|Home | About | Journals | Submit | Contact Us | Français|
Traditional antimitotic drugs for cancer chemotherapy often have undesired toxicities to healthy tissues, limiting their clinical application. Developing novel agents that specifically target tumor cell mitosis is needed to minimize the toxicity and improve the efficacy of this class of anticancer drugs. We discovered that mdivi‐1 (mitochondrial division inhibitor‐1), which was originally reported as an inhibitor of mitochondrial fission protein Drp1, specifically disrupts M phase cell cycle progression only in human tumor cells, but not in non‐transformed fibroblasts or epithelial cells. The antimitotic effect of mdivi‐1 is Drp1 independent, as mdivi‐1 induces M phase abnormalities in both Drp1 wild‐type and Drp1 knockout SV40‐immortalized/transformed MEF cells. We also identified that the tumor transformation process required for the antimitotic effect of mdivi‐1 is downstream of SV40 large T and small t antigens, but not hTERT‐mediated immortalization. Mdivi‐1 induces multipolar mitotic spindles in tumor cells regardless of their centrosome numbers. Acentrosomal spindle poles, which do not contain the bona‐fide centrosome components γ‐tubulin and centrin‐2, were found to contribute to the spindle multipolarity induced by mdivi‐1. Gene expression profiling revealed that the genes involved in oocyte meiosis and assembly of acentrosomal microtubules are highly expressed in tumor cells. We further identified that tumor cells have enhanced activity in the nucleation and assembly of acentrosomal kinetochore‐attaching microtubules. Mdivi‐1 inhibited the integration of acentrosomal microtubule‐organizing centers into centrosomal asters, resulting in the development of acentrosomal mitotic spindles preferentially in tumor cells. The formation of multipolar acentrosomal spindles leads to gross genome instability and Bax/Bak‐dependent apoptosis. Taken together, our studies indicate that inducing multipolar spindles composing of acentrosomal poles in mitosis could achieve tumor‐specific antimitotic effect, and mdivi‐1 thus represents a novel class of compounds as acentrosomal spindle inducers (ASI).
Classic antimitotic agents such as taxanes (e.g. paclitaxel) and vinca alkaloids (e.g. vinblastine) are widely used as anticancer drugs. The action of these drugs, though still not fully understood, is largely attributed to their ability to bind to tubulin. However, since their binding to tubulin is not tumor cell specific, these drugs cause devastating toxicities to both dividing and non‐dividing healthy normal cells, resulting in myelosuppression and peripheral neuropathy (Mielke et al., 2006).
Strategies that are aimed to reduce the toxicity and improve selectivity against cancer cells by taking advantage of tumor‐specific mitotic phenotypes have been emerging. Several tumor‐specific mitotic features have been identified, such as a compromised mitotic checkpoint (Kops et al., 2004) and over‐amplification of centrosomes in tumor cells (Nigg, 2002). For example, to avoid multipolar division that produces inviable progeny due to the extra centrosomes, tumor cells are able to cluster their centrosomes during mitosis to ensure bipolar division (Quintyne et al., 2005). Griseofulvin, an antifungal drug, was found able to induce centrosome declustering, resulting in consequent multipolar division and cell death in tumor cells that have over‐amplified centrosomes, but leave normal diploid cells that have normal centrosome numbers unharmed (Rebacz et al., 2007). Later, a more potent derivative of griseofulvin has been synthesized and has shown in vivo efficacy without reported toxicity (Raab et al., 2012).
In somatic cells, centrosomes are the major microtubule‐organizing center (MTOC). Each centrosome contains a pair of centrioles, which are essential for maintaining the integrity of the centrosomal structure (Nigg and Raff, 2009). Centrosomes form the poles of the bipolar mitotic spindle during prometaphase to ensure the inheritance of centrosomes to each daughter cell. Despite the fact that centrosomes mark the spindle poles during mitosis, studies have shown that centrosomes per se are not required for establishing the bipolar spindle and the progression of mitosis, but instead are required for entry into S phase of the daughter cells (Hinchcliffe et al., 2001; Khodjakov and Rieder, 2001). The importance of centrosomes during mitosis has been suggested to be critical in ensuring the fidelity of bipolar spindle assembly (Hornick et al., 2011) and cytokinesis (Khodjakov and Rieder, 2001). When centrosomes are artificially removed or their functions are inhibited, the bipolar spindle can still be established but in a non‐centrosomal mode. In addition, the non‐centrosomal pathway is also recognized as an essential mechanism for successful establishment of normal bipolar spindle even in centrosome‐containing cells (Tulu et al., 2003).
In this study, we identified that tumor cells have increased activity in the nucleation and assembly of acentrosomal microtubules. Mdivi‐1, a reported inhibitor of the mitochondrial fission protein Drp1, induces mitotic arrest and apoptosis in a tumor cell specific manner, however, independent of Drp1. We found that mdivi‐1 disrupts the integrity of centrosomal microtubules during mitosis, causing the shift of the assembly of mitotic spindles from a centrosomal to an acentrosomal mode. Formation of multipolar spindles consisting of both centrosomal and acentrosomal poles results in chromosomal segregation failure and subsequent apoptotic cell death. Our data suggest that inducing the formation of acentrosomal multipolar spindles could achieve a tumor‐specific antimitotic effect even in tumor cells that contain normal centrosome numbers.
The human breast carcinoma cell line MDA‐MB‐231 and MCF7, non‐small cell lung carcinoma H1299 and bone osteosarcoma epithelial cell line U2OS were obtained from American Type Culture Collection (ATCC). Human mammary epithelial cell line HMEC and dermal fibroblast cell line NHDF were obtained from Lonza (Walkersville, MD). Drp1 wild‐type and knockout MEF cells were established by Katsuyoshi Mihara (Ishihara et al., 2009), and kindly provided by Kasturi Mitra (University of Alabama). BJ and BJ‐hTERT cells were kindly provided by Dr. Yuan Chang and Dr. Patrick S. Moore. BJ‐SV40 and BJ‐hTERT SV40 cells were established by using a recombinant lentivirus that encodes both SV40 LT and sT. Recombinant lentivirus was produced as described previously (Houben et al., 2010). Bax/Bak wild‐type and double knockout MEF cells were established by Dr. Stanley J. Korsmeyer (Wei et al., 2001), and kindly provided by Dr. Shivendra Singh (University of Pittsburgh Cancer Institute). Cells were cultured in their corresponding media including RPMI‐1640, DMEM, MEBM or McCoy's 5A media in 5% CO2 at 37 °C.
Plasmids obtained from addgene (Cambridge, MA, USA) were: pLenti CMV/TO SV40 small + Large T (w612‐1) (Addgene plasmid 22298), H2B‐mCherry (Addgene plasmid 20972), Tubulin‐GFP (Addgene plasmid 12298) and Centrin‐2‐GFP (Addgene plasmid 29559). Plk1‐YFP plasmid was obtained from Dr. Leizhen Wei (University of Pittsburgh). Transfection was performed using FuGENE 6 (Roche Diagnostics, Indianapolis, IN) or lipofectamine 2000 (Life Technologies) according to the manufacture's instructions.
Cell synchronization and the determination of the DNA content were performed as we previously described (Qian et al., 2012).
Cells grown on glass coverslips were fixed in 4% paraformaldehyde (Electron microscopy sciences, Hatfield, PA) in PBS for 15 min at 37 °C. For tubulin regrowth assay, mitotic cells enriched by nocodazole treatment were collected, washed with fresh media, and then resuspended into prewarmed media in 37 °C containing mdivi‐1 or vehicle DMSO to allow the regrowth of microtubules. Equal amounts of 8% paraformaldehyde in PBS were then added to cells and incubated for 15 min at 37 °C for fixation. After wash with 1% BSA/PBS, cells were blocked using 3% BSA/PBS containing 0.3% Tirton‐X 100 overnight at 4 °C. For staining microtubules, cells were incubated with Alex Fluor 555‐conjugated anti‐β‐tubulin antibody (Cell Signaling Technology). For staining centrosomes, cells were incubated with anti‐γ‐tubulin antibody (Sigma) for overnight at 4 °C, followed by incubation with secondary Alex Fluor 488 goat anti‐mouse antibody (Invitrogen) for 1 h at room temperature. Actin staining was performed using Texas‐Red phalloidin (Life Technologies) according to manufacture's instructions. Slides were mounted with VECTASHIELD mounting medium containing DAPI (Vector Laboratories, Burlingame, CA). Confocal images were captured using a laser‐scanning confocal microscope, Olympus FLUOVIEW FV‐1000, with a PlanApo N 60x oil immersion objective, NA = 1.42 (Olympus).
Western blot analysis was performed as previously describe (Qian et al., 2012). Primary antibodies used were: Drp1 from BD Biosciences, p53 and SV40 T Ag from Santa Cruz Biotechnology, β‐actin from Sigma, cleaved Caspase‐3, cleaved PARP and phospho‐histone H3 (Ser 10) from Cell Signaling Technology.
RNA isolation was performed as previously described (Mao et al., 2013). 1 × 106 cells were seeded in a 60 mm cell culture dish and incubated for 24 h. Cells were then lysed with 1 mL Qiazol lysis reagent. Total RNA was extracted and purified using the Qiagen RNeasy Mini kit according to the manufacturer's instructions. After a wash with buffer RWT followed by two washes with buffer RPE, RNA products were eluted from the column with 30 μL RNase‐free water. RNA quality was determined using an Agilent 2100 Bioanalyzer at the Cancer Biomarkers Facility at the University of Pittsburgh Cancer Institute. In all sample preparations, the average RNA integrity number (RIN) was >9.0. RNA concentration was determined using a Nanodrop 2000.
The microarray analysis using the Human U219 Array Strip and the Affymetrix GeneAtlas system was performed as previously described (Mao et al., 2013). Briefly, four non‐transformed cell lines (NHDF, HMEC, IMR90, BJ) and four transformed cell lines (MDA‐MB‐231, H1299, U2OS, BJ‐SV40) were grouped into mdivi‐1 resistant and sensitive group, respectively. Each cell line within either group was considered as an individual of the biologic duplications. The comparison was between the resistant and sensitive groups. 100 ng of purified total RNA of each cell line was used as the initial material. The corresponding amplified and labeled antisense RNA (aRNA) was fragmented and hybridized to the array strip at 45 °C for 16 h. The strip was then washed and stained. The intensity of each hybridized probe was generated using the GeneAtlas Imaging Station. Raw .cel files from the Human U219 Array Strip were analyzed using the Partek GS 6.6 software. The raw data were normalized and summarized using the robust multichip average method (RMA). Each gene was represented by one or more probe sets.
Differentially expressed genes between transformed and non‐transformed cell lines were found using the bioconductor package samr (Tusher et al., 2001). Genes with a q‐value less than 5% were considered to be differentially expressed. Unsupervised hierarchical clustering was done on all differentially expressed genes and cell lines. Average linkage and Euclidean distance were used as clustering method and similarity metric, respectively. Clustering was performed using the R statistical language. Differentially expressed genes were compared to all pathways listed in KEGG and enrichment p‐value was calculated using the Fisher's exact test. p‐values were corrected by FDR. Pathways that had a FDR less than 0.05 were considered significantly enriched. Gene ontology categories were found to be enriched using DAVID (Huang da et al., 2009a; Huang da et al., 2009b).
Data were expressed as mean ± standard deviation. A Student's t test was used for the comparisons between two groups. p < 0.05 was considered statistically significant.
Mdivi‐1, which belongs to a class of quinazolinone derivative compounds, was identified from a yeast‐based chemical screen as an inhibitor of mitochondrial division dynamin Drp1 (Cassidy‐Stone et al., 2008). Mdivi‐1 has been shown previously to induce cell cycle alterations, indicating its potential in cancer therapy (Qian et al., 2012; Rehman et al., 2012). The IC50 for the inhibition of proliferation in a breast carcinoma cell line MDA‐MB‐231 has been shown to be approximately 50 μM (Qian et al., 2014). In this study, through detailed characterization of the effect of mdivi‐1 on cell cycle progression, we found that mdivi‐1 induced a dose‐dependent accumulation of M phase cells in MDA‐MB‐231 cells, as evidenced by the increase in the percentage of cells with positive staining of phosphorylated histone H3 (p‐H3) (Figure 1A). M phase accumulation induced by mdivi‐1 was also observed in breast carcinoma MCF‐7 cells, and other types of tumor cells including those from osteosarcoma (U2OS) and non‐small cell lung carcinoma (H1299) (Figure 1B). Importantly, under the treatment conditions that cause significant M phase arrest in tumor cells (50 μM for 16 h), mdivi‐1 did not induce M phase arrest in normal human fibroblasts NHDF (normal human dermal fibroblast), BJ (foreskin fibroblast), and epithelial cells HMEC (human mammary epithelial cell) (Figure 1C). These normal cells also did not show M phase accumulation after extended exposure to mdivi‐1 for up to 48 h (Supplemental Figure 1). These results suggested that mdivi‐1 induces M phase arrest in a tumor‐specific manner. Mdivi‐1 was first described as an inhibitor of Drp1 (Cassidy‐Stone et al., 2008). To investigate whether the mitotic arrest caused by mdivi‐1 is dependent on the inhibition of Drp1, we evaluated the effect of mdivi‐1 on mitosis in Drp1 knockout and wild‐type mouse embryonic fibroblasts (MEFs), which were immortalized/transformed by SV40 antigens (Ishihara et al., 2009) (Figure 1D). Both cell lines presented a similar degree of increase in the proportion of cells arrested in M phase in a time‐dependent manner, indicating the mitotic arrest induced by mdivi‐1 is Drp1 independent. Since p53 is known to play important roles in regulating mitosis, we investigated whether p53 is responsible for the effect of mdivi‐1. Mdivi‐1 induced a time‐dependent accumulation of M phase cells in both p53 knockout and wild‐type HCT116 colon cancer cells (Supplemental Figure 2), indicating p53 is not required for the mitotic arrest induced by mdivi‐1.
Mdivi‐1 induces tumor‐specific mitotic arrest independent of Drp1. (A) MDA‐MB‐231 human breast carcinoma cells were treated with increasing doses of mdivi‐1 (0, 10, 20, 50 µM) for 16 h. Flow cytometry was ...
Since the antimitotic effect of mdivi‐1 is specific for tumor cells, we then investigated whether the M phase arrest we observed in Drp1 knockout and wild‐type MEF cells was due to the expression of SV40 antigens (Ishihara et al., 2009). SV40 antigens are known tumor antigens that cause cell transformation by targeting multiple pathways through multiple protein targets such as p53 and Rb family of tumor suppressors and pp2A phosphatase (Ahuja et al., 2005). As shown in Figure 2A, we detected the expression of SV40 Large T and small t antigens in both Drp1 knockout and wild‐type MEF cells. We therefore transduced human BJ cells with lentiviruses that encode both Large T and small t antigens (Figure 2B) to test whether the expression of SV40 antigens was able to sensitize BJ cells to mdiv‐1. Following mdivi‐1 exposure, SV40‐transduced BJ cells (BJ‐SV40) showed a high degree of M phase arrest when compared to BJ cells (Figure 2D,E). Unlike human primary cells that require hTERT expression for immortalization, rodent primary cell cultures including MEF have active telomerase (Greenberg et al., 1998), showing a greater propensity to undergo immortalization/transformation following an oncogene expression like SV40 T antigen (Sedivy, 1998). We thus tested if the immortalization step by hTERT in human cells confers mdivi‐1 susceptibility. Contrary to the effect of SV40 antigens, expression of hTERT alone in BJ cells (BJ‐hTERT) did not render cells sensitive to mdivi‐1 (Figure 2F). Only the further expression of SV40 antigens in hTERT‐immortalized BJ cells (Figure 2C) could result in M phase arrest following mdivi‐1 treatment (Figure 2G). These data indicated that the tumorigenic events analogous to SV40‐mediated process and other than the inhibition of p53 and hTERT‐mediated immortalization are crucial for the antimitotic effect of mdivi‐1 and thus differentiate the response between human normal cells and tumor cells.
SV40‐related cellular events in tumor cell transformation are required for mitotic arrest induced by mdivi‐1. (A) Western blot analysis was performed to detect the expression of SV40 large T and small t antigens in both Drp1 knockout ...
We sought to understand the mechanism underlying the tumor cell specific antimitotic effect of mdivi‐1. We synchronized MDA‐MB‐231 cells at the G1/S boundary using a double thymidine block, and then released the cells into fresh media with or without the presence of mdivi‐1 (50 μM). The change in the percentage of mitotic cells was monitored by the status of p‐H3 over time. Following the release from the G1/S block, both vehicle DMSO and mdivi‐1 treated cells displayed similar progression kinetics through both the S and G2 phases (Figure 3A), indicating that mdivi‐1 did not affect the progression of the cell cycle through these phases. At 12–16 h post release, the vast majority of DMSO treated cells had transitioned through mitosis into the next interphase. In stark contrast, at 9 h post release, mdivi‐1 treated cells began to arrest in mitosis, as shown by the accumulation of cells with 4N DNA content and elevated p‐H3. The cells with positive p‐H3 reached the peak value of 57.9% at 12 h post release. Thereafter, the value of positive p‐H3 in mdivi‐1 treated cells gradually declined perhaps due to the exit of mitosis, which was abnormal and led to an increase in sub‐G1 population over the time from 16 to 24 h post release. These results indicated that mdivi‐1 specifically affects the mitotic phase of the cell cycle. To test if mdivi‐1 has a direct and instant effect on mitosis, we synchronized cells in M phase by nocodazole, and then released them into fresh media with or without mdivi‐1 (Figure 3B). The effect of mdivi‐1 on the M phase progression after nocodazole block was monitored by the levels of p‐H3 (Figure 3B) and the morphologies of spindles and nuclear (Figure 3C). Similar to the observations shown in Figure 3A, DMSO treated cells progressed though mitosis successfully, as evidenced by a sharp decrease in the number of cells with positive p‐H3 following release (Figure 3B). However, in the presence of mdivi‐1, the level of positive p‐H3 retained at 1.5 h post release. At 5 h post release, the decrease of positive p‐H3 that was accompanied with an increase of G2 and the sub‐G1 population was observed (Figure 3B). The increase of G2 cell population in the cell cycle profile indicates the possible cytokinesis failure and mitotic slippage generating tetraploidy G1 phase cells, and the increase of sub‐G1 population indicates apoptotic cell death, both of which suggest the abnormal mitosis following mdivi‐1 treatment. By examining the morphologies of microtubules and chromosomes, we found that in control cells microtubules re‐polymerized and the cells underwent bipolar division within 5 h post release (Figure 3C). However, multipolar mitotic spindles were found to persist throughout the observation period in cells treated with mdivi‐1. In addition, the chromosomes remained condensed but were not able to align properly (Figure 3C). Importantly, mdivi‐1 did not prevent the regrowth of microtubules after nocodazole block. The multipolarity of the mitotic spindles and misaligned chromosomes were also observed in mdivi‐1 treated U2OS cells that were not synchronized by nocodazole pretreatment (Figure 3D). A similar phenomenon was also observed in Drp1 knockout and wild‐type MEF cells after mdivi‐1 treatment (Figure 3E), further confirming that the antimitotic effect of mdivi‐1 is independent of Drp1. Taken together, these data indicate that mdivi‐1 directly affects mitosis progression by disrupting the assembly of the normal mitotic spindle.
Mdivi‐1 specifically and directly affects progression of mitosis by disrupting the normal assembly of the bipolar mitotic spindle. (A) MDA‐MB‐231 cells were synchronized at G1/S border by double thymidine block, and then released ...
Microtubule nucleation from centrosomes depends on γ‐tubulin ring complex. We thus stained for a centrosome marker γ‐tubulin to examine the presence of centrosome in each spindle pole. Four types of cells (U2OS, MDA‐MB‐231, H1299 and BJ‐SV40) were examined (Figure 4A and supplemental Figure 3). The location of γ‐tubulin marked the poles of bipolar spindles in DMSO treated cells. In mdivi‐1 treated cells, the lack of γ‐tubulin staining was observed in some of the multipolar spindle poles (acentrosomal spindle poles), irrespective of the numbers of centrosomes in all types of cells we have examined (Figure 4A and Supplemental Figure 3). Some spindle poles that do not contain centrosomes also seemed to be the major microtubule organizing centers, as shown by the strong and focused staining pattern of β‐tubulin. The lack of centrosomes in some spindle poles after mdivi‐1 treatment was further confirmed by the lack of centrin‐2 in U2OS cells (Figure 4B), another bona fide centrosome component essential for centriole duplication. To examine if mdivi‐1 also causes centrosome declustering and/or centriole splitting in addition to the formation of acentrosomal spindle poles, we quantified the number of U2OS cells containing both acentrosomal and centrosomal spindle poles, but with different centrin spots in each individual centrosomal poles. We found that most cells containing acentrosomal spindle poles after mdivi‐1 treatment contained two centrin spots in their centrosomal poles, only a small percentage of cells contained 1 or >2 centrin spots in their centrosomal poles. These results suggest that centrosome declustering and centriole splitting occur rarely in cells containing acentrosomal spindle poles after mdivi‐1 treatment.
Mdivi‐1 induces multipolar spindles composing of both centrosomal and acentrosomal spindle poles independent of centrosome number. (A) U2OS and MDA‐MB‐231 cells were treated with vehicle DMSO or 50 μM mdivi‐1 ...
In order to confirm that the acentrosomal poles formed after mdivi‐1 treatment are in fact functional mitotic spindle poles, we investigated the distribution pattern of Plk1 in U2OS cells after mdivi‐1 treatment (Figure 4C). In addition to the centrosomal spindle poles, Plk1 also localizes along the mitotic spindle, kinetochores, and the midbody in mitotic cells (Johmura et al., 2011; Xu and Dai, 2011). We confirmed such distribution pattern of Plk1 in DMSO treated control cells. Following mdivi‐1 treatment, multiple spindle poles were observed, with two spindle poles showing focused Plk1 signal (centrosomal spindle poles) and one pole without focused Plk1 signal (acentrosomal spindle pole). Plk1 is uniformly distributed on the acentrosomal spindle microtubules. The spindle microtubules originated from this acentrosomal pole also connected to the chromosomes through kinetochores (concentrated Plk1 spots on chromosomes). These observations suggest that the acentrosomal pole induced by mdivi‐1 is functional mitotic spindle pole that cooperates with centrosome‐containing spindle poles to segregate sister chromatids. In contrast, acentrosomal spindles were not observed in non‐transformed BJ cells following mdivi‐1 treatment (Figure 4D). These data indicated that mdivi‐1 induces spindle multipolarity through the assembly of acentrosomal spindles exclusively in tumor cells.
The mechanism underlying the assembly of acentrosomal microtubules is largely unknown. In addition, to our knowledge, the preferential formation of acentrosomal mitotic spindles in transformed cells has not been reported previously. In order to generate an overview on the genes and pathways that might contribute to the underlying molecular mechanisms of these unique phenomena, we performed genome‐wide gene expression analysis using four non‐transformed (HMEC, IMR90, BJ, NHDF) and four transformed cell lines (BJ‐SV40, MDA‐MB‐231, H1299, U2OS) (Figure 5A). Since the cell lines we have selected were originated from different organ sites and we included BJ‐SV40 cells that are distinct from other established tumor cells due to the expression of SV40 antigens, we reasoned that by analyzing the genes commonly altered in the all four different types of transformed cells versus four non‐transformed cells, we should be able to exclude the genes that are only dysregulated in a particular tumor type, and gain more focused information that is related to the assembly of acentrosomal mitotic spindles in tumor cells. We identified 908 genes that are significantly up‐regulated (Supplemental Table 1) and 102 genes down‐regulated (Supplemental Table 2) in all four transformed cell types (Figure 5A). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis on the total 1010 altered genes revealed that oocyte meiosis pathway is significantly up‐regulated in transformed cells, in addition to the well‐known pathways such as DNA replication and cell cycle (Figure 5B). Considering that spindle assembly during oocyte meiosis is acentrosomal due to the absence of centrosomes, we further analyzed the expression of genes that were known to play important roles in the nucleation and assembly of acentrosomal microtubules in both meiosis and mitosis. We found that several key genes are highly expressed in transformed cells, such as TPX2 and Ran (Figure 5C). We also confirmed the increased expression of several of these genes at the protein level by western blot (Figure 5D). These data suggest that the up‐regulation of the genes involved in oocyte meiosis and assembly of acentrosomal microtubules contributes to the preferential induction of acentrosomal mitotic spindles in tumor cells after mdivi‐1 treatment.
Genes involved in oocyte meiosis and the assembly of acentrosomal microtubules are highly expressed in tumor cells. (A) Unsupervised hierarchical clustering and heat‐map comparison of gene expression profiles of all the differentially expressed ...
We next investigated whether the elevated expression of genes in oocyte meiosis and acentrosomal microtubule assembly could lead to enhanced nucleation and formation of acentrosomal microtubule bundles during mitosis in tumor cells. BJ fibroblast cells and osteosarcoma U2OS cells were incubated with nocodazole to depolymerize their microtubules. The nucleation and the regrowth of both centrosomal and acentrosomal microtubules in mitotic cells were then examined following the removal of nocodazole. As compared to BJ cells, the nucleation of microtubules from acentrosomal sites in U2OS cells was more prominent. We then examined the assembly of spindles in U2OS cells following nocodazole removal, with or without the presence of mdivi‐1 (Figure 6B). Multiple nucleation sites were observed in both DMSO and mdivi‐1 treated U2OS cells, 5 min following nocodazole removal. At 30 min, in DMSO treated cells multiple acentrosomal nucleation sites were resolved into bipolar spindle. In contrast, in mdivi‐1 treated cells, the existence of acentrosomal microtubule nucleation sites persisted and the integrity of microtubules nucleated from centrosomes was severely impaired. Centrosomal microtubules failed to establish bipolar spindles in mdivi‐1 treated cells. Given the fact that centrosome dysfunction promotes the shift of spindle assembly from centrosomal to acentrosomal mode during mitosis, we postulate that the inhibition of the integrity of centrosome‐initiated microtubules by mdivi‐1 prevents the merging of acentrosomal microtubule bundles into centrosomal asters thus permitting the rise of acentrosomal spindle poles.
The nucleation of acentrosomal microtubules is enhanced in tumor cells during mitosis, and mdivi‐1 prevents the integration of acentrosomal MTOCs with centrosomal asters to establish bipolar spindle in tumor cells. (A) BJ and U2OS cells were ...
We sought to determine the cellular consequences of multipolar acentrosomal spindles induced by mdivi‐1 treatment. Mdivi‐1 treated cells that were able to eventually exit mitosis showed aberrations associated with abnormal karyokinesis (increased frequency of cells bearing micronuclei) and failure of cytokinesis (increase in cells containing binuclei) (Figure 7A). Such phenomena are the typical outcomes of anti‐mitosis drugs. In addition, exposure to mdivi‐1 for a longer period such as 48 h led to an increase in the number of cells with a hyperploid DNA content (>4N) (Figure 7B). Missegregation of chromosomes is a known cause of inviable aneuploidy progeny. In agreement, we observed increased apoptosis following mitotic arrest in mdivi‐1 treated cells, as characterized by the increase in the cleavage of caspase‐3 and PARP subsequent to the accumulation of p‐H3 (Figure 7C). These data indicated that the formation of multipolar acentrosomal spindles results in mitotic catastrophe. Importantly, the cleavage of caspase‐3 and PARP was only observed in tumor cell lines and in SV40‐transduced human BJ (BJ‐SV40) and BJ‐hTERT (BJ‐hTERT‐SV40) cells, but not in non‐transformed BJ and BJ‐hTERT cells. Furthermore, we found that the absence of Bax and Bak abolished the cleavage of caspase‐3 following mdivi‐1 treatment (Figure 7D), indicating the crucial role of the mitochondrial apoptotic pathway in mitotic catastrophe following the assembly of multipolar acentrosomal spindles.
Multipolar acentrosomal spindle leads to chromosomal segregation failure and bax/bak‐dependent apoptosis. (A) MDA‐MB‐231 cells were treated with vehicle DMSO or 50 μM mdivi‐1 for 24 h. Cells were ...
In this study, we identified that mdivi‐1 disrupts M phase cell cycle progression and induces subsequent mitotic catastrophe specifically in tumor cells and not in non‐transformed normal cells. The most remarkable feature we observed is the formation of multipolar acentrosomal mitotic spindles following mdivi‐1 treatment, irrespective of the number of centrosomes present in tumor cells. Analysis of the gene expression profiles comparing mdivi‐1‐sensitive transformed cells and mdivi‐1‐resistant non‐transformed cells allowed us to identify a set of genes involved in oocyte meiosis and the assembly of acentrosomal microtubules that are highly expressed in mdivi‐1‐sensitive transformed cells, thus providing a potential comprehensive molecular explanation for the highly preferential induction of acentrosomal mitotic spindle poles in tumor cells, but not in non‐transformed normal cells, after mdivi‐1 treatment. We have further found that tumor cells have elevated activity in the nucleation of acentrosomal microtubules, and mdivi‐1 prevents the integration of the acentrosomal MTOCs with centrosomal MTOCs to establish bipolar spindle during mitosis. Together, our study presents a novel therapeutic strategy of targeting mitosis selectively in tumor cell.
There are three distinct subclasses of microtubules in the mitotic spindle in animal cells: the astral microtubules, the interpolar microtubules and kinetochore‐fibers (K‐fibers). Both astral and interpolar microtubules emanate from centrosomes, and are not directly responsible for chromosome segregation. Chromosome segregation is achieved by the K‐fibers. K‐fibers attach the chromosomes to the two spindle poles. The mechanism responsible for the assembly of K‐fibers is not fully understood. K‐fibers are acentrosomal (Meunier and Vernos, 2011), and the sorting of those acentrosomal K‐fibers into centrosomal MTOCs is essential for mitotic spindle assembly and chromosomal alignment (Fu et al., 2013). In our study, we observed that among the multiple MTOCs induced by mdivi‐1 some are acentrosomal, and the microtubules originated from those acentrosomal sites are attached with chromosomes. These phenomena suggested that the acentrosomal spindle poles induced by mdivi‐1 are presumably the original nucleating sites of K‐fibers, which are normally sorted into the two centrosomal spindle poles during the assembly of bipolar spindle but failed to complete such process in the presence of mdivi‐1.
The assembly of acentrosomal spindles is a unique feature during female meiotic chromosomal segregation, due to the absence of centrosomes in oocytes. Cancer cell division shares some characteristics with oocyte meiosis such as the tendency to generate aneuploidy, MTOC clustering, and cortex softening (Breuer et al., 2010; Terret et al., 2013). Acentrosomal spindle assembly is less robust and has less fidelity, accounting for the high rate of chromosome segregation failure during meiotic division, which accounts for the aneuploidy observed in at least 10% of all embryos in human pregnancy. Recent work has proposed that tumors may gain germline‐like features, and soma‐to‐germline transition may be a general hallmark of tumorigenesis (Feichtinger et al., 2013; Janic et al., 2010; Koslowski et al., 2004). Our findings further support this hypothesis by providing evidence showing elevated meiotic genes in tumor cells. The elevated activity in acentrosomal microtubule formation may render tumor cell mitosis highly error‐prone that may favor tumor development by inducing genome instability.
The nucleation of acentrosomal microtubules can occur spontaneously around the condensed chromatin (Heald et al., 1996). However, the precise origin and the exact mechanism for the generation of acentrosomal spindle microtubules, and the composition of these acentrosomal MTOCs are unclear (Dumont and Desai, 2012; Petry et al., 2013). Acentrosomal microtubule nucleation requires the elevated activity of small GTPase Ran. The active GTP‐bound form of Ran GTPase is generated by the GTP exchange factor, regulator of chromosome condensation 1 (RCC1), which colocalizes with chromatin. Upon nuclear envelope breakdown, Ran‐GTP forms a gradient from chromatins to cytosol, resulting in the release of several specific spindle assembly factors (SAFs), including TPX2, NuMA, NuSAP, RAE1, etc., from the inhibitory effect of importins, thereby activates acentrosomal microtubule nucleation, stabilization, and organization (Dumont and Desai, 2012). TPX2, one of the important SAFs, when released from importin stimulates microtubule nucleation around chromatin, which has been shown to be essential for spindle assembly (Gruss et al., 2001; Schatz et al., 2003). Moreover, spindle assembly from acentrosomal MTOCs also requires the conserved chromosomal passenger complex (CPC) consisting of Aurora kinase B, the inner centromere protein (Incenp), and chromatin‐targeting subunits Survivin and Borealin/Dasra (Dumont and Desai, 2012). Our gene expression analysis has identified that Ran regulated pathway (Supplemental Table 3) and several key genes including Ran, RCC1, TPX2, NuSAP1, and RAE1 are highly expressed in tumor cells, underlying the mechanism of the increased microtubule nucleation from acentrosomal sites in tumor cells. Furthermore, Ran, TPX2, and NuSAP1 have been shown associated with cancer (Aguirre‐Portoles et al., 2012; Espinosa et al., 2013; Xia et al., 2008; Yuen et al., 2012), though with an unclear mechanism. Our data suggest that the high expression of these proteins in tumor cells may cause the improper assembly of acentrosomal microtubules that contributes to chromosomal instability and hence cancer.
In tumors, aberrant centrosome functions and structures have been observed, such as centrosome hypertrophy and increased microtubule nucleating capacity (Lingle et al., 1998; Pihan et al., 1998). Defects in centrosome function have been proposed to contribute to genome instability and tumorigenesis (Basto et al., 2008; Gao et al., 2012). Consistent with these reports, our gene expression analysis has also identified a set of highly expressed genes that are localized in centrosomes, microtubules and spindles in tumor cells (Supplemental Table 4). These gene products may provide a potential targeting mechanism for mdivi‐1, given the fact of the compromised centrosome‐derived microtubules preferentially observed in tumor cells after mdivi‐1 treatment. Thus targeting those genes could be of particular clinical interest for cancer therapy, as it may lead to the dysfunction of centrosome and centrosome‐derived microtubules specifically in tumor cells, which then promote tumor cells to generate multipolar acentrosomal spindles and subsequent mitotic catastrophe.
In conclusion, our study suggests that mdivi‐1 disrupts the integrity of centrosome‐derived microtubules preferentially in tumor cells. Impairment of centrosomal microtubules prevents the integration of acentrosomal microtubule bundles with centrosomal microtubule asters. Consequently, such acentrosomal asters that have already established the kinetochore attachment develop into multipolar acentrosomal spindle poles, which results in chromosomal segregation failure and mitotic catastrophe. Furthermore, the enhanced formation of acentrosomal MTOCs and enhanced activities of oocyte meiotic process may synergistically amplify the effect of mdivi‐1 in tumor cells. Mdivi‐1 thus represents a novel class of tumor cell specific anticancer agents that function as acentrosomal spindle inducers (ASI) to trigger mitotic catastrophe. Due to the known safety of mdivi‐1 in animals and its beneficial effects in protecting cells from cytotoxic stimuli (Lackner and Nunnari, 2010), the further development of this class of agents for cancer therapy through medicinal chemistry and preclinical study is warranted.
The following are the Supplementary data related to this article:
This work was funded by a grant with the Pennsylvania Department of Health PA CURE program, Magee‐Womens Foundation, P30CA047904, P50CA097190, and P50CA121973. This work was also supported by grants from the National Institute of Health (NIH) [CA148629, GM087798 and ES021116] to RWS. Jingnan Wang received funding from China Scholarship Council. We would like to thank Dr. Yuan Chang, Dr. Patrick Moore, and Dr. Leizhen Wei at University of Pittsburgh Cancer Institute for their insightful discussions. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. RWS is a scientific consultant for Trevigen, Inc. We declare that authors have no conflict of interest.
Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.molonc.2014.10.002.
Wang Jingnan, Li Jianfeng, Santana-Santos Lucas, Shuda Masahiro, Sobol Robert W., Van Houten Bennett, Qian Wei, (2015), A novel strategy for targeted killing of tumor cells: Induction of multipolar acentrosomal mitotic spindles with a quinazolinone derivative mdivi‐1, Molecular Oncology, 9, doi: 10.1016/j.molonc.2014.10.002.
Bennett Van Houten, Email: ude.cmpu@bnetuohnav.
Wei Qian, Email: ude.cmpu@wnaiq.