Birgit Simell PhD, National Public Health Institute (KTL), Department of Vaccines, Mannerheimintie 166, 00300 Helsinki, Finland. E-mail: firstname.lastname@example.org
A number of studies have shown that the ratio of IgA1 and IgA2 subclasses in secretions can depend upon the nature of the antigen inducing their production. In order to evaluate the effect of the nature of the antigen on the subclass distribution of the naturally occurring salivary IgA antibodies against Streptococcus pneumoniae, we used enzyme immunoassay to measure the levels of natural IgA, IgA1 and IgA2 antibodies to pneumococcal capsular polysaccharide type 14 (PS14) and pneumococcal surface adhesin A (PsaA) in saliva of children during their first 2 years of life. The sum of anti-PS14 and anti-PsaA IgA1 and IgA2 correlated significantly with the antigen-specific total IgA, which showed that IgA1 and IgA2 add up to IgA. IgA1 was the predominant subclass for both antigens. The median of anti-PS14 and anti-PsaA IgA1 was higher than that of IgA2, and the antigen-specific IgA1 was found in a larger proportion of samples than IgA2. The ratio of IgA1 to IgA2 (IgA1/IgA2 ratio) was lower for anti-PS14 than for anti-PsaA, suggesting that the PS antigen induced more IgA2 than the protein antigen. The possible impact of the IgA subclass distribution on protection of mucosal surfaces by natural or vaccine-induced antibodies needs to be determined.
Immunoglobulin A (IgA) plays an important role in the protection of mucosal surfaces. In external secretions, the majority of antibodies belongs to the IgA class and occurs predominantly in a polymeric form containing a secretory component. The fact that the daily production of total IgA is considerably greater than that of the other immunoglobulin classes combined highlights the significance of IgA in host defence . Human IgA occurs as two subclasses, IgA1 and IgA2. These subclasses differ from each other in several aspects, including minor differences in the primary structure, carbohydrate composition, antigenic properties and sensitivity to the proteolysis by bacterial IgA1 proteases . The ratio of IgA1- and IgA2-secreting cells varies in the lymphoid tissues of the human body. Most lymphoid tissues show a predominance of IgA1-producing cells. However, in secretory lymphoid tissues, the share of IgA2 production is larger than in the non-secretory lymphoid organs, such as the peripheral lymph nodes and spleen . Correspondingly, the IgA1 and IgA2 antibodies are distributed characteristically in body fluids: serum IgA is mainly of subclass IgA1, while in external secretions subclass IgA2 is more prominent .
A number of studies have shown that the ratio of IgA1 and IgA2 subclasses can depend on the nature of the antigen inducing their production. The naturally and vaccine-induced IgA antibodies against T cell-dependent (TD) bacterial protein antigens in saliva and in colostrum are predominantly of IgA1 subclass [1,5–7]. By contrast, type 1 T cell-independent (TI) antigens, such as lipopolysaccharides (LPS) of Gram-negative bacteria and lipoteichoic acids of Gram-positive bacteria, stimulate predominantly production of natural IgA2 antibodies in saliva and colostrum [5,6,8]. The typical type 2 TI antigens, like bacterial polysaccharides (PS), can evoke both IgA1 and IgA2 responses in external secretions [5,6]. However, systemic immunization with pneumococcal (pnc) capsular PS either in unconjugated or conjugated form induces predominantly IgA1 responses in saliva of children [9–11].
The Finnish Otitis Media (FinOM) Cohort Study was a longitudinal follow-up study that assessed the risk factors for pnc carriage and pnc acute otitis media (AOM) [12,13], and the development and role of the natural antibodies against pneumococcus in relation to the pnc culture findings from the nasopharynx and middle-ear fluid during the first 2 years of life [14–18]. As part of the FinOM Cohort Study, we have demonstrated previously that contacts with Streptococcus pneumoniae induce natural salivary IgA responses against pnc protein and PS antigens in children [16,17]. To characterize further the natural development of mucosal anti-pnc immunity, in the present study we determined the subclass distribution of natural IgA antibodies against two distinct pnc antigens: pnc capsular PS of type 14 (PS14) and pnc surface adhesin A (PsaA).
The pnc antigens for the present study were selected on the basis of our previous experience in evaluating the natural development of salivary antibodies against pnc capsular PS (types 1, 6B, 11A, 14, 19F and 23F) and pnc proteins (PsaA, PspA and pneumolysin) in saliva samples [16,17]. One reason for choosing the PS14 and the PsaA antigen was the clarity of the previous anti-PS14 and anti-PsaA antibody results in relation to the culture-confirmed pnc contacts [16,17]. Further, in contrast to all the other pnc PS types, the PS14 antigen has been shown to not contain polyreactive epitopes [19–22]. Pnc serotype 14 is also one of the serogroups associated most frequently with pnc colonization and pnc diseases in young children in industrialized countries [12,13,23,24]. PsaA, a pnc vaccine candidate, is a component of an ATP-binding cassette (ABC) Mn2+–permease complex that has been shown to play a critical role in pnc adherence and virulence . PsaA has also been included in our previous studies on serum antibodies of the FinOM Cohort Study children [26–28]. Our aim in the present study was to evaluate whether the nature of the antigen influences the subclass distribution of natural salivary anti-pnc antibodies. For the first time, we have compared the proportions of natural IgA1 and IgA2 antibodies against a pnc PS and a pnc protein antigen in saliva samples of the same individuals.
Materials and methods
Study population and saliva samples
Saliva samples for the present study were selected from the FinOM Cohort Study material used in our previous studies [16,17]. The FinOM Cohort Study population comprised 329 healthy children followed prospectively from 2 months to 2 years of age. The children were vaccinated following the standard Finnish vaccination schedule, which includes bacille Calmette–Guérin (BCG) vaccine against tuberculosis, vaccine against pertussis, diphtheria and tetanus (PDT), Haemophilus influenzae type b (Hib) vaccine against invasive infections caused by Haemophilus influenzae type b, inactivated poliomyelitis vaccine (IPV) against polio, and vaccine against mumps, measles and rubella (MMR). The Finnish vaccination schedule does not include any pnc vaccine. The FinOM Cohort Study protocol was approved by the Ethics Committees of the National Public Health Institute (KTL), Tampere University Hospital, and the Department of Social and Health Care of Tampere City. Informed consent was obtained at the time of enrolment from the parents of all children participating in the FinOM Cohort Study.
The unstimulated saliva samples were collected at the ages of 6, 12, 18 and 24 months by placing a plastic pipette in the cheek area and applying a gentle suction. The saliva samples were frozen immediately and stored at −70°C before the first analyses. In the present study, selected saliva samples were rethawed and centrifuged at 19 000 g for 10 min before determination of the antibodies. The supernatants were used for antibody measurements.
A prerequisite for a saliva sample to be selected for the present study was that it had been found to contain anti-PS14 and/or anti-PsaA IgA in the previous measurement [16,17]. The detection limits for anti-PS14 and anti-PsaA IgA in the previous measurements were OD 0·04 and 0·03 (two standard deviations of the blank), respectively. An optical density of ≥ 0·1 for anti-PS14 or/and anti-PsaA was set up as a selection criterion for the samples used in the present study. Of the FinOM Cohort Study saliva samples still available for analysis, 30 fulfilled this condition for either anti-PS14 or anti-PsaA IgA or both. Only one sample per child was included in the analyses. Of the 30 children, 29 had had a culture-proven pnc contact, i.e. nasopharyngeal or middle-ear fluid culture positive for S. pneumoniae, before the age when the saliva sample was collected. A total of seven of the 30 children (23%) had had a culture-proven contact with the pneumococcus of serotype 14. The pnc cultures in the FinOM Cohort Study were perfomed regularly at 1–6-month intervals (additional cultures were performed during respiratory infections and AOM episodes) [12,13]. Thus, some pnc contacts may have been missed due to the study design.
Pnc capsular PS antigen of type 14 was obtained from the American Type Culture Collection (Rockville, MD, USA). Cell wall polysaccharide (CWPS) used for absorption of anti-CWPS antibodies from the saliva was obtained from Statens Seruminstitut (Copenhagen, Denmark). The recombinant PsaA was prepared with the Qiaexpress system (Qiagen, Chatsworth, CA, USA) by Drs J. Sampson and E. Ades (Centers for Disease Control and Prevention, CDC, Atlanta, GA, USA). The expression host Escherichia coli was transformed with pAB247, the recombinant plasmid that carries the psaA gene from the serotype 2 strain D39 (gene sp1650; GenBank accession no. 1PSZA) cloned into pQE30. The His-tagged recombinant PsaA was purified by Ni-NTA chromatography .
Enzyme immunoassay (EIA)
The levels of IgA, IgA1 and IgA2 antibodies to pnc PS14 and PsaA in saliva were determined by EIA as described earlier . The second antibodies were monoclonal anti-human IgA (M26012; Skybio, Bedfordshire, UK), monoclonal anti-human IgA1 and monoclonal anti-human IgA2 (A89-036 and A89-038, Nordic Immunological Laboratories, Tilburg, the Netherlands). The third antibody was polyclonal alkaline phosphatase-conjugated rabbit anti-mouse IgG (H&L 315-055-045; Jackson Immunoresearch Laboratories, West Grove, PA, USA). An OD of ≥ 0·04 (two standard deviations of the blank) for all measurements was considered to be positive. Samples with undetectable anti-PS14 or anti-PsaA IgA1 or IgA2 were assigned a value equivalent to half the detection limit. The IgA1 and IgA2 results are given semiquantitatively based on OD values (OD units, OD × 1000). These were calculated from the mean OD readings of triplicate samples after subtraction of the OD readings of the phosphate-buffered saline (PBS) blank plates.
Analysis of the antibody data
spss (SPSS Institute) and Excel (Microsoft) software were used for the analysis of the antibody data. We used Pearson's correlation to evaluate the correlation between IgA and the sum of IgA1 and IgA2 (and IgA1 and IgA2 individually), between the ratios of anti-PS14 and anti-PsaA IgA1 to IgA2 (IgA1/IgA2-ratios) and between the previous and present IgA results. The log-transformed salivary IgA, IgA1 and IgA2 data have been summarized by box plots based on the median, quartile and extreme values.
A total of 30 saliva samples found to contain anti-PS14 and/or anti-PsaA IgA in our previous measurements [16,17] were selected for analyses of anti-PS14 and anti-PsaA IgA, IgA1 and IgA2 antibodies in the present study (OD ≥ 0·1 in the previous measurement used as a selection criterion). The IgA results obtained in this study correlated with the previous IgA results (r = 0·86 for both antigen specificities). Of the 30 samples, three, seven, seven and 13 had been collected at the ages of 6, 12, 18 and 24 months, respectively. Due to the small number of samples and the lack of any obvious age-specific differences in the results, the antibody results at different ages were combined.
Correlation of the antigen-specific IgA with the sum of antigen-specific IgA1 and IgA2 and with IgA1 and IgA2 individually
To see whether it was appropriate to compare the IgA1 and IgA2 levels and calculate their ratios, the sum of the OD units of anti-PS14 and anti-PsaA IgA1 and IgA2 of each sample was correlated with the corresponding value of antigen-specific IgA (Fig. 1). Irrespective of the proportions of the subclasses in the individual samples, there was a significant linear correlation of the sum of OD units of anti-PS14 and anti-PsaA IgA1 and IgA2 (IgA1 + IgA2) with the antigen-specific total IgA (r = 0·98, P < 0·01 and r = 0·98, P < 0·01, respectively). Thus, IgA1 and IgA2 were shown to add up to IgA, which allowed us to compare the IgA1 and IgA2 levels. The salivary anti-PS14 and anti-PsaA IgA1 levels mirrored the antigen-specific total IgA, which was seen by the strong correlation of antigen-specific IgA1 with the antigen-specific IgA (r = 0·96, P < 0·01 for anti-PS14 and r = 0·97, P < 0·01 for anti-PsaA). The correlation of anti-PsaA IgA2 with the antigen-specific IgA was poor (r = 0·59, P < 0·01) due to the predominance of IgA1. In line with this, the correlation of anti-PS14 IgA2 with the anti-PS14 IgA was stronger (r = 0·83, P < 0·01) than that of anti-PsaA IgA2 with the anti-PsaA IgA due to the higher anti-PS14 IgA2 levels.
Distribution of anti-PS14 and anti-PsaA IgA1 and IgA2 antibodies in saliva of children
The number of samples with detectable anti-PS14 and anti-PsaA IgA1 and IgA2 is shown in Table 1. For both antigens, IgA1 was found to be the predominant subclass. Anti-PS14 and anti-PsaA IgA1 was detected in 93% and 97% of the samples, respectively, while anti-PS14 and anti-PsaA IgA2 was found in 57% and 43% of the samples, respectively (Table 1). Anti-PS14 and anti-PsaA IgA was not detected in two and one samples, respectively.
Table 1. Number of jhsaliva samples (%) containing detectable anti-PS14 and anti-PsaA IgA1 and IgA2 antibodies in children.
No. of positive samples (%)
IgA1 and IgA2
Seven of the 30 children had had a culture-proven pnc contact specifically with serotype 14. Of the saliva samples from these seven children, five (71%) samples contained both detectable anti-PS14 IgA1 and IgA2, while two (29%) samples contained only anti-PS14 IgA1. None of the samples contained only anti-PS14 IgA2. Detectable anti-PsaA IgA1 and IgA2 was found in two (29%) of the seven samples and only anti-PsaA IgA1 in five (71%) of the seven samples. None of the samples contained only anti-PsaA IgA2. Of the saliva samples from the 22 children colonized with other pnc serotypes, 11 (50%) samples contained both detectable anti-PS14 IgA1 and IgA2, while 9 (41%) samples contained only anti-PS14 IgA1. None of the samples contained only anti-PS14 IgA2. Two (9%) samples were negative for both anti-PS14 IgA1 and IgA2. Detectable anti-PsaA IgA1 and IgA2 was found in 11 (50%) of the 22 samples and only anti-PsaA IgA1 in 10 (45%) of the 22 samples. One of the 22 samples (5%) was negative for both anti-PsaA IgA1 and IgA2.
The geometric means (GM) of anti-PS14 and anti-PsaA IgA1 levels were found to be similar to the antigen-specific total IgA (Fig. 2). The GMs of anti-PS14 IgA and IgA1 were 336 and 245 OD units, respectively, and the GMs of anti-PsaA IgA and IgA1 were 401 and 369 OD units, respectively. The GMs of anti-PS14 and anti-PsaA IgA2 were low: 60 and 33 OD units, respectively (Fig. 2). The GMs of anti-PS14 IgA1 and IgA2 were found to be relatively closer to each other than the GMs of anti-PsaA IgA1 and IgA2.
In the seven children colonized specifically with serotype 14, the GMs of anti-PS14 IgA, IgA1 and IgA2 were 918, 781 and 153 OD units, respectively, and the GMs of anti-PsaA IgA, IgA1 and IgA2 198, 189 and 25 OD units, respectively. In the 22 children colonized with other pnc serotypes, the GMs of anti-PS14 IgA, IgA1 and IgA2 were 229, 160 and 43 OD units, respectively, and the GMs of anti-PsaA IgA, IgA1 and IgA2 548, 504 and 37 OD units, respectively.
The difference in ratios of salivary anti-PS14 and anti-PsaA IgA1 to IgA2
In Fig. 3 is shown the distribution of ratios of salivary anti-PS14 and anti-PsaA IgA1 to IgA2 (IgA1/IgA2 ratios) in the individual samples. In 23 (77%) of the 30 saliva samples the IgA1/IgA2 ratio was higher for anti-PsaA than for anti-PS14 (dots above the diagonal line in Fig. 3), which indicates that proportionally more of anti-PsaA was of IgA1 subclass than of anti-PS14. This finding, with the fact that the median of anti-PS14 IgA1/IgA2 ratios (3·7) was markedly lower than the median of anti-PsaA IgA1/IgA2 ratios (13·5) (data not shown), suggests that the IgA2 levels for anti-PS14 were relatively higher than those for anti-PsaA.
The chemical nature of the immunogen and its route of exposure can affect the magnitude, the site and the quality (e.g. monomeric or polymeric, with or without secretory component, and the subclass) of the induced IgA response . In the present study, the subclass distribution of salivary IgA antibodies against two different pnc antigens (i.e. pnc PS and protein antigen) was analysed in the selected FinOM Cohort Study saliva samples known to contain natural secretory IgA antibodies against pnc capsular PS of type 14 (PS14) and/or PsaA protein [16,17].
With reference to the variable distribution of IgA1- and IgA2-secreting cells in various secretory tissues, the mucosal site in question influences the proportions of IgA subclasses in secretion. The upper respiratory tract and digestive tracts are reported to contain mainly IgA1-secreting cells, while the lower intestinal and female reproductive tracts are populated either by equal numbers of IgA1- and IgA2-producing cells or predominantly by the latter isotype [3,31,32]. In human salivary glands, IgA1-secreting cells represent about two-thirds of the IgA-secreting cells . Saliva and human milk are reported to contain 60–75% of IgA1 [4,33,34]. The results of the present study are in line with this.
Although IgA1 and IgA2 subclasses are believed to have much the same functional effects, the extended hinge region of IgA1 molecules renders them highly susceptible to the IgA1 proteases produced by several pathogenic bacteria colonizing the upper respiratory tract, including pneumococci . The susceptibility of IgA1 to bacterial proteases may affect its role in protection of mucosal surfaces. IgA1 proteases cleave the IgA1 molecule in the hinge region to Fab and Fc fragments, and it has been suggested that by coating themselves with functionally deficient Fab fragments the pathogens could turn the defensive secretory IgA1 antibodies to their own advantage . Recently, Weiser et al. have shown that pnc IgA1 protease may be able to modify the IgA1 antibodies so that they enhance markedly pnc adherence to host epithelial cells in a cell-culture colonization model .
Because the IgA2 antibodies are resistant to bacterial IgA1 proteases, pronounced production of IgA2 antibodies in secretions may offer a functional advantage for the defence of mucosal surfaces against the IgA1 protease-producing bacteria. In infants, the ratio of salivary IgA1 and IgA2 has been reported to reach the adult proportions by 6 months of age [37,38]. In the study by Tappuni et al. , the ratio of IgA1 and IgA2 did not differ significantly between the predentate children (mean age 4 months; IgA1/IgA2: 60 : 40), the dentate children (mean age 1·5 years; IgA/IgA2: 51 : 49) and the adults (mean age: 29·5 years; IgA1/IgA2: 63 : 37). This early maturation of IgA2 in saliva has been suggested to be significant as half the pioneer streptococci colonizing the oral cavity within early childhood produce IgA1 proteases [38,39].
The subclass distribution of natural anti-PS14 and -PsaA IgA antibodies may be speculated to mimic the subclass distribution of vaccine-induced IgA antibodies, i.e. there might be proportionally more IgA2 production after vaccination with a PS-based vaccine than with a protein-based vaccine. Saliva samples of children vaccinated with a protein–PS-conjugate vaccine against S. pneumoniae or Hib have been analysed for specific anti-PS IgA antibodies in several studies [10,40–43]. Relatively high proportions of anti-PS antibodies detected in saliva of the 7-month-old children after the primary-series vaccination have been of IgA2 subclass, followed by IgA1 dominance after the booster dose [10,40]. The potential effect of IgA1 proteases should be taken into account when testing mucosal vaccines that induce high concentrations of local IgA.
In conclusion, we have determined the subclass distribution of naturally occurring IgA antibodies to pnc PS14 and PsaA in saliva of children during the first 2 years of life. We found that while the predominant subclass for both salivary anti-PS14 and anti-PsaA IgA antibodies was IgA1, the proportion of IgA2 was higher for the PS antigen. The significance of the IgA subclass distribution in protection by natural or vaccine-induced antibodies needs to be determined.
We acknowledge Mika Lahdenkari for statistical advice. We also thank all parents and children who volunteered to participate in the FinOM Cohort Study. The FinOM Cohort Study was supported by Merck & Co. Inc., Sanofi Pasteur, Wyeth-Lederle Vaccines and Pediatrics. In addition, this work was supported by World Health Organization grant no. V23/181/116.