Epigenetic Regulation of p21cip1/waf1 in Human Cancer

p21cip1/waf1 is a central regulator of cell cycle control and survival. While mutations are rare, it is commonly dysregulated in several human cancers due to epigenetic mechanisms influencing its transcriptional control. These mechanisms include promoter hypermethylation as well as additional pathways such as histone acetylation or methylation. The epigenetic regulators include writers, such as DNA methyltransferases (DNMTs); histone acetyltransferases (HATs) and histone lysine methyltransferases; erasers, such as histone deacetylases (HDACs); histone lysine demethylases [e.g., the Lysine Demethylase (KDM) family]; DNA hydroxylases; readers, such as the methyl-CpG-binding proteins (MBPs); and bromodomain-containing proteins, including the bromo- and extraterminal domain (BET) family. We further discuss the roles that long noncoding RNAs (lncRNAs) and microRNAs (miRNAs) play in the epigenetic control of p21cip1/waf1 expression and its function in human cancers.


Introduction
Regulation of gene function is essential to the coordination of cellular processes and tissue homeostasis. In cancer, dysregulation of genes is commonly mediated by genetic events such as activating mutations (e.g., k-RAS, PI3K, EGFR), gene fusions (e.g., BCR-ABL, NTREK), or amplifications (e.g., FGFR). Besides these mechanisms, additional features "on top" of the genetic information, i.e., epigenetic modifications, are heavily involved in the orchestration of gene functions. The term 'epigenetics' was first used by Conrad Hal Waddington in 1942 and is now defined as inherited changes in gene expression without alterations in the DNA [1,2]. While Hanahan and Weinberg defined the hallmarks of cancer based largely on genetic alterations, an increasing body of evidence now supports the view that epigenetic mechanisms contribute to the development of the malignant phenotype and can even drive cancer formation on their own [3,4].
An epigenetic signature is mediated by enzymes that alter DNA methylation or modifications of histones via methylation, acetylation, phosphorylation, ubiquitination, and various other mechanisms [5]. Besides these so-called epigenetic writers and erasers, enzymes that "read" the chromatin state form part of the epigenetic machinery [6]. Furthermore, various noncoding RNAs have recently been identified as playing a major role in epigenetic regulation [7]. p21 cip1/waf1 is a low-molecular-weight molecule (21 kDa) that inhibits cyclin-dependent kinases. It is a key downstream target of p53 upon DNA damage and has multiple functions during cell cycle regulation, DNA repair, gene transcription, and apoptosis [8]. We have previously discussed the link between histone deacetylase inhibitors (HDACi) and p21 cip1/waf1 signaling [9]. Here, we will extend this discussion to epigenetic mechanisms, such as DNA or histone methylation, and their effects on p21 caip1/waf1 in cancer cells.

Epigenetic Writers
The epigenetic landscape of tumor cells is created by various chemical modifications at DNA or histone tails.
DNA methylation occurs in somatic cells at the so-called CpG (cytosine-phosphate-guanine) islands, mostly CG-rich DNA sequences in the gene promoter regions and usually leads to silencing of the respective gene. This modification is mediated by the DNA methyltransferase (DNMT) family [10]. In cancer, about 5-10% of promoter regions become abnormally methylated [11]. DNMT1 is considered to be a maintenance DNA methyltransferase that copies methylation patterns during DNA replication and is thus important to the inheritance of DNA methylation, while DNMT3a and DNMT3b act as de novo DNA methyltransferases during embryogenesis and in somatic tissues [12,13]. Although the promoter region of CDKN1A, the gene encoding p21 cip1/waf1 , is usually not methylated in acute myeloid leukemia (AML), decitabine is capable of inducing its re-expression via induction of DNA damage and activation of the ATM/p53 pathway [14].
Histone modifications make a major contribution to the transcriptional outcome of a specific gene sequence. Interestingly, a histone lysine residue can be modified differentially; i.e., it can be acetylated or (tri-)methylated, and each modification will contribute to cancer development in a different way and extends the language of the DNA sequence [15,16]. Lysine acetyltransferases (KATs) catalyze the transfer of an acetyl group from acetyl-CoA to the ε-amino group of a lysine. Histone acetylation is a modification that can be recognized by epigenetic readers, but it also leads to a more relaxed chromatin structure and, consequently, an enhanced binding of transcription factors. Acetylation can occur at different histone tail lysines (K9, K14, K18, K5, K8, and K12) and in non-histone proteins. TIP60 and GCN5 are further acetyltransferases that link epigenetic modifications to p21 cip1/waf1 , e.g., by acetylation of p53 and MYC, respectively [17][18][19].
H3K4 methylation in general marks a transcriptionally active gene. There are six methyltransferases (MLL1/KMT2A, MLL2/KMT2B, MLL3/KMT2C, MLL4/KMT2D, SETD1A/KMT2F, and SETD1B/KMT2G) that lead in different complexes to this active histone modification. Mutations and rearrangements of the histone-lysine N-methyltransferase 2 (MLL) family proteins play a major role in the initiation and maintenance of various cancer types [20]; p53 might be involved in the correct recruitment of the Set1 complex to DNA-damage-responsive genes and thus links this phenotype to p21 cip1/waf1 [21]. The association of methyltransferases in diseases has recently been summarized in Hyun et al. [22].
The H3K27me3 mark is associated with gene silencing. The responsible Polycomb repressive complex 2 (PRC2) is composed of four core subunits: EZH2, SUZ12, EED, and RbAp46/48. The catalytic subunit EZH2 is responsible for the mono-, di-, and trimethylation of lysine 27 on histone H3 [23]. EZH2 is upregulated in many cancers, and the inhibition of PRC2 catalytic activity has emerged as a promising therapeutic approach in cancer [24,25].
Dot1L is responsible for the active mark at H3K79 that is localized at the globular domain of histone 3. The relationship of the abovementioned epigenetic writers to p21 cip1/waf1 is summarized in Table 1 and discussed in further detail below.

Epigenetic Erasers
Modifications to DNA and histones need to be reversible to be functional regulators of gene activity and are performed by distinct enzymes that are classified as epigenetic erasers. Generally, the removal of methylation marks facilitates transcription and enhances gene expression, while deacetylation has suppressive effects.
DNA methylation can be reversed by the ten-eleven translocation (TET1-3) family of proteins [27]. TET proteins are DNA hydroxylases that convert 5-methylcytosine to 5-hydroxymethylcytosine and, if the oxidization process is repeated, to 5-formylcytosine and 5-carboxylcytosine [28][29][30]. 5-hydroxymethylcytosine is commonly found in sites of active chromatin, such as promoters and transcriptional start sites, and it has been identified as a recognition signal for epigenetic readers [31,32]. Mutations in TET2 (leading to decreased 5-hydroxymethylcytosine levels) have been associated with various hematologic malignancies [33], although the exact mechanism of how TET2 contributes to the development of such cancers remains unclear [34]. Epigenetic inactivation or reduced expression of TET3 has been demonstrated in solid tumors such as head and neck, cervical, or colon cancer [35][36][37].
Histone lysine demethylases (KDMs) consist of more than 30 enzymes that have a distinct substrate specificity and belong to two different enzyme families: the Flavin-Adenin-Dinukleotid (FAD)-dependent amine oxidase family with KDM1A and KDM1B (LSD1, LSD2) as the only two members and the Jumonji C-domain-containing family of dioxygenases with more than 30 members in several subfamilies (KDM2-8). KDM1A/B can remove mono-and dimethyl groups from histone H3K4, while the Jumonji-family enzymes can also remove trimethyl groups [38,39]. So far, no clear evidence for specific histone arginine demethylases exists. Peptidylarginine deiminase 4 (PAD4) and JMJD6 have been shown to mediate arginine demethylation, but both enzymes also possess additional properties beyond demethylation [40,41].
Histone deacetylation is removed by zinc-dependent Class I, II, and IV histone deacetylases (HDAC) or Nicotinamide adenine dinucleotide + (NAD+)-dependent sirtuins that are also referred to as Class III HDACs. HDACs remove ε-amino acetyl groups from lysine residues in histones and in non-histone proteins (especially mediated by Class II HDACs, which are also found in the cytoplasm), which explains the pleiotropic effects and rather narrow therapeutic window of HDAC inhibitors (HDACi) in humans [42,43]. HDACs are part of larger multi-enzyme complexes, including Sin3, NuRD, and Co-REST, that determine their target recognition and substrate specificity, but also have demethylase properties [44,45]. Mechanistically, removal of acetylation from histones restores a condensed and transcriptionally repressive status. While mutations in HDACs are uncommon, altered expression levels of HDACs have been described in various human cancers and proposed as prognostic markers, e.g., in colorectal, ovarian, breast, and liver cancer [46][47][48][49].

Epigenetic Readers
Epigenetic modifications are recognized by a group of proteins that are referred to as epigenetic readers. Most reader proteins specifically recognize and interact with specific epigenetic marks; however, some readers also bind to unmodified histones whose functional relevance is not well understood yet [50][51][52][53].
Histone lysine methylation marks are recognized by a plethora of different protein families that specifically interact with distinct lysine residues and their methylation pattern [55,56]. Interestingly, a cross-talk between DNA and histone methylation exists by, e.g., interaction of the ATRX, DNMT3, DNMT3L (ADD) domain of DNMT3 with the histone H3 or the binding of the DNA methylation reader UHRF1 with unmethylated H3R2 [57][58][59].
Acetylation of histones leads to an open chromatin structure that is recognized by proteins containing bromodomains or tandem plant homeodomain (PHD) domains [60,61]. In humans, more than 60 bromodomains have been identified [62]. Among them, the bromo-and extraterminal domain (BET) family proteins (BRD2, BRD3, BRD4, and BRDT) are the best studied and have been shown to regulate transcriptional elongation processes as well as cell cycle progression. Interestingly, inhibition of BET proteins leads to downregulation of MYC, which is a key regulator of cellular survival and proliferation (e.g., via p21cip1/ waf1) in hematologic and solid tumors [63][64][65]. Other acetylation readers, such as EP300 and CREBBP, possess only a single bromodomain and are also known to regulate gene transcription processes [66].

Promoter Methylation of CDKN1A
Expression of p21 cip1/waf1 is regulated by several epigenetic and other additional post-transcriptional mechanisms that are also interconnected with each other (Tables 1 and 2).
The proximal promoter of CDKN1A contains several CpG dinucleotides that can be subject to DNA methylation [67,68]. Methylation affects Sp1 and Sp3 binding activity and was shown to suppress p21 cip1/waf1 transcription independently of p53 [69,70]. Some anticancer compounds, such as curcumin, induce apoptosis by promoter demethylation of CDKN1A [68]. In prolactinoma, low p21 cip1/waf1 levels caused by CDKN1A promoter hypermethylation were shown to contribute to enhanced proliferation in invasive specimens [71].
Hypermethylation of the Sp1 and Sp3 genomic region and concomitant transcriptional repression were found in different hematologic and solid cancers, including breast cancer, and were associated with a poorer prognosis due to more aggressive phenotypes [72][73][74]. DNMTs, as mediators of DNA methylation, have thus also been shown to have elevated levels in cancer, and pharmacological inhibition of these enzymes is already a clinical standard in hematologic diseases [75]. Furthermore, DNMTs compete with p21 cip1/waf1 for PCNA binding and overexpression of DNMTs could thus release p21 cip1/waf1 from PCNA and render it more susceptible to proteasomal degradation [76,77].
Nevertheless, low p21 cip1/waf1 levels in tissues are not always associated with CDKN1A promoter hypermethylation as previously shown in osteoarthritic chondrocytes [78].
Analyzing in silico the MethHC database that integrates DNA methylation and mRNA expression data from the Cancer Genome Atlas (TCGA), we observed that there is no uniform pattern when comparing tumor with nontumor DNA for CDKN1A gene expression. Bladder cancer, colon cancer, liver hepatocellular carcinoma, and thyroid carcinoma show a lower CDKN1A expression in tumors, whereas cervical squamous cell carcinoma, lung squamous carcinoma, prostate adenocarcinoma, sarcoma, skin cutaneous carcinoma, and uterine corpus endometrial carcinoma show a higher p21 cip1/waf1 encoding mRNA expression in tumors compared to their normal counterparts ( Figure 1). Head and neck carcinoma did not even show different gene expression levels. Except for prostate and skin carcinoma, there were no impressive differences visible to the naked eye in the graphs. As expected, when analyzing DNA promoter methylation and CDKN1A gene expression, there was a negative correlation between both parameters except for uterine corpus endometrial carcinoma; however, the correlation coefficients were not remarkably high ( Figure 2). These results point to the fact that CDKN1A silencing is not epigenetically regulated by DNA methylation only. We summarize the major mechanisms in the following paragraphs.
SETD1A a histone methyltransferase that ensures that the active H3K4 methylation triggers cell proliferation by suppressing p21 cip1/waf1 in a p53-independent manner. On the other hand, SETD1A knockdown leads to a p21 cip1/waf1 -dependent senescence phenotype [93]. The fact that SETD1A is often amplified in invasive breast carcinomas might reflect the better tolerance of these aggressive tumor cells to mitotic stress. Cell cycle arrest in SETD1A knockdown senescent cells is independent of mutations in p53, Rb, and p16, which are known senescence mediators; instead, it is sustained through transcriptional suppression of SKP2, which degrades p27 and p21 cip1/waf1 . The histone methyltransferase SETD2, which is responsible for the trimethylation of H3K36, is necessary for the proper mismatch repair in vivo [35]. In this regard, it has been shown that the hMutSα subunit of the mismatch repair protein hMSH6 contains a Pro-Trp-Trp-Pro (PWWP) domain that reads the H3K36me3 histone mark [94]. SETD2 can interact with p53 to regulate the transcription of its target genes encoding for e.g. p21 cip1/waf1 , Fas, and Puma [95]. SETD2 loss could also promote tumor growth in a p53-independent manner, leading to aberrant mitosis and multinucleated giant cells in tumors.
In lymphoma, the growth factor independent 1 (Gfi1) oncoprotein mediates the recruitment of the histone lysine methyltransferase G9a in complex with the histone deacetylase HDAC1 to the CDKN1A promoter. The subsequent increase in H3K9 dimethylation leads to diminished CDKN1A transcription [36]. Recently, the COUP-TF-interacting protein 2 (CTIP2) and the histone methyltransferase SUV39H1 were shown to be recruited to the proximal SP1 site of the CDKN1A promoter, and trimethylation of H3K9 was found to lead to transcriptional repression of CDKN1A in microglial cells. In parallel, human immunodeficiency virus (HIV-1) infection of macrophages activates CDKN1A gene expression and promotes G2/M cell cycle arrest. Both actions seem to facilitate HIV-1 replication [37].
Overexpressed MYC induces acute erythroleukemia, an uncommon AML subtype that is associated with a worse prognosis. The histone methyltransferase EZH2 has been reported to be a target of the MYC oncoprotein. MYC knockdown results in reduced EZH2 expression and an increase in p21 cip1/waf1 expression by removal of the silencing EZH2-specific H3K27me3 mark at the CDKN1A promoter. The subsequently induced p21 cip1/waf1 -dependent senescence phenotype is p53-independent [38].
In non-small-cell lung cancer, the long noncoding RNA LINCO1133 recruits EZH2 and lysine-specific demethylase 1A (LSD1) to the CDKN1A promoter to repress its transcription, leading to a proliferative stimulus [39]. Whereas EZH2 sets the silencing H3K27me3 mark, LSD1 removes the active H3K4me1/2 code to shape the chromatin into a repressive conformation. This p21 cip1/waf1 suppression has also been associated with epithelial-to-mesenchymal transition, a process that is important for invasion and metastasis. Downregulation of methylation of H3K79 leads to a p21 cip1/waf1 -dependent induction of the senescence program in lung cancer, which is believed to represent a barrier to tumorigenesis. When tumor cells are deficient in the responsible methyltransferase DOT1, they show a remarkable chromosomal instability [88]. When investigating a panel of histone methyltransferases in prostate cancer, PRMT6, a Type I arginine methyltransferase that builds the H3R2 dimethylation, was highly upregulated compared to nontumor tissue [89]. PRMT6 counteracts the MLL complex at H3K4 and both marks are mutually exclusive. When the repressive H3R2me2 mark is lost, tumor aggressiveness is significantly reduced by inhibition of the oncogenic AKT pathway and senescence induction via p21 cip1/waf1 upregulation. In leukemia, methylation at H4K20 by suppressor of variegation 4-20 homolog 1 (Suv4-20h1) promotes cell cycle G1/S transition and, hence, proliferation by repressing the p21 cip1/waf1 encoding gene [90]. A direct binding of Suv4-20h1 with enrichment of histones H3K20me2 and H4K20me3 at the CDKN1A promoter was shown in that study.
The histone variant H2A.Z is enriched at p53 binding sites in the CDKN1A promoter, facilitating p53 binding and p21 cip1/waf1 expression [91]. Interestingly, H2A.Z is lost from the CDKN1A promoter in response to DNA damage followed by recruitment of TIP60 histone acetyltransferase to trigger CDKN1A gene expression [92]. Thus, H2A.Z seems to be a negative regulator of p21 cip1/waf1 . The authors show that p53 plays an active role in positioning H2A.Z to gene promoters and contributes to p21 cip1/waf1 induction in an H2A.Z-depleted setting. In contrast, MYC represses CDKN1A transcription by preferentially recruiting H2A.Z at MYC binding sites in the TATA initiator region of the CDKN1A promoter to protect the malignant cells from senescence [91,92].
These findings indicate a tight and complex crosstalk between different epigenetic mechanisms in regulating p21 cip1/waf1 expression in different human cancer cells.
Mutations of isocitrate dehydrogenase (IDH) lead to the generation of 2-hydroxoglutarate (2-HG), an oncometabolite, instead of the physiologic metabolite α-ketoglutarate (α-KG), which is a co-substrate for TET and Jumonji enzymes. 2-HG acts as a competitive inhibitor to these epigenetic modifiers and, thus, increases histone lysine methylation and decreases 5-hydroxymethylcytosine levels [107], which can lead to enhanced expression of p21 cip1/waf1 [104]. Interestingly, IDH mutations are frequently observed in glioblastoma, intrahepatic cholangiocellular carcinoma, and acute myeloid leukemia, where epigenetic alterations are commonly found, and are already part of established diagnostic and therapeutic algorithms [108][109][110].
It is known that vitamin C can increase the levels of TET enzymes [111,112]. In skin cancer cells, this treatment induced a global DNA demethylation, which was associated with elevated 5hmC levels in the CDKN1A promotor and increased p21 cip1/waf1 expression. Knockout of the enzymes TET1/TET2 inhibited p21 cip1/waf1 upregulation, suggesting that TET regulates p21 cip1/waf1 expression in skin cancer cells treated with vitamin C [113].
In agreement with our previous report [9], recent studies have continued to demonstrate the regulation of p21 cip1/waf1 expression by HDAC inhibitors (HDACi) in tumors. In non-small-cell lung carcinoma, treatment with HDACi alone or in combination with the natural flavonolignan silibinin caused increased H3 and H4 acetylation in the Sp1/Sp3 binding site at the CDKN1A promoter, activating CDKN1A transcription [105]. In colorectal cancer cells, individual treatment with different HDACi, such as Trichostatin A, SAHA, and sodium phenylbutyrate (NaPB), strongly increased p21 cip1/waf1 mRNA and protein levels. Specifically, the effect of NaPB, an inhibitor of predominantly nuclear HDACs, on p21 cip1/waf1 regulation was evaluated. Treatment increased H3K27ac occupancy at the CDKN1A promotor and the enhancer region, demonstrating that p21 cip1/waf1 expression is regulated by histone acetylation [114]. In pancreatic cancer cells, SAHA induced increased expression of p21 cip1/waf1 levels as well as increased acetylation of histone H3 globally and specifically at the CDKN1A promotor [115]. The same was observed in multiple myeloma cells treated with the novel HDAC inhibitor Scriptaid [106]. Moreover, other studies have reported the modulation of p21 cip1/waf1 expression after HDAC inhibition in lymphoma [116], leukemia [117], hepatocellular carcinoma [118], and breast cancer [119]; however, histone marks were not investigated.

The Role of lncRNAs and miRNAs in the Deregulation of p21 cip1/waf1 in Cancer
Interestingly, long noncoding RNAs (lncRNA) have been shown to be involved in the regulation of p21 cip1/waf1 (Figure 3, Table 3). Overexpression of the long noncoding RNA HOXA cluster antisense RNA2 (HOXA-AS2) is associated with tumor size, TNM stage, and a poor prognosis in gastric cancer. Interestingly, this lncRNA promotes gastric cancer cell proliferation by downregulating mRNA levels of CDKN1A in vitro and in vivo via binding with enhancer of zeste homolog 2 (EZH2) [102]. The long noncoding RNA plasmacytoma variant translocation 1 (PVT1) is upregulated in pancreatic duct adenocarcinoma (PDAC) tissues and has been shown to promote pancreatic cancer cell proliferation and migration, and epithelial-to-mesenchymal transition (EMT), by downregulating p21 cip1/waf1 [120]. The HOX Antisense Intergenic RNA (HOTAIR), which is overexpressed in a wide variety of cancers, including cervical cancer, is able to silence p21 cip1/waf1 and promote cancer radio-resistance in vitro and in vivo [121]. Other long noncoding RNAs that have been shown to silence p21 cip1/waf1 during human malignancies include the antisense noncoding RNA in the INK4 locus (ANRIL), which is upregulated in non-small-cell lung cancer tissues. Through silencing p21 cip1/waf1 expression, ANRIL promotes cancer cell proliferation and inhibits apoptosis. The mechanism of silencing includes binding to EZH2, in addition to a direct binding to the p21 cip1/waf1 promotor that mediates H3K27me3 modification [122]. Recently, the lncRNA MIR31HG was found to promote cell cycle progression and proliferation in head and neck squamous cell carcinoma (HNSCC) by targeting p21 cip1/waf1 [123]. The BRAF-activated noncoding RNA (BANCR) interacts with p21 cip1/waf1 and increases its expression to induce cell cycle arrest and apoptosis in colorectal cancer [124].  (left figure)-Mechanistic p21 cip1/waf1 repression. Several microRNAs (miRNAs) and long noncoding RNAs (lncRNAs) were shown to inhibit CDKN1A transcription by directly interacting with its promotor. The expression of these ncRNAs in tumor cells led to p21 cip1/waf1 repression. (right figure)-Mechanistic p21 cip1/waf1 activation. On the other hand, some miRNAs were shown to enhance CDKN1A transcription by remodeling histones or targeting putative sites in its promotor. The expression of these miRNAs in cancer cells led to increased levels of p21 cip1/waf1 . Moreover, the BRAF-activated noncoding RNA (BANCR) interacts with p21 cip1/waf1 to increase its expression.
MicroRNAs are small noncoding RNAs that are 18-25 nucleotides long [125]. They play an important role in gene regulation at the post-transcriptional level by targeting mRNAs and inducing their degradation and/or inhibiting their translation [125,126]. Notably, miRNAs can activate the expression of specific genes by targeting their promotors [127,128]. miRNAs have been strongly implicated in tumorigenesis and their deregulation is correlated with the hallmarks of cancer [129][130][131]. miR-125a and miR-125b have been shown to downregulate p53, thereby resulting in decreased expression of p21 [131,132]. Several other miRNAs have been reported to directly downregulate p21 cip1/Waf1 by binding to its 3'-UTR, and these include miR-17, miR-20a, miR-20b, miR-93, miR-106a, and miR-106b (reviewed by Jung et al. [133]). Subsequently, upregulation of these miRNAs promotes cellular proliferation [133]. Conversely, let-7a targets Np95 ICBP90 ring finger (NIRF) and, therefore, upregulates p21 expression; however, the mechanism by which NIRF regulates p21 remains elusive [129]. Recent studies show a complicated role of miRNAs in regulating p21 cip1/waf1 in cancer cells (Figure 1, Table 3). Plasma and tissue samples obtained from lung cancer patients showed an overexpression of miR-93-5p and a reduced expression of p21 cip1/waf1 and Bcl-w, indicating a potential role of this miRNA in suppressing senescence through p21 cip1/waf1 [130]. miR-208a has also been shown to be overexpressed in the serum of lung cancer patients who were subjected to radiation therapy. A further analysis indicated that this miRNA directly targets p21 cip1/waf1 and affects the AKT/mTOR pathway to promote cell proliferation and radio-resistance. miR-208a is predicted to bind to a putative binding site at positions 86-92 in the 3'UTR of CDKN1A mRNA. Interestingly, this miRNA is secreted by lung cancer cells in response to radiation. Moreover, it was shown that serum miR-208a can be taken up by these cells via exosomes in a time-dependent manner [134]. A recent in vivo analysis of human cervical cancer tissue showed an upregulation of miR-92a, which is associated with cervical cancer progression. A complementary in vitro analysis revealed that p21 cip1/waf1 is a direct target of miR-92a, which promotes cell cycle progression through inhibiting p21 cip1/waf1 expression in cervical cancer [135]. miR-17 5p has been reported to be upregulated in multi-drug resistant human gastric cancer cell lines and its upregulation resulted in decreased expression of p21 cip1/waf1 and drug resistance [136]. Evidence from a recent study has revealed a role of miR-95-3p in cell growth of osteosarcoma through targeting of p21 cip1/waf1 . This miRNA has been also reported to be significantly upregulated in the serum of osteosarcoma patients. The downregulation of this miRNA inhibited cell growth and induced apoptosis by enhancing the activities of caspase-3 and caspase-9 and increasing the expression of Bax/Bcl-2 proteins in osteosarcoma cells. On the other hand, the overexpression of miR-95-3p promoted cell growth, while inhibiting apoptosis, by modulating TGF-β, cyclin D1, and p21 cip1/waf1 expression levels in these cells [137]. Similarly, miR-95-3p upregulation in a hepatocellular carcinoma (HCC) cell line, HCC tissues, and HCC xenograft mouse models was found to promote tumor proliferation and growth by directly targeting p21 cip1/waf1 . An in vitro analysis showed that the 15-bp region between 1500 bp and 1515 bp from the stop codon TAA at the 3 -UTR of CDKN1A contains the 3'UTR region that is targeted by this miRNA [138]. An in vitro analysis has shown that miR-503-3p induces apoptosis of lung cancer cells by downregulating p21 cip1/waf1 and CDK4 expression [139]. miR-224 promotes chemoresistance of in vitro and in vivo models of human lung adenocarcinoma through targeting p21 cip1/waf1 and thereby modulating G1/S transition and apoptosis [140].
On the other hand, miR-3619-5p has been shown to activate p21 cip1/waf1 expression by targeting a putative site in its promotor. The same study showed that reduced expression of this miRNA and p21 cip1/waf1 is associated with cancer progression and poor survival of patients with bladder carcinoma [141]. Another study showed that miR-370, miR-1180, and miR-1236 are downregulated in bladder carcinoma tissues, and their overexpression led to p21 cip1/waf1 activation and inhibition of cancer. These miRNAs have been shown to induce the nuclear and not the cytoplasmic expression of p21 by interacting directly with its promotor. As a result, CDK4/6 and cyclin D1, which are downstream gene targets of p21 cip1/waf1 , were repressed following overexpression of the three miRNAs [142]. miR-6734 upregulates p21 cip1/waf1 expression through modifying histones in the CDKN1A promoter and induces cell cycle arrest in colon cancer cells. A detailed mechanistic analysis showed that this miRNA directly interacts with the CDKN1A promotor by binding to a target site located at −360 to −260 relative to the transcriptional start site [143]. Results from a recent study revealed that miR-200c mediates sub-G1 and G2/M phase arrest and, consequently, enhances the radio-resistance of esophageal squamous cancer cells through modulating the expression levels of several cell cycle regulators, including p21 cip1/waf1 [144]. The expression level of the lncRNA growth arrest specific transcript 5 (GAS5) has been observed to be downregulated in stomach cancer and resulted in decreased p21 cip1/waf1 expression and impaired cell cycle arrest at the G1 phase. Importantly, the downregulation of GAS5 results in the downregulation of YBX1, a transactivator of p21 cip1/waf1 in prostate cancer, thereby decreasing p21 cip1/waf1 expression [145].
All in all, these findings show a biological significance for the role of lncRNAs and miRNAs in the deregulation of p21 cip1/waf1 in cancer. However, the exact mechanism by which they affect the expression levels of p21 cip1/waf1 remains unclear and requires further investigation. Nevertheless, the use of these noncoding RNAs as molecular biomarkers offers potential diagnostic and prognostic value for the better management and treatment of cancer diseases associated with deregulated p21 cip1/waf1 expression. HOXA cluster antisense RNA2 (HOXA-AS2) GC + Epigenetically silences p21 transcription and promotes cancer cell growth [102] Plasmacytoma variant translocation 1 (PVT1) PDAC + Promotes cancer cell proliferation and migration, and epithelial-to-mesenchymal transition (EMT), by downregulating p21 [120] HOX Antisense Intergenic RNA (HOTAIR) Cervix + Promotes radio-resistance by silencing p21 expression [121] antisense noncoding RNA in the INK4 locus (ANRIL) NSCLC + Promotes cancer cell proliferation and inhibits apoptosis by silencing p21 expression [122] LnCRNA MIR31HG HNSCC + Promotes cell cycle progression and proliferation by targeting p21 [123] BRAF-activated noncoding RNA (BANCR) CRC + Interacts with p21 and increases its expression to induce cell cycle arrest and apoptosis [124] miR-93-5p Lung + in plasma and tissue samples obtained from lung cancer patients Suppression of senescence through reducing the expression of p21 [130] miR-208a Lung + in serum of lung cancer patients who were subjected to radiation therapy Reduces the expression of p21 to promote cell proliferation and radio-resistance [134] miR-92a Cervix + in human cervical tissues and Hela cells Promotes cell cycle progression and cell proliferation via inhibiting p21 expression [135] miR-17-5p Multi-drug resistant human GC + Promotes drug resistance partially through downregulating p21 [136] miR-95-3p OS, HCC + in the serum of OS patients and in an HCC cell line Promotes cell proliferation and growth by targeting p21 [137,138] miR-503-3p Lung cancer + Induces apoptosis of lung cancer cells by downregulating p21 and CDK4 expression [139] miR-224 Human lung adenocarcinoma + in cisplatin (DDP; cis-diamminedichloroplatinum II)-resistant A549 cells Promotes chemo-resistance through targeting p21 and modulating G1/S transition and apoptosis [140] miR-3619-5p BC − Activates p21 expression by targeting its putative promotor. Reduced expression of this miRNA and p21 is associated with cancer progression and poor survival of patients with BC [141] miR-370, miR-1180, and miR-1236 BC -in BC tissues Activates p21 and inhibits cancer [142] miR-6734 CC − Modifies histones in the p21 promoter and induces cell cycle arrest in colon cancer cells [143] miR-200c ESCC − Mediates sub-G1 and G2/M phase arrest and, consequently, enhances radio-resistance of esophageal squamous cancer cells through modulating the expression levels of several cell cycle regulators, including p21 [144] lncRNA growth arrest specific transcript 5 (GAS5) GC − Increases p21 expression and enhances cell cycle arrest at the G1 phase [145] Note: Upregulation-(+); Downregulation-(-); GC-Gastric cancer; PDAC-pancreatic duct adenocarcinoma; NSCLC-non-small-cell lung cancer; HNSCC-head and neck squamous cell carcinoma; CRC-colorectal cancer; OS-osteosarcoma; HCC-hepatocellular carcinoma; BC-bladder cancer; CC-colon cancer; ESCC-esophageal squamous cell carcinoma.

Conclusions
In addition to the accumulation of multiple genetic alterations during the progression of cancer, it is now widely established that global epigenetic changes also drive carcinogenesis as sequential events. Aberrant epigenetic modifications lead to the expression or silencing of oncogenes and tumor suppressor genes, respectively. The universal cyclin-dependent kinase CDK inhibitor p21, which does not act as a transcription factor per se, is deregulated in many types of cancer. By showing dual antagonism, it is involved in many hallmarks of cancer. The expression of p21 is largely mediated by epigenetic modifications, such as DNA methylation, histone H3 methylation, and histone deacetylation. Importantly, the deregulation of miRNAs and lncRNAs contributes to p21 repression or activation, resulting in enhanced proliferation or growth arrest, drug response or resistance, and apoptosis or survival, respectively. Given that epigenetic modifications are reversible, it is conceivable that the epigenetic machinery, including noncoding RNAs, might be an important regulatory mechanism for p21 in different stages of cancer or during various therapies. The use of epigenetic drugs, in combination with conventional therapy, may lead to a novel approach for chemotherapy-and radiotherapy-resistant cancers, and alterations in p21 should not be underestimated.