Abstract
The Cβ protein (Bac) of Streptococcus agalactiae (group B streptococcus; GBS) is an IgA binding protein encoded by bac, of which at least 39 sequence types have been described, based on polymorphisms in the repeated region of the cell-wall-spanning domain (‘bac sequence types’). Cβ is usually found in serotype Ib, less commonly in serotype II, and rarely in other serotypes. The aim of this study was to examine the prevalence, variety and distribution, among GBS serotypes and between invasive and superficial isolates, of bac sequence types. A total of 1101 GBS isolates were tested, from 10 countries, with a bac-specific PCR, and amplicons from all 255 (23 %) with positive results were sequenced. Ninety-seven percent (184/190) of serotype Ib and 37 % of serotype II isolates were bac positive. The Cα protein gene (bca) was present in 98 % (251/255), and insertion sequences IS1381 and IS861 in 94 % (239/255), of bac-positive isolates. The authors identified 59 bac sequence types belonging to 19 groups, based on length, from 496 to 946 bp, with up to six sequence variants (a–f) in each group. The median bac sequence length of invasive isolates was significantly shorter than that of superficial isolates overall (640 versus 586 bp; P <0.001) and specifically for serotype Ib (541 versus 676 bp; P <0.001), and invasive isolates were significantly (P <0.001) more likely to have one or more 18 bp deletions relative to the original published bac sequence (X59771). bac sequence typing is a useful addition to the previously described genotyping system, and will help to predict relative virulence among S. agalactiae serotype Ib strains.
- GBS, group B streptococcus
- mge, mobile genetic element
- MS, molecular serotype
- pgp, protein gene profile
-
The GenBank/EMBL/DDBJ accession numbers for the bac sequences determined in this study are AY672137–AY672167.
-
A phylogenetic tree inferred from 59 partial bac sequence types is available in Supplementary Fig. S1 with the online version of this paper.
INTRODUCTION
The C protein α and β antigens are immunodominant components of the surface of Streptococcus agalactiae (group B streptococcus; GBS), the most frequent cause of neonatal sepsis. Both proteins, which are mainly expressed by serotypes Ia, Ib and II, are thought to contribute significantly to virulence. The C protein β antigen (Cβ protein) binds to the Fc portion of human IgA and may be important in bacterial resistance to mucosal immune defence mechanisms (Schalen, 1993; Kreikemeyer & Jerlstrom, 1999; Berner et al., 2002; Mawn et al., 1993). It is structurally related to pneumococcal protein Hic, which binds the complement inhibitor factor H and protects the organism from opsonophagocytosis (Jarva et al., 2004). Levels of expression of bac vary (Nagano et al., 2002), and are reported to be higher among invasive than noninvasive strains of GBS serotypes Ia and Ib (Lachenauer et al., 2002). Bac has two regions with different functions: the N-terminal IgA binding region and the C-terminal Cβ antigen region (Heden et al., 1991), which contains the repeated and non-repeated regions of the cell-wall-spanning domain (Berner et al., 2002). We have previously described 39 different ‘bac sequence types' based on significant sequence polymorphism in the repeated region of the cell-wall-spanning domain (Kong et al., 2002b, 2003; Berner et al., 2002; Sun et al., 2005). Berner et al. (2002) showed a higher, but not statistically significant (P=0.08), rate of large deletions among bac sequences of neonatal invasive isolates than those from colonizing maternal isolates.
Based on PvuII ribotyping and PFGE, bac-positive GBS strains are apparently genetically homogeneous (Dmitriev et al., 2002), despite the considerable sequence heterogeneity within bac. In order to identify the extent of this heterogeneity, we screened 1101 well-characterized GBS isolates using bac-specific PCR (Kong et al., 2002b; Berner et al., 2002; Sun et al., 2005), and sequenced amplicons from all 255 bac-positive isolates.
METHODS
GBS reference strains and clinical isolates.
Our collection of 1101 GBS isolates included 27 reference strains (Kong et al., 2005) and 1074 human isolates, many of which have been previously studied, including 113 from Sydney (Kong et al., 2002a, 2003) and 121 from Melbourne, Australia, 147 from New Zealand (Kong et al., 2002a, 2003), 50 from the USA (Lachenauer et al., 1999), 19 from Canada (Martinez et al., 2000), 83 from the UK (Jones et al., 2003), 211 from Germany (Berner et al., 1999, 2002), 198 from South Korea (Lee et al., 2000) and 132 from Hong Kong.
Genotyping of GBS.
Molecular serotypes (MSs), protein gene profiles (pgps) and mobile genetic elements (mges), which comprise our three-component genotyping system for GBS, were identified as previously described (Kong et al., 2002a, b, 2003), except that, in addition, we used recently published primers for the identification of serotypes II, VII and VIII (Borchardt et al., 2004) and, for testing some isolates, a new multiplex PCR-based reverse line blot hybridization (mPCR/RLB) serotype-identification method (Kong et al., 2005). Previously published primers were used to amplify and sequence the cell-wall-spanning domain of bac (Kong et al., 2002b; Berner et al., 2002; Sun et al., 2005).
Sequencing, sequence analysis and phylogenetic trees.
WebANGIS () in ANGIS (Australian National Genomic Information Service) provided all programs used in the study; in particular, for sequence file management (WebFM), two-sequence comparison (bestfit in the Comparison program group), multiple sequence alignments (pileup and pretty in the multiple sequence analysis program group) and phylogenetic tree production (ednadist and ekitsch in the Evolutionary analysis program group).
Nucleotide sequence accession numbers.
Sequences used or referred to in this study, which have been previously reported by others (Kong et al., 2002b, 2003; Berner et al., 2002), and new bac sequences found in this study (which have been deposited in GenBank with the accession numbers AY672137–AY672167) are shown in Table 1⇓.
bac sequence types by isolate category, including gene deletions and insertions
Statistical analysis.
The analyses were performed using SSPS (Statistical Package for Social Sciences for Windows) version 11. Median base-pair lengths for the repeated region of the cell-wall-spanning domain of bac were compared, by site of isolation and country, using the Mann–Whitney U (two categories) and Kruskal–Wallace (more than two categories) tests. Proportions were compared using the chi square or Fisher's exact tests, where appropriate.
RESULTS AND DISCUSSION
Of 1101 GBS isolates studied, 255 (23 %) contained bac, including seven reference strains, 150 from superficial sites (including urine, and vaginal, wound and other swabs), 87 from normally sterile sites and 11 of unknown origin. Genotyping and bac sequence typing results for these isolates are shown in Tables 1 and 2⇑⇓.
Distribution of bac-positive genotypes, by isolate category
MS identification
The MSs of all isolates were identical to the conventional serotypes, when the latter were available. Of the 255 bac-positive isolates, 184 (72 %) belonged to MS Ib, 42 (16 %) to MS II, 13 (5 %) to MS Ia, eight (3 %) to MS V, and four or fewer isolates to each of the other MSs. Of the total 1101 GBS isolates, these represented 97 % (184/190), 37 % (42/114), 6 % (13/212) and 6 % (8/136) of the MS Ib, II, Ia and V isolates, respectively. Thus, Cβ protein (and bac) is uncommon in GBS serotypes other than Ib and II.
pgps
Most bac-positive isolates (250/255; 98 %) also had bca (Cα protein gene), i.e. their pgps were ‘AB’. Of the others, four had the Rib antigen gene rib (pgps were ‘RB’) and one reference strain had both bca and rib (pgp was ‘ARB’).
mges
All bac-positive isolates contained IS1381, and most (239/255; 94 %) also contained IS861, usually alone or with additional mges (34/255; 13 %) (Table 2⇑).
Polymorphisms in the repeated region of the bac cell-wall-spanning domain and proposed nomenclature
Using bac PCR/sequencing primers, all bac PCR products were sequenced, with satisfactory results (Table 1⇑). Sequence analysis showed genetic heterogeneity, including variable numbers of small repetitive genetic elements of 18 bp within the amplified fragment, as reported previously (Berner et al., 2002). Insertions or deletions were defined by comparison with the original published bac sequence (GenBank accession no. X59771; Table 1⇑). The phylogenetic tree derived from the bac cell-wall-spanning domain sequences (Supplementary Fig. S1) showed that closely related sequence types were often of different lengths, because of insertions and/or deletions of repetitive units, suggesting common parental strains (Berner et al., 2002). It has been shown previously that the insertion or deletion of genes or gene fragments (indels) is responsible for most genetic heterogeneity between strains (Gravekamp et al., 1998; Gupta & Griffiths, 2002). Polymorphism in the repeated region of the bac cell-wall-spanning domain appears to be mainly due to 18 bp repetitive unit indels.
Bac sequence types have previously been classified arbitrarily by letters and/or numbers (Kong et al., 2002b, 2003; Berner et al., 2002) because phylogenetic relationships, based on sequence heterogeneity, do not necessarily correlate with length. To make it easier to categorize and add new sequence types, we used amplicon length (in bp) and frequency of occurrence of each variant, within sequence length groups (designated a, b, c, etc. in descending order), as the basis of a proposed new bac sequence type nomenclature. Based on this system, there were 59 sequence types, represented by 19 groups ranging in length from 496 to 946 bp; within each sequence length group, there were up to six (a–f) sequence variants. Eleven (19 %) sequence types were each represented by five or more isolates and together accounted for 73 % (186/255) of the total (Table 1⇑). The number of heterogeneity sites (compared with the predominant variant in each group) varied from one to 43. Thirty nine of the 59 sequence types have been described previously (Kong et al., 2002b, 2003; Sun et al., 2005; Berner et al., 2002), using two different nomenclatures.
Relationship between repeated region of the bac cell-wall-spanning domain sequence length and virulence
It has been shown previously that deletion of repeats in the Cα protein enhances the pathogenicity of GBS in immune mice and that Cα gene (bca) sequences of colonizing strains are generally longer, because of higher numbers of repetitive units, than those of invasive strains. Our results suggest that this inverse relationship, between the number of repetitive units and virulence, is also true for bac. Overall, the median length of the repeated region of the bac cell-wall-spanning domain sequences for superficial isolates (640 bp) was significantly greater than for invasive isolates (586 bp; P <0.001). This was due to the high proportion (72 %) of serotype Ib, the only one for which the difference in median sequence lengths between superficial and invasive isolates was significant (676 versus 541 bp, respectively; P <0.001). However, numbers of other serotypes were small.
Superficial isolates were more evenly distributed than invasive isolates between different sequence length groups, as illustrated in Fig. 1⇓. In shorter sequence length groups (450–549 and 550–649 bp), proportions of invasive and superficial isolates were generally similar, but invasive isolates were significantly (P <0.001) less likely to be found among the longer sequence groups (⩾650 bp). Isolates with shorter sequence lengths, especially those less than 622 bp, were significantly more likely to demonstrate 18 bp deletions in the bac repeated region of the cell-wall-spanning domain (P <0.001), relative to the original bac sequence X59771, and to be invasive (P <0.001; Table 1⇑). Whether deletion of bac repetitive units in colonizing strains increases virulence and the likelihood of invasion by enhancing bac expression and/or IgA binding by Cβ locally (Gravekamp et al., 1998), or is due to a survival advantage once invasion has occurred, deserves further study. However, our results demonstrate an association between invasiveness and deletion of repetitive units in bac, as previously shown for bca (Gravekamp et al., 1998).
Numbers of isolates by bac repeated region of cell-wall-spanning domain sequence length groups and isolate category [superficial (white bars) or invasive (black bars)], and percentage of invasive isolates (solid line) in each sequence length group (with 95 % confidence intervals).
As well as median lengths, we found associations between some individual bac sequence types and whether isolates were superficial or invasive. For example, the short sequence type 532a was found significantly more commonly among invasive isolates (34 versus 17 %, P =0.002), and longer sequence types 748a (1 versus 10 %, P =0.009) and 820a (3 versus 12 %, P =0.03) significantly more commonly among colonizing isolates (Table 1⇑). It is not known whether factors other than length contribute to this difference, since we did not measure gene expression, which may also affect virulence.
Geographic distribution
There were geographic differences in the distribution of bac sequence types among both invasive and superficial isolates (Table 3⇓). Specifically, the median lengths of bac sequences among both superficial and invasive isolates from South Korea were significantly longer (P <0.001 and P =0.007, respectively) than those of the corresponding groups from other countries. The significance of this observation is uncertain, but may reflect an unusual distribution of serotypes among colonized South Korean women, characterized by a preponderance of serotype Ib (Lee et al., 2000; Uh et al., 1997).
Sequence lengths of repeated region of bac cell-wall-spanning domain by country and isolate category
For both superficial and invasive isolates, the median length of the repeated region of the bac cell-wall-spanning domainsequences for South Korean isolates was significantly greater than those of isolates from all other countries combined (superficial, P <0.001; invasive, P=0.007); the latter were not significantly different from each other (P >0.1).
Conclusion
This is the largest collection of bac-positive GBS isolates to have been studied using direct sequencing of the repeated region of the bac cell-wall-spanning domain and our genotyping system (Kong et al., 2002b, 2003). The bac-positive isolates were quite homogeneous: most belonged to MS Ib and, as well as bac, most contained bca (protein Bca or Cα) and mges IS1381 and IS861 (Dmitriev et al., 2002). They represented 23 % of all GBS isolates, with similar proportions in superficial and invasive groups. The median bac sequence length of the repeated region of the cell-wall-spanning domain of invasive isolates was significantly less than that of superficial isolates, suggesting an association between 18 bp repetitive unit deletions and increased virulence. However, we did not measure bac expression, which may affect virulence independently of or secondary to repetitive unit deletions (Nagano et al., 2002). We have proposed a new standardized nomenclature for bac sequence types, which allows the addition of new sequence types in a logical sequence.
Acknowledgments
We sincerely thank the following colleagues for allowing us to study their isolates: Drs Lawrence Paoletti and Catherine Lachenauer, Channing Laboratory, Boston, MA, USA; Dr Diana Martin, Streptococcus Reference Laboratory, Environmental Science Research Pty, Wellington, New Zealand; Professor Johan Maeland, Department of Microbiology, School of Medicine, Norwegian University of Science and Technology, Trondheim, Norway; Dr Nicola Jones, Nuffield Department of Clinical Laboratory Sciences, Institute for Molecular Medicine, John Radcliffe Hospital, Oxford, UK; Professor Yunsop Chong and Dr Kyungwon Lee, Research Institute of Bacterial Resistance, Yonsei University College of Medicine, Seoul, Korea; Catherine Satzke and Professor Roy Robins-Browne, Department of Microbiology and Immunology, University of Melbourne, Australia; Dr Margaret Ip, Department of Microbiology, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong.