Register for Mechanobiology 2016!


During exit from mitosis in Xenopus laevis egg extracts, the AAA+ ATPase Cdc48/p97 (also known as VCP in vertebrates) and its adapter Ufd1–Npl4 remove the kinase Aurora B from chromatin to allow nucleus formation. Here, we show that in HeLa cells Ufd1–Npl4 already antagonizes Aurora B on chromosomes during earlier mitotic stages and that this is crucial for proper chromosome segregation. Depletion of Ufd1–Npl4 by small interfering RNA (siRNA) caused chromosome alignment and anaphase defects resulting in missegregated chromosomes and multi-lobed nuclei. Ufd1–Npl4 depletion also led to increased levels of Aurora B on prometaphase and metaphase chromosomes. This increase was associated with higher Aurora B activity, as evidenced by the partial resistance of CENP-A phosphorylation to the Aurora B inhibitor hesperadin. Furthermore, low concentrations of hesperadin partially rescued chromosome alignment in Ufd1-depleted cells, whereas, conversely, Ufd1-depletion partially restored congression in the presence of hesperadin. These data establish Cdc48/p97–Ufd1–Npl4 as a crucial negative regulator of Aurora B early in mitosis of human somatic cells and suggest that the activity of Aurora B on chromosomes needs to be restrained to ensure faithful chromosome segregation.


The proper segregation of the replicated and condensed chromosomes during mitosis requires correct attachment of chromatid pairs to opposite poles of the mitotic spindle and subsequent accurate separation in anaphase, followed by timely decondensation and nuclear envelope formation in telophase. The kinase Aurora B plays crucial roles in various aspects of chromosome segregation (Ducat and Zheng, 2004; Ruchaud et al., 2007). In complex with its partner proteins INCENP, survivin and borealin (also known as Dasra), it associates along chromosomes in prophase where it phosphorylates histone H3 and the centromeric histone H3 homologue CENP-A (Adams et al., 2001; Giet and Glover, 2001; Hsu et al., 2000; Murnion et al., 2001; Zeitlin et al., 2001). As mitosis progresses to metaphase, Aurora B dissociates from chromosome arms and concentrates in the centromeric region, where it controls bipolar spindle attachment to the kinetochores of sister chromatids and thus the formation of the metaphase plate (Adams et al., 2001; Cimini et al., 2006; Giet and Glover, 2001; Hauf et al., 2003; Kaitna et al., 2000; Kelly and Funabiki, 2009; Tanaka et al., 2002). Aurora B also helps to maintain the spindle checkpoint that blocks anaphase onset and exit from mitosis until all chromosomes properly attach to the spindle (Ditchfield et al., 2003; Hauf et al., 2003; Kallio et al., 2002; Musacchio and Hardwick, 2002). Upon anaphase onset, Aurora B, along with its partners, transfers to the spindle midzone where it regulates spindle dynamics and function. In telophase, Aurora B persists at the midbody where it controls cytokinesis, to ensure proper division of the daughter cells (Steigemann et al., 2009).

As a consequence of its central role, RNA interference (RNAi)-mediated depletion of Aurora B, or pharmacological inhibition of its kinase activity, in mammalian tissue culture cells leads to a prominent chromosome alignment defect in prometaphase due to erroneous spindle attachment to kinetochores (Ditchfield et al., 2003; Hauf et al., 2003; Taylor and Peters, 2008). Metaphase plate formation is largely hampered, and cells display severe anaphase defects with lagging chromosomes, or they exit from mitosis before completion of anaphase with resulting micronuclei, polyploidy and multinucleated cells (Ditchfield et al., 2003; Hauf et al., 2003; Mora-Bermudez et al., 2007). How Aurora B controls and corrects spindle attachment to kinetochores and thus chromosome congression to the metaphase plate is not completely understood, but it involves regulation of a number of substrate proteins at the kinetochore with seemingly antagonizing activities (Kelly and Funabiki, 2009; Ruchaud et al., 2007). Phosphorylation of the Ndc80 (also known as Hec1) complex, the major attachment module for microtubules, is proposed to decrease its affinity for microtubules and therefore destabilizes spindle attachment (Cheeseman et al., 2006; Ciferri et al., 2008; DeLuca et al., 2006). By contrast, phosphorylation of mitotic centromere-associated kinesin (MCAK; also known as KIF2C) suppresses its microtubule-depolymerizing activity and thus might stabilize spindle attachments (Andrews et al., 2004; Lan et al., 2004; Ohi et al., 2004). The role of Aurora B in correcting spindle attachment suggests that centromeric Aurora B needs to be tightly regulated, to maintain its activity within a limited range, to allow fine-tuning of spindle attachment and consequently faithful chromatid segregation.

Previously, the AAA+ ATPase Cdc48/p97 (also known as VCP in vertebrates) has been implicated in regulation of Aurora B. Cdc48/p97 is an abundant ubiquitin-dependent chaperone involved in protein quality control and signalling events in interphase and mitosis (Woodman, 2003; Ye, 2006). Cdc48/p97 associates with alternative substrate and ubiquitin adapters, including cofactor p47 (officially known as NSFL1 cofactor p47; NSF1C) or the Ufd1–Npl4 complex, to fulfil distinct functions (Alexandru et al., 2008; Kondo et al., 1997; Meyer et al., 2000; Schuberth and Buchberger, 2008). Ufd1 and Npl4 form a stable heterodimeric complex, with both subunits containing interaction sites for Cdc48/p97, as well as binding domains for ubiquitin (Alam et al., 2004; Bruderer et al., 2004; Isaacson et al., 2007; Meyer et al., 2002). As a common activity applied in different pathways, the Cdc48/p97–Ufd1–Npl4 complex is thought to bind and extract ubiquitylated client proteins from cellular structures and to release them for either degradation or recycling (Jentsch and Rumpf, 2007; Ye, 2006). In mitosis, Cdc48/p97–Ufd1–Npl4 has been associated with different functions, such as chromosome segregation, spindle dynamics and organelle reformation, in different organisms and developmental stages (Meyer and Popp, 2008). In Xenopus laevis egg extracts, Cdc48/p97–Ufd1–Npl4 removes Aurora B from chromatin during exit from mitosis in a process that involves modification of Aurora B with ubiquitin chains (Ramadan et al., 2007). Removal of Aurora B from chromatin leads to a local decrease in activity, which then allows decondensation and nuclear envelope formation (Ramadan et al., 2007).

Ufd1–Npl4 has also been associated with chromosome segregation in HeLa cells in two independent studies. RNAi-mediated depletion of Npl4 results in anaphase defects with entangled chromosomes (Porter et al., 2007), whereas Ufd1 depletion has been shown to cause chromosome congression defects (Vong et al., 2005). On the basis of immunohistochemistry of small interfering RNA (siRNA)-treated cells, Vong and colleagues concluded that Ufd1 helps to recruit Aurora B to chromosomes. This interpretation for a positive regulation of Aurora B is in surprising contrast with our observation in Xenopus egg extracts (Ramadan et al., 2007), where Cdc48/p97 removes Aurora B from chromosomes. Two other studies using RNAi-mediated depletion of Cdc48/p97 in Caenorhabditis elegans failed to detect any role for Cdc48/p97 or Ufd1–Npl4 in mitosis (Heallen et al., 2008; Mouysset et al., 2008). The cellular relevance of the Cdc48/p97–Ufd1–Npl4 complex for regulation of mitosis and its functional relation to Aurora B has therefore remained controversial.

To address this controversy, we have re-examined the relevance of Cdc48/p97–Ufd1–Npl4 for mitosis, and its functional relation to Aurora B in HeLa cells. Here, we show that Ufd1 and Npl4 cooperate in proper chromosome segregation. Moreover, we provide evidence that Ufd1–Npl4 functions by antagonizing Aurora B during segregation. Depletion of Ufd1–Npl4 resulted in increased levels of Aurora B associated with chromosomes during prometaphase and metaphase, as shown by immunofluorescence microscopy and cell fractionation. Furthermore, Ufd1 depletion increased phosphorylation of the centromeric Aurora B substrate CENP-A. Finally, and importantly, Ufd1 depletion partially rescued the rate and efficiency of chromosome congression after attenuation of Aurora B with the inhibitor hesperadin. The data establish the Cdc48/p97–Ufd1–Npl4 complex as a negative regulator of Aurora B early in mitosis. Moreover, the data imply that restraining the activity of Aurora B on chromosomes is essential for its function in controlling proper chromosome segregation.


Ufd1–Npl4 is required for normal progression through mitosis in HeLa cells

To study the role of Cdc48/p97 in regulating progression of mitosis in HeLa cells systematically, we used an siRNA approach to deplete its two alternative adapters implicated in mitosis, p47 or the heterodimeric Ufd1–Npl4 complex (Kondo et al., 1997; Meyer et al., 2000; Schuberth and Buchberger, 2008), rather than Cdc48/p97 itself, in order to reduce pleiotropic effects. Depletion for 48 hours was efficient and did not affect Cdc48/p97 levels (Fig. 1A). Importantly, depletion of Npl4 also specifically destabilized Ufd1, but not vice versa, consistent with previous biochemical data showing that Ufd1 is unstable in the absence of Npl4 (Bruderer et al., 2004). We first monitored progression through mitosis in unsynchronized HeLa cells stably expressing histone-H2B–mRFP by time-lapse video microscopy and measured the time between nuclear envelope breakdown (scored on the basis of the loss of a defined boundary for the chromatin regions) (Beaudouin et al., 2002) and anaphase onset (Fig. 1B,C). Ufd1- or Npl4-siRNA-treated cells displayed a delay of anaphase onset (median of ~43 minutes), compared with that of control or p47-depleted cells (median of ~30 minutes), suggesting that the spindle assembly checkpoint is activated owing to compromised spindle–kinetochore attachments. Indeed, inspection of the movies revealed misaligned chromosomes at the spindle pole, and a delay in congression and the formation of a metaphase plate, in Ufd1- or Ufd1–Npl4-depleted cells (Fig. 1B). When cells eventually proceeded into anaphase, we observed segregation defects, with separating chromosomes remaining entangled and delayed retraction of chromosome arms from the spindle midzone. A fraction of the Npl4-siRNA-treated cells did not enter anaphase during the course of the experiments (Fig. 1B). These findings confirm that the Cdc48/p97–Ufd1–Npl4 chaperone complex is essential for proper mitotic progression of human somatic cells.

Fig. 1.

The Cdc48/p97 adapter Ufd1–Npl4 is required for normal progression through mitosis of HeLa cells. (A) Western blot analysis of HeLa Kyoto cells treated with non-silencing control (Ctrl) siRNA or two independent oligonucleotides (S1 and S2) each targeting Ufd1, Npl4 or the alternative adapter p47 for 48 hours. Membranes were probed with the indicated antibodies. Note that Npl4 depletion also specifically destabilizes its binding partner Ufd1. (B) Images from representative confocal time-lapse fluorescence microscopy movies of unsynchronized HeLa Kyoto cells progressing through mitosis, after treatment with indicated siRNA oligonucleotides for 48 hours. The cells are stably express histone-H2B–mRFP to visualize chromatin. Time (minutes) after nuclear envelope breakdown is indicated, closed arrowheads point to anaphase onset. Note that some Npl4-siRNA-treated cells did not enter anaphase (as shown in the bottom series of images). The open arrowhead indicates a chromosome bridge. Scale bars: 5 μm. (C) Quantification of movies, as in B, from 5 independent experiments. The time between nuclear envelope breakdown (NEBD) and anaphase onset was determined and is represented in a box-and-whisker plot [Ctrl, median 30.0 minutes, mean 31.3 minutes (n=48); Ufd1, median: 43.3 minutes, mean: 44.9 minutes (n=28); Npl4, median: 43.3 minutes, mean: 46.3 minutes (n=12); p47, median: 29.5 minutes; mean: 31.0 minutes (n=12)]. Only Npl4-siRNA-treated cells that entered anaphase within 170 minutes were quantified. Statistical significance was tested with the Mann–Whitney U-test (n.s., not significant). The siRNAs Ufd1-S2, Npl4-S2 and p47-S1 were used.

Ufd1–Npl4 depletion causes chromosome congression, as well as anaphase defects

The congression defect was confirmed in fixed cells. On average 36.8% and 46.9% of cells with an apparent metaphase plate displayed misaligned chromosomes, often several at a time, upon Ufd1 or Npl4 siRNA treatment, respectively, compared with 7.5% or 6.4% in control or p47-depleted cells, respectively (Fig. 2A,C). In Ufd1-depleted cells, the congression defect was largely rescued by overexpression of the mouse Ufd1 cDNA that is not targeted by the siRNA oligonucleotides used (Fig. 2C and supplementary material Fig. S1), showing that the effect was specific. In addition, we observed a significant (P<0.005) increase in chromosome segregation defects in anaphase from 7.8% in control and 7.7% in p47-depleted cells, to 23% and 31.4% in Ufd1- or Npl4-siRNA-treated cells, respectively (Fig. 2B,D). These defects included lagging chromosomes that probably resulted from incorrect spindle attachment. Moreover, we observed chromosome bridges between the separating chromatin masses, and often several entangled chromosomes, indicating defects in sister chromatid resolution. In particular, Npl4-siRNA-treated cells (that are depleted of the entire Ufd1–Npl4 heterodimer) displayed aberrations in nuclear morphology in interphase with micronuclei that probably originated from lagging chromosomes (Fig. 2E). Even more prominently, nuclei were often multi-lobed, a phenotype known to result from segregation errors and defective axial compaction of anaphase chromosomes upon inhibition of Aurora B (Hauf et al., 2003; Mora-Bermudez et al., 2007). These results show the shared functions of the Ufd1 and Npl4 in regulating chromosome alignment and chromosome segregation in human somatic cells.

To address whether depletion of Ufd1–Npl4, in the same manner as Aurora B inhibition, causes polyploidy, we treated cells with siRNA targeting Ufd1, Npl4 or Aurora B for 2 days and measured the DNA content by flow cytometry (Fig. 2F). Although inactivation of Aurora B caused an increase in the number of cells with 4N and even 8N DNA content, as previously reported (Hauf et al., 2003), there was no dramatic shift in the DNA content in Ufd1- or Npl4-depleted cells. Npl4 depletion did, however, cause a slight increase of cells with intermediate DNA content between 2N and 4N, at the expense of the 2N peak, which can either be explained by the aneuploidy due to the observed missegregation or by a prolonged S phase. We also counted the percentage of multinucleated cells, but did not observe any increase in Ufd1- or Npl4-siRNA-treated cells, in contrast with Aurora-B-depleted cells, indicating that cytokinesis after segregation was not affected (data not shown).

Fig. 2.

Chromosome congression and segregation defects, but not multinucleation, upon Ufd1–Npl4 depletion. (A,B) Confocal fluorescence images of unsynchronized metaphase (A) and anaphase (B) HeLa Kyoto cells that had been treated for 48 hours with the indicated siRNAs. Cells were fixed and stained with DAPI. Scale bar: 5 μm. (C) Quantification (means+s.d.) of the experiment described in A [>200 cells from three independent experiments except for the mouse Ufd1–Myc (Ufd1:myc) overexpression, which assessed >50 cells in one experiment]. Statistical significance was analysed with Student's t-test. (D) Quantification (means+s.d.) of the experiment described in B (>200 cells from three independent experiments). Statistical significance was analysed with Student's t-test. (E) Representative fluorescence overview images of interphase cells treated for 48 hours with the indicated siRNA. Cells were fixed and stained with DAPI. Note the multi-lobed nuclei and micronuclei in cells treated with the Npl4 siRNA that destabilizes the entire Ufd1–Npl4 heterodimer. Scale bar: 10 μm. (F) Analysis of the DNA content by flow cytometry of propidium-iodide-stained cells after treatment with indicated siRNAs for 48 hours.

Ufd1–Npl4 helps to remove a fraction of Aurora B from chromosomes as cells progress to metaphase

The chromosome congression defects and the segregation errors upon Ufd1–Npl4 depletion resemble the phenotype upon inactivation of the kinase Aurora B (Ditchfield et al., 2003; Hauf et al., 2003; Mora-Bermudez et al., 2007). In apparent accordance, Vong and colleagues reported a decrease of Aurora B and survivin on chromosomes upon Ufd1 depletion (Vong et al., 2005), which would suggest a positive regulation of Aurora B by Ufd1–Npl4. By contrast, we previously showed that Cdc48/p97–Ufd1–Npl4 negatively regulates Aurora B by removing it from chromatin during exit from mitosis in embryonic systems (Ramadan et al., 2007). We therefore re-examined Aurora B and survivin localization in Ufd1–Npl4-depleted cells.

To do so, we fixed unsynchronized control and siRNA-treated cells and examined the localization of Aurora B and survivin by immunostaining. Wide-field fluorescence microscopy, with identical exposure settings, revealed that the levels of chromosome-associated Aurora B increased by more than 2.5–5-fold both in Ufd1- and in Npl4-depleted cells (Fig. 3A,B). In addition, the level of survivin on the chromosomes was increased upon Ufd1–Npl4 depletion. This finding is in sharp contrast with the observation of Vong and colleagues, who reported loss of chromatin-associated Aurora B upon Ufd1 depletion (Vong et al., 2005). We therefore examined the previously published siRNA sequences (oligonucleotides V-S1 and V-S2) and compared them with a total of five Ufd1 and two Npl4 of our siRNA oligonucleotides. All our sequences upregulated the localization of Aurora B to chromosomes, although to different degrees (supplementary material Fig. S2). Intriguingly, only the published V-S2 caused the reported decrease (supplementary material Fig. S2). The conflicting results were observed despite the fact that all siRNA oligonucleotides efficiently depleted Ufd1 (supplementary material Fig. S2C). To solve this discrepancy, we performed restoration experiments. We introduced four silent point mutations in the mouse Ufd1 cDNA to render it resistant to the published V-S2, in addition to our S2 oligonucleotide, and used it to generate a resistant Ufd1–GFP expression construct (reUfd1-GFP). Overexpression of reUfd1-GFP was efficient in all cell populations without affecting depletion of Ufd1 by either V-S2 or our S2 as confirmed by western blotting (supplementary material Fig. S3A). Importantly, the increased levels of chromatin-associated Aurora B in Ufd1-S2-siRNA-treated cells were reduced to almost control levels in cells overexpressing reUfd1-GFP. By contrast, loss of Aurora B from chromosomes in cells treated with the V-S2 oligonucleotide was not restored (supplementary material Fig. S3B,C). Consistently, reUfd1-GFP restored proper chromosome alignment in S2-siRNA-treated cells, whereas it failed to do so in V-S2-treated cells (supplementary material Fig. S3B,D). This result conclusively demonstrates that the phenotype of S2 treatment was specifically caused by Ufd1 depletion, whereas V-S2 affected mitosis indirectly through an unknown mechanism.

Fig. 3.

Elevated levels of Aurora B and survivin associated with chromosomes in Ufd1–Npl4-depleted cells. (A) Cells were treated with control (siCtrl), Ufd1 or Npl4 siRNA for 48 hours, were pre-extracted to remove soluble protein and fixed. Aurora B and survivin was detected by indirect immunofluorescence using wide-field fluorescence microscopy with identical settings. (B) The intensity of Aurora B and survivin (means+s.d.) on metaphase chromatin was determined and background-subtracted as shown in A (n>21 for each condition). Data are presented for siUfd1-S2 and siNpl4-S2 (see supplementary material Fig. S2 for data from with additional oligonucleotides and supplementary material Fig. S3 for restoration experiments). (C) Western blot analysis of chromatin fractions from cells treated with control (Ctrl), Ufd1 or Aurora B (AurB) siRNA. After nocodazole arrest, mitotic cells were harvested by shake-off. Mitotic indices were comparable (see supplementary material Fig. S3 for the quantification). Equal numbers of cells were lysed with detergent, and the chromatin (pellet) separated from the soluble fractions (sup) by centrifugation. The membrane was probed with antibodies against Ufd1, Aurora B, or for phosphorylated histone H3 serine 10 (pH3). Topoisomerase II alpha (TopoIIα) and GAPDH were probed, as chromatin and cytosol controls, respectively. Note the increase in chromatin-associated Aurora B in Ufd1-depleted cells. (D) Quantification of the experiment shown in C. The intensities of the Aurora B band was normalized to the TopoII signal and presented as percentage of the control intensity for the total and pellet fractions, respectively.

The enhanced Aurora B association with chromosomes in Ufd1-depleted cells was observed by prophase and was even observed on segregating chromosomes in anaphase (supplementary material Fig. S4). To confirm this increase biochemically, we synchronized HeLa cells in nocodazole and harvested mitotic cells by shake-off. In such a way, we analyzed control and Ufd1-depleted cells, as well as Aurora-B-depleted cells for comparison. The mitotic index of the collected cells ranged between 90–95% in all preparations (supplementary material Fig. S5). Detergent-extracted cell lysates were separated by centrifugation into soluble and chromatin fractions. Western blotting confirmed a consistent increase of chromatin-associated Aurora B upon Ufd1 depletion, which was in part reflected in the total amount of cellular Aurora B (Fig. 3C,D).

Next, we ask whether Aurora B accumulated at a specific region on the chromosome in the absence of Cdc48/p97–Ufd1–Npl4 function. Analysis of Ufd1- and Npl4-depleted cells in prometaphase and metaphase, by confocal microscopy, revealed that Aurora B localized along the entire chromosome arms, whereas it was confined to centromeric regions in control cells (Fig. 4A). Aurora B on chromosome arms was active, as it was autophosphorylated on threonine 232 (Yasui et al., 2004), which we detected with a phosphorylation-specific antibody (Fig. 4A). The distribution of Aurora B along the chromosome arms was also confirmed in chromosome spreads prepared from synchronized control- and Ufd1-depleted cells (Fig. 4B). To measure the increase, we analyzed prometaphase cells by wide-field microscopy. Fluorescence intensity measurement of centromeric regions (as defined with a CREST counterstain) and areas on distal arms revealed that Aurora B levels were significantly increased in both chromosomal regions in Ufd1-depleted cells compared with its localization in control treated cells (Fig. 4C,D). From these biochemical and morphological analyses, we conclude that impairment of Ufd1–Npl4 leads to elevated levels of Aurora B associating with chromosomes.

Fig. 4.

Ufd1–Npl4 depletion leads to Aurora B spreading along chromosome arms, and its increased association at the centromere. (A) Representative confocal images of unsynchronized HeLa cells in prometaphase or metaphase after 48 hours of treatment with siRNA targeting the indicated proteins. Cells were fixed and probed with DAPI, and antibodies against Aurora B (Aur B) or phosphorylated Aurora B (pT232). Images were taken with identical laser settings. Note the Aurora B localization along chromosome arms (arrowheads) in Ufd1- and Npl4-siRNA-treated cells. Scale bar: 5 μm. (B) Representative chromosome spreads of prometaphase cells treated with control or Ufd1 siRNA. Spreads were stained with DAPI and anti-Aurora-B antibodies. The right-hand panel shows a higher magnification of a representative individual chromatid pair. Note the spreading of Aurora B along chromosome arms upon Ufd1 depletion. (C) Control-and siUfd1-treated cells were fixed, immunostained with antibodies against Aurora B and the centromeric antigen CREST, and analysed by wide-field microscopy. Scale bar: 10 μm. (D) Quantification (means±s.d.) of Aurora B intensities on chromosome arms and centromeres visualized as indicated in C.

Ufd1–Npl4 reduces chromosome-associated Aurora B activity early in mitosis

To address whether the increased levels of Aurora B on chromosomal arms and centromeres also resulted in increased Aurora B activity associated with chromosomes, we monitored phosphorylation of its substrates, histone H3 and the centromeric histone H3 homologue CENP-A, by immunofluorescence microscopy. Both proteins are first phosphorylated by Aurora B in late G2 phase: histone H3 on serine 10 and CENP-A on serine 7. We did not detect differences in their steady-state phosphorylation level in control and Ufd1-depleted cells (Fig. 5C, DMSO control; data not shown), possibly because both substrates are already nearly maximally phosphorylated in control cells. We therefore asked whether a possible increase in Aurora B activity would be reflected in a partial resistance to the chemical inhibitor of Aurora B, hesperadin.

To address this question, we first assessed whether phosphorylation of histone H3 or CENP-A is dynamic in prometaphase, as a result of repeated dephosphorylation and subsequent re-phosphorylation by Aurora B, which could then be probed in our analysis. To test this, we treated cells with 100 nM hesperadin for between 2 and 60 minutes, to inhibit possible re-phosphorylation. We then analyzed the substrate phosphorylation state by indirect immunofluorescence with phosphorylation-specific antibodies in prometaphase and metaphase cells. Given that hesperadin treatment causes morphological changes, we confirmed the mitotic stage by assessing whether the cell had broken down the nuclear envelope but not yet degraded cyclin B1, as determined by indirect immunofluorescence (data not shown). Histone H3 phosphorylation was not affected when cells were treated for up to 15 minutes (Fig. 5A,B). However, after 30 minutes, we observed a mixed population of cells with either fully, partially or non-phosphorylated chromatin, the latter cases most probably already reflecting inhibition of the initial phosphorylation in those cells that were in prophase during application of the inhibitor. This suggests that histone H3 phosphorylation is not dynamic and that the phosphorylation that occurs in late G2 is stable at least until anaphase onset. In marked contrast, CENP-A phosphorylation at the centromere was sensitive to a 15-minute incubation with hesperadin in almost all the sample cells, indicating that it is constantly dephosphorylated and needs to be re-phosphorylated by Aurora B to maintain its phosphorylation status during prometaphase.

Fig. 5.

Ufd1–Npl4 attenuates chromosome-associated Aurora B activity towards a centromeric substrate. (A,B) Phosphorylation of CENP-A, but not of histone H3 is dynamic during prometaphase. Unsynchronized cells were treated for indicated times with 100 nM of the Aurora B inhibitor hesperadin (Hesp), fixed and phosphorylation of histone H3 (serine 10) (pH3) and CENP-A (serine 7) (pCENP-A) detected by indirect immunofluorescence with phosphorylation-specific antibodies. Cells were only assessed if they had broken down the nuclear envelope, but not yet degraded cyclin B1 (as identified with an anti-cyclin-B1 antibody). A shows fluorescence micrographs and B a quantification (means+s.d.; n=40). Note that CENP-A, but not H3 phosphorylation is sensitive to 15-minute incubations with 100 nM hesperadin. (C,D) Cells were treated with increasing concentrations of hesperadin for 15 minutes, and phosphorylation of CENP-A was detected as in A. C shows fluorescence micrographs and D a quantification (means+s.d.; n>137 from three independent experiments). Note that CENP-A phosphorylation is resistant to 20–50 nM hesperadin in Ufd1-siRNA-but not control-treated cells.

We therefore investigated the sensitivity of CENP-A phosphorylation to 15-minute incubations with increasing concentrations of hesperadin (from 1 to 100 nM) in control and Ufd1-depleted cells (Fig. 5C,D). We found that phosphorylation was undetectable at 20 nM in the majority of control cells, whereas phosphorylation was resistant to 50 nM hesperadin and was even partially detectable at 100 nM hesperadin. This result demonstrates that the impairment of Ufd1–Npl4 function leads to a shift towards stronger phosphorylation of CENP-A, which is consistent with an elevated Aurora B activity associated with chromosomes. Together with the observed increase in Aurora B protein, this suggests that Ufd1–Npl4 helps to restrain the activity of Aurora B early in mitosis, by removing a fraction of it from chromosomes as cells progress to metaphase.

Ufd1–Npl4 antagonizes Aurora B during chromosome congression

Next, we asked whether regulation of Aurora B activity by Cdc48/p97–Ufd1–Npl4 had any functional relevance. We did not observe an apparent effect on chromosome condensation in Ufd1-depleted cells, even when they were challenged with hypotonic buffer (supplementary material Fig. S6), which leads to a decompaction of unstable chromosomes (Gassmann et al., 2004). To address a possible effect on chromosome congression, we analyzed the efficiency and timing of chromosome alignment both in fixed and live cells. As a first approach, we treated control or Ufd1-depleted cells with increasing concentrations of hesperadin (from 0.7 to 100 nM), fixed the cells after 1 hour and determined the percentage of prometaphase and metaphase cells with fully aligned chromosomes (Fig. 6A,B). As expected, both depletion of Ufd1 and addition of hesperadin at concentrations equal or above 5 nM caused a marked increase in misaligned chromosomes compared with that observed in control cells. Importantly, however, chromosome alignment in Ufd1-depleted cells was partially rescued by moderate inhibition of Aurora B, with 1 nM of hesperadin, suggesting that the alignment defect in Ufd1-depleted cells was, at least in part, caused by overactivation of Aurora B. Conversely, Ufd1-depletion restored proper alignment in the presence of 5–15 nM hesperadin, showing that the overactivation of Aurora B due to impaired Ufd1–Npl4 function can overcome partial inhibition of Aurora B.

Fig. 6.

Ufd1–Npl4 antagonizes Aurora B during chromosome congression. (A,B) Unsynchronized control or Ufd1-depleted cells were treated with indicated concentrations of hesperadin (Hesp) for 1 hour, and were fixed and stained with DAPI and anti-cyclin B1 antibodies. Cells were assessed if they had broken down the nuclear envelope, but not yet degraded cyclin B1, as in Fig. 5A, and the percentage of cells with fully aligned chromosomes was determined. Data are means+s.d. from three independent experiments (n=100 for each condition). (C,D) Hela cells expressing H2B–mRFP were transfected with control (Ctrl) or Ufd1 siRNA. After 48 hours, cells were treated with hesperadin (50 nM), or equivalent amounts of its solvent DMSO, and imaged by confocal time-lapse microscopy as in Fig. 1B. C shows images from representative movies and D shows a quantification of the time between nuclear envelope breakdown and metaphase plate formation. Individual cells were followed in time-lapse movies and scored when they formed a distinct metaphase plate. Scored cells were plotted cumulatively for each time point. The data are from three independent experiments (siCtrl, n=146; siCtrl + Hesp, n=138; siUfd1, n=114; siUfd1 + Hesp, n=88).

To confirm this result, we monitored metaphase plate formation in live cells (Fig. 6C,D). Control or Ufd1-depleted cells stably expressing histone H2B–mRFP were imaged in the absence or presence of 50 nM hesperadin using time-lapse video microscopy as in Fig. 1. We determined the time from nuclear envelope breakdown to the formation of a distinct metaphase plate, when no misaligned chromosomes could be detected. We then plotted the percentage of cells that had formed metaphase plates cumulatively as a function of time (Fig. 6D). Control cells readily formed a metaphase plate within 20 minutes. In accordance with our results in Fig. 1, Ufd1-depleted cells required longer, but eventually nearly all cells completed congression. By contrast, and as expected, less than 10% of the hesperadin-treated cells managed to form a metaphase plate. Importantly, Ufd1-depletion partially restored metaphase plate formation in hesperadin-treated cells, as chromosome alignment occurred in more than 50% of cells, though at a lower rate and with metaphase plates occasionally collapsing after formation. Taken together, these results demonstrate that Ufd1–Npl4 antagonizes Aurora B during chromosome congression and suggest that misregulation of Aurora B, at least in part, accounts for the congression phenotype in Ufd1–Npl4-depleted cells.


The data presented here establish that the Ufd1–Npl4 adapter of the Cdc48/p97 AAA+ ATPase is required for proper chromosome segregation in human somatic cells. Cellular depletion of Ufd1–Npl4 in HeLa cells specifically interfered with efficient chromosome alignment to the metaphase plate, delayed prometaphase and caused anaphase defects, which led to missegregation and abnormal nuclear morphology. Depletion of Ufd1 alone had a weaker effect than destabilization of the whole Ufd1–Npl4 complex after Npl4 depletion. This suggests that Npl4 alone can still partially fulfil the function of Ufd1–Npl4. Both Ufd1 and Npl4 have ubiquitin-binding activity, suggesting that one binding site might compensate for a partial depletion. Interestingly, Npl4 contains an MPN domain (residues 226–396; Kay Hofmann, Miltenyi Biotec, personal communication), which is found in certain deubiquitinating enzymes (Ambroggio et al., 2004). Although the catalytic residues are not conserved, it is possible that this domain confers a crucial function of the Ufd1–Npl4 complex.

A role for Cdc48/p97–Ufd1–Npl4 in mitosis was recently questioned, as a mitotic delay could not be observed in C. elegans embryos treated with RNAi against Cdc48/p97 or Ufd1–Npl4 (Mouysset et al., 2008). However, the RNAi in that study was fed to the animals rather than injected, which can limit efficacy. Moreover, a major delay due to chromosome misalignment is not expected in early embryos, because they have a very weak spindle assembly checkpoint under normal growth conditions, which can only delay anaphase onset moderately and only when the spindle is disrupted severely (Nystul et al., 2003). We suggest, therefore, that the evidence presented by Mouysset and colleagues (Mouysset et al., 2008) does not preclude a role for Cdc48/p97–Ufd1–Npl4 in mitosis in C. elegans.

Our new results confirm the central role of Cdc48/p97–Ufd1–Npl4 in mitosis in HeLa cells. Along with previous results from C. elegans and Xenopus (Ramadan et al., 2007), as well as yeast (Cao et al., 2003; Cheng and Chen, 2010; Frohlich et al., 1991; Moir et al., 1982), they suggest that the Cdc48/p97–Ufd1–Npl4 chaperone complex has a conserved role in maintaining chromosome stability and euploidy during proliferation. Here, we also provide several lines of evidence that Cdc48/p97–Ufd1–Npl4 antagonizes Aurora B during chromosome segregation. This is supported by the observation that Ufd1–Npl4 depletion caused an increase in the levels of Aurora B associated with mitotic chromosomes, as shown by immunohistochemistry and biochemical fractionation. This notion is based on results from a total of seven Ufd1 and Npl4 siRNA oligonucleotides. We note, however, that the depletion efficiencies and degree of segregation errors do not completely correlate with the various degrees of Aurora B accumulation for the different siRNA oligonucleotides. Nevertheless, restoration experiments with a resistant Ufd1 cDNA demonstrated that both the Aurora B accumulation and the segregation errors are a specific consequence of Ufd1 depletion. Moreover, the same experiments showed that the effect of an oligonucleotide previously reported to cause loss of Aurora B from chromosomes (Vong et al., 2005) was indirect, as it was not rescued by overexpression of the resistant Ufd1 cDNA.

In Xenopus egg extract, Cdc48/p97–Ufd1–Npl4 removes Aurora B from chromosomes at the end of mitosis (Ramadan et al., 2007). Although not shown directly in this manuscript, this suggests that Cdc48/p97–Ufd1–Npl4 might act in a similar manner at earlier stages of mitosis in HeLa cells and remove some Aurora B from chromosomes as cells progress from prophase (where Aurora B localizes along the entire chromosomes) to metaphase, thereby attenuating its activity. Given that Cdc48/p97–Ufd1–Npl4 function is triggered by ubiquitylation (Ramadan et al., 2007; Ye, 2006), this implies the requirement for a specific ubiquitin ligase activity early in mitosis. Interestingly, a cullin-3-based ligase has recently been shown to ubiquitylate Aurora B and to be required for Aurora B removal from chromosomes in prometaphase (Maerki et al., 2009; Sumara et al., 2007). Moreover, cullin-3 physically interacts with Cdc48/p97 (Alexandru et al., 2008) raising the possibility that the two factors cooperate to regulate chromatin-associated Aurora B. Independent of whether this holds true, both sets of findings concur that Aurora B might be actively removed from chromatin in a ubiquitin-regulated manner (Sumara et al., 2008). Other than in Xenopus egg extracts, where Aurora B remains stable, however, the available data suggests that the mobilized fraction of Aurora B in HeLa cells might, at least in part, be degraded. This hypothesis is supported by the increased total Aurora B levels in both Ufd1-depleted cells (Fig. 4C,D) and cullin-3-depleted cells (Sumara et al., 2007). Furthermore, proteasome inhibition leads to increased and persisting levels of Aurora B on chromosomes (Sumara et al., 2007).

More direct evidence for the antagonizing role of Cdc48/p97–Ufd1–Npl4 with respect to Aurora B is the elevated activity towards the centromeric substrate CENP-A. Ufd1-depleted cells are less sensitive to partial inhibition of Aurora B, by addition of its inhibitor hesperadin, with respect to CENP-A phosphorylation. The central question arising from these findings is, therefore, whether the activity of Cdc48/p97–Ufd1–Npl4 in antagonizing Aurora B in prometaphase has any functional impact on chromosome segregation and hence whether excessive Aurora B activity can explain the defects caused by Ufd1–Npl4 depletion. A crucial role for Ufd1–Npl4-mediated downregulation of Aurora B in chromosome segregation is supported by the similarity of the phenotypes observed upon Ufd1–Npl4 depletion and inactivation of Aurora B (severe defects in chromosome alignment, as well as lagging and entangled chromosomes in anaphase that lead to the prominent occurrence of micronuclei and multi-lobed nuclei) (Ditchfield et al., 2003; Hauf et al., 2003; Mora-Bermudez et al., 2007; Vagnarelli and Earnshaw, 2004). Segregation defects in anaphase might also be caused indirectly through inefficient DNA replication, as observed after depletion of Cdc48/p97 or Ufd1–Npl4 in C. elegans embryos (Mouysset et al., 2008). However, although we do not exclude the possibility that problems in S phase might contribute to the observed anaphase defects, we also do not detect any major inhibition of DNA replication in Ufd1–Npl4-depleted HeLa cells (Fig. 2F), as described for C. elegans embryos (Mouysset et al., 2008). Furthermore, and more importantly, our results draw a clear link between Ufd1–Npl4 and Aurora B during mitosis, as we show that they functionally interact in regulating chromosome alignment. Attenuation of Aurora B activity, by chemical inhibition with hesperadin, improved chromosome alignment to the metaphase plate in Ufd1-depleted cells. Moreover, Ufd1-depletion partially rescued congression and metaphase plate formation in hesperadin-treated cells. This functional interaction again suggests that Cdc48/p97–Ufd1–Npl4 antagonizes Aurora B during chromosome congression and that overactivation of Aurora B accounts, at least in part, for the phenotype of Ufd1–Npl4 depletion in HeLa cells.

A recent report, published during the revision of this manuscript, revealed that in Saccharomyces cerevisiae, Cdc48/p97 also counteracts Aurora (Ipl1) during chromosome bi-orientation (Cheng and Chen, 2010). Taken together with our data presented here, this suggests that the antagonizing function of Cdc48/p97 towards Aurora B is conserved during evolution. Interestingly, Cheng and Chen further suggest a role for the p47 adapter homologue Shp1. We did not find any evidence for an involvement of p47 in chromosome alignment in HeLa cells, which might point to distinct requirements for adapters in the two systems, possibly owing to differences like the open compared with the closed mitosis.

The implication that excessive Aurora B activity at the centromere, upon Cdc48/p97–Ufd1–Npl4 depletion, induces congression defects is in line with the notion that both negative and positive regulation of Aurora B is crucial for its role in controlling bipolar spindle attachment (Kelly and Funabiki, 2009). Excess Aurora B activity might either overly destabilize microtubule attachments by phosphorylation of Aurora-B-controlled anchor points, such as Ndc80 (DeLuca et al., 2006). Alternatively, it might prevent adjustment of incorrect spindle attachment to kinetochores by phosphorylation and thus inhibition of the microtubule-depolymerizing activity of MCAK (Andrews et al., 2004; Lan et al., 2004; Ohi et al., 2004). The requirement for negative regulation to limit Aurora B activity has been shown before in the case of protein phosphatase-1, which antagonizes Aurora B at the kinetochore to ensure proper chromosome segregation (Emanuele et al., 2008; Francisco et al., 1994; Pinsky et al., 2006). It will be interesting to clarify whether phosphatases and the ubiquitin system, including Cdc48/p97, cooperate to control Aurora B activity.

Materials and Methods

Antibodies and plasmids

Monoclonal (1:250; 5E2) and polyclonal (HME14) anti-Ufd1, anti-Npl4 (HME16) (all 1:500), anti-p47 (1:5000; HME22) and anti-Cdc48/p97 (1:2000; HME01) antibodies were as described previously (Meyer et al., 2000). Antibodies against the following proteins were purchased and used at the indicated dilutions: Aurora B (1:500; 611082, BD Bioscience), phosphorylated Aurora B (serine 10) (1:200; Rockland), TopoIIα (1:1000; Boehringer Ingelheim), phosphorylated histone H3 (serine 10) (1:4000; 06-570, Upstate), GAPDH (G8795, Sigma) 1:5000, phosphorylated CENP-A (1:1000; 04-792, Millipore), cyclin B1 (1:500; sc-245, Santa Cruz Biotechnology), anti-survivin (1:500; AF886, R&D Systems) and human anti-centromere (kinetochore) CREST (1:400, Antibodies).

Mouse Ufd1 cDNA was cloned into pcDNA3.1 (Invitrogen) with a sequence coding for a C-terminal Myc tag. Mouse Ufd1 cDNA was cloned into the BamHI site of pEGFP-N3 (Clontech) and the sequence stretch 5′-CTGCGGGTGATGGAGACCAAA-3′ was mutated to 5′-CTGCGGGTAATGGAAACTAAG-3′ by the Quikchange protocol to render it resistant to V-S2 Ufd1 siRNA.

Cell culture, establishment of stable cell lines and RNAi

For routine cell culture, HeLa Kyoto cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum and 1% penicillin/streptomycin (routine media). For RNAi, cells were plated 1 day before transfection with 100–200 nM siRNA (Microsynth, Switzerland) using Oligofectamine, or 10 nM siRNA using Lipofectamine RNAiMAX in OptiMEM minimal medium according to the manufacturer's instructions (Invitrogen). Cells were analysed 48 hours after transfection. As a negative control in each experiment, a non-silencing siRNA with the sense sequence 5′-UUCUCCGAACGUGUCACGUdTdT-3′ (adapted from Ambion, Austin, TX) was used. The following siRNAs, targeting Cdc48/p97 cofactors Ufd1, Npl4 and p47 were used (sense strand): Ufd1-S1, 5′-GGGCUACAAAGAACCCGAAdTdT-3′; Ufd1-S2, 5′-GUGGCCACCUACUCCAAAUdTdT-3′; Ufd1-S3, 5′-CTACAAAGAACCCGAAAGAdTdT-3′; Ufd1-S4, 5′-ACAAAGAACCCGAAAGACAdTdT-3′; Ufd1-S5, 5′-CTGGGCTACAAAGAACCCGAAdTdT-3′; Npl4-S1, 5′-CGUGGUGGAGGAUGAGAUUdTdT-3′; Npl4-S2, 5′-CAGCCUCCUCCAACAAAUCdTdT-3′; p47-S1, 5′-AGCCAGCUCUUCCAUCUUAdTdT-3′; and p47-S2, 5′-UCAGAGCCUACCACAAACAdTdT-3′. If not otherwise stated, Ufd1-S2, Npl4-S2 and p47-S1 were used for functional assays. Aurora B siRNA, 5′-GGTGATGGAGAATAGCAGT-3′ (Hauf et al., 2003), Ufd1-V-S1 and -V-S2 were as published previously (Vong et al., 2005).

Cell extracts

For western blot analysis, HeLa cells were lysed for 15 minutes on ice in 150 mM KCl, 25 mM Tris-HCl pH 7.4, 5 mM MgCl2, 5% glycerol, 1% Triton X-100, 1 mM dithiothreitol (DTT) and EDTA-free protease inhibitor cocktail (Roche), and centrifuged at 10,000 g at 4°C for 20 minutes, with the supernatant recovered for analysis. For fractionation, cells were synchronized in 330 nM nocodazole for 16 hours. Mitotic cells were harvested by shake-off and the mitotic index determined. Equal numbers of cells were incubated for 15 minutes in lysis buffer (10 mM HEPES pH 7.4, 10 mM KCl, 1.5 mM MgCl2, 340 mM sucrose, 10% glycerol, 0.25% Triton X-100, 1 mM DTT and protease inhibitors) and lysed by 20 strokes in a douncer. The chromatin was sedimented by centrifugation at 1300 g for 5 minutes and washed twice in lysis buffer; this supernatant, which contained the soluble proteins, was centrifuged at 16,000 g for 15 minutes and the supernatant harvested for analysis. Equal fractions were subjected to western blotting.

Confocal live cell imaging (mitotic timing)

Stable H2B–mRFP HeLa Kyoto cells growing on Ibidi chambers were transfected with siRNAs. Cells were cultured, for data acquisition, in minimal essential medium with Hank's F12 (MEM/F12; Gibco) supplemented with 15 mM HEPES pH 7.4. Confocal fluorescence time-lapse movies were taken with a Leica SP2 AOPS confocal microscope equipped with a 20× 0.7 NA Apofluor objective lens. Cells were kept at 37°C during imaging. For acquisition, pictures from one focal plane of living cells were taken every 1–2 minutes. Microscope settings were as follows: picture average, 6; 400 Hz scan speed; 8-bit resolution; logical size, 1024×1024, PMT 1: 0.7 Offset, 736.2 volts; PMT2: 3.5 Offset, 677.60 volts. Images were processed using Imaris (Bidplane) and ImageJ software.

Indirect immunofluorescence

Cells were grown on glass coverslips and transfected with siRNAs as described above. HeLa cells were either first treated with pre-extraction buffer (25 mM Hepes pH 7.5, 50 mM NaCl, 1 mM EGTA, 3 mM MgCl2, 300 mM sucrose and 0.5% Triton X-100) for 5 minutes at 4°C, or fixed directly with 4% paraformaldehyde in PBS for 10–15 minutes. For hypotonic treatment, cells were incubated for 5 minutes in RSB buffer (10 mM Tris-HCl pH 7.4, 10 mM NaCl and 5 mM MgCl2) before fixing. Cells were permeabilized with 0.1% Triton X-100 (Sigma) in PBS for 5–10 minutes and washed as described above. Permeabilized cells were blocked with 3% BSA in PBS for 30 minutes, followed by indirect immunostaining. Samples were mounded with Mowiol and analyzed using a Leica SP2 AOPS or a SP1 AOPS confocal microscope, or Zeiss Axiovert TV or Leica DM6000B epifluorescence microscope. Images were processed and intensities quantified with the Imaris or ImageJ software, respectively.

Statistical analysis

Datasets were quantified with Sigma Plot and Past ( software. Normal and non-normal distributed and datasets with equal and different sample numbers, were analyzed by Student's t-test, except for Fig. 1C where a Mann–Whitney U-test was used.


We thank Izabela Sumara, Matthias Peter, Monica Gotta and Patrick Meraldi for help and critical discussions. We thank Catherine Brasseur for technical help. This work was supported by a grant of the Deutscher Akademischer Austauschdienst (to O.P.), and grants of the Schweizerischer Nationalfond (3100A0-113749), the ETH (TH-03 07-1) and the Novartis Research Foundation (to H.M.). M.H.A.S. generated H2B–mRFP-expressing cells and, together with D.W.G., designed quantitative live imaging assays.


  • Accepted January 6, 2011.


View Abstract