Microarray analysis and, more recently, RNA sequencing (RNA-Seq) have become highly reliable methods for high-throughput gene expression analysis of the whole transcriptome. With a single experiment providing a quantitative readout of the expression of all genes, gene expression analysis is the most cost-effective high-throughput phenotyping method.
The high accuracy and sensitivity of the current methods for transcript quantification provide the unique opportunity to study gene expression differences in a wide range of organisms, across the different stages of a developing organism [1
], in different environments [2
], or even among individual cells [9
]. The analysis of gene expression patterns in contrasting environments is a highly promising approach to study phenotypic plasticity on a high-throughput scale [4
]. While such studies typically focus on a single or a few genotypes, another very popular line of research is the comparison of gene expression patterns among individuals with different genotypes or among different species. Such analyses have been used, for example, for the identification of genetic architectures underlying gene expression regulation [11
]. Moreover, even the evolutionary forces operating on gene expression patterns have been studied by comparing populations collected from distant habitats [13
] or contrasting evolved and ancestral populations [17
Similar to the impact of environmental and genetic variation on gene expression, dynamic changes during development also provide exciting research questions (e.g., expression profiles of the life cycle [20
]). Given the well-documented influence of these factors, it is apparent that they need to be tightly controlled for powerful and unbiased gene expression analyses. Because gene expression analyses of natural populations potentially confound genetic variation, environmental heterogeneity, developmental stage, and possibly even tissue heterogeneities [21
], it is advised to measure gene expression in settings that control most, if not all, of these factors. Hence, organisms that can be studied in the laboratory provide the potential for gene expression analyses that are much better controlled and thus more informative, in particular for small expression differences. It has become common practice to control for environmental heterogeneity by maintenance under well-defined conditions for multiple generations to minimize transgenerational effects. More challenging is the control of developmental stage, as genetic variation can also result in changes in developmental rates [22
]. Hence, the analysis of different genotypes during developmental stages with rapid transcriptomic turnover is extremely arduous. An alternative approach to account for this problem focusses on developmental stages which are characterized by low turnover in gene expression levels.
An early landmark study of gene expression in Drosophila
indicated that adult flies differ dramatically in gene expression between sexes, but, within a given sex, gene expression is rather stable across different age classes [24
]. As a consequence, many gene expression analyses in Drosophila
combine flies differing in age by up to 15 days in a single sample, for example, [11
]. Since several subsequent studies of senescence in Drosophila
demonstrated considerable gene expression dynamics during aging [29
], we were interested in scrutinizing the differences in gene expression between flies differing in age by as little as 24 h.
To provide a scale for gene expression differences which goes beyond statistical significance, we contrasted flies differing 24 h in age to adaptive expression changes after more than 100 generations in a novel temperature regime. Surprisingly, our analysis demonstrated that female Drosophila simulans flies differing 24 h in age showed about twice as many differentially expressed genes than females evolved for more than 100 generations. Our results highlight that reliable expression analysis in Drosophila adults requires accurate timing of the developmental stage to uncover biologically relevant expression changes.
Contrasting evolved populations with 24 h age difference, i.e., at day five and day six after eclosure, provides two interesting insights into short-term expression changes. First, it was surprising to find such a large number of differentially expressed genes. Second, the significant overlap of GO category enrichment and differentially expressed genes between short-term and long-term aging is remarkable (Table 1
). Hence, we conclude that in our experimental conditions flies display, already in 24 h, expression signatures of aging.
In addition, 434 of the genes with significant gene expression changes were shared in the age
contrasts. Almost half of all genes that evolved in gene expression during 100 generations of temperature adaptation were also significantly affected by age. The majority (374, 85%) of them were downregulated during evolution but upregulated by age (Figure 3
A). A PCA and cluster analysis of these genes clearly separated the 5-day-old evolved flies from the two other groups (Figure 5
), suggesting that the expression of at least a subset of genes involved in the early aging process was delayed in flies adapting for 100 generations to the novel hot environment. These genes are enriched for several developmental processes (Table S2
), including mushroom body development (GO:0016319), eye-antennal disc development (GO:0035214), and somatic muscle development (GO:0007525). Although our results suggest that temperature adaptation is mediated by or coincides with a modified aging processes, further experiments incorporating phenotypic assays at more time points in evolved and ancestral populations are needed to explicitly investigate the evolution of aging processes in the novel temperature regime.
Our study showed that already a small difference in age of only 24 h results in significant gene expression differences. This observation not only has important implications for the design of future gene expression profiling studies but also triggers re-consideration of some published results as the mixture of flies with different age corresponds to the combination of different expression patterns (Figure 4
). If such mixtures are not well-balanced among the focal groups, this can result in false positives and negatives. Unless the distribution of age classes and their specific gene expression patterns are known, it will not be possible to correct for these mixture effects. Hence, we recommend that future studies should minimize age differences between pooled specimen. Alternatively, expression profiling of individuals with documented age may provide an analytical framework, which permits the incorporation of age as an additional factor.