Our study is the first prevalence report of detection of some of the GBS virulence genes and antibiotic resistance genes in pregnant women in Zimbabwe. The 10.2% prevalence of maternal GBS colonization reported is similar to the 12.2% and 13.4% reported in Ethiopia (17) and Saudi Arabia (31), respectively. However, its lower to when compared to the 2010 study in Zimbabwe (3). The possible reasons for these discrepancies are differences in the time of sampling, specimen, sample sizes, geography, detection techniques and gestational ages among other factors.
The atr gene and the Mobile Genetic Element
The sensitivity of the atr gene amplification, in this study was 100% (43/43) (Table 4) which is similar to what is reported in other studies done by Gavino and Wang (32), de-Paris et al., (28) and Alfa et al. (33), where sensitivity was 95.8%, 100% and 90.5% respectively. This high sensitivity may be attributed to the use of selective and enriched media prior to performing PCR. Since this PCR test has shown to be reliable and robust, future studies can focus on rapidly testing pregnant women who present in labor with unknown GBS status. This test was used as a GBS confirmatory identification test. The Mobile Genetic Element IS1548 was present in only 9.3% (4/43) (Table 4). This finding is contrary to an Iranian study which showed that the gene is more prevalent in human than in bovine (77% vs 5%) GBS isolates (7). The origin of the isolates could be a possible explanation for such a difference.
Virulence Genes
Nine distinct virulence gene profiles were observed but 30/43 strains (69.8%) belonged to hly-scpB-bca-rib 37.2% (16/43), hly-scpB-bca 18.6% (8/43) or hly-scpB-rib 14.0% (6/43) profiles (Table 3). The PCR assay for detection of VFs revealed that high percentage of the GBS isolates were positive for hly 97.8% (42/43); scpB 90.1% (39/43) and bca 86.0% (37/43) (Table 4). The high prevalence of these genes in GBS has been previously reported (7,34–36).
A Zimbabwean study reported a 46.3% of Rib (rib) protein marker as well as 19.8% and 15.7% of Cα (bca) and Cβ (bac) protein markers, respectively (9). The low frequency obtained for the β protein marker agrees with the 11.6% for bac observed in this study and this can be attributed to the reported low distribution of the Ib and II CPA types (11.6% and 8.3%, respectively) in the Zimbabwean GBS population (9). However, contrary to their findings, the present study reports higher incidences of rib (69.8%) and bca (86.0%). This discrepancy could be attributed to differences in CPA types for GBS isolates in these two studies among other factors. Comparisons with this 2008 study, highlighted the importance of investigating the CPA of GBS isolates because a relationship was observed between protein markers and CPA types (9).
Lower frequencies for genes rib and bca, are reported by Manning et al., (2006) and Persson et al., (2008) with 28% and 43% for rib and 29% and 14% for bca in the USA (37) and Sweden (38) respectively. However, our higher frequencies agree with the 76.1% for rib and 88.6% for bca reported in Argentina (39). The bac gene in the present study was found at a similar frequency to the aforementioned studies in USA and Sweden: 11.6% for bac, compared with 20% and 12% for bac, in the USA and Sweden respectively. The study in Argentina observed a higher frequency of 52.3% for bac. A 2017 study in Egypt also observed high frequencies for scpB in 100% and rib in 79.2% and a lower frequency for bac (35.8%) in GBS isolates. However, none of this study’s isolates possessed the bca gene (27). Our comparison with studies from these four countries USA, Sweden, Argentina and Egypt shows that any discrepancies are possibly a result of the geographical locations among other factors.
The results also showed that the majority of the isolates had more than three virulence genes. This high prevalence of various VFs in S. agalactiae isolates from the vaginal canal of pregnant women could lead to the development of serious maternal and neonatal infections (40). However, this study was not able to follow up on the outcomes of both the pregnant mothers and their newborns. The high incidences of VFs except for bac, also suggest that GBS vaccines containing the proteins ScpB, α protein (41), Rib (9) and hyaluronate lyase (42) could potentially be effective against our population of pregnant women in Zimbabwe.
The present study observed that all the isolates 100% (n=5) carrying the bac gene also carry the unrelated bca gene, whilst 86.5% (n=32) of the bca carrying isolates did not also have the bac gene (see Additional file 2: Table S2). This finding agrees with the report that, isolates that express β tend to also express the unrelated α protein, while the α protein is often expressed on its own in GBS isolates (43,44). The majority of the Spearman rank analysis were not statistically significant and with some showing weak associations. However, a significant negative correlation (see Additional file 2: Table S4) was detected between hly and IS1548 as well as between bac and rib genes (P <0.01). This negative association is most likely because these genes are not genetically linked.
Penicillin G and Ampicillin Resistance
A high resistance to penicillin G 69.8% (30/43) and ampicillin 58.1% (25/43) (Table 2) was observed and this is contrary to most studies which report a uniform susceptibility of 100% to all GBS isolates in pregnant women (12,16,31,45,46). However, our results agree with recent findings in Ethiopia were a resistance of 77.3% to penicillin G was reported in pregnant women (17). Other studies have reported resistance to both penicillin G and ampicillin of 10.2% and 9.2% as well as 19.5% and 14.6% respectively (18,19). The high resistance reported may be due to the wide and non-prescription use of these two drugs in our study area (45,47). It may also be as a result of differences in laboratory procedures or bacterial strains. We recommend that antibiotic susceptibility testing should be performed if penicillin G or ampicillin therapy is needed in the prevention of neonatal GBS infection. Further studies could focus on identifying the genes responsible for the observed resistance.
Tetracyline Resistance
The high resistance for GBS to tetracycline 97.7% (42/43) (Table 2) can be attributed to a high presence of the tetM gene 97.6% (41/42) (see Additional file 1: Fig 13). This high tetM gene presence is as a result of the ubiquitous presence of tet genes in pathogens, opportunistic pathogens and members of the normal flora (48). The high resistance to tetracycline is also because the antibiotic is a relatively cheap, extensively used prophylaxis in the therapy of animal and human infections. It is known that bacterial strains tend to be resistant to frequently used antibiotics (47). A low presence rate of the tetO 2.4% (1/42) (Table 4) indicates that this gene is not common in GBS isolates that are resistant to tetracycline. These results agree with studies done in Canada (49) and Nigeria (50). In Kuwait, they also reported similar results with 89.5% tetracycline-resistant isolates carrying 94.5% tetM and 3.9 % tetO genes (51).
Erythromycin and Clindamycin Resistance
The current study results showed a 30.2% (13/43) and 55.8% (24/43) resistance to erythromycin and clindamycin respectively (Table 2). Such resistance decreases the options of prophylaxis in penicillin allergic women (52). This combined with the high resistance of GBS to penicillin and ampicillin, we recommend that AST should be performed if clindamycin or erythromycin therapy is needed in the prevention of neonatal GBS infection. Future studies should focus on finding suitable alternative antibiotics for GBS chemoprophylaxis in pregnant women. PCR detected 34.5% (10/29) ermB, 10.3% (3/29) ermTR and 3.4% (1/29) mefA from the intermediate and resistant erythromycin GBS (Table 4). The prevalence of the ermB determinant shows that GBS commonly use target methylation as the mechanism of macrolide resistance.
Studies done in Italy (53), South Africa (12), USA (14), Iran (13) and France (23) reported similar findings, with most of the studies also observing that the ermB gene is more prevalent in distribution than the ermTR gene among GBS strains. This study also confirmed findings by Poyart et al., (2003) that the mefA gene is rare among GBS isolates, thus efflux pumps mediated by this gene are not a common mechanism of macrolide resistance (23). Contrary to Bolukaoto et al., (2015) this current study did not find any linB genes 0% (0/35) in any of the clindamycin resistant and intermediate strains (12). In such cases, where multiple independent resistance genes can cause resistance, the observed phenotypic resistance can be attributed to any of the other known genes or genes that are yet to be discovered. This however limits the usefulness of such diagnostic tests.
It was interesting to note that 63.6% (7/11) GBS which were resistant to both erythromycin and clindamycin had the ermB gene (see Additional file 2: Table S2 and S3) and that one GBS strain which was resistant to erythromycin and susceptible to clindamycin carried the ermTR gene. Such observations are similar to other studies (14,54,55).
Genetic Linkage in Macrolide/Tetracycline Resistance Determinant
It was also observed that 100% (n=10) isolates that carried the ermB gene also carried the tetM gene. The association of erythromycin and tetracycline resistance may be due to the conjugative transposon Tn1545, which encodes erythromycin resistance via the ermB gene and tetracycline resistance via the tetM gene (56–58). This association was supported by our analysis which showed a positive correlation (see Additional file 2: Table S4) between the two genes, but this lacked statistical significance according to the Spearman Rank test. Contrary to an Egyptian study (59), we report that there is no evidence suggesting genetic linkage of tetO with ermB, ermTR or mefA in our GBS population.
Unexpressed Resistant Genotypes
An unexpected observation in our study was found in the following three (7.0%) isolates. Isolate 127 which was tetracycline sensitive, and yet the tetM gene was detected. Isolates 243 and 322, were both erythromycin sensitive however, at least one gene ermTR or mefA was found, respectively (see Additional file 2: Table S2 and S3). Similar findings were reported in GBS and other streptococcal species (58,60–62). A confirmatory test to check Minimum Inhibitory Concentration (MIC) levels of these three isolates is recommended. However, based on the disk diffusion test, this observation suggests that the Kirby Bauer disk diffusion method is inadequate for detection of resistance. Although the reason(s) behind this lack of gene expression still has to be determined, some possible explanations include gene mutations, low expression levels of the gene and the possibility of a weak, distant or absent promoter (58). This study therefore supports the idea that the resistance genotype does not always accurately predict phenotypic resistance.