Detection of Carbapenem Resistance of Proteus mirabilis Strains Isolated from Foxes, Raccoons and Minks in China

Simple Summary Proteus is found abundantly in soil and water, and it has been known to cause human urinary tract infections and food poisoning. Currently, the opportunistic pathogen Proteus mirabilis (P. mirabilis) is also found to be an emerging threat to animals, such as birds, fish, dogs, etc. In this study, we examined the antibiotic resistance genes and virulence genes of P. mirabilis isolates from raccoons, foxes and minks. Among a total of 53 P. mirabilis isolates, the proportion of bacteria resistant to three or more antibiotic classes was up to 73.58%, and the detection rate of carbapenem-resistant P. mirabilis isolates was up to 71.7%, putting human health at risk. The close evolutionary relationship between P. mirabilis isolates from animals and the farm environment suggested that multidrug-resistant P. mirabilis from animals could pose a great environmental threat. In addition, the carriage rate of virulence-associated genes was not positively correlated with P. mirabilis pathogenicity in a Galleria mellonella model, highlighting the importance of further understanding the virulence of P. mirabilis in future studies. Abstract Proteus mirabilis, an opportunistic pathogen, is found to be an emerging threat to both animals and humans for a variety of infections. However, the characteristics of P. mirabilis infections from foxes, raccoons and minks remain unclear. In this context, we identified the antibiotic resistance genes and virulence genes of P. mirabilis isolates from foxes, raccoons and minks in China. Most isolates showed resistance to florfenicol (90.57%), trimethoprim-sulfamethoxazole (73.58%), and imipenem (71.70%). A total of 73.58% of isolates were resistant to antibiotics from at least three or more classes, and were categorized as multi-drug resistant. A total of 33.33% of the isolates were resistant to antibiotics from seven classes. The most prevalent resistant were sul1 (94.34%), followed by floR, blaTEM, aac(6′)Ib-cr and blaOXA-1 with the detection rate of 88.68%, 83.02%, 71.70% and 60.38%, respectively. Among the 51 P. mirabilis isolates that were resistant to beta-lactam antibiotics, all isolates carried at least one beta-lactam gene. In addition, blaNDM and blaOXA-24 genes were firstly reported in carbapenem-resistant P. mirabilis isolates from foxes, raccoons and minks. All isolates exhibited the virulence genes ureC, zapA, pmfA, atfA and mrpA. P. mirabilis isolates carrying all detected 10 virulence genes from different animal species showed different lethal abilities in a G. mellonella larvae model. More importantly, the profiles of antibiotic resistance genes of isolates from fur animals and the environment were generally similar, and phylogenetic analysis showed that the P. mirabilis isolates from farm environment samples may have close relatedness with that from animals.


Introduction
Proteus mirabilis (P. mirabilis), a Gram-negative bacterium with flagella and fimbriae, is widely found in water, soil, and the intestinal microbiota of animals and humans. As an opportunistic pathogen, P. mirabilis is usually harmless to human health [1]. However, when the body's immunity is weakened, P. mirabilis may move to non-intestinal sites and cause serious diseases, such as cystitis, sepsis, peritonitis and meningitis [2,3]. After Escherichia coli (E. coli) and Klebsiella pneumoniae (K. pneumoniae), P. mirabilis is the third most common cause of human urinary tract infections [4]. P. mirabilis is also involved in nosocomial outbreaks in neonates in India [3]. In addition, although Proteus does not belong to the foodborne pathogens, some food poisoning cases associated with P. mirabilis have been reported [5,6]. Gong et al. [6] demonstrated that P. mirabilis from humans could induce the symptoms of both vomiting and diarrhea in a mouse model and show high gastrointestinal pathogenicity. Comparative genomic analysis showed that P. mirabilis may acquire horizontally from other microbial genomes toxin genes to exert digestion tract infection and toxicity [7]. Currently, the opportunistic pathogen P. mirabilis is found to be an emerging threat to animals. It was reported that P. mirabilis could cause large-scale mortality of fish without association with any other pathogens [8]. In addition, P. mirabilis infections have been reported in birds with reproductive failure [9], cattle and fowl with diarrhea [10,11] and dogs with chronic otitis externa [12].
Carbapenems, such as imipenem and meropenem, are bactericidal beta-lactam antimicrobials with proven efficacy in severe infections caused by extended spectrum betalactamases (ESBLs, e.g., TEM, SHV, CTX-M enzymes) producing bacteria [13]. Carbapenem resistance occurs mainly among Gram-negative pathogens such as Enterobacteriaceae, Pseudomonas spp. and Acinetobacter spp., which is mainly due to the production of carbapenemase including some class A beta-lactamases (e.g., KPC, GES), class B metallobeta-lactamases (e.g., NDM, IMP) and class D carbapenemases (e.g., OXA-23, -24, -48, -51, and -58) [14]. KPC-type producing P. mirabilis was identified for the first time in the blood culture of a diabetic patient in the United States in 2008; since then, KPC-producing P. mirabilis isolates have been found in humans in many countries [15][16][17]. NDM-producing P. mirabilis isolates have been isolated from commercial broilers in slaughterhouses, ducks, wildlife, and hospitalized patients in China [18][19][20][21]. Among the carbapenem-hydrolyzing class D beta-lactamases, OXA-48, the most frequently identified in Enterobacterales, has only been very rarely reported in Proteus spp., whereas OXA-23 and OXA-58, exclusively identified in Acinetobacter species, have recently been increasingly identified in P. mirabilis from hospital settings, the community, and cattle feces [17,22,23]. In addition, the bla OXA-24 has also been identified in P. mirabilis isolates from hospital clinical specimens in Algeria [22]. Since P. mirabilis is intrinsically resistant to tetracyclines and polymyxins, the emergence of multidrug-resistant, or even extensively drug-resistant P. mirabilis isolates carrying carbapenemase genes complicated the clinical treatment of bacterial infections, causing a significant public health concern. Most previous studies were focused on the prevalence of antimicrobial resistance of P. mirabilis in human urinary tract infections, and chicken, beef, and pork meat [19,21,23,24]. However, there are few reports on the characteristics of P. mirabilis from foxes, raccoons and minks.
The present study aimed to characterize the antibiotic resistance profiles of 53 P. mirabilis isolates obtained from fox, raccoon and mink farms in China, and to compare the genotypic and phenotypic characteristics of antimicrobial resistance and virulence factors in these isolates.

Bacterial Isolates and Growth Conditions
A total of 53 P. mirabilis isolates were used in this study. They were isolated from fox, raccoon, or mink farms in different cities of Shandong Province, China between October 2019 and November 2020 (Table 1). Detailed information on these P. mirabilis isolates is provided in Tables S1 and S2 in the Supplementary Materials. The standard Salmonella enterica serovar Enteritidis (S. Enteritidis) strain CVCC3377 was purchased from the China Veterinary Culture Collection Center (Beijing, China). All isolates were streaked on xylose lysine deoxycholate (XLD) agar plates at 37 • C for 16~24 h, and then colonies were inoculated in Luria-Bertani (LB) medium. Cultures were grown at 37 • C and 180 rpm to the stationary phase for subsequent experiments. Table 1. Fifty-three isolates of P. mirabilis from fox, mink and raccoon farms used in this study.

Isolates
Year of Isolation

Infection of Galleria Mellonella Larvae
In this study, the larvae of the Galleria mellonella (G. mellonella) wax moth was used to assess the pathogenicity of P. mirabilis as described previously with modifications [41]. Stationary bacterial cells were washed with PBS and then diluted in PBS to an optical density at OD 600 of 1, which corresponds to approximately 1 × 10 9 CFU/mL. For 50% lethal dose (LD 50 ) experiments, a series of 10-fold serial dilutions containing 10 4 to 10 8 CFU in PBS were injected into G. mellonella larvae. After surface disinfection using 70% ethanol, a Microliter Syringe (Shanghai Gaoge, Shanghai, China) was used to inject 10 µL aliquots of the inoculum into the hemocoel of each larvae via the last left proleg [42]. A group of 10 larvae were injected with 10 µL of PBS in parallel to ensure that death was not due to injection trauma. Larvae were placed in 9.2 cm Petri dishes and kept at 37 • C in the dark. Insects were considered dead when they displayed no movement in response to touch. The number of dead larvae was scored at 6, 12, and 24 h post-infection. For each strain, at least three independent experiments were performed, and LD 50 s were calculated according to the formula of Käber [43].

Determination of In Vivo Bacterial Loads
Each group containing 15 larvae was infected with approximately 1 × 10 3 CFU per larva of P. mirabilis isolates (F13, R8 and M1) or S. Enteritidis CVCC3377, respectively. Three insects in each group were collected at different post-infection time points (6 h and death time within 12 h) to determine bacterial loads. After surface disinfection with 70% ethanol, larvae were homogenized in 2 mL of PBS by use of a High-Speed KZ-II Basic Homogenizer (Servicebio, Wuhan, China). Serial dilutions of the homogenates in PBS were plated on XLD agar, and colonies were counted after incubation at 37 • C for 24 h. Three independent experiments were performed.

Phylogenetic Tree Analysis
The 16S ribosomal RNA (16S rRNA) gene sequences of 53 P. mirabilis isolates were amplified with primers 27F and 1492R [44]. The PCR products were sequenced and identified using BLAST algorithms against the NCBI databases. A total of 8 representative P. mirabilis isolates from different isolation sources (chicken, snake, milk, etc.) were selected to construct a phylogenetic tree. Multiple alignments of the 16S rRNA gene sequences were conducted with the ClustalW algorithm, and phylogenetic trees were constructed by MEGA 7.0 software (Pennsylvania State University, College Town, PA, USA) using the Neighbor-Joining method [45].

Antimicrobial Susceptibility of P. mirabilis Isolates
As shown in Table 2, all isolates were sensitive to cefepime (53/53, 100%), and most isolates were sensitive to ceftazidime (48/53, 90.57%) and ofloxacin ( (26/53, 49.06%). We found that the proportion of MDR isolates was 73.58% (39/53). Among the 39 MDR P. mirabilis isolates, isolates resistant to seven classes of antibiotics accounted for the highest proportion (33.33%, 13/39) (Figure 1), of which nine were from raccoons and four were from foxes. were from foxes. The isolates from the fox farm had generally higher resistance rate than the isolates from raccoon and mink farms ( Figure 2). Among the P. mirabilis isolates from raccoons and foxes, the significantly different resistance rates to antibiotics were as follows, respectively: to CTX, 40.9% and 94.4% (p < 0.001); to AMP, 54.5% and 88.9% (p < 0.05); to GM, 40.9% and 83.3% (p < 0.01). Meanwhile, by comparing the antibiotic resistance rates between the isolates from minks and foxes, the difference was mainly in AMP (p < 0.01), IMP (p < 0.01), GM (p < 0.001) and STR (p < 0.01). The resistant rates of P. mirabilis isolates from raccoons and foxes to CTX, CAZ and OFX were 9.1%∼94%, but no isolate from mink showed resistance to the above three antibiotics. No difference was found in ENR, SXT and FFC resistance among the isolates from the above three kinds of animals.  The isolates from the fox farm had generally higher resistance rate than the isolates from raccoon and mink farms ( Figure 2). Among the P. mirabilis isolates from raccoons and foxes, the significantly different resistance rates to antibiotics were as follows, respectively: to CTX, 40.9% and 94.4% (p < 0.001); to AMP, 54.5% and 88.9% (p < 0.05); to GM, 40.9% and 83.3% (p < 0.01). Meanwhile, by comparing the antibiotic resistance rates between the isolates from minks and foxes, the difference was mainly in AMP (p < 0.01), IMP (p < 0.01), GM (p < 0.001) and STR (p < 0.01). The resistant rates of P. mirabilis isolates from raccoons and foxes to CTX, CAZ and OFX were 9.1%∼94%, but no isolate from mink showed resistance to the above three antibiotics. No difference was found in ENR, SXT and FFC resistance among the isolates from the above three kinds of animals. Comparison of the antibiotic resistance rates of P. mirabilis isolates between raccoons, foxes and minks. The difference was analyzed by chi-squared test. * p < 0.05; ** p < 0.01; *** p < 0.001.
We also compared the detection rate of antibiotic resistance genes between samples from animals (feces, throat, anal and carcass) and the farm environment (feed and soil), and found that the detection rates of antibiotic resistance genes were generally similar ( Figure 3B). The proportions of blaTEM-positive, sul1-positive and floR-positive isolates from animal samples and farm environment samples were all above 85%. Although blaTEM, blaOXA-1, blaCTX-M, blaOXA-24, blaNDM, aac(6')-Ib-cr, qnrA, qnrC, sul1, sul3, aadA, aadB, aphA6 and floR genes were prevalent in both animal samples and farm environment samples, no significant difference (p > 0.05) was found between the positive rates of the two groups of samples. Figure 2. Comparison of the antibiotic resistance rates of P. mirabilis isolates between raccoons, foxes and minks. The difference was analyzed by chi-squared test. * p < 0.05; ** p < 0.01; *** p < 0.001.

Virulence Genes of P. mirabilis Isolates
Among the 10 virulence genes, ureC, zapA, pmfA, atfA and mrpA genes were all detected among 53 P. mirabilis isolates, followed by both atfC and hmpA genes, both with a detection rate of 98.11%, rsmA at 94.34%, rsbA at 71.70% and ucaA at 45.28% (Table 4).  By comparing the antimicrobial resistance genes from fox, raccoon and mink farms, we found that gene bla TEM , bla OXA-1 , aac(6 )-Ib-cr, sul1, and floR were common. However, bla CTX-M , qnrA and qnrC were only detected in P. mirabilis isolates from raccoons and foxes. In addition, the genes bla PSE , qnrB, aaC2, sul2 and cmlA were specific for the isolates from minks. These results show the diversity of antibiotic resistance genes among different farms.
We also compared the detection rate of antibiotic resistance genes between samples from animals (feces, throat, anal and carcass) and the farm environment (feed and soil), and found that the detection rates of antibiotic resistance genes were generally similar ( Figure 3B). The proportions of bla TEM -positive, sul1-positive and floR-positive isolates from animal samples and farm environment samples were all above 85%. Although bla TEM , bla OXA-1 , bla CTX-M , bla OXA-24 , bla NDM , aac(6 )-Ib-cr, qnrA, qnrC, sul1, sul3, aadA, aadB, aphA6 and floR genes were prevalent in both animal samples and farm environment samples, no significant difference (p > 0.05) was found between the positive rates of the two groups of samples.

Virulence Genes of P. mirabilis Isolates
Among the 10 virulence genes, ureC, zapA, pmfA, atfA and mrpA genes were all detected among 53 P. mirabilis isolates, followed by both atfC and hmpA genes, both with a detection rate of 98.11%, rsmA at 94.34%, rsbA at 71.70% and ucaA at 45.28% (Table 4). To determine the pathogenicity of P. mirabilis isolates from different animal sources, three P. mirabilis isolates (F13, R8 and M1) that carry all detected 10 virulence genes were selected. The LD 50 of P. mirabilis F13, R8, and M1 in G. mellonella larvae model were 2.5 × 10 4 CFU/larvae, 2.0 × 10 4 CFU/larvae and 3.9 × 10 5 CFU/larvae at 6 h post infection, respectively ( Table 5). The results indicate that the virulence of P. mirabilis M1 from minks seemed to be weaker than F13 from foxes and R8 from raccoons.  To further determine whether G. mellonella mortality is associated with the growth of bacteria in infected larvae, larvae were injected with 10 3 CFU of P. mirabilis. As shown in Figure 4A, infection of G. mellonella with the F13, R8, and M1 strain resulted in a 10 4~1 0 5 -fold increase in CFU at 6 h post infection, which was followed by further increases in the bacterial numbers of dead larvae within 12 h post infection. These results are similar to that of larvae infected with pathogenic S. Enteritidis CVCC3377, which caused a remarkable increase in bacterial numbers in G. mellonella over time. At 12 h post infection, larvae in P. mirabilis infection group were all dead, whereas 60% larvae survived in the S. Enteritidis infection group ( Figure 4B). The results demonstrate that P. mirabilis isolates from fox, mink and raccoon farms could replicate rapidly in vivo and displayed high virulence in G. mellonella larvae model.

Pathogenicity of P. mirabilis Isolates with Different Virulence Genes
To evaluate the effect of virulence genes on the pathogenicity of P. mirabilis isolates, three isolates, which were all isolated from fox feces but with different virulence genes, were selected to compare their virulence in a G. mellonella model. As shown in Table 6, the LD50 of F2, F3, and F6 isolates at 6 h post infection was 2.0 × 10 4 CFU/larvae, 7.9 × 10 3 CFU/larvae and 4.0 × 10 3 CFU/larvae, respectively. Although strain F2 carried all detected virulence genes, its LD50 was 2.53-and 5-fold higher than that of strain F3 lacking ucaA, rsbA and rsmA, and strain F6 lacking ucaA. The results indicate that the carriage rate of virulence-associated genes was not positively correlated with P. mirabilis pathogenicity in

Pathogenicity of P. mirabilis Isolates with Different Virulence Genes
To evaluate the effect of virulence genes on the pathogenicity of P. mirabilis isolates, three isolates, which were all isolated from fox feces but with different virulence genes, were selected to compare their virulence in a G. mellonella model. As shown in Table 6, the LD 50 of F2, F3, and F6 isolates at 6 h post infection was 2.0 × 10 4 CFU/larvae, 7.9 × 10 3 CFU/larvae and 4.0 × 10 3 CFU/larvae, respectively. Although strain F2 carried all detected virulence genes, its LD 50 was 2.53-and 5-fold higher than that of strain F3 lacking ucaA, rsbA and rsmA, and strain F6 lacking ucaA. The results indicate that the carriage rate of virulence-associated genes was not positively correlated with P. mirabilis pathogenicity in G. mellonella larvae. Table 6. Pathogenicity of P. mirabilis isolates with different virulence genes from foxes in G. mellonella larvae.

Tested Strain
Virulence

Phylogenetic Analysis of P. mirabilis Isolates
The 53 P. mirabilis isolates and the selected eight representative isolates of P. mirabilis 16S rRNA gene sequences obtained on NCBI were used to perform phylogenetic analysis to understand the evolutionary relationship among the P. mirabilis isolates ( Figure 5). The results show that the sequence similarity between the 53 isolates and eight representative bacteria was 99.5~99.9%, but the P. mirabilis isolates from fur farms have far lower relatedness with the representative P. mirabilis isolates from shrimp, chick, mastitic milk, etc. The 22 P. mirabilis isolates from foxes and 18 isolates from raccoons were co-clustered into three branches, but the 13 P. mirabilis isolates from minks formed many clades in the evolutionary phylogeny, indicating there may be multiple origins of P. mirabilis isolates from minks. Moreover, the isolates from animal samples and farm environment samples were distributed across different branches. This pattern suggested that the isolates from farm environment samples had close relatedness with those from animal samples. Figure 5. Phylogenetic tree constructed based on 16S rRNA sequences of 61 P. mirabilis isolates using the neighbor-joining method. Eight representative isolates from different sources are labeled with big parentheses, 14 isolates from farms environment in the present study are marked in turquoise, and isolates harboring blaNDM or blaOXA-24 are marked in green or yellow, respectively. All sequences were aligned using ClustalW, then the aligned data were evaluated by the neighbor-joining approach using MEGA7 software with 2000 replications of bootstrap. Branches corresponding to partitions reproduced in less than 50% bootstrap replicates are collapsed. There were a total of 1337 positions in the final dataset. Figure 5. Phylogenetic tree constructed based on 16S rRNA sequences of 61 P. mirabilis isolates using the neighbor-joining method. Eight representative isolates from different sources are labeled with big parentheses, 14 isolates from farms environment in the present study are marked in turquoise, and isolates harboring bla NDM or bla OXA-24 are marked in green or yellow, respectively. All sequences were aligned using ClustalW, then the aligned data were evaluated by the neighbor-joining approach using MEGA7 software with 2000 replications of bootstrap. Branches corresponding to partitions reproduced in less than 50% bootstrap replicates are collapsed. There were a total of 1337 positions in the final dataset.

Discussion
Bacterial antibiotic resistance is a global public health threat, and one important cause of it is the multidrug bacterial isolates of animal origins acting as an important source for human infections [46][47][48]. We investigated the characteristics of 53 P. mirabilis isolates from fox, raccoon and mink farms-the results show that the proportion of MDR isolates among all the P. mirabilis isolates was up to 73.58%, similar to that among the 32 P. mirabilis isolates from chicken carcasses in a poultry slaughterhouse in Brazil (78.13%) [47]. ESBL production is significantly associated with mortality in patients with bacteremia caused by P. mirabilis [49]. In this study, 51 of the 53 P. mirabilis isolates were resistant to betalactam antibiotics and most isolates exhibited two or more beta-lactamase genes. The most common ESBL-encoding gene among the 53 P. mirabilis isolates from foxes, raccoons and minks is bla TEM (83.02%), which is spread worldwide and is now found in many different species of the order of Enterobacterales [17]. Carbapenem resistance rates are typically extremely low in clinical P. mirabilis isolates; however, approximately 71.70% of isolates from foxes, raccoons and minks were resistant to imipenem. Among the nine detected carbapenemase genes, the class D carbapenemase gene bla OXA-24 (15.09%) and the class B metallo-beta-lactamase bla NDM (13.21%) were found for the first time in P. mirabilis isolates from foxes, raccoons and minks. Kang et al. found that 16.67% (6/54) of the P. mirabilis isolates from wild animals were resistant to meropenem, and they mainly carried the carbapenem-resistance genes bla OXA-1 and bla NDM-1 [19]. For the 35 P. mirabilis isolates obtained from 240 duck samples in Egypt, only three strains were carbapenem-resistant, two strains harbored the bla NDM-1 gene, and one strain carried the bla KPC gene [21]. The presence of bla NDM-1 in Proteus spp. clinical isolates is still episodic in China and many other countries [17]. However, the bla NDM-1 gene had been identified on a genomic island in P. mirabilis recovered from the urine of a hospitalized patient in France in 2012 [50]. The bla OXA-24 gene, mainly detected in Acinetobacter baumannii, were shown to appear in reservoirs including livestock, companion animals, and the environment [17,51]. However, the bla OXA-24 gene is only detected in clinical strains of P. mirabilis in the United States [22]. To the best of our knowledge, P. mirabilis isolates harboring bla OXA-24 were identified for the first time from animal origins. Some other mechanisms including through porin mutation with or without decreased expression of penicillin binding proteins may contribute to the high imipenem-resistance rates of P. mirabilis isolates in this study. Moreover, the prevalence of beta-lactam-resistant P. mirabilis isolates from fur animals in this study and the similar antibiotic resistance profiling between P. mirabilis isolates from fur animals and farm environment suggested that MDR P. mirabilis isolates may likely spread from fur animals to the environment and potentially humans, posing a public health threat [52].
We also found increasing prevalence and diversity of some other antibiotic resistance genes in P. mirabilis from foxes, raccoons and minks. It was reported that aac(6 )-Ib-cr conferring quinolones and aminoglycosides resistance was common in Escherichia coli [53]. Here we showed that the detection rate of aac(6 )-Ib-cr among 53 P. mirabilis isolates was up to 71.7%, similar to the 63.2% among 19 P. mirabilis isolates from a Chinese hospital by Hu et al. [54]. The detection rates of quinolone resistance genes qnrA (5.66%) and qnrB (22.64%) were higher than the 0% from chicken carcasses in Brazil by Sanches et al. [19]. For the aminoglycoside resistance of P. mirabilis isolates in this study, another frequently encountered resistance gene was aadA, which may result in high resistance rates to streptomycin in these P. mirabilis isolates. These results indicate that aac(6 )-Ib-cr and aadA gene might be common in P. mirabilis isolates in foxes, raccoons and minks of China. Florfenicol is mainly used in veterinary medicine, a key gene for florfenicol and chloramphenicol resistance, floR, coding for an efflux protein of 12 transmembrane segments, can spread among bacteria of the same and different species or genera through horizontal gene transfer [55]. In this study, we found that 90.57% of P. mirabilis isolates were resistant to florfenicol, of which 77.36% carried the floR gene. So, the emergence and dissemination of florfenicol resistance among P. mirabilis isolates will limit the use of this antimicrobial for treating bacterial infections.
All P. mirabilis isolates harbored ureC, zapA, pmfA, atfA and mrpA virulence genes, which were also prevalent in the P. mirabilis isolates from chicken carcasses [47] and pork meat [24]. Among the 10 virulence genes, the detection rate of ucaA, a major fimbrial subunit that can enhance the adhesion of P. mirabilis, was the lowest [56]. However, we showed that the virulence of a ucaA-positive strain was about 5-fold lower than that of a ucaA-negative strain in G. mellonella larvae. Moreover, the carriage rate of virulenceassociated genes in these P. mirabilis isolates did not correlate with higher pathogenicity in a G. mellonella model. Three P. mirabilis isolates carrying all detected 10 virulence genes from different animal species showed different lethal abilities in G. mellonella larvae. These P. mirabilis isolates could multiply rapidly in the hemolymph of the larvae, similarly to S. Enteritidis, resulting in a fierce but ineffective inflammatory response triggered by high bacterial burdens and hemocyte depletion [57]. These results suggest that the carriage rate of virulence genes in P. mirabilis isolates can only reflect their epidemic, the pathogenicity of bacteria should be evaluated by appropriate animal models.

Conclusions
In the present study, we firstly reported the emergence of carbapenem-resistant P. mirabilis isolates harboring bla OXA-24 (15.09%) or bla NDM (13.21%) from foxes, raccoons and minks. All P. mirabilis isolates harbored ureC, zapA, pmfA, atfA and mrpA genes, but the carriage rate of these virulence-associated genes in the isolates did not correlate with higher pathogenicity in a G. mellonella model. Moreover, phylogenetic analysis showed that the P. mirabilis isolates from farm environment samples may have close relatedness with that from animals.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/biology11020292/s1, Table S1. Characteristics of virulence genes of 53 P. mirabilis isolates from fur animal farms in this study. Table S2. Characteristics of antibiotic resistance genes of 53 P. mirabilis isolates from fur animal farms in this study. Institutional Review Board Statement: The animal study was reviewed and approved by Animal Care and Use of Shandong Agricultural University (SDAUA-2018-027). Written informed consent was obtained from the owners for the participation of their animals in this study.

Data Availability Statement:
The original contributions generated for this study are included in the article, further inquiries can be directed to the corresponding authors.