Sunitinib induces genomic instability of renal carcinoma cells through affecting the interaction of LC3-II and PARP-1

Deficiency of autophagy has been linked to increase in nuclear instability, but the role of autophagy in regulating the formation and elimination of micronuclei, a diagnostic marker for genomic instability, is limited in mammalian cells. Utilizing immunostaining and subcellular fractionation, we found that either LC3-II or the phosphorylated Ulk1 localized in nuclei, and immunoprecipitation results showed that both LC3 and Unc-51-like kinase 1 (Ulk1) interacted with γ-H2AX, a marker for the DNA double-strand breaks (DSB). Sunitinib, a multi-targeted receptor tyrosine kinase inhibitor, was found to enhance the autophagic flux concurring with increase in the frequency of micronuclei accrued upon inhibition of autophagy, and similar results were also obtained in the rasfonin-treated cells. Moreover, the punctate LC3 staining colocalized with micronuclei. Unexpectedly, deprivation of SQSTM1/p62 alone accumulated micronuclei, which was not further increased upon challenge with ST. Rad51 is a protein central to repairing DSB by homologous recombination and treatment with ST or rasfonin decreased its expression. In several cell lines, p62 appeared in the immunoprecipites of Rad51, whereas LC3, Ulk1 and p62 interacted with PARP-1, another protein involved in DNA repair and genomic stability. In addition, knockdown of either Rad51 or PARP-1 completely inhibited the ST-induced autophagic flux. Taken together, the data presented here demonstrated that both LC3-II and the phosphorylated Ulk1 localized in nuclei and interacted with the proteins essential for nuclear stability, thereby revealing a more intimate relationship between autophagy and genomic stability.

Micronuclei are small DNA-containing membrane-enveloped structures separated from the primary nuclei of the cell, containing chromosomal fragments and/or whole chromosomes based on different types of formation. 1,2 Accumulating evidence has linked the frequency of micronuclei to cell types, which reflected nuclear stability. 2,3 As micronuclei are usually destructed by cytoplasmic nucleases, they are believed to be removed through autophagy. 4,5 Autophagy, one of the two major pathways degrading the cellular components in eukaryotic cells, mainly controls the turnover of long-lived proteins and organelles. 6 Unc-51-like kinase 1 (Ulk1) and LC3 are two central components in autophagy, whereas Ulk1 participates in the induction process, 7 the phosphatidylethanolamine-conjugated form of LC3 (Atg8-PE/LC3-II) is the only reliable marker protein related to completed autophagosomes. 8 Accumulation of DNA damage has been found in autophagy-deficient mammary tumor cells, 9 and either nuclear components or nuclear lamina were degraded by the autophagic process. 10,11 In senescent cells, autophagy participates in proteolytic processing of histones, the basic structural proteins of eukaryotic chromosomes. 12 These reports indicate that autophagy has an important role in the DNA damage repairing process.
Non-homologous end joining (NHEJ) and homologous recombination (HR) are the two principal mechanisms to repair DNA double-strand breaks (DSB) in mammalian cells. 13,14 And γ-H2AX (phosphorylated histone H2AX Ser139) is commonly used as a marker for DSB. 15 Central to DSB repair by HR is Rad51, which is detected in multiple discrete subnuclear structures (foci) and promotes strand invasion and homologous pairing between the two DNA duplexes. 16,17 Many signaling molecules, such as Akt and extracellular regulated protein kinases 1/2 (ERK1/2), affected the expression of Rad51, a demonstrated negative regulator of autophagic process. [18][19][20] In addition, inhibition of autophagy enhanced radiosensitivity of nasopharyngeal carcinoma via reducing the expression of Rad51. 21 Recently, autophagy has been demonstrated to regulate chromatin ubiquitination in DNA damage response (DDR) through elimination of p62. 17,22 Besides Rad51, the DNA-binding enzyme poly(ADP-ribose) polymerase 1 (PARP-1) is also involved in modulating the activity of the DNA repair systems, 23 has a primary role in the process of poly(ADP-ribosyl)ation, and is responsible for the major poly(ADP-ribosyl)ation activity observed during DDR. PARP-1 has been demonstrated to link to the ataxiatelangiectasia mutated protein (ATM), a key signal transducer having a critical role in DDR. 24 Its over activation induces mitochondrial transition and damage, leading to cell death, 23 and its cleavage facilitates cellular disassembly and serves as a marker for cells undergoing the caspase-dependent apoptosis.
Sunitinib (ST), approved by FDA to treat renal cell carcinoma (RCC), 25 inhibits the activity of PDGFRs, c-KIT, FLT-3 and the VEGFRs. 26 It not only induces cell viability loss and cell senescence, but also causes G1-S cell cycle arrest and DDR in OS-RC-2 cells. 27 In addition, it was reported to stimulate incomplete autophagic flux in renal and bladder cancer cells, 28,29 and to induce autophagy in cardiac cells and PC12 cells. 30,31 Rasfonin is a natural product isolated from the fermentation cultures of the fungus Talaromyces sp. 3656-A1, and was named after its biological activity against the small G-protein Ras. 32 In our previous study, it was found to induce both apoptosis and autophagy. 33 In this study, we have demonstrated that both ST and rasfonin increased the level of γ-H2AX, reduced the expression of Rad51, and stimulated the formation of micronuclei. Moreover, we found that LC3-II and pUlk1 localized in nuclei, and colocalized with γ-H2AX in micronuclei. Knockdown of LC3 or Ulk1 increased the frequency of micronuclei, which was further accumulated upon challenge with ST. However, ST failed to further increase the formation of micronuclei in the p62-depleted cells. Immunoprecipitation results showed that LC3 interacted with γ-H2AX, Rad51 and PARP-1. Although both Rad51 and PARP-1 can bind to p62 and deprivation of either one completely inhibited the STinduced autophagy, they affect the micronuclei formation in different manners. Images from three independent experiments were analyzed by Image J, and mean ± S.D. were shown in histograms along with the blots. (c) Electron microscopy was performed for 786-O cells following treatment with ST for 3 and 6 h. The data of the area ratio were non-normally distributed, and presented as the mean of at least 15 cells counted for each group, the data were analyzed by Friedman test (d). (e) Immunofluorescence using the antibody of LC3 was performed for 786-O cells following treatment with ST in the presence or absence of CQ for up to 6 h (arrowheads indicated the LC3 dots in nuclei, and arrows indicated micronuclei). The numbers of the punctate LC3 in each cell were counted, and at least 50 cells were included for each group (f). Data representing the mean ± S.D. were shown in graph. *Po0.05 versus control; **Po0.01 versus control. All data were acquired from at least three independent experiments and its cytotoxicity was further confirmed by colony growth assay (Supplementary Figure 1B). Meanwhile, high doses of ST stimulated the cleavage of PARP-1 (Supplementary Figure 1C), 34 and flow cytometry analysis revealed its obvious induction of apoptotic cell death (Supplementary Figure 1D). ST has been reported to induce autophagy in a number of cell lines, 30,31 and we also found that it increased the level of LC3-II and stimulated the autophagic flux, as chloroquine (CQ) further accumulated LC3-II in the STtreated 786-O cells (Figures 1a and b). Similar results were also obtained in ACHN cells (Supplementary Figure 2A). Nuclear localization of LC3-II and the phosphorylated Ulk1. Deacetylation of nuclear LC3 has been reported to drive autophagy, 35 and LC3A-II, not LC3B-II (thereafter called LC3-II), could localize in nuclei. 36 Although LC3-II is usually considered as a cytoplasmic protein, it may localize in nuclei as suggested in recent study. 11 To confirm its nuclear localization, subcellular fractionation was performed using PARP-1 as the marker of nuclei, 37 and LC3-II was found in the nuclear fraction and its level was further increased in the presence of ST (Figure 2a). In addition, either Lamin B1 or histone H3, two often used nuclear markers, 11,38 was also detected ( Figure 2a). In contrast, glyceraldehyde phosphate dehydrogenase (GAPDH) did not appear in nuclear fraction ( Figure 2a). Consistent with the results that Ulk1 could exist in nuclei and interact with PARP-1, 39 we also observed nuclear localization of the phosphorylated Ulk1 Ser555 (pUlk1) in HEK293T cells (Figure 2b). Although the band for pUlk1 of normal molecular weight (NMW) was not found in the nuclear fraction of 786-O cells, a band for that of relatively lower MW (LMW) was observed in nuclei (Figure 2c), suggesting the cleavage of Ulk1 under certain circumstances. Actually, the LMW form of pUlk1 was also observed in HeLa and K562 cells ( Supplementary Figures 3A and B). Moreover, the bands for both NMW and LMW Ulk1 decreased in the Ulk1depleted HEK293T and HeLa cells (Supplementary Figure 3C), suggesting that the LMW one is specific for Ulk1. Although ST increased the nuclear-localized pUlk1 of NMW, rasfonin decreased its nuclear localization in HEK293T cells (Supplementary Figure 3D). However, the nuclear-localized pUlk1 of LMW appeared to be increased in both types of the treated cells (Supplementary Figure 3D). An online software, 'EMBOSS: sigcleave' (http://emboss.bioinformatics.nl/cgi-bin/emboss/sigcleave), was used to predict the cleavage sites of the proteins, and two candidate cutting sequences were predicted in Ulk1 (Supplementary Figure 3E). Considering their MWs, the sequence between serine-381 and alanine-393 could be the site of cleavage. Interestingly, LC3-II and pUlk1 were also found in the insoluble nuclear participates (Nup; Supplementary Figure 3D), which was supposed to be chromatin. 40 PARP-1 is a DNA-binding enzyme and an often used nuclear marker. 23,37 To further confirm the nuclear localization of LC3-II, immunoprecipitation was performed using the antibody of either LC3 or PARP-1, and LC3-II was found in the immunoprecipitates of PARP-1 (Figure 3a), whereas PARP-1 appeared in the immunoprecipitates of both LC3 and pUlk1 (Figures 3a and b). LC3 was found to interact with PARP-1 in both the nuclear and cytoplasmic lysates of HEK293T cells, and the interaction in nucleus was much stronger than that in cytoplasm, although much more LC3-II was detected in the cytoplasm (Figure 3c). In the immunoprecipitates of pUlk1, relatively larger amount of PARP-1 was found in the Nu fraction than the cytoplasm one extracted from HEK293T cells (Figure 3d). Although LC3 binds to less PARP-1 in cytoplasm when cells were cultured in fresh medium (N) compared with the old one (O), their interaction was enhanced in nuclei under the condition (Figure 3c). Similar to the interaction between LC3 and PARP-1, the binding of pUlk1 to PARP-1 in nuclei was increased in fresh medium ( Figure 3d).

ST inhibits cell viability and induces autophagy in
LC3 interacts with γ-H2AX and Rad51. DNA damage elements can elevate the level of γ-H2AX. 15 Both ST and rasfonin remarkably increased the protein level of γ-H2AX, but failed to augment the level of Rad51 (Figures 4a and b). 41 As the formation of Rad51 foci, not the protein level of Rad51, was increased upon DNA damage, 42 we could not completely exclude the possibility that either ST or rasfonin induced the HR. Both ST and rasfonin increased the localization of LC3 Deprivation of LC3 and Ulk1 increases the frequency of micronuclei. As the nuclear buds and nucleoplasmic bridges can also be used as the biomarkers of genotoxic events in addition to micronuclei, 2 all three types of genotoxic biomarkers were counted and labeled as micronuclei for the convenience of presentation (Supplementary Figure 4A). Either ST or rasfonin were found to significantly increase the formation of micronuclei in a dose-dependent manner (10% and 30-60% in untreated and treated groups, respectively;  Rad51 interacts with p62 and its deprivation inhibits the ST-induced autophagy. Consistently, LC3 was pulled down in 786-O cells using the antibody of Rad51 (Figure 8a). Although p62 was found to interact with Rad51, it was not detected in the negative IgG control lane, and actin was not found either in the immunoprecipitates of Rad51 (Figure 8a). Contrarily, γ-H2AX was not detected in the immunoprecipitates of p62 (Supplementary Figure 8A). Although PARP-1 was also observed in the immunoprecipitates of p62 in both 786-O and HEK293T cells ( Supplementary Figures 8A and  B), it was not detected in the GFP immunoprecipitates (Supplementary Figure 8B). Moreover, Rad51 was readily detected in HEK293T cells using the antibody of p62 in immunoprecipitation (Supplementary Figure 8C). Similar to treatment with ST, rasfonin also decreased the binding between p62 and Rad51 (Figure 8a and Supplementary Figure 8C), and the interaction between them was also observed in HeLa cells, (Supplementary Figure 8D). Unlike deprivation of p62, the loss of Rad51 did not markedly increase the frequency of micronuclei in 786-O cells, whereas ST still stimulated the formation of micronuclei in the Rad51-depleted cells (Figures 8b and c). However, depletion of Rad51 alone accumulated significant amount of micronuclei in HeLa cells ( Supplementary Figures 8E and F), suggesting that Rad51 may regulate nuclear stability in a cell PARP-1 interacts with p62 and its loss completely inhibits the ST-induced autophagy. Immunoprecipitation was performed to investigate the accumulation of micronuclei in the p62-depleted 786-O cells. Deprivation of p62 remarkably reduced the ST-induced interactions between γ-H2AX and Rad51, and PARP-1 (Figure 9a), and that between Rad51 and PARP-1 was also attenuated after treatment (Figure 9b). However, silencing of p62 increased the amount of Rad51 in the immunoprecipitates of γ-H2AX, and slightly affected the interaction between Rad51 and PARP-1 in untreated cells (Figures 9a and b). Considering that the frequency of micronuclei in the p62-silenced cells, and the recruitment of repair factors by γ-H2AX during DSB, 45 (Figures 9e and f). These data revealed a closer relationship between nuclear stability and autophagy, suggesting that PARP-1 is likely to have a role in connecting the autophagic pathway to the nuclear stability, and p62 may regulate nuclear stability by binding to PARP-1 and by affecting the interaction between PARP-1 and γ-H2AX (Supplementary Figure 9C).

Discussion
We have demonstrated that either LC3-II or pUlk1 showed the characteristics of nuclear localization, and both LC3 and pUlk1 can interact with γ-H2AX, Rad51 or PARP-1, all involved in maintaining genomic stability. Notably, both LC3-I and LC3-II were found in the immunoprecipitates of PARP-1, and Rad51 was found to interact with LC3-II. We thus propose that the nuclear-localized LC3 or pUlk1 likely functions uniquely to link genomic stability to autophagy, two important areas of biology.
LC3-II has been implied to localize in nuclei, 11,36 which has now been confirmed in our study. The autophagic initiator Nuclear LC3-II regulates genomic stability S Yan et al pUlk1 (Ser555) was also found in the nuclear lysates in significant amount. Interestingly, LC3-II was also found in the insoluble nuclear precipitates containing mainly genomic DNA, 40 we therefore assumed that LC3-II may have much broader roles than previously recognized. 8,46 Our data showed that PARP-1 bound to LC3-I and LC3-II, and both LC3 and pUlk1 interacted with Rad51. As a result, the autophagic proteins could regulate nuclear stability through either Rad51 or PARP-1. In addition, we observed a band for relatively LMW pUlk1 in the nuclear fraction. Actually, cleavage of the autophagy-targeted proteins happened not solely to Ulk1. As the results from previous studies have shown that both Atg5 and Beclin 1 can be cleaved, 47,48 and caspase 8 was found to cleave Beclin 1, leading to inhibition of autophagy, 49,50 it is important and necessary to explore the function of the LMW Ulk1 in future study.
As ST failed to accumulate micronuclei in either the p62-or p62/LC3-depleted cells, the ST-induced formation of micronuclei is obviously p62 dependent, and the basal and induced accumulations of micronuclei are likely differentially regulated because of the fact that knockdown of p62 alone increased the frequency of micronuclei. Although p62 is often used as a substrate of autophagy, mounting evidence has indicated that it has more active roles in regulating autophagy. 44 Consequently, p62 may regulate autophagy in a stimulus-or/ and cell type-dependent manner. Besides LC3 and Ulk1, p62 was also found to interact with both Rad51 and PARP-1 in this study, suggesting that it may regulate nuclear stability through either one or both. As p62 is a protein connecting the ubiquitin system and the autophagic machinery, 17 and depletion of Rad51 or PARP-1 completely inhibited the ST-induced autophagic flux, it is very likely that both the autophagic and ubiquitin systems coordinate to regulate nuclear stability. Together with the finding that ST failed to augment micronuclei in the PARP-1-deprived cells, we speculated that p62 may regulate the formation of micronuclei through PARP-1, which has been reported to vigorously participate in the regulation of autophagy. 52,53 Different from silencing of Rad51, ablation of PARP-1 resulted in nuclear instability in 786-O cells, indicating that PARP-1 is more important than Rad51 to maintain the basal genomic stability at least in this cell line. Interestingly, observation of interaction between LC3 and PARP-1 in nuclear lysates and co-immunoprecipitation of Rad51 with LC3-II implied that the nuclear localization of LC3-II may participate in maintaining genomic stability.
Although commonly used as a marker for DSB, γ-H2AX indeed actively participates in the DNA repairing process, 45,54 and it was thought to function as an adaptor for recruiting modifying factors of chromatin remodeling. 55 Here, it was found to interact with either PARP-1 or Rad51, which were disrupted by the loss of p62. Therefore, these proteins presumably formed different regulatory complexes to participate in maintaining genomic stability or other physiological processes. Although either ST or rasfonin elevated the level of γ-H2AX, both reduced the expression of Rad51 essential for HR repair, implying that ST and rasfonin induced acute DNA damage leading to cell death; on the other hand, they may cause chromatin remodeling and induce the error-prone DNA repair mechanisms, NHEJ. Given that depletion of either PARP-1 or Rad51 completely inhibited the ST-dependent autophagic flux, we propose that a balance or a switch may exist in these proteins to regulate autophagy and genomic Following treatment with STwith or without CQ for 3 h, cell lysate were prepared, immunoblotting was performed with the antibodies indicated (c), and treated with the indicated compounds for 12 h; images were obtained using fluorescence microscopy after labeling the antibodies of LC3 and p62 (white arrows indicated micronuclei). The histogram graph data representing the mean ± S.D. were shown in (e). **Po0.01 versus control, and *Po0.05 versus control; ## Po0.01 versus the Mock group, ns was short for insignificance. At least three independent experiments were performed Nuclear LC3-II regulates genomic stability S Yan et al stability. In fact, we observed that the interaction between LC3 and Rad51 was regulated by culturing condition, and the binding between LC3 and Rad51 was enhanced by fresh medium in nuclei concurring with decreased interaction in cytoplasm.
In summary, our data revealed more intimate and complicated relationship between autophagy and nuclear stability. The nuclear localization of LC3-II/pUlk1 and their interactions with PARP-1 resulted in their direct functions in this cellular organelle. Future work in this direction will provide more clues for better understanding of the mechanisms for necleophagy and non-selected autophagy. Cell culture and immunoblotting analysis. 786-O, ACHN, HeLa, HepG2, HEK293T and K562 cells were grown in DMEM media containing 10% fetal bovine serum (GIBCO, Grand Island, NY, USA) and 1% antibiotics. Cells were grown to 70-80% confluency before addition of a variety of compounds. For siRNA interference, cells of 30% confluence in the media without antibiotics were transfected using DharmaFECT (Dharmacon, T2001) according to the manufacturer's instructions. Cells were split and cultured overnight before stimulations after transfection for 48 h. Whole-cell lysates were prepared with lysis using Triton X-100/ glycerol buffer, containing 50 mM Tris-HCl (pH 7.4), 4 mM EDTA, 2 mM EGTA and 1 mM dithiothreitol, supplemented with 1% Triton X-100, 1% SDS and protease inhibitors, and then separated on a SDS-PAGE gel and transferred to PVDF membrane. Immunoblotting was performed using appropriate primary antibodies and horseradish peroxidase-conjugated suitable secondary antibodies, followed by detection with enhanced chemiluminescence (Pierce Chemical, Rockford, IL, USA). Subcellular fractionation. Cells were seeded into 100 mm dishes at 90% confluency. After the indicated treatment, cells were gathered, pelleted by centrifugation at 3000 r.p.m. for 5 min, and washed three times with cold PBS. In all, 20% cells were resuspended in Triton X-100/glycerol buffer and labeled as the total homogenate. Method A: the other cells were resuspended in 400 μl homogenization buffer A (10 mM Hepes-KOH (pH 7.9), 10 mM KCl, 1.5 mM MgCl 2 , 0.5 mM PMSF and 0.5 mM dithiothreitol) containing 0.5% NP-40, and then the homogenate was centrifuged at 3000 r.p.m. at 4°C for 5 min after static on ice for 15 min. The supernatant was collected as the nuclear cytoplasm (Cyto). After washing twice with 400 μl buffer A without NP-40, the pellet was resuspended in 60 μl buffer C (20 mM Hepes-KOH (pH 7.9), 600 mM KCl, 1.5 mM MgCl 2 , 0.2 mM EDTA and 25% glycerol). After rotating on ice for 15 min, the homogenate was centrifuged at 13 000 r.p.m. at 4°C for 15 min, and the supernatant was collected as the soluble nuclear fractions (Nu), and pellets were collected as the insoluble nuclear participates (Nup). After adding 30 μl 3 × loading buffer or 60 μl 1 × loading buffer to the Nu or Nup, respectively, the samples were boiled at 96°C for 15 min before separating on a SDS-PAGE gel. Method B: the other cells were subjected to a nuclear extraction kit (Thermo Scientific, Waltham, MA, USA; 78835), and the Cyto and Nu fractions were extracted following the instructions. Immunoprecipitation was performed in the Cyto and Nu fractions extracted using Method B.   (e and f) Cell lysates were prepared and immunoblotting was performed with the indicated antibodies for 3 h. The data represent three independent experiments. **Po0.01, and ns was short for insignificance Nuclear LC3-II regulates genomic stability S Yan et al chemiluminescence signals and quantifications were carried out using densitometry, and mean ± S.D. were shown in histograms along with the blots. The normally distributed data are shown as mean ± S.D. and analyzed using one-way analysis of variance and the Student-Newman-Keuls post-hoc test. For the non-normally distributed data in the electron microscopy, results were show as mean, and analyzed using Friedman test.