Senescence in immunity against helminth parasites predicts adult mortality in a wild mammal

See allHide authors and affiliations

Science  20 Sep 2019:
Vol. 365, Issue 6459, pp. 1296-1298
DOI: 10.1126/science.aaw5822

The decline of resistance in old age

Infection, immunity, and demography are rarely measured simultaneously, despite being intertwined. Froy et al. measured an immune marker of resistance to infection by worm parasites (helminths) in Soay sheep off the remote Atlantic island of St. Kilda (see the Perspective by Gaillard and Lemaître). They used a library of 2000 blood samples from 800 known individuals that have been left to run wild and unmanaged. Resistance declines as the sheep age, which reduces a sheep's chances of surviving the winter. Helminths are an important component of many natural systems, including humans, and may thus become an increasing burden on health with age.

Science, this issue p. 1296; see also p. 1244


Our understanding of the deterioration in immune function in old age—immunosenescence—derives principally from studies of modern human populations and laboratory animals. The generality and significance of this process for systems experiencing complex, natural infections and environmental challenges are unknown. Here, we show that late-life declines in an important immune marker of resistance to helminth parasites in wild Soay sheep predict overwinter mortality. We found senescence in circulating antibody levels against a highly prevalent nematode worm, which was associated with reduced adult survival probability, independent of changes in body weight. These findings establish a role for immunosenescence in the ecology and evolution of natural populations.

Demographic senescence, the decline in survival prospects and fertility with age, is well documented in wild vertebrates and is known to play an important role in the dynamics of natural populations (13). Biogerontologists have made great strides toward understanding the genetic and physiological processes underpinning senescence in laboratory organisms (46), but we do not yet know whether similar mechanisms drive demographic aging under natural conditions (1, 3). Declining immune function in old age, or immunosenescence, is widely observed and associated with age-related increases in morbidity and mortality in laboratory rodents and humans (79). Parasites represent a major selective force in natural populations, and the ability to mount effective immune defenses against them represents a critical determinant of fitness in wild animals (10, 11). A growing body of evidence suggests that immunosenescence also occurs in natural populations: several recent, largely cross-sectional vertebrate field studies have documented age-related variation in immune markers across adulthood (12). However, the absence of large-scale longitudinal studies simultaneously measuring parasites, immunity, health, and demography in the wild has limited our ability to test how declines in immune function affect ecological and evolutionary processes (12). There is a similar scarcity of human studies linking within-individual changes in immune phenotype in old age with clinical outcomes (9).

Gastrointestinal nematode parasites are important drivers of selection in wild vertebrate systems (13, 14). Despite the global impact of helminth parasites (15, 16), the potential for senescent declines in immune-mediated resistance to helminth infection and the consequences for human, livestock, and wildlife health have rarely been considered. Studies of lab mice, humans, and domestic ruminants show that resistance to such worm infections is dependent on T helper type 2 (TH2) immune responses (17), with a key contribution from serum antibodies (18, 19). There is evidence from wild mammals and humans that the intensity of worm burdens may increase in later adulthood (20, 21), and cross-sectional studies in laboratory mice suggest TH2 function and anti-worm antibody production may be compromised in old age (2224). In particular, two studies found that elderly mice had a reduced or delayed immunoglobulin G (IgG) antibody response to worm infection (22, 24), suggesting that this could be an important marker of immunosenescence. Deterioration in TH2-mediated immunity to worms may therefore be responsible for increasing burdens and negative health outcomes in later life, but currently longitudinal studies testing for associations among immunity, worm burden, and components of fitness are completely lacking.

We used an unmanaged population of Soay sheep on the remote St. Kilda archipelago in Scotland, which has been the subject of detailed study since 1985 (25), to test for fitness consequences of senescent declines in immunity in a wild population. First-winter lamb mortality is often high in this population (25). Amongst individuals that survive to adulthood, mean longevity is 5.5 years for females (maximum 16 years) and 2.7 years for males (maximum 10 years). Demographic senescence is well-documented in this population, with female fecundity and survival of both sexes declining progressively starting at ~5 years (26). The sheep are host to a diverse community of gastrointestinal nematode parasites, including several highly prevalent strongyle species, which have been linked to gut pathology and overwinter mortality (27). Counts of nematode eggs from fecal samples provide an important proxy of parasite burden (27). Fecal egg counts (FEC) increase with advanced age in adult Soay sheep, which could reflect a loss of immune-mediated host control of parasite infection in later life (26).

Our previous work has established levels of plasma IgG binding antigen from Teladorsagia circumcincta (IgG-Tc), a highly prevalent worm in both wild Soay and domestic sheep, as an important marker of immunity to helminths in this system. In adult Soay sheep, plasma levels of IgG-Tc in summer are weakly correlated with other immune measures, including other antibody isotypes (IgA, IgM, and IgE) binding the same Tc antigens (2830). IgG-Tc levels are also negatively associated with FEC and positively associated with subsequent winter survival, independently of other humoral and cellular immune measures (2830). Further analysis showed that IgG-Tc correlates strongly (correlation coefficient r > 0.9) with levels of IgG binding antigen from a range of strongyle species (materials and methods; fig. S1), only some of which are present on St. Kilda, suggesting a high degree of cross-reactivity. We also showed that, regardless of which strongyle species is used, levels of binding by this IgG predict subsequent survival (table S1). This implicates IgG-Tc as a potentially powerful marker of specific immunity to helminths in this study system and motivated an in-depth study of the causes of its association with adult mortality. Using measurements of IgG-Tc by enzyme-linked immunosorbent assay (ELISA) from 2215 longitudinal blood samples collected from 797 adult sheep aged 3 years or older over 26 years (1990–2015) on St. Kilda (materials and methods; table S2), we tested whether IgG-Tc showed within-individual declines in later life consistent with immunosenescence, and whether such declines were associated with parasite burden, body weight, and subsequent mortality.

We found senescence in our marker of resistance to nematode infection in wild Soay sheep. Levels of IgG-Tc declined with age [regression slope β = −0.006, 95% confidence interval (CI) = −0.010 to −0.002; table S3A], but because individuals may senesce at different rates, the number of years an individual is away from death may be a better reflection of their biological aging patterns than chronological age itself (31). Accordingly, we found that years before death explained more variation in IgG-Tc than chronological age (Table 1). Levels of IgG-Tc declined as adults approached death, and the relationship was best described by a threshold function with the decline accelerating over the final year of life (Fig. 1A and table S3B). Males had lower average levels of IgG-Tc than females (Table 2), but there was no detectable interaction between years before death and sex, indicating that the pattern of within-individual changes in IgG-Tc was consistent between the sexes. IgG-Tc levels were highly repeatable across the adult lifetimes of sheep, with 58% of the variance in our dataset explained by individual identity (repeatability = 0.576, 95% CI = 0.542 to 0.609). These results show that despite consistent among-individual differences in IgG-Tc across their lifetimes, average antibody levels declined within individuals as they approached death.

Table 1 Comparing the explanatory power of linear models with different fixed-effects structures for levels of IgG-Tc in Soay sheep.

Anti–T. circumcincta IgG was measured in plasma samples collected from adult sheep in summer 1990–2015 (n = 1869 observations of 651 individuals) All models included individual, year, and ELISA plate as random intercept terms. Sex was included as a two-level factor, age as a linear covariate, and years before death (YBD) as a threshold function with a break point at 1 year (see table S3). The number of parameters is indicated by degrees of freedom (df), and ΔAIC is the difference in the Akaike information criterion value compared with the best model (highlighted in bold).

View this table:
Fig. 1 Senescent declines in circulating anti–T. circumcincta IgG predict overwinter mortality in Soay sheep.

(A) Levels of IgG-Tc declined as Soay sheep approached death. Points and error bars show raw data medians and standard errors (black circles indicate females, gray triangles indicate males), and lines show predictions from a linear mixed-effects model (Table 2) with 95% confidence intervals (gray shading). (B) Annual overwinter survival probability was not related to an individual’s mean levels of IgG-Tc measured over adulthood, as indicated by the regression slope for the among-individual effect of IgG-Tc on survival estimated from a bivariate model (table S4). (C) Individuals with relatively low levels of IgG-Tc compared with their average were less likely to survive the winter, as indicated by the within-individual effect of IgG-Tc on survival (also estimated from the bivariate model; table S4). Points show raw data, and black lines show regression slopes with 95% credible intervals shaded in gray.

Table 2 Fixed and random effect estimates from the best model of IgG-Tc in Soay sheep.

Showing the effect of years before death (YBD) and sex (relative to females) on levels of circulating IgG-Tc in adult sheep. Estimates are given with 95% confidence intervals from 1000 bootstrap replicates.

View this table:

Senescent declines in immunity were associated with increased subsequent mortality risk. We used a bivariate mixed-effects model to estimate the covariance between IgG-Tc measured in summer and the probability of survival over the subsequent winter at three different levels: among-individual, among-year, and within-individual (see materials and methods; 2202 observations of 796 individuals over 26 years). Among-individual covariance captures the association between an individual’s average adult antibody level and their overall life span, whereas within-individual (or residual) covariance represents the association between the deviation in IgG-Tc from an individual’s mean value and that individual’s prospects of surviving the following winter. Among-year covariance reflects associations between the population’s average antibody levels and mortality rates across years. We found that covariance between IgG-Tc and survival was statistically significant only at the within-individual level, and not at either among-individual or among-year levels (Fig. 1 and table S4). The absence of any among-individual covariance reveals that consistent differences in immunity across adulthood, potentially associated with genotype or early-life environment, did not predict life span (Fig. 1B). However, the positive within-individual covariance indicates that individuals showing a within-individual decline in IgG-Tc had a reduced survival probability the following winter (Fig. 1C). These data show that longitudinal changes in a marker of immune resistance in later adulthood predict mortality risk in the wild and indicate that immunosenescence may play an important role in age-related declines in demographic rates in natural populations.

The within-individual association between IgG-Tc and survival remained when associations with FEC, our index of parasite burden, and body weight were accounted for (Fig. 2 and table S5). We ran a multivariate mixed-effects model that included all four measures as response variables, and again estimated covariance among the terms at the among-individual, among-year, and within-individual levels (see materials and methods). As expected for a marker of resistance to worm infection, FEC and IgG-Tc were negatively associated at the within-individual level (table S5). Higher levels of IgG-Tc at both among- and within-individual levels were associated with increased body weight, and weight covaried positively with survival at all three levels [table S5; accounting for variation in structural size (hindleg length) in models of body weight did not change our results, table S6]. We used this multivariate approach to test for independent effects of immunity, parasite burden, and weight on subsequent survival, while accounting for the interdependencies among these terms (analogous to a multiple regression; see materials and methods). We found that within-individual deviation in IgG-Tc was still a predictor of overwinter survival (Fig. 2). The independence of the immunity–survival relationship from body weight suggests that it was not mediated by variance in individual body condition or resource availability, and that late-life declines in body weight and immunity reflect separate physiological senescence pathways. This highlights the complex, multifaceted nature of physiological senescence in wild animals and the need for large-scale multivariate studies to understand which processes are most important for late-life fitness across taxa and ecological contexts (32).

Fig. 2 Effects of anti–T. circumcincta IgG, body weight, and fecal egg count (FEC) on overwinter survival probability in adult Soay sheep.

Regression coefficients (median of the posterior distribution with 95% credible intervals) reflect independent effects on survival, accounting for the covariance among these traits at different hierarchical levels. Effects are shown for the within-individual (dark gray circles), among-individual (black triangles), and among-year (gray squares) associations with survival. Within-individual deviation in IgG-Tc was predictive of survival probability, independent of the within- and among-individual variance in body weight and FEC. Effects were estimated from a multivariate model of IgG-Tc, weight (both Gaussian), FEC (Poisson), and survival probability (threshold) (see materials and methods; table S5). IgG-Tc and weight were standardized before inclusion in the model (mean = 0, SD = 1).

Few studies to date have investigated how the immune system changes in later adulthood in response to pathogenic, chronically infective helminth parasites. Our analyses show that associations among adult infection, immunity, and survival are not driven by constitutive among-individual differences, determined by genetics or early-life conditions, but rather by within-individual variation late in life linked to senescence. Studies in laboratory mice suggest the TH2 response to worm infection becomes compromised in old age, and that the host’s ability to resist infection declines as a result (2224). Although the observed within-individual negative correlation between FEC and IgG-Tc is consistent with a resistance function for this immune marker, our multivariate models show that longitudinal declines in IgG-Tc predict mortality independently of FEC, suggesting this relationship with mortality is not solely mediated by reduced worm burden. This may reflect the indirect and therefore inherently noisy relationship between FEC and actual worm burden. However, changes in host tolerance of worm infection, a process we have previously linked to variation in host fitness in our study system (33) or density-independent alterations in parasite behavior in response to host physiological deterioration [e.g., helminth suppression of the immune response (17)] could also explain the relationship between IgG-Tc and survival. Our results suggest that changes in the interactions between host immunity and helminth infection during adulthood could have implications for host ecological dynamics, helminth epidemiology, and host-parasite co-evolution in wild vertebrates. The focus on the development of immunity to helminths in early life in humans and livestock is understandable, but our data suggest that changes in host immune responses to worm infection occur in mammals during later life.

Supplementary Materials

Materials and Methods

Figure S1

Tables S1 to S6

References (3452)

Data S1

References and Notes

Acknowledgments: We are grateful to all those involved in the long-term study of Soay sheep on St. Kilda; J. Hadfield, A. Phillimore, and J. Pick for statistical advice and discussions; and A. Graham, A. Hayward, L. Kruuk, and R. Maizels for constructive comments on the manuscript. Thanks to the National Trust for Scotland for permission to work on St. Kilda; QinetiQ and Kilda Cruises for logistical support in the field; and D. Bartley, A. Morrison, and R. Maizels for provision of nematode larvae. Funding: UK Natural Environment Research Council, Biological and Biomedical Sciences Research Council, Medical Research Council, The Wellcome Trust (204052/Z/16/Z), and Rural and Environment Science and Analytical Services Division of the Scottish government. Author contributions: H.F., T.N.M., and D.H.N. designed the study; J.G.P. and J.M.P. collected the samples, life history data, and managed the long-term study; A.M.S., K.W., and R.S. conducted the antibody assays in the laboratory; H.F. and F.B. analyzed the data; H.F. and D.H.N. wrote the manuscript with input from all co-authors. Competing interests: The authors declare no competing interests. Data and materials availability: The data supporting the results are available in the supplementary materials.

Correction (13 December 2019): In Figure 1A, the x axis, “Years before death,” was accidentally mislabeled. It should have read (from left to right): 8, 6, 4, 2, 0. This has been corrected in the HTML and PDF online. The plot itself is not affected.

View Abstract

Stay Connected to Science

Navigate This Article