A Novel Nuclear Localization Signal in Human DNA Topoisomerase I*

DNA topoisomerase (topo) I is a nuclear enzyme that plays an important role in DNA metabolism. Based on conserved nuclear targeting sequences, four classic nuclear localization signals (NLSs) have been proposed at the N terminus of human topo I, but studies with yeast have suggested that only one of them (amino acids (aa) 150–156) is sufficient to direct the enzyme to the nucleus. In this study, we expressed human topo I fused to enhanced green fluorescent protein (EGFP) in mammalian cells and demonstrated that whereas aa 150–156 are sufficient for nuclear localization, the nucleolar localization requires aa 157–199. More importantly, we identified a novel NLS within aa 117–146. In contrast to the classic NLSs that are rich in basic amino acids, the novel NLS identified in this study is rich in acidic amino acids. Furthermore, this novel NLS alone is sufficient to direct not only EGFP into the nucleus but also topo I; and the EGFP·topo I fusion driven by the novel NLS is as active in vivo as the wild-type topo I in response to the topo I inhibitor topotecan. Together, our results suggest that human topo I carries two independent NLSs that have opposite amino acid compositions.

DNA topoisomerase (topo) I is a nuclear enzyme that plays an important role in DNA metabolism. Based on conserved nuclear targeting sequences, four classic nuclear localization signals (NLSs) have been proposed at the N terminus of human topo I, but studies with yeast have suggested that only one of them (amino acids (aa) 150 -156) is sufficient to direct the enzyme to the nucleus. In this study, we expressed human topo I fused to enhanced green fluorescent protein (EGFP) in mammalian cells and demonstrated that whereas aa 150 -156 are sufficient for nuclear localization, the nucleolar localization requires aa 157-199. More importantly, we identified a novel NLS within aa 117-146. In contrast to the classic NLSs that are rich in basic amino acids, the novel NLS identified in this study is rich in acidic amino acids. Furthermore, this novel NLS alone is sufficient to direct not only EGFP into the nucleus but also topo I; and the EGFP⅐topo I fusion driven by the novel NLS is as active in vivo as the wild-type topo I in response to the topo I inhibitor topotecan. Together, our results suggest that human topo I carries two independent NLSs that have opposite amino acid compositions.
Topoisomerase I (topo I) 1 regulates DNA topology by making single-strand breaks, allowing strand passage, and then resealing these breaks independent of ATP hydrolysis (1). Thus, topo I plays an important role in different aspects of DNA metabolism such as DNA replication, DNA recombination, transcription, and, possibly, DNA repair (2)(3)(4). In addition to its catalytic activity on DNA, the enzyme has been shown to have other activities functioning as a ribonuclease and a kinase (5-7); its kinase activity has been shown to phosphorylate RNA splicing factors (5). Topo I may also play a role in chromatid condensation (8). In lower eukaryotic organisms, topo I seems to be dispensable, in part because of the fact that other topoisomerases can subserve its role in its absence. However, mammalian topo I is essential for cell growth. Furthermore, topo I is a target for clinically important anticancer drugs such as topo-tecan and SN-38, the metabolic product of CPT-11 (9). The inhibitors target nuclear topo I and stabilize transient DNAenzyme complexes, leading to cell death (9).
Human topo I is a 765-aa protein that is exclusively localized to the nucleus because the N-terminal domain of the enzyme carries essential sequences for its nuclear localization (10,11). Although this N-terminal domain does not seem to contribute to its catalytic activity in vitro, it is essential for its in vivo activity because the enzyme exerts its functions in the nucleus (12,13).
Although all nuclear localization signals (NLSs) for nuclear proteins seem to play the same role, i.e. targeting the nucleus, they vary considerably from short peptide motifs to large protein domains (14). The classic peptide motifs are those rich in basic amino acids and are represented by a well known NLS found in SV40 T-antigen (15).
Based on their conserved amino acid sequences, four putative NLSs have been suggested to be in the N-terminal domain of human topo I (12,16,17). The first putative NLS (NLS-I) is at aa 59 -65, and the other three putative NLSs are clustered at aa 150 -198 (NLS-II, aa 150 -156; NLS-III, aa 174 -180; and NLS-IV, aa 192-198) (see Fig. 1). These NLSs consist of seven amino acids, each characterized by its high content of basic amino acids (16), and are thereby considered "classic" peptide NLSs (14).
Overexpression of human topo I in yeast revealed that a region covering 71 amino acids (aa 140 -210) of the N terminus is sufficient to direct the enzyme to the nucleus (12). By amino acid alignment, a conserved sequence (PKKIKTE) exists among human and yeast topo I and SV40 T antigen, suggesting that this is sufficient to direct the topo I to the nucleus (12). However, it is not known whether this sequence is functional in mammalian cells or whether all these putative NLSs are required for nuclear localization. Furthermore, topo I is predominantly localized in the nucleolus (18,24), but the sequence for this nucleolar localization has not been defined.
We have recently shown that overexpression of catalytically active human topo I in mammalian cells can be achieved by fusion to enhanced green fluorescent protein (EGFP) (19). Thus, the EGFP⅐topo I fusion protein provides a useful tool to examine the subcellular localization of topo I in live cells. Our data revealed that the EGFP⅐topo I closely associates with chromosomal DNA at both interphase and mitosis (19). Furthermore, our studies also revealed that the N-terminal domain is essential for the enzyme to get into the nucleus, suggesting that the functional NLS of the human topo I is located at the N terminus, consistent with findings in yeast (12). To delineate the specific sequences required for topo I to target the nucleus in mammalian cells, we expanded our deletion analysis of its N-terminal domain. To our surprise, as described herein, we found a novel NLS that does not fall into the classic basic peptide motif and is sufficient to direct topo I into the nucleus in mammalian cells.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-HeLa, COS-7, and NIH 3T3 cells were obtained from ATCC (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium (BioWhittaker, Walkersville, MD). Chinese hamster ovary cells (ATCC) were grown in RPMI. All media were supplemented with 10% fetal bovine serum, 100 units of penicillin/ml, and 100 g of streptomycin/ml. Cells were incubated at 37°C in a humidified chamber supplemented with 5% CO 2 .
Plasmids carrying the EGFP⅐topo I gene were introduced into HeLa cells by the calcium phosphate method as described previously (20). After transfection, cells were subcultured in 6-well plates with one coverslip in each well and allowed to grow for another 16 -24 h before microscopic examination. Transfection of COS-7, NIH 3T3, and Chinese hamster ovary cells was conducted by electroporation, using Electroporator with Extender II (Bio-Rad). Cells at the log phase were trypsinized, harvested, and then resuspended in Dulbecco's modified Eagle's medium without fetal bovine serum at 1 ϫ 10 7 cells/ml. An aliquot of 0.4 ml of such cell suspension was mixed with 15 g of plasmid DNA and incubated for 10 min at room temperature. Cells were subjected to an electric pulse (950 microfarads and 220 volts), as suggested by the manufacturer.
Plasmid Construction-Expression plasmids used in this study are shown in Fig. 1 and Fig. 7A. The full-length topo I⅐EGFP fusion construct (pTI-2) and the N-terminal fusion (pTI-5) have been described previously (19). pTI-2/X1 encompassed aa 1-140 and was constructed by digestion of pTI-2 with EcoRI and partial digestion with XhoI. The topo I fragment with the vector was isolated and purified by Gene Clean II (Bio 101, Vista, CA). The overhanging ends of this fragment were blunted by treatment with mungbean nuclease and then self-ligated in the presence of T4 DNA ligase.
For deletions involving other regions of topo I, where no suitable restriction enzyme sites were available in the original topo I cDNA sequence, PCR was performed to introduce appropriate enzyme sites for convenient subcloning. PCR reactions were essentially performed using a standard method (21). In general, appropriate DNA fragments were first amplified by PCR, cloned into pCR2.1 (Invitrogen, Carlsbad, CA), and then subcloned into the pEGFP-C3 (CLONTECH, Palo Alto, CA). Sequences and positions of the PCR primers are listed in Table I. pTI-28 carried aa 1-199 with a deletion of aa 148 -157 and was constructed as follows. First, a fragment that covers aa 1-147 was amplified by PCR using primers I-5.1 and I-3.3B (where a PstI site was introduced; see Table I); a second fragment that covers aa 158 -199 was amplified by primers I-5.1L (where a PstI site was also introduced; Table I) and I-3.2. These two fragments were then ligated and cloned into pEGFP-C3 so that the resultant plasmid (pTI-28) carried the N terminus of topo I with a deletion of aa 148 -157. The same strategy was used to construct pTI-29, pTI-31, pTI-32, and pTI-33 using appropriate primer sets. The control vector pT104/B1 has been described previously (22).
Western Blot-Either nuclear extract or total protein was prepared for Western blot as described previously (23). To detect the expression of the EGFP⅐topo I fusion proteins, immunoblotting was carried out using antibodies against green fluorescent protein (CLONTECH). Topo I-specific antibody TI-I (24) was used to detect both endogenous and EGFP⅐topo I fusion proteins.
Immunostaining-Immunostaining was carried out as described previously (25), but with a slight modification. In brief, HeLa cells were grown over coverslips in 6-well plates overnight. The cells were fixed in 2% formaldehyde and 0.2% glutaraldehyde in phosphate-buffered saline and permeabilized by 0.2% Triton X-100. After incubation with the primary antibody against Hsp70 (monoclonal, Amersham Pharmacia Biotech), the cells were then incubated with secondary donkey antimouse antibody labeled with fluorescein isothiocyanate (Jackson Im-munoResearch, West Grove, PA). The samples were examined by fluorescence microscopy (Carl Zeiss, Thornwood, NY) using a filter with maximum excitation at 480 nm and maximum emission at 520 nm.
Fluorescence Microscopy-To detect subcellular localization of EGFP⅐topo I, cells were subcultured after transfection with an appropriate plasmid in 6-well plates with a coverslip in each well and grown for 16 -24 h. When needed, cells were also fixed with cold fixing solution (1% paraformaldehyde in phosphate-buffered saline). For nuclear staining, the fixed cells were incubated with Hoechst dye (1 g/ml, Sigma) for 15 min at room temperature. Fluorescent signals were revealed under the fluorescence microscope using the filters for EGFP and 4Ј,6-diamidino-2-phenylindole, respectively.
Band Depletion and Growth Inhibition Assays-Band depletion and growth inhibition assays have been described previously (19).

Expression of Deleted EGFP⅐Topo I Fusion Constructs in
HeLa Cells-Our previous studies have shown that the Nterminal domain is essential for human topo I to target the nucleus in HeLa cells (19), and four putative NLSs have been proposed at this region (12). To investigate whether any or all of these NLSs are actually functional in mammalian cells, we made a series of deletion mutants for the N-terminal domain ( Fig. 1). Expression of those deleted EGFP⅐topo I fusion constructs was examined by Western blot. As shown in Fig. 2, the fusion proteins were detected at the sizes expected. For in-

62-68
a The nucleotides in bold are either introduced or mutagenized. b The starting methionine as aa 1. c s, sense strand; as, antisense strand. stance, we detected an apparent molecular mass of ϳ60 kDa for pTI-5, which is in good agreement with its predicted size ( Fig.  2). Similarly, we also detected the rest of the fusion proteins roughly within their predicted molecular masses (Fig. 2).
The N-terminal Domain of Human Topo I Carries More than One Functional NLS-Once these fusion constructs were confirmed to be expressed exogenously in HeLa cells, we followed their subcellular localization. Consistent with our previous results (19), cells transfected with pTI-5 (Clone #1 in Fig. 1) displayed green fluorescent protein in the nucleus, and was concentrated in nucleoli (Fig. 3). Consequently, experiments were designed to determine what specific sequences are required for nuclear/nucleolar localization.
Among four putative NLSs at the N terminus of human topo I, NLS-I (aa 59 -65) is well separated from the other three (aa 150 -198; see Fig. 1) (13). Thus, we attempted initially to test whether NLS-I plays any role in targeting the nucleus. Hence, we made a fusion construct, pTI-7 (Clone #4, aa 1-146), in such a way that the cloned DNA fragment ended immediately before NLS-II (aa 150 -156). The fusion protein encoded by pTI-7 showed exclusively nuclear distribution (Fig. 3), suggesting that a functional NLS exists at aa 1-146, even in the absence of the other three putative NLSs. To determine whether NLS-I (aa 59 -65) is responsible for the observed nuclear localization, we made pTI-8 (Clone #2, aa 1-67) that included NLS-I, but we deleted the rest of the sequences (aa 68 -146). The fusion protein from pTI-8 revealed cytoplasmic distribution just like that of the vector control (pT104/B1) (Fig. 3), suggesting that NLS-I is not functional for nuclear localization or is not sufficient for it.
The presence of a functional NLS at aa 1-146 suggests that human topo I might carry two NLSs, because previous studies suggested that NLS-II (aa 150 -156) is important, at least in yeast (12). To test whether NLS-II, -III, and -IV are functional in HeLa cells, we made pTI-10 (Clone #5, aa 148 -199), with no overlap with pTI-7. As expected, we found that cells transfected with pTI-10 displayed green fluorescent nuclear staining (Fig. 3). Therefore, two functional NLSs exist at the N terminus of human topo I, one located at aa 1-146, and another located at aa 148 -199. Importantly, they appear to function independently.
NLS-II (aa 150 -156) Alone Is Sufficient for Nuclear Transport-Because amino acids 148 -199 carry three putative NLSs, we made further deletions to delineate a minimal sequence that is sufficient to support nuclear localization. We first made pTI-27 (Clone #7, aa 148 -187) that includes NLS-II and -III but lacks NLS-IV. As seen for pTI-10, the fusion protein from pTI-27 displayed nuclear staining (Fig. 4), suggesting that NLS-IV is not essential. To test the role of NLS-II in nuclear localization, we deleted NLS-II from pTI-10 (aa 148 -199), resulting in pTI-13 (Clone #6, aa 157-199). Although pTI-10 retained nuclear localization capability, a deletion involving NLS-II (pTI-13) abolished its ability to target the nucleus (Fig. 4). These results indicate that NLS-II (aa 150 -156) has the capability of nuclear targeting. To further determine whether NLS-II alone is sufficient to support nuclear localization, a peptide (KPKKIKTED) was fused to EGFP (pTI-26, Clone #8). As expected, cells transfected with pTI-26 clearly revealed nuclear localization (Fig. 4). These results indicate that although three putative NLSs are clustered at aa 148 -199, in HeLa cells, NLS-II is the only functional one among them and that neither NLS-III nor NLS-IV is important for nuclear localization.

NLS-IV Plays a Role in Nucleolar Localization although It Is Not Important for Nuclear Localization-Topo I has been
shown to be predominantly in the nucleolus (18). Consistent with this finding, the fusion protein from pTI-5 was detected in the nucleus and particularly concentrated in the nucleolus (Figs. 3 and 4). To further confirm that the green fluorescent particles observed in pTI-5-transfected cells (Fig. 3) are nucleoli, we immuno-stained HeLa cells with an antibody specific to Hsp70, a well known protein that is localized to the nucleolus when cells are under heat stress (28,29). The staining patterns by the anti-Hsp70 antibody were basically identical to those of EGFP staining by pTI-5 (Fig. 4B), suggesting that the green fluorescent particles seen in cells transfected with pTI-5 are localized in nucleoli. Heat shock treatment did not change the nucleolar localization pattern of pTI-5.
Cells transfected with pTI-10 revealed both nuclear and nucleolar localization, similar to pTI-5 (Figs. 3 and 4), suggesting that the signal for nucleolar localization resided at aa 150 -199. Further deletion analyses support this notion, because both pTI-26 and pTI-27 resulted only in nuclear but not nucleolar localization of the fusion proteins (Fig. 4A). Therefore, although neither NLS-III nor NLS-IV is required for nuclear localization, at least NLS-IV is important for nucleolar localization.
The Novel NLS Is Identified within aa 117-146 -Data from Figs. 3 and 4 indicated that the N terminus of human topo I carries multiple NLSs that function independently. More importantly, NLS at aa 1-146 might be novel, because NLS-I within this region does not appear to play such a role (Fig. 3). To better understand this potential novel NLS, we first made pTI-2/X1 (Clone #3), which carried only aa 1-140, because previous studies suggested that the sequence covering aa 1-140 lacked nuclear targeting ability in yeast (12). As shown in Fig. 5, we found that green fluorescent protein encoded by pTI-2/X1 was predominantly distributed in the cytoplasm, consistent with the previous results (12), although we occasionally observed a few cells with the green fluorescent protein in the nucleus (Ͻ5%). Compared with pTI-2/X1, pTI-7 carries an additional six amino acids (DEDDAD, aa 141-146), suggesting that these six amino acids play an important role in targeting the nucleus. To test whether the sequence DEDDAD alone is sufficient to target the nucleus, we fused this fragment to pEGFP-C3 (pTI-16, Clone #13) and found that the fusion protein from this clone carried no functional NLS (Fig. 5), suggesting that the sequence DEDDAD alone is not sufficient to function as an NLS. It is obvious that the sequence DEDDAD is rich in acidic amino acids, in contrast to classic monopartite NLSs that are rich in basic amino acids (14).
To further delineate the sequence that carries a potential novel NLS, we first tested the effect of deletion of NLS-I on its nuclear localization, and thus we made pTI-9 (Clone #9, aa 69 -146) in which NLS-I (aa 59 -65) was excluded. The fusion protein from pTI-9 was localized to the nucleus (Fig. 5), as seen for pTI-7 (aa 1-146), indicating that NLS-I plays no role in targeting the nucleus. To determine the minimal sequence for the novel NLS, we made a series of deletion constructs: pTI-17 (Clone #10, aa 84 -146), pTI-18 (Clone #11, aa 117-146), and pTI-20 (Clone #12, aa 125-146). We observed nuclei stained with green fluorescent protein in the cells transfected with pTI-17 and pTI-18 (Fig. 5A), as seen for pTI-7 (Fig. 5A). A further deletion involving aa 117-125, i.e. pTI-20, drastically reduced the nuclear localization capability of the fusion protein, resulting in a predominant cytoplasmic distribution of the fusion protein (Fig. 5A). These results suggest that the minimal sequence for the novel NLS is located within aa 117-146.
Because aa 117-146 are an essential part of the novel NLS that alone is sufficient for nuclear import (Fig. 5A), these amino acids were used to search for amino acid homology. As shown in Fig. 5B, this sequence was found in topo I of mice, Chinese hamsters and chickens, in addition to humans, with an almost perfect match (26,27). A high homology was also found with topo I of frogs (Fig. 5B). Interestingly, the yeast or Drosophila topo I lacks this sequence. Because the six amino acids (DED-DAD, aa 141-146) are important for nuclear localization and are highly acidic, we thought that it would be interesting to determine whether there is any effect on the subcellular localization of the fusion protein by mutations of this region. Thus, DD at aa 143-144 in this sequence was replaced by KK, thus representing a double point mutation (pTI-19, Clone #14). Indeed, whereas pTI-18 retained its functional NLS, pTI-19 lost the ability to target the nucleus, displaying cytoplasmic distribution (Fig. 5A), suggesting that these two aspartic acids play an important role in nuclear targeting.
Although the fusion proteins from pTI-7, -9, -17, and -18 were localized in the nucleus, they showed little nucleolar green fluorescent staining (Fig. 5). This is in contrast to that of pTI-5, which was mostly localized in the nucleolus, suggesting FIG. 4. Delineation of the sequences for nuclear/nucleolar localization. The sequence KPKKIKTED (aa 148 -156 in pTI-26) at the N terminus of human topo I is sufficient to support the nuclear localization of the fusion protein in HeLa cells, and the sequence of aa 148 -199 is required for nuclear/nucleolar localization. A, images were captured from HeLa cells 16 h post-transfection with the indicated expression plasmids. Fusion proteins from all but one (pTI-13) construct displayed nuclear staining, whereas cytoplasmic distribution of the green fluorescent protein was seen for pTI-13 in which NLS-II was deleted. Note the nucleolar accumulation of the green fluorescent proteins in cells transfected with pTI-5 and pTI-28. B, nucleoli of HeLa cells revealed by immunostaining with an Hsp70-specific antibody. Untransfected HeLa cells were incubated at 43.5°C for 30 min before fixation and immunostained with Hsp70 antibody as detailed under "Experimental Procedures." The same HeLa cells without heat shock (HS) serve as a control. For comparison, HeLa cells transfected with pTI-5 were also subjected to the same heat shock treatment. that the novel NLS either plays no role in nucleolar localization or is not sufficient to do so.
To test whether the novel NLS can replace NLS-II for nucleolar accumulation in the presence of NLS-III and -IV, we made a fusion construct, pTI-28 (Clone #15; see Fig. 1), in which NLS-II was deleted from pTI-5. Transfection of HeLa cells with pTI-28 revealed a similar nucleolar accumulation to that of pTI-5 (Fig. 4A). These results indicated that NLS-II is not required for nuclear/nucleolar localization of the fusion protein if the novel NLS is present with NLS-III and NLS-IV. To further test the effect of the novel NLS on nuclear/nucleolar localization of the full-length topo I, we made three deletion constructs (pTI-31, pTI-32, and pTI-33) as seen in Fig. 6A. pTI-31 carried the full-length topo I with a deletion of NLS-II (⌬ aa 148 -157); pTI-32 carried the full-length topo I with a deletion of the novel NLS (⌬ aa 117-146); and pTI-33 carried the full-length topo I with a deletion of NLS-IV (⌬ aa 188 -198). Although not all pTI-2-transfected cells displayed nucleolar localization (Fig. 6B), which is consistent with our previous results (19), like pTI-2, pTI-31 and pTI-32 resulted in a similar ratio of nucleolar localization (ϳ50%), calculated from 200 transfected cells for each plasmid. These results suggest that the novel NLS can replace NLS-II for nucleolar localization.
To determine the role of NLS-IV in nucleolar localization, we deleted NLS-IV from pTI-28, resulting in pTI-29 (Clone #16; see Fig. 1). Like pTI-27, the fusion protein from pTI-29 displayed only nuclear localization (Fig. 4A), suggesting the importance of NLS-IV for nucleolar localization. Because pTI-29 contained only the N terminus of topo I, we asked whether deletion of NLS-IV has the same effect on nucleolar localization for the full-length protein. Therefore, we transfected HeLa cells with pTI-33. Fluorescence microscopic examinations revealed that pTI-33 had a much lower ratio (ϳ10%) of nucleolar localization compared with pTI-2 or pTI-31 and pTI-32 (Fig. 6B), further supporting the notion that NLS-IV is important for nucleolar localization.
Topo I with a Deletion of NLS-II or the Novel NLS Is as Functional in Vivo as the Wild-type EGFP⅐Topo I-The NLS of topo I has been shown to be essential for its function in vivo (12). Fluorescent microscopy revealed a nuclear localization for both pTI-31 and pTI-32 (Fig. 6B), like the wild-type pTI-2, suggesting that the fusion protein with a deletion of either NLS-II or the novel NLS might still be functional in vivo. To test this hypothesis, we performed a band depletion assay, a method commonly used to test the functionality of topoisomerases (30,31). Topo I inhibitors, such as TPT, stabilize the transient enzyme-DNA complex (9), which traps catalytically active enzyme as large molecular complexes so that the enzyme band is depleted in Western blot in response to such inhibitors. If, however, the enzyme is catalytically inactive, it will not be affected by the drug treatment. We have previously shown that wild-type EGFP⅐topo I (pTI-2) is as catalytically active as the endogenous counterpart (19). We confirmed here that the wildtype EGFP⅐topo I fusion protein (pTI-2) was depleted by treatment of 10 M TPT (Fig. 7A, lane 2, upper band); the same band deletion was observed for the endogenous topo I (Fig. 7A, lane  2, lower band). By contrast, the catalytically inactive EGFP⅐topo I (pY723F-2) (19), in which the active tyrosine residue was replaced by phenylalanine, was not depleted by the same treatment (Fig. 7A, lane 4, upper band), although the endogenous topo I was depleted under the same conditions. It is obvious that the fusion proteins encoded by pTI-31 and pTI-32 were depleted by the TPT treatment (Fig. 7A, lanes 6 and 8). As expected, the fusion protein bands migrated slightly faster than the full-length fusion proteins because of deletions. Thus, our results suggested that the topo I with a deletion of either NLS-II or the novel NLS is catalytically active in vivo in response to TPT.
We also performed growth inhibition assays to test the in vivo functionality of pTI-31 and pTI-32. Because the complexforming topo I inhibitors such as TPT act as a poison by causing DNA damage, the more topo I, or the greater activity, the more sensitive is the cell to the drug (32). As a positive control, overexpression of the wild-type EGFP⅐topo I fusion protein (pTI-2) resulted in a higher sensitivity to the drug, i.e. lower relative cell number compared with the vector control (Fig. 7B). The mutant topo I (pY273F-2) served as a negative control, because we have previously shown that overexpression of this construct has no effect on cell growth (19). As shown in Fig. 7B, overexpression of pTI-31 or pTI-32 also sensitized transfected HeLa cells to TPT. Taken together, both band depletion and growth inhibition assays suggest that, like NLS-II, the novel NLS is functional in vivo.
The Novel NLS Also Functions in Other Mammalian Cells-Finally, to examine whether this novel NLS is specific to HeLa cells, we introduced some of these constructs into nonhuman cells such as Chinese hamster ovary and mouse NIH3T3 cells as well as monkey COS-7 cells. Fluorescence microscopy revealed that all these constructs behaved as in HeLa cells (Fig.  8). For instance, we found nuclear localization for pTI-5 and pTI-7 in all three cell lines. In addition, the nucleolar localization of the fusion protein was seen in all cells transfected with pTI-5, consistent with the results in HeLa cells. Likewise, we detected cytoplasmic distribution in cells transfected with pTI-8 (Fig. 8). These results indicate that the novel NLS identified in the present study also functions in other mammalian cells.

DISCUSSION
Topo I is a well known nuclear protein, and its NLS plays an essential role in nuclear transport. Three novel findings about human topo I subcellular localization are presented in this study. First, although four putative NLSs are suggested at the N terminus of human topo I based on the consensus sequences, we found that only NLS-II (aa 150 -156) is functional in targeting the nucleus, whereas none of the other three NLSs appears to play a role in this aspect. Second, we showed that NLS-IV is important for nucleolar localization. Third, and most importantly, our results provide evidence that a novel NLS exists within aa 117-146. The novel NLS alone is sufficient to direct the enzyme into the nucleus. Moreover, the EGFP⅐topo I fusion protein driven by the novel NLS is catalytically active in vivo in response to the topo I inhibitor TPT. We conclude, therefore, that human topo I carries two independent functional NLSs.
Although NLS-II and the novel NLS function similarly, i.e. they are capable of targeting the nucleus, they have some fundamental differences in terms of their amino acid composition. NLS-II is lysine-rich and apparently falls into the family of the classic monopartite NLSs (14). The novel NLS, by contrast, carries a stretch of acidic amino acids. A deletion involving either DEDDAD (aa 141-146) or KDEPEDDG (aa 117-124) abolished or drastically reduced its nuclear targeting ability. Furthermore, site-directed mutagenesis indicated that two aspartic acids within DEDDAD are important for nuclear targeting. To our knowledge, this is the first experimentally tested example of a stretch of acidic amino acids that can direct a protein into the nucleus.
Apparently, all the peptide motifs for nuclear localization identified to date are rich in basic amino acids (14). By binding to such motifs on nuclear proteins, cognate receptors mediate transport of the nuclear proteins across the nuclear pore complexes (14). In general, these receptors recognize specific nuclear signals of their proteins, although some cross-recognition may occur under some circumstances. For instance, importing nuclear proteins carrying monopartite or bipartite NLSs into the nucleus are mediated by the importin ␣/␤ complex (14), whereas others such as histone H1 are imported into the nucleus by the importin 7/␤ complex (33). Thus, it is likely that the import of topo I into the nucleus driven by NLS-II is through the importin ␣/␤ complex, based  8. Nuclear localization of the EGFP fusion protein (pTI-7), driven by the novel NLS of human topo I, in NIH 3T3, Chinese hamster ovary, and COS-7 cells. Images were captured from cells 16 h post-transfection with pTI-5, pTI-7, and pTI-8, respectively. on the fact that the NLS-II is rich in lysine. However, whereas the nuclear receptor that recognizes the novel NLS is not known at present, the importin ␣/␤ complex probably does not play a role in this aspect. Therefore, what receptor(s) is involved in the novel NLS-mediated nuclear importing remains to be determined.
The presence of multiple functional NLSs is not uncommon for topoisomerases. For instance, the C-terminal domain of topo II␣ has been shown to carry multiple NLSs that function in mammalian cells (34); topo II␤ appears to do so as well (35). Interestingly, the NLSs in topo II␣ do not have equal potency in directing the enzyme into the nucleus (34). By contrast, the novel NLS of topo I identified in this study appears to be as effective as NLS-II, and they are functionally exchangeable in terms of nuclear/nucleolar localization. Thus, a question remains as to why multiple NLSs exist in human topo I. One plausible explanation relates to their potential additive effect. The multiple NLSs will probably make a stronger signal than will a single one. However, because the amino acid compositions for the novel NLS and NLS-II are so different, it would not be surprising to learn that the novel NLS has some yet uncharacterized physiologic role. Thus, further investigation of the novel NLS is warranted.
The nucleolus is a non-membrane subnuclear structure that is believed to be actively involved in rDNA transcription (36) and known to be a major site of ribosome biosynthesis (37). For a protein to accumulate in the nucleolus, it first must be transported into the nucleus. This may explain why many nucleolar localization signals are found along with NLSs (38 -40). Like the other nucleolar localization signals, the nucleolar localization signal of topo I also contains a stretch of basic amino acids. In particular, NLS-IV is important for nucleolar localization and is rich in lysines. The ability of topo I to target the nucleolus suggests that this enzyme may interact with components of the nucleolar complex. Nucleolin is a well known nucleolar protein (41). It has been shown that topo I and nucleolin are localized in a similar region of the nucleolus (42). Moreover, topo I interacts physically with nucleolin (43). Because nucleolin is a multi-function protein, acting as a shuttling protein between cytoplasm and nucleus, it would be reasonable to postulate that such interactions may facilitate its nucleolar localization. Interestingly, aa 166 -210 of topo I represent the minimal sequence for full binding activity to nucleolin (43). Our results indicate that the sequence for nucleolar localization appears to overlap with the nucleolin binding sequence, because aa 157-199 are important for nucleolar accumulation. However, topo I also apparently binds to other nuclear proteins (44 -46). Whether interaction of topo I with these proteins also contributes to its nucleolar accumulation remains to be determined. In addition, because topo I plays an important role in maintaining the integrity of the rDNA locus (47), we cannot exclude the possibility that topo I directly interacts with rDNA in the nucleolus, thus facilitating formation of nucleolar complexes.
In summary, we have identified a novel NLS, rich in acidic amino acids, in human topo I that alone is sufficient to support nuclear localization and that functions in human cells as well as other mammalian cells. Because the novel NLS is so significantly different from the classic peptide NLSs in amino acid composition, it is possible that translocation of the protein into the nucleus via this NLS is also different from the identified pathways (14). Further characterization of this type of NLS might lead to identification of a novel nuclear transport pathway.