Nuclear import of the four core histones H2A, H2B, H3 and H4 is one of the main nuclear import activities during S‐phase of the cell cycle. However, the molecular machinery facilitating nuclear import of core histones has not been elucidated. Here, we investigated the pathways by which histone import can occur. First, we show that core histone import can be competed by the BIB (β‐like import receptor binding) domain of ribosomal protein L23a suggesting that histone import is an importin mediated process. Secondly, affinity chromatography on immobilized core histones revealed that several members of the importin β family of transport receptors are able to interact with core histones. Finally, we demonstrate that at least four known and one novel importin, importin 9, can mediate nuclear import of core histones into the nuclei of permeabilized cells. Our results suggest that multiple pathways of import exist to provide efficient nuclear uptake of these abundant, essential proteins.
During S‐phase of the cell cycle, DNA is replicated and de novo chromatin assembly takes place. This requires nuclear uptake of newly synthesized core histones and their deposition onto DNA to form the regularly repeated unit of eukaryotic chromatin, the nucleosome (for review see Verreault, 2000). Although a number of factors have been identified aiding the process of nucleosome assembly, the molecular mechanism by which core histones are transported into the nucleus is unknown.
Transport of proteins into and out of the nucleus is a signal dependent, receptor‐mediated process that proceeds through nuclear pore complexes. Nuclear transport is largely mediated by members of a family of nuclear transport receptors referred to as importins/exportins or karyopherins. A defining feature of these receptors is their ability to interact with the small GTPase Ran in its GTP‐bound form. This interaction is instrumental in controlling the compartment specific loading and unloading of the transport substrates and hence the directionality of nucleo‐cytoplasmic transport. While binding of RanGTP to importins facilitates import substrate release, exportins need to bind RanGTP for efficient loading of the export cargo. Both events are specific to the nuclear compartment where the chromatin‐associated nucleotide exchange factor RCC1 generates RanGTP. In the cytoplasm, however, the RanGTP concentration is kept low due to the presence of the RanGTPase activating protein (RanGAP1) allowing the formation of import complexes and the disassembly of export complexes (for review see Mattaj and Englmeier, 1998; Görlich and Kutay, 1999).
A number of signals in nuclear proteins have been identified which allow their selective access to the transport machinery. The first signal to be characterized was the classical basic nuclear localization signal (NLS) which is recognized by a complex consisting of the adapter molecule importin α (Impα) and the transport receptor importin β (Impβ) (for review see Görlich and Kutay, 1999). Other signals are directly recognized by import receptors: the M9 domain of hnRNPA1 confers binding to transportin (Siomi and Dreyfuss, 1995; Pollard et al., 1996; Fridell et al., 1997), the RS domain of SR proteins binds to transportin SR (Kataoka et al., 1999), and the BIB domain of rpL23a allows its direct interaction with at least four different importins namely Impβ, transportin, importin 5 (Imp5) and importin 7 (Imp7) (Jäkel and Görlich, 1998). Impβ and transportin are exceptional in that they can recognize diverse signals. For example, transportin, through separable binding sites, interacts with signals as different as the basic BIB domain and the serine/glycine rich M9 domain. Similarly, Impβ uses distinct sites to recognize the IBB (Impβ binding) domain of Impα and BIB. In addition, Impβ can associate with Imp7 to form a heterodimer which mediates nuclear import of the linker histone H1 in vitro (Jäkel et al., 1999).
Recent studies have analysed sequence motifs involved in nuclear targeting of core histones. It has been demonstrated that both the basic N‐terminal domains and the central globular domains of all four core histones can serve as nuclear targeting signals (Baake et al., 2001). However, the transport receptors mediating core histone import have not been identified. Here, we show that multiple import pathways can contribute to nuclear import of core histones. At least five different importins can directly interact with individual core histones and support their nuclear import into the nuclei of permeabilized cells.
Nuclear import of core histones is an importin‐mediated process
To identify transport receptors involved in nuclear uptake of core histones we first characterized this process in in vitro nuclear transport experiments. Rabbit reticulocyte lysate had been shown to support nuclear accumulation of single core histones into the nuclei of permeabilized cells (Baake et al., 2001). To extend this study further, we analysed the effect of known inhibitors of importin‐mediated transport on nuclear uptake of fluorescently labelled core histones. Import of all four core histones was stimulated by the addition of reticulocyte lysate (Figure 1A). Core histones are smaller than the diffusion limit of the nuclear pore. To exclude that core histones enter the nuclei by diffusion, we further tested whether their nuclear accumulation could be abolished by pre‐incubation of the nuclei with a dominant‐negative mutant of Impβ which prevents receptor‐mediated transport through the NPC (Kutay et al., 1997b). The import of all four histones was blocked under these conditions. Biochemical studies had suggested that core histones are able to interact with Impβ (Johnson‐Saliba et al., 2000). If the interaction between histones and Impβ or related factors was required for core histone import, the import reaction should be sensitive to the addition of RanGTP which dissociates most importin/substrate complexes (Rexach and Blobel, 1995; Görlich et al., 1997). Import of all four histones was indeed sensitive to the addition of RanQ69L(GTP), a Ran mutant which is unable to hydrolyse the bound nucleotide. These results suggested that members of the importin β family mediate core histone import.
To define more specifically the pathways involved in nuclear import of histones, we performed the in vitro transport assays in the presence of known competitors of different nuclear import pathways (Figure 1B). Overall, the import of all four histones was not prevented by the addition of excess unlabelled IBB–GST or GST–M9 which target the Impα binding site of Impβ and the hnRNPA1 binding site of transportin, respectively. In contrast, the addition of GST–BIB reduced the import of core histones. The competition by BIB argues that at least one of the factors which directly bind and import BIB‐containing cargo (i.e. Impβ, transportin, Imp5 and Imp7) is required for histone import. Control experiments revealed that each of the competitors was functional and specific for its respective pathway (see Supplementary figure 1).
Identification of importins as histone binding proteins
The foregoing results suggested that core histones can be imported into nuclei by one or more members of the importin β family. To identify these transport receptors, we employed affinity chromatography on immobilized histones. The individual histones were biotinylated, immobilized on streptavidin–agarose and binding of proteins from rabbit reticulocyte lysate to the histones was compared in the absence and presence of RanQ69L(GTP). Proteins recovered on the histones in the absence of RanQ69L but whose binding was reduced in its presence were candidate import receptors. Analysis of proteins in the bound fractions showed very similar patterns of binding partners for all four histones (Figure 2). Proteins which displayed reduced interaction in the presence of RanQ69L were identified by mass spectrometry as Impβ, Imp5 and Imp7. In addition, two peptides were found which matched a novel mammalian transport receptor which we had cloned recently based on its homology to the yeast karyopherin Kap114p (for identification and cloning see Methods). We will refer to this novel protein as importin 9 (Imp9). Immunoblotting confirmed that Impβ and Imp7 were specifically enriched on the histones in the absence of RanQ69L. We could neither confirm binding of Imp5 nor test other transport receptors such as transportin by immunoblotting as the available antibodies did not detect these proteins in the reticulocyte lysate. Hence we cannot exclude that additonal importins had bound to the histones.
Binding of the identified importins could have been either direct or mediated by association with another protein present in the reticulocyte lysate. It was therefore crucial to test whether these importins were able to interact directly with the histones. For this purpose, Escherichia coli lysates were supplemented with equal amounts of purified importins and subjected to binding to immobilized histones (Figure 3). Besides Impβ, Imp5, Imp7 and Imp9, we tested whether transportin would also bind to the histones as it had been shown before to interact with BIB. All of the tested importins were able to interact directly with the histones. Two other Impβ related transport receptors, the exportins CAS and CRM1, neither bound to the histones in the absence (Figure 3) nor in the presence of RanGTP (not shown). As an additional control, with transportin as an example, we analysed whether the binding of this importin was reduced in the presence of RanGTP, which was indeed the case (see Supplementary figure 2).
Different importins can mediate nuclear import of core histones
Having established that different importins can directly interact with individual core histones, it remained to be determined whether the observed interactions are relevant to nuclear import. All of the tested importins supported nuclear uptake of fluorescently labelled histones (Figure 4). Although all the importins were able to directly interact with the core histones (see Figure 3) we observed differences in their import efficiencies towards the individual histones. Imp5 and transportin strongly supported import of all core histones. Also Impβ and Imp7 were able to mediate import of core histones and when added as heterodimer, no significant increase in import efficiency was observed. The only core histone for which we could observe strong nuclear accumulation by the Impα/Impβ complex was H2B. In contrast, Impβ mediated import of H2A, H3 and H4 was decreased by the simultaneous presence of Impα. The novel transport receptor Imp9 appeared to be most productive in H2B import.
As confirmation that the observed effects were specific to core histones, we found that histone H1 was not transported by the individual importins used in our study (Figure 4). Rather, histone H1 is efficiently imported only by an Impβ/Imp7 heterodimer (Jäkel et al., 1999). As expected, export receptors like CAS (not shown) or CRM1 did not support nuclear import of core histones.
We have shown that nuclear import of single core histones can be mediated by at least four known members of the Impβ superfamily and the novel Imp9. We, for the first time, describe a function for this novel mammalian transport receptor. Imp9 is related to the Saccharomyces cerevisiae karyopherin Kap114p, which had been previously characterized as an import receptor for TATA‐binding protein (TBP) (Morehouse et al., 1999; Pemberton et al., 1999). Like all other importins, Imp9 is a RanGTP binding protein (not shown). Although mouse Imp9 and Kap114p display 21% overall identity and 40% homology it remains to be seen if Imp9 mediates nuclear import of TBP in higher eukaryotic cells.
Besides Imp9, also Impβ, transportin, Imp5 and Imp7 directly interacted with core histones and all of these factors supported core histone import in vitro, although to differing extents. Impβ is an exceptional transport receptor in that it not only binds transport substrates directly but also forms heterodimers with import adapters such as Impα or the import receptor Imp7. In contrast to published data (Langer, 2000), we demonstrate that H2B import can be mediated by the Impα/Impβ heterodimer in vitro. One explanation for this discrepancy is that Langer used nucleoplasmin core in the import buffer; in our hands, nucleoplasmin core blocks H2B import. The heterodimer of Impβ–Imp7 has been demonstrated to be required for the efficient import of the linker histone H1 (Jäkel et al., 1999). Both Impβ and Imp7 are enriched on the histones from reticulocyte lysate. Unlike H1 import, core histones can be imported by either of the two importins on their own and we could not observe cooperativity in their binding and import of core histones. In this respect, core histone import resembles nuclear import of rpL23a which does not require cooperativity between Impβ and Imp7. Rather, nuclear import of rpL23a occurs along redundant pathways (Jäkel and Görlich, 1998).
Multiple import pathways of core histones may exist to ensure that these essential proteins can be efficiently delivered to their site of action. Core histones as well as many other nuclear proteins are highly basic in nature and often basic patches are presented as import signals. Strikingly, most characterized nuclear import receptors recognize basic import signals and to some extent can substitute for each other. This is reflected in the observation that eight of the ten yeast karyopherins known to be involved in nuclear import are encoded by non‐essential genes yet many of these receptors import substrates with essential nuclear functions. It is however difficult to decide which of the four importins defines the predominant pathway in vivo. There may even exist additional importins in higher eukaryotes that may either be dedicated to or serve as back‐up import receptors for core histones. Indeed, while our study was under review, a comprehensive analysis of nuclear import of H2A and H2B in yeast was published (Mosammaparast et al., 2001), demonstrating that the import of H2A and H2B occurs along redundant pathways in vivo.
To assemble into nucleosomes, core histones must first form H2A–H2B heterodimers and (H3–H4)2 tetramers. The question of whether the formation of histone heteromers takes place in the cytoplasm and/or in the nucleus is still not completely solved. Although in cytosolic cell extracts the existence of histone heteromers has been demonstrated (Chang et al., 1997), it is known that cytoplasmic extracts often contain proteins that are localized to the nucleus in vivo. Our data demonstrate that single core histones can be imported into the nucleus in vitro. The same holds true in vivo as shown for the nuclear import of H2A and H2B deletion mutants that are unable to form heterodimers (Mosammaparast et al., 2001). However, when binding of transport receptors to H2A and H2B monomers was compared with that to H2A–H2B heterodimers the same efficiency was observed (P. Mühlhäusser and U. Kutay, unpublished data). This suggests that the histone import signals are still accessible to importin binding when H2A and H2B are dimerized. Therefore our data does not exclude the possibility that core histones may be imported as heteromers in vivo.
Histones are very sticky, basic proteins which presumably have to be shielded from undesired interactions on their way from the ribosome to their final destination. Channelling the histones from one binding partner to the next is a likely mechanism to achieve this. On their way to the site of DNA replication histones are known to interact with histone chaperones, the acetylation machinery, chromatin assembly factors and, as we show here, nuclear transport receptors which can be considered as histone chaperones during nuclear pore passage. It will be interesting to see whether other histone chaperones play an additional role during nuclear transport of core histones.
Preparation of modified histones.
Histones H1, H2A, H2B, H3 and H4 were purchased from Roche. Each histone was dissolved in 50 mM HEPES pH 7.5 and incubated for 1 h on ice with stoichiometric amounts of either FLUOS or biotin‐aminocaproic acid N‐hydroxy succinimide ester (Roche) for fluorescence labelling or biotinylation of the histones, respectively. Reaction mixtures were then passed over NAP5 columns (Pharmacia) equilibrated in 50 mM HEPES pH 7.5 to remove unincorporated modifying agents.
Database searches revealed the existence of mouse (AC016814.5) and human (NM_018085/FLJ10402) sequence tags homologous to S. cerivisiae KAP114. The information obtained from the mouse genomic sequence and additional ESTs (AI317597.1 and BE625765.1) comprising the 5′ and 3′ ends of the coding region, respectively, allowed us to design primers for amplification of the complete coding region. Oligonucleotides (GCCTCCCGGGATGGCGGCAGCGGCGGCAG; CCGTCCCGGGTTAGATGCCGATAGTCTGTAGGACCC) were used to amplify the coding region of Imp9 by PCR from NIH 3T3 cell cDNA. The PCR product was then inserted into the XmaI site of pQE32 (Qiagen) to allow the expression of the protein with an N‐terminal His6‐tag in E. coli. Several clones were analysed to determine the consensus sequence. The novel sequence data reported in this manuscript have been submitted to the DDBJ/EMBL/GenBank database under accession number AJ309238.
The IBB–GST expression construct was obtained by insertion of the GST coding region as an NcoI–BamHI fragment into pQE60IBB (1–63 aa of Rch1).
Recombinant protein expression and purification.
RanQ69L, Impα, Impβ, Impβ (45–462), transportin, Imp5, Imp7 and CAS were purified as previously described (Görlich et al., 1996; Kutay et al., 1997a,b; Jäkel and Görlich, 1998). Untagged CRM1 was expressed from pET3a (Paraskeva et al., 1999) in BL21 DE3 at 23°C with 0.5 mM IPTG. CRM1 was purified after precipitation of the bacterial lysate with 35% ammonium sulfate and buffer exchange to 50 mM Tris pH 7.5, 100 mM potassium acetate (KOAc) on MonoQ followed by gel filtration on Superdex 200 (Pharmacia). Imp9 was expressed in XL1 blue and purified on Ni‐agarose, followed by precipitation with 37% ammonium sulfate and gel filtration.
IBB–GST was expressed in BLR(pREP4), purified on Ni‐agarose and the buffer exchanged to 50 mM HEPES pH 7.5, 250 mM KOAc, 2 mM magnesium acetate [Mg(OAc)2], and 250 mM sucrose.
Solution binding experiments.
For each binding reaction, 10 μl of streptavidin–agarose beads were presaturated with the respective histone. Binding from rabbit reticulocyte lysate (Promega) was performed such that 100 μl lysate was diluted into binding buffer (50 mM HEPES pH 7.5, 50 mM KOAc, 2 mM MgCl2) to yield a final sample volume of 600 μl.
A 100 μl aliquot of E. coli lysate was supplemented with the corresponding transport receptor to a concentration of 2 μM. Binding was performed in a final volume of 600 μl in 50 mM Tris pH 7.5, 300 mM NaCl and 4 mM MgCl2.
Subsequent binding reactions were carried out at 4°C for 4 h. Beads were then washed three times in binding buffer and bound proteins were eluted with SDS sample buffer and analysed by SDS–PAGE.
Protein identification by mass spectrometry and databank searching.
The tryptic peptide mixture was desalted by ZipTip C18 (Millipore Corporation, Bedford, MA) and analysed by nano‐electrospray mass spectrometry. A 5 μl aliquot of the peptide mixture were lyophilized and dissolved in 5 μl methanol/1% formic acid (1:1). The MS/MS measurement was performed with a nano‐electrospray hybrid quadrupole mass spectrometer Q‐Tof (Micromass, Manchester, UK). The collision gas was argon at a pressure of 6.0 × 10−5 mbar in the collision cell.
To identify the respective proteins we used the sequence tag program, which combines partial manual spectrum interpretation of about three aminoacids (sequence tag) with the residual mass N‐ and C‐terminals of the interpreted region and the peptide mass to search in a nonredundant translated nucleotide database.
The anti‐Impβ and anti‐Imp7 antibodies were a kind gift of Dr Dirk Görlich.
Permeabilized cell assay.
In vitro transport reactions were performed essentially as described previously (Kutay et al., 1997a; Jäkel et al., 1999). For core histone import with recombinant transport receptors the following buffer was used: 50 mM HEPES pH 7.5, 150 mM KOAc, 1 mM Mg(OAc)2, 0.5 mM EGTA and 250 mM sucrose. Further details are given in the figure legends.
Supplementary Figure 1
Supplementary Figure 2
We thank Drs Howard Fried, Gerd Lipowsky and Judith Erkmann for critical reading of the manuscript and Dr Alain Verreault and the members of the Kutay lab for stimulating discussions. I particularly acknowledge Dr Dirk Görlich in whose laboratory this study was initiated. We thank Nathalie Treichel for expert technical assistance and Rolf Moser for assistance with computer graphics. This work was supported by an ETH internal grant and a grant from the Swiss National Science Foundation to U.K.
- Copyright © 2001 European Molecular Biology Organization