Small heat shock protein Hsp67Bc plays a significant role in Drosophila melanogaster cold stress tolerance

Drosophila species are ectothermic animals, and plasticity in thermal tolerance is crucial to them. Resistance to cold defines fruit fly activity, reproductive success, and as a consequence, their habitat. In the laboratory, cold tolerance is generally measured in one of the following ways: acute cold stress (0°C, up to 3 h), prolonged (chronic) cold stress (0°C, 6 h or more), and long-term exposure at low temperature (+5 to +15°C, days or weeks) (Rajamohan and Sinclair, 2008; Sinclair and Roberts, 2005). Short cooling to non-lethal low temperatures causes rapid cold hardening and increases the survival of flies upon repeated exposure to cold (Overgaard et al., 2005; Vesala et al., 2012). Acute cold stress is assumed to cause cell membrane depolarization, primarily in muscle cells and neurons, thereby leading to paralysis (Andersen and Overgaard, 2019). Longer-term or sustained cold exposures differ from acute stress in the nature of the caused injury: prolonged cell membrane depolarization results in a disruption of ion and water homeostasis, thus causing more severe damage to the cell (Overgaard and MacMillan, 2017; Sinclair and Roberts, 2005). As an example, in insects, chronic cold stress results in an increase of K⁺ concentration in hemolymph, which, in combination with sub-optimal cooling, may lead to cell death (Overgaard and MacMillan, 2017). Long-term exposure to mild low temperature results in cold acclimation and increases resistance to cold in almost all insects, including Drosophila, as extensive transcriptomic and metabolomic shifts occur in acclimated individuals (MacMillan et al., 2016). Chill-tolerant insects have an improved ability to maintain homeostatic balance at low temperatures due to higher hemolymph osmolality, accumulation of cryoprotective osmolytes, and modification of cell membrane phospholipid composition (Overgaard and MacMillan, 2017).

A common method for investigating cold tolerance in Drosophila is chill coma: a state of complete neuromuscular paralysis reached by individuals at the critical thermal minimum. The immobilization is reversible because after being transferred to an environment of permissive temperature, the flies recover and stand back up. This method was first suggested by David et al. (1998) and subsequently has become widespread (e.g. Andersen and Overgaard, 2019; Colinet et al. 2010a,b,c). Chill coma onset temperature and chill coma recovery time vary across fly populations and can be used as a measure of fly cold tolerance (Andersen and Overgaard, 2019; David et al., 1998; Macdonald et al., 2004; Vesala et al., 2012). To date, chill coma onset temperature and chill coma recovery time have strictly correlated with latitudinal distribution of fly populations (Andersen and Overgaard, 2019).

Research has revealed an expression change in a great variety of genes in response to cold in Drosophila melanogaster (Qin et al., 2005; Zhang et al., 2011), with up to a third of the genome differentially regulated after prolonged cold exposure (MacMillan et al., 2016; von Heckel et al., 2016). In all instances, the most prominent transcriptional response to all the types of cold treatment was found in genes coding for heat shock proteins (HSPs) (MacMillan et al., 2016; Qin et al., 2005; von Heckel et al., 2016; Zhang et al., 2011). HSPs are present in all living organisms and are crucial for coping with various stressors, including high and low temperatures (Colinet et al., 2010c; Lindquist, 1986). Studies on other Drosophila species, namely D. ananassae, D. montana and D. virilis, have shown cold-induced expression of genes Hsp22, Hsp23, Hsp26, Hsp27, Hsp40, Hsp67Ba, Hsp67Bc, Hsp68, Hsp83 and Hsp70 (Königer and Grath, 2018; Parker et al., 2015; Vesala et al., 2012). More than half of these genes belong to the small heat shock protein (sHsp) family of HSPs.

The highly conserved sHsp family is characterized by the presence of an α-crystallin domain. The major function of sHsps in the cell is the prevention of aggregation of misfolded and damaged proteins, and sHsps are reported to be some of the most up-regulated HSPs following stress (Morrow and Tanguay, 2015). In D. melanogaster, 12 sHsp genes have been identified: CG4461, CG7409, CG13133, CG14207, CG43851, Hsp22, Hsp23, Hsp26, Hsp27, Hsp67Ba, Hsp67Bc and l(2)efl (Morrow and Tanguay, 2015). Nonetheless, only four sHsp proteins (Hsp22, Hsp23, Hsp26 and Hsp27) have been thoroughly investigated (Colinet et al., 2010c; Morrow and Tanguay, 2015). Some sHsp genes are known to be up-regulated during recovery from cold temperatures (Colinet et al., 2010a; von Heckel et al., 2016). Recently, it was demonstrated that another sHsp family member, the Hsp67Bc gene, is up-regulated in cold-acclimated flies (MacMillan et al., 2016).

In D. melanogaster, the Hsp67Bc protein has been shown to be the functional orthologue of human HSPB8 and to interact with Starvin (Stv) (Carra et al., 2010). Stv is essential for viability and muscle maintenance and is enriched at Z-discs in muscle tissue, where both HSPB8 and Hsp67Bc can be found (Arndt et al., 2010; Carra et al., 2010; Coulson et al., 2005). Stv expression was also demonstrated to be up-regulated during the recovery phase in Drosophila alongside Hsp70 expression (Colinet and Hoffmann, 2010). Hsp67Bc gene expression has been reported to increase in Drosophila embryos and S2 cells following heat stress (Leemans et al., 2000; Vos et al., 2016) and in cold-acclimated adult flies (MacMillan et al., 2016). Here, we describe a newly obtained D. melanogaster line with a deletion in the Hsp67Bc gene and investigate the effects of cold stress on these mutant flies. We found that Hsp67Bc, alongside other sHsp genes, participates in cold tolerance in D. melanogaster.

All the Drosophila melanogaster Meigen 1830 flies were raised at 24–25°C and ∼60% relative humidity on a standard cornmeal–agar medium. All the lines initially infected by endosymbiotic Wolbachia bacteria were treated with 0.42 mg ml⁻¹ doxycycline (i.e. kept on antibiotic-containing food for one generation). After that, all the fly lines were found to be PCR-negative for Wolbachia infection. The offspring were then raised on standard food.

The initial D. melanogaster lines included: (i) w¹¹¹⁸ whose DNA served as a control in classic PCR (kindly provided by Professor Scott O'Neill, University of Queensland, Australia); (ii) d06009 (genotype P{XP}Hsp67Bc^d06009), with the insertion of a mini-white gene-coding P-element {XP} of 7303 bp (GenBank: AY515149.1) 115 bp upstream of the Hsp67Bc gene on the third chromosome (Exelixis at Harvard Medical School); and (iii) transposase-coding line y w; Ki, P{Δ2-3}99B from laboratory stock.

The D. melanogaster line with a deletion in the Hsp67Bc gene was obtained by the method of imprecise excision of a P-element. For this purpose, the d06009 line was crossed to the y w; Ki, P{Δ2-3}99B line. The detailed scheme of the crosses is presented in Fig. 1. F1 males, all having both the P-element and transposase, were crossed to previously obtained D. melanogaster females with one of the third chromosomes coding P{XP}Hsp67Bc^d06009 and the other being a balancer (w; P{XP}Hsp67Bc^d06009/TM6B, Hu). From F2, males with white eyes (meaning that the P-element was cut out and lost) were selected and back-crossed to w; P{XP}Hsp67Bc^d06009/TM6B, Hu females to subsequently obtain a stable line. After that, the DNA of the F2 males was extracted and analysed by PCR to detect a desired genotype variant. Such a fly line should have a deletion affecting only the Hsp67Bc gene (the amplicon of a region including only Hsp67Bc should be shorter than that of the control w¹¹¹⁸ line) and no remains of the P-element. Of the 278 obtained lines, only one matched these criteria, while the others either had a normal-size amplicon (majority), meaning the precise excision of the P-element, or no amplicon at all, meaning that either the deletion affected an area outside the Hsp67Bc gene or that there was still a part of the P-element (without the mini-white gene) remaining. In the obtained appropriate line, the part of the third chromosome containing the Hsp67Bc gene and P-element {XP} insertion site was then sequenced. As part of the experiment, a line from which the P-element was cut out precisely was chosen (based on DNA sequence similarities with reference and mutant ones) and used as a control.

To find a mutant variant with the deletion affecting only the Hsp67Bc gene among the obtained D. melanogaster lines, classic PCR was carried out. From each line, one male was homogenized in STE buffer (100 mmol l⁻¹ NaCl, 10 mmol l⁻¹ Tris-HCl, pH 8.0, and 1 mmol l⁻¹ EDTA) with SDS and proteinase K at final concentrations of 1% and 0.2 mg ml⁻¹, respectively. The homogenate was then incubated at 56°C for 90 min, followed by 8 min at 95°C and brief centrifugation. Next, the DNA was precipitated with 0.5 volumes of 5 mol l⁻¹ sodium acetate and 2 volumes of absolute ethanol, followed by washing in 75% ethanol and dissolution in deionized water. Twenty-microlitre PCRs based on the BioMaster HS-Taq PCR-Color (2×) mix (Biolabmix, Novosibirsk, Russia) were set up with primers complementary to regions 285 bp upstream (left primer: 5′-GATTGCGCCTCTCATCTGTATC-3′) and 236 bp downstream (right primer: 5′-ATTACACTTATGGCGGGCTTTC-3′) of Hsp67Bc. The thermal cycling protocol for the amplification was as follows: 5 min denaturation at 95°C; followed by 30 cycles of 30 s at 95°C, 30 s at 57°C, and 45 s at 72°C; and finally, 5 min of elongation at 72°C. The PCR products were separated by electrophoresis on a 1% agarose gel in TBE buffer.

The primers were chosen based on the reverse complement of GenBank sequence AE014296.5.

To characterize the obtained Hsp67Bc-null fly line, sequencing of the mutant DNA fragment containing the Hsp67Bc gene was performed. DNA was extracted from 2–3 flies and amplified as described in the ‘Classic PCR’ section above. DNA quality was checked by electrophoresis. The amplified DNA was precipitated with 20% polyethylene glycol and 75% ethanol and dissolved in deionized water. The sequencing was performed with the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Waltham, MA, USA) and with the same primers as those employed for the classic PCR. The cycle sequencing program was as follows: incubation at 95°C for 1 min followed by 35 cycles of 10 s at 98°C, 10 s at 50°C, and 4 min at 60°C. The sequencing reaction mixture was then purified with 7.5 mol l⁻¹ sodium acetate and absolute ethanol, followed by washing in 75% ethanol, after which it was dried and handed over for sequencing to the Genomics Core Facility (Institute of Chemical Biology and Fundamental Medicine, SB RAS, Novosibirsk, Russia). The results were then analysed in the BioEdit Sequence Alignment Editor software, version 7.2.5 (https://bioedit.software.informer.com/7.2/).

For these experiments, wandering late third instar (L3) larvae were transferred from the walls of their rearing vials to the walls of vials with fresh food, at 20 per tube (30 in a few early tests). The larvae were then allowed to reach a developmental stage that was to be treated, in particular, immobile late L3, white prepupa (less than 1 h after pupation (Bainbridge and Bownes, 1981), with slight addition of early P2 stage prepupae with yellow puparium), P4–P5 (11–13 h after pupation), or P7–P8 (46–48 h after pupation), which correspond to the highest Hsp67Bc expression levels according to FlyBase (http://flybase.org/). The vials containing D. melanogaster individuals were then transferred to a 0°C environment (ice bath) for 2, 4 or 12 h incubation. After each treatment, the flies were kept at 24–25°C until eclosion. Survivors to the pupa stage (in case of late L3 larvae treatment) and to the adult stage (for all treatment groups) were then counted, and the resultant number was divided by the initial number of individuals in the group to assess the survival rates. Each experiment was repeated 4–16 times.

Three-day-old D. melanogaster individuals were sorted by sex and placed into vials at 15 flies per vial, one day prior to the start of the experiment. The chill coma assay was based on a method described by Colinet et al. (2010c): the vials with 4-day-old flies were placed in a 0°C (ice bath) environment for 2, 4 or 12 h incubation, after which they were transferred to a 24–25°C environment to recover. The recovery was then measured by counting the flies that were standing, every 5 min (2 and 4 h chill coma) or 10 min (12 h chill coma) after the end of the cold stress. For this experiment, three (2 and 4 h chill coma) or seven (12 h chill coma) iterations with 60–107 flies in each were carried out. Recovery curves were constructed based on averaged results. For the statistical analyses, separate experiments were compared. Two days (48 h) after the cold treatment, the surviving flies were counted. Survival rates of the flies were calculated as a ratio of the surviving individuals to the initial number of individuals.

One day prior to the start of the experiment, 3-day-old D. melanogaster individuals were placed into vials at either eight males and 10 females per each of 10 vials in the experimental group, or six males and eight females per each of four vials in the control group. The control group was kept at 24.5°C; the vials with 4-day-old individuals from the experimental group were placed in a 0°C (ice bath) environment for 12 h incubation, after which they were transferred to a 24.5°C environment. The fecundity of D. melanogaster in both groups was measured every day for one week, starting one day after the end of the exposure to cold. The females were allowed to lay eggs for 24 h in vials with fresh food, then they were transferred to new food, and the eggs were counted under a binocular microscope. The quantity of eggs in each vial was then divided by the number of females that oviposited in that very vial.

Increasing the initial number of individuals in vials, as well as number of the vials, in the experimental group compared with the control group allowed us to have similar sample sizes throughout most of the fecundity experiment because preliminary chill coma survival assays revealed that some females died during the first two days after the chill coma.

Total RNA was extracted from whole adults (three males or females per sample, or four larvae or pupae per sample) using the TRIzol Reagent analogue from Biolabmix (Novosibirsk, Russia) according to the instructions given in the Invitrogen protocol for the TRIzol Reagent. RNA concentrations were determined on a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). DNA contamination was then eliminated using DNase I with the addition of the RiboLock RNase Inhibitor (both from Thermo Fisher Scientific) in accordance with the manufacturer's protocol. Complementary DNA was synthesized by means of the oligo(dT)₁₇ primer (Evrogen, Moscow, Russia) and RevertAid Reverse Transcriptase (Thermo Fisher Scientific).

This procedure was performed on a LightCycler 96 cycler (Roche). The thermal cycling protocol for the amplification reaction was as follows: 5 min pre-incubation at 95°C; next, 40 cycles of 15 s at 95°C, 20 s at 60°C, and 30 s at 72°C; followed by a melting curve program at 65–97°C. The results were then analysed in REST 2009 software (https://www.gene-quantification.de/rest-2009.html).

Primers used in qPCR reactions are listed in Table 1. The primer sequences for αTubulin84B were taken from Ponton et al. (2011); for Hsp22, from Colinet et al. (2010c); and for Fst, from Newman et al. (2017). αTubulin84B was chosen as a reference gene based on reports of its stability across various temperature ranges (Myachina et al., 2017).

The comparison of survival rates after cold treatment between lines was performed by the chi-squared test. Recovery curves were compared within each experiment by the Kolmogorov–Smirnov test, and the conclusion about significance of a difference in each experiment within the treatment groups was based on the majority of the test results. Differences between mutant and control lines in the fecundity of females that underwent the chill coma and at 24.5°C were compared by the heteroscedastic t-test at each time point throughout the experiment to reveal possible differences in fecundity dynamics. The procedure was followed by application of the Benjamini–Krieger–Yekutieli method to control the false discovery rate. To discover differences between the mutant and control fly lines in mean fecundity throughout the experiment, total averages from all the registration points were compared by the heteroscedastic t-test.

As a result of the imprecise excision of the P-element {XP} located 115 bp upstream of the Hsp67Bc gene in parent line d06009, a D. melanogaster line with a deletion of almost the whole gene was obtained. PCR analysis revealed that in this line, the length of the amplicon between primers complementary to regions 285 bp upstream and 236 bp downstream of the Hsp67Bc gene was approximately 450 bp, unlike 1508 bp in the control (Fig. 2A). Sequencing of DNA from the obtained mutant line revealed a deletion of 1076 bp including the Hsp67Bc promoter (position 9,371,643 to position 9,372,718 on chromosome 3L in GenBank sequence AE014296.5), leaving only the last 132 bp from the 3′-end of the 987 bp Hsp67Bc gene (Fig. 2B). In addition, an insertion of 12 bp (TTTTGACTTTTT) was found at the site of the excised P-element, presumably part of the P-element. No coding sequences of adjacent genes were affected by the deletion. Additionally, we tested the expression level of a neighboring sHsp gene, Hsp22, which can share regulatory sequences with Hsp67Bc and is crucial for the maintenance of mitochondria in D. melanogaster (Morrow et al., 2004). We found no difference between the obtained Hsp67Bc mutants and the control under normal conditions (see last paragraph of the Results section for information on Hsp22 expression). The obtained flies carrying the deletion in the homozygous state were viable, fertile and phenotypically similar to the control. Hereafter in this paper, this D. melanogaster Hsp67Bc-null line is referred to as Hsp67Bc-0, whereas the control line (obtained by precise P-element excision from the same parent line) is called Hsp67Bc-2.

According to FlyBase (http://flybase.org/), mRNA and protein expression levels of Hsp67Bc are highest in late third instar larvae, and pupae of D. melanogaster. Therefore, we decided to investigate how cold treatment would affect the survival of Hsp67Bc-0 flies at larva and pupa stages of development, in addition to adults.

Drosophila individuals at a stage of wandering third instar larva (LL3), white prepupa (WPP), 12 h pupa [P (12 h)], 48 h pupa [P (48 h)] or adult (4-day-old males and females) were placed in a 0°C environment for 2 h (acute stress), 4 h (moderate stress) or 12 h incubation (prolonged stress), after which they were transferred back to 24.5°C to recover and/or continue development. The survival rates of larvae and pupae were computed from the number of eclosed individuals, and the survival rates of adult flies were determined 48 h after the cold treatment as a percentage of individuals alive. The results are presented in Fig. 3A. The response to the cold treatment depended on the stage in both control and Hsp67Bc-null flies. The survival rates declined with treatment time in larvae and pupae, whereas adult flies showed similar survival rates regardless of the duration of cold stress. Within the adult stage, strong sexual dimorphism was noted in terms of the survival after chill coma. No less than 98% of males of both lines were alive 48 h after each treatment, whereas the survival rates were lower and 1.6- to 3.0-fold different between the control and mutant females. The survival rates varied between 74.1±5.0% (2 h at 0°C) and 84.8±3.8% (4 h at 0°C) in Hsp67Bc-2 females, whereas only 27.9±1.1% (4 h at 0°C) to 45.9±5.9% (12 h at 0°C) of Hsp67Bc-null females stayed alive (χ²≥49.80, d.f.=1, P<0.001 in all the cold treatment groups). No deaths were observed after 48 h of recovery in adults.

After the acute and moderate cold stress, the survival rates of larvae and pupae were similar to those in adult males and control females, varying between 63.4±3.5% in Hsp67Bc-2 P (12 h) (moderate stress) and 95.0±2.0% in Hsp67Bc-2 P (48 h) (acute stress). Unlike adults, Hsp67Bc-null LL3 and pupae showed a tendency for an improved survival to adulthood (significant only at WPP after moderate cold exposure, χ²=7.00, d.f.=1, P=0.008). After 12 h at 0°C, the survival of LL3 and pupae decreased dramatically, compared with that after acute and moderate cold stress. It did not exceed 25.5±9.5% in the Hsp67Bc-0 larvae and 17.7±4.2% in the Hsp67Bc-2 line, and among the tested pupal stages, the highest survival rate was observed in Hsp67Bc-0 WPP and Hsp67Bc-2 P (48 h): 25.5±7.4 and 13.0±8.9%, respectively. No significant difference was found between control and mutant Drosophila at LL3 and pupae stages after the prolonged cold stress.

The death of larvae and pupae did not necessarily occur exactly during or shortly after the cold treatment. Fig. 3B shows approximate stages [LL3, pupal stage <P12 (Bainbridge and Bownes, 1981) or ≥P12] at which all the individuals died after the exposure to 0°C. The mortality was delayed to the extent that, depending on the stage at which the cold exposure occurred, 20% (Hsp67Bc-0 LL3 after 12 h at 0°C) to almost 100% [P (48 h) after 12 h at 0°C in both lines] of the deceased individuals reached the ≥P12 stage, at which the imago is almost fully formed. Mutant and control LL3 individuals had significantly different survival to the pupal stage after short and moderate exposure to cold. Almost all the LL3 larvae survived to the pupal stage in the control line regardless of cold stress duration. Hsp67Bc-0 LL3 were significantly more cold susceptible: up to 29% (4 h at 0°C) of all the deceased individuals failed to proceed to the pupa stage. The overall survival from LL3 to pupa differed between the control and mutant lines, being 100% in the Hsp67Bc-2 line and 96.4±2.1% in mutants after short cold exposure (χ²=5.95, d.f.=1, P=0.015), and 98.4±1.1% in Hsp67Bc-2 and 95.9±1.3% in Hsp67Bc-0 larvae after moderate cold stress (χ²=6.39, d.f.=1, P=0.011).

To test short-term recovery from cold stress, chill coma (0°C; 2, 4 or 12 h) experiments were conducted on 4-day-old Hsp67Bc-0 and Hsp67Bc-2 D. melanogaster individuals. Hsp67Bc-0 flies showed a slower recovery from the chill coma than the control (Fig. 4). Recovery dynamics of the mutant and control females differed significantly in all or in the majority of the experiment replicates (1.34≤λ≤3.43 in 2 h chill coma experiments, 0.001<P<0.06; λ≥1.94, P≤0.001 in 4 h chill coma experiments; λ≥1.80, P≤0.003 in 12 h chill coma experiments). Hsp67Bc-null males recovered as fast as the control flies after 2 h of chill coma. Nevertheless, they showed slower recovery after two other cold treatment durations (λ≥1.42, P≤0.036). On average, all Hsp67Bc-2 males and females recovered from the comatose state within 40 min after the end of 2 and 4 h cold treatment and within 80 min after the end of 12 h cold stress, with females recovering slightly more slowly. Hsp67Bc-0 flies manifested similar patterns of recovery but with a 5–15 min delay.

Under normal conditions (24.5°C, no treatment), Hsp67Bc-null female Drosophila showed lower fecundity than control flies (9.0±0.3 eggs per female compared with 10.8±0.3, averaged from the mean number of eggs laid by females during a 7-day period of registration, P<0.001), constituting 83.4% of that in the control line. After 12 h chill coma at 0°C, the fecundity of both Hsp67Bc-0 and Hsp67Bc-2 flies declined dramatically (Fig. 3C). Thus, the mean number of eggs laid by Hsp67Bc-2 females during the 7-day period after the end of the cold stress diminished approximately 7-fold to 1.5±0.1. In Hsp67Bc-0 females, this parameter went down more than 8-fold to as few as 1.1±0.2 eggs per female, constituting 69.5% of that in the control. Therefore, cold stress had similar negative effects on the fecundity of both control and mutant females.

Given that D. melanogaster survival and recovery were differently affected by the absence of the Hsp67Bc gene product, depending on the stage of development at which individuals were exposed to cold, we decided to measure expression levels of Hsp67Bc and some other HSP genes as well as starvin (stv) and Frost (Fst) in a search for possible candidates for Hsp67Bc function compensation. The Stv protein interacts with Hsp67Bc and is up-regulated in response to cold (Carra et al., 2010; Colinet and Hoffmann, 2010); Fst was also shown to be up-regulated after cold treatment (Colinet et al., 2010b). Among HSP genes, we selected Hsp70 because it is a prominent heat- and cold-induced HSP in Drosophila (Goto and Kimura, 1998; Parsell et al., 1993), and four sHsps: Hsp22, Hsp23, lethal (2) essential for life [l(2)efl] and Hsp67Ba. Hsp22 and Hsp23 were chosen because they were proved to be involved in cold tolerance (Colinet et al., 2010c); l(2)efl is cold inducible and is abundantly expressed in the muscles of Drosophila larvae and adults (http://flybase.org/); and Hsp67Ba is prominently expressed in the central nervous system (CNS) of fruit fly larvae and pupae and larval carcass (http://flybase.org/). Gene expression was measured under normal conditions (24.5°C, no treatment), immediately after the end of cold treatment (0 min of recovery), and 90 min after the end of cold exposure in D. melanogaster at developmental stages that had a significant difference in chill coma survival, namely LL3, WPP and adult (4-day-old male and female) stages. The duration of cold treatment was 12 h for tested adults and 4 h for LL3 and WPP as 12 h chill coma had the same effect on the survival of adults as moderate and acute cold stress, but provided an opportunity to measure changes in gene expression levels shortly after the recovery from the comatose state. The prolonged cold stress was assumed to be too severe for larvae and pupae (less than a quarter of individuals of both genotypes survived to imago); therefore, 4 h cold stress was applied to LL3 and WPP instead.

Under normal conditions, the expression levels of the majority of assayed genes were the same between the control and Hsp67Bc-null flies at all tested stages of development, the exceptions being stv (which was 1.33-fold up-regulated in mutant LL3, P=0.020), Hsp70 (which was 1.7-fold down-regulated in Hsp67Bc-0 adult females, P<0.001) and Hsp23 (which was 1.6-fold down-regulated in Hsp67Bc-0 adult males, P=0.017) (Table 2). Notable, but not significantly different, tendencies towards up- or down-regulation were also observed for Hsp70 and some other genes. The relative expression levels of the Hsp67Bc gene in the Hsp67Bc-2 line differed noticeably between the stages of development, being highest in WPP (2.8-fold higher than in LL3) and going down drastically by the adult stage (60-fold in males and 150-fold in females), in line with FlyBase data (http://flybase.org/). It is worth noting that Hsp67Bc expression significantly differed between males and females: its RNA levels were 2.5-fold lower in females (P=0.005). Sex-specific differences in expression were also present in some other assayed genes. For example, in control and mutant flies, the RNA levels of stv, l(2)efl, Hsp67Ba and Fst were higher or tended to be higher in males than in females. In females, only Hsp23 was significantly up-regulated compared with males (7.9-fold higher in the control, and 16.4-fold higher in the Hsp67Bc-null line, P<0.001). These tendencies generally persisted after the cold stress (Fig. 5). In addition to the mentioned genes, the Hsp70 level was 10-fold higher in control males than females (P<0.001) and showed a similar tendency in the mutant line (2.7-fold, P=0.062) immediately after the cold stress. Taken together, these sex-specific differences in the expression levels of the stress response genes after the exposure of flies to cold may account for the slight delay in recovery from chill coma in adult females compared with males.

The expression of the majority of investigated genes increased only at 90 min after the end of cold treatment (Fig. 5; Table 3). Expression levels of stv and Hsp70 in the mutant females were 2.4- and 3.1-fold higher than those in the control females at 0 min of recovery, respectively (P<0.001). However, the stv level in the control females increased faster by 90 min of recovery (1.7-fold higher in Hsp67Bc-2 females compared with the mutants at 40 min of recovery, P<0.001, data not shown).

Ninety minutes after the end of cold treatment, the highest expression levels of all the assayed genes were observed (Fig. 5; Table 3). The most up-regulated genes were Hsp70 and Fst in adults and Hsp22 in LL3 larvae. The greatest increase of expression was observed for Hsp70 (64-fold in the control males, P=0.065; 170-fold in Hsp67Bc-0 males, P<0.001; 99-fold in the control females, P=0.032; and 148-fold in the mutant females, P<0.001). It is worth noting that the majority of genes were differently affected by cold stress, depending on the developmental stage during the treatment. Hsp67Bc expression was higher in LL3 larvae and WPP than in adults and did not significantly change in response to cold stress, but increased 2.5-fold in males (P=0.037) and 6.1-fold in females (P=0.011). Although the expression level of Hsp67Bc was similar between males and females at 90 min of recovery, the up-regulation was delayed in females (Fig. 6). Thus, in females, Hsp67Bc RNA levels increased exponentially throughout the 25, 40 and 90 min recovery registration points, whereas in males, a close to a maximum level of RNA was reached by 25 min of recovery. The expression level of Hsp67Bc decreased between 90 min and 3 h after cold treatment (Fig. 6).

A few significant differences in RNA levels between the two Drosophila lines at 90 min of recovery were discovered for genes stv, Hsp70, Hsp22 and Fst at different developmental stages, all of which were up-regulated in the mutant line compared with the control line (Fig. 5). The Hsp70 RNA level was 1.7-fold higher in the adult males (P<0.001), the Fst level was elevated 1.2-fold in adult females (P<0.001), stv was up-regulated 1.3-fold in adult males (P<0.001) and 1.7-fold in WPP (P=0.024), and Hsp22 was 1.3-fold higher in WPP (P=0.043). It can be assumed that these increases in the expression of several cold stress response genes may be involved in compensation of the functions of the absent Hsp67Bc gene product.

In this study, we found that Hsp67Bc-null Drosophila larvae, pupae and adults have varying cold stress tolerance. Under acute (2 h at 0°C), moderate (4 h at 0°C) and chronic (12 h at 0°C) stress, the deletion in Hsp67Bc gene reduced cold tolerance in adults and wandering L3 larvae but not in white prepupae and pupae. Each stage of development showed distinct sensitivity to cold exposure. For instance, adult Hsp67Bc-null flies recovered from chill coma more slowly than the individuals with intact gene copies (Fig. 4), and females showed the most pronounced reduction in survival (1.6- to 3.0-fold lower than in the control line; Fig. 3A). Hsp67Bc-null LL3 larvae also featured a reduced survival to pupal stage after cold stress compared with the control individuals; however, the difference was not that prominent. Hsp67Bc deletion did not affect the survival rate after cold stress in late pupae and even slightly increased the survival in early pupae (Fig. 3A). We assume that this phenomenon is linked with the physiology of the corresponding stages. Larvae are characterized by extensive feeding, which is necessary for the accumulation of nutrients and energy to be used at the subsequent pupal stage; therefore, the metabolism slowdown caused by chill coma must have its consequences later on. Accordingly, we observed a slight decrease in survival to pupa at the late L3 stage and a delayed mortality at the pupal stage up to >P12 (Fig. 3B). Pupae are unable to feed, and the energy and supplies for the metamorphosis are provided via autophagic cell death, which removes most of the larval cells (Aguila et al., 2007). According to Merkey et al. (2011), metabolic rates decrease rapidly in pupae during the first 24 h and remain low until shortly before eclosion. Together with these processes, detrimental effects of cold stress (metabolic slowdown as well as ion and osmotic balance disruption that may eventually lead to cell death) either have no notable impact on the metamorphosis or have a delayed effect. Our data are consistent with the findings of Merkey et al. (2011), who reported that at the pupal stage, metabolic rates are independent of temperature fluctuations during the first two-thirds of pupal development.

Aside from survival, the imago stage can be characterized by sexual dimorphism in chill coma recovery speed. For example, females recovered more slowly than males in both the mutant and control lines, even after the shortest cold treatment (Fig. 4). We believe one of the causes to be the difference in the metabolome between males and females. A variety of studies indicate that the Drosophila metabolome is sex and age dependent, with sexual dimorphism increasing throughout development (Hoffman et al., 2014; Ingleby and Morrow, 2017). This notion is also in line with our findings about the difference in the expression of several genes between males and females (Table 2).

In addition to Hsp67Bc, we measured the expression of several genes, namely Hsp70, Hsp22, Hsp23, Hsp67Ba, l(2)efl, stv and Fst, in control and Hsp67Bc-null flies under normal (24.5°C) conditions as well as immediately after and 90 min after the end of 4 or 12 h cold treatment at 0°C (Tables 2 and 3; Fig. 5). As expected, we detected almost no statistically significant difference in the expression levels of the chosen genes between the mutant and control lines, with the exception of slight up-regulation of stv in mutant L3 larvae and down-regulation of Hsp70 and Hsp23 in adult females and adult males, respectively. The highest Hsp67Bc expression was observed at the pupal stage, whereas it was more than 60-fold lower in adults, in agreement with FlyBase data (http://flybase.org/). After the cold exposure, however, the expression of Hsp67Bc did not significantly change in pupae, tended to rise in the larvae, and dramatically increased in adults in a sex-specific manner. For instance, in males, Hsp67Bc expression was slightly elevated immediately after 12 h of the cold stress and reached a 2.5-fold increase by 90 min of recovery (Fig. 6). In females, the gene expression materialized with a lag but increased 6-fold by 90 min. This sexual dimorphism correlates with the different susceptibility of males and females to chill coma (Fig. 4). Apparently, the Hsp67Bc gene is more important for the recovery from chill coma in females, and its absence in mutants slows down the repair processes and reduces survival. The rate of recovery from chill coma depends on two processes: the recovery of CNS function and muscle polarization (Andersen and Overgaard, 2019). Although Hsp67Bc is primarily expressed in the CNS and Malpighian tubules (http://flybase.org/), the Hsp67Bc protein is also found in muscles (Carra et al., 2010). The deletion in this gene can impair both points of neuromuscular recovery. Genes Hsp23 and l(2)efl, which are abundantly expressed in the CNS and muscles, respectively, turned out to be up-regulated during the recovery phase (90 min after cold treatment in our study). By contrast, no differences in the expression of these genes were found between the control and Hsp67Bc-null lines (Fig. 5), implying their Hsp67Bc-independent functioning in chill coma recovery. Additionally, we assessed RNA levels of another sHsp gene, Hsp67Ba, which partially shares the expression pattern with Hsp67Bc, and found no significant correlation of Hsp67Ba levels with cold treatment, recovery phase, developmental stage and sex. Therefore, Hsp67Ba is unlikely to be a cold-tolerance gene.

Furthermore, we measured the expression of genes that have been shown to be involved in the cold stress response (Colinet and Hoffmann, 2010; Colinet et al., 2010b,c): Hsp22, Hsp23, Hsp70, stv and Fst (Tables 2 and 3; Fig. 5). As expected, all these genes proved to be up-regulated after cold exposure at all the fly developmental stages, although to varying degrees. The stv level was already increased during the cold exposure, reaching maximum expression at 90 min of recovery in both control flies and mutants, although with differing dynamics. Stv is a known participant of the stress response to cold exposure (Colinet and Hoffmann, 2010). These observations imply that Stv may be important for Drosophila protection from cold stress and for survival after chill coma. Genes Hsp22, Hsp23 and Fst were found to be up-regulated only during the recovery period in both control and Hsp67Bc-null lines, thereby confirming their substantial role in the recovery of flies from a comatose state (Colinet et al., 2010b,c). The total Hsp70 RNA level increased to a greater extent in Hsp67Bc-null mutants than control flies (170-fold higher in the mutant males versus 64-fold in the control males). Moreover, the up-regulation of the Hsp70 gene was already noticeable during the chill coma, whereas only tendencies towards an increase in expression were seen in the control (Table 3; Fig. 5). The drastic up-regulation of Hsp70 is consistent with other studies (Königer and Grath, 2018; MacMillan et al., 2016), and its more pronounced up-regulation in Hsp67Bc-null flies may denote its compensatory role in the response to cold stress in the absence of Hsp67Bc. It is possible that this compensatory mechanism allows Hsp67Bc-null Drosophila to survive, in contrast to Hsp22- and Hsp23-deficient flies. Colinet et al. (2010c) demonstrated that 24% of Hsp22 and Hsp23 knockout flies fail to recover from chill coma. Taken together, these data indicate that Hsp22 and Hsp23 play a more important part in chill coma recovery in Drosophila.

The presence of sex-specific differences in the expression of the assayed genes in both control and Hsp67Bc-null mutant flies is worth noting (Table 2; Fig. 5). As an example, immediately after the end of cold stress (0 min of recovery) Hsp67Bc, Hsp67Ba, Hsp70, stv and Fst had higher expression levels in males than females. This phenomenon can account for the difference in cold tolerance between males and females, as well as the dissimilarity of the sets of genes selected by the up-regulation extent after cold exposure in the studies by MacMillan et al. (2016) (males), Qin et al. (2005) (males), Vesala et al. (2012) (females) and Zhang et al. (2011) (females).

It remains unclear which functions exactly Hsp67Bc performs at the pupal stage under normal conditions. According to FlyBase (http://flybase.org/), the peak of Hsp67Bc gene expression level is reached at the pupal stage (which was also true in our study in white prepupae, compared with other assessed stages of development). It may be assumed that during metamorphosis, the Hsp67Bc gene product takes part in chaperone-assisted autophagic degradation of proteins in complex with its co-chaperone Stv (Arndt et al., 2010) or, on the contrary, in protein folding alongside other chaperones including sHsps. Apparently, these Hsp67Bc functions overlap with the roles of other proteins because Hsp67Bc-null homozygous Drosophila is viable and fertile.

Data on the involvement of Hsp67Bc in cold tolerance are contradictory. MacMillan et al. (2016) noticed a 12-fold increase in Hsp67Bc expression in D. melanogaster males after 5-day-long acclimation at +5°C and therefore classified it as a cold acclimation gene. Vesala et al. (2012) investigated the acclimation in females belonging to two species of Drosophila: cold-tolerant D. montana as well as D. virilis having a more southern distribution range. Those authors discovered that after 5 days at +5°C, Hsp67Bc was 5.9-fold up-regulated in D. virilis but not in cold-tolerant D. montana and therefore, similarly, categorized it as an acclimation gene. Nonetheless, a similar study on D. montana and D. virilis females (Parker et al., 2015) did not confirm the participation of Hsp67Bc in cold acclimation. Here, we demonstrate that the expression of Hsp67Bc increases after the cold stress and decreases after 90 min of recovery (Fig. 6), pointing to its involvement in short-term recovery from cold stress. Our data are in line with the findings of Zhang et al. (2011) who investigated gene expression in D. melanogaster females at 6 h of recovery from chill coma and detected no changes in the Hsp67Bc level. These findings allow us to regard Hsp67Bc as a cold tolerance gene. We also discovered that the deletion in this gene results in a statistically significant delay in the chill coma recovery. Moreover, survival was impaired in Hsp67Bc-null larvae and adults within 2 days after the chill coma; this pattern also characterizes a short-term stress response.

Cold acclimation lasts for days or weeks during which the concentration of sugars, polyols and amino acids increases, thereby enhancing cryoprotective capacity of cells and altering their energy exchange, proline and glutathione metabolism, lipid composition, cell membrane permeability, and ion balance, all of which contribute to the protection of the organism against cold injury (Colinet and Hoffmann, 2012; Koštál et al., 2016; MacMillan et al., 2016; Overgaard and MacMillan, 2017; Overgaard et al., 2008). The response to cold stress or chill coma includes repair of cold-induced damage, specifically DNA and protein stabilization and repair, the removal of damaged proteins, and shifts in certain parameters of metabolism involving heat shock proteins and other stress-related genes (Colinet and Hoffmann, 2012; Colinet et al., 2010a; Parker et al., 2015; Zhang et al., 2011). The increase in the mortality rate after chill coma in Hsp67Bc-null D. melanogaster occurred with a delay. That is, the individuals did not die during the cold exposure or immediately after it; rather they died within days after being transferred to normal conditions, indicating an impaired repair process. Therefore, considering all the findings, we regard Hsp67Bc as a cold stress tolerance gene rather than cold acclimation gene.

We thank Dr Yury E. Herbeck and Dmitri A. Karagodin for technical assistance with the qPCR and sequence reactions, and RNA extraction, respectively, and Dr Vadim M. Efimov for his advice on statistical methods. We also thank the Laboratory of Evolutionary Genetics for technical support with the qPCR experiments, and two anonymous reviewers for valuable comments on the manuscript. In the submitted manuscript, the English language was corrected and certified by shevchuk-editing.com.