Brain Tumor Genetic Modification Yields Increased Resistance to Paclitaxel in Physical Confinement

Brain tumor cells remain highly resistant to radiation and chemotherapy, particularly malignant and secondary cancers. In this study, we utilized microchannel devices to examine the effect of a confined environment on the viability and drug resistance of the following brain cancer cell lines: primary cancers (glioblastoma multiforme and neuroblastoma), human brain cancer cell lines (D54 and D54-EGFRvIII), and genetically modified mouse astrocytes (wild type, p53−/−, p53−/− PTEN−/−, p53−/− Braf, and p53−/− PTEN−/− Braf). We found that loss of PTEN combined with Braf activation resulted in higher viability in narrow microchannels. In addition, Braf conferred increased resistance to the microtubule-stabilizing drug Taxol in narrow confinement. Similarly, survival of D54-EGFRvIII cells was unaffected following treatment with Taxol, whereas the viability of D54 cells was reduced by 75% under these conditions. Taken together, our data suggests key targets for anticancer drugs based on cellular genotypes and their specific survival phenotypes during confined migration.


Drug availability
To predict the spatiotemporal Taxol distribution within a single 5_5 microchannel after the initial introduction of 100 nM Taxol to both the central and satellite reservoirs, we performed numerical simulations (COMSOL 4.4) for the following two cases: 1) media in the microchannel was free from Taxol and 2) two cells were conservatively positioned, each flushed against either end of the microchannel. In Case 2, cells were assumed to deform to 50 µm axially and fully obstruct the microchannel. The cell dimensions were consistent with the experimental observations of cancer cells under similar conditions of confinement. Taxol transport within the fluid-filled microchannels was taken to be a passive diffusive process governed by Fick's 2 nd law (Eq. S1), where C refers to the Taxol concentration, t is the time elapsed since drug introduction, and D is the diffusivity of Taxol in the media or in the blocking cells. For Case 1 simulation, a value of 430 m 2 /sec 53 was used for the media in the microchannel that is free from Taxol with D = D H2O . For Case 2 simulations, D cyto was used in place of D for the blocking cells, whereas D H2O was used for the center part of the microchannel between two clocking cells: The microchannel side walls were assumed to be impermeable to drug molecules. Case 2 simulation was used to help assess how cells positioned at the two ends of a microchannel could impede and affect the Taxol diffusion from the reservoirs into the media in the center of the microchannel. The overall resistance from a sitting cell to Taxol diffusion consists of that from the plasma membrane, cytoplasm, nucleus envelope and nuclear contents. No data exist that accounts for the resistance from each of these components and how they are connected to impede the Taxol diffusion in the cell. For this assessment, we chose to perform a parametric study by assuming the Taxol diffusivity D cyto in a sitting tumor cell to be either 10%, 5% or 1% that of Taxol diffusivity in water D H2O .

Doxorubicin dose response
We examined the dose response of cells cultured on a 2D surface using different Dox

Effect of low concentration Dox on the viability of GBM in physical confinements
GBM cells cultured in the devices were treated with 100 nM and 500 nM Dox for 48 hours. The viabilities of cells in different physical confinements (5_5, 15_15, and 2D) were quantified.

Virtual confirmation of the availability of Taxol to confined cells within the microfluidic device
Simulation results indicated that Taxol diffusion in the unblocked microchannels resulted in rapid saturation. Volumetric concentration contours (Fig. S1A) illustrated noticeable concentration gradients after one minute as indicated by the color variations. After five minutes, the spatial concentration gradients had disappeared and the concentration at the center of the microchannel had increased to nearly 90 nM. The contours illustrated that the concentration changes were dominant in the axial direction with little transverse variation. Figure S1B showed the minute-by-minute axial concentration profiles over the first ten minutes of simulation.
Clearly, concentration gradients disappeared in the microchannel within 10 minutes of drug introduction.
The effect of confined cells at the microchannel ends is illustrated in Figures S1C-S1E for cases where D cyto = 10%, 5%, or 1% that of D H2O . The existence of steep concentration gradients in the 50 m regions near both junctions (X = 0 m and X = 530 m) reaffirmed that resistances were much higher in the cells than in the microchannels. Diffusion was limited through the cells but drug concentrations equilibrated quickly within the microchannel interior as indicated by the smaller concentration differentials there. With assumed lower Taxol diffusivity, 10%, 5% and 1% that of value in water, the concentration gradients in the cells became increasingly steep and the drug concentration at the microchannel center continued to decrease for a given time. Target concentrations of drug (100 nM) were easily achieved through the center part of the entire microchannel within 3 hours if the Taxol diffusivity in the cell becomes 1%.
Overall, the steady state, uniform concentrations of nearly 100 nM were achieved in a couple of hours. The result indicated that even the coefficient efficiency decreased by 95%, and the desired Taxol concentration should be ready at any microchannels within one hour. Regardless of the higher number of cells inside the microchannel as well as the uptake process of Taxol by cells, the simulation results virtually confirmed that experimental Taxol doses were achieved at cell boundaries after short time windows in the microfluidic device and the experimental results should reflect the appropriate cell exposure.

Doxorubicin dose response
The dose response curves of D54, D54-EGFRvIII and GBM cells (Fig. S2A, B, and C) showed a plateau when Dox concentrations were below 10 nM. Linear responses occurred as Dox concentrations increased up to 1 µM. The estimated IC50 values were 120, 250, and 600 nM, respectively. In all cell lines, 1 µM Dox resulted in approximately less than 20% viability. As the resulting viability of cells in 2D was low, using Dox 1 µM enabled the investigation of any increasing viability in a confined environment.

Effect of low concentration Dox on the viability of GBM in physical confinement
Reduction of Dox concentrations from 1 µM to 100 nM and 500 nM resulted in overall higher viabilities ( Fig. 6 and Fig. S2D). However, for both 100 nM and 500 nM, statistical analysis showed no significant difference of viabilities among three physical confinements, which was similar to the 1 µM case. These results reinforced our finding that Dox affected cancer cells regardless of their confinements, suggesting the need for the further investigation of effective therapeutic treatment on migrating cancer cells.