Background
Mutations and aberrant methylation patterns are generally accepted as early events and important determinants of colon carcinogenesis. However, observed differences in the incidence of colon cancer seem to result primarily from the influence of environmental factors, among them oxidative stress which may be also linked to epigenetic changes [
1‐
4].
Methylation of cytosine, a key epigenetic modification, usually involving CpG dinucleotides, is closely linked to gene repression, a process that exerts a profound effect on cellular identity and organismal fate [
5]. Equally important is active DNA demethylation, a recently discovered process which results in activation of previously silenced genes. Molecular background of active DNA demethylation is still not completely understood (reviewed in [
6]). The most plausible mechanism involves ten-eleven translocation (TET) proteins that catalyze oxidization of 5-methylcytosine (5-mCyt) to 5-hydroxymethylcytosine (5-hmCyt), and then to 5-formylcytosine (5-fCyt) which is eventually converted to 5-carboxycytosine (5-caCyt) [
6,
7]. Some evidence from experimental studies suggests that TETs may be also involved in synthesis of 5-hydroxymethyluracil (5-hmUra), a compound with epigenetic function [
8].
A plethora of recent studies demonstrated unequivocally that 5-hmCyt is profoundly reduced in many types of human malignancies, including colorectal cancer [
9‐
11]. However, it is still unclear whether this phenomenon is limited solely to tumor tissue, or may occur also in surrogate materials from cancer patients, for example, leukocytes.
Moreover, it cannot be excluded that active DNA demethylation taking place under altered conditions or in a different environment, for example in presence of chronic inflammation (that may induce oxidative stress) or in malnutrition (that may influence ascorbate level), may modulate TET activity and thus, affect the level of epigenetic modifications.
Our previous research demonstrated that colorectal cancer patients present with significantly (ca. 30%) reduced levels of ascorbate [
12,
13]. Recent studies showed that ascorbate may enhance generation of 5-hmCyt in cultured cells [
14‐
17]. Also retinol has been demonstrated recently to enhance the synthesis of 5-hmCyt and to modulate the level of TETs [
18]. Consequently, it cannot be excluded that the level of epigenetic DNA modifications and the expression of TETs in leukocytes are associated with blood concentrations of ascorbate and retinol.
In this study, we used our recently developed rapid, highly-sensitive and highly-specific isotope-dilution automated online two-dimensional ultra-performance liquid chromatography with tandem mass spectrometry (2D-UPLC-MS/MS) [
19,
20] to analyze global methylation and to determine the levels of TET-mediated oxidation products of 5-mCyt and thymine: 5-hmCyt, 5-fCyt, 5-caCyt and 5-hmUra. Moreover, we analyzed the level of the best characterized marker of oxidatively modified DNA, 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxodG), as well as the expression of TETs mRNA, and plasma concentrations of antioxidant vitamins: ascorbate, retinol and α-tocopherol.
Obtaining altered tissues from patients with some pathological conditions may be challenging. However, some studies demonstrated that the analysis of non-affected tissues can provide equally informative results (reviewed in [
21]). Leukocytes are often used as an easily accessible cells carrying information about environmentally-induced DNA modifications in other tissues [
22,
23].
Despite the fact that metabolic changes closely linked to inflammation may influence 5-hmCyt and formation of its derivatives, none of the previous studies analyzed the effect of chronic inflammation on the generation of 5-hmCyt derivatives in leukocytes. Our study may also fill another knowledge gap, demonstrating how various conditions predisposing to colorectal cancer can shape TET-mediated DNA modifications and oxidatively modified DNA in an easily accessible tissue, leukocytes.
In this study, we examined leukocytes from colorectal cancer patients and individuals with two most common conditions predisposing to sporadic colorectal malignancies, colon polyps and inflammatory bowel disease (IBD).
Discussion
Although a molecular link between colon adenomas/IBD and carcinogenesis is yet to be established, it likely involves aberrant methylation and oxidative damage of DNA (reviewed [
25,
26]), and those processes were postulated to precede colonic dysplasia and colorectal cancer development [
27]. Furthermore, it cannot be excluded that aberrant methylation of DNA is somehow related to oxidative stress.
A growing body of evidence suggests that reduced content of 5-hmCyt may be characteristic not only for cancer tissue but also for precancerous lesions [
11]. This in turn implies that this process may perpetuate during tumor progression.
Our present study showed for the first time that 5-hmCyt content in leukocytes decreased according to the following pattern: healthy controls > IBD patients > polyp patients > colorectal cancer patients (Fig.
1b). This suggests that a decrease in global level of 5-hmCyt observed during the course of colon cancer development is not limited solely to the malignant tissue, but may be also observed in surrogate tissues, such as leukocytes. This in turn implies that aberrant methylation of DNA may be a systemic process, rather than a local phenomenon. Aside from a decrease in 5-hmCyt content, leukocytes from all patient groups contained significantly less 5-hmUra than the cells from healthy controls (Fig.
1e).
According to Pfaffeneder et al. [
8], 5-hmUra level undergoes changes during the course of epigenetic cell reprogramming, following the same pattern as other TET products, i.e. 5-hmCyt, 5-caCyt and 5-fCyt. This implies that 5-hmUra may have an epigenetic function, similar to other products of active DNA demethylation (reviewed in [
28]). Our hereby presented results add to this evidence, suggesting that similar to 5-hmCyt, also 5-hmUra may be an epigenetic mark of carcinogenesis.
We demonstrated that the level of 5-fCyt, a higher-order oxidative epigenetic mark, was significantly higher in IBD patients than in other study groups (Fig.
1c). Interestingly, individuals with IBD presented also with significantly higher levels of 8-oxodG, an established marker of oxidative stress (Fig.
1f).
The association between inflammation and oxidative stress is well documented, and a number of previous studies demonstrated that inflammatory conditions and infections may be associated with an increase in 8-oxodG level. Inflammatory response may result in recruitment of activated leukocytes. This may lead to a “respiratory burst”, i.e. an increase in oxygen uptake with resultant enhanced release of reactive oxygen species (ROS), such as superoxide and hydrogen peroxide; this eventually contributes to oxidative stress and DNA damage (for review see [
29]).
Noticeably, all patients participating in this study, irrespective of their underlying condition, showed statistically significant positive correlations between the leukocyte levels of all analyzed oxidized epigenetic modifications (except 5-hmdC in IBD group) and 8-oxodG content, while no such associations were found in healthy controls (Additional file
1: Figures S1, S3–S6).
Endogenous synthesis of free radicals probably does not constitute a principal reason behind the formation of epigenetic marks in cellular DNA [
30]. Rather, environment characteristic for oxidative stress, linked with the pathogenesis, may influence factors responsible for the formation of the abovementioned modifications. Indeed, recent evidence suggests that oxidative stress may contribute to post-translational modulation of TET2 [
31]. In line with those findings, we showed that the leukocyte contents of TET1 and TET2 mRNA in patients with IBD (a condition associated with more severe oxidative stress) were significantly higher than in other study groups (Fig.
3a, b). Furthermore, IBD patients presented with elevated levels of 5-fCyt. Since the structure of TET co-substrates (2-ketoglutarate, Fe
+2) depends on redox state of the cell, altered activity of these enzymes may reflect the level of oxidative stress in IBD patients, i.e. the factor that might contribute to 5-fCyt formation. Moreover, it cannot be excluded that also superoxide (O
−2), an anion radical of dioxygen and the precursor of free radicals, may play an important role in TET-mediated active DNA demethylation [
32,
33].
The level of another higher-order oxidative modification of 5-mCyt, i.e. 5-caCyt, was the highest in leukocytes from colorectal cancer patients (Fig.
1d). Recent evidence suggests that TET2 may yield 5-fCyt and 5-caCyt without the release and dilution of 5-hmCyt, and consecutive steps of the iterative oxidation were postulated to be regulated by co-substrate levels [
34]. Consequently, a persistent increase in oxidative stress may alter TET activity, promoting/down-regulating the generation of 5-fCyt and 5-caCyt during iterative oxidation of 5-mCyt. Taken altogether, this evidence suggests that the synthesis of epigenetic DNA modifications is linked to oxidative stress; however, this relationship seems to be complex and its exact character is still not completely understood.
In our present study, patients from all groups presented with significantly lower levels of 5-mCyt than the controls (Fig.
1a); the lowest 5-mCyt levels were observed in individuals with polyps and colorectal cancer. The distribution of 5-mCyt levels across the study groups followed a similar pattern as for 5-hmCyt: healthy controls > IBD patients > polyp and colorectal cancer patients. A dramatic decrease in 5-mCyt and 5-hmCyt levels may contribute to genomic instability, constituting a decisive step in colorectal cancer development. Interestingly, we found a significant inverse correlation between 5-mCyt and 8-oxodG levels in IBD patients (Additional file
1: Figures S1, S4), which constitutes another argument for a potential link between aberrant DNA methylation and oxidative stress.
A few previous studies demonstrated that ascorbate may enhance generation of 5-hmCyt in cultured cells, probably acting as a cofactor of TETs during the hydroxylation of 5-mCyt [
14‐
17]. Recently, we have reported a spectacular increase in 5-hmUra level after stimulation with ascorbate [
15]. In turn, our present study demonstrated a positive correlation between plasma concentration of ascorbate and the levels of two epigenetic modifications, 5-hmCyt and 5-hmUra in leukocyte DNA (Additional file
1: Figure S2). Moreover, we found a significant difference in the levels of these modifications in patients whose plasma concentrations of ascorbate were below the lower and above the upper quartile for the controls (Fig.
4a, b). It is of note that plasma concentrations of ascorbate may differ up to tenfold from person to person, and individuals in whom the level of this compound does not exceed the lower quartile were shown to be at increased risk of cancer mortality [
35]. Previous studies demonstrated that if blood concentration of ascorbate remains at a physiological level (above 20 µM), leukocyte concentration of this compound reaches plateau, about 3 mM [
36,
37]. Therefore, we analyzed the associations between DNA modifications in persons with higher ascorbate levels (above 40 µM) and in individuals with ascorbate concentrations below 20 µM (Additional file
1: Figure S9), in whom cellular uptake of this compound is impaired [
36,
38]. Interestingly, participants from the former group presented with significantly higher levels of DNA modifications than the persons with ascorbate deficiency.
Probably, our study provided the first in vivo evidence for the involvement of ascorbate in the generation of epigenetic DNA modifications. Our hereby presented findings suggest that ascorbate may play a role in cancer control, preventing aberrant methylation of DNA.
To the best of our knowledge, this is the first study to show that each of the analyzed groups, healthy controls, individuals with IBD and adenomatous polyps and colorectal cancer patients, presented with a characteristic pattern of epigenetic modifications in their leukocyte DNA. Therefore, an important question arises about the mechanism(s) involved in the development of disease-specific epigenetic modification profiles. Perhaps, these were the consequences of oxidative stress (differences in redox status) associated with a given pathological condition, which contributed to the alterations of cellular metabolism and interfered with iterative-enzymatic DNA modification. In this context it is worth mentioning that our previous study documented presence of oxidative stress in leukocytes from patients with colorectal cancer and polyps [
12,
13].
While the involvement of TETs in formation of all epigenetic modifications analyzed in this study raises no controversies, still little is known about the regulation of this process. Specifically, it is unclear why the oxidation of 5-mCyt either stops at 5-hmCyt stage or proceeds to 5-fCyt and 5-caCyt stages. One potential explanation is different affinity of TETs to 5-mCyt, 5-hmCyt and 5-fCyt (for review see [
39,
40]). It is also possible that different proteins/factors recognize the modifications and determine their fate [
41]. Interestingly, some recent experiments demonstrated that transcription factors, Myc and Max, and perhaps also a number of other regulatory proteins, may specifically recognize 5-caCyt, but have lesser affinity to 5-fCyt, and show only a trace of affinity to 5-mCyt and 5-hmCyt [
42]. Moreover, a recent study conducted by Xiong et al. [
43] showed that Sall4, an oncogenic protein which is overexpressed in colon cancer [
44], may further enhance TET2-catalyzed oxidation of 5-hmCyt.
Recently, various isoforms of TETs were identified (reviewed in [
45]), and it cannot be excluded that their activity may be tissue-specific. Moreover, miRNA may either upregulate or downregulate the expression of TETs mRNA [
46].
All the factors mentioned above may contribute to different activity of TETs in patients with various pathological conditions, which in turn may result in the formation of disease-specific epigenetic modification patterns.
Authors’ contributions
MK, ZB, AL recruited the participants, analyzed and interpreted clinical data. MS, EZ, MM, TD, AS, KL, JG, JS, MG, DG performed analyses of epigenetic DNA modifications, plasma ascorbate, retinol and α-tocopherol concentrations and gene expression. MS, TD, ALa collected and maintained database, analyzed data and prepared graphics for manuscript. RO, MF, DG designed the study, analyzed and interpreted data. All authors contributed in writing the manuscript. All authors read and approved the final manuscript.