Upon nuclear envelope (NE) fragmentation in the prometaphase, the nuclear and cytosolic proteomes mix and must be redefined to reinstate homeostasis. Here, by using a molecular GFP ladder, we show that in early mitosis, condensed chromatin excludes cytosolic proteins. When the NE reforms tightly around condensed chromatin in late mitosis, large GFP multimers are automatically excluded from the nucleus. This can be circumvented by limiting DNA condensation with Q15, a condensin II inhibitor. Soluble small and other nuclear localization sequence (NLS)-targeted proteins then swiftly enter the expanding nuclear space. We then examined proteasomes, which are located in the cytoplasm and nucleus. A significant fraction of 20S proteasomes is imported by the importin IPO5 within 20 min of reformation of the nucleus, after which import comes to an abrupt halt. This suggests that maintaining the nuclear–cytosol distribution after mitosis requires chromatin condensation to exclude cytosolic material from the nuclear space, and specialized machineries for nuclear import of large protein complexes, such as the proteasome.
Compartmentalization between the cytoplasm and nucleus is the defining characteristic of eukaryotic organisms, or those composed of ‘true’ cells. This separation in eukaryotic cells is afforded by the complex double-membrane structure of the nuclear envelope (NE). Exchange of materials across this barrier occurs primarily through the nuclear pore complex (NPC), which allow passive diffusion of substances up to a size of ∼70 kDa, although recent work suggests that specific NPC qualities can infer some flexibility to this cutoff (Popken et al., 2015; Timney et al., 2016). Generally, molecules above that size cutoff are disallowed entry from the cytoplasm into the nucleus, unless they possess signal sequences, such as a nuclear localization signal (NLS), or bind to nuclear import carriers to specifically license active transport via the NPC (Chook and Süel, 2011; Goldfarb et al., 2004; Timney et al., 2016; Wang and Brattain, 2007; Naim et al., 2007). In addition to the inherited genome and other nucleic acids, nuclei contain a proteome distinct from that of the cytoplasm. All nuclear proteins are synthesized by cytosolic or ER-associated ribosomes and must subsequently cross the NE barrier to reach their site of action. The resulting compartmentalization allows separation of coupled processes in space and time to support optimal functioning of cells (Hung and Link, 2011; Stelma et al., 2016; Wang and Li, 2014).
In mammalian cells (and the majority of primitive eukaryotes), the nucleoplasm/cytoplasm dichotomy is challenged at the onset of mitosis. In prophase, when the NE fragments in preparation for chromosomal alignment and segregation, soluble nuclear and cytosolic proteins can suddenly occupy the same cellular space as the barrier between these compartments is lost. The question then is how the original localization of these two protein pools is restored once the NE reforms in cytokinesis with support of endosomal sorting complexes required for transport (ESCRT)-III and spastin (Güttinger et al., 2009; Prunuske and Ullman, 2006; Vietri et al., 2016; Vietri et al., 2015; Olmos et al., 2015). One option is that, while nuclear constituents benefit from NLS-dependent targeting, those proteins and complexes normally found in the cytosol, such as, for instance, the ribosome, could become erroneously captured by the nascent nuclear space and must then find a way out (David et al., 2012). This would then imply nuclear inclusion of the tRNA machinery and more, as required for full translation, and it is likely that such mix-ups would be disfavored. Whether and how this occurs is studied here.
Chromosomal condensation ensures exclusion of soluble cytosolic material from the newly formed nucleus post mitosis
To visualize the fate of soluble proteins as a function of size, we constructed an incremental ladder of GFP multimers ranging from monoGFP to hexaGFP stably expressed in MelJuSo cells (Fig. S1A). Under steady-state conditions, monoGFP, diGFP and triGFP filled both nuclear and cytosolic compartments, while pentaGFP and hexaGFP were restricted to the cytosol (Fig. 1A). Quantification of fluorescence over the two compartments showed an equal distribution of monoGFP, diGFP and triGFP, as opposed to reduced nuclear distribution of tetraGFP and near complete nuclear exclusion of pentaGFP and hexaGFP (Fig. 1A). These observations are consistent with the accepted notion that the NPC allows passive diffusion of proteins up to 50–70 kDa, but restricts larger cargoes unless they harbor an exposed NLS (Nigg, 1997; Lange et al., 2007), as illustrated by the predominantly nuclear localization of hexaGFP fused to the NPM NLS (NPM-NLS-hexaGFP) or heptaGFP fused to the SV40 NLS (SV40 NLS-heptaGFP) (Fig. 1A).
To follow the fate of cytosolic GFP during mitosis – and their return to the appropriate compartment of residence – we co-expressed mCherry–LaminA to visualize the timing of NE breakdown and reformation in conjunction with a DNA dye, the far-red dye SiR-DNA, to visualize the mitotic process (and simultaneously differentiate mitotic stages during time-lapse imaging) (Moriuchi et al., 2016). We first followed the distribution through mitosis of the GFP dimer, which locates in both nucleus and cytoplasm during interphase (Fig. 1B) (Movie 1). We found that diGFP was excluded from condensed chromosome density up to and including anaphase (t=45 min), similar to what is observed for microinjected fluorescent Dextran (Swanson and McNeil, 1987). However, 8 min later (t=53 min), coincident with the return of the DNA-associated lamin A signal, diGFP once again filled the nuclear as well as the cytoplasmic space. The same pattern of behavior was observed for monoGFP (Fig. S1B; Movie 2). To compare mitotic behavior of proteins below the NPC size cutoff with those above this threshold, we evaluated mitotic cells expressing hexaGFP (Fig. 1C; Movie 3). Similar to its smaller counterparts, hexaGFP was excluded from the condensed DNA space during early mitosis. However, by contrast to diGFP (Fig. 1B), hexaGFP did not enter the expanding nascent nucleus after NE reformation (Fig. 1C, t=50 min). Since we visualized chromatin using SiR-DNA, we were also able to determine its volume across different phases of mitosis (Fig. 1D). This reveals that condensed chromatin during mitosis has a 3-fold smaller volume, which might explain the restricted space for small proteins to enter the condensed chromatin space. The sharp drop represents the actual chromosome division, which is then followed by an expansion of the DNA volume that will progress until it reaches a volume corresponding to that in the mother nucleus. These initial experiments reveal that soluble proteins belonging in the cytosol (small and large alike) are excluded from the space occupied by condensed chromosomes during most of mitosis. Then, in late mitosis, smaller proteins like the GFP dimer rapidly diffuse in the nuclear space following NE reformation (as timed by visualizing lamin A), concomitant with the expansion of nuclear volume (Fig. 1E), similar to what has been recently observed (Cai et al., 2018). Conversely, proteins larger than the limit of diffusion through the NPC remain excluded from the nuclear volume even as the nascent nucleus swells. Taken together, these observations suggest that cells have evolved a physical mechanism to avoid erroneous capturing of cytoplasmic materials inside the nuclei of newborn cells.
If soluble cytosolic proteins are excluded from the nuclear space during mitosis, what is then the fate of soluble nuclear proteins? To address this, we fused hexaGFP to a bipartite NLS from nucleoplasmin (NPM, also known as NPM1) and heptaGFP to the monopartite NLS of SV40 (Nigg, 1997; Kalderon et al., 1984), and followed the resulting NPM NLS-hexaGFP and SV40 NLS-heptaGFP through mitosis (Fig. 1D and Fig. S2C; Movies 4 and 5). Strikingly, the presence of the SV40 NLS enabled a small fraction of SV40 heptaGFP to mix with the condensed chromatin up to and during anaphase (t=54 min). Hence, stably expressing SV40 NLS-heptaGFP cells clearly lack the exclusion conferred by chromosomal condensation. This suggests a closer proximity of heptaGFP to the condensed chromosomes when it is decorated with an NLS. Following NE reformation, the small portion of chromatin mixed with SV40 NLS-heptaGFP is likely encapsulated in the nascent nucleus (t=56 min), while the remaining cytosolic pool of SV40 NLS-heptaGFP is rapidly imported into the nucleus. Conversely, when studying NPM NLS-hexa-GFP, we observed a slight enrichment of NPM NLS-hexaGFP on condensed chromosomal material. The NLS sequence apparently affects the interaction with chromatin components, as a variable fraction locates in the condensed chromatin space. Nevertheless, these observations suggest that, unlike soluble cytoplasmic proteins, a small fraction of soluble nuclear proteins containing SV40 or NPM NLSs are in close proximity of condensed chromatin during mitosis while the majority of NLS-hexaGFP or NLS-HeptaGFP locates outside the condensed chromatin space. When the nucleus reforms around condensed chromatin, it is followed by nuclear expansion (Lu et al., 2011) and these NLS-containing proteins then swiftly are imported in the nucleus.
Limiting chromosomal condensation through Q15 treatment compromises soluble proteome homeostasis
To test whether nuclear exclusion of larger cytosolic proteins following NE reformation (as observed in Fig. 1C) requires the preceding exclusion of these materials as a result of chromosomal density, we aimed to limit chromatin condensation and then followed the fate of hexaGFP, which is excluded from the nucleus under unmanipulated situations. To this end, we synthesized a condensin inhibitor Anilinoquinazoline 15 (Q15) (Shiheido et al., 2012) that targets human (h)CAP-G2 (also known as NCAPG2; a subunit of condensin II) and thus limits chromosomal condensation (Hirano, 2005). Incubation of MelJuSo cells expressing hexaGFP in the presence of Q15 for 48 h yielded many dead cells, which is not unexpected when manipulating chromatin condensation. However, a significantly higher number of cells that survived showed hexaGFP located in the nucleus as opposed to the cells treated with DMSO (Fig. 2A,C,D; Movies 5 and 6). Time-lapse microscopy of cells over mitosis suggests that, when compared to DMSO-treated hexaGFP-expressing cells, Q15 treatment causes a fraction of hexaGFP to localize in the periphery of the chromatin area during mitosis (Fig. 2B). Once decondensation commences, hexaGFP is allowed to diffuse in the decondensating chromatin space and is trapped in the newly formed nucleus. This assumes that Q15 does not permeabilize the NE. To rule out this possibility, we treated cells stably expressing hexaGFP with Q15 for 48 h and bleached the nuclear hexaGFP signal. If Q15 led to permeabilization the nucleus, cytosolic hexaGFP would have restored fluorescence in the nucleus. However, no recovery of fluorescence was observed over a 30 min time period post bleaching (Fig. 2E,F; Movie 8). This further suggests that a physical parameter (the condensed state of chromatin during mitosis) is critical for protein homeostasis in the nuclear and cytosolic compartments over the period of mitosis.
The 20S proteasome requires a specialized mechanism for nuclear import post mitosis
We then wondered how large protein complexes present in both compartments, such as the proteasome, are handled over cell division (Reits et al., 1997). The 20S proteasome – a protein complex of 750 kDa – would be expected to be far too large to swiftly diffuse across the NE. To test the dynamics of proteasome import in and export from the nucleus, we performed fluorescence recovery after photobleaching (FRAP) experiments to follow the kinetics of nuclear import of LMP2–GFP-labeled proteasomes in MelJuSo cells (LMP2 is also known as β1i or PSMB9). LMP2–GFP is stably incorporated into the 20S proteasome and no free LMP2–GFP subunits have been detected (Fig. 3A,B; Movie 9) (Reits et al., 1997). Following nuclear bleaching, recovery of nuclear GFP fluorescence from cytosolic proteasomes entering the nucleus was not observed over a 10 min period. In fact, FRAP experiments did not reveal nuclear import of proteasomes even over a 3 h period (Reits et al., 1997). If proteasomes are excluded from the condensed chromatin space in accordance with observations for free soluble proteins, it would imply that newly formed nuclei are devoid of proteasomes for a considerable time period post mitosis (Savulescu et al., 2011; Palmer et al., 1994). We therefore followed the fate of LMP2–GFP-labeled proteasomes over mitosis (Fig. 3C; Movie 10). As in the case of GFP multimers, proteasomes were excluded from the condensed chromatin space (t=50 min). Strikingly, and coincident with cytokinesis, a substantial fraction of labeled proteasomes rapidly entered the newly formed nucleus demarcated by lamin A (t=56 min). Quantification of the LMP2–GFP signal indicated swift nuclear import for some 20 min after nuclear reformation, followed by an abrupt termination of nuclear import. This may coincide with formation of functional nuclear pore complexes and NE reconstruction (Lafarga et al., 2002). Interestingly, nuclear import of LMP2–GFP-marked proteasomes after NE formation was even faster than that of diGFP, which enters the nucleus by free diffusion (Fig. 3D; Movie 1). Furthermore, by quantifying the nuclear signal for GFP and the cytosolic signal for NPM NLS-hexaGFP, we found that proteasomal nuclear import is slightly faster than that for NPM NLS-hexaGFP, which requires an active NLS-dependent import machinery (Fig. 3E). Of note, the fluorescence intensity of LMP2–GFP-labeled proteasome is 10–20% higher in the post-mitotic nuclei than before mitosis, likely as a result of a higher concentration present in the still expanding nucleus (see also Fig. 1E).
IPO5 conducts the post mitotic nuclear import of the 20S proteasome
The above observations can only be explained by an active nuclear import mechanism for the proteasome that is operational only for some 20 min following NE reformation. This system is then critical for rapid delivery of proteasomes in the newly formed nucleus, allowing normal propagation of proteasome-dependent degradation of a plethora of nuclear substrates. Such a system is unknown. One option is that cells possess a pre-defined nuclear proteasome subset that then simply returns to its original location post mitosis by conventional nuclear import. To examine whether this is the case, we photobleached the nucleus of MelJuSo cells containing LMP2–GFP-labeled proteasomes and followed the remaining (cytosolic) fluorescence pool through mitosis (Fig. 4A; Movie 11). While proteasomes were again excluded from condensed chromatin prior to cytokinesis (t=34 min), the original cytosolic pool of proteasomes occupies both the cytosolic and nuclear compartments post-mitosis (Fig. 4B). This excludes the presence of a nucleus-destined proteasome pool and implies a specialized mechanism is active during a short phase late in mitosis and is responsible for the rapid targeting of proteasomes to newly formed nuclei. Another option would be that the proteasome dissociates into its free subunits that then diffuse in the nucleus to form a new proteasome. Since active proteasome subunits (including LMP2–GFP) contain a prosequence that is removed when the subunit is incorporated into the active proteasome, mature proteasome subunits would not be reincorporated into proteasomes (Li et al., 2016; Jäger et al., 1999; Panté and Kann, 2002). In addition, we noted that the proteasome is entering the nucleus faster than diGFP, which enters the nucleus by simple diffusion, again suggesting active nuclear import. In combination, these data suggest that there is a unique nuclear import system for the proteasome that is active in a short time window immediately after NE formation.
Targeted nuclear import is driven by members of the karyopherin family (Chook and Süel, 2011; Goldfarb et al., 2004; Yang and Musser, 2006). Only a limited number of specific substrates have been defined for these importers, and none have thus far been described to mediate import of proteasomes to the nucleus in mammalian cells (Chook and Süel, 2011; Pumroy and Cingolani, 2015; Stelma et al., 2016; Kimura et al., 2013). It is likely that one or more karyopherins are involved in proteasome import after NE closure. To test this, we silenced importin expression using pools of small interfering RNAs (siRNA), followed by fixation and immunostaining for the endogenous proteasome. Comparison of nuclear to cytosolic fluorescence intensities identified karyopherin A1 (KPNA1) and importin-5 (IPO5) as candidates for nuclear proteasome import (Fig. S2A). We then refined the siRNA pools to confirm that it is indeed silencing of IPO5 that reduces nuclear import of proteasomes in two different cell lines (Fig. 4C; Fig. S2B). The efficiency of IPO5 depletion was validated by immunoblotting (Fig. 4D).
To assess whether expression of IPO5 is altered during mitosis, we determined IPO5 levels by western blotting, along with assessing the levels of a marker (cyclin B1) that is degraded at a later phase of mitosis. Cells were synchronized, then released and samples were analyzed at various time points post block release, as indicated (Fig. 4E). No notable effect on IPO5 expression was detected. IPO5 may also alter its location in the short time window post NE formation. We fixed and stained MelJuSo cells for IPO5 and lamin A/C and analyzed different phases in mitosis for their relative expression. In G0/G1 phase up to late anaphase, IPO5 appears to localize primarily outside the nucleus. However, the nuclear localization of IPO5 increased as soon as lamin A/C appeared around the condensed chromatin (Fig. 4G), and returned to a mainly cytosolic distribution shortly after this phase (for a quantification, see Fig. 4H). These data suggest that IPO5 alters its activity during a short phase in mitosis and could then participate in the nuclear location of proteasomes.
However, only a fraction of proteasomes are imported during this short phase where IPO5 is active. It is possible that IPO5 could be present in limiting amounts to allow for more-complete nuclear deposition of proteasomes. If so, IPO5 overexpression would be expected to increase the number of nuclear proteasomes. To test this, we ectopically expressed N-terminally GFP-tagged IPO5 in MelJuSo (Fig. 4I,J) and HeLa cells (Fig. S2C,E), followed by fixation and immunostaining for the 20S proteasome. We quantified the fluorescence signal and plotted the ratio between that in the nuclear and cytosolic compartments. Overexpression of IPO5, unlike another karyopherin family member, KPNB1 (Chook and Süel, 2011), indeed increased nuclear proteasome levels relative to that in the cytoplasm. To exclude the possibility that ectopic IPO5 expression also accelerated nuclear import of proteasomes in G0/G1 phase, we performed FRAP experiments. We expressed mCherry–IPO5 in cells expressing LMP2–GFP and bleached GFP-labeled proteasomes in the nucleus (Fig. S2E; Movie 11). No recovery of signal in the 2 h following bleaching was observed, implying that IPO5 overexpression does not accelerate nuclear import of (GFP-labeled) proteasomes in the G0/G1 phase. Collectively, this identifies one component of a nuclear import system for proteasomes that is active at a late phase in mitosis. This ensures rapid delivery of proteasomes excluded by condensed chromatin in the newly formed nucleus.
We have visualized a largely ignored aspect of mitosis: the restoration of the cytosolic and nuclear proteomes after NE reformation. Given the conserved nature of nuclear/cytoplasmic compartmentalization across species, its fidelity must be critical to the success of the progeny. With commencement of mitosis, fragmentation of the NE eliminates the barrier between the nucleus and cytosol, resulting in a mixed soluble proteome. Our data provide a simple explanation for how cells repartition soluble proteins destined for either their cytosolic or nuclear compartments. We suggest that soluble proteins are generally excluded from the condensed chromatin space in mitosis. These data are in line with recent findings suggesting that small nuclear pores are present in the reforming NE starting from anaphase onset. The NPC pre-pores would allow free diffusion of cytosolic material if condensed DNA would not act as a physical barrier (Otsuka et al., 2018). A fraction of proteins are decorated with an NLS. We tested hexaGFP with two different NLS sequences that had poor to low affinity for chromatin or nuclear import factors recognizing these signals. However, the majority of NLS-containing proteins are found in the cytosolic volume immediately after the NE is reformed, and is imported into the nucleus when nuclear expansion begins.
We propose a simple model describing the maintenance of the nuclear and cytosolic proteomes during the cell cycle. After nuclear disintegration, the soluble nuclear and cytosolic proteome mixes. The chromatin is condensed to a density that excludes even small proteins such as GFP. The NE is then formed around the condensed chromatin, physically excluding cytosolic proteins from the nuclear space. The nucleus will then start expanding to its original volume. At the same time, proteins with an NLS are rapidly re-imported from the cytosol into the nucleus, and homeostasis of the soluble proteome is restored.
In this study, we used the proteasome as an example of a macromolecule that exists in both cellular compartments. During mitosis the proteasome is also excluded from condensed chromatin and must be reimported into the nucleus to provide the machinery for protein turnover to new cells. Since nuclear import of the proteasome is slow (Reits et al., 1997), either new cells first live without nuclear proteasomes or another system for nuclear import of the proteasomes should exist. The NE is tightly sealed immediately after NE formation as illustrated by the effective exclusion of hexaGFP from the nuclear space. Proteasomes are much larger and should then also be excluded from the nucleus, at least through passive diffusion. It is also possible that a fraction of proteasomes are predestined for the nucleus. The 20S proteasome has been suggested to possess a NLS, but whether that is responsible for nuclear import of proteasomes is unclear (Nederlof et al., 1995; Ogiso et al., 2002; Knuehl et al., 1996). Our FRAP experiments exclude the possibility that there is a specific nuclear pool of proteasomes. Instead, we observe a rapid and short import of proteasomes into the nuclear space after NE formation. The import is faster than free diffusion of diGFP, which suggests that this import is achieved by an active system that is operational for only some 20 min post NE formation. This is followed by a considerably slower nuclear import event of proteasomes. We show that at least one component of the large karyopherin family, IPO5, supports nuclear proteasome import. Despite extensive attempts, we failed to identify factors specifically marking proteasomes for IPO5-mediated nuclear import due to the difficulty of ‘catching’ factors that active for only 20 min post NE formation. Yeast studies on nuclear import of proteasomes identified N-acetylation by SIRT family members as a factor controlling this process (van Deventer et al., 2015). However, yeast does not fragment its nucleus during division and we do not know whether this modification is also relevant in nuclear import of proteasomes immediately after NE formation. N-acetylation may then affect multiple substrates including IPO5. Which – if any – of these substrates is crucial in mammalian cells, is unclear. IPO5 has been implicated in nuclear import of β-catenin (Goto et al., 2013), MSI1 (Sutherland et al., 2015), CBEP3 (Chao et al., 2012) and influenza polymerase subunit PB1 (Hutchinson et al., 2011), and these proteins are likely to be delivered to the nucleus in response to specific signals or infection. Because we excluded the possibility of a dedicated nuclear proteasome fraction, dedicated biochemical events involving either the proteasome, IPO5 or other proteins would be expected to be operational during the first and final stages of mitosis that would drive the proteasome into the nucleus during the first 20 min after NE formation. IPO5 is likely rate limiting in this process but would not necessarily directly interact with the proteasome. However, it is likely that additional factors will be involved in the nuclear import of the proteasome. Of note, the proteasome may be representative of other nuclear complexes, including possibly the spliceosome, RNA polymerases or other large protein complexes without obvious NLSs, that reside and function in the nucleus. This new biology has to be developed.
We visualized an unexplored but critical aspect of the cell biology of mitosis that may be manipulated in order to control the life of mitotic cells. The nuclear exclusion of large cytosolic proteins after mitosis is the result of a physical process: the condensed state of mitotic chromatin that excludes these molecules. As the NE forms around the condensed chromatin before expansion, such cytoplasmic proteins are automatically excluded from the nuclear space. Proteins with an NLS will be located in the cytosol after NE formation and then swiftly imported back into the nucleus. The proteasome, as a prototype large protein complex without a clear NLS, then requires specific import processes, including IPO5, for nuclear import. This process is operational during a short period late in mitosis. Multiple processes then control protein complex homeostasis over the period of mitosis in cells.
MATERIALS AND METHODS
Cell culture and constructs
MelJuSo cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM; Thermo Fisher Scientific) supplemented with 10% fetal calf serum (FCS), and HeLa cells in Dulbecco's modified Eagle's medium (DMEM; Thermo Fisher Scientific) with 10% FCS. All cells were tested regularly for mycoplasma contamination and checked for contamination by other cells through morphological analysis. MelJuSo cells stably expressing LMP2 –GFP were previously described (Reits et al., 1997). A MelJuSo cell line was used to generate the stably expressing cell lines with constructs for monoGFP, diGFP, triGFP, tetraGFP, pentaGFP hexaGFP, NPM NLS-hexaGFP and SV40 NLS-heptaGFP. The cryptic start sites and methionine residues were removed following the initial start site and methionine by using the following primers: 5′-CCCAAGATCTCTTGTACAGCTCGTCCAT-3′, 5′-CCCAGGATCCGTGAGCAAGGGCGAGGAG-3′ and 5′-CCCAGAATTCTCACTTGTACAGCTCGTCCATGC-3′. These were cloned into a C1-mGFP vector using BglII, BamHI and EcoRI restriction, respectively. This was accomplished by generating a GFP without a start and stop codon and cloning it behind a wild-type GFP lacking a stop codon. This was continued until hexaGFP was reached. For NLS-hexaGFP, an NLS sequence from nucleoplasmin and from SV40 was cloned at the N-terminus of the heptaGFP coding sequence using the following primers: for NPM, 5′-CTAGCCACCATGGTGAAACGACCAGCAGCAACAAAGAAAGCAGGACAAGCAAAGAAAAAGAAGA-3′ and 5′-CTAGTCTTCTTTTTCTTTGCTTGTCCTGCTTTCTTTGTTGCTGCTGGTCGTTTCACCATGGTGG-3′ into C1 6GFP vector using NheI and SpeI restriction sites, and for 2×SV40 5′-CTAGGCCACCATGCCAAAAAAAAAAAGTTA-3′ and 5′-CTAGTAACTTTTCTTTTTTTTTTTGGCATGGTCG-3′ into C1 6GFP vector using NheI; this was repeated to obtain 2×SV40 NLS. pBABE-puro-GFP-wt-LaminA was a Addgene plasmid #17662 (deposited by Tom Misteli; Scaffidi and Misteli, 2008) and cloned into a C1-mCherry expression vector using primers 5′-CCCAAGATCTATGGAGACCCCGTCCCAGC-3′ and 5′-CCCAGAATTCGCGGCCGCTTACATGATGCTGCAGTTCT-3′. IPO5 was cloned from the ORFeome collaboration library cssbBroadEn_10938 using the primers 5′-CCCACTCGAGCCATGGCGGCGGCCGCG-3′ and 5′-CCCAGGTACCTCACGCAGAGTTCAGGAGCTCCTGAAT-3′ into mGFP-C1 and mCherry-C1 vector via the XhoI and Asp718I restriction sites. KPNB1 was cloned from IMAGE 3352610 using the primers 5′-CCCAGAGCTCCCATGGAGCTGATCACCATTCT-3′ and 5′-CCCAGGATCCTCAAGCTTGGTTCTTCAGTTT-3′ into mGFP-C1 vector using the SacI and BamHI restriction sites. Shrimp alkaline phosphatase (ANZA, Thermo Fisher Scientific) was used in cloning SV40 NLS-hexaGFP. All constructs were sequence verified.
Reagents and antibodies
Anilinoquinazoline 15 was synthesized as previously described and used at a concentration of 20 µM for 48 h (Shiheido et al., 2012). Nocodazole at a concentration of 75 ng/ml was used to synchronize MelJuSo cells (Cayman Chemicals). For live-cell DNA staining, SiR-DNA 700 was added to cells at 20 min before the start of imaging (Spirochrome, Switzerland). For detection of the human 20S proteasome by microscopy we used rabbit anti-20S (Enzo) at a dilution of 1:100. This antibody recognizes many proteasomal subunits (α5/α7, β1, β5, β5i and β7). For western blot analysis of GFP, we used rabbit anti-GFP at a dilution of 1:1000 (Rocha et al., 2009). IPO5 (karyopherin β3) was detected by western blot analyses using mouse anti-human IPO5 at a dilution of 1:1000 (cat. no. sc-55527, Santa Cruz Biotechnology). Rabbit anti-human lamin A/C, at a dilution of 1:1000 (cat. no. sc-20681, Santa Cruz Biotechnology) was used for staining a loading control. Mouse anti-cyclin B1, at a dilution of 1:1000, was used to determine mitotic phase (cat. no. sc-245, Santa Cruz Biotechnology). Secondary antibodies for microscopy were goat anti-mouse-IgG conjugated to Alexa Fluor 488, anti-rabbit-IgG conjugated to Alexa Fluor 568 and anti-mouse-IgG conjugated to Alexa Fluor 568 at a dilution of 1:400 (Life Technologies). For detection of the primary antibodies in western blotting, goat anti-mouse-IgG conjugated to 800CW IRDye (1:10,000) and goat anti-rabbit IgG conjugated to 680RD IRDye (1:10,000) (Li-COR) were used.
For ectopic expression, MelJuSo and HeLa cells were transfected using Effectene (Qiagen, Germany) according to the manufacturer's instructions. For siRNA-mediated silencing, cells were reverse transfected with DharmaFECT transfection reagent #1 (Dharmacon) and 50 nM siRNA. After 3 days, the cells were fixed, stained and analyzed by confocal microscopy or western blotting. siRNAs for the screen were extracted from a human siGENOME siRNA library genome smart pool (Dharmacon). For further studies, we used siRNA for IPO5 under code MQ-017318-02-0005 (Dharmacon).
For whole-cell lysate analyses, cells were lysed directly in SDS sample buffer (2% SDS, 10% glycerol, 100 mM DTT, 60 mM Tris-HCl pH 6.8 and 0.01% Bromophenol Blue) and boiled for 10 min. Proteins were subsequently separated by SDS-PAGE and transferred to a nitrocellulose membrane (GE Healthcare). Blocking of the membrane was performed in PBS supplemented with 5% (w/v) skim milk powder (OXOID, The Netherlands). The membranes were imaged using the Odyssey Imaging System Clx (Li-COR). Original western blot images are presented in Fig. S3.
Cells were seeded on coverslips and transfected. After 18 h, cells were fixed in 3.7% formaldehyde for 20 min and permeabilized in 0.01% Triton X-100 (Fisher Chemicals) for 15 min. Staining was performed with the antibodies described above. The cell nuclei were stained with DAPI and the cells were mounted onto coverslips with prolong-gold anti-fade reagent with DAPI (Invitrogen). Images were acquired using either a Leica TCS SP5 confocal microscope, a Leica TCS SP8 (Leica Microsystems, Wetzlar, Germany) at 63× magnification or an Andor Dragonfly 505 spinning disk confocal on a Leica DMi8 microscope (Oxford Instruments, UK). For live-cell imaging, the microscopes were equipped with a humidified climate control system at 37°C supplemented with 5% CO2. All sequential images were collected at a rate of 124 s per frame for a period of 16 h. Time-lapse images were segmented using semi-automatic procedures in ImageJ and Matlab (Mathworks). Subsequent quantification was performed in Matlab by measuring the mean fluorescent intensity. To quantify the import half-time, we fitted an exponential function to the intensity data using the curve-fitting toolbox of Matlab. All images were processed using Adobe Illustrator and ImageJ.
Statistical analysis and experimental setup
All experiments shown in the paper were performed independently at least two times. For nuclear to cytosolic ratios, numerical values were calculated and subsequently normalized using either siControl-treated samples or cells expressing GFP. Statistical significance was calculated using an unpaired Student’s t-test. Statistical values are as follows: *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.
Online supplemental material
Fig. S1 shows a western blot analysis of cells stably expressing the molecular GFP ladder, stills representing Movie 2 and NPM NLS-hexaGFP stills corresponding to Movie 5. Fig. S2 shows confocal microscopy images of MelJuSo cells treated with the siRNA molecules and the quantification of nuclear-to-cytosolic fluorescence intensities, and similar images from HeLa cells overexpressing IPO5 and the quantification of nuclear-to-cytosolic fluorescence intensities in HeLa cells overexpressing IPO5, KPNB1 or GFP, as well as confocal data and the quantification of nuclear-to-cytosolic fluorescence intensities in of MelJuSo cells stably expressing LMP2–GFP and overexpressing mCherry–IPO5 subjected to FRAP experiments. Fig. S3 shows the uncropped western blot analysis of Fig. 4D,E.
We thank the microscopy facilities of the NKI and the LUMC for their support and members of the Neefjes group for critical discussions.
Conceptualization: M.S., A.C.M.N., J.N.; Methodology: M.S., L.M.V., J.N.; Software: L.M.V.; Validation: M.S., L.M.V.; Formal analysis: M.S., L.M.V.; Investigation: M.S., L.J.J., L.M.V.; Resources: M.S., L.J.J., R.K., H.O.; Data curation: M.S., L.M.V.; Writing - original draft: M.S.; Writing - review & editing: M.S., J.N.; Visualization: M.S.; Supervision: H.O., J.N.; Project administration: M.S., J.N.; Funding acquisition: J.N.
This work was supported by a Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NOW) TOP grant and an European Research Council (ERC) Advanced Grant awarded to J.N. J.N. and H.O. are members of the Oncode Institute.
The authors declare no competing or financial interests.