Nuclear Localization of the Cell Cycle Regulator CDH1 and Its Regulation by Phosphorylation*

The anaphase-promoting complex activated by CDC20 and CDH1 is a major ubiquitination system that controls the destruction of cell cycle regulators. Exactly how ubiquitination is regulated in time and space is incompletely understood. Here we report on the cell cycle-dependent localization of CDH1 and its regulation by phosphorylation. CDH1 localizes dynamically to the nucleus during interphase and to the centrosome during metaphase and anaphase. The nuclear accumulation of CDH1 correlates with a reduction in the steady-state amount of cyclin A, but not of cyclin E. A nuclear localization signal conserved in various species was identified in CDH1, and it sufficiently targets green fluorescent protein to the nucleus. Interestingly, a CDH1-4D mutant mimicking the hyperphosphorylated form was constitutively found in the cytoplasm. In further support of the notion that phosphorylation inhibits nuclear import, the nuclear localization signal of CDH1 with two phospho-accepting serine/threonine residues changed into aspartates was unable to drive heterologous protein into the nucleus. On the other hand, abolition of the cyclin-binding ability of CDH1 has no influence on its nuclear localization. Taken together, our findings document the phosphorylation-dependent localization of CDH1 in vertebrate cells.

The timely and orderly progression of eukaryotic cell division cycle is precisely controlled through ubiquitination-induced proteolysis. The anaphase-promoting complex (APC) 1 functions as a major ubiquitin ligase in the cell, and it governs anaphase onset, mitotic exit, and G 1 events (1)(2)(3)(4). APC is a large complex of multiple subunits. It catalyzes the formation of polyubiquitin chains on cyclins and other cell cycle regulators and thereby targets them for degradation by proteosome. The components of APC are highly conserved in organisms ranging from yeast to humans. The rise and fall of APC activity is stringently regulated in a cell cycle-dependent manner. It plateaus at the transition from metaphase to anaphase, remains high throughout G 1 , and drops in S/G 2 until early mitosis (5). One important mechanism for the regulation of APC is through phosphorylation of its subunits (6 -9). In addition, CDC20 and CDH1 can act as activators and specificity factors of APC (10 -15).
CDC20 and CDH1 are WD40 repeat proteins that associate transiently with and activate APC in a substrate-specific fashion. Many substrate proteins targeted by the CDC20-and CDH1-activated APC share a sequence motif known as destruction box (5,16). Another motif termed the KEN box is recognized by CDH1-activated APC and has been found in substrates including CDC20, NEK2, HSL1, mitotic cyclins, and securin (17)(18)(19)(20). In addition, a distinct recognition domain called the A box has been shown to be required for CDH1-dependent ubiquitination and degradation of the Aurora-A kinase (21). Exactly how CDC20 and CDH1 activate APC remains to be fully understood. Existing evidence suggests that they mediate the temporal and spatial activation of APC through direct interaction with different sets of substrate proteins (19,22,23). CDC20 and CDH1 are tightly regulated during the cell cycle. The steady-state amounts of CDC20 and CDH1 transcripts peak in G 2 -M and fall dramatically at the entry into G 1 (5, 8, 24 -26). A similar expression pattern has also been described for CDC20 protein (5,24). In contrast, the abundance of CDH1 protein was shown to be constant in two earlier studies (5,25). However, experiments conducted with a highly specific antibody against CDH1 have demonstrated the oscillation of CDH1 protein as cells enter and exit mitosis (8). CDH1 is ubiquitously expressed in differentiated tissues including postmitotic neurons (27). Interestingly, a recent genetic screen in Caenorhabditis elegans has revealed a role for CDH1 in controlling cell proliferation (28). In line with this, a significant reduction of CDH1 has been documented during the malignant progression of a B-lymphoma cell line (29).
We have previously identified and characterized human mi-totic checkpoint proteins MAD1 and MAD2 (51)(52)(53). Given that CDH1 is a downstream effector of the mitotic checkpoint, we set out to characterize human CDH1 in cultured cells. In this study, we demonstrate that human CDH1 localizes to the centrosome and the nucleus. Moreover, the nuclear import of CDH1 is regulated by phosphorylation in the vicinity of the nuclear localization signal (NLS). Our findings implicate a complex cell cycle-dependent regulation of human CDH1 activity in temporal and spatial dimensions.
Plasmids-Human CDH1 cDNA with a complete coding region (Gen-Bank TM AF083810) was assembled from expressed sequence tags obtained from the American Type Culture Collection. Expression vector for human CDH1 (pHACDH1) was derived from a previously described plasmid pHA containing SV40 promoters and enhancers (54). pHA was also used to express the three CDH1 mutants CDH1-4A, CDH-4D, and CDH1-AAA. In CDH1-4A and CDH1-4D, serines 40, 151, 163, and threonine 121 were substituted with alanines and aspartates, respectively through PCR-based site-directed mutagenesis. CDH1-AAA was constructed by replacing the Arg-Val-Leu triplet (amino acids 445-447) with three alanines. Sequences of all cDNAs coding for CDH1 mutants were determined to verify that they have the desired substitutions and not any undesired changes.
To express a chimeric green fluorescent protein (GFP) with the CDH1 nuclear localization signal (NLS) at the carboxyl terminus (GF-PNLS), we PCR-amplified a 104-bp DNA fragment coding for amino acids 151-178 of human CDH1 and subcloned this fragment into plasmid pEGFP-C1 via restriction sites XhoI and BamHI (Clontech). The pEGFP-C1 vector was also used to express GFP fusion proteins GF-PCDH1, GFPNLS-2A, and GFPNLS-2D. In GFPNLS-2A and GFPNLS-2D, serines 151 and 163 near the NLS of CDH1 had been replaced by alanines or aspartates, respectively. Details for PCR primers and for plasmid construction are available upon request.
Confocal Microscopy-Laser scanning confocal microscopy was performed on a Zeiss Axiophot microscope as detailed elsewhere (51,55). Dual immunofluorescent detection was achieved with primary antibodies from different species and pre-adsorbed species-specific secondary antibodies conjugated to different fluorochromes: Cy5-conjugated goat anti-mouse immunoglobulin G (Zymax) and fluorescein-conjugated goat anti-rabbit immunoglobulin G (Zymax). GFP experiments were carried out as previously described (56).

RESULTS
Cell Cycle-dependent Localization of CDH1-The spatial regulation of APC activity is an integral part of cell cycle control. The APC components and substrates, including cyclins and CDKs, perform their specific functions at restricted subcellular locales (57,58). To shed light on the spatial control of CDH1 activity, we investigated the subcellular localization of endogenous CDH1 protein. We raised two specific antisera against CDH1 (␣-CDH1N and ␣-CDH1C) in rabbits and stained HeLa cells for CDH1 using affinity-purified antibodies. As a first step, we verified the specificity of the anti-CDH1 antibodies by immunoblotting. An example of this experiment is shown in Fig. 1A, in which HeLa cell extracts were immunoblotted with either ␣-CDH1N or ␣-CDH1N preincubated with excessive amount of immunizing peptide. A discrete protein band of 55 kDa reactive to ␣-CDH1N was observed (Fig. 1A, lane 1), and this band disappeared if the antibodies were pre-absorbed with the peptide (lane 2). Next HeLa cells were stained with either ␣-CDH1N or pre-absorbed ␣-CDH1N (Fig. 1B). The endogenous CDH1 in HeLa cells was predominantly in the nucleus (Fig. 1B, panel 1), and the peptide blocking experiment corroborated the specificity of this staining pattern (panel 2).
To assess the dynamics of CDH1 localization during the cell cycle, we co-stained CDH1 and the cell cycle marker ␣-tubulin (Fig. 1C). During the interphase, the endogenous CDH1 in HeLa cells localized to the nucleus in a nucleoli-excluded pattern. During this time CDH1 and ␣-tubulin resided in distinct subcellular compartments with minimal co-localization ( Fig.  1C, panels 1-3). Notably, as the cell entered metaphase, CDH1 concentrated into the centrosomes (Fig. 1C, panels 4 -6). Thus a significant co-localization of CDH1 and ␣-tubulin was noted. CDH1 remained in the centrosomes as the cell committed into anaphase (Fig. 1C, panels 7-9). The subcellular localizations of human CDH1 are highly reproducible and are based on two different anti-CDH1 antibodies ␣-CDH1N and ␣-CDH1C (data not shown). The dynamic changes in the localization of CDH1 provide opportunities for a precise temporal and spatial regulation of APC activity during the cell cycle.
Nuclear Accumulation of CDH1 Correlates with Degradation of Cyclin A-To further investigate the localization of CDH1 and the functional consequence of CDH1 overexpression in cultured cells, we transiently transfected HeLa cells with pHACDH1, an expression plasmid for hemagglutinin (HA)tagged CDH1. The transfected cells were stained with ␣-CDH1C ( Fig. 2A, panel 1) or anti-HA (␣-HA, panel 2) antibodies. ␣-CDH1C reacted with both native (Fig. 2A, panel 1, cells without arrow) and exogenously introduced (cell with an arrow) CDH1 protein, while ␣-HA recognized the introduced CDH1 only ( Fig. 2A, panel 2, cell with an arrow). In addition, we also examined the localization of CDH1 protein fused to GFP (GFPCDH1) in transiently transfected cells ( Fig. 2A,  panel 3). Notably, a nuclear localization pattern of CDH1 was consistently seen with ␣-CDH1C, ␣-HA, or GFPCDH1. This lends further support to the notion that human CDH1 is a nuclear protein.
The nuclear localization of CDH1 predicts that it might mediate the APC-dependent proteolysis of nuclear regulators of the cell cycle. Based on this reasoning, we asked how CDH1 might influence APC activity in the nucleus. We overexpressed HA-CDH1 in HeLa cells and co-stained for CDH1 and two nuclear cyclins, cyclin A and cyclin E (Fig. 2B). In this regard, cyclin A is a known substrate of the CDH1-activated APC (59 -61), whereas cyclin E is degraded primarily through the SCF complex (62, 63), but not through APC. Consistent with previous studies (57,64), ambient cyclin A and cyclin E are predominantly nuclear (Fig. 2B, panels 2 and 5). As expected, CDH1 co-localized with cyclin A and cyclin E to the nucleus (Fig. 2B, panels 3 and 6). If nuclear CDH1 is functional, it should induce APC-mediated destruction of cyclin A. In contrast, nuclear CDH1 might not affect the steady-state level of cyclin E, which is normally degraded by SCF and not by APC (62,63). Indeed, the forced overexpression of CDH1 correlated with the diminution of nuclear cyclin A (Fig. 2B, panel 2, compare cell with an arrow with cells without arrows), but not cyclin E (panel 5). Thus, human CDH1 functions as an APC activator in the nucleus.
Identification of a Nuclear Localization Signal in CDH1-In the process of identifying the NLS in CDH1, we noted that a CDH1 protein lacking the amino acid sequences encoded by exons 6 and 7 of human CDH1 gene localized predominantly to the cytoplasm. 2 This raises the possibility that exons 6 and 7 of the CDH1 gene could encode an NLS. Indeed, a closer examination of the amino acid sequences encoded by exons 6 and 7 identified a 28-residue stretch (amino acids 151-178) rich in arginines and lysines (Fig. 3A, 9 out of 28 residues are positively charged as highlighted by asterisks). This putative NLS is highly conserved from yeast to human and it also contains two conserved CDK phosphorylation sites (serines 151 and 163; Fig. 3A, highlighted by #).
To determine whether this is a functional NLS, we expressed in HeLa cells a GFPNLS fusion protein with the putative NLS from CDH1 at the carboxyl terminus. As a control, the GFP protein overexpressed in HeLa cells was diffusely found in both the nucleus and the cytoplasm (Fig. 3B, panel 1 and Fig. 3C, panels 1-3; compare transfected cells with an arrow to untransfected cells without arrows). In contrast, the GFPNLS protein concentrated into the nucleus (Fig. 3, B (panel 2) and C (panels 4 -6); co-localization of GFPNLS and the nuclear marker NuMA are in yellow), suggesting that the NLS from CDH1 sufficiently drove the heterologous GFP protein into the nuclear compartment. Thus, amino acids 151-178 of CDH1 harbor a functional NLS.
Nuclear Import of CDH1 Is Regulated by Phosphorylation-The phosphorylation of CDH1 by CDKs leads to inhibition of its APC-activating activity in cells (7,(31)(32)(33)(34). In this regard, CDH1 mutants CDH1-4A and CDH-4D with the serine or threonine residues at the four CDK phosphorylation sites replaced by alanines and aspartates, respectively, are very useful for the study of CDH1 because they mimic the constitutive hypo-and hyperphosphorylation states (33,60). Considered together with the fact that CDH1 and its substrates such as cyclin A and cyclin B1 are all in the nucleus, it would be of interest to see whether phosphorylation of CDH1 might influence its subcellular localization. Importantly, a more recent study in budding yeast has suggested that phosphorylation-dependent nuclear export of CDH1 contributes to efficient inactivation of APC (65). In light of this new finding, we asked whether phosphorylation of CDH1 might also regulate its nuclear import or export in cultured human cells.
We expressed HA-tagged CDH1-4A and CDH1-4D proteins in HeLa cells and assessed their localization by immunofluorescence microscopy with an ␣-HA antibody. Notably, CDH1-4A mimicking the hypophosphorylated form was almost constitutively nuclear as shown in its precise co-localization with the nuclear marker NuMA (see Fig. 4A  example representative of 79% of the transfected cells). In sharp contrast, the majority of CDH1-4D protein, which mimics the hyperphosphorylated form, was found primarily in the cytoplasm, and its staining pattern overlaps minimally with that of NuMA (see Fig. 4A, panels 4 -6 for an example representative of 74% of the transfected cells). Thus, phosphorylation of CDH1 correlates with cytoplasmic retention of the protein. Plausibly cytoplasmic sequestration of CDH1 could be a net result caused by enhancement of nuclear export and/or blockade of nuclear import.
Consistent with the previous finding that the expression of CDH1-4A but not of CDH1-4D in U2-OS cells resulted in the persistent degradation of cyclin B1 (33), we observed that the accumulation of CDH1-4A correlated with reduction of cyclin A level in the nucleus (Fig. 4B, panels 1-3; steady-state amounts of cyclin A declined in 93% of the transfected cells). This effect is more drastic than that induced by CDH1 wild-type (Fig. 2B, panels 1-3; 68% of the transfected cells had the phenotype). On the contrary, the increase of CDH1-4D in the cytoplasm did not lead to destruction of nuclear cyclin A (Fig. 4B, panels 4 -6). Our interpretation to these data is that the cytoplasmic retention of CDH1-4D is causally linked to the loss of its APCstimulatory activity. This is generally in line with the notion that the constitutively nuclear CDH1-4A is dominantly active whereas constitutively cytoplasmic CDH1-4D is a dominant negative form.
We noted that two of four conserved CDK phosphorylation sites targeted in the CDH1-4A and CDH1-4D mutants are located in the vicinity of the NLS of CDH1 (Fig. 3A). Thus, one likely mechanism for the regulation of CDH1 localization is through phosphorylation on serines 151 and 163 near the positively charged residues in NLS. To test this hypothesis, we replaced both serines in the GFPNLS with either alanines or aspartates. Then we separately expressed GFP, GFPNLS as well as the two GFPNLS-2A and GFPNLS-2D mutants in HeLa cells. The GFP-expressing cells gave a bright and homogenous fluorescent signal throughout the cells, suggesting that GFP is evenly distributed in both the cytoplasm and the nucleus (see Fig. 5A, panel 1 for one example). Thus the ratio of GFP fluorescence in the nuclear versus in the cytoplasm was close to 1 (Fig. 5B). Consistent with data shown in Fig. 3B, the ratio of GFPNLS-specific fluorescent signals in the nuclear versus in the cytoplasm was greater than 4 ( Fig. 5B), suggesting that GFPNLS targeted the GFP, which was otherwise found in both the nucleus and the cytoplasm, into the nucleus. Notably, CDH1-4D is constitutively cytoplasmic (Fig. 4, panels 4 -6). As such, GFPNLS-2D should also localize to the cytoplasm if phosphorylation near the positively charged residues in the NLS would affect the nuclear localization of CDH1. Indeed, GF-PNLS-2D was sufficiently retained in the nucleus-excluded region of the cells (see Fig. 5A, panels 3 and 4 for examples). The ratio of GFPNLS-2D-specific fluorescent signals in the nuclear versus in the cytoplasm is therefore lower than 0.5 (Fig.  5B). In line of this, the nuclear accumulation of GFPNLS-2A was even more dramatic than the wild-type GFPNLS (Fig. 5, A,  panel 2 and B). These results consistently support the notion that phosphorylation on serines 151 and 163 adjacent to the positively charged residues in NLS sufficiently impedes the nuclear import of CDH1.
Influence of Cyclin Binding on Nuclear Localization of CDH1-The phosphorylation of CDH1 by cyclin A/CDK2 has an important regulatory role in orchestrating S-phase progression with mitosis driven by stabilization of cyclin B1 (33). This functional interplay between cyclin A/CDK2 and CDH1-activated APC is determined by a conserved cyclin-binding domain in CDH1 (60). Above we have shown that CDH1 co-localizes with cyclin A to the nucleus (Fig. 2B). In this scenario, one interesting question remains unanswered as to whether the interaction with cyclin A might influence the nuclear localization of CDH1.
To address this issue, we employed one CDH1 mutant called CDH1-AAA, in which the Arg-Val-Leu triplet in the most conserved cyclin A-binding motif has been substituted with three alanines. We expressed HA-tagged CDH1-AAA in HeLa cells and co-stained for both CDH1 and cyclin A (Fig. 6). Interestingly, this CDH1-AAA mutant, which is unable to interact with cyclin A as previously described (60), retained the ability to localize to the nucleus (Fig. 6, panels 1 and 4, cells with arrows). In this setting, a significant co-localization of CDH1-AAA and cyclin was observed (Fig. 6, panels 3 and 6). In contrast to the nuclear accumulation of wild-type CDH1, which led to destabilization of cyclin A (Fig. 2B), the abundant CDH1-AAA in the nucleus did not cause increased degradation of nuclear cyclin A (Fig. 6, panels 2, 3, 5, and 6). In agreement with previous findings (60), these results indicate that binding to cyclin A is crucial for CDH1 stimulation of APC-mediated proteolysis of cyclin A itself. However, the cyclin binding did not influence the subcellular localization of CDH1. DISCUSSION Here we documented the phosphorylation-dependent subcellular localization of human APC regulator CDH1. CDH1 local- ized to nuclei during interphase and to centrosomes at mitosis (Fig. 1). The nuclear accumulation of CDH1 induced the APCmediated degradation of cyclin A, but not cyclin E, in cultured cells (Fig. 2). A functional NLS rich in positively charged residues was identified in CDH1, and it sufficiently targeted heterologous GFP protein into the nucleus (Fig. 3). Phosphorylation in the vicinity of the NLS led to constitutive cytoplasmic retention of CDH1 protein (Figs. 4 and 5). However, the disruption of cyclin-binding motif in CDH1 had no influence on nuclear targeting (Fig. 6). Our work suggests an additional level of regulation for APC activity in the spatial dimension.
Nuclear and Centrosomal Functions of CDH1-The subcellular localization pattern (Fig. 1) implicates CDH1 as a nuclear and centrosomal activator of APC. In line with this, the constitutive expression of CDH1 in the nucleus correlates with increased destabilization of cyclin A (Fig. 2B). In order to function as a substrate-specific activator of APC, CDH1 has to co-localize with APC components, APC substrates as well as regulators of CDH1. The nuclear localization of APC subunits has been documented (58,66). Meanwhile, the nuclear trafficking of CDKs and cyclins, including cyclin A and cyclin B1, has important roles in cell cycle regulation (57,64). In addition, the CDH1 phosphatase CDC14, which is sequestered in the nucleolus in budding yeast (32), is a nuclear protein in human cells (67). Thus, the nucleus appears to be one of the major subcellular compartments in which CDH1 performs its APC-stimulatory functions.
The centrosomal localization of CDH1 during metaphase and anaphase (Fig. 1) also has important implications in the regulation of APC. The centrosome is a major microtubule-organizing center that governs spindle assembly, spindle bipolarity, chromosome segregation, and mitotic progression (68,69). It duplicates only once and undergoes characteristic changes during the cell cycle. As yet the signaling pathway that regulates centrosome duplication and links the centrosome cycle to mitotic progression remains to be elucidated. Notably, many of the substrates, regulators or subunit proteins of the APC localize to the centrosome. These include cyclin B1 (57), CDC2/ CDK1 (70), CDC16/CDC27 (71), CDC20 (5,40,72), EMI1 (73), MAD1/MAD2 (40,51), BUB2 (47,48), Aurora-A kinase (21,74), and NEK2 (75). Among them CDC20, NEK2, cyclin B, and Aurora-A have been characterized as substrates of the CDH1activated APC (17,21,76). Results from this and other studies suggest that the centrosome performs a role as a relay point for signal integration, amplification, and distribution in the regulation of APC activation.
Inhibition of CDH1 Nuclear Import by Phosphorylation-Protein trafficking between the nucleus and the cytoplasm is fundamentally important to cell regulation (77). As such, nuclear import and export are pivotal in orchestrating the activities of the key regulators of the cell cycle (57). One mechanism for spatial control of cell cycle is through the retention of particular proteins in the cytoplasm or in the nucleus, thereby preventing them from physical contact with their substrates or partners. In another perspective, some proteins, such as cyclin A and cyclin E, shuttle continuously between the nucleus and the cytoplasm (64). Their apparent steady-state localization as shown by immunofluorescence microscopy reflects the relative rates of nuclear import and export.
We showed that CDH1 is apparently a nuclear protein (Figs. 1 and 2) and that it harbors an NLS (Fig. 3), whose function is regulated by phosphorylation adjacent to the positively charged amino acid residues (Figs. 4 and 5). The co-localization of CDH1 with its substrates, partners and regulators in the nucleus and the centrosome suggests that it likely functions as a nuclear and centrosomal activator of APC. Our findings that show that constitutive expression of CDH1 correlate with the destabilization of nuclear cyclin A (Fig. 2) lend some support to this notion. However, our work does not exclude the possibility that CDH1 is constantly shuttling between the nucleus and the cytoplasm. One recent study in budding yeast suggests a dynamic localization of CDH1 in both the nucleus and the cytoplasm. Interestingly, in that study the phosphorylation of yeast CDH1 has been shown to promote its nuclear export (65). It remains to be seen whether phosphorylation of human CDH1 enhances its export from the nucleus. We showed that the CDH1-4D mutant mimicking the hyperphosphorylated form is constitutively cytoplasmic in cultured human cells (Fig. 4). These results per se are also compatible with a role of phosphorylation on the nuclear export of CDH1. The inhibition of nuclear import or re-import as shown in our study (Fig. 5) and the facilitation of nuclear export as reported for yeast CDH1 (65) are not mutually exclusive. While the phosphorylation near the positively charged residues blocks the activity of the NLS of CDH1, the phosphorylation of CDH1 on other sites plausibly mediates the enhancement of nuclear export. It is noteworthy that the NLS we identified in human CDH1 is highly conserved in other organisms including budding yeast (Fig. 3). In addition, the CDK phosphorylation sites in the vicinity of the NLS of CDH1 are also conserved from yeast to humans (Fig. 3). This raises the possibility that budding yeast CDH1 might use the same NLS and phosphorylation sites for nuclear import and for regulation of nuclear import, respectively.
The regulation of nuclear import by phosphorylation in the vicinity of the NLS is not unique to CDH1, but has been reported for other proteins including SV40 large T antigen (78) and adenomatous polyposis coli protein (79). Furthermore, the dual regulation of nuclear import and export by phosphorylation also represents a common mechanism for inactivation of nuclear factors such as Pho4, NF-B, and NF-AT (77). Thus, it is not surprising that phosphorylation of CDH1 might have dual effects on its nuclear import and export, both of which led to cytoplasmic retention of the protein. It has been well documented that phosphorylation of CDH1 inhibits its APC-activating activity (7,(31)(32)(33)(34). In particular, the CDH1-4D mutant behaves as a dominant negative form which is unable to stimulate the APC-dependent proteolysis (Fig. 4B and Refs. 33 and 60). Our demonstration of the constitutive cytoplasmic localization of CDH1-4D (Fig. 4) is compatible with the model that cytoplasmic sequestration of CDH1 contributes to its inactiva-FIG. 6. Disruption of cyclin-binding motif did not influence nuclear localization of CDH1. HeLa cells were transfected with plasmid pHACDH1-AAA. Cells were fixed 36 h after transfection and co-stained for CDH1 (with ␣-HA, panels 1 and 4) and cyclin A (panels 2 and 5). In panels 3 and 6, the CDH1 (green, probed with ␣-HA) and cyclin A (red) fluorescent signals were overlaid by computer assistance. Co-localizations are shown in yellow. The same fields are shown in panels 1-3 and 4 -6. Arrows indicate transfected cells. Bar, 20 m.
tion. Nevertheless, further investigations are required to elucidate the nucleocytoplasmic shuttling of human CDH1 and the influence of CDH1 phosphorylation on nuclear export.