Current status in cancer cell reprogramming and its clinical implications

Abstract

Purpose

The technology of reprogramming a terminally differentiated cell to an embryonic-like state uncovered the possibility of reprogramming a malignant cell back to a more manageable stem cell-like state. Since the current cancer models suffer from reflecting heterogeneous tumour structure and limited to express the late-stage markers, the induced pluripotent stem cell (iPSC) technology could provide an alternative model to recapitulate the early stages of cancer. Generation of iPSCs from cancer cells could offer a tool for understanding the mechanisms of tumour initiation–progression in vitro, a platform for studying tumour heterogeneity and origin of cancer stem cells and a source for cancer type-specific drug discovery studies.

Methods

In this review, we discussed the recent findings in reprogramming cancer cells with a special emphasis on similarities between cancer cells and pluripotent cells. We presented the basis of challenges in cancer cell reprogramming including the current problems in reprogramming, cancer-specific epigenetic state and chromosomal aberrations.

Results

Cancer epigenetics represent the major hurdle before the prospective use of cancer iPSCs as a model system and for biomarker research. When the reprogramming process is optimised for cancer cell types, it might serve for two purposes: identification of the specific epigenetic state of cancer as well as reversion of the malignant phenotype to a potentially malignant but manageable state.

Conclusions

Reprogramming cancer cells would serve for our understanding of cancer-specific epigenome and elucidation of overlapping mechanisms shared by cancer-initiating cells and pluripotent cells.

Introduction

Current status in cancer cell reprogramming: Is it possible to run away from the cancer state or are we running around in circles?

The discovery of reprogramming methods enables the reversion of terminally differentiated somatic cells into cells exhibiting embryonic stem cell-like traits by ectopic expression of core transcription factors regulating the pluripotency (Takahashi and Yamanaka 2006; Yu et al. 2007). The induced pluripotent stem (iPS) cell technology holds great promise for generating patient-specific source for regenerative medicine as well as providing an invaluable tool for disease modelling, drug discovery and toxicity testing (Stadtfeld and Hochedlinger 2010). Although the iPS cell technology endures the low efficiency due to unresolved molecular mechanisms behind the reprogramming, various somatic cell types could be successfully reprogrammed into pluripotent state (Tobin and Kim 2012). Still, there is a pressing need for improving the quality of reprogramming since the cells from different origins necessitates distinct reprogramming factor combinations and the presence of reprogramming enhancers (small molecules, miRNAs, enzymes, etc.) due to the lack of universal reprogramming strategy.

The success in reprogramming a differentiated cell into an undifferentiated state has led the scientists to the idea of reprogramming malignant cells back to their very first state, well before the oncogenic transformation occurs. Generation of iPSCs from cancer cells could offer a tool for understanding the mechanisms of tumour initiation–progression in vitro, a platform for studying tumour heterogeneity and origin of cancer stem cells and a source for cancer type-specific drug discovery studies. However, the “normalisation” process of a cancerous cell remains challenging due to cancer-specific epigenetic state and chromosomal aberrations. The epigenetic memory of the starting cell type still matters in reprogramming and is closely related to the inefficient reprogramming due to a failure of resetting the DNA methylation and post-translational histone modifications to an embryonic stem cell-like state (Papp and Plath 2011). Although in practice iPSCs generated from different tissues may retain significant functional and molecular differences, continuous passaging may abrogate transcriptional, epigenetic and functional differences for mouse somatic cells (Polo et al. 2010). The heterogeneity of human iPS cells arise mostly from distinct genetic background of the donor, and cell type of origin affect the functionality of iPS cells (Phetfong et al. 2016; Rouhani et al. 2014). The fundamental reason behind the inefficient or impossible reprogramming of cancer cells is highly relevant to the epigenetic status of the starting cell population (Ivanov et al. 2016).

Epigenetics is referred as the “inherited additional changes” which regulates gene activity without accompanying DNA sequence changes (Weinhold 2006). Epigenetic mechanisms are essential for development, and dysregulation of the epigenetic process could lead to disruption of gene function and oncogenic transformation (Dawson and Kouzarides 2012; Egger et al. 2004). Since reprogramming involves resetting of the transcriptional network and chromatin state of the differentiated cell to that of pluripotent cell, the dysregulated epigenetic state of a malignant cell could be reversed by reprogramming (Hawkins et al. 2010; Kelly et al. 2010; Lister et al. 2011). Epigenetic state is of reversible nature, but attempts to reprogram cancer cells suffer from incomplete resetting of the cancer-associated epigenome due to the tumour heterogeneity and accumulated oncogenic mutations (Heng et al. 2009; Meacham and Morrison 2013; Stricker et al. 2013; Timp and Feinberg 2013). Thus, cancer cell reprogramming is currently limited to certain cancer cell lines, and cancer-specific marks in the epigenome, which impede the successful reprogramming, are not fully elucidated. Dysregulated epigenetic enzymes in cancer were determined to be functional in reprogramming events. For instance, EZH2, a histone-modifying enzyme whose transcriptional repression involves in tumour aggressiveness, also has a role in iPSC generation (Lee et al. 2015b; Tonini et al. 2008). Forced EZH2 expression enhances, while knockdown of EZH2 impairs iPSC generation (Ding et al. 2014; Villasante et al. 2011; Xie et al. 2016). BMI-1, an oncogene which is overexpressed in some cancer types, was demonstrated to increase the reprogramming efficiency by replacing the functions of KLF4 and c-MYC (Cao et al. 2011; Moon et al. 2011, 2013). The majority of cancer types exhibits aberrant epigenetic regulation of p53 tumour suppressor network (Mishra et al. 2015; Zilfou and Lowe 2009). The absence or reduced activities of p53 and p21 favour the iPSC generation (Ichida et al. 2014; Rasmussen et al. 2014; Yi et al. 2012). Studies suggest that common epigenetic regulations are involved in reprogramming and cancer. Moreover, epigenetic silencing of tumour suppressor genes like aberrant methylation of the p16 promotor was shown to be reversed by reprogramming (Ron-Bigger et al. 2010). Therefore, explanation of similar/distinct epigenetic states observed in reprogramming and cancer may facilitate the development of effective reprogramming strategies for normal cells as well as reveal the cues for cancer cell reprogramming.

When the variabilities in the reprogramming process could be surpassed, it might serve for two purposes: identification of the authentic epigenetic state of cancer as well as reversion of the malignant phenotype to a potentially malignant but manageable pluripotent state. Therefore, reprogramming of cancer cells and overcoming the barriers to pluripotency will remain as an area of active research.

Barriers to cancer cell reprogramming: being resistant to manipulation

Before the introduction of the reprogramming factors, malignant melanoma cells have been demonstrated to be reversed to a pluripotent state by nuclear transfer, and despite the presence of genetic alterations, transplanted melanoma nuclei developed into a normal-appearing blastocysts which gave rise to embryonic stem cells (Hochedlinger et al. 2004). However, other malignant cancer types including leukaemia, lymphoma and breast cancer cells could not be reprogrammed by the oocyte environment, suggesting the presence of unknown barriers to reprogramming. Similarly, embryonal carcinoma cell lines could be reprogrammed by nuclear transfer and represented normal preimplantation development, but the abnormal phenotype of the clones occurred in the developmental and tumourigenic assays, proposing that the EC lines possess unique set of modifications in the epigenetic state that could not be reprogrammed limiting the developmental potential (Blelloch et al. 2004). The presence of only a limited number of studies which reported successful derivation of iPSCs from cancer cells supports the notion that harbouring cancer-specific genetic/epigenetic state hinders successful reprogramming independent from the complexity of the technique used for reprogramming. Still, reprogramming studies either with patient-derived cancer cells or with banked cancer cells represent solid evidence for the reversion of the malignant phenotype and hold a great promise for breaking the irreversibility of the cancer state (Fig. 1).

Fig. 1.

Some cancer types could be reprogrammed into induced pluripotent stem cells (iPSCs) by a combination of four factors (4Fs): Oct4, Sox2, cMyc and Klf4 in most cases. Thus, immortal cancer cells were reprogrammed back to another potentially malignant but manageable state, namely pluripotent state. Once cancer cells reverted to an embryonic-like pluripotent state, they acquire the potential to redifferentiate into the cell type where cancer is originated in the first case. iPSC technology would enable modelling the initial stages of cancer in vitro which would serve for biomarker research, toxicity testing and drug screening. We reprogrammed bladder cancer cell line T24 into cancer iPSC-like cells where the cells lost their mesenchymal phenotype, gained epithelial morphology and formed colonies representing positive expression of pluripotency-associated markers like Oct4 (red) and Ssea-4 (green) (Iskender et al. 2016)

Patient-derived primary tumours could shed a light upon late-stage markers and differentiated cell stage, but they fail to offer clues about early stages of the disease. The current animal models also fail to serve as an ideal model system for human cancers due to the genetic differences. Although being useful models for cancer research, the established cancer cell lines also suffer from being prone to the additional genetic/epigenetic changes in the prolonged cultures in vitro which drive the cells to an unpredictable state, rendering contradicting outcomes for preclinical cancer studies. Therefore, the discovery of iPSC technology offers an adequate system for recapitulating diseases including cancer in a culture plate which is expected to augment significant clinical outcomes. The reports of fully characterised iPSCs or iPSC-like cells generated from human malignant cells have been limited to certain cancer types (Table 1). The initial studies utilised the combination of defined factors to induce pluripotency in cancer cells. However, the limited number of successful derivation of cancer cell-derived iPSCs since 2008 demonstrates a strong evidence for the difficulty of this process. Before the use of defined factors for reprogramming, embryonic stem cell-associated miRNAs or hESC-conditioned matrices are used to alter malignant phenotype of cancer cells (Lin et al. 2008; Postovit et al. 2008). Lin et al. utilised an intronic miRNA expression system to transfect melanoma (Colo) and prostate (PC-3) cell lines and demonstrated that Colo cell line highly resembled embryonic stem cell line H1 and H9 upon transfection in terms of pluripotency gene expression as well as increased demethylated genome (Lin et al. 2008). Human embryonic stem cell microenvironment was also shown to suppress aggressive melanoma and breast cancer behaviour through production of Lefty, Nodal inhibitor, which reduced the clonogenicity and tumourigenicity of cancer cells (Postovit et al. 2008). Although yielding encouraging results that rely on solely microenvironmental clues or upregulation of miRNA expression, thereby eliminating the use of oncogenes, like Klf4 and c-Myc, the subsequent cancer cell reprogramming studies heavily depended on the conventional reprogramming factors. Using a retroviral transduction, Miyoshi et al. 2010 attempted to induce pluripotency in 20 gastrointestinal cancer cell lines and demonstrated successful reprogramming of only eight cancer lines. Eight cell lines which generated iPSCs were reported to be derived from colorectal (DLD1, HT29), oesophageal (TE10), gastric (MKN45), pancreatic cancer (MIAPaCa-2, PAN-1), hepatocellular (PLC) and cholangiocellular carcinoma (HuCCT-1). Similarly, Noguchi et al. (2015) observed that PANC1 cells were susceptible to reprogramming, while three other lines (MIAPaCa-2, PSN-1 and AsPC-1) remained resistant. We have recently generated iPSC-like cells from bladder cancer cell line T24; however, another bladder cancer cell line HTB-9 did not respond to pluripotency induction and failed to generate cancer cells bearing pluripotent features (Iskender et al. 2016). Low efficiency in generating iPSC from cancer cells despite the uniformity in the induction method suggests that multiple mechanisms might be involved in the regulation of reprogramming. Indeed, Mathieu et al. (2011) observed that reprogramming factors and stable expression of HIFα accelerated iPSC induction time from lung carcinoma (A549) cell line, suggesting that reprogramming could be enhanced by cumulative effect of multiple environmental cues. Low oxygen levels have been shown to promote self-renewing capacity of stem cells (Mohyeldin et al. 2010). Exposure to hypoxic conditions activated the expression of multiple genes including c-MYC, OCT4, KLF4, NANOG and stem cell-associated miRNAs including miR-302a-b, miR-106a-b, miR-92, miR-372, miR-19b, miR-130a, miR-30e-5p and miR-195 in prostate, brain, kidney, cervix, lung, colon, liver and breast cancer cell lines which shared overlapping gene expression signature with nine established hESC lines (Mathieu et al. 2011).

Table 1.

Human cancer cell-derived iPSC lines

References	Cancer type	Cell line	Reprogramming method	Main findings
Lin et al. (2008)	Melanoma Prostate	Colo PC-3	Retroviral mir-302 s	Transfected cells resembled ESC lines H1 and H9 in terms of pluripotency marker expression and increased demethylated genome
Miyoshi et al. (2010)	Colorectal Oesophageal Gastric Pancreatic Hepatocellular Cholangiocellular	DLD1, HT29 TE10 MKN45 MIAPaCa-2, PAN-1 PLC HuCCT-1	Retroviral or lentiviral OSKM, NANOG, LIN28, BCL2, KRAS and tumour suppressor shRNAs	Cells showed slow proliferation and acquired sensitivity to differentiation inducing treatment and chemotherapeutic agents; in vivo tumourigenesis was reduced
Noguchi et al. (2015)	Pancreatic	PANC1	Sendai virus OSKM	c-Methigh cells were more susceptible to reprogramming than c-Metlow cells
Iskender et al. (2016)	Bladder cancer	T24 HTB9	Sendai virus OSKM	T24 cells were susceptible to reprogramming, while HTB-9 cells failed to generate iPSC-like cells
Mathieu et al. (2011)	Lung carcinoma	A549	Retroviral nondegradeable HIF1α and HIF2α and then lentiviral OSNL	Partially reprogrammed iPSC-like colonies
Carette et al. (2010)	Chronic myeloid leukaemia	KBM7	Retroviral OSKM	Retained the expression of BCR-ABL oncogene but lost oncogene addiction
Hu et al. (2011)	Chronic myeloid leukaemia	Bone marrow mononuclear cells of a patient	Episomal vectors expressing four factors	Cells retained unique chromosomal translocation but displayed ESC phenotype and pluripotent differentiation potential
Kumano et al. (2012)	Chronic myeloid leukaemia	Bone marrow cells of a CML patient	Retroviral OSKM	Cells acquired imatinib resistance, differentiated into haematopoetic lineage and recovered sensitivity to imatinib
Gandre-Babbe et al. (2013)	Juvenile Myelomonocytic Leukaemia	Cells of two JMML patients	Lentiviral OSKM	Cell recapitulate key pathological features of JMML, proliferate and differentiate into haematopoetic lineage
Kotini et al. (2015)	Myelodysplastic syndrome	Haematopoetic cells of two MDS patients	Lentiviral OSKM	iPSCs recapitulate disease-associated phenotypes like impaired haematopoetic differentiation
Utikal et al. (2009)	Melanoma	R545	Lentiviral OKM	iPSCs could generate chimeric mice with germline transmission
Oshima et al. (2014)	Colorectal	SW480 DLD-1	Retroviral OSK	Cells acquired cancer stem cell phenotype with enhanced tumourigenicity but failed to give rise to teratomas
Moore et al. Moore et al. (2015)	Ewing’s sarcoma	CHLA-10	Episomal OSNK	EWS-iPS cells formed tumours exhibiting disease phenotype, and cells acquired sensitivity to chemotherapeutic agents
Islam et al. (2015)	Neuroblastoma	SH-IN	Sendai virus OSKM	Partially reprogrammed cells exhibit stem cell-like colonies, differentiate into three germ layers in vitro but formed neuroblastomas in vivo, shared genomic aberrations as parental cells and exhibit resistance to cisplatin
Kim et al. (2013)	Pancreatic ductal adenocarcinoma	Patient-derived adenocarcinoma cells	Lentiviral OSKM	Cells generated tumours which recapitulated early and advanced stages of PDAC when injected into immunodeficient mice
Corominas-Faja et al. (2013)	Breast	MCF-7	Retroviral OSKM	Partially reprogrammed cells exhibited enhanced SOX2 expression and cancer stem cell characteristics
Zhang et al. (2013)	Osteosarcoma Ewing’s sarcoma Liposarcoma	SAOS2, HOS, MG63 SHNEP SW872	Lentiviral OSKM, N, L	Reprogrammed cells exhibited pluripotency features, could terminally differentiate into mature connective tissue and red blood cells and acquired reduced tumourigenicity compared to the parental cells
Mahalingam et al. (2012)	Non-small cell lung cancer	H358, H460, IMR90	Lentiviral OSKM	Reprogrammed cells reversed methylation pattern and transcription of dysregulated genes and exhibited reduced tumourigenicity

Expanding the primary samples of haematologic malignancies in culture is usually challenging; therefore, disease-specific iPSC possessing the genetic abnormalities of haematologic malignancies would provide an efficient platform for studying pathogenesis (Carette et al. 2010). Carette et al. (2010) succeeded the derivation of iPS cells from chronic myeloid leukaemia cell line, KBM7, by retroviral induction of defined factors (OSKM) which retained the expression of the BCR-ABL oncogene but lost oncogene addiction. Another study demonstrated efficient generation of iPSCs from bone marrow mononuclear cells of a patient with chronic myeloid leukaemia (Hu et al. 2011). CML-iPSCs exhibited four-way translocation between chromosomes 1, 9, 11 and 22 similar to the parental cells but displayed pluripotent potential like ESCs. Kumano et al. (2012) generated iPSCs from imatinib-sensitive CML patient samples which became resistant to imatinib despite the expression of BCR-ABL oncoprotein. When the CML-iPSCs were induced to differentiate into haematopoetic cells, they recapitulated the initial disease and recovered sensitivity to imatinib. Accumulating evidence from haematologic malignancies suggested that reprogramming of cancer cells was feasible despite the presence of genomic alterations in the parental cells, and patient-derived iPSCs from juvenile myelomonocytic leukaemia patients and patients with myelodysplastic syndrome recapitulated the disease-associated phonotypes (Gandre-Babbe et al. 2013; Kotini et al. 2015). Similar concept is widely observed in reprogramming normal somatic cells where reprogramming-induced multiple genetic/epigenetic aberrancies would not interfere with the differentiation potential of the resulting iPSCs (Hussein et al. 2011; Lister et al. 2011; Mayshar et al. 2010). Although expression of reprogramming factors was found to be successful in various primary patient samples of haematologic malignancies, Liu et al. (2014b) reported that Notch1-initiated T-acute lymphoblastic leukaemia state could not be reprogrammed into a pluripotent state. Therefore, the combination of reprogramming factors needs to be optimised for each cancer type. For instance, Utikal et al. (2009) demonstrated that R545 melanoma cell line could be reprogrammed into an iPSCs using three factors (KLF4, OCT4 and c-MYC) without ectopic SOX2 requirement. The resultant iPSCs could generate high-degree chimeric mice competent of germline transmission. In contrast, Oshima et al. (2014) reported induction of cancer stem cell properties in colon cancer cells upon transfection with three defined factors: OCT4, SOX2 and KLF4. The authors showed that a subset of colon cancer cells gained colon cancer cell properties and expressed defined colon CSC markers, carried a pathogenic genome and gave rise to tumours but not to teratomas in vivo. Failure to yield a well-characterised iPSC line in the latter study might be due to failure to condition the transfected cancer cells with the right medium as the authors indicated usage of conventional serum-containing medium. This specifically indicates that fully reprogramming could only be achieved under the circumstances where external signalling supports the intrinsic initiation of early reprogramming events.

Epigenetic remodelling of cancer cells was also demonstrated to give rise to partially reprogrammed iPSC or iPSC-like cells with reduced or increased aggressive phenotype (Corominas-Faja et al. 2013; Islam et al. 2015; Kim et al. 2013; Miyoshi et al. 2010; Moore et al. 2015; Noguchi et al. 2015; Utikal et al. 2009; Vencio et al. 2012). Despite sharing DNA methylation pattern comparable to established ESC and iPSC lines, iPSCs derived from cancer cells might exhibit distinct hypomethylation of stem cell-specific densely methylated regions (Moore et al. 2015). This result is not unexpected when the heterogenecity and differentiation hierarchy of tumour cell population is taken into account where some cells being permissive to reprogramming and the others not (Zhang et al. 2013). In another study, Stricker et al. demonstrated that highly aneuploid patient-derived glioblastoma cell lines could be reprogrammed and exhibited removal of cancer-specific DNA methylation (Stricker et al. 2013). Moreover, reprogramming reversed the DNA methylation markers that are significant for non-small cell lung carcinoma cell lines, and in vitro differentiation of NSCLC-derived iPSCs could not restore the tumour-specific epigenetic state (Mahalingam et al. 2012). iPSCs derived from glioblastoma-derived neural stem cells showed reduced capacity to infiltrate surrounding tissues, which suggests suppression of the aggressive phenotype upon reprogramming (Stricker et al. 2013). Zhang et al. (2013) reprogrammed three osteosarcoma (SAOS2, HOS and MG63), two liposarcoma (SW872) and a sarcoma of unknown origin (Ewing’s sarcoma, SKNEP) which altered epigenetic state of oncogene/tumour suppressor genes which render the cells a less aggressive tumour phenotype. On the other hand, some studies suggested acquisition of sensitivity against anti-cancer reagents in the reprogrammed iPSCs, which is not necessarily an indicator of reduced malignancy but shows increased drug response compared to the parental cells (Miyoshi et al. 2010). Reactivation of some tumour suppressor genes like p16 in iPSCs might play a role in increased chemosensitivity as well as regression of proliferation and invasiveness in reprogrammed cancer cells (Miyoshi et al. 2010). Not all cancer reprogramming studies analysed the tumourigenic potential or drug response of the resulting iPSCs, but the results from few studies suggest contradicting outcomes to those of Miyoshi et al. as discussed earlier in this review (Carette et al. 2010; Islam et al. 2015; Kumano et al. 2012).

Undoubtedly, one of the most promising outcomes of the application of iPSC technology to cancer cells is the recapitulation of the disease phenotype. Utilising patient-derived iPSCs to restage the disease would enable the identification of early markers and the underlying molecular mechanisms. Only a few studies managed to demonstrate the disease initiation and progression steps using the iPSCs derived from human cancers. In a pioneering study, Kim et al. (2013) generated iPSCs from pancreatic ductal adenocarcinoma cells (PDAC) which were then injected into immunodeficient mice and generated pancreatic intraepithelial neoplasia (PanIN) precursors within the teratomas. The PanIN-like cells were shown to secrete putative biomarkers of early-stage pancreatic cancer. In another study, iPSCs from haematopoetic cells of two patients with myelodysplastic syndrome (MDS) with loss of chromosome 7q (del (7q)) were generated which recapitulated the disease-associated phenotype like impaired haematopoietic differentiation (Kotini et al. 2015). Moreover, Gandre-Babbe et al. (2013) generated iPSCs from two juvenile myelomonocytic leukaemia (JMML) patients which recapitulated the pathological features of JMML including increased myelopoiesis and constitutive GM-CSF activation; upon induction of haematopoietic differentiation, iPSCs produced human myeloid cells with GM-CSF independence. A study from Ohnishi et al. (2014) demonstrated the close interaction between reprogramming and cancer development, suggesting that these two processes heavily depend on epigenetic changes. They created a doxycycline-inducible mouse model expressing defined factors. Incomplete expression of reprogramming factors in this model resulted in tumour development in various tissues containing undifferentiated dysplastic cells which exhibit global changes in DNA methylation patterns. The cells from tumours arose in kidneys, resembling Wilms tumour, could be reprogrammed and gave rise to normal kidney cells in chimeric mice (Ohnishi et al. 2014). The authors suggested that common epigenetic processes might be involved in reprogramming as well as development of certain cancer types. Indeed, global changes in epigenetic modifications occur in normal development and cancer which were demonstrated to be bidirectional rather than unidirectional with the nuclear transfer and reprogramming experiments (Gurdon 1962; Koche et al. 2011; Takahashi et al. 2007; Wilmut et al. 1997). Therefore, application of reprogramming technologies to cancer cells might serve for our understanding of the cancer-specific epigenome and elucidation of overlapping mechanisms shared by cancer-initiating cells and pluripotent cells.

Similarities in reprogramming and carcinogenesis: Could they both be bidirectional?

Expression of pluripotency-associated markers

Pluripotent stem cells express a unique set of transcription factors which define their pluripotent feature, reflective of their differentiation potential into every cell of an organism. OCT4, SOX2 and NANOG form a core regulatory network which sustains self-renewal while suppressing differentiation (Boyer et al. 2005; Loh et al. 2006). Acquisition of pluripotency in somatic cells requires ectopic expression of a set of transcription factors associated with pluripotency as well as maintaining embryonic stem cell growth conditions (Takahashi and Yamanaka 2006). The morphological changes occur at the initial phases of reprogramming, but the sustained pluripotency gene expression signs the fulfilled reprogramming. Although c-Myc was originally suggested among the reprogramming protocols, its role in reprogramming was defined as an enhancer rather than an indispensible transcription factor (Nakagawa et al. 2008; Wernig et al. 2008). c-Myc functions as an amplifier of gene expression during reprogramming rather than acting through proliferation-associated transcription factors (Buganim et al. 2013; Nie et al. 2012). Alternative methods or reprogramming factor combinations have been utilised in independent studies so far, but major regulatory hub for pluripotency remains unchanged for most cell types (Singh et al. 2015). The key players in reprogramming were known to be overexpressed in cancer types. For instance, c-Myc is a protooncogene which has a pivotal role in differentiation, apoptosis and cell growth and is detected to be overexpressed in most cancer types (Miller et al. 2012).

KLF4 contains various domains involved in transcriptional activation/repression and protein–protein interaction and is known to play roles in cell cycle regulation, DNA repair, apoptosis, differentiation and cell fate decision (Nandan and Yang 2009). KLF4 is known to regulate self-renewal by binding to sites throughout the genome and its role could be replaced by other family members like KLF2 or KLF5 (Chen et al. 2008). Krüppel-like factors could either act as oncogenes or tumour suppressors in certain cancer types, and they exhibited to confer therapy resistance to cancer cells. KLFs not only maintain the sustained activity of telomerase to support long-term proliferation but also enhance epithelial-mesenchymal transition of cancer cells (Wong et al. 2010; Yu et al. 2011). Overexpression of KLF4 increases cell migration and metastasis and enriches the cancer stem cell population (Vaira et al. 2013; Yu et al. 2011). However, the effects of KLF4 in cancer cells appeared to be context-dependent as in some studies, KLF4 was shown to suppress epithelial-mesenchymal transition and reduce cancer cell proliferation related to p21 upregulation (Lin et al. 2012b; Zammarchi et al. 2011).

NANOG functions together with SOX2 and OCT4 in embryonic stem cells as gatekeepers of pluripotency as well as involves in resetting of the somatic memory during reprogramming (Boiani and Scholer 2005; Takahashi et al. 2007). Mounting evidence suggests that NANOG is a critical factor for cancer stem cells to ensure their metastatic, self-renewing, drug-resistant and tumourigenic features (Jeter et al. 2015). NANOG is silenced in normal somatic cells but activated in various cancer types including breast, brain, gastrointestinal, liver, lung, ovary, pancreas and prostate (Akhavan-Niaki and Samadani 2014; Iv Santaliz-Ruiz et al. 2014; Mimeault and Batra 2011; Seymour et al. 2015; Tafani et al. 2014). Overexpression of NANOG correlates with poor prognosis and shorter patient survival in distinct cancer types (Chiou et al. 2008; Lee et al. 2015a; Lin et al. 2012a; Meng et al. 2010; Wang et al. 2014). The major pathways controlling cancer stem cell populations in distinct cancer types signal through NANOG like the platelet-derived growth factor-D (PDGR-D) signalling in prostate cancer, transforming growth factor β-1 (TGFβ-1) and epidermal growth factor (EGF) signalling in lung cancer or insulin growth factor (IGF) pathway in hepatocellular carcinoma (Kong et al. 2010; Pirozzi et al. 2011; Shan et al. 2012). Thus, all the studies imply that pluripotency transcription factor NANOG is a potential oncogene and a critical regulator in tumourigenesis.

SOX2 is a member of SRY-related high-mobility group box family proteins and significant in maintaining pluripotency in the inner cell mass (Kamachi et al. 2000). SOX2 synergises with OCT4 to activate ESC-specific genes while suppressing differentiation-related genes, and variations in SOX2 levels induce differentiation in pluripotent stem cells (Masui et al. 2007; Rodda et al. 2005). The activities of SOX2 or SOX2/OCT4 dimer are essential and sufficient for complete reprogramming (Giorgetti et al. 2010; Huangfu et al. 2008). High levels of endogeneous SOX2 expression do not necessitate ectopic SOX2 expression for reprogramming (Eminli et al. 2008). To date, many publications suggest the involvement of SOX2 in cancer cell regulation. SOX2 has been proposed to enhance proliferation, escape from apoptosis and increase migration, metastasis and invasion of various cancer types (Weina and Utikal 2014). Increased SOX2 expression is critical for the enrichment of cancer stem cell populations in different cancer types including breast, pancreatic, ovarian, cervix, gastrointestinal, prostate, lung and brain (Bareiss et al. 2013; Corominas-Faja et al. 2013; Gangemi et al. 2009; Herreros-Villanueva et al. 2013; Hutz et al. 2014; Liu et al. 2014a; Rybak and Tang 2013; Singh et al. 2012).

All these studies suggest involvement of pluripotency/reprogramming-associated factors in the regulation of carcinogenic phenotype, suggesting that overlapping pathways might regulate pluripotency and carcinogenesis. To date, our understanding of upregulated pluripotency factors in cancer types is limited to specific cancer types which are usually context dependent. Therefore, resolution of reprogramming mechanisms in both normal and malignant cells would broaden our understanding of cancer initiation.

Altered cell metabolism

Dysregulation of c-Myc in cancer cells contributes to the tendency of cancer cells to metabolise glucose via glycolysis independent from oxygen availability to support oxidative metabolism, namely Warburg effect (Dang 2007). The glycolytic phenotype is common in stem cell types, and stimulation of glycolysis enhances the reprogramming efficiency (Folmes and Terzic 2016). Somatic cells with lower oxidative capacity and greater glycolytic capacity were demonstrated to give rise to iPSCs with a higher efficiency (Panopoulos et al. 2012). Increase in glycolytic activity by the addition of glycolytic intermediates, activators, hypoxia or inhibition of p53 pathway increases the reprogramming efficiency. Similarly, c-Myc expression was shown to improve reprogramming by stimulating glycolysis (Folmes et al. 2011, 2013; Yoshida et al. 2009; Zhu et al. 2010).

Epigenome changes

Cancer is associated with the accumulation of genetic and epigenetic changes. Although the roles of genetic changes are well established in cancer, the cancer-specific epigenome is not well understood (Semi and Yamada 2015). Aberrant epigenome in cancer is characterised by global changes in DNA methylation, histone modifications and chromatin-modifying enzymes (Dawson and Kouzarides 2012; Virani et al. 2012). Epigenetic changes often act in accordance with the abnormal gene expression profiles both of which lead to the development of the disease (Munoz et al. 2012). Reprogramming also relies heavily on epigenetic modifications involving genome-wide changes in DNA methylation and histone modifications (Ladewig et al. 2013). Induction of pluripotent stem cells by ectopic expression of master regulators, OCT4, SOX2, KLF4, c-MYC (OSKM), led to the generation of iPSCs from various somatic cells yet with a low efficiency (Takahashi et al. 2007). Alternative methods have been utilised in order to improve the efficiency of iPSC induction including episomal plasmids, lentivirus, adenovirus, or Sendai virus-mediated gene transfer as well as by direct delivery of transcription factor mRNA or protein. Despite representing some success, these studies suggested that expression of all four factors alone is not sufficient for reprogramming, and thus, epigenetic events might contribute to or impede the reprogramming efficiency. Being able to generate iPSCs by using less transcription factors for generating iPSCs further support the fact that epigenetic state of a cell is a key factor for reprogramming (Eminli et al. 2009; Giorgetti et al. 2009; Kim et al. 2009; Niibe et al. 2011; Wang et al. 2013; Yu et al. 2007; Yulin et al. 2012). This led to the idea of the possibility of achieving reprogramming without the use of transcription factors. Is it possible to reprogram the cells by modifying the epigenome? Indeed, Hou et al. (2013) demonstrated that using a combination of seven small-molecule compounds is sufficient for reprogramming and exogeneous master regulators for pluripotency are dispensible for reprogramming in mouse somatic cells. Growing evidence suggests that combinations of small molecules in mouse cells could compensate for exogeneous reprogramming factors which generated iPSC-like cells with similar expression profiles and epigenetic state to embryonic stem cells (Kang et al. 2014; Kimura et al. 2015). Thus, for certain cell types epigenome editing could replace ectopic transcription factor expression for reprogramming, while for most cell types, overcoming epigenetic barriers necessitates combination of multiple mechanisms-induced forced expression of reprogramming factors. Reprogramming efficiency was shown to be improved upon treatment with small molecules including DNA methyltransferase inhibitors, histone deacetylase inhibitors, WNT signal modulators, cell senescence modulators and modulators of metabolism (Lin and Wu 2015). Cell origin thoroughly affects reprogramming efficiency as iPSC induction could not fully reset the epigenetic memory and the memory of the donor cell could be retained in the iPSCs (Hochedlinger and Plath 2009; Kim et al. 2010; Ohi et al. 2011; Polo et al. 2010; Ruiz et al. 2012). Incomplete reprogramming with inherited epigenetic memory generates iPSCs which have a tendency to differentiate towards the originated lineage (Bar-Nur et al. 2011; Sanchez-Freire et al. 2014). Although reprogramming efficiency was shown to be upgraded to 80–100 % by combinational treatment of small molecules, transcription factors and signalling pathway regulators, much effort should be put in for elucidating the mechanisms directing the terminally differentiated cells to erase their somatic state and gain pluripotency (Buganim et al. 2013; Rais et al. 2013; Vidal et al. 2014).

Concluding remarks

In this review, we described the similarities between carcinogenesis and reprogramming by visiting the recent advances in cancer cell reprogramming. iPSC technology has facilitated the generation of disease-specific models for distinct disease phenotypes, but less success has been gained for recreating carcinogenesis. The efficiency of cancer cell reprogramming is still low, and most studies raised contradicting reports on cancer-derived iPSCs for their ability to recapitulate the disease. Therefore, the complex epigenetic mechanisms behind oncogenesis and reprogramming should be uncovered before exploiting cancer-specific models which could serve for identification of early-stage markers and provide clues for cancer initiation or progression. Thus, the studies on reprogramming cancer cells or unravelling the epigenetic mechanisms beyond the nuclear reprogramming should be encouraged for the elucidation of the underlying mechanisms of carcinogenesis for developing the prospective cancer therapies.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.