Background
It has been well established that high-risk human papillomavirus (hrHPV) infections are causally related to cervical cancer development [
1]. While the two most potent viral oncogenes E6 and E7 are necessary for the initiation and maintenance of cellular proliferation [
2,
3], their expression is not sufficient for full transformation of epithelial cells. Hence, there are extensive efforts towards identifying additionally required host cell events. Chromosomal analysis has revealed that a gain of chromosome 3q is the most common event in the development of cervical squamous cell carcinoma (SCC) [
4‐
6]. In fact, a gain of chromosome 3q was found in all SCC previously analysed by microarray CGH [
6]. Candidate oncogenes on chromosome 3q include the gene encoding p110α, the active subunit phosphatidylinositol 3-kinase catalytic alpha (PIK3CA) of class I PI3-kinase. Upon activation, PI3-kinase initiates events leading to phosphorylation of PKB/AKT, which affects additional downstream signalling proteins involved in survival and cell growth. Indeed, deregulation of the PI3-kinase pathway is common in many human malignancies [
7]. In cervical carcinomas, an increased copy number of PIK3CA was positively correlated with an increase in phosphorylated PKB/AKT, one of the downstream effectors [
8]. Additionally, the level of p-PKB/AKT expression increased proportional to the histopathological grade of (pre)malignant cervical diseases [
9,
10]. Although it has been found that HPV16E7 can activate PKB/AKT in differentiating cells [
10], the relevance of PI3-kinase signalling in the process of cervical cancer development following a transforming hrHPV infection remains to be experimentally explored. Moreover, no functional studies on the specific role of PIK3CA in cervical carcinogenesis have yet been performed.
Previously we have shown that
in vitro transformation of primary keratinocytes mediated by full-length hrHPV was accompanied with a gain of chromosome 3q in immortalized descendants [
6]. This model system of HPV-transformed keratinocytes therefore provides interesting and useful source material to study the potential functional role of PI3-kinase for the various transformed phenotypes. In the present study we analysed PIK3CA expression in cervical squamous cell carcinomas. We also performed functional analyses of the contribution of PI3-kinase signalling, and specifically PIK3CA, to hrHPV-mediated transformation
in vitro.
Methods
Cell culture, LY294002 treatment and transfection
Primary human foreskin keratinocytes, HPV16 and HPV18-immortalized keratinocyte cell lines (FK16A and FK18A) as well as HPV16E6E7 containing keratinocytes were cultured as described previously [
11]. The latter cells were generated by transduction of primary human foreskin keratinocytes with the retroviral vector pLZRSneo containing HPV16E6E7, as described previously [
12].
FK16A cells between passages 45 and 62 represented immortal and anchorage dependent cells and FK16A cells between passages 99 and 189 represented anchorage independent cells [
11]. Prior to LY294002 (10 μM and 20 μM) (Cell Signaling Technology, Beverly, USA) or DMSO treatment, cells were starved overnight to ensure similar phosphorylation status. A pool of 4 siRNA sequences targeting PIK3CA (cat#L-003018-00-0005, Dharmacon, Lafayette, USA) was transfected using Dharmafect reagent 2 (Dharmacon) according to the manufacturers protocol. Pools of 4 non-targeting siRNAs (cat#D-001810-10-05) and PLK1 specific siRNAs (cat#L-003290-00-0005) were used as negative and positive controls, respectively. The use of a non-targeting siRNA pool ensures control for off-target effects. Transfection of cDNA encoding for myristoylated PIK3CA [
13] Addgene, Cambridge, USA) and cotransfections with HPV16-URR luciferase constructs into FK16A cells were performed using Effectene (Qiagen, Hilden, Germany) according to instructions. Firefly luciferase and Renilla luciferase were measured using Dual Luciferase assay (Promega, Wisconsin, USA).
Clinical material
All tissue specimens were collected during the course of routine clinical practice at the Department of Obstetrics and Gynecology at the VU University medical center. Normal epithelial control samples were obtained from histologically normal frozen biopsies of non-cancer patients undergoing hysterectomy. This study followed the ethical guidelines of the Institutional Review Board of the VU University medical center.
RNA isolation, RT-PCR and Northern Blotting
Isolation of mRNA from cell lines was performed using RNA-B reagent (Tel-Test, Friendswood, USA) and DNase treated (Promega) prior to cDNA synthesis using specific reverse primers (see below). Total RNA from micro-dissected frozen biopsies of cervical SCCs and normal ectocervical controls was isolated using Trizol reagent (Invitrogen Life Technologies, Breda, The Netherlands) as described before [
14]. Quantitative RT-PCR was performed as described previously [
15] using the following primers for PIK3CA forward 5'-CCTGATCTTCCTCGTGCTGCTC-3' and reverse 5'- ATGCCAATGGACAGTGTTCCTCTT -3' using SYBR Green PCR Master Mix (Applied Biosystems, Carlsbad, CA, USA). And for hTERT forward 5'-CACGCGAAAACCTTCCTCA -3', reverse 5'-CAAGTTCACCACGCAGCC-3' and the probe FAM-5'-CTCAGGGACACCTCGGACCAGGGT -3'-TAMRA using Universal PCR Master Mix (Applied Biosystems). To correct for RNA quality and input, we performed RT-PCR for the housekeeping gene snRNP as described before in cell line experiments [
16]. For quantification, a standard curve was established using serial dilutions of cervical cancer cell line cDNA. To determine HPV16E7 mRNA expression LightCycler real-time PCR assays were applied as described before [
17,
18] as well as Northern Blotting for HPV16. Total RNA was separated on a 1% agarose gel, blotted on nylon membranes (GeneScreen, PerkinElmer Life Sciences, Waltham, USA) and hybridized with a radioactive labelled full length HPV16 probe.
Immunoblotting
Antibodies against total (cat#9272) or phosphorylated forms of PKB/AKT (cat#4058), PIK3CA (cat#4255) and loading control beta-actin (cat#4967) (all 1:1000 from Cell Signaling Technology) were used according to the manufacturers instructions. Membranes were incubated with the appropriate horseradish peroxidase-conjugated secondary antibodies and the levels of corresponding proteins were visualized using SuperSignal West Dura Extended Duration Substrate (Pierce).
Immunohistochemical staining and immunofluorescence assays
Immunohistochemical staining was performed using 4 μm sections which were deparaffinised, rehydrated and microwave-treated (800W) for 10 min in Tris buffer (pH9), followed by incubation for 30 min in 3% H2O2 in methanol. Antibody incubation with PIK3CA (cat#4249 Cell Signaling 1:200) was performed overnight at 4°C and for detection the EnVision horseradish peroxidase system (Dako, Heverlee, Belgium) was used.
For immunofluorescence, 4 μm sections were rehydrated, treated with 10 mM citrate, pH 6.0 at 95°C for 10 min and allowed to cool to room temperature over 20 min. Slides were then treated with 3% H2O2 in water and blocked in 1XPBS containing 10% goat serum. P-PKB/AKT and Ki-67 were detected by sequential probing with respective antibodies because both were raised in rabbit. P-PKB/AKT was probed with rabbit monoclonal antibody (cat#2118-1, Epitomics Inc, 1:100 dilution) and detected by fluorescine-conjugated tyramide (NEL701001, PerkinElmer Life Science, USA) as per the manufacturers direction. Subsequently, Ki-67 was probed with rabbit monoclonal antibody (ab16667, Abcam, 1:100 dilution) and detected by Alexa Fluor 555 conjugated anti-rabbit IgG (cat#A21429, Invitrogen-Molecular Probes, USA.) as per the manufacturers protocol. P-PKB/AKT and CK10 localization was detected by concurrent probing with respective primary antibodies as they were raised in different species. Raft sections were treated with rabbit anti p-PKB/AKT (described above) and mouse monoclonal anti-CK10 (cat#ab1421, 1:150 dilution, Abcam, Cambridge, USA). Anti-CK10 was detected with Alexa Fluor 555 conjugated goat anti-mouse IgG (cat#A21424, Invitrogen-Molecular Probes, USA). Finally slides were mounted with DAPI containing media (VECTASHIELD, H1200, Vector Laboratories, USA), viewed under Olympus AX70 microscope fitted with Chroma filters. Photomicrographs were captured by Axiovision camera at 20× magnifications of objective and finally processed with Photoshop CS2 (Adobe) for documentation.
Cell Viability, Migration, Anchorage Independent Growth
Cell viability was assessed by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) dye reduction (ICN Biomedicals Inc, USA). Cells were seeded in triplicate wells in 96-wells plates, transfected or starved overnight followed by LY294002 treatment and grown for 5 days. Control conditions were set to 100%.
For cellular migration assays cells were plated at high confluence and uniformly scratched to create a cell-free gap. After 24-48 hours in serum-free keratinocyte growth medium, plates were examined and photographed to asses the migration of neighbouring cells into the gap.
To examine anchorage independent growth under the different conditions 5000 cells of each condition were plated in semi-solid agarose (as described previously [
19]). After 3 weeks colonies larger than ~50 cells were counted and pictures taken.
Organotypic raft cultures
The HPV containing keratinocytes were grown as epithelial raft tissues as described previously [
20]. For all conditions duplicate rafts were developed. Transfections with siRNAs were carried out the day before seeding on the collagen beds. Inhibitor treatment was started after seeding and continued throughout the 9 days of culturing at the liquid-air interface. After harvesting, the raft tissues were fixed in formalin and embedded in paraffin. For histological examination, 4 μm sections were stained with hematoxylin and eosin.
Statistical analysis
All statistical analyses were carried out using the T-test in the SPSS software package (SPSS 15.0, Chicago, USA).
Discussion
The PI3-kinase/PKB/AKT signalling pathway affects a wide variety of cellular characteristics such as proliferation, differentiation and cell survival and is often altered in many human malignancies [
7]. In cervical cancer, a gain of the long arm of chromosome 3, where PIK3CA is located, is often described and suggested to be a compulsory second hit for malignant transformation following an hrHPV infection [
4,
6]. However, data is scarce on the biological effects of altered PIK3CA expression and PI3-kinase signalling in cervical carcinogenesis.
In the present study we showed that mRNA expression of the catalytic subunit PIK3CA was significantly upregulated in cervical SCC. In our previous studies using arrayCGH, up to 100% of cervical SCC were shown to contain additional copies of chromosome 3q [
6]. Additionally, low levels PIK3CA amplifications were found in 40%-74% of cervical carcinomas [
5,
8,
26]. Bertelsen
et al showed that an increase in PIK3CA copy numbers (more than 3 copies) was significantly correlated to elevated p-PKB/AKT expression in cervical SCC and high-grade precursor lesions [
8]. Also, other studies reported an increase in p-PKB/AKT staining with increase of histopathological grade of cervical disease [
9,
10].
Functionally, we showed that PIK3CA and concomitant PI3-kinase signalling is involved in the different stages of HPV-mediated transformation
in vitro. Modulating PI3-kinase activity in HPV transfected keratinocytes affected proliferation, migration, anchorage independent growth, and epithelial growth and differentiation in organotypic cultures. These data are consistent with previous reports using different cellular systems. For instance, in ovarian cancer cell lines PI3-kinase pathway inhibition with LY294002 resulted in reduced proliferation via cell cycle arrest in G1 [
27] or induction of apoptosis [
28]. Proliferation measured by cell number and [
3H]-Thymidine incorporation was also reduced in rat intestinal epithelial cells as a result of LY294002 treatment [
29]. Furthermore, the involvement of PI3-kinase in colony formation has been shown in mammary epithelial cells where overexpression of PIK3CA increased colony formation while a dominant negative form of PIK3R1 (p85) lacking the PIK3CA binding domain repressed colony formation [
30]. This latter study also showed that, although PKB/AKT is the main effector of PI3-kinase, PKB/AKT activation cannot substitute PI3-kinase signalling.
Functional studies in cervical cells are restricted to the common hrHPV positive cervical cancer cell lines such as SiHa, HeLa and CaSki and did not include the explicit analysis of PIK3CA as a candidate oncogene. Treatment of SiHa with the PI3-kinase inhibitor LY294002 led to reduced proliferation and increased apoptotic DNA fragments [
5]. For HeLa and CaSki the reduction in proliferation upon LY294002 treatment was shown to be independent of apoptosis, though did sensitize the cells to radiation [
31].
None of the previous studies on cancer cells have had the benefit of a longitudinal characterization as is afforded by our
in vitro model system, which mimics the different stages of cervical carcinogenesis. We were able to show a progressive upregulation of PI3-kinase signalling during HPV-induced transformation by using the elevation of p-PKB/AKT as a reporter. Indeed, p-PBK/AKT was elevated in anchorage independent cells relative to HPV-immortalized cells. The expression levels of both PIK3CA and especially p-PKB/AKT in the latter cells were increased compared to primary keratinocytes (data not shown). This increased signalling functionally correlated with multiple attributes of transformed cells, such as cell growth, migration, and anchorage independent growth (Figure
3). The fact that specific silencing of PIK3CA using RNA interference affected each of the same transformed phenotypes in a negative manner, while PIK3CA overexpression increased proliferation (Figures
4 and
5) substantiates its function as an oncogene in cervical carcinogenesis, as has been suggested in previous studies [
5,
8,
26].
Remarkably, our data also suggest a feedback effect in which PI3-kinase regulates HPV oncogene expression. HPV16 mRNA expression and hTERT mRNA were decreased after inhibition of PI3-kinase signalling (Figure
6). The downregulation of hTERT mRNA may be a direct result of reduced HPV expression, as E6 activates hTERT transcription [
22]. Thus, the reciprocal regulation between PIK3CA/PI3-kinase and viral oncogene expression acts in concert in maintaining the transformed phenotypes [
32‐
34]. However, it is presently unknown whether the phenotypical effects of PI3-kinase inhibition seen in our model system reflect a direct consequence of E6/E7 repression or vice versa.
It has been shown that the oncoprotein E7 is able to activate PKB/AKT, which appeared to be dependent on its ability to bind and inactivate Rb gene family proteins [
10,
35]. E7 may also maintain PKB/AKT in an active state, by binding and sequestering a known binding partner, the phosphatase PP2A, thereby inhibiting dephosphorylation of p-PKB/AKT [
36]. Active PKB/AKT is also known to activate MDM2, enhancing p53 degradation and elevating CDKs leading to pRb inactivation, further enhancing the established E6/E7 effect on p53 and pRb function [
37,
38]. Moreover, activation of PKB/AKT by PI3-kinase can inhibit nuclear localization of p27kip1 and p21cip by phosphorylating the nuclear localization signal and hereby preventing nuclear transport and inducing proliferation, both of which are also affected by E7 overexpression [
35,
39]. Here we show that besides the reported effect of HPVE7 on PI3-kinase activity, there is also a feed-back effect in that PI3-kinase can regulate HPV oncogene expression. From the present data it becomes clear that both mechanisms act in concert as the increased PI3-kinase activity appears essential for maintaining HPV oncogene expression. The need for a strict regulation of HPV expression, as also implicated in our previous study [
40], suggests that yet other transformation-inducing host cell alterations contribute to the regulated expression of HPVE6/E7 as well as PI3-kinase signalling activity.
Organotypic raft cultures mimic a natural environment to evaluate the role of PI3-kinase signalling in the growth and differentiation of keratinocytes. Phosphorylation of PKB/AKT appeared to be tightly linked to differentiation, which is in agreement with a previous report showing upregulation of phosphorylated PKB/AKT in organotypic raft cultures of HPV16 expressing keratinocytes [
10]. Inhibiting PI3-kinase signalling by chemical inhibition and PIK3CA siRNA-mediated silencing during raft formation led to a dramatic reduction in p-PKB/AKT and severely hampered epithelial cell growth in both HPV16- and HPV18 immortalized cells (Figure
7). Proliferation marker Ki-67 remained detectable in residual epithelial cells, while differentiation marker CK10 was dramatically reduced. These results indicate that PIK3CA regulated p-PKB/AKT expression is important in squamous differentiation. This conclusion is in accordance with a previous report showing that mouse keratinocytes with active PKB/AKT have higher levels of differentiation markers Keratin 1, filaggrin and loricrin. Moreover, inhibition of PI3-kinase resulted in the specific death of differentiating keratinocytes [
23]. Similarly, treatment of primary esophageal keratinocytes with LY294002 resulted in an overall decrease in the number of basal keratinocytes and thickness of the epithelium [
41]. It has been suggested that PI3-kinase becomes important upon commitment of keratinocytes to differentiation, initiating the process and subsequently delivering the required survival signals [
24]. Based on the histology of our raft cultures, there were no apoptotic cells that are typified by highly condense and shrunken nuclei (Figure
7A). The thinning of the epithelium appears to stem from a severely curtailed proliferation. Hence, in organotypic raft culture systems, p-PKB/AKT appears to correlate with squamous differentiation and survival but not with proliferation, as is also evident from the lack of co-localization between p-PKB/AKT and Ki-67 (Figure
7). Also in a study by Menges et al [
10] p-AKT is seen in differentiated non-proliferating BrdU negative cells. These observation are also in line with previous studies showing that AKT knock-out MEFs reduce BrdU incorporation by only 44-61% [
42], indicating that AKT is not indispensible for proliferation, though the lack of it would slow down cell cycle progression at G1-S. We hypothesize that the balance in the dual character of PI3-kinase/PKB/AKT signalling shifts during the process of HPV-induced transformation towards tumor characteristics, rather than differentiation.
Competing interests
The authors Prof dr LT Chow, Prof dr T Broker and dr NS Banerjee are funded by NIH. Other authors have no competing interests.
Authors' contributions
FEH performed experiments and drafted the manuscript. NSB carried out the immunofluorescent stainings. RDMS, PJFS, CJLMM, LTC, TRB participated in design of the study and interpretation of data and helped to draft the manuscript with critical revisions for intellectual content. JDCA and FR participated in HPV expression analysis and critically revised the manuscript. All authors read and approved the final manuscript.