Group B Streptococcus (GBS) is a critical pathogen that causes meningitis and sepsis in neonates. Neonatal GBS isolates are mainly acquired from the maternal genitourinary tract. The most important virulence factor of GBS is its polysaccharide capsule encoded by cps. Based on the capsule, 10 serotypes (Ia, Ib, II, III, IV, V, VI, VII, VIII, and IX) with different disease-causing abilities are distinguished [1]. Serotypes and antimicrobial resistance profiles are important parameters for the characterization and treatment of invasive GBS infections [2]. Serotype III isolates belonging to clonal complex (CC)17 are associated with hypervirulence and poor disease outcomes [3^-5]. In a study in the Netherlands, most isolates obtained from neonates showing invasive infections were clustered into one of five major lineages: CC17 (39%), CC19 (25%), CC23 (18%), CC10 (9%), and CC1 (7%) [3]. The number of neonatal GBS infections caused by CC17 isolates has significantly increased [3, 4, 6]. In contrast, CC1, CC12, and CC23 are more common in pregnant women [6]. In Korea, the prevalence of GBS lineages has not yet been reported.

With the rapid advances in bioinformatics that have allowed analyzing and storing large amounts of whole-genome sequencing (WGS) data, it has become possible and feasible to obtain genetic information of bacteria in clinical microbiology laboratories. Recent studies have used WGS for molecular capsular typing and antimicrobial resistance gene typing of GBS [7^-9]. As comprehensive data, including multilocus sequence types (MLSTs), serotypes, resistomes, and virulence factors, can be extracted from a single WGS dataset, WGS is a practical and economical method as compared with conventional phenotypic methods [9, 10].

We for the first time investigated the sequence type (ST) distribution of GBS isolates collected from pregnant women in Korea and validated WGS-based antimicrobial susceptibility and capsular serotypes of the GBS isolates.

Bacterial collection and WGS

In total, 75 GBS isolates were collected from pregnant Korean women visiting Wonju Severance Christian Hospital, Korea between May 2017 and May 2019, as previously reported [11]. The study was approved by the Yonsei University Wonju Severance Christian Hospital Institutional Review Board (IRB No. CR319119). Sequencing libraries for all isolates were prepared using the Twist Library Preparation EF Kit (Twist Bioscience, San Francisco, CA, USA) according to the manufacturer’s instructions. Briefly, extracted DNA was assessed using the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA). Fifty nanograms of high-quality DNA in 40 μL of 10 mM Tris-HCl (pH 8.0) was enzymatically fragmented, and DNA ends were repaired by dA-tailing in a C1000 Touch Thermal Cycler (Bio-Rad, Hercules, CA, USA). The dA-tailed DNA fragments were ligated with Twist Universal Adapters compatible with the NovaSeq 6000 system (Illumina, San Diego, CA, USA) at 20°C for 30 minutes. The ligated libraries were PCR-amplified in six cycles using Twist UDI primers in a thermal cycler. The quantities and size ranges of the final libraries were validated using the Qubit dsDNA BR Assay Kit (Thermo Fisher Scientific) and Agilent 4200 TapeStation System (Agilent Technologies, Santa Clara, CA, USA). All libraries were equally pooled, and the pooled library was diluted to 2 nM for sequencing, denatured in 0.2 N NaOH, and diluted with 400 mM Tris-HCl (pH 8.0) to 400 pmol/L. The diluted library was sequenced using paired-end (2×150 bp) sequencing on the NovaSeq 6000 system (Illumina). De novo assembly from the FASTAQ files generated after sequencing was performed using Unicycler (v0.4.0, https://github.com/rrwick/Unicycler). The median depth of coverage was 1,050×, with maximum depths of 2,264× to 529×.

Whole-genome analysis

The FASTA files generated after assembly were analyzed using tools provided on the Center for Genomic Epidemiology website (http://www.genomicepidemiology.org/) and the PubMLST database (http://pubmlst.org) to determine MLSTs, cps genotypes, antimicrobial resistance genes, and the hemolysin operon regions of the GBS isolates. The cps genotypes were analyzed using the cpsG-K region, as previously suggested [7, 8]. We compared the WGS-based serotypes, antimicrobial resistance genes, and phenotypic characteristics with the findings of a previous study that conducted antimicrobial susceptibility testing and serotyping of GBS isolates using the Strep-B-Latex kit (SSI Diagnostica, Hillerød, Denmark) [9].

Antimicrobial susceptibility test

Antimicrobial susceptibility was tested using the MicroScan MicroSTREP Plus Panel (Beckman Coulter, Brea, CA, USA), which covers ampicillin, penicillin, cefotaxime, ceftriaxone, cefepime, meropenem, levofloxacin, clindamycin, erythromycin, tetracycline, chloramphenicol, and vancomycin. GBS isolates showing inconsistent results between WGS and previous methods were re-examined using WBS and the previously used method.

MLSTs of GBS isolates

The distribution of GBS isolates according to MLST CCs is shown in Table 1. The predominant CC was CC1 (N=28, 37.3%), followed by CC19 (N=24, 32.0%) and CC12 (N=13, 17.3%). CC1 was composed of ST1, ST2, ST667, and ST676, whereas CC19 was composed of ST19, ST27, ST335, and ST1911. Three isolates (4.0%) belonged to ST17—a hypervirulent strain. Two new STs (ST1911 and ST1912) were identified (Table 1).

Table 1 . Distribution of MLST CCs and STs and cps genotypes among 75 GBS isolates collected from pregnant Korean women

CC	ST	cps genotype
		Ia	Ib	II	III	V	VI	VII	VIII	Total
1	2								13	13
	1					6	3	1		10
	676			4						4
	667						1			1
19	19				5	8				13
	335				5					5
	27				3	2				5
	1911				1					1
12	10		5							5
	12			3						3
	654		3							3
	8		1							1
	1912		1							1
23	23	4								4
17	17				3					3
22	22			1						1
NA	498					1				1
NA	529				1					1
Total (%)		4 (5.3)	10 (13.3)	8 (10.7)	18 (24.0)	17 (22.7)	4 (5.3)	1 (1.3)	13 (17.3)	75

Abbreviations: CC, clonal complex; ST, sequence type; MLST, multilocus ST; cps, capsular polysaccharide gene; GBS, group B Streptococcus; NA, not assigned.

Latex serotypes and cps genotypes

All isolates (100%, 75/75) were cps typeable, and 89.3% (67/75) of the cps genotypes were in agreement with serotypes determined using latex agglutination. Of the 16 isolates with discordant results, eight were confirmed by retesting using latex agglutination and WGS (Table 2). The cps genotypes of the 75 isolates mainly included serotypes III (N=18, 24.0%), V (N=17, 22.7%), and VIII (N=13, 17.3%). CC19 isolates were of serotypes III and V, whereas CC12 were of serotypes Ib and II. CC1 isolates were of diverse serotypes, including II, V, VI, VII, and VIII.

Table 2 . Retyping of isolates with discordant results between latex agglutination serotyping and cpsG-K genotyping

Isolate	Latex agglutination serotype		cpsG-K genotype
	Initial	Repeat
WJ1	VII	VI	VI
WJ4	VIII	III	III
WJ10	IX	IX	III
WJ11	Ia	V	V
WJ12	VII	VII	V
WJ14	VIII	III	III
WJ21	IX	IX	Ib
WJ36	V	V	II
WJ37	V	V	II
WJ41	V	V	II
WJ47	IX	Ib	Ib
WJ50	VII	VII	V
WJ59	IX	Ib	Ib
WJ60	Ib, II	II	II
WJ64	Ib	Ib	II
WJ68	II	Ib	Ib

Abbreviation: cps, capsular polysaccharide gene.

Antimicrobial susceptibility and antimicrobial resistance genes

Erythromycin resistance was predicted by the presence of erm, which encodes a methylase (Table 3). Twenty-nine isolates (38.7%) harbored erm, including 20 harboring ermB and nine harboring ermA. Eighteen isolates carrying intact ermB were non-susceptible to erythromycin, with 13 isolates showing high-level resistance (minimum inhibitory concentration [MIC] ≥128 μg/mL). One isolate with a premature termination codon (E128stop) in ermB and one isolate with a 15-bp insertion in ermB were susceptible to erythromycin. Among the nine isolates harboring ermA, six were resistant to erythromycin, whereas the remaining three were susceptible. Forty-seven isolates (62.7%) harbored tet, including 34 harboring tetM and 13 harboring tetO. Although the 46 isolates having intact tetM or tetO genes were resistant to tetracycline, one isolate was susceptible to tetracycline because of a premature termination codon in tetO (E203stop). No β-lactam-resistant isolate was identified.

Table 3 . Antimicrobial resistance genes in erythromycin- and/or tetracycline-resistant isolates according to cps genotypes and STs

cps genotype (N isolates)	CC	ST	Erythromycin resistance			Tetracycline resistance
			MIC (μg/mL)	Susceptibility	erm mutation	MIC (μg/mL)	Susceptibility	tet mutation
V (17)	1	1	256	R	ermB	≥8	R	tetM
			256	R	ermB	≥8	R	tetM
			256	R	ermB	≥8	R	tetM
			256	R	ermB	≥8	R	tetM
			256	R	ermB	≥8	R	tetM
			256	R	ermB	≥8	R	tetM
	19	19	256	R	ermB	≥8	R	tetM
			16	R	ermA	≥8	R	tetM
			0.12	S	ermA	≥8	R	tetM
			0.06	S	ermA	≥8	R	tetM
			0.25	S	-	≥8	R	tetM
			0.12	S	-	≥8	R	tetM
			0.06	S	-	≥8	R	tetM
			0.06	S	-	≥8	R	tetM
		27	2	R	ermB	≥8	R	tetO
			2	R	ermB	≥8	R	tetO
	NA	498	0.06	S	-	≥8	R	tetM
III (18)	19	19	256	R	ermB	≥8	R	tetM
			256	R	ermB	≥8	R	tetM
			0.50	I	ermB	≥8	R	tetM
			0.06	S	-	≥8	R	tetM
			0.06	S	-	≥8	R	tetM
		27	256	R	ermB	≥8	R	tetO
			256	R	ermB	≥8	R	tetO
			0.06	S	ermB (E128*)	≥8	R	tetO
		33	2	R	ermA	≥8	R	tetM
		5	1	R	ermA	≥8	R	tetM
			1	R	ermA	≥8	R	tetM
			1	R	ermA	≥8	R	tetM
			0.06	S	ermA	≥8	R	tetM
		1911	1	R	ermA	≥8	R	tetM
		529	0.06	S	-	≥8	R	tetO
	17	17	4	R	ermB	≥8	R	tetO
			4	R	ermB	≥8	R	tetO
			0.06	S	-	≥8	R	tetM
Ia (4)	23	23	0.06	S	-	≥8	R	tetM
			0.25	S	-	≥8	R	tetM
			0.06	S	-	≥8	R	tetM
			0.06	S	-	≥8	R	tetM
II (4)	12	12	0.06	S	-	≥8	R	tetO
			0.06	S	-	≥8	R	tetO
			0.06	S	-	≥8	R	tetO
	22	22	256	R	ermB	≥8	R	tetM
Ib (2)	12	8	0.06	S	-	≥8	R	tetM
		654	0.06	S	-	≥8	R	tetO
VI (2)	1	667	128	R	ermB	0.5	S	tetO (E203*)
		1	0.06	S	ermB†	0.5	S	-
VII (1)	1	1	0.06	S	-	≥8	R	tetM

*Stop codon; ^†A 15-bp fragment inserted at position 238.

Abbreviations: ST, sequence type; cps, capsular polysaccharide gene; CC, clonal complex; S, susceptible; I, intermediate; R, resistant; MIC, minimum inhibitory concentration; NA, not assigned.

Hemolysin operon analysis

Among the 75 isolates, six were non-hemolytic. We analyzed the cyl operon (12 kb) and abx1 regulatory gene reported previously [12]. A 1-bp insertion in cylE at position 903/2,004 was detected in three non-hemolytic strains belonging to ST27. Abx1 mutations were not found in any of the isolates.

CC1 and CC19 are among the major GBS CCs associated with invasive disease and colonization in humans [6, 13, 14]. CC1 is phylogenetically close to CC19 [6, 15]. We found that 69.3% of Korean pregnant women carried CC1 or CC19 GBS. CC17 GBS strains belong to a hypervirulent lineage of homogeneous serotype III clones and are associated with a disproportionately high frequency of invasive neonatal diseases, particularly, meningitis [3, 16]. In Korea, the prevalence of neonatal GBS cases is low [17, 18]. A multicenter study showed that 157 neonatal GBS cases were identified in 14 university hospitals of Korea between 1996 and 2005 [17]. Another study reported 10 GBS cases (0.3%) among 3,862 infants during 2010–2017 [18]. In this study, only three (4.0%) isolates were of CC17, and neonatal infection with maternal CC17 GBS was not observed. The low prevalence of CC17 GBS colonization may explain the low frequency of neonatal GBS infections.

Serotype classification of GBS is based on the capsular polysaccharide, of which 10 variants are known to exist. The capsular polysaccharide is encoded on the cps locus, which comprises 16–18 genes. Kapatai, et al. [7] have suggested that molecular serotypes based on the variable cpsG-K region demonstrated the best performance in terms of typeability and concordance with latex agglutination. Therefore, we compared phenotypic serotyping using latex agglutination with molecular genotyping using the cpsG-K region. All isolates (100%) were typeable by WGS-based cpsG-K genotyping, and 89.3% of the genotypes were concordant with latex agglutination results; isolates showing discordant results were retested. Similarly, Kapatai, et al. [7] reported a concordance rate of 86.7% in initial testing and of 98.2% in retesting between cps genotyping and latex agglutination results. However, one study has indicated that the determination of GBS serotypes is often hampered by poor capsule expression [19].

Serotypes Ia, Ib, II, III, and V account for 98% of the colonizing GBS isolates identified worldwide [5]. However, serotype distribution varies with geography and ethnicity. In the United States, Europe, and Australasia, serotypes Ia, II, III, and V account for 80%–90% of clinical isolates, whereas serotypes IV, VI, VII, VIII, and IX are relatively less frequent [5, 19]. Serotype III, which is associated with invasive disease, accounts for 25% worldwide; however, it is less frequent in South American (11%) and South-Eastern Asian (12%) populations [5]. Serotypes VI, VII, VIII, and IX are common in Asia [5]. Our results demonstrated that serotypes III and V are predominant in Korean pregnant women; they were found in 24.0% and 22.7% of the women, respectively, followed by serotype VIII (17.3%). These results suggest that continuous monitoring of serotype distribution is important in epidemiological and vaccine-related studies [19, 20]. Associations between STs and serotypes have been reported in the literature, with some studies reporting strong correlation and others very weak correlation [21, 22]. Ramaswamy, et al. [21] observed correlations between serotype III and ST17 and between Ib and ST12; serotype V was found to be present in all STs, except for ST17. Similarly, we found correlations between serotype III and ST17 and between serotype VIII and ST2.

We compared antimicrobial susceptibility results with antimicrobial resistance genes detected using WGS. Tetracycline resistance was predominantly caused by tetM and tetO, whereas macrolide resistance was predominantly due to the presence of erm, with ermB being more prevalent. In a previous study, resistant and intermediate-resistant GBS isolates showed high frequencies of tetM (97.6%) and low frequencies of tetO (2.4%), ermB (34.5%), and ermTR (10.3%) [23]. In this study, 45.3% harbored tetM, 17.3% harbored tetO, 26.7% harbored ermB, and 12.0% harbored ermA. Isolates carrying intact tetM or tetO were predicted to be resistant to tetracycline. Regarding erythromycin resistance, our results are similar to those reported by Mingoia, et al. [24], in which ermB and ermA were associated with high-level and variable-level erythromycin resistance, respectively. Multidrug resistance to erythromycin, clindamycin, and tetracycline, coupled with the recovery of non-susceptible isolates resistant to antimicrobial agents such as cefazolin, penicillin G, and ampicillins indicates the importance of GBS surveillance and antimicrobial susceptibility testing [23].

Mutations are localized predominantly in the cyl operon, encoding the β-hemolytic pigment biosynthetic pathway, and in abx1, encoding a CovSR regulatory partner [12]. In the present study, three non-hemolytic strains (WJ16, WJ29, and WJ46) had 1-bp frameshift insertions at nucleotide 903 of cylE. However, abx1 mutations were not found.

This study had some limitations. First, our results are not representative of all pregnant Korean women, as we used single-center data. Second, we collected the isolates from pregnant women, not from neonates, and the strains involved in GBS infection in these two groups are not necessarily the same.

In conclusion, CC1 and CC19 GBS are prevalent in pregnant Korean women. The low prevalence of CC17 GBS, which mainly causes neonatal invasive infection, explains the low frequency of neonatal GBS infections in Korea. WGS data can predict the serotypes of GBS isolates based on cps genotypes. Detection of tetM and tetO and ermB is predictive of resistance to tetracycline and erythromycin, respectively. Therefore, WGS is a useful tool for simultaneous genotyping and antimicrobial resistance determination.

None.

Lee K and Uh Y designed the study; Uh Y, Bae HG, Won D, Yun W, Choi JK, and Lee H collected and identified clinical isolates and performed molecular studies; Lee Y analyzed the data; Lee Y, Lee K, and Uh Y wrote, edited, and reviewed the manuscript. All authors revised and accepted the final version of the manuscript.

No potential conflicts of interest relevant to this article are reported.

This study was supported by a research grant from Seoul Clinical Laboratories.