Selection of Effective Therapies Using Three-Dimensional in vitro Modeling of Chondrosarcoma

Purpose: Chondrosarcomas are a group of cartilaginous malignant neoplasms characterized by the deposition of chondrogenic extracellular matrix. Surgical resection is currently the only curative treatment option, due to their high resistance to conventional chemotherapy and radiotherapy. Novel therapeutic treatment options may improve outcome. Predominantly used cell line monolayer in vitro models lack in vivo complexity, such as the presence of extracellular matrix, and differing oxygen access. Hence, we aimed to improve pre-clinical chondrosarcoma research by developing an alginate-based 3D cell culture model. Method: An alginate scaffold was applied to generate spheroids of three chondrosarcoma cell lines (CH2879, JJ012, SW1353). Morphological, histological and immunohistochemical assessment of the spheroids were used to characterize the chondrosarcoma model. Presto blue assay, morphological and immunohistochemical assessment were applied to assess spheroid response to a panel of chemotherapeutics and targeted therapies, which was compared to conventional 2D monolayer models. Synergistic effect of doxorubicin and ABT-737 (Bcl-2 inhibitor) was compared between monolayer and spheroid models using excess over Bliss. A 3D colony formation assay was developed for assessment of radiotherapy response. Results: Chondrosarcoma spheroids produced chondrogenic matrix and remained proliferative after 2 weeks of culture. When treated with chemotherapeutics, the spheroids were more resistant than their monolayer counterparts, in line with animal models and clinical data. Moreover, for sapanisertib (mTOR inhibitor) treatment, a recovery in chondrosarcoma growth, previously observed in mice models, was also observed using long-term treatment. Morphological assessment was useful in the case of YM-155 (survivin inhibitor) treatment where a fraction of the spheroids underwent cell death, however a large fraction remained proliferative and unaffected. Synergy was less pronounced in 3D compared to 2D. A 3D clonogenic assay confirmed increased resistance to radiotherapy in 3D chondrosarcoma spheroids. Conclusion: We demonstrate that the chondrosarcoma alginate spheroid model is more representative of chondrosarcoma in vivo and should be used instead of the monolayer model for therapy testing. Improved selection at in vitro stage of therapeutic testing will increase the amount of information available for experimental design of in vivo animal testing and later, clinical stages. This can potentially lead to increased likelihood of approval and success at clinical trials.


INTRODUCTION
Chondrosarcomas are a group of malignant cartilaginous neoplasms, which account for around 20% of bone sarcomas, making them the second most common primary bone malignancy after osteosarcoma (Hogendoorn et al., 2013). The hallmark of these tumors is the deposition of extracellular matrix rich in glycosaminoglycans and collagen II. Likeliness of metastasis is best predicted by the histological grade, as an atypical cartilaginous tumor/chondrosarcoma grade I is locally aggressive but rarely metastasizes while grade III chondrosarcomas have a high metastatic rate (32-71%) and poor 10-year survival rate (26-55%) (Angelini et al., 2012;Fromm et al., 2018;van Praag et al., 2018;Bovée et al., 2020). Conventional chemotherapy and radiotherapy have very limited efficacy in patients with advanced chondrosarcoma and surgical resection is currently the only curative treatment option (Angelini et al., 2012;Van Maldegem et al., 2014). Presently there is a large variety in the treatments used in order to attempt to improve survival (van Maldegem et al., 2019). A fraction of patients are managed with drugs which have shown activity in other soft-tissue and bone sarcomas, without any previous pre-clinical information on their efficacy in chondrosarcoma (Italiano et al., 2013). Its rareness, and therefore the lack of interest from industry, hamper the design and conduct of clinical trials for these patients (Miwa et al., 2019). There is an urgent need for more representative pre-clinical chondrosarcoma models to select therapeutic strategies that might be beneficial for patients.
However, the limitation of the use of cell lines in standard 2D culture is their restricted representability, as they lack the spatial organization, architecture and extracellular matrix that is characteristic of chondrosarcoma in vivo (Bovée et al., 2020). Moreover, pathway activity is altered during culture, for instance we previously showed upregulation of IGF1R expression in CH2879 cell line in 2D culture, while this was not seen in threedimensional (3D) cultures, nor in the corresponding primary tumor (Peterse et al., 2016). Therefore, cell lines in 2D might not respond to therapy in the way cells do in vivo.
A common approach to improve the representability at the in vitro testing stage is to create 3D cell culture models. These mimic the natural organization of tissues more closely, which is especially relevant for tumors in which the spatial organization, architecture and interaction with the extracellular matrix are critical. These include chondrosarcoma, where the extracellular matrix is the defining factor, and by some even hypothesized to be involved in chemotherapy resistance (Bovée et al., 2010). Many different methods exist for the creation of 3D cell cultures of sarcoma including attachment prevention, embedded scaffold and bioreactor techniques. Each has advantages and disadvantages for sarcoma research (Colella et al., 2018).
To develop a 3D spheroid culture model for chondrosarcoma, we chose to use a biocompatible, plant derived scaffold, alginate hydrogel (Lee and Mooney, 2012;Lhuissier et al., 2017). Alginate hydrogel polymerises rapidly in the presence of divalent ions, such as Ca 2+ , therefore can be used for entrapment of single cells, which subsequently grow out into clonal spheroids. As the cells are embedded within a scaffold system, they receive physiological mechanical strain, are able to grow in three dimensions and deposit extracellular matrix (ECM) which will then remain within their surroundings. Alginate scaffolds are also suitable for drug treatment of cultures, as the pore size achieved is typically 5-200 nm and small soluble molecules are able to diffuse freely (Ching et al., 2017).
In this study, alginate chondrosarcoma spheroids were shown to produce extracellular matrix, similar to tumors in vivo, and were used for short and long-term drug testing, including assessment of synergy. Drug response within the model was monitored using cell viability assays as well as histological analysis and the model was additionally used to evaluate radiation sensitivity using a 3D colony formation assay.

3D Cell Culture
Following expansion in a monolayer, cell lines were suspended in 1.2% (w/v) medium viscosity alginic acid (Sigma-Aldrich, Darmstadt, Germany) in 0.15 M NaCl at 1 × 10 6 cell/mL density to generate spheroids (∼14,140 cells/bead). Concentrations of around 1% w/v alginate were deemed optimal for cartilageorientated in vitro and in vivo applications by several groups (Heiligenstein et al., 2011;Rey-Rico et al., 2016). Alginate beads were formed by dropping the cellular suspension in 1.2% (w/v) alginate/0.15 M NaCl (Sigma-Aldrich) through a 19-gauge needle into 0.2 M CaCl 2 solution (Sigma-Aldrich). After incubation at 37 • C for 10 min, alginate beads were washed twice with 0.15 M NaCl and washed twice in DMEM/F12 (Life Technologies Limited) complete media before being placed in RPMI medium overnight. The medium was replaced to DMEM/F12 medium where possible due to the alginate breakdown properties of phosphate buffer within RPMI medium (Lee and Palsson, 1990). The transition from RPMI to DMEM culture was made on the second day of culture in 3D to minimize transitioning stress on the cells. Preliminary tests were performed to determine the best time to transition the medium; before, during or after 3D cell culture establishment. Changing medium to DMEM 24 h after 3D cell culture was found to be the least disturbing for cell growth. The cells were cultured in DMEM/F12 for a maximum of 17 days.

Histological Staining
Day 3, 7, and 14 spheroids were fixed using 10% neutral buffered formalin, dehydrated and paraffin embedded. Sections were produced (4 µm) and, after deparaffinisation and rehydration, were stained with hematoxylin and eosin (H&E), using the Leica ST5020-CV5030 Stainer Integrated Workstation (Leica Biosystems, Amsterdam, The Netherlands), for morphological assessment by an expert bone tumor pathologist (JVMGB). Serial sections were stained with Richardson's toluidine blue or alcian blue solution, for assessment of glycosaminoglycan production; slides were incubated in boiling 10 mM citrate buffer (pH 6) for 10 minutes in order to remove the alginate scaffold, which is stained very strongly with both stains. After cooling and washing in dH 2 O, sections were incubated in toluidine blue solution (1:32 diluted in dH 2 0) for 5 min or incubated (30 min) in alcian blue solution (pH 2.5, RT). Excess dye on toluidine blue stained samples was removed in dH 2 O and the slides were dried at 58 • C for 30 min or until fully dry, before mounting. Alcian blue stained samples were washed in running tap water (2 min), rinsed in dH 2 O, counterstained with nuclear fast red (5 min), dehydrated, cleared and mounted. SJSA-1 cell line spheroids were stained as a negative control. For comparison to 2D, cell lines cultured in 2D were trypsinised, formalin fixed (30 min) and pre-embedded using the cytoblock system (7401150, Thermo Scientific, USA). Then the samples were dehydrated and paraffin embedded for sectioning. For comparison to chondrosarcoma in vivo, three high grade chondrosarcomas were retrieved from the archives of the Department of Pathology, Leiden University Medical Center, Leiden, the Netherlands. Tumor samples were fully anonymized according to the ethical guidelines described in "Code for Proper Secondary Use of Human Tissue in The Netherlands" of the Dutch Federation of Medical Scientific Societies.
The scanned IHC samples were scored using QuPath opensource software (Bankhead et al., 2017). All scoring was performed on a representative area spanning at least three biological replicates (beads). Scoring of Ki-67 was performed using the software positive cell detection tool to determine the percentage of positive nuclei among the total number of nuclei within an area larger than 800,000 µm 2 . Scoring of activated caspase-3 staining was performed by manual counting of percentage of positive cells within an area larger than 600,000 µm 2 . Scoring of IGF1R was performed as previously published for 2D cells and tissue, based on overall intensity within an area larger than 700,000 µm 2 (Peterse et al., 2016). Scoring of collagen II and pS6 was performed by scoring slides by percentage of positive spheroids within an area larger than 900,000 µm 2 . In this case a spheroid was termed positive if at least one area within the spheroid was calculated positive using the software positive cell detection tool. Setting the software detection threshold ensured reduction in observation variability. Samples were collected over at least three separate experiments and areas counted were from at least three replicates.

Assessment of Response to Antineoplastic Agents
Dose-response curves were generated for the three cell lines in 2D as well as 3D, keeping within 15 passages throughout the experiment. After trypsinisation, 2D cultured cells were seeded in 96-well plates at 7,000 (CH2879) or 3,000 (JJ012, SW1353) cells per well and left to adhere for 24 h. Treatment duration for 2D cultures was 72 h. In the case of 3D cell cultures, day 14 spheroids were used for 72 h treatment measurements and day 1 spheroids were used for 14 day treatments, including a measurement at 8 days (after 7 days of treatment) and a second dose at day 8.
After treatment, the medium was removed, wells were washed with medium and cell viability was determined using the PrestoBlue viability reagent (Invitrogen, Life-Technologies, Scotland, UK) according to the manufacturer's instructions. Plates were then incubated at 37 • C for 1 h for monolayer cultures, one and a half hours for CH2879 and JJ012 spheroid cultures and 3 h for SW1353 spheroid cultures. Fluorescence was measured with an excitation/emission wavelength of 560/590 nm Victor3V 1420 multilabel counter (Perkin Elmer, Hamburg, Germany). Selected treated samples were processed for sectioning and histological and IHC staining.
Histological staining was also performed on select treated cultures. These were stained with H&E and the colony number was counted within a representative area spanning at least three biological replicates (beads).
The drug concentration which inhibited cell viability by 50% (IC 50 ) was determined using Prism software version 7.02 after normalization to untreated control samples. A Grubbs' test was used to remove outliers from the dataset. Experiments were performed at least three times in triplicate. A Welch's ANOVA test was used to determine significant differences between the IC 50 values obtained from 2D and 3D cultures. A post-hoc Dunnett's T3 multiple comparisons test was then performed for comparison of individual cell line results in 2D and 3D.

Assessment of Drug Synergy
Increasing concentrations of doxorubicin were used to treat the chondrosarcoma spheroid cultures in combination with 3.16 µM ABT-737 as this drug combination has demonstrated synergy in previously published monolayer models (van Oosterwijk et al., 2012b). The Bliss independence model (C = A + B -A × B) was used to predict synergy, in which C represents the combined effect and A and B represent single agent effects (Borisy et al., 2003).

Assessment of Radiation Therapy Response
One day old spheroids were used for radiation treatment experiments. Five alginate beads per well were transferred to a 12-well plate. Increasing doses (0-6 Gy) of γ-radiation, using a 137 C source (YXLON, Comet Technologies, Shelton, USA), were applied to culture plates containing cell seeded alginate beads. These were cultured for 14 days, with a medium refresh at 7 days, formalin fixed using 10% formalin, dehydrated and paraffin embedded. The samples were then sectioned, stained using H&E, and the spheroid colony number and area, within an area of 3 mm 2 , was assessed using QuPath open source software (Bankhead et al., 2017). All counts were performed on a representative area spanning at least three biological replicates (beads).

Chondrosarcoma Cells Cultured Within an Alginate Model Display a Phenotype Closer to That of Chondrosarcoma in vivo
All cell lines cultured in alginate beads formed chondrosarcoma spheroids ranging from 50 to 156 µm diameter by 14 days of culture with an average diameter of 76 µm. All spheroids displayed extracellular matrix deposition, as confirmed by toluidine blue and alcian blue staining (Figure 1A), while the osteosarcoma cell line (SJSA-1) spheroid that served as control was negative as well as the cell lines cultured in 2D (Supplementary Figures 1, 2). Glycosaminoglycans were deposited throughout the spheroids and concentrated around the edges. Thus, in the alginate model, chondrosarcoma cells show some differentiation, as they produce extracellular matrix, similar to chondrosarcoma in vivo (Supplementary Figure 1). Collagen II was also expressed in the extracellular matrix and was stable or increased over time (Figures 1B,C). These features are hallmarks of chondrosarcoma but known to be absent in 2D culture (Scully et al., 2000;Gil-Benso et al., 2003;Gebauer et al., 2005) which we confirmed as the three cell lines cultured in 2D lacked glycosaminoglycans and collagen II expression, while expression was abundant in the three high grade chondrosarcoma tissue specimens (Supplementary Figure 1).
CH2879 exhibited high proliferative activity as shown by Ki-67 immunohistochemistry at 14 days of 3D culture (50%), whereas a smaller portion of JJ012 and SW1353 cultures was proliferative at this stage (8 and 11%) (Figures 2A,B). In comparison, the average proliferative activity (22.9%) was more similar to that of chondrosarcoma tissue (25.9%) than 2D cultured cell lines (82.7%) (Supplementary Figure 3). Cleaved caspase-3 levels in all three cultures were relatively high (total average = 43.9%), indicating a high level of apoptosis within the 3D cultures. For comparison, the apoptotic levels in chondrosarcoma tissue were higher than in 2D cell cultures (3.2% compared to 0.3% average) but not at similar levels to our model  Figure 3). Since we previously reported IGF1R expression to be artificially upregulated in chondrosarcoma cell lines, being low to absent in chondrosarcoma primary tumor tissue but positive in 2D cultured cell lines (Peterse et al., 2016), we assessed IGF1R in the spheroids. IGF1R expression was negative in all samples ( Figure 2B), with an adequate external positive control (Supplementary Figure 4).

Treatment Response at the Histological Level Is Highly Variable
Additionally to the global cell viability dose response curve data, histological analysis of treated spheroids revealed further information on different morphological effects on the spheroids. CH2879 spheroids are displayed as an example of drug effect (Figure 4). YM-155 treatment showed a reduction in number and clear loss of viability in smaller diameter spheroids, whereas larger spheroids remained proliferative (Figure 4,  Supplementary Figure 5). On the other hand, cisplatin treated spheroids showed a more global chemotherapy effect, with an overall decrease in spheroid size and number. The majority of ABT-737 spheroids were showing signs of apoptosis (shrunken nuclei and cleaved caspase-3 positivity) (Figure 4).

CS Spheroid Model Can Be Used to Evaluate Synergy
Chondrosarcoma spheroids were also utilized for prediction of synergy. As demonstrated by combining doxorubicin and ABT-737 the excess over Bliss values, indicating synergy, also differ between the monolayer and spheroid cultures (Figure 5). The combination of Bcl-2 inhibition using ABT-737 and doxorubicin was previously shown to reduce cell viability in vitro by inducing apoptosis and overcoming chemoresistance (van Oosterwijk et al., 2011). Synergy is less pronounced in CH2879 spheroids compared to 2D (Excess over Bliss= 22 vs. 13%) and combination was only additive for JJ012 (Excess over Bliss = 5 vs. 0%) and SW1353 (Excess over Bliss = 23 vs. 0%) spheroids ( Figure 5C). Although around 10% of total spheroids were still viable, as indicated by presto blue assay, even at 1 µM doxorubicin; histological analysis showed a higher amount of cell death in remaining spheroids for all three cell lines compared to singleagent options (Figures 3A,D, 5A,B).

CS Spheroid Model Can Be Used for Long-Term Treatment
Sapanisertib, an mTOR inhibitor, was initially used for 3 day treatment of mature 14 day old spheroids. All three cell lines were sensitive to the treatment, similar to previously published 2D results (Table 1) ( Figure 3F) (Addie et al., 2019). For longterm treatment, two doses of sapanisertib were given overall throughout spheroid growth, one at day one and one at day eight. Cell viability response measured at 7 day and 14 day treatment with sapanisertib did not significantly differ from response to 3 day treatment of mature spheroids for any of the three cell lines tested (Figure 6A), however a trend toward higher resistance to sapanisertib in two out of three mature cultures was observed ( Figure 6B). These results show that our model is suitable for up to 14 days of treatment, but that in case of sapanisertib, increase in treatment length and number of doses do not lead to lower cell viability. CH2879 fold change in cell viability at 10 nM sapanisertib treatment shows an initial delay in growth at 1 week of treatment compared to control culture followed by an increase in growth rate by 2 weeks, which may suggest outgrowth of a resistant cell population ( Figure 6C). Similar trends were observed with the other two cell lines (Supplementary Figure 6). This was confirmed by histological analysis which indicated presence of unaffected, slightly more proliferative spheroids in the treated group (65.63% Ki-67 positive cells compared to 52.32%) ( Figure 6D). The remaining spheroids were also shown to have normal expression of pS6, the activated form of a kinase downstream of mTOR, after mTOR inhibition ( Figure 6D).

CS Spheroid Model Allows for Direct Assessment of Radiation Response
Colony formation assays are typically employed to observe sensitivity to radiotherapy (Franken et al., 2006). This method is dependent on the cell clonogenic capacity, which is reduced after irradiation. As the spheroids produced using our method are clonal and cells cannot migrate freely within the scaffold, we adapted the colony formation assay for 3D. Significant changes in the surviving fraction after 2Gy (SF2) treatment are considered indicative of the in vivo response to radiotherapy (Fertil and Malaise, 1981). CH2879 spheroids appear to be the most sensitive to radiotherapy (SF2 = 0.58) and JJ012 and SW1353 spheroids only have a small reduction in clonogenic capacity (SF2 = 0.83 and 0.95, respectively) ( Figure 7B).The number of spheroids as well as their size is reduced with increased radiation dose for all three chondrosarcoma cell lines, with CH2879 shown as a histological example (Figures 7A,B). Size of colonies as well as surviving fraction is reduced with increased irradiation (Figure 7C).

DISCUSSION
Although cell lines cultured in monolayer are easy to work with, they do not adequately mimic the morphology and the behavior of tumor cells in patients (Horvath et al., 2016). For rare cancers, representative in vitro models are especially needed in order to adequately predict which compounds could provide patient benefit, as organizing clinical trials for rare cancer patients is challenging. In this paper we developed a more representative model for conventional chondrosarcoma of bone.
To date, chondrosarcoma cell lines have been cultured in 3D using non-adherent surfaces (van Oosterwijk et al., 2012b), the hanging drop method (Perut et al., 2018), collagen scaffolds (Truong et al., 2012;Hamdi et al., 2015) and alginate beads (Lhuissier et al., 2017). The attachment prevention techniques used succeeded in formation of large regular multicellular tumor spheroids (MCTS) with corresponding representative regions of quiescence, hypoxia and necrosis, whereas cells within the collagen gel model migrated throughout the gel and oxygen levels were regulated externally in order to mimic in vivo hypoxic circumstances (van Oosterwijk et al., 2012b;Hamdi et al., 2015). Use of scaffolds such as alginate, containing a mechanical strain component but where migration is limited, may be beneficial for the 3D culture of sarcoma. Morphological features similar to high grade chondrosarcoma could be observed from H&E stained spheroids and extracellular matrix components were highlighted using histological and immunohistochemical means, allowing for a suitable comparison to observations in patients. The straightforward application of methods conventionally used to assess patient tissue samples is a major advantage of 3D culture over 2D models. The development of 3D cultures for mesenchymal tumors requires a different approach as compared to the more frequent epithelial cancers, as many sarcomas are characterized by the deposition of extracellular matrix, including collagen, hyaluronan and glycoproteins. Thus, 3D sarcoma models should facilitate matrix deposition by the tumor cells, over a short culture period. Presence of glycosaminoglycan and collagen II in all chondrosarcoma cell line spheroids confirmed the deposition of a chondroid matrix, which increases model similarity to in vivo (Aigner et al., 1997). Moreover, levels of IGF1R, that we previously showed to be artificially upregulated during culture in 2D, were negative in the spheroids, also indicating a higher similarity to chondrosarcoma in vivo (Peterse et al., 2016). Culture induced upregulation of genes or pathways will affect drug response, and use of the spheroid model will reduce the number of artificially upregulated targets.
Of note, there was a high level of apoptosis observed within the spheroids, which may be a reflection of tissue remodeling related programmed cell death, which is normal for a large cell mass. This was also reported by others (Lhuissier et al., 2017). The alginate scaffold is highly biocompatible and is therefore unlikely to be the cause of apoptosis, as it has been previously used for many cell-based applications (Andersen et al., 2015). Moreover, cells are unable to migrate through the alginate, and dead cells or debris would remain trapped. All of the cultures maintained proliferation at 14 days of culture at similar levels to in vivo on average, indicating a sufficient continuation of cell growth during the therapeutic treatment period, and we showed that long-term treatment is feasible.
Overall, response to chemo-and radiation therapy in 3D chondrosarcoma spheroids were more in line with results in vivo as compared to monolayer (Table 2). Generally, higher resistance to classical chemotherapeutics, such as doxorubicin, cisplatin and temozolomide was observed. These results are in line with current in vivo and clinical data ( Table 2) as conventional chemotherapy is not effective as single-agent treatment for chondrosarcoma (Skubitz, 2003;Italiano et al., 2013;van Maldegem et al., 2019). The increased resistance in 3D might be explained by the presence of matrix components, pH and oxygen gradients as well as lowered exposure to therapy within spheroid cultures compared to monolayer, where all cells are equally fully exposed. There were also some exceptions where response in 3D and monolayer did not significantly differ, such as doxorubicin treated CH2879.
We cannot exclude that the different growth rates of the various cell lines affect response in the spheroid model. For example, doxorubicin is known to preferentially affect proliferating cells (Yang et al., 2014), therefore this could be a contributing factor to the maintained sensitivity of CH2879 spheroids to this drug, since the amount of proliferation observed by immunohistochemistry was higher in this cell line. However, overall CH2879 growth, observed by presto blue assay, was comparable to JJ012 during the 72 h treatment period, yet the JJ012 spheroids were notably more resistant to doxorubicin than monolayer cultures (2D/3D 49.6). Therefore, growth rate may not be the only contributing factor.
Growth rate dependent variation between cell lines in 2D is commonly corrected for by using a metric such as the GR 50 . This metric relies on the assumption that the cells are in exponential growth before and during application of the drug, otherwise a reliable value cannot be achieved (Brooks et al., 2019). Applying these metrics to 3D cell culture models is more difficult, since these tend to have significantly slower growth rates. Therefore, in 3D, when cell growth rate is not high enough for GR 50 calculation, differences between cell lines should still be considered separately, when observing the data by comparing the viability of control samples before and after the treatment period.
An advantage of the 3D alginate spheroid model is that in addition to monitoring cell viability, it enables morphological and immunohistochemical assessment after treatment. We previously used an siRNA screen and identified survivin as FIGURE 6 | Long-term treatment. (A) There is no significant difference in the viability of cell lines CH2879, JJ012 and SW1353 cultured in 3D if they receive 3 days of sapanisertib treatment as mature spheroids or treated over their growth period for 7 days or 14 days (doses at days one and eight). (n ≥ 3) (B) There is no significant difference in the IC50 values between mature spheroid 3 day treatment or treatment over the growth period. (C) Growth rate of CH2879 spheroids treated with sapanisertib (10 nM) increased over 2 weeks. (n = 4) (D) No significant change in colony morphology or number was observed after 10 nM treatment. Cells after 2 weeks of sapanisertib treatment were highly proliferative while apoptosis was absent and expression of pS6 was uninhibited, suggesting outgrowth of resistant colonies over time. Scale bar = 20 µm, cell line = CH2879. All results were normalized to DMSO control as applicable.
a potential target for treatment in chondrosarcoma (de Jong et al., 2016b). Here our previous results are confirmed as the survivin inhibitor YM-155 also reduced cell viability in the chondrosarcoma spheroids. However, while 2D results suggested a relation between p53 mutation status and response, this was not confirmed in the 3D spheroid model. The lack of a complete reduction in viability could also be due to reduced exposure of the cells to YM-155. Histological evaluation revealed that the spheroids which remained viable and proliferative were larger in size. Post-treatment spheroid colony number and size evaluation showed a reduced presence of smaller spheroids after treatment and no visible difference to larger spheroids. An increase in apoptosis was not observed after treatment with YM-155, which was consistent with previous data where we showed that YM-155 affected cell cycle progression. Previously published data on YM-155 sensitivity of a large panel of cancer cell lines indicated that resistance may also be associated with low SLC35F2 expression, therefore before moving onto in vivo testing expression of this efficacy biomarker could be studied within chondrosarcoma spheroids (Winter et al., 2014).
The anti-apoptotic Bcl-2 family members were earlier shown to be overexpressed and to play an important role in chemo resistance of chondrosarcoma (de Jong et al., 2018). Restoration of the apoptotic machinery, by inhibiting the Bcl-2 family members using ABT-737, induced apoptosis and reverted the resistance of chondrosarcoma cells to doxorubicin (van Oosterwijk et al., 2012b). In the present study, ABT-737 reduced cell viability and induced apoptosis in a majority of chondrosarcoma spheroids, although a limitation of this model is the presence of a moderate apoptotic signal in control samples. However, the synergy between doxorubicin and ABT-737 observed in monolayer cultures could not be confirmed in spheroids and IC 50 values were increased in comparison to doxorubicin alone. By adjusting dosage amount and times, the synergistic effect could be improved. Therefore, in future experiments, synergistic effects should also be tested within spheroid models for a more representative result and refined dosing schedules. More resistant* (Lhuissier et al., 2017) Not performed Considered resistant. Used in combination with doxorubicin, never as single agent. Increased PFS associated with combination treatment (n = ?/21 † , n = 4) (Italiano et al., 2013;van Maldegem et al., 2019) Combination of cisplatin with FT 2102 recruiting (NCT03684811)
Clinical trials, past results and ongoing studies, are also mentioned within the summary when applicable. *Current paper. † Unknown number of conventional CS subtype within cohort.
A previously performed metabolic compound screen revealed an mTOR inhibitor, sapanisertib, as an effective anti-proliferative agent for chondrosarcoma. Reduction of oxidative and glycolytic metabolism and decreased proliferation was observed with treatment as well as a 5 day delay of tumor growth in a chondrosarcoma xenograft model (Addie et al., 2019). For comparison with xenograft results, a protocol for long-term drug treatment in spheroids was established. Reduced growth rate was observed between 1 and 7 days of treatment with 10 nM sapanisertib, in comparison to untreated controls, which then increased between 7 and 14 days in all cell lines, in line with in vivo data. A phosphokinase downstream of mTOR, pS6, was still expressed in spheroids after treatment, similarly to the xenograft model, suggesting the presence of a resistance mechanism. Previously published monolayer cultures, conversely, demonstrated inhibited pS6 expression after treatment with sapanisertib, which indicates that, in the context of the mTOR pathway, the spheroid model was more representative of in vivo results. Subsequently, mTOR inhibition would have a higher effectiveness as part of a combination treatment which would eliminate the remaining cell population (Boehme et al., 2018). In clinic, no significant single-agent mTOR treatment effect has been observed in chondrosarcoma patients thus far. A phase II study using ridaforolimus, another mTOR inhibitor, as a single agent did not have a noted effect on chondrosarcoma patients in the cohort (Chawla et al., 2018). However, prolonged stable disease was achieved for chondrosarcoma patients by combining mTOR inhibitors with liposomal doxorubicin or EGFR, IGF1R, VEGF inhibitors (Katz et al., 2016;Liu et al., 2016;Trucco et al., 2018;Vlahovic et al., 2018).
High hopes for effective therapy came with the identification of specific driver mutations in the isocitrate dehydrogenase genes IDH1 and IDH2, and the development of a specific inhibitor of mutant IDH1, such as AGI-5198. However, in vitro studies showed loss of clonogenic capacity only after long-term treatment with high concentrations (20 µM), while no effect was observed at lower concentrations or shorter treatment times (Li et al., 2015;Suijker et al., 2015). The spheroid model also did not show loss of viability due to IDH mutant inhibitor treatment. A more metabolically stable version of the inhibitor, ivosidenib (AGI-120), was approved for chondrosarcoma clinical trials (Popovici-Muller et al., 2018) and showed poor progression free survival with single-agent therapy in one retrospective study (van Maldegem et al., 2019), and more durable disease control in a phase I study, with 52% of patients experiencing stable disease. However, a placebo controlled randomized phase II study with saridegib (IPI-926) stopped for futility after an interim-analysis (Bussiness Wire, 2020). The company never released the data for a formal scientific publication.
The effect of radiotherapy on chondrosarcoma monolayer models varied between cell lines in several studies Venneker et al., 2019). A special protocol for a 3D version of the commonly used colony formation assay was developed using the alginate chondrosarcoma spheroid model to determine the effect of radiation on spheroid number and size. This method is not possible within all 3D models as some scaffolds, such as collagen-based ones, allow free migration of cells, preventing formation of individual colonies, and the ECM production capabilities of chondrosarcoma also make cell extraction from scaffolds difficult (Hamdi et al., 2015). JJ012 spheroids exhibited a higher resistance to radiotherapy than previously published 2D culture results (SF2 2D = 0.54/0.55, 3D = 0.83), whereas CH2879 showed similar survival (SF2 2D = 0.62, 3D = 0.58) and SW1353 maintained resistance to radiotherapy (SF2 2D = 1.00/0.88, 3D = 0.95) Venneker et al., 2019). Overall, all three spheroid models were substantially resistant to radiation, as is reported for chondrosarcoma in patients. Advanced radiotherapy modalities may be able to overcome the observed resistance, because radio resistance has been linked to the total radiation dose applied, and modern techniques enable the safe application of a higher total dose. Current clinical data indeed show that the use of advanced radiation modalities, such as intensity-modulated radiation therapy (IMRT), volumetric modulated arc therapy (VMAT), stereotactic radiosurgery (SRS) and especially particle therapy (radiotherapy with protons or carbon ions) result in a trend toward improved survival with higher radiation doses applied for patients with positive margins after resection (Weber et al., 2016;Catanzano et al., 2019). Therefore, the 3D chondrosarcoma spheroid model provides an excellent model to further investigate the use of these modalities for chondrosarcoma.

CONCLUSION
Using a spheroid model of chondrosarcoma, representative of chondrosarcoma in vivo with regards to morphology and extracellular matrix production, we demonstrated short and long-term in vitro treatment strategies. This included assessment of synergistic effect and clonogenic capacity, in order to better predict in vivo response. Use of the alginate bead model for chemotherapeutic, radiotherapeutic and combination therapy testing would refine the selection of therapies for in vivo testing of chondrosarcoma. This approach is more efficient, cost effective, ethically beneficial, and may lead to discovery of novel treatment options for chondrosarcoma patients.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
IP optimized 3D cell cultures and IP, SV, and BA set up 2D and 3D cultures for the study. IP, SV, and BA performed antineoplastic agent experiments. IP and AV performed data analysis of cell viability assays. HG and AK performed data analysis for clinical correlation. IB performed all IHC staining. IP analyzed 3D colony counts and IHC. JB conceived and supervised the study. IP and JB wrote the manuscript. All authors read and approved final manuscript.

FUNDING
This work was financially supported by the Netherlands Organization for Scientific Research (ZON-MW VICI 170.055 to JB).

ACKNOWLEDGMENTS
We are grateful to Dr. JA Block (Rush University Medical Centre, Chicago, IL, USA), who provided us with the JJ012 cell line and Professor A Llombart Bosch (University of Valencia, Spain) for the CH2879 cell line.