BAP1 protein is required for BRCA1 expression in mesothelioma cells and tumours
BRCA1 forms a complex with BARD1 to generate an active E3 ubiquitin ligase that can auto-ubiquitinate to promote its own destruction, while BAP1 is a DUB that by deubiquitinating BARD1 can disrupt this complex and reduce its turnover 26, 46. Before setting out to explore how loss of BAP1 might regulate mitotic events in mesothelioma, we wished to examine the interdependence of BAP1 and BRCA1 protein expression in mesothelioma cells by Western blot. For this purpose, we used MSTO-211H cells that have wild-type BAP1 and NCI-H2452 cells that have an A95D mutation in the UCH domain of BAP1 that renders it catalytically inactive as a DUB 67, 75. Using two independent siRNAs, we found that BAP1 depletion led to a significant reduction in BRCA1 protein in the MSTO-211H and NCI-H2452 mesothelioma cell lines (Fig. 1A). However, there was no change in BAP1 protein expression following either siRNA-mediated depletion of BRCA1 or inducible expression of an shRNA against BRCA1 in MSTO-211H cells (Fig. 1A).
To determine whether this loss of BRCA1 in BAP1-depleted cells was a consequence of altered protein stability, we examined BRCA1 protein expression following BAP1 depletion in the presence of the proteasome inhibitor, MG132. Following depletion of BAP1 for 72 hours, treatment with MG132 for 6 hours led to a substantial rescue of BRCA1 expression in MSTO-211H and NCI-H2452 cells indicating that BAP1 inhibits the proteasome-mediated degradation of BRCA1 (Fig. 1B). Interestingly, as BAP1 is catalytically inactive as a DUB in NCI-H2452 cells, the loss of BRCA1 from these cells upon BAP1 depletion and rescue by MG132 argues that BAP1 controls the ubiquitylation state of BRCA1 in a manner that is at independent of its DUB activity.
We next examined whether the frequent loss of BAP1 in mesothelioma tumours in patients correlates with loss of BRCA1 protein expression. Previous immunohistochemical analyses for BAP1 protein have demonstrated that patient-associated deletions, insertions or point mutations all lead to loss of nuclear staining, even though cytoplasmic staining is present in some patients 45, 53, Here, we evaluated the expression of BAP1 and BRCA1 protein in tumour samples taken from chemonaive mesothelioma patients undergoing extended pleurectomy decortication (n=26) using immunohistochemistry (Fig. 1C). We identified a positive and significant correlation between BAP1 and BRCA1 protein expression (Chi-square test for association, p<0.05; Fig. 1D).
We then analysed BAP1 and BRCA1 protein expression in three primary mesothelioma cell lines (Fig. 1E). MEDCL96 is wild-type for BAP1, whereas MEDCL85 has a mutation in the HBM binding domain (R417fs) that regulates interaction with host cell factor-1 (HCF-1). MEDCL91 has a frame-shift mutation in the BRCA1 binding domain (E631fs). None of these primary cell lines have a BRCA1 mutation. The two cell lines with BAP1 mutations showed no protein expression of BAP1 and had low or no protein expression of BRCA1 compared to MEDCL96 cells where BAP1 and BRCA1 were clearly detected. Collectively, these data are consistent with the hypothesis that loss of BAP1 may impact BRCA1 expression in vivo.
BAP1 and BRCA1 are required for accurate mitotic progression in mesothelioma cells
To begin to examine the impact of BAP1 loss on mitotic progression, we first used flow cytometry to assess cell cycle distribution following BAP1 or BRCA1 depletion. Although some studies have reported that depletion of BAP1 or BRCA1 inhibits S-phase progression 33, 46, we observed no significant difference by flow cytometry in cell cycle distribution or percentage of cells in G2/M upon BAP1 or BRCA1 depletion in MSTO-211H or NCI-H2452 cells (Fig. S1A, B). These results are consistent with no observed change in cell cycle profile in skin fibroblasts taken from BAP1 heterozygous individuals that had reduced BAP1 protein levels 6, although different responses in other cell types cannot be excluded. We then performed fluorescence microscopy on cells fixed and stained for DNA to analyse the relative proportion of cells in different stages of mitosis. This revealed that depletion of either BAP1 or BRCA1 resulted in an increase in the proportion of mitotic cells in late mitosis, i.e. anaphase and telophase, compared to mock-depleted cells (Fig. 2A). However, while BRCA1 led to most mitotic cells being present in telophase, BAP1 depletion led to the highest proportion of mitotic cells being in anaphase. This suggests that while BAP1 and BRCA1 are both important for normal mitotic progression of mesothelioma cells, they are likely to have at least in part distinct functions.
Further analysis of these mitotic cells by immunofluorescence microscopy revealed that depletion of either BAP1 or BRCA1 led to chromosome congression errors with an increase in the frequency of misaligned chromosomes in metaphase (Fig.2B, and Fig. S2A). The frequency of misaligned chromosomes was similar whether cells were depleted of BAP1 or BRCA1. Depletion of either BAP1 or BRCA1 also led to a substantial increase in the number of interphase cells that contained multiple nuclei or micronuclei (Fig. 2C, D, and Fig. S2B). Staining with the centromere marker, CENP-A, revealed the presence of centromeres within these micronuclei indicating that they were likely the result of chromosome segregation errors rather than fragments of broken chromosomes arising due to DNA repair defects 49, 74.
The increased proportion of mitotic cells in late mitosis and the presence of indicators of chromosome segregation errors in interphase cells is consistent with cells slipping through the SAC before full chromosome congression. Indeed, loss of SAC integrity is known to lead to chromosome segregation errors during unperturbed mitotic progression, as well as during drug-induced mitotic arrest 36, 44. Since previous data had shown that BRCA1 controls SAC activity through regulating expression of MAD2L1, a key SAC component 69, we examined the protein levels of MAD2L1 in cells depleted of BAP1. In comparison to mock-depleted cells, there was a marked reduction in MAD2L1 protein upon depletion of either BAP1 or BRCA1 in MSTO-211H cells, or depletion of BAP1 in NCI-H2452 cells (Fig. 2E). We then analysed the expression and localization of another SAC component, BUBR1, in mitotic cells. Although there was no difference in expression of BUBR1 protein as determined by Western blotting upon BAP1 or BRCA1 depletion in MSTO-211H cells (Fig. S2C), there was reduced association with kinetochores upon depletion of either BAP1 or BRCA1 in MSTO-211H and NCI-H2542 cells arrested in prometaphase with vinorelbine (Fig. 2F, and Fig. S2D). These data indicate that BAP1 is likely to regulate the SAC in a similar manner to BRCA1 through control of MAD2L1 expression and recruitment of BUBR1 to kinetochores.
BAP1 and BRCA1 are required to maintain spindle bipolarity and normal centrosome number
The errors observed in chromosome congression in metaphase cells depleted of BAP1 or BRCA1 suggest defects in spindle organization or chromosome attachment. Examination of mitotic cells by immunofluorescence microscopy with antibodies against a-tubulin and g-tubulin revealed an increased frequency of multipolar spindles following either BAP1 or BRCA1 depletion in MSTO-211H and NCI-H2452 cells (Fig. 3A, B). The frequency of multipolar spindles in MSTO-211H cells was similar following BAP1 or BRCA1 depletion suggesting that BAP1 regulation of spindle organization may be at least in part through control of BRCA1 expression.
As multipolar spindles may result from the presence of amplified centrosomes, we next examined centrosome number in interphase cells. For this purpose, immunofluorescence microscopy analysis was performed with antibodies against g-tubulin, a marker of pericentriolar material, and C-Nap1, a marker of centrioles. Cells were scored as possessing amplified centrosomes if they had >2 centrosomes. This revealed that depletion of either BAP1 or BRCA1 led to a similar, approximate ten-fold, increase in the percentage of cells with amplified centrosomes in MSTO-211H cells (Fig. 3C, D). These results are consistent with previous studies demonstrating a role for BRCA1 in preventing centrosome amplification 59, 73. Depletion of BAP1 in NCI-H2452 cells also led to an increase in the proportion of cells with amplified centrosomes albeit with a lower fold-increase; however, parental NCI-H2452 cells have a higher basal level of centrosome amplification than MSTO-211H cells. The lower frequency of multipolar spindles in mitosis compared to amplified centrosomes in interphase in response to BAP1 or BRCA1 depletion may be explained by the ability of cancer cells to cluster amplified centrosomes to form pseudo-bipolar spindles. Nonetheless, amplified centrosomes are associated with increased levels of chromosome congression and segregation defects due to an increased frequency of merotelic kinetochore attachments 24, 25.
BAP1 loss leads to reduced centrosome size in interphase and mitosis
How BAP1 or BRCA1 controls centrosome number remains unclear. It is plausible that amplified centrosomes arise as a result of cells exiting mitosis without completing cytokinesis, in a similar manner to how multinucleated cells can be generated. Indeed, the percentage of interphase cells that were multinucleated and have amplified centrosomes upon BAP1 or BRCA1 depletion was similar. However, in complex with BARD1, BRCA1 has been reported to mono-ubiquitinate g-tubulin, which in turn is required for centrosome duplication raising the possibility that BRCA1 controls centrosome number via a more direct mechanism involving g-tubulin modification 23, 29, 59. Indeed, BRCA1 localises to centrosomes throughout the cell cycle and BAP1 co-localizes with g-tubulin at centrosomes in mitosis 8, 32, 65, 78. Furthermore, BRCA1 loss has been associated with accumulation of g-tubulin at centrosomes 57, 64. To examine whether BAP1 loss leads to a similar increase in g-tubulin at centrosomes, we performed immunofluorescence microscopy using antibodies against g-tubulin, as well as CDK5RAP2, another component of the centrosome involved in microtubule nucleation 4, 21. Consistent with these previous results, depletion of BRCA1 led to an increase in centrosome size in interphase MSTO-211H and NCI-H2452 cells as measured not only with g-tubulin but also CDK5RAP2. However, in contrast, depletion of BAP1 led to a significant decrease in centrosome size during interphase as determined with both centrosome markers (Fig. 4A, B).
At the onset of mitosis, centrosome maturation leads to an increase in centrosome size and enhanced microtubule nucleation. This occurs through recruitment of additional g-tubulin and CDK5RAP2, as well as other centrosome proteins. We therefore analysed centrosome size in mitotic cells depleted of BRCA1 and BAP1 by immunofluorescence microscopy with antibodies against g-tubulin and CDK5RAP2. Following BRCA1 depletion, there was no discernible change in g-tubulin content in mitotic MSTO-211H cells as compared to mock-depleted mitotic cells; however, there was an unexpected increase in CDK5RAP2 upon BRCA1 depletion in mitotic MSTO-211H cells (Fig. 4C, D). In contrast, BAP1 depletion led to a significant reduction in the volume of both g-tubulin and CDK5RAP2 at centrosomes in mitotic MSTO-211H and NCI-H2452 cells (Fig. 4C, D). These data reveal a striking difference in the response of cells to depletion of BRCA1 or BAP1. In turn, this raises the prospect that loss of BAP1 could have additional consequences on mitotic progression through mechanisms above and beyond the control of BRCA1 expression.
BAP1 but not BRCA1 regulates spindle length and spindle pole attachment
Further examination of mitotic cells by immunofluorescence microscopy with g-tubulin and a-tubulin antibodies revealed that bipolar spindles were less round and more elliptical in shape following depletion of BAP1 as compared to mock- or BRCA1-depleted cells (Fig. 5A). Determination of spindle length by measuring pole-to-pole distance revealed that spindles were on average 20-50% longer following depletion of BAP1 in MSTO-211H cells (depending on the siRNA sequence used) and 50% longer following depletion of BAP1 in NCI-H2452 cells (Fig. 5B). In contrast, depletion of BRCA1 in MSTO-211H cells did not affect spindle length (Fig. 5A, B). Intriguingly, following BAP1 depletion, a small number of mitotic MSTO-211H and NCI-H2452 cells were observed to have spindle poles that were clearly detached from the bulk of the spindle microtubules (Fig. 5C). This defect was never observed in mock-depleted or BRCA1-depleted cells. Astral microtubules also appeared more abundant and longer in the BAP1-depleted MSTO-211H and NCI-H2452 cells than in mock-depleted or BRCA1-depleted cells (see Fig. 5A, C). As an alternative approach to quantify the length of microtubules in mitosis, depleted cells were treated with STLC, an Eg5 inhibitor, to arrest cells in prometaphase with a monopolar spindle. Measurement of microtubule length in the monopolar spindle asters revealed a ~20% and ~80% increase in the length of microtubules upon BAP1 depletion in MSTO-211H and NCI-H2452 cells, respectively, but no change upon depletion of BRCA1 in MSTO-211H cells (Fig. 5D, E). Taken together, these results suggest that BAP1 contributes to regulation of microtubule length and spindle pole attachment during mitosis in mesothelioma cells in a manner that is independent of BRCA1.
BAP1 controls microtubule length in mitosis through regulating KIF18 expression
The temporal and spatial control of spindle microtubule organization is critical to accurate chromosome segregation in mitosis 30. Interestingly, the alterations in spindle shape and microtubule length observed in response to BAP1 depletion were highly reminiscent of phenotypes previously described to result from depletion of the mitotic kinesins, KIF18A and KIF18B, that possess microtubule destabilization activity 71. KIF18A localizes to the plus-ends of kinetochore microtubules to control spindle length and chromosome positioning, while KIF18B plays a more specific role in regulating astral microtubule length 60-62. To determine whether the elongated spindles observed after BAP1 depletion are due to perturbation of KIF18 proteins, we first analysed the localization of KIF18A on spindle microtubules by immunofluorescence microscopy in MSTO-211H and NCI-H2452 cells. This revealed substantial loss of KIF18A from kinetochore microtubules after BAP1 depletion (Fig. 6A, B). Strikingly, overexpression of GFP-KIF18A, which localised to kinetochore microtubules, in BAP1-depleted MSTO-211H cells restored spindle length to that seen in mock-depleted cells (Fig. 6C, D). BRCA1 depletion was not examined in this regard as no spindle length alteration was observed.
Previous work had shown that KIF18B, but not KIF18A, regulates the length of astral microtubules 42, 68. In this case, attempts to detect endogenous KIF18B by immunofluorescence microscopy with several different commercial antibodies proved unsuccessful. Strikingly though, the ~5-fold increase in astral microtubule volume resulting from BAP1 depletion was reduced in cells transfected with GFP-KIF18B to the level observed in mock-depleted cells (Fig. 6E, and Fig. S3A). In contrast, the increased spindle length seen upon depletion of BAP1 was not rescued by overexpression of KIF18B (Fig. S3B). As these defects in spindle and astral microtubule length mediated by BAP1 depletion could be rescued by overexpression of KIF18A and KIF18B, respectively, we examined the expression of these kinesins in BAP1-depleted cells by Western blot. This revealed a significant reduction in both KIF18A and KIF18B proteins in MSTO-211H and NCI-H2452 cells following BAP1 depletion as compared to mock-depleted cells (Fig. 6F). We also found by Western blot that expression of KIF18A was increased in NCI-H226 cells stably expressing wild-type BAP1 compared to the parental cells that are null for BAP1 (Fig. S3C). KIF18B is mutant in NCI-H226 cells and so was not analysed 54. Taken together, these data provide persuasive evidence that BAP1 regulates mitotic spindle organization through control of KIF18A and KIF18B protein expression.