NDM-1 and OXA-48-Like Carbapenemases (OXA-48, OXA-181 and OXA-252) Co-Producing Shewanella xiamenensis from Hospital Wastewater, China

Introduction

Antimicrobial resistance has become a threat for decades, and one representative example is carbapenem-resistant Enterobacterales (CRE), which have spread all over the world. CRE has received great attention in recent years due to its high morbidity, mortality, and lack of effective treatment solutions.¹ And the main mechanism of carbapenem resistance is the production of different carbapenemases, and Klebsiella pneumoniae carbapenemase (KPC), New Delhi metallo-β-Lactamase (NDM), Verona integron-mediated metallo-β-lactamase (VIM), imipenemase (IMP), and oxacillinase-48 (OXA-48) are the most common carbapenemases over the world.² Specifically, KPC and NDM are widely spread in America and China, while OXA-48-like is prevalent in Europe, North Africa, Middle East, and Indian subcontinent. OXA-48-like enzymes remain uncommon in American and most Asian countries, including China.³ In China, OXA-48-like producers have been reported in Shanghai, Jiangsu, Sichuan, Zhejiang, Guangdong. But few cases were related to foreign stay, and whether these bla_OXA-48-like genes were originated from abroad was unclear.^4–6

Surveillance has been widely regarded as the key point of administration of antibiotic resistance, which is informative for guidance on empirical treatment before antimicrobial susceptibility profile is determined.^7,8 To make surveillance data more informative, samples from both clinical patients and healthy individuals should be included. However, this is demanding and difficult to achieve, thus hospital wastewater has been increasingly recognized as an alternative of classical surveillance due to the abundance of antibiotic residual and pathogenic bacteria released from large quantity of hospitalized population.⁹ Additionally, antimicrobial resistant genes (ARG) in wastewater have also been suggested as a suitable marker to predict clinical resistance prevalence in the hospitals.¹⁰ This provides the possibility to evaluate the antimicrobial resistant bacteria (ARB) communities and ARG prevalence in less developing regions with low input of expenditure and labor. On the other hand, detection of uncommon ARG from wastewater in a non-epidemic region indicates a risk of further transmission, with probably much more severe consequences than those have already been circulating in pathogens.⁸

The genus Shewanella, facultative anaerobic, motile gram-negative bacteria, is widely distributed in various environments, such as spoilt food and aquatic environments.^11,12 In recent years, Shewanella infection cases have been rapidly increased and have been reported over the world. Most of cases are reported in tropical regions, and temperate countries and coastal cities like Australia, Spain, Martinique, Canary Islands, also have sporadic cases.^13,14 Shewanella-related infections can involve multiple parts of human body, and could be divided into several categories, including ear, nose, and throat (ENT) disorders, central nervous system (CNS) disorders, chest infections, cardiovascular diseases, bloodstream infections, intra-abdominal infections, bone arthropathy, skin and soft-tissue infections (SSTIs).¹³ Marine exposure, compromised immune system, invasive procedures and poor hygiene in hospital are risk factors of Shewanella infections.^11,15–17 Although no antimicrobial treatment guideline for Shewanella infections has been published yet, the empirical treatment could efficiently control infections.¹³ With the increase of antimicrobial resistance, however, clinicians and laboratory researchers should raise awareness to occurrence and dissemination of multi-drug resistant strains.

Here, we identified seven Shewanella xiamenensis strains that coproduce NDM-1 and OXA-48-like (OXA-48, OXA-181 and OXA-252) carbapenemases from hospital wastewater in Suzhou, China, where OXA-48-like-producing bacteria have not been detected before. We aimed to describe the risk of Shewanella xiamenensis in hospital environment through analysis of genomic comparison and resistance analysis.

Materials and Methods

Sample Collection, Culture Condition and Species Determination

To screen carbapenem-resistant bacteria in hospital wastewater, we performed the assay as described previously with minor modification.¹⁸ Briefly, water samples were collected from main sewer outlet at different depth of 0.5 to 1 m below the surface using 50mL sterile bottles. (Samples were chilled on ice during the sampling procedure and after being collected.) The samples were concentrated by 0.45 μm membrane filters, and ten-fold serial dilutions (10–1000 times) of each water sample were made in sterile saline solution. A total of 100 µL of each dilution was then plated onto LB agar plates supplemented with 2 mg/L meropenem. After incubated for 18 to 24 hours at 37°C, bacterial colonies with distinct coloration and morphologies were randomly picked and subcultured onto meropenem plates for further purification. Then each isolate was stored at −80 °C in 30% (vol/vol) glycerol for further investigation. And the species preliminary identification was determined using matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF-MS, Bruker).

Carbapenemase Gene Assay and Antimicrobial Susceptibility Testing

To determine resistance mechanism to carbapenem of these strains, common carbapenemase genes were detected (bla_KPC, bla_NDM, bla_IMP, bla_VIM, bla_OXA-48) using specific primer pairs followed by Sanger sequencing for their subtypes (Sangon, Shanghai). Primers were synthesized by Sangon and were listed in Table 1. Reaction buffer 2× Taq Mater Mix was provided by Vazyme (Nanjing).

Table 1 Primers Used in This Study

To characterize the antimicrobial susceptibility profile, we performed antimicrobial susceptibility testing. Isolates were suspended in 0.9% saline to 0.5 McFarland. Panel Phoenix NMIC/ID (BD) was used to determine MIC values on Phoenix M50 Automated System (BD, USA) and a total of 21 antibiotics were tested, including ampicillin, piperacillin, amoxicillin-clavulanate, ampicillin-sulbactam, piperacillin-tazobactam, imipenem, meropenem, cefazolin, cefepime, cefotaxime, ceftazidime, aztreonam, chloramphenicol, colistin, tetracycline, trimethoprim-sulfamethoxazole, amikacin, gentamycin, ciprofloxacin, levofloxacin and moxifloxacin. Results were interpreted according to “Other Non-Enterobacterales” in Clinical and Laboratory Standards Institute criteria (CLSI, 2020).

Whole Genome Sequencing

To further characterize the strains, we have performed genome sequencing. Bacterial genomic DNA were extracted by The Omega Bio-Tek Bacterial DNA Kit (Doraville, GA, USA) following instructions provided. Whole genome sequencing used the Illumina NovaSeq 6000 platform (Illumina, CA, USA) and then the corrected reads were assembled de novo utilizing SPAdes v3.11. In addition, sequencing of the complete genome sequence of isolate HD6416 and HD6452 were respectively performed with a sheared DNA library with average size of 10 kb on a Nanopore PromethION sequencer (Oxford Nanopore Technologies, OX, UK). The paired-end short Illumina reads were used to correct the long Nanopore reads, and then the corrected Nanopore reads were assembled de novo utilizing Unicycler v0.4.9 (https://github.com/rrwick/Unicycler).

Phylogenetic Analysis

To investigate the relationships of the bla_OXA-48-like genes, the maximum-likelihood phylogenetic tree of bla_OXA-48-like genes on Shewanella species available on Beta-Lactamase DataBase (BLDB, http://bldb.eu/) was constructed using MEGA 7.0.¹⁹

To describe the phylogenetic relationships of Shewanella xiamenensis strains, we collected all accessible genomes of Shewanella xiamenensis on NCBI to perform phylogenetic analysis. Based on recombination-free core genome, single-nucleotide polymorphisms (SNPs) was conducted with MUMmer 3.0, ClonalFrameML and MEGA7.0 as described previously.²⁰ The sequence of Shewanella xiamenensis HD6416 (GenBank accession no. CP079717) was used as the reference, and the sequence of Shewanella putrefaciens CN-32 (GenBank accession no. NC_021505) was used as the outgroup. And the phylogenetic tree was further improved on iTOL and Inkscape v1.0 (https://inkscape.org/).²¹

Bioinformatic Analysis

Species identification was performed by calculating the pairwise average nucleotide identity (ANI) value between the genome sequence of the query strain and type strain of Shewanella xiamenensis (Accession number: BMOP01000000) by using OAT 0.93.1 (cutoff value of 96%).²²

Open reading frames (ORFs) and pseudogenes were predicted using RAST 2.0²³ combined with BLASTP/BLASTN searches against the UniProtKB/Swiss-Prot database and the RefSeq database.²³ Annotation of resistance genes, mobile elements, and other features were carried out using the online databases including CARD,²⁴ ResFinder 4.0,²⁵ ISfinder,²⁶ and Tn Number Registry. Gene organization diagram was drawn in Inkscape v1.0 (https://inkscape.org/). BLAST Ring Image Generator (BRIG)²⁷ was used to compare the differences between the plasmids.

Results

Isolate Characterization

Seven Shewanella putrefaciens were identified by MALDI-TOF-MS, but they were further determined as Shewanella xiamenensis in WGS analysis. The Shewanella xiamenensis were isolated from three different spots of the sewage outlets (Depth A, B and C) (Table 2). PCR testing and Sanger sequencing showed that all isolates co-harbored bla_NDM-1 and bla_OXA-48-like carbapenemase genes (including bla_OXA-48, bla_OXA-181 and bla_OXA-252, Table 2). Antimicrobial susceptibility testing (Table 3) confirmed that all seven Shewanella strains were resistant to cefotaxime, ceftazidime, imipenem, meropenem, gentamycin and trimethoprim-sulfamethoxazole. And all of them were determined as multidrug-resistant. Only amikacin (100%), aztreonam (85.7%), tetracycline (71.4%) and levofloxacin (85.7%) retained high antibiotic activity to them. Additionally, just part of the strains were susceptible to piperacillin (14.3%), cefepime (14.3%), piperacillin-tazobactam (42.9%) and ciprofloxacin (28.6%), respectively.

Table 2 Isolation Spot and Resistance Genes of Seven Shewanella xiamenensis

Table 3 Antimicrobial Drug Susceptibility of Seven Shewanella xiamenensis*

Genome Analysis

Genetic Characterization

Whole genome sequencing (Illumina Hiseq) (Bioproject PRJNA755833) and analysis showed that bla_OXA-48-like genes were on chromosomes while bla_NDM-1 were likely on plasmids. Sequence analysis showed that HD6416, HD6424, HD6449 and HD6452 carry bla_OXA-48, and HD6443 harbors bla_OXA-252, while HD6420 and HD6446 contain bla_OXA-181. The phylogenetic analysis of bla_OXA-48-like of Shewanella spp. from NCBI showed the bla_OXA-48 genes have 3 genotypes (different nucleotide sequences) despite being the same amino acid sequences while bla_OXA-181 gene only has one genotype. In addition, plasmid-borne and chromosome-borne bla_OXA-48 displayed some nucleotide variations but bla_OXA-181 were identical (Figure 1A).

Figure 1 Phylogenetic trees of OXA-48-like genes and Shewanella xiamenensis strains. (A) Phylogenetic trees of OXA-48-like genes. blaOXA-48 genes have three genotypes despite being the same amino acid sequences while blaOXA-181 gene only has one genotype. Plasmid-borne and chromosome-borne blaOXA-48 displayed some nucleotide variations but blaOXA-181 were identical; (B) Phylogenetic trees of Shewanella xiamenensis strains. Seven S. xiamenensis genomes were distributed into different clades of the tree, and the phylogenetic clades in general correlated with the blaOXA-48-like gene sequences. Bootstrap values for each cluster of associated taxa are shown next to each branch. Bar corresponds to the scale of sequence divergence. Shewanella xiamenensis from our study were characterized in boldly red.

SNP Analysis

Core genome single-nucleotide polymorphisms (SNPs) maximum-likelihood phylogenetic tree result (Figure 1B) showed that S. xiamenensis were classified into multiple clades with different subtree lengths. For the strains from our study, HD6416 and HD6424, HD6420 and HD6446, HD6449 and HD6452 were respectively grouped together, while HD6443 was located on a different branch. In addition, the seven genomes were distributed into different clades of the tree, indicating genomic diversity of the seven S. xiamenensis isolates, despite isolating from the same hospital. The S. xiamenensis phylogenetic clades in general correlated with the bla_OXA-48-like gene sequences, but interestingly the OXA-48 producing HD6416/HD6424 and HD6420/HD6446 were located on different clusters. And a pairwise comparison of SNPs and similarities for all strains have been shown in Table S1.

Plasmid Analysis

Nanopore sequencing was further conducted in strain HD6416 and HD6452 (Bioproject PRJNA755833) to obtain completely closed chromosome and plasmid sequences. HD6416 and HD6452 both contain one circular chromosome and a plasmid. The PlasmidFinder analysis showed no known plasmid type were matched to pHD6416 and pHD6452, indicating that both plasmids belong to novel plasmid type.^28,29 And plasmid comparison was further conducted using BRIG and result (Figure 2) indicated some variation between the backbone region of bla_NDM-1-harboring plasmids. The backbone region of pHD6420, pHD6424 and pHD6443 share 99% identity with pHD6416, and pHD6446 has about 90% similarity with pHD6416 backbone, while pHD6449 and pHD6452 contain a different backbone. Notably, the multidrug-resistant regions of the seven plasmids were highly comparable with just a few mobile elements different. We then depicted gene diagrams of MDR region of two closed plasmids from our study (pHD6416 and pHD6452) for further comparison (Figure 3). The MDR region of both plasmids replaced In834 on Tn6358a, but with slight difference in resistance gene organization, so we designated them as Tn6358b (on pHD6416) and Tn6358c (on pHD6452), respectively. On Tn6358c, An ISCR-ble_MBL-bla_NDM-1 unit and an ISCR-qnrV6-bla_CMY-10 unit were inserted in the In469 on Tn6358b. And the ΔIn1357 on Tn6358b was replaced by Tn6309 and part of ΔIn2145. The MDR region of Tn6358b and Tn6358c encode resistance to various antibiotics, such as aminoglycoside (dfrA27, aadA16), disinfectant (qacE), macrolides (msr(E)), rifampin (arr-3), sulfonamide (sul1), β-lactam, carbapenem (bla_NDM-1), as well as efflux pump, such as acrABC-tolC. This suggests there were two backbone types between seven bla_NDM-1 plasmids. The MDR regions of seven bla_NDM-1 plasmids were highly similar with minor differences on resistance elements and arrangement, and they likely evolve from the same Tn element.

Figure 2 Plasmid structure comparison between blaNDM-1-harbouring pHD6416 and the other strains. The backbone region of pHD6420, pHD6424, pHD6443, pHD6446 was highly comparable with pHD6416 backbone, respectively, while pHD6449 and pHD6452 contain a different backbone. The multidrug-resistant regions of the seven plasmids were also highly comparable with just a few mobile elements different. Colored arrows indicate ORFs, with navy, yellow, Orange, olive, purple and green arrows representing plasmid backbone genes, mobile elements, plasmid transfer genes, Tn6358b/c backbone genes, the antimicrobial and heavy metal resistance genes, respectively.

Figure 3 Linear comparison of Tn6358a, Tn6358b, Tn6358c. The MDR regions from pHD6416 and pHD6452 were highly similar and could evolve from the same Tn element. Genes are denoted by arrows. Gene, mobile elements, and other features are colored based on their functional classification. Shading denotes regions of homology (nucleotide identity ≥95%). Numbers in brackets indicate nucleotide positions within the plasmid pHD6416-NDM and pHD6452-NDM. Tn6358a was used as reference and its accession number was JX141473.

Besides, oriTfinder analysis (https://tool-mml.sjtu.edu.cn/oriTfinder) failed to identify transfer origin regions (oriT) and relaxases sequences in all seven S. xiamenensis, suggesting the plasmid is non-transferable. Further conjugation assay also failed to transfer bla_NDM-1-bearing plasmids to recipients of E. coli EC600 despite multiple attempts, which is consistent with the sequence analysis results. And bla_OXA-48 genes were also not detected in recipients.

Discussion

Shewanella is an opportunistic pathogen and generally invades the immunocompromised and those who are exposed to marine environment. And the Shewanella genus has been frequently isolated from clinical specimens in recent years. According to a latest review, bloodstream infections, SSTIs and ENT disorders are most dominant Shewanella-related diseases, accounting for 37.0%, 35.9% and 27.5% of all cases, respectively. Unlike the former three, less Shewanella strains were isolated from patients of intra-abdominal infection (9.2%), bone arthropathy (6.6%), chest infection (4.4%), CNS disease (2.2%), cardiovascular infection (1.8%). And patients who suffer the infections are distributed in all generations, ranging from newborn to 92-years-old.¹³ Besides being pathogenic to human, Shewanella have been considered as a reservoir and vehicle of multiple antimicrobial resistance since the description of OXA-54 on S. oneidensis as the progenitor of OXA-48 in K. pneumoniae.³⁰ Studies have found a variety of resistance genes from Shewanella genus, such as qnrA variants, dfrA1, aadA1, sul1, sul2, and bla_OXA-48-like variants, conferring them resistance to many antibiotics, including β-lactams, quinolones, trimethoprim, aminoglycoside, sulfonamides, and carbapenems.^10,31–33 And the occurrence of ARBs and ARGs in environment raises worries due to the potential hazard to public health. In our study, we reported three types of OXA-48-like carbapenemases (OXA-48, OXA-181 and OXA-252) from hospital wastewater, where bla_OXA-48-like has not been detected before. Although the occurrence of waterborne species harboring NDM-1-encoding plasmids in China has been frequently reported,^34,35 we describe a novel type bla_NDM-1 plasmid of complicated structure in our study. We then further searched for similar bla_NDM-1 plasmids in our isolate’s storage, but no similar plasmid was found. This suggested that the bla_NDM-1 plasmids were assembled in wastewater, which could reflect the resistance gene diversity and complex genetic interaction in hospital wastewater.

The contributing factor to the ARG horizontally transmission is usually mobile element, such as insertion sequence. Insertion sequence has been also suggested accountable for the mobilization of bla_OXA-48-like genes from chromosome to plasmid in this situation. We compared the genetic environment of bla_OXA-48, bla_OXA-181 and bla_OXA-252 from our study with those on plasmids such as Tn1999 and Tn2013, respectively (Figure 4). bla_OXA-48 and lysR on plasmid borne Tn1999 shared 2.237 kb of and >95% similarity with those on chromosomes of S. xiamenensis (cHD6416, cHD6420, cHD6449 and cHD6452). The bla_OXA-48-lysR from Tn1999 in Enterobacterales plasmids were flanked by two copies of IS1999, indicating Tn1999 was likely responsible for the capture and mobilization of bla_OXA-48-lysR region from chromosomes of S. xiamenensis to plasmid. Similarly, bla_OXA-181-ΔlysR region on plasmid carried by Tn2013 shared 1.299 kb and >95% similarity with chromosome of S. xiamenensis HD6424. And bla_OXA-181-ΔlysR region from Tn2013 in Enterobacterales plasmids were adjacent to ISEcp1, thus Tn2013 was more than likely to catch bla_OXA-181-ΔlysR from chromosomes to plasmids.^36,37 This is consistent with previous study that chromosome-encoded β-lactamases of Shewanella spp., eg S. xiamenensis, are likely the progenitors of bla_OXA-48-like genes in Enterobacterale members.³⁸ Different insertion sequences have been found to be linked to bla_OXA-48-like genes on plasmids in Enterobacterales. The plasmid borne bla_OXA-48 and bla_OXA-163 are associated with IS1999 and ISEcl4, respectively, while bla_OXA-181, bla_OXA-204, and bla_OXA-232 genes are associated with ISEcp1 elements but each in slightly different arrangements.³⁸ These results suggested that bla_OXA-48-like genes have been independently mobilized from their original location in Shewanella species by different insertion elements on multiple occasions. Under specific occasions, these bla_OXA-48-like genes from the environmental S. xiamenensis may be recruited by additional mobile elements, followed by horizontal transfer and dissemination. In China, bla_OXA-48-like producers are not as common as bla_KPC producers, and so far they have been found in several provinces, including Shanghai, Jiangsu, Sichuan, Zhejiang, Guangdong, but few patients had a foreign stay before the bla_OXA-48-like producers were isolated.^4,39–41 For example, the first bla_OXA-181 gene in China was found in a clinical ST410 E. coli strain WCHEC14828 from a Chinese patient without recent travel history.⁵ In this case, the possibility that bla_OXA-48-like genes in China may originate through the horizontal gene transfer from S. xiamenensis to Enterobacterales rather than import from other countries, cannot be ruled out. Therefore, detection of OXA-48-like-producing S. xiamenensis from hospital wastewater in non-epidemic region, such as China, is rather indicative of the possibility of further transmission, which could lead to local epidemic of bla_OXA-48-like genes. According to the surveillance of CRE in our hospital, OXA-48-like-producing Enterobacterales have not been detected yet, however, bla_OXA-48-like transmission in our hospital is still possible due to the emergence of S. xiamenensis harboring bla_OXA-48-like in hospital environment. This observation raised the concern that the S. xiamenensis from the hospital wastewater may serve as a reservoir for the transmission of ARGs, suggesting the presence of naturally existing carbapenemases in the environment may be significantly overlooked. Hence, the hospital surveillance towards both clinical isolates and wastewater should afterward continue to detect bla_OXA-48-like gene prevalence in real time.

Figure 4 Linear Comparison of blaOXA-48-like-related region and related regions. Genes are denoted by arrows. Shading denotes regions of homology (nucleotide identity ≥90%). Sequences of Tn1999 and Tn2013 are acquired from pOXA-48a plasmid and pKP3-A plasmid on NCBI, respectively, and accession numbers are noted in bracket.

Wastewater-based surveillance has been recommended as a low-cost methodology to monitor antimicrobial resistance prevalence, and could be useful in monitoring environmental resistance gene and helping decrease the health risks coming from the wastewater.⁴² However, research about hospital wastewater remains rare in China because of a lack of understanding and underestimation of surveillance. Future work should be considered to establish wastewater surveillance system in China hospitals to assess antimicrobial resistance abundances and monitor the transmission of resistance between human and the environment.

Conclusion

In our study, we reported seven S. xiamenensis co-producing NDM-1 and OXA-48-like carbapenemases (OXA-48, OXA-181 and OXA-252) from hospital wastewater in China and characterized them in some profiles. And to the best of our knowledge, this is the first report of detection of three types bla_OXA-48-like genes in one hospital in China. This study proposes the possible origin of the bla_OXA-48-like cases in China and warns the risk of bla_OXA-48-like transmission in our hospital. Besides, an active antimicrobial resistance surveillance system in both clinical isolates and the hospital environment should be considered. And more attention is urgently needed on multi-drug resistant Shewanella genus.