1Laboratory studies on genetically modified strains may reveal important information on mechanisms involved in coping with thermal stress. However, to address the evolutionary significance of specific genes or physiological mechanisms, ecologically relevant field tests should also be performed.
2We have tested the importance of inducible heat shock proteins (Hsps) under different thermal conditions using two heat shock factor (Hsf) mutant lines (either able (Hsf+) or unable (Hsf0) to mount a heat stress response) and an outbred laboratory adapted wild-type line of Drosophila melanogaster under both laboratory and field conditions.
3In the field, there was a tendency towards better performance of Hsf+ flies relative to Hsf0 flies, but as compared with wild-type the performance of both mutant lines was very low.
4In the laboratory tests, Hsf+ flies had higher heat knock-down resistance relative to Hsf0 flies but in other assays on heat, cold and desiccation resistance there was either no difference between the two mutant lines or the Hsf0 line had higher performance. Also, the superiority of the wild-type flies under field conditions was trait specific.
5The results emphasize that the ecological relevance of specific molecular mechanisms should be tested under a range of conditions both in the laboratory and in the field. Genetically modified lines cannot be assumed to represent the performance of natural populations, especially for field and/or ecologically relevant studies.
6As evident in this study, ideal controls and adequate replication of genetically modified strains can be difficult to obtain. Thus, caution is needed when interpreting results comparing the performance of genetically modified lines with that of control lines.
All organisms face challenges imposed by their environment. Thus, the ability to respond and adapt to environmental conditions that are often fluctuating is crucial. Temperature is one such environmental factor that is considered important for the distribution and density of populations and species (Cossins & Bowler 1987; Hoffmann et al. 2003). The study of how organisms adapt to temperature, how climatic conditions impact on the evolution of different strategies to cope with the climate and the costs and benefits of these mechanisms is an important part of understanding the evolutionary process of thermal adaptation.
One seemingly adaptive response is the heat stress response, which involves a dramatic change in physiology and gene expression after heat hardening (Sørensen et al. 2005; Malmendal et al. 2006) and other stress exposures (Sørensen et al. 2003). Of specific importance for heat resistance is the up-regulation of heat shock proteins (Hsps). The Hsps provide an efficient defence and recovery system against the detrimental effects of protein denaturation caused by heat stress (Parsell & Lindquist 1993; Feder & Hofmann 1999), and their expression is important for heat shock survival and other stress and fitness related traits (Feder & Hofmann 1999; Sørensen et al. 2003). However, these findings are based on laboratory investigations of a few traits such as mortality after exposure to extreme thermal conditions, which animals may not be likely to experience in nature. The assumed ecological role of Hsps is thus mainly based on observations from laboratory studies, so there is a need to strengthen the links between laboratory and nature to study their importance in the field (Gibbs 1999; Harshman & Hoffmann 2000). A few studies have attempted to do this using Drosophila. Larvae and pupae express Hsp70 in nature when exposed to high and potentially lethal temperatures in sunlit fruits (Feder, Blair & Figueras 1997). Also, by inserting a heat shock promoter with a reporter gene in a genetically modified fly line Feder, Roberts & Bordelon (2000) showed that free ranging adult D. melanogaster expressed Hsp70 during hot days, but less so than caged sun exposed adults. Thus, while some individuals were heat exposed, the majority probably behaviourally avoided high stressful temperatures in the field. Therefore, there is a need to investigate the importance of Hsps both under field conditions and using multiple assays in the laboratory which may have more ecological relevance (Feder et al. 2000; Sørensen et al. 2003; Fasolo & Krebs 2004; Jørgensen et al. 2006).
New molecular techniques have opened the possibility to successfully perform functional genomics studies, which aim at identifying and describing adaptive variation from DNA to phenotype. Much progress has been made in identifying potential candidate genes and their phenotypes using knock-out or extra gene copy number stocks or by using RNAi methods (Krebs & Feder 1998; Morgan & Mackay 2006). However, the use of genetically engineered and mutant flies has potential disadvantages. Genetically modified strains may not behave as do natural populations, even in traits not directly related to the gene targeted in the modification. Similar issues are true for strongly laboratory adapted or highly inbred strains (Baldal et al. 2006; Terblanche & Chown 2007). Thus, interpretations of ecological or evolutionary significance based on laboratory results from such stocks should be done with caution.
Here we test the role of the heat shock response under natural conditions by using two genetically modified lines (Jedlicka et al. 1997). The heat stress induction of Hsps is controlled by a common transcription factor, the heat shock factor (Hsf), which is activated by the presence of denatured proteins in the cell and controls several inducible Hsps in Drosophila (Ananthan et al. 1986; Voellmy 2004). The mutant lines used were: one that has a mutation in the hsf gene which makes the gene product completely non-functional at temperatures above 30 °C and a control line with a similar genetic background containing a wild-type hsf copy (Hsf+). We use a release-recapture approach to estimate a component of fitness in the field (Markow & Castrezana 2000; Loeschcke & Hoffmann 2007; Kristensen, Loeschcke & Hoffmann 2008b). Briefly, the flies were released in a habitat with no natural resources and the ability to locate artificial resources provided was used as a fitness estimate. As D. melanogaster naturally utilizes ephemeral resources such as soft fruits, we consider this ability as closely related to fitness. To expand our knowledge of responses to thermal stress and the role of the stress response system, we assayed performance under different thermal field and laboratory conditions. To test for a possible effect of genetic modification, we included an outbred, laboratory adapted wild-type population with a different genetic background. The expectations were that all lines would show similar performance in traits where Hsps are not known to be important. Moreover, the wild-type and Hsf+ flies were expected to behave similarly, but superior to the Hsf0 in traits where the expression of Hsps is thought to be important for coping with environmental conditions.
Materials and methods
maintenance and origin of experimental flies
Three lines of D. melanogaster were used in this study: two genetically modified lines (obtained from the Bloomington Stock Center; stock number 5489 and 5490, respectively) created by Jedlicka et al. (1997) and one outbred laboratory control line (C1 line described in Bubliy & Loeschcke 2005). The genetically modified line Hsf0 (hsf 4 in Jedlicka et al. 1997) is a heat-sensitive mutant line harbouring a mutation (V57 M) in the hsf gene. The mutation completely inactivates the transcription factor at temperatures above 30 °C. Presumably, the transcription factor is fully functional at 25 °C and lower temperatures. The corresponding control line Hsf+ was derived from Hsf0, but has a normal hsf gene inserted and maintained against a balancer chromosome. The genetic background of the two lines is slightly different for the third chromosome (see Jedlicka et al. 1997 for details). In the Hsf+ line the heat shock response is fully functional and Hsps are induced normally (Jedlicka et al. 1997; Nielsen et al. 2005). The unselected wild-type line C1 (which has been kept at a high population size in the laboratory for several years (Bubliy & Loeschcke 2005)) served as an additional control. All lines were maintained under standard laboratory conditions (25 °C, agar, sugar, yeast and oatmeal medium, 12/12 h light–dark cycle) in 200 mL bottles with 30 pairs per bottle. For each experiment, we collected 0–1 day-old flies that were sexed under CO2 anaesthesia, and transferred to food vials at a density of 20 flies per vial until the experiments were performed (flies were used for one experiment only).
Field release-recapture experiments were done as described by Kristensen et al. (2008a). Briefly, three releases were undertaken at cool conditions (C1 was not tested in these releases) and two at mildly warm conditions (Table 1). Capture points in all releases were 5 m apart starting 5 m from the release site and extending up to 20 m in opposite directions from the release point. At each capture point three resource buckets (containing mashed bananas) were placed 3–4 m apart running perpendicular to the release line. Temperatures were recorded with data loggers (Tinytalk II, Chichester, UK) positioned close to the release site (Table 1). Flies were transported to the release sites in insulated styrofoam boxes kept at 24–26 °C. Just before the release, groups of 200 4–5 days-old flies were transferred into vials with 0·0015 (± 0·0005) g of fluorescent micronized dust (Radiant Corp., Richmond, CA). Dust colours were randomly assigned to the lines, and changed between releases. Vials with flies from the different lines were randomly arranged in a container and foam stoppers were removed. Only few flies (< 1%) never left the release vials and this did not differ between lines (results not shown). All releases took place in woodland, consisting mainly of spruce trees (cool releases) and a blend of pine and oak trees (mildly warm releases). Flies were captured from resources by netting and/or aspirating, starting 1 h after release and then every hour from 1200 until 1700 for 4 days. This time period represented peak activity hours in the field (only very few flies were observed in the buckets before 1200 and after 1700 on all 4 days of recapture). Captured flies were held on ice in the field to knock them out so that transfer of dust between flies was minimized. They were then transported to the laboratory and stored in a freezer until colours were scored under ultraviolet light.
Table 1. Details on release identification, number of flies released, percent captured and temperatures in the five releases performed. Lines were a wild-type control line (C1) and two mutant lines, a rescued Hsf mutant line (Hsf+) and a heat sensitive Hsf mutant line (Hsf0)
Number of flies released (per line)
Percentage captured (wild-type)
Time released (day 0)
Time of last capture (day)
R1 – mild
R2 – mild
R3 – mild
R4 – warm
Hsf+, Hsf0, C1
R5 – warm
Hsf+, Hsf0, C1
heat coma assay
Heat resistance was measured as time to entering heat coma in 4–5 day-old male and female flies set up individually in glass vials closed with plastic screw caps and heat stressed by submerging the vials in a re-circulating water tank (Rako et al. 2007). Flies were exposed to either 38·5 °C (with no prior hardening) or 39 °C (with hardening). The heat hardening treatment consisted of exposure to 35 °C for 1 h followed by recovery at 25 °C for 1 h. Twenty flies were tested per line and sex. The time for the fly to be rendered completely inactive was recorded as its ‘heat coma time’.
stimulated activity in maze
Stimulated activity at benign (25 °C) and hot (36 °C) conditions was tested by placing flies into a maze positioned horizontally. The maze consists of a large number of crossings (see Hirsch 1956; Coyne & Grant 1972 for further details). Vials with food were attached at one end and the flies were released at the other end without resources. In order to reach the food vials, the flies must migrate through the maze. Fifty 4–5 day-old flies of each sex and line were tested simultaneously. Flies from the three lines were marked with distinct fluorescent colours (as in the field releases) immediately prior to the test. Three replicates were run at 25 °C and at 36 °C, respectively. Flies completing the maze were collected from the food vials every hour starting 1 h after the beginning of the test and continuing for 5 h. A few flies died (probably due to handling effects) during the tests (< 5%) and these where discarded from the data set.
chill coma recovery
Cold resistance was assessed using a chill coma recovery assay (for details see Hoffmann, Anderson & Hallas 2002). We estimated cold resistance as the time (scored at 25 °C to the nearest second) needed for flies to recover (defined as the ability to stand) after being knocked out by exposure to –2 °C for 80 min. Flies were exposed to cold temperatures in empty 5 mL sealed plastic vials that were submerged in a water bath. Twenty-five 5 day-old flies were scored per sex and line.
Desiccation resistance was scored by exposing flies to < 5% RH in a desiccator at 25 °C. Forty 4–5 day-old flies per sex and line were assessed individually in empty 10 mL glass vials covered with gauze. Mortality was scored at hourly intervals until all flies were dead.
For release-recapture data we tested whether the total number of flies caught differed between treatments and sexes. Capture data was treated as categorical and analysed separately for each release. The difference between treatments and sexes in capture success was assessed using logit models as described by Kristensen et al. (2008a). To test the relative capture success of males vs. females and each line vs. the other lines, we computed the likelihood of capture (males and females tested separately due to sex and/or line by sex effects in the logit models). For this measure of relative capture success, values above 1 indicate that more males than females were caught.
All laboratory traits were analysed by anova using line and sex as fixed factors. For stimulated activity rate the data were calculated as proportion of flies completing the maze and these were arcsin-square-root transformed to improve homogeneity of variances. Post hoc comparisons of lines within sex were performed with Scheffe's test.
The percentage of flies captured in three releases at cool conditions varied between 4·5% and 6·9% (Table 1). In all releases, flies in the two directions were combined (e.g. flies caught 5 m from the release point in each direction). In releases R1 and R3, mostly Hsf+ flies were caught. In R2 the numbers of Hsf+ and Hsf0 caught flies were not significantly different, even though the tendency was in the same direction as for R1 and R3 (Fig. 1a; Table 2). There was no consistent difference in the likelihood of capture of males and females. In the releases performed at mildly warm conditions (R4 and R5) only 0·5% and 0·3% of the mutant Hsf flies were caught, whereas around 50% of the wild-type control flies were caught (Fig. 1b; Table 1). No significant difference in relative capture success between Hsf+ and Hsf0 flies was found, though more Hsf+ flies tended to be caught (Table 2).
Table 2. Logit models testing the effects of line and sex on the probability of recapture under cool (R1–R3) and mildly warm (R4–R5) field temperatures
Relative capture success
Line (1 or 2)
Line by Sex (2)
M vs. F
M: Hsf+ vs. Hsf0
F: Hsf+ vs. Hsf0
M: Hsf+ vs. C1
F: Hsf+ vs. C1
M: Hsf0 vs. C1
F: Hsf0 vs. C1
Relative capture success (relative risk) comparing the likelihood of capture of males (M) vs. females (F) and Hsf+ vs. Hsf0 flies, Hsf+ vs. C1 control flies and Hsf0 vs. C1 control flies are presented. Values > 1 indicate more flies caught of the first mentioned sex or line. Significance is indicated by asterisks (*P < 0·05, **P < 0·01; ***P < 0·001).
Non-heat-hardened males were more resistant to heat coma than similarly-treated females (Fig. 2a; Table 3). Females from the C1 line seemed slightly, but not significantly, more resistant to heat coma than females from either mutant line, which did not differ from each other. In contrast, Hsf0 males were significantly less resistant than the other two lines (Fig. 2a). After a hardening treatment, there were significant differences between lines (C1 > Hsf+ > Hsf0), but not between males and females (Fig. 2b; Table 3).
Table 3. Summary Mean Squares from anova analysis and results of post hoc comparisons of lines within sex (Scheffe's test) of stress resistance traits measured in a wild-type control line (C1) and two mutant lines, a rescued Hsf mutant line (Hsf+) and a heat sensitive Hsf mutant line (Hsf0)
Chill coma recovery
Significance is indicated by asterisks (*P < 0·05, **P < 0·01, ***P < 0·001).
For measures of stimulated activity at 25 °C, females were more active than males; C1 flies were most active and Hsf+ flies least active (Fig. 3a; Table 3). In the post hoc comparisons, only C1 females were significantly different from the two other lines and no significant differences between lines were found in males. At 36 °C there were no significant effects of line, sex or their interaction (Fig 3b; Table 3), though there was a tendency for the Hsf+ flies to be least active.
chill coma recovery
Lines and sexes differed in recovery time from chill coma (Fig. 4; Table 3). Females generally recovered faster than males. C1 wild-type and Hsf+ females showed similar, significantly faster recovery than Hsf0 females, while C1 wild-type males recovered significantly faster than both Hsf+ and Hsf0 line males.
Desiccation resistance differed between lines and among sexes (Table 3). Females were more desiccation resistant than males, and C1 wild-type flies were more resistant than both mutant lines (Fig. 5). Post hoc comparisons showed that desiccation resistance in the C1 wild-type line was significantly higher than in both Hsf+ and Hsf0 lines. For females, the two Hsf mutant lines did not differ from each other, but for males, desiccation resistance was significantly higher in the Hsf0 line than in the Hsf+ line.
The most striking result of this study was the extremely poor performance of both Hsf mutant lines in the field assay. This was not shown by any of the laboratory assays, not even those where the traits tested in the laboratory seemed fairly similar to the field assay, for example, the activity assay in the laboratory. Thus, no clear patterns emerged with regard to the role of Hsf under field conditions or to the relation among lines or traits (see Table 3 for an overview of post hoc comparisons). Even though the wild-type line performed better than the Hsf mutant lines in most cases, either mutant line performed as well or better than the wild-type line for some traits. Also, even though the Hsf0 line showed decreased heat resistance as expected, especially when hardened, this was not exhibited as decreased performance in activity tests or field performance at higher temperature, where this difference could be expected to be most clear. Thus, performance was trait and often also sex specific, and conclusions drawn from each assay individually could lead to very different interpretations as to the effect of the Hsf mutantion. These results are in line with studies on butterflies (Kingsolver 1999), and several other Drosophila species (Jaenike et al. 1995) showing a mismatch between laboratory and field observations.
The releases performed under cool conditions were intended to serve as a control for the planned releases under warmer conditions. In two of the three releases we caught slightly more flies from the rescued Hsf+ line, but the re-capture rates (around 5%) were in all cases much lower than has been found in earlier release studies under similar conditions (e.g. 10–30% flies recaptured over 2 days, Kristensen et al. 2008b). Thus, in all the following assays we included an outbred laboratory wild-type line for comparison and evaluation of whether the observed effects could be attributed to the heat sensitive hsf mutation directly or to an artefact of the genetic engineering of both Hsf mutant lines. Under mildly warm conditions, we recaptured an even lower proportion of the Hsf mutant lines (around 0·5%), but with a tendency for the Hsf+ line to perform slightly better. However, this difference was negligible as compared with the difference between Hsf mutant line and the C1 wild-type: just above 50% of the C1 flies were caught at the baits (around 100 times more than either Hsf mutant line). Possibly temperatures conditions were not high enough to induce Hsps or to inactivate Hsf in the Hsf0 lines. However, the temperatures reported were measured in the shade, and probably underestimate the conditions perceived by the flies as the baits at which the flies were caught were positioned in sun exposed sites between trees. Thus, the ‘true’ temperatures probably could be considered as mildly stressful and high enough to induce Hsps in the flies. Nevertheless, the results from the five releases did not confirm the hypothesis that Hsf+ flies would perform better relative to Hsf0 flies at higher temperatures only.
Based on the heat coma assay used in this study, we found that without prior hardening the Hsf+ line performed as well as the C1 wild-type control whereas the Hsf0 line performed worse. After hardening the Hsf+ line performed worse than the C1 wild-type. This suggests that the hardening treatment induced a larger cost in the Hsf+ line than in the C1 wild-type line. Thus, even though that the Hsf0 line performed worst, especially after hardening – due to the cost of heat hardening without receiving any benefits in form of Hsps – Hsf mutants in general show an about equal negative effect. Using a heat shock survival assay, Nielsen et al. (2005) found that resistance of the Hsf+ line equalled that of wild-type D. melanogaster strains (as also found in a similar assay by Sørensen & Loeschcke 2002). The heat coma assay used here has been shown to differ from both the tube heat knock-down assay (described by Huey et al. 1992) and heat shock mortality assays (Hoffmann et al. 2003) and at present the role of Hsps for heat coma resistance as measured here is unclear.
The Hsf+ line performed worst in the maze. The Hsf0 line was outperformed by the C1 wild-type at 25 °C, but this difference seemed to have disappeared at 36 °C where a cost of the heat sensitive Hsf mutation was otherwise expected. We expected that maze activity could be a laboratory alternative for field activity and results partly verified this. The C1 wild-type does not seem to be as superior in this test as in the field performance test, but this may be partly explained by the fact that most flies completed the maze in the time allowed. Inspection of the data from the individual collections (hourly) shows that the C1 wild-type flies made it through the maze faster than the other lines. Thus if collections were performed at shorter intervals a larger benefit of the C1 line would have been observed. That the Hsf0 line did better than the Hsf+ was not surprising, as earlier studies of this line have reported an increased activity level of the Hsf0 line, especially after heat hardening (Nielsen et al. 2005).
Several studies have investigated the connection between Hsp induction and cold exposure and cold adaptation. While some studies have found induction of Hsps after cold exposures, at least in Drosophila, this seems to be restricted to long-term exposures (Burton et al. 1988; Sejerkilde et al. 2003). Thus, no expression of inducible Hsps was expected in this assay. This is also the case for desiccation resistance. Even though the flesh fly Sarcophaga crassipalpis expresses two inducible Hsps (Hsp23 and Hsp70) in response to desiccation (Hayward et al. 2004), no involvement of inducible Hsps in response to desiccation has been shown in Drosophila. There were small differences between the two mutant lines for some traits. However, the most consistent pattern was the better performance of the C1 wild-type. Generally, there was no indication of a negative effect of the Hsf0 line, but a general negative effect of both mutant lines as compared to the C1 wild-type control.
Release–recapture assays have successfully been used, for example, to study dispersal in species of Drosophila (Markow & Castrezana 2000), the importance of acclimation (Loeschcke & Hoffmann 2007; Kristensen et al. 2008a) and the importance of wing size and parasite load (Jaenike et al. 1995; Kingsolver 1999). In this study, we had to extend the period of recapture to 4 days to increase the number of flies caught from the Hsf mutant lines. Still, very few flies were caught. This could not have been predicted from standard laboratory thermal resistance tests, where the mutant lines often perform equally or nearly equally as well as the C1 wild-type for some traits and worse for other traits for which Hsps are not expected to be important. Thus, most of the effects observed here are not attributable to the Hsf heat sensitive mutation and lack of inducible Hsps, but seem to be an artefact of the breeding scheme and genetic modifications used to create these lines.
Many studies of genetically modified lines suffer – like this study – from lack of true replication. The modification should ideally be crossed into an outbred background to verify the effect of the modified gene. Further, controls should ideally be created in the same backgrounds. Wild-type laboratory lines are not appropriate universal controls for the genetic modification, as the results of this study strongly suggest. Genetically engineered Drosophila have been used – seemingly successfully – to investigate costs and benefits of extracopy over-expression of Hsp70 (e.g. Krebs & Feder 1998), and to evaluate the natural expression of Hsp70 in free ranging flies (Feder et al. 2000). However, others have warned that modified or strongly inbred laboratory strains might yield misleading results, as the expression of certain phenotypes and field performance is strongly dependent on genetic background and specific conditions used (Baldal et al. 2006; Terblanche & Chown 2007; Kristensen et al. 2008b). Thus, based on these reports and the results of this study we argue that there are good reasons for and a need to utilize the possibilities of modern technology in ecological studies including the study of thermal adaptation under natural conditions. However, we also point to the many pitfalls this approach might have and recommend that such studies should always include appropriate wild-types and alternative traits, because genetically modified and strongly inbred or laboratory adapted lines might have unpredictable effects on many traits.
Authors are grateful to the Danish Research Council for financial support with frame and centre grants (VL) and postdoc positions (TNK, JGS) and to Doth Andersen and Kjeld Nygaard Kristensen for excellent help with the field work and to J.S.F. Barker and two anonymous referees for helpful comments to the ms. JGS acknowledges the hospitality of Centre of Invasion Biology (CIB), University of Stellenbosch while writing the first draft of the manuscript.