The nucleosome is the basic chromatin repeating unit, and the core histones that make up the nucleosome are small proteins. Histone modifications mainly include acetylation, methylation, and ubiquitination, and the histone modification state controls whether the transcription complex can come into close proximity with the target gene, affecting its expression activity (25–27). The quantity, position, and type of histone modifications are collectively referred to as histone codes, which play an important role in cell differentiation and maintenance (28–30). Studies have shown that abnormal expression of histone codes is an important feature of cancer tissues and is related to the heterogeneity of cancer cells (31, 32). Studies have shown that abnormal levels of the histone demethylases, KDM6A and KDM6B, are associated with pediatric acute myeloid leukemia (AML) (33). Moreover, modification of the histone proteins H3K9ac, H3K27ac, and H4K16ac plays an important role in the progression and prognosis of head and neck squamous cell carcinoma (HNSCC) (34).

DNA methylation widely exists in prokaryotes and eukaryotes and is an epigenetic mechanism controlling gene expression (35, 36). Previous studies have revealed epigenetic reprogramming during embryo development (5, 37, 38). With cell differentiation, new methylation patterns are formed to ensure the specific expression of genes in organisms (39, 40). DNA methylation is catalyzed by methyltransferases, including DNMT1, DNMT3A, and DNMT3B (41). Among these enzymes, DNMT1 is responsible for the transmission of methylation patterns during mitosis to prevent passive demethylation. Dysplasia and death have been observed in DNMT1 knockout mice (42). DNMT3A and DNMT3B can methylate unmethylated CpG sites, which is important for embryonic development and tumorigenesis (43, 44). Methylation of cytosine residues leads to gene silencing, which plays a key role in the proper regulation of gene expression, genomic imprinting, X-inactivation, and development. Interestingly, abnormal DNA methylation is often observed in clinical specimens of cancer tissues (45). During tumorigenesis, abnormally high methylation of cytosines in promoter CpG islands, as well as overall gene hypomethylation, lead to genome-wide instability and altered gene expression profiles, including silencing of oncogenes, activation of endogenous retroviruses, and upregulation of tumor antigens and oncogene expression (46, 47). Tumor-specific methylated genes can be detected in circulating tumor cells, blood, urine, and other body fluids and are therefore commonly used in the diagnosis and prognosis of early-stage tumors (48, 49).

Non-coding RNAs (ncRNAs), which are not involved in protein-coding, mainly include microRNAs (miRNAs), circular RNAs (circRNAs), ribosomal RNAs (rRNAs), transfer RNAs (tRNAs), small nuclear RNAs (snRNAs), and small nucleolar RNAs (snoRNAs) (50–52). ncRNAs mainly affect gene expression at the transcriptional and translational levels, and play an indispensable role in embryonic development, cell differentiation, damage repair, and regulation of cell function (53–55). Dicer1 is one of the most important enzymes that produce miRNAs. It has been reported that Dicer1-deficient mice have abnormal organ development or face embryonic death, which was attributed to the failure of embryos to correctly process miRNAs (56). Abnormal ncRNAs have been reported to play an important role in tumorigenesis, metastasis, and drug resistance (51, 52, 57). For instance, miRNA-143 regulates a variety of signaling pathways, including WNT/β-catenin, RAS-MAPK, and PI3K/AKT, thereby affecting tumor growth (58). miR-193a-5p promotes tumor cell metastasis by regulating the EMT signaling pathway (59). Furthermore, both germline and somatic mutations in Dicer1 have been identified in diverse types of cancer (60, 61). The errors in ncRNAs are closely related to cancer. However, the exact mechanism is still surprisingly controversial. This may be related to the complexity of the ncRNA regulatory mechanism.

By inhibiting the activity of DNMTs, the expression of tumor suppressor genes is promoted to inhibit the growth of tumor cells. The main nucleoside and non-nucleoside DNMT inhibitors (DNMTis) include azacitidine (AZA) and decita21bine (63). Azacitidine blocks cytosine methylation by noncompetitive inhibition of DNMT1, resulting in the depletion of methyltransferases and DNA hypomethylation, but it is ineffective for quiescent cells that cannot divide (64). Low-dose azacitidine and decitabin can induce reactivation of the genes that were previously silenced by methylation, thereby inducing the formation of new phenotypes, reducing proliferation, and increasing apoptosis of offspring cells. High-dose drugs have cytotoxic effects and can directly cause tumor cell death (65, 66). Azacitidine and decitabin are approved by the Food and Drug Administration (FDA) as first-line drugs for the treatment of myelodysplastic syndromes (MDS) and leukemia. However, these drugs did not show significant efficacy in solid tumors such as gastrointestinal cancer, lung cancer, breast cancer, and melanoma, and their use was limited due to their side effects and drug instability (67, 68). Dniunaite (69) observed that downregulation of miR-155-5p was significantly correlated with promoter methylation in prostate cancer. DOT1L inhibitors SYC-52221 and EPZ004777 inhibited DNMT3A-mutant cell proliferation, inducing cell cycle arrest and terminal differentiation (70).

HDAC is a highly conserved group of enzymes that removes acetyl groups from the tail of histone lysine. HDAC promotes chromatin closure and inhibits gene transcription by deacetylating histones (71). HDACi is a new antitumor drug that regulates gene expression. It has extensive effects on malignant tumors, including inhibition of cell differentiation, cell cycle growth, and angiogenesis, as well as induction of apoptosis, and immune regulation (72). In animal models, HDAC inhibition was found to inhibit tumor growth and reduce malignant proliferation by downregulating positive cell cycle regulators, such as cell cycle proteins D1, c-Myc, and AKT (73, 74).

Currently, vorinostat and romidepsin are approved for the treatment of skin T-cell lymphoma (75). However, these drugs do not achieve long-term stable efficacy (76–78). This is not only related to low drug stability and high toxicity, but also to abnormal pathway activation. Abnormal activation of the PI3K/AKT, MEK/ERK, and FAK signaling pathways has been observed in the treatment of multiple myeloma with HDACi (79, 80). 5−aza−2’-deoxycytidine promotes migration of acute monocytic leukemia cells via activation of the CCL2/CCR2/ERK signaling pathway (81).

Since the drugs target DNA methyltransferase and histone deacetylase, those non-specific alteration of cancer cell methylation and acetylation levels cannot inhibit the development of cancer. In contrast, epigenetics may form complex networks in cells, thereby affecting other genes and signaling pathways, that may be an important reason for the existing epigenetic drug resistance. According to the epigenetic landscape theory (82), it is believed that cell differentiation is a manifestation of epigenetic homeostasis. In other words, targeting epigenetic homeostasis at the molecular level is the future direction of epigenetic treatment for cancer.

Study have reported the C2H2 zinc finger transcription factor B cell CLL/lymphoma 11A (Bcl11a) is essential for lymphoid development. The deletion of Bcl11a prevents further development of hematopoietic stem cells (HSCs) into lymphocytes (97). Bcl11a^-/- HSC alters cell cycle progression. A general upregulation of cell cycle protein genes and a downregulation of the quiescent regulator Cdkn1c (p57) and G2/M markers such as Prc1, Plk1 and Mki67 (Ki-67) of Bcl11a^-/- HSC can be observed, and cells eventually appear to proliferate uncontrollably (98, 99).。Similar results can be found in the corresponding studies: DNMTi can inhibit the further differentiation of bone marrow mesenchymal stem cells into osteoblasts and chondrocytes. At the same time, the expression of the anti-senescence genes (TERT, bFGF), and the anti-apoptosis gene (BCL2) was up-regulated and the expression of the apoptotic gene (BAX) was down-regulated (100, 101). The above studies illustrate that cells fail to complete differentiation to form a differentiation lock, i.e., they do not reach epigenetic homeostasis and are transformed into cancer cells. Some mutations in tumor suppressor genes can lead to damaged DNA repair function, such as the BRCA1/2 and P53, and differentiation locks are more likely to be damaged since the gene coding sequences only account for a small proportion of their genome (102, 103). The current research may indirectly confirm our opinion that cancer is a population of cells without epigenetic homeostasis. The non-coding RNA and protein profiles of cancer cells are quite different from those of normal cells, which is also a consequence of the imbalance in epigenetic homeostasis and is often manifested as cell differentiation disorder and cellular dedifferentiation (104, 105).

The consequences of epigenetic modifications may differ according to their position in the differentiation lock. Invisible damage may occur as a result of the defects in EcDL. Cell maturation arrest caused by defective hypostatic differentiation locks (HDLs) results in cancer cell transformation under the continuous stimulation of proliferation signals. Cells with defective standard differentiation locks (SDLs) dedifferentiate and transform into cancer during proliferation, and the invisible damage may become exposed. The invisible damages may explain why the mutation frequency of oncogenes and anti-oncogenes in the human population is much higher than the incidence of cancer (106–109). A possible reason is that tissue-specific alterations in differentiation locks, and defective EcDLs may not impose an effect on the process of tissue carcinogenesis.

Any factor, including physical, chemical, and biological factors, that can damage cells may lead to differentiation lock defects, thereby increasing the incidence of cancer (110–112).

In life, a variety of damages and stresses are often met. Mild damage and stress are often dealt with by the asymmetric division of stem cells. When stress exceeds tissue tolerance, stem cells must deal with symmetric divisions (113, 114).

When the cells are asymmetrically divided, the HDL defect (if exists) will lead to the maturation arrest of the daughter cells. These cells cannot complete the next stage of differentiation, and some of the daughter cells die. However, some daughter cells survive and transform into cancer cells under the stimulation of a continuous proliferation signal. When the stem cells are symmetrically divided, the defects of the SDLs are exposed (if exists). As a result, stem cells are dedifferentiated to transform into cancer cells (115). Cell carcinogenesis is a gradual process, and epigenetic homeostasis has a certain tolerance to damage. DNMT1 mainly maintains DNA methylation pattern during DNA replication, ensuring that the pattern is inherited by the offspring. In the early stages, defects in DNMT1 may be related to the accumulation of cell mutations. When the damage exceeds the steady-state tolerance, the cells become cancerous, which may be the internal relationship between aging and cancer.

Therefore, any pressure to stimulate stem cell division can increase the possibility of defective differentiation lock exposure, including injury, infection, and chronic inflammation (116–119).

As discussed above, the essence of tumor is cells that cannot form epigenetic homeostasis. The correct differentiation process depends on information transmission between cells and the interaction between cells and stroma (120, 121). The environment required for normal development has disappeared, and cancer cells cannot form the correct differentiation lock. Dr. Tushar reported bone marrow microenvironment lead to β-catenin activation and disease progression of MDS (122). Just like without the help of molecular chaperones, peptide chains cannot form proteins properly. Under internal gene-driven and error-induced environments, cancer cells produce heterogeneous daughter cancer cells (123, 124).

Numerous studies comparing gene expression in tumor tissues with paracancerous tissues have found that a large number of genes normally silenced during cell differentiation were activated in cancer. Moreover, these differences in gene expression patterns correlated with the malignancy of the tumor (125, 126). Those studies suggest that epigenetic homeostasis makes a crucial contribution to malignancy.

The more distinct the defective EDLs from SDLs, the lower the differentiation degree of cells, and the more apparent the characteristics of malignant tumors are (127–129), and vice versa. With the further cancer development, the differentiation locks in the upper layer will change, and the cells will show the characteristics of a lower differentiation stage. At this point, cancer tissues show typical malignant tumor characteristics (130, 131).( Figure 2 ) In the late stage of cancer development, some or all genes that had been closed in the embryonic stages are open.

Epigenetic abnormalities in tumor cells have always been of interest to scientists and physicians. Scholars have found that epigenetically abnormal tumor cells often show apoptosis, which may be a target for tumor therapy. Destruction of epigenetic homeostasis can induce apoptosis of cancer cells. Existing studies have found that cisplatin can not only directly kill cancer by DNA crosslinking, but also induce epigenetic changes of cancer cells and further induce apoptosis of cancer cells (132). Destruction of epigenetic homeostasis may enhance cancer immunogenicity. Researchers have found that inhibition of the histone demethylase LSD1 enhances tumor immunogenicity and T-cell infiltration in tumors. Thus, LSD1 could be used as an anti-tumor immunosuppressant in combination with anti-PD-1 immunotherapy for tumor treatment, with encouraging results in small-scale clinical trials (133, 134). Disruption of epigenetic homeostasis may alter the drug resistance of tumor cells. Methylguanine methyltransferase (MGMT), which repairs temozolomide (TMZ)-induced O6-methylguanine (O6mG) adducts, induces drug resistance in tumor cells. New study has shown reduced production of MGMT in the presence of epigenetic instability, demonstrating that generating epigenetic instability through destruction of epigenetic homeostasis may be a viable strategy to mitigate anticancer drug resistance (135). The above studies are all based on the biological effects generated after the destruction of epigenetic homeostasis to kill tumor cells, which are the current research hotspots.

As mentioned above, imbalance of epigenetic homeostasis may be the etiology of tumors. Some scholars believe that by re-establishing epigenetic homeostasis may be a new strategy for cancer treatment. Surprisingly, some studies have reported that breast cancer cells can be terminally differentiated into adipocytes by using the rosiglitazone combined with the trametinib (136, 137). This exciting finding suggests that re-establishing epigenetic homeostasis, i.e., tumor cell transdifferentiation, may be the therapeutic direction for cancer. After transdifferentiation, tumor cells are transformed into new, solid, controlled cells, and patients may achieve durable, stable remission as a result. This fascinating therapeutic prospect has attracted the attention of scientists. However, the lack of understanding of human cell differentiation pathways and differentiation-inducing conditions continues to limit the development of cancer transdifferentiation therapies.