Electrotaxis of Glioblastoma and Medulloblastoma Spheroidal Aggregates

Abstract

Treatment of neuroepithelial cancers remains a daunting clinical challenge, particularly due to an inability to address rampant invasion deep into eloquent regions of the brain. Given the lack of access, and the dispersed nature of brain tumor cells, we explore the possibility of electric fields inducing directed tumor cell migration. In this study we investigate the properties of populations of brain cancer undergoing electrotaxis, a phenomenon whereby cells are directed to migrate under control of an electrical field. We investigate two cell lines for glioblastoma and medulloblastoma (U87mg & DAOY, respectively), plated as spheroidal aggregates in Matrigel-filled electrotaxis channels, and report opposing electrotactic responses. To further understand electrotactic migration of tumor cells, we performed RNA-sequencing for pathway discovery to identify signaling that is differentially affected by the exposure of direct-current electrical fields. Further, using selective pharmacological inhibition assays, focused on the PI3K/mTOR/AKT signaling axis, we validate whether there is a causal relationship to electrotaxis and these mechanisms of action. We find that U87 mg electrotaxis is abolished under pharmacological inhibition of PI3Kγ, mTOR, AKT and ErbB2 signaling, whereas DAOY cell electrotaxis was not attenuated by these or other pathways evaluated.

Figure 1

Design of our electrotaxis system. (a) Schematic of electrotaxis channel construction. (b) Computation validation using COMSOL with I_set = 330 μA (c,d) Biological validation using MatLyLu single cell tracking. Cells were tracked over 2 h for both no electrical field (CTL, n = 20) and under dcEF equivalent to 100 V/m (n = 22) for (c) displacement along the electrical field lines toward the cathode; and (d) displacement along the axis orthogonal to the electrical field. ****p < 0.0001 by Student’s t-test, two-tailed.

Figure 2

U87 mg spheroidal aggregates exhibit cathodal electrotaxis. (a) GFP + U87 mg control (no dcEF) aggregates showing no bias in growth direction after 8 h. (b) GFP + U87 mg aggregates exposed to 250 V/m dcEF exhibiting cathodal bias, detectable within 8 h. (a,b) are pseudo-colored for time = 0 (pink), time = 8 h (green). (c) Quantitative assessment of the outward shift of aggregate frontiers after 8 h for no electrical field (CTL, n = 19), 100 V/m (n = 19), and 250 V/m (n = 22). For cathodal frontier: *p = 0.0142; for cathodal bias: *p = 0.0448, ****p < 0.0001; by Two-way ANOVA and Holm-Sidak post-hoc test. Mean ± SEM shown. (d) Quantitative assessment of the outward shift of aggregate frontiers after 24 h for U87 mg with no dcEF (CTL, n = 39) or 250 V/m (STIM, n = 56). **p = 0.0032; ***p = 0.002; ****p < 0.0001; by Two-way ANOVA and Holm-Sidak post-hoc test. Scale: 150 μm.

Figure 3

DAOY spheroidal aggregates exhibit anodal electrotaxis. (a) Quantitative assessment of the outward shift of aggregate frontiers after 8 h for no electrical field (CTL, n = 10) and 250 V/m (STIM, n = 11). **p = 0.0032; ****p < 0.0001; by Two-way ANOVA and Holm-Sidak post-hoc test. Mean ± SEM shown. (b) Quantitative assessment of the outward shift of aggregate frontiers after 24 h for DAOY with no dcEF (CTL, n = 23) or 250 V/m (STIM, n = 29). **p = 0.0032; ***p = 0.0007; ****p < 0.0001; by Two-way ANOVA and Holm-Sidak post-hoc test. (c) Example spheroid morphology before and after experiment, with and without 250 V/m dcEF for DAOY spheroidal aggregates after 24 h. Bounding boxes are colored for time = 0 (pink), time = 24 h (green). (d,e) Proportional change in spheroid area from time = 0 to time = 8 for (d) DAOY, and (e) U87 mg spheroidal aggregates. *p = 0.0133; ****p < 0.0001; by One-way ANOVA and Holm-Sidak post-hoc test. Mean ± Tukey box plots shown. Scale: 200 μm.

Figure 4

Differential expression of transcripts from RNA-SEQ. (a–d) Volcano plots comparing fold-change and p-value of each identified transcript accession for (a,b) U87 mg and (c,d) DAOY cellular aggregates for samples exposed to 250 V/m dcEFs for (a,c) 2 h, or (b,d) 8 h compared 0 h, unexposed controls. Significance threshold is set to FDR < 0.05 and a Log₂(fold-change) of > 1 or < −1 (n = 3 for all conditions). Differentially expressed transcripts are shown in red. Clustermaps for transcripts per million of differentially expressed transcripts for: (e) U87 mg after 2 h; (f) DAOY after 2 h; (g) U87 mg after 8 h; (h) DAOY after 8 h; relative to 0 h unexposed controls. Each column represents a single replicate sample (U = U87 mg, D = DAOY). Note: for 8 h results, not all differentially expressed transcripts are shown, as Daoy significance cutoff was set to p < 0.0005 and U87 mg to p < 0.005 for improved readability. (i,j) Fold-change of most significantly differentially expressed transcripts. Comparing (i) U87 mg, or (j) DAOY differentially expressed genes for 2 h, or 8 h of dcEF relative to 0 h unexposed controls (U = U87 mg, D = DAOY). Not all differentially expressed transcripts are shown, as Daoy significance cutoff was set to p < 0.001 for improved readability.

Figure 5

Abridged pathway network for selected targets. Inhibited components are highlighted in green. Data obtained from multiple KEGG pathways.

Figure 6

Effect of PI3K/mTOR/IGF/AKT inhibitors on DAOY spheroidal aggregates undergoing electrotaxis. 24 h hours after dcEF (stim) or controls without dcEF (CTL) (a) 24 h DAOY data without inhibitor compounds, redisplayed from Fig. 3b. (b) LY294002 (2 μM) ***p = 0.0003, **p = 0.0020 (CTL, n = 10; STIM, n = 13); (c) BEZ235 (25 nM) *p = 0.0112, ****p < 0.0001 (CTL, n = 8; STIM, n = 10); (d) Rapamycin (100 nM) **p = 0.0014, *p = 0.00465, ****p < 0.0001 (CTL, n = 11; STIM, n = 10); (e) KU-0063794 (2.5 μM) **p = 0.0057, *p = 0.0351 (CTL, n = 7; STIM, n = 14); (f) OSI-906 (1 μM) **p = 0.0034 (orthogonal), **p = 0.0013 (cathodal bias), ****p < 0.0001 (CTL, n = 9; STIM, n = 8); (g) MK-2206 (2 μM) **p = 0.0040, ****p < 0.0001 (CTL, n = 10; STIM, n = 15); ns = not significant; Two-way ANOVA, Holm-Sidak post-hoc.

Figure 7

Effect of PI3K/mTOR/IGF/AKT inhibitors on U87 mg spheroidal aggregates undergoing electrotaxis. 24 h hours after dcEF (stim) or controls without dcEF (CTL) (a) 24 h U87 mg data without inhibitor compounds, redisplayed from Fig. 2d; (b) LY294002 (2 μM) *p = 0.0269 (CTL, n = 10; STIM, n = 13); (c) BEZ235 (25 nM) *p = 0.0306 (CTL, n = 14; STIM, n = 14); (d) LY294002 (20 μM) **p = 0.0015 (CTL, n = 9; STIM, n = 12); (e) BEZ235 (250 nM). (CTL, n = 8; STIM, n = 17); (f) CZC24832 (1 μM). (CTL, n = 14; STIM, n = 26); (g) Rapamycin (100 nM) (CTL, n = 15; STIM, n = 15); (h) KU-0063794 (2.5 μM) (CTL, n = 13; STIM, n = 15); (i) MK-2206 (2 μM) (CTL, n = 8; STIM, n = 16); (j) OSI-906 (1 μM) *p = 0.0479 (CTL, n = 16; STIM, n = 13); ns = not significant; Two-way ANOVA, Holm-Sidak post-hoc.

Figure 8

Effect of ErbB inhibitors on U87 mg spheroidal aggregates undergoing electrotaxis. 24 h hours after dcEF (stim) or controls without dcEF (CTL). (a) U87 mg data without inhibitor compounds, redisplayed from Fig. 2d. (b) Erlotinib (5 μM) **p = 0.0016, ****p < 0.0001 (CTL, n = 18; STIM, n = 13); (c) AZD6931 (1 μM). (CTL, n = 15; STIM, n = 16); (d) Mubritinib (1 μM).(CTL, n = 17; STIM, n = 29); (e) AZ5104 (6 nM) *p = 0.0386 (CTL, n = 3; STIM, n = 14); ns = not significant; Two-way ANOVA, Holm-Sidak post-hoc.

Figure 9

Additional inhibitors that impacted electrotaxis of DAOY (a,b) or U87 mg (c,d) spheroidal aggregates. 24 h hours after dcEF (stim) or controls without dcEF (CTL) (a) DAOY data without inhibitor compounds, redisplayed from Fig. 3b; (b) DAOY + Y-27632 (10 μM) *p = 0.0279 (anodal), *p = 0.0472 (orthogonal bias), ****p < 0.0001 (CTL, n = 10; STIM, n = 11); (c) U87 mg data without inhibitor compounds, redisplayed from Fig. 2d; (d) U87 mg + Bosutinib (1 μM) (CTL, n = 12; STIM, n = 12); ns = not significant.; Two-way ANOVA, Holm-Sidak post-hoc.

Figure 10

Revised pathway networks after findings from pharmacological inhbition studies. For (a) U87 mg; and (b) DAOY spheroidal aggregates. Inhibited components that impacted electrotactic responses are highlighted in red.