This study consists of four separate studies on pneumococcal PS responses: three phase I vaccine studies and one study on children with pneumococcal AOM. Each study was planned separately, and initially, we did not attempt to compare different vaccines. Instead we were interested in studying the mucosal response to pneumococcal conjugate vaccines, mainly based on the earlier knowledge that nasopharyngeal carriage of Hib was dramatically reduced after Hib-conjugate vaccination (Takala et al.1991, Murphy et.al. 1993, Mohle-Botani et al. 1993, Takala et al.1993). AOM was chosen to represent a local, nonsystemic pneumococcal infection in order to study the mucosal response also in a natural infection.
One of the aims in these studies was to consider how well the measurement of the ASCs in the peripheral blood (measured with the ELISPOT method), reflects the mucosal antibody response, measured as salivary antibodies (with EIA). Measuring ASCs is an indirect way to estimate the mucosal response, since the cells committed to mucosal sites transiently appear in the peripheral blood when trafficking via the circulation before homing to their destination. The present study does not provide direct data on the final distribution of the circulating ASCs into different tissues. Such data could be obtained by analysing the homing receptors expressed on the surface of the ASCs (McDermott and Bienenstock 1979, Butcher et al. 1996). However, only a few mucosal homing receptors are known: those directing the cells to the vascular endothelium of the gut lamina propria. Receptors directing the cells to respiratory mucosa are still poorly characterised, and suggested to differ both from the homing receptors of the peripheral lymph nodes (for serum antibody production), or of the gut mucosa (Abitorabi et al. 1996). Thus, optimal tools to study adhesion molecules of the ASCs committed to the respiratory mucosa do not exist. In this thesis, the ASC results were compared with the salivary antibody results, and the results demonstrated that an IgA response in saliva is related to a high IgA-ASC response.
In each of the studies, the number of circulating ASCs were counted in the peripheral blood. In addition, mucosal antibodies were measured in secretions, i.e., in saliva in the vaccine studies, and in the NPA in the AOM study. Finally, serum antibodies were also detected in order to estimate the overall immunogenicity of the vaccines in each study group. For the reasons described above, methods varied slightly between different studies. Since Study II began, all the assays in our laboratory have been conducted with the same methods and same reagents. They are therefore comparable, including Studies II and III. However, the saliva antibody results from the first vaccine study (I) cannot be directly compared to those in Studies II and III, because at the time Study I was carried out, the method was still in the process of being improved. Thus, in Study I the salivary IgA results are not expressed in relation to total IgA in saliva, which has been widely accepted as a means to compensate for the diurnal and individual variation in the protein concentration of secretory samples. In other studies, the salivary IgA results are given as specific IgA/total IgA (ng/µg). The method for measurement of the ASCs has remained essentially the same during all these studies. The antisera in the ASC measurement to detect IgG and IgM differ in Studies I and IV from those used in Studies II and III. However, the new antisera were first tested, and the ASC results did not differ from those obtained with the antisera formerly used.
The kinetics of the ASC response after pneumococcal vaccination was analyzed in Study I. Serotype-specific ASCs were not detected on day 0. Consistent with several earlier vaccine studies (Kantele et al.1986 and 1990, Tarkowski et al. 1990, Nieminen et al. 1996), the peak number of ASCs was detected on day 7, thereafter the number of ASCs decreased, and on day 14 only a few ASCs could be detected. On day 28 or 30 after immunisation, no specific ASCs existed. Occasionally the number of ASCs on day 9 exceeded the number detected on day 7. However, we adopted day 7 as the peak day to represent the ASC response in our later studies. The IgM ASC peak is regularly already seen a few days before the IgA and IgG peak, typically on day 5 (Heilmann et al.1987, Kantele 1990), as also demonstrated in Study I. Thus, the study design in II and III did not allow us to detect the IgM ASC peak. Neither were the days when serum samples were collected optimal to detect this isotype. Therefore, IgM is not discussed further in this thesis.
A major problem with saliva samples is the instability of the antibodies; enzymes in the saliva with degradative activity affect the antibodies in the sample. Therefore special precautions must be taken in transport and storage. This degradation can be prevented by addition of enzyme inhibitors and glycerol, and by storing the samples at -70°C (Butler et al. 1990). In this study, the samples were frozen immediately in dry ice, stored at -70°C, and thawed only once. Preservatives were not added to the samples, to avoid additional dilution of the samples, already containing only very low antibody concentrations. In addition to those sensitivity problems, specificity problems also existed for salivary EIA, as nonspecific binding was frequently seen. Background plates were used to control this binding.
IgG in saliva is suggested to be mostly derived from serum. Consistent with this, a positive correlation was noted between the IgG response in serum and in saliva, a finding to be discussed later in more detail. However, a considerable number of measurements in saliva remained negative for IgG, whatever the IgG concentration in serum. This could be due to saliva samples too diluted for EIA to detect the IgG antibodies, resulting from individual and diurnal variation in the salivary protein concentration (Sörensen et al. 1987) or from degradation of the immunoglobulins by the proteases in saliva. The first hypothesis is supported by our results, since most of the saliva results that remained negative for IgG were derived from a small number of vaccinees, and these vaccinees lacked salivary antibodies to three or four serotypes at the same time. On the other hand, the total IgA concentration in these saliva samples with no specific IgG antibodies was no lower than in the remaining samples, indicating that these samples were not more diluted. The explanation for this could be that IgG in saliva may be more prone to unspecific degradation by enzymes in saliva than is the dimeric S-IgA.
The fold increases in serum antibodies after immunisation are also sometimes difficult to interpret. Comparison of individual responses is disturbed by widely varying antibody concentrations even before immunisation. On the other hand, measurement of ASCs detects the response at cell level; no specific ASCs are detected before immunisation and the number of cells appearing in the blood after immunisation represents the response: indicating the magnitude of the response as the number of activated cells. Detecting the number of ASCs in the peripheral blood is therefore a useful alternative complementing more traditional serological methods for evaluation of the antibody response On the other hand, ELISPOT is labourious to perform and requires fresh blood samples for the analysis. It is therefore not useful in phase II vaccine studies, or in efficacy trials, where large populations are vaccinated, but can be adapted in phase I vaccine studies to add information on immunogenicity of the vaccines.
Parenterally administered pneumococcal vaccines, the polysaccharide vaccine PncPS, and the conjugate vaccines PncOMPC, PncD, and PncT were all able to induce mucosal and serum antibody responses. An ASC response was seen in all vaccinees. In adults, the responses were higher after the PncD and PncT conjugates than after the PncPS or PncOMPC. It was initially surprising that in Study I, the ASC and serum antibody responses in adults were actually lower after the PncOMPC than after PncPS, because the conjugate was expected to have enhanced immunological properties. This was most likely a direct consequence of the dose of the PS antigen which was radically different between the two vaccines: 1 µg/ml of each polysaccharide in the PncOMPC vaccine and 25 µg/ml in PncPS. Some studies with pneumococcal vaccines have also suggested that in adults, conjugate vaccines might not be significantly better immunogens than PS vaccines (Powers et al. 1996, Eby 1995). However, our later study (II) in adults showed that the ASC response and the humoral antibody responses were significantly higher after PncD and PncT conjugates than those seen after PncPS vaccine, indicating that the immunogenicity of a PS vaccine can be improved by conjugating the PS to a protein carrier also in adults. This is most likely due to the T-cell help provided that seems to be beneficial also in adults responding to PS antigens. This finding supports results from earlier studies showing a significant difference between PS administered as such and when conjugated to carrier proteins (Fattom et al. 1990).
Differences were also seen in the magnitude of the responses between the two conjugate vaccines PncD and PncT; both the total number of serotype-specific ASCs and the serum antibody response were higher after immunisation with PncT than after PncD, whereas salivary antibodies were seen more often after PncD than PncT. The structure of PncD and PncT vaccines is quite similar, since they both represent a PS-protein-conjugate vaccine. However, each of the conjugate vaccines still has its own characteristics. The polysaccharides in the vaccines can vary in length as well as in the terminal structure of the saccharide chains. Furthermore, different carrier proteins also have their own characteristics, and the PS/protein ratio and the coupling of the PS to the protein differs in each conjugate. These structural characteristics of the PS-protein conjugates have been shown to have an influence on the amount and the quality of serum antibody induced by the vaccine (Anderson et al. 1989, Seppälä et al. 1989, Verheul et al. 1989, van den Wijgert et al. 1991), even though the determinants of the immunogenicity are not yet understood in detail. It seems likely that the characteristics of the conjugate vaccines may also have an effect on the induction, amount, and quality of the mucosal antibody responses.
The PS antigens are able to induce only poor, if any, antibody responses in young children. The response to many of the pneumococcal serotypes improves by the age of 2 years, but for some of the serotypes, like 6B, it remains poor much longer, even until the age of 6 (Mäkelä et al. 1983, Leinonen et al. 1986). However, antibody responses to PS are seen in young children, because the PS is coupled to a protein carrier to provide T-cell help. The serum IgG concentrations detected in toddlers after a single dose of PncD or PncT (III) were comparable in magnitude to those detected in infants after primary immunisation series (three injections at the ages of 2, 4, and 6 months of age) (Åhman et al. 1998 and 1999). The booster response in infants was not dependent on the magnitude of the primary response, indicating that memory function was evoked by the primary immunisation, even if marked responses were not seen. The ASC responses seen in toddlers after PncD and PncT, on the other hand, were comparable to those we reported in adults after the PncPS, a vaccine that is widely used in adult populations to prevent invasive infections, although it is not able to induce memory. However, the ability of the conjugate vaccines to generate immunological memory is believed to be an important factor in protection against invasive infections in children (Eskola et al. 1990). The memory function is likely to play an important role in protection at the mucosal surfaces, as well. However, these studies did not evaluate memory function in the mucosal immune system.
Although the ASC responses to PncD and PncT were comparable in toddlers to those in adults after PncPS, the response in toddlers remained lower than in adults after PncD and PncT. However, salivary IgA antibody responses were seen in toddlers as frequently as in adults after the same conjugates. Thus the salivary IgA responses were seen in children as frequently as in adults despite the lower IgA ASC responses in children. This might suggest that a higher proportion of the IgA ASCs detected in the peripheral blood is committed to the mucosal sites in children than in adults. Moreover, only very low serum IgA concentrations were detected in children, which is consistent with our knowledge of the late maturation of the serum IgA production in children. Furthermore, serum IgG concentrations also remained distinctly lower in toddlers than in adults. Together, these findings suggest that mucosal IgA responses induced in adults and toddlers are comparable, although the overall ASC response in adults is higher. These findings support those of earlier studies suggesting that the mucosal compartment of the immune system matures and starts to function earlier in life than the systemic immune system (Pichichero et al.1981 and 1983).
The AOM study (IV) is the first to report a pneumococcal PS-specific ASC response induced by pneumococcal acute otitis media. The responses measured were specific to the causative pneumococcal polysaccharide type/group cultured in the MEF, since at the same time the number of ASCs. measured to other pneumococcal types remained low. ASCs could be detected in all the children studied, and their number in three of the children was surprisingly high: 260 to 1100 ASC/106 cells. These values exceeded the average response detected in the toddlers vaccinated, although more than 1 000 ASC/106 cells were occasionally detected also in children vaccinated with PncD or PncT. The response in these three children with AOM was high also when compared to other infections. The number of pathogen-specific ASCs in urinary tract infection has been less than 100 ASC/106 cells in most adult patients and even less in children with pyelonephritis (Kantele et al. 1994 and 1995). The rest of the ASC responses in AOM children, however, were lower than the average response in vaccinated children. The induction of immune responses with purified or synthetic antigen preparations is likely to differ from the situation after contact with live bacteria. The PS antigen lacks T-dependent properties, but when it is presented on bacteria the number of antigenic epitopes is higher, and PS antigens conceivably are introduced together with T-dependent antigens. Furthermore, other factors enhancing or suppressing the response can be involved, e.g., an ongoing viral infection or mixed bacterial infections that can induce inflammation and release of various cytokines and thus affect the response in a way difficult to predict.
The ASC response in children with AOM was clearly age-dependent. High ASC responses were mostly detected in children 24 months of age or older. This supports the concept that T-independent polysaccharide antigens are poor immunogens in infants and young children and that the immunogenicity of the PS antigens is improved by the age of 2 years. The mucosal IgA-response, however, was induced earlier. None of the younger children (less than 24 months) had serum antibody responses or an IgG-ASC response, but six of them had an IgA-ASC response and one a local IgA response in NPA. Measurable IgA in NPA was further detected in three of the younger children. This again supports earlier findings that a mucosal IgA response may be induced even in the first year of life, also to PS antigens that are then unable to mount systemic responses (Pichichero et al. 1981 and 1983)
In children with AOM, the highest ASC responses were seen to 19F, the serotype most commonly causing pneumococcal AOM. Furthermore, this high ASC response was related to serum and to salivary responses to the same serotype. In vaccinees, the ASC responses detected were closely similar to all four serotypes in each vaccine group, whereas each serum antibody response varied by pneumococcal serotype. The low serum responses were generally seen after PncPS and PncOMPC and mostly to serotype 6B, a poor immunogen in several studies (Leinonen et al. 1986, Mäkelä et al. 1989). After PncD and PncT, the responses to serotype 6B were clearly improved, a finding also supporting the improved immunogenicity of the conjugate vaccines as compared to that of PS vaccines. The highest responses, in general, for both serum IgG and salivary IgA, were to serotype 14 and 19F in adults and to serotype 19F in toddlers. This is in accordance with earlier findings suggesting that the structure of each PS has a profound influence on the response, both on magnitude and distribution of the immunoglobulin classes in the response (van den Dobbelsteen et al. 1992, Korkeila et al. Vaccine, in press).
After parenteral immunisation with protein vaccines considered to be T-cell-dependent antigens, the ASC response consists almost exclusively of IgG-producing cells (Lue et al. 1994, Trollmo et al. 1994). This was seen in the present study in response to the carrier proteins of the conjugates. Clearly IgG-dominant ASC responses were seen in adults and toddlers to the protein components of the vaccines. Several studies have shown that the ASC response to PncPS and other polysaccharide antigens, in contrast to that of protein antigens, is dominated by IgA-producing cells (Heilman et al. 1986, Heilman et al. 1987, Heilman et al. 1988, Lue et al.1988, Lue et al. 1990, Tarkowski et al. 1990). This was also the case in the present study. The IgA/IgG ratio of the ASC response to a PS that is conjugated to a protein carrier, on the other hand, is altered towards a pattern typical of T-cell-dependent antigens (Tarkowski et al. 1990). This was seen in adults (II) after pneumococcal conjugate vaccines, since both isotypes were observed in essentially equal numbers; this is in contrast to the response we saw to pure PS or protein antigens. However, variation in respect to the dominant isotype was observed between each serotype and different conjugates. In toddlers, on the other hand, the responses to PS components of the conjugate vaccines were clearly dominated by IgA-secreting cells.
The ASC response in children with AOM was always dominated by IgA-ASCs as well, in accordance with results from earlier studies showing an IgA-dominant ASC response after other mucosal infections such as gastroenteritis and urinary tract infection (Kantele et al. 1988 and 1994). The IgA ASC responses detected in AOM children were within the same range of magnitude as those seen in vaccinated toddlers. IgG ASC responses after AOM, on the other hand, were detected in only few children, less frequently than in the vaccinated toddlers. Thus, higher IgG ASC responses were evident in vaccinated toddlers than in children suffering from pneumococcal AOM infection.
As mentioned earlier, the ASC response in adults was higher after PncD and PncT conjugates, than after PncPS. This difference was more apparent in regard to the IgG responses, whereas the numbers of IgA ASCs were more comparable to each other, indicating that improvement in response to a conjugate vaccine, as compared to a PS vaccine, mostly concerns the systemic IgG responses. The difference in response between toddlers and adults after the conjugates was also more evident when the numbers of IgG ASCs were compared, and less difference was observed between the numbers of IgA ASCs. This supports the suggestion that although the ability to produce systemic IgG responses matures with age (and thus better IgG responses are seen in adults than in children), mucosal IgA responses comparable to those seen in adults can be induced even at an early age.
In adults some differences were also observed between responses to the two conjugate vaccines PncD and PncT, not only in the magnitude of responses, but also in the quality of responses. The number of IgG ASCs was significantly higher after PncT, whereas equal numbers of IgA ASCs were detected after PncD and PncT. However, after both conjugates, the number of IgA ASCs was still higher than in response to pure PS vaccine. Although the ASC response was not dominated by IgA- or IgG-secreting cells as clearly as in the response to PS or protein antigens, a clear difference in the IgA/IgG ratio of the ASC response was obvious between the two conjugate vaccines; IgA cells dominating in response to PncD, and IgG cells in response to PncT. Furthermore, S-IgA responses were seen in the saliva more frequently after immunisation with PncD than with PncT. This occurred despite the equal numbers of IgA ASCs seen after the two vaccines. Surprisingly, the IgG response in saliva was also higher (although not significantly) after immunisation with PncD, despite the higher serum IgG concentration after PncT. Together, these findings suggest that the PncD vaccine is able to induce better mucosal antibody responses (both IgA and IgG), whereas the PncT seems to be more efficient in inducing systemic IgG responses. In children, however, such differences between the two conjugates were not apparent. On the contrary, the responses were very similar and, as mentioned earlier, the ASC response was always clearly dominated by IgA-secreting cells.
The excellent correlation between serotype-specific IgA and serotype-specific antibodies associated with secretory component in saliva (Studies I-III) indicates that the IgA measured in saliva was secretory in nature. The lack of correlation between serotype-specific IgA in serum and in saliva further supports the fact that the IgA detected in saliva was locally produced.
In adults, the over twofold increase in specific IgA antibodies in saliva was associated with an IgA ASC response of more than 100 ASC/106 cells, and in toddlers the fold increase in IgA concentrations in saliva correlated with the number of IgA ASCs. These results suggest that a high number of IgA-ASCs detected with ELISPOT is an indicator of the mucosal IgA response after parenteral pneumococcal vaccination.
The IgG detected in saliva is mostly suggested to be derived from serum, high serum IgG concentrations leading to leakage of the antibodies into mucosal secretions (Brandtzaeg 1971, Korsud et al. 1980, Grönblad 1981). Study II showed that the IgG concentrations in saliva and serum were positively correlated, in accordance with the concept of the salivary IgG being from the same origin as is the serum IgG. In toddlers, salivary IgG was hardly ever detected, which can well be explained by the relatively low serum PS-specific IgG concentrations detected in toddlers as compared to those in adults.
On the other hand, in adults, lower salivary IgG concentrations were induced after PncT than after PncD, despite the higher serum IgG concentrations after PncT, and furthermore, in Study I, a salivary IgG response was seen more frequently after PncOMPC than after PncPS, despite higher serum IgG concentrations after PncPS than after PncOMPC. These findings suggest that salivary IgG may not be solely a result of leakage through the mucosal surfaces because of high serum IgG concentration, but that some of the salivary IgG may be produced locally. Thus, not only secretory IgA, but also IgG could be induced in secretions somewhat independently of the IgG at systemic sites. Another explanation could be a higher avidity of salivary IgG induced by PncOMPC than by PncPS; thus, lower IgG concentrations would be detected in the antibody assay after PncOMPC. The avidity could be expected to be increased after conjugated PS as compared to plain PS, as has actually been shown for serum IgG (Granoff et al. 1995, Anttila et al. 1998). However, it does not explain the differences between PncD and PncT in adults. One possible explanation is that the permeability of IgG molecules at mucosal surfaces differs individually or even diurnally in each individual.
One of the main aims of this study was to see how well the number of ASCs would correlate with the antibody responses, and this was determined also in respect to serum antibodies. In adults, no significant correlation could be shown between the number of IgG ASCs and serum IgG concentration. However, the study carried out in toddlers (III), showed a weak positive correlation. This may have been due to the low preimmunisation concentrations of pneumococcal IgG seen in toddlers. In adult volunteers widely varying concentrations of pneumococcal polysaccharide-specific IgG antibodies are often detected even before immunisation, which makes analysis of the correlation between IgG concentration and the number of IgG ASCs impossible. In toddlers the IgG concentration, as well as the fold increase in IgG concentration, represents more clearly a new induction of the IgG antibodies and can thus be compared to the number of IgG ASCs induced by the vaccine.
In this study we have attempted to evaluate the mucosal immune responses, and therefore have focussed on IgA. An interesting finding was that the pneumococcal conjugate vaccines PncD and PncT were able to induce a salivary IgA antibody response in toddlers as frequently as was reported for adults in Study II. On the other hand, it has been suggested that the reduction in Hib carriage seen after Hib immunisations is due to high IgG antibody concentrations induced by the vaccines, leading to leakage of the IgG antibodies to the mucosal surfaces (Robbins et al. 1995 and 1996, Kauppi et al. 1993). However, the serum IgG concentrations in toddlers in this study remained distinctly lower than in adults. Nevertheless, the serum IgG concentrations in toddlers were comparable to those reported in infants after a primary immunisation series (of three injections) of PncD and PncT (Åhman et al. 1998 and 1999), the same vaccines that are reported to reduce nasopharyngeal carriage of pneumococcus in infants (Dagan et al. 1996b).
The immune-mediated mechanisms behind the reduced carriage after Hib and pneumococcal conjugate vaccination still remain contradictory. Both IgA- and IgG-mediated mucosal immunity have been suggested to prevent colonization. In animal studies, intranasal administration of human anti-Hib antibodies in an infant rat model prevented colonization (Kauppi et al. 1993), suggesting that local antibodies play an important role. However, IgA and IgG were both effective, indicating that both isotypes prevent colonizations if sufficiently high concentrations are achieved in the nasopharynx. Furthermore, a high dose of IgG given intraperitoneally also can prevent colonization (Kauppi et al. 1993), suggesting that high systemic IgG concentrations would, indeed, leak into mucosal secretions. Moreover, studies with S. pneumoniae show that intralitter spread of colonization from one infant rat to another was prevented by anti-capsular antibodies given subcutaneously (Malley et al. 1998).
Human studies on passive immuno-prophylaxis have also suggested that systemic, pneumococcal PS-specific IgG may protect from pneumococcal AOM in high-risk children, even without stimulation of specific local immunity (Schurin et al. 1993). On the other hand, the presence of pathogen-specific anti-capsular S-IgA in the middle-ear fluid even at the onset of acute otitis media seems to be beneficial for the resolution of the disease (Sloyer et al. 1974, Sloyer et al. 1976, Karjalainen et al.1990). Together, these findings suggest that capsular polysaccharide specific antibodies of both isotypes, IgA and IgG, are beneficial in defence against colonization and acute otitis media; the exact role for the two isotypes still remaining somewhat obscure. Nevertheless, the improved serum IgG responses to PS antigens achieved with the conjugate vaccines in the first years of life may well explain the reduced colonization (Robbins et al. 1995 and 1996). However, these studies could not demonstrate IgG responses in salivary secretions in toddlers. Studies looking more specifically at this issue are still needed to clarify the role of IgA and IgG in mucosal protection.