Outbreak of NDM-1-Producing Klebsiella pneumoniae in a Dutch Hospital, with Interspecies Transfer of the Resistance Plasmid and Unexpected Occurrence in Unrelated Health Care Centers

ABSTRACT

In the Netherlands, the number of cases of infection with New Delhi metallo-beta-lactamase (NDM)-positive Enterobacteriaceae is low. Here, we report an outbreak of NDM-1-producing Klebsiella pneumoniae infection in a Dutch hospital with interspecies transfer of the resistance plasmid and unexpected occurrence in other unrelated health care centers (HCCs). Next-generation sequencing was performed on 250 carbapenemase-producing Enterobacteriaceae isolates, including 42 NDM-positive isolates obtained from 29 persons at the outbreak site. Most outbreak isolates were K. pneumoniae (n = 26) and Escherichia coli (n = 11), but 5 isolates comprising three other Enterobacteriaceae species were also cultured. The 26 K. pneumoniae isolates had sequence type 873 (ST873), as did 7 unrelated K. pneumoniae isolates originating from five geographically dispersed HCCs. The 33 ST873 isolates that clustered closely together using whole-genome multilocus sequence typing (wgMLST) carried the same plasmids and had limited differences in the resistome. The 11 E. coli outbreak isolates showed great variety in STs, did not cluster using wgMLST, and showed considerable diversity in resistome and plasmid profiles. The bla_NDM-1 gene-carrying plasmid present in the ST873 K. pneumoniae isolates was found in all the other Enterobacteriaceae species cultured at the outbreak location and in a single E. coli isolate from another HCC. We describe a hospital outbreak with an NDM-1-producing K. pneumoniae strain from an unknown source that was also found in patients from five other Dutch HCCs in the same time frame without an epidemiological link. Interspecies transfer of the resistance plasmid was observed in other Enterobacteriaceae species isolated at the outbreak location and in another HCC.

INTRODUCTION

In 2008, a new carbapenemase, designated New Delhi metallo-beta-lactamase 1 (NDM-1), was detected in a Klebsiella pneumoniae isolate originating from a Swedish patient who had been hospitalized in India (1). NDM-1 belongs to the Ambler class B beta-lactamases and hydrolyzes virtually all beta-lactam antibiotics, including carbapenems (2). Besides NDM-1, 13 other variants of this metallo-beta-lactamase (NDM-2 to NDM-14) have currently been described, and most of the variants contain either single- or two-amino-acid substitutions compared to NDM-1 (3). The NDM gene has been isolated from a wide range of bacterial hosts, including Vibrio cholerae, and has been located on a great variety of plasmids (4, 5).

After the first report, NDM-positive bacteria were shown to be widely present among hospitalized patients in many countries of the Indian subcontinent and in the United Kingdom in 2008 and 2009 (6). At present, NDM-positive bacteria have been isolated from many countries worldwide, but many isolates have travel-related epidemiological links, with or without hospitalization, to the Indian subcontinent (7). However, some reports showed the presence of NDM-positive isolates in patients with no history of foreign travel, suggesting that the bacteria or the NDM gene had been acquired through local transmission (8, 9).

In the Netherlands, until now, the number of NDM-positive cases has been low, with only sporadic cases reported (10). The first NDM-positive bacteria were detected in 2010 in two Dutch patients who had recently traveled to India (11). More recently, in 2015, an NDM-5-producing K. pneumoniae isolate was isolated from a Dutch patient without a history of travel abroad (12).

Here, we describe a large outbreak of NDM-1-producing Enterobacteriaceae in the Netherlands that occurred in a large teaching hospital. Unexpectedly, the same strain was also found in other health care centers with no detectable epidemiological link.

RESULTS

Transmission of NDM-1-producing K. pneumoniae.

The K. pneumoniae isolates collected from 26 cases at the outbreak location (Table 1) were NDM positive, and analysis of the next-generation sequencing (NGS) data showed that they all harbored the bla_NDM-1 gene. Three NDM-positive K. pneumoniae isolates originating from the outbreak location were not epidemiologically linked to the outbreak but also carried the bla_NDM-1 gene. The top three carbapenemase genes carried by the 133 K. pneumoniae isolates from the Dutch carbapenemase-producing Enterobacteriaceae (CPE) surveillance, which were used as context in this study, were OXA-48-like (n = 75), KPC (n = 27), and NDM (n = 26) genes (Table 2). Twenty-three of the NDM-positive surveillance isolates carried the NDM-1 variant. Two other isolates carried the bla_NDM-5 gene, and the remaining isolate harbored the bla_NDM-7 gene.

TABLE 1.

Species isolated from the outbreak location during the NDM-1 outbreak

Species isolated during the outbreak	No. of cases
K. pneumoniae	17
E. coli	2
C. freundii	1
K. pneumoniae and E. coli	7
K. pneumoniae, E. coli, K. oxytoca, and C. freundii	1
K. pneumoniae, E. coli, K. oxytoca, and P. mirabilis	1
Total no. of cases	29

TABLE 2.

Isolates collected from January 2015 to December 2016, used as context in the study (one isolate per case per year)

Species	Carbapenemase allele	No. of cases
K. pneumoniae	OXA-48	69
	OXA-232	4
	OXA-181	1
	OXA-162	1
	KPC-2	19
	KPC-3	8
	NDM-1	23
	NDM-5	2
	NDM-7	1
	NDM-1, OXA-48	3
	NDM-1, OXA-232	1
	VIM-1	1
	Total	133
E. coli	OXA-48	31
	OXA-181	7
	OXA-244	4
	KPC-2	1
	NDM-5	15
	NDM-1	7
	NDM-7	2
	NDM-6	1
	NDM-5, OXA-181	1
	VIM-2	1
	Total	70

The 26 presumed outbreak cases had meropenem MICs ranging from 0.75 to 12 μg/ml, but the majority of the isolates (n = 20) had MICs below 2 μg/ml. In contrast, 21 of the 26 NDM-positive context isolates from the CPE surveillance had MICs greater than 6 μg/ml, and only 2 isolates had MICs below 2 μg/ml.

Based on the NGS data on the K. pneumoniae isolates, all 26 presumed outbreak cases had sequence type 873 (ST873). Seven of the 133 K. pneumoniae context isolates from the Dutch CPE surveillance also were ST873, and these isolates originated from five other health care centers (HCCs). The HCCs were geographically dispersed throughout the Netherlands, and the ST873 isolates were submitted between December 2015 and October 2016. Six of the seven isolates were submitted after the outbreak period, but a single isolate was submitted during the outbreak period (Fig. 1). The other 126 isolates and the 3 unrelated isolates from the outbreak site submitted between 2015 and 2016 had various other STs.

FIG 1.

Geographic origins of the 26 K. pneumoniae isolates from outbreak cases and locations of the other five health care centers in the Netherlands. The outbreak cases were plotted based on the postal codes of the NDM carriers, and the sizes of the circles indicate the numbers of carriers with the same postal code.

The whole-genome multilocus sequence typing (wgMLST) was based on 4,978 genes, and the resulting minimum spanning tree (MST) showed that the NDM-positive isolates were scattered over the tree, with the exception of a group of 33 ST873 isolates that clustered closely together with a maximum of five genes difference between neighboring isolates (Fig. 2). The distances between this group and the nearest neighbors ranged from 2,926 to 3,666 genes. The distance from the other three epidemiologically unrelated K. pneumoniae isolates that were isolated from the outbreak location ranged from 3,781 to 3,837 genes, indicating that they did not belong to the outbreak.

FIG 2.

Minimum spanning tree of the 162 K. pneumoniae isolates. The tree was based on 4,978 genes of the K. pneumoniae whole-genome MLST scheme. Clustering was done using a categorical coefficient, and the sizes of the circles indicate the numbers of isolates with identical wgMLST profiles. The lengths of the lines between isolates represent the numbers of different alleles.

Resistome analysis of the 33 ST873 NDM-1 K. pneumoniae isolates that clustered closely together in the wgMLST analysis showed that 25 isolates had identical resistomes (Table 3). Of these, 20 were obtained from the outbreak location and 5 isolates originated from four other HCCs. The remaining 8 isolates differed in their resistance gene profiles, each lacking one or more resistance genes compared to the group of 25 isolates.

TABLE 3.

Resistance genes among the outbreak isolates identified by ResFinder software

Phenotype	Resistance gene	Presencea
		K. pneumoniae (n = 33)							E. coli (n = 11)
		V1 (n = 25)	V2 (n = 2)	V3 (n = 2)	V4 (n = 1)	V5 (n = 1)	V6 (n = 1)	V7 (n = 1)	V1 (n = 5)	V2 (n = 3)	V3 (n = 1)	V4 (n = 1)	V5 (n = 1)
Beta-lactam resistance	blaNDM-1	+	+	+	+	+	+	+	+	+	+	+	+
	blaCMY-6	+	+	+	+	+	+	+	+	+	+	+	+
	blaOXA-1	+	+	+	+	+	+	+
	blaSHV-27	+	+	+	+	+	+	+
	blaCTX-M-15	+				+	+	+		+	+
	blaTEM-1B	+					+	+		+	+		+
Aminoglycoside resistance	aac(3)-IId												+
	aadA1												+
	aadA2												+
	aadA5											+
Fluoroquinolone and aminoglycoside resistance	aac(6′)Ib-cr	+	+	+	+	+	+	+
Aminoglycoside resistance	aph(3′)-Ia	+	+	+	+	+	+	+	+	+	+	+	+
	rmtC	+	+	+	+	+	+	+	+	+	+	+	+
	strA	+					+	+		+	+		+
	strB	+					+	+		+	+		+
	aph(3′)-VIa	+	+	+	+	+	+	+	+	+	+	+	+
Rifampin resistance	arr-2	+	+	+	+	+	+	+	+	+	+	+	+
Phenicol resistance	catB3	+	+	+	+	+	+	+
	cmlA1	+	+	+	+	+	+	+	+	+	+	+	+
Trimethoprim resistance	dfrA12												+
	dfrA14	+	+	+	+	+	+	+		+	+
	dfrA17											+
Fosfomycin resistance	fosA	+	+	+	+	+	+	+
Macrolide resistance	mph(A)												+
Quinolone resistance	oqxA	+	+	+	+	+	+	+
	oqxB	+	+	+	+	+	+	+
	qnrB66	+		+	+			+
	qnrS1									+	+
Sulfonamide resistance	sul1	+	+	+	+	+	+	+	+	+	+	+	+
	sul2	+					+	+		+	+	+	+
	sul3												+
Tetracycline resistance	tet(A)	+			+					+	+	+	+

^a+, present.

Plasmid analyses showed that the cluster of 33 ST873 K. pneumoniae isolates all had identical plasmid compositions. The bla_NDM-1 gene was located on a variant of the pNDM-US plasmid (CP006661), a 140-kb plasmid belonging to the IncA/C2 group (13). Compared to the pNDM-US reference genome, the plasmid in the 33 ST873 K. pneumoniae isolates lacked a 740-bp segment that carried the aac(6′)-1B-cr (fluoroquinolone and aminoglycoside resistance) gene. This finding was corroborated by electroporation experiments, since the transformants originating from two ST873 K. pneumoniae isolates, one from the outbreak location and one from another HCC, did not carry the gene. Besides bla_NDM-1, the plasmid also harbored the beta-lactam resistance gene bla_CMY-6 and the resistance genes rmtc (aminoglycoside resistance) and sul-1 (sulfonamide resistance). In addition to the variant of the pNDM-US plasmid, all 33 ST873 K. pneumoniae isolates also carried two plasmids belonging to the IncFIB(K) and IncFII(K) groups. The three unrelated K. pneumoniae isolates from the outbreak location and the NDM-positive K. pneumoniae context isolates obtained from the national CPE surveillance did not carry the plasmid found in the ST873 isolates.

Transmission of the pNDM-US variant to E. coli.

During the outbreak, Escherichia coli isolates producing NDM-1 carbapenemase were cultured from 11 patients at the outbreak location. Nine of these patients also carried an ST873 K. pneumoniae isolate. An epidemiologically unrelated E. coli isolate submitted from the outbreak location in June 2015, 5 months prior to the outbreak, also produced NDM-1, whereas the other unrelated isolate carried a bla_NDM-5 gene. PCR of the 70 E. coli isolates originating from the national CPE surveillance that served as context showed that they carried the bla_OXA-48 (n = 42), bla_NDM (n = 26), bla_KPC (n = 1), and bla_VIM (n = 1) carbapenemase genes (Table 2). Only 7 of the 26 E. coli NDM-positive context isolates produced the NDM-1 variant, whereas 15 other isolates produced NDM-5 and the remaining isolates produced NDM-7 (n = 2) and NDM-6 (n = 1). Similar to the K. pneumoniae outbreak isolates, the meropemen MICs of the 11 E. coli isolates ranged from 0.75 to 6 μg/ml and were considerably lower than those of the E. coli context isolates.

The STs, inferred from the NGS data, revealed that each of the 11 E. coli isolates from the outbreak location had a unique ST. The two unrelated E. coli isolates submitted from the outbreak location and the 70 E. coli isolates obtained from the national CPE surveillance also showed great diversity in STs. The wgMLST analysis, based on 4,503 genes, showed that the 11 E. coli outbreak isolates were scattered over the tree and that no clustering occurred (Fig. 3). The average distance from one of the outbreak E. coli isolates to the closest neighboring E. coli isolate was 1,091 genes, with a range between 304 and 2,319 genes. The unrelated bla_NDM-1-positive E. coli isolate from the outbreak location did not cluster with any other isolate, and the distance from the closest neighbor was 3,016 genes. The other unrelated bla_NDM-5-positive E. coli isolate clustered closely with another E. coli isolate carrying bla_OXA-181, with a distance of 87 alleles.

FIG 3.

Minimum spanning tree based on whole-genome MLST of 83 E. coli isolates. The tree was based on 4,503 genes of the E. coli scheme, and clustering was done using a categorical coefficient. Each isolate in the tree is displayed as a circle. The lines between the isolates denote the distance in numbers of alleles.

Compared to the K. pneumoniae isolates, the E. coli isolates from the outbreak location showed more diversity in the resistome (Table 3). Plasmid analysis showed that all 11 E. coli outbreak isolates and the NDM-1-positive E. coli isolate, submitted in June 2015, harbored the same plasmid as the K. pneumoniae isolates, namely, the Inc A/C2 group plasmid variant of pNDM-US. A transformant of an E. coli isolate from the outbreak location confirmed that the same plasmid was present in both species. Apart from this plasmid, the E. coli isolates showed considerable diversity in their plasmid profiles, and the compositions were different in all the isolates (data not shown).

Four of the seven NDM-1-positive E. coli isolates from the national CPE surveillance also carried a plasmid belonging to the Inc A/C2 group. A single isolate carried the same plasmid variant of pNDM-US as the K. pneumoniae and E. coli outbreak isolates. This isolate originated from a patient from another HCC, and the patient was one of the seven persons who also carried an ST873 K. pneumoniae isolate (Fig. 1 and 2). The plasmid compositions of the other three NDM-1-positive E. coli surveillance isolates differed from the plasmid present in the K. pneumoniae and E. coli outbreak isolates, indicating that another plasmid from the Inc A/C2 group carried the NDM-1 gene.

Multiple NDM-1-producing species in the same patient.

From 9 of the 31 outbreak patients, two to four different NDM-1-positive Enterobacteriaceae species were isolated during the outbreak period. Seven patients carried K. pneumoniae and E. coli, one also carried Citrobacter freundii and Klebsiella oxytoca, and one patient carried Proteus mirabilis and K. oxytoca, in addition to K. pneumoniae and E. coli (Table 1 and Fig. 4). In all nine cases, the K. pneumoniae isolate was cultured before or on the same day as the other species. However, from a patient not related to the outbreak whose E. coli isolate was submitted in June 2015, prior to the outbreak, a ST873 K. pneumoniae isolate was also cultured after the outbreak in March 2016. Similar to K. pneumoniae and E. coli, all the other species carried the variant of plasmid pNDM-US lacking the aac(6′)-1B-cr gene. An NDM-1-positive C. freundii isolate was cultured from a single patient, and this isolate had the same ST and resistome as the C. freundii isolate cultured from the patient who also carried NDM-1-positive K. pneumoniae and E. coli isolates.

FIG 4.

Timeline of the outbreak for 10 patients who carried multiple NDM-1-producing Enterobacteriaceae species. The colors indicate isolates originating from the same patient. The dates below the isolates represent the sampling dates.

DISCUSSION

To date, only a few NDM-positive isolates have been described in the Netherlands, indicating a limited presence of strains producing this metallo-beta-lactamase (11, 12). Here, we report a large outbreak in a Dutch hospital comprising 26 patients colonized with NDM-1-producing K. pneumoniae. In addition, 16 NDM-1-positive isolates comprising five other Enterobacteriaceae species harboring the same plasmid were also cultured at the outbreak location.

All K. pneumoniae outbreak isolates had ST873, were identical or very closely related in the wgMLST, and contained a variant of the IncA/C2 plasmid pNDM-US (CP006661) that harbors the bla_NDM-1 gene. This indicates that the outbreak was caused by transmission of the same NDM-1-producing K. pneumoniae strain. Three epidemiologically unrelated bla_NDM-1-positive K. pneumoniae isolates collected at the outbreak location were also genetically unrelated according to wgMLST, had a completely different resistome, and did not carry the variant of plasmid pNDM-US. Consequently, they did not belong to the outbreak.

The NDM-1-positive K. pneumoniae outbreak strain was also isolated from seven other persons originating from five other HCCs geographically dispersed throughout the Netherlands. These seven isolates were submitted for the Dutch CPE surveillance between December 2015 and October 2016. Despite extensive epidemiological investigations, none of the seven cases could be connected to the cases from the hospital where the outbreak occurred. Considering their very limited genetic variation, these strains most likely come from a common source. This conclusion was supported when the observed genetic variation was compared to a recently defined cutoff for relatedness between K. pneumoniae isolates (14). Although no evidence was found for this outbreak, NDM-1-producing K. pneumoniae strains could have been introduced via food supplied to various HCCs, as exemplified by the recent study from Egypt, where 12 of 100 tested carcasses were contaminated with CPE isolates, 11 of which were K. pneumoniae (15). The contamination of meat with CPE has not been described in the Netherlands, but several reports have shown high prevalence of extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae (16, 17). Furthermore, several of the health care centers, including the hospital with the outbreak, received poultry from the same supplier. Unfortunately, an attempt to reconstruct the diets of the patients involved in the outbreak was unsuccessful at the time that food was considered a possible source.

Another scenario could be a prolonged, unnoticed circulation of this K. pneumoniae strain or of other NDM-1 strain-carrying Enterobacteriaceae in the Dutch population or in patients or health care workers at the outbreak location. This scenario is supported by the fact that an E. coli strain carrying the same plasmid as the outbreak isolates was isolated at the outbreak location in June 2015, 5 months prior to the outbreak. Moreover, an NDM-1-positive ST873 K. pneumoniae isolate was cultured from the same patient after the outbreak in March 2016. However, given the short outbreak period and the fact that no further transmission was noticed after the outbreak, the latter suggestion seems less likely.

The bla_NDM-1 gene was also present in 11 E. coli isolates collected at the outbreak location during the outbreak. The wgMLST and resistome analyses showed that the isolates were genetically unrelated. Like the K. pneumoniae outbreak isolates, all 11 E. coli isolates carried the same variant of plasmid pNDM-US, indicating multiple occurrences of horizontal gene transfer rather than transmission of the same E. coli strain. This was further supported by the fact that the transfer of the plasmid was also observed in five other isolates, comprising three other Enterobacteriaceae species, at the outbreak location. In addition, the transfer of the plasmid was also seen in another HCC where an ST873 K. pneumoniae isolate and an E. coli isolate from the same patient harbored the same plasmid as the outbreak isolates.

Our study has a number of limitations. First, our study has fully focused on bacterial isolates that harbored carbapenemase genes. The genetic background and potential spread of noncarbapenemase isolates from the different HCCs is therefore unknown. Second, our analysis comprised only isolates from the Dutch CPE surveillance. NGS data from publicly available databases containing international carbapenemase-producing K. pneumoniae isolates were not assessed and used. However, we did notice that ST873 K. pneumoniae isolates were occasionally found in recent studies from Switzerland and China, but these isolates did not produce carbapenemase (18, 19).

In conclusion, we report a hospital outbreak with an NDM-1-producing K. pneumoniae strain from an unknown source that was also found in patients from five other HCCs in the Netherlands in the same time frame. The plasmid carrying the bla_NDM-1 gene was also found in other Enterobacteriaceae species isolated at the outbreak location and in another HCC. The fact that the outbreak strain was also found in other HCCs without a clear epidemiological link shows that analysis of the NGS data on CPE isolates collected during national CPE surveillance is important to reveal the spread of carbapenemase-producing pathogens in the Netherlands.

MATERIALS AND METHODS

Hospital setting and outbreak description.

The outbreak was first noted during screening for a possible ESBL transmission in the surgical ward of a 683-bed tertiary teaching hospital in the southern Netherlands. On 23 November 2015, a carbapenemase-producing bla_NDM-positive, ESBL-positive K. pneumoniae strain was isolated during this screening. At the time of identification, the patient had already been discharged. Shortly after, ESBL screening of cultures from long-term-admitted surgical patients revealed two additional patients with NDM-producing K. pneumoniae. Contact tracing and weekly screening rounds identified additional NDM carriers. There were two wards with a single NDM carrier and seven wards with two or more NDM carriers. All the patients admitted to one of these wards from the 1October 2015 on were defined as at risk for carrying NDM. The outbreak period was defined as the date 2 weeks prior (1 October) to the admission of the initial patient until the date when all suspected NDM carriers were identified and put into contact isolation (30 December). Weekly screening rounds were continued during the first 4 weeks of January 2016 on risk wards to confirm that the outbreak was successfully controlled. All 2,964 patients admitted to one of the risk wards during the outbreak period were tested for NDM carriage. Six months after the end of the outbreak, over 95% of the patients at risk had been screened, and in total, 29 NDM carriers (1%) had been identified.

Bacterial isolates.

The first three isolates were obtained from an ESBL screening using an ESwab on ChromID ESBL agar (bioMérieux Inc., Marcy l'Etoile, France). All subsequent isolates were obtained by screening for carbapenem resistance using enrichment in tryptone soya broth supplemented with ertapenem and vancomycin. During the outbreak period, 42 isolates carrying a bla_NDM gene detected by PCR were obtained from 29 persons at the outbreak site and included in the study (Table 1). In most cases, a single NDM-producing species was found per patient. However, in nine patients, two to four distinct NDM-producing species were isolated. Most isolates were K. pneumoniae (n = 26) and E. coli (n = 11), but two C. freundii isolates, two K. oxytoca isolates, and one P. mirabilis isolate were also cultured. Retrospective analysis showed that five more isolates carrying a bla_NDM gene had been isolated at the outbreak site between 2015 and 2016 and subsequently submitted to the national reference center. These isolates (three K. pneumoniae and two E. coli) were also included in the study. For the Dutch national surveillance of CPE, medical microbiology laboratories are requested to submit Enterobacteriaceae isolates with a meropenem MIC of >0.25 μg/ml and/or an imipenem MIC of >1 μg/ml to the National Institute for Public Health and the Environment for further molecular characterization. From the Dutch CPE surveillance, NGS data from all K. pneumoniae (n = 133) and all E. coli (n = 70) carbapenemase-producing isolates obtained between January 2014 and December 2016 were included for analysis to provide genetic context. All isolates were tested for the production of carbapenemase using the carbapenem inactivation method (CIM) and subjected to PCR to detect the presence of carbapenemase-encoding genes, as previously described (20). Isolates that were shown by PCR to carry a bla_NDM gene and by CIM to produce carbapenemase are referred to as NDM positive.

Antimicrobial susceptibility testing.

At the outbreak location, all isolates were cultured and analyzed for resistance using Vitek (bioMérieux Inc., Marcy l'Etoile, France). Resistance to carbapenem was confirmed by assessing the meropenem MICs for all 250 isolates using Etest (bioMérieux Inc., Marcy l'Etoile, France).

Next-generation sequencing.

All 250 isolates were subjected to NGS using the Illumina HiSeq 2500 (BaseClear, Leiden, the Netherlands). The NGS data on the K. pneumoniae and E. coli isolates were used for MLST and wgMLST analyses using the available wgMLST schemes in SeqSphere software version 2.3.0 (Ridom GmbH, Münster, Germany). The resulting data were imported into Bionumerics version 7.6 for comparative analyses (Applied Maths, Sint-Martens-Latem, Belgium).

The resistomes and plasmid compositions of all the isolates were determined by uploading the assembled contigs into the online tools ResFinder (version 2.1) and PlasmidFinder (version 1.3) via the Center for Genomic Epidemiology website (21, 22). For ResFinder, a 90% identity threshold and a minimum length of 60% were used as criteria, whereas for PlasmidFinder, an identity of 95% was utilized. The NGS data on the presumed outbreak isolates were mapped against known NDM plasmids present in the NCBI database in May 2016.

Electroporation experiments.

To confirm that the same plasmid was transferred between species, an NDM-positive K. pneumoniae isolate and an NDM-positive E. coli isolate from the outbreak location were used to transfer the plasmid carrying the bla_NDM gene to a susceptible E. coli (DH10B) isolate by electroporation. The plasmid of another NDM-positive K. pneumoniae isolate from another health care center was also transferred to show that the same plasmid was present in multiple locations. For this, the electroporation protocol of the manufacturer was used (Invitrogen, Carlsbad, CA, USA). Transformants were subjected to PCR to confirm the presence of the bla_NDM gene, and NGS was performed on each transformant. The resulting NGS data were analyzed in CLC Genomics Workbench version 9.5.3 (Qiagen, Hilden, Germany).