Letter | Published:

VHL loss causes spindle misorientation and chromosome instability

Nature Cell Biology volume 11, pages 9941001 (2009) | Download Citation



Error-free mitosis depends on fidelity-monitoring checkpoint systems that ensure correct temporal and spatial coordination of chromosome segregation by the microtubule spindle apparatus. Defects in these checkpoint systems can lead to genomic instability, an important aspect of tumorigenesis. Here we show that the von Hippel-Lindau (VHL) tumour suppressor protein, pVHL, which is inactivated in hereditary and sporadic forms of renal cell carcinoma, localizes to the mitotic spindle in mammalian cells and its functional inactivation provokes spindle misorientation, spindle checkpoint weakening and chromosomal instability. Spindle misorientation is linked to unstable astral microtubules and is supressed by the restoration of wild-type pVHL in pVHL-deficient cells, but not in naturally-occurring VHL disease mutants that are defective in microtubule stabilization. Impaired spindle checkpoint function and chromosomal instability are the result of reduced Mad2 (mitotic arrest deficient 2) levels actuated by pVHL-inactivation and are rescued by re-expression of either Mad2 or pVHL in VHL-defective cells. An association between VHL inactivation, reduced Mad2 levels and increased aneuploidy was also found in human renal cancer, implying that the newly identified functions of pVHL in promoting proper spindle orientation and chromosomal stability probably contribute to tumour suppression.


VHL cancer syndrome is caused by inactivating germline mutations in the VHL tumour suppressor gene and is associated with an increased risk of a variety of tumours, including clear cell renal cell carcinoma (ccRCC). Biallelic inactivation of the VHL gene also occurs in 50–75% of sporadic ccRCC. Human pVHL exerts its tumour suppressor function, in part, by functionally associating with interphase and ciliary microtubules1,2. As mitotic cell division depends on proper control of microtubule dynamics, we investigated whether pVHL (A002833) also functions in mitosis. As shown by indirect immunofluorescence microscopy, endogenous pVHL associates with spindle and astral microtubules in primary mouse embryonic fibroblasts (MEFs; Fig. 1a, b) and primary human renal proximal tubular epithelial cells, U2OS and HeLa cells (data not shown). This staining was evident from prophase until cytokinesis (Fig. 1a) and was specific (Supplementary Information, Fig. S1a). Thus, pVHL is a genuine component of the mitotic spindle.

Figure 1: pVHL localizes to microtubules in mitosis and loss of pVHL leads to spindle misorientation.
Figure 1

Co-staining was performed with anti-mouse-pVHLCT and anti-α−tubulin antibodies, DAPI was used to stain DNA. (a, b) Localization is shown in all phases of mitosis (a) and to the astral microtubules of the metaphase spindle (b; higher magnifications, right). Scale bars, 10 μm. (c, d) Frame series of movies from HeLa-H2B-Tub cells transfected with control siRNA (sictrl; c) or VHL siRNA (siVHL; d). Upper panels, H2B–GFP (green) and RFP-α-tubulin (red); lower panels, RFP-tubulin only. Nuclear envelope breakdown (NEBD) was set at 0 min. Scale bars, 2.5 μm. (e) Quantification of normal, rotating and multipolar mitotic spindles of movies from HeLa-H2B-Tub cells: sictrl (n = 70), siVHL (n = 71) from four independent experiments, P = 8 × 10−4 (normal), P = 4 × 10−4 (rotating), not significant (multipolar; Student's t-test). (f, g) Time from NEBD to anaphase onset in cells described in e (f; sictrl, n = 70; siVHL, n = 71) and in siVHL-treated cells (g; normal spindles, n = 34 and rotating spindles, n = 33). (h) Scheme describing spindle angle (α) measurement. (i) Image series of mitotic spindles stained with anti-α−tubulin (green) and anti-γ−tubulin (red) antibodies and DAPI (blue) from MEFs transfected with LKO-1-ns or LKO-1-Vhlh. The position of the z stage is indicated in μm; 3D, xy projection. Scale bars, 2 μm. (j) Distribution of the mitotic spindle angle (α°) of MEFs treated as in i (from 6 slides, P = 1 × 10−14, Mann-Whitney test). (k) Average spindle angle of MEFs treated as in i (mean ± s.d., n = 6 slides, *P = 1 × 10−5, Student's t-test). (l) Spindle diameter of MEFs treated as in i (mean ± s.d., n = 6 slides). (m) Western blot of MEF cell lysates treated as in i.

Next, we assessed whether loss of pVHL function impairs mitotic division. We used time-lapse video microscopy in HeLa cells, engineered to stably express EGFP (enhanced green fluorescent protein)-tagged histone H2B and α-tubulin–mRFP (monomeric red fluorescent protein; referred to as HeLa-H2B-Tub) to mark chromosomes and microtubules, respectively3. As shown in Fig. 1c and Supplementary Information, Video S1, control short interfering RNA (siRNA)-treated HeLa cells assembled a bipolar spindle parallel to the growth plane. In contrast, siRNA targeting VHL caused spindle misorientation ('rotating spindles'; Fig. 1d; Supplementary Information, Video S2). Quantification revealed that this phenomenon occurred in 48% of pVHL-depleted mitotic cells but only in 4% of control siRNA-treated cells (Fig. 1e). Knockdown of pVHL did not delay the onset of anaphase (Fig. 1f). Neither was there a difference in the timing of anaphase onset between rotating and normally oriented spindles in pVHL-depleted cells (Fig. 1g), implying that spindle misorientation associated with pVHL deficiency does not translate into a measurable delay in mitotic progression.

We also determined the angle between the mitotic spindle axis and the growth plane in fixed cell samples (Fig. 1h, i). Compared with control non-silencing infection using LKO-1-ns (a lentiviral vector, LKO-1, expressing short hairpin RNA, shRNA, against a non-silencing control; Fig. 1m), knockdown of pVHL with LKO-1-Vhlh (which expresses mouse VHL shRNA) in MEFs broadened the distribution of the spindle angles (Fig. 1j), resulting in an increase in the mean spindle angle from 10.4° to 29.4° (Fig. 1k). Similar results were obtained in HeLa cells depleted of pVHL (Supplementary Information, Fig. S1b–d) and in VHL-deficient MEFs (Supplementary Information, Fig. S1g, h). As the theoretical average value for a random distribution of spindle angles is 32.7°, our data suggest that downregulation of pVHL causes an almost complete randomization of spindle orientation. pVHL depletion did not affect the diameter or the morphology of the mitotic spindle in MEFs or HeLa cells (Fig. 1l; Supplementary Information, Fig. S1e and f, respectively). Thus, at least one mitotic function of pVHL is to maintain normal spindle orientation.

Correct mitotic spindle orientation depends on interactions of astral microtubules with the cell cortex4,5. Quantification of the relative astral microtubule signal intensity from anti-α-tubulin antibody staining (Fig. 2b) demonstrated that depletion of pVHL leads to a reduction in astral microtubules by > 50% compared with the control (Fig. 2a left panels and 2c). The fluorescence intensity of spindle microtubules did not change in pVHL-knockdown HeLa cells (Supplementary Information, Fig. S2b). Similar results were obtained in MEFs and, after glutaraldehyde fixation, in HeLa cells (Supplementary Information, Fig. S2a). The loss of astral microtubules in pVHL-depleted cells was rescued by taxol (a microtubule stabilizing drug; Fig. 2a, right panels and 2c), indicating that the observed decrease in signal intensity was probably a consequence of unstable astral microtubules. Taxol treatment did not affect pVHL levels (Supplementary Information, Fig. S2c). These results suggest that loss of pVHL function destabilizes astral microtubules. In this regard, in vitro microtubule polymerization assays using MAP (microtubule associated protein)-rich tubulin (pure tubulin does not support pVHL–microtubule interactions; A. H. and W. K., unpublished data)6 revealed a significant shift in the equilibrium towards polymerized tubulin in the presence of purified pVHL30 (wild-type pVHL), but not pVHL30,Δ95–123, a mutant lacking the microtubule-binding domain (Fig. 2d). Supplementary Information, Fig. S2d shows that equal amounts of purified proteins were added to each reaction. Importantly, pVHL30 did not increase the number, but increased the length of microtubules as shown by microscopic examination of microtubule polymerization reactions after addition of 5% rhodamine-labelled tubulin (Fig. 2e, f). Thus, pVHL enhances microtubule stability rather than microtubule nucleation.

Figure 2: pVHL stabilizes astral microtubules.
Figure 2

(a) Staining with an anti-α−tubulin antibody in control siRNA (sictrl)- or VHL siRNA (siVHL)-transfected HeLa cells, either untreated (−) or treated (+) for 20 min with 1 μM taxol. Scale bars, 5 μm. (b) Calculation of the relative astral microtubule signal intensity (Iastral,rel): signal intensity of the astral and spindle microtubules (Itotal, red area); spindle microtubules only (Ispindle, yellow area); i = z-stack position. (c) Average Iastral,rel from HeLa cells treated as in a (mean ± s.d., n = 3 slides, 15 spindles per slide, P = 0.003, Student's t-test). (d) Polymerization curves of tubulin in the presence of pVHL30, pVHL30,Δ95–123 and MBP (maltose-binding protein, purification tag only). Corresponding bar graphs depicting the area under curves are shown below (mean ± s.d., n = 3 or 4; MBP versus pVHL30, P = 0.01; pVHL30 versus pVHL30,Δ95–123, P = 0.002; MBP versus pVHL30,Δ95–123, n.s., not significant; ANOVA with Tukey Post-Hoc analysis). (e) Microtubules were spiked with rhodamine-labelled tubulin (5%) in the presence of pVHL30, pVHL30,Δ95–123 and MBP, and polymerization was performed as in d with fixation after 10 min. The corresponding length distribution is shown from 298–333 microtubules per condition (mean ± s.d., n = 14 fields from two experiments; MBP versus pVHL30, P = 5 × 10−9; pVHL30 versus pVHL30,Δ95–123, P = 9 × 10−9; MBP versus pVHL30,Δ95–123, not significant; ANOVA with Tukey Post-Hoc analysis). (f) Average of microtubules per field in e (mean ± s.d. n = 14 fields from two experiments). (g) Angle distribution (αo) of mitotic spindles in 786-O cells lacking pVHL (control) or reconstituted with wild-type pVHL (pVHL30), pVHL30 lacking the microtubule-binding domain (pVHL30,Δ95–123) or one of four naturally occurring pVHL30 mutants (pVHL30,Y98H, pVHL30,Y98N, pVHL30,Y112H and pVHL30,Y112N) . P = 5.6 × 10−21 for the whole population (Kruskall-Wallis test, see Supplementary Information, Table S1 and for further statistical analysis, Table S2). (h) Average spindle angle of 786-O cell pools treated as in g (mean ± s.d., n = 4–9 slides; pVHL30, P = 0.008; pVHL30,Δ95–123, P = 0.03; pVHL30,Y98H, P = 0.01, pVHL30,Y112H, P = 0.05, ANOVA with Dunnett PostHoc analysis for comparison with control).

These findings suggest that inactivation of pVHL destabilizes astral microtubules, which in turn causes spindle misorientation. Indeed, the spindle misorientation phenotype associated with loss of pVHL function is linked to the microtubule stabilization function of pVHL. VHL-negative RCC 786-O cells (control) show a spindle orientation defect, which manifests as a broadened distribution of the spindle angle (Fig. 2g), with a mean spindle angle of 17.9° (Fig. 2h). Re-introduction of wild-type pVHL30 in these cells rescued the spindle misorientation phenotype, whereas the pVHL30,Δ95–123 mutant did not (Fig. 2g, h). Naturally occurring type 2A VHL mutants, pVHL30,Y98H and pVHL30,Y112H, defective in microtubule stabilization7, but with only a mild defect in ubiquitylation of HIFα8 (hypoxia inducible factor-α; pVHL targets HIFα for degradation, a tumour suppressor function) also failed to correct the spindle misorientation phenotype of 786-O cells. In contrast, type 2B VHL mutants, pVHL30,Y98N and pVHL30,Y112N, which show normal microtubule stabilization function7 but fail to degrade HIFα8, suppressed spindle misorientation (Fig. 2g, h). Immunoblotting confirmed the expression of the various pVHL species in these stable 786-O cell populations (Supplementary Information, Fig. S2e). None of the pVHL species investigated affected spindle diameter or spindle morphology (Supplementary Information, Fig. S2f and data not shown). Hence, the ability of pVHL to correct mitotic spindle orientation is disrupted in VHL disease mutants defective in microtubule stabilization.

As shown above, depletion of pVHL disrupts spindle orientation but does not lead to a delay in anaphase onset in Hela-H2B-Tub cells. We noticed, however, a small but significant increase in the rate of chromosome missegregation during anaphase (Supplementary Information, Video S3), indicative of a possible defect in spindle checkpoint function. The spindle checkpoint prevents chromosome missegregation and aneuploidy by delaying sister chromatid separation until all chromosomes have achieved bipolar kinetochore–microtubule attachment9. This delay is achieved by inhibition of the anaphase-promoting complex (APC/C) through spindle checkpoint proteins, in particular the Mad1–Mad2 complex and BubR1, which are recruited to unattached kinetochores (reviewed in ref. 10). Immunofluorescence analysis revealed strong signals for Mad1 and BubR1 at kinetochores, independent of pVHL status, in HeLa cells treated with nocodazole, a strong spindle checkpoint trigger (Fig. 3a). Mad2 was also recruited to kinetochores but its levels were reduced by 60% in pVHL knockdown cells compared with control cells (Fig. 3a). Similar results were obtained in pVHL-depleted prometaphase cells that had not been treated with nocodazole (Supplementary Information, Fig. S3a), suggesting that the spindle-checkpoint may be weakened in pVHL-deficient cells. Interestingly, immunoblotting revealed reduced levels of Mad2 (but not Mad1 or BubR1) in both interphase and nocodazole-arrested HeLa cells (Fig. 3b), implying that pVHL depletion affects Mad2 protein levels in a cell-cycle independent manner. Lower levels of Mad2 protein were also observed when Vhlh was eliminated genetically by expression of Cre-recombinase in MEFs derived from Vhlhfl/fl mice (Supplementary Information, Fig. S3b). These observations potentially explain the low amounts of Mad2 recruited to kinetochores on spindle checkpoint activation. Real-time PCR of RNA from HeLa cells showed no major changes in Mad2 (or Mad1 and BubR1) mRNAs after pVHL depletion (Fig. 3c). In addition, cycloheximide-chase and proteasome inhibition experiments revealed that pVHL does not affect Mad2 protein stability (Fig. 3d, e, respectively). Thus, VHL loss negatively affects Mad2 protein levels through a mechanism(s) yet to be identified.

Figure 3: Loss of pVHL leads to reduced Mad2 levels and aneuploidy.
Figure 3

(a) Mad1-, Mad2- or BubR1-associated kinetochore signal intensities, in HeLa cells treated with VHL siRNA (siVHL) or control siRNA (sictrl) and arrested with nocodazole, normalized to CREST (which stains kinetochores) signal intensities (mean ± s.d., n = three independent experiments, P = 0.007, Student's t-test). Scale bars, 1 μm. (b) Western blot of exponentially growing (Exp) and nocodazole arrested (Noco) HeLa cell lysates treated with siVHL or sictrl. (c) Quantitative PCR of HeLa cells treated with siVHL or sictrl using primers against VHL, Mad1, Mad2 (a and b) and BubR1, normalized against 18S rRNA (mean ± 2 × s.d., n = four independent experiments, P = 0.002, Student's t-test). (d) Western blot of HeLa cell lysates treated with siVHL or sictrl after cycloheximide (CHX; 100 μg ml−1) treatment. (e) Western blot of HeLa cell lysates treated with siVHL or sictrl after 26S proteasome inhibition for 6 h. (f) Mitotic index in HeLa cells transfected with sictrl, siVHL or siMad2 followed by nocodazole treatment (mean ± s.d., n = 3–5 independent experiments; siVHL, P = 0.01; siMad2 P = 8 × 10−6, Student's t-test). (g) Western blot of 786-O cell lysates expressing human pVHL30 or an empty vector as control. (h) Cumulative frequency of nocodazole escape by 786-O cells described in g (control, n = 38; pVHL30, n = 33,). (i) Chromosome spread from a primary MEF cultured for 5 days; diploid (2n) murine cells contain 40 chromosomes. (j) Quantification of aneuploid cells from primary MEFs treated with LKO-1-ns or LKO-1-Vhlh (mean ± s.d., n = 3 slides, P = 0.002, Student's t-test). (k) Western blot of cell lysates from MEFs treated as in j. (l) Quantification of aneuploid cells of primary MEFs co-infected with combinations of LKO-1-ns or LKO-1-Vhlh and retroviral vectors (Babe) expressing HA–Mad2 or an empty control vector (ctrl), processed at 5 days post infection (mean ± s.d., n = 3 independent experiments, P = 4 × 10−5 (ANOVA, for Tukey Post-Hoc analysis, see Supplementary Information, Tables S3 and S4). (m) Western blot of cell lysates from MEFs treated as in l.

Consistent with a failure of VHL-deficient cells to recruit normal amounts of Mad2 to kinetochores on spindle checkpoint activation, such cells are prone to mitotic slippage in that they fail to efficiently arrest in mitosis in response to nocodazole (Fig. 3f). Strikingly, VHL-negative 786-O cells, engineered to produce wild-type pVHL, expressed higher levels of Mad2 protein (Fig. 3g) and escaped a nocodazole-induced mitotic arrest less frequently than their VHL-negative counterparts, as shown by live-cell imaging (Fig. 3h; Supplementary Information, Videos 4 and 5). In the absence of nocodazole as a checkpoint response trigger, VHL-negative 786-O cells and their pVHL-producing counterparts did not show any major differences in anaphase timing (53 ± 36 min and 47 ± 25 min, respectively; Supplementary Information, Videos 6 and 7).

We hypothesized that chromosomal instability could arise due to a weakened spindle checkpoint on pVHL inactivation. We prepared chromosome spreads 5 days after primary MEFs were depleted for pVHL and determined the number of chromosomes per spread, scoring cells as diploid (2n = 40; see example given in Fig. 3i) or aneuploid (30 < 2n < 60, 2n ≠ 40). Pronounced whole-chromosome aneuploidy and reduced levels of Mad2 were observed after depletion of pVHL (Fig. 3j and k, respectively). Importantly, ectopic expression of HA–Mad2 suppressed the aneuploidy phenotype (Fig. 3l, m; Supplementary Information, Fig. S3d). No difference in the amount of tetraploid cells was observed under these conditions (Supplementary Information, Fig. S3c). These results suggest that loss of pVHL function provokes aneuploidy due to a weakened spindle checkpoint, at least in part, by negatively affecting Mad2 protein levels. Importantly, the spindle misorientation phenotype was not rescued in pVHL-depleted HeLa-H2B-Tub cells expressing HA–Mad2 (Supplementary Information, Fig. S3e, f). MEFs depleted for pVHL and expressing HA–Mad2 also showed a randomized spindle angle (data not shown). Collectively, these results indicate that the spindle misorientation and aneuploidy phenotypes associated with VHL loss reflect distinct functions of pVHL. In further support of this, type 2A VHL mutants, which failed to restore spindle-misorientation (Fig. 2g), suppressed mitotic slippage in VHL−/− 786-O cells (Supplementary Information, Fig. S3g). In contrast, type 2B VHL mutants, which restored spindle orientation (Fig. 2g), failed to suppress mitotic slippage (Supplementary Information, Fig. S3g). Mad2 protein levels were higher in those cells producing the type 2A mutants than in cells producing the corresponding type 2B mutants (Supplementary Information, Fig. S3h). We also noticed that 786-O cells expressing the pVHLY98H and pVHLY98N mutants showed increased Mad2 levels compared with pVHL wild-type or pVHLY112H and pVHLY112N mutants (Supplementary Information, Fig. S3h). The reason for this difference is unclear, but it is possible that mutations at residue Y98 of pVHL impair additional functions of pVHL that possibly relate to spindle checkpoint signalling.

Next, we asked whether reduced levels of pVHL would exacerbate the effect of treatments known to induce aneuploidy. Reduced levels of the kinetochore-associated kinesin CENP-E have been linked to aneuploidy that arises from the missegregation of one or a few chromosomes11. As shown in Fig. 4a, 62% of CENP-E- and pVHL-depleted MEFs were aneuploid (Vhlh-CENP-E). Downregulation of either CENP-E or pVHL alone (Fig. 4b) also promoted aneuploidy, but to a much lesser extent (Fig. 4a). Examination of the absolute number of chromosomes per cell revealed that most pVHL- or CENP-E-depleted cells were diploid and a subset was near-diploid (Fig. 4c), whereas in pVHL and CENP-E double-knockdown cells, the percentage of cells deviating from a diploid chromosome content increased dramatically (Fig. 4c). Thus, aneuploidy caused by reduced levels of CENP-E is further enhanced by combined inactivation of pVHL function.

Figure 4: Depletion of pVHL in cells with a weakened spindle checkpoint leads to enhanced aneuploidy due to chromosome segregation defects.
Figure 4

(a) Quantification of aneuploid cells from primary MEFs co-infected with combinations of two lentiviral vectors (LKO-1) expressing shRNA against: a non-silencing control (ns-ns), a non-silencing control and mouse VHL (ns-Vhlh), a non-silencing control and mouse CENP-E (ns-CENP-E), and mouse VHL and mouse CENP-E (Vhlh-CENP-E). MEFs were processed at 5 days post infection (mean ± s.d., n = 3 independent experiments, P = 1 × 10−6 (ANOVA, for Tukey Post-Hoc analysis see Supplementary Information, Tables S5 and S6). (b) Western blot of cell lysates from primary MEFs treated as in panel a. (c) Absolute chromosome numbers of primary MEFs treated and processed as in panel a. The bars with black (ns-Vhlh and ns-CENP-E) or red (ns-ns and Vhlh-CENP-E) borders represent the number of cells with a diploid amount of chromosomes (2n = 40). (d) Time from NEBD to onset of anaphase in HeLa-H2B-Tub cells transfected with the following siRNA: sictrl (n = 70 cells), siVHL (n = 66 cells), sictrl and siCENP-E (n = 38 cells) and siVHL and siCENP-E (n = 49 cells). (e) Quantification of cells described in d that enter anaphase with lagging chromosomes (mean ± s.d., n = 3 independent experiments, P = 2 × 10−6 (ANOVA, for Tukey Post-Hoc analysis see Supplementary Information, Tables S7 and S8). (f) Western blot of HeLa cell lysates treated as in d. (g, h) Frame series of movies acquired from HeLa-H2B-Tub cells transfected with sictrl or siCENP-E (g) or siVHL and siCENP-E (h), the panels show H2B–GFP (green) and RFP–tubulin (red). The NEBD was set as t = 0 min. Lagging chromosomes are shown with a yellow arrowhead, corresponding zooms are shown next to the frame. Scale bars, 2.5 μm.

To explore how this aneuploidy arises, we depleted HeLa-H2B-Tub cells of pVHL, CENP-E or both, using corresponding siRNAs, and monitored the onset of anaphase by live-cell imaging. Depletion of pVHL did not delay anaphase onset (t50%, anaphase onset = 50 min compared to t50%, anaphase onset = 46 min in control cells; Fig. 4d). In contrast, CENP-E downregulation caused spindle checkpoint dependent mitotic delay (t50%, anaphase onset = 210 min) (Fig. 4d)12. Of these cells 16% entered anaphase with lagging chromosomes (Fig. 4e). Consistent with the effect of loss of pVHL function in weakening the spindle-checkpoint, combined depletion of pVHL and CENP-E resulted in a partial rescue of the robust block to anaphase progression mediated by CENP-E depletion alone (t50%, anaphase onset = 76 min; Fig. 4d). As a result, an increased number of these cells (33%) entered anaphase with lagging chromosomes (Fig. 4e). Representative stills from movies taken at the indicated times after nuclear envelope breakdown (NEBD) illustrate these phenotypes (Fig. 4g, h; Supplementary Information, Videos 8 and 9 and Fig. S4a). Immunoblotting with anti-CENP-E and anti-pVHL antibodies showed that both proteins were efficiently downregulated (Fig. 4f). Although CENP-E knockdown caused some spindle misorientation, its combined depletion with pVHL did not further enhance the high rate of spindle misorientation observed in cells depleted for pVHL alone (compare Supplementary Information, Fig. S4b with Fig. 1e). Moreover, among those cells that entered anaphase with lagging chromosomes as a result of either CENP-E or CENP-E and pVHL depletion, there were similar fractions of cells with misoriented spindles (Supplementary Information, Fig. S4c). Therefore, spindle misorientation and sensitization of cells to chromosome missegregation are two distinct consequences of loss of pVHL function, and reduced levels of CENP-E and pVHL cooperate in the development of aneuploidy.

Partial loss of mitotic checkpoint control and the ensuing aneuploidy are also initiated after reduction of Mad2 levels. Mice heterozygous for this checkpoint protein are prone to tumour development and MEFs derived from Mad2+/− mice rapidly become aneuploid13. Thus, we asked whether depletion of pVHL function cooperates with Mad2 heterozygosity in the development of aneuploidy. Strikingly, we found a gradual increase in aneuploidy as Mad2 levels decreased as a result of pVHL depletion and/or deletion of one Mad2 allele (Fig. 5a, b), which was characterized by both chromosome losses and gains (Fig. 5c). Thus, as for CENP-E depletion, reduced levels of Mad2 cooperate with functional impairment of pVHL to generate elevated rates of chromosome missegregation events.

Figure 5: Mad2 is crucial to protect cells from aneuploidy.
Figure 5

(a) Quantification of aneuploid cells of primary Mad2+/fl MEFs co-infected with combinations of LKO-1-ns or LKO-1-Vhlh with (+) or without (−) adenovirus expressing Cre-recombinase (Adeno-Cre). MEFs were processed at 5 days post infection (mean ± s.d., n = 3 independent experiments, P = 2 × 10−6, ANOVA, for Tukey Post-Hoc analysis see Supplementary Information, Tables S9 and S10). (b) Western blot of cell lysates of Mad2+/fl MEFs treated as in panel a. (c) Absolute chromosome numbers of Mad2+/fl MEFs treated and processed as in a. Bars with a black (Vhlh and ns-Cre) or red (ns and Vhlh-Cre) borders represent the number of cells with a diploid amount of chromosomes (2n = 40). Cre expression in wild-type MEFs did not effect the rate of aneuploidy (data not shown). (d) Correlation of Mad2 staining (negative, positive) with VHL staining (negative, weak, strong) in ccRCC tumour specimen in a tissue microarray (n = 229, R = 0.2, P = 0.02, Kendall-β-tau test). (e) Pictures of two human ccRCC tumour specimens (spot 98 and spot 92) as described in d. Scale bars, 10 μm. (f) Correlation of Mad2 staining (negative, positive) with histological differentiation grade in ccRCC (n = 244; R = −0.2, P = 0.007, Kendall-β-tau test). (g) Number of ccRCC tumours (n = 69) with diploid or aneuploid DNA per tumour grade. Due to the low number of grade 1 tumours available (n = 4), grade 1 and 2 tumours were combined, R = 0.3, P = 0.03 (Kendall-β-tau Test).

To confirm the relevance of these findings in human ccRCC, Mad2 expression was immunohistochemically analysed in more than 200 ccRCC tissue samples and compared with pVHL expression and differentiation grade. pVHL-negative ccRCC samples showed low Mad2 expression, whereas high Mad2 expression was detected in ccRCC samples with strong pVHL expression (Fig. 5d, e), consistent with the finding that pVHL-depleted samples show reduced amounts of cellular Mad2. Progression of ccRCC is associated with histological de-differentiation from grade 1 to grade 3, combined with increasing frequencies of tumour aneuploidy. Less differentiated ccRCC are associated with a higher rate of aneuploid tumours as about two thirds of patients with grade 1 ccRCC have diploid tumours14. Interestingly, Mad2 expression was more frequent in grade 1 tumours (30%) than in grade 2 and grade 3 tumours (8–11%; Fig. 5f). When analysing a distinct subset of 69 ccRCC cases, we observed, consistent with previous reports15, increased aneuploidy with tumour grade (an increase from 40% in grade 1 and 2 tumours to 67% in grade 3 tumours; Fig. 5g). Hence, early ccRCC with diploid DNA content are characterized by higher Mad2 expression in accordance with the hypothesis that reduced amounts of Mad2 induce aneuploidy in ccRCC in vivo.

Here we identify pVHL as a component of the mitotic spindle and show that it suppresses spindle misorientation and promotes chromosomal stability. These newly identified mitotic functions of pVHL appear to be distinctly disrupted in naturally occurring pVHL mutants and are linked to microtubule-dependent and microtubule-independent pVHL activities. Spindle misorientation has been implicated in kidney cyst formation16, a hallmark of VHL disease, and as, at least in some instances, a precursor of ccRCC17; thus the promotion of correct spindle orientation may constitute a new tumour suppressive mechanism of pVHL. The development of whole-chromosome aneuploidy appears to be a consequence of reduced amounts of Mad2 protein levels actuated by pVHL loss. Human ccRCC cells that commonly display biallelic VHL inactivation are characterized by low expression of Mad2. Kidneys from patients with VHL disease contain many thousands of VHL−/− pre-neoplastic lesions, only a few of which are destined to form tumours18, reflecting the need for additional mutation(s) for tumour development. Chromosomal instability induced by VHL inactivation, possibly in combination with secondary mutations that further compromise the mitotic spindle checkpoint or decrease chromosome segregation efficiency, may thus be a critical tumour-promoting mechanism.


Cell culture and RNA interference experiments.

HeLa, HeLa-H2B-Tub, 786-O and primary MEF cells were maintained in high-glucose DMEM supplemented with 10% FCS. β-mercaptoethanol (200 μM) was added to the medium for primary MEFs. For RNA interference, HeLa or HeLa-H2B-Tub cells were plated 24 h before transfection with oligofectamine (Invitrogen) performed according to the manufacturer's instructions. The following antisense oligonucleotides were used: control (AllStars negative control siRNA, Qiagen #1027281), siVHL (5′-ACACAGGAGCGCATTGCACAT-3′), siMad2 (5′-AAGAGTCGGGACCACAGTTTA-3′) and siCENP-E (5′-ACTCTTACTGCTCTCCAGT-3′). Between 24–48 h after transfection cells were further processed. VHL retroviral vectors were generated by digestion of corresponding pcDNA3-HA–VHL mutants with BamHI and XhoI, which has been described previously7, and subsequent ligation into linearized pCMV(R)-neo19 with EcoRV. HA–Mad2 retroviral vectors were generated by digestion of pBabepuro-HA–Mad2, which has been described previously20, with NaeI and EcoRI and subsequent ligation into linearized pBabehygro with BsaAI and EcoRI. All generated vectors were verified by sequencing. To generate retroviral pools, the packaging cell lines, PT-67 (Clontech) or LinX (Openbiosystems), were transfected with pCMV(R) or pBabe constructs as described previously2. Knockdown experiments with lentiviral vectors expressing short hairpin RNA (LKO-1-shRNA) in MEFs has been described previously2. The following clones were used: LKO-1-ns (clone 10879; Addgene), LKO-1-Vhlh (clone TRC0000009735; Open Biosystems) and LKO-1-CENP-E (clone TRC0000089927, Sigma-Aldrich). Knockout of floxed alleles in MEFs with high-titer adenovirus expressing Cre recombinase was performed as described, in a wild-type, Vhlhfl/fl21 and Mad2+/fl (unpublished data; provided by P. Sorger, Harvard Medical School, MA, USA) background. For determination of the mitotic index and checkpoint protein signal intensity, HeLa cells were treated for 16 h with nocodazole (200 ng ml−1). For angle and astral microtubule intensity measurements, cells were plated on poly-lysine (Sigma-Aldrich) coated plates. Treatment with taxol (1 μM) was performed 24 h later for 20 min before processing. Treatment of HeLa cells with 26S proteasome inhibitor MG132 (25 μM; Sigma-Aldrich) was performed 48 h after siRNA treatment for 6 h, before processing. For cycloheximide chase experiments, Hela cells were supplemented with cycloheximide (100 μg ml−1). After different incubation times (1.5, 3 and 6 h), cells were washed three times with ice-cold PBS containing cycloheximide (100 μg ml−1) and then lysed in lysis buffer (50 mM Tris at pH 8.0, 150 mM NaCl, 1% IgE PAL, 10 mM MgCl2 and 100 μg ml−1 cycloheximide).

Immunofluorescence microscopy and movie acquisition.

Processing and staining of MEFs for VHL localization and control experiments have been described previously2. For tubulin staining, a mouse anti-α-tubulin antibody was used (1:2,000 dilution; Sigma-Aldrich). For determination of spindle angles, HeLa, 786-O and MEF cells were fixed for 3 min with cold methanol at -20 °C and processed as described previously2. The following antibodies were used: mouse anti-α-tubulin antibody (1:2,000 dilution; Sigma-Aldrich) and rabbit anti-γ-tubulin antibody (1:1,000 dilution; Sigma-Aldrich). For astral microtubule intensity measurements, cells were pre-extracted with PHEM buffer (60 mM Pipes, 25 mM Hepes, 10 mM EGTA, 4 mM MgSO4, 0.5% Triton X-100 and 5 μM fresh taxol) for 1 min, and then fixed for 5 min with cold methanol at -20 °C before staining with an anti-α−tubulin antibody (1:500 dilution; Abcam). For glutaraldehyde (GA) fixation Hela cells were pre-extracted with cytoskeleton buffer at pH 6.1 (10 mM MES, 150 mM NaCl, 5 mM EGTA, 5 mM MgCl2, 5mM glucose; 0.25% Triton X-100 and 0.5% fresh GA) for 1 min then fixed with 1% GA in cytoskeleton buffer for 10 min before staining with an anti-α−tubulin antibody (1:1,000 dilution; Abcam). Recruitment and quantification of the spindle check point proteins (Mad1, Mad2 and BubR1) at kinetochores in HeLa cells was performed as described previously2,22. Immunofluorescence microscopy images were obtained using an Olympus IX70 Delta Vision Spectris deconvolution microscope (Applied Precision). Obtained stacks were deconvolved using SoftWoRx 3.3.4 (Applied Precision) and processed with Imaris (Bitplane AG). Relative astral microtubule signal intensities were measured using ImageJ software (MacBiophotonics) after background subtraction. Live-cell imaging of HeLa-H2B-Tub cells and analysis of acquired movies has been described previously22. Phase-contrast live-cell imaging of 786-O cells, either arrested with nocodazole (50 ng ml−1) or untreated, was performed with an Olympus IX 81 microscope using a 20× UPlanSApo objective NA 0.75. Images were acquired every 5 min for 14–16 h.

Western blotting.

Western blotting was performed as described previously2. The following antibodies were used: anti-mouse-pVHLCT (1:500 dilution)2, anti-human-pVHLNT (1:500 dilution)7, anti-human-pVHLCT (1:500 dilution)7, anti-Mad1 (1:1000 dilution)23, anti-Mad2 (1:5,000 dilution for HeLa, 1:2,500 dilution for 786-O and 1:500 dilution for MEF cell lysates; Bethyl laboratories), anti-BubR1 (1:1,000 dilution; BD biosciences), anti-CENP-E (1:2,000 dilution)23, anti-p27 (1:2,500 dilution, BD biosciences), anti-HA (1:1,000 dilution, Babco), anti-Cre (1:2,500, Novagen), anti-α−tubulin (dilution 1:500; homemade rat hybridoma clone YL1/2. Dilution 1:5,000; Sigma-Aldrich, St. Louis) and anti-CDK2 (1:1,000 dilution; Santa Cruz Inc.).

Microtubule polymerization assays.

For expression of MBP, MBP-VHL30 and MBP-VHL30,Δ95–123 in E. coli, fragments were subcloned by digestion of corresponding pcDNA3-HA–VHL vectors, which have been described previously7, using BamHI/XhoI fragments and ligated into pMAL-C2 linearized with BamHI/SalI. MBP, MBP–VHL30 and MBP–VHL30,Δ95–123 were purified using amylose resin according to the manufacturer's instructions (New England Biolabs). The purified proteins were cleaved from MBP with factor Xa (New England Biolabs) before inactivation of factor Xa with dansyl-glu-gly-arg-chloromethyl ketone (Merck). The cleaved proteins were dialysed against BRB80 (80 mM PIPES, 1 mM EGTA and 1 mM MgCl2, at pH 6.8). Tubulin polymerization was performed in 96-well half area plates. A reaction mix was prepared containing 1.5 mg ml−1 MAP-rich tubulin (#ML113, Cytoskeleton) and 0.5 mM purified MBP, pVHL30 or pVHL30,Δ95–123 in PEM buffer (80 mM PIPES, 1 mM EGTA and 0.5 mM MgCl2, at pH 6.8). To induce polymerization, GTP was added to a final concentration of 1 mM and the optical density was monitored every 10 s at a wavelength of 320 nm at 32 °C over 30 min with a SpectraMax190 photometer (Molecular Devices). For microtubule visualization, polymerization was performed for 10 min as described above using MAP-rich tubulin spiked with 5% rhodamine–tubulin (#TL331M, Cytoskeleton). The polymerized microtubules were fixed and further processed as described previously24. Immunofluorescence microscopy images were obtained using an Olympus IX70 Delta Vision Spectris deconvolution microscope (Applied Precision).

Transcription analyses.

Total RNA was prepared from HeLa cells using the NucleoSpin RNA II kit (Macherey Nagel) and cDNA prepared using random hexamer primers (Fermentas) and Ready-to-go you-prime first-strand beads (GE Healthcare). Real time PCR analysis of cDNA was conducted using LightCycler 480 SYBR Green I Master (Roche Diagnostics) and the following primer pairs: VHL (5′-ATGGCTCAACTTCGACGGC-3′, 5′-CCAGAAGCCCATCGTGTGTC-3′), Mad1 (5′-TGGACTGGATATTTCTACCTCGG-3′, 5′-CCTCTTGTGACTCAGCTCCATC-3′), Mad2(a) (5′-GGTCCTGGAAAGATGGCAG-3′, 5′-ATCACTGAACGGATTTCATCC-3′), Mad2(b) (5′-GATGACAGTGCACCCAGAGA-3′, 5′-CCGACTCTTCCCATTTTTCA-3′), BubR1 (5′-CTGGACAGAGCAGAACTATCCT-3′, 5′-ACGCCCTAATTTAAGCCAGAGAT-3′) and 18S rRNA (5′-TGGCCGACCATAAACGATGCC-3′, 5′-TGGTGGTGCCCTTCCGTCAAT-3′).

Chromosome spreads.

MEFs at passage 2 were infected with combinations of lentiviral, retroviral and adenoviral vectors expressing short hairpin RNA (LKO-1-shRNA), HA–Mad2 or Cre-recombinase, respectively as described previously2,21. Cells were plated 4 days after infection and whole cell protein extracts were prepared 24 h later and processed for western blotting. Cells were plated at the same times and processed for chromosome spreading as described previously11, except that cells were washed four times with fixative for 10 min before spreading.


The tumours included in the renal cancer tissue microarrays have been described previously25. Two consecutive sections were incubated at room temperature one with an anti-Mad2 (dilution 1:25, BD biosciences) antibody and the other with an anti-human-pVHLCT (dilution 1:50)7 antibody. Immunostaining was performed as described. A tumour was considered Mad2-positive if an unequivocal nuclear positivity was identified. Tumours were considered VHL-positive if weak or strong cytoplasmic staining was found.

Flow cytometry.

Sixty nine ccRCC samples26 were histologically reviewed by one pathologist (H.M.) and selected for the study on the basis of hematoxylin and eosin-stained tissue sections. Tumours were graded according to the 3-tiered Thoenes grading system and histologically classified according to the World Health Organization classification. Flow cytometry was performed for DNA-measurements on 50 μm sections of formalin-fixed, paraffin-embedded tumour tissue as described previously27. Dissociated nuclei were analysed with a PAS II flow cytometer (Partec). For statistical analysis, diploid and peri-diploid tumours were grouped together and compared with aneuploid ccRCC.

Accession codes.

USCD-Nature Signaling Gateway (http://www.signaling-gateway.org): VHL, A002833.

Note: Supplementary Information is available on the Nature Cell Biology website.


  1. 1.

    et al. pVHL and PTEN tumour suppressor proteins cooperatively suppress kidney cyst formation. EMBO J. 27, 1747–1757 (2008).

  2. 2.

    et al. pVHL and GSK3beta are components of a primary cilium-maintenance signalling network. Nature Cell Biol. 9, 588–595 (2007).

  3. 3.

    , , , & The human kinetochore proteins Nnf1R and Mcm21R are required for accurate chromosome segregation. EMBO J. 25, 4033–4049 (2006).

  4. 4.

    , , , & PtdIns(3, 4, 5)P3 regulates spindle orientation in adherent cells. Dev. Cell 13, 796–811 (2007).

  5. 5.

    & Integrin-mediated adhesion orients the spindle parallel to the substratum in an EB1- and myosin X-dependent manner. EMBO J. 26, 1487–1498 (2007).

  6. 6.

    et al. The von Hippel-Lindau tumour suppressor interacts with microtubules through kinesin-2. FEBS Lett. 581, 4571–4576 (2007).

  7. 7.

    , , , & Regulation of microtubule stability by the von Hippel-Lindau tumour suppressor protein pVHL. Nature Cell Biol. 5, 64–70 (2003).

  8. 8.

    et al. Hypoxia-inducible factor linked to differential kidney cancer risk seen with type 2A and type 2B VHL mutations. Mol. Cell Biol. 27, 5381–5392 (2007).

  9. 9.

    , Cole, R. W., & The checkpoint delaying anaphase in response to chromosome monoorientation is mediated by an inhibitory signal produced by unattached kinetochores. J. Cell Biol. 130, 941–948 (1995).

  10. 10.

    & The spindle-assembly checkpoint in space and time. Nature Rev. Mol. Cell Biol. 8, 379–393 (2007).

  11. 11.

    , , , & Aneuploidy acts both oncogenically and as a tumor suppressor. Cancer Cell 11, 25–36 (2007).

  12. 12.

    et al. Gene silencing of CENP-E by small interfering RNA in HeLa cells leads to missegregation of chromosomes after a mitotic delay. Mol. Biol. Cell 15, 3771–3781 (2004).

  13. 13.

    et al. MAD2 haplo-insufficiency causes premature anaphase and chromosome instability in mammalian cells. Nature 409, 355–359 (2001).

  14. 14.

    , , & DNA ploidy is a valuable predictor for prognosis of patients with resected renal cell carcinoma. Cancer 92, 2280–2285 (2001).

  15. 15.

    et al. Nuclear DNA content-derived parameters correlated with heterogeneous expression of p53 and bcl-2 proteins in clear cell renal carcinomas. Cancer 89, 1065–1075 (2000).

  16. 16.

    et al. Defective planar cell polarity in polycystic kidney disease. Nature Genet. 38, 21–23 (2006).

  17. 17.

    & Renal pathology in von Hippel-Lindau disease. Hum. Pathol. 19, 1072–1079 (1988).

  18. 18.

    et al. HIF activation identifies early lesions in VHL kidneys: evidence for site-specific tumor suppressor function in the nephron. Cancer Cell 1, 459–468 (2002).

  19. 19.

    & Mutations of p34cdc2 phosphorylation sites induce premature mitotic events in HeLa cells: evidence for a double block to p34cdc2 kinase activation in vertebrates. EMBO J. 10, 3331–3341 (1991).

  20. 20.

    et al. The Mad1/Mad2 complex as a template for Mad2 activation in the spindle assembly checkpoint. Curr. Biol. 15, 214–225 (2005).

  21. 21.

    et al. Combined VHLH and PTEN mutation causes genital tract cystadenoma and squamous metaplasia. Mol. Cell Biol. 28, 4536–4548 (2008).

  22. 22.

    et al. The CENP-A NAC/CAD kinetochore complex controls chromosome congression and spindle bipolarity. EMBO J. 26, 5033–5047 (2007).

  23. 23.

    , & Timing and checkpoints in the regulation of mitotic progression. Dev. Cell 7, 45–60 (2004).

  24. 24.

    et al. Human spastin has multiple microtubule-related functions. J. Neurochem. 95, 1411–1420 (2005).

  25. 25.

    et al. Association of cytokeratin 7 and 19 expression with genomic stability and favorable prognosis in clear cell renal cell cancer. Int. J. Cancer 123, 569–576 (2008).

  26. 26.

    et al. Chemokine receptor CXCR4 downregulated by von Hippel-Lindau tumour suppressor pVHL. Nature 425, 307–311 (2003).

  27. 27.

    et al. DNA aneuploidy, S-phase fraction, nuclear p53 positivity, and survival in non-small-cell lung carcinoma. Virchows Arch. 431, 173–179 (1997).

Download references


We thank all members of our laboratories for helpful discussions. The authors are grateful to P. Sorger and A. Musacchio for their generous gifts of reagents, R. Carazo Salas and M. Bortfeld-Miller for helping with microtubule assays, C. Azzalin and M. Patil for helping with chromosome spreads, the Light Microscopy Centre ETH Zurich, in particular G. Csucs and J. Kusch, for help with microscopy, M. Storz and S. Behnke for preparing the tissue array and the immunostaining and J. Schelldorfer for advise in statistical data analysis. C. R. T. and A. T. are members of the Life Science Zurich Graduate School Zurich, Program in Molecular Life Sciences. P. M. is recipient of a SNF Assistant Professorship and is supported by the Swiss National Science Foundation. W. K. is supported by a grant from the Swiss National Science Foundation.

Author information

Author notes

    • Alexander Hergovich

    Current address: Friedrich Miescher Institute for Biomedical Research, 4002 Basel, Switzerland.

    • Claudio R. Thoma
    •  & Alberto Toso

    These authors contributed equally to this work.


  1. Institute of Cell Biology, ETH Zurich, 8093 Zurich, Switzerland.

    • Claudio R. Thoma
    • , Katrin L. Gutbrodt
    • , Ian J. Frew
    • , Alexander Hergovich
    •  & Wilhelm Krek
  2. Institute of Biochemistry, ETH Zurich, 8093 Zurich, Switzerland.

    • Alberto Toso
    • , Sabina P. Reggi
    •  & Patrick Meraldi
  3. Institute for Surgical Pathology, University Hospital Zurich, 8091 Zurich, Switzerland.

    • Peter Schraml
    •  & Holger Moch


  1. Search for Claudio R. Thoma in:

  2. Search for Alberto Toso in:

  3. Search for Katrin L. Gutbrodt in:

  4. Search for Sabina P. Reggi in:

  5. Search for Ian J. Frew in:

  6. Search for Peter Schraml in:

  7. Search for Alexander Hergovich in:

  8. Search for Holger Moch in:

  9. Search for Patrick Meraldi in:

  10. Search for Wilhelm Krek in:


C.R.T., A.T., P.M. and W.K. concieved and planned the project and performed data analyses; K.L.G. contributed to microtubule polymerization measurements; S.P.R. contributed to analysis of mitotic slippage in renal cancer cells; I.J.F. generated biochemical data in VHL-deficient MEFs; P.S. and H.M. provided data on human renal cancer; A.H. observed mitotic localization of pVHL and the manuscript was prepared by C.R.T. and W.K. and edited by A.T. and P.M.

Competing interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to Patrick Meraldi or Wilhelm Krek.

Supplementary information

PDF files

  1. 1.

    Supplementary Information

    Supplementary Information


  1. 1.

    Supplementary Information

    Supplementary Video S1

  2. 2.

    Supplementary Information

    Supplementary Video S2

  3. 3.

    Supplementary Information

    Supplementary Video S3

  4. 4.

    Supplementary Information

    Supplementary Video S4

  5. 5.

    Supplementary Information

    Supplementary Video S5

  6. 6.

    Supplementary Information

    Supplementary Video S6

  7. 7.

    Supplementary Information

    Supplementary Video S7

  8. 8.

    Supplementary Information

    Supplementary Video S8

  9. 9.

    Supplementary Information

    Supplementary Video S9

About this article

Publication history






Further reading