Molecular characteristics of carbapenem-resistant Enterobacteriaceae isolated from clinical infection and fecal survey samples in Southern China

Background: The identied rate of carbapenem-resistant Enterobacteriaceae (CRE) have been increasingly in the clinical infections. Here, a study of analysing the relationship of clinical infectious CRE and fecal carried CRE was performed. Methods: Clinical CRE and fecal CRE were collected from hosiptal in China. Polymerase chain reaction (PCR)-based amplication and sequencing were performed to analyse the carriage of drug-resistant genes and mobile genetic elements (MGEs), Enterobacterial Repetitive Intergenic Consensus (ERIC) technology and whole genome sequencing (WGS) were used to analysis the characteristic of genetic structure of CRE isolates. Results: 99 clinical CRE and 30 fecal CRE were collected, respectively. The top three strains present in the highest proportions were K. pneumoniae (86; 66.67%), E. cloacae (22; 17.05%), and E. coli (11; 8.53%). Most of the isolates were susceptible to colistin (98.45%) and tigecycline (98.45%). bla KPC−2 (96.03%) was the dominant carbapenemase gene in clinical CRE and fecal CRE, followed by bla NDM (52.7%); co-existence of the bla KPC−2 and bla NDM genes was detected in 63 (50.00%) strains. One K. pneumoniae isolate co-producing NDM-5 and mcr-1, and one E. coli isolate co-producing KPC-2, IMP-4, and mcr-1 were detected. Two novel gene cassettes of intI2 were dectected. ERIC genotyping and genomic analysis revealed that K. pneumoniae isolated from clinical infections and fecal survey samples were same clone. Conclusions: This is the rst report of comparing the molecular characteristics of CRE isolated from clinical infection and fecal survey samples. we found that fecal carried CRE were closely related to CRE which caused infections. and morbidity [1] . CRE have spread rapidly around the world, posing great challenges to human health. The China Antimicrobial Resistance Surveillance Report (http://www.carss.cn/), the largest survey of antimicrobial resistance in China, reported that the rate of CRE is increasing every year; in particular, the rate of carbapenem-resistant K. pneumoniae increased from 4.9% in 2013 to 10.1% in 2018. Carbapenem resistance in K. pneumoniae has been reported in all WHO regions and exceeded 50% of all isolates in two regions [2] . CRE can cause multiple infections, most frequently pneumonia, followed by urinary tract infections, primary bloodstream infections, skin and soft tissue infections, and surgical site infections [3] . Patients infected with CRE are at an increased risk of death due to the lack of appropriate antibiotics. CRE, especially carbapenem-resistant K. pneumoniae (CRKP), have a high potential to cause outbreaks in healthcare settings. Such outbreaks have been reported from several EU member states; e.g., the Czech Republic, France, Germany, Greece, Italy, Spain, the Netherlands, and the UK [4–10] . Therefore, the timely and accurate detection of CRE is important; investigating the mechanism of transmission and the route of drug resistance will also be necessary. pneumoniae detected in clinical and fecal samples showed greater than 90% similarity, indicating that fecal CRE were closely related to CRE isolated from clinical infection samples, and that they may have originated from the same clone. Clinical and fecal samples were obtained from different places in South China, suggesting that KPC-CRKP has spread widely. However, all strains were isolated from samples collected between 2016 and 2019, indicating that there was no outbreak in the short term.

The primers including the IncFII plasmid replicon [19] used for PCR are shown in Supplementary Table 1. Whole genome sequencing and analysis Three KPC-producing K. pneumoniae 156070, 158590, and NFYY0065 were submitted for WGS. The whole genome was sequenced using the Single Molecule Real-Time sequencing platform with the PacBio sequencer and Illumina HiSeq at the Health Time Gene Institute (Shenzhen, China).

ERIC-PCR
ERIC-PCR was performed on all isolates depending on the species [20] . Band comparisons were performed by clustering analysis based on the unweighted pairgroup method with arithmetic mean of isolates using NTSYS.

Identi cation of isolates
During the study period, clinical infection samples were collected from 20 departments of Nanfang hospital; the top four departments were hematology (37.4%), intensive care unit (ICU) (22.2%), rehabilitation (7.1%), and the surgical department (4.0%). A total of 99 CRE from clinical infection samples and 30 CRE from fecal survey samples were isolated. There was no obvious difference in the types of CRE between clinical and fecal samples. The species present in the highest proportions among the six different species isolated from clinical infection samples were K. pneumoniae (73.74%), E. cloacae (15.15%), and E. coli (7.07%); the species present in the highest proportions among the seven different species isolated from fecal survey samples were K. pneumoniae (43.33%), E. cloacae (23.33%), and E. coli (13.33%) ( Table 1).

Antimicrobial susceptibility testing
According to the MIC values of tested antibiotics (Supplementary Tables 2 and 3). All isolates were resistant to cephems and carbapenems, and most of the isolates were sensitive to colistin and tigecycline. Among clinical CRE, the sensitivities to amikacin, tetracycline, and sulfamethoxazole were 35.35%, 35.35%, and 32.32%, respectively. Among fecal CRE, the sensitivities to amikacin, tetracycline, and sulfamethoxazole were 56.67%, 36.67%, and 40.00%, respectively. We found that the sensitivities of different species to amikacin and tetracycline vary widely, and CRE isolated from feces were generally more sensitive to antibiotics than clinical strains with the exception of K. pneumoniae (Table 2).
The complete sequence analysis showed that chr156070 contained 5227 genes with a G + C content of 57.40%, and chr158590 contained 5262 genes with a G + C content of 57.37%. Both chr156070 and chr158590 contained blaSHV-11. BLAST analysis revealed that similarities in the chromosomes of the two KPC-CRKP were greater than 95%. Comparative genomic analysis showed that the core genes of chr156070 and chr158590 accounted for 89% of all genes, and chr156070 had no speci c resistant genes compared with chr158590. Co-linear analysis found that although there were structural differences between chr156070 and chr158590, they had a high degree of similarity and high coverage for one another (Fig. 1). They are ST11 by multilocus sequence typing. chr156070 and chr158590 were similar. BLASTn analysis revealed that chr156070 and chr158590 were highly similar to 18CPO060 and KPNIH45, and phylogenetic analysis showed that chr156070 and chr158590 had a close evolutionary relationship and high a nity to KPNIH45, which was derived from CRKP isolated from a US hospital in 2018.
The p156070-KPC is a circular plasmid containing 184 putative open reading frames (ORFs) with a G + C content of 53.95%, and p158590-KPC was a circular plasmid containing 184 putative ORFs with a G + C content of 53.95%. Both p156070-KPC and p158590-KPC had the same replication protein RepA1 of IncFII plasmids. The p156070-KPC had one unique gene relative to p158590-KPC, and p158590-KPC had two unique genes relative to p156070-KPC. Both p156070-KPC and p158590-KPC co-carried the blaKPC-2, blaCTX-M-65, and blaTEM-1b genes. BLAST analysis showed that p156070-KPC was highly similar to p158590-KPC at the nucleotide level (100% identities and 100% query coverage), and colinearity analysis revealed that the plasmid was divided into three parts: the rst segment plus chain of p156070-KPC and p158590-KPC were performed collinear comparison from 1 bp to 32195 bp, with a similarity of 99%; the second segment of p156070-KPC and p158590-KPC plasmid were performed translocation comparison from 31,374 bp to 106,077 bp, with a similarity of 100%, the phenomenon might be result from the rearrangement of small fragments that maked up the A fragment and B fragment due to MGEs; the third segment 105,256 bp to 138,959 bp of p156070-KPC and p158590-KPC were collinear comparison, with a similarity of 100% (Fig. 2).
Phylogenetic analysis showed that p156070-KPC and p158590-KPC are closely genetically related to most of the IncFII plasmids previously reported (Fig. 4). The 17 IncFII plasmids were selected from the database, and the pNFYY0Y11 plasmids were the IncFII plasmids carrying the blaKPC gene sequenced by us (Supplementary Table 5).

ERIC-PCR
Genotyping by ERIC-PCR demonstrated greater than 90% similarity of Klebsiella pneumoniae in different samples, including clinical infection and fecal survey samples ( Supplementary Fig. 2), suggesting they may have originated from the same clone. However, E. cloacae and E. coli detected in clinical and fecal samples had a high genetic diversity.

Discussion
In all isolates, the top three species were K. pneumoniae (66.67%), E. cloacae (17.05%), and E. coli (8.53%); thus, CRKP was more abundant than other species. This is consistent with Spiliopoulon et al. [21] , which showed that K. pneumoniae was present in 1,137 (77.9%) of 1,460 clinical infectious samples of hospitalized patients around the world between 2015 and 2017. The number of clinical infections caused by K. pneumoniae is increasing; in particular, the number of CRKP is increasing [22] . In our study, the species present in the highest proportions in fecal survey samples were also K. pneumoniae, E. cloacae, and E. coli. The fecal samples were primarily obtained from individuals who had not previously used related antibiotics; however, they carried a high proportion of CRE similar to clinical infection samples, indicating that CRE can also exist in the intestines of normal individuals without a history of antibiotic use and that fecal CRE may be associated with CRE causing clinical infection.
The sensitivity of CRE to antibiotics has reduced, which is a challenge for clinical treatment. In our study, clinical and fecal CRE were resistant to most of the antibiotics tested and only showed a high sensitivity to tigecycline and colistin. The susceptibilities of clinical CRE to tigecycline and colistin were 100% and 100%, and the susceptibilities of fecal CRE were 93.33% and 93.33%, respectively. Spiliopoulon et al. [21] reported that the susceptibilities of meropenem-nonsusceptible Enterobacteriaceae to colistin and tigecycline were 92.4% and 77.4%, respectively; colistin and tigecycline were shown to be active against MBLpositive isolates (susceptibilities of 92.1% and 71.9%, respectively) in Asia. Wang et al. [23] reported that 1801 CRE isolates showed high susceptibility to colistin (96.9%), followed by tigecycline (89.7%). Thus, CRE are still highly sensitive to colistin and tigecycline; in support of this, we found that colistin has high activity against CRE in Southern China, which will help in the choice of treatment of clinical infections. The susceptibility of different CRE species to amikacin varied, with a lower rate in K. pneumoniae (24.4%) than in E. coli and E. cloacae (> 72%). This result provides important data for selecting speci c drug and aminoglycoside combinations for empiric therapy of infections caused by these species.
CRE is resistant to carbapenem antibiotics for different reasons, but the high rate of carbapenemase genes detected in this study is likely the primary mechanism. Our data showed that the prevalent carbapenemases in Southern China primarily encoded KPC and NDM, and especially KPC, which may be related to a high proportion of CRKP. Notably, blaKPC is also commonly found in other species. KPC has already spread widely to different species, mainly KPC-2 in Southern China. We found a similar proportion of blaKPC between clinical and fecal CRE, especially CRKP, indicating that fecal KPC-CRKP may be closely related to KPC-CRKP isolated from clinical infection samples. In addition, we found that one K. pneumoniae isolate contained blaNDM-5 and mcr-1, and one E. coli isolate contained blaKPC-2, blaIMP-4, and mcr-1, which have not been reported previously. The mcr and carbapenemases-positive strains were isolated from fecal survey samples from individuals who had not previously used carbapenem and colistin, indicating that mcr-1 and carbapenemase genes may also be present in the gastrointestinal tract of patients who have not used related antibiotics. If individuals expressing resistance genes have infections, it will lead to the failure of antibiotic treatment, and increase the risk of transmission to other patients. Therefore, strengthening the monitoring of feces is necessary to prevent the spread of drug-resistant bacteria and drug-resistant genes.
Our data showed that clinical and fecal CRE carry MGE-related genes, which is consistent with the high prevalence of carbapenemase genes. Carbapenemaseproducing CRE strains carry resistance genes on mobile plasmids or MGEs that can shuttle between resistant and susceptible strains [24,25] , especially those encoding KPC and NDM. This suggested that MGEs were important for the spread of carbapenemase genes and multidrug resistance of strains. In addition, we identi ed two novel gene cassette structures of class II integrons, which have not been previously reported.
We found a high genetic diversity of E. cloacae and E. coli in different samples by ERIC genotyping; however, K. pneumoniae detected in clinical and fecal samples showed greater than 90% similarity, indicating that fecal CRE were closely related to CRE isolated from clinical infection samples, and that they may have originated from the same clone. Clinical and fecal samples were obtained from different places in South China, suggesting that KPC-CRKP has spread widely. However, all strains were isolated from samples collected between 2016 and 2019, indicating that there was no outbreak in the short term.

Conclusions
This was the rst study to compare the molecular characteristics of CRE isolated from clinical infection and fecal survey samples in Southern China. We found that CRKP isolated from fecal survey samples were closely related to CRKP isolated from clinical infections samples. CRE can spread between human, animals and the environment by fecal contamination and especially outbreak. Thus, Strengthening the monitoring and management of feces is essential for the prevention and control of CRE. Availability of data and materials:

Abbreviations
All data generated or analyzed in this study are included in this published article and its supplementary information les.

Competing interests:
The authors declare that they have no competing interests. This study was supported by grants from Guangdong Province Science and Technology Project (Nos. 2014A010107011 and 2015A020211011) and Guangzhou City Science and Technology (No. 201510010167). The funding body provides nancial support in designing research and collecting, analyzing, interpreting data and writing articles.

Figure 2
The p156070-KPC had one unique gene relative to p158590-KPC, and p158590-KPC had two unique genes relative to p156070-KPC. Both p156070-KPC and p158590-KPC co-carried the blaKPC-2, blaCTX-M-65, and blaTEM-1b genes. BLAST analysis showed that p156070-KPC was highly similar to p158590-KPC at the nucleotide level (100% identities and 100% query coverage), and colinearity analysis revealed that the plasmid was divided into three parts: the rst segment plus chain of p156070-KPC and p158590-KPC were performed collinear comparison from 1 bp to 32195 bp, with a similarity of 99%; the second segment of p156070-KPC and p158590-KPC plasmid were performed translocation comparison from 31,374 bp to 106,077 bp, with a similarity of 100%, the phenomenon might be result from the rearrangement of small fragments that maked up the A fragment and B fragment due to MGEs; the third segment 105,256 bp to 138,959 bp of p156070-KPC and p158590-KPC were collinear comparison, with a similarity of 100%.