Association between low concentrations of antibodies to protein α and Rib and invasive neonatal group B streptococcal infection

Abstract

Background

Infection with group B streptococci (GBS) is a serious neonatal disease. The GBS cell surface proteins α and Rib elicit protective immunity in animal models and have been suggested as potential antigens in a vaccine against human GBS disease.

Aims

To test the hypothesis that transplacentally transferred maternal antibodies to GBS proteins contribute to the protection of the neonate from GBS infection.

Methods

Thirty neonates with invasive infection were included in a case‐control study. IgG antibody concentrations were measured in sera from these neonates, their mothers, and from 60 non‐infected controls, neonates as well as mothers.

Results

A clear association was found between concentrations of antibody to proteins α and Rib in neonatal and maternal sera, indicating that transplacental transfer had occurred. Moreover, low concentrations of antibodies to α and Rib in neonatal sera were associated with invasive GBS infection caused by strains expressing the Rib protein. The odds ratio was 0.0007 (95% confidence interval 0.000 to 0.54) for antibodies to α and 0.002 (95% confidence interval 0.000 to 0.57) for antibodies to Rib.

Conclusion

These findings support the notion that antibodies to GBS surface proteins contribute to the protection against neonatal infection.

Infection with group B streptococci (Streptococcus agalactiae; GBS) is an important cause of neonatal morbidity and mortality in many countries including Sweden.1,2,3,4,5 As many as a quarter of pregnant women may be colonised with GBS during late pregnancy,6,7 with a concomitant risk of about 1% for their children to develop invasive GBS infection during the neonatal period.8

Animal studies have shown that vaccine induced antibodies to the polysaccharide capsule8 and to proteins exposed on the bacterial surface of GBS confer protective immunity.9,10,11,12 Our work focused on protective GBS surface proteins, particularly α and Rib, which are expressed on the bacterial surface by most GBS strains that cause human invasive neonatal infection.12,13,14 Immunisation of mice with highly purified preparations of α and Rib efficiently protects against lethal infection with strains expressing the corresponding protein.13,15 The α and Rib proteins have been extensively characterised and belong to a family of streptococcal proteins with extremely repetitive sequence.16,17,18 Animal antisera raised against the purified α and Rib proteins show little or no cross reactivity, although the two proteins exhibit extensive amino acid residue identity.12,13

Antibodies to type‐specific capsular polysaccharide are present in human sera, and low concentrations of such antibodies have been associated with neonatal GBS infection.19,20,21,22 Antibodies to GBS cell surface proteins also occur naturally in human sera,9,23 and it has been suggested that low serum concentrations of such antibodies may favour occurrence of neonatal GBS infection.24,25 However, studies comparing serum concentrations of antibodies to GBS proteins in infected neonates and their mothers with those in non‐infected neonates are scarce.26,27

We hypothesised that transplacentally transferred antibodies to the α and Rib proteins contribute to the protection of the neonate from GBS disease. Our objectives in this study were to compare the concentrations of antibodies to α and Rib in sera from infected and non‐infected neonates and their respective mothers.

Subjects, materials, and methods

Study design

Cases were recruited from the neonatal wards at nine hospitals in the southern part of Sweden, covering a population of 1.9 million. During the study period, October 1995 to December 1998, there were about 20 000 live births annually. In Sweden, almost all deliveries occur in hospital. Neonatal care is available for all newborns, and seven of the nine neonatal wards in the study area offer intensive care with respiratory support. General screening by culture for GBS in pregnant women is not carried out. A risk based approach is used for prevention of GBS disease. The recruitment and dimension of the study were based on the assumption of a prevalence of GBS carriage in late pregnancy of 25%.6,7 Control subjects were recruited from the neonatal ward at Lund University Hospital, the regional hospital of southern Sweden. Sera from female blood donors were used as reference representing non‐pregnant women. Analyses were performed in two steps. Firstly, comparisons were made between all cases and controls. In the next step, the cases were divided in two groups according to expression of protein α or Rib by the infecting strain and compared with the control group.

Cases

Neonates with culture positive invasive GBS infection (septicaemia and/or meningitis) within eight days of parturition and their mothers were defined as cases. Sera were collected from both neonates and mothers. The gestational age, birth weight, and sex of the neonates and the age of the mothers were recorded (table 1). The date of collection of neonatal and maternal sera was recorded; for two of the maternal sera this information was missing.

Table 1 Background characteristics of cases with invasive group B streptococcal infection and non‐infected controls.

Characteristic	Cases (n = 30)*		Controls (n = 60)†
	Median	Range	Median	Range
Gestational age (weeks)	39	26–42	34	24–43
Birth weight (g)	3320	940–4800	2165	465–4450
Mother's age (years)	28	21–40	30	15–41

*20 male and 10 female.

†33 male and 27 female.

Controls

Neonates with a non‐infectious diagnosis such as prematurity or transient tachypnoea and their mothers were recruited as controls. Inclusion criteria were no clinical signs of infection, negative blood cultures, and concentration of C reactive protein less than 11 mg/l in the neonates. Furthermore, the mothers had to show no signs of infection or have been treated with antibiotics before or during delivery. The gestational age, birth weight, and sex of the neonates and the age of the mothers were recorded (table 1).

For comparisons of antibody concentrations in pregnant and non‐pregnant women, antibodies to the α and Rib proteins were also measured in sera from 100 female blood donors of about 30 years of age, randomly sampled from blood donors at the Lund University Hospital in 1992.

GBS isolates

GBS strains isolated from blood or cerebrospinal fluid of the infected neonates were serotyped at the Clinical Microbiology Laboratory, Lund University Hospital or Statens Serum Institut, Copenhagen, Denmark. Cell surface expression of proteins was analysed with rabbit antisera specific to α and Rib.12 Rabbit antiserum specific to Rib recognises both Rib and the closely related Rib‐like protein. It has been found that the Rib‐like protein is identical with the group A streptococcal R28 protein.18,28,29 As Rib and Rib‐like protein (R28) cross react and elicit cross protective immunity in animal studies,30 we have classified them as one antigen in this study (table 2).

Table 2 Characterisation of 30 group B streptococcal isolates from neonates with invasive infection by polysaccharide capsular type and cell surface expression of proteins.

Capsular type (no of isolates)	Surface protein reacting with antibodies to
	α (n = 14)	Rib (n = 14)	None* (n = 2)
Ia (4)	4	–	–
Ib (6)	6	–	–
II (3)	2	1	–
III (7)	–	6	1
IV (2)	2	–	–
V (6)	–	5†	1
NT (2)	–	2	–

*Isolates that did not react with rabbit antibodies specific for protein α or Rib.

†These type V strains probably expressed the R28 protein, which cross reacts with Rib and has very similar structure.28,32 See text.

NT, Not typable.

Protein preparations

The α and Rib proteins were isolated from streptococci by several purification steps to ensure that they were free of contaminating polysaccharides, as described previously.12,13

Enzyme linked immunosorbent assay (ELISA)

Concentrations of IgG antibodies to the α and Rib proteins were measured by ELISA. The wells of a microtitre plate (Maxisorp Nunc‐Immunoplate; Tamrolab, Molndal, Sweden) were coated with 100 μl of a solution of purified protein α or Rib in phosphate buffered saline (PBS) (0.75 μg/ml). The test sera were diluted 1:500 in PBSAT (PBS containing 0.02% azide and 0.05% Tween 20) and tested in duplicate. Goat anti‐human IgG conjugated to alkaline phosphatase (Sigma, St Louis, Missouri, USA) diluted 1:10 000 was used to detect bound antibodies. The plate was read in a microplate reader at 405 nm. For each serum, the background value for an uncoated well was subtracted. Antibody concentrations are presented as the absorbance value.

As sera with known concentrations of antibodies to the α and Rib proteins are not available, commercial γ globulin (Gammaglobulin 165 mg/ml, Pharmacia, Uppsala, Sweden) was used to standardise the ELISA. Optimal dilutions of the reagents were determined by checkerboard titration.31 Standard curves for γ globulin (in serial dilutions in PBSAT) against each of the two GBS proteins were determined. Appropriate dilutions for γ globulin were chosen to be included as control on each microtitre plate to eliminate inter‐assay variations.

Statistical methods

Background characteristics of the cases and controls were calculated using statistical descriptive methods including mean, median, range, and proportion. Linear relations of antibody concentrations to the α and Rib proteins in sera from cases and controls were estimated using Pearson correlation coefficient and corresponding confidence interval (CI). Associations between the concentrations of antibodies to protein α and Rib in sera from mothers and female blood donors were estimated using the Wilcoxon two sample test. Associations between invasive infection and antibody concentration were estimated by odds ratios (OR) and corresponding 95% CI, adjusted for sex and gestational age using logistic regression. Statistical significance was defined as p<0.05.

Ethics

This study was approved by the ethics committee of Lund University. All mothers gave written consent after oral and written information.

Results

Study subjects

Over the study period, 35 cases of neonatal GBS infection diagnosed by positive cultures from blood and/or cerebrospinal fluid were reported from the catchment area. Two term neonates were excluded because their infection occurred several weeks after birth, and for three neonates (two term and one preterm) the bacterial isolate or both the neonatal and the maternal sera were missing. The remaining 30 neonates with invasive GBS infection were included in the study (table 1); three of them had meningitis. Twenty one (70%) showed signs of illness within 24 hours and 27 (90%) within 72 hours of birth. Sixty non‐infected neonates were included in the control group. The gestational age of cases was higher and therefore their birth weight was higher. Sixty seven per cent of the infected neonates were boys compared with 55% of the controls. The ages of mothers of infected and non‐infected neonates were similar, median 28 and 30 years, mean 29 and 30 years, respectively. Twenty nine acute sera from the 30 infected neonates and 29 sera from their mothers were collected within seven days of admittance to the neonatal ward (median 1.5 days for the neonates and 3 days for their mothers). All sera from the control group were collected within five days of admittance.

Properties of GBS strains isolated from neonates

The 30 GBS strains collected from blood and/or cerebrospinal fluid from infected neonates were analysed for expression of surface protein, using rabbit antisera against highly purified preparations of the α and Rib proteins. According to this analysis, 14 of the 30 isolates expressed the α protein and 14 isolates expressed Rib, and two strains did not express any of these proteins. There were six different serotypes (Ia, Ib, II, III, IV, and V) and two non‐typable among the strains (table 2). It is important, however, to note that strains of serotype V express a protein that is closely related to, but not identical with, the Rib protein.32 This Rib‐like protein is identical with the R28 protein (also known as Alp3), a protein with chimeric structure expressed by both GBS and group A streptococci.18,28,29 Because the R28 protein of GBS type V strains cross reacts with Rib, and because these proteins elicit cross protective immunity,30 we have classified R28 and Rib as one antigen that is recognised by anti‐Rib serum (table 2).

Antibodies to α and Rib in sera from neonates and their mothers and female blood donors

IgG antibodies that recognise purified protein α and Rib in ELISAs were detected in sera from newborn infants, their mothers, and female blood donors (table 3). A covariation was observed between concentrations of antibodies to α and Rib in sera from the neonates, both in cases and controls. These covariations were also observed in sera from their mothers and from female blood donors (table 3). These data suggest that α and Rib may cross react in humans.

Table 3 Correlation between concentrations of antibodies to protein α and Rib in sera from 29 cases with invasive group B streptococcal infection, 60 non‐infected controls, and 100 female blood donors.

	Correlation coefficient	95% CI
Anti‐α versus anti‐Rib in neonatal sera
Cases	0.82	0.65 to 0.91
Controls	0.73	0.58 to 0.83
Anti‐α versus anti‐Rib in maternal sera
Cases	0.88	0.78 to 0.95
Controls	0.80	0.69 to 0.88
Anti‐α versus anti‐Rib in sera from female blood donors	0.73	0.62 to 0.81

CI, Confidence interval.

Concentrations of serum antibodies to protein Rib in mothers of neonates infected with a Rib‐expressing strain and in mothers of non‐infected neonates were lower than in female blood donors (p  =  0.003 and p  =  0.01 respectively; table 4). For mothers of neonates infected with an α‐expressing strain, concentrations of antibodies to Rib were also lower than in blood donors. This difference, however, did not reach statistical significance. In sera from mothers of neonates infected with a Rib‐expressing strain, the concentrations of antibodies to α were lower than those of blood donors (p  =  0.003; table 4).

Table 4 Mean, median, and range of concentrations of antibodies to protein α and Rib in sera from case mothers (n_α  =  13 and n_Rib  =  14) and control mothers (n  =  60) compared with antibody concentrations in sera from female blood donors (n  =  100) using Wilcoxon two sample test.

	Mothers of neonates infected with				Mothers of non‐infected neonates		Female blood donors
	α‐expressing strain		Rib‐expressing strain
	Anti‐α	Anti‐Rib	Anti‐α	Anti‐Rib	Anti‐α	Anti‐Rib	Anti‐α	Anti‐Rib
Mean	0.35	0.34	0.16	0.21	0.29	0.34	0.31	0.45
Median	0.19	0.29	0.10	0.18	0.22	0.23	0.22	0.34
Range	0.05–1.22	0.03–1.25	0.01–0.46	0.03–0.52	0.02–1.21	0.04–1.65	0.01–0.99	0.05–1.71
p Value	0.98*	0.16†	0.003*	0.003†	0.75*	0.01†	–	–

Antibody concentrations are presented as absorbance values.

*Compared with anti‐α in sera from female blood donors.

†Compared with anti‐Rib in sera from female blood donors.

Distribution of antibodies to α and Rib in neonatal versus maternal sera

Figure 1 shows the distribution and correlation coefficients of α and Rib antibodies in neonatal versus maternal sera. A clear positive linear relation was observed for concentrations of antibodies to protein α and Rib, between neonatal sera and maternal sera, in both cases (fig 1A,B) and controls (fig 1C,D), indicating that the antibodies are transferred transplacentally.

Figure 1 Distribution and association between concentrations of antibodies to protein α and Rib in neonatal and maternal sera. (A) Anti‐α in cases with invasive group B streptococcal (GBS) infection. (B) Anti‐Rib in cases with invasive GBS infection. (C) Anti‐α in non‐infected controls. (D) Anti‐Rib in non‐infected controls. The x and y axis indicate the absorbance value at 405 nm. Pearson correlation was used.

Determinants for invasive neonatal GBS infection

The concentrations of antibodies to protein α and Rib in sera from neonates infected with GBS strains expressing Rib were associated with invasive infection (p  =  0.03 and p  =  0.03 respectively), in a model in which sex and gestation were adjusted for (table 5). The adjusted OR was 0.0007 (95% CI 0.00 to 0.54) for antibodies to α and 0.002 (95% CI 0.00 to 0.57) for antibodies to Rib. No statistically significant association was detected between invasive infection with strains expressing protein α and concentrations of antibodies to the α and Rib antigens (table 5). OR for the concentration of antibodies to Rib, however, showed a protective effect. In this study, gestational age was associated with invasive GBS infection, independent of protein expression on the bacterial surface of the infecting strain. These ORs varied between 1.21 and 1.30, and the p values were ⩽0.007. No association was observed between sex and invasive GBS infection.

Table 5 Associations, adjusted for gestational age and sex, between invasive group B streptococcal (GBS) infection and concentrations of antibodies to protein α and Rib in serum from neonates.

Infecting isolate	No of sera (cases and controls)	Antibody concentration
		Anti‐α			Anti‐Rib
		OR	95% CI	p Value	OR	95% CI	p Value
All isolates*	89	0.11	0.003 to 3.86	0.22	0.07	0.004 to 1.38	0.08
α‐expressing	73	1.17	0.02 to 67.22	0.94	0.43	0.02 to 12.00	0.62
Rib‐expressing	74	0.0007	0.000 to 0.54	0.03	0.002	0.000 to 0.57	0.03

Analyses were by multiple logistic regression including all neonates with GBS infection and controls (n  =  89), neonates with GBS infection caused by strains expressing protein α and Rib, and controls (n_α  =  73 and n_Rib  =  74).

*All GBS isolates with protein α, Rib, or neither of these proteins expressed on the bacterial surface.

OR, Odds ratio; CI, confidence interval.

Discussion

We analysed concentrations of antibodies to the α and Rib proteins in neonates with invasive GBS infection and in their mothers. The most noteworthy finding is the strong association between neonatal concentrations of antibodies to the GBS proteins α (OR 0.0007) and Rib (OR 0.002) and invasive infection caused by strains expressing Rib (table 5). For neonatal infections caused by strains expressing the α protein, no such association could be shown. There was a tendency, however, for Rib antibodies to protect against α‐expressing strains, but we do not know if this was a spurious correlation or a real effect. It is possible that a larger study would clarify the latter observation. Sex and gestational age of the neonates were considered as possible confounders and were therefore adjusted for in the statistical analyses.

Our results should be valid, as all but five cases reported to us of invasive GBS infection diagnosed by positive culture from blood or cerebrospinal fluid during the three year study period were included in the study. All recruited mothers consented to participate in the study. Two neonates with GBS infection diagnosed several weeks after birth were regarded as a separate group with different characteristics and were not further considered in this study. The control neonates were recruited from the population that generated the cases. The GBS colonisation status of the control mothers was not recorded because general screening for GBS in pregnant women is not routine in the study area. In addition, the accuracy of detecting colonisation is influenced by many factors—for example, the body sites sampled, the number of cultures obtained, and the timing of culture collection in relation to delivery.8 Moreover, we do not know if high concentrations of antibodies to the α and Rib protein would actually reduce or even eliminate maternal GBS colonisation. If the latter were the case, the recruitment of the control group would have been biased, as women with high concentrations of α and Rib antibodies would have been excluded from the study.

Characterisation of the GBS strains isolated from the 30 infected infants showed that 47% expressed protein α on the bacterial surface and 47% expressed Rib or a protein that cross reacts with Rib. This result corroborates previous reports that either α or Rib are expressed by most human invasive GBS strains.12,13,33

What is already known on this topic

• Animal studies have shown that vaccine induced antibodies to proteins exposed on the bacterial cell surface of GBS confer protective immunity

• Antibodies to GBS cell surface proteins seem to occur naturally in human sera; however, studies comparing serum concentrations of antibodies to GBS proteins from infected neonates and their mothers with concentrations from non‐infected neonates are scarce

The clear relation between the concentrations of antibodies to the GBS cell surface proteins α and Rib in individual sera (table 3) together with the finding that low concentrations of neonatal antibodies to α and Rib were associated with invasive infection caused by strains expressing Rib (table 5) may indicate that human antibodies to α and Rib show cross reactivity. In a previous experiment with mouse antibodies,13 a very weak immunological cross reactivity between the α and Rib proteins was observed. Moreover, immunisation of mice with native protein Rib conferred partial protection against experimental infections with α‐expressing strains. The cross reactivity is possibly more extensive for human antibodies. The present study measured naturally occurring human antibodies which were probably produced as an immunological response to native streptococcal proteins exposed on bacteria that colonise the mucosal surfaces in the intestinal and urogenital tracts. It is possible that these human antibodies interact with conformational epitopes shared by the native forms of α and Rib.

The concentrations of antibodies to α and Rib seemed to be generally lower in sera from mothers of infected and non‐infected neonates than from non‐pregnant women represented by female blood donors (table 4). The observed differences in antibody concentrations between pregnant and non‐pregnant women may be, at least in part, explained by the decrease in IgG caused by haemodilution during pregnancy.34

In our study, we found an association between GBS infection and gestational age. This association may reflect the increasing number of neonates born (and exposed to GBS) during late pregnancy. On the other hand, among the GBS infected neonates in this study, there were no differences between gestational age and time of first symptoms of infection in preterm and term neonates. Most of the neonates showed signs of illness during the first few hours after birth, indicating that they were probably infected in utero. These observations suggest that GBS plays an active role in onset of labour and delivery in many cases, both in early and late pregnancy. However, the association between GBS infection and gestational age needs to be studied in a separate study.

In summary, this study shows a clear association between concentrations of antibodies to GBS proteins in sera from neonates and their mothers, which indicates that these antibodies are transferred to the fetus during pregnancy. More importantly, low concentrations of neonatal antibodies to the GBS proteins α and Rib were shown to be associated with invasive GBS disease caused by strains expressing Rib. These findings indicate that naturally occurring antibodies to GBS proteins may be involved in the defence against GBS infection.

What this study adds

• Low concentrations of neonatal antibodies to the GBS proteins α and Rib are associated with invasive disease caused by strains expressing Rib on the cell surface

• The findings indicate that naturally occurring antibodies to GBS proteins are involved in defence against GBS infection and this may have implications for the development of a GBS vaccine based on protein antigens

Abbreviations

CI - confidence interval

ELISA - enzyme linked immunosorbent assay

GBS - group B streptococcus

OR - odds ratio

PBS - phosphate buffered saline