Bridging tissue repair and epithelial carcinogenesis: epigenetic memory and field cancerization

Abstract

The epigenome coordinates spatial-temporal specific gene expression during development and in adulthood, for the maintenance of homeostasis and upon tissue repair. The upheaval of the epigenetic landscape is a key event in the onset of many pathologies including tumours, where epigenetic changes cooperate with genetic aberrations to establish the neoplastic phenotype and to drive cell plasticity during its evolution. DNA methylation, histone modifiers and readers or other chromatin components are indeed often altered in cancers, such as carcinomas that develop in epithelia. Lining the surfaces and the cavities of our body and acting as a barrier from the environment, epithelia are frequently subjected to acute or chronic tissue damages, such as mechanical injuries or inflammatory episodes. These events can activate plasticity mechanisms, with a deep impact on cells’ epigenome. Despite being very effective, tissue repair mechanisms are closely associated with tumour onset. Here we review the similarities between tissue repair and carcinogenesis, with a special focus on the epigenetic mechanisms activated by cells during repair and opted by carcinoma cells in multiple epithelia. Moreover, we discuss the recent findings on inflammatory and wound memory in epithelia and describe the epigenetic modifications that characterise them. Finally, as wound memory in epithelial cells promotes carcinogenesis, we highlight how it represents an early step for the establishment of field cancerization.

Facts

• Common epigenetic mechanisms characterise tissue repair and cancer.

• A recovered antecedent lesion predisposes epithelia to carcinomas.

• Tissue repair and cancer are linked by epigenetic memory.

Open questions

• How does epigenetic field change influence the mutational burden of the developing tumour?

• Which is the role of the microenvironment in repair to cancer transition?

• How can epigenetic drugs be helpful in the modulation of memory/ field cancerization for preventive medicine purposes?

Introduction

Although the cells of our body share the same genome, their gene expression is finely tuned during development, allowing the establishment of many cell types [1]. Additionally, in both homeostatic and pathological conditions, the same cell continuously changes and adapts its transcription to the fickle stimuli that the environment provides [2]. To understand this plasticity and to connect genotype and phenotype, Waddington proposed the term epigenetics, that refers to a process that reversibly influences gene expression without altering DNA sequence [3, 4]. Epigenetic mechanisms involve DNA methylation, histone modifications, chromatin remodelling, the effects of noncoding RNA (ncRNA) as well as RNA modifications. Generally, “writers” and “erasers” refer to enzymes that respectively deposit or remove chemical groups to or from DNA or histones; “readers” are proteins that can recognize the modified DNA or histones. These factors cooperate with transcription factors (TFs) and ncRNAs, to regulate gene expression [4–6]. Importantly, since epigenetic changes are mitotically inherited in somatic cells, the environmental effects on the epigenome can have long-term consequences [6, 7]. The chromatin state is essential to coordinate spatial and temporal specific gene expression programs in homeostasis and its deregulation is evident in a multitude of pathological processes. Recent cancer genome projects highlight epigenetic dysregulation as a common feature of most tumours and epigenetic modifiers represent new targets for cancer therapy [5, 8–10].

Epithelia are highly heterogeneous tissues that line the surfaces and the cavities of our body. They consist of sheets of cells, mono or multi-layered, that lay on the basement membrane [11]. Epithelial cells exert several functions such as absorption, secretion and protection of our body from the external environment. As barriers, epithelia face acute and chronic injuries that constantly alter cell behaviour by activating repair and plasticity mechanisms needed to re-establish the homeostasis [11–14], with deep changes in the epigenome and in the chromatin landscape [15, 16]. Tissue repair and cancer share cellular and molecular processes and have histological similarities, leading to the conclusion that “tumours are wounds that do not heal” [17–20]. A close association between chronic tissue lesion and cancer has been observed: tumours develop at the site of chronic skin wounds or untreated mouth ulcers, ulcerative colitis favours colorectal cancer, chronic gastritis is linked to cancer development, and several cases of lung metastases at sites of accidental trauma have been reported [21–28]. Importantly, mutagen exposure and aging often generate pro-tumorigenic cells that predispose epithelia to cancer. This phenomenon, known as field cancerization or field change, represents the onset of epigenetic, genetic and other molecular alterations in an epithelium that does not show morphological aberrations but it is precursor to tumours [29]. Several mechanisms can sequentially occur to build a field cancerization and promote tumorigenesis. For instance, starting from a random mutation, mutant (or epigenetically altered) clones with slightly increased fitness might outcompete wild type clones over years, thus leading to cancer [30]. There is growing evidence that injuries can contribute to the establishment of a field cancerization.

In this Review, we address the common traits between carcinogenesis and damage repair in epithelia from an epigenetic perspective, pointing out the chromatin adaptations and the epigenetic mechanisms shared by epithelial cells during these two processes.

Recent studies demonstrated that a repaired epithelium (after a mechanical injury or an inflammatory event) saves a long-lasting epigenetic memory that ensures a faster reaction to secondary assaults [31, 32]. However, the memory-related epigenetic changes, beside the beneficial effect on tissue fitness, have long-term detrimental consequences, since they predispose the previously damaged epithelia to cancer development [33–36]. Therefore, inflammatory and wound memories represent a new link between repair and tumorigenesis. Moreover, healed injuries represent an early step of field cancerization [36] that, when fed with an oncogenic stimulus that causes DNA mutations, could lead to cancer.

Tissue repair and carcinogenesis are similar processes

Back to 1986, Dvorak pointed out the histopathological similarities that exist between solid tumours and injury repair, suggesting that tumours behave as wounds that do not heal [17, 37]. Today, this claim is supported even strongly, and it has been shown that cancers, and in particular epithelial malignancies, named carcinomas, use wound healing-related mechanisms to promote their own growth. Among the common cellular features shared in wound healing and cancer there are enhanced proliferation and migration of epithelial cells, activation of fibroblasts and immune cells, deposition of large amounts of extracellular matrix (ECM), angiogenesis and lymphangiogenesis, as well as deregulation of cellular energetics [18, 20, 38]. Focusing on adult epithelial cells, epithelial to mesenchymal transition (EMT) is indeed part of the physiological response to wounding, but it is also a well-established feature of cancer cells [39]. Despite significant differences, both tumours and wound healing share the activation of cellular mechanisms that enable epithelial cells to become motile. Invasive or metastatic tumour cells, in fact, can undergo a proper EMT which has multiple molecular parallels with the mechanisms that epithelial cells opt during re-epithelialization after injury [40]. EMT occurs through the activation of intracellular signalling pathways, such as Ras/ERK/MAPK and PI3K/Akt/mTOR axes, supported by shifts in epigenetic regulation, such as the modifications of chromatin-associated histones or DNA methylation that are also relevant in wound healing [40–42]. Similarly, HIF1α and glycolysis are drivers of tumour progression and metastasis but at the same time they are vital for epithelial cell migration at the wound front [43–45]. The transcriptional regulator SOX2 has the potential to reprogram epidermal cells to increase cell migration and improve wound resolution and it is also a regulator of tumour initiating potential in squamous cell carcinoma in skin, oesophagus and lung [46–49]. Other key drivers of cancer-associated proliferation such as EGF receptor, Egr1 and AP-1 members, c-Jun and c-Fos, are transcriptionally activated within a few hours after wounding [20]. However, while in tissue repair these features are temporarily activated in a controlled manner, tumour cells maintain these signals active over time, thanks to the accumulated DNA mutations [50, 51].

Common epigenetic regulators control tissue damage and tumorigenesis

Cancer initiation, progression and metastasis are driven by the cooperation between genetic and epigenetic alterations that disrupt gene regulatory programs critical to maintain specialized cellular functions [52, 53]. Importantly, as proposed by Flavahan et al. chromatin and epigenetic aberrations have the potential to confer to cells the full range of oncogenic properties represented in the classic “hallmarks” depiction of cancer [9, 54].

On the other side, tissue repair is finely coordinated by epigenetic modifiers that promote injury resolution [15, 16, 21]. Recently, in ref. [55] authors highlighted the existence of a shared chromatin landscape between wounded epidermis and cutaneous squamous cell carcinomas (SCCs). These two processes are also regulated by common TFs. Indeed, wound induces a global chromatin opening that leads to the co-expression of both SOX9 and KLF5 in the same cell, an event known as lineage infidelity, a recently established feature of carcinomas. However, while lineage infidelity is transient during repair, cancer cells do not revert this condition, culminating in the activation of oncogenic enhancers [55]. This work highlights that the cellular and molecular similarities between cancer and tissue repair are likely to be coupled by common epigenetic processes.

In the following paragraphs we will describe the epigenetic alterations that are commonly involved in cancer and repair, as summarised in Fig. 1 and Table 1. This suggests that the plasticity and the dynamics of the two processes are controlled by similar epigenetic mechanisms to regulate gene expression.

Fig. 1. Tissue repair and carcinomas share multiple cellular phenotypes and epigenetic mechanisms.

Some of the cancer hallmarks described by Hanahan and Weinberg [54] also represent reliable hallmarks for damaged tissues [64]. For each of the shared hallmarks we reported some representative examples of chromatin mechanisms involving epigenetic enzymes that can give rise to that specific hallmark.

Table 1.

Summary of a set of key epigenetic factors with known function in tissue repair and carcinogenesis.

Gene	Modification	Function in cancer	Function in tissue damage	Refs.
MLL1	Histone methylation (H3K4me3)	Proliferation MLL1 regulates intestinal tumorigenesis and cancer stemness via Wnt/β-catenin pathway.	Inflammation In skin wound healing, MLL1 modulates macrophage-mediated response via NF-κB pathway.	[58, 59]
SETD2	Histone methylation (H3K36me3)	Proliferation; Migration; Invasion Downregulation of SETD2, a tumor suppressor in gastric mucosa, promotes cancer cell proliferation, migration and invasion. It is involved in colorectal carcinogenesis by regulating Wnt pathway.	Proliferation; Migration SETD2 loss promotes cutaneous wound healing via AKT/mTOR signalling, both in vivo and in vitro.	[60, 61]
EZH2 (PRC2)	Histone methylation (H3K27me3)	Proliferation; Migration; Invasion EZH2 promotes expansion of breast cancer initiating cells via RAF1/ERK signalling. PRC2 downregulation accelerates migration and invasion of hepatocellular carcinoma cells.	Proliferation; Migration In epidermal wound healing, EZH2 is downregulated to allow the induction of repair genes.	[10, 62–67]
PRC1	Histone ubiquitination (H2AK119ub)	Migration; Metabolism Upon skin injury, a persisting decrease of H2AK119ub can favor cSCC onset by enhancing expression of key genes for migration and glucose metabolism.	Migration PRC1 deficiency impairs epidermal integrity by affecting adhesion genes in basal layers.	[36, 68, 69]
SIRT1	Histone deacetylation	Migration; Invasion SIRT1 is overexpressed in many cancers (such as prostate, breast, ovarian, gastric and colorectal cancer) where it promotes cell migration and metastasis.	Migration; Metabolism In epidermis, SIRT1 favors proper healing by regulating keratinocyte migration, TGFβ signalling and redox response.	[10, 72]
DNMT1	DNA methylation	Migration In gastric cancer, DNMT1-dependent methylation results in reduced E-cadherin expression and enhanced cancer cell migration.	Migration; Angiogenesis Corneal epithelial cells show enhanced migration upon injury when DNMT1 is expressed. Conversely, in diabetic mice upregulated DNMT1 impairs wound healing by reducing the number of macrophages; however, its inhibition promotes angiogenesis and repair.	[79–85]
DNMT3A	DNA methylation	Proliferation; Invasion Overexpression of DNMT3A in skin and vulvar SCCs promotes invasiveness. In lung cancer, it enhances proliferation and metastasis.	Proliferation DNMT3A is involved in the modulation of key repair genes of the skin, such as genes regulating keratinocyte dedifferentiation.	[86–90]
TET2	DNA demethylation	Proliferation; Migration TET2 downregulation reported in HNSCC; TET2 overexpression inhibits cancer cell proliferation and migration.	ECM remodelling In diabetic mice TET2 impairs repair by activating MMP-9.	[91–93]
MALAT-1	Non-coding RNAs	Proliferation; Migration; Invasion MALAT-1 both promotes and inhibits cancer cell proliferation, migration and EMT in various tumours.	Proliferation; Migration Upregulation of MALAT-1 enhances healing in fibroblast culture via TGFβ2/SMAD2 pathway.	[99, 100]
lncRNA-H19	Non-coding RNAs	Migration; Invasion; Proliferation; Angiogenesis; Metabolism LncRNA-H19 acts as an oncogene in various tumours.	Inflammation; Proliferation In injured skin, lncRNA-H19 enhances healing by promoting dermal fibroblast proliferation and macrophage infiltration.	[101, 102]
miR-200c	Non-coding RNAs	Migration; Invasion miR-200c inhibits cancer cell migration and metastasis by blocking EMT in HNSCC and ovarian cancer.	Migration Wound closure is promoted by anti-miR-200c, which shows pro-migratory effect in mouse keratinocytes in vitro.	[103–105]
miR-21 and miR-31	Non-coding RNAs	Migration Migration of malignant cells in cSCC and oral carcinoma is regulated by miR-21 and miR-31, respectively.	Proliferation; Migration After skin injury miR-21 is upregulated in migrating epithelial cells. miR-31 promotes both proliferation and migration of skin epithelial cells in vivo and in vitro.	[106–110]
NSUN2	RNA methylation (m5C)	Proliferation; Migration Breast, colorectal and non-small cell lung cancer cells proliferation and migration are regulated by highly-expressed NSUN2.	Proliferation; Migration NSUN2-mediated RNA methylation promotes proliferation and migration of corneal epithelial cells during repair.	[112–114]
METTL3	RNA methylation (m6A)	Proliferation; Migration; Metabolism METTL3 promotes prostate cancer cells growth and motility via Hh pathway. It enhances breast cancer cell proliferation by inhibiting tumor suppressor let-7g. MTTL3 favors colorectal cancer via GLUT1-mTORC1 axis.	Angiogenesis In diabetic foot ulcers, METTL3 regulates lymphangiogenesis promoting tissue repair in a VEGF-C-mediated manner.	[115–118]

Histone modifications

The chemical modification of a histone protein, the core unit of a nucleosome, strongly impacts transcriptional induction or repression. These reversible covalent modifications include acetylation, methylation, phosphorylation, ubiquitination, ADP-ribosylation, citrullination and SUMOylation, to allow a wide variety of chromatin rearrangements [6, 10, 56]. So far, few studies have confirmed the role of histone phosphorylation, ubiquitination, and ADP-ribosylation in wound healing, while histone methylation and acetylation are widely described [16].

Histone methylation is regulated by histone methyltransferases (HMTs) and histone demethylases (HDMs) and usually occurs on lysine or arginine residues of the histones. Several HMTs (such as MLL1, SETD2 and EZH2) and HDMs, (such as JMJD3), have been widely studied in cancer and tissue repair. In skin, the inhibition of HDMs responsible for H3K4me3, H3K9me3 and H3K27me3 removal delays injury repair [57]. MLL1 upregulation characterises many tumour types, including colon cancers [58]. In skin, MLL1 controls macrophage-mediated inflammation and MLL1-deficient mice show impaired wound closure due to the reduction of H3K4me3 at NF-κB binding sites [59]. SETD2, responsible for the deposition of H3K36me3, behaves as a potential tumour suppressor in many solid cancers [60]. SETD2 knockdown promotes proliferation and migration in vitro and accelerated re-epithelialization through the activation of AKT/mTOR signalling [61]. EZH2 is the catalytic subunit of Polycomb repressive complex 2 (PRC2), responsible for the di- and trimethylation of H3K27 (H3K27me2 and -me3). In the intestinal epithelium and in epidermis PRC2 is required for the proliferation of progenitor cells and for the correct balance in the lineage differentiation programs. EZH2 controls proliferative potential of epithelial progenitors by repressing the Ink4A-Ink4B locus, preventing its transcription [62, 63]. However, during tissue damage, EZH2 and other PRC2 components are downregulated in epidermis, whereas the demethylase JMJD3 is upregulated. Indeed, the loss of PRC2-mediated silencing contributes to the induction of repair genes, as Myc and Egfr [64]. EZH2 is among the earliest methyltransferases to be described as a functional oncogene in multiple cancers, such as breast, prostate, and bladder, although it has been proposed as a tumour suppressor as well [10, 65]. For example, overexpression of PRC2 components is linked to breast and liver cancer progression, through the enhancement of RAF1-ERK signalling [66, 67].

Beside PRC2, the Polycomb repressive complex 1 (PRC1) catalyses the ubiquitination of lysine 119 on histone H2A. It is critical for PRC2 recruitment at target genes and H3K27me3 deposition [68]. PRC1 plays a key role in the maintenance of epithelial integrity and PRC1-null mice show skin fragility phenotype, occurring between the basal layer and the underlying basement membrane [69]. In epidermal cells, decrease of H2AK119ub occurs after injury, leading to chromatin opening and enhancing the expression of key genes for migration and glucose metabolism. Importantly, the wound-mediated downregulation is maintained long-term and it is directly responsible for enhancing SCCs onset, while the treatment with a PRC1 inhibitor increases tumour frequency when combined with chronic UV-B exposure [36].

Histone acetyltransferases (HATs) and histone deacetylases (HDACs) catalyse respectively histone acetylation and deacetylation, thus regulating the opening or the closure of chromatin [16]. During wound healing, the acetylation of epidermal histone H4 is very dynamic. The hyperacetylation of H4K12 is coupled with the massive deacetylation of histone H4K5, H4K8, and H4K16, in wound site and in wound-distant tissue [70]. In 25 human cell lines from 10 different cancer types, such as breast, colon and cervix malignancies, the reduction of H4K16 acetylation, together with the downregulation of H4K20me3 deposition (a constitutive marker of heterochromatin) represents a common cancer epigenetic signature [71]. Within the histone deacetylases (HDACs), Sirtuins play relevant roles in many cancer types. Among them, SIRT1 acts as an oncogene and it is overexpressed in many cancers (i.e. prostate, breast, ovarian, gastric and colorectal) where it promotes cell migration and metabolic adaptation, although it has been described in some tumour contexts as a tumour suppressor [10]. In skin, SIRT1 is required for an efficient re-epithelialization, by enhancing cell migration, TGF-β signalling, and oxidative stress [72].

DNA methylation

DNA modification consists in the transfer of a methyl group to the C5 position of cytosine, mainly at CpG islands, and it is catalysed by methyl-transferase enzymes (DNMTs). It controls gene expression regulating critical mechanisms such as the DNA binding of TFs [73] and maintaining transcriptional fidelity [74]. While DNMT1 is responsible for the maintenance of methylated-cytosines during replication [75], DNMT3A and DNMT3B mainly play a key role in de novo methylation [76–78]. In homeostatic epidermis, DNMT1 is enriched in epidermal progenitor cells to preserve proliferation by inhibiting differentiation, and it is required for self-renewal [79]. However, its upregulation has been observed in a variety of tumours [80]. For example, while DNMT1 could not be detected in normal gastric mucosal cells, it is highly expressed in most gastric cancer, where it correlates to a decreased E-cadherin expression and to a consequent increased migration of cancer cells [81]. In corneal epithelium, DNMT1 expression is upregulated in epithelial cells during wound healing, in vivo, resulting in a global DNA hypermethylation. Among the hypermethylated targets, miR-200a and CDKN2B were identified as key regulators of epithelial cells’ migration during skin repair, and their repression through DNMT1-mediated methylation enhances wound healing in vitro [82]. On the other hand, it has been demonstrated that, in diabetic mice, the impaired healing is caused by a DNMT1-dependent dysregulation of hematopoietic cells, which directly reduces the number of wound-associated macrophages [83]. Consistently, DNMT1 deficiency improves tissue repair by enhancing the motility of macrophages [84]. Besides, it has been demonstrated by an in vivo study that the inhibition of DNMT1 promotes angiogenesis and accelerates diabetic wound healing by regulating the Ang-1/NF-κB signalling pathway [85]. Overall, these studies indicate that the reduction of DNMT1 facilitates wound healing. Also, de novo methyltransferases influence tissue repair and malignancies. Indeed, a decrease in mechanical tension within the epidermis caused by an injury leads to the nuclear translocation of DNMT3A. This event causes major alterations in gene expression by the hypermethylation of the promoters of key genes that are known to be downregulated in response to wound [86]. The effect of aberrant DNMT3A characterises many tumours. Its overexpression in skin and vulvar squamous cell carcinomas, as well as in lung cancer, promotes tumour onset and is associated with malignant characteristics such as high invasiveness and tumour recurrence [87–89]. However, in sharp contrast, it has been shown that the lack of DNMT3A facilitates lung cancer initiation and progression [90].

Beside DNMTs, the ten-eleven translocation (TET) family is responsible for DNA demethylation. TET enzymes oxidize 5-mC to 5-hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC) and 5-carboxycytosine (5-caC) [16]. Among the TET enzymes, especially TET2 is frequently mutated in various cancers, such lung and colon adenocarcinomas [91]. For example, reduced levels of TET2 characterise a large fraction of head and neck squamous cell carcinomas, correlating with tumor size and worse prognosis. Importantly, by significantly forcing TET2 expression, cancer cell proliferation is strongly inhibited as well as migration, while chemosensitivity is enhanced [92]. An opposite role was described in a diabetic mouse model, where wound healing is impaired by excessive activation of matrix metalloproteinase-9 (MMP-9), through the TET2-dependent demethylation of its promoter [93].

The genomic location of the hypermethylation (global or focally restricted to specific promoters) has functional consequences in tumours [94], and this feature might be relevant also in wound healing. However, the lack of DNA methylation omics in wound healing strongly limits these considerations.

Non-coding RNAs

An additional layer of epigenetic regulators involves ncRNAs including short regulatory non-coding RNAs, such as piwi-RNAs or microRNAs (miRNAs), and long non-protein coding RNAs (lncRNAs) [95]. Recent works highlighted that several miRNAs and lncRNAs regulate multiple aspects of tissue repair [96, 97] and cancerogenesis [98]. The metastasis-associated lung adenocarcinoma transcript 1, MALAT1, is one of the most abundant lnc-RNAs in normal epithelia. MALAT1 is a prognostic marker in lung, pancreatic, breast cancers and others, and it is involved in tumour formation and metastasis [99]. However, a recent finding correlates MALAT1 overexpression to the enhancement of TGFβ signalling, leading to an accelerated injury repair [100]. LncRNA-H19, a novel oncogene in several cancers, can promote tissue repair by increasing macrophage infiltration during healing [101, 102]. In head and neck SCCs and ovarian cancer, miR-200c limits cancer cell invasion and metastasis, through the repression of EMT [103, 104]. In accordance with these findings, the in vitro regulation of miR-200c expression in both human and mouse keratinocytes revealed inhibitory effects of miR-200c on migration, while anti-miR-200c was associated with enhanced wound closure [105]. An opposite function is instead exerted by miR-21 and miR-31. These miRNAs are upregulated in cutaneous and oral SCCs respectively and enhance cancer cell migration [106–108]. The upregulation of miR-21 and miR-31 is also relevant in the repair process: they promote injury closure through the modulation of several regulators of epithelial repair [109, 110].

RNA modifications

Epigenetic regulation is also accomplished by numerous chemical modifications of RNA molecules, including mRNAs, tRNAs and non-coding RNAs. The most widely studied modifications consist in the methylation of adenine and cytosine residues, such as m⁶A, m⁵C and m¹A. These chemical groups are reversibly deposited onto, and erased from, RNA molecules by sets of writer and eraser enzymes, which in combination with readers finely orchestrate RNA fate and gene expression. RNA modifications have been reported to play essential roles in cellular physiology, regulating processes such as cell death, proliferation, senescence, differentiation, migration, metabolism, autophagy and DNA damage response [111]. As epitranscriptomic markers of the interaction between epithelia and the external environment, RNA modifications are known to take part in the complex process of tissue repair. For example, it has been shown that RNA methylase NSUN2, that catalyzes m⁵C, enhances corneal epithelial repair upon damage by promoting human corneal epithelial cells proliferation and migration in vitro [112]. Notably, NSUN2 is highly expressed in various tumours like breast, colorectal and non-small cell lung cancer, in which RNA methylation was observed to boost proliferation and migration of tumor cells [113, 114]. N⁶-methyladenosine (m⁶A) is also reported to modulate both wound healing and cancer development. Methyltransferase-like 3 (METTL3), a well-known writer of this RNA modification, was shown to regulate lymphangiogenesis in diabetic foot ulcers, thus promoting tissue repair [115]. However, other studies revealed that METTL3 can favor prostate [116] and breast [117] cancer by enhancing tumor cell proliferation, and colorectal cancer by boosting glucose metabolism [118].

An increasing number of mRNA modifications have been discovered, including the N4-acetylcytidine (ac4C), deposited by NAT10 and able to regulate mRNA stability [119]. Several reports show that NAT10 positively regulates EMT in multiple epithelial cancers [120] and that it promotes metastasis sustaining NOTCH3 mRNA stability in an ac4C-dependent manner [121]. NAT10 has also an important function in wound healing as it promotes cutaneous re-epithelialization through a positive regulation of epithelial cell migration [122], furtherly highlighting the molecular similarities between tumour EMT and the migratory phenotype acquired by epithelial cells during tissue repair.

Tissue damage directly promotes tumorigenesis

Several observations demonstrated that tissue damage and cancer are strongly associated, and chronically injured sites frequently develop malignancies [23–27, 123]. For example, in skin, the association between chronic injury and SCCs is remarkably clear in epidermolysis bullosa patients that suffer from chronic skin blistering: almost all patients develop a malignant SCC in early adulthood at sites of antecedent lesions [124]. In the previous paragraphs we discussed the cellular and epigenetic bases of cancer and wound repair, with particular focus on the mechanisms that are shared by the two processes. However, these molecular and epigenetic similarities only offer correlative evidence of the association between carcinogens and damage healing. Recently, using genetic and non-genetic in vivo mouse models, it has been demonstrated that injury functionally enhances the risk to develop cancer through multiple mechanisms. In skin, wound accelerates the initiation, the frequency and the growth of basal cell carcinomas (BCCs) [125, 126]. The misregulation of Hedgehog (Hh) pathway is commonly associated with BCCs in humans and its genetic induction induces the formation of BCC-like lesions in mice [126]. Importantly, a wound creates an environment that accelerates the initiation frequency and the growth of BCC-like lesion, by recruiting hair follicle (HF) cells to the wound site, where the Hh signaling is represseds [126]. These two works prove that an injury can directly promote the development of superficial BCC-like tumors in vivo [125, 126]. A new direct link between wound and tumor in epidermis has been demonstrated in non-genetic SCCs models. The presence of a previous and already resolved injury is able to accelerate SCCs onset, increasing the frequency and the malignancy of the lesions in mice chronically treated with UV-B. Indeed, wound predisposes an epigenetic landscape reminiscent of the one that characterizes the carcinoma lesions, representing a first step toward carcinogenesis [36]. The connections between tissue damage ad carcinoma have been described for other epithelia. Intestinal epithelium is often subjected to frequent injuries, and it manifests robust regenerative capabilities [127]. Injury induces a spatiotemporal reprogramming of intestinal cells with histological alterations typical of metaplasia [127]. A paradigm of damage-associated carcinogenesis is pancreatic ductal adenocarcinoma (PDAC), characterised by the oncogenic KRAS in virtually all the patients. Injured pancreatic acinar cells are subjected to a wide chromatin opening, that is then transduced by the BET family member BRD4, that binds to acetylated histones. This large-scale epigenetic remodelling dictates early neoplastic commitment [35]. In general, injury-triggered epigenetic alterations, in combination with KRAS mutation and tissue damage, promote the formation of early neoplastic lesions and, ultimately, adenocarcinomas [33–35]. Hepatocellular carcinoma frequently occurs in patients with chronic liver injury [128]. Chemically induced chronic liver lesions dramatically increase the penetrance of carcinomas and significantly alter driver mutation profiles, demonstrating that a previous injury can predispose a cancer-like environment [129].

Epigenetic adaptations after tissue repair establish a memory in epithelial cells

In both mammals and plants, cells can acquire memory of a previous stress in order to react more quickly and intensely upon a second encounter with the noxious input [31, 130–132]. In mammals, this adaptive behaviour was first described for innate immune cells and named trained immunity [133]. However, these cell adaptations are not restricted to immune cells, but have been reported also in epithelial cells of the skin [36, 134–136], lung [137], intestine [138] and pancreas [33–35, 139]. Trained immunity in immune cells is well characterised, in term of establishment and maintenance. Important epigenetic regulators of this memory are the acquisition of H3K27 acetylation (H3K27ac) on gene enhancers, the increase of H3K4 mono-methylation (H3K4me1) at the promoters of stimulated genes and DNA hypomethylation [133]. On the other hand, only few molecular mechanisms controlling the memory of inflammatory events, such as an infection or a wound, in epithelia have been revealed. As immune cells, epithelial cells establish a memory that is kept at epigenetic level, involving changes in chromatin accessibility and epigenetic pattern, such as histone post-translational modifications, as described in Fig. 2 [36, 135]. In epidermis, an inflammatory assault induces histone modifications, such as H3K27ac and H3K4me1 deposition, as in the immune system [135]. An additional mechanism of epigenetic adaptation observed after wound healing is the reduction of H2AK119ub [36]. The epigenetic changes are accompanied by the binding of a stimulus-specific TF, for example STAT3, and a broad stress factor such as AP-1. The discussed histone modification changes establish the memory and are partially maintained after the resolution of the damage, when a new homeostasis is set. Overall, the emerging general mechanism involves the induction of a more accessible chromatin state that is permissive for a more rapid re-induction of memory gene transcription in case of an eventual future assault [135]. In general, as a consequence of these adaptive mechanisms, gene expression can be readily activated in response to secondary stimuli, thus enhancing tissue repair [31, 32]. Like for the immune system, distinct adaptation programs have been observed in epithelial cells acquiring memory, including trained immunity-like memory and priming (Fig. 2). In trained immunity-like memory epigenetic perturbations persist over time, when transcription of stress-induced genes is turned off; on the other hand primed cells do not restore basal levels of transcription after the injurious stimulus ceases [130]. Overall, epigenetic reprogramming confers higher plasticity to the memory cells, thus enhancing their ability to adapt to environmental stress and increasing their fitness in case of similar damages [140].

Fig. 2. Epigenetic mechanisms of wound and inflammatory memories.

The adaptive properties of epithelial cells are the result of an epigenetic rewiring. Two epigenetic mechanisms have been described for skin epidermal cells [36, 134]. Trained immunity-like memory, once acquired is maintained at the chromatin level while transcriptionally dormant [36, 135]. Priming on the other hand, represents a pre-activated state of transcription that does not turn off even when the stimulus stops [36].

Epigenetic memory directly connects tissue damage and cancer

As we discussed above, chronic injuries and their associated inflammation constitute well-known risk factors for tumour initiation [23, 24]. However, the new paradigm of memory demonstrated that the memory of a single tissue damage can be established in cells at chromatin level, with profound long-term effects (Fig. 2).

Epigenetic memory in epithelial cells is a long-lasting event. In pancreas, memory was observed until 18 weeks after damage [33, 34]. Similarly, intestinal cells display a 6-week-lasting memory, while in skin memory cells were detected even 40 weeks after the injury occurred [36, 138]. The long-term accumulation of such memory-related modifications is beneficial in terms of tissue fitness, but it may concurrently be maladaptive and lead to the development of pathological conditions, such as cancer [141]. Using in vivo and in vitro models, it has been demonstrated that, in alveolar cells, the expression of oncogenic KRAS activate AP-1 family TFs to recruit nucleosome-remodelling factors and histone modifiers, culminating in the deposit of H3K4me1 at enhancers and H3K4me3 at promoters [142]. The fact that oncogenic KRAS can trigger a chromatin remodelling that is reminiscent of memory, suggests the possible detrimental consequences of memory. Indeed, several recent works have functionally linked epigenetic memory of a tissue damage and tumorigenesis. Memory of an epithelial damage is crucial to protect pancreas from following injuries, but it is associated with an activated acinar-to-ductal metaplasia (ADM), in vivo. However, when the ADM co-occurs with the expression of oncogenic KRAS, the memory pushes the cells towards a malignant transformation [33–35]. In mice, the epigenetic memory offers to intestinal cells a protection from subsequent gut infections but mediates an increased inflammation in the context of colitis, a well described risk-factor for cancer development [138]. In epidermis too, the beneficial implications of memory are coupled with maladaptive effects. Indeed, in vivo, the epigenetic signature of memory is maintained at least for 10 months, as reported above. During this long period, a concomitant chronic exposure to UV-B increases the frequency and the malignancy of SCCs [36]. Therefore, the epigenetic adaptations that constitute the memory of a damage are not just reminiscent of cancer epigenetic alterations but they directly promote tumour onset.

Epigenetic memory as a field cancerization event

Cancer is a complex, multifactorial genetic and epigenetic disease. Particularly, the chromatin landscape of the cells of origin and of the pre-malignant niche is known to impact significantly on tumor initiation, progression and phenotype [143–145]. Epigenetic, genetic and tissue alterations accumulating in epithelial regions have been shown to potentially increase predisposition to tumor growth, a phenomenon named field cancerization (Fig. 3A). It was first reported in a seminal work on oral carcinoma by Slaughter et al. in 1953, where it was described as the preconditioning of an area in an epithelium becoming more susceptible to carcinogenesis without showing any morphological change [146, 147]. Several decades of research in cancer biology have led to a deeper understanding of the cellular and molecular features characterizing a cancerized field. In particular, tissues accumulate both genetic and epigenetic alterations which favor positive selection for mutated clones, thus allowing for their expansion into patches of potentially malignant cells [29, 148]. Although genetic lesions have been shown to contribute prominently to the formation of cancerized fields, the role of epigenetic field effect has recently gained greater attention [149]. Its influence has become evident in multiple human cancers, such as neoplasms of the vulva, head and neck, cervix, breast, colon, skin, bladder, lung, esophagus, stomach and prostate [144, 150]. Specifically, tumors associated with chronic inflammation have been revealed to carry aberrations in DNA methylation levels as critical molecular defects [151]. For example, genome-wide alterations in DNA methylation have been indicated to contribute to urothelial and gastric carcinogenesis [152–154]. Additionally, common epimutations of DNA repair genes, such as MGMT and MLH1, found in colon cancer, have been reported to be shared with their related field defects [155]. However, the epigenetic molecular mechanisms driving normal cells towards field cancerization remain mostly unclear [151, 153, 156]. New insights into the induction of such epigenetic field changes are currently being provided, suggesting a role for inflammatory molecules. For instance, it has been shown that IL1β, IFNG and NOS2 expression in chronically inflamed colonic epithelium correlates with altered DNA methylation in a mouse colitis model [151]. In addition, IL6 has been pointed out as a crucial modifier of tumor microenvironment in head and neck cancers, not only affecting epithelial cells malignant transformation but also leading surrounding fibroblasts towards a tumor-promoting role [157]. Importantly, also memory is mediated by IL1β in skin epithelial stem cells of a wound healing mouse model and by IL6 in pancreas and intestine, suggesting a similarity between memory of a tissue damage and field cancerization [33, 134, 138, 140]. Indeed, a direct link between memory and field cancerization has recently been described in epidermis. Following epithelial injury, epigenetic modifications such as a decrease in PRC1-dependent H2AK119ub, a histone repressive mark, and alterations in chromatin accessibility accumulate in a wound-primed population of HF stem cells. The reduction of H2AK119ub was shown to enhance cutaneous SCCs formation also distally from the lesion site [36].

Fig. 3. Wound memory represents an early epigenetic field cancerization.

A Field cancerization, also named field change, represents the emergence of large tissue patches within an epithelium that look morphologically normal while being characterised by genetic or epigenetic alterations that will eventually lead to cancer [169]. B Cell states path towards epithelial cancer. The transition from a “healthy” to a neoplastic cellular state can pass through an initial accumulation of epigenetic alterations, followed by DNA mutations. The arrows indicate the direction of the path. The dashed arrow between ‘Adult epithelium’ and ‘Memory/early field cancerization’ suggests the possibility of a therapeutic intervention that might be able to revert the early steps of cancerized cells toward cancer.

Thus, the epigenetic adaptations that occur after tissue repair are stored long-term in epithelial cells, and this is described as epigenetic memory. This memory represents an epigenetic field cancerization event that predisposes tissues to tumour when combined with a chronic carcinogens exposure (Fig. 3B).

Conclusions and perspectives

In barrier tissues, adult epithelial cells can acquire plasticity, that allows them to move quickly from the maintenance of the homeostatic self-renewal to become activated and restore homeostasis after inflammation or tissue damages [12]. However, the plasticity mechanisms that are transiently activated during tissue regeneration following a damage could be opted by cancer cells for initiating a tumour [158]. There is growing evidence in several epithelia that antecedent tissue lesions predispose to tumour formation. This can happen both if the carcinogenic stimulus is triggered during the acute healing phase [126] but also if it occurs during a newly established homeostasis phase, if an epigenetic memory has been established [33, 36]. Nevertheless, the underlying molecular mechanisms only started to be elucidated.

Epigenetic factors are key regulators of epithelial plasticity and characterise both tissue repair and cancer onset. Importantly, epithelial cells are able to establish a memory of an injury or an inflammatory insult. This memory is stored long-term in the epigenome of affected cells, and it promotes a faster reaction in case of a second assault, like for the immune system [31]. However, the beneficial effects are coupled with maladaptive consequences: pre-damaged tissues are more prone to develop carcinomas [33, 36].

The epigenome of the cells before their malignant transformation has a key impact on the phenotype and the prognosis of the tumour that will derive [159]. Indeed, although the genetic lesions have been shown to prominently contribute to the formation of tumours, the presence of an epigenetic cancerization before tumour onset is relevant in multiple epithelial cancers, such as oral cancer [149], gastrointestinal tumours [153], and squamous cell carcinoma [36]. Since memory primes epithelia to tumour onset, this epigenetic memory represents an early step of field cancerization.

With these premises, the understanding of the injury-induced epigenetic environment that primes the epithelial cells to tumour onset has key importance in oncological research, to better understand cancer initiation. Moreover, the epigenetic alterations that follow an injury might also influence the mutation burden of the developing tumour [129, 142]. Therefore, the use of new in vivo models [36] of epigenetic field cancerization represents a novel perspective to understand tumor onset, including pre-cancerous lesions.

Epigenetics enables us to investigate the mechanisms underlying cancer phenotypes and evolution, providing potential targets for therapy. Taking advantage of the reversibility of epigenetic marks, epigenome-targeted therapy is a promising strategy for the treatment of cancer, but also a new option in preventive medicine [8]. The emerging role of epigenetics in the establishment of the injury memory and its implications on field cancerization and pre-tumoral lesions might suggest new therapeutic targets for the preventive treatment of the pre-cancerous lesions [160].

Epithelial cells are surrounded by other cell types that strongly influence their behaviour. Thanks to the single cell approaches and spatial profiling [161–163], we can dissect the key functional cell crosstalk. In both tissue repair and tumorigenesis, tissue microenvironment plays a pivotal role, influencing disease onset, progression, and outcome [164]. Indeed, cell-cell interactions [165, 166] are likely to play a key role as fibroblasts, immune cells, but also microbiome have been demonstrated to influence wound-induced skin cancer [167–169]. In future, a deeper characterization of the effects of injuries on the microenvironment (i.e., possible epigenetic memories in non-epithelial cells or extracellular matrix memory) is needed to further understand and prevent tumour initiation.

Author contributions

CLL and GD conceptualized the idea of the review. CLL, LE, and GD prepared the initial drafts of the manuscript. CD, GP, and VP contributed to the writing and manuscript improvement. CLL, LE, and GD contributed to the creation of figures and tables. All authors reviewed the manuscript and approved to the final version of this manuscript.

Competing interests

The authors declare no competing interests.