Validation of chemoresistance phenotypes in pleural mesothelioma across 2D, 3D, and in vivo models

Abstract

Pleural mesothelioma (PM) is a highly aggressive cancer with limited treatment efficacy and poor prognosis. Conventional two-dimensional (2D) culture models fail to replicate the tumour microenvironment (TME), limiting their translational relevance. Here, we establish a three-dimensional (3D) spheroid model to investigate chemotherapy resistance across different PM subtypes. Compared to 2D cultures, 3D spheroids display enhanced resistance to cisplatin-pemetrexed, with elevated IC₅₀ values, reduced apoptosis, and altered cell cycle profiles. Seahorse metabolic analysis of 3D spheroids demonstrate a suppressed metabolic phenotype, characterised by reduced oxidative phosphorylation (OCR). However, the glycolytic capacity was not upregulated, consistent with the hypoxic and nutrient-limited conditions observed in mesothelioma lesions. In parallel, molecular profiling identifies subtype-specific miRNA signatures that closely align with patient-derived datasets. Proteomic analysis of 3D cultures identifies upregulation of PI3K/AKT and Notch/VEGF signalling, implicating these pathways in treatment resistance. Histological assessment of xenografts further confirms 3D model fidelity in capturing tumour fibrosis, necrosis, and response to therapy. These findings position the 3D spheroid system as a robust and physiologically relevant platform for modelling drug resistance and guiding therapeutic development in PM.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-38692-4.

Introduction

Pleural mesothelioma (PM) is a lethal malignancy of the mesothelial lung lining, primarily caused by asbestos exposure. The 5-year survival rate for PM is less than 10%, with a poor patient prognosis and a short median overall survival of 18 months^1,2. PM is classified into three subtypes based on their histology, and among them, sarcomatoid is the most aggressive, followed by biphasic and epithelioid^3,4. Despite cisplatin-pemetrexed combination chemotherapy improved the response rates to 41%^5,6, overall survival remains limited, with most patients experiencing recurrence within 6 to 8 months^6–9. Recent trials, including dual immunotherapy, have offered only modest survival benefits, highlighting the urgent need for more predictive preclinical models¹⁰.

Conventional 2D cell culture models fail to recapitulate the complex tumour micro-environment (TME) of PM, leading to poor translational outcomes^11,12. Although xenograft models are used^13,14, they are limited by a lengthy induction time and inefficiency in high-throughput drug screening. Emerging evidence suggests that 3D culture models more accurately mimic tumour architecture, gene expression, and drug resistance profiles observed in patients^15–20. Therefore, it is important to develop a 3D PM model to delve deeper into drug resistance mechanisms, identify novel therapeutic targets, and facilitate effective drug screening.

A deeper understanding of the molecular mechanisms outlined above can facilitate the identification of novel biomarkers and therapeutic targets for mesothelioma. The development of a 3D model that closely mimics the TME and architecture will enable the investigation of complex tumour-stroma interaction, hypoxia, and drug diffusion, which are critical for accurate reflection of the clinical behaviour of the disease. Further, such models can help elucidate molecular mechanisms such as alterations in miRNA expression, which play a pivotal role in the pathogenesis of mesothelioma, as well as regulating genes involved in drug resistance, tumour proliferation, and apoptosis. Understanding how specific miRNAs contribute to the regulation of cancer-associated proteins and their resistance pathways will not only enhance our understanding of mesothelioma biology but also pave the way for the discovery of novel therapeutic targets, ultimately improving the efficacy of current treatments.

In our previous work, we demonstrated the efficacy and cost-effectiveness of utilizing low-adhesive U-bottom plates to culture 3D mesothelioma spheroids as a preclinical model for drug screening and biomarker discovery¹⁹. Our study demonstrates a methodology for growing mesothelioma cells into self-assembled 3D spheroid structures that provide a more physiologically relevant model for drug screening. In this study, we also delve into a comprehensive comparison of chemotherapy response, miRNA and cancer-associated protein expression profiles, mitochondrial function, apoptosis, necrosis, and histological changes in mesothelioma cells cultured in 2D and 3D spheroids, both in vitro and in vivo.

Results

3D spheroids of mesothelioma demonstrate a higher resistance to chemotherapy compared to 2D monolayer

To determine the optimal cell density for spheroid formation to monitor their growth, viability, and apoptosis, a varied number of cells (2500, 10,000, and 50,000 cells/well) were tested (Supplementary Fig. 1). A cell density of 2500 cells per well produced the largest spheroid following a 3-to-6-day incubation period (Supplementary Fig. 1A and B). This seeding density was used for subsequent drug response assays. For cell cycle and apoptosis assays, a seeding density of 50,000 cells per well was chosen as they showed apoptotic response after 6 days (Supplementary Fig. 1C). For each experiment, cells were plated in same seeding density for 2D and 3D spheroids to ensure experimental consistency.

The chemotherapy response of PM cells was determined by assessing their viability. Across all subtypes, PM cells in 2D monolayers responded well to cisplatin-pemetrexed treatment, evident by the low IC₅₀ dose, while 3D spheroids were resistant, showing minimal size reduction and a high IC₅₀ dose (Fig. 1A–E). Spheroid cultures also exhibited resistance to monotherapies with cisplatin, pemetrexed, vinorelbine, and gemcitabine (Supplementary Fig. 2). Drug resistance was evident across all histological subtypes in 3D spheroids (Table 1), with sarcomatoid spheroids showing the highest resistance (IC_50, 1732 µM), followed by epithelial (IC₅₀, 215.9 µM) and biphasic (IC₅₀, 145.2 µM) (Table 2), consistent with clinical observation. In contrast, 2D cultures showed similar sensitivity across all subtypes (Fig. 1B).

Figure 1.

3D spheroids of mesothelioma cells demonstrate elevated resistance to chemotherapy compared to 2D cells and exhibit an S-phase arrest. (A) Representative images of epithelioid (1157), biphasic (MSTO), and sarcomatoid (1137) subtype of mesothelioma cells cultured in 2D monolayer and 3D spheroids with or without cisplatin and pemetrexed treatment. Drug response curve for (B) grouped subtypes, (C) epithelioid, (D) biphasic, and (E) sarcomatoid post-cisplatin and pemetrexed treatment in 2D and 3D spheroids. (F) Flow cytometry was used to assess the cell cycle distribution pre- and post-cisplatin and pemetrexed treatment. Percentage of (G) 1157, (H) MSTO, (I) 1137 in cell cycle phases. Each bar represents an average of n = 3 biological experiments. Error bars represent ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. (B) -* represents epithelioid vs sarcomatoid (3D), # represents biphasic vs sarcomatoid (3D).

Table 1.

IC₅₀ values of cisplatin and pemetrexed combination in mesothelioma subtypes cultured in 2D and 3D.

IC50 values of Cisplatin + Pemetrexed (µM)
Epithelial		Biphasic		Sarcomatoid
2D	3D	2D	3D	2D	3D
9.87	215.9	12.87	145.2	14.64	1732

Table 2.

List of cells with their histological subtype and source (p-patient-derived, c-cell line).

Histological subtype	Cells	Source
Sarcomatoid	1137 (p)	ADDRI
	2359 (p)	ADDRI
Biphasic	MSTO (c)	ATCC
	MM05 (c)	University of Queensland
	1180 (p)	ADDRI
Epithelial	H28 (c)	ATCC
	H226 (c)	ATCC
	1157 (p)	ADDRI
	1506 (p)	ADDRI
	2174 (p)	ADDRI
	VMC23 (c)	Medical University of Vienna
	H2452 (c)	ATCC
	H2052 (c)	ATCC

The drug response assay was performed on all chosen PM cells, both patient-derived and commercial. For subsequent experiments to investigate changes in cell cycle, apoptosis, and metabolic shift through Seahorse assay we chose one cell line for each histological subtype. Based on reproducible growth, capacity to form spheroids, and chemotherapy response profile, we chose the following cells to represent each subtype—1157 (epithelioid), MSTO (biphasic), and 1137 (sarcomatoid).

To investigate chemotherapy response, we analysed cell cycle changes in 2D monolayers and 3D spheroids (Fig. 1F). After cisplatin-pemetrexed treatment, all three PM subtypes: Epithelioid (1157), Biphasic (MSTO), and Sarcomatoid (1137) showed increased S phase population in 2D cultures. The percentage of cells in the S phase increased from 13.34 to 27.6% in 1157, 11.778% to 25.35% in MSTO, and 12.25% to 26.533% in 1137 (Fig. 1G–I). There was no S phase alteration in treated 3D spheroids however, a significant reduction in the G₀-G₁ phase was observed in epithelioid (2D and 3D) and biphasic (2D) cells, while no changes in the sarcomatoid cells (1137), indicating strong resistance (Fig. 1I).

Chemotherapy-resistant phenotypes in 3D spheroids were associated with apoptosis inhibition and altered miRNA expression

To investigate the cell death mechanism, we used Annexin V (AV) and Propidium iodide (PI) co-staining with flow cytometry (Fig. 2A) to identify viable (AV⁻/PI⁻), early apoptotic (AV^+/PI⁻), late apoptotic (AV⁺/PI⁺), and necrotic (AV⁻/PI⁺) cells (Fig. 2B). After cisplatin-pemetrexed treatment, a significant reduction in viable cells was observed in 2D cultures of 1157 (13%), MSTO (29%), and 1137 (10%), but not in 3D spheroids. Early apoptosis increased significantly in 2D: from 11.8% to 31.5% (1157), 10.7% to 29.3% (MSTO), and 8.7% to 21.6% (1137), while 3D spheroids showed only a 5% increase (Fig. 2C). Notably, untreated 3D spheroids of 1157 showed elevated early apoptosis compared to 2D (Fig. 2C).

Figure 2.

3D mesothelioma spheroids show inhibition of apoptosis and altered miRNA expression following treatment with cisplatin and pemetrexed. (A) Experimental workflow for assessing apoptosis/necrosis using flow cytometry. (B) 2D and 3D spheroids were treated with cisplatin-pemetrexed, followed by staining with annexin-V/PI. (C) Percentage of viable, apoptotic, and necrotic cells measured in 2D and 3D spheroids in each subtype: 1157 (epithelioid), MSTO (biphasic), and 1137 (sarcomatoid). (D) miRNA fold change in epithelioid and non-epithelioid 2D and 3D spheroids post cisplatin and pemetrexed treatment. Each bar represents an average of n = 3 biological experiments. Error bars represent ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. * 2D vs 2D cis/pem; # 2D vs 3D.

Given the important role of miRNAs in apoptosis and drug resistance, our previous study revealed that certain miRNAs are strongly associated with drug resistance. We measured the expression level of miRNAs linked to patient prognosis and chemotherapy resistance in PM, as well as in other cancers.

We observed that miRNA responses to cisplatin–pemetrexed were subtype-dependent. In non-epithelioid 3D spheroids, we observed significant post-treatment increase in chemotherapy-resistance–associated miRNAs, including miR-21 (fold-change ~ 2.45) and miR-30e (fold-change ~ 1.94). In contrast, epithelioid 3D spheroids showed attenuated or decreased expression of these miRNAs relative to 2D cultures. Two-way ANOVA confirmed a culture method × subtype interaction (F = 13.87, DF = 3, P < 0.0001), indicating that 3D-induced miRNA upregulation was enriched in non-epithelioid models (Fig. 2D). A combined summary of miRNA expression of all cell lines used in this study is listed in supplementary Fig. 3.

3D mesothelioma spheroids are less susceptible to metabolic changes compared to 2D monolayers

Hypoxia in 3D spheroids may contribute to their chemotherapy resistance. To explore this, we used the Seahorse XF analyzer to assess metabolic changes. Compared to 2D control, cells treated with cisplatin-pemetrexed showed a significant drop-in oxygen consumption rate (OCR), while the OCR profile of 3D spheroids, both control and cisplatin-pemetrexed treated, remained unaltered (Fig. 3A). In 2D cultures of epithelioid (1157) and biphasic (MSTO) cells, chemotherapy treatment caused a shift from an energetic to a quiescent state. In contrast, 3D spheroids and sarcomatoid (1137) cells stayed in a quiescent state regardless of treatment (Fig. 3B).

Figure 3.

3D mesothelioma spheroids are resistant to metabolic changes post cisplatin-pemetrexed treatment compared to 2D cells. (A) Seahorse XF Mito stress test profile for 2D and 3D spheroids with or without cisplatin-pemetrexed treatment, and (B) energy profile. Parameters of mitochondria function—(C) non-mitochondrial, (D) basal, (E) maximal, (F) ATP production, (G) proton leak, and (H) spare respiratory capacity. Each bar represents an average of n = 2 biological experiments and data represented as ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

2D treated cells also showed a significant reduction in OCR associated with basal, maximal, and non-mitochondrial respiration, proton leak, and ATP production, with spare respiratory capacity unaffected (Fig. 3C–H). While 3D cells showed a trend towards decreased respiration, the changes were not statistically significant post-treatment. These results suggest that in 2D cultures, chemotherapy treatment suppressed oxidative activity under stress, whereas 3D spheroids exhibited a metabolically quiescent profile consistent with a pre-adapted, low-OXPHOS state typical of hypoxic tumours.

We determined the extracellular pH and lactate dehydrogenase (LDH) levels in PM cells cultured in 2D and 3D spheroids (Supplementary Fig. 4). 3D spheroids had high levels of extracellular pH and elevated LDH levels compared to 2D indicating a defined metabolic and environmental difference between the two models. High LDH levels in 3D spheroids reflect an enhanced hypoxic condition within the spheroid core, a characteristic feature of tumors. Despite high LDH activity, the higher pH of 3D spheroids than 2D suggests that PM cells in 3D spheroids regulate lactate efflux while preventing excessive acidification of the extracellular environment. Together, an increase in both the extracellular pH and LDH release in 3D spheroids reflect cellular stress underscoring the limitation of 2D cultures.

3D-derived tumours are resistant to chemotherapy compared to 2D-derived tumours

To investigate drug response pathways, we implanted MSTO PM cells grown in 2D and 3D cultures into a xenograft mouse model. Mice received four cycles of cisplatin (5 mg/kg) and pemetrexed (200 mg/kg) over 28 days (Fig. 4A), with tumour growth monitored using IVIS imaging (Fig. 4B). Post-chemotherapy, we observed a decrease in the tumour size in mice implanted with 2D-derived tumours but not in the 3D-derived tumours (Figs. 4B and E).

Figure 4.

3D spheroid-derived tumours exhibit higher chemoresistance than 2D-derived tumours in PM mice model. (A) Experimental workflow in animal model and representative images for (B) tumour growth. Red arrows indicate the tumour site at the time of harvest. (C) Tumour size, (D) weight, (E) tumour volume post-animal harvest, and (F) body weight after tumour inoculation. (G) Animal response rate and (H) survival to cisplatin-pemetrexed combination treatment in 2D and 3D animal models. Data represented as mean ± SEM with n = 7–8 mice per group. *P < 0.05.

Upon animal harvest on day 32, we found that 2D-derived tumours had developed multiple nodules spreading over the peritoneal and thoracic cavity lining, whereas the 3D tumours were solid and present only in the peritoneal cavity (Fig. 4C). The average weight of 3D-derived tumours (0.34 g) was higher compared to 2D-derived (0.18 g) (Fig. 4D) and showed a significant loss of body weight in all groups of mice (Fig. 4E). Response to chemotherapy was lower in mice implanted with 3D spheroids (12.5%) compared to those with 2D (28.5%) (Fig. 4F). Notably, the two groups had no discernible difference in animal survival (Fig. 4G), and showed similar overall survival (Fig. 4H).

3D-derived tumours enhanced fibrosis and collagen deposition, contributing to the resistance phenotype

Histological analysis of the harvested tumours was carried out to understand the 2D and 3D-derived tumour response to cisplatin and pemetrexed treatment. As the fibrotic TME can contribute to the excessive production of collagen and extracellular matrix components that facilitate tumour progression, we engaged two pathologists to assess and score the degree of fibrosis and collagen deposition in the tumour samples. Tumours derived from 3D spheroids exhibited an organized histological structure with cell nests and cords compared to those grown in 2D. The tumours were subjected to Masson’s trichome and Verhoeff’s staining, which revealed enhanced collagen deposition in 3D-derived tumours (Fig. 5A). Our analysis revealed that 3D-derived tumours (57%) exhibited a poor or no response to chemotherapy, while 2D-derived tumours (71%) had an intermediate response (Fig. 5B). All animals treated with chemotherapy (both 2D and 3D) showed an increase in fibrosis (Fig. 5C). Mild fibrosis was noted in 2D-derived tumours, whereas enhanced fibrosis (moderate) was observed in the 3D group (Fig. 5C). Notably, 3D tumours showed a marked reduction in necrosis (Fig. 5D) when compared to 2D-derived tumours, with no difference in lymphocyte infiltration between the 2D and 3D mouse groups (Supplementary Table 2).

Figure 5.

Histological assessment of 2D and 3D mesothelioma spheroids derived tumours pre- and post-cisplatin and pemetrexed treatment. (A) Hematoxylin and eosin stain (H&E), Masson’s trichome, and Verhoeff’s staining on mice tumours derived from 2 and 3D cultured mesothelioma cells. Scale bar = 200 µm. (B) Tumor response to chemotherapy, (C) degree of fibrosis, and (D) tumor necrosis. N = 6 mice in each study group. * P < 0.05, ** P < 0.01, *** P < 0.0001.

Tumours derived from 3D spheroids enhance cell survival by activating the PI3K/AKT and VEGF/Notch signalling pathways

To investigate the mechanism of resistance phenotype observed in tumours derived from 3D spheroids, we first compared apoptosis in tumours using the TUNEL assay. Following treatment with cisplatin and pemetrexed, 3D-derived tumours had fewer apoptotic cells compared to 2D-derived tumours (Fig. 6A). Further, there was a significant increase in the expression of miR-30e and miR-221 in 3D tumours, with the majority of the miRNA targets being upregulated when compared to 2D tumours (Fig. 6B). This suggests that the altered miRNA expression may contribute to the observed cisplatin-pemetrexed resistance. Next, we utilized an oncology proteome array to screen cancer-associated proteins to identify any dysregulated pathways that contribute to drug resistance (Fig. 6C and D, Supplementary Table 3). Tumours derived from the 3D model, particularly following cisplatin-pemetrexed treatment, displayed a significant upregulation of proteins associated with the PI3K/AKT pathway, including BCL-X, EBRb1, EBrb3, HO-1, Osteopontin (Fig. 6E) and the VEGF/Notch signalling pathway such as DKK-1/DLL-1(Fig. 6F). The activation of these pathways enhanced mesothelioma cell survival by inhibiting apoptosis, thereby contributing to the observed resistance phenotype.

Figure 6.

Alteration in apoptosis, miRNA, and proteome profile after cisplatin and pemetrexed treatment in 2D and 3D derived tumours from mice. (A) Tunnel assay revealed a reduction in apoptosis in 3D-derived tumours post-treatment, Scale bar = 200 µm. (B) miRNAs expression in 2D/3D tumours post-chemotherapy. (C) Heat map and (D) dot blots of cancer-related proteins from tumour samples (2D/3D) pre- and post-treatment. Alteration of proteins associated with (E) PI3K/AKT and (F) Notch/VEGF signalling pathways. The numbers correspond to the dot plot position. N = 6–8 for miRNA analysis. * P < 0.05, ** P < 0.01, *** P < 0.0001.

Discussion

There is a pressing need for pre-clinical models that accurately reflect the biology of PM tumours to improve drug development and treatment outcomes¹⁴. Traditional 2D cultures fail to replicate the complex TME, contributing to poor translational success. In this study, we used a panel of PM cell lines from the three histological subtypes and cultured them in both 2D and 3D spheroid models (Table 1). Our findings show that 3D models better reflect clinical drug resistance patterns, especially in the more aggressive sarcomatoid subtype, which was less responsive to cisplatin-pemetrexed compared to epithelioid cells. These differences were absent in 2D cultures, highlighting the clinical relevance of 3D models^10,41.

Platinum-based chemotherapy, such as cisplatin, works by inducing DNA damage, cell cycle arrest, and mitochondrial dysfunction, leading to cell death^42,43. In our comparative analysis, 2D culture cells showed significant cell cycle disruption and increased apoptosis after treatment, whereas 3D spheroids showed only a mild increase in early apoptosis, likely due to hypoxia within the spheroid core. Additionally, drug penetration within our spheroid model has been previously validated, showing comparable diffusion in both small and large spheroids cultured under low-adhesion conditions. Following chemotherapy, mesothelioma cells cultured in 2D showed changes in the cell cycle that interferes with the activation of DNA repair⁴⁴ associated with a significant increase in early and late apoptosis. Using Seahorse XF analysis, we found that chemotherapy significantly reduced OCR and shifted 2D cells from an energetic to a quiescent state. In contrast, 3D cells maintained stable metabolic profiles post-treatment. This suggests that 3D spheroids may enter a senescent but metabolically active state, contributing to chemotherapy resistance⁴⁵. Although we did not directly measure extracellular lactate levels, future experiments quantifying lactate production and medium acidification will be conducted to confirm glycolytic adaptation in 3D spheroids.

Previous studies have linked certain miRNAs with chemotherapy resistance in mesothelioma⁴⁶. In our model, non-epithelioid 3D spheroids showed upregulation of miR-21, 23a, and 30e, patterns also observed in other cancers like non-small cell lung cancer (NSCLC) and ovarian cancer (Table 3). These miRNAs are known to downregulate pro-apoptotic genes, further supporting their role in promoting drug resistance and poor patient prognosis. The upregulation of these miRNAs may contribute to drug resistance and poor patient prognosis by altering the cell cycle, apoptosis inhibition, and enhanced mitochondrial function, thereby improving cell survival.

Table 3.

MicroRNA expression in different types of cancer associated with chemotherapy treatment.

miR target	Cancers			Expression
	NSCLC and mesothelioma	Ovarian	Others (hepatocellular carcinoma, breast, bladder, and oesophageal)
miR-21	21–25	26		Up
	27,28	29		Down
miR-23a	30	31	32	Up
miR-30e			33	Up
miR-31	34	35	33	Up
		36		Down
miR-221	37,38			Up
miR-222			39,40	Up

Intraperitoneal implantation of both single-cell suspensions and pre-formed 3D spheroids was performed to directly compare a conventional mesothelioma xenograft approach with a 3D-preconditioned model⁴⁷. This design enabled us to assess whether prior 3D organisation, which better recapitulates cell–cell and cell–matrix interactions, confers additional chemoresistance beyond that observed in standard single-cell xenografts, thereby enhancing the translational relevance of the in vivo findings. Consistent with our in vitro results, tumours derived from 3D spheroids grew larger, responded poorly to chemotherapy, and displayed more organised cell structures with higher collagen deposition. This fibrosis, observed through histology and Masson’s trichrome staining, likely contributes to resistance by creating physical barriers that limit drug penetration and support tumour survival. These findings mirror observations in other cancers such as colorectal and breast cancer, where the desmoplastic TME is associated with worser or worst outcomes^48–50. A dense desmoplastic TME not only establishes a physical barrier that could hinder drug penetration within the tumour but also fosters tumour growth by mechanistically altering tumour hypoxia, reducing immune infiltration, and ultimately leading to heightened chemotherapy resistance observed in spheroid-derived tumours^50–53. Our findings are consistent with existing research, where bevacizumab, a VEGF inhibitor, has been utilized to treat mesothelioma in combination with doublet chemotherapy^1,54.

Next, we analysed key signalling pathways using a cancer proteome array. Tumours derived from 3D spheroids showed increased levels of protein involved in cell survival and fibrosis, including Bcl-Xl, ErbB1/3, DLL-1, and DKK-1^55–58. These proteins are associated with activation of the PI3K/AKT, NF-κB, and Notch/VEGF pathways—all known to suppress apoptosis and promote fibrosis⁵⁷ (Fig. 7). The upregulation of miR-21 and miR-30e, alongside these proteins, likely inhibits caspase activity and mitochondrial-mediated apoptosis. Additionally, increased DLL-1 expression promotes fibrosis through the Notch/VEGF axis, further contributing to chemotherapy resistance. These findings are supported by data from The Cancer Genome Atlas (TCGA), which links high DLL-1 expression with poor survival in PM patients (Supplementary Fig. 5). Thus, a combination of Bcl-X and DLL-1 inhibitors could be a potential therapeutic approach for chemotherapy-resistant mesothelioma patients.

Figure 7.

Proposed model of chemoresistance observed in spheroids and spheroid-derived tumours. Altered miRNAs modulate BCL-XL and DLL-1 signalling, dampen apoptotic responses, and enhance fibrotic remodelling/ECM deposition, thereby limiting drug penetration and efficacy.

Overall, our study emphasises the importance of 3D spheroid model in better mimicking the tumour behaviour observed in patients. These models not only reflect clinical resistance patterns more accurately than 2D cultures but also offer a scalable and cost-effective platform for large-scale drug screening. Importantly, animal models implanted with 3D spheroids reflect patient-like responses, making them a valuable tool for validating new therapies before clinical trials. By incorporating immune cells, this 3D spheroid model can be extended to evaluate immunotherapy responses, thereby accelerating drug discovery.

Although the intraperitoneal model provides a reproducible and ethically manageable approach to monitor mesothelioma progression in vivo, differences in drug distribution and stromal composition compared with the pleural cavity may influence treatment responses. In particular, the relatively high vascular permeability and fluid turnover in the peritoneum could result in altered cisplatin exposure; however, this is unlikely to affect the comparison of tumour microenvironment composition between 2D and 3D models. Moreover, all animals were euthanized at day 32 for comparative analyses, and survival was not followed beyond this time point. As such, the absence of a survival difference reflects the study design rather than equivalent long-term outcomes. Future studies with extended follow-up and larger cohorts will be required to determine whether the enhanced chemoresistance observed in 3D-derived tumours translate to a reduced overall survival. The other limitation of our study is the absence of direct validation using patient tumour sample, due to limited tissue availability. Nonetheless, our in vitro and in vivo findings from the 3D spheroid models align closely with published patient data, reinforcing their clinical relevance.

Materials and methods

Cell culture

The pleural mesothelioma cell lines (H2052, H2452, H28, H226, and MSTO) were purchased from the American Type Culture Collection (ATCC, VA, USA). Patient-derived mesothelioma cells (1157, 1137, 1180, 1506, 2174,and 2359) were established at the Asbestos and Dust Diseases Research Institute (ADDRI) (Ethics number—2019/EHT07696, all experiments were performed in accordance with the SLHD ethics guidelines and regulations), MM05 was generated at the University of Queensland Thoracic Research Centre (The Prince Charles Hospital, Brisbane), and VMC23 was established at the Institute of Cancer Research and Walter Klepetko (Medical University of Vienna, Vienna, Austria) (Table 1). Cells were cultured in RPMI 1640 media (ATCC Modification, Gibco, Thermo Fisher) containing 10% fetal calf serum (FCS, Thermo Fisher) in a 5% CO₂ incubator at 37 °C. The inclusion of both commercial and patient-derived cell lines ensured the robustness and translational relevance of the experimental findings.

Spheroid formation and disassociation

2500, 10,000, and 50,000 mesothelioma cells were seeded in 100 µl cell culture medium per well in an ultra-low attachment (ULA) 96 well plates (Round Bottom, Costar). The plates were centrifuged for 15 min at 300g to ensure immediate close contact of cells. The spheroids were formed after 72 h, and the medium was changed every second day. For downstream analyses, spheroids were dissociated by incubation with trypsin at 37 °C for 5–10 min, followed by gentle mechanical trituration using a 1 mL pipette tip to generate a single-cell suspension.

Drug treatment

The cells were plated in a flat bottom or a U bottom low adherent 96 well plate for 2D or 3D spheroid formation, respectively. For 2D and 3D studies, 24 h and 72 h, respectively post seeding, the cells were treated with the chosen chemotherapy drugs (Merck) – cisplatin (400 µM), pemetrexed (40 µM), gemcitabine (800 nM), vinorelbine (800 nM), and a combination of cisplatin (400 µM) and pemetrexed (40 µM).

Drug response assay

2500 cells were seeded in an appropriate cell culture plate to grow in 2D and 3D conditions. To determine the drug response in a dose-dependent manner, each chemotherapy drug was serially diluted from the highest to the lowest concentration and added to the cells. The drug response between 2D and 3D models was assessed using the Alamar Blue assay to measure cell viability 72 h post- chemotherapy as previously described⁵⁹. Briefly, following the addition of 10% Alamar Blue reagent, fluorescence intensity was measured at 590–610 nm (excitation at 544 nm) using a FLUOstar Optima microplate reader (BMG LabTech, Ortenberg, Germany). Cell viability was calculated as the percentage of fluorescence intensity relative to untreated controls.

Cell cycle analysis and apoptosis assay

For cell cycle and apoptosis analyses, cells were similarly exposed to chemotherapy for 72 h prior to fixation and staining for flow cytometric analysis. Cell cycle analysis was performed using propidium iodide (PI) staining. Apoptosis assay was performed using Alexa Fluor 488 Annexin V/ Dead Cell Apoptosis Kit (Invitrogen, # 2563670) according to manufacturer’s instructions. For each sample, 10,000 events were acquired, and a consistent gating strategy was applied across all samples (Supplementary Fig. 6).

Seahorse Mito stress assay

The Seahorse XF24 Extracellular Flux Analyzer (Agilent, CA, USA) was used to measure the mitochondria function. For 2D, 8 × 10⁴ cells/well were plated in a Seahorse cell culture microplate (Agilent), incubated overnight followed by with or without cisplatin and pemetrexed (IC₅₀) treatment for 24 h. For 3D, the same cell density was seeded in a U-bottom low-adherent 96-well plate for 72 h and transferred to the Islet Capture Microplate (Agilent). A mitochondria stress test was performed as per the manufacturer’s instructions. Oligomycin (2D—1 μM, 3D—2 μM), FCCP (2D—0.5 μM, 3D—2 μM), and Rotenone/Antimycin A (0.5 μM) were added to the cartridge. To ensure drug penetration in 3D spheroids, six cycles of measurement were used for each drug.

PM xenograft mouse model

Female SCID mice (8–10 weeks old) were obtained from the Animal Resource Centre (Perth, Australia) and housed under specific-pathogen-free conditions at the ANZAC Research Institute. Human mesothelioma cells (MSTO-211H) expressing luciferase (PGL-51 luc) were cultured as 2D monolayers or 3D spheroids and implanted intraperitoneally (1 × 10⁶ cells in 200 µL RPMI). Mice (n = 5 per group) received four cycles of cisplatin (5 mg/kg) and pemetrexed (200 mg/kg) intravenously on days 7, 14, 21, and 28. Tumour progression was monitored using an IVIS Spectrum System (PerkinElmer) after luciferin injection. For each mouse, a constant region of interest (ROI) was manually defined using Living Image® software (version 4.5.4) to encompass the entire abdominal cavity. Photon emission was quantified as total photon flux (photons/sec) within this ROI. Mice were regularly monitored and sacrificed under SHLD animal ethics (2017/021 and 2019/023, all experiments were performed in accordance with the SLHD ethics guidelines and regulations). A ketamine–xylazine anaesthetic cocktail was administered intraperitoneally for euthanasia. Ketamine was used at a dose of approximately 60–100 mg/kg, and xylazine at 5–10 mg/kg, consistent with commonly accepted protocols.

Real-time PCR

Total RNA was extracted from 2D cells and 3D spheroids using an adapted TRIzol™ Reagent (#15596018). RNA isolation from mice tumours was performed using the Qiagen miRNEasy FFPE kit (# 217504) as per the manufacturer’s protocol. An equal amount of RNA (50 ng) was converted to cDNA by TaqMan microRNA reverse transcription kit (Applied Biosystems). miRNA PCR primer probes are listed in supplementary table 1.

Proteome oncology array

Tumour lysates were analysed for cancer-related proteins using the Proteome Profiler Human Oncogene Protein Array kit (R&D systems, Minneapolis, MN, USA), following the manufacturer’s instructions. Protein spots were visualized using ChemiDoc Chemiluminescent Western Blot Imaging System (Azure Biosystems) and densitometry was performed using Image J software (v1.54 g).

Histological assessment

4 µm tumour section was stained with Hematoxylin and Eosin staining kit (Abcam), Masson’s trichome staining kit (Merck), Van Gison staining kit (Abcam), and in situ cell death kit (Tunnel assay, Roche) according to the manufacturer’s instructions. Stained sections were assessed by two pathologists.

Statistical analysis

Statistical analyses were performed using GraphPad Prism v10.6.1. Data are presented as mean ± SEM unless otherwise stated. For experiments involving multiple groups (including tumour growth, cell-cycle, apoptosis, and miRNA assays), one-way or two-way ANOVA was used as appropriate, followed by Tukey’s multiple-comparisons test for post hoc pairwise analysis. Two-tailed unpaired Student’s t-test was used for comparison between two independent groups. Mixed-effects model was applied when experimental replication was limited. Statistical significance was defined as p-value less than 0.05. Significance is reported as * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

This project was initially conceived, and the experimental design developed by YYC. Later, HS and EHB contributed by adding the proteomics experiment. HS and SPS equally contributed by performing the experiments, analysing the data, drafting and editing the manuscript. RZ, LZ, and DW assisted with the experiments. BJ, VD, XD, YW, Steven K, and AL assisted with the manuscript editing. TR and Sonja K carried out the histological assessments. The manuscript was revised and finalized by YYC.

Funding

This research was supported by the iCare DDB Idea to Action Grant.

Data availability

All relevant data generated are included in this article. Other data not relevant to the results presented here are available from the first author, Dr. Shi, upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The experimental protocol for animal studies was reviewed and approved by the Sydney Local Health District (SLHD Ethics No. 2017/021 and 2019/023). All experiments were performed in accordance with the SLHD ethics guidelines and regulations.

Consent for publication

All authors agreed to the final version of the manuscript.