same for both E. coli containing the empty vector (vector) and the mrdA-overexpressing strain (mrdA-OE), with a value of 4 μg/mL in LB medium.

After induction in the specialized medium, vector and mrdA-OE showed different degrees of reduction with increasing antimicrobial peptide treatment time. The mrdA-OE strain showed stable resistance to melittin (Figures 2a,b), with an average resistance that was 1.23-fold higher than the vector strain (Figure 2b, mean resistance at each time point after melittin treatment), calculated from OD values at various time points.

Spot plate results further confirmed these observations, showing that, after melittin for 2 h, the number of colonies in the vector strain decreased approximately 5-fold compared to the initial count at 0 h, whereas mrdA-OE colonies increased 5.6-fold, exhibiting enhanced resistance to melittin (Figures 2c–e).

After 2 h of melittin treatment, the intracellular melittin content of wild-type (wt), mrdA knockout (∆mrdA), vector, and mrdA-overexpressing (mrdA-OE) strains was quantified using HPLC. The results showed that the intracellular melittin content per gram in the wt, ∆mrdA, vector, and mrdA-OE strains was 68.89 μg, 42.33 μg, 83.19 μg, and 1.54 μg, respectively. ∆mrdA intracellular melittin content was 61.45% of that in the wt strain, which was significantly different (p = 0.0008). The intracellular melittin content of mrdA-OE was only 1.85% of that of vector, and this difference was also significant (p < 0.0001) (Figure 3).

It is hypothesized that mrdA-overexpression increases the degree of peptidoglycan cross-linking and peptidoglycan thickness in host bacteria, thereby significantly restricting the entry of melittin into cells. Conversely, the lower melittin content in ∆mrdA mutant compared to the wt strain may be due to the extensive cell death of ∆mrdA observed after 2 h of melittin exposure (reflected by the 30,000-fold reduction in colony-forming unit after treatment). The high level of cell death likely resulted in a significant amount of melittin being bound to the remnants of dead cells, thereby depleting melittin in the medium, resulting in a lower intracellular concentration decrease in the surviving mutant cells.

The results of the sorption analysis showed that peptidoglycan content in the E. coli cell wall was significantly related to mrdA expression. The peptidoglycan (PG) content extracted per gram of bacterial cells was 89.08 μg, 15.00 μg, 131.67 μg, and 169.83 μg for wt, ∆mrdA, vector, and mrdA-OE strains, respectively. The peptidoglycan content in ∆mrdA was 16.84% of that in the wt strain, and the difference was significant (p = 0.0008). The peptidoglycan content in mrdA-OE was 1.29-fold higher than vector and this difference was significant (p = 0.0122) (Figure 4a).

The ability of peptidoglycan to adsorb melittin was examined after 3 h of treatment. The results showed that this ability was related to the degree of cross-linking of peptidoglycan. The amount of melittin adsorbed per mg of extracted peptidoglycan was 6.77 μg, 0.66 μg, 2.13 μg, and 4.59 μg for wt, ∆mrdA, vector, and mrdA-OE strains, respectively. ∆mrdA peptidoglycan showed 9.75% adsorption capacity for melittin compared to the wt strain, which was significantly different (p = 0.0002). The adsorption capacity of mrdA-OE peptidoglycan for melittin was 2.14-fold higher than that of vector, and this difference was significant (p = 0.0029) (Figure 4b).

SEM analysis revealed that both wt and ∆mrdA strains of E. coli were rod-shaped in the absence of melittin treatment (Figure 5a), and the sizes of these two strains were not significantly different (Figure 5b). The average length of wt and ∆mrdA strains were 1.35 μm and 1.59 μm, respectively (p = 0.1072), and 0.53 μm and 0.52 μm wide in diameter, respectively (p = 0.8895).

Both wt and ∆mrdA strains showed cell deformation and death following melittin treatment. However, the morphology and processes were notably distinct. Cells in the wt strain showed one or more localized pitting changes after melittin exposure (Figure 6, green arrows). The weakest depressions eventually ruptured, leading to cell death and the release of intracellular material (Figures 6a3,a5, orange arrows). In contrast, no obvious depressions were observed on the surface of ∆mrdA cells; instead, the entire cell became permeable except at both ends, and subsequently, multiple ruptures occurred on the cell surface and rapid cytolytic death occurred (Figures 6a2,a4,a6, yellow arrows). ∆mrdA deaths occurred rapidly and were more extensive compared to those in wt cells after 2 h of melittin treatment. The diffuse white highlights visible in the SEM images correspond to the intracellular effluent following bacterial death. These highlights were more pronounced in the picture background of ∆mrdA images compared to the wt strain images.

In addition, the proportion of intact cells treated with melittin was calculated as determined by the presence of cell rupture and cellular content leakage. As shown in Figure 6b, the average intact cell number in the wt strain was 74%, while it was only 3% in the ∆mrdA strain, and the latter was only 4.05% of the wt intact cell count, with a significant difference (p = 0.0008). These findings are consistent with the previous results from the dot-plate experiments (Figures 2c–e). It also further confirmed the speculation about the reason for the reduction of intracellular melittin content in ∆mrdA cells.

Vector and mrdA-OE E. coli strains retained their rod-shaped morphology when cultured in autoinduction medium until the pre-logarithmic growth phase (OD = 0.4) (Figure 7a). There was no significant difference in the size of the two bacteria in both the length (p = 0.8429) and width diameters (p = 0.6558) (Figure 7b). It is important to note that even though the late log phase of growth was not reached and induction had not theoretically started, a very small number of mrdA-OE bacteria exhibited leaky expression and the bacteria became longer (Figure 7a2, as indicated by the purple arrow). These elongated cells were excluded from statistical measurements in the radial line statistics.

Both vector and mrdA-OE cells remained rod-shaped (Figures 7a3,a4) when cultured in autoinduction medium up to OD = 1.0 (both overexpressed and unloaded strains had been induced). However, the long diameter of mrdA-OE increased significantly: 1.7 times higher than that of vector after induction, and 1.62 times higher than that of mrdA-OE before induction (p = 0.0067 and p = 0.0116, respectively), while the width diameter did not change significantly (p = 0.4146 and p = 0.1276, respectively). For the vector strain, there was no significant difference in either the length or width diameters of the radial lines before and after induction (p = 0.8155 and p = 0.1409, respectively) (Figure 7b).

With respect to morphological responses to melittin treatment, vector cells showed significant mortality compared to mrdA-OE when induced strains were treated with melittin for 2 h (Figure 8). In response to melittin, vector cells showed local pore formation (Figures 8a3,a5, yellow arrows), leakage of cellular contents, and cell death, during which cell morphology changed less. In contrast, mrdA-OE showed marked morphological changes, with elongated cells showing segmental collapse and shrinkage, making the rod-shaped morphology uneven in thickness (Figures 8a4,a6, cyan arrows). The vector strain images also revealed more diffuse cellular contents in the background compared to mrdA-OE images.

The proportion of intact cells treated with melittin was further analyzed. As shown in Figure 8b, the mean number of intact cells was 27% in the vector strain and 78% in mrdA-OE, which was 2.89 times higher than vector cells, with a significant difference (p-value = 0.0022).