RNA methylation and cellular response to oxidative stress-promoting anticancer agents

ABSTRACT

Disruption of the complex network that regulates redox homeostasis often underlies resistant phenotypes, which hinder effective and long-lasting cancer eradication. In addition, the RNA methylome-dependent control of gene expression also critically affects traits of cellular resistance to anti-cancer agents. However, few investigations aimed at establishing whether the epitranscriptome-directed adaptations underlying acquired and/or innate resistance traits in cancer could be implemented through the involvement of redox-dependent or -responsive signaling pathways. This is unexpected mainly because: i) the effectiveness of many anti-cancer approaches relies on their capacity to promote oxidative stress (OS); ii) altered redox milieu and reprogramming of mitochondrial function have been acknowledged as critical mediators of the RNA methylome-mediated response to OS. Here we summarize the current state of understanding on this topic, as well as we offer new perspectives that might lead to original approaches and strategies to delay or prevent the problem of refractory cancer and tumor recurrence.

KEYWORDS:

• Cellular stress response
• epitranscriptome
• redox homeostasis

Introduction

Cancer is undoubtedly a primary cause of death in the world, with approx. 10 million people deceasing from cancer every year, and 19 million new cases in 2020 [1]. Despite the efforts made by governments, law makers, and public health authorities, cancer prevention is often achievable only in part, thus leaving in the hands of physicians and oncologists, most of the responsibility for the patient’s treatment and recovery. In this context, the extraordinary efforts of generations of cancer researchers led to significant improvements in handling neoplasms and malignancies, with remarkable advances in surgical management, radio-, chemo- and immune-therapy, which are widely credited as the few effective methods that are able to delay or counteract the spreading of tumors and cancers [2–4]. Regrettably, resistance to therapy, both innate and acquired, still represents a critical complication for successful cancer patients. This phenomenon is multifactorial, but altered redox homeostatic capacity, genetic mutations, and the disruption of epigenetic regulatory mechanisms seem to be three constants in this multitude of factors [5–7]. Only in the last few decades, with the advent of advanced techniques for accurately studying a number of biologically relevant modifications occurring in RNA molecules, the role of epitranscriptomic diversity has gained a primary position on cancer researchers’ radar, especially in terms of methyl-RNA-mediated response to therapy through redox-dependent pathways. This review article aims at summarizing the current state of understanding on this topic, while attempting to provide new perspectives that might help future research to offer answers to still open questions. From a search methodology standpoint, Scopus and Web of Science were considered as the major databases for literature screening, with no publication period limits. Regarding the most specific part of the review, the process for identifying and critically appraising relevant research was conducted after using the following keywords “RNA methylation”, “methylome”, “oxidative stress”, “cancer”, “resistance”, and combinations of them.

RNA methylation and other post-transcriptional modifications

Beyond the well-known epigenetics mechanisms that control gene expression without changes to the DNA nucleotide sequence [8], over the past decades many studies have discovered that more than a hundred different functional post-transcriptional modifications (PTMs) are able to alter the chemical structure of RNA in vivo [9]. PTMs in RNA are considered a powerful means by which cells can modify the functions and fate of RNA in vivo, and such covalent chemical modifications represent critical regulatory events that are involved in both physiological and pathological conditions [10]. In particular, RNA methylation is able to significantly affect both the function and fate of virtually all the classes of RNA. For example, PTMs can alter the stability of transcripts, which is emerging as a key mechanism through which cancer cell biology can adapt to external stimuli, thus relevantly modifying the responsiveness to therapies [11].

The class of RNA that experiences the largest set of PTMs is arguably tRNA, in which about one nucleoside out of ten, on average, is chemically modified [12]. A typical modification, which is present on 67% of the 34 tRNAs, is the methylation at N(1) of adenosine 58. This type of chemical modification is phylogenetically conserved, occurring in bacterial, archaeal, and eukaryotic tRNAs [13]. N1-methylation and de-methylation of A58 is managed by tRNA methyltransferase 6 non-catalytic subunit (TRMT6) and tRNA methyltransferase 61A (TRMT61A) methylases and AlkB homolog 1, histone H2A dioxygenase (ALKBH1) demethylase, and this PTM seems to serve as a crucial determinant of tRNA stability, being its absence a cause of premature tRNA degradation [13]. Another very common modification that is often encountered in tRNAs is the isomerization of uridine to pseudouridine (ψ), which is frequently observed even in the tRNA anticodon region (also known as, nodoc), along with another non-ACUG nucleoside inosine (I), which is often detected in the wobble residue of the nodoc. Other common PTMs of tRNAs are methylations, acetylations, carbonylations and other chemical modifications that form: 2′-O-methyladenosine (Am), 2′-O-methylcytidine (Cm), 1-methylguanosine (m¹G), 2-methylguanosine (m²G), 4-acetylcytidine (ac⁴C), dihydrouridine (D), 2′-O-methylguanosine (Gm), N2,N2-dimethylguanosine (m₂²G), 3-methylcytidine (m³C), 5-methylcytidine (m⁵C), 5-methoxycarbonylmethyluridine (mcm⁵U), 5-methoxycarbonylmethyl-2-thiouridine (mcm⁵s²U), 5-carbamoylmethyluridine (ncm⁵U), 5-carbamoylmethyl-2′-O-methyluridine (ncm⁵Um), 1-methylinosine (m¹I), N6-isopentenyl adenosine (i⁶A), wybutosine (yW), N6-threonylcarbamoyladenosine (t⁶A), 2′-O-methyluridine (Um), 7-methylguanosine (m⁷G), ribothymidine (rT), and 2′-O-ribosyladenosine (phosphate) (Ar(p)) [12]. Of course, nature rarely wastes, and most tRNA modifications have been demonstrated to play some role in cell growth and survival, especially when the modified residues are missing in or in close proximity to the nodoc loop [14–17].

rRNAs are also heavily modified during their synthesis, both by small nucleolar RNA (snoRNA)-dependent and snoRNA-independent enzymes [18]. It is interesting to note that rRNA modifications cluster in and around functionally critical ribosomal sites, such as the decoding (small ribosomal subunit) and peptidyl-transferase (large ribosomal subunit) centers [19,20], where their modifications seem to be crucial to ensure accurate and efficient protein synthesis, thus allowing a fine-tuned management of gene expression, also in response to external stimuli [18]. It is also usually recognized that rRNA modifications serve as major determinants of secondary and tertiary rRNA structures, by mediating folding and increasing base stacking [18,20]. The most common rRNA modifications are 2”O-methylations on the ribose moiety (with no preference for a specific nucleotide) and the U-ψ isomerization, however, other modifications, such as m¹acp3ψ, ac⁴C, m¹A, m⁵C, m⁷G, ac⁴C, m²⁶A, and m³U were identified, for a total of almost 100 modifications found on human rRNA [21–24]. Studies on yeast mutants for enzymes that are responsible for such rRNA modifications revealed that 2”-O-methylations and pseudouridylations are fundamental, but while other modifications seem less important when taken singularly, the cumulative removal of multiple modifications leads to a strong growth impairment, thus indicating that all modifications are important for ribosomal function, but some redundancy exists [25–29].

Messenger RNAs also undergo several enzymatical PTMs during its maturation process from pre-mRNA to mature transcript [30]. More in detail, a common mRNA modification that is found in all mature mRNAs is the so-called 5’ capping, which consists in a very peculiar m⁷G methylated guanosine bound to the first nucleotide of the RNA through a reverse 5′ to 5′ triphosphate linkage. The classical cap structure also requires the first two nucleotides in the transcript to be 2’O-methylated [31]. Other common nucleotide modifications in mRNAs include added nitro groups and enzymatically-modified ribose moiety [13], but the most common modifications in mRNAs are undoubtedly methylations and U-ψ isomerization [32–34]. The latter has been proven to respond to heat shock and nutrient deprivation, thus suggesting that its occurrence is not random but may have a specific biological function, especially upon environmental changes [13]. However, while pseudouridine and other PTMs may be biologically relevant in specific circumstances, methylation represents the most prevalent internal chemical modification occurring in eukaryotic mRNA [35]. Although this was revealed almost half a century ago, thanks to the pioneering work of Desrosiers and colleagues [36], we had to wait much longer to understand to a greater extent the biological meaning of such modifications. Today, we know that mRNA methylations are common, reversible, and allow for fine-tuning control of gene expression [37]. Different types of mRNA methylation can be found in eukaryotes, both on the nitrogenous base moiety and on the ribose moiety [38]. Such modifications include N6- and N1-methyladenosine (m⁶A and m¹A, respectively), 5-methylcytidine (m⁵C), and ribose 2′-O-methylation (2′OMe, also known as Nm). Such methylations are reversible and regulated by the so-called “writers”, that include the METTL3-METTL14 methyltransferase complex [39], as well as by the “erasers”, that include the fat mass and obesity-associated protein (FTO) and the alpha-ketoglutarate-dependent dioxygenase AlkB Homolog 5 (ALKBH5) [39,40]. Importantly, the m⁶A RNA modification is enriched at the 3’UTR of thousands of transcripts and is thought to downregulate translation efficiency via the “reader” protein YTH N6-methyladenosine RNA Binding Protein 2 (YTHDF2) [41], which recruits the CCR4-NOT de-adenylase complex to the methyl-modified mRNAs, thus ultimately reducing their half-life [42]. Intriguingly, the location of m⁶A seems to be essential for the overall effect of methylation, since it has been suggested that if N6-methylated adenosines are abundant in the 5’UTR, the opposite effect is observed, with a cap-independent induction of translation mediated by eIF3 [43].

N1-methyladenosines (m¹As) are quite rare and not as finely characterized as m⁶A, partially due to the technical difficulties arising from their spontaneous isomerization to m⁶A [44]. Nevertheless, since this modification bears a positive electrostatic charge, it has been proposed as a factor that influences the secondary structure in the 5’UTR, where it more likely tends to localize, thus altering translation [45].

The biological role of m⁵C RNA methylation is still highly debated. Some of the methyltransferases involved in their generation were identified as the DNA(cytosine-5)-methyltransferase 2 (DNMT2) and the NOP2/Sun RNA methyltransferase 2 (NSUN2) [46]. Intriguingly, an impaired regulation of m⁵C RNA levels has an impact on the structural stability of the 60S ribosomal subunits [47]. In addition, experimental evidence indicated that loss of m⁵C2278 modification led to the recruitment to polysomes of mRNAs specifically linked to the oxidative stress-induced response [48], thus pointing to a possible role of m⁵C modification in the response to stressors.

The situation is even less defined for 2′OMe modifications, which have the very peculiar feature to localize on the ribose moiety of nucleotides rather than on the nucleobase. So far, little is known about the biological relevance of such modification, that was identified at least in abundant RNA species [49]. Ribosomal RNA 2′OMe was shown to be a key determinant of ribosomal heterogeneity [21], a concept that claims that not all ribosomes are equivalent in cells, with some ribosomes being regulated in terms of translation capacity and efficiencies (ribosomal plasticity) [50]. In particular, 2′OMe patterns in rRNA modulate their ability to interact with internal ribosome entry sites (IRES), as demonstrated in conditions in which U-ψ isomerization was impaired in rRNAs [51]. As a confirmation of the importance of 2′OMe-based PTM in eukaryotic cells, the inhibition of the molecular machinery responsible for such modification resulted in strongly impaired translation [26].

As mentioned above, m⁶A is the most conserved and prevalent internal chemical modification in eukaryotic RNA molecules [52], and next-generation sequencing (NGS) demonstrated that each eukaryotic mRNA has 3–5 m⁶A nucleoside modifications [53,54], thus implying that the role played by RNA methylome might be functionally relevant to cell biology. Today, the metabolism of m⁶A RNA is believed to be involved in developmental processes of vertebrates, especially in the hematopoietic milieu [55,56]. In addition, the deletion of the m⁶A “reader” protein YTHDF2 was found to promote not only the expansion of human hematopoietic stem cells (HSCs), but also their regenerative capacity upon stress conditions [57]. In addition, the repression of METTL3 in mouse fetal liver was reported to promote the generation of double-stranded RNAs (dsRNAs), whose downstream signaling via melanoma differentiation-associated protein 5- retinoic acid-inducible gene I (MDA5-RIG-I), protein kinase RNA-activated-eukaryotic initiation factor 2α (PKR-eIF2α), and 2”−5”-oligoadenylate synthetase (OAS)-RNase L eventually resulted in hematopoietic development failure [58]. Interestingly, higher m⁶A levels were observed in the central nervous system (CNS), compared to other systems and organs. Moreover, m⁶A levels increase as the embryonic brain develops, and this highly suggests that RNA methylome may play an important role during CNS development [52]. Importantly, upon hypoxic stress the downregulation of the m⁶A “eraser” ALKBH5 enhanced differentiation in postnatal mouse cerebellum via imbalanced RNA m⁶A methylation in several messenger RNAs involved in cell fate determination [59]. In addition, the loss of the fat mass- and obesity-associated (FTO) “eraser” protein resulted in a reduction of neuronal differentiation in mouse neural stem cells via disruption of the brain-derived neurotrophic factor (BDNF)-dependent signaling, leading to weakened cognitive function [60]. Other researchers confirmed that the correct development of neuronal components is crucially regulated by RNA methylome [61], and others observed that neuron regeneration is critically governed by m⁶A RNA epitranscriptome [62], thus supporting the notion of an essential participation of m⁶A RNAs in the regulation of critical neuronal physiological and stress-induced processes.

Another class of RNA that is importantly involved in regulating gene expression is that of miRNAs. Although miRNAs are already considered as crucial epigenetic regulators, it has been discovered that miRNA methylation controls their maturation. It has been demonstrated that miRNA processing is repressed by an atypical dimethylation of the 5’-P group, which is able to block the RNase III endonuclease DICER-mediated miRNA processing in vitro and in vivo, by a mechanism that is dependent on the Bicoid interacting 3 domain (BCDIN3D)-containing RNA methyltransferase [63]. Other miRNA modifications were reported to include m⁵C, which is also demonstrated to be biologically active [64]. Indeed, m⁵C is an RNA chemical modification that impairs the hybridization process in the miRNA-mRNA duplex, thus preventing miRNA/RISC-mediated silencing [64]. m⁶A is also represented in miRNAs and this chemical modification plays an important role for their processing [65,66]. Heterogeneous nuclear ribonucleoprotein A2/B1 (HNRNPA2B1) was reported to recognize pri-miRNA m⁶A methylation, and to mediate its interaction with DiGeorge syndrome critical region 8 protein (DGCR8) to promote miRNA processing, which conversely was impaired when m⁶A was lost in pri-miRNAs [65,67]. Furthermore, an opposite methylation-dependent mechanism that represses DICER-mediated pre-miRNA cleavage into mature miRNAs was also identified. In fact, BCDIN3D, which O-methylates the terminal 5′-mono-phosphate on pre-miRNAs, negatively regulates its binding with DICER, thus blocking its maturation and affecting gene expression, at least in specific miRNAs [63].

Other classes of RNAs have also been found to be methylated, such as circRNAs (m⁶A) [68,69], piRNA (2’OMe) [70], snRNAs (m⁶Am) [71], and this highlights the fact that this conserved and maybe underestimated mechanism may be of utmost importance in controlling gene expression within cells via multiple pathways.

A comprehensive examination of the whole landscape depicting the various methyl-based RNA PTMs is beyond the scope of this article, and readers interested in this general aspect of methylepitranscriptomics should consider consulting broader review articles previously published [72–83]. However, in Figure 1 we reported a schematic representation of the most common and best studied methyl-based modifications of the various classes of RNA, along with their main human methylases (“writers”), demethylases (“erasers”) and “readers”, which produce the cellular response when a methylated ribonucleotide is detected.

Figure 1. Schematic representation of the most common and best studied methyl-based modifications of the various classes of RNA, along with their main human methylases (“writers”), demethylases (“erasers”) and “readers”, which drive cellular responses when methylated ribonucleotides are detected. ALKBHs, AlkB Homolog 1, histone H2A dioxygenase; ALYREF, Aly/REF export factor; CMTR1–2, Cap methyltransferases 1 and 2; DNMT, DNA methyltransferases; FBL, fibrillarin; FTO, Fat mass and obesity-associated protein; FTSJ1–2, FtsJ RNA 2’-O-methyltransferases 1 and 2; HENMT1, HEN methyltransferase 1; IGF2BPs, insulin-like growth factor 2 mRNA binding proteins; METTLs, methyltransferase complex subunits; MRM1–3, mitochondrial rRNA methyltransferases 1–3; NML, nucleomethylin; NSUN, NOP2/Sun RNA methyltransferases; PCIF1, phosphorylated CTD interacting factor 1; RNMT, RNA guanine-7 methyltransferase; TRMT, tRNA methyltransferase; WTAP, Wilms tumor 1-associating protein; YBX1, Y-Box Binding Protein 1; YTHDCs, YTH Domain Containing proteins; YTHDFs, YTH N6-methyladenosine RNA binding proteins; ZCCHC4, zinc finger CCHC-type containing 4.

In summary, although how the methylome of tRNAs and other RNAs modulates protein and gene expression represents an emerging research field, it is reasonable to consider the hypothesis that the methyl-based epitranscriptomic landscape might play a significant role in foreseeing or driving different cell responses to external triggers and stimuli.

RNA methylation, redox homeostasis, and oxidative stress

Perturbation of redox balance and alteration of the activity of redox-dependent signaling pathways are among the major triggers of critical cell-fate decisions upon stress [84]. The so-called redox homeostasis depends on the accurate balance between the generation and elimination of redox-active radical and nonradical chemical species in a biological system [85]. Among the most biologically relevant chemical species, there are nonradical reactive oxygen species (ROS), such as hydrogen peroxide (H₂O₂) and peroxynitrite (ONOO−), whereas radical ROS include the superoxide radical anion (O₂^•−), the hydroxyl radical (OH•), the nitric oxide (NO•), and a number of alkoxyl/alkyl peroxyl (RO•/ROO•) molecules [85].

Redox homeostasis permits a precise control of local and systemic concentrations of ROS, such as H₂O₂, which, when present in strictly regulated amounts, serve as critical players that govern metabolic regulation and stress responses, thus permitting adaptation to environmental stimuli and stressors [86]. Conversely, uncontrolled buildup of ROS is today known to lead to oxidative chemical modification of all major macromolecules inside and outside cells and organelles, including lipids, proteins and nucleic acids. If prolonged, such a condition may lead to severe oxidative stress (OS), which is mechanistically linked to serious cell dysfunction and irreversible loss of homeostasis [87]. On this basis, living matter has evolved a complex network that is responsible for scavenging ROS in excess and for accurately controlling the redox homeostatic balance. The clearance of ROS involves coupled and uncoupled enzymatic systems that include superoxide dismutases (SODs), glutathione peroxidases (GPXs), catalase (CAT), thioredoxins (TRXs) and peroxiredoxins (PRDXs), together with non-enzymatic antioxidant species that include reduced glutathione (GSH), several water- and lipid-soluble vitamins and oligoelements [72,88].

Ribosomal RNAs and transfer RNAs

Collectively, the literature suggests that RNA is the most abundant cellular nucleic acid, with 10–20 times higher levels of ROS-mediated damage, as compared to DNA [73–75]. RNA is more susceptible to oxidative damage than DNA mostly due to its single-stranded structure, and due to the absence of protective proteins [76]. Oxidatively modified RNA may lead to significant changes in the most critical biological processes within the cell. For example, oxidative modification of rRNA has been linked to errors during codon-anticodon pairing [77] and altered protein synthesis [78]. Impaired protein biosynthesis could also derive from the presence of oxidatively inactivated tRNAs [79].

The presence of methylated residues in ribonucleic acids has an impact on the structure of the molecule, and some evidence indicates that RNA methylation modifies the susceptibility of RNAs to ROS. Very interestingly, Estevez and collaborators recently reported that structure-stabilizing methylation in E. coli rRNAs protected bacterial 23S ribosomal ribonucleic acids from ROS-induced damage [80]. In this context, 7-methylguanine (m⁷G), methylguanine (m²G), and guanosine ribose methylation are believed to be particularly important [80].

Remarkably, not only the presence of methylated nucleotides is able to affect RNA susceptibility to ROS, but also RNA methylation might serve as a means through which cells can respond to stimuli. Based on this assumption, and as mentioned above, changes in the epitranscriptomic pattern have been in the radar of researchers for decades. As early as the 1960s, the levels of methylated nucleotides in tRNA, which is the type of RNA on which early research was focused because of the higher abundancy and relative stability [81], were significantly altered by environmental stresses [82]. However, only in the last twenty years a great wave of interest has attracted biomedical researchers, with many groups actively investigating and fully recognizing the importance of the role of post-transcriptional internal methylation of RNAs as a central element in the dynamic regulatory network that is responsible for controlling gene expression through which cells adapt to external stimuli.

As reviewed by Wilkinson et al., the stress response in cells often involves an orchestrated pattern of RNA modifications that are mobilized to modulate a plethora of transduction pathways aimed at counteracting heat shock, metabolic imbalance, loss of proteostasis, and DNA damage [83]. Regarding this latter aspect, RNA modifications were found to participate to the maintenance of genome integrity and repair of nucleic acid lesions by modulating the formation of DNA-RNA hybrids (R-loops) at damage sites on DNA [89].

One of the most important mechanisms through which environmental agents interact with living matter is oxidative stress (OS) [90], and the integration of redox signaling processes often serve as transduction pathways through which cells modify gene expression and phenotype to respond to internal and external stimuli [91].

Unfortunately, some studies have provided potentially interesting but limited information on how RNA methylation-related metabolism is able to affect the cellular and organismal responses to OS, due to some weaknesses in the rationale or methodology (such as, the measurement of total m⁶A rather than the evaluation of the contribution of distinct RNA classes), however today much more is known as to how internal RNA methylations has a pivotal function in the cellular response to OS, as compared to just a couple of decades ago.

It is worthy of note that the catalysis of RNA methyltransferases does not rely on redox reactions, therefore the discovery of an iron/sulfur-dependent methyltransferase that methylates 23S rRNA in E. coli raised extreme interest twenty years ago. Such findings suggested that RNA methylation events could be affected by the cellular redox status and, more importantly, that the redox-responsive site of some RNA methyltransferases may undergo oxidative-dependent inactivation [92]. It was even hypothesized that methylation of 23S rRNA in E. coli could be regulated by cellular oxidative stress [93]. In 2015, Kyuma and collaborators reported that defective methylation of adenosine and cytidines at several positions in 16S rRNA was able to induce increased sensitivity to oxidative stress in S. aureus, thus leading to its attenuated virulence in a silkworm infection model [94,95]. This straightly led to the hypothesis that the regulatory activity of the methyltransferases the are required to such rRNA modifications may become important upon OS.

Other researchers outstandingly confirmed this idea in eukaryotes. Using bisulfite-based sequencing methods, Schosserer and collaborators observed that gene silencing of the “writer” NSUN5 rendered C. elegans worms resistant to the ROS-generating compound paraquat (PQ) [48]. Similar effects were documented in a H₂O₂-treated strain of S. cerevisiae with the NSUN5 homologue knocked out [48]. Such effects were linked to the lack of methylated specific cytosines within worm and yeast rRNAs, and this could be due to changes in the core ribosomal function that may allow the cells to modulate ribosome biogenesis faster, permitting a more efficient protein synthesis in case of stress [48].

Other important targets of methylation are tRNAs, and some of these chemical modifications have been critically associated with processes that govern OS-related changes in gene expression. In the early 2000s, Begley and colleagues, who have been among the most active researchers in this field, used high-throughput screening of gene deletion libraries and found that tRNA methyltransferase 9 (Trm9), an enzyme that specifically methylates tRNA, could be critical to prevent DNA damage and to increase resistance of S. cerevisiae against exposure to reactive oxygen species (ROS)-generating agents [96,97]. A few years later, Begley’s team demonstrated that some of the effects elicited by Trm9 were mediated by ribonucleotide reductases 1 and 3, which are key elements that respond to DNA damage via enhancing the production of deoxyribonucleotides needed for DNA synthesis [98]. More recently, a brilliant application of bioanalytical and bioinformatic tools by Chan et al. proposed a model which explained how the exposure of yeast to hydrogen peroxide was able to induce selective translation of mRNAs coding for proteins needed for the oxidative stress response, such as the ribosomal 60S subunit protein RPL22A, via increased levels of m⁵C at the anti-codon 5’ end of tRNA^Leu [99].

The idea that tRNA methylation was linked to cellular OS was independently confirmed by Russian researchers, who in 2009 demonstrated that the generation of m⁶A in tRNA^Val conferred a significant growth advantage upon oxidative stress to E. coli [100]. In 2011, Osterman and coworkers showed that E. coli was able to survive extreme OS only when a specific residue of 23S rRNA was methylated (i.e. m²G1835) [101], and this further research corroborated that resistance of prokaryotes to OS was mediated by RNA methylation. Later research on the opportunistic pathogen P. aeruginosa showed that an high resistance to OS could be obtained by abolishing the expression of the trmJ homolog tRNA (cytidine(32)/uridine(32)/adenosine(32)-2′-O)-methyltransferase, thus reducing the formation of Nm in tRNAs [102].

These findings led to the solid idea that the complex system of chemical tRNA modifications, and specifically methylation processes, may represent a powerful means through which cellular responses to the environment could be regulated post-transcriptionally [103].

In 2014, such idea was corroborated by the intriguing observations of Barroso and colleagues, who showed that hypomethylation of tRNA^[Ser]Sec, a particular type of transfer RNA that allows the biosynthesis of key antioxidant selenium-containing proteins, was able to induce oxidative stress in human umbilical vein endothelial cells (HUVECs) [104].

This idea was somewhat supported by the findings reported by Schaefer and colleagues, who demonstrated that Dnmt2 null mutant fruit flies showed reduced viability upon H₂O₂- and paraquat-induced OS [105]. As mentioned in the sections above, DNMT2 is an evolutionary conserved multisubstrate enzyme that is critically involved in the regulation of genomic stability via methylation of nucleic acids [106]. Originally considered as a pure DNA methyltransferase, DNMT2 was demonstrated to be a tRNA methyltransferase in 2006, when DNMT2 was mechanistically linked to methylation of tRNA^Asp [107]. In their experiments, Schaefer and coworkers provided an exquisite proof that DNMT2-mediated methylation of tRNAs was able to protect tRNA^GlyGCC and tRNA^AspGTC from stress-induced ribonuclease cleavage, which represents a conserved response to several stress stimuli in eukaryotes [105]. Taken together, these results point out to a specific role for the DNMT2-dependent regulation of tRNA structure and fate in the response of organisms to pro-oxidant conditions.

Conversely, a few years later, Mytych and coworkers found that human HeLa cells undergo oxidative stress after exposure to nanodiamond particles, possibly via upregulation of DNMT2. The same authors found also that vascular smooth muscle cells responded to a pro-oxidant insult by arresting the cell cycle, and this was associated with the upregulation of DNMT2 as a part of cellular stress response [108]. However, only a speculative stabilizing role of DNMT2 on RNA was hypothesized by the authors [108], without providing clear evidence of the actual mechanism underlying such response.

As already mentioned, NSUN2 is another important regulator of tRNA methylation [109]. Interestingly, increased apoptotic death was observed by Blanco et al. in Nsun2^−/− mice and human skin cells in response to sodium arsenite- or UV-induced OS, and this was linked to the loss of NSun2-mediated RNA methylation and to the subsequent accumulation of 5′ tRNA-derived small RNA fragments (tRFs) [110], which are known to inhibit protein synthesis [110,111]. Therefore, the Nsun2-dependent methylation of RNA should be considered as a key step in the OS-induced dynamic post-transcriptional regulation of gene expression.

In 2014, high levels of methylated tRNA^AspGUC were associated with high expression of OS-responsive proteins, such as peroxyredoxins (PRXs) [112]. More in detail, methylated C38 in tRNA^AspGUC was linked to selective biosynthesis of proteins specifically involved in the response to OS, and whose genes are highly enriched with GAC triplets [113]. The same year, Dewe and coworkers showed that human embryonic kidney cells in which the mammalian ortholog of yeast tRNA(guanine 26,N²,N²)-dimethyltransferase Trm1p (TRMT1) was knocked out or nonfunctional, exhibited hypersensitivity to several pro-oxidant compounds, such as tert-butyl-hydroperoxide (t-bu-OOH) and hydrogen peroxide, and this was associated with defective formation of dimethylguanosine (m^2,2G) in cytoplasmic and mitochondrial tRNAs [114]. More recently, S. cerevisiae strains that lacked specific tRNA methyltransferases (namely, Trms 3, 7, 13, and 44) exhibited higher sensitivity to H₂O₂ and rotenone, as compared to the wild type strain, and in some cases this was linked to increased ROS production and redox imbalance [115]. By using bioinformatics, the authors concluded that UUC-enriched gene transcripts, which encode proteins that play a functional role in recovering from OS, could be regulated by Trm at translational level [115]. Such results further confirmed that tRNA methylation may serve as a critical determinant for the response of eukaryotes to redox imbalance-promoting treatments. Recently, Trixl and colleagues inactivated the methyltransferase Nsun3 in mouse embryonic stem cells (ESCs) via a CRISPR/Cas9 technology, and found that H₂O₂ induced a weaker mitochondrial ROS production in Nsun3-mutant, with respect to that observed in wild type cells [116]. In the same year, Cosentino and coworkers showed in rat pancreatic islet beta cells that defective methylation of tRNA guanine 9 (m¹G9) was able to induce a controlled tRNA^Gln fragmentation, thus causing oxidative stress via production of both OH• and peroxynitrite [117]. These results confirmed that tRNA-targeting methylation reactions can initiate the execution of cellular redox signaling programs.

Among the pathways involved in cell adaptation to OS is the control of cell cycle progression. In particular, the cell cycle is often altered upon OS, and part of the rationale for this is that the molecular oxidative damage needs to be repaired before proceeding through the various phases [39]. Interestingly, Gkatza and colleagues have recently demonstrated that NSUN2 was essential for human dermal fibroblasts to adapt their cell cycle phases to an arsenite-induced pro-oxidant stimulus, and this was associated with an OS-induced decrease of methylated C34 in tRNA^LeuCAA [40]. Arsenite was used as a OS-promoting treatment also in other studies. Rashad et al. have recently reported that the m¹A “eraser” ALKBH1 participated to the cellular response of rat neuroblasts to sodium meta-arsenite in terms of cytosolic and mitochondrial tRNA methylation state and cleavage, even though the authors themselves claimed that further efforts were needed to better characterize the relationship between specific tRNA cleavage and cellular fate in terms of stress and death [118].

Messenger RNAs

As extensively discussed in the previous sections, ribosomal and transfer RNAs are not the only ribonucleic acids that are targeted by methylation reactions. Several methylated sites were also found in messenger RNAs. However, mRNA internal methylations are more difficult to identify and characterize due to their relatively low abundance [83]. Moreover, how specifically mRNA methylation may have an impact on transcript fate upon cellular stress is still under intense debate. In 2015, Zhou and coworkers published a pioneering work in which heat shock was reported to induce in mouse embryonic fibroblasts (MEFs) a 90-fold increase in the mRNA levels of the Hsp70-related gene hspa1a, and this was correlated to a strong increase in m⁶A levels within the 5′UTR of the corresponding transcript [119]. In the more specific context of oxidative stress, in 2012, Zhang and colleagues revealed that the response of HeLa cells to hydrogen peroxide relied on the stabilizing action of methylation on the 3′-UTR of p16 mRNA [120]. Four years later, the same research group demonstrated that oxidative stress and high glucose induced accelerated senescence in HUVECs by enhancing the translation of the adaptor protein Shc via NSUN2-dependent methylation of Shc mRNA at multiple sites, both in coding and non-coding regions [121]. Shc mRNA is known to produce three protein isoforms that are crucially involved in redox signaling and mitochondrial ROS production [122], and this strengthens the importance of RNA methylation as a key regulator operating in the complex network that governs how cells respond to OS.

A few years later, it was established that OS could influence mammalian microRNA methylome, as well. In particular, Yuan and coworkers found that OS was able to increase the methylation levels of microRNA 125b (miR-125b), reducing its capacity to recruit (RNA-induced silencing complex) RISC, thus increasing the expression of miR-125b-regulated target transcripts [123]. The transcription factor nuclear factor erythroid-derived 2-like 2 (Nrf2) is certainly one of the most important miR-125b-regulated determinant in cell response to OS [124]. This further supports the idea that post-transcriptional RNA methylation is able to affect cell phenotype in terms of redox homeostatic capacity. To the best of our knowledge, no other studies have attempted to clarify the role of methylation in mammalian microRNAs as epitranscriptomic factors serving as regulators to drive a response to OS. Hence, this definitely represents an under-investigated field that might hide secrets and potential targets that for future benefits.

More recently, Zhao and colleagues have observed increased levels of OS and higher levels of m⁶A nrf2 mRNA in Sprague-Dawley rats that were exposed to the ROS-generating compound di-(2-ethylhexyl) phthalate [125], thus corroborating the notion that the post-transcriptional control of NRF2 may link mRNA-based epitranscriptomic regulation of gene expression to the cellular effects elicited by pro-oxidant. The involvement of the Kelch like ECH associated protein 1 (KEAP1)-NRF2 axis in the mRNA methylation-dependent cytoprotective response to OS has been recently observed by Arumugam and coauthors, who found that the ROS-generating compound fumonisin B1 (FB₁) induced in immortalized human liver cells an aberrant m⁶A modification in total mRNAs via differential regulation of the major methyltransferases and demethylases that participate to m⁶A RNA metabolism (namely, METTL3/14 and ALKBH5) [126]. Such effect was shown to be causatively linked to the FB₁-induced increase in the levels of m⁶A mRNA and to the subsequent reduction in the protein expression of the repressor Keap1 [126]. Keap1 downregulation is known to facilitate NRF2 translocation to the nucleus, where in the form of a heterodimer with the small Maf protein it binds to the antioxidant responsive elements (AREs) of several antioxidant genes, thus promoting their transcriptional activation [127]. On this basis, the identification of the Keap1-Nrf2 axis as a pathway affected by OS through selective changes in the levels of specific methylated mRNAs sheds new light on the molecular details underlying the epitranscriptome-dependent regulation of the redox homeostatic capacity in mammals. Other experimental evidence supports such hypothesis. Indeed, NSUN2 and METTL3/METT14 were found to cooperatively and synergistically methylate p21 mRNA at m⁵C and m⁶A, thus enhancing p21 expression at the level of translation in a model of H₂O₂-induced cellular senescence [129,p.2]. P21 is a cyclin dependent kinase (CDK) inhibitor that exerts its protective effect against OS through multiple pathways, including the upregulation of the Nrf2 downstream signaling pathway [128,129].

Interesting results from a recent work by Zhao and coauthors showed that the underexpression of METTL3 and the subsequent reduction of m⁶A RNA levels could participate to the neurotoxic effect of Aβ oligomers in rodent and human brains, thus linking the regulation of methyl-epitranscriptome to the OS-based pathogenetic mechanisms of Alzheimer disease (AD) [130].

Hypoxia/reoxygenation (H/R) is a well-known phenomenon that causes overproduction of ROS and significant OS [131]. In this context, Song and coworkers have recently reported that rodent cardiomyocytes and hearts respond to H/R with repressed autophagic flux and increased apoptosis, as well as with a parallel upregulation of METTL3 and increased levels of METTL3-dependent methylation (m⁶A) in mRNAs [132]. These authors demonstrated that the OS-induced overexpression of METTL3 led to the hyper-methylation and destabilization of the transcriptional factor EB (TFEB) mRNA, a master regulator of autophagy, thus preventing its accumulation and translation in cardiomyocytes [132]. A few months later, Wang and colleagues shed further light on the response of the mammals’ cardiovascular system to the H/R-dependent OS. In fact, these authors observed an increased apoptotic death both in human cardiomyocytes exposed to H/R and in a rat model of myocardial ischemia/reperfusion (I/R) [133]. Interestingly, such effect was strictly dependent on the promotion of endoplasmic reticulum (ER) stress via a m⁶A-based upregulation of the activating transcription factor 4 (ATF4) mRNA [133].

Interestingly, in 2021, Pang et al. reported that increased levels of METTL14 served as a protecting factor against OS-induced reduction in m⁶A RNA after I/R insult, both in vivo and in vitro [134]. This was accompanied by the activation of the Wnt/β-catenin axis via increased levels of m⁶A in Wnt1 mRNA [134]. Considered the acknowledged role of the Wnt/β-catenin axis [135], Pang and colleagues’ results seem to suggest that the OS-triggered modulation of the m⁶A epitranscriptome occurs via regulation of one of the most important and conserved signal transduction pathways in tissue homeostasis. This idea was further confirmed by Guo and coworkers, who recently found that the inhibition of RNA demethylase activity of ALKBH5 was sufficient to improve the function of H/R-treated human trophoblast cells by activating the Wnt/β-catenin pathway [136]. These authors also found that the repression of the activity of ALKBH5 was essential to alleviate OS-dependent apoptosis in H/R-treated human trophoblast cells, and this effect was closely correlated to increased levels of m⁶A in peroxisome proliferator-activated receptor gamma (PPARγ) mRNA, as well as this effect was linked to augmented levels of the corresponding protein [136], thus strengthening the idea that in mammals there exists a sophisticated ensemble of epitranscriptome-mediated post-transcriptional regulatory mechanisms that is able to modulate cellular resistance to OS-promoting insults.

As discussed above, regarding the exact molecular mechanisms by which the rewiring of the mRNA methylation program can be functional to the response to OS, no conclusive idea has been developed so far. However, in 2018, Anders and colleagues revealed a novel pathway through which OS may lead to changes in the methyl-mRNA landscape. In fact, arsenite (AS)-treated human embryonic kidney cells exhibited increased levels of N6-methyladenosine in the 5′ UTRs and in the 5′ proximity of coding sequences of messenger RNAs, and this was mechanistically linked to the segregation of transcripts in stress granules (SGs) [137]. SGs are cytoplasmic RNA-protein complexes in which mRNAs are thought to be sequestered in a translationally stalled initiation form [138]. On this basis, recent discoveries depict a totally new function of mRNA N6-methyladenosines in the intracellular processes underlying the modulation of cell response to stress in mammals. In this context, the major m⁶A-binding proteins, YTHDF1–3, were recently indicated as possible critical players in the process of SG formation. In particular, double knockdown-based experiments on mammalian bone cells showed that YTHDF1 and YTHDF3 are essentially required to form SGs [139]. Another interesting observation was recently reported by Du and colleagues, who found that I/R-injury in liver and hepatocytes proceeds via OS and Drp1-dependent mitochondrial fragmentation due to a reduction of the enzymatic activity of N6-methyladenosine demethylase FTO [140].

lncRNAs

Another potentially interesting finding comes from a recent work by Qu et al., who reported that m⁶A-modified long noncoding RNAs (lncRNAs) may also have a role in mediating the effects of cadmium-induced oxidative damages [141]. Few other researchers have confirmed that methyl-based modification of lncRNAs may be involved in the spectrum of molecular changes activated by OS in living cells. Su and colleagues observed that exposure to the ROS-generating PQ induced a hypermethylation of total RNA in mouse neuroblasts, and this was associated with impaired expression of m⁶A methyltransferases and demethylases, along with an altered pattern of m⁶A-modified lncRNAs [142]. In particular, the authors identified two specific lncRNAs that were differentially expressed in PQ-treated cells, compared to controls: the cell division cycle 5-like (lncRNA CDC5L) and the signal transducer and activator of transcription 3 (lncRNA STAT3), and this seems to suggest that the regulation of the autophagic process could be closely involved in the downstream signaling activated by OS via lncRNA methylation [142]. Unfortunately, some of these reports tend to rely on relatively simple statistical analyses, but do not always provide a clear explanation as to how the findings observed might have biological relevance. More specifically, in some cases, a limited molecular explanation was provided as to how the participation of methyl-lncRNAs could determine the cytotoxic effects of OS or could prevent the OS-dependent biological effects, therefore researchers with documented and strong expertise in this field should study more in depth this intriguing phenomenon.

Oxidative stress and anticancer therapy

As briefly discussed above, at strictly controlled concentrations, ROS serve as functional signaling entities in cells, regulating crucial physiological processes such as proliferation, differentiation, survival, and adaptive response to external stimuli [86].

However, an unbalanced production of ROS may lead to an important loss of redox homeostasis, metabolic dysregulation, molecular damage, genetic instability, and cell dysfunction, thus contributing to the onset of several disorders and diseases, including tumor and cancer [86,143].

ROS and chronic OS not only can promote carcinogenesis, but also mediate tumor survival and progression [144,145]. In fact, an increasing number of reports have established that high levels of ROS are often required by cancer cells to escape cell death, to promote mass growth, to adapt metabolic needs, and to modify the surrounding environment to facilitate tumor expansion and migration.

For example, both superoxide anion and H₂O₂ appear to be linked to tumor and cancer cell phenotype and behavior. Specifically, in the 1990s, the metastatic capacity of rodent fibrosarcoma cells was reduced by increasing the mitochondrial production of O₂.- [146]. In addition, a strong O₂.--dependent mitogenic activity was found in fibroblasts expressing a constitutively active isoform of the oncogene p21 [147]. In more recent years, Lee et al. showed that the ROS-induced inhibition of tyrosine phosphatases was able to sustain the activation of antiapoptotic pathways in pancreatic cancer cells [148]. In addition, enhanced ROS levels were linked to drastically diminished apoptotic death in bladder cancer cells [149]. Likewise, the selective upregulation of mitochondrial ROS was demonstrated to stimulate cell proliferation and cell survival in carcinoma cells, along with epithelial – mesenchymal transition (EMT) through the activation of mitogen-activated protein kinase (MAPK) and Ras-ERK axis [150]. In 2016, Cao and coworkers reported that the overproduction of ROS and the consequent activation of ERK and p38 MAPK signaling pathways were essential for hyperglycemic stress to induce migration and invasion of pancreatic cancer cells [151]. Similarly, proliferation of colon cancer cells was found to be promoted by EGF-induced stimulation of ROS via activation of NADPH oxidase and heme oxygenase-1 (HO-1) [152]. In addition, inducible nitric oxide synthase (iNOS)-dependent NO• production boosted the growth of oral squamous carcinoma cells, also through the stimulation of angiogenesis [153]. In 2017, Aydin and colleagues reported that ROS promoted the metastatic ability of murine melanoma cells by repressing the NK cell- and lymphocyte-mediated anti-cancer activity [154]. Furthermore, the ROS-induced hyper-proliferation of human melanoma cells was found to be facilitated by the generation of high local concentrations of extracellular ROS via modulation of signaling pathways involving HIF-1α and Akt/GSK3β/p27^Kip1 [155].

On this basis, biomedical research was solicited to find a way to scavenge ROS as a tool to prevent tumor formation or, at least, to delay the progression of neoplasms. As early as the 1950s, dietary intake of the antioxidant vitamin E was found to be associated with protection from the tumorigenic effect of 3’-methyl-4-dimethylamino azobenzene [156]. Over the following years, several studies and randomized trials have confirmed that in some cases antioxidant chemicals could serve as co-therapeutics in anticancer treatments [157–159].

However, modern literature seems to support the notion that a strong and persistent increase in ROS levels leads to cell death in many cancer cell types. This explains why some of the most effective cancer therapeutics that are used today by oncologists undoubtedly rely on the ability to upregulate ROS concentrations within cancer cells [160,161]. As a corollary to this concept, it should be noted that both increased antioxidant protection and enhanced redox buffering capacity have been frequently linked to a more aggressive and therapy-resistant behavior in cancers [162]. For example, as discussed previously, the NRF2-KEAP1 system plays a critical role in preserving redox homeostasis, and the activation of such a molecular switch has a primary function to adapt to cellular OS [163]. Accordingly, a prolonged activation of NRF2 was hypothesized to promote the survival of precancerous cells harboring oncogenic mutations [164], and analogous findings help to understand why NRF2 is highly expressed in a variety of cancers [165]. The improvement of the redox homeostatic capacity is certainly one of the most important adaptive responses through which tumor progression/dissemination can be achieved, and one of the most central mechanisms by which therapy resistance traits can be developed in malignancies [161,166,167]. Other means include: i) reduced drug intake and/or enhanced drug inactivation; ii) adaptation of metabolic pathways; iii) increased DNA repair; iv) suppression of proapoptotic signaling and/or activation of prosurvival signaling; v) enhanced autophagy; vi) rewiring of mitochondrial dynamics [161,166–170].

Mitochondrial ROS production and m⁶A-modified RNAs

Mitochondria represent one of the major sources of ROS generation within cells [171]. Therefore, some researchers have investigated whether and how RNA methylation could modulate the activity of ROS-producing mitochondrial pathways. In 2019, Zhuang and colleagues reported that the FTO demethylase was able to reduce m⁶A levels in the PGC-1α transcript, thus increasing the expression of this master regulator of mitochondrial biogenesis and boosting both mitochondrial activity and ROS production in a model of clear cell renal cell carcinoma [172]. However, FTO was found to “erase” m⁶Am, not m⁶A, in cells [173,174], thus suggesting that Zhuang and coworkers’ findings might be read in terms of reversible dimethyladenosine modification to PGC-1α mRNA.

In 2019, some researchers found that dopaminergic PC12 cells overexpressing FTO were characterized by lower activities of superoxide dismutase and mitochondrial complex II, thus strengthening the idea that the dynamics of dimethyladenosines in RNA could regulate ROS-generating processes, especially in mitochondria, interfering with the activation of the apoptotic pathway [175]. A year later, Liu and coworkers demonstrated that the activity of p38 mitogen-activated protein (MAP) kinase (MAPK), which undergoes phosphorylation and subsequent activation upon mitochondrial ROS generation, was regulated by the expression of adenylate kinase (AK4), and the expression of AK4 was higher in tamoxifen-resistant MCF-7 cells with increased levels of METTL3-induced m⁶A in the 5′ and 3′ UTR of adenylate kinase (AK4) mRNA [176]. In 2021, melatonin was found to attenuate ROS generation in a reproductive model of cromium-induced mitochondrial redox imbalance, and the protective role of melatonin were exerted via restoration of METTL3-mediated m⁶A RNA modification, as well as through the activation of mitochondrial fusion and inhibition of mitophagic pathway [177]. Other researchers provided strong evidence that the regulation of the RNA methylome is able modulate mitochondrial ROS production. In fact, in 2022, Sun and collaborators found in a liver fibrosis model an impaired translation of the mRNA of peroxiredoxin 3 (PRDX3), which serves as a master regulator of mitochondrial oxidative stress, and this was mechanistically linked to a diminished interaction between m⁶A and the m⁶A “reader” YTHDF3, thus leading to increased mitochondrial ROS and hepatic stellate cell (HSC) activation [178]. A very interesting paper from Xu and colleagues reported that METTL3 was able to enhance the methylation of cytochrome c oxidase subunit (NDUFA4) mRNA, and this was sufficient to stimulate the expression of NDUFA4 in gastric cancer cells, with consequent inhibition of mitochondrial ROS accumulation [179].

These results strongly point out to RNA methylation as a powerful regulator of mitochondrial redox homeostasis and activity, and explain the raising interest of researchers toward the methylepitranscriptome as a biological means through which important mitochondrial redox alterations can occur in cells leading to significant changes in cellular phenotype. This aspect certainly deserves further investigation, as mitochondria serve as a critical crossroads in cells where ROS homeostasis meets metabolic and energy needs, thus representing a potential hub that may integrate redox sensing with adaptive changes in gene expression via RNA post-transcriptional dynamic and reversible chemical modifications, both in physiology and disease. In this context, Zhang et al. found that oxidized low-density lipoprotein (oxLDL) activate OS- and flogosis-related responses in monocytes via a METTL3-YTHDF2 cooperative interaction that causes the degradation of peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) mRNA, increased ROS leakage and, ultimately, reduced ATP production [180]. Considered that PGC-1α critically regulates mitochondrial biogenesis [181], Zhang and coworkers’ findings seem to confirm that the OS-activated pathways in mammalian cells could be intimately linked to dynamic changes in methylepitranscriptome, altered mitochondrion-dependent energy metabolism, ROS sensing and production.

RNA methylation and cancer cell response to clinically relevant OS-generating treatments: factual evidence and missing pieces of the puzzle

As discussed above, on one hand, the fine-tuned control of ROS production and metabolism, along with the redox homeostatic buffering capacity, is a key factor that contributes to define the susceptibility of neoplasms and malignancies to many anticancer therapeutics. On the other hand, adaptive processes and cellular response to OS-generating stimuli are governed by a multilevel molecular network that reacts to exogenous stimuli, via regulatory events that include changes in the RNA methylation pattern.

As discussed by others [182], it might be conceived that specific methylated RNA residues or even the overall RNA methylation diversity could predict or influence the effect of redox-based anticancer treatments on tumor and cancer cells.

5-fluorouracil

In this context, one of the first evidence was reported by Cori and coauthors in the late 70s, when the resistance of hepatoma cells to 5-fluorouracil (5-FU), a pyrimidine analog that is widely used to treat a variety of solid cancers, was reversed by adding inosine to the incubation mixture, and this was surprisingly associated with re-shaping of the methylation pattern of ribosomal RNA [183]. It is important to note that the antiblastic mechanism of action of 5-FU, which is a drug often used to treat cancers that are unresponsive to targeted therapy, is at least in part mediated by its ROS-generating capacity [184]. This should have probably pushed redox researchers to investigate this topic further. However, after such a pioneering report, this field remained relatively dormant for many decades in the shadow cast by the limited knowledge and technical difficulties surrounding the epigenetics field [185,186]. Only in April 2008, Marie Gustavsson and Hans Ronne proposed a model to explain how 5-FU-induced effects in eukaryotes may be influenced also by biological processes that involved RNA methylation. The authors hypothesized that cellular responsiveness to 5-FU could depend on the reduced activity of several tRNA modifying enzymes, among which the tRNA 5-methyluridine methyltransferase would promote RNA destabilization, especially at higher temperature [187]. Interestingly, this would explain why hyperthermia is able to enhance the effect of 5-FU in some anticancer therapies [188]. A few years later, Okamoto et al. found that the sensitivity of human cancer cells to 5-FU could be decreased by co-overexpressing both tRNA methyltransferases NSUN2 and METTL1 in HeLa cells; accordingly, the cytotoxic effect of 5-FU was increased through a double knockdown of both NSUN2 and METTL1, and such potentiated action of 5-FU was associated with a rapid degradation of tRNA^ValAAC [189]. This finding is important because it is known that cleavage of tRNAs is increasingly considered as an effective mechanism through which protein synthesis can be regulated during specific stress conditions [190–192]. A few years ago, the role of the m⁶A “writer” METTL3 in chemoresistance toward 5-FU was also investigated. In their work, Taketo and coauthors showed that the viability of 5-FU-treated pancreatic cancer cells was dramatically reduced when METTL3 was stably knocked down [193]. The extreme importance of mutant versions of the tumor suppressor protein p53 is fully acknowledged by modern molecular oncologists [194]. In this context, Uddin and coworkers recently found that human colon cancer cells with specific methylated adenosines in p53 pre-mRNA generated a R273H missense p53 mutant, thus conferring drug resistance toward 5-FU [195]. p53, which is also known as the “guardian of genome”, serves as an important regulator for dual cellular responses to OS, activating the major antioxidant defense system to remove pro-oxidants and oxidative damage, thus ensuring cell survival in mild OS, while exhibiting pro-oxidant effects that further increase the levels of stresses, leading to apoptosis, in distress conditions [196]. Therefore, the participation of p53 to the epitranscriptome-dependent regulatory network that determines the response to chemotherapeutics should motivate redox biologists to investigate further and with great effort this fascinating research field. In this regard, Ke and coworkers found that the p53 mRNA sequence presents 24 DRACH sequence motifs [197], which are the most common subset of methylation sites [35], thus strongly suggesting that methylation processes may play a strategic role in regulating p53 expression.

More recently, Ma et al. reported that higher expression levels of the lncRNA ladybird homeobox 2 antisense RNA 1 (LBX2-AS1) were correlated with lower responsiveness to 5-FU, and this was mechanistically caused by the METTL3-dependent m⁶A RNA methylation [198]. Pan and colleagues’ results confirmed that METTL3 plays a key role in resistance to 5-FU, more importantly in exosomal-mediated transfer of cancer resistance traits. In fact, the authors showed that the upregulation of METTL3‑dependent m⁶A methylation in colorectal cancer (CRC) enhanced the processing and expression of exosomal miR‑181d‑5p, which in turn inhibited the sensitivity of CRC cells to 5‑FU by targeting neurocalcin δ [199]. Unfortunately, most of these studies were carried out excluding any parameters clearly linked to redox homeostasis (e.g. DNA oxidative damage, lipid peroxidation, ROS levels etc.), therefore only suggestions can be provided as to whether the redox-responsive signaling pathways can be modulated or OS-related phenomena can be involved in the epitranscriptome-driven adaptive response triggered in cancer cells by the exposure to chemotherapeutics. In this context, LBX2-AS1 has been found to co-express with other genes frequently expressed upon OS-generating conditions, such as flogosis [200]. In addition, neurocalcin δ has been recently reported to downregulate inflammation via inhibition of the IKK/IκBα/NF-κB signaling pathway [201]. Moreover, it was recently reported that miR-181d-5p overexpression significantly repressed apoptosis in a model of ischemia/reperfusion oxidative injury [202]. On this basis, flogosis-related biological signaling might represent a putative target to be investigated further by molecular oncologists that are interested in clarifying the role of redox biology in modifying the response of cancer to chemotherapy in general, and to 5-FU in particular. This aspect gains particular importance for colorectal cancer patients, in which 5-FU is known to elicit a significant pro-inflammatory effect [203].

Doxorubicin

RNA methylation seems to be associated with resistance to other commonly used anticancer chemotherapeutics. He and coworkers reported that resistance to doxorubicin (DOXO) in human breast cancer was mechanistically due to the activation of the GlcNAc N-deacetylase/N-sulfotransferase-1 (NDST1)-dependent downstream signaling pathway, and this was found to be caused by the hypermethylation of the 5′-UTR of miR-149 and its consequent downregulation [204]. DOXO, whose pharmacological action is also elicited by promoting a condition of strongly increased oxidative load [205], is a powerful antitumor drug that is often administered to breast cancer patients. Accordingly, DOXO-treated triple-negative breast cancer cells undergo massive H₂O₂ overproduction and caspase 3/8-dependent apoptosis [206]. In this context, it is worth to note that interleukin 6 (IL-6) is one of most interesting ROS- and flogosis-promoting targets of miR-149 [207]. Therefore, the few indications found in literature might suggest that a link between cancer cell resistance to ROS-promoting chemotherapeutics and the hypermethylation of miR-149 may exist. Unfortunately, no direct evidence for such a cause-effect relationship was provided so far.

Uddin and coworkers recently reported that colon cancer cells expressing the missense mutant p53 due to RNA m⁶A methylation at codon 273 exhibited resistance toward DOXO [195]. Interestingly, when the authors silenced METTL3, drug-resistant cells were re-sensitized to DOXO, thus suggesting that METTL3 was essential for colon cancer cells to develop drug resistance through transited adenosines in the p53 pre-mRNA [195]. p53 activation is often observed as a response to ROS over-production and DNA damage [208], however Uddin and coworkers’ study did not investigate whether ROS or OS may have a role in the cytoprotective effect observed in mutant p53-haboring cancer cells, leaving this question open to further research. In 2021, Pan et al. showed in breast cancer cells that resistance to DOXO was associated with an increased expression of METTL3, that in turn upregulated the expression of miR-221-3p [209]. In their work, METTL3 over-expression was intriguingly linked to enhanced maturation of pri-miR-221-3p via increased levels of m⁶A methylation in its RNA [209]. Unluckily, no OS-related parameters or redox-specific molecular pathways were investigated in this study, however, the authors demonstrated that one major target of miR-221-3p was the homeodomain interacting protein kinase-2, which has been recently acknowledged as a critical stress-responsive kinase whose activity plays a key role in promoting cell survival upon OS [210].

Platinum-containing agents

Platinum-containing drugs (e.g. cisplatin, carboplatin, oxaliplatin, etc.) are also ROS-generating drugs that are widely employed in cancer patients with several types of solid tumors [211,212]. In this context, some Japanese researchers observed a strongly reduced viability in pancreatic cancer cells treated with cisplatin after METTL3 knockdown, and similar results were obtained treating cells with gemcitabine [193]. A microarray-based gene ontology (GO)-based analysis revealed that shMETTL3 cancer cells had several genes associated with the MAPK cascade significantly down-regulated [193]. Importantly, MAPK-dependent signaling is involved in vital biological processes, such as proliferation and stress response, along with DNA repair and apoptosis [213,214], thus highlighting the important fil rouge among RNA methylation, gene expression and response to stressors. In 2018, Li and colleagues studied biopsies from patients with esophageal squamous cell carcinoma (ESCC), and identified an overexpressed NSUN2-methylated lncRNA (NMR) that was able to promote migration and invasion of ESCC cells in vitro, and to prevent cisplatin-induced apoptosis [215]. Surprisingly, such a protective effect of NMR was presumably related to its direct binding to a component of the nucleosome remodeling factor (NURF) chromatin-remodeling complex [215]. A high resistance to cisplatin was recently revealed in non-small-cell lung cancer cells with METTL3-dependent m⁶A over-methylation within the pre-mRNA of Yes-associated protein (YAP) [216], whose aberrant activation is able to drive a broad transcriptional program intimately connected to cell cycle progression and DNA repair [217]. Also in this case, no information was unfortunately provided about the possible participation of redox-dependent pathways to the METTL3-YAP axis. This identifies an important knowledge gap, also taking into account the recognized strict co-operation between YAP and NRF2, probably the most important master regulator of cell response to OS, in contributing to chemoresistance in cancer cells [218]. In 2019, a brilliant work from Zhang and coauthors demonstrated that METTL3-induced m⁶A over-methylation in the mRNA of Chromobox 8 (CBX8) was able to enhance the stability of CBX8 mRNA, thus leading to CBX8 overexpression and development of chemoresistance in colon cancers [219]. Again, no details were provided as to whether redox-dependent factors or redox-dependent signaling pathways participated in such an effect. Interestingly, some China-based researchers recently showed that the expression of CBX8 can be regulated by redox stress [220], and this identified another potential signaling pathway through which cancer chemoresistance traits could be developed by rewiring the control program that defines the state of the RNA methylome. In this regard, Wei and others reported that seminoma cells could be rendered resistant to cisplatin by stabilizing the mRNA of the transcription factor AP-2 gamma (TFAP2C) via METTL3-mediated upregulation of m⁶A methylation levels [221]. Even though no specific OS-related endpoints were included by the authors in their study, such findings are appealing also in the context of cellular responsivity to OS-promoting conditions. In fact, in 2013, Kulak et al. showed that TFAP2C was able to transcriptionally regulate the expression of glutathione peroxidase (GPX1) through the AP-2 regulatory sequence in the promoter region of the gpx1 gene [222]. This might represent a possible link through which METTL3 overexpression could increase the redox homeostatic capacity of cancer cells, thus promoting the development of cisplatin-resistant phenotypes. In 2021, Song and colleagues confirmed that methyl-RNA “writers” had an extremely relevant role in cancer drug resistance. In their paper, the authors demonstrated that inhibition of METTL3 dramatically reduced the sensitivity of human lung adenocarcinoma cells to cisplatin, whereas its overexpression sensitized cells to the drug [223]. Very interestingly, mettl3 was proved to be a direct target gene for miR-4443, which resulted to be enriched in resistance-transferring exosomes from cisplatin-resistant non-small cell lung carcinoma (NSCLCs) cells [223]. Fascinatingly, Song and colleagues showed also that the miR-4443/METTL3 axis could drive resistance in NSCLC via increasing the level of the ferroptosis suppressor protein 1 (FSP1) [223], and this crucially links the METTL3-dependent regulation of the methylation state of RNA to the iron-dependent and lipid peroxidation-induced regulated death observed in cells exposed to certain chemotherapeutics, such as some platinum-containing drugs [224]. Others have confirmed the participation of m⁶A “writer” METTL3 in the regulation of ferroptosis. In fact, Sun et al. recently reported that METTL3-dependent methylation of the ferroptosis defense protein cystine/glutamate antiporter mRNA serves to generate binding sites for the ferroptosis suppressor NF-κB activating protein [225]. In this regard, it is worthy to note that recent reviews have extensively highlighted how the regulation of ferroptosis could represent a powerful therapeutic tool through which cancer resistance to conventional therapies and immunotherapy may be effectively reversed [226,227].

5-azacytidine

Some types of hematologic malignancies, such as acute myeloid leukemia, are treated with 5-azacytidine (5-AZA), whose antimetabolic activity leads to an effective chemotherapeutic action [228]. In 2018, Cheng and coauthors found that RNA 5-methylcytosine (m⁵C) and its modifying enzymes are strongly overexpressed in 5-AZA-resistant leukemia cells, both in vitro and ex vivo [229], however neither direct nor indirect participation of OS or ROS-involving pathways were investigated by the authors. This left open the intriguing hypothesis that the resistance of leukemia cells toward 5-AZA could be based on an epitranscriptomic-induced alteration of redox-dependent signaling and metabolism. Recent literature seems to corroborate such a hypothesis, since RNA m⁵C methyltransferases include NSUN2 and DNMT2, both of which were demonstrated to affect cell resistance to OS-promoting conditions. In fact, NSUN2 serves as a key promoter of the OS-induced downregulation of protein synthesis through the formation of tRNA-derived noncoding fragments (tRFs) [230], and DNMT2-depleted fibroblasts were found to be more sensitive to oxidative stress-generating conditions [231].

PARP inhibitors

As discussed above, the cytotoxic action of many anticancer drugs is known to be mediated by increased OS and DNA damage. This is also the case of poly (ADP-ribose) polymerase inhibitors (PARPi), whose antitumor effect in ovarian cancer cells has been shown to depend on NADPH oxidase-dependent ROS over-production [232]. In 2020, Fukumoto and coworkers reported that the response of epithelial ovarian cancer cells to PARPi was reduced when cells had m⁶A modifications in the frizzled class receptor 10 (FZD10) mRNAs [233]. This resulted in the upregulation of the Wnt/β-catenin pathway, which is known to belong to the adaptive response pathway through which cancer cells adapt to OS-promoting treatments [234,235]. Again, the study did not include endpoints related to OS or ROS-dependent signaling, thus not answering to the important question as to whether methyl-based epitranscriptome could have a role in protecting malignant cells from the redox perturbation induced by the clinically important class of PARP inhibitors, which are often used to treat BRCA mutant malignancies [236].

Tyrosine kinase inhibitors

Oxidative stress also underlies some of the antitumor effects of other compounds, such as tyrosine kinase inhibitors (TKIs). In 2006, Balko and colleagues carried out an Affymetrix GeneChip array-based gene expression study and revealed that lung adenocarcinoma cells resistant to the representative epidermal growth factor receptor (EGFR) TKI erlotinib exhibited increased levels of METTL3 mRNA [237], whose protein product is one of the major m⁶A “writer” on RNA [126]. Unluckily, no indication of the involvement of redox responsive signaling pathways emerged from the research. Resistance toward other TKIs has also been studied in terms of RNA methylation-related pathways. Sorafenib is known to promote OS and mitochondrial dysfunction in hepatocellular carcinoma cells [238]. In this context, in 2020, Lin and coworkers reported that METTL3 was down‐regulated in sorafenib‐resistant human hepatocellular carcinomas (HCCs), as well as METTL3 downregulation induced sorafenib resistance in cultured HCC cells in vitro [239]. Interestingly, such effect was tightly linked to reduced m⁶A levels in the 3′‐UTR of the FOXO3 mRNA, whose stability was subsequently impaired [239]. FOXO3 is a Forkhead transcription factor critically required by cells to respond to oxidative stress [240], also via the transcriptional activation of manganese superoxide dismutase (MnSOD, alias SOD2) and catalase (CAT), whose participation in the first line of antioxidant defense ensures adequate ROS scavenging and antioxidant protection within cells [241,242]. Apparently, and on the basis of what has been discussed so far, this may be seen as a counterintuitive finding. However, the authors demonstrated that the downregulated activity of the METTL3/FOXO3 axis induced autophagy in sorafenib‐resistant HepG‐2 cells [239], thereby pointing to the autophagic flux as a novel target to be considered in the context of epitranscriptome-dependent regulation of cellular response to OS-promoting chemotherapeutics. Another clinically relevant OS-promoting tyrosine kinase inhibitor is gefitinib [243]. In 2020, Liu and coworkers confirmed that the METTL3-dependent modulation of the autophagic process was critically involved in the cellular and molecular mechanisms responsible for resistance to gefitinib in lung cancer [244], although also in this case, no endpoint specifically linked to redox-dependent pathways was included in the study, thus leaving unexplored the potential participation of ROS-related signaling pathways to the RNA methylome-dependent re-programming of the autophagic process in cancer resistance. Extending the knowledge on this topic, Chen and colleagues found that resistant non-small cell lung cancer cells could be re-sensitized to gefitinib-induced apoptosis by upregulating the main m⁶A “eraser” FTO, and the levels of m⁶A in myc mRNA were considered as crucial factors in such epitranscriptome-dependent response to this particular TKI [245]. Taking into account the acknowledged function of MYC in the modulation of the epigenetic redox changes that are critical for the development of therapy resistance [246], this finding should be reconsidered to design more in-depth studies and ascertain whether some cancers could respond to gefitinib or other TKIs by re-programming specific redox-related pathways through a complex interplay between methyl-epitranscriptome and MYC-dependent signaling.

As discussed in the sections above, several research reports have revealed that lncRNAs are emerging as key mediators of the response that chemotherapies trigger in various cancer types [247,248]. In 2021, Chen et al. found in hepatocellular carcinoma cells that the upregulation of lncRNA NIFK-AS1 led to an increased resistance to sorafenib, along with enhanced growth and invasion potential, and this effect was mechanistically linked to the METTL3-induced m⁶A hyper-methylation and stabilization of lncRNA NIFK-AS1 [249]. Also in this case, no authors’ interest was evident in evaluating whether the major redox-responsive pathways could be involved in the effects observed. However, Chen et al. identified AKT serine/threonine kinase 1 (AKT1) as a key target gene whose induction by NIFK-AS1could be crucial upon development of drug resistance [249], and this may importantly link the METTL3/NIFK-AS1/AKT1 pathway to the molecular network responsible for redox homeostasis. In fact, some researchers showed that AKT1 crucially participates in critical prosurvival signaling pathways that ensure the restoration of adequate oxidative protection upon OS stimuli [250,251].

Tamoxifen

Tamoxifen (TMX) is often used to treat estrogen receptor-positive breast cancers, even though the development of drug resistance still remains one of the major challenges for TMX-based therapies [252]. TMX is known to generate in breast cancer cells an overproduction of ROS, whose chemical reactivity has been shown to contribute significantly to the drug-induced cytotoxic effects [253]. In 2020, Liu and colleagues reported that TMX-resistant breast adenocarcinoma cells exhibited an increased METTL3-dependent methylation at several m⁶A consensus motif sites in the 5′-UTR of adenylate kinase 4 (AK4) mRNA, and this was associated with the over-expression of its protein product [176]. AK4 is extremely important to maintain energy homeostasis in mitochondria [254], as well as it is a possible indirect activator of NRF2 [255]. In their work, Liu and colleagues found increased ROS concentrations in TMX-resistant cells, compared to their parental counterpart [176]. Unfortunately, only basal ROS levels were studied by the authors, while it would be extremely interesting to verify whether the two cell sublines may respond differently in terms of ROS production and/or scavenging after treatment with TMX. This would contribute to establish whether the epitranscriptome-dependent regulation of AK4 metabolic activity in cancer cells could result in different responsiveness profiles to OS-generating chemotherapeutics. In 2021, the same research group provided evidence that breast cancer cells resistant to TMX had their activating transcription factor 3 (ATF3) protein overexpressed, and this was due to the stabilizing effect of lower levels of m⁶A at residue 131 in the 5′-UTR of ATF3 mRNA [256,p.3]. ATF3 is widely considered as a strong modulator of the inflammatory response and as a powerful protective factor in OS-based conditions, being able to prevent NRF2 degradation [256]. Moreover, ATF3 participates in the unfolded protein response (UPR) [257,258], and it is well known that in the endoplasmic reticulum (ER) stress and oxidative stress reciprocally inter-communicate upon UPR [259,260]. For example, ROS promote chemical modifications of ER stress sensors, which, in turn, activate the NRF2-dependent antioxidative response [261]. Interestingly, Liu and coworkers proposed ATF3 as a downstream effector of AK4 activation [257]3]. On this basis, and despite the absence of a distinctive endpoints related to the redox-dependent landscape in their cancer model, Liu and colleagues have possibly identified a common redox-responsive epitranscriptome-controlled pathway responsible for the resistance of breast cancer to TMX-based antiblastic therapy. Further experimental studies would help researchers to clarify the precise molecular mediators underlying such regulatory pathways.

ITable 1 summarizes the main literature reports and findings on how changes in the RNA methylation status affect resistance to anticancer drugs potentially via interfering with the activity of pathways and regulators of ROS production and metabolism.

Table 1. RNA methylation and cancer resistance traits potentially related to redox rewiring.

Cancer	RNA methylation-related aspect	Anticancer agent	Resistance	Ref.
Pancreas	METTL3 LoF	5-FU	Increased	193
Colon	Increased mAs in p53 pre-mRNA	5-FU	Increased	195
Colon	METTL3-dependent methylation of LBX2-AS1 mRNA	5-FU	Increased	198
Colon	METTL3-dependent methylation of miR‑181d‑5p	5-FU	Increased	199
Breas	Hypermethylated 5′-UTR of miR-149	DOXO	Increased	204
Colon	METTL3-dependent A methylation in p53 mRNA	DOXO	Increased	195
Breast	METTL3-dependent hypermethylation of miR-221-3p	DOXO	Increased	209
Pancreas	METTL3 knockdown	Pt	Reduced	193
Esophagus	NSUN2-mediated hypermethylation of lncRNA	Pt	Increased	215
Lung	METTL3-dependent increase in m^6As in YAP pre-mRNA	Pt	Increased	216
Colon	METTL3-induced m^6A over-methylation of CBX8 mRNA	Pt	Increased	219
Testis	METTL3-induced m6A over-methylation of TFAP2C mRNA	Pt	Increased	221
Lung	miR-4443-dependent inhibition of METTL3	Pt	Increased	223
Leukemia	Hypermethylated cytosines	5-AZA	Increased	229
Ovary	m^6A modifications in FZD10 mRNA	PARPi	Increased	233
Lung	High levels of METTL3 mRNA	TKIs	Increased	237
Liver	Downregulation of METTL3-dependent methylation of As in 3′‐UTR of FOXO3 mRNA	TKIs	Increased	239
Lung	METTL3-mediated regulation of autophagy	TKIs	Increased	244
Lung	FTO-dependent downregulation of m^6A in myc mRNA	TKIs	Reduced	245
Liver	METTL3-induced adenosine hypermethylation of lncRNA NIFK-AS1	TKIs	Increased	249
Breast	METTL3-dependent hypermethylation of adenosines in 5′-UTR of AK4 mRNA	TMX	Increased	176
Breast	Downregulation of m^6A in 5′-UTR of ATF3 mRNA	TMX	Increased	256

Note: Abbreviations: 5-AZA, 5-azacytidine; 5-FU, 5-fluorouracil; 5′-UTR, 5′ untranslated region; DOXO, doxorubicin; LoF, loss of function; mAs, methylated adenosines; Pt, platinum-based compound; TKIs, tyrosine kinase inhibitors; TMX, tamoxifen. For abbreviations not listed here and for details on the specific redox relevance of the molecular targets indicated, see the text.

Radiation-based therapy

Along with chemotherapy, radiation-based therapy also helps clinicians to deal with the cure of several neoplasms, especially in the case of localized tumors, with no migrating cells spread to secondary sites [262]. Moreover, the combined use of radiotherapy and chemotherapy is still considered an effective approach to treat many solid tumors [263]. It is worthy of note that important cytotoxic effects of radiotherapy are mediated by the overgeneration of ROS, with the consequent activation of intracellular OS being a powerful proapoptotic challenge for tumor cells [264,265]. Therefore, and on the basis of what we discussed so far, it would be reasonable to suspect that the altered methylation patterns in RNAs could respond differently to radiation-based cancer therapies, as well as it could be possible to hypothesize that cancer cells may respond adaptively to radiotherapies by regulating the status of RNA methylome. In this regard, the role of N6-methyladenosine metabolism on tumor and cancer response to radiation-based stress was recently investigated by Kowalski-Chauvel and coworkers, who found that stem cells from human glioblastoma biopsies could be sensitized to radiotherapy by downregulating ALKBH5, one of the main m⁶A “erasers” that remove N6-methylation from adenosines in RNA [266]. Interestingly, such effect was associated with an orchestrated downregulation of genes specifically involved in DNA homologous recombination repair. Accordingly, in glioblastoma cells with alkbh5 silenced, the irradiation (IR)-dependent increase in DNA damage persisted even 24 hours post-IR, thus suggesting that radiotherapy efficacy could be improved by repressing ALKBH5 expression [266].

Future perspectives

As discussed above, the activity of several research teams has been increasing tremendously over the last two decades to clarify whether and how the RNA methylation diversity may serve as a key driver for the development of resistant traits in tumor and cancer cells.

Cancer stem cells (CSCs)

Particular attention should be paid to the role of RNA methylation in reshaping the tumor microenvironment, particularly in terms of resident and circulating CSCs. CSCs are a subpopulation of quiescent self-renewing tumorigenic cells that occupy tumor niches in most neoplasms and hematopoietic malignancies [267]. CSC subsets within neoplasms and cancers undoubtedly represent a serious challenge for oncology specialists, since a major driver of resistance to therapy resides in the genetic and nongenetic heterogeneity that characterizes CSCs. Such heterogeneity leads to phenotypic differences that often reduce the susceptibility to several anticancer therapies [268,269]. More importantly, chemotherapeutic drugs often enrich cancers for CSC population via selection of clones that are resistant to treatment [270]. Some interesting evidence has been provided regarding the emerging role of deregulated RNA methylation processes in CSCs, with important consequences on cancer biology, such as self-renewal and survival. In particular, m⁶A methylation signature has been identified as a clinical biomarker for CSC generation and cancer progression [271], with m⁶A promoting the translation of diverse specific oncogene products, such as the epidermal growth factor receptor (EGFR) and the transcriptional co-activator with PDZ-binding motif (TAZ) [272]. In 2020, Gao and coauthors showed that m⁶A levels in RNA were higher in bladder cancer stem cells (BCSCs), compared to bladder cancer cells with no stemlike phenotype [273]. In addition, the authors demonstrated that METTL3 was essentially required for self-renewal of BCSCs, via a pathway that upregulates m⁶A levels, along with the expression of AF4/FMR2 family member 4 (AFF4), thus increasing the transcription of SOX2 and MYC [273]. RNA adenosine demethylases also seem to affect stemness-related molecular factors in cancer. In fact, in 2020, some Indian researchers found in cisplatin-resistant squamous carcinoma cells an ALKBH5-dependent decrease in m⁶A levels within the nascent transcripts of FOXM1 and NANOG, and such regulation was linked to increased resistance to cisplatin [274]. FOXM1 and NANOG are two major stemness markers, as well as these two proteins are the main regulators of aggressiveness, invasive properties, and chemoresistance in several cancers [275]. [276–278]Very recently, Yang et al. found that a stem cell-like phenotype can be promoted in liver cancer by upregulating the levels of m⁶A modifications in the alpha-1,6-mannosylglycoprotein 6-beta-N-acetylglucosaminyltransferase mRNA, thus enhancing transcript stability [279]. In the same year, Wang and colleagues explored the role of m¹A methylation in tRNA in determining profound changes in cancer stem cell biology. More in detail, the authors found a TRMT6/TRMT61A-mediated m¹A hypermethylation in tRNA of liver CSCs, and this was mechanistically linked to overexpression of PPARδ and increased cholesterol biosynthesis, along with a dramatic activation of Hedgehog (HH) signaling [280]. HH targets numerous genes whose differential activation is important for stem cell maintenance, proliferation, and successful embryonic development, however, altered HH signaling has been associated with the development and progression of different cancer types [281]. Finally, Liu and coworkers have recently examined whether the methyl-RNA landscape could play a role in CSCs derived from lung cancer, finding that high levels of the p53-dependent up-regulation of m⁶A demethylase ALKBH5 in CSCs support the overexpression of critical stem hallmarks, such as NANOG and OCT4 [282]. Unexpectedly, this field of research is still underinvestigated. However, given the potential importance of this research topic in terms of translation to clinic, much greater effort should be devoted to clarify in depth whether and how the reprogramming of RNA methylation could affect CSCs biology in terms of aggressive phenotype and responsiveness to treatment.

Immunotherapy

Another promising field is represented by the application of techniques that modulate RNA methylation states with the aim of increasing the efficacy of cancer immunotherapy, which today significantly prolongs the survival of many patients affected by otherwise fatal cancers. The hypothesis to use the physiological capacity of the immune system as a tool to exploit to treat neoplasms appeared in researchers’ minds in the nineteenth century, as reviewed by Oiseth and Aziz [283], but only in the last decades specific knockout mouse models and rapidly advancing molecular and biochemical techniques have led to the identification of tumor-specific immune responses to use against tumors [4,284,285]. In this context, Li and colleagues brilliantly demonstrated that deletion of the m⁶A demethylase ALKBH5 was sufficient to sensitize tumors to cancer immunotherapy, eventually prolonging mouse survival [286], thus paving the way to potentially ground-breaking innovations for cancer medicine and healthcare providers. The capacity of RNA methylation to influence cancer response to immunotherapy was recently confirmed by Zhang and coworkers, who found that m⁶A levels in RNA could regulate the expression of metadherin [287], a well-established oncogene whose upregulation promotes EMT in various cancers [288,289]. Very recently, Yin and colleagues confirmed the idea that key enzymes for m⁶A RNA metabolism could represent a potential therapeutic target for tumor immunotherapy, finding that the loss of METTL3 led to increased tumor growth and metastasis via activation of NF-kB and STAT3 [290], with NF-kB and STAT3 being important players in cancer cell-induced immunosuppression [291]. Bioinformatics analyses and experimental data from other researchers supported the notion that METTL3-dependent signaling may serve as an important immune evasion mechanism for tumors [292].

Novel molecular pathways

As mentioned in the previous sections, the reorganization of the autophagic process and metabolic rewiring play important roles in cellular differentiation, resistance to stressors, and adaptation to (micro)environmental stimuli.

Certainly, sirtuins (SIRTs) serve as key redox-active regulators of autophagic flux and metabolic reprogramming, also due to the reliance on NAD+ for their de-ac(et)ylase activity toward multiple protein targets [293]. In fact, today the role of SIRTs in cancer resistance to therapy is acknowledged and intensely studied. In this context, to the best of our knowledge, no investigation has attempted so far to establish whether the methyl epitranscriptomic control of cancer cell response to chemotherapeutics or other anticancer interventions could be implemented by modulating the activity of sirtuins in mammals. This is a serious lack of knowledge, also taking into account that some researchers recently found that knockdown of METTL14 facilitated autophagy and alleviated apoptosis and inflammation in stressed podocytes in vitro, and such effect was mechanistically due to the upregulation of SIRT1 via m⁶A hypo-methylation of sirt1 mRNA [294]. Interestingly, SIRT1 modifies the activity of ATP-binding cassette (ABC) transporters in cancer cells, thus significantly affecting drug penetration properties [295]. In addition, SIRT1 is able to confer anti-apoptotic advantages in cancer cells via deacetylation of p53/p73, thus repressing p53/p73-dependent programmed cell death pathways [296,297]. Moreover, SIRT1 relocates to the nucleus to promote repair and cell survival upon DNA damage-promoting stress [298]. Finally, SIRT1 is a well-known activator of antioxidant protection in OS-promoting conditions, such as the ischemia/reperfusion [299]. On such a basis, it is evident that further and more targeted studies should be carried out to provide detailed information on how the adaptive responses of resistant cancer cells toward anticancer interventions could be deployed by an interplay between the methyl epitranscriptome and the SIRT-based stress-responsive molecular network.

Another pathway that is largely underinvestigated is that responsible for the clearance of methylglyoxal (MG), which is an endogenous cancer-static and glycating α-oxoaldehyde [300]. More specifically, MG is frequently overproduced by the increased glycolytic flux in tumor and cancer cells (i.e. Warburg effect), and this requires neoplastic and malignant cells to strongly upregulate the activity of the glyoxalase system [301,302]. The process of MG enzymatic removal is tightly interconnected with both redox homeostasis and antioxidant protection, mainly because glyoxalase 1 (GLO1) essentially requires glutathione to convert MG into the less toxic S-D-lactoylglutathione [303]. Accordingly, GLO1 was reported as highly overexpressed in the apoptosis-resistant human leukemia cells [304]. In addition, the knockdown of GLO1 sensitized adenocarcinoma cells toward irradiation-induced apoptosis [305]. Moreover, the inhibition of GLO1 was found to improve the antiproliferative effect of sorafenib in carcinoma [306]. On such a basis, it would not be unreasonable to hypothesize that cancer-resistant phenotypes may also develop via an epitranscriptome-dependent upregulation of glyoxalases. Surprisingly, it would seem that no study has ever attempted to verify this so far. However, Kulkarni and coworkers recently showed that alkbh7 null mice had elevated levels of GLO1 [307], thus providing a proof-of-concept that at least one of the Fe(II)/α-ketoglutarate-dependent dioxygenases that are able to de-methylate RNA can serve as important regulators of the MG-targeting system. Clearly, this should be further investigated with more specifically targeted experiments, especially in tumor and cancer cells. However, a better comprehension of such phenomena could help researchers and oncologists to circumvent or prevent the development of resistant traits in cancer cells upon antiblastic treatment.

Innovative technologies

An in-depth exploration of the role of the epitranscriptome in physiology and disease is hindered by the limited availability of techniques for accurate mapping of specific RNA modifications throughout the transcriptome. Indeed, conventional methods for RNA modifications-targeting analysis require great amounts of material and are relatively labor intensive. This is particularly true for the detection and quantitation of small RNAs, such as microRNAs (miRNAs) and Piwi-interacting RNAs (piRNAs) [308]. On this basis, over the last few years, some truly inspired researchers have presented convenient, sensitive, and reliable methods for quantifying a wide range of RNA modifications. In the standard methylated RNA immunoprecipitation and sequencing (MeRIP-Seq) method, m⁶A-specific antibodies are used to immunoprecipitate RNA, then post-reverse-transcription cDNA is deep sequenced providing high-resolution reads of m⁶A-methylated RNA [54]. Some improvements have also been developed to circumvent typical limitations of MeRIP-Seq, such as PA-m⁶A-seq, miCLIP, and m⁶A-CLIP [309]. However, some authors have recently proposed a novel high-throughput next-generation sequencing (NGS)-based approach (i.e. RiboMethSeq method) [308], whereas others have presented a technique that combines oxidative deep sequencing and stem-loop RT-qPCR to quantify accurately the levels of 3′-terminal 2′-O-methylated small RNAs in biological samples [310]. Another interesting method was recently developed by US-based researchers, who used a hydrazine-aniline cleavage sequencing (HAC-seq) to detect and profile the RNA m³C methylome at single-nucleotide resolution [311]. In addition, great interest is emerging for developing GC-MS/MS and LC-MS/MS to identify epitranscriptomic biomarker signatures that could help cancer diagnosis, detecting modified nucleosides by monitoring multiple reactions [312]. Very recently, Cui and coworkers proposed a sensitive platform for simple simultaneous detection of both m⁶A and FTO protein. This method is based on the FTO-dependent conversion of the methyl group of m⁶A to hydroxymethyl, whose covalent reaction with dithiotreitol allows the interaction with a Cd-based electrode thus generating a detectable photocurrent [313]. Further development of such innovative methods may help to circumvent the great limitations and shortcomings that still impede or delay the progress of biomedical research, thus allowing future investigations to follow complex patterns of RNA methylation, to establish and validate specific epitranscriptomic signatures, and to study how the dynamics of RNA methylation landscape could lead to different responses to stressors, including cancer chemo- and radio-therapy. This would certainly open new interesting perspectives in terms of personalized anticancer medicine.

Bioinformatics

In parallel with the escalating impact of cancer as a major global burden to human health, the interest in classifying high or low responders and in predicting the effect of chemotherapy in the perspective of personalized medicine, has been increasing formidably [314]. However, with the advent of new technologies and multi-omics-based approaches, great amounts of data have started to overwhelm the analysis capacity of biomedical researchers, thus urging innovative techniques and methods for data handling and processing. 21^st century technology and artificial intelligence (AI) may serve as a game-changer and powerful tool for this task. In this regard, in 2019, the novel approach of some French researchers allowed to develop the GECKO algorithm, which in this perspective could pave the way to the exploration of large amounts of high-throughput sequencing data to identify new sequences of interest without the intensive use of highly specialized bioinformatics workflows [315]. Three years later, Sui and colleagues created machine learning-based models that helped to identify and validate in the lab some important genes and miRNA related to the IC50 value of cisplatin [316]. Even though no gene or miRNA was clearly linked to methyl RNA metabolism, Sui and coauthors proposed a new workflow for future investigations, thus contributing to bridge the gap that still exists between bioinformatics and translational research. In this perspective, clinicians may benefit from multi-omics-based approaches, and the therapeutic arsenal of multimodal therapies would be greatly increased in view of highly personalized medicine.Translational research

RNA methylation is being explored not only by basic researchers, but also by scientific teams that are more oriented to translational research. In fact, some researchers proposed that much greater efficacy of antisense-based clinical approaches would be obtained by using methylated versions of anti-miRNA oligonucleotides. More specifically, Nahar and coworkers found that 2’-O-methyl RNAs directed against miR-21, an oncogenic miRNA whose increased activity is linked to a more aggressive behavior, was able to reduce migration capacity in metastatic adenocarcinoma cells [317]. Although being a preliminary report, this kind of approach clearly represents a suggestive indication that studies on methyl-RNA-based gene regulation in the field of oncology might be exploited in the future to improve cancer disease control or even to prevent the development of therapy resistance.

A schematic representation of the future directions research may take to reveal if and how the regulation of methylepitranscriptome could affect cancer sensitivity to treatments, especially in terms of redox-related cellular adaptive response, is reported in Figure 2.

Figure 2. Schematic representation of the possible directions that future research activities and efforts may take to reveal in detail the role of the methylepitranscriptome in affecting or determining cancer cell response to treatments, also in terms of potential participation of redox-sensitive and redox-dependent biomolecular mechanisms.

Conclusions

Over the last forty years, biomedical research has been focused on revealing the genetic and epigenetic nature of resistance traits in tumors and cancers. Nevertheless, only in the last two decades the importance of modifications and structural rearranging of RNA has gained the attention of the most important molecular oncologists and biologists worldwide. More specifically, the literature offers substantial evidence that the RNA methylome is a powerful biological tool through which cancer cells achieve a fine-tuned control of gene expression in response to anticancer treatments. Such rewiring of gene expression-related processes are realized by operating and modulating a complex molecular network of enzymes responsible for “writing”, “reading” and “erasing” methylated residues within substantially all classes of RNA, both coding and noncoding. However, it is surprising that a small number of investigations attempted to study in-depth whether the epitranscriptome-directed adaptations underlying the acquired and/or innate resistance traits in cancer could be deployed or arranged through the involvement of redox-dependent or -responsive signaling pathways. Somehow, this is curious since the effectiveness of many, if not most, antiblastic approaches available today relies on the powerful generation of reactive oxygen species within cancer and tumor cells. This renders the existing knowledge gap even more unexpected, since the alteration of the redox milieu, as well as the functional reprogramming of mitochondria, which represents a major critical determinant for cellular oxidative metabolism and a contributor to intracellular ROS formation, have already been documented as key mediators of the RNA methylome-mediated response to OS. In addition, many recent evidence suggested that a number of molecular pathways that are regulated by the RNA methylome are deeply linked to ROS-targeting enzymatic clearance and DNA repair, as well as to the expression and localization of transcription factors whose downstream signaling crucially governs the activity of key antioxidant protection systems.

On this basis, from a critical review of the existing literature corpus, we believe that a greater effort should be spent to explore and verify whether the activation of redox-active pathways may underlie the methyl epitranscriptome-controlled development of resistance to therapy in neoplasms and malignancies. Also taking into account the new high-throughput techniques and AI-based methods available today and possibly achievable in the next future, we would suggest that all researchers interested in this topic and most active and dedicated redox biologists should consider to dedicate more effort to explore this field more in depth, essentially because this topic may still reserve the possibility of breakthrough discoveries. Such striking innovations may help future biomedical researchers to understand in detail how cancer cell response unfolds upon treatment, as well as clinicians and oncologists may benefit from these advances to develop new therapeutic approaches and strategies, with the specific aim of delaying or preventing the severe problem of refractory cancer and tumor recurrence.

Disclosure statement

The authors declare that there is no relevant financial or non-financial conflict of interest to report.

Additional information

Funding

This work was financially supported by the competitive grant for Basic Research of the University of L’Aquila (D.R. no. 649/2022) to SF, and by the Associazione Italiana per la Ricerca sul Cancro (AIRC) fellowship (#25432) to MP for salary support.