Apoptotic HPV Positive Cancer Cells Exhibit Transforming Properties

Previous studies have shown that DNA can be transferred from dying engineered cells to neighboring cells through the phagocytosis of apoptotic bodies, which leads to cellular transformation. Here, we provide evidence of an uptake of apoptotic-derived cervical cancer cells by human mesenchymal cells. Interestingly, HeLa (HPV 18+) or Ca Ski (HPV16+) cells, harboring integrated high-risk HPV DNA but not C-33 A cells (HPV-), were able to transform the recipient cells. Human primary fibroblasts engulfed the apoptotic bodies effectively within 30 minutes after co-cultivation. This mechanism is active and involves the actin cytoskeleton. In situ hybridization of transformed fibroblasts revealed the presence of HPV DNA in the nucleus of a subset of phagocytosing cells. These cells expressed the HPV16/18 E6 gene, which contributes to the disruption of the p53/p21 pathway, and the cells exhibited a tumorigenic phenotype, including an increased proliferation rate, polyploidy and anchorage independence growth. Such horizontal transfer of viral oncogenes to surrounding cells that lack receptors for HPV could facilitate the persistence of the virus, the main risk factor for cervical cancer development. This process might contribute to HPV-associated disease progression in vivo.


Introduction
Epidemiological and experimental studies have highlighted that high-risk human papillomaviruses (HPV), especially HPV 16 and 18, play a major role in the induction of carcinomas of the cervix [1,2]. The mechanistic aspects of HPV-induced carcinogenesis are most often related to deletion of the E2 ORF as a consequence of viral DNA integration into the host genome [3]. This leads to a deregulated expression of viral E6 and E7 genes, which represent the main transforming genes. At the heart of this transformation are the binding of E6 to p53 and E6AP, which favors p53 degradation [4] and the E7 complex formation with the retinoblastoma protein pRb [5], resulting in the deregulation of cell cycle control, DNA repair and apoptosis.
During tumor development, a large percentage of cells is lost through apoptosis [6]. Such cell death is triggered by a variety of extracellular signals, including growth/survival factor depletion, hypoxia and a loss of cell-matrix interactions, as well as intracellular signals such as DNA damage [7]. Finally, apoptotic cells are cleared by specialized phagocytic cells that inactivate and degrade their cellular components [8]. However, apoptotic cells can also be internalized by non-specialized recipient cells. Thus, fibroblasts are able to engulf apoptotic neutrophils [9], and liver endothelial cells can bind and phagocytose liver apoptotic bodies [10]. Through this endocytic process, apoptotic cells can act as a DNA vector, and the horizontally transferred DNA may confer a selective advantage to the recipient cell.
Horizontal gene transfer (HGT) has been well documented in prokaryotes and contributes to evolution, ecology and resistance to antibiotics [reviewed in [11,12]]. While the horizontal transfer of genetic information between two eukaryotes has been reported in plants [13,14] and invertebrates [15], few studies have focused on HGT between mammalian cells. The exchange of genetic information mediated by apoptotic bodies has been shown to occur between prostate cancer cells [16]. The apoptotic bodies of transformed lymphoid cells harboring integrated copies of the Epstein-Barr virus can also transfer viral DNA sequences [17]. Similarly, HIV-1 proviral genes are transferred to cells lacking receptors for viral entry [18]. DNA has also been reported to be transferred from apoptotic H-ras V12 -and c-myc-transfected cells to p532/2 mouse embryonic fibroblasts, which leads to their transformation [19]. On the other hand, phagocytosing cells that express p53 or p21 are not transformed, suggesting a protective mechanism controlled by the p53 pathway [20]. Recently, Ehnfors J. et al. demonstrated that fibroblasts and endothelial cells are capable of acquiring and replicating H-ras V12 and c-myc DNA when apoptotic tumor cells contain the simian virus 40 large T (SV40LT) antigen [21]. These observations provided evidence that transformation efficiency is associated with the expression of SV40LT inhibiting p53 [22]. Because the majority of cervical carcinomas express the E6 viral oncoprotein, which promotes p53 degradation, as does SV40LT, we hypothesized that the horizontal transfer of HPV oncogenes could be an alternative mechanism of carcinogenesis.
Here, we present evidence that apoptotic cells derived from cervical-derived cancer cells harboring integrated copies of HPV are able to transform human primary fibroblasts (HPF). We further demonstrate that recipient tumor cells can be characterized by a high rate of proliferation and hyperploidy. In addition, the viral genetic material inhibiting the p53/p21 pathway is expressed in the transformed cells. To our knowledge, this is the first report of the transformation of human primary cells through the uptake of apoptotic bodies from HPV-infected cervical carcinoma cells.

Apoptotic cervical carcinoma cells are internalized by fibroblasts
The apoptosis of cervical carcinoma donor cells was induced by UVB irradiation and staurosporine exposure as previously described [23,24] and was documented by an analysis of phosphatidylserine exposure (annexin V staining), DNA content (propidium iodide staining) and nuclear fragmentation (DAPI staining) (Methods S1 and figure S1A and S1B). The treatment resulted in the absence of living cells capable of proliferation within the apoptotic cell suspensions (Methods S1 and figure S1C and SID). Previous studies have shown that apoptotic bodies derived from EBV-carrying B lymphocytes can transmit DNA by horizontal transfer and that EBV-integrated DNA may be preferentially transferred as compared with cellular DNA [17]. In this study, we questioned whether HPFs could engulf apoptotic cells derived from the cervical carcinoma cell lines HeLa (HPV18), Ca Ski (HPV16) and C-33 A (HPV-), regardless of virological status.
The presence of fluorescent apoptotic cells in the recipient cells was confirmed by confocal microscopy. Apoptotic HeLa cells containing DNA were entangled in the actin cytoskeleton of the HPFs within 48 h (figure 1A). Apoptotic Ca Ski and C-33 A cells were also taken up efficiently by the recipient (figure 1B). Incubation of the HPFs alone or with the supernatant of apoptotic cells did not result in CFDA, SE (5-(and 6-)-carboxyfluoresceine diacetate succinimidyl ester) staining, suggesting a link between green fluorescence and the presence of apoptotic cells (figure 1B). By tracking the fluorescent dyes at early time points (from 1 h to 3 h), we observed actin recruitment when apoptotic cells were bound to HPFs (figure 1Ci, white arrow). The fibroblast membrane expanded around both sides of the apoptotic cell through actin polymerization (figure 1Cii, white arrows). F-actin then surrounded the apoptotic cells to form a phagocytic cup and closed in a ring (figure 1Ciii). These microscopic observations are indicative of phagocytosis, although we have not specifically characterized this mechanism [25,26]. Using specific markers of intermediate filaments for each cell type, we confirmed that the apoptotic cells were epithelial cells (cytokeratin positive) that were internalized by fibroblasts (vimentin positive) ( figure 1D). Using the quantitative approach of flow cytometry, we assessed the percentage of HPFs that engulfed the stained apoptotic carcinoma cells. Regardless of the type of apoptotic cells used, the internalization efficiency was similar (12.5% with apoptotic HeLa; 13% with apoptotic Ca Ski; 14.5% with apoptotic C-33 A) ( figure 1E). However, we noted that 12 to 15% of the fibroblasts were able to take up the apoptotic cells, while the number of apoptotic cells seeded was ten times larger than that of the HPFs. This suggests that fibroblasts have a limited potential in the efficiency and/or quantity of apoptotic cell internalization. When recipient cells were co-incubated with apoptotic cells at 4uC for 48 h, the percentage of internalization dropped significantly to 2%, thus suggesting that the internalization process follows an energy-dependent pathway (figure S2). These results indicate that human primary fibroblasts can engulf apoptotic cells, independent of their virological status, through phagocytosis.

Only HPV-positive apoptotic cells efficiently transform recipient cells
Ehnfors et al. demonstrated that DNA from rat fibrosarcoma apoptotic cells transfected with H-ras V12 , c-myc and SV40LT is transferred to and transforms primary fibroblasts [21]. Because HPV oncogenes, like SV40LT, are capable of efficiently transforming infected cells and blocking the p53 pathway, among other effects, we tested whether fibroblasts cultured with apoptotic cells were able to grow with anchorage independence by measuring their ability to form colonies in a soft agar assay, as observed with the HeLa, Ca Ski and C-33 A cancer cells ( [27]. The transformation status of the HPFs was further tested by limitdilution assays. Indeed, in contrast to primary fibroblasts, the fibroblasts transformed by apoptotic HeLa (FH) and Ca Ski (FC) cells had the ability to form multilayer colonies when they were grown at low density (figure 2B). At this stage, the FH and FC began to exhibit a transformed phenotype, with some of the cells appearing rounded, unlike the control HPFs, which displayed a spindle shape (figure 2C). Moreover, the transformed fibroblasts consisted mostly of packed or aggregated small cells such as HeLa and Ca Ski cells, whereas the primary fibroblasts formed a flattened monolayer. Figure 3A shows that transformed fibroblasts, FC and FH, are not positive for cytokeratin staining, unlike epithelial HeLa and Ca Ski cells, revoking the possibility that the observed transformed cells are rare surviving cancer cells and a clonal evolution of Ca Ski cells. As illustrated in the upper left panel of figure 3B (D18S61 marker, an example of 20 markers of DNA typing used), transformed fibroblasts have a pattern of DNA that is different from the original tumor cells. Other results of DNA typing experiments showed that FC cells were different from Ca Ski cells but contain some alleles from Ca Ski DNA (TP53 and D8S264 markers), reflecting the transfer of small DNA fragments. FC DNA also contains parts of chromosomes (D17S250 markers), reflecting the transfer of large DNA fragments as described by Holmgren [17,19]. Although these results demonstrate the chromosomal rearrangement (loss and gain of alleles) of FC cells, we cannot exclude the possibility of cross-contamination, despite this being highly unlikely.
Next, we determined the effects of HPF transformation on the proliferation rate using a growth curve and the results from the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromid) proliferation assay shown in Fig. 4A and 4B, respectively. The FH had an average population doubling time of 15 h, and the FC had an average time of 16 h, while the HPF doubling time was greater by a factor of 19 (298 h) (figure 4A). The mitochondrial activity measured by the MTT proliferation assay was concordant with the growth curve results (figure 4B). These growth rate modifications support the tumorigenic potential of the newly transformed fibroblasts. We therefore tested whether the increased growth rate and the transformation of the recipient cells were associated with genetic modifications leading to hyperploidy. Cytometry assays showed that the DNA content increased with the passages of transformed fibroblasts (figure 4C). In our experiments, an HPF mean fluorescence intensity (MFI) of 154 corresponded to diploid cells (figure 4D). The MFI increased to 205 and 250 after 5 (P5) and 15 (P15) passages of FH cells, respectively. We observed similar results for the FC cells (219 at P5 and 264 at P15). The MFI at passages 15 of FH and FC represented hypertriploidy. Aneuploidy, as seen in our model, is often caused by a particular type of genetic instability and is one of the most common properties of cancers [28].

Apoptotic HPV-associated cancer cells efficiently transfer viral oncogenes to fibroblasts
Because only apoptotic HeLa and Ca Ski donor cells were able to transform fibroblasts, we hypothesized that HPV oncogenes could be transferred to the recipient cells. Our confocal microscopy analysis suggested that genetic material was transferred from the apoptotic cells to the fibroblast nuclei after 6 h of co-culture (figure 5A, white arrows). The actin cytoskeleton reorganization appeared to deliver the apoptotic cell toward the nucleus for DNA transfer, as seen at a high magnification by superimposing transmitted light with phalloidin or DAPI staining images. Following these observations, we investigated the HPV DNA transfer in fibroblast recipients using in situ hybridization (ISH) with a probe hybridized specifically to high-risk HPV DNA.
HeLa and Ca Ski cells were used as positive controls. Confirming our hypothesis, the hybridization signals were observed as purple dots in the apoptotic cells and nuclei of the transformed fibroblasts (FH and FC) whereas no signal was detected in the HPFs (figure 5B). These data validate the hypothesis of horizontal transfer of viral oncogenes. A second approach consisting of amplifying the E6 DNA of HPV 16 and 18 confirmed the presence of viral DNA in the transformed fibroblasts (figure 5C). We further analyzed the expression of E6 HPV16 and E6 HPV18 by reverse transcriptase followed by real-time quantitative PCR. Figure 6A illustrates that the E6 transcripts were detected in the transformed FH and FC cells with however lower levels than in the parental HeLa and CaSki cells. These data suggest that the transfer of viral oncogenes is efficient and functional. To more thoroughly scrutinize the role of the transferred E6 oncogenes as inhibitors of p53 expression, we immunoblotted for p53 and one of its targets, p21. Accordingly, the p53 and p21 levels of the transformed HPFs decreased substantially, similar to the decrease in donor cancer cells ( figure 6B). Overall, these results emphasize a critical role of viral oncogene transfer in the transformation of primary cells, a process that bypasses the p53 pathway.

Discussion
The results of this study provide direct evidence of the oncogenic potential of HPV positive apoptotic cells. The transfer of viral DNA derived from HPV-positive cervical cancer cells through HGT promoted the growth and transformation of HPFs and could represent an alternative mechanism for HPV-associated  [19]. In their model, the donor cells were primary rodent fibroblasts modified to overexpress c-myc and H-ras V12 and a hygromycinresistant gene that permitted a highly stringent selection of transformed recipient cells after HGT.
In our study, the internalization rates of apoptotic cancer cells were similar, regardless of the HPV status. However, the finding that only HeLa and Ca Ski cancer cells carrying naturally integrated HPV oncogenes, but not HPV-negative C-33 A cells are able to transform phagocytosing fibroblasts provides support for the hypothesis that viral DNA transferred by apoptotic cells can be reused and expressed by recipient cells.
The in vitro recipient cell transformation by horizontally transferred DNA, facilitated by DNA fragmentation, was shown to be dependent on p53 [29]. Indeed, p53-or p21-deficient cells, but not wild-type p53 fibroblasts, became tumor-like after their uptake of c-myc and H-ras V12 oncogenes, indicating that the Chk2/ p53/p21 signaling pathway protects cells against the propagation of potentially harmful DNA [19,20,30]. Moreover, SV40LT, which facilitates p53 degradation, can overcome this genetic surveillance both in vitro and in vivo [21]. Like SV40LT, the E6 oncoproteins of HPV type 16 and 18 are physically associated with wild-type p53 and favor its proteosomal degradation [reviewed in [31]]. An analysis of the transformed fibroblasts revealed that they contained E6 HPV16 or E6 HPV18 DNA. In addition, ISH showed that HPV DNA from the apoptotic cancer cells was transferred into the nuclei of recipient fibroblasts. Once the DNA was transferred, the expression of HPV E6 genes was detected at the RNA level up to 15 weeks after the start of co-culture experiments. The E6-expressing recipient cells exhibited decreased levels of p53 as well as its target, p21, which might partly explain the alterations in growth control circuitry.
We intended to confirm by genotyping that transformed fibroblasts were unique and were derived from primary fibroblasts and parental cancer cell lines. By studying several microsatellites, we found that they actually harbored some alleles identical to those identified in apoptotic cervical cancer cells. However, due to the loss of primary fibroblasts to senescence, we were unable to confirm that they were derived from primary fibroblasts. We cannot therefore completely exclude the possibility of cross-contamination with an unrelated cell line, even if this possibility appears improbable.
However, virally altered fibroblasts that were able to form colonies demonstrated peculiar morphological characteristics, a high proliferation rate and aneuploidy. The transition from diploidy to aneuploidy, a hallmark observed in virtually all cancers [28], has been noted in early high-risk HPV-associated lesions of the cervix [32] and has been attributed to the synergistic effect of E6 and E7 oncoproteins [reviewed in [31]]. However, high-risk HPV immortalized cells are non-tumorigenic, and the activation of cellular oncogenes c-myc, H-ras and c-fos is necessary to completely overcome the anti-oncogenic function of p53 and to result in cervical cancer development [33,34,35]. Further study of the expression of these possibly activated cellular oncogenes will aid in understanding the mechanism of fibroblast transformation. Nonetheless, HPV oncogene transmission could have a role more crucial than considered, since the expression of HPV16 E6 oncogene in HPV negative C-33 A cells confers an aggressive phenotype as shown by the radiation resistance in transplanted tumors [36].
Escape from immune surveillance mechanisms may represent the main risk factor for HPV DNA persistence and lesion progression, whereas the viral transfer from apoptotic bodies to surrounding cells lacking receptors for HPV in vivo could facilitate the persistence of HPV in the cervix. Moreover, such transfer might explain the spread of HPV to mesenchymal cells, as observed by ISH in cervical carcinomas [37,38], and the possibility of a stromal reservoir for HPV. Furthermore, in human solid tumors, a subset of cancer cells, called cancer stem cells, that are likely initiated as a result of HGT cause very aggressive cancers with a high propensity toward metastatic dissemination [39]. This might explain the positive association between the rate of intratumoral apoptosis and several cervical tumor parameters such as tumor size, lesion grade, metastatic phenotype and patient survival [7,40,41,42].
In addition, the increased prevalence of apoptotic cells following chemo-and/or radiation therapy could be partly responsible for the recurrence of HPV lesions by inducing the transformation of new cells at the same or different anatomic locations months or years after remission [43,44]. Investigations of this possibility are warranted, in addition to efforts to find new therapeutic strategies targeting E6/E7 oncogenes to limit their horizontal transfer and to control tumor development.

Cell culture and apoptosis induction
HPFs were isolated from adult human skin after abdominoplasty as previously described [45] and were grown in complete DMEM (Lonza) containing 10% FBS (Lonza), penicillin/streptomycin and L-glutamine. HPFs extracted from surgical residues are not subject to validation from an ethics committee and the patient's consent in accordance with the law L.1245-2 of the ''Code de la santé publique'' applied in France. Moreover, the laboratory of skin engineering of Prof. Philippe Humbert's dermatology departement, providing human fibroblasts, has manuscript documents stating the patient's non-opposition to the use of his surgical residues to medical research in accordance with the law L.1211-2. Human cervical carcinoma cell lines (ATCC), HeLa (HPV18, wild-type p53) and C-33 A (HPV negative, mutated p53) were grown in complete EMEM (Lonza), and Ca Ski cells (HPV16, wild-type p53) were grown in complete RPMI (Lonza). They were monitored monthly and found to be free of mycoplasms. Twelve hours prior to apoptosis induction, the carcinoma cells were seeded at 2610 4 cells/cm 2 . They were then treated with 20 mJ/cm 2 UVB irradiation followed by 300 nM staurosporine (STS) (Sigma Aldrich) for 48 h. The apoptotic cells were harvested after centrifugation of the supernatant at 300 g for 10 min. The apoptosis detection was conducted as described in the supplemental information (Methods S1). The apoptotic cells were incubated with HPFs at a ratio of 10:1. This ratio was chosen because a higher ratio causes fibroblast death and a lower ratio decreases the rate of internalization.

Apoptotic cell internalization analysis
The apoptotic cells were stained with 1 mg/ml CFDA, SE (Invitrogen Ltd) diluted in DMEM with 2% FBS for 13 min at 37uC. After washing, they were incubated for 48 h with recipient cells. For cytometry analysis, the fibroblasts incubated with the apoptotic cells were harvested by trypsinization and analyzed using a Cell Lab Quanta TM SC flow cytometer (Beckman Coulter). Apoptotic cells were labeled with CFDA, SE and HPFs were distinguished from apoptotic cells by their diameter as evaluated by flow cytometry. Events with small diameters (,13 mm) and positive for CFDA SE, were considered apoptotic cells;, events with large diameters (.13 mm) and negative for CFDA, SE, were HPFs, and events with large diameters and positive for CFDA, SE, were HPFs with engulfed apoptotic cells.

Colony formation assay, cell growth and aneuploidy analysis
Soft agar assays were conducted in 24-well plates in semi-solid media (DMEM, 10% FBS, 0.35% agar; Invitrogen) with 8610 3 and 3610 5 cells/ml on a media base layer (RPMI, 10% FBS, 0.5% agar). The cells were grown for 21 days, and the colonies were observed using an Axiovert 25 inverted microscope (Zeiss) [27]. The colonies were then harvested by scraping the surface of the soft-agar and two cell lines of fibroblasts transformed by apoptotic HeLa (FH) and Ca Ski (FC) cells were derived and used for further experiments.
Primary fibroblast transformation was also tested by limitdilution cultures in 6-well plates at 5610 2 cells/well for 21 days. The colonies were then stained using a purple crystal solution (0.1% purple crystal (w/v), ethanol 5%) and photographed with a Nikon Coolpix 4500 digital camera (Nikon) [46].
The proliferation of the transformed cells was monitored by counting the total number of cells in each individual well daily for 10 days with a Cell Lab Quanta TM SC flow cytometer. The proliferation was also monitored using the MTT test with the Cell Proliferation Kit I (MTT) (Roche) for 5 consecutive days. The formation of purple formazan crystals was quantified using the scanning multi-well spectrofluorimeter EnVisionH 2102 Multilabel Reader (Perkin Elmer). For the aneuploidy analysis, 10 6 living cells were collected and fixed overnight in 70% (v/v) cold ethanol. After two washes, the cells were stained with a propidium iodide solution (0.1 mg/ml propidium iodide, Sigma Aldrich; 20 mg/ml RNaseA DNase-free, ABgene). After 15 min at RT, 20,000 events were analyzed by an FC500 flow cytometer (Beckman Coulter). The mean fluorescence intensity (MFI) of the G0/G1 peak was evaluated using the CXP TM cytometer software and expressed in arbitrary units.

In situ hybridization
Cells were cultured overnight on poly-L-lysine microscope slides (Thermo Fisher Scientific), fixed with 3.7% paraformaldehyde for 15 min at RT and permeabilized with PBS Tween 1% for 10 min at RT. The high-risk HPV DNA detection was conducted using the INFORMH HPV family probe (Ventana Medical Systems) and the BenchMarkH XT Automated Slide Stainer (Ventana Medical Systems) as described by the manufacturer. The cells were also counterstained with eosin.

Amplification of microsatellites and viral genes
Total DNA was extracted from 10 6 cells using the QIAampH DNA Mini Kit (Qiagen, Courtaboeuf, France). For typing, extracted DNA was amplified by fluorescent PCR as described previously [47]. A panel of 20 microsatellites was used. Primers were obtained from the Genome Data Base (www.gdb.org) or Genemap'99 (www.ncbi.nlm.nih.gov/genemap99/). Amplified fragments were analyzed on an ALF Sequencer (Amersham-PharmaciaH, Piscataway, NJ, US), allowing for a very sensitive and quantitative evaluation of the allele ratio by measuring the peak height of both alleles.
For viral gene detection, PCRs targeted the albumin, E6 HPV16 and E6 HPV18 genes, using 500 nM of the corresponding primers (Eurogentec, Seraing, Belgium). The sequences of the albumin and E6 HPV16 primers have been previously described by Laurendeau  [48,49]. After a hot-started reaction at 94uC for 5 min, the target DNAs were amplified for 30 cycles for 30 sec at 94uC, 30 sec at the annealing temperature (57uC for albumin, 51uC for E6 HPV16 and 55uC for E6 HPV18) and 20 sec at 72uC, followed by a 7-min extension at 72uC. The PCR products were analyzed by agarose gel electrophoresis.

Detection of viral transcripts
Total RNA was extracted from 2610 6 cells using the QIAampH RNA Blood Mini Kit (Qiagen). After DNAse I treatment (Invitrogen), the reverse transcription of 500 ng of RNA was performed using the MMLV-Reverse Transcriptase (Invitrogen). Twenty five nanograms of cDNA were pre-amplified using 45 nM E6 HPV16, E6 HPV18 and human b-2-microglobulin primers in the TaqMan PreAmp Master Mix (Applied Biosystems) as follows: a 10-min step at 95uC and 10 amplification cycles (15 sec at 95uC, 4 min at 60uC). The quantification of the pre-amplified products was performed with a 7500 Real Time PCR System (Applied Biosystems) in the TaqMan Gene Expression Master Mix (Applied Biosystems), using 500 nM of each primer (Eurogentec) and TaqMan Probes (100 nM for b-2-microglobulin: 59FAM-cctccatgatgctgcttacatgtctcgatccc-BHQ1-39; 250 nM for E6 HPV16: 59-FAM-aggagcgacccagaaagttaccacagtt-BHQ1-39 or E6 HPV18: 59-JOE-caacacggcgaccctacaagctacc-BHQ1-39, Eurogentec) according to the manufacturer's thermal cycling protocol. Standard curves were obtained by serial dilutions over a range of six log concentrations of the pBR322-HPV16 and pBR322-HPV18 diluted in 50 ng/ml salmon sperm DNA.

Supporting Information
(DOC)