Molecular characteristics of group B Streptococcus isolates from infants in southern mainland China

Background Invasive group B Streptococcus (GBS) disease in Chinese infants has gradually gained attention in recent years, but the molecular epidemiology of the pathogen is still not well known. Methods This multicenter study retrospectively investigated distribution of capsular serotypes, sequence types (STs), and hypervirulent GBS adhesin gene (hvgA) in clinical GBS isolates that caused invasive disease in infants aged < 3 months of age in southern mainland China between January 2013 and June 2016. Genes for antibiotic resistance to tetracycline, erythromycin, and clindamycin were also examined. Results From a total of 93 GBS isolates taken from 34 early-onset disease (EOD, 0–6 days after birth) and 59 late-onset disease (LOD, 7–89 days after birth) cases, four serotypes were identified: serotypes III (79.6%), Ib (12.9%), Ia (4.3%), and V (3.2%). Serotype III accounted for 73.5% of EOD and 83.1% of LOD and was responsible for 75.5% of cases involving meningitis. Fifteen STs were found, with the majority being ST17 (61.3%), ST12 (7.5%), ST19 (7.5%), and others (23.7%). 96.8% of STs belonged to only five clonal complexes (CCs): CC17 (64.5%), CC10 (12.9%), CC19 (9.7%), CC23 (6.5%), and CC1 (3.2%). The hvgA gene was detected in 66.7% of GBS isolates and 95% of CC17 isolates, all of which were serotype III except one serotype Ib/CC17 isolate. A large proportion of GBS isolates were found to be resistant to tetracycline (93.5%), clindamycin (65.5%), and erythromycin (60.2%). Genes of tetO (74.7%) and tetM (46.0%) were found in tetracycline resistant isolates, linB (24.6%) in clindamycin resistant isolates, and ermB (87.5%) and mefA (3.6%) in erythromycin resistant isolates. Conclusion Our results reveal higher prevalence of serotype III, ST17, CC17, hvgA expressing, and antibiotic resistant GBS isolates than previously reported in southern mainland China. This study provides guidance for appropriate measures of prevention and control to be taken in the future.

7-89 days after birth, respectively [6]. GBS is a highly diverse organism and commonly colonizes the human gastrointestinal and genitourinary tracts. Based on capsular polysaccharides (CPS), GBS can be classified into 10 serotypes including Ia, Ib, and II through IX. However, only limited subtypes have been identified in infants with invasive infection, indicating that variation in expressed surface molecules is related to pathogenicity of strains. The Madrid et al. meta-analysis found that serotype III accounted for the highest proportion of serotypes globally (61.5%) [4], while our group's recent study found the proportion of serotype III in Guangdong province was 77.9% [7], much higher than the global level. Jones et al. analyzed GBS isolates with multilocus sequence typing (MLST) and reported that serotype III strains have been categorized to ST17 and belong to complex clone 17 (CC17) [8]. In the CC17 lineage, which is strongly associated with neonatal meningitis, the hvgA gene encoding a surface-anchored protein named hypervirulent GBS adhesin (HvgA) has been identified [9]. The HvgA surface-anchored protein not only acts as an important adhesion medium for destroying and penetrating the intestinal and blood-brain barriers, but also mediates the migration of bacteria into the bloodstream and the central nervous system, which leads to infection [9,10].
In addition to expressing many factors which help it evade immune detection and clearance, acquiring antimicrobial resistance genes is another strategy observed in GBS to improve survival within the host. Penicillin G and ampicillin are recommended by U.S. CDC as the first-line drugs for prophylaxis and treatment of neonatal GBS infection, while clindamycin was the recommended alternative for individuals with a β-lactam allergy [6,11]. Unfortunately, the proportion of isolates reported to be resistant to erythromycin and clindamycin has increased in recent years, and especially high macrolide resistance in GBS have been detected among isolates circulating in both infected infants and adult female carriers in China [11][12][13].
Although invasive GBS infection in Chinese infants has gradually gained attention in recent years, no standardized or official guidelines for GBS prevention and control exists and the molecular epidemiology of the pathogen remains not well known. Furthermore, published studies have shown that geographical and ethnic differences in hosts effect the molecular epidemiology of invasive GBS [9,[14][15][16][17]. By determining serotypes, sequence types (STs), hvgA expression and antibiotic resistance of GBS isolates from infants with invasive infections in four hospitals in Guangdong province, China between January 1, 2013 and June 30, 2016, this study aims to elucidate the molecular characteristics of neonatal GBS disease prevailing in southern mainland China. This work will be helpful for exploring preventive and treatment strategies for GBS infections in southern mainland China and for the development of serotypebased vaccines in the future.

Study design
The study population was comprised of infants younger than 90 days of age with GBS isolated from a normally sterile medium including blood, cerebrospinal fluid (CSF), synovial fluid, and/or bone marrow, from the following four hospitals in Guangdong, China, between January 1, 2013 and June 30, 2016: Guangzhou Women and Children's Medical Center, Dongguan Tungwah Hospital, Guangdong Women and Children's Hospital, and Zhongshan Boai Hospital. We retrospectively collected demographic information, clinical signs and symptoms, and laboratory data for each GBS patient from Hospital Information Systems (HIS) and medical records at each study site.
GBS sepsis was defined as neonates in whom GBS was cultured from blood with clinical symptoms or signs including fever, respiratory distress, apnea, cyanosis, poor feeding, jaundice, lethargy, and seizure. Additionally, meningitis was diagnosed if a) CSF was cultured positive for GBS, or b) GBS negative but with a cellular reaction of more than 20 leukocytes/μL in CSF associated with GBS positive blood culture and consistent clinical manifestation.

GBS isolates
All GBS strains were confirmed by the four study hospitals to be S. agalactiae using VITEK 2 COMPACT microbiology systems (BioMerieux, Marcy L'Etoile, France). GBS isolates collected during the study period were centralized in the microbiology laboratory of Guangzhou Women and Children's Medical Center for analysis, where they were cultured at 37°C in 5% CO 2 in trypticase soy agar supplemented with 5% sheep blood. We used ATCC 2592 and ATCC 49619 as quality control bacteria.

Capsular serotyping
GBS serotyping was performed using Strep-B-Latex® rapid latex agglutination test kit (Statens Serum Institute, Hillerød, Denmark), according to the manufacturer's instructions. The capsular genotype was verified using a previously described multiplex PCR assay [18].

Multilocus sequence typing (MLST)
Genomic DNA was prepared from overnight GBS cultures by a standard protocol for Gram-Positive bacteria, using Bacterial Genomic DNA kit (Takara Code NO.9164, Japan). Typing was performed by sequencing the internal fragments of seven house-keeping genes (adhP, pheS, atr, glnA, sdhA, glcK, and tkt). PCR was used to amplify genomic DNA using the primer pairs indicated at http://pubmlst.org/sagalactiae/info/primers. shtml. PCR products were purified and sequenced in both directions by BGI Tech Solutions Co. Ltd. Alleles and sequence types (STs) were determined using the S. agalactiae MLST website (http://pubmlst.org/sagalactiae/). Sequence types that shared five or more alleles of the seven loci were clustered into a clonal complex (CC) using eBURST software. The term "singleton ST" refers to an ST that did not cluster into a CC. BioNumerics software version 5.1 (Applied Maths, Belgium) was used to create minimum spanning trees to illustrate relationships between MLST and CCs.

Detection of hypervirulent GBS adhesion (hvgA) gene
The hvgA gene was amplified by PCR using previously described primers ST17S and ST17AS [18]. Amplification products were purified and sequenced by BGI Tech Solutions Co. Ltd.

Antibiotic resistance
We used VITEK 2 COMPACT microbiology systems to determine the minimum inhibitory concentrations (MICs) of the following 11 antibiotics: penicillin, ampicillin, ceftriaxone, vancomycin, linezolid, erythromycin, azithromycin, levofloxacin, ofloxacin, ciprofloxacin and clindamycin. The criteria of the Clinical and Laboratory Standards Institute (CLSI), 2016 edition, were applied for interpretation. S. pneumonia ATCC 49619 and Staphylococcus aureus ATCC25923 were used as quality control strains in each set of tests to ensure accuracy of results. Erythromycin resistant genes erm(A), erm(B), mef(A), and tetracycline resistant genes tet(L), tet(K), tet(O), tet(m), being the most common antimicrobial genes, were detected by PCR. The primers and PCR conditions followed previously described methods [19,20].

Statistical analysis
Continuous data was expressed in terms of numbers and percentages and Fisher's exact test was used for comparison of categorical variables. Statistical analyses were performed using SPSS software (ver. 21.0; SPSS Inc., U.S). A p-value of < 0.05 was considered statistically significant. All antibiotic susceptibility data were extracted using WHONET 5.6 software, as recommended by the World Health Organization (WHO).

Results
A total of 93 non-duplicate GBS isolates which caused invasive disease in infants were recovered; among them were 65 from Guangzhou Women and Children's Medical Center, 12 from Dongguan Tungwah Hospital, 8 from Guangdong Women and Children's Hospital and Health Institute, and 8 from Zhongshan Boai Hospital. GBS identification by the Vitken Compact 2 system were confirmed for all isolates using mass spectrometry. A total of 34 cases (36.6%) were diagnosed as EOD and 59 (63.4%) were diagnosed as LOD (Table 1). There were 49 GBS cases which were diagnosed with meningitis, 47% (16/34) of which were observed in EOD cases and 55.9% (33/59) in LOD cases. Overall, there were 43 GBS cases which were diagnosed with sepsis, and one case with arthritis.

Hypervirulent GBS adhesion gene
The hvgA gene was detected in 66.7% of the invasive GBS isolates (62/93,

Antimicrobial resistance phenotypes and genotypes
Universal antimicrobial susceptibility to β-lactam antibiotics (penicillin and ampicillin) was detected, while low levels of resistances were observed for levofloxacin (8.6%), ofloxacin (2.2%), and azithromycin (1.1%). 93.5% of isolates (87) were resistant to tetracycline, owing to the tetO gene in 74.7% (65/87) and tetM gene in 46.0% (40/87) of GBS isolates, while the tetL and tetK genes were not observed (Table 3). Moreover, both tetO and tetM genes were detected in 2 of the 6 tetracycline-susceptible isolates. Erythromycin resistance (60.2%) was mostly due to modifications of the ribosomes, linked to the presence of ermB (87.5%, 48 isolates), and in 3.6% (2 isolates) of cases to an efflux mechanism encoded by the genetic determinant mefA. The ermA gene was not found, but ermB was detected in 30.0% (6/20) of erythromycin-susceptible isolates. Resistance to clindamycin was observed in 61 isolates (65.6%) and there were another 12 isolates (12.9%) exhibiting erythromycin-induced clindamycin resistance. The linB gene was identified in 24.6% (15/61) of clindamycin-resistant isolates. The resistance to tetracycline, erythromycin, and clindamycin was not significantly different between serotypes or CC17 isolates, but resistance to clindamycin was higher in GBS isolates with hvgA than the isolates without (p < 0.05).

Discussion
GBS remains a leading cause of infant sepsis and meningitis worldwide within the first 90 days of life. The recent meta-analysis by Madrid et al. showed the incidence of invasive GBS disease in infants was 0.49 per 1000 live births worldwide, with serotype III (61.5%) being the dominant type and with 97% of cases caused by just five serotypes (Ia, Ib, II, III, and V) [4]. In the present study, serotypes Ia, Ib, III, and V were detected, and serotype III was the dominant serotype observed (79.6%), which is higher than the global average of 61.5% cited above, but similar to estimates from France [9], Italy [20] and other parts of China [7,13].GBS vaccines covering five serotypes (Ia, Ib, II, III and V) have already entered Phase III clinical trials [16]; an approved vaccine would cover 95% of global newborn GBS infection [2,5,21], and would protect from all serotypes detected in this study. Additionally, in contrast to the approximately 50% of GBS-related meningitis cases caused by serotype III in infants worldwide [21,22], serotype III accounted for 62.5 and 81.8% in EOD and LOD cases with meningitis in this study group, respectively. In the present study, we obtained 15 STs from 93 invasive GBS isolates, of which 61.3% were caused by the highly pathogenic genotype ST17. The five types of CCs detected in this study (CC17, CC1, CC10, CC19, and CC23) are also the most common clones globally. Among 93 isolates, more than half of neonatal GBS infection were caused by the highly pathogenic strain CC17 (ST17, ST357, ST188). We found serotype III was more commonly associated with the most virulent clone, clonal complex (CC17). CC17 accounted for 78.4% of serotype III strains, and mainly caused the observed meningitis. Studies have shown that less than 12% of GBS strains with ST17 colonize adults, whereas CC17, in contrast, presents as highly pathogenic and virulent among both infants with under-developing immune systems and immunocompromised patients [10]. In the process of occurrence and development of neonatal invasive infection, CC17 strains enhance the pathogenicity and invasiveness of GBS through a special virulence mechanism, which is the main cause of meningitis in these patients (> 80%) [9,10,23,24]. CC17 colonization is the main risk factor for neonatal GBS infection [25,26]. A study from the Netherlands found that the main contributing factor to the increase in incidence of GBS infection between 1987 and 2011 was the increase in prevalence of CC17 [25]. Manning SD et al. [23] found ST17 and ST19 were the local dominating genotypes of neonatal invasive infections in Canada, and that ST1, ST12, and ST23 were the primary genotypes among pregnant women. In China, similar to the results of that study, ST17 was found as the main genotype for newborns infected with GBS, and ST19 for pregnant women [27]. CC17 and CC19 are the most common important gene pedigrees of GBS serotype III, and likely for that reason serotype III is the most common serotype causing GBS invasive disease in infants.
Our findings showed around 95% of serotype-III/CC17 isolates harbouring the hvgA gene, which suggested that hvgA is involved in the pathogenesis of invasive GBS Table 3 Comparison of resistance to tetracycline, erythromycin, and clindamycin of GBS isolates with different serotypes, CC17 and hvgA  Compared to the hvgA(+) isolates, P < 0.05 infectious diseases. The ST17 strain without the selective hvgA gene presents low adhesion to the host cell, which is comparable to that of the non-ST17 strain [10]. The hvgA gene was previously known to be carried only by ST17, while our study found the hvgA gene was mainly detected from strains with CC17 (92%,57/62), while CC10, CC19, and CC23 strains also carried the hvgA gene. Moreover, a strain with serotype-Ib/ST357/CC17 had the hvgA gene as well. Our findings also indicated that most of the hvgA genes have been detected in serotype-III/ST17 strains, which is consistent with other reported findings [18,28]. Antimicrobial susceptibility testing of GBS isolates is critical for selecting appropriate antibiotic treatment. We found all isolates were susceptible to β-lactam antibiotics over the study period. Some studies have found that there was a decrease in the susceptibility of GBS to β-lactam (such as penicillin and cephalosporin) [29][30][31]. This study found that 93.5% of GBS isolates showed resistance to tetracycline, comparable to France (90, 95.5%) [24,26], Italy (93.3%) [20], and Beijing (100%) [13]. The rate of erythromycin resistance was 60.2%, which is higher than previous results from France (16.7, 13.8%) [24,26], Italy (12%) [20], and the United States (32%) [2]. This high resistance to macrolide antibiotics including erythromycin and clindamycin is possibly due to the overuse of antibiotics in animal husbandry and humans in China in the past decades. We found 85.7% of the erythromycin-resistant strains carried the resistance gene ermB. In this study, tetracycline resistance was mainly mediated by tetO and tetM, with carrier gene rates of 75 and 46%, respectively, whereas European countries were mainly dominated by tetM, with gene rates of 95% in France [24,26] and 97.1% in Italy [20]. Furthermore, it was observed that a small number of antibiotic-sensitive GBS strains carried resistant genes, such as tetracycline-sensitive strains and erythromycin-sensitive strains. We also found linB is an important reason for clindamycin resistance; among 61 strains of clindamycin-resistant strains, 15 (24.6%) strains had the linB gene. This is lower than observations in South Africa where 38% (11/29) of strains had linB [32]. In contrast, linB was not detected in some investigations from Korean [33] and Iran [34]. Our group's previous work revealed multi-drug resistance clusters (tetracycline, aminoglycoside, macrolide/lincomycin, and others) in GBS CC17 hypervirulent isolates by using genomic analysis [35]. This suggests that the antibiotics recommended by the intrapartum antibiotic prophylaxis program to prevent neonatal GBS in southern China are in need of updating. On the other hand, in addition to strengthening the rational use of antibiotics, and taking into account the higher proportion of CC17 of neonatal infection at the local level, we suggest more attention should be put on GBS multi-drug resistance control.
Until now, many studies have reported on the resistance spectrum of GBS strains among colonized pregnant women, but by contrast, fewer studies have been performed on the resistance spectrum of neonatal GBS infection [2,13,20,24,26]. In this study, genotypes exhibiting resistance to tetracycline and erythromycin were significantly correlated with a drug-resistant phenotype. However, antibiotic resistance was not limited to a specific serotype or sequence type. There was no correlation between GBS resistance and serotype, gene sequence type, and type of disease presentation. A study by Poyart C. et al. demonstrated similar results [24], but other studies have found the opposite. Joubrel, C. et al. reported that GBS serotype had a correlation with resistance, with the serotype-III strain having a low resistance rate and a high resistance rate to serotype-V strain [26]. One possible reason might be due to differences in the study population.
Our study has several limitations. GBS isolates were only from one province in China, Guangdong, and the sample size is relatively small. Therefore, these findings may not uniformly represent the whole China. We also did not collect data of population-based incidence rate and maternal information, which prevented limited further population-level analysis.

Conclusions
The molecular epidemiologic characteristics of GBS neonatal invasive infections in southern China were mainly comprised of serotype-III/CC17. In Guangdong, resistance to tetracycline was mainly associated with tetO and tetM genes, erythromycin resistance was associated with the presence of ermB, and clindamycin resistance was associated with linB. It is recommended to establish corresponding prevention and control measures accordingly. Further large scale studies and routine GBS surveillance are required to learn more about this infection and to prevent it from occurring.

Funding
This study was supported by grants from Guangzhou Science Technology and Innovation commission (grant number 201804010447), and Health and Family Planning Commission of Guangzhou Municipality (grant number 20151A010034). The funding supported consumables for conducting molecular biology studies.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate This study was approved by the ethical committees of Guangzhou Women and Children's Medical Center (reference number 2017021915), and written informed consents were obtained from parents or guardians of all infants enrolled by the care providers.

Consent for publication
Not applicable.

Competing interests
We declare that we have no conflicts of interest. David J McIver (DJM) works at Metabiota Inc., Nanaimo, Canada. This does not alter our adherence to the journal policies on sharing data and materials. Metabiota Inc. provided in-kind support in the form of participation of author [DJM], but did not have any additional role in the study design, data collection and analysis, or decision to publish. No conflict of interest exists in the submission of this article. I declare on behalf of all my co-authors that the article has not been published, and is not under consideration for publication elsewhere. All the listed authors have approved the enclosed manuscript.