Centrosome Aberrations as Drivers of Chromosomal Instability in Breast Cancer

Abstract

Chromosomal instability (CIN), or the dynamic change in chromosome number and composition, has been observed in cancer for decades. Recently, this phenomenon has been implicated as facilitating the acquisition of cancer hallmarks and enabling the formation of aggressive disease. Hence, CIN has the potential to serve as a therapeutic target for a wide range of cancers. CIN in cancer often occurs as a result of disrupting key regulators of mitotic fidelity and faithful chromosome segregation. As a consequence of their essential roles in mitosis, dysfunctional centrosomes can induce and maintain CIN. Centrosome defects are common in breast cancer, a heterogeneous disease characterized by high CIN. These defects include amplification, structural defects, and loss of primary cilium nucleation. Recent studies have begun to illuminate the ability of centrosome aberrations to instigate genomic flux in breast cancer cells and the tumor evolution associated with aggressive disease and poor patient outcomes. Here, we review the role of CIN in breast cancer, the processes by which centrosome defects contribute to CIN in this disease, and the emerging therapeutic approaches that are being developed to capitalize upon such aberrations.

In 2011, Hanahan and Weinberg defined genome instability and mutation as an enabling characteristic that facilitates the acquisition of cancer hallmarks (1). Several forms of genome instability can occur, one of which is chromosomal instability (CIN) (2, 3). This refers to defects in chromosome segregation resulting in gains, losses, and alterations in portions of, or entire, chromosomes (a genomic state known as aneuploidy). These errors ultimately induce gene dosage changes for oncogenes and tumor suppressors (4). Importantly, CIN contributes to the advent of distinct phenotypes by promoting tumor evolution. CIN during early cancer development is associated with poor patient prognosis (5, 6). Furthermore, continuous evolution of genomically unstable tumors leads to more aggressive cancers. Thus, a more comprehensive understanding of the mechanisms leading to CIN should provide targets that can ultimately be leveraged for developing therapies for patients with genomically unstable tumors. Due to their essential role in mitosis, centrosomes provide 1 such target. Abnormalities affecting centrosome number, structure, and nucleation of the primary cilium have been identified as drivers of CIN. This review discusses the various centrosomal defects that contribute to CIN and how they impact breast cancer evolution, patient prognosis, and treatment.

Chromosomal Instability and Cancer

For decades, a high incidence of CIN-induced aneuploidy has been observed in most types of cancer. Technological advances such as cytogenetic and chromosomal array comparative genomic hybridization (CGH) have revealed that the vast majority of tumor cells contain 1 or more chromosomal aberrations (3, 7-10). For example, in breast cancer, it is not uncommon for a tumor to have lost over half of its alleles (11). The vast collection of genetic alterations caused by CIN leads to genetically heterogeneous cell populations within a tumor that continue to diversify with subsequent cell divisions (12). The genomic complexity of these populations contributes to significant intratumoral heterogeneity, resulting in tumor cells that are more adaptive, metastatic, and resistant to treatment (13-15); in other words, CIN imparts a fitness advantage (Fig. 1). Previously, CIN was thought to be tangentially related to the hallmarks of cancer (16). However, it has become evident that the extensive genomic variability induced by CIN produces discrete populations of cells that display an enhanced ability to sustain cancer hallmarks (1, 17). Hence, it has been recently proposed that the processes that foster CIN, including DNA damage/replication stress and mitotic stress should be included in the list of cancer hallmarks (17, 18), highlighting the importance of this enabling phenotype in tumor initiation and progression.

Figure 1.

Centrosome abnormalities induce CIN in cancer cells. Acquired centrosomal abnormalities impart a cellular fitness advantage through CIN, allowing cancer evolution and the generation of tumor cells that are more adaptable, metastatic, and treatment resistant. By inhibiting centrosome function using agents such as kinase inhibitors or centrosomal declustering agents, tumors already experiencing CIN can be further pushed to develop excessive CIN. This leads to cell senescence or death and ultimately, tumor stasis or regression.

Overview of CIN

Imprecision during chromosome condensation, spindle assembly checkpoint, centrosome duplication, DNA replication, and cytokinesis often result in CIN (19, 20), ultimately leading to both large- and small-scale genomic changes. Large genomic alterations, or whole CIN (wCIN), consists of a gain or loss of entire chromosomes causing variations in gene dosage of oncogenes and tumor suppressors (4). In contrast, structural CIN (sCIN) involves translocations, deletions, inversions, insertions, and fragmentations of subchromosomal regions (19). Both wCIN and sCIN events result in unstable aneuploid cells, a dynamic genomic state associated with evolving karyotypes across the tumor cell population. It is notable that a single cell division error can lead to progressive aneuploidy (21). Moreover, focal CIN can occur that results from the shattering and reformation of a single chromosome. This is known as chromothripsis and is a rapid form of karyotypic evolution (22) that leads to dysregulated and disrupted genes within the chromothriptic region.

CIN and Cancer Cell Fitness

As indicated above, the constant flux in chromosome composition in unstable aneuploid cells, along with associated changes in gene expression, causes extensive cell-to-cell variation in tumors (23, 24). This heterogeneity can enhance tumor aggressiveness (25-27). Paradoxically, CIN can also be detrimental to cancers. This paradigm can be explained by biological optimization or the Goldilocks Principle. In the framework of this theory, when CIN levels are “just right” cancer cells have various fitness and oncogenic advantages, including disabled cell cycle checkpoints and loss of tumor suppressor expression. However, cells with CIN levels at the extreme margins (excessive or low CIN) exhibit no evolutionary advantage. Cells with low CIN are unable to adapt. In contrast, excessive CIN can act as a vulnerability where the loss of essential genes/chromosomes leads to cell senescence or death and ultimately, tumor stasis or regression (5, 28, 29) (Fig. 1). Indeed, numerous studies have demonstrated that high CIN enhances sensitivity to cytotoxic therapies and radiotherapy (30-34). Hence, drugs that induce mitotic defects can push modestly genomically unstable cells from intermediate CIN to intolerable CIN (35). To maximally utilize this approach, it is first necessary to identify which cancer types and targets can be leveraged to shift cells to an intolerable state of CIN.

Breast Cancer and CIN

Breast cancer encompasses a number of diseases that are molecularly, morphologically, genomically, and clinically distinct. Clinically, breast cancers are classified by expression of estrogen (ER) or progesterone (PR) receptors and amplification of epidermal growth factor receptor 2 (HER2). Tumors lacking these receptors are referred to as triple negative breast cancer (TNBC). Breast cancers can also be categorized molecularly using gene expression profiles. Molecular subtypes include Luminal A (typically ER/PR+, HER2–), Luminal B (typically ER/PR/HER2+), HER2 (ER/PR–, HER2+), or basal-like (ER/PR/HER2–), with basal-like being further subclassified. The majority of TNBC are the basal subtype (36-38). Despite their discrete attributes, all breast cancer subtypes exhibit some degree of CIN. Indeed, in a study of 2201 tumor samples from 24 cancer types, breast cancers have one of the highest proportions of altered genomes as measured by copy number variation and chromosome segmental alterations (3). The heterogeneity in this disease has also been defined using cytogenetics, chromosomal, and microarray-based CGH, and massively parallel sequencing (14, 39-47). Strikingly, approximately 60% to 80% of breast tumors deviate from a normal diploid karyotype, with about 50% exhibiting genome doubling (48). Breast tumors with higher rates of CIN are associated with poorer patient outcomes, therapeutic resistance, metastasis, and recurrence (33, 36, 49).

CIN and TNBC Outcomes

Among breast cancer subtypes, basal (TNBC) breast cancers rank the highest in several categories of CIN (Fig. 2). This is associated with whole genome doubling or tetraploidization events (37, 38) and is prognostic of TNBC patient outcomes (50). A seminal study by Carter and colleagues developed a gene signature indicative of aneuploidy (51). By conducting a meta-analysis of 1944 tumors spanning 9 cancer types, the authors correlated the expression of individual genes with total functional aneuploidy levels (a proxy measure of aneuploidy that utilizes DNA copy number profiles and spectral karyotyping data). This analysis revealed that tumors with functional aneuploidy were enriched for the expression of 70 genes, now commonly known as the CIN70 gene signature. Birkbak et al. divided TNBC patients into quartiles based on gene expression levels of these CIN70 signature within their tumors and these groups were associated with differences in outcomes. Patients with intermediate (third quartile) CIN70 expression exhibited decreased recurrence-free survival when compared with patients with either extreme (fourth quartile) or low (first and second quartiles) CIN (50). These results exemplify the CIN and cancer fitness paradigm discussed previously. Underscoring the utility of targeting intermediate CIN to induce excessive genomic disruption, TNBC is often treated with taxanes and radiotherapy, both of which induce DNA and chromosomal damage (52-54). Radiotherapy induces excessive sCIN and causes chaotic mitoses that can lead to wCIN (30, 55), whereas taxanes stabilize microtubules and prevent proper chromosome segregation, also leading to wCIN (53, 54). By inducing extreme CIN, these therapies cause senescence, apoptosis, and, ultimately, tumor regression (28, 31, 56-59). Moreover, elevated CIN sensitizes breast cancers to taxanes. While such therapies are initially highly effective, resistance and recurrence is common, suggesting that the level of CIN is insufficient to uniformly induce cell death or irreversible senescence (60-62). Identifying druggable factors that further drive intolerable CIN in a tumor cell–selective manner should lead to highly effective therapies for breast cancer and other malignancies with intrinsic CIN.

Figure 2.

Basal breast tumors are genomically unstable. Breast cancer samples within The Cancer Genome Atlas (TCGA) classified as Luminal A, Luminal B, HER2+, or Basal were queried for (A) aneuploidy score, (B) fraction genome altered (FGA) using CBioportal. n = 945. (C) Expression level of the CIN70 gene signature in breast cancer patients within the TCGA. Note: 63/70 of the CIN70 genes were detected and reported in the dataset. *P < .05 when comparing each subtype with basal (148, 149).

The Centrosome and CIN

Centrosomes play key roles in regulating CIN and can be leveraged to treat breast cancers and other tumor types. Normally, these organelles are precisely regulated to maintain faithful chromosome segregation and cytokinesis. However, cancers often harbor centrosome aberrations that can lead to abnormal mitoses, CIN, and intratumoral heterogeneity (63) (Fig. 3). Notably, many of the CIN70 genes are regulators of centrosome function such as TPX2, NEK2, and TTK. Structural, numerical, and functional defects of centrosomes have been observed in breast cancers and are correlated with elevated CIN and poorer patient prognoses (64-67). The centrosome has also been implicated in driving the development of CIN in early breast lesions and inhibition of centrosome function can promote intolerable CIN. Importantly, the centrosome is composed of and regulated by many enzymatic components that are inherently druggable. For these reasons, the centrosome is a promising therapeutic target in cancers with high CIN.

Figure 3.

Centrosome function is necessary to maintain genome stability and tissue architecture. (A) Centrosome structure, number, and functionality are necessary for maintaining chromosome stability by nucleating mitotic spindles. The centrosome also defines cell polarity via its role in nucleating the primary cilium. (B) Increased centrosome volume and/or centrosome amplification (CA) are well documented occurrences in breast cancer that have been linked to CIN and intratumoral heterogeneity. (C) Loss of cell polarity due to primary cilium defects are less well defined, but is a common feature in cancer that can contribute to dysregulated cell signaling, chromosome missegregation, and disorganized tissue architecture, phenotypes commonly associated with aggressive cancer types.

Overview of the Centrosome

The Centrosome and Mitosis

The centrosome is an organelle that critically regulates cell polarity as well as chromosomal segregation during mitosis (68-70) (Fig. 3A). This multifunctional, microtubule-based structure comprises 2 centrioles surrounded by an electron-dense matrix known as pericentriolar material (PCM). The centrosome is the microtubule-organizing center; that is, it is the nucleating site for microtubule networks established during interphase and mitosis (65, 71, 72). The centrosome defines the number, polarity, and organization of microtubules. Microtubule nucleation is regulated by serine/threonine phosphorylation events that either promote or restrict the PCM from recruiting microtubule-organizing proteins (71, 73). Normally, centrosomes are tightly regulated by a number of proteins that control their number and spatial positioning throughout the cell cycle (74-77). During G₁ of the cell cycle, normal cells have a single centrosome comprised of a pair of centrioles, a “daughter” and a “mother,” designated solely by the timing of their formation in the prior duplication cycle (76). During S phase, the mother and daughter centrioles both duplicate to generate 2 pairs of centrioles. At the onset of prophase, the centrosomes begin to separate and move to opposite poles of the dividing cell (78). At their respective poles, they nucleate bipolar mitotic spindles (79, 80) that are essential for chromosome alignment on the metaphase plate, sister chromatid separation, and cytokinesis (80). Upon forming the mitotic spindle, microtubules emanating from the centrosome elongate toward, and ultimately connect to, kinetochores within chromosomes. Once a centrosome-originating microtubule attaches to a kinetochore in the correct orientation, the microtubule is stabilized (81). Forces generated by centrosome-microtubule-kinetochore attachments pull the sister chromatids to opposite poles of the cell during anaphase. Following sister chromatid separation, the mitotic spindle will also determine the point of cell cleavage during cytokinesis, ensuring that both daughter cells inherit a complete and equal set of chromosomes within symmetrical cells. As the core of the mitotic spindle, centrosomes dictate correct chromosome segregation, cytokinesis, and cell cycle signaling, with centrosome defects being associated with prolonged mitosis, CIN, and failure to complete subsequent mitotic events (82, 83).

The Centrosome and the Primary Cilium

While the centrosome’s role in mitosis is well known, this organelle is also essential for formation of the primary cilium, a specialized, solitary, and nonmotile projection found on the surface of most animal cells that senses and transmits extracellular cues and establishes cell polarity (84, 85). Following mitosis, the mother centriole migrates toward the cell surface to form a basal body that acts as a scaffold for microtubule nucleation. In this case, the microtubules form the interior of the primary cilium (86, 87). Microtubule minus-ends anchor to the basal body near the apical membrane of the cell, adjacent to the lumen, while the plus-ends extend to the basal compartment, adjacent to the basement membrane (88). Through this directional anchoring, microtubules emanating from the centrosome dictate apical–basal polarity. Loss of cell polarity due to abnormalities in the primary cilium disrupt normal tissue architecture, a common hallmark of cancer.

Centrosome Dysfunction and CIN

Centrosome Amplification

Centrosome amplification (CA) refers to an abnormal increase in the number of centrosomes in the cell. This includes centrosomes that contain more than 4 centrioles and/or the acquisition of more than 1 pair of centrosomes when cells are not actively engaged in the cell cycle (64-66, 87). CA is highly prevalent in solid tumors and cell lines, with ≥75% of breast cancers displaying CA (90). Compared with the NCI-60 panel of cell lines originating from 9 different tissue types, breast cancer cell lines across all subtypes have significantly higher CA (90). CA arises through several processes, including hormone signaling perturbation (91, 92), excessive recruitment of PCM (93-95), abnormal kinase regulation, mutation-induced dysfunction of centrosome proteins, centrosome gene dysregulation, cytokinesis failure, or cell–cell fusion (96, 97). Downstream, CA induces CIN by initiating multipolar spindles during prophase, leading to chromosome missegregation (98) (Fig. 3B). Abnormal chromosomal attachments arising from multipolar spindles cause an array of mitotic defects including lagging or broken chromosomes that generate micronuclei, chromosome rearrangements, and chromothripsis (98, 99); all of which contribute to CIN.

Similar centrosome defects are observed in patient samples. In a cohort of 362 breast cancer patients, CA was assessed by immunohistochemistry for pericentrin, a component of PCM, and polyglutamylated tubulin (γ-tubulin), a marker of centrioles. The number of centrosomes per cell in tumor tissue was nearly double that of normal breast epithelium. The population of patients with the highest number of centrosomes were those in the HER2+ and TNBC cohorts (97). These data affirm results from 2 smaller studies, both using similar immunohistochemistry methods to quantify centrosome number (64, 66). In a separate study, basal breast cancers were also shown to have significantly elevated CA when compared with receptor-positive subtypes (Luminal A/B, HER2+) in cell lines and patient tumor samples (90, 100). Overall, patients with higher centrosome number exhibit higher stage and grade tumors (97), with the most aggressive breast cancer subtypes (TNBC, HER2+) having more extensive CA. Patients with tumors that harbor CA are also more likely to have decreased progression-free survival, or shorter time to recurrence or metastasis, compared with patients with less centrosome aberrations (100). On a molecular level, overexpression of 5 centrosome-related genes, CETN2 (centrin-2), TUBG1 (γ-tubulin), PCNT2 (pericentrin), PLK4 (polo-like kinase 4), and CCNE1 (cyclin E1), are associated with decreased overall survival regardless of breast cancer subtype (100).

Several features of aggressive breast cancers are associated with CA, including loss of estrogen-dependence and p53 function, elevated epidermal growth factor receptor expression, tumor dedifferentiation, and increased metastatic potential (64, 97, 101). In an independent study using METABRIC and TCGA datasets, a signature of 20 centrosome structural genes, named the CA20, was developed to assess the association of CA with clinical outcome in breast cancer patients. This study found that tumors with a high CA20 signature were associated with decreased overall patient survival compared with tumors that had a lower CA20 score (102). The CA20 score was also highly correlated with a smaller subset of the CIN70 gene signature, known as CIN25 (90). Together, these data indicate that CA may drive malignant phenotypes of breast cancer and affirm the relationship between CA and CIN.

The aggressive nature of tumors harboring CA may be due to its ability to drive aneuploidy, CIN, and intratumoral heterogeneity. In the previously mentioned cohort of 362 patients, fluorescence in situ hybridization analyses revealed that CA positively correlates with tumor ploidy and an increase in nonmodal number of chromosomes, both indicators of CIN (97). Unsurprisingly, patients with both high CA and high CIN had worse overall survival (97). In cell line models, higher rates of CA are also associated with CIN, including abnormal mitotic events characterized by multipolar spindles, lagging chromosomes, and cytokinetic defects (66, 97). Breast tumors comprising cells with severe CA are also more likely to have aberrant spindle morphology, a common precursor of CIN, and higher mitotic indices, which correlate with worse prognosis (66).

Direct evidence that CA impacts various stages of tumor development, including tumor initiation and progression, has been obtained using rat and mouse models of hormonally controlled mammary tumorigenesis (91, 92, 103-107). Matthews and colleagues found that glucocorticoid receptor (GR) localization at the mitotic spindle, especially in metaphase and anaphase cells, is necessary for mitotic progression. Specifically, loss of GR at the mitotic spindle led to CA, chromosome missegregation, aneuploidy, and increased incidence of tumor formation (91). Importantly, the disruption of the DNA-binding domain of GR alone was insufficient to induce CA and mitotic defects. Hence, these data suggest that GR may have a nontranscriptional role in regulating mitotic progression. In a second model of rat tumorigenesis, long-term estrogen stimulation elevated the expression of the centrosome protein Aurora Kinase A (AURKA), CA, CIN, and tumorigenesis (92). These results contrast with another study utilizing a rat model of pregnancy-induced protection from breast cancer. In this case, short-term estrogen treatment coupled with progesterone, which mimics pregnancy, prevented AURKA overexpression, CA, and CIN. This approach revealed that parity, or hormone signaling resembling parity, caused resistance to carcinogen-induced tumorigenesis in rats (104). These studies indicate the important role of hormones and their timing in the regulation of genomic stability and how modulation of these control mechanisms can impact breast tumorigenesis. Further supporting the tumorigenic role of CA, a study involving overexpression of AURKA in the murine mammary gland demonstrated that CA led to premature sister chromatid segregation, chromosome tetraploidization, CIN, and the formation of mammary tumors (103). A second centrosome associated protein, ninein-like protein (Nlp) has been found to be overexpressed in 80% of breast cancer patients (107). Transgenic mice that globally overexpress Nlp also form spontaneous mammary tumors, with CA occurring in tumors as well as mouse embryonic fibroblasts (89, 107). These findings revealed that dysregulated centrosome-associated proteins lead to CA and that this may induce breast tumorigenesis. These in vivo models highlight the impact that CA has on breast cancer initiation and demonstrate that CA can occur prior to detectable lesions.

This same ability of CA to induce CIN also drives tumor progression through the acquisition of intratumoral heterogeneity. As a genomically unstable tumor develops, clonal outgrowth contributes to branched evolution. In the MDA-MB-231 model of claudin-low TNBC, cells with CA demonstrated increased migratory capacity compared with cells devoid of CA (100), a phenotype associated with invasive, metastatic disease. To assess whether CA drives aggressive phenotypes, this same study overexpressed another centrosomal kinase, PLK4, in the nontransformed mammary epithelial cell line MCF10A. PLK4 also induced CA with subsequently elevated invasion and migration (100). These findings indicate that CA is an important driver of aggressive phenotypes in cancer. This could be due to elevated CIN and subsequent evolution and intratumoral heterogeneity. Alternatively, the centrosome also controls cell polarization, intracellular positioning of the small GTPase, Rac1, and cell migration (108-110). Thus, CA may also directly stimulate migration and invasion of cancer cells. This possibility is supported by the ability of PLK4 and NEK2, another centrosome kinase, to induce CA and migration when overexpressed. This occurs in a short time frame before selection of promigratory genomic changes (111, 112), thus elevated migration could be a direct effect of altering centrosome function rather than its indirect impact on CIN. The presence of CA is associated with breast tumor initiation and tumor progression, thus CA has the potential to be both a diagnostic and prognostic feature of breast cancer.

Centrosome Structural Abnormalities

Maintaining the structure of the centrosome and its centrioles is as vital as controlling centrosome number. Structural abnormalities can be classified as defects in centriole shape or in the amount of surrounding PCM. Mechanisms underlying such abnormalities are less well studied than CA; however, altered expression of genes that control centrosome structure may be a key factor (113). These abnormalities can present in a few ways, but the most common phenomenon in breast cancer is overelongated centrioles (90, 100). Centrioles are the microtubule-based cylinders that constitute the centrosome and are typically ≈450 nm in length. In the NCI-60 cell line panel, breast cancer cell centrioles have a median length of ≈546 nm, making them the cancer type with the longest centrioles across the panel (90). Overly elongated centrioles are prone to fragmentation, resulting in the formation of ectopic procentrioles. When centrosome elongation occurs asymmetrically, 1 centriole nucleates more microtubules and becomes unstable. Subsequently, centriole fragmentation causes CA and increases the number of sites for microtubule nucleation (90). This ultimately leads to CIN (Fig. 3B). The second mechanism through which CA can occur is as a result of de novo ectopic procentriole formation (90). Rather than fragmenting, overly long centrioles can generate more than 1 daughter centriole along its length, compared with normal cells where each centriole typically only nucleates the formation of a single daughter (90). As a result, ectopic procentriole formation increases the number of centrioles (>4) and centrosomes per cell. Lastly, overly long centrioles can cause the centrosome to be overactive, leading to excessive microtubule-nucleation in the absence of overt fragmentation. This is not considered CA because a normal centriole number is observed, but it does cause the same defect in microtubule organization. The clinical relevance of increased microtubule-nucleating capacity observed in cell lines was affirmed using breast adenocarcinoma tissue samples (114), where microtubules were found to originate from multiple regions within tumor cells (114). Centrosomes with increased microtubules can generate unbalanced forces on chromosomes, multipolar spindles, and chromosome missegregation, all of which can induce CIN and intratumoral heterogeneity. Confirming a direct link between centriole overelongation and CIN, overexpression of CPAP, a promoter of centriole elongation, was found to induce centriole overelongation in a luminal A breast cancer cell line (MDA-MB-435) and this resulted in chromosome missegregation and CIN (90). These results indicate that centriole overelongation contributes to intratumoral heterogeneity in breast cancer either due to secondary CA or as a result of increased microtubule-nucleating capacity.

A second form of centrosome structural abnormality involves elevated centrosome volume, or centrosome hypertrophy. Comparison of centrosome volume using γ-tubulin immunostaining in breast cancer samples revealed a significant increase in centrosome size compared with normal adjacent breast epithelium. Strikingly, centrosome size was also found to be subtype specific, with TNBC samples having larger centrosomes by volume than non-TNBC samples (100). Similar results were obtained using centrin immunostaining. In this case, centrosomes in breast tumor cells were approximately 7 to 10 times larger than those in normal breast (65). Centrosome hypertrophy may contribute to CIN and intratumoral heterogeneity by disrupting normal mitoses. In all reported cases of structural abnormalities, features of CIN were present, indicating varied irregular centrosome pathologies can independently contribute to mitotic dysfunction and chromosome segregation errors.

Primary Cilium Aberrations

While the centrosome’s role in mitosis has been extensively studied, this organelle also acts as the basal body, which anchors microtubules to form the primary cilium and establishes apical–basal cell polarity. The directionality of the microtubules emanating from the basal body establishes the polarity of the cell (Fig. 3A) and loss of cellular polarity due to centrosome aberrations can disrupt normal tissue architecture (Fig. 3C). This is particularly important in secretory tissues such as the mammary gland that comprises alveoli that produce milk and ducts that transport it during lactation. The ducts and alveoli are lined by 2 layers of cells: the luminal epithelial layer which is closest to the lumen of the duct and the myoepithelial layer attached to the basement membrane (115). Proper cellular polarity is necessary for the formation of complex branching units within the mammary gland (115). The cells themselves are also highly organized and oriented such that the nucleus resides at the basal portion of the cell while the centrosome is positioned at the luminal portion of the mammary duct, providing directionality for milk secretion. Loss of this organizational structure, including apical–basal polarity, is often observed during breast cancer progression (115). Because the centrosome dictates this polarity, it is not surprising that centrosome aberrations can induce irregular breast tissue architecture in the setting of breast cancer.

In normal breast tissue and tissue adjacent to breast tumors, epithelial cells form highly organized ring-like structures with centrin/centrosomes being localized near the luminal membrane (65). In contrast, breast cancer cells containing centrosome abnormalities display an architecture that is chaotic and disarrayed, losing the luminal structure that is characteristic of the terminal ductal lobular units (97). Disrupted tissue organization is often associated with a relocalization of centrosomes to regions near the nucleus rather than at the apical pole of the cell (116). This abnormal positioning prevents proper apical localization of the primary cilium and impedes normal polarization. In other cases, centrin can be diffusely localized throughout the cytoplasm of cancer cells, resulting in the inability to coalesce a functional centrosome. This also results in the absence of a primary cilium, disruption of polarity, and architectural disarray (65). Notably, loss of the primary cilium occurs in most breast cancers (117). In cell lines, the loss of the primary cilium was more prominent in transformed and aggressive derivatives than in parental nontransformed cells (117). Moreover, in vitro studies of mammary epithelial cells grown in 3D cultures demonstrated that loss of tissue architecture was associated with increased chromosome missegregation, perhaps due to disrupted cell polarity. Mature mammary spheroids that had an organized tissue architecture similar to normal breast tissue lacked CIN phenotypes. However, immature mammary spheroids that lacked tissue architecture had elevated rates of chromosomal missegregation (118). While incompletely understood at a mechanistic level, these data indicate that the loss of cellular polarity cues imparted by the primary cilium is associated with inaccurate chromosome segregation. Further studies will be necessary to elucidate how primary cilium defects result in disorganized tissue architecture and how this contributes to CIN. This is especially important because the morphological complexity in invasive ductal breast carcinoma is linked to both disease-specific and overall survival and is more strongly associated with outcomes than other standard prognosticators such as nodal status or grade (119).

Therapeutic Potential of Targeting Centrosomes

As indicated above, CIN and its associated intratumoral heterogeneity contribute to significant disease burden, cancer cell fitness, and therapeutic recurrence in breast cancer (120, 121). However, CIN can also be a therapeutic vulnerability that can be leveraged to induce cell death and tumor regression or suppression (5, 28, 29, 59) (Fig. 1). CIN-elevating strategies typically exacerbate existing instability by inducing extreme levels of chromosome missegregation or DNA damage and loss of vital genomic information (35). CIN-targeting therapies typically disrupt mitosis and chromosome dynamics. For example, taxanes, including paclitaxel and docetaxel, are one of the most established classes of chemotherapy. These drugs stabilize microtubules, resulting in mitotic arrest due to activation of the spindle assembly checkpoint (54, 122). A number of other drugs function in a similar capacity to inhibit microtubules and induce chromosome missegregation. Furthermore, drugs targeting mitotic kinases, chromatin modifications, and microtubule-associated motor proteins are also inducers of excessive CIN (35). An additional druggable target related to CIN in breast cancer is the centrosome itself. Given the high incidence of centrosomal abnormalities in this disease and their relationship to chromosome segregation, targeting centrosomes is a rational avenue for cancer therapy. This can involve several different approaches including therapeutics that are directed at CA, primary cilium defects, phosphorylation dynamics, upstream activating kinases, nucleating capacity of the centrosome, or overexpression of proteins involved in centrosome structure or function (123). Several therapies that directly or indirectly impact these mechanisms have been developed and ongoing efforts focus on assessing their efficacy as single agents and in combination with other therapies.

Centrosomal Kinase Inhibitors

Given the diverse roles of various kinases in cancer initiation and progression, suppressing kinase activity has been a favorable approach for cancer drug development (124). Several centrosome kinase inhibitors are already in clinical trials, including those that target polo-like kinases (PLK), cell cycle related kinases, and Aurora kinases (125-127). Given the frequency of centrosome aberrations in breast cancers, centrosome kinase inhibitors should be highly effective in this disease. As a recent example, inhibitors of PLK4 have been assessed in breast cancers with TRIM37 amplification (128). TRIM37 limits the accumulation of PCM proteins, thus preventing the formation of noncentrosomal PCM foci. These foci are essential for chromosome segregation and maintenance of euploidy in cells that are viable but lack centrosomes. Thus, the overexpression of TRIM37 provides a vulnerability of tumor cells to agents that impact centrosomes. Notably, the inhibition of PLK4 causes centrosome depletion and is synthetically lethal with TRIM37 overexpression in breast cancer cells (128). PLK4 inhibition using CFI-400945 CFI-400945 is also effective in breast cancer cells with centrosome abnormalities due to its ability to induce catastrophic defects in centrosome function (129).

Aurora kinases are also important targets in breast cancer that control centrosomes or microtubules. AURKA acts at centrosomes to permit spindle bipolarity by regulating microtubule assembly and centrosome organization. In contrast, Aurora kinase B (AURKB) is a member of the chromosome passenger complex that localizes to chromosomes and assists in kinetochore–microtubule attachment during metaphase (130). Both AURKA and AURKB are overexpressed or amplified in breast cancer and are associated with poor outcomes, making them promising therapeutic targets (131). As such, a number of Aurora kinase inhibitors are currently being evaluated in breast cancer clinical trials (126, 132-136). In addition to the goal of treating primary disease, centrosome kinases may also be leveraged to combat therapeutic resistance. Taxane resistance is common in breast cancer patients and remains a significant therapeutic obstacle. We recently reported that paclitaxel-resistant cells have CA and express elevated levels of NIMA-related kinase 2 (NEK2), a kinase that controls centrosome separation and bipolar spindle formation in mitotic cells (137). These findings indicated that NEK2 may be a potential target to combat paclitaxel resistance. Indeed, inhibiting NEK2 with small molecule inhibitors or siRNA resensitized breast cancer cell lines to taxanes. Moreover, the combination of paclitaxel with NEK2 inhibitors suppressed tumor growth more so than either drug alone in multiple in vivo models of breast cancer (138). These data support the use of centrosome kinase inhibitors as either first line, targeted, or combination therapy in breast cancer.

Centrosome Declustering Agents

Tool compounds have also been developed that target compensatory mechanisms utilized by cells to sustain viability within the context of CA. As stated above, CA is a common feature of breast tumors that can lead to multipolar spindles and CIN. To overcome CA and prevent chaotic and fatal mitoses, these cells undergo an adaptive process called centrosome clustering that involves the coalescence of multiple centrosomes into 2 functional spindle poles comprised of megacentrosomes (139). This process is beneficial in tumor cells for 2 reasons: (1) it prevents excessive CIN and (2) it still enables low-grade chromosome missegregation and sustains tumor evolution (139, 140). Centrosome clustering is common in centrosome-amplified breast cancer tumors, with more than 75% of cells having this observable phenotype in tumors and cell lines (141). Multiple drugs and targets have been identified to disrupt centrosome clustering. Kinesin Family Member C1 (KIFC1, also known as HSET) is a nonessential kinesin motor protein that is overexpressed in breast cancer cells. Along with its binding partner CEP215, KIFC1 is essential for centrosome clustering and survival in breast cancer cell lines (142). Genetic depletion of either KIFC1 or CEP215 significantly increases multipolar spindles, centrosome scattering, and death in cell lines with CA (143, 144). Multiple small molecule inhibitors of KIFC1 have been developed, including CW069 and AZ282, and both phenocopy the effects of KIFC1 genetic depletion. Indirect inhibition of KIFC1 using a small molecule inhibitor of poly(ADP-ribose) polymerase (PARP) (PJ34) also blocks breast cancer cell growth due to a loss of centrosome clustering (142, 145). Lastly, KIFC1 depletion sensitizes breast cancer cells to ionizing radiation (IR) (146). IR induces CA and protective centrosome clustering, therefore genomic depletion of KIFC1 provides a selective vulnerability, making these cells more sensitive to IR (146). Griseofulvin (139) and noscapinoids (147) are additional examples of microtubule-inhibiting drugs that also prevent centrosome clustering. Discerning whether drugs targeting centrosome clustering are effective in patients awaits the development of clinically relevant molecules.

Conclusions and Future Directions

Intratumoral heterogeneity and CIN contribute to evolving genomic landscapes that result in aggressive breast cancers. A limited understanding of the many mechanisms that lead to CIN have precluded efforts to develop strategies that combat tumor evolution. In this regard, the centrosome offers a promising prognostic tool and therapeutic target in breast cancer. Centrosome aberrations are a driving force in breast cancer that cause chromosome missegregation, CIN, and tumorigenesis. While abnormal centrosomes participate in tumor evolution and are associated with worse outcomes in breast cancer, they also represent an attractive vulnerability. Many therapies are at different stages of development for targeting aberrant centrosomes in breast cancer. However, a major looming limitation of anticentrosome therapies is the advent of drug resistance. Generating additional CIN could expedite stochastic evolution as a result of chromosome missegregation. Cells able to evade death from centrosome targeting will likely have elevated CIN compared with the remaining tumor bulk and have the ability to resist further treatment. Thus, it will be essential for centrosome-targeting drugs to induce intolerable levels of CIN that ensure cell death. Elucidating mechanisms of resistance that could develop in response to these therapies is necessary before wide-scale implementation in breast cancer patients. It is likely that evoking durable responses to centrosome-targeting drugs will require their combination with additional CIN-inducing therapies to prevent adaptive and resistant disease. Discovering effective biomarkers predictive of susceptibility to centrosome-targeted therapies would also facilitate their clinical development and deployment. These may include direct assessments of centrosome number and size as well as using molecular profiling to assess the CA20 score (102). Overall, it remains imperative to understand how to therapeutically induce excessive CIN while simultaneously preventing further evolution. Dysfunctional centrosomes are a partial and promising answer to this challenging issue in breast cancer.

Glossary

Abbreviations

CA

centrosome amplification

CIN

chromosomal instability

ER

estrogen receptor

GR

glucocorticoid receptor

HER2

epidermal growth factor receptor 2

PCM

pericentriolar material

PR

progesterone receptor

TNBC

triple negative breast cancer

Additional Information

Disclosure Statement: The authors have nothing to disclose.

Data Availability

Some or all data generated or analyzed during this study are included in this published article or in the data repositories listed in References 148 and 149.