Smurf2-Mediated Stabilization of DNA Topoisomerase IIα Controls Genomic Integrity

DNA topoisomerase IIα (Topo IIα) is the major form of the Topo II enzyme in cycling vertebrate cells that acts to untangle chromosomal catenanes forming during the duplication of genetic material. Topo IIα plays a pivotal role in chromatin organization, dynamics, and unperturbed chromosome inheritance. The failure of Topo IIα to maintain DNA supercoiling homeostasis and properly disentangle daughter chromosomes can lead to the formation of pathological chromosome bridges, chromosomal instability (CIN), and ultimately to cancer (1–4). Furthermore, due to the high abundance of Topo IIα in rapidly proliferating cells and its vital roles in mitotic processes, Topo IIα is a core target of several anticancer drugs (4, 5). Despite the importance of Topo IIα, the mechanisms governing its cellular levels remain largely unknown.

Recently, we reported that HECT-type E3 ubiquitin ligase Smurf2 operates in mammalian cells as a critical regulator of chromosome integrity, and acts to prevent the CIN, and carcinogenesis (6). We showed that genomic ablation of Smurf2 leads to accumulation of chromosomal aberrations, including Robertsonian translocations, undefined translocations and marker chromosomes. Furthermore, we demonstrated that mice knockout for Smurf2 develop a wide spectrum of tumors in different organs and tissues. These and other studies established Smurf2 as an important regulator of genomic integrity, whose inactivation results in carcinogenesis (7). However, the mechanisms underlying tumor-suppressor functions of Smurf2 are far from understood.

Here, we report a novel mechanism by which Smurf2 is involved in genome integrity regulation—via stability regulation of Topo IIα, and prevention of anaphase bridge formation. We also show that cellular levels of Smurf2 delineate Topo IIα protein levels in both mouse and human normal and cancer cells and tissues, and could determine cell sensitivity to Topo II poison etoposide.

Smurf2 knockout (Smurf2^−/−) mouse embryonic fibroblasts (MEF) and wild-type cells derived from littermate control embryos were originally established in the laboratory of Dr. Ying Zhang (National Cancer Institute, NIH, Bethesda, MD), and obtained from her laboratory in 2012 (6). These cells were obtained at passages 30–35, and used between passages 40 and 55. Human cell lines used in this study were generously provided by Prof. Yosef Shiloh (Tel Aviv University, Tel Aviv, Israel). These cells were obtained in 2013, and originated from the ATCC. These cell lines were propagated, frozen, and used in culture for up to 12 passages. All cell lines were maintained in high glucose DMEM medium (4,5 g/L d-Glucose, Gibco) supplemented with 10% (v/v) FBS, 2 mmol/L l-glutamine, and 1% (v/v) penicillin–streptomycin, and incubated at 37°C with 5% CO₂. Cell strains were selectively tested for mycoplasma at the Faculty Core Facility using a PCR-based approach. The latest date for testing of our leading cellular models, which include U2OS wild-type cells, Smurf2 knockout cells, Smurf2-knockdown strains, and cells expressing Smurf2 catalytically inactive (Cys716Gly) and wild-type forms, was January 24, 2017. Cell line authentication was not conducted.

Human tissue microarrays (TMA) were purchased from U.S. Biomax, Inc. Mice tissues were fixed in 4% formalin and 5-μm tissue sections were prepared. IHC was conducted as described previously (6). All comparable samples were sampled on the same slide, and all staining procedures were conducted on slides positioned horizontally. Histologic evaluations and TMAs scoring were conducted by a board-certified pathologist at the Galilee Medical Center (Nahariya, Israel).

GST fusion proteins were prepared from E. coli using glutathione-Sepharose beads (Amersham); purified full-length human Topo IIα was purchased from TopoGen (TG200H). An in vitro binding assay was performed as described previously (6). Briefly, GST or GST-Smurf2 were incubated with Topo IIα in binding buffer for 15 minutes at 37°C and GST-Smurf2 was pulled-down using Glutathione Sepharose 4B beads (GE Healthcare). The beads were then washed four times with ice-cold binding buffer and proteins were eluted with 5× SDS sample buffer.

In vivo and in vitro ubiquitination assays were performed as previously described (6, 8) with a few modifications. In brief, for the in vivo ubiquitination assay cells were lysed with either RIPA buffer supplemented with 5 mmol/L NEM or in 1% SDS followed by an immediate boiling of samples for 15 minutes. Following the boiling, cell lysates were equilibrated with RIPA/NEM buffer to reduce SDS concentration in the samples down to 0.1%. Cell lysates were then sonicated, FLAG-Topo IIα was immunoprecipitated, and its ubiquitination pattern analyzed.

For the in vitro ubiquitination assay, 500 ng of Topo IIα were incubated with 250 ng of GST or GST-Smurf2, 5 μg of HA-ubiquitin protein, E1 (UBE1; 100 ng) and E2 enzyme (UbcH5c; 150 ng), and 100 mmol/L ATP-Mg in the E3 ligase reaction buffer (Boston Biochem) for 2 hours at 37°C. RIPA buffer was added to the reactions and Topo IIα was pulled down using anti-Topo IIα antibody (Abcam) and protein G-Sepharose beads.

Nuclear extracts for DNA decatenation assay were prepared as described previously (9), with a few modifications. In brief, cell pellets were resuspended in TEMP buffer (10 mmol/L Tris-HCl, pH 7.5, 1 mmol/L EDTA, 4 mmol/L MgCl₂, 0.5 mmol/L PMSF), and incubated on ice for 15 minutes. Subsequently, 0.6% Nonidet-P40 substitute was added, samples vortexed for 10 seconds, and centrifuged (20,000 × g, 30 seconds). The supernatants, representing the cytoplasmic fraction, were collected and saved for Western blot analysis. The remaining nuclei were then washed in ice-cold TEMP, resuspended in a TEP buffer (10 mmol/L Tris-HCl pH 7.5, 1 mmol/L EDTA, 500 mmol/L NaCl, 0.5 mmol/L PMSF), vortexed at 4°C for 30 minutes, and centrifuged for 20 minutes at 20,000 × g. The supernatants, representing the nuclear fraction, were collected and saved for subsequent analyses. To verify the complete extraction of Topo IIα from the samples, the remaining insoluble fractions were solubilized in TEP buffer using a sonication, and subsequently analyzed in immunoblots.

DNA decatenation assay was performed using nuclear extracts and kinetoplast DNA (kDNA; TopoGEN) in a complete buffer assay. In particular, each reaction contained 50 mmol/L Tris HCl pH 8, 150 mmol/L NaCl, 10 mmol/L MgCl₂, 2 mmol/L ATP, 400 ng of kinetoplast DNA, 2 μg of nuclear extract and double-distilled water to bring the final volume up to 40 μL. Reactions were incubated at 37°C for 30 minutes, and stopped by adding 8 μL of 5×Stop Buffer (TopoGEN). Subsequently, samples were incubated with RNase (40 μg/mL) at 37°C for 15 minutes and with Proteinase K (150 μg/mL) for an additional 15 minutes at 37°C. Finally, samples were separated by electrophoresis through a 1% agarose gel stained with SYBRS Safe DNA Gel Stain (Invitrogen), and visualized in the SyngeneG:BOX.

All other methods and reagents we used in this study are detailed in Supplementary Methods Section.

To identify novel Smurf2-binding partners, we employed immunoprecipitation coupled with mass spectrometry (MS) analyses. These analyses were conducted on Smurf2-deficient MEFs reconstituted either with a full-length FLAG-tagged Smurf2 or with an empty vector as a control. FLAG-Smurf2 immunoprecipitates were resolved in SDS-PAGE, and protein bands were visualized using Coomassie staining (Fig. 1A). Bands specifically associated with Smurf2 immunoprecipitates were excised from the gel and submitted for identification in MS. To deduct a background, the bands were cut side-by-side from both positive (Smurf2-reconstituted cells) and control lanes (empty vector). Using this approach, we identified Topo IIα as a novel Smurf2 interactor with a high degree of reliability: Topo IIα-specific protein sequence was detected in 146 peptides in Smurf2-immunoprecipitated samples versus 0 peptides in control samples (immunoprecipitates from cells expressing an empty vector; Fig. 1A).

To validate the MS results, we conducted several lines of experiments. First, we demonstrated the interactions between Smurf2 and Topo IIα by coimmunoprecipitation experiments in MEFs, the cell model that has been used to discover the Smurf2-Topo IIα protein interaction (Fig. 1B). Next, we show that this interaction is also preserved in human cells. Immunoprecipitation of MYC-tagged Smurf2 from HEK-293T cells detected endogenous Topo IIα in complex with Smurf2, and vice versa, immunoprecipitation of FLAG-tagged Topo IIα detected endogenous Smurf2 complexed with Topo IIα (Fig. 1C). Finally, using these cells we demonstrated the interaction between both endogenous Smurf2 and Topo IIα (Fig. 1D). Essentially, the discovered interaction between Smurf2 and Topo IIα appears to be direct, because we were also able to detect the interaction between purified human Topo IIα and Smurf2 in the tube (Fig. 1E).

To visualize the interaction between Smurf2 and Topo IIα in cells, we expressed GFP-Smurf2 in human osteosarcoma U2OS cells. These cells have a large nucleus and are commonly used in localization microscopy to detect molecular interactions, in particular in the nuclear compartment. Following cell fixation and immunostaining with the Topo IIα-specific antibody, we evaluated the colocalization between GFP-Smurf2 and Topo IIα under the confocal microscope. The specificity of the anti-Topo IIα antibody used in our studies was rigorously validated using Topo IIα knockdown cells. The results (Fig. 2A) show that in interphase cells Smurf2 occupies both nuclear and cytoplasmic compartments, whereas Topo IIα exhibits exclusively nuclear localization. In the nucleus, Smurf2 showed a high degree of colocalization with Topo IIα, in particular with its nucleolar fraction, as evident from the images recorded in the Nomarski imaging mode.

To validate the interaction between Smurf2 and Topo IIα, we recorded cell images of Smurf2/Topo IIα-stained cells at different focal planes through the cell volume (Z-stack analysis). The results we obtained corroborated the interaction between Smurf2 and Topo IIα, and indicated that in interphase cells Smurf2 associates with Topo IIα through the nuclear volume (Supplementary Fig. S1).

Next, an in situ proximity ligation assay (PLA), which enables determining the protein–protein interactions directly within the cell, provided further evidence that Smurf2 and Topo IIα physically interact with each other (Fig. 2B and C). Immunoprecipitation studies conducted in U2OS cells further corroborated the interaction between Smurf2 and Topo IIα in this cell model (Fig. 2D).

During mitosis, Topo IIα is tightly associated with chromatin, and follows the chromatin movement pattern. The examination of Topo IIα and Smurf2 biodistribution in U2OS cells going through the unperturbed mitosis revealed that while Topo IIα associates with mitotic chromosomes, Smurf2 is mainly excluded from the interactions with chromatids (Fig. 2E). Moreover, in mitotic cells Smurf2 was mostly found at the periphery of Topo IIα-chromatin templates, including the spindle midzone. The observed biodistribution of Smurf2 and Topo IIα in interphase and mitotic cells is in accordance with the previous studies, which determined the localization of each of these proteins in human cells, but independently of each other (6, 10, 11). Collectively, these findings establish Topo IIα as a novel Smurf2 interactor.

To gain an insight into the biological significance of the complex formation between Smurf2 and Topo IIα, we first examined the levels of Topo IIα in Smurf2 knock-out (KO) vs. wild-type MEF cells. These cells were derived from littermate control embryos. We found that the steady-state levels of Topo IIα were profoundly diminished in Smurf2KO MEFs (Fig. 3A, Left). To determine whether the loss of Smurf2 alone is responsible for the decrease in the cellular levels of Topo IIα, we restored Smurf2 expression in KO MEFs and found that upon Smurf2 reconstitution the cellular levels of Topo IIα were significantly increased (Fig. 3A, Right). Essentially, IHC and western blot analyses conducted on the tissue samples of Smurf2-deficient vs. control mice revealed that diminished Topo IIα protein levels were also a characteristic of Smurf2-ablated tissues (Fig. 3B and C; Supplementary Fig. S2A and S2B). The levels of Topo IIβ, which activities are mostly associated with transcription regulation (2), were comparable between Smurf2KO and control tissues.

Similar results were also obtained in different human cell models: in U2OS osteosarcoma cells, HCT116 colon carcinoma cells, and DU145 and PC-3 prostate carcinoma cells. In all these cells, decrease of Smurf2 expression levels either through acute (using siRNAs) or stable knock-down (using lentiviral-based shRNAs) reduced the steady-state levels of Topo IIα (Fig. 3D). These effects were monitored through the use of four different si/shRNAs: designed to target Smurf2 mRNA either at 3′UTR or its coding sequence.

Next, to validate our biochemical data at a single-cell resolution, we performed immunofluorescence staining of Topo IIα in Smurf2 knockdown cells. The immunofluorescence studies were conducted concomitant to the Western blot analyses performed on WCL, as well as on the fractionated samples: cytosolic, nucleoplasmic, and chromatin fractions (Fig. 3E; Supplementary Fig. S2C). The results we obtained corroborated that the steady-state levels of Topo IIα in Smurf2-depleted cells are significantly diminished. Of note, the observed Topo IIα-staining pattern and its decreased levels in Smurf2 knock-down cells were independent of the cell fixation procedure, and monitored in both formaldehyde- and methanol/acetone-fixed cells (Supplementary Fig. S2C).

Furthermore, using the CRISPR/Cas9-based gene-editing system, we generated Smurf2^CRISPR U2OS cell line, in which we succeeded to decrease Smurf2 cellular levels down to 10%–15% of the original level (Fig. 3F). Using these cells, we demonstrated that targeting SMURF2 for inactivation at the genome level produced in human cells results similar to both mouse Smurf2^−/− cells and tissues (Fig. 3A–C; Supplementary Fig. S2B), and to human cell lines knocked-down for Smurf2 with RNAi (Fig. 3D and E). Of note, both the levels of Topo IIβ and DNA topoisomerase 1 (Top1), which also operates in DNA transition processes to maintain supercoiling homeostasis (12), were unaffected by Smurf2 depletion (Fig. 3F). Furthermore, we conducted bi-parametric FACS analyses on cells in which Smurf2 was depleted either through CRISPR/Cas9 gene editing or using RNAi. These analyses provided evidence that targeting of Smurf2 for inactivation does not affect cell-cycle distribution, although a moderate decrease in the mitotic population of Smurf2-knockdown cells was observed (Supplementary Fig. S3).

To determine whether Smurf2 affects Topo IIα protein levels by modulating its gene expression, we analyzed the mRNA expression levels of Topo IIα in Smurf2 knockdown and control U2OS and HCT116 cells using real-time qRT-PCR. The data showed that mRNA levels of Topo IIα were unaffected by Smurf2 depletion (Fig. 3G).

Finally, we demonstrated that overexpression in these cells of MYC-Smurf2 increases Topo IIα protein levels proportional to the amount of Smurf2 transduced to the cells (Fig. 3H). Altogether, these data reveal Smurf2 as an essential regulator of the steady-state levels of Topo IIα, and suggest that Smurf2 regulates Topo IIα protein levels posttranslationally.

To investigate whether Smurf2 manages Topo IIα protein levels by regulating its degradation, we conducted several lines of experiments. First, we treated Smurf2-deficient and -overexpressing cells with the proteasome inhibitor MG-132 and found that the stability of Topo IIα is regulated through the proteasome-mediated degradation, and is under Smurf2 control (Fig. 4A and B; Supplementary Fig. S4A and S4B). Following cell treatment with a proteasome inhibitor, the Topo IIα levels in Smurf2-proficient cells were notably increased, yet slightly affected in Smurf2 knockdown cells (Fig. 4A, lane 2 vs. 4). In contrast, overexpression of Smurf2 in these cells dramatically increased the levels of Topo IIα in both untreated and MG-132–treated cells as compared with control samples transduced with an empty vector (Fig. 4B, lanes 3 and 4 vs.1 and 2). Noteworthy, MG-132 cell treatment did not lead to further stabilization of Topo IIα in Smurf2-overexpressing cells. This finding suggests that high Smurf2 levels were sufficient to prevent Topo IIα proteasomal degradation. Finally, the coexpression of FLAG-Topo IIα together with MYC-Smurf2 protected Topo IIα from the proteasome-mediated degradation in Smurf2 dose-dependent manner (Supplementary Fig. S4A). Of note, we did not find significant changes in the levels of Topo IIα in chloroquine-treated versus control cells (Supplementary Fig. S4B). Taken together, these data suggest that Smurf2 stabilizes Topo IIα by inhibiting its proteasomal degradation.

Smurf2 is a HECT-type E3 ubiquitin ligase in which active-site cysteine 716 (Cys716) is crucial for its catalytic activity. To determine whether E3 ligase functions of Smurf2 are required for Topo IIα stability regulation, we analyzed Topo IIα protein levels in U2OS cells expressing either Smurf2WT or its catalytically inactive form, in which active-site cysteine was substituted with glycine (Cys716Gly; Smurf2CG). Cells transduced with an empty vector served as a control. The data (Fig. 4C) show that only the expression of Smurf2WT was able to increase the Topo IIα protein levels. Similar results were also obtained in Smurf2-ablated MEF cells reconstituted with Smurf2WT or its E3 ligase mutant form (Fig. 4D).

To determine whether Smurf2 is capable of ubiquitinating Topo IIα, we conducted in cellulo ubiquitination assay in which FLAG-Topo IIα was coexpressed together with HA-tagged ubiquitin and Myc-tagged Smurf2. FLAG-Topo IIα was then immunoprecipitated and its ubiquitination pattern was analyzed. To perform this analysis, we used an anti-HA antibody, which specifically recognized the HA-tagged ubiquitin attached to Topo IIα. This experimental design allowed us to identify that Topo IIα undergoes ubiquitination (Fig. 4E, lanes 3–5 vs. 1 and 2). This design also allowed us to discover that the addition of catalytically active Smurf2 significantly enriched Topo IIα monoubiquitination (Fig. 4E, lanes 4 vs. 3). Moreover, the results show that the substitution of Smurf2WT with its mutant form failed to produce this ubiquitination phenomenon (Fig. 4E, lane 5 vs. 4). The data also show that in the cells, which expressed Smurf2CG, the Topo IIα is polyubiquitinated. The observed polyubiquitination of Topo IIα, whereas evident in Smurf2CG-expressing cells, was less visible in cells transfected with an empty vector (Fig. 4E, left; lane 5 vs. 3). We assumed that this finding due to the incomplete inactivation of cellular deubiquitinases upon cell lysis. To clarify this point, we conducted the Topo IIα ubiquitination experiment again, but to deactivate the cellular deubiquitinases more efficiently, in the second round, the cells were lysed directly in 1% SDS, followed by an immediate sample boiling. Similar to our previous findings, the addition of Smurf2WT, but not Smurf2CG or an empty vector, significantly enriched the monoubiquitinated fraction of Topo IIα (Fig. 4E, right). In addition, we verified that the expression of catalytically active Smurf2 switched the ubiquitination code on Topo IIα from poly- to monoubiquitination, as compared with samples expressing an empty vector or Smurf2CG (Fig. 4E, right; lane 4 vs. 3 and 5).

Our data indicate that the protein stability of Topo IIα is regulated through the proteasome-mediated degradation. This type of proteolysis relies on the formation of a particular type of ubiquitination–K48–linked polyubiquitination. To examine whether Smurf2 protects Topo IIα from the degradation by modulating this particular type of ubiquitination, we coexpressed FLAG-Topo IIα with Smurf2 together with either wild-type ubiquitin or its mutant form (K48-only ubiquitin). In this mutant ubiquitin, all the lysines except K48 were mutated to arginines. These mutations abolish the ability of ubiquitin to form polyubiquitin chains other than through the K48-linked chain. Both wild-type and mutant forms of ubiquitin were HA-tagged. FLAG-Topo IIα was then immunoprecipitated and its ubiquitination pattern analyzed using anti-HA antibody. The results show that similar to the previous findings the expression of Smurf2 together with a wild-type form of ubiquitin switched the Topo IIα ubiquitination from poly- to monoubiquitination (Fig. 4F, lane 2 vs. 1). However, when the wild-type ubiquitin was substituted to K48-only ubiquitin, and expressed together with Smurf2, the K48-linked polyubiquitination of Topo IIα was completely abolished (Fig. 4F, lane 4 vs. lane 3). These data suggest that Smurf2 operates as a molecular editor that modifies the Topo IIα ubiquitination code to protect Topo IIα from the degradation-promoting–K48 polyubiquitination. The data also indicate that the E3 ligase activities of Smurf2 are required for the switch.

Finally, we performed a ubiquitination reconstitution assay using purified ubiquitin-activating enzyme (E1), ubiquitin conjugase (E2), HA-ubiquitin, Smurf2 (either wild-type or E3 mutant form), and human Topo IIα. Using these experimental settings, we demonstrate that Smurf2 ubiquitinates Topo IIα directly (Fig. 4G). The specificity of this reaction is demonstrated by the unique ubiquitination pattern of Topo IIα monitored only in the presence of Smurf2WT. In addition, we validated that the observed ubiquitination pattern belongs to Topo IIα, and not to Smurf2 or any other components used in the ubiquitination assay (Supplementary Fig. S4C and S4D).

Topo IIα is a key player in the decatenation checkpoint, and its inability to unwind chromosomal entanglements can lead to the formation of pathological chromosome bridges—one of the major sources of chromosomal translocations. On the basis of these findings, and our data establishing Smurf2 as a positive regulator of Topo IIα, we hypothesized that cells depleted of Smurf2 will exhibit a phenotype similar to the Topo IIα-depleted cells. Specifically, we hypothesized that these cells will exhibit the reduced ability to untangle chromosomal catenanes, and will show the exacerbated formation of anaphase bridges.

To test this hypothesis, we first examined the Topo II decatenation activity in nuclear extracts prepared from Smurf2-knockdown and control cells using the decatenation assay. Nuclear extracts prepared from Topo IIα and Topo IIβ knockdown cells were also incorporated in the analysis, and served as additional controls. The data (Fig. 5A and B; Supplementary Fig. S5A) show that DNA decatenation was significantly reduced in Smurf2-knockdown cells, and was comparable with the cells knockdown for Topo IIα. In addition, using Topo IIα and Topo IIβ knockdown cells, we demonstrate that Topo II decatenation assay used in our study is specific to monitor activities of Topo IIα. This finding is in agreement with results previously published by Bower and colleagues (13), and showing that Topo II assay mainly measures activities of Topo IIα. Furthermore, by assaying Topoisomerase I activity, we demonstrate that the extracts derived from Smurf2-knockdown cells are otherwise normal, and have unaltered Top1 activity (Supplementary Fig. S5B). The sensitivity of Top1 assay was validated using Top1 knockdown cells (Supplementary Fig. S5C and S5D).

Next, we analyzed and compared the occurrence of chromosome bridges, and lagging chromosomes, in Smurf2- and Topo IIα-depleted cells. To this end, we used the U2OS cell model. These cells have been previously shown to exhibit an increased DNA bridge formation after Topo IIα knockdown (14). The data (Fig. 5C and D; Supplementary Fig. S5E) show that Smurf2 depletion significantly increased the formation of DNA bridges in mitotic cells (P = 0.000212, χ² test). Moreover, the incidence of DNA bridges in Smurf2-depleted mitotic cells was highly similar to the incidence of these bridges in Topo IIα knockdown cells (Fig. 5C). This finding suggests that Smurf2 depletion phenocopies the Topo IIα depletion in the resolving of DNA catenanes. Essentially, we demonstrated that Smurf2 inactivation had no significant effect on the occurrence of lagging chromosomes (Fig. 5E and F). These chromosomes are formed through the mechanisms distinct from the mechanisms underlying anaphase bridge formation, and are independent of Topo IIα (14–16).

Finally, we reconstituted Smurf2^CRISPR U2OS cells with mCherry-tagged Topo IIα (or with an empty vector as a control) and, following the generation of stable cell lines, analyzed the formation of anaphase bridges and lagging chromosomes in these cells. The data (Fig. 5G; Supplementary Fig. S5F) show that in Smurf2-depleted cells transduced with Topo IIα the cellular phenotype was rescued, as evident by a significant decrease in the population of mitotic cells with chromatin bridges compared with control (P = 0.014, χ² test). The incidence of lagging chromosomes in these cells remained unaffected (Fig. 5H). Of note, decatenation assay conducted on Topo IIα-reconstituted cells suggested that the N-terminally tagged mCherry-Topo IIα is catalytically active (Supplementary Fig. S5G); similar results were reported by Lane and colleagues (17), showing that mCherry-tagged Topo IIα can functionally complement the loss of endogenous protein.

The cellular levels of Topo IIα are an important determinant of tumor cell sensitivity to Topo II-targeting drugs, in particular to etoposide: higher Topo IIα levels, higher sensitivity, and vice versa (18–21). Topo II poison etoposide acts to stabilize transient Topo II-bridged break and lead to generation of highly cytotoxic DNA damage–double-strand breaks. To determine whether and how manipulations with Smurf2 cellular levels affect cell sensitivity to etoposide, we assessed the sensitivity of Smurf2^CRISPR and control wild-type U2OS cells to this drug. The data (Fig. 5I) show that Smurf2^CRISPR cells were more resistant to etoposide treatment than control cells. Similar results were also observed in Smurf2KO versus WT MEF cells (Supplementary Fig. S5H).

To determine whether Smurf2 is also a relevant determinant of Topo IIα protein levels in human tissues, we conducted IHC analyses on two TMAs: prostate (PR1921), which included both normal (n = 31) and cancer tissues (n = 149); and breast TMA (BR10010) containing primary and metastatic carcinoma tissues (n = 98). These TMAs were stained with anti-Topo IIα and anti–Smurf2 IHC-specific antibodies, and counterstained with hematoxylin. The subsequent histopathologic examination coupled with the Spearman's rank correlation analysis revealed that the expression levels of Smurf2 and Topo IIα are positively correlated. This correlation was statistically significant for all analyzed tissues: P < 0.001 for breast and prostate carcinoma tissues, and P < 0.05 for normal prostate tissues (Fig. 6A–D; Supplementary Table S1).

In this study, we identified a novel mechanism by which Smurf2, an HECT-type E3 ubiquitin ligase and recently discovered tumor suppressor, is involved in the regulation of genome integrity. Recently, we reported that cells depleted of Smurf2 accumulate multiple chromosome abnormalities in their genome, where translocations were the most notable hallmark (6). However, the mechanisms underlying this phenomenon are far from understood.

The formation of pathological chromatin bridges is considered one of the major causes of chromosomal translocations, which are viewed as an outcome of the compromised decatenation checkpoint mediated by Topo IIα (16, 22). Inability of Topo IIα to decatenate DNA (e.g., due to its reduced cellular levels and/or activities) can lead to CIN and, ultimately, to cancer. Moreover, in established tumors, CIN can drive tumor progression by accelerating the gain of oncogenic loci and the loss of tumor-suppressor loci (2, 4).

Using different human and mouse cellular models, Smurf2-targeting approaches, human tissue arrays, genetically modified systems and rescue experiments, we provided confirmatory insights that Smurf2 acts as a key cellular factor that stabilizes Topo IIα and prevents the formation of pathological DNA bridges. Mechanistically, we demonstrated that Smurf2 physically interacts with Topo IIα and protects it from the proteasome-mediated degradation in Smurf2 dose- and catalytically dependent manners. We showed that Smurf2 operates as a molecular switcher that modifies the Topo IIα ubiquitination code to reduce its degradation-promoting K48 polyubiquitination and increase monoubiquitination. Essentially, we demonstrated that Smurf2 is capable to directly bind to and ubiquitinate Topo IIα. Finally, we provided evidence that Smurf2 is a relevant determinant of Topo IIα protein levels in mouse and human normal and cancer tissues, and that Smurf2 levels could affect the cell sensitivity to Topo II poison and anticancer drug etoposide.

Currently, there is limited information about the mechanisms that regulate Topo IIα protein levels and stability. Studies published to date mainly report that following cell treatment with topoisomerase II poisons, Topo IIα undergoes proteasome degradation, which enables repair of DNA lesions (23–26). Our study demonstrates that Topo IIα undergoes proteasomal degradation also in undamaged cells, and it is under the control of Smurf2. This finding is intriguing because Smurf2 is primarily known as a degradation promoting E3 (7, 27). Nonetheless, our data strongly suggest that Smurf2 is an authentic positive regulator of Topo IIα that protects it from the proteasome-mediated proteolysis in Smurf2 dose- and E3 ligase-dependent manners. The data obtained in IHC studies conducted on tissues derived from Smurf2-deficient and wild-type mice, as well as on human TMAs (278 normal and tumor tissues), provided further support that Smurf2 is also a positive regulator of Topo IIα in tissues.

The functional and cellular assays conducted in this study provided evidence that Smurf2-depletion phenocopies Topo IIα depletion. This finding is quite intriguing especially in light of mouse phenotypes reported for Topo IIα and Smurf2: Topo IIα-deficient mice are embryonic lethal (28), whereas Smurf2KO mice are viable and exhibit no overt developmental defect during embryogenesis (6, 29). There are a few possible explanations that could reconcile these mouse phenotypes. First, as we demonstrated in a panel of different human and mouse cells and tissues (Fig. 3), the depletion of Smurf2 significantly reduced, but not ablated Topo IIα levels. It is possible that the remaining levels of Topo IIα were sufficient to complete embryogenesis. Second, during embryogenesis other members of the NEDD4 E3 ligase family (nine in total) could compensate for Smurf2 deficiency. For example, mice with a targeted disruption of either the Smurf2 or Smurf1 allele are viable and survive to adulthood (6, 29, 30); however, disruption of both Smurf2 and Smurf1 alleles leads to embryonic lethality (31). The possibility that Smurf2 does not affect the cellular levels of Topo IIα during early embryogenesis, where it appears to be particularly important (28), also exists.

Collectively, our findings establish a novel functional link between an E3 ubiquitin ligase Smurf2 and Topo IIα, and propose a new paradigm in the stability regulation of Topo IIα, and genome integrity maintenance. Our data also suggest an intriguing role of Smurf2 as an E3 ligase that can protect some of the key cellular proteins from degradation, and provide a novel insight into the tumor suppressor functions of Smurf2.

No potential conflicts of interest were disclosed.

Conception and design: M. Blank

Development of methodology: P. Koganti, G. Levy-Cohen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Emanuelli

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Apel-Sarid, M. Blank

Writing, review, and/or revision of the manuscript: M. Blank

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Emanuelli, A.P. Borroni, P.A. Shah, D. Manikoth Ayyathan, G. Levy-Cohen

Study supervision: M. Blank

Other (carried out most of the experimental studies in this article): A. Emanuelli

Other (generation of cell lines): P. Koganti

We thank Meir Shamay for helpful discussions during this article preparation and Basem Hijazi for statistical analysis. We are also grateful to Ying E. Zhang for providing Smurf2-deficient mice and for other support.

This work was supported by several grants, including ICRF 00636, Marie-Curie FP-7 CIG 612816, and Dayan Family Foundation award to M. Blank.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.