Using Tumor-Like Spheroids to Study the Effect of Anti-Cancer Drugs In Vitro

Abstract

Cell culture techniques have evolved in the last decades and allow now testing anti-cancer drugs using tumor-like spheroids. We describe here issues and trouble-shooting solutions when generating spheroids from three human melanoma cell lines (A375, WM115 and WM266). A375 cells generated irregular shape spheroids that were difficult to study due to their fragility. Spheroids generated from all cell lines initially reduced their diameter and increased compacity before increasing in size overtime. Cells present at the periphery of the spheroids showed higher metabolic activity than cells present in the core of the spheroids. When grown as spheroids, a smaller fraction of the A375 and WM115 cells was sensitive to the chemotherapeutic agent temozolomide as compared to cells grown on flat surface. However, this difference was not observed with WM266 cells. Although the presence of spheroids resulted in a smaller fraction of WM155 cells sensitive to the anti-cancer agent vemurafenib, the opposite was observed with A375 cells. Among the cells, WM266 cells were the most resistant to vemurafenib. In conclusion, our study suggests that cell lines behave differently in terms of spheroid formation, and that the effect of the 3D cellular architecture on drug effect is cell type and drug dependent.

1. Introduction

For many decades, cell culture experiments have been performed with cells attached to the flat surface of plastic wells. These experiments have provided valuable information in many aspects of cell biology, from the identification of new signaling pathways to the screening of new therapeutic anti-cancer agents [1, 2, 3]. However, the outcomes of many experiments performed in these in vitro conditions could not be reproduced in pre-clinical studies or in clinical trials [4]. One of the reasons for this failure is that two-dimensional (2D) cell culture growth conditions do not mimic the complex three-dimensional (3D) architecture of tumors [5, 6]. Indeed, when cells are grown on flat surfaces, they form only limited contacts with other cells, and the cellular architecture is different from the complex 3D structures observed in tumors (Figure 1). For instance, in 2D growth conditions all cells receive similar levels of oxygen from the environment. However, when grow as spheroids, diffusion of oxygen and nutrients vary according to the position of the cells in the spheroid: cells at the center of the spheroid do not receive the same levels of nutrient and oxygen and are often subject to hypoxia [7, 8].

Figure 1.

Schematic representation of cells grown in standard 2D growth conditions (left) or as spheroids (right). A denser extracellular matrix is present around the cells in 3D growth conditions.

To overcome the limitations of 2D cell culture conditions, researchers have developed 3D in vitro cell culture systems [9, 10, 11]. Experimental conditions have been optimized to allow the formation of monoculture or co-culture of 3D assemblies (spheroids) of cells that mimic tumors. Studies have shown that spheroids are valuable 3D cellular models that possess features shared with tumors [12, 13, 14, 15]. These features include the presence of gradients of nutrients, waste and gazes (oxygen and CO₂) throughout the spheroids, the presence of layers of cells with different metabolic activities (proliferative, quiescent and necrotic cells), as well as stronger deposition of extracellular matrix proteins as compared to 2D cultures of cells [5, 6, 16, 17].

Two main approaches are used for the generation of spheroids: scaffold-based and scaffold-free approaches [12, 13, 14, 15, 18]. Each approach has its advantages and disadvantages and the choice of the approach to use is guided by the type of assays the investigator will perform with these spheroids. In the scaffold-based approach, the cells are seeded onto artificial matrices with different porosities, resulting in the growth of cells both around and inside the matrices. This approach is not the most appropriate for drug studies because of the low reproducibility between batches of spheroids, the possible adsorption of the drugs with the matrices, the presence of non-transparent matrices that disturb spectroscopic analysis and the difficulties in detaching and extracting the cells from the scaffold when performing Western blot or flow cytometry analysis [18]. The second approach and the one that was chosen in this study is the generation of scaffold-free spheroids. Several methods deriving from the scaffold-free approaches have been developed and include the hanging-drop method, the agitation-based approach and the liquid overlay technique [9, 10, 11, 18]. This later method is one of the simplest, reproducible and cheapest methods to produce spheroids, and was chosen in the present study for these reasons [19, 20]. In this study, we report the generation of 3D spheroids from three different human melanoma cell-lines (WM115, WM266-4 and A375) using the liquid overlay technique, with the purpose of determining how the growth of cells in 3D conditions affect their sensitivity towards two different types of anti-cancer agents: temozolomide and vemurafenib. To assess the effect of the anti-cancer drugs, we used the Alamar Blue assay that uses changes of oxidative state of resazurin as an indication of changes in metabolic activities. We discuss issues and trouble-shooting solutions throughout our study.

2. Materials and methods

2.1 Cell lines

The three human melanoma cell-lines, WM115, WM266.4 and A375 were purchased from ATCC (Manas, VA) and were grown in Dulbecco’s Modified Eagle’s Medium (DMEM, ATCC) supplemented with 10% FBS and penicillin (100 U/ml) and streptomycin (100 μg/ml). The absence of mycoplasma contamination in the cells prior to all experiments was confirmed by PCR. Note that cells in our laboratory are typically tested every 2 months for possible contamination with mycoplasma by PCR using the method described in Tang et al. [21]. For simplicity, WM226.4 cells will be referred to as WM266 throughout the rest of the manuscript.

2.2 Cell culture plates

For the generation of spheroids using the liquid overlay technique, cells are typically seeded in round-bottom 96-well plates. These plates are made of plastic polymer. Standard tissue culture plates are made from two types of plastic polymer: polystyrene or polypropylene [22]. Because these polymers are hydrophobic in nature, they need to be modified or “treated” by surface functionalization to become more hydrophilic, therefore allowing cells to adhere to the surface of the plates [22]. The generation of spheroids using round-bottom plates requires the use of non-tissue culture treated plastic plates, because cells should not adhere to the plastic but rather form multi-dimensional assemblies between them. However, it is still possible to use tissue-culture treated multi-well plates for the generation of spheroids. In this case, the surface of the wells should be coated with a polymer, such as poly 2-hydroxyethyl methacrylate (poly-HEMA), to prevent attachment of the cells to the plastic [23]. However, multi-well tissue culture plates with low adhesive properties designed for the generation of spheroids are now available from many suppliers.

In the present study, we compared the generation of spheroids from four different types of round-bottom 96-well plates provided by different manufacturers (Table 1). One type of plate was made of tissue-culture treated polystyrene (Plate 1). To prevent cell adhesion to this specific plastic surface, this plate was coated with poly-HEMA as described below. The three other plates (Plates 2, 3 and 4) were not tissue-culture treated. Two plates were made of polystyrene (Plates 1 and 2) and one plate of polypropylene (Plate 3). One type of plate (Plate 4) was specifically marketed for the generation of spheroids.

Number of the plate referred to in this paper	#1	#2	#3	#4
Round-bottom 96-well plate. Cat. #	#353917	#650185	#3879	#174929
Manufactu-rer	BD Falcon	Greiner Bio-One Cellstar	Corning	ThermoScientific
Polymer type	Polystyrene	Polystyrene	Polypropylene	Polystyrene; Nunclon-sphera treated*
Tissue-culture treatment	Yes	No	No	No
Poly-HEMA coating requirement	Yes	No	No	No

Table 1.

Characteristics of the round-bottom plates used for the generation of spheroids.

*

Nucleon-sphera treated plates have low adhesive properties following a non-disclosed proprietary process.

2.3 Poly-HEMA coating

The wells of the tissue-culture treated 96-well plate 1 (Table 1) were coated with poly-HEMA by adding 200 μl of 5 mg/ml poly-HEMA (Sigma-Aldrich, Saint-Louis, MO), prepared in 95% ethanol, to each well of a round bottom 96-well plate [23]. After addition of the poly-HEMA solution to the wells, the plates were left in a Biosafety cabinet for 72 h, allowing ethanol to evaporate completely.

2.4 Methylcellulose preparation

Methylcellulose was added to the cells to promote spheroid formation by acting as a cell repellent additive [24]. Methylcellulose is an inert polymer that forms a viscous and gel-like solution that promotes cell aggregation [25, 26, 27]. The methylcellulose solution was prepared after modification of the procedure described in Longati et al. [26]. Briefly, a 1.2% (w/v) methylcellulose stock solution was prepared by resuspending 6 g autoclaved methylcellulose (M1314, Spectrum) in 50 ml DMEM containing 10% FBS, penicillin (100 U/ml) and streptomycin (100 μg/ml). The solution was stirred overnight at 4° C and centrifuged for 2 h at 7500 RPM. After centrifugation, 20 ml of the clear viscous solution was added to 80 ml of DMEM containing 10% FBS, penicillin (100 U/ml) and streptomycin (100 μg/ml), resulting in a methylcellulose final concentration of 0.24%.

2.5 Cell seeding

Cells grown to 80–90% confluence in T-75 cell culture flasks were detached with 0.05% trypsin, centrifuged for 2 minutes at 500 g and resuspended in DMEM media supplemented with 10% FBS, penicillin (100 U/ml), streptomycin (100 μg /ml), and 0.24% methylcellulose. Cells were then counted using an hematocytometer and added to the wells of the four different 96 well-plates (Plates 1, 2, 3 and 4) at two different seeding densities: 2500 and 5000 cells per well. The wells were observed every day (0, 24, 48, 72, 96, 120, 144 and 168 h) by microscopy, images of spheroids were taken from 12 different wells, and the total number of formed spheroids per plate was counted after 8 days.

2.6 Resazurin assay

Resazurin or 7-hydroxy-10-oxidophenoxazin-10-ium-3-one is a water soluble and permeable dye [28]. After entering cells, resazurin is reduced by accepting electrons from multiple intracellular electrons donors such as NADPH, FMNH, NADH or cytochromes, without altering the normal transfer of electrons in cells [28]. Non-viable cells will not reduce resazurin. Resazurin is generally non-toxic for many cell-types, but cellular toxicity has been reported [29]. Oxidized resazurin has a dark blue color and has little intrinsic fluorescence, however, reduced resazurin or resofurin, is pink and highly fluorescent with measured excitation and emission fluorescence wavelengths typically of 540 nm and 590 nm, respectively.

Resazurin or Alamar Blue (AB) (Amresco/VWR/Avantor, Radnor, PA.) was prepared at 0.1 mg/ml in 50 mM Phosphate buffer pH 7.4 containing 150 mM NaCl (PBS), sterile filtered using a 0.2 um filter, and stored at 4° C in the absence of light. When added to cells, 10% (vol./vol.) resazurin was added to each well of the plate. Control wells contained resazurin and media only. When measuring changes in metabolic activity of cells grown in flat surface, the AB fluorescence signals were detected after 6 h incubation with the dye. However, after adding AB to the spheroids, we waited a 24 h incubation period before measuring the AB signal.

2.7 Staining of spheroids by immunofluorescence

Ten-days old spheroids generated from A375, WM115 and WM266 cells were fixed with 4% paraformaldehyde (PFA) for 30 minutes and then stained with 0.1% eosin Y (Sigma-Aldrich, Saint Louis, MO) in 80% ethanol to facilitate the visualization of spheroids in the paraffin block. The spheroids were then paraffin embedded using a Leica ASP300S processor. Briefly, the spheroids were washed with a series of ethanol solutions, starting with 95% ethanol followed by two washes with 100% ethanol. The spheroids were then cleared by two 100% xylene baths and finally embedded in paraffin blocks (Tissue-Tek® Base Mold 4162, 15 x 15 x 5 mm) using a Sakura Tissue-Tek embedding station. The paraffin blocks were sectioned at 5 μm using Thermo Microm 355S microtome. The sections were then floated on a water bath, placed on VWR VistaVision TM Histobond © charged slides and air dried. The spheroid sections were treated with three 100% xylene washes to remove the wax from slides and rehydrated through a series of 100–70% ethanol washed followed by a last wash in deionized (DI) water. An antigen retrieval step was then performed for 20 minutes at 120°C in citrate buffer (at pH 6.0) using a Retriever 2100 (Diagnostic Technology). For the staining procedure, the spheroid sections were first washed with TBS-T for 3 min and blocked for 1 h at room temperature (RT) with 5% NGS in TBS blocking buffer (Vector laboratories, Newark, CA). The sections were then incubated with a Ki67 rabbit polyclonal primary antibody (AbCam # ab15580, Waltham, MA) at a 1:500 dilution in 1% BSA overnight at 4°C. After 3 min washes with TBS-T, the sections were further incubated with CF 633 labeled Goat anti-rabbit secondary antibody (Biotium #20122, Fremon, CA) at a 1:250 dilution in 1% Bovine Serum Albumin (BSA) in TBS for an hour at room temperature. Finally, the sections were washed for 1 min with DI water and stained with 4,6-diamidino-2-phenylindole (DAPI) (Biotium #40043) before being mounted using the EverBrite™ Hardset Mounting Medium (Vector laboratories #23003). The slides were imaged using a Zeiss Axio Observer Z1 LSM 700 microscope using 20x 0.8NA lens. (Carl Zeiss Microscopy LLC, White Plains, NY).

In a different experiment, 72 h old spheroids were stained for the cell viability marker calcein acetoxymethyl (AM) ester (2 μM; Sigma-Aldrich), the dead cell marker Ethidium Homodimer-1 (EthD-1), (3 μM; Invitrogen/Thermo Fisher Scientific, Waltham, MA) and the nuclear marker Hoescht 33,342 (33 μM; Thermo Fisher Scientific) according to a procedure described in [30]. The spheroids were stained simultaneously with the three dyes in the 96 well plate for 30 min at 37°C. After incubation, the spheroids were carefully transferred to a glass bottom dish (Ibidi USA, Fitchburg, WI) for imaging with a Zeiss Axio Observer Z1 epifluorescence microscope equipped with a LSM 700 confocal scanning head, and using a 20×, 0.8 NA objective (Carl Zeiss, Thornwood, NY). Images were captured after excitation/emission of 494/517 nm for the detection of calcein-AM positive cells, 528/617 nm for the detection of EthD-1-positive cells and 360/461 nm for the detection of Hoescht 33,342 positive cells.

2.8 Statistical analysis of the data

Twelve spheroids were imaged for each cell line in each plate. Figure 2 shows only a representative image from one of these 12 wells. The diameter of 12 independent spheroids were measured each day, for a period of 8 days, using ImageJ software [31]. The mean and the standard deviation were used for the graphs shown in Figure 3. A student’s t-test was performed to indicate the significance between spheroids’ diameters at different time points with *p < 0.05; **p < 0.001 and ***p < 0.001. Figures 4 and 5 show representative images of stained spheroids. The cell-based assays were repeated at least three independent time. The data points from the titration with vemurafenib and temozolomide were fit using Kaleidagraph version 5.0 software (Synergy Software).

Figure 2.

Images of spheroids formed from A375 cells (A), WM115 cells (B) and WM266 cells (C) at different time points: 0, 24, 48, 72, 96, 120, 144 and 168 h scale bar: 250 μm.

Figure 3.

Diameters of spheroids formed from 2500 (A) and 5000 (B) A375 cells, from 2500 (C) and 5000 (D) WM115 cells, and from 2500 (E) and 5000 (F) WM266 cells at different time points (0 h, 24 h, 48 h, 72 h, 96 h, 120 h, 144 h and 168 h). ** p < 0.001; *** p < 0.001.

Figure 4.

Staining of spheroids formed from 2500 (A) and 5000 (B) A375 cells, 2500 (C) and 5000 (D) WM115 cells, and 2500 (E) and 5000 (F) WM266 cells with the proliferative marker Ki-67 (pink). The nuclei are stained in blue with Hoescht 33,342. Shown on the figure are the merged images. Scale bar: 50 μm.

Figure 5.

Staining of spheroids formed from 2500 (A) and 5000 (B) A375 cells, 2500 (C) and 5000 (D) WM115 cells, and 2500 (E) and 5000 (F) WM266 cells with calcein-AM (green) and EthD-1 (pink). The nuclei are stained in blue with Hoescht 33,342. Shown on the figure are the merged images. Scale bar: 50 μm.

3. Results and discussion

3.1 The morphology, shape and growth rate vary among cell lines

Our first goal was to determine if the three different cell lines could form spheroids using the scaffold-free liquid overlay technique. For each cell line, we tested the formation of spheroids in four commercial plates. These plates were made of polystyrene (Plates 1, 2 and 4) or polypropylene (Plate 3) and were either tissue-culture treated (Plate 1) or not (Plates 2, 3 and 4). In case of plate 1, to avoid cell adhesion to the plastic surface, the plate was coated with poly-HEMA prior to seeding the cells. The surface of plate 4 had already been pre-treated by the manufacturer using a non-disclosed proprietary process to enhance spheroid formation. Two different cell densities (2500 and 5000 cells) were seeded for each type of plate. The presence of spheroid was assessed every day for 7 days (0, 24, 48, 72, 96, 120, 144 and 168 h) by microscopy.

We first compared the yield of spheroids produced on the different plates for each cell line. We noticed that in our experimental conditions, a larger number of spheroids was generated using the plates P2, P3 and P4, whereas the poly-HEMA coated plate produced lower yields of spheroid for each of the three cell lines examined (Table 2).

Cell line	Cells/well	Spheroid yield
		Plate 1 (%)	Plate 2 (%)	Plate 3 (%)	Plate-4 (%)
A375	2500	68.3	100	100	100
	5000	48.1	100	100	100
WM115	2500	35	100	100	100
	5000	77.7	100	100	100
WM266	2500	78.3	100	100	100
	5000	54	100	100	100

Table 2.

Number of spheroids formed per number of examined wells (54 or 60). Poly-HEMA coated plate 1 produced the smallest yield of spheroids. All other plates generated spheroids in every well.

Observation of the spheroids by microscopy every day during the time course of 8 days (168 h), revealed that the spheroids generated with the different cell lines had different morphologies and shapes (Figure 2). Differences in the ability of cancer cells to form spheroids has been previously reported [32]. A375 cells did not produce spherical spheroids but rather irregular-shape spheroids with small assemblies of cells protruding in different directions. WM155 cells produced spheroids with the most regular shape (spheres) whereas WM266 cells generated more oval-shaped spheroids. WM115 cells produced the smallest size spheroids from all three cell lines (Figure 2). Note that the WM115 and WM266 cell lines originate from the same patient: whereas WM115 was established from a primary tumor, WM266 was established from a metastatic tumor [33]. Our observation suggests that the choice of the cell line for the generation of spheroid is an important aspect for assuring the success of the study. The size of regular-shape spheroids is more easily measured than that of irregular-shape spheroids. We indeed experienced issues when measuring the diameter of many A375 spheroids as well as some WM266 spheroids whose shapes were not spherical. We also noticed that it was very difficult observing spheroids generated in the P3 polypropylene plate. The plastic presented some opacity that hindered the microscopic observation of the spheroids. If a study aims to follow changes in spheroids’ size overtime by brightfield microscopy, these aspects are important to consider, as previously reported by other groups [34].

To assess the changes in spheroids’ size overtime, we measured the diameter of 12 spheroids every day for a period of 7 days, in plates 2, 3 and 4. Because the yield of spheroid formation was very variable from plate 1, we chose not to study the change in spheroids’ size overtime from this plate. The first measurement was performed 24 h after seeding the cells in the round-bottom wells. The average diameter for these 12 spheroids over time is presented in Figure 3. We observed similar growth changes in the size of the spheroids in all three plates. Overall, the size of the spheroids tended to decrease for the first 48 h to 72 h, before slowly increasing over the next 4 days. Observation of the spheroids by microscopy showed that immediately after seeding and centrifuging the cells in the wells, the spheroids initially consisted of an assembly of cells loosely connected. The space between cells seemed then to reduce within the first 3 days, resulting in an apparent shrinkage of the spheroids. The size of the spheroids then increased in the next 4 days, probably because of cell division (Figures 2 and 3). This observation was similar for each cell line in each of the three plates. For conciseness, we give below numerical data of the changes in diameter observed for each cell-line in plate 4. As mentioned above, the changes were similar in plates 2 and 3.

From all three cell lines, WM115 cells showed the largest initial size reduction between 24 h and 72 h, with a reduction of 37.5% and 36.1% in the average diameter for spheroids generated in Plate 4 with 2500 and 5000 cells respectively. Between 72 h and 168 h, these spheroids grew and increased their size by 15.5 and 9.5%, respectively. WM115 generated spheroids with the smallest size from the three cell lines examined. After reaching their lowest size at 72 h with a decrease of 5.3 and 13.9% for spheroids generated with 2500 and 5000 WM266 cells respectively, these spheroids later increased their size between 72 h and 168 h by 30.7 and 13.6% respectively (Plate 4). A375 spheroids also saw a decrease in their diameters between 24 h and 72 h 0f 10.9% and 16.3% for spheroids generated from 2500 and 5000 cells respectively, and an increase between 72 h and 168 h by 30.7% and 12.1% respectively. For all three cell lines examined, the growth increase between 72 h and 168 h was larger with 2500 cells than with 5000 cells (Figure 3). This suggests that cell seeding density is an important factor to consider if the study aims to use changes of spheroid size when investigating the effect of drugs on spheroid growth. Another important point to consider when performing a drug study is the choice of the age of the spheroids for initiating the drug study. As we just showed, the spheroids generated from the three cell lines all “shrank” in the first 72 h. Starting a drug study earlier than 72 h could hamper the analysis of the data, if one would aim to investigate changes in spheroid size as an effect of the drug. Initial changes of spheroid size have been recently explained by the effect of cell-cell adhesion forces that pull cells towards each other’s resulting in more compact structures [35]. In a different study, Thakuri et al. did not observe the initial size reduction in spheroids when dispensing dextran-respuspended HT-29 cells in a poly-ethylene glycol solution [36], suggesting that experimental conditions can greatly modulate the behavior of cells within spheroids.

3.2 Spheroids possess proliferative and metabolically active cells at their periphery and edges

We imaged 8–10 days old spheroids for the presence of proliferative cells using antibodies against the proliferative marker Ki-67. Ki-67 is a nuclear protein that is present in all stages of the cell cycle but is absent in quiescent cells and has been previously used for the staining of spheroids [37]. We first observed differences in abundancy in Ki-67 positive cells in the spheroids generated from the three different cell types. WM266 spheroids showed the highest amount of Ki-67 positive cells (Figure 4E and F). WM115 showed a lower percentage of Ki-67 positive cells, both with spheroids generated from 2500 and 5000 cells (Figure 4C and D). The presence of Ki-67 positive cells at the edges of the spheroid was clearly evidenced with the irregular-shape A375 spheroids generated from 5000 cells (Figure 4A and B).

To further evaluate the viability of cells within the spheroids, we stained 72 h old spheroids with calcein-AM, ethidium homodimers and Hoescht 33324 as described in [30, 38]. In these conditions, nuclei from both viable and dead cells are labeled with Hoescht 33342, the viable cells are labeled with calcein-AM (green) and the dead cells with ethidium homodimer-1 (pink). Imaging of the stained spheroids revealed that the calcein-AM positive cells were at the periphery of the spheroids (Figure 5). After the staining procedure and during the transfer of the spheroids from the round-bottom well of the 96 well plate to the imaging dish, many spheroids from A375 and WM115 cells broke, and only incomplete fragments could be imaged (Figure 5). However, these fragments do show that the metabolically active and calcein-AM positive cells are located at the edge of the spheroids in agreement with the observation of Ki-67 positive proliferative cells present at the edge of the spheroids as well (Figure 5). Furthermore, the non-fragmented WM266 spheroids also showed the presence of metabolically more active cells at the periphery of the spheroids. Staining of the spheroids with EthD-1 revealed only few dead cells (pink) in the 72 h old spheroids (Figure 5). The presence of the external layer of proliferative and metabolically active cells in the spheroids generated from all melanoma three cell lines (Figures 4 and 5) is in agreement with previous studies (reviewed in [39]).

3.3 Sensitivity of melanoma cells towards temozolomide and vemurafenib in 2D and spheroid 3D growth conditions

After identifying the optimal properties for the generation of the spheroids from the three different melanoma cell lines, our goal was to compare the sensitivity of the three melanoma cell types, grown in standard 2D growth conditions and as spheroids, towards two cancer drugs, temozolomide (TMZ) and vemurafenib. TMZ is an alkylating agent that alkylate guanine bases in DNA leading to DNA cross-linkages and impaired cell division, thus reducing melanoma cell growth [40].

When comparing the effect of TMZ on cells either grown in 2D conditions in cell culture flasks and as spheroids, we observed similar patterns of sensitivity. In general, A375, WM115 and WM266 all showed some resistance to TMZ as the highest concentration tested (1000 μM) only reduced the signal by 50% with WM115 cells (Figure 6D). For A375 and WM266, the signal observed at 1000 μM was only 37.5% and 30% lower than that in the absence of TMZ (Figure 6A and G). In our experimental conditions, we were not able to increase the TMZ concentration at higher levels due to the toxic effect of DMSO. For A375 and WM115, when comparing the effect of TMZ on the spheroids to that on cells, we observed that TMZ effected a smaller proportion of cells grown as spheroids than on flat surfaces (Figure 6B and C for A375 spheroids and E and F for WM115 spheroids). Only WM266 spheroids appeared to be slightly more sensitive towards TMZ with only 30% signal reduction by 1000 μM TMZ in cells and about 50% and 45% reduction by the same concentration of TMZ in the spheroids generated from 2500 and 5000 cells, respectively (Figure 6G, H and I). These results reflect the pattern and distribution of the proliferative marker Ki-67 in the A375, WM115 and WM266 spheroids. Indeed, because TMZ affects cell division, it is only effective in cells that divide, thus being most effective in proliferative cells, such as those staining positive for Ki-67. As observed in Figure 5, the spheroids generated from WM266 are those that possess the largest percentage of proliferative cells, from the three cell lines examined, and therefore, should be the spheroids that respond the most to TMZ. However, we also observed that a larger fraction of the WM266 cells responded to TMZ when assembled as spheroids than as monolayers (Figure 6G, H and I).

Figure 6.

Effect of temozolomide on: 2D grown A375 cells (A) and 3D A375 spheroids formed from 2500 (B) and 5000 (C) cells; 2D grown WM115 cells (D) and 3D WM115 spheroids formed from 2500 (E) and 5000 (F) cells; and 2D grown WM266 cells (G) and 3D WM266 spheroids formed from 2500 (H) and 5000 (I) cells.

The second drug that we tested in our study was vemurafenib. Vemurafenib is a BRAF kinase mutant inhibitor [41]. This small molecule drug only inhibits the mutant form of the kinase and is therefore only effective when the melanoma cells carry the appropriate BRAF V600E mutation. By blocking the BRAF kinase, vemurafenib inhibits the growth of the melanoma cells by inhibiting the BRAF/Map kinase pathway. All three cell lines used in our study carry the BRAF V600E mutation and were thus suitable for the study [33, 42].

We first observed that when testing vemurafenib on cells grown on 2D flat surfaces, WM115 showed the strongest IC50 for vemurafenib with a calculated IC50 = 0.057 ± 0.021. Based on the shape of the curve, it appeared that most cells in the dish were affected by vemurafenib (Figure 7D). This was not the case for A375 (Figure 7A) and WM266 (Figure 7G) where only a portion (less than 50%) of the cells seemed to be affected by the BRAF inhibitor. A similar effect of vemurafenib on only a portion of the cells was also observed in spheroids generated from WM115 and WM266 cells (Figure 7E, F, H and I). Interestingly, vemurafenib appeared to affect most of the cells within the A375 spheroids, with the fluorescence signal decreasing by up to 80% at the highest vemurafenib concentration used (Figure 7B and C).

Figure 7.

Effect of vemurafenib on: 2D grown A375 cells (A) and 3D A375 spheroids formed from 2500 (B) and 5000 (C) cells; 2D grown WM115 cells (D) and 3D WM115 spheroids formed from 2500 (E) and 5000 (F) cells; and 2D grown WM266 cells (G) and 3D WM266 spheroids formed from 2500 (H) and 5000 (I) cells.

Vemurafenib inhibits the BRAF/Map kinase pathway, and by doing so, affects cell proliferation [41]. Based on the larger population of proliferative cells in the WM266 spheroids (Figure 5E and F) when compared to the A375 and WM115 spheroids, it was expected that these spheroids would respond the best towards vemurafenib. However, we observed that for these spheroids, the fluorescence signal decreased by about 37 and 25% in the presence of the highest concentration of vemurafenib, for spheroids generated from 2500 and 5000 cells respectively. We are currently investigating the reasons of this discrepancy. As future studies, it would be particularly interesting to stain the vemurafenib treated spheroids with Ki-67, as well as with calcein-AM and ETHD-1 to identify the regions of the spheroids affected by the drug.

Our study showed that the response of the cancer cells towards the two cancer agents TMZ and vemurafenib was affected by the architecture of the cells (3D spheroids versus 2D monolayers), but that the response to the drugs depended on both the cell type and the drug itself. Indeed, although the A375 cells appeared to be more resistant towards TMZ when present as 3D spheroids than as 2D monolayers (Figure 6A–C), they were more sensitive to vemurafenib when present as 3D spheroids than as monolayers (Figure 7A–C). In an earlier study, Filipiak-Duliban et al. had also shown that the drug sensitivity of cancer cells was affected by the mode of culture of the cells [43]. In their study, the authors showed that the B16F10 murine melanoma cells were more resistant to the three anti-cancer drugs everolimus, doxorubicin and cisplatin when grown as spheroids than as 2D monolayers [43]. However, the authors also showed that the effect of the 3D structure on drug sensitivity was not observed when studying a renal cancer cell line [43]. These data, in agreement with our own observations, strongly suggest that the effect of the 3D structure of the cell assemblies depends on both the cell type and the drug that is studied. In addition, the size of the spheroids has also been shown to affect drug resistance, with spheroids of larger sizes being in general more resistant than spheroids of smaller sizes [44]. In our study, we also observed that spheroids generated from 5000 WM115 cells were more resistant towards TMZ than spheroids generated from 2500 cells. Similarly, Spheroids generated from 5000 WM266 cells were more resistant to vemurafenib than spheroids generated from 2500 cells. However, the size of the spheroids did not seem to affect the sensitivity of WM115 cells towards vemurafenib, or the sensitivity of WM266 cells towards TMZ, making the relationship between each cell line and each drug unique.

4. Conclusions and perspectives

Research performed in the last decade has clearly demonstrated that cells grown as 3D spheroids represent a more accurate model of tumors than cancer cells grown as monolayers in cell culture flasks [11, 39]. As shown in our study, the spheroids generated from the three human melanoma cells all showed features of tumors. For instance, all three types of spheroids exhibited a layer of proliferative and metabolically active cells at the edge of the spheroids, as shown by their positive staining with Ki67 and calcein-AM. However, we are also aware that our spheroid-based model presents weaknesses and needs to be improved to better mimic tumors. One of the weaknesses is that our spheroids were generated from cancer cells only. Tumors are formed not only from cancer cells, but also from other cell types such as fibroblasts, macrophages, pericytes and endothelial cells. All these cells communicate with each other’s in tumors [45]. Therefore, in tumors, responses to anti-cancer drugs can be very different than the response to the same drugs in spheroids generated from a single type of cells. To solve this issue, new generation of multicellular spheroids are being developed and studied [10, 46, 47, 48, 49]. For example, Klicks et al. recently generated spheroids composed of a mixture of fibroblasts, keratinocytes and melanoma cells [48]. However, challenges remain when using multicellular spheroids for the study of anticancer drugs. One of those challenges reside in the generation of consistent spheroids of homogeneous size and shape [50, 51, 52]. As we have shown in our study, certain cell lines such as A375, generate irregular shape spheroids that are difficult to study. In addition, in an effort to better mimic the complexity of tumors’ architecture and cellular heterogeneity, spheroids are now combined with bioprinting technologies, resulting in additional challenges due to, among other factors, the mechanical instability of spheroids [53, 54]. However, with the fast advancement of new technologies, these challenges will be surely overcome in a near future [55, 56, 57, 58, 59].

Acknowledgments

We would like to thank Virginia Montgomery from the Advanced Imaging Microscopy (AIM) Core Facility at NDSU for her assistance in staining the spheroids. This study was supported by NIH-NIGMS Award number U54GM128729 (DaCCoTA Scholar Award to E.L.), the National Science Foundation under NSF EPSCoR Track-1 Cooperative Agreement OIA#1946202 and the College of Health Professions at NDSU.

Conflict of interest

“The authors declare no conflict of interest.”