A pharmacological screen for compounds that rescue the developmental lethality of a Drosophila ATM mutant

Dominant modifier screens in flies have been extremely successful in identifying genes involved in specific biological processes, including genes involved in ATM-mediated neuroprotection [14]. The key feature of dominant modifier screens is a sensitized, intermediate phenotype that permits heterozygous mutations in genes that function in relevant biological processes to dominantly suppress or enhance the phenotype. Based on this paradigm, we set out to identify a sensitized, intermediate ATM-dependent phenotype where compounds that affect the function of ATM or its downstream targets would suppress or enhance the phenotype. It was previously documented that in ATM⁸ flies, ATM kinase activity and development to adulthood is temperature-sensitive [12, 15, 16]. Indeed, we found that when heterozygous ATM⁸/TM3 flies (TM3 is a balancer chromosome that contains a wild-type ATM allele) were self-crossed at 25°C, no ATM⁸ F1 progeny eclosed, indicating that ATM kinase activity is below the threshold needed for viability (Table 1). In contrast, at 18°C, 24% of the F1 progeny were ATM⁸, which approaches the Mendelian ratio of 33% expected for fully viable ATM⁸ flies. Moreover, an intermediate temperature of 21°C produced an intermediate ratio of ATM⁸ F1 progeny, 16%. These data suggest that 21°C is a sensitized condition where compounds that affect ATM kinase activity or signaling through ATM pathways might alter the level of ATM⁸ progeny up to as high as 33% or down to as low as 0%.

To identify compounds that affect the function of ATM, we performed a blinded screen of 2400 compounds from the Spectrum Collection (MicroSource Discovery System) for those that altered the percent eclosion of ATM⁸ flies at 21°C. The library included FDA approved drugs, natural products, and bioactive components. Details of the screening protocol are described in the Materials and methods and illustrated in Fig 1. In brief, for each compound, 200 μl of 0.2 mM compound in 2% dimethyl sulfoxide (DMSO) was evenly applied to the surface of newly prepared molasses food in vials. This concentration was selected based on other screens in flies that used 0.05 or 0.1 mM compounds that were homogenously mixed in food [10, 11, 24]. 50–60 young ATM⁸/TM3 flies were added to the vials, and the flies were cultured at 25°C for 5–6 days to allow egg-laying, at which time the parental flies were removed and the vials were transferred to 21°C for about 12 days, enough time for eggs to have developed to adults. This protocol allowed compounds to be consumed ad libitum by parental flies as well as F1 larvae. ATM⁸ and ATM⁸/TM3 F1 adults were counted and the percent ATM⁸ flies was determined. In the primary screen, we only tested each of the 2400 compounds once, but, in a secondary screen, we retested the top 100 compounds two more times to identify compounds that had a reproducible effect.

Table 2 provides an overview of the findings, and S1 Table provides data for each compound. 147 compounds completely blocked the production of ATM⁸ progeny. Of these, 23 compounds, including several insecticides, also blocked the production of ATM⁸/TM3 flies either by killing the parental flies and preventing egg-laying or by inhibiting the development of progeny. The other 124 compounds that preferentially blocked production of ATM⁸ progeny may provide insights into ATM function; however, there were no obvious commonalities among these compounds. 892 compounds had little or no effect on the development of ATM⁸ relative to ATM⁸/TM3 flies leading to 10–20% ATM⁸ progeny, which was not substantially different than 14% ATM⁸ progeny for the control (i.e., 2% DMSO) (Table 3). 819 compounds were detrimental to the development of ATM⁸ relative to ATM⁸/TM3 flies, reducing ATM⁸ progeny to 1–10%. In contrast, the remaining 542 compounds were beneficial to the development of ATM⁸ relative to ATM⁸/TM3 flies, increasing ATM⁸ progeny to >20%. In summary, the primary screen served the desired purpose of broadly categorizing the collection of 2400 compounds into those that were detrimental, neutral, or beneficial to the development of ATM⁸ flies.

We performed a secondary screen of the 100 compounds that produced the highest percent ATM⁸ progeny in the primary screen (S2 Table). Each compound was tested twice under the same conditions as the primary screen. However, for unknown reasons, controls in the secondary screen had 1% rather than 14% ATM⁸ progeny observed in the primary screen (Table 3). Despite these more stringent conditions, 18 compounds substantially increased the percent ATM⁸ progeny in one of the two trials. Moreover, 10 compounds, listed in Table 3 and hereafter referred to as the top 10 compounds, substantially increased the percent ATM⁸ progeny in both trials. Thus, in three independent biological trials, the top 10 compounds reproducibly suppressed the developmental lethality due to reduced ATM kinase activity.

Three of the top 10 compounds, Ronnel (also known as fenchlorphos), estragole (also known as p-allylanisole), and stigmasterol, are inhibitors of the enzyme acetylcholinesterase (AChE), which catalyzes breakdown of the neurotransmitter acetylcholine (ACh) to choline [25–27]. Inhibition of AChE leads to accumulation of ACh at synapses and stimulation of nerves and muscles in nervous systems. AChE inhibitors have been used in the treatment of Alzheimer’s and Parkinson’s disease patients who have reduced levels of ACh in the brain [28, 29]. Thus, we retested AChE inhibitors (i.e., donepezil hydrochloride, tacrine hydrochloride, dichlorvos, galantamine, and rivastigmine tartrate) from the primary screen that were not included in the secondary screen. In the primary screen, donepezil hydrochloride, tacrine hydrochloride, and dichlorvos weakly suppressed the developmental lethality of ATM⁸ flies, but neither these nor the other AChE inhibitors affected ATM⁸ lethality in the retest (S3 Table). Similarly, four of the AChE inhibitors failed to suppress ATM⁸ lethality at other concentrations, ranging from 0.002 to 2 mM or when used in combination (S4 Table). Nevertheless, direct evaluation of AChE activity and ACh levels are needed to definitively determine the extent to which inhibition of AChE leads to suppression of ATM⁸ lethality.

Ronnel may also act through other mechanisms, as it is a member of the organophosphate class of compounds that not only inhibit AChE but also have non-cholinergic effects, including disruption of mitochondrial oxidative phosphorylation and mitochondrial membrane potential [30]. Interestingly, six of the other top 10 compounds cause mitochondrial dysfunction through a variety of mechanisms; hypoxanthine inhibits oxidative phosphorylation [31], quassin and usnic acid cause mitochondrial depolarization [32, 33], sodium thioglycolate (also known as 2-mercaptoacetate) and avocatin B inhibit mitochondrial fatty acid oxidation [34, 35], and tyramine inhibits mitochondrial respiration [36]. Thus, based on the finding that elimination of defective mitochondria by mitophagy in ATM-deficient neurons and animals has beneficial consequences [8], the identified compounds may trigger mitophagy of partially dysfunctional mitochondria in ATM⁸ cells by further reducing their function and thereby improve the mitochondrial network’s integrity and functionality.

To investigate the mechanisms by which the top 10 compounds suppress the developmental lethality of ATM⁸ flies, we examined their effect on other recessive lethal ATM alleles, ATM³ and ATM⁴, that were identified in the same chemical mutagenesis screen that identified ATM⁸ [15]. ATM³ contains a nonsense mutation that truncates the ATM protein at residue 600 out of 2429, and ATM⁴ contains a missense mutation that changes leucine 284 to histidine. By genetic criteria, ATM³ and ATM⁴ are both null alleles. In three independent tests of each compound at 0.2 mM, we found that none of the top 10 compounds rescued the developmental lethality of even a single ATM³ or ATM⁴ fly. These data argue that some level of ATM activity is required for suppression of developmental lethality by the top 10 compounds.

Among the top 10 compounds, Ronnel (also known as fenchlorphos) stood out because >90% of the F1 progeny from self-crosses of ATM⁸/TM3 flies were ATM⁸, which is considerably higher than the maximum expected 33% if both ATM⁸ and ATM⁸/TM3 flies are fully viable (Table 3). These data indicate that Ronnel is both beneficial to the development of ATM⁸ flies and toxic to the development of ATM⁸/TM3 flies. In support of this conclusion, the number of ATM⁸ flies that eclosed per vial with Ronnel was similar to the number for the other top 10 compounds, but the total number of eggs from which the flies developed was substantially lower because Ronnel was lethal to the parental flies (Table 3). In other words, since a similar number of ATM⁸ flies eclosed from fewer eggs, it means that ATM⁸ flies that usually would have died during development were able to survive, indicating rescue by Ronnel rather than tolerance to Ronnel. Furthermore, the effect of 0.2 mM Ronnel during fly development is dependent on the level of ATM activity; Ronnel was unable to rescue the lethality of flies with no ATM activity (i.e., ATM³ and ATM⁴ at 25°C), it rescued the lethality of flies with low ATM activity (i.e., ATM⁸ flies at 21°C), and it increased the lethality of flies with higher ATM activity (i.e., ATM⁸/TM3 flies at 21°C and ATM³/TM3, ATM⁴/TM6B, and wild-type flies at 25°C).

To further characterize Ronnel, we used the screening protocol to examine the effect of different concentrations of Ronnel on the developmental lethality of ATM⁸ flies. As observed in the primary and secondary screens, 0.2 mM Ronnel produced >90% ATM⁸ progeny (Table 4). A 10-fold lower concentration of Ronnel (0.02 mM) reduced ATM⁸ progeny to 59% and even lower concentrations (0.002–0.00002 mM) reduced ATM⁸ progeny to 14–19%, which was similar to the 20% ATM⁸ progeny observed in DMSO controls. Fewer total progeny were produced at high concentrations of Ronnel because it killed most of the parental flies within 48 hrs thereby reducing the number of eggs laid. Moreover, the percent ATM⁸ flies at 0.02 mM Ronnel was lower than at 0.2 mM Ronnel because of increased viability of ATM⁸/TM3 progeny rather than reduced viability of ATM⁸ progeny. These data indicate that molecular events targeted by Ronnel that suppress ATM⁸ lethality and enhance ATM⁸/TM3 lethality are sensitive to different concentrations of Ronnel.

To investigate whether treatment with Ronnel during the development of ATM⁸ flies has long-lasting effects during adulthood, we determined the lifespan of ATM⁸ and ATM⁸/TM3 flies at 25°C after being raised at 21°C on molasses food, molasses food with 0.2% DMSO (i.e., DMSO), or molasses food with 0.02 mM Ronnel in 0.2% DMSO (i.e., Ronnel). We used 0.02 mM rather than 0.2 mM Ronnel for this and subsequent analyses of Ronnel because it was less toxic to parental ATM⁸/TM3 flies and ATM⁸/TM3 progeny, which led to more ATM⁸/TM3 flies to analyze (Table 4). This study showed that Ronnel had no effect on the lifespan of either ATM⁸ or ATM⁸/TM3 flies (Fig 2). Kaplan-Meier analysis revealed that the mean lifespan of ATM⁸ and ATM⁸/TM3 flies was not significantly different when raised with or without Ronnel (S5 Table). Furthermore, with and without Ronnel, ATM⁸ flies had similar mean lifespans that were significantly shorter than those of ATM⁸/TM3 flies (P<0.01). Thus, the mechanism by which Ronnel affects the developmental of ATM⁸ and ATM⁸/TM3 flies does not cause lasting damage or repair that affects the lifespan of adult flies.

The innate immune response is hyperactivated in glial cells of ATM⁸ flies, and genetic manipulations that reduce the innate immune response prevent neurodegeneration and increase the lifespan of ATM⁸ flies [16, 17]. Thus, Ronnel-mediated suppression of the developmental lethality of ATM⁸ flies might involve effects on the innate immune response. To test this hypothesis, we monitored activation of the innate immune response by quantifying mRNA levels of AMP genes Diptericin B (DiptB), Metchnikowin (Mtk), and Attacin C (AttC). At 24 hrs after development of ATM⁸ flies on molasses food containing 0.2% DMSO (i.e., DMSO) or 0.02 mM Ronnel in 0.2% DMSO (i.e., Ronnel), Mtk and AttC expression was equivalent between DMSO- and Ronnel-treated flies and DiptB expression was about 2-fold lower in Ronnel-treated flies (Fig 3A). Similarly, the expression of all three AMP genes was not affected in ATM⁸ flies fed Ronnel as adults, that is, raised on molasses food at 18°C and transferred to molasses food containing DMSO or Ronnel for 24 hrs (Fig 3B). Longer treatments of ATM⁸ adult flies with 0.02 mM Ronnel were not possible because it killed too many flies within 48 hrs. These data indicate that the innate immune response is unlikely to play a role in Ronnel-mediate suppression of the developmental lethality of ATM⁸ flies or the lethality of adult ATM⁸ flies.

The primary function ascribed to ATM is in the recognition and repair of DNA double-strand breaks [1–3]. Abnormally high levels of DNA damage have been reported in tissue culture cells and organisms, including flies, that contain ATM mutations, and DNA damage may be responsible for triggering neurodegeneration and other phenotypes in A-T patients [12–17]. To test the possibility that DNA damage in the brain contributes to the developmental lethality of ATM⁸ flies, we measured DNA damage using the Comet assay, which involves lysing cells in low-melt agarose and electrophoresing the released DNA [37]. Fragments of damaged DNA migrate faster than intact chromosomes during gel electrophoresis and when visualized with a fluorescent dye resemble a comet (Fig 4A). The amount of DNA in the comet tail relative to the total amount of DNA (i.e., percent tail DNA) is a measure of the amount of DNA damage. We found that cells from dissociated brains of 0–3 day old ATM⁸ flies had a significantly higher percent tail DNA than w¹¹¹⁸ flies (a standard laboratory strain), whereas ATM⁸/TM3 flies had the same percent tail DNA as w¹¹¹⁸ flies (Fig 4B and 4C). Therefore, ATM activity is required in the brain to prevent the accumulation of DNA damage, and the level of ATM activity determines the extent of DNA damage.

To confirm that the Comet assay quantitatively detects DNA damage in the brain, we used it to analyze cells from dissociated brains of 3 day old flies 1 hr after exposure to 50 Gy of ionizing radiation (IR), which induces DNA double-strand breaks. In all of the flies examined, w¹¹¹⁸, ATM⁸/TM3, and ATM⁸, IR significantly increased the percent tail DNA relative to unirradiated flies (Fig 4B and 4C). Thus, the Comet assay is able to detect a wide range DNA damage levels in brain cells.

To investigate whether Ronnel suppresses the developmental lethality of ATM⁸ flies by affecting the level of DNA damage, we used the Comet assay to determine the level of DNA damage in 0–3 day old ATM⁸ and ATM⁸/TM3 flies that were raised at 21°C on molasses food, molasses food with 0.2% DMSO (i.e., DMSO), or molasses food with 0.02 mM Ronnel in 0.2% DMSO (i.e., Ronnel). As observed in Fig 4, ATM⁸ flies had higher percent tail DNA than ATM⁸/TM3 flies under all of the culturing conditions (Fig 5). Moreover, relative to molasses food and DMSO, Ronnel increased the percent tail DNA in both ATM⁸ and ATM⁸/TM3 flies, indicating that Ronnel promotes DNA damage. This is consistent with studies showing that organophosphate compounds induce DNA damage in multiple cell types [38, 39]. Since Ronnel had the same effect on DNA damage in ATM⁸ and ATM⁸/TM3 flies but had opposite effects on the development of these flies, the effects of Ronnel on DNA damage and development are unlikely to be mechanistically related.

Fly stocks were maintained on standard cornmeal-molasses food at 25°C, unless otherwise stated. ATM⁸/TM3,Sb, ATM³/TM3,Sb, and ATM⁴/TM6B,Hu flies (referred to as ATM⁸/TM3, ATM³/TM3, and ATM⁴/TM6B, respectively) were obtained from the Bloomington Drosophila Stock Center. To generate ATM⁸ flies, ATM⁸/TM3 flies were raised at 18°C or 21°C. Heterozygous ATM⁸/TM3 flies were maintained at 25°C.

2400 compounds from the Spectrum Collection (MicroSource Discovery Systems) were obtained in a 96-microplate format, 30 plates of 80 compounds (columns 1 and 12 empty), and were stored at -80°C. Each plate was thawed at room temperature in light protective packaging for 24 hrs before use. After use, all plates were flushed with nitrogen gas, immediately capped, and stored at -80°C to prevent degradation. All compounds were provided in 100% DMSO at a concentration of 10 mM. 6 μl of each compound was diluted in 300 μl of deionized, distilled H₂O (ddH₂O) to a final concentration of 0.2 mM in 2% DMSO. 200 μl of 0.2 mM compound in 2% DMSO or control 2% DMSO was evenly distributed over the top of freshly made molasses food [42] in vials before the food completely solidified. Compounds were allowed to absorb into the molasses food overnight at room temperature. As described in Fig 1, one plate was tested at a time. Compounds were blindly tested through identification by plate number, row, and column position. For each plate screened, four vials of control 2% DMSO were screened. 0–7 day old ATM⁸/TM3 flies raised at 25°C were allowed to lay eggs on molasses food with compound for 5–6 days at 25°C. ATM⁸/TM3 parents were removed from the vials, the vials were incubated at 21°C for ~12 days, and the percent ATM⁸ progeny was determined. For the primary screen, each compound was tested once. For the secondary screen, the 100 compounds with the highest percent ATM⁸ from the primary screen were retested in duplicate.

ATM⁸ and ATM⁸/TM3 flies were raised on molasses food, molasses food with 200 μl of 0.2% DMSO, or molasses food with 200 μl 0.02 mM Ronnel in 0.2% DMSO at 21°C, collected at 0–5 days old, transferred to fresh molasses vials at approximately 20 flies per vial, and aged at 25°C. The day of collection was designated day 1. Surviving flies were counted daily until all flies had died. Flies were transferred to new vials approximately every 3 days. Lifespan assays for the different genotypes and compounds were performed at the same time. >186 flies were examined for each assay condition. The number of flies examined is provided in the legend for Fig 2. Kaplan-Meier survival analysis, standard t-tests, and chi square tests were performed using Oasis online application for survival analysis (sbi.postech.ac.kr/oasis/) and Rstudio statistical software.

To determine the effect of Ronnel on AMP gene expression when fed during development, ATM⁸ flies were raised on molasses food with 0.2% DMSO or molasses food with 0.02 mM Ronnel in 0.2% DMSO at 21°C, collected at 0–7 days post-eclosion, transferred to molasses food vials without compound or DMSO at approximately 50 flies per vial, and incubated at 25°C for 24 hrs. Total RNA was isolated from 50 whole flies for each experimental condition, ATM⁸ DMSO (n = 13) and ATM⁸ Ronnel development (n = 14). To determine the effect of Ronnel on AMP gene expression when fed to adult flies, ATM⁸ flies were raised on molasses food at 18°C, collected at 0–7 days old, transferred to fresh molasses food with 2% DMSO or molasses food with 0.02 mM Ronnel in 0.2% DMSO at approximately 50 flies per vial, and incubated at 25°C for 24 hrs. Total RNA was isolated from 50 whole flies per experimental condition, ATM⁸ DMSO (n = 15) and ATM⁸ Ronnel (n = 12). RNA was isolated using the RNeasy Plus Mini Kit (Qiagen). cDNA was generated by reverse transcription (RT) with the iScript cDNA Synthesis Kit (Bio-Rad). Real-time PCR (qPCR) was carried out as described by Katzenberger et al. (2006) [42]. Primer sequences for DiptB, Mtk, AttC, and Rpl32 are described in Katzenberger et al. (2016) [42].

The Comet Assay Kit (Trevigen, Catalog #4250-050K) was used with the following modifications, fly brains were dissected in cold phosphate-buffered saline (PBS) and placed in 2 ml Eppendorf tubes at a concentration of 1 brain/10 μl of PBS. A minimum of 6 brains was dissected per sample. Samples were maintained on ice until they all were dissected. Samples were homogenized with plastic pestles for Eppendorf tubes, 10 μl of brain homogenate was added to 75 μl of LMAgarose that had been boiled and cooled to 40°C in a heat block. 50 μl of LMAgarose/homogenate mix was immediately added to a CometSlide that had been warmed at 37°C. A 200 μl pipette tip was used to spread the sample evenly across the sample area. Slides were placed flat, at 4°C for 30 min in the dark to cool and harden. Slides were immersed at room temperature in Lysis Solution and incubated at 4°C overnight in the dark. Slides were removed from the Lysis Solution and washed by immersion twice for 10 min in 50 ml 4°C tris-borate-EDTA (TBE). Slides were then placed in an electrophoresis unit side by side lengthwise with the white slide label at the top. The electrophoresis box was filled with 4°C TBE to just covering the sample on the slide and electrophoresed at 23V for 10 min. Excess TBE was drained from the slides, slides were washed twice for 5 min in ddH₂O and then in 70% ethanol once for 5 min, and dried overnight at room temperature in the dark. DNA was stained by placing 100 μl of SYBR Gold solution (30 ml tris-EDTA (TE) buffer and 1 μl SYBR Gold (Molecular Probes, S11494)) on each circle of dried agarose for 30 min at room temperature. Excess SYBR Gold solution was drained from the slides and the slides were quickly rinsed in 50 ml ddH₂O and dried for 20–25 min at 37°C. Comets were imaged using a Zeiss LSM 510 confocal microscope with an Imager.M1 module using a 20X/0.8 NA Plan-APOCHROMAT objective. Fluorescence was excited using an Argon laser at 488 nm. Scan mode was set to frame, frame size was 2048x2048, scan speed was 4 sec, and bit depth was 16 bit. For each sample, >200 comets were imaged and analyzed using CometScore Pro Software (TriTek Corporation). One-way ANOVA and Tukey’s post hoc test was performed using RStudio statistical software to compare percent tail DNA between pairs of samples.

Three day old flies were placed in empty vials at approximately 20 flies per vial. Vials were exposed to 50 Gy (5,000 rad) of gamma rays using a Cesium-137 irradiator (J. L. Sheppard Mark I unit). Brains were dissected and processed for the Comet assay 1 hr after irradiation.