Akt-mediated phosphorylation increases the binding affinity of hTERT for importin α to promote nuclear translocation

Telomeres, the nucleoprotein structures found at the physical ends of eukaryotic chromosomes, are essential for chromosome protection (Blackburn, 2001; Smogorzewska and de Lange, 2004). Mammalian telomeres comprise duplex tandem TTAGGG repeats that terminate in a 3′ single-stranded G overhang (Griffith et al., 1999), and they are coated with the shelterin complex (de Lange, 2005). Functional inhibition of shelterin has been shown to induce telomere loss, chromosome end-to-end fusion and inappropriate DNA repair reactions (Palm and de Lange, 2008); this inhibition also evokes a telomere damage response (Sfeir and de Lange, 2012). Telomere elongation is dependent on telomerase, a ribonucleoprotein enzyme that comprises the telomerase reverse transcriptase (hTERT in humans), the telomerase RNA component (TERC) and several additional proteins (Autexier and Lue, 2006; Bianchi and Shore, 2008). Telomerase offsets telomere erosion that results from a DNA-end replication problem by adding telomere repeats to the chromosome termini (Lingner et al., 1995). In humans, telomerase activity occurs primarily during the early stage of embryogenesis but, later in life, is strongly repressed in most normal somatic cells (Cong et al., 2002). However, telomerase is highly active in a majority of human cancers, indicating that the activation of telomerase is crucial for tumorigenesis (Kim et al., 1994).

Telomerase activity is regulated by hTERT at both the transcriptional and post-translational levels, implying that hTERT is a key component of the telomerase complex (Nakamura and Cech, 1998; Wang et al., 1998). For the assembly of catalytically active telomerase, cytoplasmic hTERT must be translocated to the nucleus through the nuclear pore complex (NPC). Proteins smaller than ∼40 kDa in size can pass through the NPC by simple diffusion, whereas nuclear localization of larger proteins requires an active nuclear transport pathway (Poon and Jans, 2005; Stewart, 2007). In the classic pathway, importin-α proteins recognize a nuclear localization signal (NLS) displayed by a cargo protein. The heterodimeric importin-α–NLS-cargo complex interacts with importin-β proteins and is translocated across the NPC through the interaction with nucleoporins (Cingolani et al., 1999,, 2002; Chook and Süel, 2011). Once translocation through the NPC is complete, RanGTP binds to importin-β proteins and induces dissociation of importin β from the importin-α–NLS complex, subsequently dissociating the cargo from importin α (Forwood et al., 2008; Lonhienne et al., 2009). Importin-α proteins can interact with two major classes of NLS – a monopartite NLS comprising a single cluster of basic amino acids and a bipartite NLS comprising two clusters of basic amino acids (Lange et al., 2007). The architecture of the importin-α protein comprises armadillo repeats, which includes three α-helices. Importin-α proteins contain two NLS-binding pockets – the major site and the minor site (Peifer et al., 1994; Conti et al., 1998). The monopartite NLS primarily binds to the major site, whereas the bipartite NLS interacts simultaneously with both sites (Fontes et al., 2000; Conti and Kuriyan, 2000). The NLS-binding pockets of importin α can also be occupied by the N-terminal importin-β-binding (IBB) domain, which contains basic amino acids (Harreman et al., 2003; Kobe, 1999), suggesting that NLS binding to importin α is regulated by an internal autoinhibitory mechanism.

Because the size of hTERT (∼124 kDa) precludes a passive nuclear transport mechanism, it should undergo active nuclear import, thereby regulating cellular telomerase activity. We have previously reported that hTERT contains a bipartite NLS (residues 222–240) and that Akt-mediated phosphorylation of residue S227 on hTERT plays an important role in the nuclear translocation of hTERT (Chung et al., 2012). Interestingly, S227 is located between two clusters of basic amino acids in the bipartite NLS. Thus, it is of interest to investigate whether phosphorylation of S227 affects the interaction between the hTERT NLS and the import receptors. Recently, importin 7 has been shown to interact with the hTERT NLS and to mediate nuclear transport of hTERT (Frohnert et al., 2014). However, the molecular mechanism regulating the nuclear import of hTERT remains unclear.

In this work, we demonstrate that nuclear import of hTERT is dependent on importin α and importin β proteins, and the cellular GTPase Ran. The hTERT NLS interacts with multiple importin-α-family members, and depletion of individual importin-α proteins inhibits the nuclear transport of hTERT. Crystallography analysis of a complex comprising importin-α5 (also known as KPNA1) and the NLS of hTERT reveals that the bipartite NLS binds to both the major and minor binding sites of importin α5. We also show that Akt-mediated phosphorylation of S227 increases the binding affinity of hTERT for importin-α proteins, resulting in effective nuclear localization of hTERT. These results provide a hTERT-NLS-binding mode in importin α5 and a nuclear transport control of hTERT through Akt activation.

We have previously identified a bipartite NLS (222-RRRGGSASRSLPLPKRPRR-240) that is responsible for the nuclear import of hTERT (Chung et al., 2012). Mutations in the two clusters of basic amino acids in the bipartite NLS results in reduced nuclear localization of hTERT. We have also shown that Akt-mediated phosphorylation at S227 is necessary for efficient nuclear translocation of hTERT (Chung et al., 2012). To determine whether nuclear localization of hTERT is affected by different cellular backgrounds, several cell lines – including MCF7, H1299 and U2OS cells – were transfected with Flag–hTERT and various mutant constructs (Flag-7A, Flag-S227A and Flag-S227E, see Fig. 1A), and then subjected to immunofluorescence analyses using an antibody against Flag (supplementary material Fig. S1). The results indicated that both the bipartite NLS and Akt-mediated phosphorylation at S227 are required for efficient nuclear import of hTERT, suggesting that the mechanism of nuclear localization of hTERT is not cell-type specific.

To determine whether the hTERT NLS contains all of the determinants that are sufficient to confer nuclear import, the extended bipartite NLS (residues 212–250) was fused to green fluorescent protein (GFP, 26 kDa) and β-galactosidase (β-gal, 112 kDa) to construct the GFP–NLS–β-gal expression vector (Fig. 1A). Because the expected size of the fusion protein (∼143 kDa) prevents passive diffusion across the NPC, any nuclear accumulation of the protein should be the result of active transport by the soluble receptors. To determine the subcellular localization of the fusion protein, MCF7 cells were transfected with various GFP–NLS–β-gal constructs and subjected to immunofluorescence staining (Fig. 1A). The majority of the GFP–β-gal signal was detected in the cytoplasm (86.7% cytoplasmic), whereas the fusion protein that included the wild-type NLS (GFP–NLS–β-gal) showed a pronounced nuclear localization (88.6% nuclear; Fig. 1B,C). When the two clusters of basic amino acids in the NLS were mutated to alanine residues (7A and S227A/7A), the fusion protein signals were mostly found in the cytoplasm (81.2% cytoplasmic for 7A, and 89.51% cytoplasmic for S227A/7A). When S227 was mutated to alanine, the nuclear fluorescence signals were considerably decreased (49.5% nuclear for S227A). Interestingly, the phosphorylation-mimicking mutant S227E was mostly localized to the nucleus (79.1% nuclear). Taken together, these results demonstrate that the NLS sequence plays a major role in the nuclear translocation of hTERT.

To examine whether Akt-dependent phosphorylation at S227 is required to mediate nuclear import of hTERT, we determined the effect of an Akt inhibitor (LY294002, which inhibits all Akt isoforms) on subcellular localization of the fusion protein. When cells were treated with the inhibitor, nuclear accumulation of GFP–NLS–β-gal was substantially reduced (Fig. 1D,E). By contrast, subcellular localization of GFP–S227A–β-gal was not affected by treatment with the inhibitor (Fig. 1F), highlighting the requirement of Akt-dependent phosphorylation at S227.

It has been well documented that a protein containing a classic basic NLS is imported by a heterodimeric import receptor comprising importin-α and importin-β subunits (Chook and Süel, 2011; Cingolani et al., 2002,, 1999). Seven members of the importin-α family have been described in humans; these are classified into three subfamilies, and each is encoded by a different gene (Kelley et al., 2010). Because the nuclear import of hTERT is mediated by a classic bipartite NLS, we wanted to determine whether hTERT directly interacts with importin-α proteins. Importin α1, importin α3, importin α5 and importin β1 (encoded by KPNA2, KPNA4, KPNA1 and KPNB1, respectively) were fused to glutathione S-transferase (GST); the fusion proteins were purified and then used in an in vitro binding assay. GST–importin-α proteins, but not the control GST, bound to Flag–NLS, indicating that hTERT interacts with importin-α proteins through the NLS (Fig. 2A). We also found that Flag–NLS was precipitated by GST–importin-β1, suggesting that the NLS of hTERT can form a complex with importin-α–importin-β1 dimers. Interestingly, the S227A/7A mutation severely impaired the interactions between hTERT and importin-α proteins (Fig. 2A). Likewise, GST–NLS, but not the control GST or GST–S227A/7A mutant, precipitated hemagglutinin (HA)-tagged importin α1, importin α3 and importin α5 when they were expressed in MCF7 cells (Fig. 2B). Consistently, importin β1 was also found to associate with GST–NLS. To determine whether the NLS of hTERT associates with importin-α proteins in vivo, MCF7 cells were transfected with Flag–NLS and HA-tagged importin-α proteins, and then subjected to immunoprecipitation. All importin proteins were immunoprecipitated by Flag–NLS (Fig. 2C). Importin β1 was also recovered by Flag–NLS.

Because the importin-α–NLS complex interacts with importin β1, we further verified the formation of a complex comprising the importin-α–NLS heterodimer and importin β1. As shown in Fig. 2D, HA–importin-β1 was immunoprecipitated by Flag–NLS only when one of the importin-α proteins was co-expressed. To confirm direct interactions with endogenous proteins, we performed immunoprecipitation analyses using MCF7 cells and found that endogenous importin α1, importin α3, importin α5 and importin β1 could be immunoprecipitated by endogenous hTERT (Fig. 2E). Taken together, these results suggest that the nuclear import of hTERT is mediated by the importin-α–importin-β pathway.

To examine the involvement of importin-α proteins in a more physiological setting, endogenous importin-α proteins were depleted using two different small interfering (si)RNA duplexes specific to each importin α. Depletion of importin-α proteins led to a clear reduction in the level of endogenous hTERT (Fig. 3A). This reduction in the level of hTERT was rescued by treatment with MG132, suggesting that degradation of hTERT is mediated by the proteasome. To investigate these findings in more detail, we determined the effect of depleting importin-α proteins on the subcellular localization of hTERT. Nuclear and cytosolic extracts were separately collected from MCF7 cells and subjected to immunoblot analysis to assess the level of endogenous hTERT. In the presence of MG132, the majority of hTERT was localized to the nucleus in the control siRNA cells (Fig. 3B). When importin-α proteins were depleted, substantial amounts of hTERT were detected in the cytoplasm. These results indicate that nuclear import of hTERT is inhibited by the depletion of importin-α proteins. We also examined whether the depletion of importin α is involved in regulating the half-life of hTERT. MCF7 cells were transfected with Flag–hTERT and treated with cycloheximide to inhibit new protein synthesis. The stability of hTERT was monitored by immunoblotting with an antibody against Flag. Depletion of importin-α proteins substantially reduced the half-life of Flag–hTERT, as compared with that in control siRNA cells (supplementary material Fig. S2A,B). The protein levels were rescued when the transfected cells were treated with MG132.

To investigate whether hTERT is ubiquitylated before it is degraded in a proteasome-dependent manner, MCF7 cells were co-transfected with Flag–hTERT and HA–ubiquitin. To visualize ubiquitin-modified hTERT, immunoprecipitations of the Flag tag were evaluated by immunoblotting with an antibody against HA. Ubiquitylated hTERT was barely detected in the control siRNA cells in the presence of MG132, whereas depletion of any of the importin-α proteins resulted in a marked increase in hTERT ubiquitylation (Fig. 3C). Because mutations in the two clusters of basic amino acids and S227 (7A/S227A) inhibit nuclear import of hTERT, the resulting cytoplasmic hTERT could be ubiquitylated before it is degraded by the proteasome. To test this possibility, MCF7 cells were co-transfected with HA–ubiquitin and either Flag–hTERT or Flag–7A/S227A. Ubiquitylation of the 7A/S227A construct was higher than that of wild-type hTERT in the presence MG132 (supplementary material Fig. S2C). These results demonstrate that, when nuclear import is inhibited by depletion of importin-α proteins or by mutations in the NLS, the resulting cytoplasmic hTERT is rapidly ubiquitylated and degraded by the proteasome.

We next examined whether depletion of importin-α proteins influences telomerase activity. As shown in Fig. 3D, telomerase activity was considerably reduced in cells in which importin-α proteins had been knocked down compared with that in control siRNA cells. To determine the effect of importin-α depletion on the nuclear localization of hTERT, we performed an indirect immunofluorescence assay using a confocal laser-scanning microscope. Endogenous hTERT predominantly localized to the nucleus in the control siRNA cells, whereas depletion of importin-α family members resulted in a substantial decrease in the total intensity of hTERT fluorescence (Fig. 3E). This reduction in hTERT fluorescence could be rescued by treatment with MG132. We also found that cytoplasmic accumulation of endogenous hTERT was increased in the presence of MG132 (Fig. 3E). These results support the idea that hTERT fails to be translocated into the nucleus when importin-α proteins are depleted and that hTERT is subsequently degraded by the proteasome. Because the specificity of the antibody against hTERT is crucial to the interpretation of the data presented, we verified that small hairpin (sh)RNA-mediated depletion of hTERT reduced the immunofluorescence staining pattern (supplementary material Fig. S2D). Depletion of importin β1 was also found to inhibit nuclear import of hTERT and to reduce telomerase activity (supplementary material Fig. S4A,B), suggesting that importin β1 is required for nuclear localization of hTERT. By contrast, overexpression of importin-α proteins and importin β1 had no effect on the nuclear import efficiency of hTERT and telomerase activity (data not shown).

Conventional importin-α–importin-β-dependent nuclear transport through the NPC is dependent on the nucleotide-bound state of the small GTPase Ran, which affects the disassembly of importin–cargo complexes (Forwood et al., 2008; Lonhienne et al., 2009). To examine the dependence of the subcellular localization of hTERT on Ran, a dominant-negative mutant of Ran (RanQ69L) that is unable to hydrolyze GTP to GDP was used to inhibit importin-α–importin-β-dependent nuclear protein import (Dickmanns et al., 1996; Palacios et al., 1996). MCF7 cells expressing V5-tagged Ran (V5–Ran) or V5-tagged RanQ69L (V5–RanQ69L) were subjected to immunofluorescence staining in order to monitor the localization of endogenous hTERT. A clear accumulation of hTERT in the nucleus was observed in cells that expressed the empty vector or V5–Ran (Fig. 4A). By contrast, when RanQ69L was overexpressed, the hTERT signal was detected in a diffuse pattern throughout the nucleus and cytoplasm in the presence of MG132. These results demonstrate that Ran plays an essential role in the nuclear import of hTERT.

We next determined the effect of RanQ69L on the stability of hTERT. Overexpression of RanQ69L resulted in a clear reduction in the level of Flag–hTERT compared with that in cells expressing the empty vector or wild-type Ran (Fig. 4B). However, the RanQ69L-induced reduction in hTERT was rescued in the presence of MG132. Because the hTERT NLS interacts with importin-α family members and importin β1 for its nuclear import, we determined whether overexpression of RanQ69L affects the formation of heterotrimeric complex. Overexpression of RanQ69L severely impaired the interactions of hTERT with importin-α proteins and importin β1, indicating that Ran is required for the nuclear import of hTERT (Fig. 4B). When cells that expressed RanQ69L were analyzed for the telomerase activity, overexpression of RanQ69L was seen to lead to a substantial reduction in telomerase activity compared with that in the control cells (Fig. 4C). Moreover, this reduction in telomerase activity was not rescued by treatment with MG132, suggesting that cytoplasmic hTERT is catalytically nonfunctional for telomerase activity.

Telomerase undergoes a highly elaborate, stepwise process for the assembly of the telomerase holoenzyme (Venteicher et al., 2008,, 2009; Lee et al., 2014). After translocation to the nucleus, hTERT associates with telomerase RNA component (TERC) and dyskerin to generate catalytically active telomerase. We determined the effect of importin-α depletion on the telomerase ribonucleoprotein (RNP) assembly. The protein level of hTERT was reduced by depletion of importin-α proteins, whereas the protein level and nuclear localization of dyskerin were unaffected (Fig. 5A). We also found that the interaction between hTERT and dyskerin was dramatically reduced by depletion of importin-α proteins (Fig. 5A). We next examined whether dyskerin associates with catalytically active telomerase when importin α is depleted. Importin-α-knockdown cells were immunoprecipitated with a dyskerin-specific antibody and analyzed for telomerase activity. Depletion of importin-α proteins resulted in a substantial reduction in telomerase activity compared with that of the control siRNA cells (Fig. 5B).

The results presented here demonstrate that depletion of importin-α proteins inhibits the nuclear translocation of hTERT and that the resulting hTERT is degraded through the proteasome but accumulates in the cytoplasm when cells are treated with MG132. To determine whether cytoplasmic hTERT is assembled with the TERC–dyskerin RNP to form catalytically active telomerase, nuclear and cytosolic extracts were collected from importin-α-knockdown cells in the presence of MG132. As shown in Fig. 5C, telomerase activity was exclusively detected in the nuclear extracts, demonstrating that cytoplasmic hTERT is not assembled into active telomerase. When immunoprecipitates of dyskerin were analyzed for telomerase activity, similar results were observed (Fig. 5C). These findings suggest that depletion of importin α causes a reduction in telomerase activity by inhibiting the association of hTERT with dyskerin in the nucleus.

We have previously shown that Akt-mediated phosphorylation at S227 is required for efficient nuclear translocation of hTERT (Chung et al., 2012). To explore the molecular mechanism by which phosphorylation of S227 is involved in the nuclear import of hTERT, we used isothermal titration calorimetry (ITC) to determine the binding affinity of importin α1 and importin α5 for an NLS peptide that was phosphorylated at S227 (phospho-S227) compared with that for wild-type NLS peptide. Phosphorylated NLS peptide (at S227, bold, ²²⁰GARRRGGpSASRSLPLPKRPRRGA²⁴²) was injected into an ITC cell containing 20 μM of the importin-α proteins, and the enthalpy of binding was carefully monitored. As a control experiment, wild-type NLS peptide was injected against the importin-α proteins under identical experiment conditions. The binding affinity of NLS peptide to importin α5 was increased approximately 2-fold when S227 was phosphorylated and decreased about 1.4-fold when S227 was mutated to alanine (Fig. 6A–C). Although the binding affinity of importin α1 is lower than that of importin α5, a similar binding trend was obtained for the binding affinity of NLS peptides to importin α1 (Fig. 6D–F). These findings clearly indicate that the phosphate moiety of S227 functions as a direct binding determinant for importin-α proteins.

To provide further evidence for the molecular mechanism of hTERT nuclear import, we solved the complex structure of importin α5 bound to the hTERT NLS peptide at 2.5-Å resolution (Fig. 7A). The results showed a canonical bipartite binding mode of importin α5 and NLS (Fig. 7B). Major interactions of the side chain of the basic amino acid residues in both the minor (²²²RRR²²⁴) and major (²³⁶KRPRR²⁴⁰) sites are within acidic pockets of importin α5 and supplemented by additional hydrogen bonds (Fig. 7B). In the minor site, residue R222 interacts with residue Q399 through a salt bridge and interacts with residue W402 through a hydrophobic interaction (Fig. 7C). Residue R223 interacts with residues G284 and T325 through hydrogen bonds. Residue R224 interacts with residue Q357 through a salt bridge. In the linker region between the major and minor sites, residue R230 interacts with residue N322 through hydrogen bonds. The carbonyl oxygen of residue S231 binds to R318 through hydrogen bonds. Residue P233 interacts with residue W276 through van der Waals interaction, and its carbonyl group interacts with residue R241 through hydrogen bonding. In the major site, the ε-amino group of residue K236 interacts with the carboxyl groups of residues A155 and G157, and the side chains of residues T162 and N199 as a core of the hydrogen bond network (Fig. 7D). The amino and carboxyl groups of residue R237 interacts with residues N195 and W191, respectively, through hydrogen bonds. The side chain of residue R237 stacks on W234. The amino and carboxyl groups of residue R239 interact with residues N153 and W149 through hydrogen bonding, and its side chain interacts with residues Q188 and W191 through hydrogen and hydrophobic interactions, respectively. Residue R240 interacts with residues Q113 and S111 through electrostatic and hydrophobic interactions, respectively. The amino acid residues 225–228 of NLS are not visible in the electron density, presumably owing to flexibility. These interactions have been reported in previous structures of importin-α–NLS complexes (Conti and Kuriyan, 2000; Tarendeau et al., 2007; Fontes et al., 2003).

Although phospho-S227 was not seen in the electron density map because of the disordered region, a possible structural model can be proposed by using molecular dynamics and Autodock tools (Morris et al., 2009). In this structural model, phospho-S227 appears to interact with residue R395 of importin α5 through hydrogen bonding (Fig. 8A–C). To confirm the importance of this interaction, we performed site-directed mutagenesis of importin α5, mutating residue R395 to lysine (R395K) or glutamate (R395E). In the ITC analysis, we found that the binding affinity of the R395K and R395E mutants for phosphorylated NLS peptide decreased approximately 4.3-fold (K_d, 202.9 nM) and approximately 14.1-fold mutant (K_d, 666.7 nM), respectively, compared with that of wild-type importin α5 (K_d, 47.2 nM) (Fig. 8D–F). As would be expected based on our model, the binding affinity for unphosphorylated NLS peptide remained unchanged regardless of mutations in R395 (supplementary material Fig. S3A–C). These findings indicate that the positive charge of R395 is crucial for binding to the phosphate moiety of S227, and that Akt-mediated phosphorylation of S227 increases the binding affinity of hTERT for importin α5 (supplementary material Fig. S3D).

In order for telomeric ends to be elongated by telomerase in vivo, newly synthesized hTERT has to be translocated into the nucleus and assembled with the RNA component to generate the active telomerase holoenzyme (Venteicher et al., 2008,, Venteicher et al., 2009; Lee et al., 2014). The hTERT NLS has been shown to be bipartite, comprising the first basic amino acid residues 222-RRR-224, a 12-residue spacer and the second residues 236-KRPRR-240 (Chung et al., 2012). In this work, we have studied the molecular mechanism by which hTERT enters the nucleus. The extended NLS can confer nuclear import on a heterologous protein through the classical importin-α–importin-β-dependent pathway. In humans, there are at least seven genes that encode importin α proteins (Kelley et al., 2010). These subtypes share the same general structure and exhibit a broad functional redundancy. The bipartite NLS of hTERT is able to interact with importin α1, importin α3 and importin α5. Crystallography analysis of importin α5 in complex with the hTERT NLS reveals the ability of the basic elements to interact with both the major and minor sites of importin α5. We also show that overexpression of a dominant-negative mutant of Ran (RanQ69L) inhibits nuclear import of hTERT, suggesting that nuclear import of hTERT is dependent on the cellular GTPase Ran and soluble import receptors – importin α and importin β1.

Although the classic NLS-receptor complex of importin α can mediate nuclear import of hTERT, importin α cannot be the sole import receptor for hTERT. Importin 7 has been recently shown to interact with hTERT and to mediate nuclear transport of hTERT (Frohnert et al., 2014). To confirm these results in our system, we examined whether importin 7 binds to the hTERT NLS, and whether its depletion affects the levels and subcellular localization of hTERT and telomerase activity. We found that endogenous importin 7 can interact physically with the hTERT NLS (supplementary material Fig. S4C,D). When importin 7 was depleted, telomerase activity was slightly reduced compared with that of the control siRNA cells (supplementary material Fig. S4E). Depletion of importin 7 led to a clear reduction in the nuclear accumulation of hTERT (supplementary material Fig. S4F). We also found that the total intensity hTERT fluorescence was reduced upon importin 7 depletion and that this reduction in hTERT level was rescued by treatment with MG132. Taken together, these results suggest that importin 7 mediates nuclear localization of hTERT. Thus, it is likely that hTERT can be imported into the nucleus through multiple nuclear import pathways. Although redundancy in nuclear import pathways is a usual phenomenon and is observed for many abundant cargo substrates, such as ribosomal proteins (Jakel and Gorlich, 1998) or histones (Muhlhausser et al., 2001), our results offer no clear indication of the comparative transport efficiencies of the different pathways in vivo. Therefore, it will be of interest to investigate how these pathways are specifically regulated and how they are activated under different conditions.

Akt kinase has been shown to facilitate the nuclear import of hTERT through phosphorylation and to enhance telomerase activity (Chung et al., 2012). Two putative Akt phosphorylation sites, S227 and S824, have been identified previously in hTERT (Kang et al., 1999). Although both serine residues have been shown to be phosphorylated by Akt kinase, phosphorylation of S227 appears to be important for nuclear import of hTERT (Chung et al., 2012). Because residue S227 is located between two clusters of basic amino acids in the bipartite NLS sequence, Akt-mediated phosphorylation of S227 might contribute to the interaction of the hTERT NLS with importin-α proteins. Structural characterization of importin α5 bound to the hTERT NLS showed that the phosphate moiety of S227 interacts with R395 of importin α5 through hydrogen bonding. When arginine was replaced by lysine or glutamate at position 395, the binding affinity for phosphorylated NLS peptide was substantially decreased. These findings suggest that the interaction between the phospho-S227 and R395 residues of importin α5 contributes to the efficient nuclear localization of hTERT by increasing the binding affinity of hTERT for importin α5. In addition, biochemical studies using chemically synthesized peptides and ITC showed that phosphorylation of S227 increases the binding affinity for importin α5, compared with that of the unphosphorylated NLS. Thus, the phosphate moiety within the NLS functions as a direct binding determinant for importin α5.

There are several well-documented examples of importin cargos that are regulated by phosphorylation (Nardozzi et al., 2010). One example is the Epstein-Barr virus (EBV) nuclear antigen 1 (EBNA-1) protein (Kitamura et al., 2006). Phosphorylation of residue S385 within the EBVA-1 NLS upregulates the nuclear transport efficiency by increasing its binding affinity for importin α5. In the case of the hepatitis B virus (HBV) core particle, protein kinase C (PKC)-mediated phosphorylation at residue S172 causes a conformational change that exposes the C-terminal NLS on the surface. Upon phosphorylation, the HBV core particle is recognized by importin α1 and is then imported into the nucleus (Kann et al., 1999). Phosphorylation by the protein kinase casein kinase 2 and double-stranded DNA-dependent protein kinase upstream of the classic SV40 NLS promotes the nuclear import of large T-antigen by enhancing its recognition by importin α1 (Hübner et al., 1997; Xiao et al., 1997). Therefore, phosphorylation-mediated regulation of nuclear transport could provide a general mechanism for many more nuclear proteins. Although nuclear accumulation of a cargo is dependent on the NLS, the rate of nuclear import is regulated through phosphorylation within or near the NLS of the cargo (Nardozzi et al., 2010). Thus, it is likely that the nuclear import efficiency of a cargo is determined by the addition and removal of a phosphate moiety, kinases and phosphatases – the activities of which are tightly regulated by diverse biological pathways in different cellular environments. On the basis of a previous report (Chung et al., 2012) and our data presented here, we propose that two functional determinants serve as molecular switches controlling nuclear localization of hTERT – the bipartite NLS and Akt-mediated phosphorylation of S227. The interplay between these two signals is likely to ensure the proper cellular function of telomerase activity by determining the nuclear import efficiency of hTERT.

When importin α is depleted or RanQ69L is overexpressed, hTERT fails to be translocated into the nucleus, and the resulting cytoplasmic hTERT is degraded through ubiquitin-dependent proteolysis. Ubiquitin-dependent degradation of hTERT was efficiently inhibited by treatment with MG132, suggesting that proteasomal degradation of hTERT occurs in a step separate from that of inhibition of its nuclear import. Several E3 ubiquitin ligases, including CHIP (Lee et al., 2010), MKRN1 (Kim et al., 2005) and Hdm2 (Oh et al., 2010), have been identified as targeting hTERT for ubiquitylation. Although all E3 ligases are each capable of promoting ubiquitylation of hTERT, the functional similarity raises many important questions about the physiological significance of multiple pathways that exert negative control over hTERT. The Hsp90 and p23 molecular chaperones have been shown to associate with hTERT to maintain its conformation, enabling nuclear translocation (Holt et al., 1999; Forsythe et al., 2001). By contrast, the binding of CHIP and Hsp70 to hTERT causes a dissociation of p23 from hTERT, resulting in a cytoplasmic accumulation of hTERT that cannot be translocated into the nucleus (Lee et al., 2010). Cytoplasmic hTERT is rapidly ubiquitylated by CHIP and other ubiquitin ligases as well. Based on these findings, we propose a model for chaperone-mediated nuclear import of hTERT – the hTERT–CHIP–Hsp70 complex favors degradation of hTERT in the cytoplasm, whereas the hTERT–Hsp90–p23 complex is preferentially bound by nuclear import receptors and then translocated into the nucleus. Thus, p23 and the CHIP–Hsp70 complex might compete for binding to hTERT in the cytoplasm, and the outcome of this competition is likely to determine the fate of hTERT – nuclear translocation or degradation.

In this study, we show that the nuclear localization of hTERT requires the classic nuclear import machinery, comprising importin α, importin β and GTPase Ran. Crystallography analysis demonstrated that the basic amino acid clusters within the bipartite NLS interact with both the major and minor sites of importin α in a manner that is characteristic of that of other bipartite NLSs (Fontes et al., 2003). Importantly, we show that Akt-mediated phosphorylation of hTERT is an important step that regulates its nuclear availability, which directly affects telomerase activity. Although many important questions about how hTERT phosphorylation is regulated under diverse physiological conditions remain, Akt-mediated phosphorylation of hTERT represents a new pathway for determining proper cellular telomerase activity by regulating its nuclear import efficiency.

The human breast cancer cell line MCF7 was cultured in Dulbecco's modified Eagle's medium (DMEM), the human lung carcinoma cell line H1299 was cultured in RPMI 1640 medium, and the human osteosarcoma cell line U2OS was cultured in McCoy's 5A medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin and 100 μg/ml streptomycin under 5% CO₂ at 37°C. The expression vectors were transiently transfected using Lipofectamine-PLUS reagent according to the manufacturer's protocol (Invitrogen).

To construct the GFP–NLS–β-gal expression vectors, the wild-type and mutant NLS fragments were generated by using PCR and subcloned into the XhoI and HindIII sites of the pEGFP-C3-β-gal vector (BD Biosciences). The expression vectors for HA–importin-α proteins, HA–importin-β1, GST–importin-α proteins and GST–importin-β1 were constructed by inserting the whole coding regions into the MluI and NotI sites of the pRK5-HA vector (Stratagene) and the BamHI and NotI sites of the pGEX-5X-1 vector (Amersham Biosciences), respectively. The expression vectors for V5–Ran and V5–RanQ69L were constructed by inserting cDNAs encoding wild-type and mutant Ran into the EcoRI and XhoI sites of pcDNA3.1/V5-His vector (Invitrogen). N-terminally truncated importin α5 containing amino acid residues 66–512 were mutated to generate the R395K and R395E mutants, and subcloned into the EcoRI and XhoI sites of the pET28a-TEV vector.

GST pulldown, immunoprecipitation and immunoblot analyses were performed as described previously (Lee et al., 2004). Immunoprecipitation and immunoblotting were performed using anti-Flag (cat. no. F3165, Sigma-Aldrich), anti-HA (cat. no. sc-805, Santa Cruz Biotechnology), anti-GST (cat. no. sc-138, Santa Cruz Biotechnology), anti-Myc (cat. no. sc-40, Santa Cruz Biotechnology), anti-importin-α1 (cat. no. NB 100-79807, Novus Biologicals), anti-importin-α3 (cat. no. NBP1-31260, Novus Biologicals), anti-importin-α5 (cat. no. sc-101292, Santa Cruz Biotechnology), anti-importin-β1 (cat. no. sc-11367, Santa Cruz Biotechnology), anti-importin-7 (cat. no. GTX 106408, GeneTex), anti-hTERT (cat. no. 600-401-252, Rockland), anti-dyskerin (cat. no. sc-48794, Santa Cruz Biotechnology), anti-actin (cat. no. A2066, Sigma-Aldrich), anti-tubulin (cat. no. sc-8035, Santa Cruz Biotechnology) and anti-lamin (cat. no. sc-6215, Santa Cruz Biotechnology) antibodies as specified. All the immunoblots are representatives of at least three experiments that demonstrated the similar results.

After transfection, cells were harvested and resuspended in hypotonic buffer (10 mM HEPES-KOH, pH 7.9, 1.5 mM MgCl₂, 10 mM KCl, 0.5 mM DTT, 0.2 mM phenyl methylsulphonyl fluoride) for 10 min on ice, followed by the addition of 0.5% NP-40 for 10 min. After centrifugation at 14,000 g for 5 min at 4°C, the supernatant was used as the cytoplasmic fraction. The nuclear pellets were resuspended in nuclear extraction buffer (20 mM HEPES-KOH, pH 7.9, 25% Glycerol, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM DTT, 0.2 mM phenyl methylsulphonyl fluoride) for 20 min on ice and centrifuged at 14,000 g for 20 min at 4°C. The supernatants were saved as the nuclear fraction.

Cells that had been grown on glass coverslips were fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100 in PBS. Cells were then blocked in PBS containing 0.5% bovine serum albumin and 0.2% gelatin, and incubated with mouse anti-Flag (cat. no. F3165, Sigma-Aldrich), rabbit anti-HA (cat. no. sc-805, Santa Cruz Biotechnology), rabbit anti-hTERT (cat. no. 600-401-252, Rockland) or rabbit anti-dyskerin (cat. no. sc-48794, Santa Cruz Biotechnology). After thorough washing with PBS, cells were incubated with Alexa-Fluor-594-conjugated goat anti-mouse immunoglobulin (cat no. A11005, Molecular Probes) and Alexa-Fluor-488-conjugated goat anti-rabbit immunoglobulin (cat no. A11008 , Molecular Probes). DNA was stained with 4,6-diamino-2-phenylindole (DAPI) (Vectashield; Vector Laboratories). Immunofluorescence images were captured using a confocal laser-scanning microscope (Carl Zeiss).

The siRNA duplexes were transfected into MCF7 cells using RNAiMax transfection reagent (Invitrogen). The scrambled sequence (5′-AATCGCATAGCGTATGCCGTT-3′) was used as a control and did not correspond to any known gene in the data bases. Importin α1-1 siRNA, 5′-TAGCATGTGGCTACTTACGTA-3′; importin α1-2 siRNA, 5′-TAGGAGCTTCTGAATTGCCAA-3′; importin α3-1 siRNA, 5′-ATGGGATCTTGCTGCTGCATTA-3′; importin α3-2 siRNA, 5′-TAGGCTGCCCATATAAGTCAA-3′; importin α5-1 siRNA, 5′-ATGGGATCTGCTGCTGCATTA-3′; importin α5-2 siRNA, 5′-AAAGATCTTTGCATGAGGTAA-3′; importin β1 siRNA, 5′-GTGCAACTCTTCAGAATGT-3′; importin 7-1 siRNA, 5′-GATGGAGCCCTGCATATGA-3′; importin 7-2 siRNA, 5′-GGCAGGTGTTATCTATCTG-3′.

MCF7 cells were co-transfected with HA–ubiquitin and Flag–hTERT, and treated with 10 μM MG132 to inhibit proteasome function. Lysates were subjected to immunoprecipitation with an anti-Flag antibody, followed by immunoblotting with an anti–HA antibody to show ubiquitin-modified hTERT.

The telomeric repeat amplification protocol (TRAP) was used as previously described (Lee et al., 2004).

ITC experiments were performed at 25°C in a VP-ITC calorimeter (Microcal Software) (Pierce et al., 1999). The hTERT NLS peptides (wild-type, phospho-S227 and S227A) were dissolved in ITC buffer (20 mM Tris-pH 7.5, 150 mM sodium chloride) and injected in 10-μl increments into a calorimetric cell containing 1.8 ml of the importin-α proteins. The interval between injections was 220 s. In separate experiments, the NLS peptides were titrated into the importin-α proteins, as described above. Titration data were analyzed using the Origin 7.0 data analysis software (Microcal Software). Injections were integrated following manual adjustment of the baselines. Heats of dilution were determined from control experiments with the ITC buffer and subtracted prior to curve fitting using a single set of binding site models.

For crystallization, purified importin α5 (residues 66–512) was concentrated to 25 mg/ml using a Centricon-30 filter (Amicon) and mixed with phospho-S227 NLS peptide in a 1:5 molar ratio. Crystallization conditions were screened by using micro-bath plates. The best crystals of the importin-α5–NLS complex were obtained with 17% poly(ethylene glycol) 10 K as the precipitant in 100 mM HEPES (pH 7.5) in hanging drop plates. Diffraction data were collected from single crystals with 20% ethylene glycol and flash frozen at 100 K in a nitrogen stream. Data to 2.5-Å resolution were measured on beam line BL5A at the Photon Factory synchrotron using X-rays of wavelength 1.0 Å, and were auto-indexed and processed with the HKL suite. The complex crystals of importin–peptide exhibit monoclinic (space group C2). The complex was located in the asymmetric unit by molecular replacement using PDB 2JDQ (importin α5 with PB2 subunit) as the search model (Tarendeau et al., 2007). Coot was used for visualization of electron density maps and manual rebuilding of atomic models (Emsley and Cowtan, 2004). The initial model was refined using REFMAC5 (Vagin et al., 2004) at 2.5-Å resolution.

We are grateful to Joonyoung Her, Yu Young Jeong, and Juyeong Hong for technical assistance and helpful comments on the manuscript and to the staff scientists at the BL-5A beamline of the Photon Factory.