As the core enzyme involved in DNA methylation, interfering with DNMTs could be the most effective way for its function disturbance. Many lncRNAs were identified that physically interact with DNMTs to regulate methylation on target genes (Figure 2A). Merry et al. (2015) discovered 148 lncRNAs that interact with DNMT1 in colon cancer by using the RNA immunoprecipitation sequencing (RIP-seq) method. Among these, one named DACOR1 (DNMT1-associated colon cancer repressed lncRNA 1), which presents downregulated expression in colon cancer, was identified to interact with DNMT1 and recruit this macromolecular complex at specific genomic sites to influence DNA methylation and gene expression. Induction of DACOR1 in colon cancer cells results in global hypermethylation at multiple loci without changing the DNMT1 expression level, many of the hypermethylated regions are associated with genes that participate in cancer related pathways, such as TGF-β/BMP signaling (Somasundaram et al., 2018). Similarly, another lncRNA SAMD12-AS1 was found highly up-regulated in gastric cancer. SAMD12-AS1 may facilitate the repression of p53 by recruiting DNMT1, thus promoting the progression of gastric cancer (Lu et al., 2021). In chronic myelocytic leukemia (CML), the lncRNA HOTAIR was found to enhance the methylation of PTEN promoter by recruiting DNMT1. Overexpression of HOTAIR could facilitate the proliferation, invasion, and migration of CML cells (Song et al., 2021). Besides PCGs, lncRNAs associated DNA methylation dysregulation are also widely found in promoters of other types of ncRNAs, such as miRNA. In hepatocellular carcinoma (HCC), miR-122 was identified as the methylation target of HOTAIR, the downregulated expression of miR-122 by HOTAIR leads to the activation of oncogene Cyclin G1 and promotion of tumorigenesis in HCC (Cheng et al., 2018). Another example is TINCR, this lncRNA can recruit DNMT1 to the promoter of miR-503 gene in breast cancer. Overexpression of TINCR could increase methylation and suppress the transcription of miR-503-5p. Of note, TINCR can also act as a ceRNA for miR-503-5p to regulate EGFR and interfere with JAK2–STAT3 signaling (Wang et al., 2021).

Besides the recruitment mechanism, lncRNA also sequester DNMTs from particular genome loci by a competitive interaction mode (Figure 2B). A lncRNA arising from the CEBPA gene locus termed ecCEBPA could compete with DNMT1, thus inhibit methylation of CEBPA gene and facilitate CEBPA expression in leukemic cells. (Di Ruscio et al., 2013). This lncRNA was later identified to interact with DNA strand by forming a DNA:RNA triple helices and protect regions near its binding site from methylation (Ogunleye et al., 2021). Another lncRNA, named 91H which located at the H19/IGF2 locus and transcribed in H19 antisense orientation, is overexpressed in breast cancer and prevent the maternal allele at the H19/IGF2 locus from DNA methylation, by this mechanism to induce overexpression of oncogenic H19 (Vennin et al., 2017). LncRNA HOTAIRM1 (HOX antisense intergenic RNA myeloid 1), which is located between the HOXA1 and HOXA2 genes, could interact with DNMTs and other epigenetic factors to sequester them away from HOXA1 promoter in glioblastoma multiforme (GBM). Upregulation of HOTAIRM1 could lead to reduced methylation levels of HOXA1 and finally to its upregulation of expression (Li et al., 2018). A similar observation was found in dental follicle stem cells (hDFSCs), in which HOTAIRM1 binding to the CpG islands of the HOXA2 promoter and reduce the binding of DNMT1 at the HOXA2 promoter, resulting in HOXA2 hypomethylation and deviant induction (Chen et al., 2020). These examples indicate that this regulatory mechanism by HOTARIM1 within the HOXA cluster could be universal across tissues and diseases.

LncRNAs are also found to interact with other DNA methyltransferases in addition to DNMT1 to influence the methylation pattern of target genes. For instance, lncRNA HOTAIR was shown to recruit DNMT3B to increase HOXA5 promoter methylation and silence its expression in acute myeloid leukemia (AML). HOTAIR silence and HOXA5 activation were found to induce apoptosis and reduce proliferation of AML cells (Wang et al., 2019d). Another lncRNA MROS-1 was found to modulate tumor suppressor PRUNE2 expression by interacting with DNMT3A in oral squamous cell carcinoma (OSCA). Higher methylation levels of PRUNE2 promoter induced by MROS-1 were associated with cell migration and metastases (Su et al., 2021). The lncRNA TTTY15 could interact with DNMT3A and prevent its binding to TBX4 promoter in non-small cell lung cancer (NSCLC), the lower expression level of TTTY15 and the associated downregulation of TBX4 is connected with metastasis and worse prognosis of NSCLC patients (Lai et al., 2019).

Besides interacting with DNMT proteins, lncRNAs could also regulate their expression level with different mechanisms (Figure 2C). For instance, one lncRNA named Dnmt3aos (DNA methyltransferase 3A, opposite strand) located on the antisense strand of DNMT3A was found to participate in the regulation of DNMT3A expression. Dnmt3aos is highly expressed in M(IL-4) macrophages, which leads to the highly coordinated expression of this sense-antisense pair of DNMT3A and Dnmt3aos. Elevated expression of Dnmt3aos and DNMT3A results in global DNA methylation changes in M(IL-4) macrophages (Li et al., 2020). In small cell lung cancer (SCLC), HOTAIR was found to inhibit expression of DNMT1 and DNMT3B, thus regulating the methylation of HOXA1 to mediate chemoresistance of SCLC (Fang et al., 2016). Whereas in AML patients, HOTAIR present up-regulated expression, which leads to downregulation of PTEN via DNMT3B-dependent pathway, and lead to doxorubicin resistance (Zhou et al., 2021).

LncRNAs have long been recognized to regulate gene expression via the ceRNA mechanism, by which lncRNAs act as a “sponge” to combine with miRNAs and sequester their interactions with mRNAs to de-repress the expression of targets. Many examples have been found for lncRNAs that regulate the expression of DNMTs as ceRNA (Figure 2D). In laryngeal squamous cell carcinoma (LSCC), H19 was found to be the sponge for miR-148a-3p, through which to regulate DNMT1 expression. Overexpression of H19 in LSCC leads to elevated expression of DNMT1 and genome wide change of DNA methylation, including MGMT (Wu et al., 2016). Similar observation was also found for the RP11-159K7.2—miR-206 – DNMT3A axis in LSCC. Overexpressed RP11-159K7.2 could interact with miR-206, which binds with DNMT3A 3′-UTR. Interestingly, DNMT3A was also found to inhibit the expression of miR-206 via a DNA methylation-dependent manner, thus a feedback loop is maintained between DNMT3A and miR-206 to keep its internal balance (Wang et al., 2020). In hepatocytes, HOTAIR was found as sponge of miR-29b, which also regulates the expression of DNMT3B to regulate the methylation level of PTEN (Yu et al., 2020). Besides interactions with miRNAs, lncRNAs are also found to regulate the mRNA level of DNMTs by interacting with other proteins. For instance, the RMST, a lncRNA capable of upregulating DNMT3B expression by interaction with the RNA binding protein HuR, leads to alterations in global methylation in cancers (Peng et al., 2020).

LncRNAs could also function to regulate protein expression for DNMTs, such as by mechanism of ubiquitination (Figure 2E). In HCC, lncRNA linc-GALH overexpression could enhance the ubiquitination of DNMT1 to accelerate its degradation. In this way, linc-GALH reduces the methylation level of Gankyrin to promote its expression (Xu et al., 2019). In another example, lncRNA LUCAT1 was found to interact with DNMT1 but now to inhibit the ubiquitination in esophageal squamous cell carcinoma (ESCC). Upregulated LUCAT1 thus stabilizes DNMT1 to enhance the methylation and inhibit the expression of tumor suppressors (Yoon et al., 2018). In addition, lncRNAs could also regulate local concentration of DNMTs by interfering with its subcellular location (Figure 2F). For example, the lncRNA CCDC26 could promote DNMT1 localization from cytoplasm to nucleus. In absence of CCDC26, DNMT1 is found mis-located in the cytoplasm, resulting in global hypomethylation (Jones et al., 2021). Examples of lncRNAs that interact with DNMTs to regulate methylation of downstream genes and their functions in cancers are summarized in Table 2.

DNMTs are responsible for catalyzing the conversion of cytosine to 5-mC whereas TET enzymes catalyze the successive conversion of 5mC to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) to promote locus-specific removal of methylation. DNA demethylation can be achieved either as a process in the absence of functional DNA methylation maintenance mechanism during DNA replication or through TET-mediated 5mC oxidation. In this case, regulation to TET family affects the methylation level of downstream genes as well. Studies have discovered many lncRNAs interact with TETs to regulate methylation process (Table 2). For example, lncRNA MAGI2-AS3 (MAGI2 antisense RNA 3) which is transcribed from the antisense strand near the MAGI2, acts as cis-acting factor to downregulate the DNA methylation level of the MAGI2 promoter by interaction with TET1 and promotes apoptosis by activating the Fas/FasL signaling pathway in breast cancer (Xu et al., 2021). In AML, MAGI2-AS3 recruits TET2 to the LRIG1 promoter region in trans and causes DNA demethylation of LRIG1. Downregulation of MAGI2-AS3 suppresses the self-renewal capacity of leukemic stem cell by promoting LRIG1 expression (Chen et al., 2020). LncRNAs are also found to recruit TET enzymes in an indirect mode. The lncRNA TARID (TCF21 antisense RNA inducing demethylation) could interact with both the TCF21 promoter and GADD45A protein, whereas GADD45A in turn recruits TET1 to activate the expression of TCF21 (Arab et al., 2014). The authors further show that TARID combine to TCF21 promoter to form an R-loop of DNA–RNA hybrids, which is recognized by GADD45A and then triggers TET1-dependent DNA demethylation (Arab et al., 2019).

TETs are also found to be regulated by lncRNAs at the transcriptional, posttranscriptional, and protein expression levels. In cervical cancer, the HOTAIR could regulate TET1 expression, which leads to promoter hypermethylation of Wnt/β-catenin signaling related genes. In Hela cells, upregulated HOTAIR leads to the decreased TET1 expression, which is associated with the transcriptional activity of Wnt/β-catenin pathway genes, such as PCDH10, SOX17, AJAP1, and MAGI2 (Salmeron-Barcenas et al., 2019). At the posttranscriptional level, TET1 is found to be regulated by lncRNA H19 via miRNA let-7 with ceRNA mode, TET1 expression alteration due to upregulation of H19 promotes TGF-β signaling related endothelial–mesenchymal transition in endothelial cells of atherosclerotic coronary arteries (Cao et al., 2020). A similar observation was found for TET3 in uterine leiomyomas, a H19—let-7—TET3 axis was identified for methylation regulation of fibroid-promoting gene and to drive proliferation of leiomyoma cells (Cao et al., 2019). At the protein expression level, a multifunctional lncRNA TETILA was found in diabetic skin that play a key role in wound healing. Zhou et al. (2019a) indicated this lncRNA could regulate TET2 stability through the ubiquitin-proteasome pathway and also promote TET2 nuclear translocation. In addition, TETILA also acts as a scaffold to recruit thymine-DNA glycosylase (TDG), which simultaneously interacts with TET2 at the promoter of MMP-9 for its demethylation and transcriptional activation.

One of the most intriguing observations have recently emerged in epigenetics is the subtle crosstalk between DNA methylation and other epigenetic modifications. Accumulating literature has revealed complex mechanisms underlying the interplay between DNA methylation and histone modification. Many partners of DNMTs have been found that involved in both of the DNA methylation and histone modification. In addition, DNA methylation status within genome present concomitant presence with other repressive marks, such as histone deacetylation. For example, HDAC1 has the ability to bind DNMT1, the histone deacetylase activity is required for DNMT1 related DNA methylation maintenance in heterochromatin (Fuks et al., 2000). DNMTs have also been identified to interact with G9a, which is responsible for mono-, di-and slowly trimethylation of histone H3 lysine 9 (H3K9). This interaction has been shown to play a role in the establishment of DNA methylation pattern for key genes in ES cells (Xin et al., 2003; Esteve et al., 2006). In addition, the PRC2 system, which has histone methyltransferase activity for H3K27me3, is connected to DNA methylation related gene silencing at specific loci. The PRC2 core component EZH2-dependent recruitment of DNMT3A was found to be associated with H3K27me3 and DNA methylation (Jin et al., 2009; Rush et al., 2009; Li et al., 2021c). This explains how lncRNAs interact with epigenetic factors to regulate DNA methylation at particular loci (Figure 3A). For instance, the PYCARD-AS1, which is antisense to the pro-apoptotic gene PYCARD, functions to induce DNA methylation and H3K9me2 modification of PYCARD promoter by recruiting the chromatin-suppressor proteins G9a and DNMT1 in breast cancer (Miao et al., 2019). Another example is the lncRNA KCNQ1OT1, which binds and recruits the heterochromatin protein HP1α, and finally lead to DNA methylation and H3K9me3 modification in the genome. One repeat-rich region within KCNQ1OT1 is identified mainly responsible for Hoogsteen base pairing with double-stranded DNA, by which to fulfill the function of protein recruitment. This observation demonstrates an example for lncRNA to induce and maintain epigenetic silencing at repetitive DNA elements, in order to safeguard against genome instability (Zhang et al., 2022). In pancreatic cancer, the upregulated LINC01133 was found to recruit EZH2 to for histone methylation and also to promote the promoter methylation of DKK1, thus activate Wnt signaling (Weng et al., 2019). LncRNA HOXB13-AS1 is found upregulated in glioma and negatively correlated with its surrounding gene HOXB13, this lncRNA could increase DNMT3B-mediated methylation of HOXB13 promoter by binding with EZH2 (Xiong et al., 2018). Similar examples include the regulation of LZTS1 by lnc-LALC during liver metastasis of colorectal cancer (Zhang et al., 2021), regulation of CXXC4 and SFRP2 by LUCAT1 in gastric cancer (Byun et al., 2020). In addition, lncRNA could also regulate promoter methylation of miRNA genes by interacting EZH2. For instance, lncRNA SNHG22 was found to recruit DNMT1 to miR-16-5p DNA promoter through EZH2 and inhibited miR-16-5p transcription via DNA methylation (Zhang et al., 2021c). LncRNA GIHCG physically associates with EZH2 and recruits EZH2 and DNMT1 to promoter regions of the miR-200b/a/429, which lead to changes of H3K27me3 and DNA methylation levels in the miR-200b/a/429 promoter, and dramatically silences their expression (Sui et al., 2016).

It is worth noting that many miRNAs regulated by lncRNA through promoter DNA methylation are also found to regulate the upstream lncRNAs or other epigenetic factors, by which a feedback loop formed to control the internal gene expression. For example, lncRNA SChLAP1 was found to recruit EZH2 and DNMT3A to repress multiple miRNA expression in prostate cancer, including the miR-340-5p/miR-143-3p/miR-145-5p, these miRNAs in turn regulate DNMT3A expression (Huang and Tang, 2021). In gastric cancer, EZH2 along with the histone demethylase LSD1 and DNMT1 were recruit by the lncRNA HOXA11-AS, this lncRNA also acts as sponge for miR-1297, antagonizing its ability to repress EZH2 protein translation (Sun et al., 2016). In glioblastoma, LINC00470 could enhance the expression of ELFN2 through adsorption of miR-101, and also affect the methylation level of ELFN2 by decreasing H3K27me3 occupancy (Liu et al., 2018). The above examples indicate that lncRNAs are able to control genes at the transcriptional level or post-transcriptional level through a variety of different mechanisms to achieve accurate regulation of expression levels for downstream target genes.

All DNA methyltransferases are known to use S-adenosylmethionine (SAM) as the methyl donor and generate S-adenosylhomocysteine (SAH) as by-product. The methyl donor SAM is synthesized from ATP and methionine by the methionine adenosyltransferase (MAT) (Lu and Mato, 2012), whereas SAH could be eliminated by S-adenosylhomocysteine hydrolase (SAHH), SAH also acts as feedback inhibitor of DNMTs (Lyko, 2018). Regulation on the genes involved SAM synthesis or SAH degradation by lncRNAs may lead to malfunction of DNMTs to interference DNA methylation (Figure 3B). The H19 for instance, could bind to SAHH and inhibits its function of SAH hydrolyzing, then give rise to genome-wide methylation alteration (Zhou et al., 2015). This mechanism was further observed in liver of metformin-exposed fetuses to induce hypomethylation and increased expression of HNF4α (Deng et al., 2017), and also in tamoxifen-resistant breast cancer to induce the upregulation of Beclin1 (Wang et al., 2019c), as well as in human lung tissue to regulate the LINE-1 methylation (Fu et al., 2018).

Interference to MAT may result in the alteration of the SAM concentration and disturbance of DNA methylation process. This has been confirmed by the interaction between lncRNA SNHG6 and MAT family members of MAT1A and MAT2A. On one hand, SNHG6 was found to upregulate MAT2A expression by act as sponge for miR-1297, on another hand, this lncRNA also downregulate MAT1A translation by suppressing the nucleus-cytoplasmic shuttling of MAT1A mRNA, thereby regulate genome wide methylation in hepatoma cells of HCC (Guo et al., 2018). Another lncRNA LINC00662 was identified to induce decay of MAT1A mRNA and also the degradation of SAHH protein by ubiquitination mechanism, in this way to reduce SAM and enhance SAH levels, which finally leads to global hypomethylation (Guo et al., 2020). It is worth mentioning a dual functional lncRNA PARTICLE in response to low-dose irradiation. Over expressed PARTICLE upon irradiation recruits the PRC2 to the promoter region of MAT2A in a DNA-RNA triplex form, in this way to regulate MAT2A expression via methylation. The altered expression level of MAT2A lead to changed concentration of SAM, which further influence the methylation level of downstream genes (O'Leary et al., 2015) (Figures 3A,B). This triplex-mediated expression regulation based on interaction between lncRNA PARTICLE and DNA strand was further proved to be widespread in the human genome (O'Leary et al., 2017). In summary, these studies indicate that lncRNAs could regulate methylation level of downstream genes by regulating the SAM related pathway genes.

Studies have indicated that lncRNAs could modulate gene for degradation and/or elimination of endogenous and exogenous toxins or medicines, by which they are able to exert their effects on drug metabolism and response to treatment (Table 2). For example, LINC00261 was found to recruit DNMTs to the promoter of the dihydropyrimidine dehydrogenase (DYPD), which is mainly responsible for 5-fluorouracil (5-FU) degradation. Increased LINC00261 promotes the methylation level within the DPYD promoter region and leads to its downregulation in esophageal cancer. As a result, 5-FU degradation is inhibited, finally results in an elevated sensitivity to 5-FU of the cancer cell (Lin et al., 2019). Similar observations were also found for the effect of LINC01419-GSTP1 regulation in esophageal cancer (Chen et al., 2019). In prostate cancer, regulation of KLF4 promoter methylation by LINC00673 is associated with paclitaxel resistance (Jiang et al., 2020). In lung adenocarcinoma, vincristine resistance is meditated by promoter methylation of LAMA3 induced by LINC00628 (Xu et al., 2019). In thyroid cancer, LINC00607 mediates doxorubicin resistance through the regulation of CASP9 methylation (Li et al., 2021). These observations lead to the thought that the chemical drug effectiveness can be improved for better treatment by regulating the expression level of these lncRNAs.

Another possible direction for cancer treatment is to interfere with lncRNAs involved in DNA methylome regulation by using gene editing methods. One example is the lncRNA 91H which is reasonable for inducing methylation of CDK4 promoter, knockdown of this lncRNA could suppress the tumorigenesis of osteosarcoma (Cheng et al., 2021). Some small molecules directly interfering lncRNAs responsible for methylation regulation could also be efficient treatment targets. For instance, metformin was found to induce H19 repression and the genome-wide DNA methylation alterations by modulating the activity of H19—SAHH axis, this observation provides a novel explanation for the mechanism and function of the metformin for the epigenetic regulation effect in cancer (Zhong et al., 2017). In addition, some chemical compound that interrupts the HOTAIR—EZH2 interaction are found to inhibit cancer cell invasion and migration, which was thought to be a potential approach for targeted therapy of cancers (Ren et al., 2019; Wang et al., 2021). In summary, lncRNAs involved in DNA methylation regulation are promising targets for applications in cancer therapy. Representative lncRNAs currently identified that are involved in DNA methylation regulation, and the associated cofactors, interaction mode, as well as target genes are listed in Table 2. This comprehensive summary revealed us a complex interaction network based on epigenetic regulatory mechanisms that remains to be further explored. In-depth analysis of non-coding RNA and other epigenetic regulatory elements including DNA methylation at the systemic level will help us to reveal the underlying mechanisms of tumor development and development, thus providing a new perspective for personalized tumor therapy.

In recent years, circRNAs have been revealed for their crucial role during the onset and progression of human disease by their important regulatory effect. The capacity of circRNAs interact with proteins involved in epigenetic modification manifests itself the ability for the transcriptional regulation on target genes (Table 3). Examples include a circRNA termed ACR (autophagy related circular RNA), which directly binds to DNMT3B and block DNMT3B-mediated DNA methylation of Pink1 promoter. Pink1 further brings about phosphorylation of the downstream target FAM65B, and finally inhibits autophagy and cell death in the heart (Zhou et al., 2019). An exosome derived circRNA circ_6790 from bone marrow mesenchymal stem cell was found to increase the nuclear translocation of CBX7, by this indirect interaction mode to recruit DNMTs and induce the methylation of S100A11 in pancreatic ductal adenocarcinoma (Gao et al., 2022). Many circRNAs are found to regulate the expression level of DNMT genes and finally influence the downstream target methylation. For example, hsa_circ_0012919 is downregulated in CD4⁺ T cells of systemic lupus erythematous (SLE) and results in the increased the expression of DNMT1 and finally leads to the hypermethylation of CD70 and CD11a (Zhang et al., 2018). A similar example is the circ-Amotl1, which interacts with STAT3 and facilitate its nuclear translocation and the binding to the promoter of DNMT3A gene, the activated DNMT3A further induce miR-17 promoter methylation and decrease its expression (Yang et al., 2017). In addition, a multi-functional circRNA was found that regulate downstream methylation by different mechanisms. The circRNA derived from FLI1 termed FECR1 is able to recruit TET1 to the promoter of the host gene and lead to the hypomethylation in cis, in addition, this circRNA could also bind to the DNMT1 promoter, where it downregulates DNMT1 transcription in trans. In this manner, this circRNA regulator controls tumor growth and metastasis of breast cancer (Chen et al., 2018).

The ceRNA mechanism is also widely involved in the processes of methylation regulation by circRNAs. For example, hsa_circ_0040809 regulates cell proliferation of colorectal cancer by upregulating DNMT1 via targeting miR-515-5p (Mao et al., 2021). Another example is from HCC, the circSOD2 was activated by promoter modification of H3K27ac and H3K4me3, the activated circSOD2 inhibits miR-502-5p expression and rescues miR-502-5p target gene DNMT3A expression (Zhao et al., 2020). Similar observations include the circMEMO1—miR-106b-5p—TET1 axis (Dong et al., 2021) and circTRIM33–12—miR-191—TET1 axis (Zhang et al., 2019), which play key roles for controlling cell proliferation, migration and immune evasion. This ceRNA mechanism for downstream target methylation regulation was also found during SLE development (Wang et al., 2018). Interestingly, miRNA genes are also found to be the methylation targets of circRNA regulators. For instance, the circ-ATAD1 leads to miR-34b gene methylation in AML to increase the cell proliferation (Wu et al., 2021). This very circRNA was found to regulate miR-10a gene methylation in endometrial cancer (Yang et al., 2021). Other similar examples are also identified in many types of diseases (Table 3) (Wu et al., 2021; Zhou et al., 2021; Zhang et al., 2021; Wan et al., 2021; Guo et al., 2022). However, the detailed mechanism on how circRNA influence the methylation of miRNA gene promoters are largely unknown and remains to be further investigation.