We previously generated a dPGAM5 null allele, PGAM5¹, and found that PGAM5¹ hemizygous (PGAM5¹/Y) flies, designated here as PGAM5¹ flies, were viable and fertile but showed a slightly longer lifespan than control (y¹w¹/Y) flies [11]. Recently, we observed that PGAM5¹ flies showed a similar lifespan to that of control flies when maintained on a glucose-based diet instead of the molasses-based diet that had previously been used (Fig. S1). This condition was favorable for phenotypic analyses of PGAM5¹ flies, since differences in lifespan may bias the evaluation of the responses of PGAM5¹ flies to environmental stressors.

Although PGAM5¹ flies showed neither resistance nor sensitivity to oxidative stress (H₂O₂) and starvation, they were sensitive to HS stress and died sooner than control flies (Fig. 1A), suggesting that dPGAM5 is required for proper response to HS. Since it was previously found that introduction of the PGAM5¹ allele into a loss-of-function allele for Drosophila PINK1 (dPINK1), PINK1^B9, ameliorated pathological changes induced by dPINK1 deficiency [11], we examined the genetic interaction between dPGAM5 and dPINK1 in the context of HS response. As shown in Fig. 1B, PINK1^B9 flies were also sensitive to HS stress and died even sooner than PGAM5¹ flies. Importantly, however, PGAM5¹, PINK1^B9 double mutant flies were more sensitive to HS stress than either PGAM5¹ flies or PINK1^B9 flies, suggesting that dPGAM5 and dPINK1 are involved in HS response but act independently of each other.

To confirm and further analyze the involvement of dPGAM5 in HS response, we used transgenic flies harboring the gene encoding an inverted repeat (IR) RNA for dPGAM5, which specifically inhibits dPGAM5 expression, downstream of the UAS binding sequence for the transcription factor GAL4 (UAS-dPGAM5 IR). When these flies were crossed with daughterless (da)-GAL4, a driver strain that expresses GAL4 ubiquitously, the progeny (da>dPGAM5 IR) in which dPGAM5 was knocked down in the whole body was also vulnerable to HS (Fig. 1C, D). To exclude the possibility that persistent deficiency of dPGAM5 in flies throughout development affects their HS sensitivity, we transiently knocked down dPGAM5 prior to the HS treatment using heat shock (hs)>dPGAM5 IR flies in which dPGAM5 was knocked down by four times transient treatment of flies at 37°C each for 30 min. These flies were vulnerable to sustained HS, whereas the sensitivity of the flies without HS pretreatment, thus not expressing dPGAM5 IR, was similar to that of control hs>LacZ IR flies (Fig. 1E). This result suggests that dPGAM5 is directly involved in HS response in adult flies.

We then examined whether or not the tissue-specific knockdown of dPGAM5 could also induce vulnerability of flies to HS. Whereas dPGAM5 knockdown in various tissues such as the fat body (yolk-GAL4), hemocytes (pxn-GAL4), eye (sev-GAL4), bristle (sca-GAL4) and the mediodorsal parts of the thoracic and abdominal segments (pnr-GAL4) exerted no effect on the HS response (Fig. S2A–E), dPGAM5 knockdown in the MB using three different driver strains (c739-GAL4, OK107-GAL4 and 201Y-GAL4) [12], [13] induced vulnerability in the flies to HS similar to that observed in PGAM5¹ flies (Fig. 2A–C).

Consistent with the fact that the MB is an insect brain structure composed primarily of neuronal cell bodies and neurites [14], dPGAM5 knockdown using the pan-neuronal elav^c155-GAL4 driver induced the vulnerability of flies to HS (Fig. 2D). On the other hand, dPGAM5 knockdown in dopaminergic neurons using the tyrosine hydroxylase (TH)-GAL4 driver had no effect on the sensitivity of the flies to HS (Fig. S2F), suggesting that the MB, but not necessarily the entire system of neuronal tissues, plays a major role in the HS response in dPGAM5-deficient flies. These findings, taken together, indicate that a deficiency of dPGAM5 in the MB may be responsible for the vulnerability to HS.

To determine whether or not the vulnerability of PGAM5¹ flies to HS might indeed be attributable to a loss of dPGAM5 in the MB, we expressed dPGAM5 selectively in the MB of PGAM5¹ flies using the c739-GAL4 driver (PGAM5¹; c739>dPGAM5 flies). The vulnerability of these flies to HS was attenuated compared to that of PGAM5¹ flies (Fig. 3A), indicating that the loss of dPGAM5 in the MB was responsible for the vulnerability of PGAM5¹ flies to HS. On the other hand, the phosphatase-inactive mutant of dPGAM5 (H94A) [10], when expressed at a similar level to wild-type dPGAM5 in the MB, had no effect on the sensitivity of PGAM5¹ flies to HS (Fig. 3A and B). These results suggest that, depending on its phosphatase activity, dPGAM5 in the MB plays a critical role in the protection of flies from HS.

To elucidate the role of dPGAM5 in the MB, we next examined HS-induced histological changes in the MB. The MB consists primarily of a cluster of intrinsic neurons referred to as Kenyon cells (KCs), together with their dendritic branches (calyx) and axonal bundle (pedunculus). At the end of the pedunculus, the axons bifurcate dorsally and medially to form the vertical and medial lobes, respectively [14]. When the lobes and calyxes were observed by MB-specific expression of a fusion protein between a transmembrane protein CD8 and GFP (mCD8-GFP), which labels the cell surface, HS-induced morphological changes of these regions were induced in neither control flies nor PGAM5¹ flies (Fig. 4A–H). However, when the nuclei of a subset of KCs were visualized by c739-GAL4-dependend expression of the nucleus-targeting Histone2B::ECFP, which was previously shown to detect nuclear condensation and morphological changes that accompany cell death [15], the number of KC nuclei decreased in PGAM5¹ flies, but not in control flies, upon 90-min HS treatment (Fig. 4I–M). Consistent with this, Hoechst 33258 staining of the nuclei of KCs in PGAM5¹ flies was reduced compared to that in control flies already after 75-min HS treatment, which is when PGAM5¹ flies started to die (Fig. 5C and D). These results suggested that HS induced a rapid degeneration of the nuclei of the MB neurons in the absence of dPGAM5, although the overall structures of the neurons were relatively preserved at the same time point.

Based on the previous finding that HS induced apoptosis in the wing imaginal disc of third-instar larvae [6], we examined whether or not apoptosis is involved in the HS-induced degeneration of the MB neurons of adult flies. When HS-induced apoptosis in the MB was examined by TUNEL staining, TUNEL-positive KCs were observed in PGAM5¹ flies, but not in control flies, after 75-min HS treatment (Fig. 5A and B). At the same time point, TUNEL-positive cells in other brain regions we examined, such as the optic lobe, were detected in neither control flies nor PGAM5¹ flies (Fig. S3), suggesting that TUNEL-positive cells were almost restricted to the MB in HS-treated PGAM5¹ flies. Even after 150-min HS treatment, i.e., when approximately 20% of the control flies had died, only a negligible amount of TUNEL-positive cells was detected in the MB of control flies (data not shown). These results indicated that apoptosis was induced in the MB of PGAM5¹ flies to a much greater extent than in that of control flies. Nevertheless, the number of TUNEL-positive cells in the MB of PGAM5¹ flies was much smaller than we expected from the obvious reduction in the number of Histone2B::ECFP-labeled nuclei after 90-min HS treatment (Fig. 4); this may be because each apoptotic cell only transiently reacted to TUNEL according to a rapid degeneration of the nuclei in the absence of dPGAM5. Taken together, apoptosis in the MB may be involved in HS-induced death of individual PGAM5¹ flies but not of control flies.

We next examined whether or not apoptosis in the MB is indeed responsible for the vulnerability of PGAM5¹ flies to HS. To this end, we suppressed HS-induced apoptosis in the MB by selective expression of the caspase inhibitor protein p35 in the MB of PGAM5¹ flies. We found that expression of p35 in the MB of PGAM5¹ flies decreased their vulnerability, and the resulting flies exhibited a similar response to HS to that of control flies (Fig. 5G). On the other hand, expression of p35 in the MB of control flies did not affect their response to HS (Fig. 5H), further supporting that apoptosis in the MB was not involved in HS-induced death of individual control flies. Taken together, the findings indicate that dPGAM5 does appear to exert a protective effect against HS by preventing apoptosis in the MB.

As regards the role of the MB in the whole-body HS response, two possibilities were considered: 1) the MB exerts an apoptosis-sensitive protective function against HS, and 2) the MB acquires an apoptosis-dependent toxic function. To address these possibilities, we chemically ablated the MB by treating flies with hydroxyurea (HU) at the larval stage. Flies lacking the MB are known to develop without any obvious abnormalities [16]–[18]; here, we confirmed that the MB was indeed ablated in adult flies pretreated with HU (Fig. 6A). Interestingly, PGAM5¹ flies lacking the MB were more resistant to HS than those possessing the MB (Fig. 6B), whereas control flies lacking the MB exhibited a similar response to HS to that of flies retaining the MB (Fig. 6C). These results once again demonstrated the critical role played by the MB in the whole-body HS response and suggest that the MB gains a toxic function in the absence of dPGAM5, instead of losing a protective function in response to HS.

Taking account of the function of the MB as a neuronal tissue, we raised the possibility that the toxic function of the MB in HS-treated PGAM5¹ flies might depend on synaptic signaling from KCs to other neuronal cells. To address this possibility, we used a temperature sensitive allele of the Drosophila dynamin gene shibire (shi^TS), which loses its activity at 30°C and above and suppresses synaptic neurotransmission [19]–[21]. However, suppression of neurotransmission during the HS treatment by expression of shi^TS in the MB of PGAM5¹ flies did not decrease but rather increased their vulnerability to HS (Fig. 6D), suggesting that the toxic function of the MB does not depend on synaptic signaling. Thus, the toxic function gained by the MB in response to HS in the absence of dPGAM5 may depend, not on synaptic transmission, but rather on the passive release of toxic factor(s) from apoptotic neurons.

Since the MB have been shown to influence locomotor activity, as well as higher brain functions such as olfactory learning and memory [17], [18], we examined locomotor activity of PGAM5¹ flies during the HS treatment using the Drosophila activity monitor (DAM) system (Fig. 7A). We found that control flies exhibited increased locomotor activity immediately after they were subjected to HS, which peaked around 30 min and then decreased within 60 min. On the other hand, PGAM5¹ flies were overall less active than control flies, and their activity peaked immediately after they were subjected to HS and gradually decreased thereafter within 60 min. We next examined whether or not this reduced locomotor activity of PGAM5¹ flies was restricted to their HS response (Fig. 7B). To this end, we monitored locomotor activity of control and PGAM5¹ flies under normal unstressed conditions for three days. We found that activity of PGAM5¹ flies was constantly lower than that of control flies, suggesting that locomotor activity of PGAM5¹ flies was reduced under both unstressed and stressed conditions. Although it is difficult from these data to determine if the reduced HS-induced locomotor activity affects the increased vulnerability of PGAM5¹ flies to HS, dPGAM5, possibly in the MB, may influence locomotor activity of flies even under unstressed conditions.

Fly cultures and crosses were carried out on normal growth medium (10% glucose, 4% dry yeast, 4% cornmeal and 0.9% agar) at 25°C, unless otherwise stated. Transgenic flies harboring UAS-dPGAM5 and UAS-dPGAM5 H94A were generated by standard P element-mediated transformation (BestGene Inc.). The following fly strains were used in this study: PINK1^B9, revertant PINK1^REV [24]; PGAM5¹, PINK1^B9 [11]; UAS-LacZ IR [25]; pnr-GAL4 [26]; pxn-GAL4 [27]; sca-GAL4 [28]; yolk-GAL4 [29]; TH-GAL4 [30]; UAS-shi^TS [19]; sev-GAL4 [31], repo-GAL4 [32], OK107-GAL4 [13], 201Y-GAL4, c739-GAL4 [12], UAS-mCD8::GFP [33], UAS-Histone2B::ECFP [15], da-GAL4, hs-GAL4 and elav^c155-GAL4 (Bloomington Drosophila Stock Center); and UAS-dPGAM5 IR (R-6) (NIG-FLY stock center).

Adult flies (aged 3–5 days) were subjected to various stressors, and live flies were counted on the basis of movement following manipulation of vials as previously described [34]. For the oxidative stress condition, flies were placed in vials containing a medium containing 0.1% H₂O₂, 1% sucrose and 1.3% agar at 25°C, and live flies were counted daily. For the starvation condition, flies were placed in empty vials at 25°C, and live flies were counted every 3 h. For HS, flies in vials containing normal growth medium were placed in a 37°C water bath, and live flies were counted every 30 min. For the assay of hs>LacZ-IR and hs>PGAM5-IR flies, IR RNA was induced in flies (aged 12–17 days) two days prior to sustained HS by two cycles of HS pretreatment, each of which was composed of 37°C for 30 min, 25°C for 5 hr, 37°C for 30 min and 25°C for 18 hr.

Flies were lysed with RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP40, 0.5% deoxycholate and 0.1% SDS), and lysates were resolved on SDS-PAGE and electroblotted onto PVDF membranes. After blocking with 5% skim milk in TBS-T (50 mM Tris-HCl, pH 8.0, 150 mM NaCl and 0.05% Tween 20), the membranes were probed with dPGAM5 polyclonal antibody [11] and actin monoclonal antibody (Sigma). The antibody-antigen complexes were detected using the ECL system (GE Healthcare).

Immediately after hatching, larvae were fed a heat-killed yeast suspension containing 25 mM hydroxyurea (HU; Invitrogen) for 4 h. Larvae were then washed in distilled water and cultured on normal growth medium at 25°C [16].

Brains dissected from adult flies were fixed with 4% paraformaldehyde in 0.3% Triton X-100/PBS (PBT) for 20 min at room temperature (RT). After three washes with a buffer containing 54 mM NaCl, 40 mM KCl, 7.4 mM MgSO₄, 2.4 mM CaCl₂, 4.8 mM tricine, 0.36% (w/v) glucose, 1.7% (w/v) sucrose and 0.3% Triton X-100 (20 min each), the brains were incubated with 5 µg/ml proteinase K (Nacalai Tesque) for 5 min, followed by three washes with PBT (20 min each) at RT. TUNEL staining was performed using the In Situ Cell Death Detection Kit TMR-Red (Roche) for 1 h at 37°C. Brains were then counterstained with 400 ng/ml Hoechst 33258 for 3 h at RT. Fluorescence images were acquired with an LSM510 confocal microscope (Carl Zeiss). For counting the KCs, Z-series confocal images were collected to cover the KC perikarya cluster (0.75 µm virtual sections), and the total number of KC nuclei visualized by Histone2B::ECFP was calculated by summing up the number counted manually in every tenth section.

Male flies of each genotype (day 5–10) were entrained to a 12 hr light: 12 hr dark cycle (LD12:12) at 25°C for at least 2 days. Flies were transferred individually to glass tubes containing food at one end. Locomotor activity was monitored by recording infrared beam crossings by individual flies using the Drosophila activity monitoring (DAM) system (Trikinetics) [35], [36]. For HS conditions, flies in glass tubes were placed in an air incubator at 37°C, and their locomotor activity was monitored every 5 min. For unstressed conditions, data were collected continuously for at least 3 days under LD12:12 conditions at 25°C.