Rapid Identification of Different Escherichia coli Sequence Type 131 Clades

ABSTRACT

Escherichia coli sequence type 131 (ST131) is a pandemic clonal lineage that is responsible for the global increase in fluoroquinolone resistance and extended-spectrum-β-lactamase (ESBL) producers. The members of ST131 clade C, especially subclades C2 and C1-M27, are associated with ESBLs. We developed a multiplex conventional PCR assay with the ability to detect all ST131 clades (A, B, and C), as well as C subclades (C1-M27, C1-nM27 [C1-non-M27], and C2). To validate the assay, we used 80 ST131 global isolates that had been fully sequenced. We then used the assay to define the prevalence of each clade in two Japanese collections consisting of 460 ESBL-producing E. coli ST131 (2001-12) and 329 E. coli isolates from extraintestinal sites (ExPEC) (2014). The assay correctly identified the different clades in all 80 global isolates: clades A (n = 12), B (n = 12), and C, including subclades C1-M27 (n = 16), C1-nM27 (n = 20), C2 (n = 17), and other C (n = 3). The assay also detected all 565 ST131 isolates in both collections without any false positives. Isolates from clades A (n = 54), B (n = 23), and C (n = 483) corresponded to the O serotypes and the fimH types of O16-H41, O25b-H22, and O25b-H30, respectively. Of the 483 clade C isolates, C1-M27 was the most common subclade (36%), followed by C1-nM27 (32%) and C2 (15%). The C1-M27 subclade with bla_CTX-M-27 became especially prominent after 2009. Our novel multiplex PCR assay revealed the predominance of the C1-M27 subclade in recent Japanese ESBL-producing E. coli isolates and is a promising tool for epidemiological studies of ST131.

INTRODUCTION

The global increase in resistance to third-generation cephalosporins and fluoroquinolones among extraintestinal pathogenic Escherichia coli (ExPEC) isolates is a public health emergency due to the importance of these drugs in the treatment of serious infections (1). The increase in antimicrobial resistance among ExPEC isolates is mainly due to the recent expansion of a pandemic clonal group known as E. coli ST131, which is usually fluoroquinolone resistant and often associated with extended-spectrum β-lactamases (ESBLs) responsible for third-generation cephalosporin resistance (2, 3).

Recent studies using whole-genome sequencing (WGS) analysis revealed that ST131 consists of different lineages or clades: A/H41, B/H22, and C/H30 (4–6). The H numbers indicate the fimH allele, which mostly corresponds to each clade (4–6). Since the 2000s, clade C has become the most dominant lineage among ST131 isolates (currently, up to 80% of global ST131 isolates belong to clade C) (7). Clade C mainly consists of the subclades C1 and C2 that are fluoroquinolone resistant with characteristic quinolone resistance-determining region (QRDR) mutations (4–6). Interestingly, C2 (also known as H30Rx) is associated with bla_CTX-M-15, while C1 is often negative for ESBLs. We recently described a novel global C1 subclade named C1-M27, which is associated with bla_CTX-M-27 and is present among Japanese ESBL-producing ST131 isolates (8). The other clade C subclade, C0, is considered an ancestor of subclades C1 and C2, and the clade B subclade B0 is considered an ancestor of subclade C0 (6). These C0 and B0 subclades were regarded as belonging to an “intermediate” clade because they had intermediate characteristics between those of clades B and C (6).

The detection of ST131 and its clades is important for epidemiological studies. WGS- or sequence-based strain typing using multilocus sequence typing is considered the gold standard for identifying ST131; however, this approach is expensive and time-consuming. Simple conventional PCR assays to detect ST131, O16-H41, H30, or H30Rx (9–13) are helpful and are often used because they are rapid and cheap, but they cannot detect other clades or subclades. Reliance on the fimH alleles for classification of ST131 clades has a risk due to the presence of alternative fimH alleles (6, 14). Therefore, a more convenient and reliable multiplex method with the ability to detect all of the major ST131 clades, including the C subclades, is required for the optimal detection of ST131 clades for global surveillance studies. Here, we designed a novel multiplex PCR assay to detect ST131 clades in a single reaction. We used this method to identify ST131 clades in two different Japanese E. coli collections.

RESULTS AND DISCUSSION

Definition of clades and subclades of the 390 ST131 genomes.

A total of 390 ST131 genomes were used for the identification of the genomic regions and single nucleotide polymorphisms (SNPs) specific to each clade and subclade. A recombination-free maximum-likelihood phylogeny made from 6,667 core SNP sites represented three major clades, A, B, and C, including four subclades within C (C0, C1, C1-M27, and C2) in agreement with previous studies (6, 8) (see Fig. S1 in the supplemental material). The 390 genomes included clades A (n = 46), B (n = 77), and C (n = 267) and subclades C0 (n = 7), C1-nM27 (n = 85), C1-M27 (n = 29), and C2 (n = 146) (see Data Set S1 in the supplemental material).

The C1-M27 subclade-unique M27PP1 region was found exclusively in a cluster of 29 isolates within the C1 clade. This cluster included all 19 C1-M27 subclade isolates from the original C1-M27 study (8), which had included isolates with bla_CTX-M-27 from the Petty et al. study (5), eight isolates from the Stoesser et al. study (14), one isolate sequenced in this study, and one isolate from the Price et al. study (4). Based on the phylogeny and the M27PP1 result, these 10 isolates not derived from the original C1-M27 study (8) were defined to belong to the C1-M27 subclade. They are from Laos (n = 4; 2007 to 2009), Thailand (n = 2; 2009 to 2011), Cambodia (n = 1; 2010), Japan (n = 1; 2014), Germany (n = 1; 2010), and the United States (n = 1; 2010). The original C1-M27 study focused on isolates with ESBLs and the C1-M27 subclade that uniformly had bla_CTX-M-27 (8). Among these ten new isolates, six had bla_CTX-M-27, one had bla_CTX-M-24, and three lacked any ESBL genes. The discovery of C1-M27 in European (Germany) and Asian countries (Laos and Cambodia) further supports its global distribution.

Identification of genomic regions and SNPs specific to each clade and subclade.

The pangenome analyses found four clade A-specific regions (see Table S1 in the supplemental material), but no other regions specific to clade or subclade B, C, C1, C1-M27, or C2 were found. We designed a primer that targeted region 4 because this region was absent in non-ST131 E. coli genomes (Table 1; see also Table S1 in the supplemental material).

TABLE 1.

Primers used for the detection of E. coli ST131 clades

Assay and primer	Nucleotide sequence (5′-3′)	Primer concn (μM)	Target	PCR product size (bp)	Reference or source
ST131 clade assay
CladeAspe4-YF5	TGACGGGACGTGAGCAAATTA	0.15	Clade A specific (region 4)	707	This study (Table S1)
CladeAspe4-YR5	AGTCAGACCTAGCCACCCTT	0.15
ST131_R19-YF1	AGCAACGATATTTGCCCATT	0.15	ST131 specific (region 19)	580	11; this study
ST131_R19-YR1	GGCGATAACAGTACGCCATT	0.15
prfC-1615spe0-YF1	CAACGTTGAAGCAGTGTATGAG	0.08	prfC SNPs specific for the clade B	442	This study (Table S2)
prfC-d2034-YR1	TGACAATCGACGGCTTTAGA	0.08
C1-578spe-YF1	GGCCCCACAAATTGCTT	0.1	Clade C1	337	This study (Table S2)
C1-898-YR1	CGCACCTCCGATACCAAA	0.1
M27PP1C-YF1	TGAATCAAAGGTCCGAGCTG	0.08	M27PP1 region specific for the C1-M27 subclade	232	8; LC209430a
M27PP1C-YR1	TATGGCTGGCAGATGCTTTA	0.08
nrdI-534spe2-YF1	ACGGATTCAGGTAGACGATT	0.25	nrdI SNP specific for the C2 clade	164	This study (Table S2)
nrdI-678R	CCTCACCAAAGTTGCGATTAC	0.25
C-SNP1-700spe-YF1	CGCTGGCCAGTTATCTGAAAT	0.2	mgtA SNPs specific for the C clade	103	This study (Table S2)
C-SNP1-762spe-YR2	CCTTTCACCAACTGGGTTACT	0.2
C1-M27 subclade assay
ST131_R19-YF1	AGCAACGATATTTGCCCATT	0.15	ST131 specific (region 19)	580	11; this study
ST131_R19-YR1	GGCGATAACAGTACGCCATT	0.15
M27aer-spe-YF1	GCCGATGGGCTTTCCT	0.15	aer SNP specific for the C1-M27 subclade	140	This study (Table S2)
M27aer-YR2	GTCACCGCGTCTTCCAGT	0.15

^aGenBank/ENA/DDBJ accession number.

Analysis of SNPs specific for clades and subclades B, C, C1, C1-M27, and C2 found 4, 16, 1, 5, and 1 SNP sites, respectively (see Table S2 in the supplemental material). We used some of these SNPs to design primers for the ST131 clade and the C1-M27 subclade assays (Table 1; see also Table S2 in the supplemental material). The only C2-specific SNP found in this study was situated in nrdI, although the PCR assay that has often been used for the detection of the C2 clade utilized ybbW SNP (9). One C2 genome sequenced in this study (BRG145; marked with an arrow in Fig. S1 in the supplemental material) lacked the ybbW SNP.

Validation of the ST131 clade and C1-M27 subclade assays.

We validated two PCR assays using the WGS collection that consisted of 80 isolates. Both assays produced the expected amplicons (Fig. 1 and Table 2; see also Fig. S2 in the supplemental material) for all isolates. Thus, the ST131 assay correctly identified clades and subclades A, B, C1-M27, C1-nM27, and C2. The three clade C0 isolates were classified to clade C other than subclades C1 or C2, as expected. The C1-M27 assay correctly identified the C1-M27 subclade.

FIG 1.

Agarose gel electrophoresis of the ST131 clade PCR amplicons. Lanes: M, 100-bp DNA ladder; 1, strain SNEC15, clade A; 2, strain KFEC6, clade B; 3, strain KFEC8, subclade C1-M27; 4, strain SNEC5, subclade C1-nM27 (C1-non-M27); 5, strain ONEC14, clade C2; 6, strain KSEC7, subclade C0; 7, strain BRG210, subclade I1; 8, strain BRG28, non-ST131; 9, no-template control. All strains underwent WGS except for BRG28.

TABLE 2.

Identification of ST131 clades and subclades using the ST131 clade assay

ST131 clade/subclade	Amplicons of the ST131 clade assay
	A specific	ST131 specific	B specific	C1 specific	C1-M27 specific	C2 specific	C specific
A	+	+	–	–	–	–	–
B	–	+	+	–	–	–	–
C
C1
C1-M27	–	+	–	+	+	–	+
C1-nM27 (non-C1-M27)	–	+	–	+	–	–	+
C2	–	+	–	–	–	+	+
C other than C1 or C2a	–	+	–	–	–	–	+
Unclassifiedb	–	+	–	–	–	–	–
Non-ST131	–/+	–	–/+	–/+	–/+	–/+	–/+

^aIsolates found in this study belonged to subclade C0 or C3.

^bIsolates found in this study belonged to subclade I1, belonging to the intermediate clade between clades B and C.

Application of the ST131 clade assay for the ESBL and ExPEC collections.

Table 3 shows that the ST131 clade assay correctly detected the ST131 status of the ESBL and ExPEC collection isolates (n = 460 and 105, respectively). In agreement with previous ST131 studies (5, 14), the O serotypes and fimH types of O16-H41, O25b-H22, and O25b (or O nontypeable) H30 perfectly corresponded to clades A (n = 54), B (n = 23), and C (n = 483), respectively. The other five ST131 isolates that contained O25b-H54 were unclassified by the ST131 clade assay and further investigated (detailed below). Among the clade C isolates, the H30Rx status defined by the ybbW SNP-based PCR assay corresponded to subclade C2, except for 1 subclade C2 isolate in the ExPEC collection (BRG145, as mentioned above). These 86 C2 isolates were associated with bla_CTX-M-15 (94.2%; whose presence was defined in previous studies [15, 16]), confirming their relationship (4). The 202 C1-M27 and 180 C1-nM27 subclade isolates were associated with bla_CTX-M-27 (98.0%) and bla_CTX-M-14 (61.1%), as expected from our previous studies (8, 15, 16). The other 15 clade C isolates other than C1 or C2 were considered to belong to subclade C0 according to the current understanding of the clade C lineages (4–6). To confirm this expectation, we further investigated these isolates (detailed below).

TABLE 3.

Results of the ST131 clade PCR assay

Strain collection and characteristicsa	Total	ST131							Non-ST131
		A	B	C				Unclassified
				C1		C2	C other than C1, C2
				C1-M27	C1-nM27
ESBL (n = 460)
ST131	460	45	13	178	140	66	15	3
fimH41 group,b O16	45	45
fimH22 group,c O25b	13		13
fimH30, O25b/ONT,d H30Rx negative	333			178d	140d		15
fimH30, O25b, H30Rx	66					66
fimH54, O25b	3							3
blaCTX-M-27	179	2		176	1
blaCTX-M-14	157	31	4	1	104e	3f	11	3
blaCTX-M-15	72	3			5	64f
Other ESBLs	56	9	9	1	33		4
ExPEC (n = 329)
Non-ST131	224								224
ST131	105	9	10	24	40	20		2
fimH41 group, O16/O12g	9	9
fimH22 group,h O25b	10		10
fimH30 group,i O25b, H30Rx negative	65			24	40i	1j
fimH30, O25b/ONT,k H30Rx	19					19
fimH54, O25b	2							2
blaCTX-M-27	25			22	1				2
blaCTX-M-14	17	2	1		6	1f			7
blaCTX-M-15	20				1	17f			2
Other ESBLs	10				3				7
Non-ESBL	258	7	9	2	29	3		2	206

^aH30Rx status was defined by PCR detection of ybbW SNP.

^bOne isolate had fimH89 (1 SNP to fimH41) and was included in the WGS collection (clade A).

^cTwo isolates had fimH376 (1 SNP to fimH22) and were included in the WGS collection (clade B).

^dThree C1-M27 and three C1-nM27 isolates were O nontypeable (ONT).

^eTwo isolates had both bla_CTX-M-14 and bla_CTX-M-3. One isolate had both bla_CTX-M-14 and bla_CTX-M-2.

^fOne isolate had both bla_CTX-M-14 and bla_CTX-M-15.

^gTwo isolates had fimH488 (1 SNP to fimH41), one of which was included in the WGS collection (clade A). One isolate with fimH41 was O12.

^hTwo isolates had fimH338 (1 SNP to fimH22), one of which was included in the WGS collection (clade B).

ⁱOne isolate had fimH497 (2 SNPs to fimH30) and was included in the WGS collection (subclade C1-nM27).

^jThis isolate (BRG145) included in the WGS collection belonged to the C2 subclade and lacked H30Rx-specific ybbW SNP.

^kOne isolate was ONT.

C1-M27 subclade detection by the ST131 clade and the C1-M27 subclade assays.

The results from the ST131 clade and the C1-M27 subclade assays for the detection of the C1-M27 subclade in the ESBL and ExPEC collections were completely in agreement, which supports the presence of the C1-M27 subclade in the 188 isolates (of 202 isolates found) that did not undergo WGS. Thus, we concluded that a single ST131 clade assay is sufficient for the detection of the C1-M27 subclade.

Blinded assay validation.

A blinded validation study in the Shiga laboratory using the WGS and ExPEC collections yielded the same results except for one isolate in the ExPEC collections. The isolates did not produce any bands and were initially regarded as non-ST131. A repetition of the experiment produced the expected clade B amplicons, so we suspected that pipetting errors of the DNA template were responsible for the anomalous results.

Prevalence of the ST131 clades and subclade in Japanese ESBL-producing E. coli.

Figure 2 shows the rates of ST131 clades and subclades between 2001 and 2014 calculated from the results of the ST131 clade assay in the ESBL and ExPEC collections with ESBL genes. The C1-M27, C1-nM27, and C2 subclades contributed to the recent expansion of ESBL-producing E. coli in Japan. The C1-M27 subclade was the most prevalent ST131 subclade after 2009. These results clarified that the main contributor of the recent ESBL-producing E. coli in Japan is the C1-M27 subclade, which was regarded as the CTX-M-27-producing H30R group in our previous study (15).

FIG 2.

Yearly rates of ST131 clades and subclades in Japanese ESBL-producing E. coli. Rates of ST131 and subclades C1-M27 and C2 increased yearly (P < 0.001 each). In 2014, the rates of clades A and B and subclades C1-M27, C1-non-M27, and C2 were 3, 1, 31, 15, and 24%, respectively.

New subclades I1 and C3 were found by genome analysis of isolates unclassified by the ST131 clade assay.

All of the 5 O25b-H54 isolates unclassified by the ST131 clade assay underwent WGS (see Data Set S2 in the supplemental material). The phylogenetic tree analysis indicated these isolates belonged to a new lineage between B0 and C0 in the intermediate clade (Fig. 3) and were presumptively placed in a new subclade designated subclade I1 (Fig. 3). The genome analysis indicated that the subclade I1 isolates lacked all of the clade B or C characteristic SNPs, including the sites used for the ST131 clade assay (n = 4 and n = 16, respectively; see Table S2 in the supplemental material). The subclade I1 isolates contained the prophages Phi2 and Phi4 and the parC-1a genotype characteristic of subclades B0 and C0 (6). Among the genomic markers for clade C, including subclade C0 (6), the prophage Phi1 and the ST131 genomic island GI-pheV were present, whereas GI-leuX and ISEc55 insertions in fimB were absent. These findings suggest that subclade I1 shared an ancestor with subclade C0 and underwent subsequent recombination events in the core genome, including fimH. The five subclade I1 isolates were susceptible to ciprofloxacin, had O25b-H54 had emerged in 2010, and were collected from five Japanese hospitals. Three isolates had bla_CTX-M-14.

FIG 3.

Recombination-free core SNP-based phylogeny of 401 ST131 genomes. This maximum-likelihood phylogenetic tree is rooted by using the clade A isolates. A total of 6,881 core SNP sites were used after excluding the 18,693 core SNP sites that were located within the 1,955,761-bp recombination region. Branches that had >90% bootstrap support from 100 replicates are highlighted in blue. The bootstrap values for the root of subclades I1 and C3 were 100 and 99%, respectively. Genomes of the WGS collection isolates are marked in red. Eleven genomes (marked in blue) were sequenced to investigate their phylogeny in addition to the 390 genomes (see Fig. S1 in the supplemental material).

Of fifteen clade C isolates other than C1 or C2 indicated by the ST131 clade assay, three (KSEC7, KKEC3, and KUN5823) had been sequenced and defined as subclade C0 (see Fig. S1 in the supplemental material) without the C1/C2-characteristic QRDR mutations of gyrA-1AB and parC-1aAB (6). Six of the remaining twelve isolates underwent WGS (see Data Set S2 in the supplemental material). The genome analysis confirmed that these isolates lacked the subclade C1- and C2-specific SNPs (see Table S2 in the supplemental material). The phylogenetic tree analysis indicated that these isolates belonged to a new lineage in clade C and they were presumptively named subclade C3 (Fig. 3). Subclade C3 diverged from subclade C0 and had contained gyrA-1AB and parC-1aAB genotypes, similar to subclades C1 and C2. Among the 401 ST131 genomes in Fig. 3, the average number of SNPs between subclades C3 and C1 (n = 54) or C2 (n = 55) were similar to that between subclades C1 and C2 (n = 50) and was larger than that among C1 (n = 40), C2 (n = 38), and C3 (n = 26), which supports the hypothesis that subclade C3 is independent of C1 or C2. The other six isolates that did not undergo WGS also contained the gyrA-1AB and parC-1aAB genotypes, suggesting subclade C3 status. The subclade C3 isolates were not susceptible to ciprofloxacin, had O25b-H30 and ESBLs (bla_CTX-M-14, n = 10; bla_CTX-M-2, n = 2), had emerged in 2005, and were found in 2.6% of the ESBL-producing E. coli collection from five Japanese hospitals. Further analysis of both subclades I1 and C3 is needed to define their characteristics and their importance in evolutional history and global epidemiology.

The study limitations include the lack of extensive validation studies in other laboratories. The strengths include the use of large numbers of well-characterized ESBL-producing isolates and ExPEC isolates without any selection bias, which encompass the study population with which the assay will likely be used.

Conclusions.

To our knowledge, this is the first study to develop a PCR assay (the ST131 clade assay) to detect ST131 clades and subclades in a single reaction. This assay will elucidate the global molecular epidemiology of ST131 and provide potential knowledge to establish infection control policies for ESBL-producing E. coli. Application of this assay to Japanese isolates and subsequent genomic analysis identified two novel ST131 lineages that may be important to understand the evolutional history of ST131. We also showed that the C1-M27 subclade with bla_CTX-M-27 is responsible for the expansion of ESBL-producing E. coli in Japan, as well as the underlying global importance of this clade. We urgently need well-designed epidemiological and molecular studies to understand the dynamics of the transmission, risk factors, and reservoirs for C1-M27.

MATERIALS AND METHODS

Bacterial isolates.

We included three collections of clinical E. coli isolates for the PCR assays: the WGS collection for the validation of the method, as well as the ESBL and ExPEC collections for application. The WGS collection included 80 isolates, whose clades were defined using WGS-based analysis and phylogeny: A (n = 12), B (n = 12), C0 (n = 3), C1-M27 (n = 16), C1-nM27 (C1-non-M27; n = 20), and C2 (n = 17). These clades were derived from our previous study (8) (n = 53) and the present study (n = 27). Part of the WGS collection of isolates was selected from the ESBL (n = 55) and ExPEC collections (n = 15).

We assessed the prevalence of the ST131 clades and subclades using the ESBL and ExPEC collections of E. coli clinical isolates. The ESBL collections comprised 460 ESBL-producing ST131 isolates that were collected in 10 acute-care hospitals in the Kyoto and Shiga prefectures in Japan from 2001 to 2012 (15). The ExPEC collection comprised 329 isolates that were collected in ten acute-care hospitals in five prefectures in Japan during December 2014 (all clinical isolates during the collection period) and included seven major ST complexes containing ST131 (32%), ST95 (13%), ST73 (8%), ST357/1876 (5%), ST69 (4%), ST38 (3%), and ST1193 (2%) (16). These isolates from the two collections were previously defined for ciprofloxacin susceptibility, the presence of ESBL genes, ST131 status using PCR detection of specific SNPs in mdh and gyrB (12) or fumC-fimH typing (17), O typing (rfb alleles) (18, 19), fimH alleles using sequencing or H30-specific PCR (10), and H30Rx status using PCR detection of ybbW SNP (9). The clade C isolates that were negative for the C1 or C2 subclades according to the developed ST131 clade assay underwent Sanger sequencing to define the gyrA and parC genotypes (3).

Genome analysis.

We performed WGS using Illumina MiSeq as previously described (8) for the 27 isolates from the WGS collection and 11 isolates from the ESBL and ExPEC collections whose clades or subclades had not been classified by the ST131 clade assay we developed. In silico multilocus sequence typing, fimH typing, gyrA and parC typing, O serotyping, and C2-specific ybbW SNP typing were performed using the assembled draft genomes as previously described (8).

To design the PCR assay, we used data on 390 ST131 genomes from the three major ST131 studies reported to date (Petty et al. [5], n = 95; Stoesser et al. [14], n = 119; and Ben Zakour et al. [6]—originally by Price et al. [4], n = 90), 6 ST131 draft or complete genomes, and our WGS data set (see Data Set S1 in the supplemental material). The reads were assembled using SPAdes v3.10.1 (20). First, we created a recombination-free, core SNP-based phylogenetic tree to define the clades and subclades. We identified the SNPs by mapping the reads or aligning the genomes against a subclade C2 reference genome of EC958 (21) as previously described (8). Recombination sites identified according to Gubbins (22) were excluded and recombination-free core SNPs, sites that are present in all genomes, were included to create a maximum-likelihood tree using RAxML with the GTR GAMMA substitution model (23). The tree was visualized using iTOL v3 (24). We defined the C1-M27 subclade based on the phylogenetic tree and the presence of the C1-M27 subclade-specific prophage-like region (M27PP1) based on the original study (8). Second, to identify a genomic region specific to each clade or subclade, we performed a pangenome analysis using Panseq (25). The unique core SNPs of each clade or subclade were determined using an in-house Perl script. The genomes of 11 isolates mentioned above from unclassified clades or subclades were analyzed to determine their phylogeny, by following the same steps as above, for specific SNPs and the ST131 genomic islands (6).

ST131 clade PCR assay.

Bacterial DNA was isolated using a QIAamp DNA minikit (Qiagen, Hilden, Germany). The assay utilized seven primer sets designed in the present study (Table 1). Table 2 shows the expected amplicons for each clade. Amplification was performed with 1× PCR buffer, 1.5 mM MgCl₂, 0.2 mM deoxynucleoside triphosphates, each primer, 0.5 U of TaKaRa Ex Taq DNA polymerase (TaKaRa Bio, Inc., Otsu, Japan), and 1 μl of purified DNA in a total volume of 20 μl using a TaKaRa PCR Dice Touch thermal cycler (TaKaRa Bio). The cycling protocol was as follows: 30 cycles of 98°C for 10 s, 57°C for 20 s, and 72°C for 40 s. The PCR products were loaded on a 2.5% agarose gel with GelRed (Biotium, Hayward, CA) and photographed under UV light.

C1-M27 subclade PCR assay.

We developed another assay to detect only the C1-M27 subclade by using a primer set designed for the C1-M27 subclade-specific SNP in its core genome, which is different from the accessory genes target (M27PP1) used in the ST131 clade assay (Table 1), in order to support the presence of the C1-M27 subclade in the ESBL and ExPEC collections.

Blinded assay validation.

All of the PCR experiments were performed in the Kyoto laboratory. The ST131 clade assay was validated in another laboratory in Shiga using the WGS and ExPEC collections in a blinded manner.

Statistical analysis.

The yearly prevalence of isolates was analyzed using a chi-square test for linear trends. A P value of <0.05 was considered statistically significant. We conducted our statistical analysis using R version 3.3.2 (R Foundation for Statistical Computing [www.r-project.org]).

Accession number(s).

The sequences determined in this study have been deposited in the GenBank/ENA/DDBJ Sequence Read Archive database (DRA005691 and DRA005809).