Despite the spectacular progress in cancer treatment made during recent years, some types of aggressive cancer are still able to spread easily and become resistant to anticancer treatment. One of the reasons for this could be the phenomenon of therapy-induced senescence (TIS).
TIS, which halts cancer cell proliferation instead of inducing cell death, became a desirable outcome of cancer treatment [1
]. However, it eventually turned out that the senescence of cancer cells can have an adverse effect of radio/chemotherapy. Indeed, senescent cells secrete many factors that modify the microenvironment, which in turn favor cancer development [3
]. Moreover, the senescence of cancer cells can be reversible. Interestingly, it seems that the reversibility of cancer cell proliferation arrest is associated with their therapy-induced polyploidization [4
]. The progeny of polyploid senescent cells regain the ability to proliferate, together with depolyploidization. Thus, TIS could represent a mechanism of evasion from the toxicity of chemotherapy and radiation, facilitating cancer recurrence [14
Senescent cancer cells, beside division arrest and secretory activity, known as the senescence-associated secretory phenotype (SASP), are, similarly to senescent normal cells, characterized by many other features, such as the increased activity of senescence-associated β-galactosidase (SA-β-gal), lipofuscin and lipid droplet accumulation, altered morphology (flattening and increased granularity), an increased level of cyclin-dependent kinase inhibitors, such as p16INK4A and p21WAF1/CIP1 [15
], increased lysosomal mass [16
], morphologically and functionally altered mitochondria [17
] and DNA double-strand breaks (DSBs). The latter induces the so-called DNA damage response (DDR) signaling pathway [18
]. The most frequently analyzed key proteins of the DDR, which include DNA DSB sensors, mediators and executors, are: γH2AX, 53BP1, p-ATM, p-ATR, p-p53 and p-CHK2 [6
]. Cellular senescence, characterized by increased metabolism, is closely interrelated with autophagy, however, senescence may be a result of autophagy impairment or, on the contrary, senescence may lead to autophagy dysfunction. Autophagy in senescent cells, especially cancer senescent cells, is highly dependent on the cell type and context [19
]. It has been postulated by Erenpreisa et al. [20
] that transiently senescent cancer cells acquire additional DNA repair capacity through mitotic slippage and entering a sequence of ploidy cycles, which facilitate the repair and sorting of damaged DNA, ultimately promoting the genesis of mitotically competent daughter cells following final depolyploidization. It has been stated that autophagy is required to fuel this process [21
Autophagy is a catabolic process in which macromolecules and organelles are degraded and recycled, thus providing metabolites to maintain the energy supply in the cell. A characteristic feature of macroautophagy (herein referred to as autophagy) is the formation of vesicles, called autophagosomes, that enclose the degradation-bound cargo and, subsequently, fuse with lysosomes, giving rise to autolysosomes, wherein the cargo is degraded and recycled. The formation of the autophagosome includes phagophore nucleation, elongation and vesicle completion, which are tightly regulated by various autophagy-related proteins, e.g., the ULK1/2 complex, the class III PtdIns3K complex and LC3-II [22
]. Autophagy is regulated at each stage, i.e., initiation, vesicle fusion or cargo degradation, by external factors or by endogenous modulators, e.g., Rubicon, the protein present in the Beclin complex, and inhibiting autophagy [16
]. As autophagy is a highly dynamic process, for proper estimation of autophagy efficiency, it is crucial to measure the autophagic flux, which determines the degradation activity [22
Although the modulation of autophagy is a very important part of cancer therapy [23
], no therapies are currently available that specifically focus on autophagy modulation. Moreover, there are many controversies in the literature concerning the role of autophagy in cancer cell senescence and, as has been pointed out [24
], it is difficult to judge whether and how these two processes are interconnected.
Accordingly, in this study, we aimed to answer the question about autophagy modulation in breast cancer cells induced to senescence following doxorubicin treatment. We chose MDA-MB-231 and MCF-7 cells. MDA-MB-231 cells are triple negative breast cancer and have a mutated form of p53 [25
], whereas MCF-7 cells possess an estrogen receptor and WT p53 [26
]. Moreover MDA-MB-231 cells, in contrast to MCF-7, have a very low basal autophagy level [28
]. Despite this, we expected the senescence process in both types of cells to be connected with polyploidization and senescence escape [29
]. Thus, we were interested in propensity to undergo senescence and autophagy activity in the escapers.
The induction of polyploid giant cells by anticancer therapy and their role in cancer resistance and metastasis has been well described, however, the issue of polyploid cell senescence has still not been extensively explored by researchers [8
]. The depolyploidization of polyploid/multinucleated giant cells was first reported almost twenty years ago [45
]. These and subsequent studies by Erenpreisa and colleagues demonstrated that treated cells first undergo polyploidization, ultimately resulting in the emergence of mitotically dividing para-diploid progeny [5
]. Similarly to therapy-induced polyploidization (TIP), it is now generally accepted that therapy-induced senescence (TIS) favors escaping from division arrest and re-emergence into an actively reproductive state. Indeed, many anticancer treatments induce cell senescence both in vitro and in vivo, posing a potential threat to the effectiveness of therapy (reviewed in [48
]). Although studies which show the coupling of TIP and TIS are still scarce (reviewed in [44
])—it seems logical to assume that these two processes occur simultaneously or sequentially in cancer cells subjected to anticancer therapy and that DNA over-replication in senescent cells is a driving force leading to atypical cell divisions. Indeed, in this study and in a previous one [38
], we showed that MDA-MB-231 and MCF-7 breast cancer cells displayed the features of cell senescence, such as SA-β-gal activity and became polyploid. Additionally, the hallmarks of cell senescence, including DNA damage and DNA damage response, were detected several days after dox treatment and they preceded cell polyploidization. Eventually, polyploid cells disappeared from the culture and, in their place, small cells appeared.
DNA damage, especially DSBs and the subsequent DDR, are almost universal features of radio- and chemotherapy treatment [49
] and cell senescence [18
]. The key player in the DDR and senescence is the p53 protein, which transactivates the CDKN1A gene, producing the main cell cycle inhibitor, p21WAF1/CIP1 [6
]. Therefore, we used two different breast cancer cell lines, namely MDA-MB-231 and MCF-7, with different p53 statuses. MCF-7 breast cancer cells express WT p53, whereas MDA-MB-231 cells possess a mutated form of TP53 (R280K). In MDA-MB-231 cells, we observed a high level of p21WAF1/CIP1 until day D1+19, when the majority of cells were again proliferating; the escapers, however, reverted to the state of non-detectable p21WAF1/CIP1. The R280K mutation affects the DNA-binding domain [50
] and results in the decreased activation and repression of p53 target genes, including CDKN1A [51
]. Although we observed p53 phosphorylation, as well as phosphorylation of its upstream activator, ATM, the signal could not be transduced downstream from p53. Thus, the upregulation of p21WAF1/CIP1 observed by us was p53 independent, which is what we previously observed in dox-treated p53-deficient colon cancer cells [52
Dox-treated MDA-MB-231 cells had a very high number of γH2AX foci, sensing DSBs, and a relatively low number of 53BP1 foci. Both proteins, involved in the DDR, are markers of senescence and take part in homologous recombination (HR), an error-free process dependent on properly functioning autophagy [53
]. A higher number of 53BP1 than γH2AX foci was considered as a marker of the initiation of DNA repair [55
]. However, 53BP1 is also involved in alternative lengthening of telomeres (ALT) [56
], which has been shown to participate in the senescence/polyploidization of dox-treated MDA-MB-231 cells [29
]. Thus, insufficient autophagy may be the cause of the inability of the MDA-MB-231 cells to repair DNA lesions. However, the Ku70- DNA-dependent protein kinase (DNA PKs) axis, belonging to the non-homologous end joining (NHEJ) pathway, performs DNA repair independently from autophagy [53
]. In MDA-MB-231 cells, the protein level of Ku70 was slightly upregulated on days D1+4 and D1+9 (Figure 2
a), suggesting an active DNA repair process and possibly telomere stabilizing ALT [29
]. However, NHEJ is an error-prone mechanism of DNA repair. It was documented that damaged DNA is sorted to the cytoplasm, and probably digested in the process of active autophagy [29
]. The very low autophagic index we observed suggested to us abortive autophagy, but does not exclude, however, the active selective autophagy of damaged DNA in senescent/polyploid cells. Nonetheless, this issue needs more studies.
The crucial question concerns the molecular and cellular mechanisms of polyploidization/depolyploidization of breast cancer cells and the phenotype of the escapers. It seems that the paper already published in a Special Issue finally confirmed the role of mitotic slippage in reversible polyploidization in cancer cells induced to senescence. Moreover, the results of that study refreshed the old idea about the role of recapitulation in the amoeba-like agamic lifecycle, decreasing the mutagenic load and enabling the recovery of recombined, reduced progeny for a return to the mitotic cycle [29
]. Using a time-lapse, we were able to confirm atypical divisions of giant MDA-MB-231 cells. Furthermore, we also proved that the escapers were not cells that simply avoided senescence/polyploidy and restarted cell division after the drug withdrawal. Live cell images and movies support our claim that escapers are derived from giant polyploid/senescent cells.
The interesting question emerges whether reversible senescence leads to the production of progeny with a different phenotype than parental cells. To date, published data have mainly focused on the differences in stemness and aggressiveness between mother cells and their progeny [4
The main question we posed in these studies concerned the role of autophagy in breast cancer cells undergoing reversible senescence/polyploidization. In particular, there is a plethora of evidence showing that the majority of conventional therapies used to combat breast cancer induce autophagy [61
]. As mentioned previously, Erenpreisa’s group suggested autophagy as a crucial factor in depolyploidization [20
] and the emergence of vital daughter cells via mitotic slippage [21
]. Recently, Jakhar et al. documented that postmitotic slippage, leading to tetraploidy formation in cancer cells, depended on autophagy induction. Furthermore, the pharmacologic inhibition of autophagy or the silencing of an autophagy-related gene, ATG5, led to a bypass of G1 arrest senescence, reduced SASP-associated paracrine tumorigenic effects and increased DNA damage after S-phase entry with a concomitant increase in apoptosis [62
]. Similarly, Was et al. showed that the pharmacological inhibition of autophagy (by bafilomycin A1) in cancer cells induced to senescence increased cell death, but they claimed that polyploid/senescent cells were resistant to bafilomycin A1 treatment [63
]. Moreover, it seems that escaping from senescence needs autophagy reactivation. We showed that senescent/polyploid cells with a very low autophagic index gave rise to either progeny with autophagic indexes that were relatively high (in MDA-MB-231 cells) or similar to the control (in MCF-7 cells). We suggest that the senescence of breast cancer cells is intertwined with insufficient autophagy, whereas autophagy activation is indispensable for the appearance of vital progeny.
Indeed, we have found that general autophagy, estimated by the autophagic index, was impaired during the senescence/polyploidy of breast cancer cells. On the other hand, the transient increase of phosphorylated AMPK and the decrease in phosphorylated mTOR observed in MDA-MB-231 cells suggested autophagy induction, but it was not accompanied by a decrease in mTOR-dependent ULK1/2 phosphorylation (involved in autophagy induction [64
]). Altogether, this could suggest that, in senescent/polyploid cells, contradictory processes can be involved in autophagy regulation. Namely, the activation of signals for autophagy induction and the impairment of autophagic flux, leading to a sort of abortive, poorly functioning autophagy process. However, this does not necessarily exclude the activity of selective autophagy, especially in the process of the elimination of damaged DNA, which was confirmed by a low, but still present, autophagic flux. Additionally, in the case of insufficient autophagy degradation, autophagic vesicles with cargo can be removed by extrusion. Moreover, the autophagic index was measured in the entire population of senescent/polyploid cells and the presence of a limited population of cells with high autophagic flux, which produced the progeny with fully functional autophagy, cannot be excluded. Nevertheless, we can conclude that the appearance of escapers in the population of senescent cancer cells is associated with mTOR phosphorylation, however, without the transduction of signal to its substrate, which may be a cause of TFEB translocation to the nucleus. Moreover, the level of the endogenous autophagy inhibitor, Rubicon, also showed transient changes in dox-treated breast cancer cells. The significant increase in Rubicon in senescent cells and then its decrease to or below the level observed in control cells clearly revealed autophagic flux blockage followed by its resumption during escape from senescence.
The relationship between autophagy and cell senescence seems to be complicated, especially in senescent cancer cells as, generally, cancer cells have different basal levels of autophagy, including high and insufficient autophagy [65
]. Indeed, MDA-MB-231 cells have a very low autophagic index in comparison to MCF-7 cells and normal fibroblasts [28
]. Interestingly, the inhibition of autophagy can promote [66
], delay [67
] or alleviate senescence [68
]. However, our results demonstrated that, regardless of basal autophagic flux, during dox-induced senescence/polyploidization, autophagy was impaired and then increased in escapers. The autophagic index was elevated to the level observed in parental cells in MCF-7 cells or significantly higher in the case of MDA-MB-231 cells.
Moreover, we showed that MDA-MB-231 and MCF-7 parental cells, as well as escapers, underwent drug-induced senescence, although with different efficiencies. That allowed us to conclude that autophagy is not indispensable for senescence induction in MDA-MB-231 cells (with mutated p53 and a very low autophagic flux) and MCF-7 cells (with wild-type p53 and a high autophagic flux). Importantly, our results allowed us to conclude that unlocked autophagy was necessary to release descendants. Furthermore, we proved that the escapers were different from parental cells in terms of autophagy functionality, which is an obvious novelty of our studies. It is worth mentioning that the first study that pinpointed the involvement of autophagy in depolyploidization was published by Erenpreisa’s group [69
]. The authors hypothesized that autophagy was responsible for the degradation of chromosome bridges and the release of daughter cells, due to the distribution of cathepsin B in the central cytoplasmic area between subnuclei.
In short, our studies suggest that breast cancer cells can undergo drug-induced senescence, independently from the autophagy status. Furthermore, senescence is intertwined with insufficient autophagy. In turn, transient senescence ensured favorable conditions for the appearance of polyploid cells. However, the appearance of vital progeny is interconnected with functional autophagy. The mechanism of that phenomenon still needs to be unraveled. Although escapers had a similar DNA index to parental cells, they were characterized by a different phenotype. We are the first group to report that reversible polyploidization, intertwined with senescence escape, stably activates autophagic flux due to TFEB translocation to the nucleus and the reduction of the autophagy inhibitor, Rubicon.
4. Materials and Methods
Primary antibodies are listed in Table 1
. Secondary antibodies: anti-rabbit Alexa 488, anti-mouse Alexa 488, anti-rabbit Alexa 555, anti-mouse Alexa 555 from Life Technologies, (Carlsband, CA, USA), (1:500; A11008, A110296, A21428 and A21422) and anti-guinea pig Alexa 594 from Jackson ImmunoResearch (Cambridgeshire, UK), (1:500; 103-605-155).
4.1. Cell Culture and Treatment
MDA-MB-231 (HTB-26) cells, obtained from the European Collection of Authentic Cell Cultures (ECACC, Wiltshire, UK), were kindly provided by Prof. Jekaterina Erenpreisa (Latvian Biomedical and Research Centre, Riga, Latvia). MCF-7 cells (HTB-22) were purchased from the American Type Culture Collection (ATCC). Cells were grown under standard conditions (37 °C, 5% CO2) in DMEM low-glucose (MCF-7; Sigma-Aldrich, St. Louis, MO, USA, D5546) or DMEM high-glucose (MDA-MB-231; Biowest, Nuaillé, France, L0104) medium supplemented with 10% fetal bovine serum (FBS) (Cytogen, Zgierz, Poland, S181H), antibiotic–antimycotic solution (Sigma-Aldrich, St. Louis, MO, USA, A5955) and, in the case of, DMEM low-glucose medium, 2 mM l-Glutamine solution (Sigma-Aldrich, St. Louis, MO, USA, G7513) was added. The cells were seeded 24 h before treatment at a density of 1 × 104 cells/cm2. To induce senescence, cells were treated with 100 nM dox (IC30; Sigma-Aldrich, St. Louis, MO, USA, D1515) for 24 h and then cultured in fresh medium without the drug for several days (1 + n). Every third day, the medium was replaced by a fresh one. The escaper cell line was established after a monthly culture of small cells collected on D1+19 for MDA-MB-231 cells and on D1+13 for MCF-7 cells in four independent experiments.
4.2. Western Blotting Analysis
Alive, adherent cells were harvested and subjected to the procedure described previously [11
Staining was performed as described in [70
]. According to requirements, F-actin was stained by additional incubation with phalloidin (1:50; Thermo Fisher Scientific, Waltham, MA, USA, A12379) for 30 min.
4.4. Detection of Senescence-Associated β-galactosidase
Staining was performed according to Dimri et al. [31
4.5. Lipofuscin Staining
To detect lipofuscin accumulation, SenTraGor (Arriani Pharmaceuticals, Athens, Greece, AR8850020) staining was performed as described by Evangelou et al. [72
4.6. Lipid Staining
Cells grown on coverslips were fixed in 4% paraformaldehyde (PFA) (Sigma-Aldrich, St. Louis, MO, USA, P6148) for 15 min at room temperature. Then, cells were washed with ddH2O, dehydrated (5 min incubation in 60% isopropanol) and stained with 0.3% Oil Red O (Sigma-Aldrich, St. Louis, MO, USA, O-0625) in 60% isopropanol for 7 min. The Oil Red O solution was discarded, stained cells were washed with ddH2O and covered with mounting medium.
4.7. Cytokine Measurement
To assess the secretion of IL-8, IL-6 and VEGF proteins, the culture medium was collected and subjected to analysis by the DuoSet ELISA Development Kit, according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN, USA, DY208-05, DY206-0, DY293B-05). Absorbance was measured using a Tecan Sunrise microplate reader (Tecan Group Ltd., Männedorf, Switzerland).
4.8. DNA Content Evaluation by Toluidine Blue Staining
For DNA content analysis, stoichiometric DNA staining with toluidine blue (TB; Fisher Scientific, Waltham, MA, USA, T161-25) was performed as described previously [73
]. As a reference for locating normal 2C peaks, the nuclei of leukocytes were used. The estimated integral error of the method was lower than 10%.
4.9. Bromodeoxyuridine Incorporation Assay
DNA synthesis assay was performed as described previously [52
4.10. Cell Size and DNA Index Estimation by Flow Cytometry Analysis
Cell size and cell granularity were determined with the use of flow cytometry as described previously [52
]. DNA index was determined with the use of peripheral blood mononuclear cells as a diploid reference.
4.11. TEM Sample Preparation
Cells growing on a 35-mm glass bottom dish (MatTek, P35G-1.5-14-CGRD) were fixed with 2% paraformaldehyde (Sigma Aldrich, P6148) and 1% glutaraldehyde (EMS, EM grade) in 0.2 M HEPES pH 7.3 and prepared for electron microscopy according to a published protocol, with minor changes [74
]. Briefly, cells were post-fixed with 1% aqueous solution of osmium tetroxide (Agar Scientific, AGR1023) and 1.5% potassium ferrocyanide (Sigma Aldrich, St. Louis, MO, USA, P3289) in PB for 30 min on ice. Then, samples were immersed in 1% aqueous thiocarbohydrazide (Sigma Aldrich, St. Louis, MO, USA, #88535) for 20 min, post-fixed with 2% aqueous solution of osmium tetroxide for 20 min (all at room temperature) and incubated in 1% aqueous uranyl acetate at 4 °C overnight. The next day, samples were exposed to 0.66% lead aspartate for 30 min at 60 °C, dehydrated with increasing dilutions of ethanol, infiltrated with Durcupan resin (Sigma Aldrich, St. Louis, MO, USA, #44610), embedded using a BEEM capsule according to a published protocol [75
] and hardened at 70 °C for at least 72 h. The resin blocks were cut with an ultramicrotome (ultracut R, Leica) and ultrathin sections (70 nm) were collected on formvar-coated copper grids, mesh 100 (Agar Scientific, AGS138-1).
4.12. Population Doubling (PD)
To determine the cell number parameters, cells were counted with a Neubauer camera by trypan blue dye exclusion. The quantification of PD as a measure of cell growth for each cell line was carried out on the basis of the total number of viable cells.
4.13. Double Staining
The double staining method with Hoechst 33342/PI [76
] was used in the investigation of the effect of doxorubicin on the MDA-MB-231 cells.
4.14. Autophagic Index
To quantify autophagic flux in the subsequent days following treatment, we calculated the autophagic index (AI). This was achieved by establishing the difference between the ratio of the LC3B II/LC3B I protein level (ΔLC3B) of cells treated with 200 nM bafilomycin A (Sigma-Aldrich, St. Louis, MO, USA, B1793) or 50 μM chloroquine (Lab Empire, Rzeszów, Poland, CHL919) for the last 3 h of culture and untreated ones and normalizing it to the ΔLC3B of untreated ones according to the following equations: for bafilomycin A, AIBAF = (ΔLC3BBAF − ΔLC3BNT)/ΔLC3BNT and for chloroquine, AICQ = (ΔLC3BCQ − ΔLC3BNT)/ΔLC3BNT.
4.15. Live Imaging
To pinpoint the origin of escaper cells, live imaging techniques for the division of giant polyploidy MDA-MB-231 cells were employed. Two independent techniques were used: a holographic microscope, HoloMonitor4 (LabSoft, Warsaw, Poland) and a spinning disc confocal microscope Zeiss Axio Observer Z.1 Inverted Microscope (Zeiss, Oberkochen, Germany) with Yokogawa CSU-X1 Spinning Disc (Yokogawa, Tokyo, Japan), with objective: C APO 40x/1.20 Water. Film acquisition took 3–5 days, time between each frame: 15–30 min.
4.16. Image Acquisition
Immunofluorescence (IF) specimens were visualized either with a Nikon Eclipse Ti (Tokio, Japan), a fluorescent microscope with a 40×/0.6 Nikon lens, or a confocal laser scanning microscope, Leica TCS SP8 (Wetzlar, Germany), usingHC PL APO CS2 63x/1.40 Oil immersion lens. Electron microscope (EM) specimens were imaged with a transmission electron microscope, JEM 1400 (JEOL Co., Tokyo, Japan, 2008), equipped with a 11 megapixel TEM camera MORADA G2 (EMSIS GmbH, Münster, Germany). Digital images of nuclei stained with toluidine blue were collected using a Sony DXC 390P color video camera.
4.17. Quantitative Analysis
Computational analyses of cell, foci number and fluorescence intensity were performed using ImageJ (FiJi) software. The ratio of the intensity of TFEB fluorescence in the nucleus versus the cytoplasm was measured as corrected total cell fluorescence (CTCF), according to the equation: CTCF = integrated density of selected area (nucleus/cytoplasm) − (selected area (nucleus/cytoplasm)) × mean fluorescence of background readings).
DNA content was measured as the integral optical density (IOD), using Image-Pro Plus 4.1 software (Media Cybernetics, Rockville, MD, USA). More than 50 cells were counted per sample in each analysis.
4.18. Statistical Analysis
Sample size was chosen according to previous observations, in which similar experiments were performed in order to see significant results, in this case with heterogeneous biological replicates. Therefore, all presented data concerning parental and escapers cells are shown as an average of at least three or four independent experiments. All biological replicates without exclusion were used to perform statistical analysis. In the case of bar graphs, the error bars represent the SEM, whereas in ANOVA graphs, the vertical bars indicate a 0.95 confidence interval. Statistical analysis was performed with the use of the STATISTICA 11 program (TIBCO Software Inc., Palo Alto, CA, USA) or GraphPad Prism 8 (San Diego, CA, USA). ANOVA (analysis of variance), analysis was used for the analysis of differences among three or more groups, followed by post hoc analysis (Tukey’s honest significant difference test; HSD test). Multiple comparisons were done after a homogeneity test for variance. Variance was similar between the groups that were being statistically compared. Normal distribution of the data was tested with a Shapiro–Wilk test. Statistical significance in relation to the control is marked with an asterisk (* or $), whereas that between subsequent days of treatment is shown with a hash (#). The p-value is stated as: $ p < 0.051, * 0.01 < p < 0.05, ** 0.001 < p < 0.01, *** p < 0.001.