From 2022 to 2023, a total of 230 fecal swab samples were collected from swine farms across various districts of Kunming. Specifically, 34 samples were collected from the Luquan district of Kunming in March 2022, 22 samples from the Jinning district in June 2022, 52 samples from the Yiliang district in August 2022, 33 samples from the Anning district in March 2023, 39 samples from Songming in June 2023, and 42 samples from the Xundian district in August 2023. The samples were placed in sterilized centrifuge tubes containing Brain Heart Infusion (BHI) agar, refrigerated, and promptly transported to the Yunnan Joint International R&D Center of Veterinary Public Health for bacterial culture. Initial characterization of the samples began on the day they arrived at the laboratory. This formula, n = (z)²p(1–p)/d², was applied to calculate the required sample size. In the formula, the ‘n’ represents the required sample size; ‘z’ the level of confidence according to a standard normal distribution (for a level of confidence of 95%, z = 1.96); the ‘p’ the expected prevalence, and the ‘d’ allowable error (here it was set to be 5%). The expected prevalence was approximately 18.03% for P. mirabilis, according to the report by Rui et al. (2016).

The fecal swabs were inoculated in buffered peptone water (BPW, Huankai) and incubated in a constant temperature shaker at 37°C, 140 rpm/min for 12 h. Subsequently, 100 μL of the culture was transferred onto Xylose Lysine Deoxycholate (XLD) agar plates and incubated under the same conditions for an additional 12 h. Suspected colonies were selected, re-inoculated onto fresh XLD agar, and incubated at 37°C for 18 h. Finally, a single colony from suspected isolates was then purified by culturing on lysogeny broth (LB) medium. To confirm purity, all isolates were triple-passaged to obtain fresh colonies, followed by Gram staining and biochemical tests, including TSI (Triple Sugar Iron), urea, indole, phenylalanine deaminase, ornithine decarboxylase, and lactose utilization.

Preliminary identification was conducted by amplifying the sequences based on the GenBank-registered ids gene cluster of P. mirabilis. Bacteria with positive ids gene clusters were further analyzed by blasting the 16S rRNA genes. Primers are shown in Table 1.

A total of 30 isolates were subjected to antimicrobial susceptibility testing using the disk diffusion method and minimum inhibitory concentration (MIC) assays. Both methods were conducted according to the guidelines of the American Clinical and Laboratory Standards Institute (CLSI). MIC results were interpreted according to the recommendations of the Clinical Laboratory Standard Institute guidelines (CLSI, 2018; Humphries et al., 2021). Isolates were classified as multidrug-resistant if they were resistant to three or more antimicrobial drugs in the panel (Dargatz et al., 2016). The 19 antibiotics from 8 categories included: aminoglycoside antibiotic have Neomycin (NEO, 30 μg) and Streptomycin (STR, 10 μg); Beta-lactam antibiotics have Cefaclor (CEC, 30 μg), Cefotaxime (CTX, 30 μg), Ceftriaxone (CRO, 30 μg), Cefepime (FEP, 30 μg), Penicillin (PEN,10 units) (for Staphylococcus spp.) or 1 unit (for Streptococcus spp.), Ampicillin (AMP, 10 μg), Imipenem (IPM, 10 μg); tetracycline class of antibiotics have Tigecycline (TGC, 15 μg) and Tetracycline (TC, 30 μg); Fluoroquinolones have Norfloxacin (NOR, 10 μg) and Ciprofloxacin (CIP, 5 μg); Amphenicols have Chloramphenicol (CHL, 30 μg); Macrolide antibiotics have Erythromycin (ERY, 15 μg); lincosamide antibiotics have Clindamycin (CLI, 2 μg); Sulfonamides and trimethoprim have Cotrimoxazole (SXT, 1.25/23.75 μg); While polymyxin antibiotics contain polymyxin E (colistin) and polymyxin B, they were tested using the MIC. The antimicrobial susceptibility results were evaluated based on the CLSI M100-S23 criteria.

The inc/rep PCR method was used to detect replicons on reference plasmids (Carattoli et al., 2005). Eighteen pairs of primers were designed for 18 Inc plasmid genes, including rep and par. Primers are shown in Table 2. After the amplification reaction, the results were observed using 1% agarose gel electrophoresis.

The characteristic gene-positive isolates from the 30 strains were used as donor strains in a conjugation test. The recipient strain was an engineered Escherichia coli DH5α. Donor and recipient strains were mixed in a 4:1 ratio and inoculated into LB medium, followed by incubation at 37°C for 4 h. Subsequently, 100 μL of the donor, recipient, and conjugated cultures were plated on MacConkey agar containing antibiotics to which all strains were resistant. The plates were then incubated at 37°C for 24 h.

Isolates exhibiting severe MDR and containing transfer elements were selected for whole-genome sequencing. DNA was extracted using SDS combined with a purification column. Then, the genome was quality-checked. Qualified DNA was randomly fragmented using Covaris. The ONT SQK-LSK109 and EXP-NBD104/114 kits (Oxford Nanopore Technologies¹) were used for online library construction: the qualified DNA samples were purified using magnetic beads to select DNA fragments with an average size of 200–400 bp.

The 1 μL sample was taken for Qubit quantification, damage repair, end-repair, barcode labeling, pooling library preparation, and sequencing junction ligation. Their whole genomes were sequenced using Illumina MiSeq and Oxford Nanopore MinION platforms. The genomes of this study were assembled using the NECAT software, which used long-read data from the Nanopore. Subsequently, the final high-quality assembly was achieved using Pilon software (v.1.22) to correct the Illumina MiSeq sequencing data based on NGS reads. Multiple assembly versions were mirrored to create the final map.

This study employed the complete level whole genome data from both self-isolated and sequenced strains, as well as 126 publicly available strains. The public complete genome data for 126 strains of P. mirabilis were downloaded from the NCBI database on March 17, 2024. The whole genome data of the self-isolated strains were annotated using the GO, KEGG, and RAST² databases.

PubMLST³ was used to analyze the genomic relationship of selected P. mirabilis and the published strains based on their location and sampling time. The 16 representative strains from various years and regions were selected. Gene loci were plotted using Entro-wgMLST v1.0. The phylogenetic trees were graphed using ggplot2 in R.

The resistome (the entire set of ARGs) of P. mirabilis was analyzed using the Comprehensive Antibiotic Resistance Database (CARD⁴).

The expression of resistance gene types and the phylogenetic analysis of ARGs based on sequence site information across the whole genome datasets were conducted.

IslandPath 1.0.6 software was employed to predict the genome islands and potential horizontal gene transfers using the sequence composition prediction method. The expression of genome island-related genes in the whole-genome-sequenced strains in this study was analyzed. SnapGene Viewer 6.1.2 software was used for the gene island visualization.

The annotation results of these genomes were examined to determine the expression of transfer-related genes. The proportion of Inc plasmid structure genes in P. mirabilis was determined. The presence or absence of repA and repC, along with their sequence site information, was analyzed for evolutionary relationships.

The Ids gene was amplified by PCR. The positive P. mirabilis product was 829 bp. The 16S rRNA gene PCR amplification produced a band at 1,300 bp, which was sequenced and compared using NCBI’s BLAST software.⁵ It revealed that all isolates had more than 98% nucleic acid sequence homology with P. mirabilis. Ultimately, 30 isolates were obtained from 230 samples, resulting in a prevalence of 13.04% (see Figure 1).

The results were evaluated according to the CLSI M100-S23 criteria by the diameter of the inhibitory zone. The summary results are presented in Figure 2. Overall, among all P. mirabilis isolates (n = 30), the highest frequency of resistance was observed against SXT, ERY, PEN, AMP, CHI, STR, TC, and TGC (100%), followed by CEC (97%), CLI (97%), and CIP (70%). All 30 isolates exhibited MDR. Specifically, 4 isolates were resistant to 16 drugs is strain 16, 21, 22, 25 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-NEO-CET-CRO-NOR-CIP-CHI), 3 isolates to 15 drugs is strain 13, 17 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-CET-CRO-NOR-CIP-CHI), and strain 15 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-NEO-CRO-NOR-CIP-CHI), 1 isolate to 14 drugs is strain 8 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-CET-FEP-CIP-CHI), 8 isolates to 13 drugs is strain 3, 23, 24 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CLI-NEO-CET-CIP-CHI); and strain 7, 9 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-FEP-CIP- CHI); and strain 7 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-NOR-CIP-CHI); and strain 10 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CET-CRO-CIP-CHI); and strain 26 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-NEO-CRO-CHI); 8 isolates to 12 drugs is strain 2, 4, 5, 19, 20 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-CIP-CHI); and strain 1 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CLI-NOR-CIP-CHI); and strain 14 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-CET-CHI); and strain 30 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-NOR-CHI), 6 isolates to 11 drugs is strain 11, 12, 18, 27, 28, 29 (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI-CHI).

For antimicrobial resistance, 14 resistance patterns were identified, with the most common being the 10-resistance pattern (AMP-ERY-PEN-SXT-STR-PMB-PMC-TC-CEC-CLI), accounting for 20%, followed by 12-resistance (17%), 15-resistance (14%). The MDR rate (resistant to over nine types of antimicrobials) reached 100%, as shown in Figure 2. According to Magiorakos et al. (2012), the isolated strains were classified into MDR, XDR, and PDR categories, which can be viewed in the attached table.

The results indicated that 15 isolates were positive for the IncQ1α plasmid repC gene, specifically isolates 4–9, 11, 15, 19–21, 23, 24, 26, and 27, with a detection rate of 50%. The repC gene amplification results for IncQ1α were shown in Figure 3, with amplified bands appearing at 802 bp, consistent with the expected results. The remaining isolates were negative for the Inc marker genes. According to the antimicrobial susceptibility results, two multi-drug-resistant and growth-dominant strains were selected for subsequent experiments. Considering the number of multi-drug-resistant in the repC-positive isolates, it can be inferred that NO. 15 and NO. 21 were suitable subjects for further research.

IncQ plasmid-positive isolates formed colorless and transparent colonies on MacConkey’s medium supplemented with 10 μg/mL ampicillin, 30 μg/mL tetracycline, 5 μg/mL ciprofloxacin, and 15 μg/mL erythromycin. The control strain, E. coli DH5α, did not grow under these conditions. The control strain, E. coli DH5α, did not grow under these conditions. No transfer events were observed in any of the conjugation experiments.

Isolates NO. 15 and NO. 21 were both 16-drug-resistant strains and tested positive for the IncQα1 plasmid repC gene.

The assembled genome sequences and functional annotation were mapped using the BLAST Ring Image Generator (BRIG 0.95). The whole genome landscapes of strains 15 and 21 are shown in Figure 4.

This study analyzed two strains of P. mirabilis and 16 strains from diverse global sources using whole-genome phylogenetic tree analysis. The results revealed three major strain groups. Interestingly, no correlations were observed among P. mirabilis strains from human, food, or animal origins. This indicates a close relationship between human-derived P. mirabilis and those found in food, animals, and pets. The two strains studied closely resembled porcine-derived P. mirabilis, reported in 2019 in Henan, China (NCBI genome assembly number GCA_013358795.1) (see Figure 5).

The sequences of strains NO. 15 and 21 were compared with the CARD databases using the Resistance Gene Identifier (RGI). Resistance gene annotation and statistical analysis were conducted for sequences with an identity greater than 95%. The antimicrobial resistance gene information for strains 15 and 21 is presented in Table 3.

Strain NO. 15 carried 33 ARGs, with qacJ present in two subtypes. Ten were efflux pump genes, 14 genes in synthesizing antibiotic-inactivating enzymes, five genes in altering therapeutic targets, and three genes in substituting therapeutic targets. One gene was involved in protecting the therapeutic target. These genes confer resistance to a wide range of antibiotics, such as tetracyclines, fluoroquinolones, aminoglycosides, sulfonamides, polypeptides, chloramphenicols, cephalosporins, lincosamides, macrolides, and 2-aminopyrimidine.

Strain NO. 21 carried 38 ARGs, including two subtypes of qacJ, sul1, catA1, and APH (6)-Id genes. Among the ARGs, 11 were involved in efflux pumps, 16 in synthesizing antibiotic-inactivating enzymes, 7 in target alteration, 3 in target substitution, and 1 in target protection. These genes mediate resistance to tetracyclines, fluoroquinolones, aminoglycosides, sulfonamides, polypeptides, chloramphenicols, cephalosporins, lincosamides, macrolides, and 2-aminopyrimidines.

To identify the most prevalent resistance genes in P. mirabilis, a meta-resistome analysis based on 128 strains (the 126 publicly available data and two strains of this study) of P. mirabilis was performed. According to the annotation results, 105 antimicrobial resistance genes were identified. These genes resisted various antibiotics, including aminoglycosides, beta-lactams, cephalosporins, chloramphenicols, chloromycetin, diaminopyrimidines, elfamycin, glycopeptides, oxazolidinone, phenicol, pleuromutilin, macrolide, lincosamide, streptogramin, monobactam, carbapenem, cephalosporin, penam, nucleoside, phosphonic acid, polypeptides, quinolone, rifamycins, and tetracyclines. The MFS, SMR, and RND efflux pumps were also discovered. There were 12 genes expressed in 90% of the strains. They were cephalosporins ARG PBP3; chloramphenicols ARG catA4; elfamycin ARG EF-Tu; glycopeptides ARG vanG; polypeptides ARG ArnT; quinolone ARG gyrB; MFS efflux pump-related genes KpnE, KpnF; the RND efflux pump-related genes adeF, rsmA, and CRP, the SMR efflux pump-related gene qacJ. Notably, the polypeptide gene ArnT was present in 99.2% of P. mirabilis strains. Alternatively, it might elucidate the principal cause of its intrinsic resistance to polymyxin. Details are shown in Figure 6. Supplementary material presents the phylogenetic relationships of the primary resistance genes (vanG, rsmA, PBP3, KpnH, KpnF, gyrB, EF-TU, CRP, catA4, ArnT, and adeF), including those from 2 strains in this experiment and 16 representative various region strains from public data.