Selection for cold resistance alters gene transcript levels in Drosophila melanogaster

Abstract

Microarrays have been used to examine changes in gene expression underlying responses to selection for increased stress resistance in Drosophila melanogaster, but changes in expression patterns associated with increased resistance to cold stress have not been previously reported. Here we describe such changes in basal expression levels in replicate lines following selection for increased resistance to chill coma stress. We found significant up- or down-regulation of expression in 94 genes on the Affymetrix Genome 2.0 array. Quantitative RT-PCR was used to confirm changes in expression of six genes. Some of the identified genes had previously been associated with stress resistance but no previously identified candidate genes for cold resistance showed altered patterns of expression. Seven differentially expressed genes that form a tight chromosomal cluster and an unlinked gene AnnX may be potentially important for cold adaptation in natural populations. Artificial selection for chill coma resistance therefore altered basal patterns of gene expression, but we failed to link these changes to plastic changes in expression under cold stress or to previously identified candidate genes for components of cold resistance.

Introduction

Microarray studies are widely used to identify transcript profiles of genes and pathways involved in adaptive responses, and they have proven successful in several cases particularly when expression studies are undertaken across populations adapted to different conditions (Fisher and Oleksiak, 2007, Jensen et al., 2007, Larsen et al., 2007). One approach for identifying relevant genes is to compare patterns of gene expression between selected and control lines. In Drosophila a very large study of lines selected for increased levels of stress resistance recently detected repeatable changes in the basal expression of genes in replicated lines (Sørensen et al., 2007). There was also substantial overlap among up- and down-regulated genes selected for different stresses. In particular lines selected for altered starvation and heat resistance showed strong and replicable patterns of changes in expression.

Despite the overlap in gene expression patterns among replicate lines selected for the same traits or replicated sets of individuals exposed to the same conditions, there is evidence that patterns of changes in gene expression can show limited repeatability across studies. This is particularly the case for studies that consider responses to the same types of environmental conditions. In Arabidopsis, a comparison of expression changes across studies for plants exposed to drought stress shows rather limited overlap in sets of genes with common expression patterns (Bray, 2004). An additional challenge lies in defining candidate genes whose expression patterns are consistent across platforms and laboratories (Stevens and Doerge, 2005). Low inter-study reproducibility of expression patterns limits extrapolations across populations and species in identifying adaptive gene sets (Hoffmann and Willi, 2008). It is therefore important to establish how often different sets of lines selected for the same traits show common patterns of gene expression. Traits need to be carefully defined in establishing consistent patterns, because expression patterns will depend on the types of physiological changes that underlie a trait.

In Drosophila, some measures of adult cold resistance vary in a predictable manner across species depending on the climate where they come from (Gibert and Huey, 2001, Kimura, 2004) and there is also good evidence for adaptive variation within Drosophila melanogaster for recovery from a chill coma (Hoffmann et al., 2002) as expected based on geographic patterns of cold stress encountered by insects generally (Chown and Nicholson, 2004). However, cold resistance is a complicated trait because it consists of different components that appear to be at least partly physiologically independent (Rako and Hoffmann, 2006) and show different responses to acclimation (Jensen et al., 2007) and rates of temperature change (Terblanche et al., 2007). When flies encounter cold conditions, they become incapacitated and eventually enter a coma. They can recover from this coma partly by acclimation or if temperatures increase again, but eventually cold conditions cause mortality (Rako and Hoffmann, 2006). There are sub-lethal effects of cold associated with a decrease in fecundity (Jenkins and Hoffmann, 1999). Adult mortality after cold stress is also influenced by reproductive diapause entered by adults when they overwinter in temperate conditions (Schmidt and Conde, 2006).

Despite the ecological importance of responses to cold conditions, very little is known about the genetic basis of adaptive variation in different measures of cold resistance. One early study on D. melanogaster suggested the importance of genes on the second chromosome (Tucic, 1979) and a recent effort to map variation in chill-coma recovery time located three QTL to Ch 2 and one to Ch 3 (Morgan and Mackay, 2006). Despite the location of QTL peaks, the list of candidate genes for cold resistance is small (Hoffmann et al., 2003) and mostly derived from single-gene expression studies that have identified genes upregulated under cold conditions (Goto, 2001, Sinclair et al., 2007). In contrast, multiple-gene expression studies have been unsuccessful in identifying candidate genes. Sørensen et al. (2007) investigated differences in expression levels between lines selected for cold mortality following acclimation and control lines (Bubliy and Loeschcke, 2005) and found no significant differences for any genes when lines were compared without acclimation and after they had been reared under control conditions. This suggests that basal levels of gene expression were not differentiated by selection. The physiological basis of adult female survival in Drosophila after a cold shock seems to involve an altered membrane state and change in composition of lipid components (Overgaard et al., 2006) although other traits like increased levels of proline (Misener et al., 2001) and increased trehalose and glucose sugars also appear to be important (Overgaard et al., 2007). Pupal cold resistance has also been recently linked to the expression of some heat shock proteins in a number of insects (Rinehart et al., 2007).

Here we compare gene expression patterns between replicate lines that have increased levels of resistance to cold conditions in both chill coma and mortality assays as described previously (Anderson et al., 2005). The selection regime involved a different assay than the lines tested by Sørensen et al. (2007), in that flies were selected without a prior acclimation period and were scored for recovery from a cold shock rather than for mortality, so that the level of stress experienced by the flies was lower. Divergence in the selected lines also increased survival following a cold shock (without acclimation) (Anderson et al., 2005) and influenced starvation as well as cold resistance (Hoffmann et al., 2005). We used a more recently developed Affymetrix gene array than the one used by Sørensen et al. (2007). We tested whether there was clustering of genes because clustering has previously been found in Drosophila subobscura genes differentially expressed in populations cultured for 3 years at different non-stressful laboratory temperatures (13 °C, 18 °C and 22 °C) (Laayouni et al., 2007). The existence of close chromosomal clusters of similarly expressed gene in Drosophila is now well established (Spellman and Rubin, 2002, Kalmykova et al., 2005) and is a general phenomena in eukaryotic genomes (Michalak, 2008).