Sumoylation of Topoisomerase I Is Involved in Its Partitioning between Nucleoli and Nucleoplasm and Its Clearing from Nucleoli in Response to Camptothecin*

Previous studies identified a small fraction of putatively sumoylated topoisomerase I (TOP1) under basal conditions (∼1%), and anticancer camptothecins that trap the TOP1-DNA covalent intermediate markedly increase the sumoylation of TOP1 (≤10%). To study the role of the sumoylation of TOP1, we mutated sites on green fluorescent protein (GFP)-TOP1 corresponding to the consensus sequence for protein sumoylation (ΨKXE, where Ψ is a hydrophobic residue) and assayed the mutants for basal and camptothecin-induced sumoylation. Only one of the eight mutants, K117R, located in the highly charged NH2-terminal region, showed a substantial reduction (∼5-fold) in basal and camptothecin-induced sumoylation; thus, Lys-117 appears to be the major sumoylation site. A triple mutant having the ΨKXE sequences flanking K117R additionally mutated (K103R/K117R/K153R) showed little if any sumoylation, but was degraded like wild-type GFP-TOP1 during camptothecin treatment. However, K103R/K117R/K153R-GFP-TOP1 was markedly concentrated within nucleoli, depleted from the remainder of nucleus, and failed to be cleared from nucleoli in response to camptothecin treatment. These data are consistent with a model wherein basal transient sumoylation of the NH2-terminal, highly charged, disordered region prevents TOP1 binding to sites in nucleoli, thus driving it to bind in the nucleoplasm; and camptothecin treatment, which increases TOP1 sumoylation, further shifts the binding resulting in delocalization of TOP1 from nucleoli to nucleoplasm.

Topoisomerase I (TOP1) 1 is the molecular target of the camptothecin (CPT) class of cancer chemotherapeutics. CPT stabilizes the covalent TOP1-DNA intermediate by inhibiting the religation step of the DNA cleavage-religation mechanism. Previous studies have demonstrated that treatment of cells with CPT causes TOP1 to be hypersumoylated, ubiquitinated, and destroyed in proportion to the quantity of the covalent TOP1-DNA intermediate (1)(2)(3). TOP1 is also known to delocalize from nucleoli following CPT treatment (4 -6), which has been suggested to involve sumoylation (7). TOP1 sumoylation increases rapidly after CPT treatment and has been hypothesized to be related to the ubiquitination, proteasome-dependent degradation, or the delocalization of TOP1 from nucleoli (1)(2)(3)7).
Sumoylation has been shown to alter the cellular distribution, activity, or stability of a number of proteins (8 -16). In the case of TOP1, a putative low steady state level of sumoylation of TOP1 (Ͻ1%) has been observed in the absence of CPT treatment (1)(2)(3), which suggests that sumoylation of TOP1 may be important for regulating its activity, localization, or abundance under basal conditions.
To explore the role of TOP1 sumoylation under basal conditions and in response to camptothecins, we used site-directed mutagenesis of sumoylation consensus sequences to identify sumoylation sites on TOP1. The major sumoylation site was identified to be Lys-117 in the amino-terminal region, as only this mutant (K117R) showed a reduction in both basal and CPT-induced sumoylation. A triple mutant with two sumoylation site mutations besides K117R (K103R/K117R/K153R) showed an even greater reduction in CPT-induced sumoylation. This triple mutant almost totally lacks sumoylation, although it was degraded similarly to wild-type GFP-TOP1 (WT-GFP-TOP1). Surprisingly, a striking difference in the localization of K103R/K117R/K153R-GFP-TOP1 was observed in comparison to WT-GFP-TOP1; K103R/K117R/K153R-GFP-TOP1 was much more highly concentrated in nucleoli, was less abundant in the remainder of the nucleus, and failed to delocalize from nucleoli upon CPT treatment.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The plasmid for expression of GFP-TOP1, pTI-2, was constructed as previously described (17). Lys to Arg mutations of potential sumoylation sites (KXE) on pTI-2 (called GFP-TOP1 from here on) were created using the QuikChange mutagenesis kit (Stratagene) with the primer pairs shown in Table I Band Depletion Assay-Cells treated with CPT (25 M) for 20 min at 37°C in suspension were pelleted and immediately lysed in SDS-PAGE sample buffer to trap TOP1 in covalent linkage with DNA. After boiling for 10 min, 2 ϫ 10 5 cells/lane were subjected to SDS-PAGE (5% polyacrylamide), and Western blotting was carried out as previously described (2). Nitrocellulose membranes were blocked for 1 h in 5% milk and then incubated with GFP antibodies (1:2,000, Abcam, ab290-50) for 3 h to overnight in 1% milk in Tris-buffered saline containing 0.1% Tween 20. The membranes were then incubated with horseradish peroxidase-conjugated secondary antibodies (1:5,000, Amersham Biosciences) for 1 h at room temperature. Chemiluminescence was detected using either SuperSignal ® or Dura ® (Pierce) according to the protocol from the manufacturer.
Assay of CPT-induced TOP1 Degradation-Cells were treated and processed as for the band depletion assay except that, following a 3-h CPT treatment at 37°C in complete medium, the cells were washed free of CPT and re-incubated for 30 min at 37°C to allow cleavable complexes to reverse before cell lysis with 1ϫ SDS sample buffer. Cell lysates (ϳ1 ϫ 10 5 cells/lane) were then subjected to SDS-PAGE and Western blotting using GFP antibodies to detect the 130-kDa GFP-TOP1 protein as described above. Duplicate membranes were probed with TOP1 antibodies to detect the 100-kDa endogenous TOP1.
Fluorescence Microscopy of Wild-type and Sumoylation Site Mutants-HeLa cells were transiently transfected with WT or mutant GFP-TOP1. Sixteen hours later, cells were either treated with the solvent control (Me 2 SO, 1%) or with 25 M CPT for 1 h. Cells were washed three times with phosphate-buffered saline, fixed with 4% paraformaldehyde or methanol:acetic acid (1:1) for 10 min at room temperature, washed, and mounted with Fluoromount-G (Southern Biotechnology Associates, Inc.) containing 0.5 g/ml 4Ј,6-diamidine-2phenylindole. Images were captured using an Olympus IX70 inverted fluorescence microscope with a Princeton Instrument cooled CCD camera integrated with Scanalytics IP-Lab software.

TOP1 with Lys 3 Arg Mutations of Sumoylation Consensus
Sequences (KXE) Are Trapped by CPT in the TOP1-DNA Covalent Complex Similar to Wild-type TOP1-The consensus site for SUMO conjugation is a lysine residue in a sequence KXE. The residue preceding the lysine is predominantly Ile, Leu, or Val, although Ala, Phe, or Pro have also been found in this position (Table II). TOP1 contains 18 KXE sites. Eight of 18 of these KXE sequences were mutated (Lys 3 Arg) on a GFP-TOP1 plasmid (Fig. 1). Additionally, because we anticipated that multiple KXE mutations might be required to visualize reduced sumoylation resulting from mutations of sumoylation consensus sequences, we created double, triple, quadruple, and quintuple Lys 3 Arg mutant GFP-TOP1 (see "Experimental Procedures").
These KXE mutants were transfected into HEK293 cells to determine whether they retained TOP1 activity by determining their ability to form the TOP1-DNA covalent complex using the band depletion assay. In this assay, SDS lysis of CPT-treated cells causes TOP1 to be denatured in covalent linkage with DNA, decreasing its electrophoretic mobility, which results in depletion of TOP1 from its normal position on the Western blot. In the band depletion assay, all the transfected GFP-TOP1 mutants were indistinguishable from WT-GFP-TOP1 in their ability to form the TOP1-DNA complex ( Fig. 2A, detected with GFP antibodies; Fig. 2B, detected with TOP1 antibodies; results are shown for only K117R and K103R/K117R/K153R). Furthermore, the band depletion of wild-type and mutant GFP-TOP1 were similar to endogenous TOP1 (Fig. 2B).
Reduced Sumoylation of K117R and K103R/K117R/K153R-Basal and CPT-induced sumoylation were assessed on the various Lys 3 Arg GFP-TOP1 mutants in transfected HEK293 cells. Following lysis of cells in the presence of CPT, Western CCATATTCTTTCAGTAATTTTTCATTCTCCTCTcgGATTTTCAGTTTCTCTTCCcgGCTCATCTGTTTCC KIKEENE a Each lysine to arginine mutation as well as the wild-type amino acid sequence is shown. b Primers are oriented 5Ј to 3Ј and nucleotide changes are shown in lowercase. c A single primer pair was used to create the double mutation.
blotting was used to detect the pattern of GFP-TOP1 sumoylation as previously reported (2,3,7). Seven of the eight GFP-TOP1 mutants showed a sumoylation pattern that was indistinguishable from WT-GFP-TOP1 (data not shown). One mutant, K117R, showed an ϳ5-fold reduction in sumoylation when assayed by two different methods. In the first method, endonuclease digestion was used to remove DNA linked to TOP1 prior to SDS-PAGE and Western blotting with GFP antibody (Fig 3A). By this method, WT-GFP-TOP1 shows an immunoreactive band above it having an M r that corresponds to the mobility shift of a single SUMO conjugate, as previously demonstrated (2,3,7). This putative singly sumoylated GFP-TOP1 is present without CPT treatment (basal conditions) and is dramatically induced by CPT treatment (2,3). In the K117R mutant, both the basal and the CPT-induced mono-sumoylation are significantly reduced (Fig. 3A, top panel). The triple mutant with additional Lys 3 Arg changes of the KXE sites flanking Lys-117 (K103R/K117R/K153R) is almost devoid of sumoylated forms (Fig. 3A, top panel). A longer exposure of this Western blot (Fig. 3A, lower panel) results in saturation of the signal of the first SUMO conjugate of WT-GFP-TOP1, and reveals multi-SUMO conjugates of WT-GFP-TOP1 and K117R-GFP-TOP1. The longer exposure also reveals a doublet of closely migrating K103R/K117R/K153R-GFP-TOP1 bands that appear to be residual mono-sumoylated forms.
Only mutants containing the K117R mutation showed a reduction in sumoylation. Mutants without K117R, including those having multiple Lys 3 Arg mutations, did not show a detectable reduction in sumoylation. This result suggests that Lys-117 is the major site of sumoylation on TOP1.
Sumoylation of GFP-TOP1 was also assessed without nuclease digestion before SDS-PAGE and Western blotting. With this procedure, DNA remains covalently linked to TOP1 causing it to smear upward ( Fig. 3B (this panel is an overexposure of Fig. 2A)). Against the background smear, discrete GFPimmunoreactive bands are evident above WT-GFP-TOP1 from CPT-treated cells. Two of these bands are evidently sumoylated forms as they are reduced or not found on K117R-or K103R/ K117R/K153R-GFP-TOP1 (Fig. 3B, labeled GFP-TOP1-S). The absence of DNA linked to these sumoylated GFP-TOP1 species suggests that they were not covalently linked to DNA at the time of cell lysis.
Other bands of WT-, K117R-, and K103R/K117R/K153R -GFP-TOP1 are visible that have an intermediate size between the non-sumoylated and sumoylated bands. These bands are absent from samples treated with endonuclease, possibly because they may have been proteolyzed in the lysates during the endonuclease treatment (compare CPT-treated lanes of wildtype and mutant GFP-TOP1 in Fig. 3, A and B).
To confirm the identity of the sumoylated GFP-TOP1, lysates of cells transfected with WT-GFP-TOP1 and untransfected cells were probed with antibodies to TOP1, GFP, or SUMO-1 (Fig. 3C). Probing of a replicate membrane with TOP1 antiserum shows approximately equal amounts of endogenous TOP1 and WT-GFP-TOP1, and their putative sumoylated forms are more intense with CPT treatment. Probing of another replicate membrane with GFP antibodies confirms WT-GFP-TOP1 and its putative sumoylated forms. Probing of a third replicate membrane with SUMO-1 antibodies confirms the identity of the sumoylated forms of endogenous TOP1 and WT-GFP-TOP1.
Lack of an Effect of Lys 3 Arg Mutations on CPT-induced Degradation-CPT treatment was previously shown to cause a rapid increase in the sumoylation and the degradation of TOP1 that were both prevented at the restrictive temperature for the thermolabile ubiquitin-activating enzyme in ts85 cells (3). CPT has also been shown to induce ubiquitination of TOP1 and degradation of TOP1 that is blocked by proteasome inhibitors (2, 3). These results and recent findings of apparent co-regulation of other proteins by ubiquitination and sumoylation (10, 18) prompted us to study the relationship between the CPT-  induced sumoylation of TOP1 and its proteasome-dependent degradation. Plasmids expressing WT-, K117R-, or K103R/ K117R/K153R-GFP-TOP1 were transfected into three cell lines: HEK293 and CHO, which efficiently degrade endogenous TOP1 during CPT treatment; and a colon cancer line, KM12, which degrades endogenous TOP1 poorly or not at all during CPT treatment. 2 The degradation of WT-GFP-TOP1 and the sumoylation-deficient mutant K103R/K117R/K153R-GFP-TOP1 were compared in at least three independent experiments per cell line (Fig. 4 shows representative results, and Table III shows the statistical analysis of the collective experiments). In HEK293 and CHO cells, more than half of both WTand K103R/K117R/K153R-GFP-TOP1 was degraded during 3 h of CPT treatment, which was similar to the amount of their endogenous TOP1 that was degraded. Thus, K103R/K117R/ K153R-GFP-TOP1, which nearly lacks sumoylation, was reproducibly degraded like WT-GFP-TOP1 in these two cell lines. In KM12 cells, endogenous TOP1 was not degraded in response to CPT treatment (Fig. 4C) and, consistently, neither was WTnor K103R/K117R/K153R-GFP-TOP1. Overall, the degradation of the transfected GFP-TOP1 species mimicked the degradation of the endogenous TOP1, and the degradation of the transfected GFP-TOP1 was unaffected by the sumoylation state. (4 -6, 19). We determined the localization of endogenous TOP1 by immunostaining of cells that were fixed with paraformaldehyde or methanol:acetic acid (data not shown). In cells fixed with paraformaldehyde, endogenous TOP1 localized diffusely throughout the nucleus and nucleoli with little evidence of enrichment in nucleoli in the majority of cells. In cells fixed with methanol:acetic acid, TOP1 was evidently extracted more from nucleoplasm than from nucleoli because it was highly concentrated in nucleoli relative to nucleoplasm in nearly all cells. WT-GFP-TOP1 showed a similar strong nucleolar concentration in nearly all cells fixed with methanol:acetic acid, but not in cells fixed with paraformaldehyde. We used paraformaldehyde fixation because it did not significantly alter the distribution of WT-GFP-TOP1 or K103R/K117R/K153R-GFP-TOP1 from that seen in live cell images (data not shown). We found three classes of cells with regard to the localization of WT-GFP-TOP1: 1) diffuse nuclear (62%), 2) nucleolar or pe-2 P. D'Arpa, unpublished data.

FIG. 3. Basal and CPT-induced sumoylation of WT and Lys 3 Arg mutants of GFP-TOP1. Basal (Ϫ) and CPT-induced (ϩ) sumoylation of WT-, K117R-, or K103R/K117R/K153R-GFP-TOP1 in transfected HEK293 cells were assayed by Western blot analysis with GFP-antibodies. A,
DNase was used to remove covalently linked DNA from TOP1 prior to SDS-PAGE and Western blotting. The top panel is a short exposure of the membrane and shows GFP-TOP1 and mono-sumoylated GFP-TOP1 (short exposure). A longer exposure of the same membrane reveals additional sumoylated species of GFP-TOP1 of lower abundance (long exposure). Treatments and sample processing were performed in duplicate and quadruplicate for untreated (Ϫ) and CPT-treated samples (ϩ), respectively. B, cell lysates were not treated with nuclease before SDS-PAGE and Western blotting. After this procedure, much of the TOP1 is covalently linked to DNA and is present as a smear. However, conjugates of WT-GFP-TOP1 are evident, and some of these are missing from K117R-and K103R/K117R/K153R-GFP-TOP1 (GFP-TOP1-S1 and -S2). Furthermore, the WT-GFP-TOP1 as well as both of the mutant GFP-TOP1 show additional bands that are dependent upon CPT treatment and are roughly intermediate in size between GFP-TOP1 and GFP-TOP1-S1 (labeled GFP-TOP1-?). C, confirmation of the identity of sumoylated GFP-TOP1 on Western blots probed with antibodies to SUMO-1, TOP1, or GFP. Cells transfected with WT-GFP-TOP1 (WT) and untransfected cells (Ϫ) were treated with the solvent control (Ϫ) or CPT (ϩ) and were processed for Western blotting as in A. CPT-induced bands recognized by SUMO-1 antibody correspond to those recognized by TOP1 and GFP antibodies. The experiments were repeated multiple times, and a representative experiment is shown. rinucleolar enrichment (9%), or 3) nucleolar depleted (29%). As described previously, CPT treatment causes TOP1 to be cleared from nucleoli (4 -7). Consistent with this, we observed a dramatic clearing of WT-GFP-TOP1 from nucleoli during CPT treatments of 5-60 min (Fig. 5 shows nucleolar clearing after a 60-min CPT treatment).
As compared with WT-GFP-TOP1, K103R/K117R/K153R-GFP-TOP1 was strongly concentrated within nucleoli in nearly all cells and its intensity was significantly reduced in the remainder of the nucleus. In response to CPT treatment, unlike WT-GFP-TOP1, K103R/K117R/K153R-GFP-TOP1 was not cleared from nucleoli. Thus, as compared with WT-GFP-TOP1, K103R/K117R/K153R-GFP-TOP1 was more concentrated within nucleoli and relatively less concentrated within the remainder of the nucleus, and in response to CPT treatment, it was not cleared from nucleoli. DISCUSSION Sumoylation of TOP1 and TOP2 increases within minutes after treatment of cells with drugs that specifically target the respective enzymes (2, 20 -22). Basal sumoylation of endogenous TOP1 (i.e. without CPT treatment) has been observed in all the cell types we have analyzed and mainly consists of singly sumoylated TOP1, although a lesser abundance of dou-bly and triply sumoylated TOP1 is sometimes visible (e.g. see Fig. 3A, compare WT (Ϫ) and (ϩ) CPT; only mono-and disumoylated TOP1 are visible without CPT treatment). CPT treatment induces the sumoylation of TOP1 manyfold, and we and others have observed TOP1 with as many as six SUMO conjugates (2,21).
For both basal and CPT-induced sumoylation, the monomer is the most abundant, the dimer is less abundant, and so on with the abundance of multiply sumoylated forms being inversely proportional to the number of sumoylations (Fig. 3 and see Refs. 2 and 21). Here we have shown that within the highly charged amino-terminal region of TOP1, a conservative mutation of Lys-117 to Arg prevented the majority of both basal and CPT-induced sumoylation. However, residual sumoylations were observed on K117R-GFP-TOP1. To identify the sites of the residual sumoylations, the lysines of KXE sites flanking Lys-117 were additionally mutated. These double mutants, K103R/K117R-GFP-TOP1 and K117R/K153R-GFP-TOP1, exhibited significant residual sumoylation, and their partitioning between nucleoli and nucleoplasm was similar to WT-GFP-TOP1 and K117R-GFP-TOP1; they were also cleared from nucleoli during CPT treatment like WT-GFP-TOP1 and K117R-GFP-TOP1 (data not shown). Only the triple mutant, K103R/K117R/K153R-GFP-TOP1, which showed very little sumoylation, was found to be concentrated in nucleoli and depleted from nucleoplasm. None of the single mutants, except K117R, nor any of the double mutants having wild-type Lys-117 showed a detectable reduction in sumoylation. These results suggest that Lys-117 is the major sumoylation site on TOP1, and that the sumoylation of the sites flanking Lys-117 is sufficient to enable TOP1 to be cleared from nucleoli in response to CPT and to prevent the accumulation of TOP1 in nucleoli under basal conditions. Mutation of only a single lysine (Lys-117) eliminated the majority of sumoylated TOP1 despite the fact that up to six SUMO molecules have been seen attached to TOP1 (2,20). These observations can be explained by more than one model. First, the KXE sites may be differentially favored for sumoylation and de-sumoylation. If Lys-117 is the most favored sumoylation site with respect to its steady state sumoylation, it could account for the majority of mono-sumoylated TOP1. One of the sumoylation sites flanking Lys-117, either Lys-103 or Lys-153 may be the next favored; thus, sumoylation of Lys-117 plus Lys-103 or sumoylation of Lys-117 plus Lys-153 may account for the majority of di-sumoylated TOP1. Because monosumoylated TOP1 is severalfold more abundant than di-sumoylated TOP1, reduced sumoylation may only be observable when Lys-117 is mutated. This model of multiple mono-sumoylations predicts that the abundance of mono-sumoylated K117R-GFP-TOP1 might be roughly similar to di-sumoylated WT-GFP-TOP1. Our data are roughly consistent with this model (see Fig. 3, A and B). Our data do not exclude the possibility of a SUMO chain attached at Lys-117. SUMO chains consisting of SUMO-2/3 or SUMO-1 have been observed in vitro (23,24), and a low abundance of apparently di-sumoylated HDAC4 has been observed in cells co-transfected with HDAC4 and SUMO-2 (23). For consistency with the chain model, it would be necessary to postulate that a chain forms on a less favorable lysine when Lys-117 is mutated to explain the multiple sumoylations still visible on K117R-GFP-TOP1. Currently, the multiple monosumoylation model appears to be most consistent with our experimental data and with the fact that single SUMO conjugates on four closely spaced (IVL)KXE sites have been mapped on Cdc3 (25). Notably, using antibodies specific to SUMO-1 or  SUMO-2/3, we and others have observed SUMO-2/3 and SUMO-1 conjugated to TOP1 (20,21). 3 The close juxtaposition of multiple sumoylation sites on TOP1 (Lys-103, Lys-117, Lys-153) is similar to that found on the Saccharomyces cerevisiae septins Cdc3 (Lys-4, Lys-11 , Lys-30, Lys-63) and Shs/Sep7 (Lys-426, Lys-437) (25), the cytomegalovirus protein IE2 (Lys-175, Lys-180) (26), HDAC1 (Lys-444, Lys-476) (27), and GRIP1 (Lys-731, Lys-788) (28) (see Table I).
Multiple closely spaced sumoylation sites may be a redundancy mechanism for ensuring sumoylation. In the case of TOP1, the closely spaced sumoylation sites may ensure sumoylation of the amino terminus, which is largely disordered (29) and in cells may dynamically switch among conformational forms.
Sumoylation and ubiquitination sites in close proximity or involving the same lysine have been previously reported and suggested to serve a co-regulatory function. For example, sumoylation of the ubiquitin-protein isopeptide ligase Mdm2 appears to modulate its self-ubiquitination and destruction and thus the ubiquitination and proteasome-dependent degradation of p53 (10). In the case of IB␣, sumoylation may compete with ubiquitination for the same acceptor site lysine (18). These examples prompted us to test the hypothesis that the sumoylation of TOP1 might somehow regulate its proteasomedependent degradation. However, we did not detect a difference between the degradation of transfected WT-and K103R/ K117R/K153R-GFP-TOP1 ( Fig. 4 and Table III). We did find that the degradation of both WT-and K103R/K117R/K153R-GFP-TOP1 mimicked the degradation of the endogenous TOP1 in the cell lines tested. In conclusion, a near complete lack of sumoylation of TOP1 had no effect on the CPT-induced degradation of TOP1.
Sumoylation has been shown to alter the cellular distribution of several proteins (11-13, 15, 30). In every cell type we have tested (at least 10), we identified putative sumoylated forms of endogenous TOP1 under basal conditions (i.e. in the absence of CPT treatment), albeit at markedly lower abundance than after CPT treatment. We also observed basal sumoylation on WT-GFP-TOP1, but these forms were significantly reduced on K117R-GFP-TOP1 and K103R/K117R/ K153R-GFP-TOP1, indicating that a small fraction of TOP1 is normally sumoylated in the absence of CPT. Suggesting a role for basally sumoylated TOP1, WT-GFP-TOP1 and K103R/ K117R/K153R-GFP-TOP1 distribute differently within the nucleus. WT-GFP-TOP1 distributes to both nucleoplasm and nucleoli similar to endogenous TOP1 immunostaining; in contrast, the K103R/K117R/K153R-GFP-TOP1 concentrates in nucleoli and is depleted from the nucleoplasm (Fig. 5, A and B). Interestingly, the single mutant K117R-GFP-TOP1 and the double mutants K103R/K117R-GFP-TOP1 and K117R/K153R-GFP-TOP1 localized similarly to WT-GFP-TOP1, which may be the consequence of their residual sumoylation. This result suggests that basally sumoylated TOP1, whose steady state levels are only ϳ1%, plays a role in partitioning TOP1 between nucleoli and the nucleoplasm.
TOP1 is well known to be cleared from nucleoli after CPT treatment (4 -6). We found that TOP1 clears from nucleoli during CPT treatment in the presence of proteasome inhibitors, suggesting that relocalization to the nucleoplasm rather than degradation of nucleolar TOP1 is responsible for the clearing (data not shown). Recent studies have shown that CPTinduced nucleolar clearing of TOP1 correlated with changes in total protein sumoylation and TOP1 sumoylation, which were experimentally varied by heat shock or the expression of wildtype or dominant negative Ubc9 (7). Here we have shown that WT-GFP-TOP1 is cleared from nucleoli within 5 min of CPT treatment, and its distribution within the nucleoplasm remained unchanged up to the longest CPT treatment time (60 min). After CPT treatment, WT-GFP-TOP1 localized to perinucleolar structures as have been previously observed (31). In contrast, K103R/K117R/K153R-GFP-TOP1 failed to clear from the nucleoli even after 60 min of CPT treatment (Fig. 5, A and  B). K103R/K117R/K153R-GFP-TOP1 also failed to clear from nucleoli following treatment with 5,6-dichlorobenzimidazole riboside or actinomycin D, which caused endogenous TOP1 and WT-GFP-TOP1 to rapidly clear from nucleoli (data not shown). Thus, these results suggest that sumoylation of TOP1 is required for its clearing from nucleoli during exposure to CPT, 5,6-dichlorobenzimidazole riboside, or actinomycin D.
The major sumoylation sites were found on the amino-terminal region of TOP1. The amino-terminal region of TOP1 has 3 K. Hashimoto and P. D'Arpa, unpublished data. several notable features indicating an important role in regulating the cellular localization of TOP1. First, it contains the nuclear localization signals. Second, it contains the nucleolin binding site (32), and in Saccharomyces, the nucleolin ortholog, Nsr1p, physically interacts with yeast TOP1; and null mutants of NSR1 exhibited altered localization of TOP1 (33). Third, the NH 2 terminus of Drosophila TOP1 was sufficient for recruitment to active sites of transcription (34). Finally, the NH 2terminal region per se fused to GFP has been shown to concentrate in nucleoli and to clear from nucleoli in response to CPT (7,35) or ultraviolet light (35), which indicates that formation of the covalent TOP1-DNA intermediate is not required for nucleolar clearing. We have also found that CPT-induced clearing of WT-GFP-TOP1 was not changed after mutagenizing the active site tyrosine to phenylalanine (data not shown). Together these results suggest that the major reason TOP1 clears from nucleoli during CPT treatment is not a result of slower nucleoplasmic versus nucleolar mobility resulting from preferential nucleoplasmic formation of the TOP1-DNA covalent intermediate as recently suggested (31).
WT-GFP-TOP1 was recently shown to move continuously between nucleoli and nucleoplasm (31) similar to other nucleolar proteins (36 -38). Basic stretches of amino acids have been identified as elements that may mediate nucleolar retention of proteins through interaction with polyanions (39), and in the case of TOP1, the 120 amino acids centered about Lys-117 are 35% Lys and Arg (28% Asp and Glu). It is possible that SUMO conjugates may disrupt this highly charged region, effectively dislodging TOP1 from binding sites in nucleoli and enabling a large part to continuously partition into the nucleoplasm under basal conditions. CPT treatment may cause an equilibrium shift toward lesser nucleolar retention and greater nucleoplasmic localization because of its known effect of increasing TOP1 sumoylation.
Whether cancer cells can exploit the sumoylation-dependent partitioning of TOP1 to evade the cytotoxic effects of CPT requires further investigation.