Temozolomide may cause MGMT-positive glioblastoma cells to become more invasive through Snail-1 and Cx43 pathways

U87 and T98G cells display considerable differences in the reactivity to TMZ. In contrast to T98G cells, their U87 counterparts were highly reactive to TMZ. This is illustrated by a prominent inhibition of U87 proliferation (Figure S1A), the prolongation of U87 doubling-time (Figure 1A), and distinct TMZ-induced apoptotic response (Figure 1B; cf. Figure S1B). These effects were observed within seven days of incubation in the presence of TMZ administered at the concentrations of 5 and 25 μM, i.e., within the range of its therapeutically relevant doses. Consequently, the survival rate of U87 cells under TMZ stress was considerably lower than that of T98G cells, as shown by IC₅₀ values that were estimated with the clonogenic assay (103.5 μM (T98G) vs. 6.33 μM (U87); Figure 1C). Relatively high TMZ-resistance of T98G cells (Figure 1A,C) was correlated with the relatively sparse fractions of apoptotic cells in TMZ-treated T98G populations (Figure 1B). The differences in TMZ-resistance between T98G and U87 cells can be ascribed to the MGMT-negative phenotype of U87 cells (Figure 1D). Actually, T98G expressed high MGMT levels, whereas we did not observe any significant increase of the MGMT levels in U87 cells after the TMZ treatment (Figure 1E). Despite these differences, an induction of U87 and T98G invasiveness was observed after TMZ application (Figure 1F). These data indicate that TMZ induces the microevolution of GBM invasiveness in a manner independent of MGMT.

We focused on the potentially pro-invasive phenotypic shifts in TMZ-treated U87 populations to further identify the consequences of TMZ stress for the phenotype of GBM cells. The short-term administration of 25 μM TMZ increased the motility and displacement efficiency of U87 cells, followed by the gradual reversal of this effect starting after 48–72 h (Figure 2A). An enriched sub-population of highly motile U87 cells seen after 72 h of TMZ treatment remains in concordance with the data from transmigration assays, which revealed increased the fraction of invasive U87 cells (Figure 1E) that apparently constitutes the U87 “invasive front”. These effects were followed by the shifts of TMZ-treated U87 cells towards an “epithelioid” morphology (Figure 2B). Namely, spindle-like U87 cells were replaced by increasing the fraction of well spread, hypertrophic cells, reminiscent of the dormant “giant” cells that were observed elsewhere [49].

A corresponding, although less pronounced, effect was observed in the presence of 5 μM TMZ (Figure S2). Concomitantly, gradual phenotypic shifts of TMZ-treated U87 cells were accompanied by the rearrangements of their cytoskeleton architecture, incl. the maturation of vinculin⁺ focal adhesions (Figure 2C). Even though we could not observe any changes of Snail-1 levels in TMZ-treated U87 cells (Figure 2D), our data show a transient TMZ-triggered expansion of invasive U87 lineages. It is apparently attenuated by the cytotoxic TMZ effects in MGMT^low cells (cf. Figure 1A,B). Thus, the MGMT function is necessary for TMZ-induced microevolution of functional GBM lineages.

Further studies were performed to estimate the role of “innate” MGMT-dependent drug-resistance in the long-term microevolution of GBM cell invasiveness under TMZ stress. A comparison of U87 and T98G morphology/motility revealed a lower motility of MGMT^high T98G cells (Figure 3A).

In contrast to homogeneous, predominantly spindle-shaped MGMT^low U87 cells, a co-existence of epithelioid and fibroblastoid/mesenchymal cells was observed in MGMT^high T98G populations under the control conditions. This makes them a suitable model for the microevolutionary studies. The heterogeneous phenotype of T98G cells was accompanied by a sub-population of relatively fast-migrating T98G cells. Similarly to their U87 counterparts (cf. Figure 2), T98G cells reacted to the short-term TMZ treatment with an acceleration of their motility (Figure 3B). The prolongation of TMZ treatment did not result in T98G cell hypertrophy/flattening/senescence (as in the case of U87 cells). Instead, phenotypic shifts towards “fibroblastoid” morphology were detected by NIC and immunofluorescence microscopy (Figure 3C). This is summarized by morphometric analyses, which revealed an increased abundance of rear-front polarized cells in the long-term (30 days) TMZ-treated T98G populations (Figure 3D). Concomitantly, we observed an increased invasiveness and a considerable Cx43 up-regulation in these cells, as manifested by transmigration (Figure 3E) and immunoassays, respectively (Figure 3F). These pro-invasive effects of the long-term TMZ treatment were accompanied by a relatively low fraction of apoptotic cells in the populations that were cultivated in the presence of 25 μM TMZ (Figure 3G). Thus, TMZ induces the microevolution of invasive subsets of Snail-1⁺/Cx43⁺/MGMT^high T98G cells.

Concomitant induction of the invasiveness and Cx43 levels in MGMT^high T98G cells that were subjected to the long-term TMZ stress may suggest the involvement of Snail-1-dependent signaling in the microevolution of GBM invasive front. Therefore, we looked into the involvement of Snail-1 in TMZ-induced pro-invasive phenotypic shifts of T98G cells. After a transient siRNA-induced down-regulation of Snail-1 in these cells, its levels reached ca. 60% of the control value (Figure 4A). It was sufficient to evoke the morphological transition of T98G cells towards “epithelioid” phenotype and significantly attenuate their motility (Figure 4B).

Snail-1 up-regulation experiments further confirmed the notion on the crucial role of Snail-1 in the regulation of T98G invasiveness. Ectopically increased Snail-1 levels in T98G cells correlated with their shifts towards fibroblastoid morphology (Figure 4C) and with their considerably enhanced motility (Figure 4D). Furthermore, Transwell assays revealed the correlation between Snail-1 levels and the transmigration index (TMI) of T98G cells. This is illustrated by a gradual decrease of this parameter, along with Snail-1 down-regulation and the opposite effect after Snail-1 up-regulation (Figure 4E). Notably, we did not observe any effect of TMZ on the transmigration efficiency of T98G cells after Snail-1 silencing, whereas TMZ considerably augmented this parameter after Snail-1 up-regulation. These effects were accompanied by the correlation between Snail-1 levels and TMZ-resistance of T98G cells, as illustrated by the reduced viability of TMZ-treated T98G cells undergoing Snail-1 knock-down (Figure 4F). Collectively, our data confirm that Snail-1-dependent signaling participates in TMZ-induced invasiveness of T98G cells.

T98G and U87 cells constitute a suitable experimental tool for the analyses of the interrelations between MGMT and Snail-1 during TMZ-induced phenotypic GBM microevolution due to the differences in MGMT levels. Additionally, native U87 cells showed considerably higher levels of Cx43 than their T98G counterparts (Figure 5A). The application of TMZ considerably up-regulated Cx43 in U87 cells, as illustrated by cytofluorimetric analyses (Figure 5B) and immunoblotting (Figure 5C). However, an ectopic down-regulation of Cx43 had no effect on the motility of U87 cells (Figure 5D). Concomitantly, it reduced their transmigration potential in the presence of TMZ (Figure 5E), even though we observed an increase of otherwise very low U87 transmigration in its absence (cf. Figure 1F). In turn, morphological shifts of T98G cells towards epithelioid morphology and the inhibition of their motility was observed upon Cx43 down-regulation in T98G cells (Figure 5F). This effect corresponded to that observed after Snail-1 down-regulation (cf. Figure 4), and it was accompanied by increased sensitivity of T98G cells to TMZ (Figure 5G). It is illustrated by the attenuation of their invasiveness and viability after concomitant Cx43 down-regulation/TMZ-treatment. These data unequivocally show the involvement of Cx43 in the regulation of T98G invasiveness at the onset of its microevolution.

The effects of Cx43 up-regulation on the phenotype of T98G cells and their responsiveness to TMZ further confirmed this notion. Namely, transient Cx43 up-regulation in T98G cells (Figure 6A) increased their motility (Figure 6B). Concomitantly, TMZ exerted an enhancing effect on T98G transmigration potential (Figure 6C) after Cx43 up-regulation. These data remained in contrast to the negligible reactivity of MGMT^low U87 cells to the Cx43 signaling. This is illustrated by the lack of the effects of Cx43 up-regulation on their motility (Figure 6D,E) and their low invasiveness after concomitant Cx43 up-regulation and TMZ treatment (Figure 6F), even though an increase of otherwise very low U87 transmigration could be observed in its absence (cf. Figure 1F). Finally, we did not see any effects of Cx43 up-regulation on the viability of T98G and U87 cells (Figure 6G). Collectively, these observations show the MGMT-dependent involvement of Cx43 in the regulation of GBM invasiveness under TMZ stress.

Human GBM T98G (ATCC, CRL-1690) and U87 cells (ATCC, CRL-1690) were routinely cultivated in high glucose (4500 g/L) DMEM medium (Sigma, St. Louis, MO, USA), which was supplemented with 10% fetal bovine serum (FBS; Gibco) and 1% Antibiotic-Antimycotic Solution (Sigma) [59,60]. For endpoint experiments, the cells were harvested with 0.25% trypsin in Ca²⁺/Mg²⁺-free PBS (Corning, Corning, NY, USA) supplemented with 0.5 mM EDTA (UltraPure^TM; Invitrogen, Carlsbad, CA, USA), counted with Z2 particle counter (Beckman-Coulter, Brea, CA, USA) and then seeded into multi-well tissue culture plates (Falcon). For some experiments U373 (human astrocytoma; ATCC, HTB-17), U118 (ATCC, HTB-15), Ln229 (ATCC, CRL-2611), Ln18 (ATCC, CRL-2610) and F98 cells (rat glioblastoma; ATCC, CRL-2397) were cultivated, as described above. TMZ (Sigma; No. T2577) was dissolved in DMSO at the concentration of 100 mM, stored in −80 °C, and administered at the concentrations of 5 or 25 μM. Unless stated otherwise, the medium containing 0.25‰ DMSO (corresponding to 25 μM TMZ) was used as a vesicle control.

T98G and U87 cells were seeded into 12-well plates (Falcon) at the density of 5.3 × 10⁴ cells/cm² and allowed to attach, before the administration of 5 or 25 µM TMZ-supplemented medium. Subsequently, the cells were harvested every 24 h for the next four days (96 h) and counted in a Z2 particle counter. The obtained data were processed with GraphPad Prism software to calculate the kinetics of cell proliferation (incl. the doubling times). For the analyses of apoptosis, the cells were cultured in T-25 flasks (Falcon) in the presence of 25 µM TMZ for up to 30 days, trypsinized, re-suspended in Ca²⁺/Mg²⁺-free PBS, and subjected to AnnexinV/propidium iodide staining according to the manufacturer’s protocol (BD Pharmigen, Franklin Lakes, NJ, USA). Flow cytometric detection of apoptotic cells was performed with ImageStreamX imaging cytometer (Merck, Darmstadt, Germany). For each condition, at least 5000 singlets were collected. The data were processed using IDEAS 6.2 software (Merck; [59]). The biability of cells was analyzed using Trypan blue assay (Sigma; No. T8154) assay, as described before [37]. For the clonogenic assay, the cells were seeded into six-well plates (500 cells/well) and then incubated in the presence of 0–500 µM TMZ (or 0.5% DMSO) to form colonies for two weeks. Subsequently, the samples were fixated with formaldehyde (FA; 3.7%; 20 min at 37 °C), washed, stained with 0.01% Crystal Violet solution in EtOH, dried, and the colonies were counted. The data were processed with GrapPad Prism software to calculate IC50.

T98G/U87 cell movement was recorded with the Leica DMI6000B time-lapse video microscopy system, equipped with an incubation chamber (37 °C ± 0.2 °C)/(5% CO₂), IMC contrast optics, and DFC360FX CCD camera. The cells were seeded into 12-well plates at the density of 500 cells/cm², immersed in TMZ-containing medium for 48 h and their movements were recorded for the next eight hours (time step—10 min). The cell trajectories were constructed from a sequence of cell centroid positions recorded for eight hours at 300 s time intervals (using a dry ×20, NA—0.4 objective) to estimate the length of single cell trajectory (Distance; μm) and the single cell displacement (Displacement; μm). Cell trajectories were pooled to calculate the averaged values of Distance and Displacement at the population level (at least three independent experiments; n = 50 [61]). Invasive potential of T98G and U87 cells was examined with transmigration assay using the Transwell^TM inserts (8 µm micropores; Corning). The cells were seeded onto the top of microporous membrane at the density of 1 × 10⁴ cells/24-well plate compatible insert and allowed to transmigrate for 24–96 h in the absence/presence of TMZ, allowed to grow for the next 72 h, and then harvested and counted with Z2 particle counter. Transmigration index (TMI) was calculated as the percentage of transmigrating cells/population normalized against the time of transmigration (24–96 h) and cell proliferation rates [51].

Immunolocalization of Cx43, Snail-1, MGMT, vinculin, and α-tubulin was performed in formaldehyde/Triton X-100 fixed/permeabilized cells (FA; 3.7%; 20 min at 37 °C)/Triton X-100 (0.1%; 10 min at RT). Non-specific binding sites were then blocked with 2% BSA (30–45 min, RT), followed by the incubation of specimens in the presence of primary antibodies: mouse anti-vinculin IgG (NV9131, Sigma, 1:300), mouse anti-α-tubulin IgG (Sigma, No. T9026, 1:500) rabbit anti-Cx43 IgG (Sigma; No. C6219; 1:500), and goat anti-Snail-1 N-terminal IgG (Sigma; No. SAB2501370; mouse anti-O6-Methylguanine-DNA Methyltransferase (Sigma). Subsequently, the cells were labeled with AlexaFluor488-conjugated donkey anti-mouse IgG (No. A21202, Invitrogen, Carlsbad, CA, USA) or AlexaFluor647-conjugated chicken anti-rabbit IgG (Invitrogen; No. A21443), AlexaFluor546-conjugated donkey anti-mouse IgG (Invitrogen; No. A10036), and AlexaFluor488-conjugated chicken anti-goat IgG (Invitrogen; No. A21467). In places, the cells were counterstained with Alexa546-conjugated phalloidin (Invitrogen; No. A22283) and Hoechst 33258 (No. B2883, Sigma) [62]. The images were acquired with Leica DMI6000B fluorescence microscope equipped with a DFC360FX CCD camera, total internal reflection fluorescence (TIRF) module (DMI7000 version; Leica Microsystems, Wetzlar, Germany), Nomarski interference contrast (DIC), and IMC modules using ×40, NA-1.47 or ×100 oil immersion objectives, and the Leica Application Suite Advanced Fluorescence software [37]. The raw images were additionally processed (contrast adjustment, background substraction, fluorimetric analysis) with ImageJ software. When indicated, cells were suspended with 0.2% EDTA in Ca/Mg-free PBS, fixed, permeabilized, and stained in suspension, as described above, and flow cytometric detection of Snail-1/Cx43 cells was performed with ImageStreamX imaging cytometer (Merck). For each condition, at least 5000 singlets were collected. The data were processed using IDEAS 6.2 software.

T98G/U87 cells were were seeded at the density of 5000 cells/cm² in a 12-well plate in antibiotic-free DMEM medium that was supplemented with 10% FBS and grown at 37 °C for 24 h. Subsequently, the medium was replaced with Opti-MEM Reduced Serum Medium (31985070, Gibco-Life Technologies, Waltham, MA, USA) and cells were treated with complexes of 1.5 μL Lipofectamine™ 2000 Reagent (11668-019, Invitrogen) and 1 μg plasmid DNA/small interfering RNA (siRNA) in Opti-MEM, according to the manufacturer’s protocol. Afterwards, the medium was replaced by standard DMEM medium with 10% fetal bovine serum and 1% Antibiotic-Antimycotic Solution and the efficiency of Cx43/Snail-1 up/down-regulation was analyzed with immunofluorescence-assisted cytofluorimetry. The following siRNA/plasmids were used: SNAI-1 siRNA and siRNA-A (sc-38398 and sc-37007, Santa Cruz Biotechnology, Dallas, TX, USA), Snail_pGL2 (#31694, Addgene, Watertown, MA, USA), Cx43_FLAG (#17664, Addgene), MISSIONR esiRNA GJA1 (Sigma-Aldrich, St. Louis, MO, USA), and Cx43 shRNA (sc-29276-SH, Santa Cruz Biotechnology). The cells were subjected to endpoint experiments within the next 96 h. Cx43/Snail-1 expression was monitored with flow cytometry (see Section 2.4).

For the analyses of intracellular Snail-1/Cx43/MGMT levels, the cells were incubated in the absence/presence of TMZ, as indicated in the text, harvested with cold (~4 °C) Ca²⁺/Mg²⁺—free PBS/EDTA solution, centrifuged, and dissolved in a Triton X-100 cell lysis buffer with protease inhibitor cocktail (followed by freeze-thaw procedure and sonication). Bradford assay was applied for the estimation of total protein content in the lysates. The samples (20 µg of protein) were separated on 12% polyacrylamide gel (SDS-PAGE electrophoresis; Laemmli protocol; [37]), followed by the electrotransfer of proteins into PVDF membranes (Immun-Blot^® PVDF Membrane, #1620177; Bio-Rad, Hercules, CA, USA). Membranes were then blocked with skimmed milk/TBST solution and then incubated in the primary antibodies: monoclonal goat anti-Snail-1 IgG (Sigma; SAB2501370; 1:500), polyclonal rabbit anti-Cx43 (Sigma; No. C6219; 1:3000) or monoclonal mouse anti-MGMT IgG (Sigma; MAB16200; 1:500). α-tubulin was labeled with monoclonal IgG mouse anti-α-tubulin antibody (Sigma; No. T9026; 1:1000) and it used as a reference protein. Signal detection was performed with HRP-conjugated goat anti-rabbit IgG (No. 31466, Thermo Fisher Scientific, Waltham, MA, USA) and goat anti-mouse IgG (Thermo Fisher Scientific; No. 31430) using the chemiluminescence technique (Merck, Luminata Crescendo; No. WBLUR0500). For membrane imaging, MicroChemii system (DNR Bio-Imaging System) was used.

Statistical significance of differences was tested with non-parametric Mann–Whitney U-test. Analyses were performed using Origin 2020 software.

The data that supports the findings of this study are available in theof this article or available from the corresponding author upon reasonable request. supplementary material