A Benzodiazepine-Derived Molecule That Interferes with the Bio-Mechanical Properties of Glioblastoma-Astrocytoma Cells Altering Their Proliferation and Migration

Abstract

Glioblastoma multiforme (grade IV glioma) is characterized by a high invasive potential, making surgical intervention extremely challenging and patient survival very limited. Current pharmacological approaches show, at best, slight improvements in the therapy against this type of tumor. Microtubules are often the target of antitumoral drugs, and specific drugs affecting their dynamics by acting on microtubule-associated proteins (MAPs) without producing their depolymerization could affect both glioma cell migration/invasion and cell proliferation. Here, we analyzed on a cellular model of glioblastoma multiforme, the effect of a molecule (1-(4-amino-3,5-dimethylphenyl)-3,5-dihydro-7,8-ethylenedioxy-4h2,3-benzodiazepin-4-one, hereafter named 1g) which was shown to act as a cytostatic drug in other cell types by affecting microtubule dynamics. We found that the molecule acts also as a migration suppressor by inducing a loss of cell polarity. We characterized the mechanics of U87MG cell aggregates exposed to 1g by different biophysical techniques. We considered both 3D aggregates and 2D cell cultures, testing substrates of different stiffness. We established that this molecule produces a decrease of cell spheroid contractility and it impairs 3D cell invasion. At the same time, in the case of isolated cells, 1g selectively produces an almost instantaneous loss of cell polarity blocking migration and it also produces a disorganization of the mitotic spindle when cells reach mitosis, leading to frequent mitotic slippage events followed by cell death. We can state that the studied molecule produces similar effects to other molecules that are known to affect the dynamics of microtubules, but probably indirectly via microtubule-associated proteins (MAPs) and following different biochemical pathways. Consistently, we report evidence that, regarding its effect on cell morphology, this molecule shows a specificity for some cell types such as glioma cells. Interestingly, being a molecule derived from a benzodiazepine, the 1g chemical structure could allow this molecule to easily cross the blood-brain barrier. Thanks to its chemical/physical properties, the studied molecule could be a promising new drug for the specific treatment of GBM.

Keywords: GBM; anticancer drug; biomechanics; invasion; microtubule; spheroid.

Figure 1

(a) A normalized equatorial area expansion of a control (DMSO) U87MG spheroid (black curve) and a spheroid of the same cells treated with 20 μM 1g (red curve). The spheroids were embedded inside a Matrigel^® matrix as described in Section 4: Materials and Methods; (b) images of a control and of a 1g-treated spheroid at different time points. The line highlighting the spheroid expansion has been obtained using the Analyze_Spheroid_Cell_Invasion_In_3D_Matrix tool from FIJI (bar = 200 μm for all six images). (c) A comparison of the cell shape between the control sample and the U87MG spheroid exposed to 20 μM 1g. The frames are representative of the spheroids at the beginning of the experiment and after 68 h. (bar = 50 μm for all four images).

Figure 2

The traction force microscopy of U87MG spheroids. (a) Frames at different time steps of a control U87MG spheroid and of a U87MG spheroid in 15 μM 1g. The spheroids were embedded in a Matrigel^® matrix (10 mg/mL protein content, Young’s modulus of about 400 Pa) with 5 μm diameter latex beads (bar = 200 μm for all six images); (b) a representation of the traction force at the end of the experiment for a control and 1g-treated spheroids. The arrows show the direction and intensity of the bead displacement; (c) a plot of the mean contractility as a function of time for spheroids in the different conditions (two experiments for each of the conditions are reported).

Figure 3

The micropipette aspiration of U87MG spheroids. (a) The sequence of three images during the aspiration process of an U87MG spheroid exposed to 20 μM 1g. L represents the length of the tongue to be used for the fitting procedure (bar = 50 μm for all images); (b) the representative aspiration and relaxation curves for a control spheroid (black squares) and a 1g-treated spheroid (red circles). In the plot, the regions exploited to obtain v_in and v_out (see text) are highlighted. The continuous lines represent the fit of Equation (1) to the data; (c) the rheological model exploited to obtain Equation (1). According to this model, in parallel with a Voigt element (a spring characterized by a Young modulus k₂ in series with a damper η₂ representing a dissipation coefficient related to the initial instantaneous spheroid deformation) there is a spring with Young modulus k₁ to account for the instantaneous elastic deformation (the region is highlighted in the plot) and the overall model is in series with another dashpot η₁ (viscous dissipation of the spheroid tongue) to account for the long-term flow of the cell aggregate.

Figure 4

The effect of 1g on the morphology and adhesion of U87MG cells. (a) Examples of U87MG cells before the injection of 20 μM 1g into the culture medium and their corresponding morphology 1 h after the injection (bar = 20 μm in each column of images); (b) a statistical analysis of the elliptical shape defined as the ratio of the two axes resulting from an elliptical fit of the cell morphology; (c) a cell adhesion assay for the U87MG cell line. Cells, pre-treated or not with 1g 20 μM, were grown on BME and incubated with EMEM without FBS alone (Ctrl and pre-treated with 1g 20 μM) or with 5, 20, or 50 μM of 1g and expressed as rates of U87MG cell adhesion to culture plates (abs = 540 nm). **** p < 0.0001, *** p < 0.001, ** p < 0.01 and * p < 0.05 vs. respective; one way ANOVA and Tukey’s as post test.

Figure 5

The effect of 1g on cell duplication. (a) A plot of the duplication time (measured as the time cells remain rounded in mitosis) as a function of the 1g concentration. The red line represents a linear fit to the data. (b) An example of cells exiting mitosis and producing more than two daughter cells (bar = 20 μm for both images); (c) examples of the presence of more than two asters in the organization of microtubules in mitosis in the presence of 20 μM 1g. The actin cytoskeleton has been marked in red and the microtubules in green (bar = 5 μm); (d) a time-lapse imaging analysis of U87MG cells exposed to 20 μM 1g in mitosis. For 26 cells observed by the time lapse-imaging, the time spent in mitosis has been recorded. The large majority (78%) of the cells exposed to 20 μM 1g are able to perform mitotic slippage.

Figure 6

Wound healing assay for control cells and cells exposed to 1g. (a) An optical microscopy image of a 2D wound healing assay immediately after the formation of the scratch and after 24 h since the formation of the wound both in the absence and in the presence of 20 M 1g (bar = 200 μm in all four images); (b) the time evolution of the still-present wound area; and (c) the percentage of the wound repair closure after 24 h in the absence of 1g and in the presence of 20 μM 1g (p = 0.0049 (**), unpaired t test).

Figure 7

An analysis of the MSD of U87MG cells on substrates of different Young moduli and exposed to increasing concentrations of 1g. For the migration analysis, we considered cells that did not interact by physical contact with other cells and we stopped tracking the cells once they reached the typical globular configuration of the mitotic stage. The MSD vs. time is reported in a Log–Log plot with two lines representing the ballistic and the pure Brownian motion cases. In the inset of each plot, the MSD value at the end of the corresponding plot is reported as a function of the substrate stiffness. The plots are limited to the time intervals for which the standard deviation of the data does not become too large (data for the longest time intervals are averaged over a smaller set of measurements). The presence of a maximum value of the MSD for an intermediate value of the substrate stiffness for the control case is evident. By increasing the 1g concentration, the sensitivity of the MSD value to the substrate stiffness strongly decreases.

Figure 8

Single-cell mechanical properties measured by AFM: (a) an AFM image of control U87MG cells; (b) an AFM image of a U87MG cell exposed to 1g 20 μM for 24 h; (c) an AFM image of a U87MG cell exposed to nocodazole 20 μM for 24 h. (bar = 20 μm for all AFM images); (d) distributions of the values obtained for the Young modulus of U87MG cells before (black squares) and after 24 h in 20 μM 1g (blue triangles). The points in the plot are derived from the distribution histogram of the value. The Young modulus was obtained by fitting the force curves with a modified Hertz model; (e) elastic (G′, red squares) and viscous (G″, black squares) components of the shear modulus G* as a function of the frequency of the sinusoidal signal applied to the cantilever base by the piezoactuator in control U87MG cells; and (f) a comparison of the modulus of the complex shear modulus G*, as a function of frequency, of U87MG cells for the untreated, exposed to 20 μM 1g for 24 h and exposed to 20 μM nocodazole for 24 h. Data represent the average value and the standard deviation of n = 7 cells for each condition. The slope of the linear fit to the control cell data is 0.33.

Figure 9

(a) The expression of ERM and pERM proteins in U87MG cells exposed to 20 μM 1g. The analysis shows the strong increase of the pERM/ERM ratio; (b) the immunofluorescence of typical U87MG cells 30 min after the injection of 20 μM 1g: pERM (red), microtubules (green), and DNA (blue) (bar = 20 μm). **** p < 0.0001 and * p < 0.05 vs. Control; one way ANOVA and Tukey’s as post test.

Figure 10

The effect of 1g on proteins involved in cell adhesion and motility. (a) Representative western blots of a-actinin, paxilin, vinculin, FAK, Rac-1, cofilin, P-cofilin, and LIMK at different time points; (b) a densitometric evaluation of protein levels in U87MG cell lysate after incubation with 20 μM of 1g for 5 or 12 h. Densitometry values were normalized to the protein loading control, beta-actin. The values are expressed as the mean ± SD of three independent experiments (n = 3 per group). *** p < 0.001, ** p < 0.01 and * p < 0.05 vs. untreated cells (Ctrl), using one-way ANOVA with Dunnett’s as a post test.