RCC1 is the guanine nucleotide exchange factor for Ran GTPase. Generation of Ran-GTP by RCC1 on chromatin provides a spatial signal that directs nucleocytoplasmic transport, mitotic spindle assembly and nuclear envelope formation. We show that RCC1 is expressed in human cells as at least three isoforms, named RCC1α, RCC1β and RCC1γ, which are expressed at different levels in specific tissues. The β and γ isoforms contain short inserts in their N-terminal regions (NTRs) that are not present in RCC1α. This region mediates interaction with chromatin, binds importin α3 and/or importin β, and contains regulatory phosphorylation sites. RCC1γ is predominantly localised to the nucleus and mitotic chromosomes like RCC1α. However, compared to RCC1α, RCC1γ has a greatly reduced interaction with an importin α3-β and a stronger interaction with chromatin that is mediated by the extended NTR. RCC1γ is also the isoform that is most highly phosphorylated at serine 11 in mitosis. Unlike RCC1α, RCC1γ supports cell proliferation in tsBN2 cells more efficiently when serine 11 is mutated to non-phosphorylatable alanine. Phosphorylation of RCC1γ therefore specifically controls its function during mitosis. These results show that human RCC1 isoforms have distinct chromatin binding properties, different molecular interactions, and are selectively regulated by phosphorylation, as determined by their different NTRs.
The small GTPase Ran is highly conserved in eukaryotic cells and has roles in the control of nucleocytoplasmic transport (Melchior et al., 1993; Moore and Blobel, 1993), mitotic spindle assembly (Carazo-Salas et al., 1999; Kalab et al., 1999; Ohba et al., 1999; Wilde and Zheng, 1999; Zhang et al., 1999) and nuclear envelope assembly (Hetzer et al., 2000; Zhang and Clarke, 2000). Proper spatial coordination of these processes relies upon the distribution of GTP-versus GDP-bound forms of Ran, which is in turn controlled by the localisation of its regulators (Clarke and Zhang, 2001; Dasso, 2002; Görlich and Mattaj, 1996; Hetzer et al., 2002). The only known guanine nucleotide exchange factor (GEF) for Ran, RCC1, is localised to chromatin throughout the cell division cycle (Bischoff and Ponstingl, 1991; Moore et al., 2002; Ohtsubo et al., 1989). By contrast, RanGAP1, the sole GTPase-activating factor that stimulates GTP hydrolysis by Ran (Bischoff et al., 1994), is localised to cytoplasm in interphase, in particular the cytoplasmic face of nuclear pore complexes, and to the mitotic spindle and kinetochores in mitosis (Joseph et al., 2002; Mahajan et al., 1997; Matunis et al., 1996). Together with the concentration of Ran in the nucleus by its active import, the distinct localisation of its regulators results in a high concentration of Ran-GTP in the nucleoplasm and a low concentration of Ran-GDP in the cytoplasm, whereas in mitosis, Ran-GTP is concentrated around chromosomes (Caudron et al., 2005; Kalab et al., 2006; Kalab et al., 2002).
Nucleocytoplasmic transport is carried out by the importin β family of transport proteins, or karyopherins, which carry cargo across the nuclear envelope via nuclear pores, with directionality conferred by the high concentration of Ran-GTP in the nucleoplasm (Görlich and Mattaj, 1996). As import cargo complexes enter the nucleus through nuclear pore complexes, Ran-GTP binding to importin β causes disassembly of importin-cargo complexes, releasing cargo into the nucleoplasm (Stewart, 2007). Conversely, Ran-GTP is an essential component of export complexes involving the export receptor, Crm1 (Fornerod et al., 1997). These export complexes disassemble on entry into the cytoplasm as RanGAP1 catalyses hydrolysis of GTP by Ran, converting it into Ran-GDP (Bischoff et al., 1994).
During mitosis, when the nuclear envelope breaks down in animal cells, similar mechanisms are involved in the orchestration of spindle assembly by Ran. Ran-GTP generated around mitotic chromosomes binds to importin β and causes the release of spindle assembly factors such as TPX2 from inhibitory complexes with importins, thereby promoting microtubule nucleation and stabilisation in the vicinity of chromosomes (Gruss et al., 2001; Nachury et al., 2001; Wiese et al., 2001). It has been proposed that a gradient of Ran-GTP concentration, declining at greater distances from mitotic chromosomes, provides a spatial signal that is important for orchestrating activity of different factors required for spindle assembly (Hetzer et al., 2002).
The localisation of RCC1 to chromatin is crucial for the spatial organisation of Ran-GTP and is therefore important for the functions of Ran throughout the cell division cycle. The crystal structure of RCC1 reveals it to form a seven-bladed propeller, from the seven characteristic RCC1 sequence repeats (Renault et al., 1998). The association of RCC1 with chromatin involves interactions between its core region and histones H2A and H2B (Nemergut et al., 2001). The binding of RCC1 to chromatin in vitro is promoted by nucleotide-free or GDP-bound Ran, and inhibited by Ran-GTP (Li et al., 2003; Zhang et al., 2002). Ran interacts separately with histones H3 and H4 (Bilbao-Cortes et al., 2002) and the affinity of RCC1 with chromatin is reduced in a D182A mutant (Moore et al., 2002; Hutchins et al., 2004), which disrupts the interaction with Ran (Azuma et al., 1999). However, the localisation of RCC1 to chromatin is critically dependent on a flexible N-terminal region (NTR) (Moore et al., 2002) which is likely to extend beyond the core structure (Renault et al., 1998). Deletion of the NTR disperses RCC1 from mitotic chromosomes and causes a variety of spindle abnormalities consistent with disruption of the proper spatial organisation of Ran-GTP (Moore et al., 2002). The NTR is required for binding of RCC1 directly to DNA in vitro via residues 21-25 (Seino et al., 1992), although it is not known if this binding occurs during the interaction with chromatin.
The NTR of RCC1 also mediates interactions with other proteins and is the target for post-translational modification. There are two clusters of basic residues in the NTR that form a bipartite nuclear localisation signal (NLS) that binds to importin β via the adaptor importin α3 (Kohler et al., 1999; Nemergut and Macara, 2000; Talcott and Moore, 2000). Lysines 21 and 22 in the second cluster are essential for importin α3 binding whereas the first cluster strengthens the interaction (Friedrich et al., 2006). The N-terminal methionine of RCC1 is removed and the α-amino group of serine 2 is mono-, di- or tri-methylated, which is required for interaction with chromatin, but does not appear to be regulated during the cell cycle (Chen et al., 2007). By contrast, phosphorylation of serines 2 and 11 by CDK1-cyclin B1 during mitosis disrupts importin binding and regulates the dynamic interaction with mitotic chromosomes (Hutchins et al., 2004; Li and Zheng, 2004).
Although animal cells contain only one RCC1 gene, human and hamster cells express RCC1 proteins of different lengths that are probably generated by alternative mRNA splicing. However, no apparent difference in activity or cellular function of these novel isoforms compared to the shorter, more abundant isoform was found (Miyabashira et al., 1994). No further consideration of the possible importance of different RCC1 isoforms has been made.
Here we show that human cells express at least three isoforms of RCC1, which we name RCC1α, RCC1β and RCC1γ. Normal human tissues show different levels of expression of these isoforms. We show that RCC1γ has increased affinity for chromatin in cells, reduced binding to importins, and is the isoform that is most highly phosphorylated at serine 11 during mitosis. In a cell proliferation assay, RCC1γ promotes cell proliferation more efficiently when phosphorylation at serine 11 is abolished by mutation. This work demonstrates that isoforms of RCC1 in mammalian cells have different molecular interactions and divergent mechanisms of regulation, suggesting that they have distinct functions during the cell division cycle and in differentiated cells.
Multiple isoforms of RCC1 are expressed in mammalian cells
Alignment of the amino acid sequences of mammalian RCC1 molecules derived from cDNA sequences (Fig. 1A) indicates that three isoforms are expressed in human cells. These correspond to the originally identified short isoform, a second isoform with a 31 amino acid insert previously identified by Miyabashira et al. [(Miyabashira et al., 1994) RCC1-I], and a third isoform with a 17 amino acid insert that we also identified by RT-PCR amplification of RCC1 cDNA from HeLa cells. We named these three isoforms RCC1α, RCC1β and RCC1γ, respectively. The insert in the novel human isoform, RCC1γ, consists of the first 17 residues of the insert in RCC1β. RCC1γ is closely related to the hamster and mouse RCC1-I proteins previously identified (Miyabashira et al., 1994), and is predicted in rhesus monkey, whereas RCC1β is predicted in chimpanzee. This suggests that the expression of all three isoforms is conserved in primates.
Examination of the genomic sequence of human RCC1 indicates that the latter part of exon 6′ that is present in RCC1β but not RCC1γ contains a GT-AG splice donor and acceptor sites, and also a potential branch site, suggesting that the differences between all three RCC1 transcripts could be due to alternative splicing (Fig. 1B). Comparison of the genomic sequence of mouse with human shows the sequence conservation of the 6′ exon of RCC1 (supplementary material Fig. S1), including the alternative splice sites identified in human, indicating that homologues of all three isoforms may be expressed in non-primate mammals as well as in primates. An isoform containing a four amino acid insert is found in Xenopus laevis, indicating that multiple isoforms are also present in other vertebrates.
To examine expression of RCC1 proteins in human cells, two antibodies were generated (Fig. 1C). The first antibody, anti-R-INS, was raised to a peptide corresponding to the insert region of RCC1γ, and so should detect both RCC1β and RCC1γ. The second antibody, anti-R-DIS, was raised to a peptide corresponding to the region either side of the insert site. This epitope was predicted to be disrupted when the insert is present, and therefore should only detect RCC1α. A similar strategy was used successfully by Miyabashira et al. (Miyabashira et al., 1994). These antibodies were tested by immunoblotting of proteins retrieved from HeLa cell extracts using glutathione S-transferase (GST) expressed as a fusion with RanT24N, a mutant that forms a stable complex with RCC1 (Fig. 1D). Anti-R-DIS detected a single band precipitated specifically by GST-RanT24N that corresponded to the major polypeptide detected by an antibody raised against the invariant C terminus of RCC1, consistent with it being RCC1α (45 kDa). Anti-R-INS detected two bands corresponding to the two higher molecular mass bands detected with the total RCC1 antibody. The migration of these larger polypeptides is consistent with the predicted sizes of RCC1γ and RCC1β of 46.7 and 48.2 kDa, respectively, indicating that both are expressed in HeLa cells, with RCC1γ being expressed at a higher level than RCC1β. In addition, a very minor band with an apparent molecular mass less than that of RCC1α was detected by the total RCC1 antibody in the GST-RanT24N precipitate but not by either anti-R-DIS or anti-R-INS antibodies. This may represent an additional shorter isoform that lacks the anti-R-DIS epitope.
Immunoblotting of lysates from a range of human cell lines showed expression of both RCC1α and insert-containing isoforms in all of those tested (Fig. 1E). When the relative levels of expression of RCC1α and RCC1γ were analysed during the cell cycle, no significant changes in the amount or the ratio of the two isoforms was observed. However, the relative expression levels of RCC1 isoforms differed between HeLa and U2OS cells (supplementary material Fig. S2). Analysis of the expression of RCC1 isoforms in normal human tissues showed remarkable differences (Fig. 1F). RCC1α was the major isoform expressed in most tissues, but the level of expression varied widely. RCC1α was not detected or only weakly expressed in small intestine, lung and stomach. The expression levels of insert-containing isoforms also varied between tissues, but the pattern differed from that of RCC1α. RCC1β and γ were expressed in all tissues tested, but were the predominant isoforms in stomach and were also strongly expressed in lung. The ratio of β and γ isoforms also varied between tissues, but γ was the more abundant isoform except in skeletal muscle. In addition, a low molecular mass band was detected in skeletal muscle and testis which might correspond to the low molecular mass form weakly expressed in HeLa cells (Fig. 1D).
Localisation of RCC1α and RCC1γ in cells
When RCC1γ was expressed in cells as a fusion with green fluorescent protein (GFP) it was found to be mainly nuclear in interphase and localised predominantly to chromosomes during mitosis, similar to GFP-RCC1α (Fig. 2A). However, when the localisation of the two isoforms was quantified in live U2OS cells, more cells expressing GFP-RCC1γ than GFP-RCC1α showed a fluorescent signal in the cytoplasm, albeit at much lower levels than in the nucleus (Fig. 2B). The proportion of cells expressing GFP constructs that were in mitosis was reduced when a higher concentration of DNA was used for transfection, and this effect was greater for GFP-RCC1γ (2.1% and 0.7% at 0.4 and 0.8 μg DNA/well, respectively) than for GFP-RCC1α (3.4% and 1.4% at 0.4 and 0.8 μg DNA/well, respectively). This suggests that overexpression of RCC1 is detrimental to cell cycle progression, particularly with the γ isoform.
RCC1γ has a stronger interaction with chromatin
The interaction of RCC1 with chromatin in cells is dynamic (Cushman et al., 2004; Hutchins et al., 2004; Li et al., 2003; Li and Zheng, 2004). To examine whether the dynamics of this interaction varied among the isoforms, we used fluorescence recovery after photobleaching (FRAP). A laser was targeted at the nuclei of live interphase U2OS cells expressing GFP-RCC1α or GFP-RCC1γ in order to bleach a spot of about 1 μm in diameter. The recovery of fluorescence within the spot was monitored as a measure of the mobility of GFP-RCC1 proteins, from which the stability of their interaction with chromatin can be inferred (Fig. 3A). These experiments showed that the half-time of signal recovery for GFP-RCC1γ (1.62±0.36 seconds) was approximately twofold greater than that for GFP-RCC1α (0.71±0.19 seconds), indicating that RCC1γ has a more stable interaction with chromatin than RCC1α. Consistent with this finding, when subcellular fractionation of U2OS cells was used to separate the cytoplasm and soluble nuclear material (supernatant) from insoluble nuclear material including chromatin (pellet), a greater proportion of RCC1β or γ than RCC1α was present in the pellet (Fig. 3B).
Since the localisation of RCC1α to mitotic chromosomes is strongly dependent on the NTR (Moore et al., 2002), we compared the localisation of the isolated NTRs of RCC1α and RCC1γ attached to a GFP-GST (GG) fusion. Only a small proportion (14%) of mitotic cells expressing GG-NTRα showed concentration of the fusion protein on chromosomes, similar to previous results (Moore et al., 2002), whereas in the majority of mitotic cells expressing GG-NTRγ, the fusion was localised to chromosomes (Fig. 3C). Thus, the extended NTR of RCC1γ is sufficient to target a fusion protein to mitotic chromosomes. Together, these results show that RCC1γ has a stronger interaction with chromatin in cells than RCC1α and that this difference is due to its extended NTR.
RCC1γ has greatly reduced binding to importins
The inserts of RCC1β and γ are located adjacent to the NLS within the NTR, and introduces an acidic aspartate residue next to the second cluster of basic residues of the NLS that are critical for importin-α3 binding (Friedrich et al., 2006). We therefore tested whether the presence of the insert could affect binding of importin α3-β to RCC1γ. We used GST-RCC1α or GST-RCC1γ coupled to beads to precipitate endogenous importin β from asynchronous HeLa extracts, with or without addition of His6-importin α3. In contrast to the strong importin α3-dependent association of importin β with GST-RCC1α, little or no importin β was associated with GST-RCC1γ (Fig. 4A), indicating a greatly reduced interaction. When GST-importin β-coated beads were used to precipitate endogenous RCC1 from HeLa cell extracts, much less RCC1γ was precipitated than RCC1α, again showing a strongly reduced interaction of RCC1γ with importins compared to RCC1α (Fig. 4B). These results provide an explanation for the greater proportion of cells with cytoplasmic GFP-RCC1γ than GFP-RCC1α (Fig. 2B), since the nuclear localisation of RCC1α is in part dependent on its active import mediated by importin α3-β which is disrupted in RCC1γ.
The interaction of RCC1 with importins does not affect exchange activity (supplementary material Fig. S3); however, the interaction of the NTR with an importin α3-β dimer might influence the dynamic interaction of RCC1 with chromatin (Hutchins et al., 2004; Li and Zheng, 2004). To test the possible effect of importin α3-β on the interaction of RCC1 isoforms with chromatin, we incubated recombinant RCC1 proteins with the resuspended chromatin pellet from subcellular fractionation of HeLa cells with or without GST-importin β and His6-importin α3. We found that importin α3-β efficiently competed with the association of RCC1α but not RCC1γ with chromatin in vitro (Fig. 4C).
The NTR of RCC1 is also required for DNA binding in vitro (Seino et al., 1992), although it not yet clear if this binding is important in the interaction with chromatin. To test the ability of importin α3-β to compete with DNA for binding of RCC1, we incubated HeLa extract with DNA-cellulose with or without GST-importin β and His6-importin α3. Consistent with its ability to bind RCC1α selectively, the importin α3-β dimer strongly inhibited the binding of RCC1α but not RCC1γ to DNA (Fig. 4D).
These results indicate that the more stable interaction of RCC1γ with chromatin compared to RCC1α (Fig. 3) is in part due to reduced competition from importin α3-β for the NTR. It remains possible that RCC1γ also has an intrinsically higher affinity for chromatin than RCC1α.
RCC1γ is the isoform most highly phosphorylated at serine 11 in mitosis
Human RCC1 is regulated by phosphorylation in mitosis at sites in the NTR close to the insert site (Hutchins et al., 2004; Li and Zheng, 2004). Previous studies have shown that RCC1α can be phosphorylated at these sites by CDK1-cyclin B1, although the identity of the endogenous phosphorylated RCC1 isoform(s) is unknown.
When we compared the separation of RCC1 isoforms between the chromatin pellet and supernatant in lysates of mitotic and interphase (asynchronous) HeLa cells, we found that most of RCC1α was released from the pellet in mitosis whereas a greater proportion of RCC1γ was retained (Fig. 5A). A major RCC1 band was detected by a phospho-specific antibody against the serine 11 site in lysates of mitotic, but not asynchronous HeLa cells. In addition, a less abundant phosphorylated RCC1 band of lower molecular mass was detected. A further very weak band of higher molecular mass was also detected, but this latter band was not depleted by a total RCC1 antibody and is therefore unlikely to be an RCC1 isoform (data not shown). We found that the major phosphorylated form of RCC1 was strongly enriched in the chromatin pellet of mitotic cells (Fig. 5A). This major band exactly comigrated with RCC1γ (Fig. 5B) and is therefore very likely to represent this isoform. This indicates that phosphorylated RCC1γ is predominantly associated with chromatin, suggesting that phosphorylation of this isoform promotes this interaction.
The less abundant phosphorylated RCC1 band is most likely to be RCC1α. Interestingly the phosphorylated form of RCC1α was not associated with chromatin but was released into the supernatant (Fig. 5A). This indicates that phosphorylation of RCC1α does not promote its interaction with chromatin, in contrast to phosphorylation of RCC1γ.
These results indicate that RCC1γ is the endogenous isoform of RCC1 that is most highly phosphorylated at serine 11 during mitosis, although both RCC1α and RCC1γ can be phosphorylated in cells when over-expressed (supplementary material Fig. S4) (Hutchins et al., 2004). The differences in phosphorylation between RCC1α and RCC1γ in cells could be due to differential regulation of the isoforms, rather than intrinsic differences in the effectiveness of the molecules as a CDK1-cyclin B substrate. However, when recombinant RCC1 isoforms were incubated in mitotic HeLa extracts, RCC1γ was more highly phosphorylated at serine 11 than RCC1α (supplementary material Fig. S5). Furthermore, RCC1γ was more highly phosphorylated than RCC1α at serine 11 by CDK1-cyclin B semi-purified from mitotic HeLa cell extracts (Fig. 5C). Therefore the RCC1γ molecule is an intrinsically better substrate for the kinase than RCC1α.
The function of RCC1γ in cells is regulated by phosphorylation at serine 11
To determine whether RCC1γ is functional like RCC1α in cells, we compared the ability of the isoforms to replace endogenous RCC1 in tsBN2 cells (Fig. 6). This temperature-sensitive cell line carries a single point mutation in RCC1 (S256F), which causes the protein to be rapidly degraded when cells are shifted from the permissive temperature (32°C) to the restrictive temperature (39.7°C) (Uchida et al., 1990). We transiently expressed GFP-RCC1α and GFP-RCC1γ in tsBN2 cells, along with the non-phosphorylatable mutants GFP-RCC1α S11A and GFP-RCC1γ S11A in order to investigate the contribution of phosphorylation at serine 11 (supplementary material Fig. S6). All four constructs complemented tsBN2 cell proliferation at the restrictive temperature, as assessed by numbers of GFP-positive cells (Fig. 6A) or total numbers of cells (Fig. 6B). However, GFP-RCC1γ supported tsBN2 cell proliferation less well than GFP-RCC1α, whereas GFP-RCC1γ S11A worked as well as either GFP-RCC1α or GFP-RCC1α S11A (Fig. 6). Because the levels of GFP expression per cell declined during the time course of the experiment (supplementary material Fig. S6), we confirmed that GFP-RCC1γ was less efficient at supporting cell proliferation by analysing the number of cells on each day as a percentage of those transfected at day 0 (Fig. 6C). These results demonstrate that phosphorylation of RCC1γ at serine 11 regulates its cellular function and its ability to support cell proliferation.
RCC1 is the only known GEF for Ran GTPase, and as such plays a central role in cellular organisation and trafficking. We have identified and characterised a novel isoform of human RCC1 which we named RCC1γ. RCC1γ and a generally less abundant isoform, RCC1β, differ from RCC1α in the length of their N-terminal regions (NTR) and are likely to be generated by alternative splicing during pre-mRNA processing. The NTR has been previously identified in the context of the α isoform as the domain that is critical for interaction of RCC1 with chromatin in cells, and that binds importin α3-β in an interaction controlled by phosphorylation during mitosis. Our results show that the extended NTR of RCC1γ confers specific properties compared with RCC1α, namely that it has a much weaker interaction with importin α3-β and a stronger interaction with chromatin. It is also a better substrate for phosphorylation by CDK1-cyclin B and we show that phosphorylated RCC1γ preferentially associates with mitotic chromatin. Additionally, our cell proliferation assay shows that regulation of RCC1γ by phosphorylation plays an important role in its cellular function. By contrast, RCC1α is only weakly phosphorylated at serine 11 during mitosis and this phosphorylation does not promote chromatin enrichment of RCC1α. Thus, we show that RCC1 isoforms have divergent mechanisms of regulation and may play specialised cellular roles.
Sequence analysis of the human RCC1 gene suggests that RCC1γ is produced using a previously unidentified splice donor site within exon 6′. This results in expression of a protein that contains an insert of 17 amino acids after residue 24 compared to RCC1α. Alternative splicing around and within the 6′ exon of the RCC1 gene appears to be an evolutionarily conserved mechanism. The expression of RCC1α, RCC1β and RCC1γ in mammals, and the likely presence of multiple isoforms in other vertebrates, strongly suggests that they have distinct and conserved functions. In the cultured human cells that we analysed, all three isoforms are expressed in order of decreasing abundance α>γ>β. However, in normal human tissues, we find clear differences in the relative expression levels of the isoforms, with some tissues apparently lacking RCC1α. This strongly suggests specific roles for the different isoforms in differentiated cells.
We show that the NTR of RCC1γ is sufficient to localise a fusion protein to chromosomes, strongly supporting the proposal that differences in chromatin association between the full-length isoforms are due solely to their different NTRs and is not dependent on differences in Ran- or histone-binding ability. This difference is likely to be due, in part, to the intrinsically higher affinity of the NTR of RCC1γ for chromatin, which possibly involves a direct interaction with DNA. In addition, the binding of RCC1γ to chromatin or DNA is not competed by the binding of importin α3-β, in contrast to RCC1α, and this may promote the relatively more stable association of RCC1γ with chromatin in cells. The reduced binding of importin α3-β to RCC1γ also provides an explanation for the slightly greater proportion of cells with cytoplasmic GFP-RCC1γ than GFP-RCC1α (Fig. 2B), since the nuclear import of RCC1 is partly dependent on this interaction, although it is clear that RCC1α can also be localised to the nucleus in an NLS-independent manner (Nemergut and Macara, 2000; Moore et al., 2002), and this may account for the relatively small effect of the insert on RCC1γ localisation.
In mitosis, we find that RCC1γ also differs from RCC1α in its phosphorylation status at serine 11. The major phosphorylated endogenous isoform is RCC1γ, most likely because this isoform is an intrinsically better substrate for CDK1-cyclin B. Previous work on RCC1 phosphorylation indicated that mitotic phosphorylation disrupts the interaction of RCC1 with importin α3-β and also affects the dynamics of its interaction with chromatin. Phosphorylation of RCC1 was proposed to be required for the generation of a Ran-GTP gradient in mitosis (Hutchins et al., 2004; Li and Zheng, 2004). In the case of our previous experiments conducted using transfected RCC1α (which does become phosphorylated; supplementary material Fig. S4), the data indicated that phosphorylation of this isoform releases the protein from importin α3-β but also destabilises its dynamic interaction with mitotic chromatin in cells (Hutchins et al., 2004). This is supported by our new observation that endogenous RCC1α phosphorylated at serine 11 is released from mitotic chromatin during cell fractionation (Fig. 5A). However, we now show that the major phosphorylated isoform in mitotic cells is RCC1γ, and phosphorylation of this isoform at serine 11 is associated with stable interaction with chromatin, in contrast to RCC1α. Thus, phosphorylation of serine 11 differentially regulates RCC1 isoforms in mitosis.
Surprisingly, our experiments using non-phosphorylatable alanine mutants indicate that phosphorylation of serine 11 in the context of RCC1γ restrains tsBN2 cell proliferation, which is otherwise prevented at the restrictive temperature due to the loss of endogenous RCC1 isoforms. This demonstrates that mitotic phosphorylation at this site strongly regulates the function of RCC1γ. One explanation for RCC1γ being less efficient at supporting tsBN2 cell proliferation compared to its S11A mutant is that phosphorylation of this isoform at serine 11 enhances its activity to the extent that it is detrimental when the protein is expressed in the absence of other isoforms or at a higher level than normal. If this is the case, then phosphorylation would enhance the function of RCC1γ specifically during mitosis by stabilising its interaction with chromatin, where it would generate Ran-GTP in a highly localised manner. This would suggest that RCC1γ has a particularly important role during mitosis, whereas RCC1α may be the most important isoform during interphase. It remains conceivable, however, that phosphorylation of RCC1γ rather inhibits its function during mitosis, although the mechanism would be unclear.
To summarise, we have identified a novel isoform of mammalian RCC1, RCC1γ, which contains an insert in the NTR adjacent to the NLS. This insert reduces importin binding, increases the stability of its interaction with chromatin and makes it a better substrate for CDK1-cyclin B1 in mitosis. Phosphorylation of RCC1γ at serine 11 regulates its function in mitotic cells. The distinct biochemical properties of RCC1 isoforms, their different expression patterns in normal human tissues and the conservation of their expression in other mammals suggest that they have distinct roles in vivo.
Materials and Methods
Amplification of RCC1
RCC1α and RCC1γ cDNAs were amplified from HeLa cells by RT-PCR and cloned into the pENTR/D-TOPO entry vector of the Gateway system (Invitrogen) and transformed into E. coli DH5α. Colony PCR using primers designed to amplify a small section including the spliced region (5′-GCA TAG CTA AAA GAA GGT CCC-3′ and 5′-CCA ATC ACA CCG TTA TTG TCC-3′) was used to screen for different isoforms, as insert-containing isoforms were predicted to be of lower abundance. The LR Clonase reaction was used to transfer open reading frames (ORFs) into pDEST15 for bacterial expression with a GST tag. ORFs were also subcloned into pEGFP.C3, pEGFP.N1 and pGEX6p1. The linker region between GFP and the RCC1 coding sequences differs from that in a previously used GFP-RCC1α construct; the current constructs contain sequence encoding SGRTQISR between GFP and RCC1, whereas in the previously published construct this encoded YSDLE (Moore et al., 2002). The first 27 codons of RCC1α and codons 1-44 of RCC1γ (the NTRs) were amplified from plasmid template and subcloned into a modified pEGFP.C1 vector containing an ORF for GST in the multiple cloning site, to give GFP-GST-tagged constructs. Phosphorylation site mutants were generated with the QuikChange site-directed mutagenesis kit (Stratagene). RFP-Histone H2B was a gift from J. Swedlow, University of Dundee.
Recombinant proteins were expressed in E. coli BLR (DE3) cells as described previously (Moore et al., 2002). GST-RCC1 and GST-importin β proteins were purified using glutathione Sepharose 4B (GS4B, GE Healthcare) as described previously (Hutchins et al., 2004). For GST-tagged proteins with a cleavable tag, the tag was removed using PreScission protease (GE Healthcare) while protein was bound to GS4B. His6-importin α3 was purified on Ni-NTA agarose (Qiagen) as described previously (Hutchins et al., 2004).
Sequence alignment was performed using the ClustalW method. Accession numbers are as follows: Homo sapiens RCC1α, NM_001269; H. sapiens RCC1β, NM_001048194; H. sapiens RCC1γ, NM_001048195; Pan troglodytes RCC1-I, XP_513256; Macaca mulatta RCC1-I, XR_013696.1; Mesocricetus auratus RCC1-I (insert sequence only), P23800; Mus musculus, AAH57645; and Xenopus laevis RCC1-I, P25183. All sequences are available through NCBI.
Antibodies against the C terminus of RCC1 (Santa Cruz Biotechnology, C20), Importin β (Transduction Labs), Ran (Transduction Labs), actin (Sigma), GST (Molecular Probes), lamin B (Calbiochem), Crm1 (Transduction Labs), phospho-S10 histone H3 (Upstate), anti-caspase 3 (Santa Cruz Biotechnology, N-19) and GAPDH (Ambion) are all commercially available. Anti-phospho-serine 11 RCC1 antibody was purified from rabbit serum as before (Hutchins et al., 2004). For isoform specific antibodies, peptides derived from RCC1 sequences were synthesised: `R-INS', DTRAAASRRVPGARS and `R-DIS', CPKSKKVKVSHRSHST (Cancer Research UK, London Institute). Peptides were conjugated to KLH and used to raise polyclonal antibody sera in rabbits (Moravian Biotechnology, Brno, Czech Republic). Sera were negatively selected against the other isoform (RCC1α used for anti-R-INS and RCC1γ for anti-R-DIS) coupled to CNBr-activated Sepharose (GE Healthcare), then affinity purified against the appropriate peptide immobilised on Reacti-gel beads (Pierce Biotechnology).
U2OS and HeLa cells were obtained from Cancer Research UK London Research Institute. tsBN2 cells were a generous gift from Prof. T. Nishimoto (Fukuoka University, Japan) via Prof. H. Ponstingl (DKFZ, Heidelberg, Germany). Cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM, Invitrogen), supplemented with 10% foetal calf serum (Biosera), 2 mM L-glutamine (Invitrogen), 50 IU/ml Penicillin G (Invitrogen) and 50 μg/ml streptomycin (Invitrogen). Cells were grown at 37°C with 5% CO2, apart from tsBN2 cells which were grown at 32°C and shifted to 39.7°C to remove endogenous RCC1 protein. U2OS cells were transfected with 0.8 μg DNA/2 cm well (unless otherwise stated) using Superfect (Qiagen), whereas tsBN2 cells were transfected with 2 μg DNA/2 cm well using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions.
For live cell microscopy, cells were transfected and grown in glass-bottomed dishes (Willco) in phenol red-free DMEM (Invitrogen). Cells were imaged on a Zeiss Axiovert 200M microscope, in a 37°C Incubator XL-3 with 5% CO2, and photographed through a 63× or 100× lens using a cooled CCD camera (Hamamatsu) under the control of Volocity software (Improvision). Images for z stacks were taken at 1 μm intervals and deconvolved using Volocity. FRAP experiments were performed and analysed at the Light Microscopy Facility, College of Life Sciences, University of Dundee, as described previously (Hutchins et al., 2004). For microscopy using fixed cells, cells were fixed in 1:1 methanol acetone at –20°C as described previously (Hutchins et al., 2004) and counterstained with DAPI. Cells were visualised on a Zeiss Axioplan 2 microscope. Figures were assembled using Photoshop (Adobe).
For mitotic samples, cells were arrested in prometaphase using 100 ng/ml nocodazole for 17 hours, mitotic cells were collected by wash-off, and when required the remaining G2-arrested cells were collected by trypsinisation. For asynchronous samples, adherent cells were collected by trypsinisation. To arrest cells in G1-S phase, 3 mM hydroxyurea was added to the medium and cells incubated for 16 hours before harvesting by trypsinisation.
Mitotic or asynchronous cells were harvested, then washed three times with cold PBS and once with cold STM buffer [50 mM Tris-HCl pH 7.5, 250 mM sucrose, 5 mM MgCl2, 10 mM iodoacetamide, 0.1 mM phenylmethanesulphonyl fluoride (PMSF), 0.1 mM benzamidine, 1 μg/ml each of aprotinin, leupeptin and pepstatin A, and 1 μM okadaic acid (Biomol)] (Batchelor et al., 2004). Cells were resuspended in STM-N buffer (STM plus 0.5% NP40), then incubated for 5 minutes on ice. Half of the suspension was removed into sample buffer; the remaining suspension was centrifuged at 1000 g for 15 minutes to pellet nuclear material. The supernatant was removed into sample buffer, representing the soluble fraction. The pellet was resuspended in 1 ml STM buffer (STM without protease inhibitor or okadaic acid) and layered onto a 200 μl sucrose cushion (STM plus 40% sucrose), then centrifuged at 16,000 g for 15 minutes to re-pellet nuclear material. The pellet was resuspended in STM-N buffer to the same volume as the other samples, then added to SDS sample buffer. Equal volumes of each fraction were analysed by SDS-PAGE and immunoblotting.
HeLa cell extracts were made in EBS buffer (80 mM β-glycerophosphate, 20 mM EGTA, 15 mM MgCl2, 100 mM sucrose, 1 mM dithiothreitol (DTT), and 1 mM PMSF) as described previously (Hutchins et al., 2004).
HeLa extracts were supplemented with an ATP regenerating system (10 mM creatine phosphate, 40 μg/ml creatine kinase, and 1 mM ATP) and diluted in EBS to 7.5 mg protein/ml extract. For experiments in which recombinant importins were added, extracts were pre-incubated for 30 minutes at 30°C with 5 μM GST-importin β and/or 5 μM His6-importin α3. For GST-RCC1 precipitations, GS4B was loaded with GST-RCC1 proteins or GST in HBS (20 mM Hepes, pH 7.5, 150 mM NaCl) plus 2 mM DTT. GST-RCC1 beads were incubated with HeLa extract for 30 minutes at 30°C with shaking. Beads were washed in 20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100 and 2 mM EDTA. For GST-importin β precipitations, GS4B beads were loaded with GST-importin β or GST in IPB (50 mM Tris-HCl pH 7.5, 200 mM NaCl, 2 mM MgCl2, 5 mM β-mercaptoethanol) then incubated with HeLa extract for 45 minutes at 30°C with shaking. Beads were washed in IPB. For GST-RanT24N precipitations, GS4B beads were loaded with GST-RanT24N or GST in HBS plus 2 mM DTT, then incubated with HeLa extracts diluted 2.5 fold in HBS plus 2 mM DTT overnight at 4°C with rotation. Beads were washed in HBS plus 2 mM DTT. For DNA precipitations, DNA-cellulose beads (Sigma; at 0.5 μg/μl) in DCB (50 mM Tris pH 7.5, 50 mM NaCl, 10% v/v glycerol) were incubated with HeLa extract for 45 minutes at 30°C with shaking. Beads were washed in DCB. In each case, proteins were eluted in SDS sample buffer and analysed by SDS-PAGE and immunoblotting.
Chromatin-binding competition assays
The pellet fraction from subcellular fractionation of typically four 15 cm dishes of mitotic HeLa cells was resuspended in 200 μl HBS + 5 mM DTT and distributed in 20 μl aliquots. 10 μl of premix containing RCC1α, RCC1γ or GST; and His6-importin α3, GST-importin β or GST, as required was added. This gave final concentrations of 5 μM of each importin or 10 μM GST plus 2.2 μM RCC1 or 2.2 μM GST. The reaction was incubated for 15 minutes at 30°C. The reaction was resuspended in 100 μl HBS + DTT and spun through a 50 μl 40% sucrose cushion. The resulting pellet was resuspended in SDS sample buffer and analysed by SDS-PAGE and immunoblotting.
tsBN2 cell proliferation assay
This assay was similar to that used previously (Li and Zheng, 2004; Miyabashira et al., 1994). Cells (3×105) were seeded in 2 cm wells for transfection the next day. Cells were transfected and 1/8 of each well seeded onto coverslips. Cells were grown at 32°C for 24 hours to allow expression of constructs, then shifted to 39.7°C for the specified period, and the number of GFP-and DAPI-positive cells in 21 fields of view under 63× magnification were counted at each timepoint. Experiments were repeated to give four values for each data point, from which mean values were determined as a percentage of the number of cells at day 0 for that transfection.
Guanine nucleotide exchange assays
Guanine nucleotide exchange assays were performed as described previously (Nicolas et al., 2001) using 1.8 pmole RCC1 (3.6 nM) and 36 nM each of GST-importin β and His6-importin α3 where indicated, in an exchange reaction with 50 pmoles GST-Ran loaded with [8-3H]GDP.
FRAP was carried out at the University of Dundee Light Microscopy Facility with help from Sam Swift. Thanks to Sonia Lain and Oliver Staples for use of the microscope for live cell imaging. Thanks also to Helen Sanderson and Lindsey Allan. This work was supported by the Biotechnology and Biological Sciences Research Council, a Medical Research Council Studentship (F.E.H.) and a Royal Society-Wolfson Research Merit Award (P.R.C.).