In this study, we evaluated how genetic differences impact the age-related ability to clear a bacterial infection using 183 genotypes of Drosophila melanogaster. We used a GWAS to identify genes and genetic networks that influence the age-specific immune response, resulting in three key findings.
First, we found significant variation among genotypes at each age (Fig. 1). These results are consistent with previous findings by Lesser et al. (2006) and Felix et al. (2012)5,8. We also found a significant effect of age, with clearance ability worsening, on average, by just over 70%. Two previous studies found different overall effects of age on bacterial clearance. Lesser et al. reported a moderate improvement (17%) in bacterial clearance ability with age when averaged across the 25 genotypes used in their study. Felix et al observed no general effect of age on clearance in their study, which used 20 genotypes. Comparisons of the effects of age in our study with those of the previous studies is slightly misleading for two main reasons. First, due to technical issues, we had to use two different methods of injection. If flies were infected with a higher titer at old age in our study, this could bias our results and overestimate the effect of age on clearance ability. Second, these earlier studies used different genotypes and a much smaller number of genotypes than the current study. Of course, one of the main goals of the current and prior studies was to look at the genotype specific effects of age on immune function. When we do this, by comparing the reaction norms of each genotype and the age by genotype interaction terms from the ANOVAs, each study reached the same general conclusion - the effect of age on bacterial clearance ability differed dramatically among genotypes.
Our second key finding is that we identified several polymorphisms and candidate genes that contribute to variation in clearance ability at each age (Table 2; Supp File S2), suggesting that clearance ability is age-specific, as no genes are overlapping between ages. While this could be due to lack of power, additional lines of evidence suggest that this is not the entire explanation. The first is that age affected clearance ability in a genotype dependent manner. This was reflected in the significant age by genotype interaction and the crossing reaction norms of the clearance ability of each line across ages (Fig. 2). If, instead, the polymorphisms had the same relative contribution to clearance ability at both ages, the rank order of clearance ability of the lines would not be expected to change so dramatically with age. The second line of evidence is that the genetic correlation of clearance ability across ages was fairly low (0.221; Table 1). While the correlation is not 0, it is also not close to 1, indicating that the ability of a genotype to clear a bacterial infection at one age is not predictive of its ability to clear infection at the other. If the magnitude of the effect of a polymorphism on clearance ability changes with age, then these two outcomes would be predicted.
The third key finding was that we were able to map many of the genes to a global network (Fig. 4). GO analysis of the candidate genes revealed several different processes, with the most significantly represented GO term enriched with genes involved in cell adhesion, with 6 candidate genes (CG42594, CG43373, DIP-alpha, Rh7, dpr8, and kirre) being of those identified at 5 weeks of age (Table 3; Supp. File S3). Other significantly overrepresented GO terms that included 5 week old candidate genes included cellular components such as the cell/plasma membrane (atl, CG10550, CG42269, CG42594, CG8852, CG42594, CG43373, con, DIP-alpha, Rh7, dpr8, Hs6st, side-VIII, and kirre), or molecular functions such as transcriptional regulation (Lim1, EcR, pnt, NelfE, and CG9899) and compound eye development (a, cindr, Lim1, and NetA) (Table 3; Supp. File S3).
This lack of overlap prompted us to compare our GWA results to the list of candidate genes identified by Felix et. al. (2012). Although Felix et. al. did not use the DGRP lines, they did look at differences in clearance ability of 20 different inbred fly lines with age and assessed the association between gene expression in response to infection using 12 of those lines. They measured the expression of more than 1300 genes and found that about half (678) of the genes were upregulated and the other half (665) were downregulated with age5. Of those, 139 upregulated genes were associated with a reduced clearance ability, while 105 downregulated genes were associated with an improved, or better, clearance ability with age5. Surprisingly, only 10 candidate genes overlapped with those identified by Felix et. al, with 7 of them (CG13492, CG10550, CG31087, lim1, upSET, NetA, and Nelf-E) overlapping with our 5 week candidate genes; the other 3 genes, Wbp2, sens-2, and CG34417, overlapped with our 1 week candidate genes. All 10 overlapping genes were of those whose expression was upregulated, with higher expression levels of NetA, Lim1, and upSET being associated with a reduced clearance ability with age5.
lim1, Nelf-E and upSET are components of the nucleus involved in regulating transcription, while NetA is found within the cytosol and functions as a developmental protein. This lack of overlap in candidate genes affecting clearance ability across ages and between the study mentioned, could be due, in part, to the regulatory control of gene expression declining with age21–23. However, we cannot make a definitive statement about the magnitude and direction in which the expression of these genes are regulated as we did not measure expression levels, and the previous study did not use the DGRP lines. Future studies are needed to test this hypothesis.
It is important to note that our study was conducted using virgin females. Using males and or mated females may have yielded significantly different results. Other studies that have used males9,24,25 and/or mated females21,24, 26–29 have observed differences in immune response, suggesting that sex and mating status, also have an effect and underlying genetic contribution on immune response. For example, Short and Lazzaro looked at differences in virgin vs. mated females and identified many differentially expressed immune-response genes in response to infection between the two27. Similar findings on age-dependent genetic effects have been observed in other organisms, such as mice30–32, zebra fish33,34, and even cows15, indicating the importance of identifying the underlying mechanisms that contribute to age-related changes, not only in immune response, but other biological processes as well. Future studies using mated flies of both sexes, and/or a different pathogen, will be necessary to gain a more complete understanding of the genes involved in regulating the age-dependent immune response.