Resolution of Sister Centromeres Requires RanBP2-Mediated SUMOylation of Topoisomerase IIα

Generation of Mutant Mice with Low Amounts of RanBP2

We created a series of mice in which expression of RanBP2 is reduced in a graded fashion from normal to zero by the use of various combinations of wild-type (RanBP2⁺), hypomorphic (RanBP2^H), and knockout (RanBP2⁻) alleles. The RanBP2^H allele was generated by inserting a neomycin resistance cassette into the third intron of the RanBP2 gene via homologous recombination (Figures 1A and 1B). The RanBP2⁻ allele was established by removing exon 3 from the RanBP2^H allele via Cre-mediated recombination (Figures 1A and 1B). As previously described (Aslanukov et al., 2006), RanBP2^−/− mice died during embryogenesis. Death occurred prior to day 13.5 of development (data not shown). In contrast, RanBP2^−/H and RanBP2^H/H mice were viable and overtly indistinguishable from RanBP2^+/−, RanBP2^+/H, and RanBP2^+/+ mice. Western blot analysis of mouse embryonic fibroblast (MEF) lysates revealed that RanBP2^+/H, RanBP2^+/−, RanBP2^H/H, and RanBP2^−/H cells contained ~90%, 44%, 31%, and 26%, respectively, of the RanBP2 protein level present in RanBP2^+/+ MEFs (Figures 1C and S1).

RanBP2^−/H Cells Have No Overt Transport-Related Defects

Next, we investigated whether nucleocytoplasmic transport might be impaired in RanBP2^−/H cells. In situ hybridization with an oligo(dT)50-mer probe revealed that the intracellular distribution of polyadenylated mRNA was indistinguishable between RanBP2^+/+ and RanBP2^−/H MEFs, indicating that nuclear export of bulk mRNA was not affected by decreased RanBP2 expression (Figure S2A). Furthermore, NLS-mediated protein import and NES-mediated protein export were both equally efficient in RanBP2^+/+ and RanBP2^−/H MEFs as measured using established in vivo transport assays (Figures S2B–S2E). Collectively, these results suggest that reduction of RanBP2 protein levels to about a quarter of normal levels has no overt impact on nucleocytoplasmic trafficking.

RanBP2 can sumoylate RanGAP1 in vitro, but whether it does so in vivo is unclear (Pichler et al., 2002). RanBP2 has further been proposed to protect SUMO1-modified RanGAP1 from de-sumoylation by SUMO isopeptidases such as SENP2 (Zhang et al., 2002). Western blot analysis of RanBP2^+/+ and RanBP2^−/H MEF lysates for SUMO1 showed that SUMO1-RanGAP1 levels remained constant in cells with reduced RanBP2 (Figure S2F). The same holds true for other SUMO1-conjugated proteins and SUMO2/3-conjugated proteins level (Figure S2G), suggesting that hypomorphism for the SUMO E3 ligase RanBP2 does not affect the global patterns of SUMO modification in MEFs.

Mice and MEFs with Low Amounts of RanBP2 Develop Severe Aneuploidy

To determine whether RanBP2 insufficiency leads to chromosomal instability in the context of an adult mouse, we collected splenocytes from RanBP2^+/+, RanBP2^+/H, RanBP2^+/−, RanBP2^H/H, and RanBP2^−/H mice at 5 months of age and performed karyotype analyses. Chromosome counts showed that 0% of RanBP2^+/+ and RanBP2^+/H splenocytes were aneuploid (Table 1A). In contrast, splenocytes from RanBP2^+/−, RanBP2^H/H, and RanBP2^−/H mice had 1%, 5%, and 15% incidence of aneuploidy, respectively, revealing an inverse correlation between the level of RanBP2 protein and the percentage of aneuploidy in this cell type. Moreover, the range of abnormal chromosome numbers broadened substantially with decreasing expression of RanBP2 protein. The extent of aneuploidy was also determined in MEFs with graded reduction in RanBP2 expression. As shown in Table 1B, in passage 5 MEFs, aneuploidy increased with declining RanBP2 expression levels. These data establish that RanBP2 is a protein that acts to prevent chromosome missegregation and accumulation of aneuploid cells.

RanBP2 Insufficiency Induces Anaphase Bridges

To determine the nature of the chromosome missegregation defects underlying the aneuploidy observed in cells with low levels of RanBP2, the chromosome movements of RanBP2 mutant MEFs were followed by timelapse live microscopy. To visualize DNA, MEFs were transduced with a retrovirus expressing YFP-tagged histone 2B. We found that 9% of RanBP2^−/H and RanBP2^H/H cells failed to progress from metaphase to anaphase, a defect observed in only 2% of RanBP2^+/−, RanBP2^+/H, and wild-type MEFs (data not shown). Of the RanBP2^−/H and RanBP2^H/H cells that were able to progress into anaphase, 30% and 28% formed chromatin bridges, respectively (Figures 2A–2C). Most cells had one or two bridges, while a small fraction of cells had more than two (data not shown). These bridges resolved in telophase or during cytokinesis (Figure 2A). The percentage of chromatin bridges was much lower in RanBP2^+/− and RanBP2^+/H anaphases (16% and 12%, respectively) but still significantly higher than in RanBP2^+/+ anaphases (7%) (Figure 2C). In addition, a small but statistically significant increase of centrophilic chromosomes was observed in RanBP2^−/H and RanBP2^H/H MEFs in comparison to RanBP2^+/+ MEFs (Figure 2C). There was a small increase of lagging chromosomes in these MEFs that did not reach statistical significance (Figure 2C). Together, the above findings suggest that anaphase-bridge formation is the main chromosome segregation defect in RanBP2-insufficient cells.

To examine whether chromatin-bridge formation caused structural chromosome damage, spectral karyotype (SKY) analysis was performed on RanBP2^−/H and RanBP2^+/+ MEFs. Although this analysis confirmed that RanBP2^−/H cells had increased numerical chromosome instability, there was no evidence for increased breakage or fusion of chromosome arms (Figure S3A). Giemsa-stained metaphase spreads from splenocytes and MEFs with various levels of RanBP2 were also inspected for structural chromosomal abnormalities. Again, there was no evidence for increased chromosome breakage or end-to-end fusion in RanBP2 mutant cells, although a small but significant increase in the incidence of chromosome gaps was observed in RanBP2^H/H and RanBP2^−/H splenocytes and MEFs (Figures S3B–S3D). These data suggest that chromosome-bridge formation causes aneuploidy in the absence of extensive structural chromosome damage. However, the possibility that MEFs with structural damage have reduced viability and thus go unnoticed cannot be excluded.

Previous studies of HeLa cells have documented that depletion of RanBP2 by RNA interference perturbs mitotic spindle formation and loading of certain mitotic checkpoint proteins onto kinetochores in early mitosis (Joseph et al., 2004; Salina et al., 2003). However, no such mitotic defects were detectable in RanBP2 hypomorphic MEFs (Figure S4).

Topo IIα Targeting to Inner Centromeres in Mitosis is Dependent on RanBP2

Formation of chromatin bridges in anaphase is a distinctive feature of cells in which DNA decatenation is impaired by mutation or chemical inhibition of Topo IIα (Clarke et al., 1993). Topo IIα is active in S phase and at the metaphase-anaphase transition, where it disentangles sister centromeres to enable their separation in anaphase (Bhat et al., 1996). To examine whether ana-phase-bridge formation in RanBP2 hypomorphic MEFs might be due to Topo IIα insufficiency, we examined lysates from asynchronous MEF cultures of various RanBP2 genotypes for Topo IIα expression by immunoblotting. As shown in Figure 3A, Topo IIα expression levels were independent of RanBP2 status. RanBP2^+/+ and RanBP2^−/H MEFs that were arrested in mitosis by nocodazole also contained similar amounts of Topo IIα (Figure 3B).

Next, we hypothesized that RanBP2 deficiency might cause anaphase-bridge formation by interfering with proper localization of Topo IIα to inner centromeres of mitotic chromosomes. To test for this possibility, RanBP2^+/+ and RanBP2^−/H MEFs were arrested in prometaphase with monastrol, an inhibitor of the mitotic kinesin Eg5, and immunostained with anti-Topo IIα and anti-centromeric antibodies (ACA). We found that Topo IIα concentrated at inner centromeres in most RanBP2^+/+, RanBP2^+/H, and RanBP2^+/− cells (Figures 3C, 3D, and S5). In contrast, the majority of RanBP2^H/H and RanBP2^−/H MEFs failed to accumulate Topo IIα at inner centromeres. These results demonstrate that proper targeting of Topo IIα to inner centromeres of mitotic chromosomes is dependent on RanBP2 and suggest that anaphase-bridge formation in RanBP2 hypomorphic cells is caused by Topo IIα mislocalization.

RanBP2 Binds to Topo IIα in Mitosis

Because Topo IIα localization in mitosis is RanBP2 dependent, we suspected that the two proteins might form a complex. To investigate this, RanBP2 was immunoprecipitated from interphase and mitotic HeLa cell extracts and examined for Topo IIα coprecipitation. As shown in Figure 3E, about 4%–5% of the cellular Topo IIα pool coimmunoprecipitated with RanBP2 from mitotic extracts but not from G1 or G2 extracts. In the reverse experiment, RanBP2 immunoprecipitated with Topo IIα from mitotic extracts but again not from interphase extracts (data not shown). To determine whether the abundance of RanBP2-Topo IIα complexes in mitosis was affected by RanBP2 insufficiency, mitotic extracts from RanBP2^−/H and RanBP2^+/+ MEFs were subjected to coimmunoprecipitation with anti-RanBP2 antibodies. Although Topo IIα coprecipitated with RanBP2 from RanBP2^−/H cell extracts, the amount of protein coprecipitated from RanBP2^+/+ cell extracts was considerably greater (Figure 3F). These data suggest that RanBP2 and Topo IIα form a complex in mitosis and that RanBP2-insufficient cells have reduced amounts of this complex. RanBP2 does not colocalize with Topo IIα at inner centromeres (data not shown). Both RanBP2 and Topo IIα, however, are present in the mitotic cytosol, suggesting that interaction between them takes place in this subcellular compartment.

The RanBP2 SUMO E3 Ligase Domain Regulates Proper Chromosome Segregation and Targeting of Topo IIα to Inner Centromeres

In S. cerevisiae, SUMO modification of topoisomerase II (Top2p), which is catalyzed by the E3 ligases Siz1p and Siz2p, targets the protein to the pericentromeric regions of mitotic chromosomes (Takahashi et al., 2006). Studies in Xenopus egg extracts and human cells suggest that Topo IIα is also subjected to sumoylation in higher eukaryotes (Azuma et al., 2003, 2005; Mao et al., 2000). Since RanBP2 is a nucleoporin with SUMO E3 ligase activity (Pichler et al., 2002), we speculated that this activity might be required for accumulation of Topo IIα at inner centromeres and proper separation of sister chromatids. To test this hypothesis, we expressed influenza hemagglutinin (HA) epitope-tagged RanBP2(2553–2838), a 33 kDa fragment spanning the RanBP2 SUMO E3 ligase domain (Pichler et al., 2004) (Figure 4A), in RanBP2^−/H MEFs and measured its ability to correct ana-phase-bridge formation and Topo IIα localization. Three catalytic mutants of the SUMO E3 ligase domain, HA-RanBP2(2553–2838)^{L2651A,L2653A}, HA-RanBP2(2553–2838)^{F2657A,F2658A}, and HA-RanBP2(2553–2838)^{L2651A,L2653A,F2657A,F2658A}, and two regions flanking the SUMO E3 ligase domain, HA-RanBP2(1950–2553) and HA-RanBP2(2839–3224), were also expressed in RanBP2^−/H cells as controls (Figure 4A). Western blotting for HA revealed that all RanBP2 fragments were expressed at similar levels (Figure S6). Consistent with our earlier results, 28% of RanBP2^−/H MEFs expressing empty vector formed anaphase bridges (Figure 4B). A similarly high percentage of anaphase bridges was observed in RanBP2^−/H MEFs expressing the catalytic mutants or the flanking regions of the RanBP2 SUMO E3 ligase domain. By contrast, only 11% of RanBP2^−/H MEFs expressing the fully functional RanBP2 SUMO E3 ligase domain (2553–2838) formed anaphase bridges, a percentage that was similar to that recorded for RanBP2^+/+ cells expressing the empty vector. Ectopic expression of the HA-RanBP2(2553–2838) had no overt effect on global sumoylation patterns as measured by western blotting (data not shown). Importantly, overexpression of PIASy, a SUMO E3 ligase required for sumoylation of Topo IIα in Xenopus egg extracts (Azuma et al., 2003, 2005), did not correct anaphase-bridge formation (Figures 4B and S6).

Complementary immunostaining experiments with antibodies to Topo IIα and centromeres revealed that the expression of RanBP2 SUMO E3 ligase domain in RanBP2^−/H cells greatly improved the targeting of Topo IIα to inner centromeres of mitotic chromosomes (Figures 4C and 4D). The percentage of cells with diffuse Topo IIα localization was reduced to only 25% in RanBP2^−/H cells expressing RanBP2(2553–2838) compared to 50% in RanBP2^−/H cells expressing the empty vector. No such improvement was observed when catalytic mutants or regions flanking the RanBP2 SUMO E3 ligase domain or PIASy were overexpressed (Figures 4C and 4D). Together these data establish a specific requirement for RanBP2 SUMO E3 ligase activity in targeting Topo IIα protein to inner centromeres in early mitosis to allow for decatenation of sister centromeres prior to anaphase onset.

To investigate whether the SUMO E3 ligase domain might exert its corrective effects by targeting Topo IIα protein, we determined whether it formed a complex with Topo IIα in mitosis. To this end, we expressed HA-RanBP2(2553–2838) in RanBP2^+/+ MEFs and precipitated it with an antibody against HA from mitotic extracts. As shown in Figure 4E, Topo IIα indeed coimmunoprecipitated with HA-RanBP2(2553–2838) from these extracts. In contrast, Topo IIα failed to coimmunoprecipitate with HA-RanBP2(1950–2553) and HA-RanBP2(2839–3224). Thus, the SUMO E3 ligase domain of RanBP2 is sufficient for complex formation with Topo IIα in vivo.

RanBP2 Promotes Sumoylation of Topo IIα In Vitro and In Vivo

Next, we used an in vitro sumoylation assay to address whether Topo IIα is a substrate for SUMO modification by RanBP2. We incubated full-length purified human Topo IIα with recombinant SUMO E1, Ubc9, His-HA-RanBP2(2553–2838), SUMO1, and ATP for 1 hr. Topo IIα sumoylation was then analyzed by immunoblotting for Topo IIα. As shown in Figure 5A, Topo IIα was efficiently sumoylated by His-HA-RanBP2(2553–2838). No such modification occurred in the absence of His-HA-RanBP2(2553–2838) or ATP, or when catalytic mutant His-HA-RanBP2(2553–2838)^{L2651A,L2653A} was used instead of His-HA-RanBP2(2553–2838). We found that Topo IIα was also efficiently sumoylated by His-HA-RanBP2(2553–2838) when SUMO1 was replaced by SUMO2 or SUMO3 (Figure S7A). Collectively, these data demonstrate that Topo IIα is a substrate of RanBP2 for SUMO modification in vitro.

If RanBP2 indeed functions to sumoylate Topo IIα in mitosis, one would expect mitotic RanBP2^−/H MEFs to have reduced levels of SUMO-conjugated Topo IIα. To test whether this is the case, we stably expressed HA-tagged SUMO1 in RanBP2^+/+ and RanBP2^−/H MEFs (Figure 5B), harvested mitotic cells by mitotic shake-off, and analyzed lysates prepared from these cells for the presence of SUMO-conjugated Topo IIα by western blotting for Topo IIα. In addition to unmodified Topo IIα, RanBP2^+/+ MEFs possessed a higher-molecular-weight SUMO1-conjugated Topo IIα band (Figure 5B). The presence of HA-SUMO1 in the upper band was confirmed by anti-HA immunoblotting. SUMO1-conjugated Topo IIα was not detectable in RanBP2^−/H MEFs. Importantly, expression of HA-RanBP2(2553–2838), but not HA-RanBP2(2553–2838)^{L2651A,L2653A}, was able to restore sumoylation of Topo IIα in these MEFs. To determine whether Topo IIα sumoylation was impaired under physiological conditions, lysates from mitotic RanBP2^+/+ and RanBP2^−/H MEFs were analyzed for the presence of SUMO-Topo IIα with anti-Topo IIα antibody. As shown in Figure 5C, SUMO-modified Topo IIα was detectable in RanBP2^+/+ lysates but not in RanBP2^−/H lysates. As expected, SUMO-modified Topo IIα was not observed when RanBP2^+/+ MEFs were lysed in the absence of the isopeptidase inhibitor N-ethylmaleimide (NEM; see Figure S7B). The observation that RanBP2 hypomorphism leads to decreased sumoylation of Topo IIα in mitosis suggests that Topo IIα is an in vivo substrate of RanBP2 during this cell-cycle stage.

To further test whether Topo IIα is the relevant target of RanBP2 SUMO E3 ligase activity in mitosis, we expressed a Topo IIα construct encoding a nonhydrolyzable SUMO1 in linear N-terminal fusion with Topo IIα-EGFP in RanBP2^−/H MEFs and determined its ability to restore Topo IIα localization and efficient separation of sister chromatids in these cells. SUMO1-Topo IIα-EGFP fusion protein indeed targeted with high efficiency to inner centromeres of RanBP2 hypomorphic MEFs in mitosis (Figure 5D) and caused a dramatic reduction in ana-phase-bridge formation in these cells (Figure 5E). These data support the idea that the mitotic defects seen in RanBP2 hypomorphic cells are due to impaired SUMO modification of Topo IIα.

Depletion of PIASy in Xenopus egg extracts or HeLa cells has been shown to inhibit Topo IIα targeting to inner centromeres and block chromosome segregation (Azuma et al., 2003, 2005; Diaz-Martinez et al., 2006), suggesting that RanBP2 and PIASy E3 ligases might have overlapping functions in mitosis. To investigate this, we analyzed the subcellular localization of Topo IIα in MEFs derived from PIASy^−/− mice (Roth et al., 2004). As shown in Figure 5F, PIASy loss had no impact on Topo IIα targeting to inner centromeres in mitosis. MEFs from an independently generated PIASy^−/− mouse strain yielded the same results (Wong et al., 2004) (data not shown). Furthermore, anaphase-bridge frequencies and aneuploidy were not significantly elevated in PIASy^−/− MEFs (Figures 5G and S8). Western blot analysis of mitotic lysates from PIASy^+/+ and PIASy^−/− MEFs for Topo IIα revealed that SUMO modification of Topo IIα was unperturbed by the deletion of PIASy (Figure 5H). These data demonstrate that, at least in MEFs, RanBP2 rather than PIASy regulates accumulation of Topo IIα at inner centromeres and proper chromosome segregation.

RanBP2 Insufficient Mice Are Prone to Carcinogen-Induced and Spontaneous Tumors

As RanBP2 insufficiency leads to chromosome number instability, a condition that has been linked to cancer development, we sought to determine whether mice with reduced RanBP2 expression are prone to tumorigenesis. To this end, we performed a tumor bioassay with 7,12-dimethylbenz[a]anthracene (DMBA), a carcinogen that predisposes wild-type mice to lung tumors, skin tumors, and thymic lymphomas when applied to the skin (Serrano et al., 1996). Pups from RanBP2^H/H × RanBP2^−/H and RanBP2^+/− × RanBP2^+/H intercrosses were given a single application of 50 μl 0.5% DMBA to the dorsal surface on postnatal day 5. Treated animals were then monitored for development of overt tumors over a period of 5 months. The incidence of skin tumors was dramatically increased in RanBP2^H/H and RanBP2^−/H mice compared to RanBP2^+/+, RanBP2^+/H, and RanBP2^+/− mice (Figures 6A and 6B). Seventy-seven percent of RanBP2^−/H and 60% of RanBP2^H/H mice developed skin tumors compared to 31% of RanBP2^+/−, 24% of RanBP2^+/H, and 13% of wild-type mice. Moreover, the latency with which skin tumors formed was much shorter in RanBP2^H/H and RanBP2^−/H mice than in RanBP2^+/+, RanBP2^+/H, and RanBP2^+/− mice (Figure 6B). Compared with RanBP2^+/+ mice, the skin tumor burden was 16- and 23-fold increased in RanBP2^H/H and RanBP2^−/H mice, respectively (Figure 6C). In RanBP2^+/− and RanBP2^+/H mice this increase was only 3-fold. At 5 months of age, the animals were sacrificed and their internal organs were screened for tumors. Lung tumors were identified in 58% of RanBP2^+/+ mice (Figures 6D and 6E). Small increases in the incidence of lung tumorigenesis were observed in RanBP2^+/− and RanBP2^+/H mice, but they were not statistically significant. In contrast, the increases in lung tumor incidence were much larger in RanBP2^H/H and RanBP2^−/H animals, with 95% of RanBP2^H/H and 100% of RanBP2^−/H animals having this tumor type. Similarly, the lung tumor burden in RanBP2^H/H and RanBP2^−/H mice was considerably higher than in RanBP2^+/−, RanBP2^+/H, and RanBP2^+/+ mice, with RanBP2^H/H and RanBP2^−/H animals developing on average 17 and 14 tumors per mouse, respectively, compared to three or less in the other genotypes (Figure 6F).

Furthermore, to determine if RanBP2 insufficiency promotes spontaneous tumorigenesis, cohorts of RanBP2^+/+, RanBP2^H/H, and RanBP2^−/H mice were established and monitored biweekly for signs of overt tumors or ill health for up to 2 years. Moribund animals were sacrificed and screened for tumors. Tumors were collected and embedded in paraffin for histopathology. RanBP2^−/H and RanBP2^H/H mice formed tumors with increased incidence and shorter latency than RanBP2^+/+ mice (Figure 6G). Lung adenocarcinomas were the most prevalent tumor type in RanBP2^−/H and RanBP2^H/H mice (Figures 6H and 6I). Other frequently observed tumor types in these mice were hepatocellular carcinomas and sarcomas. Chromosome counts and interphase fluorescence in situ hybridization (FISH) using probes for chromosomes 4, 7, 9, and 12 demonstrated that tumors of all three types consistently displayed severe aneuploidy (Figures 6J and S9). Thus, when RanBP2 expression drops below a critical threshold level, mice become prone to spontaneous and carcinogen-induced tumorigenesis. Furthermore, these studies establish RanBP2 as a protein with tumor-suppressive activity and reveal a correlation between aneuploidy and tumorigenesis.

RanBP2 Sumoylates Topo IIα to Mediate Sister-Chromatid Separation

Siz1p and Siz2p, two proteins that provide the major SUMO (Smt3p) E3 ligase activity in budding yeast, execute SUMO modification of Top2p in mitosis (Takahashi et al., 2006). This modification seems to function as a signal for Top2p targeting to pericentromeric regions of mitotic chromosomes. Seven observations reported here suggest that, in mammals, RanBP2 catalyzes the SUMO modification of Topo IIα in mitosis to direct this protein to inner centromeres for accurate chromosome separation prior to anaphase onset (Figure 5I). First, catalytic inhibition of Topo IIα causes formation of chromatin bridges in anaphase due to incomplete DNA decatenation (Clarke et al., 1993). Ana-phase-bridge formation is the primary phenotype of RanBP2 hypomorphic cells. Second, the RanBP2 hypomorphic cells fail to accumulate Topo IIα at inner centromeres of mitotic chromosomes at high incidence. Third, ectopic expression of the RanBP2 SUMO E3 ligase domain in RanBP2 hypomorphic cells by itself is sufficient to restore proper localization of Topo IIα to inner centromeres and prevent anaphase-bridge formation. These corrective effects require SUMO conjugation, as they were not observed in RanBP2 hypomorphic cells expressing catalytic mutants of the RanBP2 E3 ligase domain. Unlike Topo IIα, the RanBP2 SUMO E3 ligase domain and full-length RanBP2 do not concentrate at inner centromeres of mitotic chromosomes, suggesting that Topo IIα is sumoylated by RanBP2 in the mitotic cytosol and, from there, translocates to inner centromeres of duplicated chromosomes without RanBP2 (Figure 5I). Fourth, RanBP2 forms a complex with Topo IIα specifically in mitosis. Importantly, the abundance of RanBP2-Topo IIα complexes is substantially reduced in RanBP2 hypomorphic cells. Fifth, the RanBP2 SUMO E3 ligase domain, which is sufficient for Topo IIα binding, efficiently catalyzes SUMO conjugation of Topo IIα in an in vitro assay using recombinant proteins. We show that Topo IIα can be modified by all three SUMO species in this assay. Sixth, RanBP2 hypomorphic cells have substantially lower levels of SUMO-modified Topo IIα than cells expressing a full complement of RanBP2. Furthermore, ectopic expression of the RanBP2 SUMO E3 ligase domain in RanBP2 hypomorphic cells is sufficient to restore SUMO conjugation of Topo IIα in mitosis. Seventh, ectopically expressed SUMO1-Topo IIα fusion protein corrects both Topo IIα mislocalization and anaphase-bridge formation in RanBP2 hypomorphic MEFs. As expression of the catalytically active SUMO E3 ligase domain has similar corrective effects in these cells, it is reasonable to conclude that Topo IIα is the relevant substrate of RanBP2 in mitosis.

Previous studies in Xenopus egg extracts have implicated PIASy in the sumoylation of Topo IIα (Azuma et al., 2003, 2005). This combined with the observation that depletion of PIASy from these extracts results in metaphase arrest led to speculation that sumoylation of Topo IIα is essential for sister-chromatid separation in mitosis. The observation that siRNA-mediated depletion of PIASy from HeLa cells interferes with targeting of Topo IIα and blocks separation of duplicated chromosomes further supported this idea (Diaz-Martinez et al., 2006). However, here we demonstrate that Topo IIα sumoylation, Topo IIα accumulation at inner centromeres, and proper separation of sister chromosomes are unperturbed in PIASy^−/− MEFs. Consistent with this, PIASy^−/− mice are viable, fertile, and devoid of any overt phenotypes (Roth et al., 2004; Wong et al., 2004). Furthermore, PIASy overexpression was found to be unable to correct Topo IIα mislocalization and chromosome missegregation in RanBP2 hypomorphic MEFs. One interpretation is that the role of RanBP2 and PIASy in Topo IIα sumoylation is species, cell type, and/or transformation status dependent. It is also possible that off-target effects of siRNA oligos drive aberrant Topo IIα localization and chromosome segregation in the PIASy-depleted HeLa cells. Furthermore, it is unknown whether depletion of PIASy in HeLa cells correlates with impaired SUMO modification of Topo IIα.

RanBP2 in Chromosomal Instability and Tumorigenesis

Our finding that RanBP2 hypomorphic mice are highly susceptible to carcinogen-induced tumors and prone to various spontaneous tumors reveals a novel and important role for RanBP2 in suppression of tumorigenesis. What could be the tumor-suppressive function(s) of RanBP2? Given that RanBP2 hypomorphism induces aneuploidy and that the incidence of aneuploidy is high in human cancers, it is tempting to speculate that RanBP2 exerts its tumor-suppressive effect by ensuring accurate segregation of duplicated chromosomes in mitosis. The finding that aneuploidy is high in tumors from RanBP2 hypomorphic mice supports this notion. Yet, whether aneuploidy is causally implicated in cancer development has been a subject of intense investigation and debate. Recent studies of mutant mouse strains with defects in mitotic checkpoint genes that increase the incidence of missegregation of whole chromosomes suggest that aneuploidy indeed has the ability to promote tumorigenesis (Weaver and Cleveland, 2006). Structural chromosome defects have also been implicated in tumor development. Despite having high rates of chromatin-bridge formation, RanBP2 hypomorphic MEFs showed no evidence of increased gross chromosomal aberrations. However, this does not exclude the possibility that structural chromosome damage from improper decatenation contributes to tumor development in RanBP2 mutant mice. For instance, it is possible that transformed cells may be more resistant to cell death induced by structural damage than nontransformed cells.

Although RanBP2 is a well-established component of the mammalian NPC, cells that are hypomorphic for this protein displayed no overt defects in NLS-dependent protein import, NES-dependent protein export, or mRNA export. These findings suggest that tumor development in RanBP2 hypomorphic mice is unlikely due to impairments in some of the major transport pathways. However, it cannot be excluded that specific NLS- or NES-containing proteins that are important for controlled cell proliferation or induction of apoptosis are expressed at inappropriate levels or unable to reach their intracellular sites of action in RanBP2 hypomorphic mice. Moreover, only those transport pathways that were functionally impaired by RanBP2 depletion in HeLa cells were included in our analysis (Bernad et al., 2004; Hutten and Kehlenbach, 2006). Thus, it is possible that certain transport pathways are not optimally functioning in RanBP2 hypomorphic cells.

To explore whether RanBP2 might have a role in human cancer, we compared the relative expression of RanBP2 in normal tissue versus tumors using human gene expression data from the Oncomine database (http://www.oncomine.com). RanBP2 hypomorphic mice are particularly sensitive to spontaneous and carcinogen-induced lung tumors. Consistent with these data, two independent studies revealed that RanBP2 transcript levels are substantially reduced in human non-small cell lung cancers (Beer et al., 2002; Garber et al., 2001). These findings, along with data showing that RanBP2 expression is frequently downregulated in human lung cancer cell lines and primary lung tumor samples (D. Baker and J.M.v.D., unpublished data), suggest that RanBP2 downregulation is a frequent event in human lung tumorigenesis.

Generation of Hypomorphic and Knockout Mice and Tumor Susceptibility Studies

The gene targeting procedure used to generate the hypomorphic RanBP2 allele (H) was as previously described (Dawlaty and van Deursen, 2006). Correctly targeted ES cell clones were injected into blastocysts and RanBP2^+/H offspring were obtained from the resulting chimeras through standard procedures. RanBP2^+/− mice were established by Cre-mediated excision of RanBP2 exon 3 in the female germline by the use of MMTV-Cre transgenic mice (this excision causes out-of-frame fusion of exons 2 and 4). All mice were maintained on a 129Sv/E × C57BL/6 genetic background. Mice in tumor susceptibility studies were monitored daily. Moribund mice were killed and their major organs screened for overt tumors. Tumors were processed for histopathology by standard procedures. Prism software was used for the generation of tumor-free survival curves and for statistical analyses. DMBA treatment was as described (Serrano et al., 1996).

Western Blotting and Coimmunoprecipitation

Western blot analyses and coimmunoprecipitations were essentially done as described (Kasper et al., 1999). Lysis buffer for detection of SUMO-modified Topo IIα by western blotting consisted of 50 mM Tris HCl (pH 7.5), 150 mM NaCl, 0.1% NP40, 0.5% SDS, protease inhibitor cocktail (Roche, Cat. #11-836-170-001), and 20 mM NEM. Samples were kept on ice for 15 min and vortexed at ~2 min intervals and then boiled in Laemmli buffer. Antibodies were rabbit anti-hRanBP2(2500–3224) and rabbit anti-hRanBP2(2550–2837) (Joseph et al., 2004); rabbit anti-Topo IIα (Topogen); mouse anti-β-actin (AC-151, Sigma); rat anti-HA (3F10; Roche); and mouse anti-phosphohistone H3 (Ser10) (Upstate).

Indirect Immunofluorescence and Live-Cell Imaging

MEFs were cultured on 10-well glass slides for 24 hr, arrested in prometaphase with 100 μM monastrol for 3 hr, and fixed with 3% paraformaldehyde in PBS for 12 min. Antibody incubations were as described (Kasper et al., 1999). Antibodies were rabbit anti-Topo IIα (Topogen); human anti-centromeric antibody (Antibodies Incorporated); rabbit anti-Mad1 (Dr. T. Yen); and rabbit anti-Cenp-E (Dr. D. Cleveland). Live-cell imaging was performed as described (Jeganathan et al., 2005).

In Vitro Sumoylation Assays

Sumoylation reactions were carried out in a total volume of 20 ml. Reactions contained 50 mM HEPES (pH 8), 100 mM NaCl, 1 mM DTT, 1 mM Mg-ATP, 50 nM E1 (110 ng), 25 μM SUMO1, 2, or 3 (5 μg), 125 nM Ubc9 (50 ng), 100 ng recombinant His-HA-RanBP2(2553–2838) or His-HA-RanBP2(2553–2838)^{L2651A,L2653A}, and 20 ng of purified full-length Topo IIα (Topogen). Reactions were performed at 37°C for 1 hr, stopped with equal volume of 2×Laemmli buffer, resolved by SDS-PAGE on 5% Tris-HCl gels, and analyzed by immunoblotting for Topo IIα. Recombinant E1, Ubc9, and SUMO1–3 were from Boston Biochem (SUMO1 conjugation kit; Cat. #K-710). To produce recombinant RanBP2 His-HA-RanBP2(2553–2838) and His-HA-RanBP2(2553–2838)^{L2651A,L2653A}, their corresponding cDNA constructs were cloned into BamH1 and Xho1 sites of pET28a(+) (Novagen). Proteins tagged with His were expressed in BL21 (DE3) bacteria at 15°C and purified from bacterial lysates with Ni²⁺ agarose.