Cervical cancers, a malignancy associated with oncogenic papilloma viruses, remain a major disease burden in the absence of effective implementation of preventive strategies. CD66+ cells have previously been identified as a tumor-propagating subset in cervical cancers. We investigated the existence, differentiation state, and neoplastic potential of CD66+ cells in a precancer cell line harboring HPV31b episomes. The gene expression profile of CD66high cells overlaps with differentiated keratinocytes, neoplastic mesenchymal transition, cells of the squamocolumnar junction, and cervical cancer cell line–derived spheroids. There is elevated expression of DNMT1, Notch1, and the viral gene product E1⁁E4 in CD66high cells. Thus, CD66high cells, in the absence of differentiating signals, express higher levels of key regulators of keratinocytes stemness, differentiation, and the viral life cycle, respectively. We also find a striking association of neoplastic traits, including migration, invasion, and colony formation, in soft agar with CD66high cells. These properties and a distinct G2–M–enriched cell-cycle profile are conserved in cells from cervical cancers. Principally, using a precancerous cell line, we propose that CD66high cells have an intermediate differentiation state, with a cellular milieu connected with both viral replication and neoplastic potential, and validate some key features in precancer lesions. Such pathophysiologically relevant systems for defining cellular changes in the early phases of the disease process provide both mechanistic insight and potential therapeutic strategies. Collectively, our data provide a rationale for exploring novel therapeutic targets in CD66+ subsets during cancer progression. Cancer Res; 74(22); 6682–92. ©2014 AACR.

In tumors such as breast cancers, glioblastomas, and colorectal cancers, tumorigenic subpopulations have been identified and are thought to underlie resistance to therapy and recurrence of tumor (1–6). Such subsets often upregulate the expression of pluripotency factors, Oct4, Nanog, Sox2, and cell survival pathways such as Notch signaling (4, 7–10). We have recently identified a subset of cells in cervical cancers with enhanced tumorigenic and metastatic functions (10). These cells are sustained by Notch signaling and are distinct in their expression of CD66. The transmembrane protein CD66, a member of the carcinoembryonic antigen family, has been implicated in invasive functions in different solid tumors, including ovarian cancer and estrogen-deprived breast cancer cells (11–14). CD66+ cells in cervical cancers have higher expression of the pluripotency factors Oct4 and Nanog, as well as drug transporters (10). Other groups have reported CD49f, a marker of basal undifferentiated keratinocytes, Sox2, one of the induced pluripotency genes, and CD44+ cytokeratin 17+ subsets to be linked to tumorigenic traits and subsets in cervical cancer (15, 16). The identification of these subsets has raised new and unresolved questions about the origin of functional heterogeneity. For instance, it is unclear whether these cells represent a deregulation of a stem cell pool or the induction of a stem-like state in relatively differentiated cells (4). Recent evidence from different systems suggests that differentiated cells can become tumorigenic subsets by hijacking the self-renewal machinery (4,17,18). There is accumulating evidence that these stemness and survival pathways can be invoked in the context of stress response, such as hypoxic niches and the process of epithelial-to-mesenychmal transitions accompanying wound healing (4, 19,20). It is therefore likely that some populations in a tumor can evolve distinct functional features even in the absence of genetic insults, possibly by epigenetic mechanisms.

Currently, an issue that remains unexplored is whether the subsets of cells with unique tumorigenic functions are present and functionally important in the early stages of tumorigenesis (4,21). Cervical precancers or cervical intraepithelial neoplasias (CIN) arise due to persistent infection with the high-risk papilloma viruses, including HPV16, 18, 45, and 31 (22–24). Here, we use the CIN-612 culture system to analyze a putative tumorigenic population in early cervical lesions and ascertain mechanistic links using primary keratinocytes transfected with papilloma virus genomes. CIN-612 cells are derived from a natural infection with HPV31b (25). They represent an early phase of the disease process as they maintain low copies of the viral genome as epsiomes (25, 26). The unique property of this cell line is its ability to support viral replication upon differentiation; thus, these cells exist as an undifferentiated pool, with similarities to CIN1 lesions (25, 26). This cell line is therefore amenable for the study of papilloma virus–related changes to keratinocytes, such as regulation of genes required for the viral life cycle, in the critical, clinically relevant, window of early disease.

Cell culture and reagents

CIN-612 9E cells, primary keratinocytes (HFKs), HFKs transfected with HPV genomes, and CaSki spheroids have been described before (10,25–29). CIN-612 cells and HFKs were cultured in E medium supplemented with EGF [mouse EGF (BD Biosciences) in Fig. 3, human recombinant EGF (Peprotech) in Figs. 1, 2, and 4–6], and differentiated as described previously (26–29). CIN-612 cells were routinely (at the time of the experiments) characterized for episomal HPV31 maintenance and differentiation potential in rafts. Cultures were used within 25 passages of acquisition for CaSki and SiHa (bought from ATCC), early passages for HFKs, and transfected HFKs. All cultures were routinely tested for mycoplasma and retention of described morphology and growth features. Furthermore, CaSki explants in mice generate tumors as per previous reports (10). CIN-612 cells were transduced as described earlier (30) with virions carrying shRNAs against DNMT1 (Origene for Supplementary Fig. 3G and plasmids described in ref. 30; Fig. 6J). The HPV16 E6 mutants have been described before (29).

Figure 1.

CD66high cells in CIN-612 express higher levels of Notch1, DNMT1, and p63. A, undifferentiated CIN-612 cells (10,000 cells) were analyzed for CD66 expression by flow cytometry. Density plots show CD66 expression (dark gray) and negative/isotype control (light gray). B, box plot of percentage of CD66high cells. N = 20; median = 24.10. C, sorted CD66low and CD66high cells were examined for DNMT1, NICD, and p63 proteins. GAPDH was used as a loading control. N = 5, means with SEM are shown. D, density plots show DNMT1 expression in 8,000 CD66high (dark gray), CD66low (light gray) cells, and the secondary control (white) in undifferentiated CIN-612 cells by flow cytometry. E, density plots show Notch1 expression in 10,000 CD66high (dark gray), CD66low (light gray) cells, and secondary control (white) in undifferentiated CIN-612 cells by flow cytometry.

Figure 1.

CD66high cells in CIN-612 express higher levels of Notch1, DNMT1, and p63. A, undifferentiated CIN-612 cells (10,000 cells) were analyzed for CD66 expression by flow cytometry. Density plots show CD66 expression (dark gray) and negative/isotype control (light gray). B, box plot of percentage of CD66high cells. N = 20; median = 24.10. C, sorted CD66low and CD66high cells were examined for DNMT1, NICD, and p63 proteins. GAPDH was used as a loading control. N = 5, means with SEM are shown. D, density plots show DNMT1 expression in 8,000 CD66high (dark gray), CD66low (light gray) cells, and the secondary control (white) in undifferentiated CIN-612 cells by flow cytometry. E, density plots show Notch1 expression in 10,000 CD66high (dark gray), CD66low (light gray) cells, and secondary control (white) in undifferentiated CIN-612 cells by flow cytometry.

Close modal
Figure 2.

CD66high cells are differentiated along with progenitor/stemness profiles. A, genes upregulated in CD66high cells were subjected to Gene Ontology–based classification using the DAVID gene term classification tool. The top four enriched clusters are shown with their Benjamini-corrected EASE scores (−log10). B, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with 216 genes upregulated in keratinocytes upon differentiation in Ca2+-containing media (30). ES = 0.63605547; NES = 1.9540734; nominal P < 0.001; FDR q < 0.001; FWER P < 0.001. C, expression of pluripotency factors in sorted CD66high versus CD66low cells from undifferentiated CIN-612 cells by real-time PCR. D, gene expression of indicated genes in sorted CD66high versus CD66low cells from CaSki spheroids. C and D, N = 3 sorts; error bars, SEM. Fold change of 1 is shown as a horizontal line.

Figure 2.

CD66high cells are differentiated along with progenitor/stemness profiles. A, genes upregulated in CD66high cells were subjected to Gene Ontology–based classification using the DAVID gene term classification tool. The top four enriched clusters are shown with their Benjamini-corrected EASE scores (−log10). B, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with 216 genes upregulated in keratinocytes upon differentiation in Ca2+-containing media (30). ES = 0.63605547; NES = 1.9540734; nominal P < 0.001; FDR q < 0.001; FWER P < 0.001. C, expression of pluripotency factors in sorted CD66high versus CD66low cells from undifferentiated CIN-612 cells by real-time PCR. D, gene expression of indicated genes in sorted CD66high versus CD66low cells from CaSki spheroids. C and D, N = 3 sorts; error bars, SEM. Fold change of 1 is shown as a horizontal line.

Close modal

Cell sorting

CIN-612 cells were sorted using the Pluriselect Kit. Pluriselect Universal Mouse S beads (Pluriselect) were conjugated to the CD66 antibody (BD Pharmigen, 551354) according to the manufacturer's instructions. Sorting of CD66 cells from SCCs and from CaSki spheroids was performed as described previously (10).

Flow cytometry and immunofluorescence

Live cell staining was performed with PE-conjugated anti-CD66 (BD 553457) or isotype control (BD 555574, BD 551480) at room temperature for 30 minutes in HBSS buffer. Cells with intensity greater than mean + 3 SDs of the isotype (live staining)/secondary control (for fixed cells) were considered CD66high and the rest considered CD66low. Annexin V staining was performed according to the manufacturer's instructions (BD Pharmigen, 556420). For immunofluorescence of fixed cells, 0.5 × 106 cells were fixed in 4% PFA, blocked with blocking solution (10% serum, 0.2% Triton X-100, NaAzide in PBS), stained for 1 hour (primary), followed by 30 minutes (secondary) at room temperature with antibodies against Notch1 (Santa Cruz Biotechnology, sc-6074), DNMT1 (Abcam, ab19905), HPV31 E1⁁E4 (L.A. Laimins laboratory), PML (Santa Cruz, sc-966), CD66 and Alexa Fluors 488 and 647 (Molecular Probes). Cells were analyzed by flow cytometry on BD FACSCalibur and imaged on Zeiss LSM510 Meta confocal microscope. Analysis was performed using FlowJo, R, Zeiss LSM viewer and ImageJ. Immunohistochemistry was performed after antigen retrieval (sodium citrate, pH 6.0), using the Novolink Polymer Detection System (Leica Biosystems) and imaged on Nikon ECLIPSE TE2000-S. Antibodies against CD66 and pan HPV E1⁁E4 (gift from Dr. Sally Roberts) were used.

Western blots

Immunoblots were performed using antibodies against the following DNMT1 (Abcam, ab19905), Notch1 (Santa Cruz Biotechnology, sc-6074), CHK2 (Santa Cruz Biotechnology, sc-5278), pCHK2, Thr68 (Cell Signaling Techonology, 2661), NICD (Cell Signaling Technology, 2421), p63 (Santa Cruz Biotechnology, sc-8431), GAPDH (Santa Cruz Biotechnology, sc-47724), KRT14 (Abcam, ab7800), cleaved caspase-3 (Cell Signaling Technology, 9661), PML (Santa Cruz Biotechnology, sc-966), CDC2 (Santa Cruz Biotechnology, sc-137034), pCDC2, Tyr15, (Cell Signaling Technology, 9111). Densitometry was performed using ImageJ. Cropped blots are shown.

Cell-cycle analysis

A total of 0.4 × 106 cells was stained with the CD66 primary antibody on ice for 30 minutes, after which they were fixed in ice-cold methanol for 15 minutes. Cells were stained with Alexa Fluor 488 followed by DRAQ5 (Biostatus) and acquired on BD FACSCalibur.

Soft agar assay

Five-thousand sorted CD66high and CD66low cells were seeded for colony formation in soft agar as described before (10,31). Cells were treated with 5 μmol/L 5-azadeoxycytidine (5AZA-dC; Sigma)/DMSO for 72 hours before sorting. The number of colonies in 10 random fields was counted under the 10× objective (Olympus CK40).

Microarray analysis

RNA was extracted from 0.5–0.6 × 106 sorted cells using the RNeasy Kit (Qiagen). RNA quality was checked using the Bioanalyzer, labeled using the Agilent's Quick-Amp labeling Kit, and hybridized to the Human 4 × 44K Array. Intra-array normalization was performed using the GeneSpringGX 11.5 Software (Genotypic Inc). The gene expression data have been deposited in the GEO database (accession number –GSE59322). Gene ids (upregulated >1.5-fold in CD66high cells) were mapped to the Unigene database and gene term enrichment analysis was performed using DAVID web tools using only the Gene Ontology databases. Enrichment threshold of 2 and P value cutoff of 0.05 were used to make representative plot of the top clusters. Gene Set Enrichment Analysis (GSEA) was performed using the GSEA tool against available or custom datasets described in the text (33). GSEA enrichment plots are shown and details can be found in the documentation of the GSEA tool.

Gene expression and HPV DNA content by real-time PCR

RNA was isolated using TRIzol (Invitrogen) and treated to remove DNA (Fermentas DNaseI). cDNA synthesis was performed using MuMLV (Invitrogen). RPLP0 or GAPDH were used as a control for equal loading. DNA was isolated either by detergent lysis or using the DNeasy Kit (Qiagen). RRP40 was used as a control for equal amount of genomic DNA. Real-time PCRs were performed on the ABI7500 machine using the Kapa SYBR Fast qPCR Kit. Primer sequences are listed in the Supplementary Material and Methods.

Migration and invasion assays

Migration and invasion assays were performed as described before (10). Five percent FBS for CIN-612, 10% FBS for SiHa cells, and intact CaSki spheroids were used as chemoattractant. The outer side of the membrane was stained, and the number of nuclei was counted.

Detection of a CD66high subset with higher protein levels of Notch1, DNMT1, and p63 in CIN-612 cells

We examined CIN-612 cells for the surface expression of the CD66 protein (Fig. 1A). Using an antibody against CD66, we defined a CD66high subset with a frequency of around 24% (Fig. 1B). To establish whether the CD66high cells were differentiated or progenitor-like (basal), we estimated the protein levels of DNMT1, p63, KRT14, and intracellular Notch1 in sorted CD66high and CD66low cells (Fig. 1C and Supplementary Fig. S1A). While DNMT1 and p63 are known to maintain keratinocyte stemness, KRT14 is a marker of basal keratinocytes, whereas Notch signaling is a key regulator of keratinocyte differentiation (30, 34–37). We find that CD66high cells have higher protein levels of DNMT1, p63, cleaved intracellular Notch1 (NICD), and comparable levels of KRT14 to CD66low cells (Fig. 1C and Supplementary Fig. S1A). We confirmed these results by costaining CIN-612 cells for CD66 and Notch1/DNMT1 (Fig. 1D and E Supplementary Fig. 1B–F). Collectively, CD66high cells are Notch1/NICDhigh, DNMT1high, and p63high in CIN-612 (Fig. 1C–E and Supplementary Fig. S1B–S1F). The expression of p63, DNMT1, and KRT14 is normally restricted to the basal (undifferentiated) cells and Notch1 to the partially differentiated suprabasal cells in the epithelium (30, 34–37). The coexpression of these proteins in the CD66high subset predicts a state consistent with both self-renewal and differentiation.

CD66high cells are differentiated along with progenitor/stemness profiles

To resolve the differentiation state that we found in the CD66high cells, we combined gene expression analysis with the transcript analysis of the reprogramming factors, Oct4, Nanog, and Sox2 (38). Gene expression profiling was performed using microarrays on CIN-612 cells sorted on the basis of CD66 expression. A total of 924 transcripts were upregulated (>1.5-fold) and 204 downregulated (<1.5-fold) in the CD66high cells versus CD66low cells. The genes upregulated in the CD66high cells were clustered using the DAVID functional annotation clustering tool (32). Keratinocyte differentiation–related genes were among the top enriched clusters (Fig. 2A and Supplementary Fig. S2A). This was supported by GSEA showing enrichment of genes expressed in keratinocytes upon differentiation in CD66high cells (Fig. 2B; refs. 30, 33). These data place the CD66high cells in a differentiated keratinocyte state. In contrast, gene expression analysis by real-time PCR revealed that CD66high subset also had increased levels of Oct4, Nanog, and Sox2 (Fig. 2C). These data demonstrates a component of stemness in the CD66high subset, which is supported by the higher levels of DNMT1 and p63 protein in these cells (Fig. 1C and D and Supplementary Fig. S1B–S1D). We also find this mixed pattern of the expression of stemness- and differentiation-related genes in sorted CD66high cells from spheroids of CaSki, a cervical cancer cell line (Fig. 2D and Supplementary Fig. S2B and S2C) and from primary cervical cancers (Supplementary Fig. S2C). We thus establish that the CD66high subset has features consistent with both stemness and keratinocyte differentiation.

Papilloma virus genomes enhance protein expression of both Notch1 and DNMT1

Papilloma virus genomes are known to regulate host proteins; for instance, p63 is a known target of the papilloma virus genome and is required for the viral late gene expression and viral amplification (27). On the basis of our results from Fig. 1, we asked whether papilloma virus genomes regulate the levels of Notch1 and DNMT1, and whether their expression is sustained upon differentiation. Activated Notch1 has been shown to cooperate with the papilloma virus oncogenes in transformation assays (31, 39). Recent evidence suggests that papilloma virus genomes alter methylation patterns in the host via DNMT1, whereas DNMT1 itself plays a role in carcinogenesis (40, 41).

Keratinocytes transfected with HPV31 genomes upregulate cleaved intracellular Notch1 and DNMT1 proteins (Fig. 3A and B). Upon differentiation of these cells in methylcellulose, although NICD levels decrease, DNMT1 levels are maintained (Fig. 3A and B). When these cells are differentiated in calcium, DNMT1 levels decrease but are still higher than the corresponding untransfected keratinocytes (Supplementary Fig. S3A). Differentiation of primary keratinocytes is accompanied by a reduction in DNMT1 and Notch1 protein levels (Fig. 3A and B).

Figure 3.

HPV31 genomes enhance both Notch1 and DNMT1 levels. A, expression of DNMT1 in primary keratinocytes (HFK) compared with HFKs transfected with HPV31 genomes (HFK + HPV31), in undifferentiated (U) and after differentiation (D) in methylcellulose. B, expression of cleaved intracellular Notch1 (NICD) in HFKs compared with HFK + HPV31, in undifferentiated (U) and after differentiation (D) in methylcellulose. C, expression of DNMT1 in undifferentiated CIN-612 cells (UD) and after differentiation in Ca2+ for 24 hours [D(24)] or 48 hours [D(48)]. D, immunoblot of DNMT1 and NICD levels in CIN-612 cells in undifferentiated (U) conditions and after differentiation in methylcellulose for 24 hours [D(24)] or 48 hours [D(48)]. E, immunoblot of Notch1 levels in CIN-612 cells before (U) and after differentiation in methylcellulose for 24 hours [D(24)] and 48 hours [D(48)]. Notch1* represents full-length and Notch1** represents intracellular levels of Notch1. A–E, GAPDH was used as a loading control. N = 3; mean with SEM are shown below the blots.

Figure 3.

HPV31 genomes enhance both Notch1 and DNMT1 levels. A, expression of DNMT1 in primary keratinocytes (HFK) compared with HFKs transfected with HPV31 genomes (HFK + HPV31), in undifferentiated (U) and after differentiation (D) in methylcellulose. B, expression of cleaved intracellular Notch1 (NICD) in HFKs compared with HFK + HPV31, in undifferentiated (U) and after differentiation (D) in methylcellulose. C, expression of DNMT1 in undifferentiated CIN-612 cells (UD) and after differentiation in Ca2+ for 24 hours [D(24)] or 48 hours [D(48)]. D, immunoblot of DNMT1 and NICD levels in CIN-612 cells in undifferentiated (U) conditions and after differentiation in methylcellulose for 24 hours [D(24)] or 48 hours [D(48)]. E, immunoblot of Notch1 levels in CIN-612 cells before (U) and after differentiation in methylcellulose for 24 hours [D(24)] and 48 hours [D(48)]. Notch1* represents full-length and Notch1** represents intracellular levels of Notch1. A–E, GAPDH was used as a loading control. N = 3; mean with SEM are shown below the blots.

Close modal

A similar trend is seen in CIN-612 cells, where there is a decrease in the protein levels of both NICD and DNMT1 upon differentiation in methylcellulose (Fig. 3D and E). In calcium-containing medium, although there is an overall decrease in DNMT1 protein levels, it is still maintained at a high level in a fraction of cells (Fig. 3C and Supplementary Figs. S3A and S1D).

In line with previous reports, we also find an increase in DNMT1 expression with HPV16 and 18 genomes (Supplementary Fig. S3B and S3C; ref. 40). As sustained expression of E6 and E7 is critical to papilloma virus mediated tumor progression (22), we examined whether the viral oncogenes can also upregulate DNMT1. We find that while E6 and E7 from HPV31, HPV18, and E7 from HPV16 are able to upregulate DNMT1 (Supplementary Fig. S3D and S3E) an E6 mutant, I128T (29) is unable to do so (Supplementary Fig. S3F). This mutant is defective in its interaction with the ubiquitin ligase E6-associated protein (E6AP). As a result, these cells have higher steady-state levels of p53. These data argue for a role for HPV16 E6 in maintaining DNMT1 levels.

We tested whether DNMT1 is required for the viral life cycle and found that cells expressing shRNA against DNMT1 are defective in differentiation-dependent expression of viral gene E1⁁E4 (Supplementary Fig. S3G). Treatment of CIN-612 cells with the DNA methylation inhibitor 5AZA-dC also resulted in lower clonogenic ability and decrease in the amount of papilloma virus DNA (Supplementary Fig. S3H and S3I).

These data implicate the papilloma virus genomes in the upregulation of DNMT1 and Notch1 proteins, sustaining DNMT1 levels even upon differentiation, and reveal a role for DNA methylation in the papilloma virus life cycle. It also prompts an examination of the CD66high cells, which are DNMT1high and Notch1high for links with the viral life cycle.

CD66high cells show a differentiation-dependent link with the papilloma virus life cycle

The cells of origin in cervical cancer are thought to reside in the squamocolumnar junction (SCJ) region of the cervix (22, 23,42). Persistent infection with the high-risk papilloma viruses is known to induce cervical cancers predominantly in this region (22, 23, 42). Papilloma viruses require keratinocyte differentiation for the completion of their life cycle (22–24). In particular, the expression of the late gene E1⁁E4 and amplification of the viral genome occur only upon differentiation (22, 24, 43). Viral replication centers, which are formed in a subset of cells upon differentiation, coincide with PML bodies or ND10 structures and recruit molecules such as CHK2 and H2AX (44–48). On the basis of the results in Figs. 1–3, we asked whether CD66high cells in CIN-612 were in fact linked to the papilloma virus life cycle and to the cells in the SCJ.

To address the link to the viral life cycle, we analyzed three features: the amount of HPV DNA, expression of viral replication–associated proteins, and the viral protein E1⁁E4. CD66high cells have higher amounts of HPV DNA both before and after differentiation (Fig. 4A). Sorted CD66high cells from undifferentiated CIN-612 cells have higher protein levels of PML and marginally higher levels of CHK2 and pCHK2 (Fig. 4B). Costaining of undifferentiated CIN-612 cells shows PML positive nuclear structures in CD66high cells (Fig. 4C). We assessed E1⁁E4 levels by costaining with CD66 followed by flow cytometry. We find that E1⁁E4high cells are CD66high and confirm this by imaging (Fig. 4D and F and Supplementary Fig. S4A and S4B). We find a positive correlation between the expression of E1⁁E4 and CD66 in undifferentiated cells (R2 > 0.5), which decreases upon differentiation (Fig. 4D and E). In organotypic rafts, the bottom compartment with the undifferentiated cells has cells that coexpress CD66 and E1⁁E4 (Supplementary Fig. S4C and S4D). In the top compartment with the differentiated cells, CD66high and E1⁁E4high cells are often distinct (Supplementary Fig. S4C). Collectively, we find a CD66high E1⁁E4high subset in undifferentiated conditions with features of viral replication.

Figure 4.

CD66high cells are linked to the differentiation-dependent papilloma virus life cycle. A, HPV31 DNA content in sorted CD66high versus CD66low cells. Left, undifferentiated cultures of CIN-612; right, differentiated in Ca2+, N = 3, error bars, SEM; **, P = 0.004, Welch t test. B, expression by Western blot analysis of PML, pCHK2, and CHK2 in sorted CD66high versus CD66low cells from undifferentiated CIN-612 cells. GAPDH was used as a loading control. C, expression of PML by immunofluorescence in undifferentiated CIN-612 cells, CD66c (red), PML (green), DAPI (nuclei, blue). Left, CD66high cell; right, CD66low. Scale bar, 10 μm. D and E, scatter plots showing E1⁁E4 and CD66 expression in 10,000 CIN-612 cells, undifferentiated (Undiff; D) anddifferentiated (Diff; E) in Ca2+. R2 = Pearson coefficient of correlation. F, relative expression levels of E1⁁E4 in CD66low and CD66high cells. N = 3; 8,000 cells/experiment. Error bars, SEM. *, P = 0.038, Welch t test. G, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with a 68 gene signature of cells of the SCJ (42). ES = 0.6349658; NES = 1.7570925. Nominal P < 0.0002; FDR q value < 0.0002; FWER P < 0.0002.

Figure 4.

CD66high cells are linked to the differentiation-dependent papilloma virus life cycle. A, HPV31 DNA content in sorted CD66high versus CD66low cells. Left, undifferentiated cultures of CIN-612; right, differentiated in Ca2+, N = 3, error bars, SEM; **, P = 0.004, Welch t test. B, expression by Western blot analysis of PML, pCHK2, and CHK2 in sorted CD66high versus CD66low cells from undifferentiated CIN-612 cells. GAPDH was used as a loading control. C, expression of PML by immunofluorescence in undifferentiated CIN-612 cells, CD66c (red), PML (green), DAPI (nuclei, blue). Left, CD66high cell; right, CD66low. Scale bar, 10 μm. D and E, scatter plots showing E1⁁E4 and CD66 expression in 10,000 CIN-612 cells, undifferentiated (Undiff; D) anddifferentiated (Diff; E) in Ca2+. R2 = Pearson coefficient of correlation. F, relative expression levels of E1⁁E4 in CD66low and CD66high cells. N = 3; 8,000 cells/experiment. Error bars, SEM. *, P = 0.038, Welch t test. G, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with a 68 gene signature of cells of the SCJ (42). ES = 0.6349658; NES = 1.7570925. Nominal P < 0.0002; FDR q value < 0.0002; FWER P < 0.0002.

Close modal

We then examined the relationship between the CD66high cells in CIN-612 and cells in the SCJ, which are thought to be the targets of papilloma virus–mediated transformation, using GSEA analysis. There is an enrichment of the SCJ gene signature (42) in CD66high cells (Fig. 4G). The similarity between junctional cells, which are thought to be the cells of origin of cervical cancer and CD66high cells is consistent with neoplastic potential.

CD66high cells have pro-oncogenic features of survival and growth in vitro

We proceeded to assess the neoplastic abilities of the CD66high cells from CIN-612. The cell-cycle profile of CD66high cells, showed an enrichment of CD66high cells in the G2–M phase of the cell cycle, together with higher levels of pCDC2 (Fig. 5A and B and Supplementary Fig. S5A). This property is conserved in CD66high cells from SiHa, a cervical cancer cell line (Fig. 5C). In CaSki spheroids, CD66high cells are slowly dividing cells (Supplementary Fig. S5D). The lower sub-G1 fraction in the CD66high cells prompted us to look at markers of apoptosis in CD66low cells (Fig. 5A and Supplementary Fig. S5A). CD66low cells had higher levels of AnnexinV staining (Fig. 5D and Supplementary Fig. S5B) and elevated cleaved caspase-3 protein levels (Supplementary Fig. S5C). CD66high cells also form more colonies in soft agar (Fig. 5E). CIN-612 cells treated with 5AZA-dC before sorting form significantly less colonies than DMSO-treated sorted cells (Fig. 5F). Collectively, these assays demonstrate that CD66high cells have features of enhanced survival and transformation, thus revealing marked neoplastic potential in the precancer stage itself.

Figure 5.

CD66high cells have pro-oncogenic features of survival and growth in vitro. A, CD66high cells and CD66low cells have distinct cell-cycle profiles. Undifferentiated CIN-612 cells were stained for CD66 expression and the DNA-binding dye DRAQ5. Representative plots of distribution of DNA (<2n, 2n, 4n, >4n) in CD66low (left, 19,427 cells) and CD66high (right, 19,193 cells) cells. B, expression of CDC2, pCDC2 (Tyr15) in sorted CD66low and CD66high cells from undifferentiated CIN-612 cells. GAPDH was used as a loading control. C, SiHa cells were stained for CD66 expression and the DNA-binding dye DRAQ5. Representative plots of distribution of DNA (<2n, 2n, 4n, >4n) in CD66low (left, 1381,660 cells) and CD66high (right, 17,046 cells) cells. D, binding of AnnexinV in 3,000 CD66high (dark gray) and CD66low (light gray) cells is shown in the density plots. E, sorted cells from undifferentiated CIN-612 cells were seeded for colony formation in soft agar. Ratio of colonies from CD66high and CD66low cells after 10, 14, and 21 days (10, 14, 21 days, respectively). N = 3 sorts; error bars, SEM; *, P = 0.0163, Welch t test. F, undifferentiated CIN-612 cells were treated with 5AZA-dC or DMSO before sorting and seeded for colony formation in soft agar. Number of colonies from treated (5AZA-dC, 5 μmol/L, 72 hours)/untreated (DMSO) CD66high and CD66low cells is plotted. Colonies were counted on day 10, 14, and 21. N = 3; error bars, SEM; **, P = 0.02; *, 0.032; Welch t test.

Figure 5.

CD66high cells have pro-oncogenic features of survival and growth in vitro. A, CD66high cells and CD66low cells have distinct cell-cycle profiles. Undifferentiated CIN-612 cells were stained for CD66 expression and the DNA-binding dye DRAQ5. Representative plots of distribution of DNA (<2n, 2n, 4n, >4n) in CD66low (left, 19,427 cells) and CD66high (right, 19,193 cells) cells. B, expression of CDC2, pCDC2 (Tyr15) in sorted CD66low and CD66high cells from undifferentiated CIN-612 cells. GAPDH was used as a loading control. C, SiHa cells were stained for CD66 expression and the DNA-binding dye DRAQ5. Representative plots of distribution of DNA (<2n, 2n, 4n, >4n) in CD66low (left, 1381,660 cells) and CD66high (right, 17,046 cells) cells. D, binding of AnnexinV in 3,000 CD66high (dark gray) and CD66low (light gray) cells is shown in the density plots. E, sorted cells from undifferentiated CIN-612 cells were seeded for colony formation in soft agar. Ratio of colonies from CD66high and CD66low cells after 10, 14, and 21 days (10, 14, 21 days, respectively). N = 3 sorts; error bars, SEM; *, P = 0.0163, Welch t test. F, undifferentiated CIN-612 cells were treated with 5AZA-dC or DMSO before sorting and seeded for colony formation in soft agar. Number of colonies from treated (5AZA-dC, 5 μmol/L, 72 hours)/untreated (DMSO) CD66high and CD66low cells is plotted. Colonies were counted on day 10, 14, and 21. N = 3; error bars, SEM; **, P = 0.02; *, 0.032; Welch t test.

Close modal

The neoplastic features of CD66high cells include an association with migration and invasion

We have noted that CD66+ cells are enriched in CaSki spheroids and have enhanced migratory and invasive properties (10). A comparison of the gene expression in CD66high and CD66low cells from CIN-612 with CaSki spheroids showed a striking correlation between the genes highly expressed in CaSki spheroids and in the CD66high cells from CIN-612 (Fig. 6B). This was underscored by GSEA analysis showing that gene expression of CD66high cells positively correlates with genes in the “mesenchymal transition signature” common to all invasive cancer types (Fig. 6A; ref. 49). We therefore reasoned that CD66high cells in CIN-612 may have functional features thought to be acquired only in late stages of carcinogenesis. Consistent with this idea, we find that CIN-612 cells are capable of migration and invasion and that CD66high cells form a major fraction of both migratory and invasive cells (Fig. 6C–F). DNMT1-depleted CIN-612 cells show a mild defect in migration (Fig. 6J). The migratory cells from CaSki and SiHa, like CIN-612 are predominantly CD66high (Fig. 6G and H). In addition, the edges of primary tumors and CINs are enriched for CD66high cells (Fig. 6I and Supplementary Fig. S6A–C).

Figure 6.

The neoplastic features of CD66high cells include an association with migration and invasion. A, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with the indicated gene set from the MsigDB (M2572). ES = 0.571464; NES = 1.5786825; Nominal P = 0.008526187; FDR q value = 0.008526187; FWER P = 0.007. B, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with top 297 genes upregulated in CaSki cells grown as spheroids (10). ES = 0.55261254; NES = 1.741556; Nominal P < 0.001; FDR q value < 0.001; FWER P < 0.001. C, CD66 staining in CIN-612 after migration; left, secondary control; right, CD66 (green), DAPI (nuclei, blue). Scale bar, 10 μm. D, fraction of CD66high cells in the migratory cells from CIN-612. E, CD66 staining in CIN-612 after invasion; left, secondary control; right, CD66 (green) staining; DAPI (nuclei, blue). Scale bar, 10μm. F, fraction of CD66high cells in the invasive cells from CIN-612. For D and F, N = 3; 100 cells/experiment; error bars, SEM; P = 0.00012 and 0.0010, respectively, Welch t test. G, representative image of CD66 expression in the migratory cells from CaSki spheroids after 24 hours of migration. CD66 (right, green); DAPI (left, nuclear stain, blue). Scale bar, 100 μm. H, representative image of CD66 expression in the migratory cells from SiHa after 24 hours of migration. CD66 (green); DAPI (nuclear stain, blue). Scale bar, 10 μm. I, primary cancer section showing immunofluorescence of DAPI (left, nuclear stain, blue) and CD66 (right, green). Scale bar, 20 μm. J, effect of DNMT1 depletion on migration of CIN-612 cells. Number of migratory cells from 10 random fields were counted and fold change was calculated against untransduced control (control) and plotted for scrambled control (Scram) and two shRNAs to DNMT1, shDNMT1 1, and shDNMT1 2. N = 3; error bars, SEM. *, P = 0.0361, Welch t test.

Figure 6.

The neoplastic features of CD66high cells include an association with migration and invasion. A, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with the indicated gene set from the MsigDB (M2572). ES = 0.571464; NES = 1.5786825; Nominal P = 0.008526187; FDR q value = 0.008526187; FWER P = 0.007. B, GSEA analysis of gene expression profiles of CD66high and CD66low cells compared with top 297 genes upregulated in CaSki cells grown as spheroids (10). ES = 0.55261254; NES = 1.741556; Nominal P < 0.001; FDR q value < 0.001; FWER P < 0.001. C, CD66 staining in CIN-612 after migration; left, secondary control; right, CD66 (green), DAPI (nuclei, blue). Scale bar, 10 μm. D, fraction of CD66high cells in the migratory cells from CIN-612. E, CD66 staining in CIN-612 after invasion; left, secondary control; right, CD66 (green) staining; DAPI (nuclei, blue). Scale bar, 10μm. F, fraction of CD66high cells in the invasive cells from CIN-612. For D and F, N = 3; 100 cells/experiment; error bars, SEM; P = 0.00012 and 0.0010, respectively, Welch t test. G, representative image of CD66 expression in the migratory cells from CaSki spheroids after 24 hours of migration. CD66 (right, green); DAPI (left, nuclear stain, blue). Scale bar, 100 μm. H, representative image of CD66 expression in the migratory cells from SiHa after 24 hours of migration. CD66 (green); DAPI (nuclear stain, blue). Scale bar, 10 μm. I, primary cancer section showing immunofluorescence of DAPI (left, nuclear stain, blue) and CD66 (right, green). Scale bar, 20 μm. J, effect of DNMT1 depletion on migration of CIN-612 cells. Number of migratory cells from 10 random fields were counted and fold change was calculated against untransduced control (control) and plotted for scrambled control (Scram) and two shRNAs to DNMT1, shDNMT1 1, and shDNMT1 2. N = 3; error bars, SEM. *, P = 0.0361, Welch t test.

Close modal

CINs have a distinct pattern of CD66 expression overlapping with regions that express E1⁁E4

We have previously shown that CD66high cells are present in primary invasive cervical cancers and at their metastatic sites (10). Here, we asked whether CD66 expression could be detected in the early stages of cervical cancers. Consistent with expected prevalence, we noted about 30% HPV16 and 10% HPV31 positivity in the CIN lesions that we screened (Supplementary Fig. S6E). In a limited analysis of sections from CINs, we find CD66 expression mostly in the top third of the epithelium with differentiated cells, with a few CD66-expressing cells in the lower layers with undifferentiated cells. To test the hypothesis that CD66high cells are E1⁁E4high, we stained serial sections with antibodies against CD66 and a pan HPV E1⁁E4 (Fig. 7A and B). Collectively, we find overlapping staining of CD66 and E1⁁E4, analogous to our results in Fig. 4. We also note that in the histologically normal cervical epithelium, CD66 expression is restricted only to the top layers in cells with a definite flattened, stratified squamous morphology (Supplementary Fig. S6D). In CINs, CD66 expression is seen in the bottom layers, which contain the undifferentiated cells, consistent with both ectopic expression and movement of these cells (Fig. 7A–C).

Figure 7.

CD66 expression localizes to areas that also express E1⁁E4 and extends to the undifferentiated compartment in CINs, A and B, immunohistochemical analysis of CIN sections for the expression of CD66 and pan HPV E1⁁E4. Top, regions of E1⁁E4 expression (brown); bottom, regions of CD66 expression (brown) from the same lesion (serial sections). Sections were counterstained with hematoxylin (blue, nuclear stain). Scale bar, 50 μm. Vertical black arrows represent the corresponding regions from the two images CD66high and E1⁁E4high. Horizontal blue arrow shows CD66 staining in the bottom layers. C, schematic of CD66 and E1⁁E4 expression in the CINs, left, productive viral life cycle stage; right, precancerous stage.

Figure 7.

CD66 expression localizes to areas that also express E1⁁E4 and extends to the undifferentiated compartment in CINs, A and B, immunohistochemical analysis of CIN sections for the expression of CD66 and pan HPV E1⁁E4. Top, regions of E1⁁E4 expression (brown); bottom, regions of CD66 expression (brown) from the same lesion (serial sections). Sections were counterstained with hematoxylin (blue, nuclear stain). Scale bar, 50 μm. Vertical black arrows represent the corresponding regions from the two images CD66high and E1⁁E4high. Horizontal blue arrow shows CD66 staining in the bottom layers. C, schematic of CD66 and E1⁁E4 expression in the CINs, left, productive viral life cycle stage; right, precancerous stage.

Close modal

Tumor heterogeneity has been analyzed in different cancers from two main perspectives, first, the frequency and properties of subsets (1–5) and, second, mechanisms by which these subsets are generated and sustained (4, 6, 8, 17–21). Our focus in this study has been on defining the existence of CD66+ cells and their properties in the early phase of human cervical cancers. There are two striking observations in this study. A CD66high subset with neoplastic properties emerges in the early phase of cervical cancers (Figs. 5–7) and this subset is linked to the papilloma virus life cycle (Figs. 4 and 7).

Our findings suggest a basis for these properties, namely, a unique state of differentiation. We find that CD66high cells have components of both keratinocyte differentiation and stemness as characterized by gene expression profiling, expression of reprogramming factors, and the pattern of expression of Notch1, DNMT1, and viral protein E1⁁E4 in these cells (Figs. 1, 2, and 4). It is likely that the stemness component is required for the development of neoplastic properties. The simultaneous overlapping differentiated state is likely to be permissive for differentiation-dependent viral life cycle events, which are also required for transformation.

In line with this idea, we find that CD66high cells from the precancer derived CIN-612 are capable of anchorage-independent growth, make up the major migratory and invasive fraction in these cells, and show a dependence on DNA methylation for these properties (Figs. 5 and 6). We find a striking G2–M fraction in the CD66high cells (Fig. 5). This can be explained in part by high levels of E1⁁E4 (43, 50); however, the conservation of this signature in SiHa requires further analysis. Our data extend recent observations on the CD66high subset in advanced human cervical cancers where we have noted a relationship with tumor-propagating potential and metastasis (10). As our approach has involved using CD66 as a marker to sort and study subsets, a functional role for CD66 itself needs further evaluation.

In conclusion, our data provide novel insights into the generation of critical subsets at the intersection of the papilloma virus life cycle and cellular transformation. The mixed differentiation–stemness state in the CD66high cells suggests the possibility of a reprogramming event (Supplementary Fig. S7). Future work should be aimed at evaluating whether differentiated CD66high cells, which have already undergone some viral replication, are reprogrammed to a stem-like state by the upregulation of molecules like DNMT1 and p63 by the HPV genomes.

No potential conflicts of interest were disclosed.

Conception and design: C. Pattabiraman, S. Hong, V.K. Gunasekharan, J. Bajaj, V.G. Giri, L.A. Laimins, S. Krishna

Development of methodology: C. Pattabiraman, V.K. Gunasekharan, J. Bajaj, S. Srivastava, L.A. Laimins, S. Krishna

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Pattabiraman, S. Hong, A. Pranatharthi, H. Krishnamurthy, V.G. Giri, S. Krishna

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Pattabiraman, S. Srivastava, H. Krishnamurthy, L.A. Laimins, S. Krishna

Writing, review, and/or revision of the manuscript: C. Pattabiraman, S. Hong, L.A. Laimins, S. Krishna

Other (performed experiments): A. Ammothumkandy

The authors thank Dr. Sally Roberts for the E1⁁E4 antibody, Dr. Geeta Mukherjee for histology, and C. Jamora, M. Rosbash, J. Dhawan, S. Dalal, A. Dutta, and T. Rajkumar for critical comments and discussions.

This work was supported by grants from DST, DBT, NCBS-TIFR (S. Krishna), an IUBMB Travel Award (C. Pattabiraman), and grants from the NCI (CA142861 and CA59655; L.A. Laimins, S. Hong, and V.K. Gunasekharan).

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.
Al-Hajj
M
,
Wicha
MS
,
Benito-Hernandez
A
,
Morrison
SJ
,
Clarke
MF
. 
Prospective identification of tumorigenic breast cancer cells
.
Proc Natl Acad Sci U S A
2003
;
100
:
3983
8
.
2.
Singh
SK
,
Hawkins
C
,
Clarke
ID
,
Squire
JA
,
Bayani
J
,
Hide
T
, et al
Identification of human brain tumour initiating cells
.
Nature
2004
;
432
:
396
401
.
3.
Dalerba
P
,
Dylla
SJ
,
Park
I-K
,
Liu
R
,
Wang
X
,
Cho
RW
, et al
Phenotypic characterization of human colorectal cancer stem cells
.
Proc Natl Acad Sci U S A
2007
;
104
:
10158
63
.
4.
Visvader
JE
,
Lindeman
GJ
. 
Cancer stem cells: current status and evolving complexities
.
Cell Stem Cell
2012
;
10
:
717
28
.
5.
Ricci-Vitiani
L
,
Lombardi
DG
,
Pilozzi
E
,
Biffoni
M
,
Todaro
M
,
Peschle
C
, et al
Identification and expansion of human colon-cancer-initiating cells
.
Nature
2007
;
445
:
111
5
.
6.
Sharma
S V
,
Lee
DY
,
Li
B
,
Quinlan
MP
,
Takahashi
F
,
Maheswaran
S
, et al
A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations
.
Cell
2010
;
141
:
69
80
.
7.
Sikandar
SS
,
Pate
KT
,
Anderson
S
,
Dizon
D
,
Edwards
RA
,
Waterman
ML
, et al
NOTCH signaling is required for formation and self-renewal of tumor-initiating cells and for repression of secretory cell differentiation in colon cancer
.
Cancer Res
2010
;
70
:
1469
78
.
8.
Roesch
A
,
Fukunaga-Kalabis
M
,
Schmidt
EC
,
Zabierowski
SE
,
Brafford
PA
,
Vultur
A
, et al
A temporarily distinct subpopulation of slow-cycling melanoma cells is required for continuous tumor growth
.
Cell
2010
;
141
:
583
94
.
9.
Perumalsamy
LR
,
Nagala
M
,
Sarin
A
. 
Notch-activated signaling cascade interacts with mitochondrial remodeling proteins to regulate cell survival
.
Proc Natl Acad Sci U S A
2010
;
107
:
6882
7
.
10.
Bajaj
J
,
Maliekal
TT
,
Vivien
E
,
Pattabiraman
C
,
Srivastava
S
,
Krishnamurthy
H
, et al
Notch signaling in CD66+ cells drives the progression of human cervical cancers
.
Cancer Res
2011
;
71
:
4888
97
.
11.
Blumenthal
RD
,
Leon
E
,
Hansen
HJ
,
Goldenberg
DM
. 
Expression patterns of CEACAM5 and CEACAM6 in primary and metastatic cancers
.
BMC Cancer
2007
;
7
:
2
.
12.
Blumenthal
RD
,
Hansen
HJ
,
Goldenberg
DM
. 
Inhibition of adhesion, invasion, and metastasis by antibodies targeting CEACAM6 (NCA-90) and CEACAM5 (Carcinoembryonic Antigen)
.
Cancer Res
2005
;
65
:
8809
17
.
13.
Lewis-Wambi
JS
,
Cunliffe
HE
,
Kim
HR
,
Willis
AL
,
Jordan
VC
. 
Overexpression of CEACAM6 promotes migration and invasion of oestrogen-deprived breast cancer cells
.
Eur J Cancer
2008
;
44
:
1770
9
.
14.
Duxbury
MS
,
Ito
H
,
Benoit
E
,
Zinner
MJ
,
Ashley
SW
,
Whang
EE
. 
Overexpression of CEACAM6 promotes insulin-like growth factor I-induced pancreatic adenocarcinoma cellular invasiveness
.
Oncogene
2004
;
23
:
5834
42
.
15.
López
J
,
Poitevin
A
,
Mendoza-Martínez
V
,
Pérez-Plasencia
C
,
García-Carrancá
A
. 
Cancer-initiating cells derived from established cervical cell lines exhibit stem-cell markers and increased radioresistance
.
BMC Cancer
2012
;
12
:
48
.
16.
Feng
D
,
Peng
C
,
Li
C
,
Zhou
Y
,
Li
M
,
Ling
B
, et al
Identification and characterization of cancer stem-like cells from primary carcinoma of the cervix uteri
.
Oncol Rep
2009
;
22
:
1129
34
.
17.
Auffinger
B
,
Tobias
AL
,
Han
Y
,
Lee
G
,
Guo
D
,
Dey
M
, et al
Conversion of differentiated cancer cells into cancer stem-like cells in a glioblastoma model after primary chemotherapy
.
Cell Death Differ
2014
;
21
:
1119
31
.
18.
Kim
J
,
Villadsen
R
,
Sørlie
T
,
Fogh
L
,
Grønlund
SZ
,
Fridriksdottir
AJ
, et al
Tumor initiating but differentiated luminal-like breast cancer cells are highly invasive in the absence of basal-like activity
.
Proc Natl Acad Sci U S A
2012
;
109
:
6124
9
.
19.
Sahlgren
C
,
Gustafsson
M V
,
Jin
S
,
Poellinger
L
,
Lendahl
U
. 
Notch signaling mediates hypoxia-induced tumor cell migration and invasion
.
Proc Natl Acad Sci U S A
2008
;
105
:
6392
7
.
20.
Mani
SA
,
Guo
W
,
Liao
M-J
,
Eaton
EN
,
Ayyanan
A
,
Zhou
AY
, et al
The epithelial-mesenchymal transition generates cells with properties of stem cells
.
Cell
2008
;
133
:
704
15
.
21.
Driessens
G
,
Beck
B
,
Caauwe
A
,
Simons
BD
,
Blanpain
C
. 
Defining the mode of tumour growth by clonal analysis
.
Nature
2012
;
488
:
527
30
.
22.
Zur Hausen
H
,
de Villiers
EM
. 
Human papillomaviruses
.
Annu Rev Microbiol
1994
;
48
:
427
47
.
23.
Doorbar
J
,
Quint
W
,
Banks
L
,
Bravo
IG
,
Stoler
M
,
Broker
TR
, et al
The biology and life-cycle of human papillomaviruses
.
Vaccine
2012
;
30
:
F55
70
.
24.
Moody
CA
,
Laimins
LA
. 
Human papillomavirus oncoproteins: pathways to transformation
.
Nat Rev Cancer
2010
;
10
:
550
60
.
25.
Bedell
MA
,
Hudson
JB
,
Golub
TR
,
Turyk
ME
,
Hosken
M
,
Wilbanks
GD
, et al
Amplification of human papillomavirus genomes in vitro is dependent on epithelial differentiation
.
J Virol
1991
;
65
:
2254
60
.
26.
Meyers
C
,
Frattini
MG
,
Hudson
JB
,
Laimins
LA
. 
Biosynthesis of human papillomavirus from a continuous cell line upon epithelial differentiation
.
Science
1992
;
257
:
971
3
.
27.
Mighty
KK
,
Laimins
LA
. 
p63 is necessary for the activation of human papillomavirus late viral functions upon epithelial differentiation
.
J Virol
2011
;
85
:
8863
9
.
28.
Moody
CA
,
Laimins
LA
. 
Human papillomaviruses activate the ATM DNA damage pathway for viral genome amplification upon differentiation
.
PLoS Pathog
2009
;
5
:
e1000605
.
29.
Hebner
C
,
Beglin
M
,
Laimins
LA
. 
Human papillomavirus E6 proteins mediate resistance to interferon-induced growth arrest through inhibition of p53 acetylation
.
J Virol
2007
;
81
:
12740
7
.
30.
Sen
GL
,
Reuter
JA
,
Webster
DE
,
Zhu
L
,
Khavari
PA
. 
DNMT1 maintains progenitor function in self-renewing somatic tissue
.
Nature
2010
;
463
:
563
7
.
31.
Rangarajan
A
,
Syal
R
,
Selvarajah
S
,
Chakrabarti
O
,
Sarin
A
,
Krishna
S
. 
Activated Notch1 signaling cooperates with papillomavirus oncogenes in transformation and generates resistance to apoptosis on matrix withdrawal through PKB/Akt
.
Virology
2001
;
286
:
23
30
.
32.
Huang
DW
,
Sherman
BT
,
Lempicki
RA
. 
Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources
.
Nat Protoc
2009
;
4
:
44
57
.
33.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
34.
Truong
AB
,
Kretz
M
,
Ridky
TW
,
Kimmel
R
,
Khavari
PA
. 
p63 regulates proliferation and differentiation of developmentally mature keratinocytes
.
Genes Dev
2006
;
20
:
3185
97
.
35.
Rangarajan
A
,
Talora
C
,
Okuyama
R
,
Nicolas
M
,
Mammucari
C
,
Oh
H
, et al
Notch signaling is a direct determinant of keratinocyte growth arrest and entry into differentiation
.
EMBO J
2001
;
20
:
3427
36
.
36.
Blanpain
C
,
Lowry
WE
,
Pasolli
HA
,
Fuchs
E
. 
Canonical notch signaling functions as a commitment switch in the epidermal lineage
.
Genes Dev
2006
;
20
:
3022
35
.
37.
Fuchs
E
. 
Finding one's niche in the skin
.
Cell Stem Cell
2009
;
4
:
499
502
.
38.
Takahashi
K
,
Tanabe
K
,
Ohnuki
M
,
Narita
M
,
Ichisaka
T
,
Tomoda
K
, et al
Induction of pluripotent stem cells from adult human fibroblasts by defined factors
.
Cell
2007
;
131
:
861
72
.
39.
Lathion
S
,
Schaper
J
,
Beard
P
,
Raj
K
. 
Notch1 can contribute to viral-induced transformation of primary human keratinocytes
.
Cancer Res
2003
;
63
:
8687
94
.
40.
Leonard
SM
,
Wei
W
,
Collins
SI
,
Pereira
M
,
Diyaf
A
,
Constandinou-Williams
C
, et al
Oncogenic human papillomavirus imposes an instructive pattern of DNA methylation changes which parallel the natural history of cervical HPV infection in young women
.
Carcinogenesis
2012
;
33
:
1286
93
.
41.
McCabe
MT
,
Low
JA
,
Daignault
S
,
Imperiale
MJ
,
Wojno
KJ
,
Day
ML
. 
Inhibition of DNA methyltransferase activity prevents tumorigenesis in a mouse model of prostate cancer
.
Cancer Res
2006
;
66
:
385
92
.
42.
Herfs
M
,
Yamamoto
Y
,
Laury
A
,
Wang
X
,
Nucci
MR
,
McLaughlin-Drubin
ME
, et al
A discrete population of squamocolumnar junction cells implicated in the pathogenesis of cervical cancer
.
Proc Natl Acad Sci U S A
2012
;
109
:
10516
21
.
43.
Doorbar
J
. 
The E4 protein; structure, function and patterns of expression
.
Virology
2013
;
445
:
80
98
.
44.
Nakahara
T
,
Lambert
PF
. 
Induction of promyelocytic leukemia (PML) oncogenic domains (PODs) by papillomavirus
.
Virology
2007
;
366
:
316
29
.
45.
Wang
H-K
,
Duffy
AA
,
Broker
TR
,
Chow
LT
. 
Robust production and passaging of infectious HPV in squamous epithelium of primary human keratinocytes
.
Genes Dev
2009
;
23
:
181
94
.
46.
Gillespie
KA
,
Mehta
KP
,
Laimins
LA
,
Moody
CA
. 
Human papillomaviruses recruit cellular DNA repair and homologous recombination factors to viral replication centers
.
J Virol
2012
;
86
:
9520
6
.
47.
Moody
CA
,
Laimins
LA
. 
Human papillomaviruses activate the ATM DNA damage pathway for viral genome amplification upon differentiation
.
PLoS Pathog
2009
;
5
:
e1000605
.
48.
Swindle
CS
,
Zou
N
,
Van Tine
BA
,
Shaw
GM
,
Engler
JA
,
Chow
LT
. 
Human papillomavirus DNA replication compartments in a transient DNA replication system
.
J Virol
1999
;
73
:
1001
9
.
49.
Anastassiou
D
,
Rumjantseva
V
,
Cheng
W
,
Huang
J
,
Canoll
PD
,
Yamashiro
DJ
, et al
Human cancer cells express Slug-based epithelial-mesenchymal transition gene expression signature obtained in vivo
.
BMC Cancer
2011
;
11
:
529
.
50.
Davy
CE
,
Jackson
DJ
,
Raj
K
,
Peh
WL
,
Southern
SA
,
Das
P
, et al
Human papillomavirus type 16 E1 E4-induced G2 arrest is associated with cytoplasmic retention of active Cdk1/cyclin B1 complexes
.
J Virol
2005
;
79
:
3998
4011
.