Background
Group B Streptococcus (GBS) is a recognized cause of infant sepsis and meningitis globally and is a leading cause of morbidity and mortality in Africa [
1]. GBS is also a common commensal colonizer of the gastrointestinal and urogenital tracts of women and maternal colonization is a major factor in mother-to-child GBS transmission. Early onset (< 7 days) GBS disease has been well characterized, whereas the epidemiology of late onset disease (LOD, 7–89 days) is less well understood [
2].
Penicillin G is the drug of choice for intrapartum prophylaxis [
3] and GBS isolates remain mostly susceptible to penicillin. Its prophylactic use has been instrumental in significantly reducing the incidence of early-onset diseases in neonates [
4]. However, GBS isolates with reduced susceptibility to penicillin have been reported for more than a decade [
5]. Erythromycin and clindamycin have been used as alternatives to prevent vertical transmission of GBS among pregnant women who are allergic to penicillin, but resistance to these antibiotics has emerged in several countries, including reports from Africa [
6,
7].
GBS strains are subdivided according to type-specific capsular polysaccharides into 10 unique serotypes which are also a major focus of vaccine development [
8]. Serotypes I to V account for about 98% of colonizing GBS isolates worldwide with serotype III usually associated with invasive disease and less common among colonization isolates. GBS serotype distribution is not uniform across different geographic regions and temporal variations have also been described [
9]. As information on GBS from their genomes has accumulated, molecular methods have also proved very useful for investigating the population structure of GBS and discriminating differences between strains isolated from different sources [
5,
10‐
12]. For example, the ST17 lineage has been associated with neonatal infections, particularly with late-onset disease [
11]. There are previous reports showing diversification and shifts in serotypes [
5,
12,
13] and abundant evidence of past capsular switching within several MLST-based lineages [
12,
14] which also pose potential challenges for the development of a vaccine. Global data on circulating GBS strains is important for disease control and for informing the development of effective vaccines. However relatively few studies are available from Sub-Saharan Africa [
1] and there are limited data on serotypes and strain characterization [
7].
Ethiopia is an important country with a substantial birth cohort in Africa, and while there are previous data published from Ethiopia on antibiotic susceptibility patterns for GBS, there are limited available detailed descriptive data on circulating strains [
7,
15]. Here we used whole genome sequencing to investigate serotype distribution, clonal relationships, lineage distributions, virulence factor determinants, and antimicrobial susceptibility patterns of
S. agalactiae strains recovered from pregnant mothers and their newborns attending three hospitals in Ethiopia.
Methods
Study population
This prospective, cross-sectional study was conducted at 3 hospitals in Ethiopia: Adama Hospital Medical College (AHMC), Hawassa University Comprehensive Specialized Hospital (HUCSH) and Tikur Anbessa Specialized Hospital (TASH) between June 2014 and September 2015. The three hospitals were selected based on their convenience. AHMC is a rural hospital located in Adama City in the Oromia regional state. It is located 100 km due east of Addis Ababa and has a total population of 220,212. HUCSH, a rural hospital, is located in Hawassa which is the capital city of Southern Nations Nationalities and Peoples Region (SNNRP) and is located 275 km south of Addis Ababa. The total population of Hawassa town is 235,000. TASH is an urban hospital located in Addis Ababa, the capital city of Ethiopia, with a large population size of 3,384,569.
Eligibility criteria
Pregnant women who were admitted for delivery along with their newborn were included. Pregnant women with cesarean section delivery and those who were on antibiotic treatment for the last 2 weeks prior to data collection were excluded.
Newborns who were suspected of neonatal disease, those with signs and symptoms of neonatal disease (breathing problem, reduced movement, reduced suckling, seizure, slow or increased heart rate, vomiting, increased or reduced body temperature).
Isolation of Bacteria
Recto-vaginal swabs from 840 pregnant women, samples from the nasal area, external ear, umbilical cord or throat area of 857 newborns and blood samples from newborns suspected of neonatal disease were collected. The detail are as follows: Number of pregnant women and their newborn at AHMC, Adama were 280 and 282 respectively; data collection period was from June 2014 to October 2014. Number of pregnant women and their newborn at HUCSH, Hawassa were 280 and 292 respectively; data collection period was November 2014 to March 2014. Number of pregnant women and their newborn at TASH, Addis Ababa were 280 and 283 respectively; data collection period was March 2015 to August 2015. One hundred seventy-six newborns suspected of early onset disease (defined as occurring during the first week of life) were included from TASH, Addis Ababa; data collection period was from March 2015 to August 2015.
At each study site, to isolate GBS from pregnant women and newborns, recto-vaginal swabs from mothers and samples from the nasal area, external ear, umbilical cord or throat area of newborns were placed into Lim broth (BD Diagnostics, USA) and incubated for 18–24 h at 37 °C in CO2 enriched atmosphere. Then sub-cultured onto sheep blood agar (BD Diagnostics, USA) and incubated in CO2 enriched atmosphere at 37 °C for 18–24 h. If there was no growth, blood agar plate was re-incubated and examined after 48 h.
To isolate GBS from newborns suspected of early onset disease, about 1 ml blood was inoculated into Tryptone Soy Broth (BD Diagnostics, USA). All blood cultures were incubated aerobically at 37 °C and inspected daily for 7 days for the presence of visible microbial growth by observing any of one of the following changes: turbidity, hemolysis, gas production and coagulation of broth. Blood cultures with sign of microbial growth were sub-cultured onto blood agar (BD Diagnostics, USA). The blood agar plate was incubated aerobically in CO2 enriched atmosphere at 37 °C for 24–48 h.
To identify GBS, hemolytic reaction on BAP (beta-hemolytic or non-hemolytic), Gram reaction, catalase test, CAMP (Christie, Atkins, and Munch-Petersen) test and Strep B Grouping Latex (Remel, USA) were used. Isolates were stored at − 70 °C in medium containing skim milk, tryptone, glucose, and glycerol [
16] and transported to the Streptococcus Laboratory at the Centers for Disease Control and Prevention for confirmation and characterization. For several analyses, where more than one isolate was available for mothers and/or babies and they had the same serotype and MLST type, only a single isolate was selected so as not to duplicate results.
DNA extraction and whole genome sequencing
At CDC, isolates were cultured on Trypticase soy agar supplemented with 5% sheep blood (BAP). A positive CAMP test and Strep. B Grouping Latex (Remel, USA) were used to confirm isolates as S. agalactiae.
For whole genome sequencing (WGS), GBS isolates were cultured on BAPs and incubated overnight at 37 °C. Genomic DNA was extracted manually using a modified QIAamp DNA mini kit protocol (Qiagen, Inc., Valenica, CA, USA) (
https://www.cdc.gov/streplab/downloads/pcr-body-fluid-dna-extract-strep.pdf). Nucleic acid concentration was quantified by Qubit assay (Thermo Fisher Scientific Inc., Waltham, MA, USA) and samples were sheared using a Covaris M220 ultrasonicator (Covaris, Inc., Woburn, MA, USA) programmed to generate 500-bp fragments. Libraries were constructed on the SciCloneG3 (PerkinElmer Inc., Waltham, MA, USA) using an Ovation Rapid multiplex library preparation kit with 96 dual indices (NuGEN, San Carlos, CA, USA) and quantified by KAPA qPCR library quantification method (Kapa Biosystem Inc., Wilmington, MA, USA). Short read sequences were generated with MiSeq v2 500 cycle kit (Illumina Inc). Isolate identifiers, pipeline features, and assembly metrics are listed in Supplementary Table
1. For the 119 isolates that yielded high quality sequencing metrics with contig size below 500, sequences are available in the NCBI repository and accession number provided in Supplemental Table
1.
Serotyping and surface protein detection
A multiplex PCR assay was initially used for the direct identification of the capsular serotypes (Ia to IX) of GBS [
17]. In addition, serotype was confirmed using the CDC short-read WGS bioinformatics pipeline [
5,
12]. The presence of hypervirulent GBS adhesion determinant (
hvgA), serine-rich repeat gene (
srr), one of four alpha-family surface protein genes (rib, alpha, alp1, alp2/3) and the pilus islands (PI-1, PI-2a, and PI-2b) were also queried through the CDC pipeline (
https://github.com/BenJamesMetcalf) [
12].
Multilocus sequence typing
Seven locus MLST to assign sequence type (ST) was facilitated from whole genome sequence data with the CDC pipeline using SRST2 and database at
http://pubmlst.org/sagalactiae/ [
5]. eBURST was used to group isolates into lineages or clonal complexes (CCs), based upon sharing at least six of 7 MLST alleles with one or more other members [
18]. The relations between STs and serotypes of GBS isolates were illustrated by the minimum spanning tree (PHYLOVIZ software version 2.0; PHYLOViZ team, Lisbon, Portugal).
Antimicrobial susceptibility testing
Antimicrobial susceptibility patterns of GBS were tested by broth microdilution. The antimicrobials tested included penicillin, cefotaxime, erythromycin, clindamycin, levofloxacin, vancomycin, daptomycin, tetracycline, and linezolid. Isolates were classified as sensitive, intermediate, or resistant according to Clinical Laboratory Standards Institute (CLSI) guidelines [
19]. Strains were determined to be multidrug resistant if resistant to ≥3 different antibiotic classes. Phenotypic MICs were compared with WGS-predictions for non-β-lactam antibiotics (except daptomycin) using sequence queries and a bioinformatics pipeline (
https://github.com/BenJamesMetcalf/Spn_Scripts_Reference) for detection of resistance determinants provided in a previous study [
5]. The PBP2x typing scheme used serves to flag missense mutations within the
pbp2x gene for subsequent isolate MIC testing for beta-lactam antibiotics.
Discussion
Vaccination of pregnant women against GBS is a promising strategy to prevent invasive GBS disease in their infants [
8]. Vaccine candidates include protein-based formulations and serotype-specific polysaccharide-protein conjugates [
20] and thus an understanding of serotype and surface-protein antigen distribution in maternal colonization and infant disease worldwide is important. While there have been several reports published from Ethiopia on maternal and infant GBS colonization and disease over the past decade, little information on GBS strain characteristics has been described. In this study, serotype II was the most common predicted serotype, with five types (Ia, Ib, II, III and V) accounting for the vast majority of isolates. This is consistent with maternal colonization serotypes described from other studies in Africa and globally [
7,
9], however, serotype II is not usually the predominant serotype but rather serotypes III and V [
9]. Globally, the vast majority of invasive and colonizing GBS isolates are grouped into five CCs (CC1, CC10, CC17, CC19, and CC23) [
21,
22]. In this study, 88% of isolates were grouped into one of these CCs, although no isolates belonging to CC1 and CC17 were identified, emphasizing the diversity of
S. agalactiae in human isolates and highlighting the potential for local geographic differences. A recent study from Northwest Ethiopia grouped GBS in four CCs (CC1, CC10, CC19 and CC23) [
23]. Similar to findings by others [
21,
22] we also saw evidence of past capsular locus replacement events within 2 genetic lineages (ST19 and ST196) which may have implications for vaccine development strategies.
Surface protein antigens play an important role in the pathogenesis of GBS infection, and several of these antigens have been documented as promising vaccine targets [
24]. Our data are consistent with past work indicating that a vaccine containing the 3 pilus protein components could be effective in preventing disease caused by GBS as all strains carried at least one of the pilus proteins [
25]. Similarly, in all of the isolates one of four highly related Alpha family proteins (Alpha, Rib, Alp2/3, Alp1) were detected suggesting possible broad coverage of the fused N terminal domain Rib-Alpha (GBS-NN) vaccine tested in phase I clinical trials in pregnant women [
26]. The HvgA surface-anchored protein has been found to be critical for GBS intestinal colonization and translocation across the blood brain barrier during the onset of meningitis [
27]. There were no isolates from meningitis cases in this study but a small proportion of GBS colonization isolates (2.4%) contained the
hvgA gene. The
hvgA gene has been detected in previous studies, primarily among highly virulent
S. agalactiae belonging to serotype III/ST17 [
27] and rarely found among non-serotype III isolates [
14,
28]. Two
hvgA positive isolates in our sampling were of serotype Ia/ST934, which is genetically divergent from other characterized GBS. The closest MLST match within the global database was the triple locus variant ST1232 from an invasive,
hvgA-negative serotype II strain recovered during 2017 [
12http://pubmlst.net].
S. agalactiae have generally been considered universally susceptible to penicillin although there have been several reports of isolates with mutations associated with decreased susceptibility over the last decade [
12,
29,
30]. Here, PBP 2X types were restricted to types, 1, 4 and 5, commonly seen in penicillin-susceptible US isolates [
5] and all isolates were sensitive to penicillin and cefotaxime. This supports data published from Jimma, Ethiopia [
31], however, several previous colonization studies in Ethiopia have documented large numbers of GBS strains resistant to penicillin [
15,
32]. Differences in these reported rates warrants further investigation to determine if these rates are indeed real or due to challenges with appropriate and accurate laboratory testing for GBS species identification and antibiotic resistance. The proportion of isolates with in vitro resistance to both erythromycin and clindamycin was similar to rates described by Mengist et al. [
31] but lower than that reported from other regions of Ethiopia [
7,
32] and other countries [
7,
12,
33]. Combined resistance to erythromycin and clindamycin in GBS is most commonly due to 23S rRNA methylases encoded by different
erm genes which supports our findings with
ermTR and
ermB determinants predominantly associated with macrolide and lincosamide resistance [
5]. The high level of tetracycline resistance, associated with
tetM resistance determinant, may strengthen the hypothesis that current globally circulating
S. agalactiae strains in humans were selected by tetracycline usage in 1940s [
34]. The proportion of isolates resistant to levofloxacin were similar to that from United States [
5,
12], Taiwan [
35], Italy [
36] and Brazil [
37] and lower compared to data reported from China [
38] and Canada [
39].
A key strength of our study was the molecular characterization of GBS isolates from a region of the world with limited data on this subject. A major challenge and limitation were the reduced recovery rate of GBS strains at CDC allowing only half of the isolates available for additional testing. The inability to recover GBS from the initial frozen stocks could partly be due to loss of viability during storage and transport, and a number of samples were also heavily contaminated which may have contributed also to reduced recovery.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.