SUMO-1 Conjugation to Human DNA Topoisomerase II Isozymes*

Topoisomerase I-mediated DNA damage induced by camptothecin has been shown to induce rapid small ubiquitin-related modifier (SUMO)-1 conjugation to topoisomerase I. In the current study, we show that topoisomerase II-mediated DNA damage induced by teniposide (VM-26) results in the formation of high molecular weight conjugates of both topoisomerase II a and II b isozymes in HeLa cells. Immunological characterization of these conjugates suggests that both topoisomerase II a and II b isozymes are conjugated to SUMO-1. The involvement of SUMO-1/UBC9 in the modification of topoisomerase II isozymes is also supported by the demonstration of physical interaction between topoisomerase II and SUMO-1/UBC9. Surprisingly, ICRF-193, which does not induce topoisomerase II-mediated DNA damage but traps topoisomerase II into a circular clamp conformation, is also shown to induce similar SUMO-1 conjugation to topoisomerase II isozymes. In addition, we show that both oxidative and heat shock stresses, which can cause protein damage, rapidly increase nuclear SUMO-1 conjugates. These studies raise the ques-tion on whether SUMO-1 conjugation to topoisomerases is an indirect result of a DNA damage response or a direct result because of protein conformational changes. of Wuerzburg, Germany). Rabbit antisera against hTOP2 ab were obtained from Dr. Jaulang Hwang (Academia Sinica, Taiwan). Staphylococcal S7 nuclease was purchase from Roche Molecular Biochemicals. Cloning of hTOP2 b cDNA— The full-length of hTOP2 b cDNA was cloned from human U937 cells by reverse transcriptase-polymerase

Topoisomerase I-mediated DNA damage induced by camptothecin has been shown to induce rapid small ubiquitin-related modifier (SUMO)-1 conjugation to topoisomerase I. In the current study, we show that topoisomerase II-mediated DNA damage induced by teniposide (VM-26) results in the formation of high molecular weight conjugates of both topoisomerase II␣ and II␤ isozymes in HeLa cells. Immunological characterization of these conjugates suggests that both topoisomerase II␣ and II␤ isozymes are conjugated to SUMO-1. The involvement of SUMO-1/UBC9 in the modification of topoisomerase II isozymes is also supported by the demonstration of physical interaction between topoisomerase II and SUMO-1/UBC9. Surprisingly, ICRF-193, which does not induce topoisomerase II-mediated DNA damage but traps topoisomerase II into a circular clamp conformation, is also shown to induce similar SUMO-1 conjugation to topoisomerase II isozymes. In addition, we show that both oxidative and heat shock stresses, which can cause protein damage, rapidly increase nuclear SUMO-1 conjugates. These studies raise the question on whether SUMO-1 conjugation to topoisomerases is an indirect result of a DNA damage response or a direct result because of protein conformational changes.
Topoisomerase-mediated DNA damage represents a unique form of DNA damage in which topoisomerases are covalently linked to the broken DNA strands (1). Many antitumor drugs are known to induce topoisomerase-mediated DNA damage (reviewed in Ref. 2). However, the mechanism for repair of topoisomerase-mediated DNA damage is still largely unclear. Topoisomerase I (TOP1) 1 -mediated DNA damage has been shown to activate the ubiquitin/26 S proteasome pathway resulting in degradation of TOP1 (TOP1 down-regulation) (3). Recent studies have also demonstrated that TOP1-mediated DNA damage induced by camptothecin results in rapid accumulation of SUMO-1-TOP1 conjugates (4). Both TOP1 downregulation and SUMO-1 modification of TOP1 have been suggested to be repair responses to TOP1-mediated DNA damage (3,4).
In this communication, we show that TOP2-mediated DNA damage induced by teniposide (VM-26), a TOP2-specific poison (27), also results in rapid accumulation of SUMO-1-TOP2 conjugates. In addition, ICRF-193, which is known not to induce TOP2-mediated DNA damage but to trap topoisomerase II into circular clamp conformation, is also shown to induce rapid accumulation of SUMO-1-TOP2 conjugates. Furthermore, heat shock stress, which is known to cause protein unfolding, also significantly increases SUMO-1 conjugation to nuclear proteins. These results suggest that the signal that triggers SUMO-1 conjugation to topoisomerase II may result from the conformational change (the circular clamp conformation) of topoisomerase II.

EXPERIMENTAL PROCEDURES
Materials-ICRF-193 was purchased from ICN Biomedicals. Protein A-Sepharose bead 4B and GST-Sepharose beads were purchased from Amersham Pharmacia Biotech. The Matchmaker II yeast two-hybrid system and human placenta cDNA library were purchased from CLON-TECH. Mouse monoclonal anti-SUMO-1 antibody (anti-GMP1) was purchased from Zymed Laboratories Inc. Laboratories. Rabbit antisera against the hTOP2␤ isoform were obtained from Dr. Fritz Boege (University of Wuerzburg, Germany). Rabbit antisera against hTOP2␣␤ were obtained from Dr. Jaulang Hwang (Academia Sinica, Taiwan). Staphylococcal S7 nuclease was purchase from Roche Molecular Biochemicals.
Cloning of hTOP2␤ cDNA-The full-length of hTOP2␤ cDNA was cloned from human U937 cells by reverse transcriptase-polymerase chain reaction. The full-length hTOP2␤ cDNA was cloned into yeast hTOP2␣ expression vector YEpWob6 by replacing hTOP2␣ cDNA (28). Both hTOP2␣ and hTOP2␤ were overexpressed and purified according to the procedure described in Ref. 28. The full-length hTOP2␤ cDNA was also cloned into plasmid pAS2-1 (Matchmaker II) as a bait for screening the human placenta cDNA library (Matchmaker II). The active site tyrosine at amino acid 821 was mutated to Phe by polymerase chain reaction based site-directed mutagensis. The mutation was confirmed by sequencing. The cDNA containing the Y821F mutation was similarly cloned into pAS2-1 for yeast two-hybrid assay.
Yeast Two-hybrid Assays-The yeast two-hybrid screen was performed using full-length hTOP2␤ cDNA and Matchmaker II human placenta cDNA library (CLONTECH). The large scale screening assay was done as described in the Matchmaker II protocol. The filter and quantitative liquid ␤-galactosidase assay were done according to the protocol described by the Matchmaker II system (CLONTECH).
GST Pull-down Assay-The full-length hUBC9 cDNA, which was excised from the plasmid recovered from the positive yeast clone obtained from the yeast two-hybrid screen, was constructed to fuse inframe with GST in the pGEX-2T vector (Amersham Pharmacia Biotech). GST and GST-hUBC9 fusion protein were overexpressed in E. coli and purified by batch elution from affinity glutathione-Sepharose 4B beads (Pharmacia Biotech). The GST pull-down assay was performed exactly the same as described previously except that hTOP2 rather than hTOP1 antibodies were used for immunoblotting (4).
Co-immunoprecipitation Assay-This assay was performed exactly the same as described previously except that anti-hTOP2 rather than anti-hTOP1 antibodies were used in co-immunoprecipitation (4).
Detection of SUMO-1-hTOP2 Conjugates in HeLa Cells-HeLa cells in subconfluency were treated with 100 M VM-26 or 100 M ICRF-193 for 30 min. Subsequently, cells were lysed by an alkali solution as described previously (3). Following neutralization, lysates were digested with Staphylococcal S7 nuclease at room temperature for 15 min and then analyzed by 5% SDS-polyacrylamide gel electrophoresis. Immunoblotting with anti-hTOP2␣, anti-hTOP2␤, and anti-SUMO-1 antibodies was performed as described (3).
Fractionation of HeLa Cells-HeLa cells were seeded in 60 ϫ 15-mm dishes. Cells in subconfluency were subject to heat shock treatment. Following heat shock, cells were trypsinized and pelleted. Cell pellets were resuspended in 2 ml of buffer N (50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 0.5% Nonidet P-40, 1 mM NaF, 5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, and 10 g/ml pepstatin A) and centrifuged at 100 ϫ g for 5 min at 4°C. The supernatant and the nuclear pellet fractions were collected separately after centrifugation. Proteins in the supernatant fraction were subject to trichloroacetic acid precipitation. Both supernatant and nuclear pellets were then resuspended in an alkaline solution as described previously (3). Following neutralization, the samples were digested with Staphylococcal S7 nuclease and then subject to immunoblotting analysis.
Immunoprecipitation of hTOP2-SUMO-1 Conjugates-HeLa cells treated with 100 M ICRF-193 for 50 min were trypsinizied, and cell pellets were washed once with phosphate-buffered saline and resuspended in 0.5 ml of radioimmune precipitation buffer (phosphate-buffered saline, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS and protease inhibitors) with brief sonication. 150 l of lysates were precleared by incubating with 2 mg of protein A-Sepharose beads for 2 h at 4°C. The precleared lysates were then mixed with another 2 mg of protein A-Sepharose beads and anti-hTOP2␣ or anti-hTOP2␤ antibodies for 2 h at 4°C. After incubation, beads were washed twice with 600 l of radioimmune precipitation (RIPA) buffer and once with rinse buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride). The beads were resuspended in 45 l of 2ϫ SDS buffer and boiled for 10 min. The samples were analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotted with anti-hTOP2 antibodies.

RESULTS
hTOP2␣ and hTOP2␤ Interact with hUBC9 -Using hTOP2␤ as the bait in a yeast two-hybrid screen (see "Experimental Procedures"), we have identified 12 positive clones. Sequencing analysis of these clones have revealed that 9 of 12 of these clones encodes full-length hUBC9. The interaction between hUBC9 and hTOP2␤ was also confirmed by quantitative liquid ␤-galactosidase assay. The interaction between hTOP2␤ (both wild type and active site mutant Y821F) and hUBC9 gave rise to ␤-galactosidase activity, which was about 100-fold higher compared with hUBC9 alone without the bait (Table I). However, no positive interaction between the C-terminal domain of hTOP2␤ (amino acids 991-1621) and hUBC9 was demonstrated in the filter assay. UBC9 is the E2 enzyme for SUMO-1 conjugation (16). It has been reported that some proteins, which are covalently modified by SUMO-1 also interact with SUMO-1 noncovalently (9,29). Consequently, we have also tested the interaction between hTOP2␤ and SUMO-1 using the yeast two hybrid assay. As shown in Table I, hTOP2␤ also interacts strongly with SUMO-1 in the ␤-galactosidase filter assay.
To confirm the interaction between hUBC9 and hTOP2␤, we constructed and purified a GST-hUBC9 fusion protein and performed a GST pull-down assay and co-immunoprecipitation assay. Using the GST pull-down assay, we showed that both hTOP2␤ and hTOP2␣ interacted with hUBC9 ( Fig. 1A). GST-hUBC9 pulled down both hTOP2␣ and hTOP2␤ from nuclear extracts and purified protein preparations (Fig. 1A, lanes 3 and 4). As a control using GST alone, neither hTOP2 isozyme was pulled down (Fig. 1A, lanes 1 and 2). This result suggested that the interaction between hTOP2 and GST-hUBC9 was specific. This specific interaction between hTOP2 and hUBC9 was further confirmed by the co-immunoprecipitation assay. In this assay, GST-hUBC9 was mixed with the nuclear extract prepared from human 2RA cells (WI38 cells transformed with SV40 T antigen). As shown in Fig. 1B (lane 3), antibodies that recognize both hTOP2␣ and hTOP2␤ isoforms co-immunoprecipitated GST-hUBC9 from nuclear extract (Fig. 1B, lane 3). In contrast, HA antibodies (used as a control) did not immunoprecipitate GST-hUBC9, which remained in the supernatant fraction (Fig. 1B, compare lanes 2 and 4).
hTOP2 Isozymes Are Covalently Modified by SUMO-1 in HeLa Cells Treated with VM-26 -To demonstrate a functional interaction between hTOP2 and SUMO-1/UBC9, we have tested the possibility that hTOP2 can be covalently modified by SUMO-1. As shown in Fig. 2A (lane 1), no SUMO-1-hTOP2 conjugates in HeLa cells were detectable by immunoblotting with anti-hTOP2␣␤ antibodies (antibodies that recognize both isoforms of hTOP2). However, upon VM-26 treatment, some higher molecular weight species (see bands marked by *) were detectable in the immunoblot (Fig. 2A). The same blot was stripped and reblotted with anti-SUMO-1 antibodies. The same higher molecular weight species were detected (see bands marked by * in lanes 3 and 4 of Fig. 2A). These experiments suggest that these high molecular weight species could be SUMO-1-hTOP2 conjugates.
To test whether both hTOP2 isoforms were covalently modified by SUMO-1, antibodies specific for each hTOP2 isoforms were used to perform immunoblotting analysis. As shown in Fig. 2B, high molecular species were detectable in HeLa cells treated with VM-26 using isozyme-specific antibodies against either isoform.   (30). Unlike VM-26, which induces topoisomerase II-mediated DNA cleavage, ICRF-193 inhibits the catalytic activity of hTOP2 by trapping TOP2 into circular protein clamps in the presence of ATP (31). Interestingly, covalent modification of hTOP2 by SUMO-1 was also observed in HeLa cells treated with ICRF-193 (Fig. 3A). As shown in Fig. 3A (compare lanes 1 and 2), a similar group of high molecular weight species (see the bracket labeled hTOP2-SUMO-1 conjugates) were detected by anti-hTOP2␣␤ antibodies using lysates prepared from HeLa cells treated with ICRF-193. The same membrane filter was stripped and reblotted with anti-SUMO-1 antibodies (Fig. 3A, lanes 3 and 4). Again, similar high molecular weight species were observed (see bands in the bracketed region in Fig. 3A), suggesting that these high molecular weight species could be covalent SUMO-1-hTOP2 conjugates.
To verify that those conjugates induced in HeLa cells treated with ICRF-193 were covalent SUMO-1-hTOP2 complexes, we immunoprecipitated hTOP2 isozymes from HeLa cell lysates using isozyme-specific antibodies (Fig. 3B). The immunoprecipitates were then immunoblotted with either anti-hTOP2␣␤ antibodies (Fig. 3B, lanes 3-6) or anti-SUMO-1 antibodies (Fig.  3B, lanes 9 -12). The same high molecular weight species (see bracketed regions) in immunoprecipitates were detectable using either hTOP2␣␤ antibodies or anti-SUMO-1 antibodies, suggesting that these high molecular weight species contained both SUMO-1 and hTOP2. In the aggregate, these results suggest that the high molecular weight species induced by either ICRF-193 or VM-26 are SUMO-1-hTOP2 conjugates.
SUMO-1-conjugated hTOP2 in Cells Treated with ICRF-193 Is Not Covalently Linked to DNA-In our current study, we have shown that both ICRF-193 and VM-26, which represent two distinct types of TOP2 inhibitors, induce SUMO-1 conjugation to hTOP2. Previous studies have demonstrated that the majority of SUMO-1-TOP1 conjugates in cells treated with camptothecin are covalently linked to DNA (3,4). To test whether SUMO-1-hTOP2 conjugates in cells treated with either ICRF-193 or VM-26 were covalently linked to DNA, the effect of nuclease (Staphylococcal S7 nuclease) treatment of cell lysates was tested (Fig. 4). As shown in Fig. 4A (lanes 1 and 2), in the absence of nuclease treatment, no SUMO-1 conjugates were detectable in HeLa cells treated with VM-26. The decrease in the band intensities of hTOP2␣ and hTOP2␤ (Fig. 4A,  lane 2) was because of the formation of significant fractions of hTOP2 into covalent complexes with DNA, which were too large to enter gel (32). Upon nuclease treatment, these covalent complexes (either retained in the well or migrated as smears) were released from DNA, and consequently more hTOP2 species including those high molecular weight SUMO-1-hTOP2 conjugates were detected as distinct bands (Fig. 4A, lane 4). 1. Both hTOP2␣ and hTOP2␤ interact with hUBC9 in GST pull-down and co-immunoprecipitation assays. A, hTOP2 interacts with GST-hUBC9 in the GST pull-down assay. The GST pull-down assay was performed as described under "Experimental Procedures." Samples from all bead fractions were immunoblotted with anti-hTOP2␣␤ antibodies. Lanes 1 and 2 were samples from GST bead fractions. Lanes 3 and 4 were samples from GST-hUBC9 bead fractions. NE indicates that nuclear extracts were used in the pull-down assay (lanes 1 and 3). TOP2 indicates that purified recombinant hTOP2␣ and hTOP2␤ were used in the pull-down assay (lanes 2 and 4). B, coimmunoprecipitation of hTOP2 and GST-hUBC9. The co-immunoprecipitation assay was performed as described under "Experimental Procedures." Samples from both bead and supernatant fractions were immunoblotted with anti-GST antibodies. Purified GST-hUBC9 is shown in lane 1. Lanes 2 (bead) and 4 (supernatant), anti-HA antibody (control antibody) was used in co-immunoprecipitation. Lanes 3 (bead) and 5 (supernatant), anti-hTOP2␣␤ antibodies were used in co-immunoprecipitation.  lanes 3 and 4). B, VM-26 induces SUMO-1 conjugation to both hTOP2 isozymes. Cell lysates prepared from HeLa cells with (ϩ) or without (Ϫ) VM-26 treatment were analyzed by immunoblotting using hTOP2 isoform-specific antibodies. Lanes 1 and 2, immunoblotted using anti-hTOP2␣␤ antibodies. Lanes 3 and 4, immunoblotted using anti-hTOP2␣ antibodies. Lanes 5 and 6, immunoblotted using anti-hTOP2␤ antibodies.

SUMO-1 Conjugation to Topoisomerase II
However, when a similar experiment was performed using ICRF-193 instead of VM-26, those high molecular weight species were detectable with or without nuclease treatment (Fig.  4B, compare lanes 2 and 4). In addition, the amounts of unconjugated hTOP2 species (hTOP2␣ and hTOP2␤) were about the same with or without nuclease treatment (Fig. 4B, compare   lanes 2 and 4), indicating that ICRF-193, in contrast to VM-26, does not induce the formation of covalent hTOP2-DNA complexes. These experiments suggest that SUMO-1-hTOP2 conjugates formed in cells treated with VM-26 but not with ICRF-193 are covalently linked to chromosomal DNA. Consistent with this interpretation, we have also shown that the formation  3 and 4). B, HeLa cells were treated with ICRF-193 and then lysed by radioimune precipitation (RIPA) buffer. The lysates were immunoprecipitated with either anti-hTOP2␣ or anti-hTOP2␤ antibodies as described under "Experimental Procedures." Samples were analyzed by 5% SDS-polyacrylamide gel electrophoresis and immunoblotted using either anti-hTOP2␣␤ antibodies (lanes 1-6) or anti-SUMO-1 antibodies (lanes 7-12). Lane 1, lysates from HeLa cells without ICRF-193 treatment (Ϫ). Lane 2, lysates from HeLa cells treated with ICRF-193 (ϩ). Lane 3, immunoprecipitate of lysate from lane 1 using hTOP2␣ antibodies. Lane 4, immunoprecipitate of lysate from lane 2 using anti-hTOP2␣ antibodies. Lanes 5 and 6, the same as lanes 3 and 4, respectively, except that anti-hTOP2␤ antibodies were used for immunoprecipitation. The same membrane filter was stripped and reblotted with anti-SUMO-1 antibodies (lanes 7-12).

FIG. 4. VM-26-but not ICRF-193-induced SUMO-1-hTOP2 conjugates are covalently linked to chromosomal DNA.
A, HeLa cells were treated with VM-26 and then lysed by the alkali lysis procedure as described in Fig. 2. Following neutralization, lysates were treated with Staphylococcal S7 nuclease and subject to immunoblotting with anti-hTOP2␣␤ antibodies. B, the same as in A except that the treatment was with ICRF-193. AЈ and BЈ, the same membrane filters as shown in A and B, respectively, were stripped and reblotted with anti-hTOP1 antibodies. Treatment of HeLa cells with either VM-26 or ICRF-193 did not induce the formation of hTOP1-SUMO-1 conjugates. of SUMO-1-hTOP2␣ conjugates occurs in CEM cells but not in VM-26-resistant CEM/VM-1 cells (Fig. 5). CEM/VM-1 cells are known to express a mutant form of hTOP2␣, which is highly resistant to VM-26-induced formation of TOP2␣ cleavable complexes (33). The presence of a 160-kDa form of hTOP2␣ in CEM/VM-1 cells has been reported previously (33,43).
Heat Shock Stress Induces SUMO-1 Conjugation to Nuclear Proteins-The above experiment suggests that SUMO-1 conjugation to hTOP2 may be triggered by a direct signal from hTOP2 (e.g. an altered conformation of hTOP2 in the presence of either VM-26 or ICRF-193) rather than an indirect signal from the broken DNA strands. To test this notion, we have monitored the formation of SUMO-1 conjugates in cells treated with heat shock as well as agents known to damage either protein or DNA. As shown in Fig. 6A, heat shock treatment of HeLa cells at 45°C for 10 min resulted in a greatly elevated level of SUMO-1 conjugates. Interestingly, concomitant with the increase in the level of SUMO-1 conjugates, the unconjugated SUMO-1 pool decreased to an undetectable level (Fig. 6A,  lower panel). These results suggest that, upon heat shock, the unconjugated SUMO-1 pool is rapidly mobilized to form SUMO-1 conjugates. The increase in the levels of SUMO-1 conjugates was also observed in HeLa cells treated with hydrogen peroxide (Fig. 6B) or N-ethylmaleimide (NEM) (Fig. 6C). No increase in the level of SUMO-1 conjugates was observed in HeLa cells treated with cisplatin, which is known to damage DNA (Fig. 6E). Except at very high doses (200 and 2000 J/m 2 ), UV irradiation resulted in little increase in the level of SUMO-1 conjugates (Fig. 6D).
To determine the cellular distribution of SUMO-1 conjugates upon heat shock treatment, HeLa cells with and without heat shock were fractionated into cytoplasmic (C) and nuclear (N) fractions (Fig. 7). Without heat shock, except for a 90-kDa protein which is SUMO-1-conjugated RanGAP1 (7), nearly all SUMO-1 conjugates were in the nuclear fraction (Fig. 7A, compares lanes 4 and 5). Upon heat shock, the level of SUMO-1 conjugates in the nuclear fraction (N) was greatly increased, whereas SUMO-1 conjugates in the cytoplasmic fraction remained undetectable (Fig. 7A, compare lanes 2 and 3). To ensure that the lack of SUMO-1 conjugates in the cytoplasmic fraction is not because of a loss of cytoplasmic proteins during fractionation, a duplicate gel was stained with Coomassie Blue and shown in Fig. 7C. We have also monitored the levels of ubiquitin conjugates in HeLa cells with or without heat shock treatment. The same gel as shown in Fig. 7A was stripped and reblotted with anti-ubiquitin antibodies. As shown in Fig. 7B, heat shock greatly increased in the levels of ubiquitin conjugates in both the cytoplasmic (C) and nuclear (N) fractions.
The effect of heat shock on the increase in SUMO-1 conjugates may be because of two possibilities; one is that heat shock could partially unfold some nuclear proteins making them better substrates for SUMO-1/UBC9. The other is that heat shock could either stimulate the activity of UBC9 and/or inhibit the activity of the de-SUMOylation enzyme, resulting in the increase in the steady state level of SUMO-1 conjugates. To test the latter possibility, we have performed the following experiment (Fig. 8). Previous studies have demonstrated that TOP1mediated DNA damage induced by camptothecin specifically induces accumulation of SUMO-1-hTOP1 conjugates (4). As shown in Fig. 8, HeLa cells treated with camptothecin resulted in the formation of a ladder of bands, which have previously been identified to be SUMO-1-hTOP1 conjugates (Fig. 7, lane  2). These SUMO-1-hTOP1 conjugates did not increase in abundance upon heat shock as would be expected if the latter possibility were true. They in fact dramatically decreased in their abundance (Fig. 8, compare lanes 2 and 4). The decrease in abundance of SUMO-1-hTOP1 conjugates is probably because of heat (45°C)-induced reversal of hTOP1 cleavable complexes followed by de-SUMOylation (data not shown). This result thus argues against the latter possibility and favors the former possibility that unfolding of nuclear protein induced by heat shock directly signals SUMO-1 modification. DISCUSSION SUMO-1, a ubiquitin-like protein, has been identified to be conjugated to many important cellular proteins (for review, see Ref. 34). Consequently, SUMOylation has been suggested to be involved in diverse biological functions such as regulation of ubiquitin conjugation in the case of IB␣ (35), regulation of protein transport in the case of RanGAP1 (36 -38), the formation of PML oncogenic domains (PODs) (nuclear bodies) in the case of PML (39,40), regulation of transcriptional activation in the case of p53 (41,42), and regulation of apoptosis in the case of the FAS (9). Our recent studies have suggested that SUMO-1-hTOP1 conjugates may be involved in the repair of topoisomerase I-mediated DNA damage (4).
In this study, we show that VM-26 induces SUMO-1 modification of both hTOP2 isozymes, hTOP2␣ and hTOP2␤, in HeLa cells. The role of UBC9 and SUMO-1 in this system has been further corroborated by the demonstration of physical interactions between hTOP2 and UBC9 and between hTOP2 and SUMO-1. VM-26 is a TOP2 poison, which traps DNA-hTOP2 in a ternary complex in which TOP2 is covalently linked to the 5Ј-phosphoryl ends of the double strand break (27). We have also demonstrated that VM-26-induced SUMOylation of hTOP2 is greatly reduced in CEM/VM-1 cells, which are defective in VM-26-induced formation of the ternary complex (Fig.  5). Consequently, the effect of VM-26 on SUMOylation of hTOP2 is because of the formation of these ternary complexes rather than other unidentified side effect(s) of VM-26. However, it is unclear whether SUMO-1 conjugation to hTOP2 is signaled by hTOP2 because of its conformational change and/or an indirect response because of a DNA damage response.
To differentiate between these two possibilities, another TOP2 inhibitor, ICRF-193, was used. ICRF-193 belongs to a different class of TOP2 inhibitors that inhibit the catalytic activity of TOP2 without trapping the covalent ternary complex (i.e. no topoisomerase II-mediated DNA damage). In the pres- ence of ATP, ICRF-193 is known to lock TOP2 into a circular clamp conformation (31). In this study, we have shown that ICRF-193 also induces SUMO-1 conjugation to hTOP2. As expected, the SUMO-1-hTOP2 conjugates induced by ICRF-193, unlike those induced by VM-26, were not covalently linked to DNA. This result suggests that the signal for SUMO-1 conjugation to hTOP2, at least in the case of ICRF-193, is not from DNA damage.
Previous studies have also demonstrated that in the presence of ATP or AMP-PNP, TOP2-mediated DNA cleavage induced by VM-26 is greatly (over 50-fold) stimulated (44). Consequently, it was suggested that the ATP-or AMP-PNP-bound form of TOP2 is the major target of VM-26. This ATP-or AMP-PNP bound form of TOP2 has also been shown to be a circular protein clamp (31). It seems possible that it is the circular protein clamp conformation(s) of TOP2 that signals SUMO-1 conjugation to TOP2 in cells treated with either ICRF-193 or VM-26. Our studies have also demonstrated that hTOP2 physically interacts with both SUMO-1 and UBC9. One could speculate that the conformational change associated with the formation of the circular protein clamp could either enhance hTOP2 interaction with SUMO-1 and/or UBC9 or bring SUMO-1 and UBC9 into proper geometry on the enzyme surface for effective SUMO-1 conjugation to hTOP2. Alternatively, the circular clamp conformation could increase the k cat of the SUMO-1 conjugation reaction. We are in the process of testing these possibilities in an in vitro SUMOylation system. Clearly, further experiments are necessary to establish whether SUMO-1 conjugation to hTOP2 is signaled by TOP2 conformational change(s) or an unknown mechanism(s) (e.g. a DNA damage response).
We have also examined the effect of some stress conditions on SUMO-1 conjugation in HeLa cells. These stress conditions are known to cause protein unfolding (e.g. heat shock) and/or protein damage (e.g. H 2 O 2 and NEM). Among the various stress conditions, heat shock and H 2 O 2 treatment are most effective in increasing the levels of SUMO-1 conjugates in the nucleus. NEM exhibited less effect compared with heat shock and H 2 O 2 . Within 10 min of heat shock at 45°C, the levels of SUMO-1 conjugates in HeLa cells were greatly elevated. Concomitantly, the unconjugated SUMO-1 pool was rapidly exhausted. Except for the 90-kDa SUMO-1-conjugated RanGAP1, all SUMO-1 conjugates were shown to be in the nuclear fraction. The simplest explanation is that heat shock induces partial unfolding of some nuclear proteins. These unfolded proteins trigger their SUMO-1 conjugation. However, it is also possible that SUMO-1 conjugates may exist in a dynamic state being constantly formed and de-SUMOylated. In this case, heat shock might increase the steady state level of SUMO-1 conjugates by affecting the enzymes (e.g. UBC9 and the SUMO-1specific protease) involved in SUMOylation and de-SUMOylation. If this were true, the levels of all SUMO-1 conjugates should be elevated upon heat shock. However, careful inspec-tion of Fig. 6A has revealed that heat shock appears to induce additional SUMO-1 conjugates without affecting some of the existing SUMO-1 conjugates. In addition, the level of camptothecin-induced SUMO-1-hTOP1 conjugates was decreased rather than increased after heat shock treatment. These results argue against an effect of heat shock on SUMOylation and/or de-SUMOylation enzymes. Thus, it seems plausible that heat shock may cause unfolding of certain nuclear proteins that directly signal their SUMO-1 conjugation. The effect of H 2 O 2 can be similarly explained. H 2 O 2 is known to damage both DNA and protein via reactive oxygen species. These damaged proteins may somehow signal SUMO-1 conjugations. This result may be particularly significant, because cells are constantly under oxidative stress. SUMOylation may thus be evolved as a defense mechanism for repair/removal of damaged nuclear proteins. The effect of UV irradiation on SUMOylation is interesting. At low doses of UV, little effect on SUMOylation was observed. However, at high doses (over 200 J/m 2 ), SUMOylation of nuclear proteins was significantly increased. It has been reported that UV irradiation increases SUMOylation of p53 (40,41). Our preliminary study has also demonstrated that UV induces SUMOylation of hTOP1. 2 UV irradiation is known to damage both nucleic acids and proteins (45). Whether the effect of UV on SUMOylation of nuclear proteins is because of DNA damage or protein damage requires further clarification. It should be pointed out that whereas our studies have demonstrated the involvement of SUMO-1 in modifying topoisomerases and nuclear proteins, we cannot rule out the possibility that SUMO-2 and SUMO-3 may also be involved (8,46). In fact, we have shown that camptothecin can induce SUMO-2/3-hTOP1 conjugates in addition to SUMO-1-hTOP1 conjugates in mammalian cells. 2 The fate of SUMO-1-modified nuclear proteins remains unclear. SUMO-1 conjugates could be destined for degradation by interacting with the ubiquitin/26 S proteasome pathway or for repair/refolding by some unknown mechanisms. Our current studies have demonstrated that topoisomerases are attractive models for studying SUMO-1 function because of the availability of specific inhibitors of different classes. Clearly, further studies are necessary to reveal the function of SUMO-1 in the 2 Y. Mao and L. F. Liu, unpublished results. repair of topoisomerase-mediated DNA damage and more generally in regulating nuclear protein functions.