All bacterial strains used in this experiment were stored at −80°C in culture media containing 25% glycerol. The bacteria were revived in LB agar medium and prepared for experiments. The clinical isolates used in this study were obtained from the Chinese Academy of Agricultural Sciences. Detailed descriptions of these bacteria can be found in Supplementary Table S1.

Halicin was synthesized and identified using NMR in our laboratory (Purity, 99.5%). Carboxymethyl cellulose sodium (CMC-Na) was purchased from Shanghai Chemical Reagent Co., Ltd. Cyclophosphamide was purchased from Beyotime Biotechnology Co., Ltd. Ciprofloxacin was purchased from Energy Chemical Shanghai Co., Ltd.

The test strains were inoculated on the corresponding solid medium and incubated overnight at 37°C. A single colony was picked from the medium and inoculated into Luria Broth (LB), and then incubated overnight at 37°C with shaking at 220 rpm. The bacterial suspension was diluted to 0.5 McFarland turbidity with Mueller Hinton broth (MHB), and then dilute 100 times for later use. Halicin was added to the 96-well plate, and diluted it to concentrations of 128, 64, 32, 16, 8, 4, 2, 1, 0.5, 0.25, 0.125, and 0 μg/mL by gradient. The 96-well plate was incubated at 37°C for 16 h, followed by observing and recording the results. The MIC was determined as the concentration at which there was no visible colony growth was observed.

According to the determination method of MIC, the bacterial cells were diluted to different concentrations and mixed with halicin, then incubated overnight. Next day, the sample close to the MIC were selected for re-culture, the concentration was adjusted to 1 × 10⁶ CFU/mL, and the above steps were repeated. Meanwhile, the strains were preserved daily, and the changes of MIC were recorded.

Briefly, Escherichia coli (ATCC 25922), Salmonella (CVCC 3377), S. aureus (ATCC 29213), Klebsiella pneumoniae (ATCC 4352) and Proteus (CMCC (B) 49027) (1 × 10⁶ CFU/mL) were transferred into MHB. Different concentrations of halicin were added to achieve final concentrations of 4 ×, 2 ×, 1 ×, 1/2 ×, and 0 ×MIC. The optical density at 600 nm was determined.

After dissolving 2 mg of halicin in 1 mL DMSO, the solution was serially diluted in distilled water. The resulting dilutions were mixed with melted top agar to obtain final halicin concentrations of 45, 22.5, 11.2, and 5.6 μg/plate. Positive controls were used with and without the mammalian microsomal enzyme test (S9). The Ames test was conducted as previously described with modifications (Mello Silva Oliveira et al., 2016; Wahab et al., 2018; Fan et al., 2022). The experiments were performed in triplicate.

Both female and male Sprague Dawley (SD) rats and ICR mice were provided by the Shandong Pengyue Experimental Animal Breeding Center. All experimental animals were housed in standard rodent housing conditions, with a temperature of 20°C–24°C, a relative humidity of 55% ± 10%, and a 12/12-h light/dark photoperiod. They had free access to water and food. All animals underwent a 2-week quarantine period and a physical examination before being provided. Prior to the experiment, animals were allowed a 3-day adaptation period. All animals were fasted overnight before the administration of treatment but had free access to water.

Fifty male and female ICR mice weighing 18–22 g were randomly divided into five groups to conduct an acute oral toxicity test and determine the LD₅₀. The mice were given 4,000, 2000, 1,000, 500, and 0 mg/kg (control group) halicin in 0.1 mL per 10 g of body weight using the gavage method. Following administration, their health status was monitored for a period of 7 days. Survival rates and signs of toxicity were recorded. Furthermore, a post-mortem examination was conducted to observe any pathological changes in the organs of deceased animals (Wang et al., 2017).

Fifty male ICR mice, aged 6–8 weeks and weighing 25 ± 2 g, were randomly divided into the following five groups (n = 10 mice per group): the high-dose group (1009.2 mg/kg), the medium-dose group (504.6 mg/kg), the low-dose group (252.3 mg/kg), the negative control group (0 mg/kg), and the positive control group (40 mg/kg cyclophosphamide). The mice in the experimental groups were orally administered halicin via gavage. The treatment was conducted every 24 h for five consecutive days. After 35 days from the initial administration, the mice were euthanized by cervical dislocation. Mouse spermatozoa were smeared and then stained using eosin staining solution (Beyotime, China). Under low magnification, non-overlapping areas with a uniform distribution were selected. High magnification was then used to observe and record the percentage of sperm abnormalities and the number of different types of abnormalities present (OECD, 2018; Dong et al., 2022).

Forty ICR mice, with a 1:1 male: female ratio weighing 18–22 g were selected. The dose group settings were the same as 2.9. The mice received three consecutive administrations, spaced 24 h apart. Two to 4 hours before euthanasia, the mice were treated with 2 mM colchicine through intraperitoneal injection. After euthanasia, the bilateral femurs were promptly extracted and cleansed of any blood. Following the guidelines for veterinary drug experiments, bone marrow cells were collected for slide preparation. Samples were stained using Giemsa staining solution (Beyotime, China). A total of 100 cells with well-separated chromosomes in the metaphase stage were randomly selected from each sample for observation. Chromosomal aberrations, such as chromosome or chromatid breaks and chromosome deletions, were documented (OECD, 2016; Akagi et al., 2023).

Fifty ICR mice (weight 18–22 g), were divided into the following five groups (n = 10 mice per group). The dose group settings were the same as 2.9. The mice received two consecutive doses, separated by 24 h. All medications were given through tube feeding with 0.5% CMC-Na as the solvent. Six hours after the second administration, the mice were sacrificed. Both femurs were removed, the bone marrow cells were washed with 0.2 mL of calf serum, and the tablets were smeared. Slides were smeared and then stained using eosin staining solution (Beyotime, China). Using a double-blind procedure, 1,000 polychromatic erythrocytes cells were observed from each mouse (OECD, 2016; Guo et al., 2021).

Forty rats (weight 100 ± 10 g) were randomly divided into four groups (n = 10 rats per group). Each rat was housed in a separate cage. The experiment comprised three treatment groups and one negative control group (0.5% CMC-Na). The dosage concentrations for the treatment groups were determined based on the LD₅₀ results obtained from the acute oral toxicity test. The high-dose group received a concentration of 201.8 mg/kg, the medium-dose group received 100.9 mg/kg, and the low-dose group received 50.5 mg/kg. Administration was conducted via oral gavage, with a dosage of 1 mL per 100 g of body weight, for a continuous period of 13 weeks. Throughout the study period, body weight, food intake, and water consumption were recorded daily. Additionally, clinical observations and behavioral assessments were conducted. On day 45, which marked the mid-term of the study, blood samples were collected from the rats through the jugular vein. To minimize external interference with the experimental results, the rats were subjected to a 16-h fasting period before blood collection. Whole blood samples were stored in EDTA anticoagulant tubes, and routine hematological parameters were analyzed using a Mindray hematology analyzer.

Clinical chemistry parameters were measured using the ELLIPSE (YSBAERT) clinical chemistry analyzer. At both the mid-term (Day 45) and the end of the experiment (Day 90), five rats from each group were euthanized by cervical dislocation for necropsy. The hearts, livers, spleens, lungs, and kidneys of the rats were embedded in paraffin, stained with eosin and hematoxylin, and observed under a microscope to assess the impact of the treatment on their organs (OECD, 2018; Guefack et al., 2022).

Under controlled laboratory settings, wild-type zebrafish of strain AB were maintained under a 14/10-h light/dark photoperiod at a constant temperature of 27°C. Approximately 4–5 h prior to the embryotoxicity assay, zebrafish ova were harvested. We primarily utilized ova that had progressed to the blastula stage of embryonic development (OECD, 2013; Adam et al., 2021). Approximately 4–5 h after the fertilization of zebrafish oocytes, halicin was introduced into the culture medium. We employed a 32-well plate format, allocating 20 oocytes to each experimental group. Concentrations of halicin in the culture medium were 128, 64, 32, 16, and 8 μM. As a control, 0.2% DMSO was used. The cultures were maintained at a constant ambient temperature of 27°C. After drug administration, survival of the zebrafish embryos was recorded for 72 h. The LC₅₀ of halicin on zebrafish was obtained by this method.

For embryonic malformation studies, groups of 20 zebrafish embryos each were treated with halicin. The concentrations of the drug were set at 4, 2, 1, and 0.5 μM, in accordance with established protocols for zebrafish embryo acute toxicity test. Observations were conducted at 24, 48, and 72 h after fertilization using a microscope. The embryonic assessment focused on six specific teratogenic endpoints: pericardial edema, yolk sac edema, cardiac malformations, morphological body shape deformities, tail deformities, and craniofacial anomalies, as well as otolith abnormalities. To minimize the impact of random variability, these experiments were performed in triplicate (OECD, 1998).

Six-week-old BALB/c mice, with an equal distribution of males and females (n = 10 per group), were selected for the study. The mice were divided into three groups: control group, halicin group, and ciprofloxacin (CIP) group. Administration was conducted via oral gavage at a dose of 10 mg/kg. The treatment was given once daily, with a 24-h interval, for a continuous period of 5 days. Fecal samples were collected from the mice every day to analyze the changes in the gut microbiota following the drug treatment. The fecal samples were stored at −80°C on day 0, day 4, day 6, day 10, day 18, and day 26 for subsequent 16S rDNA analysis. Genomic DNA from the ground fecal samples was extracted using the Fecal Genomic DNA Extraction Kit (Beijing, Solarbio) for library construction and analysis. Amplification was performed using the primers 515F (5′-GTGCCAGCMGCCGCGGTAA-3′) and 907R (5′-CCG​TCA​ATT​CCT​TTG​AGT​TT-3′). Microbiome bioinformatics analysis was performed using QIIME 2 2019.4 (Bolyen et al., 2019). The sequences were subjected to quality filtering, denoising, merging, and removal of chimeras using the DADA2 plugin (Callahan et al., 2016). The amplicon sequence variants (ASVs) were classified using the classifier (classify-sklearn naive Bayes) from the Feature Classifier plugin, based on the Greengenes 13_8 99% OTU reference sequences (McDonald et al., 2012).

Eight SD rats weighing 200 g on average were used. They were split into two groups, each of which had two males and two females. The drugs were dissolved in a 0.5% CMC-Na aqueous solution to achieve the desired concentration and mixed thoroughly. The medication was given as a single oral gavage at a dose of 1 mL per 100 g body weight. A high-dose group (50 mg/kg) and a low-dose group (10 mg/kg) were set. Blood samples were taken at 0, 1, 2, 4, 8, 12, and 24 h after administration. Liquid chromatography-mass spectrometry (PerkinElmer, Qsight 210) analysis was conducted for the determination of drug levels in the plasma (Q1 262, Q3 133). The 0.1% formic acid in water and acetonitrile were used as the mobile phase. The time-plasma concentration data in the plasma were analyzed and calculated using the PK Solver plugin in Excel, and noncompartmental modeling was employed to calculate various PK parameters.

We prepared an in vivo infection model using 24 BALB/c mice (n = 6 per group) weighing 18–22 g. A single colony of Clostridium perfringens Type A was inoculated into 10 mL of liver broth gastric digestion medium. The culture was grown in an anaerobic chamber for 16 h, and the pellet was resuspended in PBS buffer. Bacterial counts were determined using the agar plate count method. Based on preliminary experiments, the mouse infection model was established through oral administration (10⁹ CFU/mL). One hour after bacterial challenge, halicin was injected intraperitoneally. The mice were divided into four groups: the high-dose group (10 mg/kg), the low-dose group (5 mg/kg), the negative control group (0 mg/kg), and the positive control group (metronidazole, 10 mg/kg). The survival state of mice was observed and recorded.

The LD₅₀ of halicin was calculated using the Karber formula and visualized using GraphPad Prism 8.0.2. The data for daily weight gain, food consumption, and organ weights were presented as mean ± standard deviation. Analysis was performed using one-way analysis of variance (ANOVA) to determine significant differences. Pairwise comparisons between all groups and the control group were conducted using the t-test. Significance was reported for results with a p < 0.05.

As shown in Table 1, the in vitro activity test results demonstrated that halicin exhibited broad-spectrum antibacterial activity. Except for Bacillus subtilis and Pseudomonas aeruginosa, the MIC of halicin against other bacteria was less than 8 μg/mL.

At the same time, we found that halicin had a good inhibitory effect (MIC, 8 μg/mL) on C. perfringens. We conducted in vitro antimicrobial experiments using 10 clinical isolates of C. perfringens strains and 36 clinical isolates of E. coli strains. The results showed that the MICs of halicin ranged from 0.5 to 16 μg/mL against clinical isolates of C. perfringens strains, and from 4 to 16 μg/mL against clinical isolates of E. coli strains (Figure 1). This indicates that halicin has a strong antibacterial activity against both bacteria. Interestingly, we found that E. coli with sulfonamide resistance genes showed an increase in the minimum inhibitory concentration (MIC, 16–32 μg/mL) when faced with halicin. This may imply that the mechanism of action of halicin is similar to that of sulfonamide drugs, and that sulfonamide resistance genes affect the sensitivity of resistant strains to halicin.

After 90 days of continuous exposure to subminimal inhibitory concentration (0.5 × MIC) of halicin (Figure 2), the MIC of S. aureus (ATCC 29213) increased from 8 to 64 μg/mL. Resistant strains were isolated and sequenced using a second-generation sequencing scheme to identify the main cause of the increase in MIC. No clear genomic differences were observed between susceptible S. aureus (ATCC 29213, MIC 8 μg/mL) and resistant S. aureus strains (SAR, MIC 64 μg/mL), indicating that halicin resistance in SAR may be mediated by small chromosomal changes. Compared with susceptible S. aureus (ATCC 29213), SAR revealed 18 SNPs in the coding regions (Table 2). SNP analysis of SAR suggested that resistance to halcin may be related to bacterial protein synthesis (priA, prmA and rot), transport (metC and agcS), methylation regulation (miaB) and nitroreduction (nadE). nadE mutations are important for metronidazole resistance (Danecek et al., 2011). The conversion of cytosine to thymine at nucleotide position 1,955,821 results in the conversion of arginine (codon CGA) to glutaminic acid (codon CAA) at amino acid position 188 of the nadE protein.

The main active group of metronidazole is 5-nitro group, and its antibacterial activity largely depends on the reduction reaction (Oliveira et al., 2018). The nadE gene encodes the NAD⁺ synthase, a crucial enzyme in NAD⁺ biosynthesis that is essential for maintaining intracellular NAD⁺ levels. NAD⁺ is a vital cofactor involved in various biochemical reactions, and affects the cellular reduction environment (Santos et al., 2020). Mutation or expression changes in the nadE gene can lead to adjustments in NAD⁺ levels, which alters the state of cellular reduction, affects the activity of metronidazole and increases bacterial resistance to metronidazole. Given the structural similarity between the nitro group of halicin and metronidazole, we speculate that this is also the main reason for SAR resistance to halicin. Therefore, before halicin is applied in the clinical treatment of infections with multidrug-resistant pathogens, investigating the effect of altered NAD⁺ levels on halicin activity and how variations in the nadE gene contribute to the formation of resistance is a key step in developing effective treatment protocols and preventing the development of resistance.

Growth kinetics curve assays showed that Salmonella, Proteus and K. pneumoniae were completely inhibited by halicin at 1/2 × to 4 × MIC (Figures 3A–C). Interestingly, S. aureus and E. coli still proliferated significantly in vitro under halicin treatment at 4 × MIC after 20 h. We compared the growth rates of the induced resistant bacteria (SAR and ECR) and the wild-type strains S. aureus (ATCC 29313) and E. coli (ATCC 25922) (Figures 3D–G). A specific phenomenon was observed where halicin-treated S. aureus started to grow rapidly after 16 h at 1 × MIC. This result confirmed that halicin was a bacteriostatic rather than a bactericidal agent against S. aureus. A similar phenomenon was observed in E. coli (ATCC 25922) and its resistant strain ECR. However, ECR exhibited a higher bacterial growth rate. At 1/2 × MIC, the growth of E. coli (ATCC 25922) was significantly inhibited during the initial 16 h. However, ECR exhibited noticeable growth at the 8 h.

The in vitro activity and toxicity data suggest that halicin is suitable as a potential antibiotic for the treatment of intestinal bacterial infections. However, the effect of halicin on the intestinal microbiota has not been reported. Therefore, we investigated the effects of 15 mg/kg halicin to conduct intestinal microbiota (Figure 9A). The richness and diversity of the microbial community are reflected in its alpha diversity. Ciprofloxacin (CIP), a commonly used drug for the treatment of intestinal bacterial infections, was used as a positive control. We found that halicin and CIP increased Shannon (p < 0.05) and Faith_pd levels (p < 0.01) 1 day after withdrawal. The results indicated that the diversity and richness of the intestinal microbiota increased significantly after withdrawal, which may be because halicin and CIP are broad-spectrum antibiotics, which have a great impact on the intestinal microbiota after administration, resulting in intestinal microbiota disorder. In the halicin group, the Shannon and Faith_pd values returned to normal levels in the first recovery period, whereas in the CIP group, the Shannon and Faith_pd values recovered in the second recovery period (Figure 9B). We used PCoA methods to analyze beta diversity (Figure 9C). Halicin-treated samples showed no significant separation as the experiment progressed, whereas CIP-treated samples showed significant changes in microbial community. These results indicated that although halicin had a significant effect on gut microbiota, it made them recover faster. CIP treatment caused significant changes in the structure of gut microbiota, which did not return to normal levels 21 days after withdrawal.

The relative proportions of the different taxa were evaluated at the phylum, class, and genus levels (Figure 10). At the phylum level, there was no clear change before and after administration, and phyla mainly included Bacteroidetes, Firmicutes and Proteobacteria, accounting for 94.7%–99.59% of the total proportion, among which Bacteroidetes had the highest proportion (Figure 10A). At the genus level, the main taxon was S24-7, Prevotellaceae, Bacteroidaceae, Lachnospiraceae, Ruminococcaceae, Paraprevotellaceae, Rikenellaceae, Odoribacteraceae, Helicobacteraceae, and Lactobacillaceae. Halicin and CIP caused a significant decrease in the content of Prevotellaceae and Lactobacillaceae after administration. Halicin had less effect on the Prevotellaceae and Lactobacillaceae abundance than CIP, which did not restore these two genera to normal levels during the third recovery period (Figure 10B).

To further explore the effect of halicin on the intestinal bacterial structure at different time points, we performed Lefse analysis to identify different species at different time points (Figure 11A; Supplementary Figure S5). In the fecal samples collected during administration, there was no significant difference between the CIP-treated and halicin-treated groups at the phylum level; however, we found a special species in each group. In the halicin group, the abundance of g_Allobaculum, which has been reported to be associated with beneficial effects in organisms, was increased. In the CIP group administration, we found g_Mucispirillum, which has a protective effect on the intestinal mucosa, but mainly attaches to the intestinal epithelial mucosa and is not often found in feces (Supplementary Figure S5). We speculate that this may be related to serious damage to the mouse intestinal mucosal layer after CIP treatment, which affects the colonization of g_Mucispirillum in the intestinal mucosal layer. In the halicin group, we identified 15 marker species during the second recovery period, including p_Proteobacteria, o_Campylobacterales, c_Epsilonproteobacteria, f_Helicobacteraceae, g_Helicobacter, c_Gammaproteobacteria, c_Deltaproteobacteria, f_Desulfovibrionaceae, o_Desulfovibrionales, g_Flexispira, g_Novosphingobium, f_Alteromonadaceae, g_Alteromonas, f_Tissierellaceae, f_Clostridiaceae, g_Clostridium. Interestingly, most of these marker species belonged to the phylum Proteobacteria. In the first recovery period, the CIP group had more marker species than the halicin group, totaling more than 50 specific species, most of which belonged to the phylum Proteobacteria. However, among these marker species, we identified many bacteria that have not been previously reported in the intestinal microbiota. We believe that these bacteria appeared because CIP seriously damaged the mouse intestinal microbiota structure, causing it to lose its resistance to external bacteria colonization. We performed cluster analysis on Re-HL-1 and Re-CIP-1 groups, and showed obvious cluster differences between the two groups (Figure 11B). This result showed that CIP severely disrupted the homeostasis of the gut microbial system in mice during this period.

Further studies are necessary to fully elucidate the mechanism of action and the potential clinical applications of halicin. The pharmacokinetics (PK) of a drug determines its method of use and range of applications. Therefore, we performed a preliminary evaluation of the oral PK of halicin.

We used a non-compartmental method (NCA) for PK modeling and PK parameter calculations (Table 7; Figure 12) (Shikov et al., 2020). Halicin had a peak plasma concentration (C_max) of 64.25 ± 1.98 ng/mL, and a T_1/2λz between 8–10 h. The drug concentration in the blood was ≤5 ng/mL at 24 h after drug administration. The PK parameters indicated that halicin has a high elimination rate and low blood concentration, making it unsuitable for the treatment of systemic infections. Multiple doses may be required to increase the drug concentration in the blood during treatment. According to the PK parameters and in vitro activity experiments, halicin is suitable for intestinal treatment.

C. perfringens was an anaerobic, gas-producing, spore-forming bacterium that was commonly found in the intestines of humans and animals (Mehdizadeh Gohari et al., 2021). Due to its ability to produce toxins and cause tissue necrosis, C. perfringens infections could lead to severe intestinal diseases. In vitro studies, we found that C. perfringens was sensitive to halicin. We established a mouse intestinal infection model using Ciprofloxacin-resistant C. perfringens isolated from clinical samples via gavage (10⁹ CFU/mL) and evaluated the therapeutic effect of halicin on it.

The mice in the control group (0 mg/kg halicin) had a high mortality rate and appeared wrinkle and abdominal distension 1 day after administration (Figure 13). Within 4 d, 66% of the mice died, and the intestines of the dead mice were inflated. All mice in the 10 mg/kg halicin group survived, and more than 80% of the mice in the 5 mg/kg halicin group survived, and most of the mice did not show ruffled fur. Halicin was much more effective in treating intestinal anaerobic bacteria infections than 10 mg/kg metronidazole. Considering the dosing and LD₅₀ doses, we believe that halicin is a potential drug for the treatment of C. perfringens infections.