Revealing the epigenetic effect of temozolomide on glioblastoma cell lines in therapeutic conditions

Temozolomide (TMZ) is a drug of choice in glioblastoma treatment. Its therapeutic applications expand also beyond high grade gliomas. However, a significant number of recurrences and resistance to the drug is observed. The key factor in each chemotherapy is to achieve the therapeutic doses of a drug at the pathologic site. Nonetheless, the rate of temozolomide penetration from blood to cerebrospinal fluid is only 20–30%, and even smaller into brain intestinum. That makes a challenge for the therapeutic regimens to obtain effective drug concentrations with minimal toxicity and minor side effects. The aim of our research was to explore a novel epigenetic mechanism of temozolomide action in therapeutic conditions. We analyzed the epigenetic effects of TMZ influence on different glioblastoma cell lines in therapeutically achieved TMZ concentrations through total changes of the level of 5-methylcytosine in DNA, the main epigenetic marker. That was done with classical approach of radioactive nucleotide post-labelling and separation on thin-layer chromatography. In the range of therapeutically achieved temozolomide concentrations we observed total DNA hypomethylation. The significant hypermethylating effect was visible after reaching TMZ concentrations of 10–50 μM (depending on the cell line). Longer exposure time promoted DNA hypomethylation. The demethylated state of the glioblastoma cell lines was overcome by repeated TMZ applications, where dose-dependent increase in DNA 5-methylcytosine contents was observed. Those effects were not seen in non-cancerous cell line. The increase of DNA methylation resulting in global gene silencing and consecutive down regulation of gene expression after TMZ treatment may explain better glioblastoma patients’ survival.

Recently we showed that temozolomide modulates 5-methylcytosine (m 5 C), the main epigenetic marker, contents in cancer cells' DNA [17]. An increasing TMZ concentration in the short time induced an increase of total m 5 C content in DNA (hypermethylation), but at extremely high (1000 μM) TMZ concentration and longer exposition time DNA hypomethylation was observed. DNA hypermethylation can be only explained by activation of DNA methyltransferase (DNMT), because TMZ is not a substrate for DNMT [24]. The other TMZ effect on the cell is chemical and oxidative stress. In such conditions 5-methylcytosine is a target for oxidation, what results with the decrease (demethylation, hypomethylation) of its total genomic contents [25]. Therefore changes in DNA methylation provide an appropriate picture of the disease [26]. In our previous studies we have also shown that m 5 C is a good marker for the diagnosis of brain tumors, hypertension, and other diseases [27][28][29]. However, the concentrations of TMZ used (100-1000 μM) in our previous study [17] didn't correspond well with those reached in regular therapy in brain tissue and cerebrospinal fluid [19,[21][22][23]. Therefore in the present analysis we decided to focus on the concentrations potentially achieved in the brain during TMZ therapy (mean 3 μM±1.5 μM [23]).
The aim of the current project was to evaluate the epigenetic influence of temozolomide treatment explaining its effects in glioblastoma, and other pathologies, in therapeutic concentrations and mimicking the therapeutic schemata.

Chemicals
Temozolomide (Sigma-Aldrich/Merck) was freshly dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich/Merck) at a concentration of 0.103 M. TMZ stock solution was used to prepare the required concentration with the complete medium.
Safety procedures were used with cell lines to avoid contamination from other cell lines and microbes. Low-passage (up to 25) cell lines were used. Plasmocin™ Prophylactic (InvivoGen), a broad-spectrum anti-mycoplasma reagent, was used at 5 μg/ml in liquid media to prevent mycoplasma contamination. It is also active on a broad range of Gram-positive bacteria, such as Staphylococcus species, and Gram-negative bacteria. Medium with fresh Plasmocin was changed every 3 days. Plasmocin™ Prophylactic (5 μg/ml) exhibits no toxicity in eukaryotic cells. Cell cultures were monitored with PlasmoTest™ (InvivoGen). Mycoplasma was not detected in our cell cultures.
The control cells were treated with H 2 O or DMSO only. After appropriated (indicated above) time with TMZ, the cells were washed with PBS, trypsinized and collected by centrifugation at 3500 rpm for 5 min. The cellular pellets were quickly frozen and stored at-20˚C for DNA isolation.
MTT assay for TMZ was carried out during our previous experiments [17]. It was shown that TMZ treatment in the range of 1-2000 μM triggers cell death apoptosis independent.

DNA isolation from cell cultures
DNA from tissue samples was extracted with Genomic Mini kit (A&A Biotechnology, Poland). Shortly, tissue samples were incubated with RNase A, and then with proteinase K. After centrifugation (15000 rpm for 3 min), the supernatant was applied to mini column and DNA bound to the column was eluted with Tris-buffer pH 8.5 and stored at -20˚C for further analysis. The purity and concentration of DNA preparations was checked by measuring of UV absorbance at 260 and 280 nm. The A 260 /A 280 ratio was 2.0-2.1.

DNA hydrolysis, labelling and TLC chromatography
1 μg of dried DNA was dissolved in a succinate buffer (pH 6.0) containing 10 mM CaCl 2 and digested with 0.001 units of spleen phosphodiesterase II and 0.02 units of micrococcal nuclease in 3.5 μl total volume for 5 h at 37˚C. 0.17 μg of DNA digest was labeled with 1 μCi [γ-32 P]ATP (6000 Ci/mM; Hartmann Analytic GmbH) and 1.5 units of T4 polynucleotide kinase in 3 μl of 10 mM bicine-NaOH pH 9.7 buffer containing 10 mM MgCl 2 , 10 mM DTT, and 1 mM spermidine. After 0.5 h at 37˚C 3 μl of apyrase (10 units/ml) in the same buffer were added and incubated for another 0.5 h. The 3' nucleotide phosphate (from [ 32 P]dNp) was cleaved off with 0.2 μg RNase P1 in 500 mM ammonium acetate buffer, pH 4.5. Identification of [ 32 P]dN was performed with a two-dimensional thin-layer chromatography on cellulose plates (Merck, Germany) using solvent system: isobutyric acid:NH 4 OH:H 2 O (66:1:17 v/v) in the first dimension and 0.1 M sodium phosphate (pH 6.8)-ammonium sulfate-n-propyl alcohol (100 ml/60 g/ 1.5 ml) in the second dimension. Radioactive spot analysis was done with the Phosphoimager Typhoon Screen (Pharmacia, Sweden) and ImageQuant Software (GE Healthcare, USA). For precise calculation we used the amount of material in spots corresponding to m 5 dC (5-methylcytosine), dC (cytosine) and dT (thymine). The total m 5 C contents was calculated as R = (m 5 dC/(m 5 dC+dC+dT))×100 (Fig 2).

Statistical analysis
Microsoft Excel 2010 software with Data Analysis package was used for statistical analysis of all data. The data are the result of three independent experiments. Descriptive statistics function was used to generate mean and SD (S1-S3 Tables). One-tailed t-test was used to calculate significant differences of R values for tested samples as compared with control in experiments 1-3.

Results
We analyzed changes in 5-methylcytosine (epigenetic marker) contents in DNA of three types of human glioblastoma cell lines (T98G, U138, U118), and one human keratinocyte cell line (HaCaT) after treatment with 0.5-100 μM of TMZ in the time range between 3 hrs and 7 days in different treatment regimens. These TMZ concentrations were chosen to cover the possible concentration of TMZ in the brain during chemotherapy (0.6±0.3μg/ml, corresponding to 3 ±1.5 μM) [19,[21][22][23], and perform the observation after reaching the peak concentration in the brain (2.0±0.8 hrs) [23], as well as standard chemotherapy course time [1]. In the first experiment total DNA methylation in cell lines (T98G, U138, U118, HaCaT) was estimated after treatment with single doses of TMZ (0.5, 1, 3, 5, 10, 20, 30, 50, 100 μM) and incubation for 3, 12, 24 and 48 hrs (Figs 3-6). Analyzing the time factor (in every set of bars of the same TMZ concentration) in glioblastoma cell lines one can clearly see DNA demethylation in all glioblastoma cell lines increased with prolonging time, and evident extreme at 48 hrs. That effect is little influenced by increasing TMZ dose to a degree depending on the line (more in U118 and T98G). Taking into account the concentration factor we observe initial DNA hypomethylation (in comparison to control) in glioblastoma cell lines in concentration range 0.5-10 μM (depending on the cell line, p values are presented in S1 Table). That effect is particularly interesting because that range covers the concentrations that are most probably reached in the brain during TMZ treatment. After reaching the concentration of 10 μM (3 μM in case of U118) dose dependent DNA hypermethylation is observed. The dependence on dose increase in all cell lines was statistically significant ( ��� p<0.001), whereas dependence on time was observed at significant level up from doses: 3 � /5 �� /10 ��� μM in T98G (Fig 3), 1 ��� μM in U138 (Fig 4), 0.5 � /1 �� /3 ��� μM in U118 (Fig 5), and only for TMZ 10 � μM in HaCaT (Fig 6). Both phenomena were not seen in HaCaT cell line, where in general hypermethylation is detected. The results of the first experiment show that therapeutically achieved TMZ concentrations can be too low to result in hypermethylation, which is regarded as a prognostically better state [30]. The next experiment was intended to partially mimic the Stupp scheme for adjuvant TMZ therapy [1], and therefore to explore the effect of repetitive TMZ doses on DNA methylation. TMZ was administered to cell lines (T98G, U138, U118, HaCaT) at 0.5, 1, 3, 5, 10 μM concentrations for 5 days, wherein after 24, 48, 72 and 96 hrs cells were washed with PBS and placed in fresh medium with indicated TMZ concentrations. The total DNA methylation was estimated after 5 th day of treatment (Fig 7). The general TMZ dose-dependent hypermethylation is observed. Even the lowest concentration (0.5 μM) resulted with the increase of DNA methylation. That is not the case with HaCaT cell line, where rise in R level is scarce, and not clearly related to TMZ dose. One can see that repeated dosing of TMZ (every day) allows bypassing the hypomethylation effect that occurred after single therapeutically achieved TMZ doses in previous experiment (Figs 3-6). The results (S2 Table) are statistically significant for all cell lines ( ��� p<0.001).
The goal of the final experiment was to evaluate the time effect on total DNA methylation during prolonged TMZ treatment within concentrations reachable in the brain interstitium. Cell cultures (T98G, U138, U118, HaCaT) were treated with the single dose of TMZ (0.5, 1, 3, 5, 10 μM) and were left for 1-7 days. The global DNA methylation was estimated after 1 st , 2 nd , 3 rd , 4 th , 5 th , 6 th , and 7 th day of incubation (Figs 8-11). For glioblastoma cell lines we observed general hypomethylation (when compared with day 1) tendency during time with some stabilization of R level between days 2-6, and clear drop down at day 7. The hypomethylating effect is overcome with increasing TMZ dose. In line U118 (Fig 10) all TMZ concentrations induced hypermethylation over control condition (DMSO treatment), while in T98G (Fig 8) and U138 (Fig 9) that was the case only in highest concentrations (5 and 10 μM). That effect lasted until day 6 at maximum. At day 7 general loss of DNA methylation below the control level was observed. The results of that experiment suggest time limitation of pause in chemotherapy to 4-5 days (depending on the line). The effect of time and initial TMZ dose on HaCaT (Fig 11) culture is scarce, the R values change in a very small range. The numeric results of that experiment are presented in S3 Table.

Discussion
The aim of our work was to evaluate the effects of temozolomide action on the total contents of m 5 C in DNA of glioblastoma cell lines in conditions mimicking clinical treatment situations. The choice of the research area was driven by the fact that epigenetic modifications, particularly DNA methylation, are very sensitive to environmental (chemotherapy) changes and react faster than genetic ones, therefore that level of observation seems to be most promising in such an etiologically complex phenomenon as cancer [31,32]. In cancer, global hypomethylation and local hypermethylation are observed [33]. Global hypomethylation is regarded as a result of unspecific, broad oxidative DNA damage (all bases including m 5 C), which is associated with cancer formation and progression by many different pathways, whereas local hypermethylation is regarded as a carcinogenic event through repression of promoters of tumor suppressor genes, thereby facilitating cancer formation [34-37]. Nevertheless, globally decreased DNA methylation, which appears to correlate with carcinogenesis, is an adverse phenomenon in cancer treatment [35,37].
Decreasing amount of m 5 C in DNA of cells treated TMZ can be explained straightforward by oxidative removal of methyl group from m 5 C (demethylation) [38,39] or by demethylating activity of DNA Methyltransferases (DNMTs) in redox state, which can be induced by cancer chemotherapeutics [40,41]. Although there are data that the oxidation of 5-methylcytosine to 5-hydroxymethylcytosine can be done with oxo-reductases (TET enzymes), the mechanism of cellular DNA damage by induction of reactive oxygen species (ROS) is also plausible, The proof for that is the synthesis and presence of 8-oxo-deoxyguanosine in DNA. Newly published data provide evidence, that also in the case of loss of TET function, general DNA demethylation occurs [42]. Recently we showed fluctuations in m 5 C level in DNA and parallel 8-oxo-guanosine synthesis, which is the product of hydroxyl radical reaction with guanosine [25]. Therefore, a general conclusion is that TMZ induced genomic demethylation (hypomethylation) is a result of global non-specific oxidative stress and damage of DNA components. The level of hypomethylation depends on time and dose of treatment.
5-methylcytosine is a product of enzymatic reaction catalyzed by DNA Methyltransferases, where the only methyl donor is SAM (S-adenosyl methionine) [24]. Neither TMZ, nor TMZ degradation products are the substrates for DNMT. Therefore, the only mechanism through which a drug can increase m 5 C contents can be through the induction of DNMTs activation. The induction of DNA hypermethylation by drugs, hormones, and other biological compounds was already observed [43][44][45]. In our experiments we showed TMZ dose-dependent total DNA hypo-and hypermethylation, as well as repeating-dose-dependent hypermethylation. That can argue with the fact that several drugs that inhibit DNA methylation have been studied for the reactivation of tumor suppressor genes (from silencing with promoter hypermethylation) and repression of cancer cell growth [46], and also hypomethylation is regarded as an event helping with the chemotherapy effect. The problem is, the drugs don't act selectively on hypermethylated promoters only, but on the entire DNA, resulting in global DNA hypomethylation, the hallmark of cancer, aging, and other pathologic processes [29,33,37,47,48], as well as result with mutations in epigenetic enzymes (e.g. DNMTs), that are important for response to hypomethylating treatment [49]. Moreover, DNA damage response that is activated after treatment with radiation and different classes of chemotherapeutic drugs can result both in hypersensitivity or resistance of tumors to therapy and can be exploited for improvement of cancer treatment [50]. The concept of hypomethylation as the goal of treatment has to be probably reevaluated, especially when some results show better survival in globally hypermethylated gliomas [30]. What was also observed, even if MGMT promoter methylation was stable, LINE1 (global methylation surrogate) methylation status was not, resulting with decreased overall survival and GBMs post-treatment evolution with newly reprogrammed epigenetic status [51].
Besides our previous study [17] currently there are no data exploring the subject on total DNA methylation changes during TMZ therapy. In a study analyzing the impact of temozolomide in concentrations 0-200 μM during 24-72 hrs on T98G cell line Jakubowicz-Gil et al. showed that significant level of apoptosis is observed over TMZ concentration of 50 μM. That Temozolomide effects in therapeutic conditions effect was potentiated with increasing incubation time, and accompanied in higher concentrations with necrosis [52]. That is in concordant with our previous results [17], and supports the results of the present study-with no significant impact on cell apoptosis, necrosis, and autophagy in the concentrations under 50 μM, the total DNA hypomethylation is observed. Similar results were showed on U118 cell line [53].
In the experiment partially mimicking the Stupp scheme (only chemotherapy factor) we observed dose-dependent increase of total DNA methylation (m 5 C), however never up to the level previously estimated to that of less malignant tumors or healthy individuals [29]. Exploring the prolonged time effect we showed that any hypermethylating treatment effect of TMZ in our experimental cell lines lasted over 5-6 days after drug intake, then having a definite demethylating trend. In the last decades a new concept in TMZ administration emergedmetronomic chemotherapy-involving low doses administered on a frequent schedule, without taking a prolonged break. Such approach appeared to be effective for primary [54] and recurrent glioblastoma [55][56][57]. Also prolonged (> 6 cycles) standard TMZ therapy was proven to be successful [58,59]. There are data showing that this result does not depend on MGMT promoter methylation [55,60], but other indicate such connection [8]. However, there are also trials that present no significant difference in overall and progression free survival after implementation of a dose-dense protocol for primary glioblastoma [60]. The positive therapeutic effect of dose-dense treatment was shown on animal models [61], however the best results were achieved for daily doses approximately 4x higher (recalculated for human conditions) than those given in a standard TMZ therapy in humans [62,63].

Temozolomide effects in therapeutic conditions
The question concerning a real TMZ concentration achieved in the target tissue (brain) is only partially answered [19,[21][22][23]]. The doses we tested in our experiments (0.5-10 μM), being the result of previous studies measuring real TMZ contents in the brain [19,[22][23], showed DNA hypomethylation tendency compared to control. That effect was only overcome by repetitive TMZ doses. Even if not analyzed on the epigenetic background before, that effect (inadequate drug amount in target tissue) can be the reason for progression of the disease / chemotherapy resistance.
Many strategies are taken to increase the temozolomide concentration in the CNS, overcoming the boundaries of BBB [64][65][66][67][68]. What was shown earlier, simple implementation of radiotherapy, which is the damaging factor of BBB [69], improves the efficacy of TMZ [70]. That clinical action supports our results that increased TMZ dose results in total hypermethylation [present work, 17], and on clinical level effects in longer progression free and overall survival, as well as less general drug toxicity (because of better penetration of the given dose without the need to increase the intake).

Conclusions
We observed dose-dependent DNA hypermethylation in glioblastoma cell lines and lack of that effect in non-cancerous cell line. A high TMZ concentration induced a significant increase of m 5 C contents in DNA even in the short time. Therefore an increase of DNA methylation, resulting in genes' silencing and their expression down regulation, can be the reason for better glioblastoma patients' survival. However, in the range of concentrations achieved in target tissue during standard treatment, loss of total DNA methylation is observed, what can explain the resistance to TMZ therapy and tumor recurrence. That effect is somehow overcome by repeated TMZ doses in short periods of time.
Our results show the epigenetic response to the therapeutic doses and regimens of temozolomide chemotherapy in glioblastoma. They provide a new insight into the molecular background of the treatment effects, and possible therapy modifications.