Karyopherin-alpha2 protein interacts with Chk2 and contributes to its nuclear import.

Chk2 is a nuclear protein kinase involved in the DNA damage-induced ataxia telangiectasia mutated-dependent checkpoint arrest at multiple cell cycle phases. Searching for Chk2-binding proteins by a yeast two-hybrid system, we identified a strong interaction with karyopherin-alpha2 (KPNA-2), a gene product involved in active nuclear import of proteins bearing a nuclear localization signal (NLS). This finding was confirmed by glutathione S-transferase pull-down and co-immunoprecipitation assays. Of the three predicted Chk2 NLSs, located at amino acids 179-182 (NLS-1), 240-256 (NLS-2), and 515-522 (NLS-3), only the latter mediated the interaction with KPNA-2 in the yeast two-hybrid system, and in particular with its C terminus. Unlike mutations in NLS-1 or NLS-2, which left the nuclear localization of Chk2 unaffected, mutations in NLS-3 caused a cytoplasmic relocalization, indicating that the NLS-3 motif acts indeed as NLS for Chk2 in vivo. Finally, co-transfection experiments with green fluorescent protein (GFP)-Chk2 and wild type or mutant KPNA-2 confirmed the role of KPNA-2 in nuclear import of Chk2.

To find out Chk2-interacting proteins, we have undertaken yeast two-hybrid screens using either full-length or deletion forms of Chk2 as baits and a human cDNA library from HeLa cells. Here we report that Chk2 interacts with karyopherin-␣2 (KPNA-2), a family member of molecules involved in the nuclear translocation of target proteins that carry a nuclear localization signal (NLS) motif characterized by short stretches of basic amino acids (21,22). NLSs can be monopartite if containing a single stretch of basic amino acids, bipartite if containing two stretches of basic amino acids separated by a spacer region. The nuclear translocation of target proteins by KPNA-2 requires the binding to karyopherin-␤, which allows passage of the complex through the nuclear pore (23,24). Once in the nucleus, the binding of Ran-GTP to karyopherin-␤ causes the dissociation with KPNA-2, thereby allowing it to release its cargo and to finally recycle to the cytoplasm via interaction with export proteins (25). Here we show that the C terminus of KPNA-2 binds the NLS-3 sequence of Chk2 and that this interaction has a role in the nuclear localization of Chk2 in cells.

EXPERIMENTAL PROCEDURES
Cells and Transfections-The lymphoblastoid cell line LCL-N from a normal donor was cultured in RPMI 1640 medium (BioWhittaker, Walkersville, MD) supplemented with 15% heat-inactivated fetal calf serum, penicillin (100 units/ml), and streptomycin (100 g/ml), whereas the U2OS osteosarcoma cell line was maintained in Dulbecco's modified Eagle's medium with 10% serum. Cells were transfected by the calcium phosphate method. Irradiations were performed in an IBL437CO instrument equipped with a 137 Ce source emitting a dose of 8 grays/min.
Yeast Two-hybrid Screening-The screening in yeast was performed as described (26). Briefly, Chk2 c-DNA deleted of aa 76 -213 (obtained by digestion of full-length Chk2 c-DNA with AocI) was cloned in the EcoRI-SalI sites of pLexA vector in-frame with the Lex A DNA binding domain and used as a bait to transform the EGY48 yeast strain. A selected clone was then transformed with a HeLa c-DNA library cloned in the pJG4 -5 vector in-frame with the B42 transactivation domain. About 2 ϫ 10 6 clones were obtained and tested for their capacity to grow in a medium without leucine and by the X-gal filter method. Plasmid DNA was then extracted from positive clones, and the c-DNA inserts were amplified by PCR and analyzed by sequencing. To map the regions of interaction between Chk2 and KPNA-2, deletion mutants for the two proteins were prepared by PCR and cloned in the pLexA and in pJG4 -5 vector. The constructs were then used to transform the EGY48 strain and tested for the interaction using a medium containing X-gal or without leucine.
GST Pull-down Assays-The GST-Chk2 recombinant protein was purified from bacteria transformed with the full-length Chk2 cDNA cloned in the EcoRI-NotI sites of pGEX-4T-1 vector. LCL-N cells were lysed in ice-cold ELB buffer (150 mM NaCl, 50 mM Hepes, pH 7.5, 5 mM EDTA, 0.1% Nonidet P-40), and 1 mg of cell extracts were incubated for 3 h at 4°C with 10 g of GST-Chk2 or GST together with glutathione-Sepharose. The beads were extensively washed in lysis buffer, sizefractionated by SDS-PAGE, and immunoblotted for KPNA-2 using a specific monoclonal antibody (Clone 2, Transduction Laboratories).
Immunofluorescence Analysis-U2OS cell lines were transiently transfected with pCDNA3-carrying wild type or mutant forms of FLAGtagged Chk2. After 48 h, the cells were cytocentrifuged on glass slides, fixed with 3.7% paraformaldehyde, permeabilized in 0.2% Triton X-100, and immunostained with the anti-FLAG-M2 monoclonal antibody (Sigma). Binding of primary antibodies was revealed with a fluorescein isothiocyanate-labeled F(abЈ) 2 anti-mouse antibody (Jackson Laboratories). Nuclei were stained with 4Ј,6 diamidino-2-phenylindole, and coverslips were mounted with an anti-fade solution. Slides were analyzed using a Zeiss Axioskop (Oberkochen, Germany) fluorescence microscope and digital imaging.
GFP-Chk2/FLAG-KPNA-2 Co-expression-To establish the role in vivo of Chk2/KPNA-2 interaction, Chk2 cDNA was cloned in the EcoRI-SalI sites of pEGFP-C vector, whereas full-length or deletion mutants of KPNA-2 cDNAs were cloned in the EcoRI-XhoI sites of pcDNA3-FLAG vector. FLAG-KPNA-2 and GFP-Chk2 constructs were co-transfected in U2OS cells at a ratio of 6:1, respectively, and 48 h later, the subcellular localization of GST-Chk2 was analyzed by fluorescence microscopy.
In Vitro Chk2 Kinase Assays-Kinase reactions were performed using a catalytically active GST-Chk2 recombinant protein (2 g) incubated for 30 min at 30°C with 1 g of either GST-KPNA-2 or GST-Cdc25C substrate (the latter used as positive control) in 30 l of kinase buffer (20 mM Tris-HCl, 75 mM KCl, 5 mM MgCl 2 , 0.5 mM EDTA, 2 mM dithiothreitol, 50 M ATP, and 15 Ci of [␥-32 P]ATP). The reaction products were separated by SDS-PAGE and autoradiographed. The gels were then stained with Coomassie Blue to visualize the amount of loaded substrate per lane.

RESULTS
Yeast Two-hybrid Screenings for Chk2 Interactors-The yeast two-hybrid system was employed to search for Chk2 interactors. The bait consisted of Chk2 deleted of aa 76 -213 rather than full-length Chk2 (Fig. 1, a and b), because in preliminary tests, the latter partially inhibited the growth and consequently the transformation efficiency of yeast (data not shown). Approximately two million colonies were screened by the X-gal filter method and for their capacity to grow without leucine. Of 45 sequenced positive clones, 33 encoded the full length or fragments of the karyopherin-␣2 (KPNA-2) protein, the shortest of which represent aa 285-529. KPNA-2 (21, 22) is a nuclear transporter of proteins bearing an NLS, structurally characterized by an N terminus importin-␤ binding (IBB) domain necessary for nuclear translocation, eight armadillo repeats involved in the binding and recognition of NLSs, and a C-terminal acidic domain (Fig. 1c).
Chk2 Interacts with KPNA-2 Both in Vitro and in Vivo-The physical interaction between Chk2 and KPNA-2 was initially examined in pull-down assays in which extracts from normal lymphoblastoid cells (LCL-N) were incubated with Sepharosebound recombinant GST-Chk2. Under these conditions, the endogenous KPNA-2 bound GST-Chk2 but not GST (Fig. 2a), thus confirming the interaction seen in yeast. To verify this interaction in vivo, we determined whether Chk2 immunoprecipitates with KPNA-2. The results (Fig. 2b, lane 3) showed that Chk2 from LCL-N cells co-precipitated with KPNA-2, and the specificity of this finding was demonstrated by the absence of a KPNA-2 signal when immunoprecipitating with a control antibody (Fig. 2b, lane 2). As Chk2 is involved in the response to DNA damage (1, 2), we tested the effect of IR on the interaction between Chk2 and KPNA-2. In cells treated with 10 grays of IR and harvested 3 h later, Chk2 co-precipitated KPNA-2 to the same extent as in untreated cells (Fig. 2b, lane  7). As Chk2 protein could not be detected in KPNA-2 immunoprecipitates (possibly because the epitope recognized by the monoclonal anti-KPNA-2 antibody sits in the region binding to Chk2, see below), we transiently co-transfected U2OS cell lines with FLAG-KPNA-2 and HA-Chk2 constructs and analyzed Chk2 in KPNA-2 immunoprecipitates obtained with anti-FLAG antibody. Under these conditions, we were able to find HA-Chk2 protein associated with KPNA-2 (Fig. 2c, lane 4). Altogether, these results demonstrate an interaction between Chk2 and KPNA-2 in human cells.
Identification of the Sites of Interaction between Chk2 and KPNA-2-To map the region of interaction between Chk2 and KPNA-2, the full-length and deleted versions of these proteins were tested in the yeast two-hybrid system. Initially, three Chk2 deletions were made, carrying the TQ/SQ-rich region plus the FHA domain (Chk2 1-209 ), the kinase domain only (Chk2 210 -543 ), or the C terminus (Chk2 450 -543 ). Besides fulllength Chk2, Chk2 210 -543 and Chk2 450 -543, but not Chk2 1-209 , interacted with full-length KPNA-2 (Fig. 3a), indicating that the KPNA-2 binding site is located at aa 450 -543 of Chk2. Because the shortest KPNA-2 sequence found in the initial two-hybrid screening encoded the C-terminal 244 amino acids, it was clear that this sequence mediated the interaction with Chk2. Thus, to narrow down the region of interaction, mutants deleted of the acidic domain (KPNA-2 1Ϫ460 ) or acidic domain plus the two terminal armadillo repeats (KPNA-2 1-369 ) Chk2 Nuclear Import by KPNA-2 were tested in yeast against full-length Chk2. As neither KPNA-2 1-460 nor KPNA-2 1-369 interacted with Chk2 (Fig. 3b), it can be concluded that aa 461-529 of KPNA-2 are required for binding to Chk2. The expression of KPNA-2 1-460 and KPNA-2 1-369 proteins in yeast was demonstrated by Western blot with an anti-LexA antibody. Of note, these mutant proteins were, in contrast to full-length KPNA-2, unreactive with the anti-KPNA-2 antibody (Fig. 3c), suggesting that aa 461-520 contain the anti-KPNA-2 binding epitope.
Identification of Chk2 NLS-As KPNA-2 mediates the nuclear import of proteins carrying the NLS sequence (22), we determined the possible involvement of such an NLS of Chk2 in the interaction with KPNA-2. A PSORT II computer program analysis of Chk2 sequence identified three putative NLSs, a monopartite NLS-1 (KRRP, aa 179 -182), a bipartite NLS-2 (RKTCKKVAIKIISKRKF, aa 240 -256), and a monopartite NLS-3 (PSTSRKRPR, aa 515-522) (Fig. 4a). To study the function of these putative NLSs, the localization of various Chk2 NLS mutants expressed in transiently transfected U2OS cells was assessed by immunofluorescence. Like full-length FLAG-Chk2, FLAG-Chk2 deleted of aa 76 -213 (and therefore of NLS-1) was exclusively expressed in the nucleus, thus suggesting that NLS-1 is dispensable for Chk2 nuclear localization. By contrast, a cytoplasmic localization became evident (Fig. 4b) when the lysine 520 and arginine 521 of NLS-3 were substituted with alanine (Table I). These changes are thought to destroy the putative consensus sequence. Mutations in NLS-2 motif had no effect on Chk2 nuclear localization (Table I and Fig. 4b). In accordance with these findings, KPNA-2 failed to interact in yeast with Chk2 mutated in NLS-3 (Fig. 4c). These results underscore the role of NLS-3 in the interaction with KPNA-2 and nuclear localization of Chk2.
Function of the Chk2/KPNA-2 Interaction in Vivo-To determine the in vivo function of Chk2/KPNA-2 association, we examined the effect of ectopically expressed KPNA-2 on the nuclear translocation of Chk2. U2OS cells were transiently co-transfected with plasmids encoding GFP-Chk2 and wild type or mutant KPNA-2 and examined by fluorescence microscopy. Of the fluorescent cells detected in co-transfections with GFP-Chk2 and an empty vector, 51% showed a nuclear labeling, whereas the remaining, in addition, showed a cytoplasmic labeling whose intensity was in some cases much weaker than the nuclear fluorescence. After co-transfection with wild type KPNA-2, the number of cells with nuclear GFP-Chk2 (Fig. 5) rose to 66%, whereas after co-transfection with KPNA-2 deleted of either the Chk2-interacting region (KPNA-2 1-460 ) or the IBB domain (KPNA-2 70 -529 , unable to translocate to the nucleus), the number dropped to 41%, indicating a role for KPNA-2 in the nuclear translocation of Chk2 kinase.
Since certain kinases modulate their nuclear import by direct phosphorylation of KPNA-2 (28), we assessed whether KPNA-2 is a target substrate for Chk2 kinase. For this purpose, in vitro kinase assays were performed using catalytically active recombinant GST-Chk2 and full-length GST-KPNA-2 as a substrate. However, whereas GST-Chk2 phosphorylated, as expected, Cdc25C, it failed to phosphorylate KPNA-2 (Fig. 6a), thus excluding a phosphorylative event in the interaction of these proteins. This finding is concordant with the lack of a Chk2 consensus substrate motif within KPNA-2 (as determined by Scansite analysis) and with the normal nuclear localization of ectopically expressed Chk2 kinase-dead (Fig. 6b). DISCUSSION Two-hybrid screenings in yeast have enabled us to identify a new physical and functional interaction between Chk2 kinase and KPNA-2, a family member of proteins involved in the active transport from the cytoplasm to the nucleus of cargo proteins containing an NLS (21,22). Two types of NLS motifs exist; one, exemplified by the SV40 large T antigen PKKKRKV, contains a short stretch of basic amino acids (monopartite), and the other, first reported in nucleoplasmin (KRPAATKK-AGQAKKKKLDK), contains two clusters of basic amino acids separated by a spacer region of 10 -12 amino acids (bipartite). To cross the nuclear membrane and deliver its NLS-tagged cargo, KPNA-2 has to bind karyopherin-␤ (23,24). In the nucleus, these complex components, including the cargo, are released, allowing the recycling of karyopherins to the cytoplasm (25).
We confirmed the interaction seen in yeast between Chk2 and KPNA-2 also in pull-down assays between recombinant GST-Chk2 and endogenous KPNA-2 from total cell lysates. We have additionally demonstrated that Chk2 immunocomplexes from unirradiated or irradiated cells, the latter expressing hyperphosphorylated Chk2 (1, 2), contain KPNA-2. The detection by mass spectrometry of KPNA-2 in FLAG-Chk2 immuno-   complexes extracted from cells 2 lends further support to these results. Although we were unable to co-precipitate Chk2 with anti-KPNA-2 antibodies, reasonably because the epitope that it recognizes maps within the region of interaction with Chk2 (as we have shown in Western blots of KPNA-2 deletion mutants expressed in yeast) and may thus be unavailable when bound to Chk2, we could nevertheless demonstrate Chk2 in FLAG-KPNA-2 complexes immunoprecipitated with an anti-FLAG antibody from cells ectopically expressing FLAG-KPNA-2 and HA-Chk2.
Having confirmed the interaction between these two molecules, we next determined whether this was mediated by the NLSs. As the PSORT analysis of Chk2 cDNA sequence scored three NLS motifs localized at aa 179 -182 (NLS-1; monopartite), aa 240 -256 (NLS-2; bipartite), aa 515-522 (NLS-3; monopartite), we examined in the two-hybrid system various missense and/or deletion mutants of Chk2 and KPNA-2 and found that only the C-terminal aa 450 -543 of Chk2, which comprises NLS-3, associates with KPNA-2.
To determine the functional role in vivo of the three putative NLSs, the intracellular distribution of FLAG-tagged Chk2 constructs transiently transfected in U2OS cells was analyzed by immunofluorescence. We have shown that mutations in NLS-3, but not in NLS-1 or NLS-2, cause a cytoplasmic redistribution of Chk2. This finding, together with the fact that NLS-3 mutations disrupt the interaction with KPNA-2 in yeast (see above), underscores the importance of NLS-3 for the binding to KPNA-2 and nuclear translocation of Chk2.
KPNA-2 contains an IBB, eight armadillo repeats typically involved in recognition and binding to NLSs, and a C-terminal acidic domain (22). We have established that the C-terminal aa 461-529 of KPNA-2 are indispensable for binding to Chk2. This result is in apparent contrast with the finding that the armadillo repeats mediate the binding to NLS (22), but concordant with others showing that proteins such as RAG-1, BSAP, Stat1, and the export protein CAS interact with the C-terminal region of the karyopherin family members (29 -32). It is speculated that the use of C-terminal region for interactions allows karyopherins to bind more than one molecule at the time, giving rise to a scaffold for protein complexes (31).
To establish the role played by KPNA-2 in the nuclear transport of Chk2, the subcellular localization of GFP-Chk2 coexpressed with wild type or mutant versions of KPNA-2 was analyzed by fluorescence microscopy. We have shown that only 51% of GFP-Chk2 cells have a nuclear localization, the remaining exhibiting both nuclear and cytoplasmic fluorescence of various intensities. It is worth noting that this cytoplasmic localization was not observed in cells expressing FLAG-Chk2, suggesting that the conformation of GFP-Chk2 might impair its nuclear transport. Interestingly, when co-expressed with wild type KPNA-2, the fraction of cells with nuclear GFP-Chk2 fluorescence rose to 66%, whereas when co-expressed with KPNA-2 mutated in the Chk2 binding site or in the IBB domain (thus unable to reach the nucleus), this number fell to 41%. Altogether, these results indicate that overexpression of wild type KPNA-2 enhances the nuclear localization of GFP-Chk2, whereas the mutant forms of KPNA-2 do not significantly modify the subcellular distribution of Chk2. These data may not be unexpected since there is no evidence indicating that the KPNA-2 mutants could act as dominant negative. Moreover, given the redundancy of the nuclear import/export pathways, other members of the karyopherin isoforms (33) could possibly complement KPNA-2 defects. This hypothesis is supported by experiments showing that the selective inhibition of KPNA-2 protein expression by small interfering RNA treatment only induces a 2-fold increase in cells with cytoplasmic GFP-Chk2, rather than an abrogation of nuclear GFP-Chk2 import, 3 indicating that other karyopherin family members could partake in Chk2 nuclear import, like certain ribosomal and core histone proteins whose nuclear transport is mediated by at least five different importins (34). It should be noticed, however, that additional two-hybrid screenings in yeast with various Chk2 baits and different cDNA libraries have never revealed interactions with the karyopherin family other than KPNA-2.
As the Itk kinase stimulates its nuclear import through the phosphorylation of KPNA-2 (28), we tried to establish whether Chk2 operates in a similar way. However, we were unable to demonstrate, at least in vitro, any phosphorylation of GST-KPNA-2 by Chk2, indicating that the nuclear transport of Chk2 by KPNA-2 is independent of its enzymatic activity, which actually concurs with the appropriate nuclear localization of the kinase-inactive form of Chk2.
In conclusion, we have shown that Chk2 kinase carries a functional NLS located at aa 515-522 indispensable for KPNA-2 binding and that this interaction contributes to Chk2 nuclear translocation. As proteins of the nuclear import machinery, including KPNA-2, appear to have a role in other physiological processes, e.g. mitotic spindle formation, RNA export, and chromatin structure (35)(36)(37), we cannot exclude that Chk2-KPNA-2 association solely regulates Chk2 nuclear import. Enzymatically active recombinant GST-Chk2 was tested for its capacity to phosphorylate KPNA-2 or GST-Cdc25c (the latter used as positive control). No KPNA-2 phosphorylation was detected (a). Normalization with Coomassie Blue staining is shown. b, normal nuclear localization of FLAG-Chk2-KD (kinase-dead) transiently expressed in U2OS cells.