Introduction

Glioblastoma (GBM) is the most common malignant primary tumor of the brain in adults and associated with a particular poor prognosis [1]. The current standard of care includes surgery, radiotherapy, and the monofunctional alkylating agent temozolomide (TMZ). Although TMZ has improved overall survival, most patients still develop tumor recurrence within a period of 7 months [2]. Therapy failure is often due to resistance evolution processes against TMZ, one of the major obstacles in GBM treatment [3].

Amongst DNA adducts created by monofunctional agents like TMZ, O6-methylguanine assumedly is the most important lesion mediating TMZ toxicity. These adducts are repaired by O6-methylguanin-DNA-methyltransferase (MGMT); if MGMT is absent, base mispairing triggers repetitive but unsuccessful mismatch repair (MMR) leading to subsequent DNA strand breaks, cell cycle arrest, and apoptosis [4, 5]. Resistance to TMZ at the time of diagnosis is mostly due to high levels of MGMT [6, 7]. In contrast, tumors with methylated MGMT promotor are very sensitive to TMZ, but resistance almost inevitably develops during treatment. Several authors suggested MMR deficiency being responsible for acquired TMZ resistance in MGMT-methylated tumors [6, 8,9,10,11,12,13] and strategies to restore the effect of MMR system have been postulated to improve the effect of TMZ [14].

Aside from TMZ, lomustine (CCNU) is another alkylating agent with proven efficacy in GBM therapy [15, 16]. While a proficient MMR is vital for the expression of TMZ cytotoxicity, it is supposedly inversely related to CCNU toxicity as interstrand links caused by bifunctional agents such as CCNU are repaired by MMR. MMR deficiency indeed was shown to increase sensitivity to bifunctional agents in different MMR-deficient non-glioma cell lines [17,18,19,20].

Clinical trials showed promising results for the combination of CCNU and TMZ in patients with newly diagnosed GBM, with the largest clinical benefit found for MGMT-methylated patients [21, 22]. In addition, CCNU monotherapy demonstrated good efficacy for recurrent cases [15]. However, the underlying mechanisms have not been elucidated to date.

We hypothesized that acquired resistance to TMZ in MGMT-methylated GBM cells is mediated by MMR deficiency and, therefore, might be accompanied by increased sensitivity to CCNU. We, therefore, sought to investigate the combination and the differential effects of TMZ and CCNU in the human GBM cell line U251 and a TMZ-resistant line U251/TMZ-R with regard to the role of MMR and MGMT.

Materials and methods

Cell lines and primary culture

The human glioblastoma cell line U251 was obtained from Cell Line Service (CLS; Eppelheim, Germany). Cells were maintained and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; PAN-Biotech GmbH, Aidenbach, Germany) supplemented with 10% fetal bovine serum (FBS; Biochrom AG, Berlin, Germany), 100 U/ml penicillin and 100 μg/ml streptomycin (Invitrogen, Darmstadt, Germany) and cultured at 37 °C in a 5% CO2 incubator.

Drugs

TMZ, CCNU, and CdCl2 were obtained from Sigma-Aldrich (St Louis, USA). TMZ was dissolved in dimethyl sulfoxide (DMSO), CCNU in ethanol, and CdCl2 in sterilized water. Stock solutions were stored at −20 °C.

Generation of a TMZ-resistant cell line

U251 cells were cultured in 75 cm2 cell flasks (Cellstar; Greiner BioOne, Nürtlingen, Germany) and allowed to adhere overnight. Cells were treated with 100 µM TMZ. Cell treatment was repeated every 24 h for 5 consecutive days. After those 5 days, exposure to the fresh TMZ was repeated every 3 days to a total of 3 weeks. This procedure has been previously described for U251 [13].

Treatment with drugs and irradiation

Cells were seeded in 25 cm2 flasks at a density of 2.0 × 105 cells. The chemotherapeutics were added 48 h later. CdCl2 was added 2 h before chemotherapy treatment for pre-incubation according to a protocol introduced by Yamauchi et al. [23]. 1 h after chemotherapy treatment, cells were irradiated with 2 Gy, corresponding to the daily dose employed in clinical practice [24]. Irradiation was performed with an X-ray generator (120 kV, 22.7 mA, variable time; GE Inspection Technologies, Ahrensburg, Germany).

Cell death detection

We used APC Annexin V (AxV)/7-amino-actinomycin D (7AAD) staining and flow cytometry to investigate TMZ and CCNU-induced cell death. After harvesting the cells 72 h after treatment by trypsinization, cell suspension (100 µl, 1 × 105 cells) and 5 µl of AxV and 7AAD (both BD Biosciences, Franklin Lakes, USA), respectively, were combined with 400 µl Ringer solution (B. Braun Melsungen AG, Melsungen, Germany) and incubated at 4 °C for 30 min. Cell death was determined using flow cytometry (Gallios, Beckman Coulter, Brea, USA) and its associated Kaluza 1.3 Software (Beckman Coulter, Krefeld, Germany). For each sample, a minimum of 2 × 104 events was assayed. Experiments were performed at least thrice with two replicates per run. AxV/7AAD-double-negative cells were considered to be viable cells, AxV-positive/7AAD-negative cells early apoptotic cells, and AxV/7AAD-double-positive cells late apoptotic/necrotic cells [25,26,27].

Cell cycle analysis

Cell cycle distribution was analyzed by Hoechst 33342 staining (BD Biosciences, Franklin Lakes, USA) and flow cytometry. 72 h after treatment, cells were harvested. 2 ml of cell suspension (≙ 2 × 106 cells) were combined with 10 ml of 70% ethanol (Carl Roth, Karlsruhe, Germany) at 4 °C for at least 2 h to fix the cells. Following incubation, cells were resuspended in 1 ml Ringer solution, combined with 3 μl of Hoechst, and incubated at 4 °C for 20 min. In each sample, 2 × 105 cells were assayed using flow cytometry. The cell cycle phase distribution was determined with the Kaluza 1.3 software.

Clonogenic assay

The effects of irradiation were determined with clonogenic assays. Cells were plated in 60-mm dishes (Nunc Thermo Fisher, Waltham, USA) with 300–1600 cells per dish. After 6–12 h, cells were treated with drugs and irradiated with increasing doses of 2, 4, 6, 8, and 10 Gy (see above). After incubation for 10–14 days in drug-free fresh, medium cells were fixed with methylene blue for 30 min. Subsequently, colonies containing >50 cells were counted. The survival fraction (SF) was calculated as follows: SF at a given condition = colonies counted of the given condition/(cells seeded of the given condition × plating efficiency/100). Plating efficiency = percentage of untreated cells seeded that grow into colonies.

Immunostaining

To assess MMR protein expression, immunostaining and subsequent image analyses were performed following a standard protocol [28] with slight modifications. The following mouse monoclonal antibodies were used: anti-MLH1, anti-MSH6, anti-PMS2 (all BD Biosciences), and anti-MSH2 (Merck Millipore, Darmstadt, Germany). Primary antibodies were applied at a dilution of 1:100 and Alexa labelled secondary antibodies (Invitrogen; Life Technologies GmbH, Darmstadt, Germany) at 1:400. Images were captured by fluorescence microscopy (Leica DM 6000). Overlays were built using an image-processing software (Biomas 3.3 10/2004 MSAB). MLH1 foci were counted as previously described [28, 29].

Pyrosequencing for promotor status determination

In both cell lines, quantitative methylation analyses of the MGMT promotor were performed by pyrosequencing (PyroMark Q24 MGMT-Kit [Qiagen]) in the Institute of Neuropathology, Erlangen.

Statistical analysis

If not indicated otherwise, results are expressed as the mean ± standard deviation (SD) of three independent experiments. Statistics were performed with IBM SPSS Statistics 22.0 for Windows (IBM Corporation, New York, USA) using the two-sided t test and the non-parametric Mann–Whitney U test. Significant differences with a p value of ≤0.05 are marked as *, very significant differences (p ≤ 0.01) as **, and highly significant differences (p ≤ 0.001) as ***.

Results

U251/TMZ-R was more resistant to TMZ than U251

Cytotoxicity of 500 µM TMZ for 72 h was determined by analyzing cell viability, apoptotic, and necrotic cell death using AxV/7AAD staining (Fig. 1a). Following exposure to TMZ, the percentage of viable U251/TMZ-R cells (AxV−/7AAD−) was significantly elevated compared to parental U251 cells (92.0 ± 2.0% for U251/TMZ-R and 69.2 ± 8.8% for U251, p = 0.03). Both early apoptotic cells (AxV+/7AAD−) and late apoptotic/necrotic cells (AxV+/7AAD+) were notably decreased in U251/TMZ-R compared to U251 (mean difference (MD) in apoptosis: 0.8 ± 0.3 vs. 2.7 ± 0.5%; p = 0.02; MD in necrosis: 5.9 ± 1.5 vs. 22.3 ± 6.63%; p = 0.04) (Fig. 1a, b).

Fig. 1
figure 1

Flow cytometric analysis of cytotoxicity of TMZ, CCNU, and combination treatment. Cells were treated with 500 µM of TMZ, 38.5 µM of CCNU, or a combination of both. Cell viability, apoptosis, and necrosis were detected by 7AAD/AxV staining. When U251 was compared to U251/TMZ-R, cell viability was calculated as cell viability (treatment condition)/cell viability (control). To compare apoptosis and necrosis rates between both cell lines, mean differences (MD) were calculated as the differences between apoptosis/necrosis rates of treatment condition and matching control of the relevant cell line. a Representative flow-cytometric histograms; b cell viability, apoptosis and necrosis after mock treatment, TMZ, CCNU or CCNU+TMZ for U251 and U251/TMZ-R. c Cell viability, apoptosis, and necrosis after treatment with increasing doses of CCNU. Note: U251/TMZ-R showed increased cell viability after treatment with TMZ but reduced cell viability after CCNU compared to U251. Combination treatment with CCNU and TMZ resulted in highly significantly reduced cell viability in U251 and U251/TMZ-R compared to CCNU or TMZ alone. Accordingly, apoptosis and necrosis rates were decreased after TMZ therapy and increased after CCNU/CCNU+TMZ (a, b). The differential effects of CCNU on parental and resistant cells were apparent for various concentrations of CCNU (c)

Full size image

Cell cycle distribution after exposure to 500 µM TMZ was analyzed by Hoechst 33342 staining (Fig. 2a). TMZ induced a G2/M block in parental U251 cells (G2/M fraction; Controls: 11.0 ± 1.8%, TMZ: 45.3 ± 15.4%; p = 0.05) and a distinct reduction of cells in G1 phase (76.5 ± 1.6 vs. 34.7 ± 7.6%; p = 0.03). U251/TMZ-R cells, however, did not accumulate in G2/M after TMZ treatment, and the cell cycle distribution remained largely unchanged (Fig. 2b).

Fig. 2
figure 2

Cell cycle effects of TMZ, CCNU, or combination treatment. Cells were treated with 500 µM of TMZ, 38.5 µM of CCNU, or a combination of both. Cell cycle distribution was analyzed using Hoechst staining and flow cytometry. a Representative flow-cytometric histograms; b cell cycle distributions after mock treatment, TMZ, CCNU or CCNU+TMZ for U251 and U251/TMZ-R. Note: TMZ did not induce a G2/M block in U251/TMZ-R, whereas CCNU and combination treatment induced a highly significant G2/M block. TMZ, CCNU, and CCNU+TMZ induced G2/M block in U251

Full size image

Growth inhibition was assessed using colony formation assay. TMZ suppressed colony formation in U251, but not in U251/TMZ-R (SF: 41.3 ± 11.6 vs. 100.0 ± 2.8%; p = 0.003).

U251/TMZ-R cells did not develop cross resistance to CCNU and cytotoxicity of CCNU was enhanced in U251/TMZ-R

Cytotoxicity of CCNU was analyzed using AxV/7AAD staining and flow cytometry as described for TMZ (Fig. 1a). After exposure to 38.5 µM CCNU for 72 h, cell viability (AxV−/7AAD−) was decreased in U251/TMZ-R compared to U251 (56.4 ± 6.5 vs. 80.8 ± 4.6%; p = 0.02). Late apoptosis/necrosis (AxV+/7AAD+) was notably increased in U251/TMZ-R (30.7 ± 5.7 vs. 12.3 ± 3.3%; p = 0.04). Thus, cells were very sensitive to CCNU-induced cell death with an even more cytotoxic effect of CCNU in U251/TMZ-R cells than in U251 cells (Fig. 1b, c). CCNU induced G2/M arrest (Fig. 2b) and suppressed colony formation in both cell lines (data not shown), strengthening the conclusion that TMZ-R cells were not cross-resistant to CCNU.

Combination of CCNU and TMZ led to increased cytotoxicity in both cell lines and overcame resistance

The combination of CCNU (38.5 µM) and TMZ (500 µM) had stronger effects on cell viability and late apoptosis/necrosis than each single drug for U251 and U251/TMZ-R (Fig. 1b). The effects of CCNU+TMZ were even stronger in resistant cells (cell viability: 26.0 ± 2.7 vs. 39.7 ± 1.1%; p = 0.01, MD in necrosis: 60.6 ± 2.9 vs. 48.1 ± 0.6%; p = 0.02). Similarly, the combination of TMZ and CCNU led to an increase in G2/M arrest (Fig. 2b).

Inhibition of MMR increased sensitivity to CCNU in U251

Cells were pre-incubated with CdCl2, an inhibitor of MMR, at a concentration of 2.5 µM, chosen for its minimally toxic effects (Fig. 3a). 72 h after exposure to 4.8 µM CCNU, cells pre-incubated with CdCl2 showed significantly increased rates in late apoptosis/necrosis (3.3 ± 0.9 vs. 8.2 ± 1.0% for CCNU vs. CCNU + CdCl2; p = 0.02) and decreased rates in cell viability (93.6 ± 1.8 vs. 86.5 ± 1.8%; p = 0.05, Fig. 3d).

Fig. 3
figure 3

Inhibition of DNA MMR by CdCl2. 2 h before drug treatment, cells were pre-incubated with CdCl2, an inhibitor of MMR, at a concentration of 2.5 µM considered to have no effects on cell viability (a) and cell cycle distribution (b). CCNU was applied at a minimally toxic concentration of 4.8 µM to avoid additive cytotoxic effects, TMZ was applied at 500 µM. Cell cycle phases (b), cell viability, and late apoptosis/necrosis rates (c, d) were compared. Cells treated with CCNU and TMZ but without CdCl2 served as controls. Note CCNU-induced G2/M-block was more distinct in cells pre-incubated with CdCl2 than in controls. Rate of necrosis was significantly increased in CCNU treated cells after pretreatment with CdCl2 compared to CCNU or CdCl2 alone and single treatment with either CCNU or CdCl2 did not result in significantly elevated necrosis rate compared to control levels

Full size image

Correspondingly, G2/M arrest after exposure to CCNU was significantly more distinct in cells pre-incubated with CdCl2 (G2/M fraction: 15.7 ± 2.2 vs. 23.5 ± 1.4% for CCNU vs. CCNU+CdCl2; p = 0.04, Fig. 3b). Although TMZ toxicity was slightly decreased in cells pre-incubated with CdCl2 with regard to cell viability (60.7 ± 8.0 vs. 65.1 ± 1.0%; p = 0.64), late apoptosis/necrosis rate (30.1 ± 4.7 vs. 25.8 ± 2.6%; p = 0.53), and G2/M arrest (61.1 ± 2.3 vs. 58.1 ± 4.8%; p = 0.61), no significant effects were observed (Fig. 3c). Our results indicate that MMR inhibition sensitized cells to CCNU but not to TMZ.

Radio-sensitivity

To investigate effects of irradiation, clonogenic assays were performed. No significant differences between parental and resistant cells were detected for doses of 2 and 6 Gy with regard to growth restriction. However, significantly reduced colony formation was observed in TMZ-resistant cells for radiation doses of 4, 8, and 10 Gy, although the observed effects were minimal (Fig. 4). Furthermore, we did not observe differences in radio-sensitivity between both cell lines with regard to cell viability, apoptosis, and cell cycle distribution (data not shown).

Fig. 4
figure 4

Effects of radiation on U251 and U251/TMZ-R. Radiation with 2–10 Gy suppressed colony formation at an effective level in both parental U251 and resistant U251/TMZ-R with dose-dependent toxicity. SF for U251 vs. U251/TMZ-R: 2 Gy 66.1 ± 8.64 vs. 81.25 ± 4.03%, 4 Gy 51.13 ± 3.32 vs. 59.13 ± 1.46%, 6 Gy 37.69 ± 1.54 vs. 36.66 ± 2.08%, 8 Gy 14.32 ± 0.96 vs. 18.79 ± 0.29%, 10 Gy 3.53 ± 0.05 vs. 5.25 ± 0.15%

Full size image

Molecular characterization of U251/TMZ-R

U251 and U251/TMZ-R were tested for MGMT promotor methylation by pyrosequencing and promotor methylation was detected in both cell lines (Fig. 5). When tested for expression of MMR proteins, MLH1 was only activated in U251 (17.7 ± 1.1 vs. 0.4 ± 0.0 foci/cell; p = 0.04). For MSH2, MSH6, and PML1, no countable foci were detected in both cell lines. Our results suggest that TMZ resistance acquired in U251/TMZ-R was not related to changes in MGMT status but to deficiency of the MMR protein MLH1 (Fig. 6).

Fig. 5
figure 5

MGMT promotor methylation status in U251 and U251/TMZ-R as assessed by pyrosequencing. U251 and U251/TMZ-R both showed methylation of the MGMT promotor

Full size image
Fig. 6
figure 6

Expression of MMR protein MLH1. U251 and U251/TMZ-R were stained by anti-MLH1 (green) and DAPI (blue) (a). MLH1 foci were counted and the number of foci per cell was calculated (b). MLH1 expression was significantly decreased in U251/TMZ-R compared to U251

Full size image

Discussion

In our study, TMZ-resistant MGMT-methylated GBM cells showed increased resistance to TMZ-induced cell death, but intriguingly increased sensitivity to CCNU-induced cell death compared to parental U251 cells.

Beyond that, the combination of CCNU and TMZ was more effective than each single agent in both cell lines regarding drug-induced cell death. It is especially interesting that combination of CCNU und TMZ was even more effective in resistant cells than in non-resistant parental cells.

Our observations are well supported by clinical trials. Herrlinger et al. investigated the combination of CCNU and TMZ in the single arm phase II UKT-03 trial. In an updated analysis after an extended follow-up, Glas et al. reported long-term survival especially in the subgroup that received intensified TMZ and CCNU dose [21, 22]. In the recurrent setting, the REGAL trial showed an impressive overall survival of 9.8 months in the arm that received CCNU alone [15].

Apart from repairing DNA adducts caused by monofunctional agents like temozolomide, MGMT also repairs the O6-chloroethylguanine residues induced by bifunctional agents such as CCNU; thus, elevated MGMT levels lead to cross resistance between both drugs [4, 30, 31]. When MGMT capacity is saturated by an excess of O6-methylguanine produced, TMZ treatment causes base-pair mismatches and replication errors, triggering repetitive but unsuccessful MMR. This leads to continuous DNA strand breaks, cell cycle arrest in G2 phase, and subsequent apoptosis in MMR-proficient cells [4, 5, 32]. In contrast, cells with MMR deficiency possess relative resistance to monofunctional agents such as TMZ and a correlation between MMR deficiency and GBM recurrence has been reported [10,11,12, 33]. In line with these previous studies, changes in MGMT promotor methylation were not involved in the acquisition of resistance in our study. Conversely, expression of the MMR protein MLH1 was strongly decreased in the resistant cell line. A decrease in MLH1 expression has been reported to occur early during acquisition of TMZ resistance in vitro and in vivo [12]. MLH1 expression was significantly decreased in recurrent GBM [33]. This proves the importance of MLH1 for a proficient MMR response in TMZ-sensitive GBM cells, supporting our results.

Our findings of increased CCNU toxicity in TMZ-resistant, MMR-deficient GBM cells are mechanistically supported by previous preclinical studies in non-glioma cell lines: Interstrand links caused by bifunctional agents like CCNU were shown to be repaired by MMR, leading to resistance to bifunctional agents [17,18,19]. To further examine the effects of MMR inhibition in U251, we used CdCl2, a substance targeting several proteins involved in MMR, at a concentration of 2.5 µM. At this concentration, CdCl2 effectively inhibits MMR [34, 35] and had minimally toxic effects (Fig. 3a). Sensitivity to CCNU was increased in cells pre-incubated with CdCl2 which is in line with results from Yamauchi et al. who reported that BCNU-resistant leukemia cells were partially sensitized to CCNU as a result of CdCl2-mediated MMR inhibition [23]. Although we would have expected decreased sensitivity to TMZ in cells pre-incubated with CdCl2, this did not occur at a significant level in our experiments. This might be due to the low CdCl2 concentration chosen on account of high toxicity levels; at a concentration of 100 µM, MMR efficiency is reduced by approximately 95%, but at 1 µM only by 2,7% [34]. A concentration of 2.5 µM might be just enough to cause MMR-deficiency-mediated sensitization to CCNU, yet not enough to induce TMZ resistance.

Aside from chemotherapy, the current standard of care includes radiotherapy [2]. Re-irradiation for recurrent GBM modestly increases overall survival [36, 37]. Accordingly, we did not detect meaningful differences in the effectiveness of radiotherapy between both cell lines. Both cell lines were sensitive to irradiation with dose-dependent growth restriction. The role of MMR in radio-sensitization has been controversially discussed [11, 38, 39]. However, in our study, MMR deficiency did not affect radio-sensitivity in U251, confirming the importance of radiotherapy in recurrent GBM.

To the best of our knowledge, this preclinical investigation is the first to show that TMZ resistance mediated by MMR deficiency in MGMT-methylated GBM cells is accompanied by increased sensitivity to CCNU and to combined CCNU and TMZ. Radiosensitivity was preserved in resistant cells. These findings have important clinical implications as acquired TMZ resistance is one of the major obstacles in the treatment of GBM [3] and increased sensitivity to CCNU could be a work-around. As CCNU resistance is mediated by up-regulation and TMZ resistance by downregulation of MMR, we speculate that TMZ resistance evolution might even be preventable by concomitant CCNU administration. This may also explain the promising results for combined CCNU and TMZ in MGMT-methylated patients [21, 22].

Conclusion

This study showed promising results for CCNU in MMR-mediated TMZ resistance, indicating that further pre-clinical and clinical research is clearly warranted. Beyond that, our findings may provide the missing link between already well-described clinical and preclinical observations. Upcoming clinical trials like the German NOA-09 will provide further answers on the role of combined TMZ and CCNU.