p53 and cyclin G cooperate in mediating genome stability in somatic cells of Drosophila

One of the key players in genome surveillance is the tumour suppressor p53 mediating the adaptive response to a multitude of stress signals. Here we identify Cyclin G (CycG) as co-factor of p53-mediated genome stability. CycG has been shown before to be involved in double-strand break repair during meiosis. Moreover, it is also important for mediating DNA damage response in somatic tissue. Here we find it in protein complexes together with p53, and show that the two proteins interact physically in vitro and in vivo in response to ionizing irradiation. In contrast to mammals, Drosophila Cyclin G is no transcriptional target of p53. Genetic interaction data reveal that p53 activity during DNA damage response requires the presence of CycG. Morphological defects caused by overexpression of p53 are ameliorated in cycG null mutants. Moreover, using a p53 biosensor we show that p53 activity is impeded in cycG mutants. As both p53 and CycG are likewise required for DNA damage repair and longevity we propose that CycG plays a positive role in mediating p53 function in genome surveillance of Drosophila.

been shown to interact directly 14 . Moreover, in response to IR-stress the demethylase UTX acts as a specific epigenetic co-factor of p53 in the transcriptional upregulation of the DNA repair gene Ku80, but is not required for the p53-mediated expression of either hid or rpr 15 . This implies that Drosophila p53 might recruit different co-factors to fulfil its specific activities in response to various stressors. Here we describe the role of Cyclin G (CycG) as a co-factor of p53-mediated genome stability in Drosophila. We have shown earlier that CycG is involved in DSB sensing and repair during meiosis 16 , and now find that it is also important for combating genotoxic stress in somatic tissue. Although Drosophila CycG is not a transcriptional target of p53 like its mammalian counterpart cyclin G1, it physically interacts with p53 and is essential for p53 mediated DNA damage response. We provide evidence that p53 activity is hampered in the absence of cycG suggesting that CycG and p53 function together in the process of DNA damage repair.

Results
Loss of CycG compromises transposon-induced DSB repair. The fact that cycG mutants are impeded in meiotic DSB repair 16 prompted us to investigate its involvement in somatic DSB repair. To gain first insights we decided to employ the Drosophila P{w a }-element based system, which allows to uncover DSB repair-defects after P-element transposon excision 17 . This elegant assay is based on a X-linked P-element P{w a } that carries the apricot allele (w a ) of the white (w) gene. The w a allele is characterized by a copia insertion in an intron of w, decreasing the expression of the w gene. Thus flies with only one copy of w a are identified by yellow eye colour, whereas those with two copies have apricot-coloured eyes 17 . DSBs are induced by mobilizing the P{w a }-element with help of the Δ2-3 transposase in male flies and female progeny is judged by eye colour (Supplementary Fig. S1). Depending on the repair mechanism, eye colour may be apricot (exact repair by homologous recombination), yellow (absent or defective repair like non-homologous end joining), or red (reflecting loss of the copia-element) 17 .
To employ this assay we made use of the cycG CreD null allele being deficient for white 18 , combined it with the P{w a }-element and transposase and analysed the female offspring. Heterozygous cycG mutants were similar to control flies: about 93% of the progeny had apricot-coloured eyes, whereas about 3% failed to repair DSBs properly (yellow-coloured eyes) ( Fig. 1 and Supplementary Fig. S1). In contrast, only 83% of the homozygous cycG mutant female progeny had apricot-coloured eyes, whereas the percentage with either red-or yellow-coloured eyes was significantly increased with more than twofold of the controls or the heterozygotes ( Fig. 1 and Supplementary Fig. S1). This indicates that in the absence of CycG, somatic repair of double-strand breaks in the DNA is compromised. cycG mutants are hypersensitive to genotoxic stress. In order to narrow down the role of CycG in sustaining genome stability, we next analysed cycG mutants' sensitivity towards ionizing irradiation (IR) or the DNA damaging agent methyl methanesulfonate (MMS). Both genotoxic stressors directly or indirectly generate Figure 1. P-element based DSB repair assay with cycG CreD mutants. The P{w a }-element was introduced in a Oregon-R control, a cycG CreD heterozygous or homozygous background, mobilized by transposase and backcrossed with P{w a } (for details see Supplementary Fig. S1, and 17 ). Eye colours were scored in the F 2 -female offspring; the fraction with apricot eye colour, red and yellow eye colour, respectively, was determined. The total number of analysed females (n) was 6113 for the control, 6020 for cycG CreD /+ heterozygotes and 6022 for cycG CreD homozygotes. Error bars show standard error. Frequency of apricot, red and yellow eye coloured offspring was not significantly different between the control and the cycG CreD heterozygotes (n.s., p-values 0.34, 0.22 and 0.64 determined by Students T-test, respectively), whereas the respective fractions of the cycG CreD homozygotes varied significantly from control (**p-values 0.0005, 0.0022 and 0.010, respectively).
SCIENtIFIC REPORtS | (2017) 7:17890 | DOI:10.1038/s41598-017-17973-z DNA single-stranded or double-stranded breaks thereby enforcing a DNA damage response (overview 1,2 ). To this end homozygous cycG HR7 mutant larvae were exposed to 16 Gy IR or to food containing 2 mM MMS, and compared with likewise treated wild type Oregon-R or okra (okra AA /okra RU ) mutant animals. okra mutants served as positive control as they are sensitive towards a variety of genotoxic stressors 19,20 . The survival index, i.e. percentage of flies emerging from treated versus untreated larvae was determined for each genotype and related to the control. We found that cycG HR7 mutants were sensitive to IR with about 60% survival rate of the control, but not as sensitive as the okra mutants with no survivors (Fig. 2a). MMS exposure uncovered an even higher sensitivity of the cycG HR7 mutants with only about 30% survival rate relative to wild type control, but again lesser compared to okra mutants (Fig. 2b). These data show that CycG is important for a DNA damage response not only Figure 2. cycG mutants are sensitive to genotoxic stress. (a,b) Surviving genotoxic stress. Wild type control (Oregon-R), cycG HR7 and okr AA /okr RU mutant larvae were exposed to (a) IR-stress (40 Gy) or to (b) MMS (final ~2 mM). The survival index was determined as fraction of treated vs. untreated flies emerging from larvae and is given as % of the wild type control. The experiments were done in duplicate or triplicate (n = total number of animals for each genotype analysed in the assay); standard deviation is given and significance determined by Student's T-test (***p < 0.001). (c-f) Examples of metaphases in larval neuroblasts un/exposed to IRstress (12.5 Gy). (c) Wild type metaphase. Chromosomes are labelled. Scale bar: 10 μm in all panels. (d-f) Examples of aberrant metaphases. Arrowheads point to examples (see also insets): chromosome breaks (d), single telomere fusion (e) or multifusions (f). (g) Frequency of normal (dark grey bars) and aberrant (light grey bars) metaphases in wild type and cycG HR7 /cycG HR7 mutant neuroblasts prepared from un-irradiated (unirr) or irradiated animals. Time point of preparation after irradiation is given underneath the bars in hours, sample size (n) within the bars. Statistical significance was determined with Student's T-test (ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001). Compared to control, cycG HR7 mutant neuroblasts show a higher incidence of chromosomal aberrations in response to irradiation (12.5 Gy), and take considerably longer to recover from IR-stress. with UAS-p53 overexpression driven by Gmr-Gal4 were hybridized with the probe indicated, i.e. either p53 to confirm p53 expression (arrow), reaper (rpr) as a known target gene which is indeed turned on (arrow), or cycG which does not respond to p53 overexpression (arrow). Quality of the cycG probe was monitored on Gmr::cycG eye imaginal discs and on control discs. Size bar, 50 μm in all panels. (b) RT-PCR analyses for p53, cycG and beta-tubulin (Tub56D) transcripts from homozygous adults of the respective genotype with or without irradiation (40 Gy). Reactions were performed with (+) and for control without (−) reverse transcriptase. cycG mRNA is present in p53 mutants and vice versa, independent of irradiation. (b') Expression levels of cycG in p53 5A-1-4 homozygotes (left panel) and of p53 in cycG eoC homozygotes (right panel) were quantified by qRT-PCR; no significant differences (n.s.) were detected compared to expression levels of control flies (y 1 w 67c23 ) (p-value 0.461 and 0.472, respectively). As reference genes to cycG (0.98), βTub56D (0.98) and eRF1 (0.98) were used; and cyp33 (0.95) and Tbp (0.93) as reference for p53 (0.93). Target efficiency (given above in parentheses) was taken into account for the relative quantification 51 . (c,c') Yeast two-hybrid interaction assays reveal physical interaction between Cyclin G and p53 protein, as visualized by blue yeast colonies activating SCIENtIFIC REPORtS | (2017) 7:17890 | DOI:10.1038/s41598-017-17973-z in meiotic but also in somatic tissue. In meiotic tissue, CycG is associated with the 9-1-1 complex 16 , comprising Rad9, Rad1 and Hus1 proteins, which plays a key role in DNA integrity surveillance and damage response 21,22 . We were, however, unable to co-precipitate CycG protein together with Rad9 and Rad1 or BRCA2 in irradiated flies ( Supplementary Fig. S2), unlike in female germ cells 16 . This suggests that in somatic cells CycG is not involved in DSB sensing but rather conveys chromosomal stability by protecting chromosomes and/or by assisting in the rapid repair of damaged DNA.
Increased incidence of mitotic abnormalities after irradiation in cycG mutant larval brain cells. To further address the role of the cycG gene in chromosome stability, we determined the sensitivity of cycG HR7 mutants to the induction of chromosomal breaks by irradiation. IR-induced chromosome breaks can be detected in mitotic larval neuroblasts as chromosomal aberrations like fragmentation or telomere fusions 23 . To this end, we examined DAPI-stained preparations of larval brains of either wild type or cycG HR7 mutants irradiated with 12.5 Gy, and determined the frequency of chromosomal aberrations in a time course of 4, 8 and 24 hours after irradiation. In un-irradiated larvae of either homozygous cycG HR7 mutants or wild type control, less than 3% of metaphase chromosomes displayed aberrations (Fig. 2c,g). In contrast, 4 hours after IR-treatment, about 30% of cycG HR7 mutant and 25% of wild type metaphases disclosed damaged chromosomes on average ( Fig. 2d-g). This number declined over time, however, to a lesser degree and therefore much more slowly in cycG HR7 mutant neuroblasts (Fig. 2g). While approximately 14% of the analysed metaphases have defects at 24 hours after irradiation in the wild type, the number is nearly doubled in the cycG HR7 mutants where about 26% of the analysed metaphases display aberrations, implying a retardation and/or erroneous DNA repair in the absence of CycG.

CycG is not a transcriptional target of p53 but physically interacts with p53 protein. Presumably,
CycG relies on additional partner(s) to pursue its role in safeguarding DNA in somatic tissue. One of the most central players in the control of events following DNA damage is p53. The single Drosophila p53 gene, for example, was shown to mediate the response to a multitude of stressors by triggering the transcriptional activity of different target genes 7 . So far we know that Drosophila CycG is involved in DNA damage recovery after IR-stress, similar to its vertebrate counterpart cyclin G1 (Ccng1) 24 . Interestingly Ccng1, one of the two mammalian cyclin G homologues, is a transcriptional target of p53 25 . Here, Ccng1 together with Mdm2 is involved in p53 negative regulation 10,11 . To elucidate if a comparable scenario also exists in Drosophila we addressed the p53-dependence of cycG transcription. To this end we overexpressed p53 specifically in larval eye imaginal discs (Gmr-Gal4; UAS-p53), sufficient to induce expression of the p53 target gene reaper (rpr) 13 (Fig. 3a). Accumulation of cycG transcripts, however, was not observed upon p53 overexpression (Fig. 3a). We wondered whether a response of cycG expression to p53 may depend on genotoxic stress, which we analysed by RT-PCR. Expression of cycG and p53 was analysed in wild type and null mutant p53 5A-1-4 animals, unchallenged or irradiated with 40 Gy. No apparent difference was observed in cycG expression in p53 5A-1-4 null mutants versus control, indicating that the transcriptional activation of cycG in Drosophila is independent of p53 unlike in mammals ( Fig. 3b,b'). Moreover, no changes were observed upon IR-stress (Fig. 3b). Also, p53 transcription was unaffected in cycG eoC mutants ( Fig. 3b,b').
Though no transcriptional interdependence exists between p53 and cycG in Drosophila, interactions at the protein level remained to be tested. Interaction between Drosophila p53 and CycG proteins was assayed in a yeast two-hybrid assay revealing a robust binding between the two proteins ( Fig. 3c,c'). We were able to narrow down the p53-interaction domain in CycG to the Cyclin domains, and to the C-terminal part of p53 containing the oligomerization domain (Fig. 3c,c'). To test for in vivo interactions, we performed immunoprecipitation with an epitope-tagged version of p53 (UAS-p53_3xHA) overexpressed during eye development using Gmr-Gal4. Indeed, we were able to co-precipitate p53-HA tagged protein together with concurrently overexpressed CycG (Fig. 3d), however only upon IR-stress and not in unstressed animals. Together, these data clearly reveal that CycG and p53 proteins physically interact, suggesting that CycG may directly modulate the activity of p53 during DSB repair.
Genetic interactions between cycG and p53. According to our hypothesis CycG may be an auxiliary factor of p53 in DNA damage response. In this case we would expect that p53 activity is modulated by cycG gene dose which should be uncovered by genetic interactions. Previous studies have shown that ectopic p53 expression affects Drosophila eye development: smaller adult eyes with a disturbed ommatidial patterning result from growth and differentiation defects 26 . This phenotype develops when p53 is constantly induced in cells anterior to the morphogenetic furrow, i.e. within dividing cells. If p53 activity depends on functional CycG in this developmental context, the effects may be modified in a cycG mutant background. To test this assumption, the eye phenotype of flies overexpressing p53 (ey::p53) was investigated in the presence or absence of CycG, using two the lacZ reporter. Above a scheme of Cyclin G (c) and p53 (c') depicts relevant protein domains. Conserved Cyclin domains (AS 287-500, magenta) are relevant for interaction with p53 protein. The p53 protein contains a transactivating domain (TA green, AS 1-84), a DNA-binding domain (DNA BD blue, AS 85-252), and an oligomerization domain (Oligo grey, AS 253-385) which is relevant for CycG binding. Empty vectors served as controls. (d) CycG and p53 proteins can be co-precipitated after IR in vivo. HA-tagged p53 proteins were immunoprecipitated (IP) from head extracts (Gmr-Gal4 UAS-CycG::UAS-p53_3xHA) 2 hrs after IR using anti-HA antibodies (upper box, arrow). The input lane contained 10% of the protein extract used for the IP. No antiserum was used as mock control. CycG protein can be co-precipitated when flies were irradiated (middle box, arrows), but not without irradiation (lower box). The asterisk labels unspecific IgG signals. Size is given in kDa. Uncropped blots are shown in Supplementary Figure S3.
SCIENtIFIC REPORtS | (2017) 7:17890 | DOI:10.1038/s41598-017-17973-z different cycG null alleles cycG HR7 and cycG eoC 16,27 . The adult progeny was subdivided into four different categories representing the severity of eye size reduction (Fig. 4a). As severity frequently differs between left and right eye, the eyes were scored separately. As a consequence of ectopic p53 expression in the developing eye disc, nearly half of the adult eyes were severely reduced in size ('pinhead size') or were considerably smaller, representing categories 1 or 2 (Fig. 4a,b). In total, about 90% of all ey::p53 eyes were smaller than wild type (Fig. 4b,c). The p53 mediated growth defects were clearly ameliorated in the absence of CycG: in the cycG mutant background more than half of the eyes had a wild type size, and the most severe categories were rarely observed, independent of the cycG allele used (Fig. 4b). As we have no indication that transcriptional regulation of ey is under the control of CycG ( Supplementary Fig. S4), we conclude that CycG supports p53 activity at the protein level. We expected that a combined overexpression of p53 with CycG may enhance the p53 eye phenotype, which was however not observed (Fig. 4c), suggesting that CycG requirement for p53 activity is not strictly dose sensitive. Rather CycG appears to be a limiting factor for p53 activity.
Adult eye phenotypes reflect the whole developmental process, and a large part of tissue loss may be attributed to increased cell death in pupal stages although overexpression was restricted to the proliferating part of the tissue. To address the specificity of CycG activity regarding p53 mediated growth defects, we analysed the effects at the time of occurrence, i.e. directly in the eye imaginal discs. Moreover, we monitored not only the apparent tissue loss but also influences on cell proliferation. In the developing eye disc, proliferating cells can be visualized by EdU-and Phospho-Histone H3 (PH3) labelling, detecting either cells in the S-phase or mitosis of the cell cycle. Cells anterior to the morphogenetic furrow are vividly proliferating, whereas cells in the posterior part start their differentiation program towards photoreceptor fate and are assembled into ommatidia, visualized by the  Overexpression of UAS-p53 with ey-Gal4 results in a strongly reduced eye field as visualized by Elav which is decreased or even absent. A weak phenotype is characterized by remains of proliferating eye tissue (closed arrows). Tissue loss is ameliorated in the background of cycG mutants, either cycG HR7 or cycG eoC , and the discs develop an almost normal eye field. (b) The size of the eye field was categorized based on Elav expression as similar to wild type (wt), as clearly smaller than wild type (less) or as absent (no), and discs of the given genotypes were grouped into the respective categories. The number of analysed discs is given above the columns. Eye field size is significantly different between control and all other genotypes, as well SCIENtIFIC REPORtS | (2017) 7:17890 | DOI:10.1038/s41598-017-17973-z Elav marker. About 50% of the analysed ey::p53 eye-antennal discs lacked the posterior part and showed no Elav expression, whereas in the other half the Elav positive part was much smaller (Fig. 5a,b). EdU and PH3 staining, however, appeared similar to wild type in the remaining tissue. In contrast, most of the eye discs from ey::p53 cycG mutant animals resembled wild type, using either cycG HR7 or cycG eoC allele. Almost 90% had an Elav positive compartment, i.e. differentiating ommatidia, which was wild type in size in about a third of the discs (Fig. 5a,b). Again, no influence on either EdU or PH3 staining was apparent suggesting that the loss of CycG has no primary effect on cell division. Taken together, our data imply that CycG is a positive mediator of p53 activity also in this developmental context. DNA repair processes and initiation of apoptosis are delayed in wing discs of cycG mutants similar to p53 mutants. DNA damage repair is a highly conserved cellular response involving activation of p53. In Drosophila, p53 not only instructs apoptotic elimination of irreversibly damaged cells but is also required for DNA repair and the maintenance of regenerative potential 13,28,29 . One of the first events that allows to visualize DNA damage is phosphorylation of the histone variant H2Av (γ-H2Av) in Drosophila, which helps to recruit further DNA repair proteins to the damaged site 3,5 . γ-H2Av can be traced in vivo with respective antibodies, and is detected at discrete foci of DSB sites which disappear when DNA repair processes are completed. Third instar larvae mutant for either cycG allele cycG HR7 or cycG eoC , or mutant for the p53 null allele p53 5A-1-4 , were irradiated with 40 Gy, and γ-H2Av accumulation was monitored in wing imaginal discs relative to control (Fig. 6a-c). No differences were seen between mutants and control one hour after irradiation, demonstrating that sensing and initiation of DNA damage repair are unaffected by a loss of CycG (Fig. 6b). In contrast, γ-H2Av foci still persisted 25 hours after IR-stress in the cycG mutants, at a time when wild type cells had completed the repair process (Fig. 6a,c). A similar result was seen also for the p53 mutant, as described earlier 29 (Fig. 6a-c). Moreover, cell death induction was comparably hampered in cycG and p53 mutant discs: six hours after irradiation a robust induction of apoptosis was observed in the wild type, whereas only rarely detectable in p53 and cycG mutant discs, respectively (Fig. 6d,e). These results provide evidence for an impaired repair process in cycG and p53 mutants alike. We propose that CycG is required to resolve IR-induced DNA damage presumably as co-factor of p53.
DNA damage notably by oxidative stress is a major inducer of aging with p53 being a key coordinator 30 . Accordingly, p53 mutant flies have a shortened life span compared to the wild type 9 . We compared the life span of the homozygous p53 5A-1-4 flies with that of cycG HR7 and cycG eoC , wild type Oregon-R and methuselah mth 1 as example of a long lived stock 31 . As is expected for a common role of p53 and CycG in the context of DNA damage repair, the life span of respective mutants was likewise reduced (Supplementary Fig. S5).

Activation of the p53 biosensor by IR-stress is reduced in cycG mutant germaria. Based on our
genetic data we favour the idea that in Drosophila CycG assists p53 function during DNA damage response. To follow p53 activity directly in vivo we made use of a p53R-GFPnls biosensor, where p53 activation triggers nuclear GFP expression 32 (Fig. 7a). This sensor is activated in response to genotoxic stress in the female germline exclusively in germline stem cells (GSC) and their immediate progeny, the cystoblasts (CB) 33 . To determine, whether CycG is required for p53 activity in response to IR-stress, we introduced the p53R-GFPnls biosensor into a cycG HR7 mutant background and measured the GFP reporter activity 24 hours after irradiation (Fig. 7b,c). As described earlier 33 , GFP nuclear localization was detected in region 2a/b of control germaria, where cross-over events are repaired, whereas in response to IR-stress nuclear GFP was detected in gonadal stem cells in more than 80% of germaria (Fig. 7b,c). In the cycG HR7 mutant background, however, less than 20% of the irradiated germaria showed the expected nuclear GFP accumulation (Fig. 7b,c). These findings strongly corroborate our genetic data and support our hypothesis that CycG is crucial for p53 mediated response to genotoxic stress.

IR induced p53 biosensor activity is decreased in salivary glands of cycG mutants. The
p53R-GFPnls reporter faithfully reports p53 activation after genotoxic stress also in Drosophila embryos 32 . To expand our in vivo analyses to somatic tissue, we hence exposed p53R-GFPnls third instar larvae to irradiation (40 Gy), and monitored GFP expression in polytene nuclei of salivary glands 2 hours after irradiation. IR-stress caused a strong nuclear GFP signal, reflecting IR-induced activation of p53 in our test system (Fig. 8a). In the background of cycG mutant alleles cycG HR7 or cycG eoC , the intensity of nuclear GFP dropped dramatically to levels of about 30% of the control value (Fig. 8a,b). We conclude that not only in the germline but also in the soma CycG is a pivotal component for either the activation of p53 or for p53 transcriptional activity in response to IR-stress.

Discussion
Earlier we have shown that Drosophila CycG is important for the meiotic recombination checkpoint in the female germline. In cycG mutant germaria, DSB repair is delayed, and CycG protein is found in conjunction with the 9-1-1 complex suggesting that it may be involved in DSB sensing 16 . We have now extended our analysis to somatic tissue, where again we note problems in DNA damage repair as detected by persistent γ-H2Av signals in irradiated cycG mutants. This indicates that in the absence of CycG, repair of double-strand breaks in the DNA is compromised. Accordingly, cycG mutants fail to repair DSBs with the fidelity of wild type, display more chromosomal aberrations upon irradiation, and are hypersensitive to genotoxic stress. We found no evidence for an as between ey::p53 and ey::p53 in any cycG mutant background as determined by ANOVA two-tailed Tukey-Kramer approach (***p < 0.001). Blue: Elav staining similar to wild type (wt); orange: reduced Elav field (less); magenta: no Elav staining (no). Genotypes are: UAS-lacZ/+; ey-Gal4/+ control, UAS-p53/+; ey-Gal4/+, UAS-p53/+; ey-Gal4 cycG HR7 /cycG HR7 , UAS-p53/+; ey-Gal4 cycG eoC /cycG eoC . Figure 6. DNA damage response is similarly altered in p53 and cycG mutant discs. Larvae of the given genotype were irradiated (40 Gy) and analysed at the given time (aIR, after irradiation). (a) γ-H2Av signals mark foci of damage response; signals are hardly detected in wild type wing discs that have mostly completed repair at 25 hrs after IR. In contrast, γ-H2Av signals are still elevated in cycG HR7 or cycG eoC mutant wing discs. Persistent signals are similarly observed in p53 mutant wing discs (p53 5A-1-4 ). (b,c) Quantification of γ-H2Av signals was performed by measuring integrated density in 9 to 11 wing discs of each genotype either 1 h (b) or 25 hrs (c) after irradiation using ImageJ. Standard deviation is shown; ***p < 0.001 according to Student's T-test. (d) Anti Caspase-3 act staining of wing discs 6 hrs after irradiation revealed a robust induction of apoptosis in the wild type, whereas cell death was barely detectable in cycG or p53 mutant discs. (e) Quantification of Caspase-3 act signals by measuring integrated density in 13-15 wing discs of the respective genotype using ImageJ. Standard deviation is shown; ***p < 0.001 according to Student's T-test. Size bar represents 100 μm in all panels A and D.
SCIENtIFIC REPORtS | (2017) 7:17890 | DOI:10.1038/s41598-017-17973-z involvement of CycG in DSB sensing in somatic cells, however. Instead, CycG appears to perform its role through modulating the activity of p53. Since we observe a retardation and/or erroneous DNA repair in the absence of CycG, we propose that CycG is required to resolve IR-induced DNA damage presumably as co-factor of p53. From our genetic data we conclude that CycG is a positive mediator of p53 activity, and indeed mutants in either gene resemble each other not only in life span but also in radiation sensitivity. The physical interaction of CycG and p53, however, strongly suggests that CycG directly promotes p53 activity, regardless of whether it may also regulate downstream or upstream components of the DNA damage repair machinery.
Unlike in vertebrates, Drosophila cycG is not under the transcriptional control of p53. Instead we see a robust protein-protein interaction in a yeast two-hybrid assay between p53 and CycG proteins, involving the cyclin repeats of CycG and the tetramerization domain of p53. Direct binding in vivo, however, required genotoxic stress. We propose that complex formation, rather than being permanent, occurs only in response to DNA damage and perhaps requires additional factors and/or protein modification/s. With the help of a p53 biosensor we showed that CycG is crucial for p53 mediated transcriptional response to genotoxic stress in the germline as well as in somatic tissue, suggesting that CycG may be involved in the activation or stabilization of p53 itself, or in the assembly of active transcriptional complexes.
The CycG-p53 axis might have been expected given their close interrelationship in the mammalian system. Here, the two cyclin G homologues Ccng1 and Ccng2 have been involved in growth control to genotoxic stress 24,34,35 . Ccng1 but not Ccng2 is a direct transcriptional target of p53 25 . Both are found in complexes with protein phosphatase 2A, and together with Mdm2 Ccng1 is involved in Mdm2 mediated degradation of p53 11,36,37 . As the two mammalian cyclin G proteins appear to act differently on cell proliferation, a lot of work has been invested to understand their respective roles. More recently it was proposed that observed discrepancies may arise from dose dependency of Ccng1 34 . In fact, also Drosophila tissues and cells appear to respond differentially to the dose of CycG, as for example overexpression may impact the cell cycle in a dominant negative manner, and RNAi downregulation causes effects different from the gene deletion phenotypes 16,18,27,38,39 . The role of Ccng1 in response to genotoxic stress has been analysed in quite some detail. Here, Ccng1 not only forms a complex with Mdm2, resulting in destabilizing p53. Moreover, it also interacts with ARF, thereby stabilizing and activating p53 40 . It hence has been proposed that Ccng1 is required for a timely and proper response to genotoxic stress, first for the activation of p53 to allow for DNA damage repair, and then for p53 degradation to protect cells from apoptosis that have recovered from the initiating stress 34,40 .
Intensive searches in the Drosophila genome failed to uncover Mdm2 or ARF homologues to date. Recently, however, a Mdm2 analogue called Corp has been identified that shares several Mdm2 properties 14 : Corp is a transcriptional target of p53 in response to genotoxic stress, it binds to p53 protein and results in reduced p53 protein levels presumably by proteolytic degradation. Hence, like Mdm2 Corp acts in a negative feed back loop on p53 activity 14 . Whether Corp is likewise inactivated by phosphorylation and/or an ARF-like molecule remains to be shown. Moreover, it will be interesting to see, whether Corp can recruit CycG, and whether PP2A plays any role in its regulation. We know already that Drosophila CycG also binds to the PP2A-B' subunit, similar to the two mammalian CycG proteins 27,36,37,41 . Unlike in vertebrates, however, it acts negatively on PP2A activity by genetic means 27,41 . Despite the similarity of the respective components and the processes they are involved in, there is not a 1:1 conformity when comparing flies and mammals. Perhaps, the manifold feed back loops weaved into the system, obstruct our view and elude genetic analyses. Perhaps, like in mammals 34,40 , Drosophila CycG forms protein complexes with disparate activities depending on tissue, cell cycle phase, or phase of response to DNA damage -repair or apoptosis.

Materials and Methods
Fly stocks and genetic work. Three null alleles of cycG were used, cycG HR716 , cycG CreD18 and cycG eoC27 . cycG HR7 was generated by ends in homologous recombination and carries the white + marker in between two defective cycG copies. From this allele, cycG CreD was derived lacking the white + marker. cycG eoC was generated by ends out homologous recombination. All three alleles are null mutants as judged by molecular analyses and lack of protein expression. They all produce homozygous, sterile offspring at variably low frequency, with cycG eoC producing the lowest numbers 16,18,27 . In this work most analyses were performed with the allele cycG HR7 . Flies Figure 8. Absence of CycG affects p53 biosensor response to IR-stress in salivary glands. p53R-GFPnls reporter activity was analysed in larval salivary glands from either control, cycG HR7 or cycG eoC mutants irradiated with 40 Gy. (a) IR-stress results in p53 activation and subsequent expression of nuclear GFP from the p53 biosensor in the control, however to a much lower degree in cycG mutant glands. Size bar represents 100 μm in all panels. (b) Nuclear intensity of GFP signals was recorded using ImageJ by measuring individual nuclei (30-40 per gland; 6-8 glands per genotype). In both cycG mutant alleles nuclear GFP staining is reduced to about one-third of the value measured in the wild type. Significance was tested by ANOVA two-tailed Tukey-Kramer approach (***p < 0.001; ns, not significant).
SCIENtIFIC REPORtS | (2017) 7:17890 | DOI:10.1038/s41598-017-17973-z were reared on standard fly food at 18 °C. Crosses were conducted at 25 °C. The following stocks were used: mth 1 (BL27896), p53 5A-1-4 (BL6815), okr RU /CyO (BL5098) 19 and UAS-lacZ 42 . They were obtained from the Drosophila stock center Bloomington. UAS-CycG is described 16 . UAS-Flag-Rad9 20 and UAS-GFP-Rad1 43 were obtained from U. Abdu, and okr AA /CyO 19 was a gift of T. Schüpbach. Ey-Gal4 44 , Gmr-Gal4 45 , UASp-p53 46 , and p53R-GFPnls 32 , were gifts from U. Walldorf, U. Abdu and J. Abrams. UAS-p53_3xHA (#F000091 FlyORF, Zürich, Switzerland) was obtained from the Zurich ORFeome Project. Oregon-R and y 1 w 67c23 , respectively served as wild type control. Flies were combined and recombined according to standard genetics and genotypes were verified by PCR. For fly selection, GFP or Tb 1 markers on balancer chromosomes were used. Adult eyes were documented as described earlier 27,47 . Lifespan analysis. Freshly hatched flies were collected under mild anaesthesia with CO 2 , and groups of 10 females and 10 males each per vial were collected and reared on standard fly food at 25 °C. Flies were transferred onto fresh food every other day. The number of dead flies was recorded at least every other day. A minimum of 200 flies were analys ed per genotype.
Genetic reporter assay for monitoring DNA repair efficiency. Crosses were done according to 17  Neuroblast chromosome squashes. Whole brains of late 3 rd instar larvae were dissected in 3 ml 0.7% sodium chloride and, after addition of 6 µl 10 µM colchicine (Serva; Heidelberg, Germany), incubated for 40 min at room temperature. They were next transferred for 9 min into a hypotonic solution of 0.5% sodium citrate and fixed for 5 min in 3% paraformaldehyde dissolved in 50% acetic acid. Brains were squashed on Polysine TM coated slides (Gerhard Menzel GmbH; Braunschweig, Germany) under cover slips coated with Gel Slick TM (AT Biochem; Malvern, Pennsylvania, USA). The slides were frozen in liquid nitrogen, and the cover slips were removed. After several washing steps in TBS (20 mM Tris-HCl pH 7.4, 150 mM NaCl) and TBST (TBS plus 0.5% Triton X-100) the preparations were mounted in Roti ® -Mount FluorCare with DAPI (Roth; Karlsruhe, Germany). Metaphases were analysed with an Axiophot 2 plus (Zeiss; Göttingen, Germany) coupled to a Power Shot A95 camera (Canon; Krefeld, Germany).

Protein expression analysis in tissue.
Immunostaining of imaginal discs, germaria and salivary glands was done according to standard protocols using the following antibodies: mouse anti-γ-H2Av Unc93-5. . For labelling cells undergoing DNA synthesis, the Click-iT EdU Alexa Fluor 488 Imaging Kit (Invitrogen; Eugene, Oregon, USA) was used, and the tissue was treated as described 48 . Anti γ-H2Av staining on larval wing discs was performed 1 h and 25 hrs after irradiation of wandering third instar larvae at 40 Gy. Frequency of remaining DSBs was determined by measuring the integrated density of signals in the wing discs with ImageJ. Secondary antibodies with minimal cross-reactivity coupled to DTAF, Cy3 or Cy5 generated in goat were purchased from Jackson Immuno-Research Laboratories (Dianova; Hamburg, Germany). Dissected tissue was embedded in Vectashield (Vector Laboratories; Burlingham, CA, USA) and documented with a Zeiss Axiophot microscope (Carl Zeiss AG; Oberkochen, Germany) coupled to a Bio-Rad MRC1024 scan head using LaserSharp 2000 imaging software. Pictures were assembled with Corel Photo Paint and Corel Draw software.