The Size of Glioblastoma Spheroids Influences Patterns of Invasion and Temozolomide Efficacy

Abstract

Glioblastoma (GBM) is one of the most common malignant brain tumors, with patient mortality driven by invasion into the surrounding brain microenvironment and drug resistance. Multicellular spheroids are an increasingly common model to study GBM invasion and drug response in engineered biomaterials. However, a key design feature of tumor spheroid studies is the size of each spheroid (number of cells, diameter). Given the heterogeneous growth of GBM cells at the surgical margin, spheroids of different sizes may also have clinical relevance. Here, we define shifts in behavior and drug response of wild-type (WT) and temozolomide (TMZ)-resistant GBM spheroids as a function of initial spheroid size. GBM spheroids ranging from 1,000 to 10,000 cells in size were embedded into a methacrylamide-functionalized gelatin hydrogel. GBM spheroid size had an inverse relationship with the number of apoptotic cells. We observed significant spheroid-size-dependent effects on TMZ efficacy for both TMZ-resistant and WT cells. Interestingly, high single doses of TMZ were more effective in reducing three-dimensional migration from smaller spheroids than metronomic dosing, while high single dose and metronomic dosing were equally effective in reducing invasion for large TMZ-resistant spheroids. Our study highlights the importance of considering and reporting spheroid size for cancer tissue engineering studies considering invasion and drug resistance. It also informs future studies of residual GBM at the tumor margins most responsible for patient relapse and mortality.

Introduction

Glioblastoma (GBM) is the most common, aggressive, and lethal form of primary brain cancer.^1,2 While advances in radiotherapy and chemotherapy have extended median survival from 12 to ~15 months, overall survival (<5% after 5 years) remains poor and largely unchanged over the past two decades.^3–10 Current standard of care includes maximal surgical resection followed by radiation and chemotherapy with the alkylating agent temozolomide (TMZ).⁹ However, mortality is driven by rapid diffusive spreading of GBM cells beyond the surgical margin.^11–14 GBM tumors recur rapidly (6.9 month post debulking) and in close proximity (>90% within 2 cm) of the original tumor.^15,16 It is therefore important to develop strategies that target GBM cells infiltrating the tissue microenvironment beyond the surgical margin.¹⁷ While studies of GBM progression and therapeutic response are commonly performed using in vivo assays,^18–20 such models are less well-suited to examine both the kinetics of cell invasion and for nuanced studies of the role of the tumor microenvironment on progression.

In vitro experiments using both two-dimensional cell cultures^21,22 and three-dimensional biomaterial models of the brain tumor microenvironment^14,23–25 offer a lens to elucidate the instructive nature of the composition, mechanical properties, and multicellular nature of the GBM tumor microenvironment. Recent efforts have identified key features of the GBM microenvironment such as chemical and mechanical gradients^26–28 as well as fluid flow.^29–31 Our group has developed a methacrylamide-functionalized gelatin (GelMA) hydrogel to study the role of matrix stiffness, hypoxia, and hyaluronic acid content on GBM progression using both immortalized cell lines and patient-derived xenografts.^32–34 In these studies, metrics of GBM response can be gathered from low-density cell populations dispersed in, or high-density cell spheroids embedded into, the hydrogel network. Both modes of study are important because single cells and clusters of GBM cells may persist in the tumor margins after debulking.^35–38 We used these hydrogel models to show angiocrine signals influence GBM invasion and TMZ resistance.^39–42 More recently to study the emergence of TMZ resistance⁴³ in response to physiologically relevant single and metronomic^44,45 TMZ doses using isogenic pairs of TMZ-resistant and responsive GBM cell lines developed by Tiek et al.⁴⁶ Metronomic dosing, repeated administration at a dosage below the maximum tolerated dose, has been suggested to offer better penetration and potentially circumvent resistance mechanisms seen with high single drug doses^44,47 and is more consistent with the clinical use of TMZ. While difficult to examine in vivo, tissue engineering approaches offer the opportunity to explore the biophysical basis by which the aggregated nature of cancer spheroids may influence therapeutic response to metronomic dosing.

We previously used GBM spheroids to profile patterns of cell invasion in response to changes in matrix composition, hypoxia, and the presence of GBM stem cell subpopulations.^34,48–51 However, previous studies of the role of confinement and biophysical signals on liver development^52–54 and other cancer stem cell phenotypes⁵⁵ suggest that spheroid size may be an essential design principle. Larger spheroids have been hypothesized to affect the emergence of hypoxia and necrosis as well as to induce changes in gene expression that may underlie invasion and chemoresistance.⁵⁶ However, given the likelihood that multiple scales of GBM spheroids are present in the tumor margins, it is essential to understand the degree to which spheroid size influences metrics of GBM phenotype.

The goal of this project is to define the influence of GBM spheroid size on cell activity, invasion, and drug response within a model GelMA hydrogel (Fig. 1). We fabricated spheroids from both TMZ-resistant or wild-type (WT) GBM cell lines, with sizes ranging between 100 and 12,000 cells per spheroid. We report the effect of spheroid size on GBM cell growth, gene expression, apoptosis, and invasion. We define the effects of single versus metronomic TMZ dosing on GBM spheroids as a function of size and TMZ-resistant status. Together, these findings establish design principles regarding spheroid sizes necessary to evaluate therapeutic compounds and establish thresholds (e.g., spheroid surface area to volume) that alter GBM phenotypes in vitro.

FIG. 1.

Conceptual illustration of experimental plan. (a) GBM spheroids were formed from isogenically matched pairs of wild-type (42MGBA-WT) or TMZ-resistant (42MGBA-TMZres) cells across a range of spheroid sizes (100–12,000 cells/spheroid). (b) Schematic depiction of the TMZ treatments. GBM spheroids encapsulated in GelMA hydrogel are subjected to a DMSO control, metronomic treatment (100 μM per day for 5 days) via TMZ, or single high TMZ dose (500 μM at day 1). Cell phenotype, invasion, and gene expression patterns were traced for up to 5 days after initial TMZ exposure. DMSO, dimethyl sulfoxide; GBM, glioblastoma; GelMA, methacrylamide-functionalized gelatin; TMZ, temozolomide.

Materials and Methods

Cell culture

WT (42MGBA-WT) and TMZ resistant (42MGBA-TMZres) 42MGBA cell pairs were acquired from the Riggins lab at Georgetown University.⁴⁶ 42MGBA cells were cultured using Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin; media was changed every 48 h. Cells were maintained in incubators at 37°C and 5% CO₂ and used within 10 passages.

Spheroid formation

GBM spheroids were generated using two well-established protocols. Ultralow attachment plate method: WT (42MGBA-WT) and TMZ-resistant (42MGBA-TMZres) GBM cells were formed into spheroids using the nutation method.^42,57 Briefly, GBM cells were seeded in Corning^® spheroid microplates, incubated at 37°C on a rotating incubator shaker (Thermo Fisher, Canoga Park, CA) at 60 rpm for 48 h. AggreWell^™ microwell method: WT (42MGBA-WT) and TMZ-resistant (42MGBA-TMZres) GBM spheroids were also generated in larger quantities required for flow cytometry analysis. Briefly, AggreWell 400 or 800 microplates (STEMCELL Technologies, Vancouver, Canada) were pretreated with antiadherence rinsing solution (STEMCELL Technologies, Vancouver, Canada) for 5 min. The antiadherence rinsing solution was then removed and replaced with 2 mL of cell culture media seeded with the desired number of cells to form each spheroid. A rotating shaker is not used for the formation of tumor spheroids using AggreWell^™ 400 or 800.

Immunostaining

Samples were fixed using 4% formaldehyde solution (Sigma-Aldrich, Burlington, MA), permeabilized using 0.5% Tween-20 (Fisher Scientific, Fairlawn, NJ) in PBS, and blocked using 2% donkey serum (Sigma-Aldrich) in PBS-T (0.1% Tween-20 in PBS) for 2 h at room temperature. Ki-67 recombinant rabbit monoclonal antibody SP6 (Thermo Fisher, Canoga Park, CA) was diluted at a 1:200 ratio in 2% donkey serum in PBS-T. Samples were incubated in primary antibody at 4°C overnight. Alexa 555 was used as a secondary antibody (Thermo Fisher, Canoga Park, CA) at a 1:500 ratio in PBS-T. Hydrogels were washed with PBS-T between antibody incubation. Hoechst 33342 (Thermo Scientific, Rockford, IL) was used as a nuclear marker at a 1:1000 ratio in PBS-T.

Live and dead staining

Briefly, spheroids collected from the AggreWell^™ 400 and 800 were treated with 0.05% trypsin/EDTA and incubated at 37°C for 20 min; dispersed GBM cells incubated with 0.05% trypsin/EDTA for 20 min showed no significant decrease in cellular viability when assessed with 0.4% Trypan Blue solution and Countess II FL automated cell counter (Thermo Fisher, Canoga Park, CA) (data not shown). Dissociated spheroids were centrifuged at 300 g for 5 min, then washed with PBS twice. Working concentration of Calcein-AM and bobo-3 iodide (Thermo Fisher, Canoga Park, CA) was added to the cell suspension and incubated for 15 min at room temperature. Calcein-AM and bobo-3 iodide solution were removed, and the cells were washed with PBS twice before further experiments.

Flow cytometry

All flow cytometry measurements were performed using a FACSymphony A1 configuration (BD Biosciences, Franklin Lakes, NJ). Briefly, spheroids were recovered from biomaterial culture, dissociated using 0.5% trypsin, then resuspended in 500 μL 5% FBS for analysis in biological triplicate. Live cells were gated by forward scatter and side scatter. Fluorescence data were collected for GFP (excitation laser: 488 nm; bandpass filter: 530/30 nm) and PE-Texas Red (excitation laser: 561 nm; bandpass filter: 610/20 nm), with a minimum of 50,000 singlet events collected for each sample. Flow cytometry data were analyzed using FACSDiva^™ Software (BD Biosciences, Franklin Lakes, NJ) and then graphed using Prism 8 software (GraphPad). Apoptotic cells were gated as PE-Texas Red (bobo-3 iodide, Thermo Fisher Scientific, Waltham, MA) and FITC (CalceinAM; Thermo Fisher Scientific, Waltham, MA) positive. Necrotic cells were gated as PE-Texas Red (bobo-3 iodide, Thermo Fisher Scientific, Waltham, MA) positive and FITC (CalceinAM) negative (Supplementary Fig. S2).

GFP transduction of 42MGBA cell lines

42MGBA cells were transduced using the CMV-rFLuc-T2A-GFP-mPGK-Puro lenti-labeler^™ plasmid (Systems Bioscience, Palo Alto, CA) per manufacturer’s instructions. Briefly, 7500 42MGBA cells were seeded in 48-well plates with both TransDux Max and Max enhancers diluted in culture media at a final concentration of 1×. The multiplicity of infection (MOI) used was calculated based on the infectious forming units provided by the manufacturer, with lenti-labeler added to the cells at an expected MOI of 50. Cells were incubated at 37°C and 5% CO₂ for 72 h before sorting (BigFoot Invitrogen Thermo Fisher Spectral Cell Sorter, Thermo Fisher, Canoga Park, CA) for GFP-positive cells. Isolated GFP-expressing cells were expanded for two passages before experiments.

GelMA synthesis

GelMA macromers for all hydrogel cultures were synthesized as previously described.⁴⁸ Briefly, porcine gelatin type A, 300 bloom (Sigma-Aldrich, St. Louis, MO), was dissolved in carbon-bicarbonate buffer (pH 9.4) at 65°C. Fifteen microliters of methacrylic anhydride (Sigma-Aldrich, St. Louis, MO) was added dropwise per gram of gelatin, and the reaction proceeded for 2 h with continuous stirring (250 rpm). The reaction was quenched with preheated deionized water and dialyzed for 7 days against deionized water with daily exchange. The product was then frozen at −20°C and lyophilized for 7 days. Proton nuclear magnetic resonance (HNMR) was used to determine the degree of functionalization (DOF). GelMA with a DOF of 85% (data not shown) was used in this study.

Spheroid encapsulation and outgrowth measurements

GFP-expressing 42MGBA-WT or 42MGBA-TMZres spheroids were suspended in a 5 wt.% GelMA precursor solution and then pipetted into 20 μL volume (cylindrical) Teflon molds (1 spheroid per hydrogel construct). GelMA hydrogels were polymerized in the presence of lithium acylphosphinate (0.1% w/v) and exposed to UV light for 35 s using a UV lamp (λ = 365 nm, 5.79 mW/cm²).⁵⁸ GBM cell invasion assay was performed using a DMi8 Yokogawa W1 spinning disk confocal microscope with a Hamamatsu EM-CCD digital camera (Leica Microsystems, Buffalo Grove, IL) per previously described methods.⁴² Overall GBM cell outgrowth area was quantified by using ImageJ (National Institutes of Health [NIH], Bethesda, MD). Briefly, sequential slices (5 μm thick) were imaged along the hydrogels’ depth. Maximum intensity projections were assembled from approximately 50 slices, with the composite image then binarized and analyzed with the analyze particle function on ImageJ (NIH, Bethesda, MD). Six independent spheroid outgrowth measurements were gathered per experimental group as a biological replicate.

TMZ treatment

We performed TMZ studies in GelMA hydrogels as previously described.⁴³ Briefly, TMZ (Calbiochem via Millipore Sigma, Burlington, MA) was dissolved in dimethyl sulfoxide (DMSO) at a concentration of 200 mM. DMSO alone (≦0.5% v/v) was used as a vehicle control for all experimental studies. Hydrogels containing GBM spheroids were cultured for 24 h prior to the administration of TMZ. Hydrogels were subsequently treated with one of the following TMZ dose regimens: a single dose of 500 μM or five daily doses of 100 μM (a cumulative dose of 500 μM). Media was replaced for both groups daily with either the appropriate concentration of TMZ or DMSO control.

RNA extraction

Collagenase IV (Worthington Biochem Corporation, Lakewood, NJ) was reconstituted with PBS to a working concentration (10 U/μL) before being added to individual hydrogel specimens. Hydrogels were incubated at 37°C for 20 min. Dissolved hydrogels and media were centrifuged at 300 g at 4°C for 5 min. After the removal of supernatant, 500 μL of tri-Reagent was added to 6–8 spheroids. A 20% volume of chloroform was then added to the tri-reagent mixture and centrifuged at 14,000 ×g at 4°C for 1 h. The aqueous phase was collected then RNA precipitation was performed by adding an equivalent amount of isopropyl alcohol. RNA was then extracted using RNA Clean and Concentrator Kit RC5 (Zymo Research, Irvine, CA).

Quantitative reverse transcription polymerase chain reaction

RNA was reverse transcribed to cDNA reagents using SuperScript^™ IV Reverse Transcriptase (Invitrogen, Carlsbad, CA) and a Bio-Rad MyCycler thermal cycler (Bio-Rad, Hercules, CA). Quantitative reverse transcription polymerase chain reaction (RT-qPCR) was performed using Applied Biosystems QuantStudio 7 Flex PCR system (Thermo Fisher, Waltham, MA) with 2× Universal SYBR Green Fast qPCR mix (ABclonal Technology, Woburn, MA). All primers (Supplementary Table S1) were designed and ordered using IDT PrimerQuest tool (Integrated DNA Technologies, Newark, NJ). Fold change was calculated using the ΔΔCT method and normalized to expression of the DMSO control.

Statistics

Statistics were performed using SPSS. Normality of data was determined using the Shapiro–Wilk test, and equality of variance was determined using Levene’s test. For normal data, comparisons between two groups were performed using a t-test, while comparisons between multiple groups were performed using a one-way ANOVA followed by Tukey’s post hoc test when assumptions were met. In the case where data were not normal, or groups had unequal variance, comparisons between two groups were performed using a Mann–Whitney test, while comparisons between multiple groups were performed using a Kruskal–Wallis test with Dunn’s post hoc test. Significance was determined as p < 0.05. All quantitative analyses were performed on hydrogels or microfluidic assays set up across at least three independent experiments.

Results

Library of GBM spheroids as a function of cell size

We first created spheroid libraries for both WT (42MGBA-WT) and TMZ-resistant (42MGBA-TMZres) GBM cell lines, ranging in size from 100 to 12,000 cells (Fig. 2a). We quantified the resultant size and morphology of the spheroids via overall projected surface area and circularity metrics from bright-field microscopy images. Spheroid projected surface area scale with number of cells, and broadly, all spheroids displayed consistent circularity indices of greater than 60%. Spheroids generated from TMZ-resistant GBM cells were significantly larger than those generated from the same number of WT GBM cells (Fig. 2b, Supplementary Table S3). Despite being significantly larger than spheroids formed from WT GBM cells, TMZ-resistant GBM spheroids made from greater than 2500 cells displayed similar levels of circularity as WT, a metric of the cohesive nature of the spheroid and its appropriateness to assess collective properties such as invasion. While not statistically significant, TMZ-resistant spheroids made from up to 2500 cells displayed a trend toward increased circularity versus GBM WT spheroids (Fig. 2c) suggesting an increased cohesive nature of small TMZ-resistant spheroids. We characterized the packing of GBM cells using fluorescent images obtained from CMV-rFLuc-T2A-GFP-mPGK-Puro lenti-labeler^™ labeled WT and TMZ-resistant cells formed into 10,000 cell spheroids (Supplementary Fig. S1). TMZ-resistant spheroids appear to be more densely packed with a brighter GFP core, whereas the fluorescent signal for the WT spheroids displays a more diffuse organization.

FIG. 2.

A library of GBM spheroids as a function of cell size (a) Bright field image of 42-MGBA spheroids initiated with different numbers of cells after 48 h (n = 8/group). Scale bar: 200 μm. (b) Projected surface area of GBM spheroids initiated as a function of size (n = 8/group). *p < 0.05: statistical significance exists between WT and TMZres groups at each seeding density, as well as between each seeding density for WT and TMZres groups. (c) Circularity of 42MGBA spheroids initiated at different seeding densities (n = 8/group). Statistical analysis was done by Kruskal–Wallis test followed by Mann–Whitney U test.

Patterns of GBM proliferation and apoptosis are influenced by spheroid size and TMZ-resistant status

We subsequently examined patterns of cell proliferation and death within GBM spheroids as a function of spheroid size and TMZ-resistant status (42MGBA-WT vs. 42MGBA-TMZres). Immunostaining was performed 48 h after encapsulation of spheroids into GelMA hydrogels. Broadly, we observed Ki-67+ proliferative cells in all 42MGBA-WT spheroids regardless of spheroid size (Fig. 3a). We observed higher levels of Ki-67 staining in 42MGBA-TMZres spheroids increasingly at the spheroid periphery, with the effect most pronounced in larger spheroids (Fig. 3b). We also interrogated patterns of cell death in GBM spheroids as a function of spheroid size and TMZ-resistant status 48 h after formation (Fig. 3c,d). Broadly, bobo-3 iodide positive (dead) GBM cells were observed in spheroids of all sizes for both 42MGBA-WT and 42MGBA-TMZres cells (Fig. 3c,d).

FIG. 3.

Patterns of GBM cell proliferation and apoptosis 48 h post spheroid formation. Representative fluorescent images of (a) wild-type versus (b) TMZ-resistant spheroids labeled with DAPI (nuclei) and Ki-67 (G1/S/G2/M phases of the cell cycle) as a function of spheroid size. Representative fluorescent images of (c) wild-type versus (d) TMZ-resistant spheroids labeled with Calcein-AM (live cells) and bobo-3 iodide (dead cells) as a function of spheroid size. Scale bar: 200 μm.

To better assess whether spheroid size influenced the emergence of an apoptotic or necrotic GBM population, large quantities of spheroids of each size were generated using an AggreWell^™ microwell plate process. The spheroids generated with AggreWell^™ microwell plates displayed similar morphology and circularity as those generated via the nutation method. Small spheroids (<200 cells) showed higher variation in size and circularity, likely due to the overall low seeding density (Supplementary Tables S2 and S3). These spheroids were pooled by size, then dissociated for flow cytometry analysis of the cellular content (Supplementary Fig. S2). Dissociation with 0.5% trypsin showed no effects on the GBM cell viability (data not shown). Disaggregated cells were analyzed for expression of bobo-3 iodide (PE-TexasRedA) and CalceinAM (FITC-A), with cells characterized as apoptotic (bobo-3+/CalceinAM+) versus necrotic (bobo-3+/CalceinAM−). We observed few (<5%) overall necrotic cells and no significant difference in the fraction of necrotic cells in spheroids as a function of spheroid size or TMZ-resistant phenotype (Fig. 4a). Interestingly, we observed significant effects of both spheroid size and TMZ-resistance on the emergence of apoptotic GBM cells (Fig. 4b). We observed significantly higher fractions of apoptotic cells in 42MGBA-WT versus 42MGBA-TMZres spheroids for almost all spheroid sizes, with higher fractions of apoptotic cells in smaller (≤5000 cells) than larger (10,000, 12,000 cells) spheroids regardless of TMZ-resistant status. For all subsequent studies, we chose two spheroid conditions (1000 vs. 10,000 cells) that displayed different levels of proliferative capacity and apoptosis.

FIG. 4.

Percentage of apoptotic versus necrotic GBM cells as a function of spheroid size and TMZ-resistance. Spheroids were dissociated and analyzed after cultured for 48 h. (a) The percentage of necrotic cells in spheroids as a function of spheroid size and TMZ-resistant status (n = 3). (b) The percentage of apoptotic cell population in spheroids as a function of spheroid size and TMZ-resistant status (n = 3). Statistical analysis was completed by one-way ANOVA followed by Tukey’s Honestly Significant Difference (HSD) test. Different letters (a, b, c, d) denote p < 0.05 between wild-type and TMZ resistant spheroids as a function of size. WT1K, 42MGBA-WT, 1000 cell spheroid; WT10K, 42MGBA-WT, 10,000 cell spheroids. TMZ1K, 42MGBA-TMZres, 1000 cell spheroids; TMZ10K, 42MGBA-TMZ, 10,000 cell spheroids. 100 μMx5, treated with five daily (metronomic) doses of 100 μM TMZ. 500 μM, treated with one dose of 500 μM TMZ.

Single high dose and metronomic TMZ reduce GBM invasion for WT, but not TMZ-resistant spheroids

Single spheroids initiated with 1000 or 10,000 cells were encapsulated in individual GelMA hydrogel and subjected to treatment via TMZ. TMZ exposure was either a single exposure to 500 μM TMZ, replaced after 24 h with complete medium containing an equivalent concentration of DMSO, or metronomic dosing via 100 μM TMZ each day for 5 consecutive days, with all treatments compared with a DMSO control (Supplementary Fig. S3). Projected surface area, calculated by quantifying the area of GFP signal from transduced GBM cells, was quantified on day 3 and 5 post-TMZ treatment. Small (1000 cells/spheroid) and large (10,000 cells/spheroid) 42MGBA-WT spheroids displayed significant invasion into the surrounding GelMA hydrogel, marked by an increase in projected surface area with time. The effect of a single high dose or metronomic dosing on small 42MGBA-WT spheroids did not become significant until 5 days after treatment, with significantly reduced invasion in response to metronomic dosing (vs. control) and a single high TMZ dose (vs. metronomic, control; Fig. 5a,c). The effects of TMZ treatment on larger (10,000 cells/spheroid) GBM spheroids followed similar trends, though with a single high TMZ dose significantly reducing cell invasion as early as 1 day post-treatment and with metronomic dosing or a single high dose both significantly reducing invasion by day 5 (Fig. 5b,d).

FIG. 5.

The effect of a single high TMZ dose or metronomic dosing on invasion of WT spheroids. (a) Representation image of small (1000 cells) 42MGBA-WT spheroids treated with metronomic (left) versus a single high TMZ dose (right) 5 days post-treatment. (b) Representation image of large (10,000 cells) 42MGBA-WT spheroids initiated with 10,000 cells treated with metronomic (left) versus a single high TMZ dose (right) 5 days post-treatment. (c) Quantifying cell outgrowth for small (1000 cell) spheroids over the course of 5 days after TMZ treatment (n = 3/group). (d) Quantifying cell outgrowth for large (10,000 cell) spheroids over the course of 5 days after TMZ treatment (n = 3/group). Statistical significance is represented as *p < 0.05. WT1K, 42MGBA-WT, 1000 cell spheroid; WT10K, 42MGBA-WT, 10,000 cell spheroids. 100 μMx5, treated with five daily (metronomic) doses of 100 μM TMZ. 500 μM, treated with one dose of 500 μM TMZ.

We observed limited invasion of 42MGBA-TMZres GBM cells for both small and large spheroids, consistent with earlier studies of 42MGBA-TMZres invasion in 3D hydrogels.^43,59 Smaller spheroids (initiated with 1000 cells) did not show any changes in cell invasion in response to either a single high dose or metronomic dosing (Fig. 6a,c). While both TMZ treatment groups significantly reduced projected surface area on day 3 and 5 for large spheroids (Fig. 6b,d), there was no difference between TMZ treatments, and overall invasion was small compared with WT GBM cells.

FIG. 6.

The effect of a single high TMZ dose or metronomic dosing on invasion of TMZ-resistant spheroids. (a) Representation image of small (1000 cells) 42MGBA-TMZres spheroids treated with metronomic (left) versus a single high TMZ dose (right) 5 days posttreatment. (b) Representation image of large (10,000 cells) 42MGBA-TMZres spheroids initiated with 10,000 cells treated with metronomic (left) versus a single high TMZ dose (right) 5 days post-treatment. (c) Quantifying cell outgrowth for small (1000 cell) TMZ-resistant spheroids over the course of 5 days after TMZ treatment (n = 3/group). (d) Quantifying cell outgrowth for large (10,000 cell) TMZ-resistant spheroids over the course of 5 days after TMZ treatment (n = 3/group). Statistical significance is represented as *p < 0.05. TMZ1K, 42MGBA-TMZres, 1000 cell spheroids; TMZ10K, 42MGBA-TMZ, 10,000 cell spheroids. 100 μMx5, treated with five daily (metronomic) doses of 100 μM TMZ. 500 μM, treated with one dose of 500 μM TMZ.

Difference in gene expression across spheroids of different sizes

We performed RT-qPCR on gene targets related to cell death (caspase-3 [CASP3], TGF-β), cellular growth (PTEN, TIMP-1), invasion (MMP2), and drug resistance (CD44, HIF1α) (Fig. 7a). WT GBM cells in larger spheroids showed increased expression of PTEN, STAT3, CASP3, TGF-β, TIMP-1, and MMP2, but reduced CD44 expression. Interestingly, a different expression pattern was observed in TMZ-resistant cells, where larger spheroids demonstrated increased expression of TIMP-1 and CD44 but reduced expression of TGF-β. Interestingly, quantitative analysis of TIMP-1 (ΔCT normalized to actin expression) showed trends toward higher TIMP-1 expression in larger versus smaller spheroids for both WT and TMZ-resistant cells, but with TMZ-resistant cells displaying significantly higher TIMP-1 expression versus WT cells (Fig. 7b).

FIG. 7.

Shift in gene expression as a function of GBM spheroid size, TMZ treatment, and TMZ-resistant status. (a) Heatmap of relative gene expression of wild-type versus TMZ-resistant spheroids of different sizes (1000 or 10,000 cells) in conventional culture. WT1K and TMZ1K are set to 1.0 for fold-change comparisons. (b) ΔCT value of TIMP-1 of wild-type and TMZ-resistant spheroids of different sizes (1000 vs. 10,000 cells). All data were normalized to ΔCT value of housekeeping gene actin (n = 3/group). (c) Heatmap of relative gene expression of wild-type versus TMZ-resistant spheroids of different sizes (1000 vs. 10,000 cells) and as a function of TMZ treatment relative to DMSO control (single 500 μM dose vs. 5 days of 100 μM metronomic treatment). (d) TIMP-1 expression for GBM spheroids as a function of size, TMZ-resistant status and TMZ dose, relative to the DMSO control (n = 3/group). (e) CASP3 gene expression for GBM spheroids as a function of size, TMZ-resistant status, and TMZ dose, relative to DMSO control (n = 3/group). Statistical analysis was completed by one-way ANOVA followed by Tukey’s HSD test. Different letters denote significance level at p < 0.05. WT1K, 42MGBA-WT, 1000 cell spheroid; WT10K, 42MGBA-WT, 10,000 cell spheroids. TMZ1K, 42MGBA-TMZres, 1000 cell spheroids; TMZ10K, 42MGBA-TMZ, 10,000 cell spheroids. 100 μMx5, treated with five daily (metronomic) doses of 100 μM TMZ. 500 μM, treated with one dose of 500 μM TMZ.

Change in gene expression on day 5 after TMZ treatments

We subsequently examined changes in gene expression of GBM spheroids of different sizes after 5 days of metronomic or single high TMZ treatment. In general, WT cells showed increased expression (vs. DMSO control) of CASP3, TIMP-1, HIF1α, and TGF-β in response to either TMZ treatments (Fig. 7c). TIMP-1 expression was significantly higher in the WT cell line compared with the TMZ-resistant cell line after TMZ treatment (Fig. 7d); while less pronounced, similar trends were also observed for increased CASP3 expression in WT GBM cells after TMZ treatment (Fig. 7e). Interestingly, small WT spheroids showed higher expression of PTEN, STAT3, MMP2, TGF-β, and CD44 compared with the larger spheroids when treated with either metronomic or single high TMZ doses. Although there was no statistical significance, several trends were observed in changes in gene expression for the TMZ-resistant cells. Notably, metronomic dosing increased expression of CASP3, PTEN, and STAT3 compared with a single high TMZ dose, while larger spheroids trended toward increased expression of MMP2, TIMP-1, TGF-β, and CD44 in response to single high TMZ doses (Fig. 7c, Supplementary Fig. S4).

Discussion

GBM is marked by significant invasion into the surrounding brain parenchyma. Cancer tissue engineering platforms offer the opportunity to dissect the role of multiple components of the tumor microenvironment on metrics of progression and therapeutic response. Many first-generation studies have used GBM cells distributed within a hydrogel biomaterial, simulating low-density cancer cell populations to assess how matrix composition,^48,60,61 stiffness,^48,62–65 and structure⁶⁶ influence invasion and drug response. Diffuse GBM invasion into the brain microenvironment is often modeled in vitro as radial invasion of GBM spheroids encapsulated in a hydrogel matrix, where composition, stiffness, and inclusion of other niche-associated factors can strongly affect invasion.⁶⁷ Our lab has extensively described a GelMA hydrogel platform to study GBM cell phenotype and therapeutic response. While not perfectly mimetic of the brain, it provides a well-characterized material that supports GBM as well as GBM stem cell invasion, proliferation, and drug response studies. For example, patient-derived Epidermal growth factor receptor (EGFR) mutant GBM cells in this hydrogel showed clinically relevant changes in activity and gene expression in response to cyclic erlotinib exposure.^48,68 This hydrogel also provides a route to incorporate brain-mimetic hyaluronic acid to consider the role of hyaluronic acid molecular weight,⁵⁰ mode of exposure (soluble vs. matrix-bound),³⁴ as well as the inclusion of microvascular cells and microglia to form neurovascular models of GBM progression.^42,51,69 Critical for this study, we recently used the GelMA hydrogel to benchmark TMZ responses for a series of paired TMZ-resistant and responsive GBM cell lines (including the 42MGBA variant used here),⁴³ establishing the technical basis for efforts here to evaluate the role of spheroid size on TMZ response and invasion. While important, the multicellular nature of GBM clusters in vivo may also significantly influence cell activity. Yet while multicellular tumor spheroids are increasingly used in a wide range of cancer tissue engineering studies, spheroid sizes can differ drastically among studies.^70,71 Larger spheroids are often hypothesized to lead to greater diffusional gradients for nutrients and oxygen that may contribute to the formation of necrotic core, regional changes in cell proliferative activity that promote chemoresistance, and shifts in gene expression associated with a more aggressive phenotype.^72,73

The goal of this study was to define the role of spheroid scale on metrics of GBM progression (invasion, drug response) traditionally used in cancer engineering models. Because TMZ sensitivity depends on the presence of proliferative cells, tissue engineering models of TMZ response have often relied on supraphysiological TMZ doses. We recently validated a tissue engineering workflow, using an isogenically matched pair of TMZ-resistant and responsive cells with disparate MGMT expression levels generated by Tiek et al.,⁴⁶ to assess TMZ response at physiological and metronomic dosing levels. While primarily for cells suspended in a GelMA hydrogel matrix, these studies confirmed that 42MGBA-TMZres cells showed reduced TMZ sensitivity in GelMA hydrogels. Our current study adds to this narrative, showing spheroid size is an important design characteristic that influences GBM cell behavior.

While we described spheroids from a fabrication perspective, we created GBM spheroids from both 42MGBA-WT and 42MGBA-TMZres cells that range in scale from <130 μm (WT 100 cells/spheroid) to >600 μm (TMZ-resistant 12,000 cells/spheroid) in diameter, which are relevant for surgical resection and recurrence perspectives. Spheroid size increased with number of cells per spheroid and showed a high degree of circularity. However, spheroids created from TMZ-resistant cells were significantly larger, more compact, and trended toward a higher degree of circularity than WT GBM cells. Qualitative fluorescent images suggested that Ki-67+ (proliferative) cells were more likely to be found toward the periphery of WT and TMZ-resistant spheroids, and that 42MGBA-TMZres were broadly more proliferative than WT counterparts, consistent with prior reports of a proliferative advantage for the TMZ-resistant variant.⁴⁶ Potential differences in the overall and spatial pattern of GBM cell proliferation in GBM spheroids suggest opportunities for future investigations that bring multiple avenues of spatial biology analyses (e.g., high-resolution fluorescence imaging, spatial transcriptomic analyses) to better quantify the formation of proliferative, apoptotic, and necrotic zones within spheroids of different sizes and in response to TMZ therapy. Flow cytometry analysis of spheroids showed that a higher fraction of apoptotic cells was present in WT (vs. TMZ-resistant) cells, particularly in smaller (<5000 cells/spheroid) spheroids. While previous work using mammary carcinoma cells⁷⁴ has suggested the formation of a necrotic core in tumor spheroids, we observed only a small fraction of necrotic cells. However, the increased fraction of apoptotic cells we observed in small GBM spheroids suggests that a relationship between spheroid size and apoptosis and necrosis may be more variable across cell types. This result motivates future studies to expand our understanding of GBM apoptosis within cancer spheroids. The absence of a quantifiable necrotic core in this study may be a result of insufficient culture time or poor probe diffusion. As a result, there is an opportunity in future studies to combine histological sectioning and specific necrotic marker such as Annexin V to further define spatial, intraspheroid patterns of cellular behavior. In addition, it will be important to more deeply consider molecular mechanisms that govern GBM apoptosis. Bobo-3 iodide is a membrane-impermeant fluorescent nucleic acid dye used to mark cell death by penetrating fragmented cell membranes. However, cancer cells have been previously shown to recover from a compromised membrane.⁷⁵ There is also the possibility to define the emergence of expression levels of antiapoptotic proteins (e.g., Bcl-2 family protein) or other known inhibitors of apoptosis (e.g., XIAP, cIAP1, cIAP2).^75–77 However, the increase in Ki-67 staining observed in GBM spheroids, particularly in larger spheroids containing TMZ-resistant GBM cells, was also consistent with increased 42MGBA-TMZres proliferation seen previously in 2D culture.⁴⁶

GBM invasion versus proliferation (“go vs. grow”) plasticity suggests a process by which GBM cells can navigate complex selection pressures in the tumor margins.^78,79 Interestingly, we observed that while 42MGBA-TMZres proliferation was higher in spheroid culture, both large and small spheroids formed from 42MGBA-WT cells displayed significantly increased invasion into GelMA hydrogels. This finding was also consistent with our prior work that showed for 5000 cell spheroids, 42MGBA-WT cells were significantly more invasive than TMZ-resistant cells.⁴³ Importantly, our work here extends the range over which this observation is consistent (1000–10,000 cells/spheroid), suggesting that 42MGBA-WT cells exhibit an invasive phenotype across a wide range of biophysical environments. More interestingly, invasion of 42MGBA-WT cells was significantly reduced by both metronomic and single high doses of TMZ, though the effect took 3–5 days after treatment. This delay corresponds to the known mechanism of action for TMZ, which relies on inducing DNA damage in proliferating cells.^80,81 Proliferation is also likely an important regulatory element in the context of GBM go (invasion) versus grow (proliferation) plasticity. Prior efforts have suggested that TMZ treatment can increase the proliferation and stemness of cancer stem cells, potentially contributing to TMZ resistance.^82–84 While cancer stem cells are absent in an immortalized cell line like the 42MGBA isogenic pair used here, there is a significant opportunity to examine the influence of GBM spheroid size and TMZ treatment on the proliferative versus invasive phenotype of patient-derived GBM specimens, which are known to possess a cancer stem cell subpopulation. Interestingly, 42MGBA-TMZres invasion was largely insensitive to TMZ treatment except in large (10,000 cells/spheroid) spheroids. Our prior work showed 42MGBA-TMZres cells (compared with 42MGBA-WT) displayed reduced G2/M and increased G1 phase in 2D culture⁴⁶ and no change in metabolic activity in response to 100 μm TMZ in 3D GelMA culture.⁴³ Our current findings suggest that 42MGBA-WT cells also exhibit reduced invasion in response to TMZ as well as the emergence of spatial patterns of cell proliferation versus invasion in large 42MGBA-TMZres spheroids that may be consistent with go-versus-grow plasticity. These findings are different from prior work with ovarian cancer cells that showed larger (5000 cells) spheroids were more resistant to taxol and cisplatin compared with smaller (500, 200, 100, or 50 cells) spheroids.⁸⁵ Given that larger spheroids have also been suggested to exhibit poor drug biotransport,⁷² our findings suggest the need for future studies to more closely examine the spatial-dependent shifts in GBM cell behavior and drug response. The observed differences in patterns of cell proliferation, spheroid-size-dependent effects on apoptotic fraction, and the relative dearth of publications addressing the role of spheroid size in engineered cancer models, suggest the need for future studies that quantify radial patterns of cell proliferation, apoptosis induction, and metronomic TMZ exposure as a function of spheroid size and time.

Finally, our analysis of GBM gene expression revealed that the largest changes were due to differences between TMZ-resistant and WT cells, rather than spheroid size. TMZ-resistant (vs. WT) cells displayed significantly upregulated expression of the MMP inhibitor TIMP-1. TIMP-1 has a multitude of functions and has been implicated in suppressing cell proliferation and metabolic activity in mesenchymal stem cells⁸⁶ but promoting cell proliferation and invasion via the TIMP1/FAK/Akt pathway in colorectal cancer cells.⁸⁷ Here, we observed increased TIMP-1 expression in large (10,000 cells) versus small (1000 cells) spheroids for both WT and TMZ-resistant lines and increased TIMP-1 expression in TMZ-resistant versus WT spheroids. We also observed phenotypic shifts in the expression of CASP3, a protease implicated in programmed cell death. CASP3 expression was elevated in WT spheroids after both metronomic and single high TMZ doses, especially in small spheroids. Interestingly, although we did not see a difference between metronomic versus a single high TMZ dose on invasion in TMZ-resistant cells, the combination of metronomic dosing and small spheroids induced the greatest increase in CASP3 expression in TMZ-resistant cells (Fig. 7e). We observed an inverse effect of spheroid size on MMP-2 expression in 42MGBA-WT specimens as well as a modest increase with spheroid size for 42MGBA-TMZres specimens. These findings, while broadly consistent with prior studies using drug-resistant (vs. WT) GBM cell lines that did not consider spheroid size,^46,88 suggest that smaller GBM spheroids may exhibit the most robust responses to TMZ treatment in tissue engineering models. This again argues for more careful consideration of spheroid size as a critical design factor in tissue-engineered cancer models.

Of critical importance is our use of tissue engineering methods to explore aspects of GBM tumor therapeutic response that are largely intractable in vivo. A recent review on metronomic dosing highlighted that metronomic dosing does not consistently increase overall survival⁸⁹ but rather provides benefit via reduced toxic side effects.^89,90 As a result, subsequent tissue engineering efforts may also provide valuable insights into the role of metronomic dosing on cancer spheroids but also on changes in the surrounding microenvironment such as models that incorporate vascular, stromal, and/or immune cell components.^{26,31,42,67,69,91} There is significant opportunity to use more complex tissue models as well as novel chemical imaging tools⁹² to develop a more clinically relevant model to assess therapeutic responses.

Conclusions

We report the adaptation of a well-characterized GelMA hydrogel system to evaluate the combined effect of TMZ dosing strategy (metronomic vs. single high TMZ dose), TMZ-resistant status, and the size of GBM spheroids on metrics of GBM invasion and drug response. While spheroids embedded into a hydrogel matrix are increasingly common in studies of GBM cell behavior, explicit consideration of the role of spheroid scale is essential to interpret these findings. TMZ-resistant and WT GBM cells maintain their TMZ response phenotypes, regardless of spheroid size. Analysis of the spatial pattern of cell proliferation as well as underlying gene expression data suggests that smaller GBM spheroids may exhibit an increased fraction of apoptotic cells and a more robust response to TMZ treatment. These findings suggest that spheroid size should be rigorously reported and that future investigation must also consider the role of spheroid size on matrix remodeling (TIMP-1) and apoptosis-associated (CASP3) proteases.

Supplementary Material

Supplementary Data

Impact Statement.

Glioblastoma (GBM) is a highly aggressive and recurrent brain cancer characterized by diffuse invasion at the tumor margins. Multicellular spheroids are an increasingly common model to study GBM invasion and drug response in engineered biomaterials. We report shifts in behavior and drug response of wild type and temozolomide (TMZ) resistant GBM spheroids as a function of initial spheroid size. This study highlights the importance of considering and reporting spheroid size for cancer tissue engineering studies considering invasion and drug resistance.