The effects of anticancer treatments on cell heterogeneity and their proliferative potential play an important role in tumor persistence and metastasis. However, little is known about de-polyploidization, cell fate, and physiologic stemness of the resulting cell populations. Here, we describe a distinctive cell type termed "pregnant" P1 cells found within chemotherapy-refractory ovarian tumors, which generate and gestate daughter generation Gn cells intracytoplasmically. Release of Gn cells occurred by ejection through crevices in the P1 cell membrane by body contractions or using a funiculus-like structure. These events characterized a not yet described mechanism of cell segregation. Maternal P1 cells were principally capable of surviving parturition events and continued to breed and nurture Gn progenies. In addition, P1 cells were competent to horizontally transmit offspring Gn cells into other specific proximal cells, injecting them to receptor R1 cells via cell–cell tunneling. This process represents a new mechanism used by tumor cells to invade surrounding tissues and ensure life cycles. In contrast to the pregnant P1 cells with low expression of stem cell markers despite their physiologic stemness, the first offspring generations of daughter G1 cells expressed high levels of ovarian cancer stem cell markers. Furthermore, both P1 and Gn cells overexpressed multiple human endogenous retroviral envelope proteins. Moreover, programmed death-ligand 1 and the immunosuppressive domain of the retroviral envelope proteins were also overexpressed in P1 cells, suggesting effective protection against the host immune system. Together, our data suggest that P1 oncogenerative cancer cells exhibit a not yet described cell biological mechanism of persistence and transmission of malignant cells in patients with advanced cancers.

Significance: P1 oncogenerative cell entities express low levels of CSC markers, which are characteristic of their histological origin. Cancer Res; 78(9); 2318–31. ©2018 AACR.

Embryonic stem cells (ESC) derive from the undifferentiated inner mass of blastocysts. As pluripotent cells, they can differentiate into all derivatives of the three primary germ layers (1). Adult stem cells (ASC) are undifferentiated multipotent cells with the capacity of self-renewal and generation of fully differentiated daughter cells. They are virtually tissue resident, help maintain cellular homeostasis, and express high levels of ABC transporters. ASCs are capable of cell transdifferentiation, that is, generating differentiated cell types of other histologies. ASC transdifferentiation into pluripotent stem cells (iPSC) can be induced via Yamanaka's quadriga of transcription factors (2, 3).

The source of tumor cell heterogeneity is still under debate because the "cell of origin" has remained elusive. However, multiple conclusive lines of evidence suggest the existence of cancer stem cells (CSC) with characteristics that are normally associated with ASCs, including the capacity for long-term self-renewal and multipotency, which contributes to tumor cell heterogeneity, a hallmark of stem cells (4, 5). Despite their localized origin, CSCs can disseminate to distant sites. Another ASC-like characteristic is the overexpression of ABC transporters and DNA repair mechanisms, which play a critical role in untreatable malignancies. Moreover, under therapeutic treatment conditions, differentiated cancer cells may change their phenotype and alter their biology toward a more stem-like state to ensure survival (4–6).

A common phenomenon associated with cytostatic- and irradiation-induced cell heterogeneity is believed to be the emergence of multinucleated giant cell entities with CSC properties (7, 8). These cells can be generated via two different mechanisms: (i) endoreplication coupled with karyokinesis - endomitosis (plasmodium formation); and (ii) syncytin-mediated cell–cell fusion (syncytium formation). Endoreplication can be seen as a deviant form of the normal mitotic cell cycle in which mitosis is completely suppressed. Endomitosis is an abortive from of mitosis that does not result in cell division but leads to the formation of multinucleated giant cells (9, 10). Another multinucleated cell type arises from specific cell–cell fusion events that are chiefly mediated by the fusogenic properties of envelope proteins from exogenous or endogenous retroviruses (HERV), that is, HERV-WE1 and HERV-FRD1. Cell–cell fusion via syncytins may generate both normal (syncytium trophoblasts during placental development) and pathologic forms, as observed, for example, in high-grade Reed–Sternberg cells, chemoresistant tumors or after viral infection, for example, with the syncytial respiratory virus (11, 12).

Although the two mechanisms differ vastly at the physiologic level, the final fate of many giant cells is the reverted multinucleation, which leads to the generation of daughter cells, thus promoting tumor initiation and development (8, 13). The theory of cellularization states that in multinucleated cells, the nuclei may have developed internal perinuclear membrane partitions facilitating the formation of new cells. This process has been described in Drosophila melanogaster embryos (14). However, to the best of our knowledge, the molecular mechanism behind this process has not been studied in detail in cancer cells.

Here, we document a distinctive cell type termed “pregnant” P1 cells found within chemotherapy-refractory ovarian tumors that generate daughter G1 cells intracytoplasmically by cellularization, which, upon gestation, are released into the immediate surroundings. Strikingly, P1 cells can inject G1 cells into specific recipient cells (R1) via cell–cell tunneling, which represents a horizontal transmission of the entire genome between cells, resulting in new tetraploid cell entities.

P1 appears to be a distinctive tumor cell type with intriguing stem cell biology. Phenotypic characterization shows that P1 express very low levels of CSC markers in spite of their stem cell biology (15). Interestingly, this is contrasted by the first generation of progeny cells, which express a distinct signature of canonical CSC markers. Contrarily, P1 cells differentially produce more transcripts of several stem cell markers in comparison with their offsprings.

Both P1 and G1 cells overexpress different human endogenous retroviral (HERV) envelope proteins containing the immunosuppressive domain (ISD; ref. 16). The programmed death-ligand 1 (PD-L1) immune checkpoint was found to be overexpressed too, suggesting the cells may be able to circumvent immune attack in a way observed in both cancer and pregnancy (17). Thus, P1 cells may provide a sheltered immune- and therapy-resistant environment to guarantee the generation and gestation of new tumor cells. These findings represent an as-yet undocumented mechanism of conferring drug resistance and persistence to malignant cells and may provide a target for novel therapeutic strategies.

Cell lines, patient samples, and ethical considerations

All cells were obtained from the cell and tumor bank of the Marienhospital Herne (Herne, Germany). This study was reviewed and approved by the Ethics Committee of the Ruhr University of Bochum, Medical School (Herne, Germany; register numbers: 5235-15, 5416-15, 17-6114).

Generation of 3D spheroids from single P1 and G1 cells

To analyze P1 cell heterogeneity, we cultivated single P1 and G1 cells in hanging drops to enable clonal proliferation in nonadherent conditions and simulate the natural tumor architecture (18). Cells of different sizes (<80 μm) were picked using a TransferMan 4r micromanipulator (Eppendorf) and washed in 25 μL DMEM. Next, the cells were placed in approximately 25 μL on a dish lid, inverted onto the PBS-filled bottom chamber, and cultured in 5% CO2, 95% humidity, 37°C until morula-like spheres had formed.

Studies on the migration capacity of P1 and G1-Gn cells by Transwell migration assay

A Transwell migration assay was used to track and quantify the migration of P1 and offspring cells. Tissue culture inserts with a translucent 8 μm PET-membrane (Sarstedt) precoated with collagen types I/IV (Thermo Fisher Scientific) were used. A total of 2 × 104 ovarian carcinoma cells from ascites were deposited on the Transwell upper chamber, and DMEM containing 20% FBS was added as a chemoattractant to the lower chamber. After 24 hours, cells were fixed with TCA 10% at 4°C for one hour and stained with SRB 0.4% in 1% acetic acid for one hour. Depending on the side of the membrane analyzed, the stained cells on the other side were removed mechanically. Migration was documented using an inverted Nikon Eclipse TS100 microscope or quantified by immersing the membranes in Tris-HCl pH 10.5, measuring the dissolved proteins at 570 nm in a Tecan Infinite M200 microtiter plate reader. Different P1/G1-Gn cell surface structures like CXCR4, HERV-Fc1, HERV-KISD, and EpCAM were targeted using specific antibodies at 10 μg/mL for 24 hours to study their role in cell migration.

Differential expression of stem cell markers and immune checkpoints in P1 progenies as analyzed by qPCR

Total RNA purification and cDNA synthesis.

P1 progenies were grown in hanging drops and harvested to study the expression of CSC markers. Total RNA was dually extracted with Trizol (Life Technologies) and purified on RNeasy mini columns (Qiagen) according to the manufacturer's instructions (19).

Single-cell RNA isolation and cDNA amplification and real-time PCR.

The REPLI-g WTA Single Cell Kit (Qiagen) was used to generate and amplify cDNA from single cells following the manufacturer's instructions. Cells were obtained from ovarian ascites and sorted in Transwell chambers after migration in FCS gradients as reflected above. Briefly, 100 cells were picked and lysed for 5 minutes at room temperature. gDNA was removed prior to the WTA process. Transcripts were amplified using random and oligo-dT primers. The synthesized cDNA was ligated and then amplified by MDA technology, using the REPLI-g SensiPhi DNA polymerase, in an isothermal reaction for 2 hours. The amplified cDNA was quantified with a Pico Green dsDNA Reagent Kit from Invitrogen. cDNA was employed for qPCR using 100 ng per reaction.

CSC markers, HERV and PDL1 expression was measured by qPCR with validated primers and probes from Integrated DNA Technologies Inc. Sequence and polarity of the primers are shown in Table 1. 100 ng of RNA was amplified in triplicate in a CFX96 Real-Time System (Biorad Laboratories).

Table 1.

Real-time PCR primers and probes

GeneAccessionSense (5′–3′)Probe (5′–3′)Antisense (5′–3′)
18s NR_003286 GGACATCTAAGGGCATCACAG GGACATCTAAGGGCATCACAG GAGACTCTGGCATGCTAACTAG 
c-Myc NM_001354870.1 AGAGGCTTGGCGGGAAA AGGGAGGGATCGCGCTGAGTATAA CTGCCTCTCGCTGGAATTACTA 
KLF4 NM_001314052.1 AGGAGCCCAAGCCAAAG TAATCACAAGTGTGGGTGGCGGTC GCCTTGAGATGGGAACTCTT 
Nanog NG_004093.3 CAGGACAGCCCTGATTCTTC CAGTCCCAAAGGCAAACAACCCAC GTTTCTTGACCGGGACCTT 
Oct4 NM_002701.5 TATGGGAGCCCTCACTTCA TACTCCTCGGTCCCTTTCCCTGAG TCAGTTTGAATGCATGGGAGA 
Sox2 NM_003106.3 CGGACAGCGAACTGGAG AGAGGAGAGTAAGAAACAGCATGGAGA TTTGAGCGTACCGGGTTT 
EpCAM NM_002354.2 GGTGATGAAGGCAGAAATGAATG CCCATCATTGTTCTGGAGGGCC TCATCGCAGTCAGGATCATAAAG 
HERV-Fc1 AJ507128.1 GCCTCATCAGTCCTTTCAGATT TCGCCCAGAGAATGGACAAGCAT AAACACCAGGGACAGCTTATC 
HERV-KISD DQ360584.1 TGTTCTAAGCTCATGAGTCTGTCT TGATCTTAGACAAACTGTCATTTGGATGGG CAGTCAGTAAACTTTGTTAATGATTGGC 
PDL1 AY291313.1 GCTGTCTTTATATTCATGACCTACT ACGCATTTACTGTCACGGTTCCCA TGTCATATTGCTACCATACTCTACC 
GeneAccessionSense (5′–3′)Probe (5′–3′)Antisense (5′–3′)
18s NR_003286 GGACATCTAAGGGCATCACAG GGACATCTAAGGGCATCACAG GAGACTCTGGCATGCTAACTAG 
c-Myc NM_001354870.1 AGAGGCTTGGCGGGAAA AGGGAGGGATCGCGCTGAGTATAA CTGCCTCTCGCTGGAATTACTA 
KLF4 NM_001314052.1 AGGAGCCCAAGCCAAAG TAATCACAAGTGTGGGTGGCGGTC GCCTTGAGATGGGAACTCTT 
Nanog NG_004093.3 CAGGACAGCCCTGATTCTTC CAGTCCCAAAGGCAAACAACCCAC GTTTCTTGACCGGGACCTT 
Oct4 NM_002701.5 TATGGGAGCCCTCACTTCA TACTCCTCGGTCCCTTTCCCTGAG TCAGTTTGAATGCATGGGAGA 
Sox2 NM_003106.3 CGGACAGCGAACTGGAG AGAGGAGAGTAAGAAACAGCATGGAGA TTTGAGCGTACCGGGTTT 
EpCAM NM_002354.2 GGTGATGAAGGCAGAAATGAATG CCCATCATTGTTCTGGAGGGCC TCATCGCAGTCAGGATCATAAAG 
HERV-Fc1 AJ507128.1 GCCTCATCAGTCCTTTCAGATT TCGCCCAGAGAATGGACAAGCAT AAACACCAGGGACAGCTTATC 
HERV-KISD DQ360584.1 TGTTCTAAGCTCATGAGTCTGTCT TGATCTTAGACAAACTGTCATTTGGATGGG CAGTCAGTAAACTTTGTTAATGATTGGC 
PDL1 AY291313.1 GCTGTCTTTATATTCATGACCTACT ACGCATTTACTGTCACGGTTCCCA TGTCATATTGCTACCATACTCTACC 

FACS analysis of cell cycle and polyploidic populations

Cell-cycle distributions were analyzed combining propidium iodide (PI; Sigma-Aldrich) staining and 5-bromo-2′-deoxyuridine (BrdUrd) incorporation, as described previously (20). BrdUrd incorporation and DNA content were measured using the CytoFLEX Research Cytometer B5-R0-V0 (Beckman Coulter Biosciences).

Immunocytochemical and IHC staining

Immunocytochemical (ICC) and IHC staining was performed according to standard protocols (21).

Primary antibodies for CSC-phenotyping as for CD24, CD44pan, CD133, SCF, and vimentin were purchased from Cell Signaling Technology; ESA or Ep-CAM/TROP1 from R&D Systems; CD44 variants from Bio-Rad; primary antibodies for the detection of HERV-WE1, HERV-FRD1, HERV-V3.1 from Biorbyt and for HERV-K form BioSS. The anti-ISD was generated using a 17 amino acid consensus fragment corresponding to the ISD of different envelope proteins of the HERV-K family. Conjugated secondary antibodies were purchased from Cell Signaling Technology.

Preparation of ultrathin tumor sections and transmission electron microscopy

Samples for transmission electron microscopy were prepared according to our unmodified protocol (7). A Zeiss transmission electron microscope (EM 902A) was used applying 80 kV at magnifications from ×3,000 to ×140,000. Digital images were taken with a MegaView II slow-scan CCD camera and aITEM 5.0 software (Soft Imaging Systems).

Morphologic studies of transmigrating P1 and G1-Gn cells by scanning electron microscopy

Cells on Transwell membranes were fixed using glutaraldehyde 2.5% in PBS for 12 minutes at room temperature, maintained in PBS, and then dehydrated in increasing concentrations of ethanol (1× in 25%, 50%, 75%, 95%, and 3× in anhydrous ethanol, 10 minutes each). The samples were subjected to critical point drying in a CO2 atmosphere to minimize shrinkage, mounted on a holder using LeitC-Tabs adhesive, and sputtered with gold in vacuum using an Edwards Sputter Coater S150B. Samples were scanned in a FEREM DSM 982 Gemini electron microscope (Zeiss) using an electron beam of 15 kV.

Statistical analysis

All experiments were performed at least in triplicate. Intergroup comparison of medians was performed by ANOVA. Statistical analysis was performed with Sigma Plot 12 (Systat Software Inc.). Significance was accepted when P < 0.05. Electron microscopy, Western blot, ICC, IHC, and video microscopy studies were descriptive and therefore not analyzed statistically; the results shown are representative of at least three independent experiments.

P1 cells are endoreplicative entities with virus-like life cycles

In this study, we characterized a distinctive cell type termed pregnant (“P1”) cell and hypothesized it to be the primordial generative cell type. Figure 1A–I and Supplementary Videos S1 and S2 illustrate the morphology of P1 cells and how they give rise to the first generation of progeny cells (“G1”). P1 are hypertrophic mono- and binucleated (in very early stages) or multinucleated (later stages) cells that can reach diameters of more than 200 μm in adherent cultures. The P1 cell type was found both in tumors of different histologies and in primary cell cultures (Supplementary Fig. S1A). In untreated patients, we found that about 1% of all cells are P1, depending on the cellularity grade of the ascites. In pretreated and mostly chemoresistant cancers, viable P1 cells were increased by 2% to 3%, indicating an influence of chemotherapeutics on the selection/induction (Supplementary Fig. S1B).

Figure 1.

Cellularization events linked to the generation of daughter cells by oncogenerative P1 cells. AI, P1 cells are giant cell entities with the ability to breed daughter (G1) cells intracytoplasmically. The release of G1 cells from P1 starts with the formation of crevices through which the G1 cells are ejected via mechanical contraction of P1′s cell body, leaving an aperture that will be repaired afterward. The connection between P1 and G1 is sustained for some time by a funiculus cellularis (Fu), analogous to placental organisms. The sequence of pictures was extracted from Supplementary Video S2. JL, Ultrathin structure of P1 and G1 in ovarian carcinoma tissues. J and K, P1 oncogenerative cell that holds several G1 progeny cells. Note the aperture left by G1 cells once released from P1 (K). The separation of G1 cells from the P1 parental cell starts with the formation of crevices through which G1 are ejected (L). Mitochondria-rich plasma is observed in TEM image (L). These events are representative of several tumors.

Figure 1.

Cellularization events linked to the generation of daughter cells by oncogenerative P1 cells. AI, P1 cells are giant cell entities with the ability to breed daughter (G1) cells intracytoplasmically. The release of G1 cells from P1 starts with the formation of crevices through which the G1 cells are ejected via mechanical contraction of P1′s cell body, leaving an aperture that will be repaired afterward. The connection between P1 and G1 is sustained for some time by a funiculus cellularis (Fu), analogous to placental organisms. The sequence of pictures was extracted from Supplementary Video S2. JL, Ultrathin structure of P1 and G1 in ovarian carcinoma tissues. J and K, P1 oncogenerative cell that holds several G1 progeny cells. Note the aperture left by G1 cells once released from P1 (K). The separation of G1 cells from the P1 parental cell starts with the formation of crevices through which G1 are ejected (L). Mitochondria-rich plasma is observed in TEM image (L). These events are representative of several tumors.

Close modal

The ejection of G1 from the parental P1 starts with the formation of funiculus-like structures (Fig. 1B and G; Supplementary Fig. S2A and S2B) and crevices in the P1 cell membrane through which G1 are released by contraction of the P1 parent cell (Fig. 1A; Supplementary Fig. S2C and S2D). The resulting local aperture is then patched up with cellular content rich in HERV envelope proteins (Supplementary Fig. S2C and S2D). P1 and G1-Gn ultrastructure as well as the apertures left by the G1 upon ejection from P1 cells are clearly discernible in tumor ultrathin sections (Fig. 1J–L; Supplementary Videos S3 and S4).

The development and cellularization of G1 cells inside P1 cells and their subsequent release by P1 contraction represents a unique, novel form of cell division (Fig. 1; Supplementary Videos S5 and S6; Supplementary Fig. S2E andS2F). P1 cells undergo long quiescent phases, during which, they do not appear to generate and eject G1 cells (Supplementary Fig. S1A; Supplementary Video S7). These quiescent phases vary significantly and may extend up to 6 weeks.

Another striking feature is the division of P1 cells as a whole cell entity, which appears to imply a cytokinesis-like process. During cell division, mitochondria accumulate around the subsequent P2-Pn de novo cytoplasm delimiting frontiers (Supplementary Fig. S3A and S3B; Supplementary Video S8).

P1 actively injects G1 cells into apparently specialized receptor cells

Endomitotic events of G1 cells inside P1 oncogenerative cells, with subsequent karyokinesis, are common (Fig. 2A–J, red arrows). We noticed another cell type in bidirectional interaction with P1 and termed this second cell type as receptor cell (“R1”) to account for its capability of accepting G1 cells. First, a R1 in close proximity to a P1 polarizes by accumulating material in the membrane area facing the P1 cell (Fig. 2A–J, white arrows), which, in response, extends a protrusion toward the polarized R1 cell connecting both cells. Via this protrusion, or injector, the P1 cell inoculates a G1 into R1. The resulting heterotic tetraploid entity is a “P2” cell with the same set of characteristics as the P1 polyploid ancestor (Fig. 2K–S; Supplementary Video S1). Analysis of these injectors revealed a composition rich in actin and tubulin proteins as described previously (7).

Figure 2.

A P1 oncogenerative cell transmits stemness features by injecting progeny cells into selected adjacent receptor cells. AJ, Events of R1 polarization before the reception of a G1 cell (white arrows). In the sequence, it is possible to observe the development of a sulcus (red arrows) indicative of karyokinetic-like processes. KS, Bidirectional interaction between P1 and surrounding cells and the transmission of stemness to a R1 cell. An adjacent cell is seen to be polarizing toward the P1 oncogenerative cell. These events are representative of several tumors.

Figure 2.

A P1 oncogenerative cell transmits stemness features by injecting progeny cells into selected adjacent receptor cells. AJ, Events of R1 polarization before the reception of a G1 cell (white arrows). In the sequence, it is possible to observe the development of a sulcus (red arrows) indicative of karyokinetic-like processes. KS, Bidirectional interaction between P1 and surrounding cells and the transmission of stemness to a R1 cell. An adjacent cell is seen to be polarizing toward the P1 oncogenerative cell. These events are representative of several tumors.

Close modal

Replication fitness, mitochondrial fission, and the spherogenicity of P1 and G1-Gn

The proliferative status of P1 and G1-Gn cells is reflected in Fig. 3A and B. Cyclin D2 and Poliota replicative enzymes are highly expressed in both P1 and G1 cells. Also, proliferating cell nuclear antigen is highly expressed in P1 cells and many tetranuclear cells surrounding P1 (Supplementary Fig. S4A). Moreover, we observed high expression of HERV envelope proteins in different mitotic phases of G1 cells (Supplementary Fig. S4B). P1 oncogenerative cells do not show an apoptotic phenotype (Fig. 3C). Moreover, these high replication rates are supported by the presence of large quantities of mitochondria in both cell types (Fig. 3D).

Figure 3.

Replication fitness, apoptotic status, mitochondrial fission, and the spherogenic capacity of P1 and G1 cells. AD, Replication fitness and apoptotic status of the oncogenerative and their offspring cells. P1 and G1-Gn cells do not express apoptotic phenotype and several replicative enzymes that denote strong replicative events (AC). D, High mitochondrial fission in G1-Gn cells. EK, 3D cultures of P1 and G1 cells display different growth patterns. E–G, I, and J, 3D culture of P1 and G1 from ovarian tumor cells grown in hanging drops. P1 and G1 cells were aspirated after accutase detachment, washed in media, and resuspended in 25 μL media for 3D cultures. Single P1 oncogenerative cells grown in hanging drops show similar growth patterns as in 2D culture. They are able to form several spheroids during culturing, recapitulating the adjacent colony formation. Contrarily, G1-Gn cells grow uniformly into morula-like structures. Difference in migration patterns of P1 (H) and G1-Gn (K) cells cultured in 3D was observed after being subjected to chemotactic gradient using Transwell chambers as visualized by SEM. LQ, Cytometric analysis of ovarian carcinoma primary cell cultures by FACS. Cell-cycle distribution (L, N, and P) reveals that G1-Gn small cells are near diploids (N), whereas giant cell population shows polyploid genomes (P). L, N, and P represent the BrdUrd incorporation and M, O, and Q represent the DNA content as measured by PI. Pictures are representative of 10 patient ascites analyzed.

Figure 3.

Replication fitness, apoptotic status, mitochondrial fission, and the spherogenic capacity of P1 and G1 cells. AD, Replication fitness and apoptotic status of the oncogenerative and their offspring cells. P1 and G1-Gn cells do not express apoptotic phenotype and several replicative enzymes that denote strong replicative events (AC). D, High mitochondrial fission in G1-Gn cells. EK, 3D cultures of P1 and G1 cells display different growth patterns. E–G, I, and J, 3D culture of P1 and G1 from ovarian tumor cells grown in hanging drops. P1 and G1 cells were aspirated after accutase detachment, washed in media, and resuspended in 25 μL media for 3D cultures. Single P1 oncogenerative cells grown in hanging drops show similar growth patterns as in 2D culture. They are able to form several spheroids during culturing, recapitulating the adjacent colony formation. Contrarily, G1-Gn cells grow uniformly into morula-like structures. Difference in migration patterns of P1 (H) and G1-Gn (K) cells cultured in 3D was observed after being subjected to chemotactic gradient using Transwell chambers as visualized by SEM. LQ, Cytometric analysis of ovarian carcinoma primary cell cultures by FACS. Cell-cycle distribution (L, N, and P) reveals that G1-Gn small cells are near diploids (N), whereas giant cell population shows polyploid genomes (P). L, N, and P represent the BrdUrd incorporation and M, O, and Q represent the DNA content as measured by PI. Pictures are representative of 10 patient ascites analyzed.

Close modal

In 3D cultures, the G1-Gn populations grew faster than P1 cells, as measured by the size of the spheroids, indicating replicative fitness. Moreover, P1 cells were not able to form spheres in hanging drops, resembling the growth patterns seen in highly chemotherapy-refractory ovarian cancers growing in 3D (Fig. 3E–K). This indicates that it might be P1 cells that give rise to the heterogeneous cell populations that make up the bulk of a tumor. In contrast, G1-Gn cells were able to form symmetrical spheres (Fig. 3F, I, and J). When the content of hanging drops from P1 cells and G1-Gn was placed on Transwell membranes in an FCS gradient, P1-derived cells migrated rapidly in comparison with G1-Gn spheres (Fig. 3H and K).

In an attempt to investigate the polyploidy status of these heterokaryotic multinucleated ovarian cells, we undertook cytometric studies of the cell-cycle distribution using PI and BrdUrd incorporation. Small and large cells from the whole population were sorted, revealing high numbers of large polyploid populations alongside the near diploid small cells (Fig. 3L–Q).

P1 and G1-Gn cells differentially express CSC markers typical of their histologic origin

We found that P1 oncogenerative and G1-Gn cells do express CSC markers differentially. P1 cells, despite their physiologic stemness, produced low levels of CSC antigens, thus presenting an example of CSC marker–low phenotypes that can initiate tumor formation. Contrarily, G1-Gn expressed high levels of CSC markers. The differential expression of different CSC markers (CD24, CD44 variants, CD133, ESA/TROP1, Oct-4, SCF, Nanog, and SUZ12) by P1 and G1 in a primary ovarian carcinoma is depicted in Fig. 4A–H and Supplementary Fig. S5.

Figure 4.

P1 and G1 cells show differential expression of specific CSC markers. A–H, These pictures reflect the ICC analysis of different CSC markers in a primary ovarian carcinoma. A, Different forms of the release of G1 cells, which are predominantly CD24 positive. This same pattern is noted for CD44PAN and CD133 as observed in B and C, respectively. Expression of CD44 variant 5 is similar in P1 and G1-Gn cells (F). This CSC expression pattern was found in several primary ovarian carcinomas. D, Ejection of a SCF-positive cell. E and F, Expression of c-Kit and Oct-4, respectively. G and H, Expression of Nanog and SUZ12, respectively. I, Genetic expression of c-Myc, KLF4, Nanog, Oct4, and Sox2 stem cell transcription factors as well as EpCAM in ovarian carcinoma cells obtained from ascites. Up represents the giant cells that are not able to migrate, whereas down represents the small highly migrating cells. J, Comparison of the expression of CSC markers between patient material and SKOV3 ovarian carcinoma wild-type and chemotherapy-resistant models. Significant expression in patient material was observed for Nanog. Experiments are representative of 10 patients analyzed.

Figure 4.

P1 and G1 cells show differential expression of specific CSC markers. A–H, These pictures reflect the ICC analysis of different CSC markers in a primary ovarian carcinoma. A, Different forms of the release of G1 cells, which are predominantly CD24 positive. This same pattern is noted for CD44PAN and CD133 as observed in B and C, respectively. Expression of CD44 variant 5 is similar in P1 and G1-Gn cells (F). This CSC expression pattern was found in several primary ovarian carcinomas. D, Ejection of a SCF-positive cell. E and F, Expression of c-Kit and Oct-4, respectively. G and H, Expression of Nanog and SUZ12, respectively. I, Genetic expression of c-Myc, KLF4, Nanog, Oct4, and Sox2 stem cell transcription factors as well as EpCAM in ovarian carcinoma cells obtained from ascites. Up represents the giant cells that are not able to migrate, whereas down represents the small highly migrating cells. J, Comparison of the expression of CSC markers between patient material and SKOV3 ovarian carcinoma wild-type and chemotherapy-resistant models. Significant expression in patient material was observed for Nanog. Experiments are representative of 10 patients analyzed.

Close modal

G1-Gn cells show high migration rates

We examined cell populations containing both P1 and G1-Gn cells on Transwell and analyzed the populations that were capable of unilateral migration. More than 90% of cells that invaded the lower side of the chamber were less than 25 μm measured as planar. Contrarily, in the upper chamber, more than 90% of cells were greater than 25 μm in size, proving that the G1-Gn cells in particular may have dissemination capacity via intra/extravasation (Supplementary Fig. S6).

Gene expression of stem cell transcriptional factors and CSC markers is not proportionate to protein expression as analyzed by qPCR

We studied the gene expression of pivotal stem cell transcription factors like Oct4, Sox2, Nanog, c-Myc, KLF4 as well as EpCAM in migrating cells. P1 and hypertrophic cells did not migrate through an 8-μm pore membrane, whereas G1-Gn did unimpedingly. We isolated the RNA from both populations with a high sensitive kit, which allows the isolation of RNA from single cells. Figure 4I shows a representative gene expression pattern of both cell populations [large nonmigrating (up) and small migrating (down) cells] isolated from ovarian carcinoma patients.

After separation of both cell types, we noted that multinucleated giant cells differentially produced more transcripts of stem cell markers than their offsprings (Fig. 4I). This inversely mirrors the protein expression of these markers as seen by ICC studies (Fig. 4A–H). Moreover, stem cell transcription factors associated with pluripotency like Oct4, Sox2, Nanog, and KLF4 at the protein levels were found to be expressed by approximately 2% of the cells in the entire population. Contrarily, c-Myc was expressed by approximately 60% of the cells (Supplementary Fig. S7A).

We next compared the gene expression of the above-mentioned markers in the chemotherapy-refractory ovarian carcinoma cell model SKOV3 and cells from ovarian carcinoma patients. Interestingly, expression of Nanog and Sox2 was significantly higher in patients than in chemotherapy-naïve ovarian models (Fig. 4J).

To get a hint about the differential gene expression of different stem cell markers in adherent as well as 3D cultures and to compare this with the same markers in patient material, we employed the SKOV3 wild-type, MDR+, and CP-resistant cells and detected that the multinucleated P1 cells isolated from patients reflect similar behavior as the cells growing in 3D, especially those cells that are highly resistant to cytostatics (Supplementary Fig. S7B).

P1 and G1 cells express HERV envelope proteins

P1 cells expressed low levels of HERV-derived envelope proteins, indicating that P1 cells do not originate from cell–cell fusion. In contrast, G1 cells were very high in HERV protein expression (Fig. 5), indicating that G1-Gn cells retain fusogenic properties. Four different envelope proteins from HERVs (HERV-WE1, HERV-FRD1, HERV-V3.1, and HERV-KISD) were differentially expressed in P1 and G1-Gn cells. Notably, various HERV proteins were found to be strongly overexpressed in different mitotic phases (Supplementary Fig. S4B).

Figure 5.

HERV envelope proteins are widely expressed in P1 and G1 cells obtained from primary ovarian tumors. A–D, ICC analysis of the expression of HERVs. At least four different envelope proteins from endoretroviruses like HERV-WE1, HERV-FRD1, HERV-V3.1, and HERV-K are differentially expressed in P1 and G1-Gn cells. HERV envelope proteins are principally localized in the perinuclear cell compartments, which are rich in mitochondria. A, Release of G1 cells. The expression of HERV-WE1, also known as syncytin 1, is similar in P1 and G1-Gn cells. B, Perinuclear distribution of the protein HERV-FRD1, also known as syncytin 2. The lateral cell division is observable by the expression of EpCAM. In this picture, a repaired aperture is clearly recognizable. C, P1 cell separation (cytokinesis-like event), suggesting that P1 cells are complete, replicative entities. A dense accumulation of mitochondria in the perinuclear area (mitoplasma or mitochondrial plasma) is seen in HERV-K positive G1-Gn cells (D). E–H, Cytometric analysis of HERV expression. Ovarian cancer cells express several envelope proteins derived from endogenous retroviruses as seen by FACS (E–H). Pictures are representative of 10 patient ascites analyzed.

Figure 5.

HERV envelope proteins are widely expressed in P1 and G1 cells obtained from primary ovarian tumors. A–D, ICC analysis of the expression of HERVs. At least four different envelope proteins from endoretroviruses like HERV-WE1, HERV-FRD1, HERV-V3.1, and HERV-K are differentially expressed in P1 and G1-Gn cells. HERV envelope proteins are principally localized in the perinuclear cell compartments, which are rich in mitochondria. A, Release of G1 cells. The expression of HERV-WE1, also known as syncytin 1, is similar in P1 and G1-Gn cells. B, Perinuclear distribution of the protein HERV-FRD1, also known as syncytin 2. The lateral cell division is observable by the expression of EpCAM. In this picture, a repaired aperture is clearly recognizable. C, P1 cell separation (cytokinesis-like event), suggesting that P1 cells are complete, replicative entities. A dense accumulation of mitochondria in the perinuclear area (mitoplasma or mitochondrial plasma) is seen in HERV-K positive G1-Gn cells (D). E–H, Cytometric analysis of HERV expression. Ovarian cancer cells express several envelope proteins derived from endogenous retroviruses as seen by FACS (E–H). Pictures are representative of 10 patient ascites analyzed.

Close modal

The localization of HERV envelope proteins in the perinuclear cell compartment is of particular relevance too (Fig. 5A–D). Cytometric analysis for HERV envelope proteins revealed the expression of these viral elements to be high in ovarian cancers (Fig. 5E–H).

P1 cells express high levels of HERV-KISD and PD-L1 to ensure immune escape

The expression of ISD as depicted in Fig. 6A is expected to confer some immune escape capability to P1 and G1-Gn cells. In the next step, we sought to identify a second immune escape mechanism by investigating the expression of PD-L1 (or CD274) in heterogenic tumor populations. As illustrated in Fig. 6B, P1 is the predominant cell type to overexpress PD-L1, which appears to be delivered to the intercellular space in extrasomal vesicles. Immune escape mediated by ISD in P1 and G1-n cells in ovarian carcinoma tumors was found very high as seen in Fig. 6C. Interestingly, normal ovarian tissues do not express this protein (Fig. 6D).

Figure 6.

Expression of immune escape markers in P1 cells and patient paraffin sections of ovarian carcinoma. A–D, The expression of HERV-KISD in patient paraffin sections and primary ovary cells obtained from ascites (A, C, and D). The programmed death-ligand 1 (PD-L1) or CD274 in heterogenic tumor populations was monitored by ICC in primary ovarian carcinoma cells. A, Expression of HERV-KISD in P1 and G1-Gn cells. Parallel expression of both immune modulator proteins was noted in several tumors of the same histology. PD-L1 is overexpressed in P1, as observed in B. C, Expression of the HERV-KISD domain in ovarian carcinoma paraffin sections. D, Expression of the ISD domain in normal ovarian tissues. E and F, Genetic expression of EpCAM, HERV-Fc1, HERV-KISD, and PDL-1 (E). EpCAM and PDL-1 were found to be significantly overexpressed in ovarian cells isolated from ascites in comparison with wild-type SKOV3 cells (E). The influence of blocking antibodies against CXCR4, EpCAM, HERV-Fc1, and HERV-KISD on cell migration was significant at 10 μg/mL concentration (F) using Transwell migration assay. Migration capacity was measured by SRB proliferation assay. Pictures are representative of 10 patient ascites analyzed.

Figure 6.

Expression of immune escape markers in P1 cells and patient paraffin sections of ovarian carcinoma. A–D, The expression of HERV-KISD in patient paraffin sections and primary ovary cells obtained from ascites (A, C, and D). The programmed death-ligand 1 (PD-L1) or CD274 in heterogenic tumor populations was monitored by ICC in primary ovarian carcinoma cells. A, Expression of HERV-KISD in P1 and G1-Gn cells. Parallel expression of both immune modulator proteins was noted in several tumors of the same histology. PD-L1 is overexpressed in P1, as observed in B. C, Expression of the HERV-KISD domain in ovarian carcinoma paraffin sections. D, Expression of the ISD domain in normal ovarian tissues. E and F, Genetic expression of EpCAM, HERV-Fc1, HERV-KISD, and PDL-1 (E). EpCAM and PDL-1 were found to be significantly overexpressed in ovarian cells isolated from ascites in comparison with wild-type SKOV3 cells (E). The influence of blocking antibodies against CXCR4, EpCAM, HERV-Fc1, and HERV-KISD on cell migration was significant at 10 μg/mL concentration (F) using Transwell migration assay. Migration capacity was measured by SRB proliferation assay. Pictures are representative of 10 patient ascites analyzed.

Close modal

In patient material, PDL-1 and EpCAM were significantly overexpressed at the genetic level (Fig. 6E) in comparison with cytostatic-sensitive ovarian cancer cells. Expression of the ISD motif of HERV-K was also higher than the immune checkpoint PD-L1, regardless of the cells' resistance status. The relevance of these antigens as immunotargets lies in their capacity to inhibit the migration of small cells as seen in Fig. 6F for EpCAM and ISD, respectively.

P1 cells undergo several replicative and cellularization events simultaneously

Figure 7A depicts a cartoon illustrating the major replication events experienced by P1 oncogenerative cells and Fig. 7B–M represents each event captured individually. P1 cells spawn vast number of functional progenies (G1-Gn) by lateral (Fig. 7B) and crevice-mediated eclosions (Fig. 7C), which sometimes leave plicae and apertures in cell membranes (Fig. 7D and E). P1 cells have the ability to inoculate G1 cells via injectors (Fig. 7F) into adjacent polarized (Fig. 7G) receptor R1 cells (Fig. 7H). G1 cells released from P1 (Fig. 7I) fuses (Fig. 7J) with R1 receptor cells. The admixture of G1-R1 (Fig. 7K) forms a heterotic tetranuclear P2 cell with the same stemness characteristics as P1. Two additional forms of G1-Gn release from P1 oncogenerative cells are by budding (Fig. 7L) and via funiculus-like structures (Fig. 7M).

Figure 7.

Cartoon of the cellular events taking place in and around P1 oncogenerative cells. P1 cells are multinucleated giant cells that can reach up to 200 μm diameter with nuclei of about 20 μm diameter. G1 cells are generated by P1: (i) by peripheral cell eclosion or (ii) by intracytoplasmic cell division with subsequent release of the daughter cells by contraction of P1 cells. An aperture or foramen remains once G1 is released. P1 have the ability to inoculate receptor (R1) cells with G1 cells via an injector that docks to the R1 cells. Once the connection is established, the G1-daughter cell detaches from the parent cell and fuses with R1, resulting in a P2 cell with the same stemness characteristics as P1. G1 cells replicate via symmetric division, giving rise to G2-Gn cell generations that will form the mass of the tumor. A–M illustrate the events observed in this work.

Figure 7.

Cartoon of the cellular events taking place in and around P1 oncogenerative cells. P1 cells are multinucleated giant cells that can reach up to 200 μm diameter with nuclei of about 20 μm diameter. G1 cells are generated by P1: (i) by peripheral cell eclosion or (ii) by intracytoplasmic cell division with subsequent release of the daughter cells by contraction of P1 cells. An aperture or foramen remains once G1 is released. P1 have the ability to inoculate receptor (R1) cells with G1 cells via an injector that docks to the R1 cells. Once the connection is established, the G1-daughter cell detaches from the parent cell and fuses with R1, resulting in a P2 cell with the same stemness characteristics as P1. G1 cells replicate via symmetric division, giving rise to G2-Gn cell generations that will form the mass of the tumor. A–M illustrate the events observed in this work.

Close modal

Oncogenesis and CSCs

Several theories of oncogenesis have been put forward. In the early 1900s, it was noted that processes of germ cell development and oncogenesis share similar characteristics (22). In fact, John Beard proposed the trophoblastic theory of cancer. More recently, Vinnitsky suggested the oncogerminative theory of tumor formation after which the malignant transformation of somatic cells is based on the activation of embryogenic programs that confer phenotypic germ cell features (23). In particular, the “oncotrophoblastic cells” and their role in cell division, migration, host–tissue conditioning for angiogenesis, and immune tolerance seem to be of particular relevance in tumorigenesis and metastatic spread (24). Another definition, which refers to the embryonic theory of tumorigenesis, was given by Lloyd J. Old, who claimed that cancer is somatic cell pregnancy (22).

Cell heterogeneity is a hallmark of cancer, especially in the parenchymal structures. The cells constituting a malignant tumor are thought to have a common origin and exhibit a rather multipotent phenotype (25). These oncogenerative cells are hard to characterize based on a set of markers that define CSC phenotypes. Moreover, little effort has been made to determine for how many generations after the very stem cells these CSC antigens continue to be expressed. Interestingly, some tumors have been demonstrated to arise from CSC-negative cells (26). These reports have kindled a debate over the interpretation of the results and the reliability of the known CSC markers as an instrument to identify stemness characteristics (26, 27).

P1 and Gn life cycles and cellularization

The neosis theory introduced by Rajaraman seeks to explain the replication mechanism of multinucleated cells. Neosis is thought to occur in postsenescent multinucleated cells, and it is characterized by karyokinesis via nuclear budding leading to aneuploid mononuclear cells with transient stem cell features, while the polyploid mother cells die (28). Although attractive, the neosis concept does not provide a satisfactory explanation for a few major aspects of our observations: P1 oncogenerative cells do not necessarily die upon G1 generation. Moreover, they are not phenotypically senescent as evidenced by the lack of β-galactosidase expression and appear to be highly viable according to the abundant presence of antiapoptotic effectors like Bcl-2, which are markers of cellular fitness. Furthermore, P1 cells show a mode of cytokinesis that allows them to divide as a whole system (Supplementary Video S8), recently described as cytofission (29).

Our studies established that P1 cells might not meet the current biological criteria of endoreplication (endomitosis and endocycling) in a strict sense per se (9, 10). The way P1 cells generate the first (i.e., proximal) daughter cell generation shows some analogy to cells “infected” (P1) by “giant viruses” (G1) that cycle through a lysogenic phase without killing the host cell as described for viral life cycles. In fact, the contribution of viral proteins to multinucleation different from classic cell–cell fusion has been described previously (20).

P1 oncogenerative cells may drive tumor evolution

Endoreplication and the resulting polyploidy is an adaptive mode with far-reaching consequences for the evolution of many biological systems (30). Although polyploidy may reduce overall cell fitness, it has enormous adaptive potential, as its genetic or epigenetic changes (amplification of oncogenes, new acetylation, or methylation patterns) can alter physiology, metabolism, and, finally, morphology of the cell. This leads up to new phenotypes that contribute to tumorigenesis, drug resistance, and metastatic spread (7, 8, 31).

We found that P1 oncogenerative cells have the ability of inoculating receptor cells (which probably represent a specialized cell type for which P1 has a “tropism”) with G1 via a connective tube or injector. This process has not been described until now, as it is an active, targeted act rather than an unspecific fusion event. It implies the horizontal transmission of the entire genomes, epigenetic patterns, or amplified genes.

The mixing of genetic material (heterosis) in the same cell system (autopolyploidy) or among cells of divergent structure (allopolyploidy) leads to increased allelic diversity that may result in hybrid, more robust de novo cells that are superior to the parental cells in terms of growth rate, robustness, etc., and the ability to occupy new niches (evolutionary selection; ref. 30). The mixing of genetic material between P1 and R1 represents an event of heterosis that may lead to a more adaptive phenotype that in turn supports tumor spread and persistence.

Polyploidy and multinucleation may arise from tetraploidy intermediates. It has been demonstrated that neither a tetraploid condition nor centrosome or cell size anomalies or the failure of cytokinesis lead to G0–G1 cell-cycle arrest. This indicates that there are no efficient checkpoint restrictions in the cell cycle of mammalian cells to avoid tetraploidy (32). Tetraploid cells are thought to be unstable intermediates with aberrant proliferation, probably by the loss of caretaker genes of cell-cycle checkpoint controls. These cell entities are often with restricted distribution but resisting eradication even in the face of antireplicative therapies. Importantly, multinucleation via tetraploidization may lead to induced embryonic-like stemness as proposed by Erenpreisa and colleagues (31, 33). In tumors, endoreplication and subsequent depolyploidization is thought to protect the tumor cells from stress, thus increasing cellular fitness and contributing to the emergence of highly resistant cancer entities (8, 13).

P1 oncogenerative cells are somatic cells in “pregnancy” generating and gestating a prole of cells intracytoplasmically that are cellularized into immediate surroundings, forming new colonies. They have the capacity of a horizontal transmission of the entire genome to other cells, resulting in new tetraploid cell entities.

We expect the striking size difference between P1 and G1-Gn cells to be of crucial importance in the metastatic spread process. Invasiveness via transendothelial migration (tumor–vessel lumen–target tissue) may be different for P1 and G1-Gn cells. We hypothesize that large cells like P1 will remain confined to the local tumor, whereas the smaller G1-Gn cells are likely to be more mobile and therefore possess a higher metastatic potential.

The stemness status of P1 oncogenerative cells

P1 oncogenerative cells display few CSC marker characteristics for multipotency despite their physiologic stemness as judged by their capacity to generate and gestate de novo cells. Interestingly, this is contrasted by the first generation of progeny cells (G1), which clearly express a signature of canonical ovarian CSC markers (15, 27). Importantly, both P1 and G1-Gn express EpCAM/TROP1, a marker of trophoblastic lineage that matches the characteristic of oncotrophoblastic cells. Although Oct-4A (one of the essential stem cell transcriptional factors, among c-Myc, Sox2, and KLF4) was expressed in P1 cells to some extent, this alone does not qualify for pluripotency. c-Myc is thought to be another essential stem cell transcriptional factor, but is also upregulated in several tumors. Thus, the quadriga of stem cell transcriptional factors that leads to pluripotency is not complete. This indicates a rather multipotent phenotype (1, 4).

In general, the presence of pluripotent cells in tumors has remained obscure. The existence of such cells implies at least theoretically, that such pluripotent CSCs may simultaneously transdifferentiate into tumors of dissimilar histologies. This possible scenario is not supported by clinical evidence, because such tumors are uncommon. Biologically speaking, a pluripotent cell has a particular signature amidst the expression of some factors necessary for development and maintenance of this quality. Normal cells can be reprogrammed to totipotent stem cells using the Yamanaka cocktail stem cell transcriptional factors simultaneously (2). Interestingly, efforts directed toward reprogramming cancer cells into stem cells of higher orders have resulted in normal cells (34).

Our studies revealed a significant discrepancy in the levels of mRNA–protein pairs of specific stem cell markers, which demands a further explanation. On the one hand, we know that multinucleated oncogenerative cells are highly polyploid. Polyploid cells commonly show a near-global mRNA increase that is proportional to overall genome content and partially driven by epigenetic regulations. The increase in mRNA transcripts observed for several stem cell factors seems to be not a requirement for P1 cells, because it does not lead to protein synthesis. But, what could be the fate of this overabundant mRNA? We hypothesize that (i) P1 cells may translocate mRNA transcripts to de novo G1 cells during cellularization to drive their development, or (ii) that in these cells the translational/posttranslational events are the dominant factors controlling protein turnover and abundance for the analyzed transcripts (35, 36). The first assumption may explain the high expression of CSC markers, provided that robust protein synthesis in G1-Gn cells is robust. The latter may explain the low expression of CSC markers at the protein level observed in P1 cells. On the other hand, the offspring of P1 cells are nearly diploid and therefore expected to have a more regular transcriptome with less genetic redundancy (30). G1-Gn de novo cells express much higher levels of CSC markers at the protein level than P1. This divergence supports the view that these new cells are initially primed to face the conditions they are about to encounter.

Genome-wide correlations between mRNA and protein expression levels are notoriously inconsistent in human cancers. In fact, only some 40% of the cellular protein levels can be predicted from mRNA measurements (37). Recently, van Velthoven and colleagues described the balance between synthesis and degradation of transcripts in quiescent muscle stem cells (36).

Tumor immune escape may be facilitated by endoretroviral elements

One of the central tenets of the immune system is the active surveillance for malignant transformation and elimination of cancer cells but cancer cells present “self” antigens, and as a consequence, autoreactive immune cells are usually eliminated. On the other hand, tumors generate an immunosuppressive microenvironment that prevents their infiltration by immune cells.

Interestingly, all P1 and G1-Gn cells overexpress some endogenous retroviral proteins, in particular, HERV-coded envelope proteins that are mostly intact. HERV-WE1 and FRD1 are key fusogenic factors that mediate syncytial assembly, a collateral effect attributed to a number of different viruses (38, 39). This raises questions as to the role of the reactivation of endogenous retroviral elements and its implication in cancer development and metastatic spread (40, 41). In this context, HERV envelope proteins may contribute to multinucleation by virtue of their fusogenic properties and, as self-antigens, help mask tumor cells against immune attack in a virus-like fashion (42, 43). Interestingly, HERV elements were overexpressed in mitotic cells in different phases of cell division, indicating an important role in these populations. Of note is the biological function of the ISD introduced by viral envelope proteins, which may help the tumor escape the immune response (16). This means that blocking ISD may be killing two birds with one stone, first by staging a powerful anticancer immune response, and second by killing ISD-rich tumor cells directly (44).

Our data show that P1 cells are high in PD-L1 expression, which, together with the presence of ISD introduced by the HERV proteins, may help to ensure tumor escape from immunologic surveillance.

Another critical observation is that perinuclear mitochondrial fission activity and crowding is extremely high in both P1 and G1-Gn cells and especially the latter, which are particularly rich in mitochondria. The dense perinuclear mitochondrial mass seen in G1-Gn appears to form a plasmatic shield or mitoplasm, which delimits the G1 cell boundaries. Their high mitochondrial content, together with their small size, could explain the very high dividing rates and distinct infiltration potential of G1-Gn, as this process places enormous energy requirements on the cells (19, 45).

In conclusion, the identification and description of P1 oncogenerative cells and G1-Gn daughter cells in ovarian cancer sheds new light on tumorigenesis and tumor persistence and may open up new avenues for targeted cancer therapies in the future.

No potential conflicts of interest were disclosed.

Conception and design: D. Díaz-Carballo, A. Tannapfel

Development of methodology: D. Díaz-Carballo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Díaz-Carballo, S. Saka, J. Klein, T. Rennkamp, A.H. Acikelli, H. Jastrow, C. Tempfer, I. Schmitz

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Díaz-Carballo, H. Jastrow, G. Wennemuth, I. Schmitz

Writing, review, and/or revision of the manuscript: D. Díaz-Carballo, S. Malak, H. Jastrow, G. Wennemuth, C. Tempfer, D. Strumberg

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Tannapfel, D. Strumberg

Study supervision: D. Strumberg

Other (final approval): D. Strumberg

The authors want to thank to Marienhospital Herne from the Elisabeth Group for supporting this investigation. Special thanks to the staff of the Gynaecology Department for the coordination of samples collection. This investigation was supported by grants afforded by institutional funds from Marienhospital Herne.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Martello
G
,
Smith
A
. 
The nature of embryonic stem cells
.
Annu Rev Cell Dev Biol
2014
;
30
:
647
75
.
2.
Schmidt
R
,
Plath
K
. 
The roles of the reprogramming factors Oct4, Sox2 and Klf4 in resetting the somatic cell epigenome during induced pluripotent stem cell generation
.
Genome Biol
2012
;
13
:
251
.
3.
Scott
EW
. 
Stem cell reviews and reports. Adult stem cells and tissue regeneration section
.
Stem Cell Rev
2017
;
13
:
2
.
4.
Aponte
PM
,
Caicedo
A
. 
Stemness in cancer. Stem cells, cancer stem cells, and their microenvironment
.
Stem Cells Int
2017
;
2017
:
5619472
.
5.
van Niekerk
G
,
Davids
LM
,
Hattingh
SM
,
Engelbrecht
A-M
. 
Cancer stem cells. A product of clonal evolution?
Int J Cancer
2017
;
140
:
993
9
.
6.
López-Lázaro
M
. 
The migration ability of stem cells can explain the existence of cancer of unknown primary site. Rethinking metastasis
.
Oncoscience
2015
;
2
:
467
75
.
7.
Diaz-Carballo
D
,
Gustmann
S
,
Jastrow
H
,
Acikelli
AH
,
Dammann
P
,
Klein
J
, et al
Atypical cell populations associated with acquired resistance to cytostatics and cancer stem cell features. The role of mitochondria in nuclear encapsulation
.
DNA Cell Biol
2014
;
33
:
749
74
.
8.
Puig
P-E
,
Guilly
M-N
,
Bouchot
A
,
Droin
N
,
Cathelin
D
,
Bouyer
F
, et al
Tumor cells can escape DNA-damaging cisplatin through DNA endoreduplication and reversible polyploidy
.
Cell Biol Int
2008
;
32
:
1031
43
.
9.
Fox
DT
,
Duronio
RJ
. 
Endoreplication and polyploidy. Insights into development and disease
.
Development
2013
;
140
:
3
12
.
10.
Lee
HO
,
Davidson
JM
,
Duronio
RJ
. 
Endoreplication: polyploidy with purpose
.
Genes Dev
2009
;
23
:
2461
77
.
11.
Larsson
L-I
,
Bjerregaard
B
,
Wulf-Andersen
L
,
Talts
JF
. 
Syncytin and cancer cell fusions
.
Scientific World Journal
2007
;
7
:
1193
7
.
12.
Rengstl
B
,
Newrzela
S
,
Heinrich
T
,
Weiser
C
,
Thalheimer
FB
,
Schmid
F
, et al
Incomplete cytokinesis and re-fusion of small mononucleated Hodgkin cells lead to giant multinucleated Reed-Sternberg cells
.
Proc Natl Acad Sci U S A
2013
;
110
:
20729
34
.
13.
Erenpreisa
J
,
Salmina
K
,
Huna
A
,
Kosmacek
EA
,
Cragg
MS
,
Ianzini
F
, et al
Polyploid tumour cells elicit paradiploid progeny through depolyploidizing divisions and regulated autophagic degradation.
Cell Biol Int
2011
;
35
:
687
95
.
14.
Morgan
DO
. 
“Specialization of Cytokinesis in Animal Development.” In Cell Cycle: Principles of Control
,
170
1
.
London:
Publisher: New Science Press Ltd,
2007
.
15.
Ottevanger
PB
. 
Ovarian cancer stem cells more questions than answers
.
Semin Cancer Biol
2017
;
44
:
67
71
.
16.
Denner
J
. 
Immunosuppressive properties of retroviruses
.
Eur J Immunol
2016
;
46
:
253
5
.
17.
Robainas
M
,
Otano
R
,
Bueno
S
,
Ait-Oudhia
S
. 
Understanding the role of PD-L1/PD1 pathway blockade and autophagy in cancer therapy
.
Onco Targets Ther
2017
;
10
:
1803
7
.
18.
Weihua
Z
,
Lin
Q
,
Ramoth
AJ
,
Fan
D
,
Fidler
IJ
. 
Formation of solid tumors by a single multinucleated cancer cell
.
Cancer
2011
;
117
:
4092
9
.
19.
Díaz-Carballo
D
,
Klein
J
,
Acikelli
AH
,
Wilk
C
,
Saka
S
,
Jastrow
H
, et al
Cytotoxic stress induces transfer of mitochondria-associated human endogenous retroviral rna and proteins between cancer cells
.
Oncotarget
2017
;
8
:
95945
64
.
20.
Patel
D
,
Incassati
A
,
Wang
N
,
McCance
DJ
. 
Human papillomavirus type 16 E6 and E7 cause polyploidy in human keratinocytes and up-regulation of G2-M-phase proteins
.
Cancer Res
2004
;
64
:
1299
306
.
21.
Diaz-Carballo
D
,
Acikelli
AH
,
Klein
J
,
Jastrow
H
,
Dammann
P
,
Wyganowski
T
, et al
Therapeutic potential of antiviral drugs targeting chemorefractory colorectal adenocarcinoma cells overexpressing endogenous retroviral elements
.
J Exp Clin Cancer Res
2015
;
34
:
81
.
22.
Old
LJ
. 
Cancer is a somatic cell pregnancy
.
Cancer Immun
2007
;
7
:
19
.
23.
Vinnitsky
VB
. 
Oncogerminative hypothesis of tumor formation
.
Med Hypotheses
1993
;
40
:
19
27
.
24.
Vinnitsky
V
. 
The development of a malignant tumor is due to a desperate asexual self-cloning process in which cancer stem cells develop the ability to mimic the genetic program of germline cells.
Intrinsically Disordered Proteins
2014
;
2
:
e29997
.
doi: 10.4161/idp.29997
.
25.
Prasetyanti
PR
,
Medema
JP
. 
Intra-tumor heterogeneity from a cancer stem cell perspective
.
Mol Cancer
2017
;
16
:
41
.
26.
Huang
S-D
,
Yuan
Y
,
Tang
H
,
Liu
X-H
,
Fu
C-G
,
Cheng
H-Z
, et al
Tumor cells positive and negative for the common cancer stem cell markers are capable of initiating tumor growth and generating both progenies
.
PLoS One
2013
;
8
:
e54579
.
27.
Abbaszadegan
MR
,
Bagheri
V
,
Razavi
MS
,
Momtazi
AA
,
Sahebkar
A
,
Gholamin
M
. 
Isolation, identification, and characterization of cancer stem cells. A review
.
J Cell Physiol
2017
;
232
:
2008
18
.
28.
Rajaraman
R
,
Rajaraman
MM
,
Rajaraman
SR
,
Guernsey
DL
. 
Neosis–a paradigm of self-renewal in cancer
.
Cell Biol Int
2005
;
29
:
1084
97
.
29.
Niu
N
,
Zhang
J
,
Zhang
N
,
Mercado-Uribe
I
,
Tao
F
,
Han
Z
, et al
Linking genomic reorganization to tumor initiation via the giant cell cycle.
Oncogenesis
2016
;
5
:
e281
.
doi: 10.1038/oncsis.2016.75
.
30.
Comai
L
. 
The advantages and disadvantages of being polyploid
.
Nat Rev Genet
2005
;
6
:
836
46
.
31.
Erenpreisa
J
,
Cragg
MS
. 
MOS, aneuploidy and the ploidy cycle of cancer cells
.
Oncogene
2010
;
29
:
5447
51
.
32.
Wong
C
,
Stearns
T
. 
Mammalian cells lack checkpoints for tetraploidy, aberrant centrosome number, and cytokinesis failure
.
BMC Cell Biol
2005
;
6
:
6
.
33.
Salmina
K
,
Jankevics
E
,
Huna
A
,
Perminov
D
,
Radovica
I
,
Klymenko
T
, et al
Up-regulation of the embryonic self-renewal network through reversible polyploidy in irradiated p53-mutant tumour cells
.
Exp Cell Res
2010
;
316
:
2099
112
.
34.
Lang
J-Y
,
Shi
Y
,
Chin
YE
. 
Reprogramming cancer cells. Back to the future
.
Oncogene
2013
;
32
:
2247
8
.
35.
Jiang
Q
,
Crews
LA
,
Holm
F
,
Jamieson
CHM
. 
RNA editing-dependent epitranscriptome diversity in cancer stem cells
.
Nat Rev Cancer
2017
;
17
:
381
92
.
36.
van Velthoven
CTJ
,
de Morree
A
,
Egner
IM
,
Brett
JO
,
Rando
TA
. 
Transcriptional profiling of quiescent muscle stem cells in vivo
.
Cell Rep
2017
;
21
:
1994
2004
.
37.
Maier
T
,
Güell
M
,
Serrano
L
. 
Correlation of mRNA and protein in complex biological samples
.
FEBS Lett
2009
;
583
:
3966
73
.
38.
Gonzalez-Cao
M
,
Iduma
P
,
Karachaliou
N
,
Santarpia
M
,
Blanco
J
,
Rosell
R
. 
Human endogenous retroviruses and cancer
.
Cancer Biol Med
2016
;
13
:
483
8
.
39.
Soygur
B
,
Sati
L
. 
The role of syncytins in human reproduction and reproductive organ cancers
.
Reproduction
2016
;
152
:
R167
78
.
40.
Ohnuki
M
,
Tanabe
K
,
Sutou
K
,
Teramoto
I
,
Sawamura
Y
,
Narita
M
, et al
Dynamic regulation of human endogenous retroviruses mediates factor-induced reprogramming and differentiation potential
.
Proc Natl Acad Sci U S A
2014
;
111
:
12426
31
.
41.
Santoni
FA
,
Guerra
J
,
Luban
J
. 
HERV-H RNA is abundant in human embryonic stem cells and a precise marker for pluripotency
.
Retrovirology
2012
;
9
:
111
.
42.
Blinov
VM
,
Krasnov
GS
,
Shargunov
AV
,
Shurdov
MA
,
Zverev
VV
. 
Mechanisms of retroviral immunosuppressive domain-induced immune modulation
.
Mol Biol
2013
;
47
:
707
16
.
43.
Izsvak
Z
,
Wang
J
,
Singh
M
,
Mager
DL
,
Hurst
LD
. 
Pluripotency and the endogenous retrovirus HERVH. Conflict or serendipity?
Bioessays
2016
;
38
:
109
17
.
44.
Codd
AS
,
Kanaseki
T
,
Torigo
T
,
Tabi
Z
. 
Cancer stem cells as targets for immunotherapy
.
Immunology
2018
;
153
:
304
14
.
45.
Farnie
G
,
Sotgia
F
,
Lisanti
MP
. 
High mitochondrial mass identifies a sub-population of stem-like cancer cells that are chemo-resistant
.
Oncotarget
2015
;
6
:
30472
86
.

Supplementary data