In this study we observed that, at known cytostatic doses, mifepristone induced a strikingly similar change in shape of four highly metastatic and aggressive cancer cell lines. This included shrinkage of the cell body with long, thin, neurite-like cellular extensions. Morphology changes were found to be dependent on time of exposure to the drug, with confirmed and observable phenotypic changes occurring after 48 h. These alterations were not influenced by the cellular density of culture, suggesting that mifepristone has an intrinsic effect upon cellular structure, and that this change in morphology is not a result of lack of perceived cell number. In addition, treatment with mifepristone was not associated with changes in total expression of β-actin and α-tubulin.
Mifepristone is known to exert anti-progestin activity by blocking the action of progesterone on progesterone receptors (PR) . In cancer cells expressing high levels of PR, nanomolar concentrations of mifepristone have been shown to be sufficient to block cell growth ( and references cited therein). In the present work, in cell lines not expressing canonical PR , we used concentrations of mifepristone in the micromolar range, higher than those required to bind PR, suggesting that a different mechanism not involving classical PR is at work. Such mechanism, which still needs to be unveiled, may be relevant to target with mifepristone cancers not expressing standard PR.
Potential mediators of the effect of mifepristone are the glucocorticoid receptors (GR). It was reported that mifepristone binds GR when used at concentrations higher than those needed to bind PR . We previously addressed that the only commonality among the four cell lines studied in the present work was the expression of the beta isoform of the GR (GR-β) . This is of interest as GR-α is considered the driver of glucocorticoid effects upon regulation of gene transactivation, whereas GR-β has been mostly considered a dominant negative isoform . There are however reports indicating that in cells forced to express only GR-β, mifepristone was the only one out of more than 50 steroids capable of binding the receptor  and of regulating the activity of a reporter gene .
When operating at micromolar concentrations, mifepristone has a distinctive effect not shared by other natural and synthetic steroids. For instance, when used at equimolar concentrations, mifepristone was a much potent inhibitor of ovarian cancer cell growth than progesterone, medroxyprogesterone acetate or levonorgestrel . Furthermore, equimolar concentrations of the GR agonist dexamethasone did not inhibit growth when compared to mifepristone or to two other related antiprogestins, ORG31710 and CDB2914 . At the concentration utilized, dexamethasone was able to down regulate the expression of GR-α and GR-β isoforms, whereas the antiprogestins did not, suggesting that they have different mechanisms of action despite reaching the cells at micromolar levels .
Important in studying mifepristone as a possible treatment option in oncology was to address its long-term effect on the cancer cells. Interestingly, upon removal of mifepristone following exposure for 3 days, cell morphology and proliferation returned to that of untreated cells in all cell lines except for the LNCaP prostate cancer cells. In these cells, the non-proliferative effect of mifepristone remained as long as 9 days following treatment removal, suggesting an irreversible growth arrest. Previous research has shown the propensity of LNCaP cells to enter irreversible growth arrest and senescence in response, for instance, to treatment with doxorubicin or to culture in androgen-free media [29, 40]. We confirmed that LNCaP cells exposed to mifepristone become senescent upon removal of the drug as indicated by the increase in the activity of perinuclear SA-β-galactosidase. One possible explanation as to why, from a panel of 4 cells lines, only LNCaP cells underwent senescence following mifepristone, relies in the likely induction of the tumor suppressor gene p16Ink4A. While SKOV-3 and U87MG cell lines are null for p16Ink4A[41–43], and MDA-MB-231 has a homozygous deletion of p16Ink4A[44, 45], LNCaP cells retain the p16Ink4A gene . Cells that undergo senescence have been reported to upregulate p16Ink4A[47–49], and not to regrow in response to overexpression of oncogenic Ras . The phenomenon of senescence has been studied both in vitro and in vivo, and pro-senescence therapy has rapidly become a target for cancer treatment . Whereas it was shown that senescence occurs naturally in benign tumors of melanocytes , it was also found that cellular senescence can be induced in vivo and block, for example, prostate tumorigenesis . Also, the use of chemotherapy to induce senescence has been shown to be successful in mice models, leading to an anti-tumor effect with a corresponding increase in p16Ink4A. The ability of mifepristone to induce senescence in p16Ink4A-positive prostate cancer cells provides yet another rationale for its potential use as an anti-cancer agent, in particular in cells carrying wild type versions of the p16Ink4A tumor suppressor gene.
An alternative explanation for the senescence induced by mifepristone in LNCaP cells is the possible mediation by androgen receptors (ARs). LNCaP is the only cell line in the studied cohort that expresses ARs , which have been found able to bind mifepristone . Thus, the role of both p16 Ink4A as well as ARs in the mediation of mifepristone-induced senescence in LNCaP cells deserves to be investigated.
Mifepristone likely altered cellular morphology as a consequence of the dysregulation of the cytoskeletal structure, which was observed via fluorescent staining of actin fibers (F-actin) and tubulin filaments; these were both found to rearrange in response to mifepristone. Actin fibers were found reorganized to sites located at the ends of tubulin-rich extensions. As this was seen multiple times in different directions within individual cells, there appears to be a loss of cell polarity following mifepristone treatment. Upon mifepristone treatment, tubulin filaments were mainly located in the neurite-like extensions, in contrast to their original localization throughout the cell body with particular intensity around the nucleus observed in controls.
F-actin is formed by polymerization of actin molecules that assemble at different times and locations, depending upon the extracellular environment . These actin-based structures interact with one another or with microtubule-based structures, reflecting the complexity of the dynamics of the cytoskeleton . Among the actin-based structures are: i) cortical actin, which mostly defines the shape of the cell; ii) finger-like protrusions termed filopodia that are adhered in some manner to a substratum or another cell, and are believed to function as directional sensors; iii) stress fibers that are contractile actomyosin bundles essential for cell adhesion to the substratum and for changes in cell morphology during migration; iv) lamellipodia, which are surface-attached sheet-like, membrane protrusions with weak adherence, and observed during cell motility and spreading; and v) ruffles, which are sheet-like membrane protrusions or flat membrane folds from the cortical cytoskeleton that do not attach whatsoever to the extracellular matrix. Ruffles are formed as a consequence of inefficient integrin-ligand interaction at the leading edge of lamellipodia and contain densely packed arrays of thin actin filaments . A high frequency of ruffle formation is usually associated with low lamellipodia formation and inefficient cell adhesion and migration . We propose that mifepristone induces the accumulation of membrane ruffles and a reduction in lamellipodia, thus destabilizing the formation of cell substrate adhesions by integrins that connect the cytoskeleton with the extracellular matrix . Under mifepristone treatment, the formation of adhesions may be inefficient because lamellipodia may not have the appropriate anchorage, becoming detached and retracted toward the main cell body, thus forming membrane ruffles. The behavior of mifepristone-treated cells supports the data of Born et al. , suggesting that high ruffling rates are indicative of inadequate adhesion, whereas low ruffling rates are associated with optimal adhesion. While the majority of nascent adhesions undergo rapid turnover such that their components can be incorporated into newly formed adhesion sites, a few mature behind the leading edge in response to tensile stress and increase in size . Adhesion turnover may be blocked by mifepristone leading to the accumulation of actin ruffles that mature and do not adhere to the substratum. We suggest that the more ruffles, the less surface area mifepristone-treated cells would have to actually develop the needed focal adhesion complexes to link the cytoskeleton, the integrins and the extracellular matrix.
We observed that cells under the stress of mifepristone are easily de-adhered from the extracellular surface than untreated cells when exposed to a sub-optimal concentration of trypsin. Because an inverse relationship between de-adhesion time and cell contractility assessed by trypsin-induced de-adhesion has been demonstrated , by altering cytoskeletal dynamics, mifepristone may interfere with the molecular link between the actin cytoskeleton and the extracellular matrix. In addition, owing to the fact that changes in shape and cytoskeletal remodeling are coupled to the cell cycle machinery governing the G1/S transition [60–62], it is possible that the cell growth inhibition caused by mifepristone, which we previously demonstrated to be associated with G1-S cell cycle arrest, blockage of cyclin dependent kinase 2 activity, and accumulation of cyclin dependent kinase inhibitors p27kip1 and p21cip1[14, 17, 26], may be secondary to a primary effect on the cytoskeleton.
Since the survival, movement and invasiveness of cancer cells in the organism require great plasticity in the distribution of F-actin, our data suggest the mifepristone may interfere with such actin polymerization dynamics, disturbing the metastatic process. Blocking actin plasticity with mifepristone can be therapeutically beneficial to reduce the seeding at secondary sites by cells that had detached from a primary tumor.
Microtubules are important to maintain cell shape, play a key role in the polarized distribution of signals within a cell, and have been implicated in the asymmetric regulation of adhesion dynamics; in particular, they promote adhesion disassembly triggering the destabilization first, and then the detachment of adhesion components. At the same time, adhesions can be pulled off the substrate by stress fibers, which contract in response to microtubule depolymerization . We observed a lack of radial distribution of tubulin filaments from the center of the cells in response to mifepristone; instead we visualized an increase of tubulin fibers located in the neurite-like extensions of the cells where the membrane actin ruffles became abundant, suggesting a connection between redistribution of microtubules and dysregulated adhesion. Usually an intact microtubule network with dynamic properties that are asymmetric has the full potential to coordinate adhesion dynamics in different regions of the cells, allowing directional migration . Mifepristone may disrupt this dynamic equilibrium, blocking adhesion dynamics, and, tentatively, migration as well. Further studies are necessary to elucidate the relationship between mifepristone treatment, membrane ruffling, tubulin rearrangement, and cellular adhesion.