Oncotarget. 2015 Apr 30;6(12):10487-97.

Rampant centrosome amplification underlies more aggressive disease course of triple negative breast cancers.


Pannu V1, Mittal K1, Cantuaria G2, Reid MD3, Li X3, Donthamsetty S1, McBride M1, Klimov S1, Osan R4,5, Gupta MV6, Rida PC1, Aneja R1,7.
  • 1Department of Biology, Georgia State University, Atlanta, GA 30303, USA.
  • 2Department of Gynecologic Oncology, Northside Hospital Cancer Institute, Atlanta, GA 30342, USA.
  • 3Department of Pathology, Emory University Hospital, Atlanta, GA 30322, USA.
  • 4Department of Mathematics and Statistics, Georgia State University, Atlanta, GA 30303, USA.
  • 5Neuroscience Institute, Georgia State University, Atlanta, GA 30303, USA.
  • 6Clinical Pathology & Anatomic Pathology, West Georgia Hospitals, LaGrange, GA 30240, USA.
  • 7Institute of Biomedical Sciences, Georgia State University, Atlanta, GA 30303, USA.



Centrosome amplification (CA), a cell-biological trait, characterizes pre-neoplastic and pre-invasive lesions and is associated with tumor aggressiveness. Recent studies suggest that CA leads to malignant transformation and promotes invasion in mammary epithelial cells. Triple negative breast cancer (TNBC), a histologically-aggressive subtype shows high recurrence, metastases, and mortality rates. Since TNBC and non-TNBC follow variable kinetics of metastatic progression, they constitute a novel test bed to explore if severity and nature of CA can distinguish them apart. We quantitatively assessed structural and numerical centrosomal aberrations for each patient sample in a large-cohort of grade-matched TNBC (n = 30) and non-TNBC (n = 98) cases employing multi-color confocal imaging. Our data establish differences in incidence and severity of CA between TNBC and non-TNBC cell lines and clinical specimens. We found strong correlation between CA and aggressiveness markers associated with metastasis in 20 pairs of grade-matched TNBC and non-TNBC specimens (p < 0.02). Time-lapse imaging of MDA-MB-231 cells harboring amplified centrosomes demonstrated enhanced migratory ability. Our study bridges a vital knowledge gap by pinpointing that CA underlies breast cancer aggressiveness. This previously unrecognized organellar inequality at the centrosome level may allow early-risk prediction and explain higher tumor aggressiveness and mortality rates in TNBC patients.

KEYWORDS: centrosome amplification; disease prognosis; metastasis; triple negative breast cancer

PMC4496369; PMID: 25868856



CA is the presence of centrosomes that are excessively large or numerous. The centrosome is a non-membrane-bound organelle that consists of two barrel-like centrioles surrounded by a highly ordered, approximately toroidal matrix of pericentriolar material (PCM). The centrosome serves as the primary microtubule-organizing center and a signaling hub in animal cells. In normal mammary epithelial cells, the centrosome is <1 µm3 (based on staining for the PCM protein γ-tubulin) and exists in only one or two copies (before and after S phase, respectively). Usually centrosome duplication is tightly coupled to DNA replication, but if this coupling is perturbed, CA may ensue. CA can also arise from mitotic slippage, cytokinesis failure, cell-cell fusion, and de novo formation of centrosomes. CA is a hallmark of human tumors, both solid and hematologic, and has been suspected as a potential cause of cancer for over a century. A majority of breast tumors exhibit CA along with defects in centrosome morphology and abnormal centriole numbers. Induction of CA results in chromosomal instability, aneuploidy, cellular invasion, and accelerated tumor development in certain models, so CA may promote tumorigenesis and aggressive phenotypes in human cancers.


The literature contains conflicting data regarding the association between CA and clinicopathology in breast cancer, with some studies reporting associations with histological grade, tumor size, and nodal metastasis, and others finding no such associations. However, in most of these studies, CA was not rigorously quantitated both in terms of centrosomal volume and number, or sample sizes were small. Furthermore, breast cancer is a highly heterogeneous disease, and yet no study had systematically examined differences in CA between triple-negative breast cancer (TNBC) and non-TNBC prior to ours. Given that TNBC is an aggressive subtype, we hypothesized that it would be characterized by a greater proportion of cells with CA, both numerical and structural, as compared with non-TNBCs. Cells with CA can be selectively targeted with centrosome declustering drugs; so determining which breast tumor subtype exhibits the greatest CA could inform future clinical trials of these agents, which include non-toxic drugs like griseofulvin and noscapine derivatives. It is important to determine to what extent TNBC and non-TNBC cell lines recapitulate patient breast tumors in terms of CA so that preclinical studies can be improved. Finally, no group had previously examined CA-related gene expression profiles among breast tumor subtypes, which is another potential manner of quantitating CA that would enable validation of immunohistochemical data along with capitalization on the wealth of breast cancer microarray data that are publically available. Our work, as published in Oncotarget, represents our efforts to fill these knowledge gaps by comprehensively characterizing centrosomal profiles by gene expression and immunohistochemistry in triple-negative and non-triple-negative breast tumors from a large patient cohort and by immunofluorescence imaging and Western blotting in TNBC and non-TNBC cell lines.


We began our study with in silico analysis of expression levels of genes involved in CA and their correlation with overall survival using data from The Cancer Genome Atlas and the Gene Expression Omnibus. Our overarching goal was to create a centrosome amplification index (CAI) that correlated with clinical outcomes and thus could potentially assist in prognostication. We hypothesized that patients with higher expression levels of CA-associated genes would exhibit worse overall survival, suggesting these patients require more aggressive treatment regimens. Towards this end, we assessed expression levels of the following genes: CETN2, TUBG1, PCTN2, PLK4, and CCNE1, the functions of which we briefly review below to explain their inclusion in the score, and also why we chose to immunolabel and Western blot their proteins as markers of CA in other experiments.


CETN2 encodes centrin-2, a highly conserved, calmodulin-like protein that localizes to the distal end of both full-length centrioles and nascent centrioles (also called “procentrioles”), interacts with many centrosomal proteins, and is required for centriole duplication and ciliogenesis in mammalian cells. In addition to its many roles as a centriolar protein, centrin-2 also contributes to nucleotide excision repair. It is considered a marker for bona fide centrosomes, especially when it is found to colocalize with PCM proteins, which suggests the PCM puncta are not merely centrosome fragments or ectopic PCM accumulations. TUBG1 encodes γ-tubulin, a ubiquitously expressed PCM protein widely labeled as a centrosomal marker. (TUBG2 encodes a paralog whose expression is restricted to the brain in mice.) This tubulin superfamily member exists in highly organized large, multiprotein complexes called γ‑TuRCs, which nucleate and stabilize centrosomal microtubules and thus are critical for mitotic spindle assembly and cell cycle progression. Tubulin α/β-heterodimers polymerize on top of the helical γ-tubulin scaffold in the γ‑TuRC. In normal cells, 1 or 2 γ‑tubulin puncta with a well-defined, approximately circular morphology are easily identified by confocal microscopy. PCTN2 (also known as PCTN) encodes pericentrin, a PCM protein that recruits and anchors γ-tubulin and other centrosomal proteins to the PCM, regulates timely centrosome separation, and is required for nucleation of astral microtubules in mitosis and thus is necessary for correct spindle positioning. Like γ-tubulin, pericentrin is present in PCM throughout the cell cycle and thus is a good centrosomal marker. Similar to centrin-2, pericentrin moonlights as a DNA-damage-response protein. PLK4 encodes the master regulator of centrosome duplication, polo-like kinase 4 (PLK4), a member of the polo family of serine/threonine kinases. PLK4 localizes to the mother centriole and recruits centriolar proteins, namely STIL and SAS6, to initiate procentriole formation. Without PLK4 centrioles do not form, and when PLK4 is overexpressed supernumerary centrioles (and thus supernumerary centrosomes) form. Simultaneous overexpression of PLK4 and depletion of p53 in the developing epidermis of the mouse results in the accumulation of aneuploid cells and acceleration of squamous cell carcinoma development. CCNE1 encodes cyclin E1. Cyclin E (also encoded by the paralog CCNE2) is a G1/S-phase cyclin that licenses centrosomes for duplication by phosphorylating centrosomal proteins like nucleophosmin, CP110, and Mps1. Simultaneously, cyclin E initiates DNA replication, thereby coupling this process to centrosome duplication. Cyclin E overexpression induces CA in p53-deficient bladder cancer cells.


Given that they are integral structural and functional components of the centrosome and their levels are expected to increase with CA, centrin-2, γ‑tubulin, and pericentrin were selected for inclusion in the CAI. We also included PLK4 and cyclin E due to their pivotal roles in centriole duplication and licensing, respectively, and well-defined activities in promoting CA. We computed CAI by summing the log2-transformed, robust multi-array normalized expression levels of the five corresponding genes. We rationalized that multiple genes should be surveyed because CA can arise from deregulation of distinct pathways (e.g., PLK4 or CCNE1 overexpression) and the centrosome is composed of hundreds of proteins, not all of which are necessarily amplified in abundance in the setting of CA. Thus, a multigene score may have a greater sensitivity in detecting CA than a single gene. We found that CAI was higher in TNBCs (n=138) than grade- or stage-matched non-TNBCs (n=466). Moreover, using another dataset, higher CAI (n=78) was associated with worse overall survival than lower CAI (n=84) in Kaplan-Meier analysis, with the optimal cutpoint found using the FINDCUT macro for SAS. These data suggest that CA is associated with an aggressive breast cancer subtype and poor clinical outcomes, so it may be a useful prognostic biomarker in breast cancer.


We substantiated our in silico findings in grade-matched breast tumor specimens. We immunolabeled the tissues for γ-tubulin to quantitate centrosome numbers and volumes by confocal microscopy. We found that ~60% of TNBC cells exhibited >2 centrosomes, whereas only ~30% of non-TNBC cells exhibited >2 centrosomes. We assessed centrosomal volumes using the Zeiss three-dimensional measurement module, and we discovered that centrosomal volumes were larger in TNBC cells (~7 µm3) than non-TNBC cells (~4 µm3) and normal breast cells (<1 µm3). Approximately two-thirds of TNBC cells exhibited numerical and/or structural CA (the latter defined as a volume greater than the upper bound of the normal range, 0.76 µm3) vs. less than half of non-TNBC cells. Compared with grade-matched non-TNBCs, TNBCs exhibited a greater abundance of centrin-2 and γ-tubulin than grade-matched normal breast tissue by Western blotting, supporting the in silico gene expression and confocal imaging data. Having confirmed that numerical and structural CA was more severe in the more aggressive breast cancer subtype, we set out to establish whether the percentage of cells with CA (numerical and/or structural) could risk-stratify patients. We found that tumors exhibiting >20% of cells with CA were associated with worse progression-free survival than tumors with <20% of cells with CA. Collectively, these data reinforce the notion that CA may serve as a valuable prognostic biomarker in breast cancer.


We were next interested in characterizing centrosomal profiles in breast cancer cells lines of different subtypes routinely employed in the laboratory to determine how well they reflect those in clinical specimens. TNBC cell lines (MDA-MB-231 and MDA-MB-468) demonstrated increased levels of pericentrin, PLK4, and cyclin E compared with a non-TNBC cell line (MCF-7) by Western blotting. We also immunofluorescently imaged centrosomes, with representative micrographs displayed. The MDA-MB-231 cell line exhibited a higher proportion of interphase cells with CA (~25%) than the other two cell lines (~5-7%), and both the MDA-MB-231 and MDA-MB-468 cell lines exhibited a greater proportion of mitotic cells with multipolar spindles (~33% and ~18%, respectively) than the non-TNBC cell line (~3%), as determined by confocal imaging of γ-tubulin and α-tubulin. These CA levels were lower than those found in breast tumor samples, which ranged from 15-80% of cells, indicating that commonly used breast cancer cell lines may not faithfully recapitulate the centrosomal status of patient breast tumors, although among the cell lines we tested MDA-MB-231 cells came closest. Altogether, our findings suggest that caution is warranted in generalizing CA-related experimental data from cell lines to the clinical scenario.


Lastly, we wondered how elevated centrosome numbers might contribute to the efficiency of directional cell migration, which could promote metastatic dissemination and poor outcomes. To answer this question, we employed a GFP-centrin-expressing MDA-MB-231 cell line and live cell imaging. We found that cells with >4 centrioles (i.e., >2 centrosomes) exhibited greater net displacement, net displacement rate, and track velocity than cells with £4 centrioles. We also induced CA in MCF7 cells via PLK4 overexpression; this resulted in higher levels of γ-tubulin, centrin, and vimentin (a marker of mesenchymal phenotype), faster wound closure in a scratch assay, and a greater number of migrating cells in a Boyden chamber transmigration assay, than when cells expressing PLK41-608 (a truncated form that retains kinase activity but does not induce CA) were employed in similar assays. These data suggest that CA can promote directional migration of TNBC and non-TNBC cells and may thereby contribute to metastatic dissemination and disease progression in breast cancer.


Our study is an important contribution to the literature because it offers a comprehensive, quantitative portrait of the landscape of CA in patient breast tumors and routinely employed breast cancer cell lines of distinct subtypes. It bolsters the notion that CA is a feature of more aggressive breast tumors and thus may offer insight into the disease course of breast cancer patients, suggesting which ones may benefit from more aggressive treatment regimens. Ultimately, our work helps to settle the debate over whether CA is a cause or consequence of tumorigenesis in breast cancer, namely by revealing a previously unreported mechanism by which CA can fuel tumorigenesis (enhanced directional migration), which occurred even in a wildtype-p53 background (MCF7 cells).



Multiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier SchönmannMultiselect Ultimate Query Plugin by InoPlugs Web Design Vienna | Webdesign Wien and Juwelier Schönmann