Alterations in Cell Motility, Proliferation, and Metabolism in Novel Models of Acquired Temozolomide Resistant Glioblastoma

To better recapitulate acute temozolomide (TMZ) resistance models, we created cell lines that were resistant to 200 µM TMZ and were challenged in 100 µM TMZ for all experiments. The overall time to acquired resistance varied from approximately 2 months for the 8MBGA cell line to ~3 months for the 42MBGA cell line. Cells were defined as resistant when each no longer showed a G2/M arrest or apoptosis in response to TMZ treatment (Fig. 1a–d). While both parental cell lines (WT) exhibited a significant increase in the SubG1, or apoptotic, fraction following TMZ treatment, this was lost in the TMZres variants (Fig. 1e). However, the shape of the SubG1 curves differed between the two cell lines – 8MGBA-WT cells treated with TMZ showed a defined SubG1 peak typically associated with apoptosis, while 42MGBA-WT cells did not (Supp Fig. 1a,b). This was corroborated by immunoblotting for cleaved poly (ADP-ribose) polymerase (PARP), which was observed in 8MGBA-WT but not 42MGBA-WT cells (Fig. 1g,h). Expression of the O6-methylguanine methyltransferase (MGMT) was detectable in both resistant cell lines, but absent in both parental lines (Fig. 1f).

Acquired resistance to the previous standard of care for GBM, bis-chloroethylnitrosourea (BCNU), leads to an increased sensitivity to TMZ¹⁴. Because BCNU can still be a component of second-line treatment for GBM^15–18, we tested whether our TMZres cells exhibited altered responsiveness to BCNU. Since the resistant cell lines are populations selected from the wild type cell lines they have a more reproducible phenotype as the variability or heterogeneity of the cell line has decreased. Therefore, the TMZres cells still showed a statistically significant G2/M arrest upon BCNU treatment. However the biological effect is dramatically reduced, with a 74% increase in G2/M vs 8% increase in G2/M for the 42MBGA-WT vs. 42MBGA-TMZres cells compared to controls, respectively (Fig. 1a,b). This is similar to 8MGBA-WT and 8MGBA-TMZres cells, where the increase in G2/M fraction is 86% vs. 7% compared to controls, respectively (Fig. 1c,d).

The mechanism of action for TMZ is to induce DNA damage and give rise to cell cycle arrest in the G2/M phase⁵. As cells have already copied their DNA at this point in the cell cycle, if a cell is to survive TMZ-treatment they may also be retaining the extra copies of chromosomes that have already been duplicated. The 42MGBA-WT cell line has a hypertetraploid karyotype (88–95 chromosomes) with 8% polyploidy, while the 8MGBA-WT cell line has a hyperdiploid karyotype with 15% polyploidy and contains 47–52 chromosomes (DSMZ, https://www.dsmz.de/↗). 42MGBA-TMZres cells showed a significant increase in total nuclear area (Fig. 2a), and their nuclear morphology became multi-lobed (Fig. 2b). Total nuclear area was also significantly increased in 8MGBA-TMZres cells, though their nuclei retained an oblong or circular morphology (Fig. 2a,b). Given the strikingly different nuclear morphology of the two resistant variants, we cultured both TMZres lines in the absence of TMZ for three weeks to test the stability of the resistant phenotype. After re-challenging with TMZ treatment for 72 hours, we observed no significant change in cell cycle profile or the SubG1 peak in the 42MBGA-TMZres cell line, while the 8MBGA-TMZres line underwent a slight G2/M arrest and showed a unique SubG1 peak in both untreated and treated conditions suggesting a less stable phenotype (Supp Fig. 2a–d).

To determine whether the increase in the total nuclear area in TMZ-resistant cells was associated with chromosomal gain, we assessed the overall chromosomal number in metaphase spreads from the TMZres cell lines compared to their respective parental lines (WT). We observed an increase in overall chromosome number in the majority of the 42MBGA-TMZres cells in contrast to the 8MBGA-TMZres cells where only a small subpopulation of cells had an increase in overall chromosome number (Fig. 3b). We also assessed the change in copy number of a representative chromosome in each cell line pair (chromosome 17 in the 8MGBA pair and X in the 42MGBA pair), in interphase nuclei and metaphase spreads (200/cell line) using fluorescent in situ hybridization (FISH) (Fig. 3a). The choice of the two representative chromosomes was made based on reported karyotype analysis of the 2 parental cell lines showing a mostly diploid count for chromosome 17 in the 8MGBA line and X in 42MGBA (DSMZ, https://www.dsmz.de/↗). We observed that 96% of the 42MBGA-TMZres cells had three or more copies of the X chromosome compared to only 7% of the 42MBGA-WT cells (93% of those cells had 2 copies). In contrast, this dramatic shift was not observed in 8MBGA-TMZres cells, where only a small subpopulation of cells showed an increase in the number of chromosomes 17 (18% had 3 or more copies) compared to the parental cells (6% had 3 or more copies). Taken together, these findings tracked with the stability of TMZ-resistance, with the 42MBGA-TMZres cells showing a more stable phenotype compared to 8MBGA-TMZres cells (Fig. 3c).

We then determined how TMZ-resistance affected cell size and proliferative vs migratory phenotypes. 42MBGA-TMZres cell size was not changed vs 42MGBA-WT, though their basal growth rate was dramatically increased (Fig. 4c, Sup Fig. 3a,b). They also showed a modest but nonsignificant reduction in cell migration (Fig. 4a, images in Sup Fig. 4). In contrast, 8MBGA-TMZres cell size was significantly increased when compared to its parental cell line, while the basal growth rate was unchanged (Fig. 4d, Sup Fig. 3a,b). 8MGBA-TMZres cells were significantly more migratory than 8MGBA-WT cells (Fig. 4b). Enhanced cell migration correlated with increased F-actin stress fiber thickness in both TMZres models. There was no significant change in F-actin thickness in the 42MBGA-TMZres compared to 42MGBA-WT cells, while it was significantly increased in the more migratory 8MGBA-TMZres when compared to 8MBGA-WT cells (Fig. 4e,f).

An advantage of selecting these resistant models was the ability to observe changes that occurred during and continued to persist after resistance was acquired. One observation was the number of intracellular puncta and extracellular vesicles that were expelled into the media as the cells became resistant to TMZ (Fig. 5c). As there was no longer a dramatic increase in the SubG1 fraction (Fig. 1e, Sup Fig. 1), or cleaved PARP (Fig. 1h) upon resistance, we were able to rule out general apoptotic bodies. We therefore wanted to test if there was an increase in overall vesicle production in the TMZres vs WT cell lines. To accomplish this, we stained for the endosomal marker early endosome antigen 1 (EEA1). After analysis, we saw that there was a significant increase in the total number of endosomes in both TMZres cell lines compared to their parental lines (Fig. 5a–c). Interestingly in two independent datasets – the Chinese Cancer Genomics Consortium (CCGC) and REpository for Molecular BRAin Neoplasia DaTa (REMBRANDT) – EEA1 expression increases from normal brain to GBM, and survival is significantly better for patients with tumors having lower expression of EEA1 (Fig. 5d–f). However, it was not only intracellular, or early endosome, vesicles that we observed to be different, but extracellular vesicles as well. We therefore performed untargeted liquid chromatography and gas chromatography time-of-flight-based mass spectrometry (LC-MS, GC-TOF-MS) on cell culture media from the 8MGBA-WT and 8MBGA-TMZres cell line to identify differences in metabolites that may be associated with this phenotype. Putative results from the LC-MS showed an increase in the extracellular lipids LysoPC(16:0) and LysoPE(16:0) (Sup Fig. 5). Interestingly, indoleacetaldehyde, a product of tryptophan metabolism, was also shown to be putatively increased in the 8MGBA-TMZres media. Tryptophan metabolism disruption has been shown previously as a feature of treatment adaptation in GBM cells¹⁹. Consistent with our observation of increased extracellular vesicles, there was an increase in palmitic acid and stearic acid lipids in the GC-TOF-MS metabolites from the 8MBGA-TMZres cell line conditioned media (Fig. 6a–c). The top hit from GC-TOF-MS was the non-aminogenic amino acid aminomalonate, an inhibitor of aminolevulinic acid synthase, the enzyme that catalyzes the rate limiting step of the protoporphyrin IX (PpI IX)/heme synthesis pathway²⁰. This pathway is deregulated in many cancers, including GBM, and can be exploited to identify tumor area during surgery^21,22. The last significant metabolite was citric acid, which has been associated with a decrease in overall survival in GBM patients with higher citric acid in their cerebral spinal fluid²³.

The enrichment of lipid metabolites and aminomalonate identified in conditioned media from resistant cells by LC-MS and GC-TOF-MS, respectively, could suggest dysregulation of mitochondrial metabolism. We therefore determined whether the expression of electron transport chain (ETC) subunits is altered between sensitive and resistant lines. There were no changes between ETC subunits in the 42MBGA-WT and -TMZres lines. However, there was an increase in three ETC subunits in the 8MBGA-TMZres cell line as compared to its WT control: MT-CO1 (Complex IV), SDHB (Complex II), and NDUFB8 (Complex I) (Fig. 6d). Increased expression of Complex IV and Complex II subunits is consistent with prior studies²⁴, while in other malignancies, inactivating mutations in SDHB (Complex II) are implicated in TMZ responsiveness^25,26.

42MBGA and 8MGBA cell lines were provided by Dr. Jeffrey Toretsky (Lombardi Comprehensive Cancer Center (LCCC), Georgetown University, Washington DC). 42MBGA-TMZres and 8MGBA-TMZres were developed by our lab with constant exposure of the parental cell lines to increasing concentrations of TMZ, from 12.5, 25, 50, 100, 200 µM, over the course of 2–3 months. All cells tested negative for Mycoplasma contamination, and were maintained in a humidified incubator with 95% air: 5% carbon dioxide. All cell lines were fingerprinted by the LCCC Tissue Culture Shared Resource to verify their authenticity using the standard 9 STR loci and Y-specific amelogenin. Both the 42MBGA-TMZres and 8MGBA-TMZres fingerprinted the same as their parental cell line. 42MGBA and 8MBGA cells were grown in DMEM with 10% FBS. 42MBGA-TMZres and 8MBGA-TMZres cells were grown in DMEM with 10% FBS and 100 µM TMZ. TMZ (Selleckchem, Catalog No. S1237) was dissolved in DMSO to 130 mM and used at concentrations indicated. Bis-chloroethylnitrosourea (BCNU) was a kind gift from Dr. Esther Chang, and was dissolved in ethanol to 100 mM before use at concentrations indicated.

On day 0, cells were seeded at 100,000–150,000 cells per well in 6-well plastic tissue culture dishes one day prior to treatment with the indicated concentrations of drug. For experiments with TMZ or BCNU cells were treated for 72 hours. After 72 hour treatments, cells were collected, ethanol-fixed, stained with propidium iodide, and analyzed for cell subG1 (fragmented) DNA content and cell cycle profile by fluorescence activated cell sorting.

Cells were lysed in RIPA buffer for protein extractions and separated by a 4–12% gradient gel (Novex by Life Tech, NP0321BOX). They were then transferred onto Nitrocellulose membranes (Invitrogen, IB23001) with the iBlot 2 and probed with antibodies against MGMT (1:1000, Cell Signaling, 2739S), Beta-Tubulin (1:10,000,Sigma Aldrich, T7816), PARP (1:1000, Cell Signaling, 9542L), Actin (1:5000, Sigma Aldrich, A5316), and Electron Transport Chain antibody cocktail (1:500, Abcam, ab110413). Proteins were detected with horseradish peroxidase-conjugated secondary antibodies (1:5000, GE Healthcare Life Sciences) and enhanced chemiluminescent detection HyGLO Quick Spray Chemiluminescent (Denville) for dark room development.

Metaphase and interphase slide preparation was done using a standard protocol⁴⁴. For metaphase chromosomal copy number determination, chromosomes were stained with 4′,6-diamidino-2-phenylindole (DAPI). FISH analysis was performed using a standard protocol^45–47. A 17q21.1-21.2 locus-specific FISH probe (Stat5A/B locus) was used to evaluate chromosome 17 copy number change in the 8MGBA-TMZres and WT cell lines, as described earlier⁴⁵. For the X chromosome enumeration in the 42MGBA-TMZres and WT cell lines, an X chromosome centromeric probe obtained from Empire Genomics (Buffalo, NY) was used according a manufacturer’s protocol. Scoring of cells and digital image acquisition were performed using a Leica DMRBE microscope (Leica Microsystems, GmbH, Wetzlar, Germany) equipped with the appropriate optical filters and a CCD camera⁴⁵.

Cells were plated at 150,000–200,000 cells/well and allowed 48 hours to create a monolayer. After monolayer formation, a P200 tip was used to make the scratch. Images were taken at 0 hr, 24 hr, and 48 hr time points. Analysis was done in ImageJ⁴⁸ to determine percent closed with 0% being at 0 hr.

FIJI (ImageJ⁴⁸) was used to measure the thickness of F-actin fibers. Grayscale images of phalloidin-stained cells were converted to binary using the Threshold tool (black/white). The BoneJ plugin⁴⁹ was then used to measure fiber thickness in pixels. The mean thickness per cell is reported for 28-29 individual cells per cell line.

8MGBA-WT and 8MGBA-TMZres cells were seeded at a density of 50,000 cells onto 18 mm coverslips in 12-well dishes. On the following day, the media was removed and cells were fixed and permeabilized in 3.2% paraformaldehyde (PFA) with 0.2% Triton X-100 in PBS for 5 minutes at room temperature. Three washes were performed with PBS in the 12-well plate, then coverslips were inverted onto 100 μl of primary antibody in the antibody block (0.1% gelatin with 10% normal donkey serum in PBS) on strips of parafilm and incubated for one hour. Coverslips were first stained with EEA1 (Cell Signaling, 3288), an early endosomal marker (1:100 dilution, 1 hr). Phalloidin (polymerized F-actin, Invitrogen, A12379, 1:200 dilution), and DAPI (DNA, 1:500 dilution) stains were added to the secondary antibody mixtures. Each coverslip was washed three times with PBS after primary incubation and then were inverted onto 100 µL of the appropriate secondary antibody, DAPI dihydrochloride, and (where appropriate) ActiStain-488-phalloidin (Cytoskeleton, Denver, CO) in antibody block in the dark for 20 minutes. Coverslips were again washed 3x with PBS, then gently dipped three times into molecular biology-grade water before inversion onto one drop of FLUOROGEL (Electron Microscopy Sciences, Hatfield, PA) then allowed to air-dry in the dark for at least 10 minutes. Slides were stored at 4 °C until image collection on the LCCC Microscopy & Imaging Shared Resource’s Leica SP8 microscope with the 63x oil objective.

Cells were seeded at a density of 1,000 cells per well in 3, 96-well plastic tissue culture dishes per cell line on day 0. On day 1, one plate was stained with crystal violet (Sigma,C0775). For staining, plates were rinsed 1 time with 1X PBS to remove excess cellular debris. After, 100 µL of 3.2% PFA was added to each well and incubated at room temperature for 5 minutes followed by three washes in 1X PBS. Then, 200 µL of 0.5% crystal violet in 25% methanol was added to each well and incubated at 4 C for 10 min. The stain was then removed and the plate was rinsed 4–6 × with diH2O to remove excess stain. The plates were left to air-dry overnight. On day 8 and 10, all plates were rehydrated with 100% methanol and read at an absorbance of 550 nm.

Cell size was determined by the Countess 2 (Invitrogen) using trypan blue dye exclusion of dead cells. The Countess 2 was also used to determine number of cells for 72-hour growth assays. 100,000 cells were seeded per 6-well on day 0. 24, 48, and 72 hours post-seeding, cells were collected and total cell number was determined by trypan blue dye exclusion.

Results are represented as mean values ± standard error (SD) and considered statistically significant where p-value < 0.05. Three independent biological replicates were done for each experiment, with the exception of metabolomics where six independent replicates were performed. Statistical significance has been calculated using GraphPad Prism version 7.00 for Mac, GraphPad Software, La Jolla California USA (www.graphpad.com) for the Student’s t-test, Mann Whitney U test, one-way ANOVA with Dunnett’s correction, or χ2 analysis of 2 × 3 contingency table. The RStudio packages survival, GGally, ggplot2, readr, and magrittr were used for violin plots and big data sets (CCGA, REMBRANDT)⁵¹.

The authors declare no competing interests.