Macrorestriction analysis of Streptococcus agalactiae (group B Streptococcus) isolates from Malaysia

Group B streptococci (GBS) are an important cause of morbidity and mortality among neonates. In adults, these bacteria are associated with a variety of septic diseases but are also frequent colonizers of the gastrointestinal and genital tracts. Approximately 30 % of women of child-bearing age carry GBS in the rectovaginal compartments (Park et al., 2001). At birth, about 50 % of infants born to colonized mothers will also become colonized on their mucosal surfaces and on skin (Berner et al., 2002). The mechanism of transmission from mother to offspring is still not very clear. A study by Berner et al. (2002) hypothesized that the transmission of certain strains from mother to infant depends on a genetic polymorphism in the bacterial cell wall anchoring domain of the C protein ß-antigen with its ability to bind human IgA on the mucosal surface.

GBS are traditionally serotyped based on their capsular polysaccharide and protein antigens. Serotype V has been reported to be the most common serotype recovered from non-pregnant adults with invasive GBS disease (Blumberg et al., 1996) and serotype III has been reported to be the most common serotype in neonatal invasive disease, accounting for 3050 % of all cases (Berner et al., 2001). Although serotyping is of great importance in epidemiological studies of GBS disease, it is not very discriminative and requires specific reagents that may not be available in every laboratory. Hence molecular approaches which are reproducible and highly discriminatory for differentiating individual strains of bacterial pathogens are often applied to complement serotyping. These approaches include multilocus enzyme analysis (Quentin et al., 1995), ribotyping (Huet et al., 1993; Chatellier et al., 1996), random amplified polymorphic DNA analysis (Chatellier et al., 1997; Limansky et al., 1998; Martinez et al., 2000) and PFGE (Fasola et al., 1993; Gordillo et al., 1993; Rolland et al., 1999; Le Thomas-Bories et al., 2001; Benson et al., 2002; Moyo et al., 2002), analysis of insertion sequences (Rolland et al., 1999; Dmitriev et al., 2004), analysis of mobile genetic elements (Kong et al., 2003) and, more recently, multilocus sequence typing (Jones et al., 2003). Of these, PFGE is the most versatile and powerful technique for the study of chromosomal relatedness among bacterial strains as it is reproducible and very discriminative. Many researchers have demonstrated the discriminatory ability and reliability of PFGE in further differentiation of GBS strains of the same serotype (Rolland et al., 1999; Le Thomas-Bories et al., 2001; Benson et al., 2002; Moyo et al., 2002).

In this study, we examined a collection of 123 clinical strains of GBS by PFGE with the objective of assessing the genetic diversity of these strains, to determine the clonal relationship of GBS strains isolated from mothers and their infants and also the genetic relatedness of the strains obtained from different body sites from GBS-positive women. To the best of our knowledge, this is the first report on the molecular characterization of GBS from women and neonates in Malaysia.

Bacterial strains.
Over 3 months in 2001, all women admitted to the labour ward of the University Hospital, Kuala Lumpur, and their newborns subsequently delivered were sampled for GBS carriage. Vaginal, urethral and rectal swabs were taken from the women in labour, after obtaining informed consent. The specimens were placed in transport medium for transportation to the laboratory on the day of collection. External ear, mouth and umbilical swabs were taken from newborns after bathing but before transfer to the newborn nursery.

Swabs were incubated overnight in ToddHewitt broth (Oxoid) containing nalidixic acid and subcultured onto blood agar plates for the isolation of GBS. Isolates were identified by colonial morphology, Gram stain reaction, the CAMP test (Christie et al., 1994) and Streptex latex agglutination (Murex Biotech). Isolates were kept at 80 °C for up to 9 months until used for PFGE analysis.

DNA preparation and PFGE.
Chromosomal DNA from each strain was prepared by the modified method as previously described (Thong et al., 2002). Slices of DNA-containing agarose plugs were digested overnight with 10 units SmaI (Promega) at 25 °C, then electrophoresed on a CHEF DRII/III system (Bio-Rad Laboratories) for 24 h at 6 V cm¹, with ramped pulsed times of 140 s in a 1.0 % (w/v) 0.5x TBE gel. A lambda DNA concatemer PFGE marker (New England BioLabs) was used as a DNA size standard. The gel was stained with ethidium bromide (1 µg ml¹; Sigma) for 15 min, destained in distilled water for 15 min, and photographed under UV illumination.

Data analysis.
DNA fragment patterns were assessed visually, and distinct profiles were assigned an arbitrary PFGE profile (PFP). Dice coefficients of similarity (F) were calculated to compare the macrorestriction patterns. This coefficient, F, expresses the proportion of shared DNA fragments in two isolates and was calculated by the formula F = 2n_XY/(n_X + n_Y), where n_X is the total number of DNA fragments from isolate X, n_Y is the total number of DNA fragments from isolate Y, and n_XY is the number of DNA fragments that were identical in the two isolates. By this assessment F = 1.0 indicates complete identity and F = 0 indicates complete dissimilarity. Clustering was based on the unweighted pair group average method (UPGMA) and was performed with GelCompar II version 2.0 (Applied Maths, Kortrijk, Belgium).

PFGE analysis of SmaI-digested chromosomal DNA of GBS strains gave reproducible and discernible profiles. Among the 123 strains studied, 48 distinct PFPs were identified. The PFPs consisted of 819 DNA fragments each, with F values of 0.151.00, indicating wide genetic variation in these strains.

Representative SmaI profiles are shown in Fig. 1. A schematic representation of all the PFPs and the deduced genetic relationship of 48 PFPs are depicted in Fig. 2. Five distinctive clusters were identified, with each cluster consisting of different yet closely related PFPs. The most common profile obtained was S1, as seen in 37.4 % of all strains studied.

(77K): Fig. 1. Representative SmaI PFGE profiles of Malaysian GBS strains. Lanes 1, 11, 21, 31 and 39, lambda DNA concatemer PFGE marker; lanes 3 and 25, DNA degraded; lanes 4, 6-7, 10, 12-13, 26, 29-30, 34-35, SmaI PFGE profile S1 from isolates GBS 5, 7-8, 12-14, 36, 44-45, 50-51; lane 2, profile S2 (GBS 1); lane 5, profile S4 (GBS 5); lanes 8-9, profile S5 (GBS 9-10); lane 14, profile S6 (GBS 15); lanes 15 and 20, profile S9 (GBS 25); lanes 16 and 22, profile S7 (GBS 16 and 26); lanes 17-18, profile S11 (GBS 19-20); lane 19, profile S8 (GBS 21); lane 23, profile S10 (GBS 27); lane 24, profile S14 (GBS 30); lane 27, profile S16 (GBS 38); lane 28, profile S18 (GBS 40); lane 32, profile S20 (GBS 47); lane 33, profile S21 (GBS 48); lane 36, profile S23 (GBS 53); lane 37, profile S24 (GBS 59); lane 38, profile S25 (GBS 61); lane 40, profile S46 (GBS 122).

(24K): Fig. 2. Dendrogram showing the cluster analysis of the different SmaI PFGE profiles from 123 clinical isolates of GBS strains, generated by the GelCompar program using the UPGMA method, based on the matrix of F values.

Distribution of PFPs in relation to the source of GBS strains
Motherinfant pairs. Twelve sets of paired motherinfant strains were analysed. Among these, 10 different SmaI profiles were seen (Table 1; Fig. 1, lanes 6 and 7, 8 and 9, 12 and 13). Only two motherinfant pairs showed dissimilar profiles (S1 and S4 and S6 and S7). Most of the neonatal isolates were clustered in one group (Fig. 2). The indistinguishable profiles between mother and infant in each pair reinforced earlier belief that most newborns acquire GBS from contamination by the maternal vaginal flora during the birth process.

Table 1. SmaI PFGE profiles of motherinfant GBS pairs All the isolates were recovered from high vaginal swabs.

Genetic diversity of strains from different body sites within an individual. Twenty-three specimens (11 vaginal, 10 urethral and 2 rectal) from 11 women were analysed to determine the genetic variability of strains isolated from different body sites within an individual (Table 2). In nine women who gave both vaginal and urethral swabs, the vaginal isolates were indistinguishable from the urethral isolates, indicating autoinoculation as the likely mechanism of spread in the individual colonized at more than one body site (Fig. 3). In the tenth woman, the vaginal isolate differed from the rectal isolate by five bands, and in the eleventh woman from whom three isolates were obtained from the vagina, urethra and rectum separately, the vaginal and rectal isolates were indistinguishable while the urethral isolate was different (Table 2). These results suggest that some women may be colonized by multiple clones of GBS.

Table 2. SmaI PFGE profiles of isolates from different body sites within an individual (GBS-positive mother)

(120K): Fig. 3. Representative SmaI PFGE profiles of paired GBS strains recovered from different sites from individuals. Lanes: 1, lambda DNA concatemer PFGE marker; 2 and 3, paired vaginalurethral strains from patient 7 (S1); 4 and 5, paired vaginalurethral strains from patient 9 (S25); 6 and 7, paired vaginalurethral strains from patient 6 (S23); 8 and 9, paired vaginalurethral strains from patient 5 (S22).

GBS from neonates of culture-negative mothers. GBS was isolated from 17 neonates whose mothers were culture-negative for that organism (Table 3). These 17 isolates yielded 9 different PFPs with S1 as the most common profile (47.1 %). In these infants, the possibility of an environmental source for their GBS cannot be excluded although it is also likely that the negative maternal cultures were due to technical errors.

Table 3. SmaI PFGE profiles of GBS isolates from neonates of culture-negative mothers

Genetic diversity of strains from pregnant women. From 59 clinical specimens taken from pregnant women whose newborn infants were culture-negative for GBS, 30 profiles were obtained (F = 0.201.0) with profile S1 again the most common PFP (41 %) (Table 4). Except for PFPs S10 and S20S24 (which were also seen in other individuals Table 2), most of the profiles were unique and were represented by one isolate each. This shows the wide genetic diversity of Malaysian GBS strains.

Table 4. SmaI PFGE profiles of strains from pregnant women with culture-negative infants

Molecular typing of GBS has distinct advantages over the traditional serotyping used to identify sources of infection in early and late-onset neonatal disease and for the investigation of nosocomial outbreaks. Restriction endonuclease analysis of chromosomal DNA (REA) and ribotyping have been shown to be reliable in the demonstration of epidemiological relationships and are able to differentiate among strains of the same serotypes (Blumberg et al., 1992). Harrison et al. (1995) used REA to distinguish relapse from reinfection in recurrent GBS disease. Limansky et al. (1998) also showed that GBS reactivation and reinfection could be differentiated by RAPD. Gordillo et al. (1993) found that PFGE patterns obtained with SmaI digestion of GBS strains were more easily discerned and interpreted than patterns generated by conventional electrophoresis with HindIII or EcoRI plus BglII. Similarly, Fasola et al. (1993) and Ferrieri et al. (1997) obtained more SmaI PFGE patterns than REA with HindIII or HindIII followed by EcoRI and conventional agarose gel electrophoresis. Many researchers who used PFGE analysis of GBS strains have documented considerable genetic diversity among strains of similar serotypes (Fasola et al., 1993; Rolland et al., 1999; Le Thomas-Bories et al., 2001; Moyo et al., 2002).

In this study, we have examined GBS strains isolated from mothers and newborn infants by using PFGE analysis with SmaI digestion. The results illustrated the wide genetic diversity in strains of GBS in the female genital tract, motherinfant transmission during the birth process, the co-existence of different clones of GBS in the same individual and the clonal spread of strains to different body sites in the same individual.

At our hospital, GBS from the urogenital tract had previously been studied by serotyping with the capillary precipitin test, and serotype 111 was shown to be the predominant type (50.9 %) among both infants and adults, with identical serotypes found in mother and infant pairs (Ngeow & Puthucheary, 1987). In the present study on carrier strains, there was also a predominant profile (S1) occurring in 46/123 (37.4 %) strains. Although pathogenic strains of GBS may be genetically different from carrier strains, it is expected that PFGE patterns of isolates from GBS-infected patients would similarly show a much wider genetic diversity than what was previously determined by serotyping.

The higher discriminatory power of PFGE analysis identified a few non-identical PFGE patterns among motherinfant pairs, thus raising the possibility of environmental contamination of the infant by GBS occurring soon after birth. The discrepancy in PFGE patterns found in motherinfant pairs can also be the result of transmission of only one of several clones of GBS colonizing the mother or, more likely, the laboratory selection (from each culture examined) of only one of several clones of GBS present in both mother and infant. These possibilities can be further examined by the selection of more colonies from both maternal and neonatal cultures for genotyping.

Overall, our data confirmed the other researchers reports of the heterogeneous nature of individual GBS serotypes by PFGE analysis (Gordillo et al., 1993; Ellis et al., 1996). Overall, the 123 Malaysian strains of GBS were defined by 48 profiles. Thirty PFPs were unique, with each represented by a single isolate. This demonstrates the high discriminatory ability of PFGE in differentiating very closely related strains. The existence of different PFGE types in the same individual may be the result of mutation or co-colonization with strains from different sources, e.g. from different sex partners. This concurrent carriage of different genotypes has to be borne in mind when genotypic information is used to establish proof of epidemiological linkage in disease transmission. It may be necessary for the laboratory to test several isolates from a carrier or case before the individual can be excluded as the source of an outbreak of infection on the basis of a difference in genotype pattern.

The changing epidemiology of GBS colonization reported in recent years (Hickman et al., 2001) underlines the importance of continued GBS surveillance for the design of GBS vaccines. The PFGE analysis we used fulfilled the criteria of a useful typing method in that it is reproducible, has sufficient discriminatory power, is able to type all strains and is practical for the hospital laboratory that has to carry out nosocomial infection investigations. Compared to the newer approach of multilocus sequencing typing (Jones et al., 2003), PFGE is relatively cheaper and more accessible to most laboratories. It should also be an appropriate typing method for the study of the association of genetic types with disease manifestation, development and response to treatment.

The authors would like to thank Drs Tetty Aman Nasution and Yi Yi for technical assistance. Funding for the study comes from the University of Malaya F-Vote and IRPA grant 06-03-02-01119, 06-02-03-0750 and 06-03-02-1007 from the Ministry of Science, Technology and Environment, Malaysia.