Subcellular translocation signals regulate ... - Wiley Online Library

Biol. Cell (2006) 98, 363–375 (Printed in Great Britain)

Research article

doi:10.1042/BC20060007

Subcellular translocation signals regulate Geminin activity during embryonic development Aline Boos, Amy Lee, Dominic M. Thompson, Jr and Kristen L. Kroll1 Department of Molecular Biology and Pharmacology, Washington University School of Medicine, 660 S. Euclid Ave., St. Louis, MO 63110, U.S.A.

Background information. Geminin (Gem) is a protein with roles in regulating both the fidelity of DNA replication and cell fate during embryonic development. The distribution of Gem is predominantly nuclear in cells undergoing the cell cycle. Previous studies have demonstrated that Gem performs multiple activities in the nucleus and that regulation of Gem activation requires nuclear import in at least one context. In the present study, we defined structural and mechanistic features underlying subcellular localization of Gem and tested whether regulation of the subcellular localization of Gem has an impact on its activity in cell fate specification during embryonic development. Results. We determined that nuclear localization of Gem is dependent on a bipartite NLS (nuclear localization signal) in the N-terminus of Xenopus Gem protein. This bipartite motif mapped to a Gem N-terminal region previously shown to regulate neural cell fate acquisition. Microinjection into Xenopus embryos demonstrated that importdeficient Gem was incapable of modulating ectodermal cell fate, but that this activity was rescued by fusion to a heterologous NLS. Cross-species comparison of Gem protein sequences revealed that the Xenopus bipartite signal is conserved in many non-mammalian vertebrates, but not in mammalian species assessed. Instead, we found that human Gem employs an alternative N-terminal motif to regulate the protein’s nuclear localization. Finally, we found that additional mechanisms contributed to regulating the subcellular localization of Gem. These included a link to Crm1-dependent nuclear export and the observation that Cdt1, a protein in the pre-replication complex, could also mediate nuclear import of Gem. Conclusions. We have defined new structural and regulatory features of Gem, and showed that the activity of Gem in regulating cell fate, in addition to its cell-cycle-regulatory activity, requires control of its subcellular localization. Our data suggest that rather than being constitutively nuclear, Gem may undergo nucleocytoplasmic shuttling through several mechanisms involving distinct protein motifs. The use of multiple mechanisms for modulating Gem subcellular localization is congruent with observations that Gem levels and activity must be stringently controlled during cell-cycle progression and embryonic development.

Introduction During embryonic development, pathways that regulate cell cycle progression, acquisition of cell fate and 1 To

whom correspondence should be addressed (email [email protected]). Key words: Cdt1, embryonic development, Geminin, nuclear localization signal (NLS), Xenopus . Abbreviations used: DAPI, 4 ,6-diamidino-2-phenylindole; (h/x)Gem, (human/Xenopus ) Geminin; GFP, green fluorescent protein; LMB, leptomycin B; NES, nuclear export signal; NLS, nuclear localization signal; mNLS, mutated NLS; SV40, simian virus 40.

cellular differentiation must be co-ordinated temporally and spatially throughout the embryo. The activity of molecules that regulate these processes must therefore be under particularly stringent control. One such molecule, Geminin (Gem), is a novel coiled-coil protein originally identified through two activities that are localized to physically separate domains. Gem can regulate the fidelity of DNA replication via its central coiled-coil and C-terminus (McGarry and Kirschner, 1998). Gem activity also

www.biolcell.org | Volume 98 (6) | Pages 363–375

363

A. Boos and others

has an impact on cell fate: in Xenopus and Drosophila embryos, Gem converts ectoderm into neural tissue, at the expense of other ectodermal cell types, such as epidermis and neural crest (Kroll et al., 1998). In Xenopus, this activity was mapped to a minimal Nterminal domain (amino acids 38–90). Recently, two additional activities for Gem during embryonic development have been defined. Gem binds to and antagonizes the activities of homeodomain-containing transcription factors which are required to drive body segmentation (Luo et al., 2004) or retinogenesis (Del Bene et al., 2004). To date, the major mechanism described for regulation of Gem activity is proteolytic degradation. This occurs through a D-box (destruction box) in the Gem N-terminus, which mediates targeting for polyubiquitination and degradation by the anaphasepromoting complex at the metaphase-to-anaphase transition in mitosis (McGarry and Kirschner, 1998). Gem has been demonstrated to localize to the nucleus in multiple species (Kroll et al., 1998; McGarry and Kirschner, 1998; Wohlschlegel et al., 2000, 2002; Nishitani et al., 2001; Quinn et al., 2001; Bermejo et al., 2002; Mihaylov et al., 2002; Shreeram et al., 2002; Yanagi et al., 2002; Kulartz et al., 2003), and several of the activities of Gem are known or presumed to occur in the nucleus (Kroll et al., 1998; Wohlschlegel et al., 2000; Tada et al., 2001), thus presumably relying on the nuclear localization of Gem. The ability of Gem to affect DNA replication in Xenopus was also shown to be regulated independent of Gem proteolysis through entry of deactivated Gem into the nucleus (Hodgson et al., 2002). Together, these features suggest that regulation of the subcellular localization of Gem represents an additional means of controlling Gem activity, both for its roles in regulating DNA replication and potentially in cell fate acquisition. The aim of the present work was to identify structural and mechanistic features that regulate the subcellular localization of Gem, and to test whether the subcellular localization of Gem has an impact on its activity in cell fate specification during embryonic development. Our data define several new structural and regulatory features of Gem, suggest differences in Gem regulation between mammalian and nonmammalian vertebrate species, and demonstrate that the activity of Gem in regulating cell fate is dependent upon control of its subcellular localization.

364

C

Portland Press 2006 | www.biolcell.org

Figure 1 Gem is a predominantly nuclear protein In HEK-293 cells and Xenopus embryo cells at the blastula stage, endogenous Gem was predominantly detected in the nucleus, whereas expression of a Myc–GFP epitope-tag vector was not restricted to the nucleus. In subsequent Figures, subcellular localization of Myc–GFP–xGem fusion constructs was compared with expression of the Myc–GFP vector (pan-localized) and with endogenous Gem (nuclear).

Results Localization of the Gem protein is predominantly nuclear

The endogenous Gem protein was detected in nuclei of HEK-293 cells and in cell nuclei of blastulastage Xenopus embryos using previously characterized antibodies directed against hGem (human Gem) and xGem (Xenopus Gem) respectively (Figure 1). Interestingly, Gem was also detected in the cytoplasm

Localization signals regulate Geminin activity

of some cells of blastula-stage Xenopus embryos and in murine P19 cells undergoing neuronal differentiation, as detected by immunostaining and by immunoblotting for Gem following subcellular fractionation (Supplementary Figure 1, http://www. biolcell.org/boc/098/boc0980363add.htm). To characterize structural features regulating subcellular localization of Gem, we used both Xenopus embryos and human HEK-293 cells. In each experiment, we assayed localization of Myc–GFP (green fluorescent protein) epitope-tagged Gem variants by comparison with the pan-localized negative control (Myc–GFP protein expressed from the plasmid used for protein fusions in subsequent experiments) and with the predominantly nuclear localization seen for endogenous Gem (Figure 1). A nuclear import-related sequence maps to the N-terminus of xGem

xGem is a 217-amino-acid protein with an Nterminal D-box (residues 34–42) and a central coiled-coil domain (residues 118–152) that mediates homodimerization and interaction with protein partners (Figure 2A). To determine the structural motifs necessary for nuclear localization of Gem, we initially constructed N- and C-terminally truncated variants of xGem fused with an N-terminal Myc–GFP tag. These constructs were transfected into HEK-293 cells or microinjected into Xenopus embryos to assess changes in localization. Expression patterns of the fusion proteins were scored as nuclear, cytoplasmic or pan-localized. For each construct, the data derived from at least three independent experiments are shown in Figure 2(A). Representative images of critical constructs for nuclear localization (highlighted in green) are shown after transfection into HEK-293 cells (Figure 2B). Full-length Myc–GFP–xGem recapitulated the nuclear localization pattern observed for endogenous Gem (xGem1−217 ). Serial truncations from the Cterminal end to residue 90 did not impact on the ability of Gem to localize to the nucleus, whereas further truncations resulted in loss of exclusive nuclear staining and ultimately in pan-localized expression similar to Myc–GFP control (Figures 2A and 2B, and data not shown). N-terminal truncations up to residue 42 did not impact on nuclear localization, whereas further N-terminal truncations caused fusion proteins to lose exclusive nuclear staining and to localize to the

Research article cytoplasm instead (Figures 2A and 2B, xGem72−217 ). Finally, a short C-terminal fragment produced panlocalized expression similar to that of Myc–GFP alone (Figures 2A and 2B, xGem184−217 ). These results suggested that the residues required for Gem nuclear localization were contained within amino acids 42–90 [a region previously described as a minimal neuralizing domain (Kroll et al., 1998)]. Consistent with these results, transfection of a peptide comprising amino acids 42–90 fused to Myc–GFP was sufficient to confer nuclear localization upon the normally pan-localized Myc–GFP control protein (Figures 2A and 2B, xGem42−90 ). Conversely, internal deletion of residues 44–90 caused cytoplasmic accumulation of xGem (Figures 2A and 2B, xGem44−90 ). Together, these results defined these N-terminal amino acids as both sufficient and required for xGem nuclear localization. These data are consistent with previous demonstrations of nuclear localization activity within the xGem N-terminus (Benjamin et al., 2004; Yoshida et al., 2005). Amino acids comprising a bipartite NLS (nuclear localization signal) in xGem

Having shown that residues 42–90 of Gem are sufficient and required for xGem nuclear localization, we attempted to determine whether this sequence contained a NLS and to localize the critical residues. NLSs are generally comprised of basic amino acids (lysine and arginine), consisting of a single motif (monopartite NLS), for example PKKKRKV for the SV40 (simian virus 40) large-T antigen (Kalderon et al., 1984), or of two basic amino acid clusters approx. 10 amino acids apart (bipartite NLS), for example KRPAATKKAGQAKKKKL for nucleoplasmin (Robbins et al., 1991). The xGem fragment sufficient for nuclear localization (residues 42–90) contains a number of lysine and arginine residues. Single and combinatorial point mutations of these amino acids were made and tested for loss of nuclear localization (Figure 2C, red arrows). Single point mutations had no or little effect (data not shown), whereas mutation of multiple amino acids revealed two motifs that were essential for nuclear localization (highlighted in black). Alanine substitution of either a basic amino acid-rich motif found at residues 60–62 or at 71–72 and 74 caused partial loss of nuclear localization such that some protein was now seen in the cytoplasm (xGemKRK → AAA and xGemKKAK → AAAA ;


365

A. Boos and others

Figure 2 For legend see facing page

data not shown), whereas combined alanine substitution of these two motifs [xGemm(mutated)NLS ] caused the Myc–GFP-fusion protein to localize predominantly to the cytoplasm, rather than the nucleus (Figure 2C, left-hand panels). Similar results were obtained by testing critical constructs in Xenopus embryos (Figure 2C, right-hand panels). Using measurement of relative fluorescence in nuclei compared with the cytoplasm of transfected HEK-293 cells, we confirmed our observations that Gem truncations lacking N-terminal amino acids accumulated in the

366

C


cytoplasm (Figure 3A, for example, compare 1–90 with 91–217 and 44–90). Alanine substitution in the xGem NLS also rendered the protein predominantly cytoplasmic, rather than pan-localized (Figure 3A, compare 1–217 with 1–217mNLS ). These data indicate that the KRK and KKAK motifs constitute a bipartite NLS, with the two motifs eight amino acids apart, which is required for nuclear localization of xGem. This xGem bipartite NLS differs from that previously described by Benjamin et al. (2004) (RTK + KRK; see the Discussion for further


Research article

Figure 2 An N-terminal bipartite NLS is required for nuclear localization of full-length xGem in HEK-293 cells and Xenopus embryos Serial amino acid truncations revealed an N-terminal xGem region sufficient and required for Gem nuclear localization. (A) Upper panel shows the xGem protein with structural features and amino acid numbers shown. xGem contains an N-terminal D-box (grey), and a central coiled-coil domain (black). The N-terminus also contains a minimal fragment with the ability to modulate ectodermal cell fate in Xenopus embryos (neuralizing region, as indicated) Lower panel shows the Myc–GFP–xGem variants transfected into HEK-293 cells and scored for subcellular localization. N, nuclear; P, pan-localized; C, cytoplasmic. Letters in parentheses denote that expression was enriched in, but not exclusive to, the compartment indicated. Constructs for which images are in (B) are highlighted in green. (B) Representative immunofluorescent images of critical constructs transfected into HEK-293 cells and immunostained to detect the Myc–GFP epitope tag (GFP). Corresponding nuclei were visualized by DAPI staining. (C) Upper panel: the amino acids of the minimal nuclear fragment (xGem42−90 ) are shown. Candidate residues for a NLS (Lys and Arg) were mutated (red arrows) by alanine substitutions. Mutants were scored for subcellular localization as in (A). Mutation of two basic amino acid motifs at residues 60–62 (KRK → AAA) and 71–74 (KKAK → AAAA) (black boxes) had a substantial effect on localization and, when in combination (mNLS), reproduced the cytoplasmic pattern of the internal deletion (xGem44−90 ) shown in (B). Lower panels: images are shown for full-length xGem1−217 with intact and mutated NLS. Left: immunostaining for the epitope-tag for constructs transfected into HEK-293 cells. Right: expression of xGem plasmids injected into Xenopus embryos at the two-cell stage and analysed in early gastrulae. Localization was visualized by immunostaining against the epitope tag. A punctate nuclear pattern was observed with full-length xGem, whereas mutation of the bipartite signal within the full-length protein (xGemmNLS ) caused loss of nuclear localization and predominantly cytoplasmic staining.

details). However, these data are consistent with work by Yoshida et al. (2005) demonstrating that xGem amino acids 59–78 are sufficient to confer nuclear localization activity upon a heterologous protein. We used amino acid substitution to demonstrate that the residues KRK and KKAK, located in this region, are required for xGem nuclear localization. Gem localization is affected by Crm1-dependent export

To analyse whether cytoplasmic localization of xGem variants could be related to nuclear export mechanisms, we treated cells expressing import-deficient Gem variants with leptomycin B (LMB), a specific inhibitor of Crm1-dependent nuclear export (Kudo et al., 1998). Such treatments eliminated cytoplasmic localization, and instead caused import-deficient Gem variants to show pan-localized expression similar to control (Myc–GFP epitope tag) protein expression (Figure 3B, 44–90 + LMB). By contrast, treatment with ethanol (vehicle) had no effect on xGem44−90 localization. An internal deletion of amino acids 139–184 resulted in pan-localized distribution similar to the effects of LMB treatment, indicating that the deleted region was required for the Crm1-dependent export-related activity of Gem (Figure 3B). In conclusion, our data demonstrate that cytoplasmic accumulation of xGem occurs in a Crm1dependent manner, and that sequences required for

this phenomenon map to residues 139–184 of the xGem protein. Cross-species comparison of Gem protein sequences

We compared the Gem protein across a number of vertebrate species to determine whether the motifs regulating Gem subcellular localization show crossspecies conservation. These data are shown in a Gem protein sequence alignment, with the hGem (human Gem) and xGem sequences highlighted by asterisks (Supplementary Figure 2, http://www.biolcell.org/ boc/098/boc0980363add.htm). Amino acids 139– 184 of xGem are highlighted in grey, with relevant amino acids in red. We noted that the Gem export-related region exhibits a high degree of crossspecies conservation. Moreover, the coiled-coil and C-terminus of the xGem protein include stretches of hydrophobic residues that are typical of NESs (nuclear export signals), suggesting that these residues may represent an additional localization motif within Gem. Leucine and isoleucine residues, amino acids characteristic of a NES, were highly conserved between various Gem orthologues in the coiled-coil domain. Thus Gem may possess a NES. An alternative amino acid motif is required for nuclear localization of hGem

Our work above defined a bipartite NLS (KRK/ KKAK) in the xGem N-terminus. Although the


367

A. Boos and others

368

Figure 3 Amino acids 139–184 mediate Crm1-dependent cytoplasmic accumulation of xGem (A) The ratios of the nuclear to nuclear + cytoplasmic fluorescence for the various xGem variants tested for subcellular localization in HEK-293 cells are shown as the means + − S.D. (B) HEK-293 cells were transfected with Myc–GFP-tagged xGem44−90 and immunostained to detect the GFP tag. Nuclei were visualized using DAPI staining (blue). Cytoplasmic accumulation was lost when HEK-293 cells, transfected with xGem44−90 for 12–16 h, were subsequently treated with the export inhibitor LMB for 3 h. Control treatment with ethanol (vehicle) had no effect on xGem44−90 subcellular localization. Deletion of amino acids 139–184 had a similar effect to LMB treatment (44–90 + 139–184).

threonine residues in the position of the KKAK motif in the mammalian species we assessed. These data indicated that different structural features were responsible for conferring nuclear localization upon nonmammalian compared with mammalian Gem orthologues. Since endogenous hGem and mouse Gem are predominantly nuclear proteins but do not possess the xGem bipartite NLS, we tested truncated hGem variants to define residues that could represent an alternative NLS in mammalian species. Unlike the full-length xGem, which was robustly nuclear under various experimental conditions, full-length hGem and mouse Gem N- or C-terminal fusion proteins with several epitope tags showed mis-localization to the cytoplasm upon overexpression (Figure 4A and Supplementary Figure 3, http://www.biolcell.org/ boc/098/boc0980363add.htm). However, N-terminal fragments of hGem were predominantly nuclear when expressed as GFP-fusion proteins in HEK293 cells (Figure 4A, hGem1−130 and hGem1−109 ). This nuclear localization was lost upon truncation of amino acids 98–109 (Figure 4A, hGem1−97 ). We observed that another lysine- and argininerich motif (RRK) was present at residues 106–108 within the N-terminal half of hGem and that this motif was well conserved across species (Supplementary Figure 2, box 3, http://www.biolcell.org/ boc/098/boc0980363add.htm). It was demonstrated that this motif was neither necessary nor sufficient for nuclear localization of the xGem protein (Figure 2, xGem42−90 and xGem91−217 , and data not shown). However, hGem truncations that eliminated the RRK motif resulted in loss of hGem nuclear localization (Figure 4A, hGem1−97 ). Likewise, alanine substitution of the RRK residues within hGem1−130 abolished the nuclear enrichment seen with intact hGem1−130 (Figure 4A, 1–130mRRK ). Together, these data indicate that the RRK motif is essential for nuclear localization of hGem.

first part of the bipartite signal (KRK) was present in all available Gem sequences (Supplementary Figure 2, box 1, http://www.biolcell.org/boc/098/ boc0980363add.htm), the second part of the motif (KKAK) was restricted to non-mammalian vertebrates considered (Supplementary Figure 2, box 2, http://www.biolcell.org/boc/098/ boc0980363add.htm). Instead, we found serine and

Cdt1 rescues nuclear localization of cytoplasmic Gem variants

C


Direct binding of Gem to Cdt1, a protein required for DNA replication initiation, has been established in previous studies and this interaction involves residues within the Gem coiled-coil domain and N-terminal to it (Tada et al., 2001; Lee et al., 2004; Saxena et al., 2004). Therefore, we also tested whether


Research article

Figure 4 Alternative motifs and mechanisms for Gem nuclear localization (A) GFP-tagged hGem fragments transfected into HEK-293 cells were analysed for subcellular localization in comparison with transfection of GFP vector alone and endogenous hGem (see Figure 1). Nuclei were visualized using DAPI staining (lower panels). Truncation or alanine substitution of a basic amino acid motif (RRK; amino acids 106–108) within the N-terminus of hGem resulted in loss of nuclear localization, indicating that in hGem the RRK motif is required for nuclear localization activity. A scheme of hGem protein is shown, indicating the location of the RRK motif relative to other features. (B) Cdt1 mediates the translocation of cytoplasmic Gem into the nucleus. HEK-293 cells were co-transfected with Myc–GFP-tagged xGem44−90 or GFP–hGem and human (h)Cdt1 (FLAG-tagged). To determine the resulting subcellular localization of the Gem variants, transfected cultures were immunostained for the GFP tag. Nuclei were visualized using DAPI staining. Cdt1 co-transfection conferred nuclear localization upon both NLS-deficient xGem and mislocalized hGem. Cdt1 co-transfection had no effect on the pan-cellular localization of the epitope-tag alone (data not shown).

Cdt1 could have an impact on the subcellular localization of Gem by co-expressing Gem variants with FLAG-tagged Cdt1 in HEK-293 cells. Although the nuclear localization pattern of full-length xGem was unaffected by Cdt1 co-expression (data not shown), we found that Cdt1 was able to rescue nuclear localization of NLS-deficient xGem (Figure 4B, compare xGem44−90 with or without Cdt1). Nuclear localization of Gem in response to Cdt1 coexpression required the Gem coiled-coil and residues situated N-terminal to this, which are regions of Gem previously shown to mediate Cdt1–Gem interaction

(Supplementary Figure 4, http://www.biolcell.org/ boc/098/boc0980363add.htm). Cdt1 co-transfection had no effect on the Myc–GFP control (data not shown). Thus co-transfection of the binding partner of Gem, Cdt1, was able to rescue nuclear localization of NLS-deficient xGem. As described above, overexpression of hGem led to cytoplasmic mislocalization of various fusion protein constructs (Figure 4A and Supplementary Figure 3, http://www.biolcell.org/boc/098/ boc0980363add.htm). This phenomenon might be due to a limiting factor involved in hGem


369

A. Boos and others

localization. Based on our finding that Cdt1 co-expression could rescue nuclear localization of NLSmutated xGem variants, we assessed whether Cdt1 co-expression could also restore nuclear localization to cytoplasmically mislocalized hGem variants by cotransfecting hGem together with Cdt1 in HEK-293 cells. Indeed, Cdt1 co-expression caused hGem to localize to the nucleus, resembling localization of endogenous hGem protein (compare Figure 4B with Figure 1A). Thus addition of exogenous Cdt1 could also translocate epitope-tagged human Gem into the nucleus. A similar result was recently observed by Yoshida et al. (2005). Gem localization to the nucleus is required to regulate ectodermal cell fate

Microinjection of xGem has previously been shown to change competent Xenopus ectoderm to neural tissue at the expense of non-neural tissues, such as neural crest and epidermis. The mechanistic basis of this phenomenon is not known (Kroll et al., 1998). The NLS identified in the present study maps to the minimal N-terminal domain previously shown to be sufficient for the activity of Gem in neural cell fate acquisition (residues 38–90, shown in Figure 2A). We therefore tested whether nuclear localization was required for the ability of Gem to affect ectodermal cell fate during Xenopus embryogenesis. N-terminal Gem variants with an intact versus mutated NLS were introduced into Xenopus. Figure 5 shows the subcellular localization of ectopically expressed Gem proteins, as revealed by immunostaining against the Myc–GFP tag (Figures 5A, 5D and 5G). Effects of Gem mRNA injection on ectodermal tissue were observed by double in situ hybridization for neural and epidermal marker genes (blue stain) and for injected Gem (pink stain) (Figures 5B, 5C, 5E, 5F, 5H and 5I). As seen in HEK-293 cells, the Gem Nterminal domain (xGem42−90 ) localized to the nucleus following injection into Xenopus embryos (Figure 5A). xGem42−90 effectively suppressed expression of epidermal-specific keratin (Figure 5B) and caused expansion of neural tissue, indicated by Sox2 (Figure 5C). Mutation of the NLS of Gem prevented both nuclear-specific localization (Figure 5D), and also both of the activities observed above for xGem42−90 (Figures 5E and 5F). Conversely, nuclear localization and both of these activities were restored by fusion of a heterologous NLS from SV40 protein to the N-

370

C


terminus of NLS-mutant Gem (Figures 5G–5I). Note that we also observed a suppression of Sox2 expression where the lineage label crossed the neural plate in wild-type or rescue-construct-injected embryos (Figures 5C and 5I). This activity was also absent in the NLS-mutated form (Figure 5F). Together, these data indicate that although the mechanism by which Gem has an impact on neural cell fate is unknown, localization to the nucleus is essential for Gem to carry out this activity.

Discussion Localization signals in the xGem protein

We have identified several amino acid motifs that can regulate subcellular localization of Gem. A minimal nuclear fragment of xGem (amino acids 42–90) was sufficient to translocate a heterologous protein (GFP) to the nucleus, demonstrating that this fragment had independent import activity. We subsequently identified a bipartite NLS located between amino acids 60 and 74 in the xGem N-terminus. This NLS was required for nuclear localization of xGem. Intriguingly, NLS deficiency caused epitope-tagged full-length Gem proteins to localize mainly to the cytoplasm, unlike the pan-localized control protein (Myc–GFP tag alone). This observation suggested the presence of additional regulatory motifs involved in the subcellular localization of Gem, such as an NES, a binding site for an exported interaction partner or a cytoplasm-tethered binding partner. Indeed, treatment with the export inhibitor LMB caused cytoplasmic NLS-deficient xGem to distribute throughout the cell, demonstrating the involvement of a Crm1-dependent export-related activity in control of the subcellular localization of Gem. Further deletion within NLS-deficient Gem localized the exportrelated region to a stretch of hydrophobic residues between amino acids 139 and 184 of xGem. However, structural studies of Gem have shown that the leucine and isoleucine residues within the hGem and mouse Gem coiled-coil domains are located within the Gem–Gem dimer interface (Lee et al., 2004; Okorokov et al., 2004; Thˆaepaut et al., 2004), and therefore are not likely to be accessible to Crm1 binding. By contrast, many leucine and isoleucine residues C-terminal to the coiled-coil domain were present across species and potentially accessible to an export receptor. This possibility is not mutually exclusive with the alternative of indirect export by Gem


Research article

Figure 5 Bipartite NLS is required for the ability of xGem to regulate ectodermal cell fate Xenopus embryos were injected with xGem variants. The top panels (A, D, G) show the subcellular localization of constructs tested functionally in the panels below. Effects on epidermal tissue (indicated by the epidermal-specific marker keratin) and on neural tissue (marked by Sox2) were visualized by in situ hybridization (blue stain). Cells containing injected xGem mRNA were lineage labelled by in situ hybridization (pink stain). Dashed lines in (B), (E) and (H) indicate borders of the epidermal keratin-stained territory. Brackets in (C), (F) and (I) mark the width of Sox2-expressing territory on each side of the embryonic midline. Representative embryos injected unilaterally at the two-cell stage and raised to gastrula stages are shown. Injected territory (pink stain) is oriented to the left. The numbers in (B), (C), (E), (F), (H) and (I) indicate the fraction of embryos that displayed the effect shown. (A–C) xGem42−90 was localized to the nucleus when ectopically expressed in Xenopus (A). Injection of xGem42−90 mRNA suppressed epidermal-specific keratin (B), and caused expansion of the Sox2-positive territory (C). (D–F) Mutation of the bipartite NLS caused the minimal nuclear fragment to lose specific localization (D). Ectopic xGem42–90mNLS expression overlapped with the epidermal keratin domain without suppressing the epidermal marker (dark purple colour, highlighted by arrow; E). No expansion of the Sox2-positive territory expansion was seen (F). (G–I) Addition of an exogenous NLS from SV40 large-T antigen rescued nuclear localization of the mutant (SV40NLS-xGem42-90mNLS ; G). The phenotypes observed for xGem42−90 were also rescued via the heterologous NLS (H and I). Panels (B, E, H) show the animal hemisphere views of stage 11.5 embryos with the dorsal side orientated towards the bottom of the page; (C, F, I) show dorsal-facing views of the prospective neural plate at stage 12, with the anterior orientated towards the bottom of the page.

interaction with a NES-containing binding partner. In summary, our results demonstrate that xGem localizes to the nucleus through a bipartite NLS, and accumulates in the cytoplasm in a Crm1-dependent manner in the absence of its NLS.

A bipartite NLS in xGem

During the preparation of the present manuscript, a study was published describing a different bipartite NLS for xGem (Benjamin et al., 2004). We identified a bipartite NLS consisting of the KRK and


371

A. Boos and others

KKAK motifs that was both necessary and sufficient to confer the nuclear localization of xGem. Although Benjamin et al. (2004) also found the KRK motif to be necessary for nuclear localization, the truncated and mutated constructs used in that work (which left the KKAK motif intact) showed nuclear plus cytoplasmic, rather than strongly cytoplasmic, localization. These data are in agreement with ours, in suggesting that the KRK motif is not solely responsible for conferring nuclear localization. The other motif (RTK) included in the NLS found by Benjamin et al. (2004) was required for some aspect of the interaction of Gem with the import-receptor, importin-α, but Gem variants mutated for KRK alone compared with RTK and KRK did not differ substantially in their extent of nuclear localization. Amino acids 45– 60, which encompass the RTK motif, were also shown to be involved in maintaining xGem protein stability. We also note that the threonine and lysine residues of this motif are specific to Xenopus subspecies and not conserved among other vertebrates (Supplementary Figure 2, http://www.biolcell.org/ boc/098/boc0980363add.htm). The N-terminal amino acid fragment we found to be necessary and sufficient for xGem nuclear localization (xGem42−90 ) contains both our bipartite NLS (KRK and KKAK) and amino acids 45–60, which include the RTK motif. Since mutation of KRK and KKAK resulted in strongly cytoplasmic localization in our present work, we suggest that the RTK motif may play a role in xGem subcellular localization by stabilizing xGem or by regulating binding to importin-α, but is less likely to contribute to a bipartite NLS. Yoshida et al. (2005) reported that amino acids 59–78 of xGem can confer nuclear localization to a heterologous protein (BSA), in agreement with our identification of specific residues of the bipartite NLS of Gem in this region. Differential Gem localization in mammalian and non-mammalian vertebrates

Although the first half of the bipartite NLS we identified in xGem (KRK) was broadly conserved across species, the second part (KKAK) was conserved in all non-mammalian vertebrates we considered, but was not present in mammals. Instead, we found that nuclear localization of hGem depended upon an Nterminal RRK motif adjacent to the coiled-coil domain. This motif may represent an alternative NLS

372

C


or, among other possibilities, may represent an interaction site for a protein partner that mediates hGem nuclear localization. Intriguingly, a recent study suggested that the RRK residues that we found to be critical for hGem nuclear localization are also part of a Gem domain critical for interaction of Gem and Cdt1 (Lee et al., 2004). In keeping with this, we found that co-expression of Cdt1 with import-deficient xGem and mislocalized hGem can allow translocation of cytoplasmic Gem proteins into the nucleus of HEK293 cells. This observation may be due to overexpression of the protein, but alternatively suggests Cdt1mediated nuclear translocation as another mechanism for Gem nuclear entry, particularly for mammalian Gem proteins that do not possess the strong xGem bipartite NLS. Together, these data provide a potential connection between the Cdt1–Gem interaction and Gem nuclear localization. Interestingly, Yoshida et al. (2005) also recently observed that hGem added to Xenopus egg extracts was unable to cross the nuclear envelope of sperm nuclei unless recombinant Cdt1 was added, whereas, in agreement with our data, nuclear import of xGem was not dependent on addition of Cdt1 (Yoshida et al., 2005). Functional relevance of Gem subcellular localization to cell fate determination

Studies in Xenopus (Kroll et al., 1998) and Drosophila (Quinn et al., 2001) demonstrated a role for Gem in development of the nervous system. The NLS identified in the present study maps within the minimal N-terminal domain of Gem able to affect neural cell fate (Kroll et al., 1998). Mutation of the bipartite NLS resulted in loss of the ability of Gem to suppress epidermal cell fate and to expand the neural plate territory, activities which are rescued by fusion of NLS-mutant Gem to a heterologous NLS. Although the molecular basis of the ability of Gem to affect ectodermal cell fate is not yet known, we have demonstrated that nuclear import is a requirement for this activity. Nuclear import of Gem is likely to impact on the activity of the protein in other developmental contexts, as Gem has been shown to negatively regulate the activities of homeodomain proteins during retinogenesis and embryonic axial patterning (Del Bene et al., 2004; Luo et al., 2004). These phenomena are likely to require regulation of nuclear Gem protein levels and activity. Thus it will be of interest to determine whether signalling pathways that


regulate embryonic development also modulate nuclear import of Gem. Functional relevance of cytoplasmic Gem and/or Gem export

Gem has been described as a nuclear protein, and the functional relevance of cytoplasmic Gem protein has only been discussed in the context of Gem degradation and of somatic cells undergoing mitosis (McGarry and Kirschner, 1998; Hodgson et al., 2002). In these cases, when cytoplasmic Gem is observed, the nuclear envelope has already been disassembled, allowing protein exchange between the nuclear and cytoplasmic compartments. In the present study, we demonstrated that xGem accumulated in the cytoplasm in the presence of a distinct nucleus when import-related amino acids were deleted (xGem44−90 ), and it was found that this phenomenon involved Crm1-dependent nuclear export. Although the functional relevance of retaining Gem in the cytoplasm is unknown, we suggest that this activity may represent a mechanism for controlling the subcellular localization of Gem and/or to activity in post-mitotic cells, which lack periods of nuclear envelope disassembly. Interestingly, in multiple developmental contexts, we detected Gem protein outside the nucleus (Supplementary Figure 1, http://www.biolcell.org/boc/098/ boc0980363add.htm), which is consistent with the notion that Gem shuttles between subcellular compartments. Together, our observations of cytoplasmic Gem and our findings of multiple regulatory mechanisms for the subcellular localization of Gem suggest that the activity of Gem may be controlled by nucelocytoplasmic shuttling. Conclusions

In conclusion, we have identified new structural motifs of Gem underlying control of the subcellular localization and function of Gem. In Xenopus, a bipartite NLS is essential for controlling the activity of Gem in cell fate regulation, and presumably in regulating the fidelity of DNA replication. Differences between residues critical for xGem and hGem localization indicate that Gem is regulated differently across species. Possession of a functional NLS, together with sensitivity to the nuclear export inhibitor LMB, suggests that Gem may undergo nucleocytoplasmic shuttling. Our findings have determined a

Research article need for considering structure and species specificity in experiments employing Gem variants. It will be of importance to account for these factors both in drawing conclusions from previous work and in designing future experimental studies. It is likely that additional mechanisms underlying control of Gem activity will be uncovered in the future; however, the complex interplay of mechanisms already known to regulate Gem activity underscores its position as a critical mediator of multiple processes needed for cellular function and embryogenesis.

Materials and methods Subcloning and mutagenesis

All PCR-based xGem fragments were cloned from EcoRI to XbaI sites into pCS2PMTeGFP. hGem was cloned from EcoRI to XhoI sites into pCS2 + eGFP (a similar fusion construct lacking the Myc6 tag) or from EcoRI to XbaI sites into pCS2 + MTeGFP (both tags N-terminal to insert). Internal deletion of xGem amino acids 44–90 was carried out by a two-step PCR method; the product was then subcloned into pCS2PMTeGFP (EcoRI and XbaI sites). Internal deletion of amino acids 139–184 was created with primers flanking the desired deletion, and amplifying away from the deletion around the entire xGem44−90 plasmid. PCR products were then ligated through a unique restriction site introduced in the primers (e.g. SphI, Sac, EcoRV). Point mutations were obtained using the QuikChange® sitedirected mutagenesis kit (Stratagene), with minor modification of the manufacturer’s protocol. Primer sequences and additional descriptive information regarding all of the constructs used in the present study are available from authors upon request. Tissue culture and transfection

HEK-293 cells were grown in standard DMEM (Dulbecco’s modified Eagle’s medium) with 10% FBS (fetal bovine serum) and incubated at 37◦ C and 95% air/5% CO2 . For imaging purposes, cells were plated on to poly(L-lysine)-coated (0.1 mg/ml; Sigma) glass coverslips placed in 6-well plates. One day prior to transfection, cells were seeded at 6 × 105 per well and grown to approx. 70% confluency overnight. On the following day, cells were transfected (by adaptation of the manufacturer’s protocol) with 1 µg of DNA using 12 µl of Polyfect (Qiagen) and incubated overnight. Immunocytochemistry, microscopy and relative fluorescence intensity analysis

After 10 min of fixation with 4% paraformaldehyde in PBS, cells transfected with either GFP- or Myc–GFP-tagged constructs were stained with the rabbit polyclonal anti-GFP antibody (1:500; Clontech), followed by Alexa488 goat anti-rabbit secondary antibody (1:500; Molecular Probes). For detection of Myc-tagged constructs, we used the mouse monoclonal anti-Myc antibody (9E10; 1:500; Santa Cruz Biotechnology), followed by Alexa488 goat anti-mouse secondary antibody (1:500; Molecular Probes). Coverslips were then mounted with Vectashield and stained with DAPI (4 ,6-diamidino-2-phenylindole) (Vector Laboratories) to visualize cell nuclei. Endogenous hGem


373

A. Boos and others

was detected using the rabbit polyclonal antibody FL-209 (1:100; Santa Cruz) and secondary antibody was detected as described above for GFP. Fluorescent staining was observed under a Zeiss fluorescent Axioscope, and images taken with a digital camera using Axiocam software. Relative fluorescence was determined as described previously (Plafker and Macara, 2002) and analysed using ‘Quantity One’ software (Bio-Rad). Relative intensity of nuclear luminescence is expressed as a ratio of the total luminescence (nuclear + cytoplasmic), with data representing the means + − S.D. from at least 10 cells. Xenopus embryo injections, immunohistochemistry, in situ hybridization and light microscopy

Endogenous xGem was detected using a rabbit polyclonal antibody directed against xGem (McGarry and Kirschner, 1998). Gem variants were introduced into Xenopus embryos as either plasmids or capped mRNA by microinjection into one cell at the two-cell stage, as described previously (Kroll et al., 1998). For confirmation of subcellular localization, 75–100 pg of plasmid was injected. Embryos were devitellinized at stage 12/13 and fixed in MEMFA [0.1 M Mops (pH 7.4), 2 mM EGTA, 1 mM MgSO4 , 3.7% formaldehyde] for 2 h, then placed for over 1 h in methanol. Constructs were detected with the anti-Myc antibody 9E10 (1:500; Santa Cruz), and were visualized with an HRP (horseradish peroxidase)-conjugated secondary antibody (1:500) (Jackson ImmunoResearch Laboratories), using diaminobenzidine as the colour substrate (Fast DAB Tablet Set, Sigma) or with an alkaline phosphatase antibody, using Nitro Blue Tetrazolium (Roche) and 5-bromo-4-chloro-3-indolyl phosphate (Roche) together as the colour substrate. For mRNA injections, 500 pg of capped mRNA (prepared using the mMesssage mMachine kit; Ambion) was injected into one cell at the two-cell stage. In situ hybridization was performed as described previously using digoxigenin- and fluoresceinlabelled RNA probes, the appropriate alkaline phosphataseconjugated secondary antibodies (Roche Diagnostics) and the colour substrates 5-bromo-4-chloro-3-indolyl phosphate and MagentaPhos (Biosynth International, Inc) (Kroll et al., 1998). Embryos were analysed and imaged using an Olympus stereomicroscope and digital camera. LMB treatment

Cells were transfected with Polyfect (see above) and allowed to express transfected plasmids for 12–16 h before removal of medium. LMB (Sigma) was diluted to a final concentration of 20 nM in complete medium and added to the transfected cells for 3 h. An equal volume of ethanol was added to control cells. Cells were then fixed and stained for the appropriate tag as described above. Co-transfection of Gem and Cdt1 in HEK-293 cells

Epitope-tagged Gem variants were co-transfected into HEK293 cells with FLAG-tagged full-length human Cdt1 (plasmid generously provided by Dr Seongjin Seo, Department of Molecular Biology and Pharmacology, Washington University School of Medicine, St Louis, U.S.A.) at a DNA ratio of 1:1 using Polyfect transfection reagent as described above. After transfection (12–24 h), cells were fixed in 4% formaldehyde for 10 min, following imunocytochemical detection of the Gem epitope tag (as described above).

374

C


Acknowledgements We thank Brent-Brower Toland for participation in the NES characterization, Diane Redmond for graphic art assistance, the thesis committee for A.B. (David Beebe, Ross Cagan, Phyllis Hanson, Greg Longmore and Helen Piwnica-Worms), Seongjin Seo, Cathy Collins and Genova Cook for helpful discussions, and Irving Boime and Jason Weber for critical reading of the manuscript prior to submission. This work was funded by grants from the National Institutes of Health (R01 GM6681501), the American Cancer Society (IRG-58-010-44), the Pharmacia/Washington University Biomedical Research Program, the March of Dimes and the Howard Hughes Medical Institute Faculty Development Award to K.K. References Benjamin, J.M., Torke, S.J., Demeler, B. and McGarry, T.J. (2004) Geminin has dimerization, Cdt1-binding, and destruction domains that are required for biological activity. J. Biol. Chem. 279, 45957–45968 Bermejo, R., Vilaboa, N. and Cales, C. (2002) Regulation of CDC6, geminin, and CDT1 in human cells that undergo polyploidization. Mol. Biol. Cell 13, 3989–4000 Del Bene, F., Tessmar-Raible, K. and Wittbrodt, J. (2004) Direct interaction of geminin and Six3 in eye development. Nature (London) 427, 745–749 Hodgson, B., Li, A., Tada, S. and Blow, J.J. (2002) Geminin becomes activated as an inhibitor of Cdt1/RLF-B following nuclear import. Curr. Biol. 12, 678–683 Kalderon, D., Roberts, B.L., Richardson, W.D. and Smith, A.E. (1984) A short amino acid sequence able to specify nuclear location. Cell (Cambridge, Mass.) 39, 499–509 Kroll, K.L., Salic, A.N., Evans, L.M. and Kirschner, M.W. (1998) Geminin, a neuralizing molecule that demarcates the future neural plate at the onset of gastrulation. Development 125, 3247–3258 Kudo, N., Wolff, B., Sekimoto, T., Schreiner, E.P., Yoneda, Y., Yanagida, M., Horinouchi, S. and Yoshida, M. (1998) Leptomycin B inhibition of signal-mediated nuclear export by direct binding to CRM1. Exp. Cell Res. 242, 540–547 Kulartz, M., Kreitz, S., Hiller, E., Damoc, E.C., Przybylski, M. and Knippers, R. (2003) Expression and phosphorylation of the replication regulator protein geminin. Biochem. Biophys. Res. Commun. 305, 412–420 Lee, C., Hong, B., Choi, J.M., Kim, Y., Watanabe, S., Ishimi, Y., Enomoto, T., Tada, S. and Cho, Y. (2004) Structural basis for inhibition of the replication licensing factor Cdt1 by geminin. Nature (London) 430, 913–917 Luo, L., Yang, X., Takihara, Y., Knoetgen, H. and Kessel, M. (2004) The cell-cycle regulator geminin inhibits Hox function through direct and polycomb-mediated interactions. Nature (London) 427, 749–753 McGarry, T.J. and Kirschner, M.W. (1998) Geminin, an inhibitor of DNA replication, is degraded during mitosis. Cell (Cambridge, Mass.) 93, 1043–1053 Mihaylov, I.S., Kondo, T., Jones, L., Ryzhikov, S., Tanaka, J., Zheng, J., Higa, L.A., Minamino, N., Cooley, L. and Zhang, H. (2002) Control of DNA replication and chromosome ploidy by geminin and cyclin A. Mol. Cell. Biol. 22, 1868–1880

Research article


Nishitani, H., Taraviras, S., Lygerou, Z. and Nishimoto, T. (2001) The human licensing factor for DNA replication Cdt1 accumulates in G1 and is destabilized after initiation of S-phase. J. Biol. Chem. 276, 44905–44911 Okorokov, A.L., Orlova, E.V., Kingsbury, S.R., Bagneris, C., Gohlke, U., Williams, G.H. and Stoeber, K. (2004) Molecular structure of human geminin. Nat. Struct. Mol. Biol. 11, 1021–1022 Plafker, S.M. and Macara, I.G. (2002) Ribosomal protein L12 uses a distinct nuclear import pathway mediated by importin 11. Mol. Cell. Biol. 22, 1266–1275 Quinn, L.M., Herr, A., McGarry, T.J. and Richardson, H. (2001) The Drosophila Geminin homolog: roles for Geminin in limiting DNA replication, in anaphase and in neurogenesis. Genes Dev. 15, 2741–2754 Robbins, J., Dilworth, S.M., Laskey, R.A. and Dingwall, C. (1991) Two interdependent basic domains in nucleoplasmin nuclear targeting sequence: identification of a class of bipartite nuclear targeting sequence. Cell (Cambridge, Mass.) 64, 615–623 Saxena, S., Yuan, P., Dhar, S.K., Senga, T., Takeda, D., Robinson, H., Kornbluth, S., Swaminathan, K. and Dutta, A. (2004) A dimerized coiled-coil domain and an adjoining part of geminin interact with two sites on Cdt1 for replication inhibition. Mol. Cell 15, 245–258

Shreeram, S., Sparks, A., Lane, D.P. and Blow, J.J. (2002) Cell type-specific responses of human cells to inhibition of replication licensing. Oncogene 21, 6624–6632 Tada, S., Li, A., Maiorano, D., Mechali, M. and Blow, J.J. (2001) Repression of origin assembly in metaphase depends on inhibition of RLF-B/Cdt1 by geminin. Nat. Cell Biol. 3, 107–113 Thaepaut, M., Maiorano, D., Guichou, J.F., Augae, ˆ ˆ M.T., Dumas, C., Maechali, M. and Padilla, A. (2004) Crystal structure of the ˆ coiled-coil dimerization motif of geminin: structural and functional insights on DNA replication regulation. J. Mol. Biol. 342, 275–287 Wohlschlegel, J.A., Dwyer, B.T., Dhar, S.K., Cvetic, C., Walter, J.C. and Dutta, A. (2000) Inhibition of eukaryotic DNA replication by geminin binding to Cdt1. Science (Washington, D.C.) 290, 2309–2312 Wohlschlegel, J.A., Kutok, J.L., Weng, A.P. and Dutta, A. (2002) Expression of geminin as a marker of cell proliferation in normal tissues and malignancies. Am. J. Pathol. 161, 267–273 Yanagi, K., Mizuno, T., You, Z. and Hanaoka, F. (2002) Mouse geminin inhibits not only Cdt1–MCM6 interactions but also a novel intrinsic Cdt1 DNA binding activity. J. Biol. Chem. 277, 40871–40880 Yoshida, K., Takisawa, H. and Kubota, Y. (2005) Intrinsic nuclear import activity of geminin is essential to prevent re-initiation of DNA replication in Xenopus eggs. Genes Cells 10, 63–73

Received 5 January 2006/1 February 2006; accepted 8 February 2006 Published as Immediate Publication 8 February 2006, doi:10.1042/BC20060007


375