Presence of stromal cells in a bioengineered tumor microenvironment alters glioblastoma migration and response to STAT3 inhibition

Abstract

Despite the increasingly recognized importance of the tumor microenvironment (TME) as a regulator of tumor progression, only few in vitro models have been developed to systematically study the effects of TME on tumor behavior in a controlled manner. Here we developed a three-dimensional (3D) in vitro model that recapitulates the physical and compositional characteristics of Glioblastoma (GBM) extracellular matrix (ECM) and incorporates brain stromal cells such as astrocytes and endothelial cell precursors. The model was used to evaluate the effect of TME components on migration and survival of various patient-derived GBM cell lines (GBM10, GBM43 and GBAM1) in the context of STAT3 inhibition. Migration analysis of GBM within the 3D in vitro model demonstrated that the presence of astrocytes significantly increases the migration of GBM, while presence of endothelial precursors has varied effects on the migration of different GBM cell lines. Given the role of the tumor microenvironment as a regulator of STAT3 activity, we tested the effect of the STAT3 inhibitor SH-4-54 on GBM migration and survival. SH-4-54 inhibited STAT3 activity and reduced 3D migration and survival of GBM43 but had no effect on GBM10. SH-4-54 treatment drastically reduced the viability of the stem-like line GBAM1 in liquid culture, but its effect lessened in presence of a 3D ECM and stromal cells. Our results highlight the interplay between the ECM and stromal cells in the microenvironment with the cancer cells and indicate that the impact of these relationships may differ for GBM cells of varying genetic and clinical histories.

Fig 1. Experimental setup to analyze GBM migration in 3D co-culture with stromal cells.

(A). Cell populations are expanded in 2D liquid culture, recovered, fluorescently labeled and homogenously encapsulated within a 3D matrix formed by polymerization of an oligomer collagen-hyaluronan solution. (B). The position of individual glioblastoma cells (green) is tracked by confocal microscopy every 1.5 h during 15 h. (C). The trajectories of GBM single cells are analyzed to obtain accumulated migration distance, net displacement, migration velocity and directionality. Z-stack: 200 μM, step: 12 μM. Bar 100 μM.

Fig 2. Astrocytes increase the migration GBM cells in a 3D-brain-like model.

(A). Accumulated distance of migration, measured over 15 h, increased for all three GBM cell lines. (B). Net displacement between initial (0 h) and final points of migration (15 h) also increased for all cells when co-cultured with human primary astrocytes. (C). The velocity of migration is relatively stable for GBAM1 (CD133⁺) and GBM43 but decreases overtime for GBM10. Data represents a population of 250–2500 individual cells, from at least 2 independent repetitions. Comparison between groups was done by Kruskal-Wallis test. * Represents statistical significant difference at α = 0.05.

Fig 3. Effect of ECFCs and astrocytes-ECFCs on 3D GBM migration.

(A). Accumulated distance of migration over 15 h. (B). Net displacement between initial (0 h) and final points of migration (15 h). Net displacement of GBM cell lines increased due to presence of stromal cells (+ECFC+Ast). (C). The highest migration velocity was observed at the beginning and decreased over time. Migration data represents a population of 250–2500 individual cells from at least 2 independent repetitions for GBM only and from 3 replicates for ECFC and ECFC-Ast. Comparison between groups was done by Kruskal-Wallis test. * Represents statistical significant difference at α = 0.05.

Fig 4. Presence of stromal cells in the tumor microenvironment increases intrinsic migration directionality of GBAM1 (CD133+).

Data represents a population of 250–2500 individual cells. GBM only and GBM+Ast correspond to at least 2 independent repetitions per group, GBM+ECFC and GBM+Ast+ECFC correspond to 3 replicates. Comparison between groups was done by Kruskal-Wallis test and Tukey-Kramer mean comparison * Represents statistical significant difference at α = 0.05.

Fig 5. GBM exhibits basal activation of STAT3 when cultured in a 3D matrix that recapitulates characteristics of GBM ECM.

Presence of astrocytes in a 3D environment increased the basal STAT3 phosphorylation of GBAM1 compared to 3D culture without astrocytes (see materials and methods for GBM protein recovery during co-culture). STAT3 status of GBM cell lines in 2D culture was tested for at least 2 independent samples with analogous results. Expression of STAT3 from GBM in 3D culture correspond to one sample. Phosphorylation ratio correspond to the images shown.

Fig 6. SH-4-54 decreased STAT3 migration of GBM43 but not of GBM10 in a 3D model of GBM microenvironment.

GBM43 migration distance decreased with SH-4-54 and STAT3 siRNA knockdown treatments (KD-STAT3) while GBM10 migration was not reduced by the inhibitor SH-4-54 or by STAT3 siRNA treatment. Data represents a population of 250–2500 individual cells from 2 replicates. Comparison between groups was done by Kruskal-Wallis test. * Represents statistical significant difference at α = 0.05. Western blots correspond to cells recovered after 48 h of transfection with STAT3 siRNA or control siRNA A.

Fig 7. Presence of a 3D brain-like matrix reduces the cytotoxic effect of SH-4-54 in GBAM1 Treatment with SH-4-54 for 72 h drastically reduced cell viability of GBAM1 in 2D liquid culture, but only decreased viability slightly in 3D culture.

GBM10 and GBM43 showed a slight decrease in viability after SH-4-54 treatment in 2D culture. The effect of treatment in 2D culture was not significantly different than 3D culture for GBM10 and GBM43. Cell viability was normalized with viability of cells treated with DMSO. Bars indicate Mean ± SE, n> = 3 independent repetitions. Comparison between groups was done by t-test. * Represents statistical significant difference at α = 0.05.

Fig 8. Presence of stromal cells decreased the cytotoxic effect of SH-4-54 on GBAM1 in 3D culture.

Astrocytes and ECFCs within the 3D matrix reduced the effect of SH-4-54 (5μM) on GBAM1 viability compared to 3D culture without stromal cells but has no effect on GBM10 viability. Cell viability was normalized with cells treated with DMSO for each of the culture models used. Bars indicate Mean ± SE, n> = 3 independent repetitions. For GBM in co-culture viability analysis was performed as triplicates, each assay with a population of 200–900 individual cells (see materials and methods). Comparison between groups was done by Kruskal-Wallis test and Tukey-Kramer mean comparison * Represents statistical significant difference at α = 0.05.