Roles of NRF2 in DNA damage repair

Abstract

Purpose

The transcription factor NF-E2-related factor 2 (NRF2) is a master regulator widely involved in essential cellular functions such as DNA repair. By clarifying the upstream and downstream links of NRF2 to DNA damage repair, we hope that attention will be drawn to the utilization of NRF2 as a target for cancer therapy.

Methods

Query and summarize relevant literature on the role of NRF2 in direct repair, BER, NER, MMR, HR, and NHEJ in pubmed. Make pictures of Roles of NRF2 in DNA Damage Repair and tables of antioxidant response elements (AREs) of DNA repair genes. Analyze the mutation frequency of NFE2L2 in different types of cancer using cBioPortal online tools. By using TCGA, GTEx and GO databases, analyze the correlation between NFE2L2 mutations and DNA repair systems as well as the degree of changes in DNA repair systems as malignant tumors progress.

Results

NRF2 plays roles in maintaining the integrity of the genome by repairing DNA damage, regulating the cell cycle, and acting as an antioxidant. And, it possibly plays roles in double stranded break (DSB) pathway selection following ionizing radiation (IR) damage. Whether pathways such as RNA modification, ncRNA, and protein post-translational modification affect the regulation of NRF2 on DNA repair is still to be determined. The overall mutation frequency of the NFE2L2 gene in esophageal carcinoma, lung cancer, and penile cancer is the highest. Genes (50 of 58) that are negatively correlated with clinical staging are positively correlated with NFE2L2 mutations or NFE2L2 expression levels.

Conclusion

NRF2 participates in a variety of DNA repair pathways and plays important roles in maintaining genome stability. NRF2 is a potential target for cancer treatment.

Introduction

DNA lesions occur at a rate of 10,000 to 1,000,000 molecular lesions per cell every day [1]. Unrepaired or improperly repaired DNA damage can lead to severe genomic aberrations or mutations that alter cell proliferation or affect cell survival [1]. DNA damage can consist of double-stranded breaks (DSBs) and single-stranded breaks (SSBs), chemical modifications such as oxidation and methylation, formation of apurinic/apyrimidinic (AP) sites, intra- and inter-strand crosslinks, and protein-DNA adducts. All of these events activate a network of events collectively referred to as the DNA damage response (DDR) [2, 3]. The DDR is a collection of mechanisms that includes DNA damage recognition, checkpoint activation, cell cycle arrest, and three final outcomes of repair, apoptosis, or immune clearance [3, 4]. These mechanisms mainly consist of direct repair, base excision repair (BER), nucleotide excision repair (NER), mismatch repair (MMR), non-homologous end joining (NHEJ), and homologous recombination (HR) [1, 3].

NF-E2-related factor 2 (NRF2), encoded by NFE2L2 gene, is a member of the cap ‘n’ collar (CnC) family of basic leucine zipper transcription factors [5]. NRF2 consists of seven NRF2-ECH homology (Neh) domains [6]. Among them, Neh1 is the CnC-leucine zipper (bZIP) domain that interacts with small musculoaponeurotic fibrosarcoma (sMAF) proteins. Neh2 possesses DLG and ETGE motifs that interact with the C-terminal Kelch domain of Kelch ECH-associated protein 1 (KEAP1) [6, 7]. Under unstressed conditions, KEAP1 dimers bind to NRF2 and accelerate its ubiquitination, resulting in its proteasomal degradation and low-level intracellular expression [8]. When exposed to electrophilic chemicals or reactive oxygen species (ROS), KEAP1 is inactivated. Newly synthesized NRF2 can then escape capture by KEAP1, becoming stabilized and translocated into the nucleus to activate its downstream genes [8]. Induced NRF2 is widely involved in essential cellular functions such as redox homeostasis, metabolism, cell survival, proliferation, autophagy, and proteasomal degradation [6]. In recent years, NRF2 has been proven to participate in a variety of DNA repair pathways and play important roles in maintaining genome stability [5, 6, 9].

The roles of NRF2 in DNA damage repair

NRF2’s primary contributions to DNA damage repair consist of its functions in genome integrity maintenance, cell cycle regulation, and antioxidation. By directly or indirectly regulating DNA repair proteins, NRF2 is widely involved in a variety of DNA repair pathways and maintains genomic integrity. In addition, when DNA is damaged, NRF2 also causes G2 or G1 arrest, thereby buying time for different DNA repair pathways [10, 11]. It’s worth noting that NRF2-mediated antioxidant pathways and DNA damage repair pathways often work together to protect DNA [12, 13]. However, there are not many studies coupling the antioxidant function of NRF2 with DNA repair. When the antioxidant function of NRF2 is inhibited, it influences the expression of DNA repair proteins, which may affect DNA repair [14].

NRF2 is mainly involved in the regulation of DNA damage repair by regulating gene transcription through antioxidant response elements (AREs). Nuclear-localized NRF2 binds to AREs in the promoter regions of a variety of genes to activate their transcription [15]. AREs are highly versatile. They can be found located in the promoter region of coding and non-coding genes, in the same strand of the genes and in the opposite strand, and promote as well as inhibit gene expression [16–18]. The core ARE consensus amino acid sequence was initially defined as 5ʹ-TMAnnRTGAYnnnGCRwwww-3ʹ (where the ‘core’ is highlighted in bold, and ‘n’ is used to denote residues considered to be redundant) [16]. A subsequent study defined 5ʹ-gagTcACaGTgAGtCggCAaaatt-3ʹ as the complementary nucleotide sequence, with the two cytosine residues in bold being essential for enhancer activity and the two guanine residues shown in bold being dispensable [16]. Table 1 summarized currently reported DNA repair genes containing AREs. NRF2 can interact with many other transcription factors, including MAF-F, -G, and -K, BACH1, ATF4, c-Fos, and c-Jun [15, 19], but the research on the interaction between NRF2 and DNA repair-related proteins such as ATM and Rad3-related (ATR), is still in its infancy [10]. In Fig. 1, we described the roles of NRF2 in DNA repair in both cancer and non-cancer studies.

Table 1.

AREs of DNA repair genes

Gene	Species	Sequence	Position	Description of AREs	Description of genes/products	References
NFE2L2	Murine	TGACTCCGC	−574 to −403	a perfect ARE	transcription cis-regulatory region binding; transcription coregulator binding; response to oxidative stress	[95]
	Murine	TGACTGTGGC	−848 to −684	has one more base before the GC box
MGMT	Human	-	617 to 595	AP1 recognition sequence overlaps with the ARE	DNA binding; methyltransferase activity; DNA dealkylation involved in DNA repair; positive regulation of double-strand break repair	[18]
	Human	-	584 to 562	-
OGG1	Human	-	−29	the binding site is “TTCC”	nucleotide-excision repair; negative regulation of double-strand break repair via single-strand annealing; damaged DNA binding; endonuclease activity; base-excision repair, AP site formation; 8-oxo-7,8-dihydroguanine DNA N-glycosylase activity	[96]
FEN1	Human	TGGCGGAGC	−95 to −86	on the reverse strand, harbors a “GG” in place of a “GA” motif	damaged DNA binding; double-stranded DNA exodeoxyribonuclease activity; double-strand break repair via homologous recombination; base-excision repair, gap-filling; positive regulation of sister chromatid cohesion	[16]
ATF3	Human	TGACACAGC	−3406 to −3315	-	DNA-binding transcription factor activity, RNA polymerase II-specific	[42]
	Human	TGACAATGC	−3406 to −3315	-
RAD51D	-	GTGACTCAGCAG	15,822 to 15,835	-	DNA binding; ATP-dependent DNA damage sensor activity; double-strand break repair via homologous recombination; regulation of cell cycle; telomere maintenance via recombination	[56]
RAD52	-	GTGACTCAG	1819 to 1827	-	DNA double-strand break processing involved in repair via single-strand annealing; regulation of nucleotide-excision repair; DNA recombination;	[56]
MDM2	Murine	TGACTCAGC	−360	located in the first intron, capable to affect promoter activity	regulation of cell cycle; ubiquitin-dependent protein catabolic process; p53 binding	[57]
	Murine	-	−1699	in 50-flanking upstream from the first exon
	Murine	-	−1801
	Murine	-	−2316
	Murine	-	−2709
DMC1	-	GTGACTCAGCA	6614 to 6624	-	DNA strand exchange activity; double-stranded DNA binding; chromosome organization involved in meiotic cell cycle; double-strand break repair involved in meiotic recombination; homologous chromosome pairing at meiosis	[56]
SHFM1	-	TGACTCAGCAG	6108 to 6118	-	double-strand break repair via homologous recombination;	[56]
RBBP8	-	GGTGACTCAGCA	1878 to 1867	-	single-stranded DNA endodeoxyribonuclease activity; double-strand break repair via homologous recombination; DNA double-strand break processing involved in repair via single-strand annealing; mitotic G2/M transition checkpoint; DNA strand resection involved in replication fork processing	[56]
XRCC3	-	GGTGACTCA	2406 to 2398	-	four-way junction DNA binding; crossover junction DNA endonuclease activity; double-strand break repair via homologous recombination; interstrand cross-link repair; double-strand break repair via synthesis-dependent strand annealing	[56]
BRCA1	-	-	−218 to −210	-	regulation of cell cycle; double-strand break repair via homologous recombination; postreplication repair	[69]
53BP1	Human	AGTGACTCAGCT	Chr15: 43,808,509–43,808,520	on the reverse strand	double-strand break repair via nonhomologous end joining; negative regulation of double-strand break repair via homologous recombination; DNA damage checkpoint signaling; site of double-strand break; transcription coregulator activity	[58]
	Human	TTGCTGAATCAC	Chr15: 43,818,989–43,819,000	on the forward strand
	Human	CTGCTGAGTCCT	Chr15: 43,822,481–43,822,492	on the forward strand
SQSTM1	Human	GCTGCTGAGTCACGCTT	-	conserved in human, mouse, rat, and elephant species	protein serine/threonine kinase activity; selective autophagy	[97]
	Mouse	ACTGCTGAGTCACGCTT	-
	Rat	GCTGCTGAGTCATGCTT	-
	Elephant	GCTGCTGAGTCATACGC	-
MIR140	-	-	−7844 to −7835	-	miRNA-mediated gene silencing; RISC complex	[17]

“-“ means not mentioned in the article, description of genes/products is referenced from the GO database (http://geneontology.org/).

Fig. 1.

Roles of NRF2 in DNA Damage Repair. The left side shows the roles of NRF2 in DNA repair reported in non-cancer studies, and the right side shows the roles of NRF2 in DNA repair reported in cancer studies. The upper part shows the activation modes of NRF2 involved in DNA repair, while the lower part shows the mechanism of NRF2 involved in BER, NER, MMR, NHEJ, and HR

CNC proteins also include Nuclear Factory Erythroid derived 2 (NFE2), NRF1, and NRF3 [20]. Among them, there has been no research exploring the role of NRF3 and NFE2 in DNA repair, and a small amount of literature has reported the role of NRF1 in DNA repair. NRF1 can also bind to AREs and selectively bind to estrogen response elements (ERE) [18]. There is an ERE located < 300 bases upstream of the 5ʹ end of DNA repair gene O6-methylguanine DNA methyltransferase (MGMT) exon 1, which NRF1 uses as a recognition site [18]. The binding of NRF1 to the MGMT promoter is highly evident in MGMT rich cell lines [18]. xeroderma pigmentosum group C (XPC) is a key factor in initiating global genome NER by identifying DNA damage and recruiting downstream factors [21]. NRF1 enhances the expression of XPC by increasing the availability of glutathione [21]. Adding XPC or glutathione can restore the DNA repair ability of NRF1 inhibited cells [21].

NRF2 participates in direct repair

Direct repair of damaged DNA usually relies on a single protein to remove DNA damage, which is simple and less prone to errors, and has efficiency in maintaining genetic information [1]. MGMT can directly remove O6-alkylated adducts in DNA in a one-step reaction, which is one of the three main direct DNA repair mechanisms [22, 23]. This reaction transfers an alkyl group from the O6 position of guanine to a cysteine residue within the active site pocket of MGMT, thereby restoring guanosine to its undamaged state [22, 23]. Al-Obaide et al. found two AREs between positions 617–595 and 584–562 of the MGMT promoter, and found that enhanced NRF2 expression significantly increased MGMT mRNA and protein levels in glioma cells [18]. Xu et al. found that the guanylate binding protein 3 (GBP3)/stimulator of interferon genes (STING)/ sequestosome 1 (SQSTM1, p62)/ NRF2 axis plays a role in temozolomide (TMZ)-induced DNA damage, with NRF2 promoting MGMT expression and then promoting TMZ resistance [24]. Their findings indicated that TMZ induces the upregulation of GBP3 and STING, and that the interaction between them may be involved in preventing TMZ-induced p62 degradation, leading to upregulation of p62 protein and mRNA levels in glioblastoma cells [24]. NRF2 overexpression can, in turn, increase the expression of MGMT, GBP3, and STING and thus reduce the DNA damage caused by GBP3 downregulation [24].

NRF2 participates in excision repair

Excision repair includes BER, NER, and MMR with multiple sub-pathways [25–27]. NRF2 is mainly involved in excision repair through its function of genome integrity maintenance, which promotes or inhibits different molecules related to BER, promotes NER, and inhibits MMR [16, 28–30].

NRF2 in BER

BER is a highly conserved mechanism for dealing with oxidative damage generated by respiration, natural hydrolysis, and alkylation reactions, all of which occur in each cell many thousands of times a day [31]. DNA base damage is first recognized by DNA glycosidases, which excise the damaged base to form apurinic/apyrimidinic (AP) sites [32]. The endonuclease then cleaves the DNA backbone to form a SSB with a 3’ -hydroxyl group [32]. DNA polymerase removes 5’-deoxyribosephosphate (dRP) and inserts the correct undamaged nucleotide into the repair gap, and then DNA ligase seals the gap [32]. In humans, at least 30 proteins are involved in short-patch BER (SP-BER)—the BER that removes a single damaged base—and long-patch BER (LP-BER), the BER that synthesizes 2–8 nucleotides to replace the damaged region [31]. The main way for NRF2 to regulate BER is to combine with AREs of BER-related genes to regulate their expression [6].

8-oxoguanine DNA glycosylase 1 (OGG1) acts as both a glycosidase and DNA endonuclease, enabling it to not only excise 7,8-dihydro-8-oxoguanine (8-oxoG) and formamidopyrimidine (Fapy-G), but also cleave the DNA backbone [32, 33]. The OGG1 promoter region contains AREs, which can be recognized and regulated by NRF2 [34]. Habib et al. found that mammalian target of rapamycin complex 1 (mTORC1) and mTORC2 negatively regulate OGG1 by inhibiting NRF2 expression. mTORC1 and 2, in turn, are themselves respectively inhibited or activated by AMPK and AKT, thereby respectively activating or inhibiting DNA repair [28]. Moreover, tuberin, inhibited by AKT and activated by AMPK, has an inhibitory effect on mTORC1 [28]. mTORC1 can be activated by elevated glucose levels, which induce the phosphorylation-mediated inactivation of tuberin protein through the PI3K/AKT pathway and reduce the phosphorylation-mediated activation of AMPK, leading to the activation of mTORC1 and the impairment of DNA repair [28]. In addition, it was observed in the experiment that promoter activity and protein expression were not completely parallel, suggesting that post-transcriptional and post-translational modification may influence the expression of NRF2 and OGG1 [28]. Studies have shown that the polyphenolic compound 7,8-dihydroxyflavone (DHF) or purified Feverlife extract treatment can induce BER and other DNA repair pathways through the PI3K-dependent NRF2/ARE pathway [34, 35]. This indicates that, under different stimulatory conditions, the regulation of AKT on NRF2/OGG1 is bidirectional. Other drugs, such as antioxidants, vitamin C, and butylated hydroxyanisole (BHA), can also regulate NRF2/OGG1 to promote BER [36].

When 5’-dRP or other factors become resistant to the dRP lyase activity of polymerase β (Polβ), LP-BER is activated. This process produces a 5’-flap structure, which is produced by flap endonuclease-1 (FEN1) resection [32, 37]. On the reverse strand of the FEN1 promoter region, from −95 bp to −86 bp, there is an ARE-like sequence with the sequence TGGCGGAGC. NRF2 can bind to this and down-regulate the expression of FEN1 [16]. This may be because the ARE-like sequence in the FEN1 promoter – which contains a “GG” substitution for the previously reported “GA” motif forming the core ARE sequence – is not a precise ARE sequence [16]. Moreover, other transcription factors, such as Ets-like protein (ELK) or c-Ets, can also bind to the ARE-like sequence. NRF2 may interact with these transcription factors to mediate the downregulation of FEN1 expression [16]. Notably, there is an ARE within the MIR140 promoter region (−7844/−7835). Ionizing radiation (IR) promotes NRF2 nuclear translocation and subsequent MIR140 transcriptional activation in human lung fibroblasts [17]. miR-140 can impede DNA repair by targeting FEN1 [38]. This may also be one of the mechanisms by which NRF2 regulates the downregulation of FEN1.

NRF2 in NER

NER is capable of eliminating the most extensive structure-independent DNA lesions [39]. NER that deals with lesions spanning the entire genome is called global genome repair (GGR), whereas NER that repairs template strand lesions in transcribed regions is called transcription-coupled repair (TCR) [40, 41]. NER is divided into three major steps: damage recognition, which includes initial recognition and damage verification; dual incisions and release of excision products; and gap filling, which includes repair synthesis and DNA junction [41].

Liu et al. found that NRF2-regulated genes were induced after sulforaphane (SF) treatment in vivo, followed by a doubling of the repair activity against pyridyloxobutyl (POB)-DNA adducts in mouse liver extracts [29]. This means that NRF2 may increase repair activity by regulating NER proteins [29]. However, the amount of p53 protein and the level of histone H3 protein acetylation (H3ac) in the extract were not significantly affected, and the binding of XPA, B, and C to DNA was not significantly changed [29]. This suggests that changes in NER activity may be related to NRF2’s activation of other proteins outside the scope of the study, or its effect on post-translational modifications of NER-associated proteins [29].

Activating transcription factor 3 (ATF3), which binds to p53 and stabilizes p53 protein, is required for DNA repair [42]. Cui et al. proposed that ATF3 may promote UV-induced H3ac and subsequently increase the accessibility of damaged DNA to NER proteins by promoting the interaction between p53 and p300, leading to the enhancement of p53-mediated DNA repair capacity [43]. Wei et al. identified two putative AREs, TGACACAGC and TGACAATGC, at positions −3406 to −3315 of the proximal promoter of the ATF3 gene [42]. They found that ZnO nanoparticle treatment significantly induced ATF3 expression in human bronchial epithelial cells through ROS/NRF2 signaling [42].

NRF2 in MMR

The MMR system is primarily responsible for error monitoring and correction during DNA replication, repair, and recombination [44]. MMR can be divided into three steps: mismatch recognition, excision, and resynthesis [45]. During DNA replication, the heterodimer MutSα of MutS homolog 2 (MSH2)-MSH6 is loaded onto DNA to look for mismatches [46]. MutLα subsequently cleaves DNA with its endonuclease activity to form SSBs [45, 46]. Then, exonuclease 1 (Exo1) is recruited to the SSB and excises the discontinuous strand [46]. Finally, Pol δ fills the gap and Lig I catalyzes nick ligation [46].

A certain concentration of ROS can promote cell growth and activate the cellular MMR system [30]. When tumor cells are stimulated by chemotherapy, NRF2, as an important transcription factor against oxidative stress, can significantly inhibit the generation of ROS and inhibit MMR [30]. However, Liu et al. found that ROS levels in NRF2-overexpressing cells had no significant effect on MSH2 expression, suggesting that high expression of NRF2 inhibits MSH2 in a ROS-independent manner [30]. NRF2 suppresses the expression of MSH2 by activating the JNK/c-Jun signaling pathway, leading to MMR deficiency and increasing tumor mutational burden, which can induce gene instability-dependent drug resistance [30].

NRF2 participates in DSB repair

DSBs, including double-ended DSBs and single-ended DSBs, can be formed by topoisomerase inactivation, transposon excision, IR, site-specific endonucleases, and problems in replication [47]. The two main DSB repair mechanisms are HR and NHEJ [48]. The PI3K-related kinase (PIKK) family, which includes ataxia telangiectasia mutated (ATM), ATR, and DNA-dependent protein kinase (DNA-PK), drives the DDR signal cascade [10, 49, 50]. Among them, ATM can promote both HR and NHEJ, ATR mainly promotes HR, and DNA-PK mainly promotes NHEJ [49, 50]. All three kinases are able to phosphorylate S139 at the carboxyl terminus of the histone variant H2AX to form the γH2AX chromatin domain, which is one of the earliest events following injury [50]. γH2AX subsequently diffuses to several megabases around the DSB, initiating an interaction network around γH2AX that coordinates the retention of many DDR proteins at the DSB site [50, 51].

The p62/NRF2 axis maintains the expression levels of H2AX and ATM. This is a critical function, as H2AX-deficient cells and mice exhibit genomic instability and suffer from DDR deficiency [14, 52, 53]. Gruosso et al. found that cells lacking JunD or NRF2 exhibit an insufficient antioxidant response, resulting in the continuous accumulation of ROS and the reduction of the H2AX protein level [14]. This effect may be due to the ubiquitination of H2AX by RING finger protein 168 (RNF168) and subsequent proteosomal degradation [14]. The accumulation of γH2AX in JunD- and NRF2-deficient cells was also affected due to the downregulation of H2AX [14]. This may affect DSB repair and is an example of overlap between the antioxidant and DNA repair functions of NRF2 [53].

NRF2 in HR

HR employs homologous DNA sequences located on sister chromatids or homologous chromosomes as templates and can repair double-ended as well as single-ended DSBs [47]. HR repair of double-ended DSB must undergo several steps, including excision of damaged DNA ends to form ssDNA, homology search to form a synaptic complex, invasion of ssDNA strands on one side of the DSB site to form a D-loop, and DNA synthesis extending the D-loop [54]. Finally, repair is completed via the synthesis-dependent strand annealing pathway (SDSA) or the formation of a double Holliday junction (dHJ) [54, 55]. Break-induced replication (BIR), which is capable of repairing single-ended DSBs, is a special kind of HR [47].

Jayakumar et al. and Todoric et al. found that several genes related to HR proteins—RAD51D, RAD52, mouse double minute 2 (MDM2), disrupted meiotic cDNA 1 (DMC1), split-hand/foot malformation 1 (SHFM1), C-terminal binding protein interacting protein (CtIP), and X-ray repair of cross-complementary 3 (XRCC3)—that have AREs in their promoters [56, 57]. (Table 1) However, some literature has pointed out that no ARE sequence was found in the CtIP or RAD51 promoter, nor was there found any site for the binding of MAF-K to the RAD51 promoter [58, 59]. Among them, RAD51D can catalyze homologous pairing between ss and dsDNA, and RAD52 can bind ssDNA ends and mediate the DNA-DNA interaction required for the annealing of complementary DNA strands [56]. MDM2 inhibits HR through its interaction with the MRE11, Rad50 and Nbs1 (MRN) complex, and its release from the MRN complex through an ATP-dependent response initiates efficient HR [60]. Jayakumar et al. also pointed out that NRF2 inhibition leads to a significant reduction in the expression of the RAD51 gene, which was also confirmed in subsequent studies [10, 56]. However, there is no direct evidence to suggest whether NRF2 can bind to AREs in these DNA repair gene promoters to further regulate their expression.

Sun et al. found that NRF2 activates the ATR-checkpoint kinase 1 (CHK1)-cell division cycle protein 2 (CDC2) signaling pathway to induce G2 arrest, promoting HR and maintaining genome stability [10]. NRF2 contains a Trp residue at position 188 and a messy hydrophobic fragment, similar to ATR activation domain (AAD) of DNA topoisomerase 2-binding protein 1 (TOPBP1) and Ewing’s tumor-associated antigen 1 (ETAA1) [10]. This region of NRF2 is highly conserved in evolution [10]. NRF2 interacts with ATR at the site of DNA damage and promotes the phosphorylation and activation of ATR through its AAD-like domain [10].

In a follow-up study, Sun et al. found that NRF2 was decreased by metformin in NSCLC cells, thereby inhibiting the initiation of DNA damage repair pathways and affecting G2/M phase arrest after radiation [61]. This reduction of NRF2 by metformin was AMPK- but not KEAP1-dependent, relying on increased ubiquitination and proteasomal degradation of NRF2 [61].

The studies by Jayakumar and Sun collectively reveal two mechanisms by which NRF2 promotes HR through its functions of genome integrity maintenance and cell cycle regulation. Increased ROS accumulation due to NRF2 inhibition may lead to increased DNA damage in irradiated cells [10, 56]. Both studies used N-acetyl cysteine (NAC) to reduce the excessive accumulation of ROS caused by NRF2 inhibition before IR and found that DNA damage was still elevated, suggesting that NRF2 can act through an ROS-independent mechanism to regulate DNA repair to maintain genome integrity [10, 56].

Chen et al. found that mechanical injury upregulates NRF2 through the PI3K/AKT pathway, which then upregulates the expression of RAD51 to promote HR [62]. Following mechanical injury, serine 9 of glycogen synthase kinase (GSK) 3β and serine 473 of AKT become inhibitively phosphorylated, leading to an increase in the expression of NRF2 [62]. Combined with earlier findings that focal adhesion complexes sense mechanical signals, activate internal focal adhesion kinase (FAK), and then activate the PI3K/AKT pathway, Chen et al. was able to compile a complete pathway in which NRF2 promotes HR in mechanically injured cells [62]. It was also found that scratched fibroblasts after exposure to IR are arrested primarily at the G2/M phase and mechanical injury accelerates fibroblast recovery from IR-induced cell cycle arrest [62]. The study did not explore whether this change in cell cycle distribution is mediated by NRF2. However, this G2/M phase arrest is consistent with Sun’s results.

NRF2 in NHEJ

NHEJ is the main repair mechanism for any type of DSB in most cell cycles and is critical for the development of antigen receptors [63]. Aberrant NHEJ is a major source of genome rearrangements and chromosomal translocations, leading to genome instability [64]. The NHEJ process is mainly divided into three steps: DSB identification, processing, and connection [65]. Different terminal structures direct different sets of NHEJ proteins to arrange two DNA ends into an end-to-end configuration, which constitute sub-pathways of NHEJ [66]. For example, some DNA ends can be joined together using only the ligase complex, but others require the action of polymerases or nucleases [66].

Qin et al. found that ROS activates NRF2 in irradiated triple-negative breast cancer cells [67]. NRF2 knockout partially hinders DNA repair by inhibiting DNA-PK, and NRF2 agonists can activate NRF2 by inhibiting KEAP1, upregulating p62-related autophagy, and through various post-transcriptional mechanisms [67]. Although total DNA-PK is only slightly reduced (at both the protein and mRNA level) following NRF2 knockdown in cells, the phosphorylation level of DNA-PK (S2056) is significantly reduced [67]. These results suggest that NRF2 knockdown may inhibit DNA repair by inhibiting DNA-PK phosphorylation through a post-translational mechanism [67].

Deville et al. found that KEAP1 regulated NHEJ in an NRF2-dependent and cell cycle-independent manner, while HR activity was not affected [68]. Notably, ROS levels following X-ray irradiation were not significantly different between cells either untreated or treated with KEAP1 inhibitor ML344, suggesting that NRF2-involved DSB repair may be ROS-independent [68]. The radiosensitivity of KEAP1-inhibited cells after irradiation was enhanced, and the S2056 phosphorylation of DNA-PKcs was defective, indicating that KEAP1 inhibition impairs DSB repair [68]. The studies of Qin and Deville come to different conclusions as to whether NRF2 induction depends on ROS and whether KEAP1 inhibition or NRF2 can enhance DNA-PK phosphorylation, which may be the result of differing experimental conditions, such as cell lines and irradiation parameters.

The 53BP1 promoter contains three AREs, all of which can be bound by NRF2 [58]. NRF2 inhibition reduces 53BP1 foci formation and protein expression in irradiated cells [56, 67]. Yang et al. found that activated Kirsten rat sarcoma viral oncogene homologue (KRAS) with the G13D mutation promotes the transcription and nuclear translocation of NRF2. NRF2 binds to the 53BP1 promoter element to a greater extent in KRAS mutant cells than wild-type cells, stimulating 53BP1 transcription and initiating NHEJ-mediated repair activity [11]. This enhanced DNA repair activity is mediated in part by an increase in the proportion of G1 phase cells [11]. Overexpression of 53BP1 in KRAS mutant cells increases phosphorylation of ATM and CHK2 early after IR, increasing the proportion of cells in G1 phase and enhancing their radiation resistance [11]. Brusatol can not only inhibit NRF2, but also inhibit the activation of RAS-ERK signaling, indicating that there may be a feedback signal loop between RAS-ERK and NRF2 antioxidant responses [11].

Crosstalk between different damage repair pathways

Many downstream molecules and pathways of NRF2 participate in multiple DNA repair pathways. Here, we explore the BRCA1, sirtuin 1 (SIRT1), and autophagy pathways as examples.

BRCA1 is widely involved in a variety of DNA damage repair mechanisms. BRCA1’s functions include regulation of cell cycle progression, DNA repair, maintenance of genome integrity, and transcriptional regulation [69, 70]. BRCA1 can interact with NRF2 in multiple ways and may form positive feedback loops with it [69, 70]. In HR, the expression of BRCA1 can be promoted by NRF2. There is an ARE site between positions −218 and −210 on the BRCA1 promoter; and the NRF2 transcription complex can bind to this ARE and activate the transcription of BRCA1 by inducing histone acetylation [69]. BRCA1 can also be recruited to the xenobiotic responsive element (XRE) of the NFE2L2 promoter to affect NFE2L2 transcription [70]. In response to DSBs, the BRCA1-BRCA1-associated RING domain protein 1 (BARD1) is recruited to DSB sites [71]. Subsequently, BRCA1-BARD1 is involved in the process of end excision to generate ssDNA, RAD51 exchange of RPA bound to ssDNA, and D-loop formation [72, 73]. In addition, Bau et al. found that the expression of BRCA1 was precisely and causally related to NHEJ ability in vitro and in vivo. Based on this observation, they speculated that BRCA1 participates in NHEJ through the MRN complex [72]. Xu et al. also showed that enhanced BRCA1 expression promotes DSB repair through NHEJ [70]. However, no studies have shown whether NRF2 affects NHEJ through BRCA1.

SIRT1 is an important player in DDR, acting both as a histone deacetylase at the site of DNA damage and as a deacetylase for proteins involved in DNA repair and DDR [74]. SIRT1 is widely involved in the HR, NHEJ, BER, NER, and MMR repair pathways [74]. Ku70, forkhead transcription factor 1 (FOXO1), ATM, Werner syndrome protein (WRN), XPA, and many other DNA repair proteins can be modified by SIRT1 to change their activity [74]. NRF2 can induce the degradation of p53 by regulating the expression of MDM2; conversely, inhibiting NRF2 increases p53 and reduces the expression of SIRT1 [75]. However, whether the regulation of SIRT1 by NRF2 affects SIRT1-depedent DNA damage repair pathways is still inconclusive. SIRT1 can also induce NRF2 deacetylation, disengaging NRF2 from AREs and leading to transcription termination and nuclear export of NRF2 [75, 76]. This suggests that there may be a negative feedback regulation mechanism between SIRT1 and NRF2.

RNF168-mediated histone ubiquitination is a prelude to the activation of DSB repair mechanisms in HR and NHEJ, as p62 inhibits HR and NHEJ through physical interaction with RNF168 [77–79]. The ROS-KEAP1-NRF2 pathway contributes to the constitutive expression of p62 in viral latency cells and the activation of p62-mediated selective autophagy, thereby promoting DNA repair and making virus-transformed cells insensitive to DNA damage under oxidative stress [77]. Upregulation of CHK1, RAD51, and 53BP1 during viral latency is inversely correlated with p62, and inhibition of autophagy promotes proteasomal degradation of DNA repair proteins RAD51 and CHK1 in a manner dependent on p62 accumulation in the nucleus [77].

NRF2 in the repair of DNA damage caused by IR

DNA damage induced by IR may cause all the DNA repair pathways mentioned above. However, the research on NRF2 regulating direct repair, BER, NER, and MMR generally do not use IR to cause DNA damage, while the research on its regulation of HR and NHEJ usually require IR. The type and distribution of IR-induced DSBs, chromatin state, DNA end structure, and DNA end excision are the main determinants of the choice of IR-induced DSB repair pathways [80].

The chromatin environment can regulate DNA end resection. Taking BRCA1 and 53BP1 as examples, 53BP1 restricts DNA end resection and promotes NHEJ repair through two downstream pathways mediated by RAP1-interacting factor 1 (RIF1)-revertibility protein 7 homologue (REV7) and Pax transactivation domain-interacting protein (PTIP)-Artemis [63, 81]. BRCA1, on the other hand, promotes end resection to mediate HR repair by promoting dephosphorylation and inactivation of 53BP1 [63, 81]. Both BRCA1 and 53BP1 have AREs, and NRF2 inhibition has been reported to reduce BRCA1 and 53BP1 phosphorylation levels after IR exposure, suggesting that both are regulated by NRF2 after IR [10, 58, 69]. It is currently unknown whether the regulation of NRF2 on BRCA1 and 53BP1 in the IR background affects the selection of cells for DNA end excision and further causes cells to undergo HR or NHEJ. Both Sun and Yang’s studies found evidence of mutual antagonism between BRCA1 and 53BP1, but under what circumstances NRF2 preferentially regulates one rather than the other remains unclear [10, 11]. Whether this is mediated by active selection by NRF2 or caused by post-transcriptional or post-translational modification of BRCA1 and 53BP1 by other mechanisms needs further clarification [10, 11].

The cell cycle can also play roles in DNA excision. This is because CDK activity can increase significantly when cells enter S phase and G2 phase, with CDK activating DNA repair proteins to perform DNA end resection and promote HR [80]. However, during G1 phase, CDK activity is significantly reduced to limit DNA end excision, thus favoring NHEJ rather than DNA end excision-dependent repair pathways [80]. Current studies have shown that, in the context of IR, NRF2 activates the ATR-CHK1-CDC2 signaling pathway in a BRCA1-independent manner, promoting G2 arrest and HR, while the KRAS(G13D)/NRF2/53BP1 axis increases ATM and CHK2 phosphorylation, increasing the proportion of cells in G1 phase and promoting NHEJ [10, 11]. Whether the decisive step of cell cycle arrest selection occurs before or after the activation and induction of ATR or 53BP1 by NRF2, whether there are damage detection factors regulating NRF2, and whether there are epigenetic modifications or post-translational modification changes are currently unknown.

NRF2 and cancer

NFE2L2 mutations in cancer

Using cBioPortal online tools (http://www.cbioportal.org/), we analyzed the mutation frequency of NFE2L2 in different types of cancer and found that esophageal carcinoma (ESCA), lung cancer, and penile cancer had the highest overall mutation frequency of NFE2L2 gene (Fig. 2A). The highest frequency of mutations detected by whole exon sequencing was observed in ESCA, penile cancer, and head and neck cancer. Among these mutations, the most ones occurred in the ETGE and DLG motifs which had been identified as cancer hotspots, and missense was the main type of mutation (Fig. 2B). Disrupted KEAP1 binding to NRF2 causes aberrant accumulation of NRF2 in all cancers analyzed so far [82]. ESCA, lung cancer, uterine endometrioid carcinoma, and ovarian epithelial tumor have the highest frequency of copy number variations (CNV) mutations. Moreover, in cancers with CNV mutations, amplification rather than deep deletion was the main type of mutation, which may also lead to an increase in NRF2 expression levels. Except for carcinoma of uterine cervix and ESCA, which have very few structural variants, NFE2L2 rarely undergoes this type of variation in other types of cancer.

Fig. 2.

The relationship between NFE2L2 mutations and DNA repair. (A) shows the situation of NFE2L2 mutations in different cancers. 86,614 samples from databases including TCGA were analyzed using cBioPortal, and each cancer had cases greater than 10 and mutation frequency greater than 2%. The horizontal axis represents the type of cancer, and the vertical axis represents the frequency of mutations. Color represents the type of mutation. (B) shows the mutation sites on NRF2. The horizontal axis represents the amino acid number of NRF2, the vertical axis represents the total number of mutations, and the color represents the type of mutation. (C) demonstrates the correlation between gene expression levels of DNA repair systems and NFE2L2 mutations, NFE2L2 expression levels, as well as clinical staging. The correlation between genes was analyzed using the Spearman coefficient, while the correlation between mutations and stages, as well as the correlation between mutations or stages and genes, were analyzed using the Kendall coefficient. The heat map shows genes that are positively or negatively correlated with mutations (r > 0.25 or r < −0.25, p < 0.05). (D) shows the Venn diagram of genes that are positively correlated with mutations and NFE2L2 overexpression (r > 0.1, p < 0.05), but negatively correlated with staging (r < −0.1, p < 0.05). (E) shows the Venn diagram of genes that are negatively correlated with mutations and NFE2L2 overexpression (r < −0.1, p < 0.05), but positively correlated with staging (r > 0.1, p < 0.05). (F) shows the expression distribution of NFE2L2 in tumor tissues and normal tissues. The abscissa represents different groups of samples, and the ordinate represents the expression distribution of the gene, different colors represent different groups, top-left represents the significance p-value, *p < 0.05, **p < 0.01,***p < 0.001, asterisks (*) stand for significance levels. G1 represents stage I and II, G2 represents stage III and IV. The statistical difference of two groups was compared through the Wilcox test, significance difference of three groups was tested with Kruskal-Wallis test. The data from TCGA and GTEx databases were analyzed using aclbi (https://www.aclbi.com/static/index.html#/)

We obtained 2036 genes from four gene sets: DNA repair (GO: 0006281), regulation of DNA repair (GO: 0006282), cell cycle (GO: 0007049), and regulation of cell cycle (GO: 0051726) in the GO database (http://geneontology.org/), and used them as DNA repair systems. At the same time, we obtained somatic mutation data including single nucleotide polymorphism (SNP) and INDEL, clinical information of patients, and gene expression data of ESCA in the TCGA database (https://portal.gdc.com). Subsequently, we analyzed the correlation between gene expression levels of the DNA repair system and NFE2L2 mutations, NFE2L2 expression levels, and clinical staging (Fig. 2C). The results showed that the NFE2L2 mutation was positively correlated with the expression of 381 genes (r > 0.1, p < 0.05), and negatively correlated with 207 genes (r<−0.1, p < 0.05). Moreover, many genes that are positively or negatively correlated with NFE2L2 mutations are also positively correlated (326 of 381) or negatively correlated (82 of 207) with the expression level of NFE2L2 genes. The NFE2L2 mutation is also positively correlated with the expression level of NFE2L2.

Clinical staging was positively correlated with 52 genes in the DNA repair system (r > 0.1, p < 0.05), and negatively correlated with 58 genes (r<−0.1, p < 0.05). Moreover, a significant proportion of genes (50 of 58) that are negatively correlated with clinical staging are positively correlated with NFE2L2 mutations or NFE2L2 expression levels (Fig. 2D). Among the genes positively correlated with clinical staging, some (14 of 52) are negatively correlated with NFE2L2 mutations or NFE2L2 expression levels (Fig. 2E). We speculate that this may reflect the dual roles of NRF2 in cancer. On the one hand, as shown in Fig. 2F, the expression levels of NRF2 in stage I and II or stage III and IV are significantly higher than that in normal tissues. The high expression of NRF2 in ESCA indicates poor prognosis, which may be due to the DNA repair and antioxidant functions of NRF2 leading to the resistance of tumors to treatments [83, 84]. On the other hand, in ESCA patients, the expression levels of NRF2 in stages III and IV are significantly lower than those in stages I and II, and the gene expression of some DNA repair systems is also negatively correlated with clinical staging (Fig. 2C and F). This may be because the DNA repair function and antioxidant function of NRF2 inhibit the occurrence of more malignant mutations in tumors. Further research is needed to determine whether these correlations imply actual regulation or impact.

Research progress of NRF2 in cancer

NRF2 has a dual role in the occurrence and development of cancer. On the one hand, due to the detoxification effect and antioxidant capacity, NRF2 can inhibit cancer [85]. The absence of KEAP1 and NRF2 can lead to tumor development [85]. On the other hand, excessive activation of NRF2 can promote cancer, increase tumor resistance, and lead to poor prognosis in patients [82, 86]. The molecular mechanism of activation of NRF2-KEAP1 pathway in cancer is different, which mainly includes somatic mutation in KEAP1 or NRF2; epigenetic modification of KEAP1 and NRF2 promoters; post-transcriptional activation of NRF2; carcinogenic signaling and hormonal activation [85, 87]. Zimta et al. summarized the different roles of activated NRF2 in different cancers [88]. For example, the activation of NRF2 in the bladder cancer leads to cisplatin resistance and increases resistance to taxol and anchorage dependent growth [88]. The activation of NRF2 stimulates the cancer stem cell phenotype in glioma and increases chemoresistance in lung cancer [88]. According to the TCGA and GTEx databases (https://gtexportal.org), NRF2 is significantly overexpressed in 9 types of cancer, including cholangiocarcinoma, lung squamous cell carcinoma, glioblastoma, kidney clear cell carcinoma, kidney papillary cell carcinoma, lower grade glioma, liver cancer, pancreatic cancer, and ESCA(Fig. 3). Kitamura et al. summarized that the high expression of NRF2 or its target genes among 17 types of cancer, including glioma, lung cancer, esophageal carcinoma, head and neck squamous cell carcinoma, breast cancer, hepatocellular carcinoma, and bladder cancer are associated with poor prognosis [86]. Literature has summarized the role of NRF2 in hallmarks of cancer, such as genetic instability, avoid immune construction, and proteotoxic stress [6, 89, 90]. Among them, research on the role of NRF2 in DNA repair has focused on glioblastoma, blast cancer, colon cancer, acute myeloid leukemia, head and neck square cell cancer, and non-small cell lung cancer.

Fig. 3.

The expression distribution of NFE2L2 in tumor tissues and normal tissues. The abscissa represents different tumor tissues, and the ordinate represents the expression distribution of NFE2L2, different colors represent different groups. *p < 0.05, **p < 0.01, ***p < 0.001, asterisks (*) stand for significance levels. The statistical difference of two groups was compared through the Wilcox test. The data was analyzed using aclbi

NRF2 modulators are one of the most attractive strategies for anticancer therapy, but their clinical application is still challenging due to factors such as the short half-life of NRF2 [82]. In general, NRF2 inducers can protect normal cells from carcinogen, while NRF2 inhibitors can inhibit the proliferation of cancer cells with abnormal NRF2 activation or NRF2 addiction [82]. Sova et al. summarized the research and development strategies as well as types of NRF2 modulators while Robledinos Ant ó n et al. summarized the NRF2 modulators under clinical development [91, 92].

Summary

The DNA repair capacity of tumor cells is associated with drug resistance in chemoradiotherapy, targeted therapy, and immunotherapy [93, 94]. Assessing the DNA repair phenotype prior to treatment may be of great value in the clinical management of therapeutic agents or modalities [93, 94]. Here, we have reviewed how NRF2 regulates DNA damage repair through its functions in maintaining genomic integrity, regulating the cell cycle, and reducing oxidative damage. We have also explored the possible role of NRF2 in DSB pathway selection under the background of IR. Much of the research on the regulation of DNA damage repair by NRF2 is in its initial stages. For example, HR can be subdivided into multiple pathways such as SDSA, dHJ, and BIR. NHEJ also has 12 sub-pathways with a variety of steps and molecules involved [54, 63]. However, the research on NRF2 has not gone deep into the specific branches and steps, which is a direction worth studying. Some evidence has shown that there are post-transcriptional and post-translational modifications in the process of NRF2 regulating DNA repair, but the specific mechanism for this remains unclear [28, 29, 67]. Whether pathways such as RNA modification, ncRNA, and protein post-translational modification affect the regulation of NRF2 on DNA repair is still to be determined. We hope that by clarifying the upstream and downstream links of NRF2’s regulation of DNA repair, we will draw attention to NRF2 as a target for cancer treatment.

Author contributions

All authors contributed to the outline and content of the article. The first draft of the manuscript was written by Jiale Li and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (32171239 and 82273580), CAMS Innovation Fund for Medical Science (2021-I2M-1-042), CIRP Open Fund of Radiation Protection Laboratories (CIRP-DTRI20220202).

Data Availability

Not applicable.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

Not applicable.

Statements and Declarations

The authors have no relevant financial or non-financial interests to disclose.