The NRF2 transcription factor plays a dual role in colorectal cancer: A systematic review

Background Colorectal cancer is one of the most common cancers worldwide, and is influenced by the interplay of various factors, including a very strong genetic component. For instance, incorrect mitochondrial biogenesis is correlated with increased risk of developing colorectal cancer. Thus, it is important to understand the consequences of changes in both the expression and the correct function of the transcription factors that regulate mitochondrial biogenesis, namely NRF2. Objectives The main objective of this paper is to characterise the relationship between NRF2 and colorectal cancer by compiling data from an exhaustive literature search. Methods Information was obtained by defining specific search terms and searching in several databases. After a strict selection procedure, data were tabulated and the relationships between articles were assessed by measuring heterogeneity and by constructing conceptual maps. Results and discussion We found a general consensus in the literature that the presence of oxidizing agents as well as the inhibition of the NRF2 repressor Keap1 maintain NRF2 expression at basal levels. This predominantly exerts a cytoprotective effect on cells and decreases risk of colorectal cancer. However, if NRF2 is inhibited, protection against external agents disappears and risk of colorectal cancer increases. Interestingly, colorectal cancer risk is also increased when NRF2 becomes overexpressed. In this case, the increased risk arises from NRF2-induced inflammation and resistance to chemotherapy. Conclusion The proper basal function of NRF2 and Keap1 are essential for preventing oncogenic processes in the colon. Consequently, any disruption to the expression of these genes can promote the genesis and progression of colon cancer.


Objectives
The main objective of this paper is to characterise the relationship between NRF2 and colorectal cancer by compiling data from an exhaustive literature search.

Methods
Information was obtained by defining specific search terms and searching in several databases. After a strict selection procedure, data were tabulated and the relationships between articles were assessed by measuring heterogeneity and by constructing conceptual maps.

Results and discussion
We found a general consensus in the literature that the presence of oxidizing agents as well as the inhibition of the NRF2 repressor Keap1 maintain NRF2 expression at basal levels. This predominantly exerts a cytoprotective effect on cells and decreases risk of colorectal cancer. However, if NRF2 is inhibited, protection against external agents disappears and risk of colorectal cancer increases. Interestingly, colorectal cancer risk is also increased when NRF2 becomes overexpressed. In this case, the increased risk arises from NRF2induced inflammation and resistance to chemotherapy. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
Colorectal cancer (CRC) is one of the most common cancers worldwide, and causes more than half a million deaths per year [1]. This type of cancer is a useful model for studying both carcinogenesis and tumour progression, as colonocytes follow a systematic process of proliferation, differentiation and transformation (adenoma to a carcinoma) [2]. CRC arises from stem cells in the colon [3], and its development is partly due to functional changes in the mitochondrial genome (mtDNA) [4].
Mitochondria play a crucial role in physiological processes, and their deregulation contributes to the genesis and progression of CRC. This is known as the Warburg effect, which consists of metabolic reprogramming that is characteristic of tumour processes. Thus, it is important that all genes involved in mitochondrial biogenesis are expressed according to a standard pattern. This process is highly coordinated and depends on the activity of several key proteins, including Nuclear Respiratory Factor 1 (NRF1) and Nuclear Respiratory Factor 2 (NRF2), which regulate the expression of respiratory chain proteins, and act as anti-oxidants [5].
NRF transcription factors control the functionality of the respiratory chain, the main pathway responsible for producing reactive oxygen species (ROS). ROS are extremely harmful for cellular micromolecules because they induce tumorigenesis through non-specific reactions with nucleic acids, proteins, and lipids [6].
NRF2 is considered to be a main defence mechanism and regulator of cell survival [7]. Under healthy conditions, NRF2 protects against tumorigenesis and cancer progression by attenuating genotoxic compounds that emerge both intrinsically and extrinsically. However, activation of the NRF2 defence response can promote the survival of both normal and cancer cells by creating an optimal environment for cell growth. Furthermore, NRF2 protects tumour cells from oxidative stress, chemotherapeutic agents, and radiotherapy, promoting tumour genesis and progression, and also metabolic reprogramming to anabolic pathways [8]. Fig 1 shows a simplified diagram of the mechanisms that can activate the transcription factor NRF2. NRF2 is sequestered by the inhibitor Keap1 and is transcriptionally inactive until Keap1 is dissociated by ROS. This dissociation allows NRF2 to form transcriptionally active complexes with other proteins, such as Mafs, and, after the transcription of cytoprotective and antioxidative genes, to remove ROS [9].
In this study, we provide a qualitative systematic review to summarize the role of the NRF2 transcription factor. Our goal was to assess the effects of differential expression of NRF2 on CRC.
independently assessed by two individuals to minimize bias (one was an expert on this topic and the other had no knowledge of it).

Design of search strategies and use of Boolean operators
For each search, synonyms, related terms and spelling variations were taken into account, and Boolean operators were used to relate the terms. Searches were performed on eight different databases (PubMed, Scopus, Web of Science (WOS), Cochrane library, PysCinfo, PLoS One, Scielo and PubMedCentral (PMC)) using the following terms: To retrieve optimal results when using the Scopus database, it was necessary to modify the search criteria. Thus, only the following more specific terms were used: 'CRC' AND 'NRF2 OR NFE2L2' AND 'Warburg' 'CRC' AND 'NRF2 OR NFE2L2' AND 'OXPHOS'. We included only scientific reports written in English, and published after 2006, to obtain a current review of this issue. We collected articles using all combinations of search terms.

Selection process
The articles returned by the searches were manually screened to identify relevant articles to be included in this review: 1. Search results were integrated using informatics programs. We used the EndNote reference manager to remove duplicate articles. 2. Titles and abstracts were assessed to remove irrelevant articles. Only articles satisfying the following criteria were included: (i) the article considers CRC or other related diseases; (ii) the title of the article includes the word NRF2, or the name of another related gene; (iii) the article discusses the relationship between CRC and either mitochondrial biogenesis, OXPHOS or the Warburg effect. 3. The full texts of potentially relevant articles were retrieved and analysed to verify that they met the inclusion criteria. Fig 2 shows a flow chart summarising each step in the search process. Reports that did not meet the inclusion criteria were excluded from the analysis. S1 Table lists the articles excluded in the last step of the search process.

Assessment of results
Tabulation. We tabulated the following key data from each article: first author, publication year, organism/cell type, primary results and conclusion (S2 Table).
Heterogeneity. To assess the heterogeneity of the selected articles, we analysed the following aspects: expression of NRF2, relationship to CRC, use of cell culture, NRF2 as a chemotherapeutic agent, apoptosis, the inflammatory process and mitochondrial function. To more easily view the specific characteristics of each article, and to evaluate the homogeneity among them, we represented each article in a stacked bar chart (Fig 3).
Concept map. We constructed three maps to visualize the links between the selected articles. We aimed to relate three key concepts behind NRF2 expression and its role in the development of CRC so that we could draw conclusions and suggest ideas for future research.

Article selection
The search described above identified relevant articles only in the Pubmed, Web of Science, and Scopus databases. Removing duplicated articles from these search results resulted in a total of 104 articles. We reviewed the titles of these articles and discarded those not related to our topic, reducing the number of articles to 82. We then reviewed the abstracts, further reducing the number of articles to 60. Finally, we reviewed the full text, and selected 40 articles for the review. This article selection process is summarized in Fig 2. Articles excluded in the last step are summarized in Supplementary Information 1.

Tabulation
Complete data from the 40 articles selected for this review are shown in Supplementary Information 2. All selected papers were published after 2006. The geographic settings of the studies vary, with nine from in the USA, nine from China, seven from Korea, four from Germany, four from Malaysia, three from India, two from Japan, and one each from the United Kingdom, Norway, and Australia. Most studies were based on human cancer cell lines. Only 13 studies used animal experimentation (mice), and four considered patients with ulcerative colitis or Crohn's disease. The most important results from the selected articles are shown in Supplementary Information 2. Titles, author and publication date of the included articles are shown in Table 1.

Heterogeneity
The characteristics of each report are shown in

Building a concept map
To interconnect the ideas gathered from the articles included in this review, we constructed three conceptual maps focused on the expression of NRF2 and its effect on colon cancer. The

First author Primary results
Kim et al., 2011 [6] Hypoxia-induced angiogenesis can be blocked by inhibiting Nrf2. Inhibition of Nrf2 reduced tumour growth and decreased angiogenesis in mice xenografts, and was associated with a lower accumulation of HIF-1 α under hypoxic conditions. Hypoxia cannot activate HIF-1α in Nrf2-inhibited colon cancer cells.
PGC1-α and NRF1 levels were unaffected by inhibition of Nrf2. However, destabilization of HIF-1α is associated with a weakened mitochondrial function in colon cancer cells.
Pandurangan et al., 2014 [12] Nrf2 is activated by luteolin Kim et al., 2013 [13] Catechol residues are essential for activating Nrf2.  [35] Nrf2 is activated when cells are exposed to inflammatory macrophages. This increases proteosomal activity and the expression of proteosomal genes in a Nrf2-dependent manner.

Yokoo et al., 2016[36]
In Nrf2 -/mice treated with KBrO3, atypical hyperplasia, adenoma and adenocarcinoma were all observed in the upper small intestine. These same mice displayed an increase in the number of aberrant crypts compared with their Nrf2 +/+ counterpart. CRC in humans, while altered expression of this gene promotes tumour genesis and progression (Figs 5 and 6).

Discussion
The articles selected show that research on the role of NRF2 in CRC can be classified into three broad groups: i) those that analyse standard expression and observe a cytoprotective effect on CRC [11][12][13][14][15][16][17], and those that analyse variations in expression, either ii) overexpression or iii) inhibition, both leading to increased CRC risk. Below, we discuss the findings in each of these three lines of research. The first concept map (Fig 4) shows the protective effect of induced expression of the NRF2 gene on CRC following a moderate accumulation of ROS due to specific external agents [18][19][20][21][22]. When NRF2 activates, it exerts a cytoprotective effect by decreasing genotoxic damage and promoting autophagy [23]. Sometimes, NRF2 expression leads to reduced cell proliferation and increased apoptosis of colon cancer cells [24][25][26][27]. In all cases, the authors observed the same final result, namely reduced risk of CRC. Another example of this behaviour was reported by Yang et al. (2014), in which digitoflavonoids were found to act upon p38 and to induce NRF2 target genes and antioxidants. They reported a reduction in both the number and size of tumours, as well as a decrease in the number of aberrant crypts, further pointing to a protective effect of NRF2 on CRC [28]. Therefore, expression of NRF2 and its target genes, such as glutathione synthase, likely play a crucial role in cytoprotective mechanisms against CRC [29][30][31].
In the studies mentioned up to this point, NRF2 expression is directly induced by the external agents. However, NRF2 expression can also be promoted indirectly by other mechanisms, via the repressor Keap1. For example, Jung and Kwak, 2013 silenced Keap1 and found that induction of NRF2 promoted the expression of AKR members, which are involved in the detoxification process, leading to reduced risk of CRC [32]. In another study, Lu et al., 2016 used an inhibitor of its interaction with NRF2, CPUY192018, and observed that activated NRF2 acted as a protective factor upon exposure to DSS, decreasing CRC risk [33].
In all of these cases, standard NRF2 expression levels function as a protective factor against CRC by providing cytoprotection and by activating antioxidant target genes. These results are corroborated by the review by Pandurangan and Esa, (2014) and Pandurangan et al (2014aPandurangan et al ( , 2014b, who also support the idea that expression of NRF2 decreases the risk of CRC [12,29,50]. As shown in the second concept map (Fig 5), several studies have reported that NRF2 overexpression increases risk of CRC. This overexpression can occur for a number of reasons, including constitutive mutations in the Keap1 repressor gene or in the NRF2 gene itself. Stachel et al. (2014) found that excessive levels of ROS induced overexpression of NRF2, leading to inflammation of the colon tissue and promoting tumorigenesis [34]. In another study, Sebens et al. (2011) reported overexpression of NRF2 in tissues exposed to inflammatory macrophages, and a simultaneous increase in the activity of proteosomal genes, leading to reduced apoptosis and uncontrolled proliferation. This in turn results in greater risk of CRC [35]. Other authors have observed similar results when NRF2 is overexpressed as a consequence of silencing, or as a result of epigenetic changes in this gene [36][37][38]51]. Another way to induce overexpression of NRF2 is using chemical compounds. Kim et al. (2011) found that T-BHQinduced overexpression of NRF2 resulted in increased O 2 consumption, which is correlated with higher HIF-1α and VEGF signalling, which in turn directly increases angiogenesis and tumour growth [6]. Therefore, as mentioned above, overexpression of the NRF2 gene is closely linked to increased risk of CRC [39,40]. In this case, CRC risk can increase by promoting colonic inflammation, decreasing apoptosis, or promoting angiogenesis and uncontrolled cell proliferation. Notably, NRF2 overexpression has also been linked to increased resistance to the chemotherapeutic agent 5-FU via the activity of demethylases and methyltransferases. In addition to the increase risk of CRC, this implies a poorer response to treatments based on this agent. [37,41].
The third concept map (Fig 6) shows that inhibited expression of NRF2 also has consequences for this type of cancer, since one of NRF2's functions is to protect against damaging and carcinogenic substances [42,43]. Various studies highlight the role of NRF2 on CRC via its inhibition. Li et al. (2008) related Nrf2 inhibition with increased risk of CRC [44]. Yokoo et al. (2016) found that silencing Nrf2 resulted in decreased expression of the COX2 gene, leading to an increase in the number of aberrant crypts, resulting in the formation of adenoma, adenocarcinoma and ultimately CRC [36]. Other studies have compared the effects of silencing the Nrf2 gene in mice and exposing them to treatments that induce ulcerative colitis (DSS) or colon carcinogenesis (AOM) [45][46]. Nrf2 knockout mice with ulcerative colitis suffer prolapse, rectal bleeding, inflammation, and as in the previous case, show an increase in the number of aberrant crypts. Together, these symptoms contribute to the development of CRC. This information is corroborated by reviews from Saw and Kong in 2011 [47] and Cheung et al. in 2014 [48].
Analysing the behaviour of this gene, not at a single time point, but throughout the various stages, Chang et al. (2013) suggested that the differential expression profile of NRF2 in tumour versus normal tissue could be an interesting new target for CRC treatment [49]. Similarly, Arlt et al. (2009) observed that NRF2 expression is beneficial during early stages, but can contribute to tumorigenesis in the colon at later stages [43]. These data are also supported by Menegon et al (2016) and Sporn M and Liby K (2012), who described NRF2 not only as an oncogene but also as a tumour suppressor gene [52,53]. In addition, few studies have analysed the role of Single Nucleotide Polymorphisms (SNPs) in this gene and its promotor, although Yokoo et al. (2016) suggested that some SNP alleles in the NRF2 gene could increase CRC risk based on previous studies of these polymorphisms and the risk of ulcerative colitis [36].
Finally, another important factor is the type of cell used in the studies included in this review. Most of the articles focus on colon stem cells, which need more energy and therefore require greater expression of NRF2 to obtain the necessary ATP for proliferation, migration and differentiation, such that altered expression of NRF2 increases genetic instability leading to increased CRC risk. In vivo, these cells migrate to the apical part of the crypt and become mature epithelial cells, and altered expression of NRF2 during migration and differentiation, will lead to an accumulation of mutations that promote tumorigenesis due to absence of cytoprotection. Thus, NRF2 function is probably essential in both the stem cells of the colon and in mature epithelial cells [54].
Although we searched several databases using an exhaustive list of relevant terms, our study has some limitations. We only selected articles published in English, which could result in selection bias. In addition, the use as search terms such as oxidative phosphorylation, mitochondria or Warburg effect, could limit the selection of articles that relate NRF2 to colorectal cancer, although searches performed without these concepts in several databases did not return any additional relevant articles.
In conclusion, it is clear that the NRF2 transcription factor has a very complex role in the cell, and that its expression can be strongly affected by external agents. Any disruption of its standard expression, be it overexpression or silencing, can promote the genesis and progression of CRC. Further studies are needed to clarify the role of NRF2, particularly to analyse the role of various SNPs that could generate variability in the expression and function of the protein in humans, as well as to study the pathways regulated by this gene, and possible interactions with environmental and behavioural factors, although such studies require large sample sizes.
Supporting information S1 File. Guidance on the Conduct of Narrative Synthesis in Systematic Reviews. (PDF) S2 File. Check list. (DOCX) S1