Background
Ovarian cancer is the fifth most frequent cause of cancer death in women. It is estimated that 21,410 new cases and 13,770 deaths occurred in the United States in 2021 [
1]. Ovarian cancer is a heterogenous disease that is classified into serous, endometroid, mucinous, and clear cell subtypes based on distinct histology and genetics. Ovarian high-grade serous carcinoma (HGSC) is the most common and deadly histotype and is responsible for approximately 70% of ovarian cancer cases and deaths [
2,
3]. This most lethal female reproductive cancer is nicknamed the “silent killer” as patients are frequently diagnosed at advanced stages with metastatic disease [
4]. The standard-of-care for HGSC patients includes cytoreductive surgery and chemotherapy, usually carboplatin and paclitaxel, but many patients experience platinum-resistant relapse and 5-year survival rates are less than 50% [
1]. While cancer research has seen profound progress in many areas over the past five decades, only marginal increases in overall- and disease-free survival in ovarian cancer patients have occurred and improved therapies are critically needed.
Ploidy is more flexible in cancer cells than non-transformed cells. HGSC tumors are triploid or higher in whole-genome average ploidy in 53–56% of cases [
5,
6]. Single-cell sequencing has revealed untreated on-average diploid HGSC tumors exhibit 2–4% of epithelial cells in a triploid or higher ploidy [
7]. Ploidy gains can be induced in cancer cells, often by chemotherapy or other forms of stress. Following a lethal dose of chemotherapy, most cells in a population undergo cell death, but some cells are able to enter a quiescent, therapy-induced state and survive. Remarkably, many of these cells have been observed in histological sections to be polyploid, including in ovarian cancer [
8]. Such polyploid cells may contain either a single much-enlarged nucleus or an amalgamation of diploid or larger sized nuclei. Polyploid giant cancer cells (PGCCs) are defined as cancer cells with tetraploid or higher ploidies (4 N) that express markers of or have properties of stem cells [
9‐
12].
PGCCs exhibit unique life cycle characteristics which implicate their important roles in chemoresistance and tumor evolution. Polyploidy initially forms by a variety of mechanisms which can include cell–cell fusion or endoreplication (duplication of the genome without mitosis). Chemotherapies such as DNA-damaging platinum agents and microtubule-stabilizing taxanes induce formation of PGCCs. These PGCCs are temporarily arrested in the cell cycle and express senescence markers such as p21 [
13]. After a period of days to weeks, PGCCs re-enter the cell cycle and repopulate the tumor with drug-resistant progeny [
14,
15]. This can occur partially through symmetric division of polyploid cells, but more substantially occurs via asymmetric budding of lower-ploidy daughter cells from the originating PGCC, which then re-enter the cell cycle. The latter daughter-budding process is termed “neosis,” which we adopt here [
11]. As the PGCC progeny resemble the original parental cells, the entire process of PGCC formation and subsequent progeny generation is referred to as the PGCC life cycle [
9,
10]. PGCC progeny are resistant to the therapies that originally induced their formation, recapitulating the development of drug-resistant cancers [
12,
13,
16]. Progeny of ovarian cancer PGCCs have highly variable karyotypes, providing a source of genetic diversity which may enable the evolution of chemoresistance [
17].
Aneuploidy is unusually high in HGSC. HGSC has ~ 16,000 genes altered in dosage by copy number alterations (CNAs) in the median tumor due to a high degree of aneuploidy and focal (sub chromosome arm-level) copy number alterations [
18]. Specifically, the tumor suppressor p53 is mutated in essentially all (96%) HGSCs [
19,
20], enabling aneuploid cells to survive. Using genetic pathway analyses of HGSC CNAs, we discovered the autophagy cellular recycling pathway is the most downregulated pathway by CNA losses with 98% of tumors having multiple heterozygous deletions of autophagy genes. Yet, autophagy remains critical for these cancer cells. Autophagy is a stress response mechanism required for drug resistance in ovarian cancer [
21‐
23]. Autophagy is upregulated during the formation of PGCCs [
24,
25].
We previously discovered that autophagy is a targetable vulnerability, as drugs disrupting autophagy killed both chemo-sensitive and chemo-resistant ovarian cancer cells
in vitro and
in vivo [
18,
26]. HGSC growth was inhibited by autophagy inhibitors, chloroquine or nelfinavir, as well as autophagy inducers, such as the mTORC1 inhibitor rapamycin. However, neither we nor the PGCC field has examined whether autophagy drugs impinge on the life cycle of PGCCs. We hypothesized that autophagy-modulating-therapeutics may interfere with the chemotherapy-induced PGCC life cycle in ovarian cancer cells. Here, the impact of these autophagy modulators on chemotherapy-induced PGCC formation and neosis by PGCCs was investigated.
Methods
Cell culture
CAOV3 and OVCAR3 human ovarian cancer cell lines were from ATCC and were cultured in RPMI-1640 supplemented with L-glutamine, 10% fetal bovine serum, sodium pyruvate, and penicillin–streptomycin. Cells were incubated at 37 °C with 5% CO2.
Chemotherapy-induced polyploid giant cancer cell induction and neosis
Carboplatin- and docetaxel-induced PGCC formation and subsequent daughter cell formation were studied over the span of 14 days. CAOV3 cells were seeded at 100,000 cells/mL and OVCAR3 cells were seeded at 250,000 cells/mL, 24 h later cells were treated with 10 µM carboplatin or 5 nM docetaxel for 3 days, followed by 3 days of recovery. For experiments testing effects of autophagy-targeting therapeutics on PGCC development, cells were treated with 33 µM hydroxychloroquine, 10 µM nelfinavir, or 10 nM rapamycin alone or concurrently with carboplatin or docetaxel, and after 3 days of drug treatments and 3 days of recovery, cells were fixed, stained, imaged, and nuclear content was quantified as described below. For studies of daughter cell formation by PGCCs, on day 7 PGCCs were isolated based on size-exclusion with pluriSelect™ cell strainers of 30 µm for CAOV3 cells and 10 µm for OVCAR3 cells. Then cells were re-plated, allowed to rest for 24 h, and treated with 33 µM hydroxychloroquine, 10 µM nelfinavir, or 10 nM rapamycin for a total of six days with a media change containing fresh drugs in the middle. Finally, colonies which arose from PGCCs were fixed, imaged, and quantified through crystal violet staining as described below.
Nuclear Quantification
DNA staining with Hoechst 33342 was used to quantify changes in CAOV3 and OVCAR3 cell nuclear content. Specifically, cells were fixed with ice-cold methanol at -20 °C for 7 min, permeabilized with 0.1% Triton X-100 for 2 min, blocked with 5% bovine serum albumin (BSA) / 5% goat serum in phosphate buffer saline (PBS) at room temperature for 45 min, and incubated with the primary antibody mouse anti-E-Cadherin (BD, #610182) overnight. Secondary anti-rabbit Alexa Fluor 594 (Fisher Scientific) was used at 1:1,000 and Hoechst 33342 (Fisher Scientific, #A11029) was used at 1:10,000 and were diluted into 5% BSA / 5% goat serum and incubated for 90 min. The immunofluorescent cells were then imaged using the Lionheart FX automated microscope (BioTek) and NIH ImageJ (Fiji) software was utilized in addition with a custom macro to measure nuclear area and intensity using Hoechst 33342 staining. Fifty representative cells were counted in each of two independent experiments, and the data were normalized and aggregated. The median nuclear area X intensity of the control CAOV3 and OVCAR3 cells was designated as “normal ploidy”, and to exclude cells undergoing normal mitotic processes (normal—2X normal ploidy), a threshold DNA content ≥ 4.5X normal ploidy was used to classify cells as PGCCs. Using the total number of cells classified as “normal” or PGCCs, contingency tables were generated, and Fisher’s exact tests were conducted to test for significant differences between treatment groups.
Western blotting
Western blotting was performed as described previously [
18] to confirm that autophagy-targeting therapeutic treatment affected the expression of the autophagy markers GRP78 and LC3B-II. As above, CAOV3 cells were treated with 33 µM hydroxychloroquine, 10 µM nelfinavir, or 10 nM rapamycin alone or concurrently with carboplatin or docetaxel for 72 h, then cells were lysed in ice-cold RIPA buffer supplemented with a protease inhibitor cocktail (Sigma-Aldrich). After centrifugation at 10,000 g for 10 min at 4 °C, protein concentration in the supernatants was quantified by bicinchoninic acid assay (BCA; Pierce #23235). For each sample, 30 µg was resolved on 4 – 20% gradient polyacrylamide gels (Biorad #4561093), transferred to nitrocellulose membranes, blocked using 5% milk in PBS, and incubated overnight with β-actin (Thermo Fisher #MA515739), GRP78 (Cell Signaling #3177), or LC3B (Novus Biologicals #NB100-2220) primary antibodies at 1:1000 dilutions. Horseradish peroxidase conjugated goat anti-mouse (Sigma #12–3349) and goat anti-rabbit (VWR #100244–772) secondary antibodies were incubated in TBST at 1:5000 dilutions for 45 min. Enhanced chemiluminescence (ECL, Biorad #1705060), a Chemidoc Imaging System (Biorad), and NIH ImageJ software were used to visualize the results.
PGCCs were isolated after three days of chemotherapy treatment followed by three days of rest. Specifically, CAOV3 and OVCAR3 cells were trypsinized and filtered through 30 µm size-exclusion cell strainers for CAOV3 cells (pluriSelect, #43-50030-03) and with cell strainers of 10 µm size-exclusion for OVCAR3 cells (pluriSelect, #43-50010-03). After rinsing the strainers with 10 mL of media, the cell strainers were inverted and PGCCs were gathered and re-plated. After one week of colony outgrowth, CAOV3 and OVCAR3 cells were fixed in methanol, stained with 0.2% crystal violet in PBS at room temperature for 20 min, washed twice with PBS, and representative brightfield images were acquired with a Lionheart FX microscope (BioTek). After imaging, crystal violet was resuspended in 10% glacial acetic acid and absorbance at 600 nm was read in an Epoch 2 spectrophotometer (BioTek).
Statistical analysis
For the data in PGCC formation assays, statistical significance was calculated using Fisher’s exact tests. For the data in colony quantitation assays, a two-tailed, Student’s t-test was used to calculate statistical significance. P < 0.05 was considered statistically significant.
Discussion
Two decades of research have well-established that autophagy plays a role in chemoresistance. Remarkably, none of these previous studies have tested if autophagy inhibitors modulate the PGCC life cycle, despite the clear formation of these cells following chemotherapy and radiotherapy. Although it is known that autophagy is elevated in PGCCs [
24,
25], to our knowledge this is the first study which directly examines the interaction of autophagy modulating drugs and PGCCs. We observed that while autophagy inhibitors do not prevent PGCC formation, autophagy inhibitors are able to reduce the amount of progeny which arise specifically from PGCCs. Given that PGCCs are a source of chemoresistant cancer cells and a source of random karyotype shuffling and therefore intra-patient genetic diversity, autophagy inhibitors may be promising to pursue in the clinic for HGSC patients exposed to carboplatin or docetaxel chemotherapy.
Exceptionally large cancer cells with large, abnormal nuclei have been described in the scientific literature since the 1850s (reviewed in [
32]), and PGCCs are present at low levels in many cancer cell lines and in virtually all types of cancer including HGSC [
8,
12,
13,
33,
34]. Cells with these abnormal morphologies were traditionally regarded as dying or irreversibly senescent, but there is growing recognition that some PGCCs are able to overcome senescent cell-cycle arrest and spawn near-diploid progeny, enabling cancer cells to survive senescence induced by therapy or other stresses encountered in the tumor microenvironment such as hypoxia or nutrient deprivation [
35‐
37]. Further, the reversible polyploidization process facilitates genome instability, which “underlies the hallmarks of cancer” [
38]. The PGCC life cycle enables the development of aneuploidy and the myriad copy number alterations accompanying that process [
34]. This genome shuffling yields karyotype diversity that is a substrate upon which selection acts during tumor evolution, especially during the development of drug resistance [
39].
Here, CAOV3 and OVCAR3 ovarian cancer cell lines were treated with chemotherapies to induce the PGCC life cycle: formation of PGCCs and subsequent neosis and colony formation by PGCCs. Autophagy-modulating therapeutics were added either during PGCC formation or during the time that PGCCs were producing progeny by neosis. We found that these autophagy-modulating therapeutics had minimal effects or actually increased the number of PGCCs that were formed in response to CPt and DTx. In marked contrast, treatment with autophagy-modulating therapeutics had a strong inhibitory effect on colony formation after PGCCs were already formed. Breast cancer PGCCs display elevated markers of autophagy LCB-II and p62/SQSTM1 but low autophagic flux, whereas progeny derived from these PGCCs have elevated rates of autophagy [
25]. Elevated autophagy during PGCC progeny generation may be necessary to rid cells of irreversibly damaged DNA, organelles, and proteins. The lack of an effect of autophagy-modulating therapeutics on chemotherapy-induced PGCC formation and the inability of these therapeutics to effect basal PGCC levels suggests that autophagy is not critical for PGCC generation. In contrast, during neosis autophagy is elevated and treatment with the autophagy-modulating therapeutics HCQ or NFV significantly decreased PGCC progeny survival.
The similar effect of Rapa on reduced colony formation post-PGCC formation can be interpreted in a few ways, based on previous literature. The first, autophagy-independent explanation would be that Rapa inhibits cell growth processes via mTORC1 inhibition, resulting in fewer daughter cells. Previous observations make this interpretation somewhat unlikely; cell growth inhibition using this same dose of rapamycin in a panel of ovarian cancer cells exhibited 10–30% reduction in growth rates [
18,
26], not the 52–84% inhibition of colony formation observed in the current study. However, autophagy-independent roles of mTORC1 may nonetheless be uniquely important after PGCCs have formed and start to re-seed tumors. The second interpretation is that Rapa creates a stress on autophagy, just as HCQ and NFV stress autophagy, so the similar effects would not be surprising. This is consistent with the observation that treatment with all three drugs results in HGSC accumulating aberrant vesicles and proteotoxic aggregates, as observed by electron microscopy [
18]. This model further explains why Rapa actually worsens cytotoxicity of chloroquine and NFV, rather than ameliorating cell death caused by their administration. Future studies are warranted to better understand the molecular mechanism of these observations.
Aneuploidy was first hypothesized to cause cancer over a hundred years ago, but the discovery of oncogenes and tumor suppressor genes led to a “gene-centric” view of the etiology of cancer [
40]. More recently, however, it has been suggested that a “genome-centric” view may be similarly appropriate [
39]. Chromosomal instability is a high rate of chromosome mis-segregation that gives rise to aneuploidy. Aneuploidy is a hallmark of cancer, and ~ 90% of solid tumors have some degree of aneuploidy at the level of whole chromosomes [
40,
41]. In addition, chromosome arm-level alterations and more focal copy number alterations are common in cancer and these chromosomal alterations and are included here in the term ‘aneuploidy.’ In most contexts, aneuploidy is associated with substantial fitness costs, but the pervasiveness of chromosomal instability and aneuploidy in cancers suggests that aneuploidies drive tumorigenesis, presumably by increasing genetic diversity. It is suggested that whole-genome duplications often precede the development of aneuploidy [
42‐
46]. Whole genome-duplications are present in 37% of all cancers and 53–56% of HGSC [
5,
6]. Both ovarian cancer cell lines used here, CAOV3 and OVCAR3, exhibit a hypotriploid karyotype, indicating that they likely evolved from a whole genome doubling event followed by chromosome loss [
47]. HGSC is characterized by extensive aneuploidies, and the degree of aneuploidy correlates with malignancy and poor prognoses [
48‐
50]. Due to extensive aneuploidy including focal copy number alterations, ~ two-thirds of genes are altered in dosage in a typical HGSC tumor [
18]. The autophagy pathway is the most downregulated pathway by copy number alterations in HGSC with 95% of tumors having multiple heterozygous deletions in at least four autophagy genes. In addition, loss of autophagy genes in HGSC is shown to cause genomic instability [
51]. Further, a cocktail of drugs including chloroquine, NFV, and Rapa which affect several nodes in the autophagy pathway simultaneously demonstrated remarkable efficacy in killing ovarian cancer cells in preclinical studies [
26]. By virtue of having reduced capacity for functional autophagy, it appears that ovarian cancer cells have a unique vulnerability to drugs targeting this pathway.
It is generally accepted that tumor cell populations evolve, and that intra-tumor genetic heterogeneity is one source of variation upon which selection acts. In addition to genetic diversity at the level of gene mutations, genetic diversity arises from genomic diversity caused by different aneuploidies, and tumor heterogeneity also manifests at the epigenetic and phenotypic levels. Initially in tumorigenesis, gradual clonal evolution of a population of cells with various oncogene or tumor suppressor gene mutations results in sustained proliferation and decreased responsiveness to DNA damage as well as other cancer hallmarks. Next, excessive endogenous stress in the tumor microenvironment or stress induced by chemotherapeutics may lead to chromosomal instability, causing extensive aneuploidies which change the expression levels of thousands of genes in one generation. These profound chromosomal alterations may enable rapid punctuated evolution of cancer genomes and be instrumental in tumor progression, including the development of chemo resistant relapsed disease. The polyploidization process may underlie the ability of cancers to evolve resistance to virtually all current therapies. Following genome doubling, more potential for beneficial mutations is created because extra intact copies of genes are available if an allele is mutated deleteriously. Further, there is more genomic material available to participate in DNA repair processes and having additional copies of genes can result in increased expression of proteins involved in stress responses. These attributes facilitate the survival and evolution of cancer cell populations.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.