Published ahead of print on November 7, 2008, doi:10.1164/rccm.200803-438OC
© 2009 American Thoracic Society doi: 10.1164/rccm.200803-438OC
Interleukin-10/Interleukin-5 Responses at Birth Predict Risk for Respiratory Infections in Children with Atopic Family History1 Telethon Institute for Child Health Research, and Centre for Child Health Research, Faculty of Medicine and Dentistry, University of Western Australia, Perth, Australia Correspondence and requests for reprints should be addressed to P. G. Holt, M.D., Division of Cell Biology, Telethon Institute for Child Health Research, P.O. Box 855, West Perth, WA 6872, Australia. E-mail: khaas{at}ichr.uwa.edu.au
Rationale: Respiratory infections in early life are associated with risk for wheezing bronchiolitis, especially in children at high risk of atopy. The underlying mechanisms are unknown, but are suspected to involve imbalance(s) in host defense responses against pathogens stemming from functional immaturity of the immune system in this age group. Objectives: To assess the contribution of eosinophil-trophic IL-5, and the potent antiinflammatory cytokine IL-10, to risk for infection in early life. Measurements and Main Results: We prospectively monitored a cohort of 198 high-risk children to age 5 years, recording every acute respiratory infection episode and classifying them by severity. We measured cord blood T-cell capacity to produce IL-10 and IL-5, and related these functions to subsequent infection history. IL-10 and IL-5 were associated, respectively, with resistance versus susceptibility to infections. The greatest contrasting effects of these two cytokines were seen when they were considered in combination by generating IL-10/IL-5 response ratios for each subject. The low IL-10/high IL-5 T-cell response phenotype was strongly associated with susceptibility to all grades of acute respiratory infection, relative to the more resistant high IL-10/low IL-5 phenotype. Conclusions: Excessive production of IL-5 by T cells at birth is associated with heightened risk for subsequent severe respiratory infections, and this risk is attenuated by concomitant IL-10 production. The underlying mechanisms may involve IL-10–mediated feedback inhibition of IL-5–dependent eosinophil-induced inflammation, which is a common feature of host antiviral responses in early life.
Key Words: infectious diseases • cord blood • cytokines
Infancy and early childhood are recognized as periods of maximal risk for infectious diseases (1). In westernized countries the principal infections during this life phase are respiratory, and the list of relevant pathogens is dominated by rhinovirus and respiratory syncytial virus (RSV) (2, 3). Most of these infections are of short duration and remain restricted to the upper respiratory tract, but in a subset of subjects they spread to the lower airways and give rise to severe febrile and/or wheezing symptoms that necessitate hospitalization (4). Severe infections of this nature, particularly during the first 2 years of life, have been associated in epidemiologic studies with markedly increased risk for subsequent development of atopy and asthma (5, 6). Genetic risk for these inflammatory diseases has been in turn linked to a variety of underlying developmental deficiencies in adaptive and/or innate immune functions, in particular those associated with helper T (Th)-cell–associated mechanisms (7). However, the relationship between these developmental deficiencies in immune competence and risk for severe respiratory infections per se is less well understood, and was the subject of the present study. It is clear from the results of many investigators that overall immune function during the early postnatal period is constitutively skewed toward Th2 cytokine production (8, 9), representing the persistence of a phenotype that serves in utero to protect the fetoplacental unit from the highly toxic effects of Th1 cytokines (10). However, one potential consequence of the postnatal persistence of this fetal immune phenotype is that host defense mechanisms that rely principally on Th1 effector mechanisms, particularly antiviral immunity, may become distorted during this life phase by the presence of an excessive component of Th2 cytokine production, which has the potential to antagonize the development of effective Th1-dependent sterilizing immunity (11). The relatively high frequency of bronchiolitis associated with eosinophilia in RSV-infected infants relative to other age groups (12–15) represents one likely example, whereby the prominent IL-5 component in the Th2-biased antiviral response of this age group appears to drive eosinophil responses in the airways (16, 17). Similarly, the results from infant mouse models of RSV infection (18, 19) have linked the degree of functional immaturity (and hence Th2 bias) of the immune system at the time of first infection with ensuing morbidity, and also with morbidity at subsequent infection; in both cases linking infection outcome with degrees of ensuing eosinophilia. These findings are consistent with the general model that inflammation resulting from Th2-driven airway eosinophilia represents a major cause of severe bronchiolitis in early life, and that susceptibility is directly related to the developmentally determined Th1/Th2 balance at the time of infection (8). We aimed to extend this model to consider the converse immunologic process, the contribution from mechanisms associated with protection against inflammation. The archetypal antiinflammatory mechanism in the immune system is represented by the activities of the cytokine IL-10, which, in humans, is produced by multiple cell types within the innate and adaptive arms of the immune system (20). Adaptive immune responses in newborns are dominated by this cytokine, which is produced at disproportionately high levels by T cells during early life (21). We hypothesized that high IL-10 responsiveness during infancy may provide protection against the downstream inflammatory consequences of respiratory infections in Th2-biased neonates and infants, and that infection severity may be related to the balance between capacity to produce IL-10 and proinflammatory Th2 cytokines, exemplified by IL-5. We tested this hypothesis in a prospective cohort study of children at high genetic risk of atopic disease, in whom the degree of Th2 bias during early life would be expected to be high relative to the population at large.
Subjects Subjects comprised 238 children from a cohort of 263 recruited in a prospective study on acute respiratory infections (ARI), 198 of whom were monitored to age 5 years (detailed in Reference22). This cohort is at high risk for developing atopic disease, as at least one parent of each subject had doctor-diagnosed asthma, hay fever, or eczema.
Infection Episode Assessment
Cytokine Responses
Microarray and Quantitative Reverse-Transcriptase–Polymerase Chain Reaction
Statistical Analysis At first, the relationship between levels of cytokines in cord blood (IL-10 and IL-5) and infection incidence was assessed by Spearman correlation. Univariate Poisson regression models were used to investigate associations between infection incidence and the loge values of individual cord blood cytokines and cytokine response ratios, as well as cytokine response data derived from peripheral blood mononuclear cells collected at 6 and 12 months. To estimate the relative risk of infection as a function of cytokine response capacity, incidence rate ratios (analogous to odds ratios generated in logistic regression models) were computed via these Poisson regression models. Incidence rate ratios in this context provide a relative measure of the effect of a given level of cytokine response capacity on risk for infection, designated in the text below as relative risk (RR). To allow for changes of rates with age and repeated measures over time on the same children, Poisson regression with random effects was used to estimate RRs of the loge(IL-10/IL-5) ratios on respiratory infections in the first 5 years of life, including a categorical variable for each year of age and with each number of infections in each year as the dependent variable and time spent in the study in that year as denominator (or "offset"). Possible differences in effects on infection rates of the ratios with age were also investigated, using interaction terms. All analyses were also adjusted for the possible confounding effects of daycare attendance, number of children at home, and passive smoking. The statistical methods used for the analyses of the microarray data (25–27) are detailed in the online supplement. Descriptive analyses were done with SPSS software (SPSS, Chicago, IL) and all other analyses (except for microarray data) were performed in Stata (StataCorp, College Station, TX). Although we looked at many combinations of cytokine levels and disease outcomes, the results of all are presented, so no formal adjustment to the P values was performed (28).
A total of 238 children were included in the study, with 134 of them boys (56.3%). Table E1 in the online supplement and Figure 1 show infection incidence rates during the first 5 years of life. The incidence rate for ARI was highest during Years 1 and 2 (about four episodes per person-year) and declined thereafter. The proportion of infections spreading to the lower respiratory tract annually averaged one third, and this proportion remained relatively constant throughout the observation period.
Table E2 illustrates the spectrum of IL-5 and IL-10 response capacities in CBMC samples from the study group after stimulation with the polyclonal T-cell mitogen PHA. Comparable variation (data not shown) was observed with a second T-cell stimulant, SEB. PHA-induced IL-5 and IL-10 responses were highly variable, and these variations in production levels were associated with fluctuations in risk for subsequent infections. Cord blood cytokine response capacity correlated with the number of ensuing respiratory infection episodes. For example, during the first year, positive correlations were found between the number of FRIs and levels of IL-5 production in response to the polyclonal T-cell stimulant PHA or SEB, with P values of 0.018 and 0.002, respectively. There was also a correlation between IL-5 (SEB) production and the number of wLRIs (P = 0.034). In the second year, significant positive correlations were found between FRI and SEB-stimulated IL-5 (P = 0.046). In contrast, negative correlations were observed between infections and IL-10 responses sporadically across the time course. Notably, a negative correlation was observed between ARI in Year 5 and PHA- and SEB-stimulated IL-10 (P = 0.001 for both), and between FRI and PHA-induced IL-10 (P = 0.014). A similar weak negative association was observed between LPS-stimulated IL-10 and ARI in Year 2. We next assessed cumulative risk for infection associated with individual cord blood cytokine responses by calculating RR for infection over the full 5-year observation period, using Poisson regression (Table 1). Significant negative correlations were observed between PHA- and SEB-stimulated IL-10 responses and estimated RR for total ARI over the 5 years (P < 0.001). In contrast, IL-10 production in response to the innate immune stimulus LPS was unrelated to infection incidence (data not shown). The association with PHA- and SEB-induced IL-10 did not extend to LRI and/or symptoms, suggesting that it was related to susceptibility to primary infection as opposed to ensuing infection severity. In contrast, PHA- and SEB-stimulated IL-5 production were associated with increased risk for all categories of infection.
The contrasting positive and negative associations observed between IL-5 and IL-10 response capacity and ARI suggested that susceptibility to respiratory infection may be related to the balance between these two cytokines during host defense responses. In absolute terms IL-10 was produced at higher levels in all subjects, but the relativity between IL-10 and IL-5 levels varied widely within the population (Table E2). To address questions concerning the quantitative relationship between these two cytokine responses in individual children and their subsequent infection histories, we expressed individual IL-10 and IL-5 production levels as ratios that were then loge transformed to estimate RR in terms of log(IL-10/IL-5) response ratios. Figure 2 shows the RR of respiratory infections for the loge ratios of PHA-stimulated (Figure 2A) versus SEB-stimulated (Figure 2B) IL-10/IL-5. Children with the higher IL-10/IL-5 ratios appeared to have lower risks for virtually all of the four categories of respiratory infections in each of the five 1-year periods spanning the study. Notably, these analyses generated 40 subgroups within the overall study population (Figure 2), and the IL-10/IL-5 ratios were protective against respiratory infections with 39 RRs in the 40 subgroups being below the reference value of 1.0 and among them 16 RRs being significant. The most consistent infection category yielding statistically significant P values across the observation period was ARI, but FRI and LRI additionally displayed repeated significant associations, suggesting that relative IL-10/IL-5 production capacity may also be related to infection severity as well as to overall infection susceptibility. When these analyses were repeated with IL-10 response data obtained from the innate immune stimulus LPS in conjunction with PHA-induced IL-5, no comparable associations were observed (data not shown).
We also tested IFN-  , IL-13, and tumor necrosis factor- responses in combination with IL-10, and the combination of IL-5 and IFN-, but observed only weak and inconsistent relationships and these cytokines were not pursued further. To further investigate the association between relative IL-10 and IL-5 response capacity and the incidence of ARI in the first 5 years of life, we reestimated RR for cumulative infections across the 5-year observation period in relation to cord blood loge(IL-10/IL-5) response ratios (Table 2). IL-10/IL-5 response ratios were strongly associated with reduced overall risk for respiratory infections over the first 5 years of life. Similar findings resulted from follow-up analyses on T-cell IL-10 versus IL-5 response profiles in samples collected from the cohort at 6 and 12 months of age (data not shown). In contrast to T-cell cytokine response profiles, ratios of IL-10 to IL-5 production elicited by the innate immune stimulus LPS were not predictive of subsequent infection risk (data not shown). To confirm the findings, Poisson regression models with random effects were used to estimate RRs for respiratory infections in the first 5 years of life (see Table 2), allowing for changes over time in infection rates. The RRs changed slightly and P values increased, as expected. However, the protective effects of the IL-10/IL-5 ratios on respiratory infections were still significant in the IL-10/IL-5 [PHA] ratios for ARI (P = 0.017) and LRI (P = 0.046) and in the IL-10/IL-5 [SEB] ratios for both ARI and FRI (P = 0.012 and 0.013, respectively). The protective effects were consistent across all categorical infections, with some insignificant RRs due to insufficient statistical power. We also tested the interactive effects of these ratios with age on respiratory infections and these interactions were not large or significant (all P values > 0.09).
Expression Profiling Microarray analysis was used to investigate whether additional genes were hyperexpressed in mitogen-stimulated CBMCs from the infection-resistant IL-10hi/IL-5low individuals. With a 2.0-fold–change cutoff, hyperexpressed genes were identified in PHA responses from this group after an empirical Bayes t test (see the online supplement), including a subset of genes that, in addition to IL-10, are known to be associated with immunologic function. These were selected for additional qRT-PCR analysis within the individual RNA samples employed for microarray analysis, employing material from unfractionated CBMCs and from CD4+ and CD8+ cells purified from the CBMCs. Among these, three major immune function–associated genes in addition to IL-10 differed significantly between the groups (Table 3), notably genes encoding the transcription factor c-Maf, which plays a number of roles in T-cell differentiation; and the pleiotrophic cytokines IL-2 and IL-21. Expression of c-Maf in purified CD4+ T cells correlated strongly with IL-10 (Pearson r2 = 0.86, P < 0.0001) and also with IL-2 (r2 = 0.47, P = 0.0095) and IL-21 (r2 = 0.88, P < 0.0001).
This study has shown that risk for repeated respiratory infections during infancy and early childhood is markedly influenced by the relativity between IL-10 and IL-5 response capacity. Moreover, the relevant cytokine responses in this context are within the adaptive immune system, as substitution of T-cell–derived IL-10 response data in these analyses with data obtained with the innate immune stimulus LPS did not yield similar findings. The summary data in Figure 3 illustrate this tripartite relationship between T-cell–derived IL-10 and IL-5 and cumulative infection risk over the first 5 years of life. Figure 3A presents the predicted infection incidence rates of ARI estimated with a Poisson regression model using loge PHA-induced cord blood IL-10 and IL-5 response data (as individual independent variables) and 5 years of ARI episodes from individual children (as the dependent variable). It can be seen that relative risk for ARI during the period varies over at least a twofold range between the two extremes of the cytokine response spectrum, namely, IL-10hi/IL-5lo versus IL-10lo/IL-5hi. For LRI (Figure 3B) it varies over at least a threefold range across the response spectrum.
The inference from these findings is that risk for infection during early life is associated with excessive IL-5 production and that this is attenuated by parallel production of IL-10, and further that IL-10 itself may have a positive role in resistance to early infections. It is possible that increased infection risk in the IL-5hi group may be due in part to underlying IL-4 production, which is a necessary prerequisite for IL-5 responses in these children, that is, via IL-4–mediated immune deviation (and hence attenuation) of antiviral Th1 responses. However, it is also feasible that IL-5 may contribute directly to the high-risk phenotype via its effects on eosinophil recruitment and activation. Eosinophil function during infancy is known to be increased throughout the whole population relative to that in older age groups (29, 30), reflecting the generalized Th2 bias characteristic of the immune system during this life phase (10), and the subset of IL-5hi children identified here as at maximal risk of infection are at the extreme of the Th2 bias spectrum. As such, the contribution of eosinophils to host defense responses (including to infections) within this subgroup is likely to be larger than within the population at large, and if this occurs in the context of respiratory viral infections it may translate into the intense symptomatology that is observed in susceptible infants (12–15). However, findings suggest that the appearance of eosinophils at infection sites may not always represent an immunologic aberration per se. Notably, one animal model study has demonstrated accelerated respiratory viral clearance in hypereosinophilic (IL-5 transgenic) mice, which suggests that inflammatory effector mechanisms used by these cells can also potentially contribute positively to host defense in the airways (31). This contribution may be particularly significant during early infancy, when many other inflammatory effector mechanisms are developmentally attenuated, in contrast to the relative hyperactivity of the eosinophil pathway (29, 30). Nevertheless, findings linking airway eosinophilia with bronchiolitis symptoms during early life (12–15) indicate that if this alternative pathway for viral elimination indeed operates in human infants then it is prone to dysregulation; it has intrinsic potential to overshoot in intensity (i.e., airway eosinophilia), and as a consequence can itself contribute to local inflammation and resulting pathology. This suggests a failure of whatever endogenous homeostatic mechanisms normally operate to limit local inflammation. In this context, one of the key functions of IL-10 is downregulation of inflammation (20), including IL-5–dependent eosinophil-associated airways inflammation (32). The relationship between these two cytokines within the T-cell compartment and infection risk, which is illustrated in Figure 3, is consistent with such a damping role for IL-10 during infancy.
The data in Figure 3 also leave open the possibility that relative resistance to infection in the IL-10hi group may be due in part to other and more direct effects of IL-10 on viral elimination. IL-10 is a highly pleiotrophic cytokine with multiple immunoregulatory functions, which also include direct or indirect stimulation of a range of important immunoinflammatory effector mechanisms that are central to host defense. In particular, IL-10 drives the functional maturation of CD8+ cytotoxic T cells (33–36) and associated allograft rejection responses (37), and also activation of CD8+ NK cells (26, 32, 35, 36). Of particular note are reports demonstrating the role of IL-10 in enhancing the IFN- As noted previously, we also acknowledge the possibility that the covert expression of additional effector genes may be central to the relative resistance to infection observed in children expressing the IL-10hi/IL-5lo phenotype. As a preliminary step to investigate this possibility we have performed a case–control investigation to compare gene expression profiles in stimulated CBMCs from two groups of neonates at the extremes of the IL-10/IL-5 response spectrum, focusing on genes with known immune-associated functions that are differentially expressed in the resistant IL-10hi/IL-5lo subgroup. Prominent among the hyperexpressed genes were those encoding transcription factor c-Maf, and the pleiotrophic cytokines IL-2 and IL-21. IL-2 has multiple immunoregulatory functions of relevance to this discussion, in particular in the promotion of antiviral immunity via enhancement of NK-cell and CD8+ T-cell activity (43, 44) and in expansion of T-regulatory cells (45), which play a central role in limiting the intensity and duration of inflammatory responses within the airway mucosa. IL-21, which is itself IL-2 dependent (32), is also hyperexpressed in the IL-10hi/IL-5lo subgroup. This cytokine has potent effects in stimulation of NK-cell and CD8+ T-cell effector functions (46–50), and has also been implicated in the regulation of dendritic cells (51). One report has also demonstrated that IL-21 can mediate an alternative pathway for induction of proinflammatory Th17 cells (52). Of additional interest was the finding that production of these three pleiotrophic cytokines in CBMCs from the IL-10hi/IL-5lo neonates was highly correlated with expression of the transcription factor c-Maf. Earlier interest in this transcription factor has focused mainly on its role in the regulation of IL-4/IL-5 balance in Th2 differentiation (53), but more recently it has become evident that it plays a broader role in T-cell function. In particular, c-Maf responsive elements have been demonstrated in the IL-10 promoter (54), overproduction of IL-10 is observed in c-Maf transgenic T cells (55, 56), and elevated expression of c-Maf is a feature of IL-10–producing regulatory T cells (57). In conclusion, although the balance between the pro- and antiinflammatory consequences of IL-5 and IL-10 production may play a direct role in resistance to (and subsequent intensity of) respiratory infections in the IL-10hi/IL-5lo subgroup in this cohort, it appears likely that genes encoding other cytokines and associated regulators, exemplified by this group of IL-2, IL-21, and c-Maf with functions overlapping and complementing IL-10, may also be involved in the expression of this phenotype. Such findings are not unexpected, given the accumulating evidence from genetic studies indicating the involvement of multiple genes in the pathogenesis of airway diseases. It is also pertinent to note the report relating to the Childhood Origins of Asthma (COAST) Project on high-risk children in the United States, which has also investigated relationships between cord blood cytokine responses and susceptibility to viral infections during the first years of life. No comparable relationships were detected between IL-10 and IL-5 response capacity and susceptibility to subsequent infection in that cohort (42). However, in that study a clinical score was used to determine whether episodes would be logged as infections and specimens collected. Symptoms needed to achieve a score of 5 (note: wheeze scored 5, and therefore all wheeze-associated infections would have been logged) before specimens were collected. Thus, infections in the COAST project excluded milder and upper respiratory infections and were more similar to those classified as severe (i.e., LRI/wLRI) in our study. In addition, IL-5 responses above the limits of detection were observed in only approximately 50% of the COAST cohort (58), in contrast to the present study in which responses were detected in at least 80% of subjects. These differences, which may be due to variations in cell-handling/culture techniques and/or to differing patterns of gene x environment interactions within the relevant Australian/U.S. populations, limit the capacity to make direct comparisons between the studies and underscore the necessity for follow-ups in other independent cohorts.
Supported by the National Health and Medical Research Council of Australia. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200803-438OC on November 7, 2008 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form March 20, 2008; accepted in final form November 5, 2008
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
