Knocking down expression of Hsp22 and Hsp23 by RNA interference affects recovery from chill coma in Drosophila melanogaster

Insects have evolved a range of molecular adaptations to cope with seasonal exposure to stressful (high and/or low) temperatures (Doucet et al., 2009). Heat shock proteins (Hsps) are considered prime candidates for thermal tolerance and adaptation in organisms, including the vinegar fly Drosophila melanogaster (Feder and Hofmann, 1999; Hoffmann et al., 2003; Sørensen et al., 2003; Michaud and Denlinger, 2010). Most of the focus of research on these proteins has been on their role in providing heat resistance, while their potential role in non-freezing cold-stress resistance has received less attention (Norry et al., 2007; Sørensen and Loeschcke, 2007), with the exception of Hsp70 (Michaud and Denlinger, 2004; Sørensen and Loeschcke, 2007; Clark and Worland, 2008). Recently, the effect of low temperatures and diapause on other Hsps has been examined in insects. It appears that a wealth of additional Hsps are responsive to both low temperature and diapause, particularly genes/proteins from the small heat shock protein family (sHsp) (Qin et al., 2005; Li et al., 2007; Rinehart et al., 2007; Huang et al., 2009; Colinet et al., 2010a; Michaud and Denlinger, 2010). A key feature of the response to heat shock is its suppression following the restoration of normal environmental conditions (Parsell and Lindquist, 1993), whereas the response to cold stress is generally observed during the recovery phase (Colinet et al., 2010a; Colinet and Hoffmann, 2010). The molecular mechanisms behind recovery from cold stress are complex and it seems that more genes/proteins are activated during the recovery phases than during the period of the cold stress itself (Colinet et al., 2007; Clark and Worland, 2008).

RNA interference (RNAi)-mediated gene silencing is a powerful tool for exploring gene function. So far only a few studies have used this method to understand how specific genes respond to cold stress in insects. Rinehart et al. (Rinehart et al., 2007) found that suppression of Hsp23 and Hsp70 expression by RNAi resulted in a loss of cold tolerance in the flesh fly Sarcophaga crassipalpis. Furthermore, Kostál and Tollarová-Borovanská reported that RNAi targeting Hsp70 negatively affected repair of chilling injuries in the firebug Pyrrhocoris apterus (Kostál and Tollarová-Borovanská, 2009). Finally, Colinet et al. found that silencing Frost expression impaired recovery from chill coma in D. melanogaster (Colinet et al., 2010b).

Drosophila melanogaster is a useful model for understanding the molecular basis of thermal adaptations, as it is found in a range of different thermal environments. Previous results on adult flies suggest that upregulation of cold-responsive Hsps during recovery from cold stress might be related to some undefined repairing functions (Colinet et al., 2010a), supporting the ideas of Kostál and Tollarová-Borovanská (Kostál and Tollarová-Borovanská, 2009). In D. melanogaster, there are 11 sHsp genes (Li et al., 2009), although only four members (Hsp22, Hsp23, Hsp26 and Hsp27) have been studied in detail (e.g. Joanisse et al., 1998; Michaud et al., 2002; Morrow et al., 2006) and these are all upregulated during recovery from cold temperatures (Colinet et al., 2010a). In this study we used the GAL4/UAS system to knock down expression of two of these genes, Hsp22 and Hsp23. We tested whether suppression of this upregulation response affects recovery ability after a prolonged chill coma.

RNAi-mediated gene silencing was achieved using the GAL4/UAS system (Duffy, 2002). The UAS-Hsp lines were obtained from the Vienna Drosophila RNAi Center (Hsp22, transformant ID 43632; Hsp23, transformant ID 111816) (Dietzl et al., 2007). The actin5C-GAL4 line (Bloomington Drosophila Stock Center, #4414) was used to drive the expression of the UAS-Hsp and resulted in ubiquitous Hsp mRNA knockdown. Progeny (act-GAL4/UAS-Hsp) were tested in cold recovery assays. To control for genetic background effects, the same GAL4 driver line was crossed to the w¹¹¹⁸ line (from the Bloomington Drosophila Stock Center) and their progeny (act-GAL4/+) was assayed alongside their act-GAL4/UAS-Hsp counterparts. Fly stocks were maintained in 250 ml bottles in uncrowded conditions. Bottles were kept at 25°C, 70% relative humidity and continuous light on a standard fly medium as previously described (Hoffmann and Shirriffs, 2002).

To verify the extent of gene silencing, Hsp mRNA levels were measured in: (1) untreated flies, kept at 25°C (i.e. basal expression), and (2) treated flies, recovering for 2 h at 25°C after 12 h of cold stress at 0°C (i.e. during Hsp upregulation). RNA extractions were performed using the RNeasy RNA extraction kit and the RNase-Free DNase Set (Qiagen, Doncaster, VIC, Australia), as described by Colinet et al. (Colinet et al., 2010a). cDNA was synthesized using the Superscript III First-Strand Synthesis System (Invitrogen, Mulgrave, VIC, Australia), according to manufacturer's instructions. Hsp primers were designed with the Primer3 module (http://biomanager.info/) [Hsp22, 5′-GCCTCTCCTCGCCCTTTCAC-3′ (forward) and 5′-TCCTCGGTAGCGCCACACTC-3′ (reverse); Hsp23, 5′-GGTGCCCTTCTATGAGCCCTACTAC-3′ (forward) and 5′-CCATCCTTTCCGATTTTCGACAC-3′ (reverse)]. Quantitative real-time PCR (qRT-PCR) was performed on the LightCycler 480 system (Roche Diagnostics, Castle Hill, NSW, Australia) following the method previously described (Colinet and Hoffmann, 2010). The percentage knockdown (i.e. relative expression ratio) was calculated using the 2^–ΔΔCt method (Livak and Schmittgen, 2001). The ratio of target gene (Hsp) expression in the act-GAL4/UAS-RNAi line relative to in the act-GAL4/+ control line was calculated and normalized using the housekeeping reference gene RpS20 [primers described by Colinet et al. (Colinet et al., 2010a)]. Three biological replicates of 20 males each were used and ratios were compared with control values using Student's t-test.

All tests were performed using synchronized 4-day-old males, sexed visually without CO₂ anaesthesia using an aspirator. Experiments were carried out only on males, in order to relate to expression data of a previous experiment that found that Hsp22 and Hsp23 are upregulated during recovery from cold stress in adult males (Colinet et al., 2010a).

Flies were placed in 42 ml glass vials immersed in a 10% glycol solution cooled to 0°C for 12 h (chill coma) before being returned to 25°C to recover. To test whether suppression of Hsp gene expression affects recovery abilities, we used three different measures based on the method previously described (Colinet et al., 2010b). Briefly, ‘short-term recovery’ compared recovery times (i.e. the time to stand up) at 25°C. Recovery curves were compared between RNAi and control lines using Mantel–Cox analysis with a censoring factor for individuals that did not recover at the end of the experiment. Forty-five flies were monitored for each line. ‘Medium-term recovery’ assessed climbing activity during a period of 8 h following the end of cold stress. In this negative geotaxis assay, males were individually transferred to a 9.5 cm plastic vial and the height reached within 7 s after a mechanical stimulation was noted. Flies were divided into three categories: (a) ‘injured’, no climbing; (b) ‘recovering’, slow climbing without reaching the top of the vial within 7 s; (c) ‘fit’, fast climbing and reaching the top of the vial within 7 s. The time of observation was chosen based on preliminary assays (see Colinet et al., 2010b). This test was performed repeatedly on the same individuals after 2, 4, 6 and 8 h of recovery at 25°C. Chi-square contingency tests were carried out to compare numbers of flies in the three categories between RNAi and control lines. Seventy flies were tested for each line. Flies were maintained on food during this period.

Finally, ‘long-term recovery’ measured mortality 24 h after the end of the cold stress. Mortality rates were calculated based on 150 flies per line. Chi-square contingency tests were used to compare mortality rates between RNAi and control lines. All statistical tests were performed using Prism V 5.01 (GraphPad Software Inc., 2007).

The mRNA level was significantly reduced in the act-GAL4/UAS-Hsp22 line compared with in the act-GAL4/+ control line in both untreated (t=75.36, d.f.=2, P=0.001) and recovering (t=557.3, d.f.=2, P=0.001) flies (Fig. 1A). The percentage knockdown reached 84% in untreated males and 92% in treated males. The level of expression was also significantly reduced in the act-GAL4/UAS-Hsp23 line compared with in the act-GAL4/+ control line in both untreated (t=22.14, d.f.=2, P=0.0002) and recovering (t=141.9, d.f.=2, P=0.001) flies (Fig. 1B). The percentage knockdown reached 52% in untreated males and 90% in treated males.

Short-term recovery was significantly affected in the act-GAL4/UAS-Hsp22 line compared with in the act-GAL4/+ control line (Fig. 2A), resulting in significantly different recovery curves (χ²=5.30, d.f.=1, P=0.021). After 60 min, all the act-GAL4/+ flies had recovered, whereas 24% of the act-GAL4/UAS-Hsp22 flies remained in coma. The short-term recovery was also significantly affected in the act-GAL4/UAS-Hsp23 line compared with in the act-GAL4/+ control line (χ²=24.69, d.f.=1, P=0.001; Fig. 2B). After 80 min, all the act-GAL4/+ flies had recovered, whereas 24% of the act-GAL4/UAS-Hsp23 flies still had not recovered. All flies eventually recovered and no mortality was observed at the end of the experiment.

The medium-term recovery tests revealed significant differences in mobility (climbing) between the act-GAL4/UAS-Hsp22 and act-GAL4/+ lines (Fig. 3A). Differences were manifested after 2 h (χ²=7.083, d.f.=2, P=0.029) and 4 h (χ²=10.397, d.f.=2, P=0.006) of recovery, with, respectively, 74% and 40% of flies still injured in the act-GAL4/UAS-Hsp22 line compared with 52% and 15% in the act-GAL4/+ line. The proportions within each category (fit, recovering, injured) were similar between the act-GAL4/UAS-Hsp22 and act-GAL4/+ lines (Fig. 3A) after 6 h (χ²=2.917, d.f.=2, P=0.233) and 8 h (χ²=5.629, d.f.=2, P=0.06) of recovery.

There was initially (after 2 h recovery) no difference in the proportion of injured and recovering flies between the act-GAL4/UAS-Hsp23 and the act-GAL4/+ lines (χ²=2.386, d.f.=1, P=0.122) (Fig. 3B). In the act-GAL4/+ line, flies recovered progressively with an increasing proportion of fit and a decreasing proportion of injured flies. By contrast, flies from the act-GAL4/UAS-Hsp23 line showed a reduction in their recovery ability with time, which resulted in significant differences between the two lines after 4 h (χ²=15.586, d.f.=2, P=0.001), 6 h (χ²=18.186, d.f.=2, P=0.001) and 8 h (χ²=10.096, d.f.=2, P=0.006) of recovery (Fig. 3B).

In the long-term recovery assay, no difference in mortality was observed between the act-GAL4/UAS-Hsp22 and the act-GAL4/+ lines (χ²=0.667, d.f.=1, P=0.414), with very low mortality rates in both lines (Fig. 4A). Similarly mortality rates were low and similar between the act-GAL4/UAS-Hsp23 and the act-GAL4/+ lines (χ²=3.447, d.f.=1, P=0.067; Fig. 4B).

In D. melanogaster, chill coma starts around 7°C as a result of neuromuscular dysfunctions (Hosler et al., 2000). At low temperature, chilling injuries accumulate because of various physiological perturbations, such as thermotropic damage to membranes, complex metabolic disorders, ion homeostasis imbalance and oxidative stress (reviewed by Chown and Terblanche, 2006; Lee, 2010). Within limits, these physiological damages are reversible (Chown and Terblanche, 2006), but an understanding of the molecular mechanisms behind recovery is lacking.

The present results suggest that members of the sHsp gene family are involved in the recovery process. Some sHsp genes are known to be upregulated during the recovery from cold stress in D. melanogaster but are not modulated during the cold stress itself (Colinet et al., 2010a). In this study we have now shown that knocking down the upregulation response of Hsp22 and Hsp23 by RNAi affects recovery ability. In the case of Hsp22, the time to recover was slightly affected and individuals displayed reduced mobility after 2 and 4 h of recovery when compared with controls. However, after 6 h of recovery the Hsp22-knockdown flies were as active as their control counterparts. This suggests that the kinetics of the chill coma recovery process was impaired, at least during the first 4 h period immediately after the cold stress. In the case of Hsp23, recovery time was also affected, and the reduction of mobility parameters was still manifested even after 8 h of recovery. In general, the phenotypic consequences of Hsp23 knockdown were more severe than those of Hsp22 knockdown. Given that the transcript knockdown efficiencies were similar (90–92%) for both genes, our results imply that Hsp23 might be more strongly involved in cold recovery than Hsp22. However, no mortality effect was observed for either gene. Our findings show that upregulation of Hsp22 and Hsp23 is required for recovery from prolonged chill coma, at least during the early stages of the recovery process (0–8 h).

The sHsps represent the least conserved subfamily of Hsp genes, and most of the sHsp genes are species specific (Li et al., 2009), so that direct comparison between insects must be made with care. Nevertheless, there are similarities between S. crassipalpis and D. melanogaster; both of these dipterans express Hsp70 and Hsp23 during diapause, whereas Hsp90 shows an opposite pattern of expression (Rinehart and Denlinger, 2000; Rinehart et al., 2007; Baker and Russell, 2009). Suppressing the expression of Hsp23 in S. crassipalpis by RNAi did not alter the tendency to enter diapause but, as in the case of D. melanogaster, there was a negative effect on cold tolerance.

Most organisms have multiple sHsps but their precise identities are often unknown, which makes any functional investigation difficult (Michaud and Denlinger, 2010). Although the identities of D. melanogaster sHsps are known, the question as to why there are several structurally similar sHsps remains unclear. Hsp22 and Hsp23 are coordinately upregulated during/after heat stress (Bettencourt et al., 2008) and after cold stress (Colinet et al., 2010a). During development, each sHsp gene shows specific stage and cellular patterns of expression without coordination (Michaud et al., 1997; Michaud et al., 2002; Morrow and Tanguay, 2003). This might denote a common function(s) under stress conditions but a more specific role(s) during development. In spite of this, Hsp22 and Hsp23 show different chaperone activity efficiencies, which suggests different modes of action (Morrow et al., 2006). Moreover, overexpression of Hsp22 and Hsp23 both lead to increased lifespan but with different intensity, supporting the idea of specific modes of action (Morrow and Tanguay, 2003; Morrow et al., 2004).

The fact that RNAi directed against Hsp22 affected recovery less than RNAi directed against Hsp23 suggests that the two genes contribute to cold tolerance in slightly different ways. The two chaperone products of these genes differ in their subcellular localization, Hsp22 being found in the mitochondria and Hsp23 in the cytoplasm (Joanisse et al., 1998). It is not known whether the proteins Hsp22 and Hsp23 have specific or overlapping activities during the recovery from cold stress, and which functions are required to maintain proper recovery ability. Increased Hsp22 and Hsp23 mRNA levels correlate with increased stress resistance in selected D. melanogaster lines (Kurapati et al., 2000). A body of evidence supports the idea that Hsp22 is an oxidative stress response gene (e.g. Bhole et al., 2004; Morrow et al., 2004). Both Hsp22 and Hsp23 are involved in protection from the disturbance of normal redox state in D. melanogaster. Indeed, Hsp23 plays important role in hypoxia tolerance (Azad et al., 2009), whereas Hsp22 has a protective role in hyperoxia (Gruenewald et al., 2009). Maintenance of the oxidant equilibrium might be important during recovery, especially because perturbation of the antioxidant system is known to be related to chilling injury in insects (Rojas and Leopold, 1996; Grubor-Lajsic et al., 1997; Jing et al., 2005). In addition, Hsp22 expression in motoneurons allows the maintenance of locomotion activity (Morrow et al., 2004). Similarly, Hsp23 expression within the CNS might confer neuroprotection (Michaud and Tanguay, 2003). These protective functions might also be important during the recovery because chill coma results from neuromuscular dysfunctions (Hosler et al., 2000).

It is clear that molecular chaperones are involved during the recovery from cold stress (Colinet et al., 2010a; Michaud and Denlinger, 2010). Moreover, several other genes show expression changes concomitant with those of Hsps during the recovery phase. This is the case for Starvin and DnaJ-1, two co-chaperones that interact with members of the Hsp70 family, although the expression of Starvin declines more sharply than that of the Hsps after 4 h of recovery (Colinet et al., 2010a; Colinet and Hoffmann, 2010). Frost which has no known chaperoning functions also shows an expression pattern similar to those of the Hsp genes, although its expression is particularly high during cold recovery; knocking down Frost expression results in a loss of recovery ability in a fashion similar to that observed following Hsp22 and Hsp23 knockdown (Colinet et al., 2010b). It is unclear whether these genes affect cold recovery via independent pathways. Future experiments could aim to simultaneously suppress two or more of these genes and assay for cold recovery performances. The molecular mechanisms behind recovery from cold stress are complex and poorly understood. Genes/proteins involved in stress responses are conserved in all organisms and are related to various key functions, such as cell cycle control, protein chaperoning, DNA stabilization and repair, the removal of damaged proteins, and certain aspects of metabolism (Kütlz, 2003; Kütlz, 2005).

In conclusion, this study provides evidence that Hsp22 and Hsp23 are important for chill-coma recovery in adult D. melanogaster. The mechanisms whereby they influence this trait are still unclear, but the chaperone products of these genes might target different cellular and tissue-specific functions important in cold recovery. The role of other sHsp genes upregulated during cold recovery (Colinet et al., 2010a) also remains to be explored.

We are grateful to Phillip Daborn and Philip Batterham for providing access to the PC2 facility (Melbourne University, Australia), and we thank Steve McKechnie (Monash University, Australia) for assisting in the importation of fly lines.