A Screen for Caffeine-sensitive Eye Mutants Reveals Three Loci on Chromosome 3R

The compound eyes of Drosophila are ideal tissues to detect defects in proliferation and apoptosis as they are not essential for survival, but they are sensitive to developmental perturbations and easy to score for mutant phenotypes. To identify novel genes functioning in DNA damage response pathways that are redundant with ATM and ATR, we previously performed a genetic screen to identify conditional eye phenotypes in adult flies fed 2 mM caffeine and 3 mM hydroxyurea (HU) throughout larval development [31]. While caffeine inhibits ATM and ATR, HU stalls replication forks through inhibition of dNTP production, eventually generating single strand or double strand DNA breaks, thereby activating DNA damage responses regulated by ATM and ATR. At the drug concentrations used, there were no phenotypic effects in wildtype flies. In this screen, we used the “EGUF, GMR-hid” (EGUF) system to produce homozygous mutant clonal cells in the entire adult eye of an otherwise heterozygous fly [32]. This screen identified a single caffeine-sensitive locus (huc95E) on chromosome arm 3R, here renamed java no jive (jnj), which we mapped to cytological region 95E by complementation testing with chromosomal deficiencies [31]. Flies that were mosaic hemizygous for jnj in the eye exhibit caffeine-dependent small, rough eyes associated with increased apoptosis. To identify novel DNA damage pathway components, we have now carried out a new screen of chromosome arm 3R for conditional caffeine-sensitive eye phenotypes. By screening 9098 males, we identified three loci on chromosome arm 3R including six additional alleles of jnj, two mutant alleles of a locus called sleepless in seattle (sst), and one allele of a novel locus called double double trouble (ddt), that has not yet been linked to a specific gene (Fig. 1A, Fig. S1). All hemizygous jnj, sst and ddt mutants exhibit caffeine-dependent pupal lethality (Fig. 1B–D and data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Eye phenotypes in caffeine-sensitive mutant flies.

(A) Caffeine-dependent eye phenotype of Smc6 (jnj) and MAGE (sst) mutants. Fly genotypes are as follows. Control: EGUF/+; FRT82B +/FRT82B GMR-hid. Smc6 (loss of Smc6 in eye cells): EGUF/+; FRT82B jnj^R1/FRT82B GMR-hid. MAGE (loss of MAGE in eye cells): EGUF/+; FRT82B sst^RZ/FRT82B GMR-hid. (B-D) Smc6, MAGE or Smc5 homozygous, trans-heterozygous or hemizygous mutants have reduced survival when raised in media with caffeine. Bars represent the survival index (p) and error bars represent SEM. “□” indicates flies eclosed from the same cross. Absence of a bar indicates no surviving flies. Wildtype control flies are w¹¹¹⁸. (B) Smc6 mutants are sensitive to caffeine. R1 (jnj^R1) is an allele from the caffeine screen, X1 (jnj^X1) was generated by an imprecise excision of a P-element adjacent to the 5′UTR of Smc6, and Df (Df(3R)Exel6198) is a deficiency chromosome uncovering the Smc6 locus. (C) MAGE mutants are sensitive to caffeine. RZ (sst^RZ) is an allele from the caffeine screen, XL (sst^XL) is a targeted knockout, and Df (Df(3R)Antp¹) is a deficiency chromosome uncovering the MAGE locus. (D) Smc5 mutants are sensitive to caffeine. Both P5 (Smc5^{P{GSV1}GS3245}) and P7 (Smc5^{P{GSV6}GS14577}) contain P-element insertions in a coding exon of Smc5, and Df (Df(3L)BSC418) is a deficiency chromosome uncovering the Smc5 locus.

https://doi.org/10.1371/journal.pone.0059866.g001

Mutations in Smc6 Cause Caffeine-dependent Defects in java no jive Mutant Flies

Deletion mapping indicated that all of the caffeine-sensitive jnj alleles were viable in hemizygous combinations with deletions uncovering region 95E, indicating that the homozygous lethality of most jnj alleles was caused by second site mutation(s). Homozygotes for one allele, jnj^R1, were viable on regular media, but died at the pupal stage when raised in media containing caffeine (Fig. 1B). Sequencing of candidate genes in the jnj region identified a four base pair deletion in exon two of the FlyBase annotated gene CG5524 (del_ATCT at position 334–337 bp from the presumptive start codon), creating a frameshift resulting in a stop codon at position 133 of the presumptive 1122 amino acid protein (Fig. 2A). The predicted CG5524 protein has highest amino acid identity with SMC6 (Structural Maintenance of Chromosomes 6) in other species. SMC6 regulates chromosome stability in yeasts [7], [8], [9], and is implicated in heterochromatic DNA repair in Drosophila [27]. We tested CG5524 (hereafter called Smc6) and four neighboring genes for levels of expression by quantitative RT-PCR of RNA from whole flies. Levels of Smc6 RNA were greatly reduced with all seven alleles of jnj, ranging from 9% to 24% of control levels (Fig. S2A) whereas nearby genes showed little change in expression. Despite extensive sequencing efforts, we were not able to identify the nature of jnj alleles other than jnj^R1, suggesting that these unmapped mutations reside in as yet unidentified regulatory regions of Smc6. To be certain that our jnj alleles corresponded to Smc6, we generated additional Smc6 lines by imprecise excision of the P-element present in line NP2592, including the new line jnj^X1 that lacks exon 1 and sequences up- and downstream of this exon (Fig. 2A). We tested caffeine sensitivity in all of the jnj allelic combinations and found that raising larvae on 0.5 mM caffeine resulted in almost complete lethality (Fig. 1B). Using RNAi to deplete Smc6 expression in developing eye discs also resulted in a caffeine-dependent rough eye phenotype (Fig. S2B). Collectively, the presence of a frame shift mutation in Smc6 in jnj^R1, the reduced expression levels of Smc6 in all seven alleles of jnj, the caffeine-dependent lethality of the deletion allele jnj^X1, and caffeine-dependent eye phenotypes induced by Smc6 RNAi all implicate CG5524/Smc6 as the relevant gene in jnj mutants.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Overview of Smc6, MAGE, and Smc5 gene location, structural organization and mutant alleles.

(A) Smc6 is a 14 exon gene located on 3R:95E8–95F1. jnj^R1 contains a 4 bp deletion in the 2nd coding exon. jnj^X1 contains a 473 bp deletion of sequences upstream of exon 1 (196 bp), the entire exon 1 (252 bp), and a portion of intron 1 (25 bp), with a 12 bp vestige of the original P element remaining. Smc6 genomic locus (3R:20,014,770.20,019,145 [−]) is shown. (B) MAGE is a single exon gene located on the right arm of the 3rd chromosome at position 84C7–84C7. sst^RZ has a point mutation that converts a glutamine at position 109 to a stop codon. sst^XL carries a targeted deletion of the entire coding sequence of MAGE. MAGE genomic locus (3R:2,979,960.2,980,898 [−]) is shown. (C) Smc5 is a 16 exon gene located in 78D6–78D7 of the left arm of the 3rd chromosome. Exons encoding the longest transcripts are shown. Both P{GSV1}GS3245 and P{GSV6}GS14577 are inserted in the second coding exon. The Smc5 genomic locus (3L:21,562,309.21,566,623 [+]) is shown. CDS, coding sequence.

https://doi.org/10.1371/journal.pone.0059866.g002

Caffeine-sensitivity in sleepless in seattle Mutants is Due to Mutations in the MAGE Gene

The sst^RZ mutation exhibits caffeine-dependent pupal lethality in combination with a chromosomal deficiency (Df(3R)Antp¹, Fig. 1C) but sst^RZ homozygotes are not viable on regular media, presumably because of a second site mutation. Further deletion mapping refined the position of the caffeine-sensitive sst locus to a region containing seven candidate genes, each of which were sequenced. We identified a glutamine to stop mutation affecting the MAGE gene [33] in sst^RZ, at position 109 of the 232 amino acid Mage protein (Fig. 2B). In previous studies, depletion of MAGE mRNA using double strand RNA injection suggested that MAGE was essential for viability during early embryogenesis, whereas conditional knockdown at later developmental stages suggested a role in postembryonic neuronal survival and proliferation [34]. Moreover, DNA fibers connecting mitotic cells were observed after RNAi-mediated depletion of Smc5 or Smc6 in S2 cells, suggesting that the Smc5/6 complex could be essential for mitosis in Drosophila [27]. We therefore initially reasoned that sst^RZ was a partial loss-of-function allele, since hemizygous sst^RZ flies were viable. To test this idea we synthesized a knockout allele by homologous recombination [35]. In this new allele (sst^XL) the complete coding sequence of MAGE was deleted (Fig. 2B). Surprisingly, homozygous sst^XL flies displayed no increased lethality or obvious mutant phenotype when raised on media without caffeine. As with sst^RZ hemizygotes, sst^XL flies reared in caffeine media were inviable, but they were less sensitive to a lower dose of caffeine (0.5 mM) than jnj mutants (Fig. 1C). About 15% of predicted sst^XL homozygous flies survived 2 mM caffeine exposure and the surviving flies often had small or rough eyes, similar to sst^RZ mutants (Fig. 1A). Transheterozygous sst^RZ/sst^XL progeny were also viable on normal media, but only 6% survived on 2 mM caffeine (Fig. 1C). Using polyclonal antibodies directed against Mage [36] we found that Mage was absent from protein lysates derived from sst adult flies (Fig. S3). In addition, caffeine-dependent lethality of sst^XL can be complemented by a genomic MAGE transgene (Table S1) that includes the full coding region of MAGE and 3 kb sequence upstream and expresses Mage protein at normal levels (Fig. S3). Collectively, the identification of a stop mutation in the MAGE gene (sst^RZ), the caffeine-sensitivity of a MAGE knockout allele sst^XL, the loss of Mage protein in sst flies and the rescue of caffeine sensitivity by a MAGE transgene all implicate MAGE as the mutated gene in sst flies.

Smc5 Mutant Flies are also Caffeine Sensitive

In yeasts and mammalian cells, all known SMC6 functions involve SMC5 [23], [37], so we predicted that loss of Smc5 activity would also cause caffeine sensitivity in flies (Fig. 3A). We tested two P insertion alleles predicted to affect Smc5 for caffeine sensitivity, namely Smc5^{P{GSV1}GS3245}, referred to as Smc5^P5, and Smc5^{P{GSV6}GS14577}, referred to as Smc5^P7 [38]. As predicted, both Smc5 mutants were sensitive to caffeine (Fig. 1D). Both of these alleles have P-element insertions within the second exon of Smc5 and the insertion sites are very close to the putative start codon (Fig. 2C). Therefore, they are very likely to be null alleles. To rule out the possibility that caffeine-sensitivity of Smc5 flies was caused by second site mutations, we generated fly lines in which the P-elements in both alleles were excised by a transposase, either restoring the wild-type sequence or resulting in an insertion or deletion of the original P element insertion in the coding exon of Smc5. We therefore predicted that some excision lines would no longer be caffeine-sensitive while others would retain the mutant phenotype. As expected, of 13 independent fly lines produced by the excision of P7, seven lines were no longer caffeine sensitive (Table S2A). Similar results were obtained from the excision of P5 (Table S2B). In conclusion, as with Smc6 and MAGE, loss of Smc5 function results in caffeine-dependent lethality.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Mage is part of the Drosophila Smc5/6 complex.

(A) Diagram of a generic Smc5/6 complex in S. pombe (adapted from [70]). The structure in S. cerevisiae is different in that Nse5/6 were found to bind at the hinge. (B) Mage interacts with Nse4 when both proteins are co-expressed in S2 cells. HA-Nse4 co-immunoprecipitated (co-IP) with FLAG-Mage from an S2 cell lysate when two proteins were co-expressed; FLAG-Mage co-IPed with HA-Nse4 from the S2 cell lysate when two proteins were co-expressed. (C) Recombinant Mage interacts with Nse4 and Nse1 directly. Immobilized maltose binding protein (MBP)-fused MAGE or MBP were incubated with ³⁵S-methionine labeled Mage, Nse4, Nse1, or luciferase (as a negative control), respectively. Proteins that were associated with immobilized MBP-Mage or MBP were resolved with SDS-PAGE and visualized by autoradiography. Results show that Mage, Nse4, and Nse1 each interact with MBP-Mage but not with MBP and luciferase does not interact with either of these proteins. (D) Coomassie staining of protein immobilized on 10 µl of amylose beads showed that approximately equal amounts of MBP-Mage and MBP proteins were immobilized on resin beads.

https://doi.org/10.1371/journal.pone.0059866.g003

Caffeine Sensitivity is Mediated through Smc5/6

At the whole organism level, a higher proportion of MAGE mutants were able to survive exposure to 0.5 mM caffeine throughout larval development than Smc6 and Smc5 mutants. Indeed all genetic combinations of MAGE mutant flies had some survivors on media containing 2 mM caffeine, while there were essentially no survivors among the Smc5 or Smc6 mutants raised on 2 mM caffeine (Fig. 1B–D). This suggests that the Mage protein is less important for caffeine resistance than the Smc5 and Smc6 proteins. To further test this hypothesis, we measured the viability of flies carrying mutations in two different components of the protein complex when raised on media containing caffeine. Flies deficient for both Mage and Smc6 were more sensitive to caffeine than flies deficient for Mage alone, but were similar in sensitivity to flies deficient for Smc6 alone (Table S3). This suggests that the Smc5/6 heterodimer has a more critical role in caffeine resistance than does the sub-complex containing Nse1-Mage, consistent with observations in yeasts [1].

Drosophila Smc5/6 Components Form a Protein Complex

In yeasts, the Smc5/6 complex consists of Smc5, Smc6 and six Nse (non-Smc element) subunits [14], four of which were also identified in humans [15], [23]. In searches of Drosophila genome databases, we uncovered a set of putative transcription units that appear to correspond to SMC5/6 complex subunits in yeasts (Table S4). Of these, MAGE has previously been described as a homolog of yeast Nse3 and human MAGEG1 [23]. In Drosophila, Mage protein was shown to interact with Drosophila Nse4 (Nse4) using a yeast two-hybrid system [39]. When we examined the Gene Expression Omnibus (GEO, [40]) to compare gene expression profiles, we found that these two genes have very similar expression patterns across different tissues, supporting the idea that the encoded proteins function in a complex (Fig. S4). Fission yeast Nse1 has been detected in the same sub-complex as Nse3 and Nse4, as part of the larger Smc5/6 complex (Fig. 3A) [11]. We first tested for a physical interaction between Drosophila Mage and Nse4 in cell culture, by generating epitope-tagged plasmid constructs that produce HA-tagged Nse4 or FLAG-tagged Mage, and co-transfecting them into Drosophila Schneider 2 (S2) cells. We were able to co-immunoprecipitate HA-Nse4 and FLAG-Mage from S2 cell lysates (Fig. 3B). We then performed in vitro pull down experiments to show that this interaction is likely direct, and that Mage also interacts with Nse1 directly (Fig. 3C). These results indicate that the three Drosophila proteins (Nse1, Mage and Nse4) form a sub-complex analogous to that found in yeast, consistent with conservation of structure across species.

Loss of Function for Smc6 or MAGE Sensitizes Imaginal Cells to Caffeine-induced Apoptosis

Previous examinations of jnj^huc95E hemizygous mutants were based on the EGUF eye mosaic system [31]. In this experiment, we observed caffeine-dependent defects in ommatidial patterning and increased apoptosis in the eye discs. Larvae mutant for Smc6 or MAGE die at the pupal stage when raised long term on caffeine-containing media. Remarkably, upon dissection of these larvae we noticed that the imaginal discs were severely damaged or altogether absent, suggesting increased cell death as the cause of this defect. To test this hypothesis, we dissected eye imaginal discs from late third instar larvae and labeled them with antibodies against activated caspase 3 to mark apoptotic cells. We detected minimal labeling of apoptotic foci in eye discs of control larvae, regardless of caffeine exposure (Fig. 4). In contrast, dramatically increased labeling of apoptotic foci were seen in the eye discs of Smc6 or MAGE mutant third instar larvae after short term (12 hours) caffeine exposure. Apoptotic labeling was markedly enhanced in a band of cells immediately anterior to the morphogenetic furrow, where cells become synchronized in G1 phase [41]. These results suggest that caffeine-induced apoptosis in developing imaginal discs likely underlies caffeine-dependent pupal lethality in MAGE and Smc6 mutant flies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Caffeine exposure results in apoptosis in eye discs of MAGE and Smc6 mutants.

(A) Anti-cleaved-caspase-3 antibody staining of eye discs from third instar larvae of control (WT, FRT82B), MAGE (sst^RZ/sst^XL), and Smc6 (jnj^X1/jnj^R1) genotypes raised in either standard media (0 mM caffeine) or media supplemented with 2 mM caffeine for 12 hours before dissection. Images are single stacks of confocal images. More cleaved-caspase-3 foci in eye discs of sst^RZ/sst^XL and jnj^X1/jnj^R1 larvae were observed after caffeine exposure. A narrow band of apoptotic cells (white arrow heads) anterior to the presumptive morphogenetic furrow are most noticeable. Scale bar represents 50 µM. (B-D) Quantification and comparison of cleaved caspase-3 staining levels in WT (B), MAGE (C) or Smc6 (D) eye discs, comparing the no caffeine and 2 mM caffeine groups. Data represent mean area stained from multiple eye discs for each genotype per treatment. A maximum projection of all stacks of a confocal image was used to quantify the signal intensity of staining. This value was divided by the area of each eye disc to obtain a ratio representing the relative amount of immunostaining. Error bars represent SEM. A non-paired two-tailed t-test was used to determine statistical significance. **, P = 0.006, ***, P<0.0001.

https://doi.org/10.1371/journal.pone.0059866.g004

Smc5/6 Mutant Flies are Hypersensitive to Genotoxic Stress

The DNA damage response is a multi-step process that involves sensing of damage, cell cycle arrest, and repair of the damaged DNA. Yeast with hypomorphic mutations affecting Smc6, Nse1, Nse2, Nse3 or Nse4 are hypersensitive to gamma irradiation, UV light, MMS, camptothecin (a topoisomerase I inhibitor), and inhibition of DNA replication by HU [26]. All of these genotoxic stresses directly or indirectly generate DNA single-stranded or double-stranded breaks. To explore whether Drosophila Smc5/6 provides similar responses to genotoxic stress, we analyzed the effects of ionizing radiation, camptothecin, HU or MMS on viability. Exposure to 40 Gy ionizing radiation caused increased lethality in MAGE, Smc6 and Smc5 mutants compared to controls (Fig. 5). Moreover, all three mutants were hypersensitive to camptothecin, HU and MMS, compared to controls (Figs. S5 and data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Smc5/6 mutants are hypersensitive to ionizing radiation.

(A–C) Smc6, MAGE or Smc5 homozygous, trans-heterozygous or hemizygous mutants have reduced survival when exposed to 40 Gy of IR. Bars represent the survival index (p) ± SEM. “□” indicates flies eclosed from the same cross. Absence of a bar indicates that no flies survived at that IR dose. (A) Smc6 mutants are hypersensitive to IR. R1 (jnj^R1) and X1 (jnj^X1) are Smc6 alleles. Df (Df(3R)Exel6198) is a deficiency chromosome uncovering the Smc6 locus. (B) MAGE mutants are hypersensitive to IR. RZ (sst^RZ) and XL (sst^XL) are MAGE alleles. Df (Df(3R)Antp¹) is a deficiency chromosome uncovering the MAGE locus. (C) Smc5 mutants are hypersensitive to IR. P5 (Smc5^{P{GSV1}GS3245}) and P7 (Smc5^{P{GSV6}GS14577}) are Smc5 alleles. Df (Df(3L)BSC418) is a deficiency chromosome uncovering the Smc5 locus.

https://doi.org/10.1371/journal.pone.0059866.g005

Loss of Smc5/6 Function does not Compromise G2/M and S Phase Checkpoints Induced by Genotoxic Agents

Studies in Drosophila have proven to be valuable for the study of proteins and pathways controlling DNA repair and checkpoint responses, which are remarkably well conserved among flies and other organisms [27], [42]. In S. pombe, nse3-1 hypomorphic mutants activate a DNA damage checkpoint that arrests cells in late S phase/G2 [7], and Smc6 (Rad18) is required for maintenance but not activation of the G2 checkpoint [43], [44]. We therefore tested whether cell cycle checkpoints important for DNA damage response pathways were perturbed in caffeine-sensitive MAGE or Smc6 mutant flies. To assess G2/M checkpoint function we used ionizing radiation (IR) to determine if IR exposure decreased the number of mitotic cells [45]. We dissected eye imaginal discs from late third instar larvae and labeled them with anti-phospho histone H3 antibodies to mark mitotic cells. The number of mitotic cells in un-irradiated eye imaginal discs of jnj^R1 (Smc6) or sst^XL (MAGE) larvae was comparable to that of control eye discs (Fig. 6A). Larvae were exposed to 40 Gy of IR and dissected eye discs were examined from 15 to 120 min. after exposure. Phospho-histone H3 foci disappeared after 30 or 60 min in wild-type (Iso) controls, jnj^R1/X1 (Smc6) and sst^XL/RZ (MAGE) eye discs (Fig. 6A), demonstrating that neither Mage nor Smc6 is required for activation of the G2/M checkpoint.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Smc5/6 genes are not required for G2/M and S phase checkpoints induced by genotoxic agents.

(A) Wandering third instar larvae were irradiated with 40 Gy of ionizing radiation and the eye-antenna discs were dissected and fixed 15 minutes, 30 minutes, 1 hour or two hours after radiation, with discs from unirradiated larvae serving as controls. Representative images of PH3 staining for mitotic cells in eye-antenna discs from control (WT, FRT82B) and Smc6, (jnj^R1/jnj^X1) transheterozygous larvae are shown. (B) Eye-antenna discs from wandering third instar larvae were incubated with or without HU before adding BrdU to the incubation solution. Representative images of BrdU staining for cells in S phase in eye-antenna discs from control (WT, FRT82B), transheterozygous Smc6 (jnj^R1/jnj^X1) or transheterozygous MAGE (sst^RZ/sst^XL) eye-antenna discs are shown.

https://doi.org/10.1371/journal.pone.0059866.g006

The caffeine sensitive ATM/ATR kinases are important mediators of DNA damage checkpoints [28]. In S. pombe, the SMC5/6 complex is recruited to and stabilizes stalled replication forks after Rad3 (ATR homolog) activation [46]. To investigate whether the S phase checkpoint was intact in jnj^R1/X1 (Smc6) and sst^XL/RZ (MAGE) mutant flies, we monitored BrdU incorporation pattern in eye imaginal discs before and after treatment with HU, which induces the S phase checkpoint [47]. We observed many S-phase cells incorporating BrdU in control untreated eye discs, however incorporation was abolished upon exposure to HU. BrdU incorporation was also abolished by HU treatment in jnj^R1/X1 and sst^XL/RZ mutant discs (Fig. 6B), demonstrating that Mage and Smc6 are also not essential for S phase checkpoint activity in Drosophila.

Smc6 and MAGE Genetically Interact with Proteins Required for DNA Damage Responses

Caffeine inhibits ATR and ATM kinase activity [29], [30], raising the possibility that partial loss of ATM or ATR function could be contributing to the caffeine-induced defects that we observed in Smc5/6 mutant flies. We therefore examined whether genetically reducing ATM or ATR function in an Smc6 mutant background would cause synthetic lethality. The Drosophila homolog of ATR is Mei-41 [48] and mei-41 mutants are homozygous viable but not caffeine-sensitive on their own [31]. To test for genetic interactions between mei-41 and Smc6, we generated double mutants and measured the proportion that survived to adulthood when raised on caffeine-free media. There was no increased lethality associated with mei-41;Smc6 double mutants (Table S5), implying that the inhibition of ATR alone by caffeine was not the main cause of caffeine-dependent lethality of Smc6 homozygotes. To further examine genetic interactions between ATR and MAGE or Smc6, we used the EGUF system as a more sensitive system for detecting mutant phenotypes than lethality. Raised on standard media, adult flies with homozygous MAGE mutant eyes were indistinguishable from control flies (Fig. 7). Raised on 2 mM caffeine, however, MAGE mutant eyes were moderately rough relative to control eyes. ATR RNAi alone caused no observable roughness in the eye but when ATR RNAi was expressed in MAGE-deficient eyes, moderate to severely rough caffeine-dependent eye defects were observed that were not seen on caffeine-free media (Fig. 7, quantification in Fig. S6). We then tested whether ATM plays a role in caffeine sensitivity. Drosophila ATM (tefu) null mutants are non-conditional pupal lethal [49], so we used the EGUF system to examine these interactions as well. ATM-RNAi knockdown alone produced a normal looking eye, either in the absence or presence of caffeine. When MAGE mutant eyes were combined with ATM-RNAi, however, we observed a range of caffeine-dependent rough eye phenotypes, similar to eye defects caused by ATR-RNAi in MAGE-deficient eyes (Fig. 7, S6). ATR-RNAi knockdown alone produced a normal looking eye, either in the absence or presence of caffeine. We noted differences in expressivity between the MAGE-deficient eyes (compare Figs. 1A and 7A) that could be caused by slight differences in the genetic background (the genetic interaction study used CyO balancers while the original screen had wild type chromosomes) or the accumulation of genetic modifiers. We propose that the caffeine-induced partial loss of function of both ATM and ATR causes the rough eye phenotype in the MAGE-deficient background, and that further loss of either ATM or ATR increases the severity of this phenotype, We also examined interactions with NBS1, a component of the MRN (Mre11, Rad50 and Nbs1) complex that collaborates with ATM in DNA repair and telomere maintenance [50]. While NBS1-knockdown alone produced no effect, a dramatic caffeine-dependent enhancement of the rough eye phenotype was observed when NBS1-RNAi was combined with eye-specific MAGE mutants (Fig. S7). These striking caffeine-dependent genetic interactions between MAGE and ATR, ATM, and NBS1 suggest that these proteins act together in maintaining genome stability. Similar genetic interactions were observed between ATR and ATM in Smc6 eye-specific mutants, supporting this conclusion (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Caffeine-dependent genetic interaction of MAGE with ATM, ATR and Rad51(SpnA).

(A) Representative eye phenotypes of MAGE (EGUF/+; FRT82B sst^RZ/FRT82B GMR-hid, loss of MAGE in eye cells), ey>ATMi (knockdown of ATM in eye cells), ey>ATMi;MAGE (EGUF/UAS-ATM-RNAi;FRT82B sst^RZ/FRT82B GMR-hid, loss of MAGE and knockdown of ATM in eye cells) and ey>ATRi;MAGE (EGUF/UAS-ATR-RNAi;FRT82B sst^RZ/FRT82B GMR-hid, loss of MAGE and knockdown of ATR in eye cells) flies that were reared on either standard media or media containing 2 mM caffeine. The EGUF system carrying the eyeless-Gal4 driver was used to drive the UAS-RNAi transgenes in the eye and also makes the eye homozygous for MAGE (sst^RZ). Controls for the effects of each eyeless-driven RNAi alone were carried out for ATM and ATR resulting in wild type appearing eyes, but only the results of ATM RNAi are shown here as an example. (B) Representative eye phenotypes of MAGE knockdown (eyeless-Gal4/+;UAS-MAGE-RNAi/UAS-Dicer2, knockdown of MAGE in eye cells) and MAGE Rad51 double knockdown (eyeless-Gal4/UAS-SpnA-RNAi;UAS-MAGE-RNAi/UAS-Dicer2, knockdown of MAGE and Rad51 in eye cells) flies that were reared on either standard media or media containing 2 mM caffeine.

https://doi.org/10.1371/journal.pone.0059866.g007

Drosophila MAGE RNAi Caffeine Sensitive Phenotype is Rescued by Rad 51 Knockdown

In Drosophila and other organisms, Smc5/6 functions in the homologous recombination repair pathway in DNA double strand break repair [26], [51], [52]. Rad51 is a key component of the homologous recombination pathway, regulating the rate-limiting step of homology searching and strand invasion. In Drosophila, Smc5/6 prevents precocious Rad51 loading onto irradiation damaged heterochromatin region before it moves outside of the HP1a domain for proper repair [27]. In yeast, Smc5/6 mutants accumulate unresolved DNA structures, and Smc5/6 actively resolves DNA mediated sister chromatin linkages [53], [54], [55]. We therefore tested whether the caffeine-dependent rough eye phenotype of Smc5/6 mutants is related to deregulated Rad51 activity. Knocking down Rad51 in the MAGE-RNAi background rescued the rough eye phenotype of MAGE-RNAi flies in 80% of the double RNAi flies raised on 2 mM caffeine (Figs. 7B, S8). Taken together, these data indicate that the caffeine sensitivity of the Smc5/6 complex or at least of MAGE mutants is largely attributable to improper Rad51 activity. It is also possible that Rad51 action is normal during HR, but the Smc5/6 complex mutants are unable to complete HR repair or resolve HR intermediates.

Drosophila Stocks and Husbandry

All crosses were carried out at 25°C, and flies were maintained on media formulated at the Bloomington Drosophila Stock Center at Indiana University (BDSC) with p-Hydroxy-benzoic acid methyl ester or propionic acid as the fungicide. Stocks were obtained from the BDSC, the Vienna Drosophila RNAi Center (VDRC), or the Drosophila Genetic Resource Center at Kyoto (DGRC) or generated in our laboratories where specified. Fly stocks used were:

y¹ w*; P{70FLP}11 P{70I-SceI}2B sna^Sco/CyO, S2.

w¹¹¹⁸; P{70FLP}10; Sb¹/TM6, Ubx.

y¹ w^67c23 P{Crey}1b; D*/TM3, Sb^1.

P{GawB}NP2592.

w*; Dr¹/TMS, P{Delta2-3}99B.

P{GSV1}GS3245.

P{GSV6}GS14577.

P{ey3.5-GAL4.Exel}2.

C(1)DX, y [1] f [1]/w [1] mei-41[D3].

UAS-ATR-RNAi.

UAS-ATM-RNAi.

UAS-NBS1-RNAi.

UAS-SpnA-RNAi.

UAS-MAGE-RNAi/CyO (TRiP).

Ethyl Methanesulfonate (EMS) Screen for Caffeine-sensitive Mutants on Chromosome 3R

The isogenized fly stock FRT82B carries a transgenic Flippase Recognition Target (FRT) site inserted at polytene segment 82B on chromosome 3R and was used to screen for caffeine sensitivity. Adult male flies were mutagenized by feeding with 15 mM EMS dissolved in 1% sucrose for 12 h. After a one day recovery period, mutagenized males were crossed to EGUF; FRT82B GMR-hid, CL/TM3, Sb virgin females. Three to five F1 progeny EGUF/+; FRT82B/FRT82B GMR-hid, CL males with normal eye morphology were crossed to EGUF; FRT82B GMR-hid, CL/TM3, Sb virgin females. The F2 progeny were raised in media with 2 mM caffeine. Individual male non-balancer F2 flies displaying abnormal eye morphology in both eyes were backcrossed to EGUF; FRT82B GMR-hid, CL/TM3, Sb virgin females, and the F3 progeny were raised in media without caffeine to identify any flies with caffeine-independent eye defects. Once the caffeine-dependence of the eye phenotype was confirmed, each mutation was mapped by complementation with the original jnj^huc95E allele [31] or using the Drosophila 3R deficiency kit (BDSC). Both the jnj^R1 and sst^RZ lines emerged from this screen.

Sequencing of Candidate Genes

Targeted re-sequencing of mapped caffeine-sensitive loci was used to identify mutations in candidate genes. Genomic DNA from 50 adult flies was extracted using DNAzol reagent (Invitrogen, Burlington, ON, Canada). Overlapping PCR fragments about 10 kb in size were amplified using a Long Range PCR kit (Invitrogen). These fragments covered each region predicted to contain a mutation and 10 kb on either side. The PCR products were sequenced using Illumina technology and data was analyzed with Bowtie software (Illumina Inc., San Diego, CA) [66]. Mutations were confirmed by Sanger sequencing with BigDye v3.1 (Applied Biosystems, Carlsbad, CA). Restriction digestion (BpmI) of a genomic PCR fragment was used to confirm the mutation in jnj^R1.

Generation of the MAGE Allele sst^XL Using Gene Targeting

The “ends-out” method [35] was used to produce a targeted deletion of MAGE. Specifically, 3 kb genomic regions upstream and downstream of the MAGE genomic locus were amplified by PCR from a Drosophila BAC clone (BACPAC Resources Center, RP98-3E11), using the following PCR primers 5′-ATTCATGCGGCCGCCGAAACTCAAACGCAGCGAA and 5′-ATTCTAGGTACCGAGAAGTGCTAGCCATTTCGAG or 5′-ATTCTAGGCGCGCCGGAGTAAACGCGGAGTAGAATACC and 5′-ATTCATCGTACGGGAAGGGGATCAGGATTGAA. The two PCR fragments were subcloned into the NotI-KpnI (Acc65I) or AscI-BsiWI sites of the ends-out vector P[w25.2] to produce a donor construct P[w25.2]_NK_AB. Seven transgenic lines were generated by P element transformation of a w¹¹¹⁸ strain using P[w25.2]_NK_AB (BestGene Inc, Chino Hills. CA). The three lines in which the P[w25.2]_NK_AB was located on chromosome 2 were tested for efficient excision by crossing to a line carrying the FLP recombinase (w¹¹¹⁸; P{ry+t7.2 70FLP}10; Sb¹/TM6, Ubx). One of the three transgenic lines (6030-1-6M) with the highest excision efficiency was chosen as the donor line, and crossed to y¹ w*; P{70FLP}11 P{70I-SceI}2B sna^Sco/CyO, S2 (BDSC #6934). The parents were allowed to lay eggs for two days in a vial, and on the third day the larvae were heat-shocked for 1 h in a 38°C water bath. F1 virgin females were collected and crossed to w¹¹¹⁸; P{70FLP}10; Sb¹/TM6, Ubx (BDSC #6938) males. About 100 F2 progeny were selected by screening for nonwhite flies from about 1000 independent crosses. Each of these progeny was crossed to w¹¹¹⁸; P{70FLP}10; Sb¹/TM6, Ubx to make stocks. Twenty five independent lines were identified that exhibited correct targeting as detected by PCR of genomic DNA and loss of Mage protein expression by immunoblotting with a guinea-pig anti-Mage antibody [36]. The white marker of these lines was removed by crossing to a line carrying a Cre recombinase (y¹ w^67c23 P{Crey}1b; D*/TM3, Sb¹ (BDSC #851). The resulting lines were tested for heterozygote and homozygote viability under normal conditions, yielding the line named sst^XL.

Generation of a Genomic Rescue Construct for MAGE on Chromosome 2

Genomic DNA was isolated from the isogenized strain P{ry[+t7.2] = neoFRT}82B to PCR amplify (Sequal Prep Long PCR Kit, Invitrogen) a 4 kb fragment spanning from 3 kb upstream of the MAGE gene (genomic locus 3R:2983898, based on the predicted transcription start site), to 206 bp downstream MAGE stop codon (genomic locus 3R: 2979891). The PCR product was digested with the restriction enzyme XbaI and cloned into the pCasper-hs vector. Transgenic flies were generated by BestGene Inc.

Generation of Additional Smc6 Alleles by P-element Mediated Excision

The Smc6 deletion allele jnj^X1 was generated by imprecise excision of a P element in P{GawB}NP2592 (DGRC #104251). This insertion, hereafter referred to as NP2592, is located 7 bp upstream of the putative transcriptional initiation site of CG5524 (Smc6) (3R:20,014,770.20,019,145). Its location was confirmed by genomic PCR using primers flanking the NP2592 locus. To excise out NP2592, NP2592 virgin females were crossed to w*; Dr¹/TMS, P{Delta2-3}99B (BDSC #1610) males carrying a Δ2–3 transposase. Single virgin F1 females of genotype ΔNP2592/TMS,{Δ2–3}99B were crossed to Ly/TM3, Sb males. Single F2 males of genotype ΔNP2592/TM3, Sb were crossed to virgin Ly/TM3, Sb virgin females to establish balanced lines. About 200 candidate lines were produced and subsequently tested for sensitivity to 2 mM caffeine. Six lines were found to be homozygous viable but caffeine-dependent lethal. Genomic PCR was used to confirm that there were deletions around the original P insertion sites in these stocks. One of the resulting lines was renamed jnj^X1.

Molecular Characterization of Smc5 Alleles

The location of P{GSV1}GS3245 (BDSC #200582) and P{GSV6}GS14577 (BDSC #205862) within coding exon 2 of the Smc5 gene was confirmed by genomic PCR using primers 5′-CGTTTCCACGATTTGTTACTGACA and 5′-CGTTTTTGCTTCTTAACCAGATCAC. These lines were renamed Smc5^P5 and Smc5^P7, respectively. Df(3L)BSC418 (BDSC #24922) is a sequence mapped chromosome deletion (78C9;78E1) that includes the Smc5 locus and nearby genes.

Embryo Collection, Drug Administration and Ionizing Radiation (IR) Treatment

Parental flies were allowed to lay eggs in collection cages on apple juice agar plates with yeast paste for 20 h. The eggs were gently removed from the agar plates using distilled water and a brush and collected using a small cloth-bottomed basket, and then arrayed on new apple juice agar plates. For each drug or radiation treatment, at least 100 embryos were transferred with a thin layer of agar underneath into each of 3 vials containing medium. Drug stocks were pre-added into the media to the appropriate working concentration, with the exception of methyl methanesulfonate, which was added into the medium 48 hours after transferring the embryos. For drugs dissolved in DMSO, an equal amount of DMSO alone was added into medium fed to control flies. The following drugs were used: caffeine (Sigma-Aldrich, St. Louis, MO, stock 1 M in water, final concentration 0.25–2 mM); camptothecin (Sigma-Aldrich, stock 25 mM in DMSO, final concentration 0.025 mM), methyl methanesulfonate (Sigma-Aldrich, stock 99%, final concentration 0.005–0.015%) and HU (Sigma-Aldrich, stock 1 M, final concentration 4–8 mM). For IR, third instar larvae were irradiated at doses of 20 and 40 Gray using an irradiator (Gammacell 220–Cobalt-60, Atomic Energy of Canada, 1979). The survival index (p) of a given genotype was calculated by dividing the number of adult survivors of the genotype resulting from media with a given reagent concentration or treatment (n) by the number of adult survivors of the same genotype resulting from media without that reagent or treatment (N).

Immunoblotting

For each sample, ten 3–4 day-old adult flies were collected, frozen in liquid nitrogen and ground using a pestle in a 1.5 ml eppendorf tube. Mild lysis buffer (50 mM Tris, 150 mM NaCl, and 1% Triton X-100, pH 8.0) was then added (10 µl per fly) to solubilize the tissue. The suspension was centrifuged at 20,000g for 10 min. at 4°C and the supernatant was mixed and boiled with 2X Laemmli Buffer. Proteins were resolved by SDS-PAGE and transferred onto PVDF membranes for immunoblotting. A 1∶2500 dilution of guinea pig anti-Mage serum was used to detect Mage protein [36].

Genetic Interactions of ATM, ATR, NBS1 and RAD51 Loss-of-function with MAGE and Smc6

Double mutants of ATR and Smc6 used mei-41^D3 [48] and Smc6 alleles jnj^X1 and jnj^{Df(3R)Exel6198}. Knockdown of ATM, ATR or NBS1 function in MAGE or Smc6 homozygous mutant eye clones was achieved using the EGUF system, which uses the eyeless-Gal4 driver to express transgenes throughout eye development [32]. The EGUF system also ensures that all ommatidia of the adult eye are homozygous for either Smc6 or MAGE mutant alleles, because of an eye-specific GMR-hid transgene that eliminates non-mutant ommatidia. RNAi knockdown of MAGE alone or double RNAi of MAGE and Rad51 ortholog SpnA in the eye was achieved by crossing appropriate RNAi constructs containing males to UAS-Dcr2/CyO; ey-Gal4/TM3,Ser virgin females. For each genotype, five to nine specimens were photographed, and representative phenotypes are shown.

cDNA Clones, Cell Culture, Transfections, and Co-immunoprecipitation

Full-length cDNA clones for Nse1 (GM14348) and Nse4 (IP09347) were obtained from the Canadian Drosophila Microarray Centre, the MAGE (RE25453) clone was obtained from the Drosophila Genomics Resource Center (DGRC, Indiana University). Drosophila S2 cells (from the DGRC) were grown at 25°C in TNM-FH medium (SH30280.02, Thermo Scientific, Waltham, MA) supplemented with 10% fetal bovine serum. Expression constructs for transfection of S2 cells were created by inserting relevant full-length coding sequences into the Drosophila Gateway destination vectors (obtained from the DGRC). S2 cells were transfected with relevant expression constructs using dimethyldioctadecyl-ammonium [67]. Cells were harvested 24 h after transfection, washed once in phosphate buffered saline, pH 7.2, and re-suspended in the mild lysis buffer supplemented with a protease inhibitor cocktail (Roche Applied Science, Indianapolis, IN). The lysate was centrifuged for 10 min. at 20,000g at 4°C, and the supernatant transferred to a fresh tube. 200 µl of supernatant was mixed with 20 µl of protein G agarose beads (GE Healthcare Life Sciences, Piscataway, NJ) pre-bound with 5 µg of antibody in 800 µl mild lysis buffer. The agarose beads were then incubated for 1 h at 4°C with rocking, washed six times using mild lysis buffer and the bound proteins analyzed on immunoblots.

In vitro Pulldown Assays

pMBP-Mage was previously described [36] and the control pMBP construct was supplied with a Maltose binding protein (MBP) purification kit (New England Biolabs, Ipswich, MA). Expression constructs were produced by inserting relevant full-length coding sequences into a Gateway pDEST-14 expression vector. MBP fused Mage (MBP-Mage) was expressed in Escherichia coli (ER2523, New England Biolabs) and immobilized onto amylose resin (E8200S) according to the manufacturer’s directions. ³⁵S labeled probe proteins were expressed from Gateway pDest14 vectors using the TNT-coupled in vitro transcription-translation system (Promega, Madison, WI). For the in vitro binding assay, ³⁵S-labeled probe proteins were incubated with immobilized MBP-Mage proteins in 500 µl of buffer (20 mM Tris, 100 mM NaCl, 0.5 mM EDTA, 10% glycerol, and 1% Tween-20, pH 7.6) containing 0.25% bovine serum albumin (BSA) and protease inhibitor cocktail [68] overnight at 4°C with end-over-end mixing. The resin was washed six times in 500 µl of the same buffer, and the bound proteins were resolved by SDS-PAGE and detected by autoradiography.

Scanning Electron Microscopy (SEM) and Immunohistochemistry

Adult heads were prepared for SEM according to the HMDS method described in Drosophila Protocols [69] and imaged using a Scanning Electron Microscope (FEI (XL30), Philips, Hillsboro, OR). Dissection, fixation, BrdU labeling, and antibody staining of third larval instar eye-antennal discs were also carried out as described in Drosophila Protocols. Antibodies for immunohistochemistry included anti-cleaved caspase 3 (1∶1600 dilution, Cell Signaling Technologies, Beverly, MA), anti-BrdU (1∶200 dilution, Pharmingen San Jose, CA), and anti-phospho-histone H3 (Cell Signaling, 1∶1000 dilution). Secondary antibodies were used at a dilution of 1∶1000 (Alexa Fluor 488 and 586, Invitrogen). For the detection of apoptosis in third instar imaginal discs with an anti-cleaved caspase 3 antibody, embryos were collected at one hour intervals on grape juice plates and larvae were reared on yeast paste plates until the L3 molt. They were then transferred to 2 mM caffeine medium 32 h after the L3 molt and allowed to develop for a further 12 h before dissection. Images of the dissected discs were acquired using a LSM 700 confocal microscope (Carl Zeiss Inc., Thornwood, NY) and processed using Zen (Carl Zeiss). A maximum projection of all stacks of a confocal image was used to quantify the signal intensity of staining using a lower threshold to eliminate background staining. This value was divided by the area of each eye disc to obtain a ratio representing the relative amount of immunostaining. Data represent at least 7 eye discs per genotype per treatment.