Skip to main content
NCBI home page
Microsystems & Nanoengineering logoLink to Microsystems & Nanoengineering
. 2026 Aug 31;12:310. doi: 10.1038/s41378-026-01437-4

Microarray chip to fabricate 3D PDAC heterospheroids with architectural organization towards tumor invasion and chemoresistance study

Xiaoyun Wei 1,2,✉, Xingrun Lan 1, Yiqian Pan 1, Ling Wang 1,2, Keke Chen 1,2,✉, Mingen Xu 1,2,✉
PMCID: PMC13530237  PMID: 42675045

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive solid tumor, closely associated with its unique tumor microenvironment (TME), which is characterized by a dense desmoplastic stroma. Abundant stromal cells, primarily fibroblasts, constitute the majority of cells in the tumor mass and exhibit pronounced spatial heterogeneity. Importantly, the spatial distribution of tumors and fibroblasts is vital for shaping the TME and critically influencing therapeutic responses. Here, we present a facile microarray chip for generating architecturally defined 3D PDAC heterospheroids. This platform enables us to mimic the dynamic interactions between tumor and stromal cells and to investigate how spatial organization influences stroma heterogeneity, tumor invasion and chemoresistance. The chip incorporates square concave microstructure array allowing controllable and reproducible production of uniform-sized spheroids. By simply altering the cell seeding sequence, we successfully constructed heterospheroids with distinct spatial distributions of cancer cells and fibroblasts. We further demonstrated that these organizational patterns modulate tumor-stroma crosstalk and ultimately regulate tumor invasive behavior. Furthermore, the heterospheroids with defined patterns exhibited distinct drug responses, and the potential for combination therapy evaluation was also verified. Beyond providing a robust platform for engineering heterospheroids with controllable tumor-stroma architectures, this system offers a robust 3D co-cultured model for advancing cancer research and drug screening.

graphic file with name 41378_2026_1437_Figa_HTML.webp

Open in a new tab

Subject terms: Engineering, Materials science

Introduction

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most challenging malignancies, characterized by an exceptionally poor prognosis and a low five-year survival rate1. Accumulating evidence suggests that the pancreatic stroma, which constitutes up to 50–80% of the tumor mass, plays a critical role in shaping the tumor microenvironment (TME) and promoting PDAC progression2–4. Cancer-associated fibroblasts (CAFs), which are activated from fibroblasts, represent the predominant cell type within the stromal compartment5. Highly abundant CAFs, along with the excessive extracellular matrix (ECM) proteins they produce, drive the fibrotic process and ultimately form a physical barrier that impedes drug delivery6–8.

It has been recognized that the spatial organization of CAFs within the TME remains highly heterogeneous9–12. This heterogeneity gives rise to distinct tumor-stroma organizational patterns, ultimately leading to the formation of specialized niches associated with specific histopathological features and clinical outcomes. CAFs located adjacent to tumor cells often exhibit high expression of α-smooth muscle actin (α-SMA) and are classified as myofibroblastic CAFs (myCAFs). These myCAFs contribute to matrix deposition and contraction, thereby promoting desmoplasia. In contrast, CAFs situated farther from tumor cells, within stromal regions, are characterized by their secretion of inflammatory mediators and are termed inflammatory CAFs (iCAFs)13,14. The phenotypic diversity and plasticity of CAFs present both challenges and opportunities for developing effective cancer therapies. Consequently, there is a critical need to develop advanced in vitro PDAC models that incorporate tumor-stroma spatial complexity, enabling the exploration of their dynamic interactions within the TME and facilitating the discovery of novel therapeutic strategies.

Current in vitro 3D PDAC models aim to recapitulate the complex stromal architecture and cell-cell interactions observed in vivo15. Commonly employed platforms include patient-derived organoids (PDOs)16–18, multicellular spheroids19–21, organ-on-a-chip systems22,23, and bioprinted hydrogel constructs24–26. These models allow for the co-culture of pancreatic cancer cells with stromal components such as CAFs, endothelial cells, and immune cells, thereby preserving key aspects of TME heterogeneity and signaling. In particular, 3D spheroids are capable of mimicking critical structural features of solid tumors, including tight cell-cell contacts and cell-ECM interactions, making them widely adopted in recent years for disease modeling and anticancer drug screening27–30. However, most current pancreatic cancer spheroid models rely on random co-culture techniques, which involve simply mixing tumor and stromal cells together. This results in heterogeneous spheroids formed through a largely uncontrolled aggregation process, limiting their physiological relevance and reproducibility.

In this study, we developed a versatile microarray platform for facile generation of architecturally defined 3D PDAC spheroids. The generated 3D heterospheroids with core-shell morphology were achieved by coculturing of cancer cells and fibroblasts with a sequential cell seeding strategy, enabling systematic investigation into stroma spatial heterogeneity and their crosstalk with cancer cells. As illustrated in Fig. 1a, the spheroid fabrication process allows precise control over cellular organization on the microarray chip. By sequentially depositing these two cell types, we engineered two distinct spheroid configurations: tumor-core/stroma-shell and stroma-core/tumor-shell. For comparison, randomized mixed spheroids were also produced through simultaneous co-seeding of tumor cells and fibroblasts. We further demonstrate the utility of this platform in elucidating the correlation between spatial architecture and tumor-stroma interactions, ultimately establishing a robust and physiologically relevant in vitro model for mechanistic studies and drug screening applications.

Fig. 1. 3D PDAC heterospheroids fabrication through the designed microarray chip.

Fig. 1

Open in a new tab

a Schematic representation of the microarray chip to fabricate 3D PDAC heterospheroids with architectural organization. b Images of the PDMS chip and the chip integrated into a well of 48-well plate. c Cross-sectional image of the PDMS chip. Scale bar is 200 μm. d Width and height measurement of the square concave structure (n = 30). e Representative image of the generated tumor spheroids and the size distribution. Scale bar is 200 μm. f Representative image of the generated stroma fibroblast spheroids and the size distribution. Scale bar is 200 μm

Results and discussion

Design and fabrication of microarray chip for spheroid generation

The microwell structure has been extensively utilized for the production of uniform cell spheroids due to its ability to ensure high consistency and reproducibility31,32. In this study, we developed a transparent polydimethylsiloxane (PDMS) based microarray chip incorporating square microwells for efficient cell trapping. The overall diameter of the fabricated chip was designed to be 10 mm to ensure compatibility with standard 48-well plates (Fig. 1b). Cross-sectional analysis demonstrated that the chip architecture was faithfully replicated from the master mold, exhibiting high morphological fidelity (Fig. 1c). Individual microwells featured a side length of approximately 300 μm and a depth of around 220 μm (Fig. 1d), dimensions optimized for promoting spheroid formation and growth at physiologically meaningful size.

The capacity of the designed microarray chip to efficiently produce uniformly sized cell spheroids was evaluated. Tumor cells (3 × 105 cells/chip) and normal fibroblasts (6 × 105 cells/chip) were separately seeded onto the chip. Prior to cell seeding, single-cell suspensions of PANC-1 cells and fibroblasts were thoroughly mixed to ensure uniform distribution. After seeding, cells were allowed to settle into the microwells via gravity sedimentation for 30 min. After one-day culture, both cell types had successfully aggregated and formed spheroids with narrow size distributions (Fig. 1e, f). Notably, the fibroblast-derived spheroids exhibited better structural roundness compared to those formed by tumor cells. This difference in morphology may be attributed to the stronger cell-cell adhesion and higher movement frequency inherent to fibroblasts, which promote more compact cellular assembly within each microwell. Collectively, the square microwell design of our microarray chip provides several distinct advantages for culturing spheroids over traditional well geometries. First, the square microwells are arranged in a tightly-packed configuration, which significantly reduces cell loss during seeding and culture. A 10 mm-diameter chip allows approximately 470 spheroids to be generated in the same batch. Further, the geometric confinement provided by the square microwells guided consistent spheroid formation with reduced size variability. Besides, unlike conventional V-shaped microwells, the flat-bottomed square architecture promotes a more uniform cell distribution and avoids stress concentration often found at narrow or curved bottoms. Particularly for co-cultures, this structural feature facilitates enhanced cell distribution throughout the formed spheroids, thus allowing the development of heterospheroids with exact spatial structure.

On-demand generation of architecturally tunable PDAC heterospheroids

To further recapitulate the stromal compartment of the PDAC TME, co-cultured heterospheroids comprising tumor cells and fibroblasts were established. As noted, CAFs in the pancreatic TME originate from various sources, such as pancreatic stellate cells (PSCs), resident fibroblasts, and bone marrow-derived mesenchymal stem cells. And the quiescent PSCs are considered the predominant precursors of CAFs in PDAC33. Studies have employed normal fibroblasts co-cultured with pancreatic cancer cells to investigate how fibroblasts acquire CAF-like phenotypes34–36, while their role in tumor-stroma crosstalk remains worthy of further investigation, particularly in spatially defined contexts. In this study, we specifically co-cultured PDAC cells with normal fibroblasts under controlled architectural configurations, in order to systematically explore how spatial organization influences tumor-stroma interactions and spheroid phenotype.

Using a sequential cell seeding strategy, we efficiently generated heterospheroids with tunable spatial configurations. Tumor cells and fibroblasts were pre-labeled with DiI (red) and DiO (green), respectively, to enable clear visualization of their distribution. The self-assembly process of the spheroids was monitored via time-lapse imaging to assess growth dynamics and morphological characteristics (Fig. 2a). After forming mono-culture spheroids of either tumor or stromal cells, the other cell type was seeded and trapped into flat-bottom microwells, facilitating their organization around the pre-formed spheroids. Fluorescence imaging revealed that well-defined heterospheroid architectures, specifically tumor-core/stroma-shell and stroma-core/tumor-shell, were successfully achieved following two days of culture. In contrast, randomly mixed co-culture spheroids exhibited a disordered distribution of both cell types. Diameter measurements of the generated PDAC heterospheroids with different configurations were conducted to access the reproducibility of the microarray chip. The sizes of the spheroids varied little among different chips for three types of heterospheroids, suggesting the robustness of the proposed system for generating heterospheroids with configurations (Fig. S1). The inclusion of fibroblasts significantly enhanced the compactness and sphericity of the heterospheroids compared to mono-culture tumor spheroids. This pronounced morphological effect is likely mediated by the abundant deposition and remodeling of ECM components by fibroblasts, which provides structural support and facilitates cell-cell adhesion37. To confirm structural stability, we acquired fluorescence images of the heterospheroids at day 5 for each configuration. All three spatial architectures remained stable for at least 5 days, although a limited number of fibroblasts were observed to infiltrate the tumor cell region in the tumor-core/stroma-shell heterospheroids (Fig. S2). This phenomenon aligns with previous reports indicating that fibroblasts possess higher motility and aggregation propensity, which can drive partial reorganization within heterospheroids over time38.

Fig. 2. On-demand generation of architecturally tunable PDAC heterospheroids with stromal components.

Fig. 2

Open in a new tab

a Representative fluorescent and bright field images showing the architecturally organized heterospheroids respectively in tumor-core, stroma-core and random pattern in different time points. Tumor cells and stroma fibroblasts were respectively labeled with red and green dye. Scale bar is 100 μm. b Images showing the live/dead analysis of the heterospheroids in three different architectural patterns at day 3. Scale bar is 100 μm. c The corresponding measured cell viability. Values are expressed as the means ± SD (n = 3)

Viability was assessed using a live/dead assay to further characterize the developed 3D PDAC heterospheroids. Representative fluorescence micrographs and quantitative analysis revealed that all three spatial configurations of co-cultured spheroids maintained high cell viability, exceeding 80%, after 5 days in culture (Fig. 2b, c). Notably, there was no distinct necrotic core formation within these three types of heterospheroids, indicating the efficacy of the microarray chip in supporting 3D tumor spheroid culture with stability and high viability.

Tumor-stroma interactions in PDAC heterospheroids

The tumor-stroma interactions within the heterospheroids were evaluated through immunofluorescence staining based on cell type-specific markers. Specifically, α-SMA, a typical marker for CAFs was used to assess whether normal fibroblasts underwent activation during the co-culture period6. Cytokeratin 19 (CK19) was employed to label PDAC cancer cells23, enabling clear distinction between cell types within the heterospheroids. As demonstrated in Fig. 3a, fluorescence signals of α-SMA were obvious observed in all the three configurations of heterospheroids, indicating that the normal fibroblasts were activated and gained the CAF-like phenotype through the co-culturing with PDAC cancer cells. The tumor-core/stroma-shell configuration exhibited the strongest fluorescence signal among the three architectures (Fig. S3). Furthermore, the spatial organization of each cell type corresponded well with the intended design. In the tumor-core/stroma-shell group, the majority of activated fibroblasts were localized to the outer layer, whereas in the stroma-core/tumor-shell group, tumor cells predominantly surrounded the stromal core.

Fig. 3. Characterization of three different PDAC heterospheroids.

Fig. 3

Open in a new tab

a Immunofluorescence images showing the expression of proteins including α-SMA and CK19 on three kinds of heterospheroids at day 3. Scale bar is 100 µm. b Quantification of mRNA expression levels of α-SMA, FAP and MMP-2 in three different PDAC heterospheroids at day 3 with the random heterospheroids regarded as control. c Quantification of E-cadherin and Vimentin mRNA expressions with the mono-tumor regarded as control. All results are expressed as the means ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

Gene expression analysis was further conducted to elucidate the molecular mechanisms underlying tumor-stroma interactions across the three heterotypic spheroids. We assessed the mRNA expression levels of α-SMA, and fibroblast activation protein (FAP), both well-established markers of CAFs. As previously reported, α-SMA characterizes highly contractile myCAFs, which typically reside adjacent to tumor cells and contribute to a dense physical barrier through ECM production and mechanical contraction. In contrast, FAP-positive CAFs demonstrate strong proteolytic activity and immunomodulatory functions, and are generally distributed in stromal regions distant from tumor cells4,7,13,14. As noted, the tumor-core/stroma-shell heterospheroids showed significantly higher α-SMA expression than the other configurations, whereas stroma-core/tumor-shell spheroids exhibited elevated FAP expression (Fig. 3b). Furthermore, this stroma-core/tumor-shell configuration also demonstrated the highest expression level of matrix metalloproteinase-2 (MMP-2), a key matrix-degrading protease. This elevated MMP-2 level may be attributed to the concentrated CAF core, which creates a localized microenvironment facilitating efficient, high-concentration paracrine signaling from CAFs to tumor cells39.

To further investigate the functional consequences of these distinct CAF phenotypes, we evaluated their impact on epithelial-mesenchymal transition (EMT), a key process promoted by the intercellular crosstalk between CAFs and cancer cells that enhances motility and invasion40,41. Analysis of classic EMT markers revealed that tumor-core/stroma-shell heterospheroids displayed significantly reduced E-cadherin and elevated Vimentin expression, indicating a pronounced shift toward a mesenchymal state (Fig. 3c). Importantly, all co-culture models enhanced tumor cell invasiveness compared to the monocultured tumor cells, highlighting the critical role of tumor-stroma interactions in driving malignant progression. Collectively, these findings demonstrate that spatial organization governs CAF differentiation into distinct functional subtypes. The stroma-core/tumor-shell structure enriched both FAP and MMP-2, defining a proteolytically active CAF phenotype specialized in ECM remodeling. In contrast, the tumor-core/stroma-shell organization promoted a myCAF-rich barrier and strongly induced EMT. These results demonstrate that architectural context not only determines CAF phenotypic heterogeneity but also directly regulates their capacity to drive malignant progression in cancer cells.

Drug response of the PDAC heterospheroids

The potency of the engineered PDAC heterospheroids as a preclinical drug-screening platform was then evaluated. We tested their response to gemcitabine, a first-line chemotherapeutic agent for PDAC42. On day 3 of culture, heterospheroids were treated with gemcitabine at concentrations ranging from 25 to 100 μM, while control groups were maintained in medium containing 0.1% DMSO. After 48 hours of treatment, cell viability was assessed using the AlamarBlue assay. The results revealed that at the drug concentrations below 25 μM, all three types of heterospheroids maintained viabilities above 80% with comparable drug responses. However, distinct differential responses emerged at higher concentrations. The random mixed spheroids showed the highest sensitivity, with viability dropping below 50% at 75 μM. Conversely, both core-shell structured spheroids exhibited significantly enhanced drug resistance, maintaining higher viability even at 100 μM. Especially the tumor-core/stroma-shell configuration demonstrated the highest resistance among all models, retaining the greatest proportion of viable cells (Fig. 4a). Live/dead staining results corroborated these findings, showing the most extensive cell death in random spheroids and the best preservation of viability in tumor-core/stroma-shell spheroids (Fig. 4b). The superior resistance observed in structured spheroids, particularly those with tumor-core organization, suggests that the spatial architecture of tumor-stroma interactions contributes significantly to chemoresistance. Studies have shown that the fibrotic stroma in PDAC poses a significant challenge to the effective delivery of nanomedicines43,44. In particular, the dense collagenous matrix produced by CAFs is known to hinder drug penetration45. Besides, CAFs contribute to chemoresistance in PDAC by promoting EMT in tumor cells, which was verified through a pancreatic organoid-fibroblast co-culture system46. Thus, the enhanced drug resistance observed in the tumor-core/stroma-shell configuration potentially arises from mechanisms involving stromal protection and EMT-induced survival advantages.

Fig. 4. Effects of the drug on the three different PDAC heterospheroids.

Fig. 4

Open in a new tab

a Cell viability of PDAC heterospheroids on day 3 post gemcitabine treatment at varying concentrations, assessed using the AlamarBlue assay. b Live/dead fluorescence images of PDAC heterospheroids treated by gemcitabine at different concentrations for 48 h. Scale bar is 100  μm. c Cell viability of the tumor-core/stroma shell heterospheroids treated by pirfenidone at varying concentrations for 24 h. d Cell viability of the tumor-core/stroma shell heterospheroids under monotherapy and combination therapy. e Fluorescence images showing the expression of Collagen I and α-SMA proteins of the tumor-core/stroma shell heterospheroids after drug treatment. Scale bar is 100 μm. All results are expressed as the means ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Pirfenidone, a well-characterized antifibrotic and antioxidative agent, has been shown to inhibit fibroblast activation and reduce extracellular matrix deposition in various fibrotic diseases and cancer models47,48. In this study, we selected pirfenidone as a model anti-fibrotic agent for combination therapy based on the established role of fibrosis-driven desmoplasia in PDAC chemoresistance. The tumor-core/stroma-shell heterospheroids, which demonstrated the highest level of drug resistance in previous assays, were first treated with a concentration gradient of pirfenidone for 24 hours. A dose-dependent decrease in cell viability was observed, confirming the pharmacological activity of pirfenidone in our model system (Fig. 4c). To evaluate the utility of our spheroid platform for screening combination therapies, the tumor-core/stroma-shell heterospheroids were pretreated with 40 μM pirfenidone for 24 h, followed by gemcitabine exposure for 48 h. Viability assays revealed that the combination treatment resulted in significantly stronger growth inhibition compared to gemcitabine monotherapy (Fig. 4d). To elucidate the mechanism underlying this synergistic effect, immunofluorescence staining was performed. The results showed a marked reduction in both Collagen I and α-SMA expression in the combination treatment group, demonstrating that pirfenidone alleviates stromal fibrosis in the tumor microenvironment (Fig. 4e and Fig. S4). This attenuation of fibrotic components appears to enhance drug penetration and/or efficacy, thereby potentiating the cytotoxic effect of gemcitabine. These findings not only validate the relevance of our heterospheroid models for evaluating stromal-targeting therapies but also provide mechanistic insight into how antifibrotic agents can overcome microenvironment-mediated chemoresistance in PDAC.

Invasive outgrowth exploration of the PDAC heterospheroids post drug treatment

To model the clinical context of neoadjuvant therapy, where assessing the invasive and metastatic potential of residual tumor cells after chemotherapy or targeted therapy that is directly linked to tumor recurrence, we embedded the tumor-core/stroma-shell heterospheroids previously treated with either monotherapy or combination drugs, within gelatin methacryloyl (GelMA) hydrogel to assess their post-treatment invasive outgrowth. Untreated spheroids served as controls. Following 5 days of subsequent culture, all groups exhibited proliferation and outward invasion into the surrounding hydrogel matrix (Fig. 5a). We measured the spheroid area within the hydrogel at day 1 and day 5 post encapsulation. The control group exhibited a greater increase in spheroid area over time, indicating more extensive invasive outgrowth, whereas the combination therapy group showed the smallest expansion (Fig. S5). Immunofluorescence analysis revealed that both monotherapy and combination treatment significantly reduced CK19 expression compared to controls, with the most pronounced suppression observed in the combination group. Concurrently, expression levels of fibroblast activation markers including α-SMA and FAP were markedly downregulated in all treatment conditions, most substantially in the combination therapy group (Fig. 5b). These findings demonstrate that the therapies not only diminish epithelial tumor cell presence but also effectively attenuate stromal activation, with the combinatorial strategy yielding the strongest dual effect on both tumor and stromal compartments.

Fig. 5. Characterization of the drug-treated tumor-core/stroma-shell heterospheroids after hydrogel culture.

Fig. 5

Open in a new tab

a Bright-field and fluorescence images of the drug-treated tumor-core heterospheroids cultured using GelMA hydrogel at day 1 and day 5 time points. Scale bars are 100 μm. b Quantification of mRNA expressions of α-SMA and FAP on the drug-treated tumor-core/stroma-shell heterospheroids after 5 days of cultivation in the hydrogel. All results are expressed as the means ± SD (n = 3). *p < 0.05, **p < 0.01, ***p < 0.001

Conclusions

Overall, we establish a robust and physiologically relevant 3D heterospheroid microarray platform that recapitulates the spatial heterogeneity of PDAC stroma based on a PANC-1/dermal-fibroblast co-cultured model. Importantly, these heterospheroids with architectural organization reveal how distinct tumor-stroma configurations drive malignant progression and therapy resistance through architecture-driven CAF phenotypes. Specifically, the tumor-core/stroma-shell configuration promotes a myCAF-like phenotype characterized by elevated expression of α-SMA. The stroma-core/tumor-shell configuration induces a distinct CAF state with upregulated FAP and MMP‑2 expression, thereby facilitating matrix remodeling. By deconstructing the architectural regulation of CAF heterogeneity and its differential contributions to tumor EMT and drug response, we provide a mechanistic framework for understanding how tumor-stroma organization influences PDAC pathobiology. Furthermore, we validate the translational utility of this platform by demonstrating that stromal-targeting therapy can sensitize resistant tumors to conventional chemotherapy. This integrated approach offers a predictive preclinical tool for evaluating combination strategies and underscores the critical importance of targeting the tumor-stroma ecosystem in developing more effective treatments for pancreatic cancer.

Materials and methods

Cell source and culture

PANC-1, a human pancreatic cancer cell line with highly metastatic ability, and human normal fibroblasts from dermal tissues were generously provided by Hangzhou Regenovo Biotechnology Co., Ltd. Cells were tested using Myco-Lumi™ Luminescent Mycoplasma Detection Kit (Beyotime, China) and confirmed to be negative. Both PANC-1 and fibroblast cells were cultured in high-glucose Dulbecco’s Modified Eagle Medium (DMEM, Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS, Excell Bio). Cells were maintained in T75 flasks at 37 °C in a humidified incubator with 5% CO₂, and passaged when they reached approximately 80–90% confluence. The fibroblasts were used for the following experiments at passage 6.

Design and fabrication of the microarray chip

The microarray chip was fabricated using a replica molding technique. An SU-8 mold with a square protrusion array (300 µm in side length) was first designed and prepared. Then, a defined amount of PDMS prepolymer (Dow Corning, SYLGARD 184 Silicone Elastomer Kit) was poured onto the mold, degassed to remove air bubbles, and cured in an oven at 80 °C for 2 h. After curing, the PDMS layer was peeled off from the mold, and cut into circular chips (10 mm in diameter) to fit the 48-well plates. The obtained chips with square concave array were sterilized under UV irradiation for at least 1 h, and subsequently transferred into a 1% (v/v) F-127 solution overnight to achieve surface modification.

Cell fluorescent labeling

For visualization of cell distribution, PANC-1 and normal fibroblasts were fluorescently pre-labeled using a cell plasma membrane staining kit (Beyotime). Specifically, DiI and DiO dyes were diluted in phosphate-buffered saline (PBS, Gibco) at a ratio of 1:1000 to prepare working solutions. PANC-1 cells were stained with DiI, while fibroblasts were stained with DiO, and the cells were incubated at 37 °C for 35 min. Prior to seeding, the cells were thoroughly washed with PBS.

Generation of architecturally tunable PDAC heterospheroids

The structure of PDAC heterospheroids was modulated by altering the seeding sequence of tumor cells and fibroblasts. For the tumor-core heterospheroids, PANC-1 cells at the amount of 3 × 105 cells/chip were seeded on day 0, and after self-assembly into tumor spheroids on day 1, fibroblasts at the amount of 6 × 105 cells/chip were subsequently seeded, resulting in heterospheroids with a stroma shell on day 2. Conversely, by reversing the seeding sequence, heterospheroids with stroma-core were easily obtained. Specifically, random mixed heterospheroids were generated by co-seeding PANC-1 cells and fibroblasts respectively with cell amounts of 3 × 105 and 6 × 105 cells/chip simultaneously.

Live/Dead staining

Cell viability of the heterospheroids generated over 3 days was assessed using a live/dead staining kit (Beyotime) according to the manufacturer’s instructions. Briefly, heterospheroids on the chips were stained and incubated at 37 °C for 35 min, followed by washing with PBS. The samples were then observed under an immunofluorescence microscope (Nikon, A1RHD25, Japan). Fluorescent images of Calcein-AM and propidium iodide (PI) were processed with ImageJ software and converted to grayscale for quantitative analysis. Cell viability was calculated as the ratio of the area of Calcein-AM to the sum of the areas of Calcein-AM and PI.

Drug exposures

Heterospheroids were cultured for 3 days prior to drug treatment. Gemcitabine and pirfenidone (MedChemExpress) powder were dissolved in dimethylsulfoxide (DMSO, Sigma), and working solutions were prepared with final concentrations of 25, 50, 75, and 100 μM for gemcitabine, and 20, 40, 60, and 100 μM for pirfenidone. For single-agent assays, heterospheroids were treated with 1 mL of drug-containing medium at the indicated concentrations in each well, with medium containing 0.1% DMSO serving as the control. After 48 h of drug treatment, cell viability was assessed. To determine the effective concentration of pirfenidone for combination therapy, heterospheroids with a tumor-core structure at day 2 were first exposed to pirfenidone at concentrations of 20, 40, 60, and 100 μM for 24 h to evaluate the impact of pirfenidone on spheroids. For combination treatment, tumor-core heterospheroids were pretreated with pirfenidone (40 μM) for 24 h, followed by gemcitabine (50 μM) exposure for an additional 48 h, after which cell viability was analyzed. The mono-therapy was conducted by treating tumor-core heterospheroids at day 3 using gemcitabine (50 μM) for 48 h.

AlamarBlue assay

For the drug screening assay, AlamarBlue kit (Thermo Fisher Scientific, USA) was used to access the viability of the heterospheroids post drug treatment. Briefly, the AlamarBlue solution was diluted with basic DMEM medium at a ratio of 1: 10, and then incubated with heterospheroids for 3 h at 37 °C. After incubation, the culture supernatants were transferred to 96-well plates at 100 µL per well for detection. Fluorescence intensity was measured using a fluorescence microplate reader at excitation/emission wavelengths of 544/590 nm.

Immunostaining characterization

On day 3 of culture, heterospheroids were subjected to immunofluorescence staining for α-SMA and CK19. Spheroids were fixed with 4% (v/v) paraformaldehyde for 2 h, permeabilized with 0.5% (v/v) Triton X-100 for 30 min, and blocked with 3% (v/v) BSA at room temperature for 45 min to reduce nonspecific binding. Subsequently, the samples were incubated overnight at 4 °C with primary antibodies against α-SMA (1:200, Invitrogen) and CK19 (1:100, HUABIO). The following day, spheroids were incubated for 4 h at room temperature with secondary antibodies conjugated with Alexa Fluor 594 IgG and 488 IgG (1:200, Invitrogen). Finally, cell nuclei were counterstained with DAPI (10 μg/mL, Beyotime) for 10 min, and fluorescence images were acquired using a confocal fluorescence microscope (Nikon, A1RHD25, Japan). After monotherapy and combination therapy, heterospheroids were fixed and subjected to immunofluorescence staining. Type I collagen was detected using an anti-collagen I antibody (1:200, Invitrogen). Following drug treatment, samples were encapsulated in hydrogels and cultured for an additional 2 days prior to F-actin staining with phalloidin (1:100, Beyotime). Samples were rinsed three times with PBS and imaged using a confocal fluorescence microscope.

Fluorescence intensity analysis

Images were acquired using a confocal microscope with identical acquisition settings across all heterospheroid configurations. Mean fluorescence intensity quantification was performed using ImageJ (FIJI) software. Confocal Z-stack images of each spheroid were first acquired and then projected into a single 2D image using maximum intensity projection. The resulting images were exported and subsequently imported into ImageJ for analysis. For each image, individual fluorescent channels were split for separate quantification. To define positive staining regions, ImageJ’s automatic threshold function (Default) was applied to generate binary masks with the Dark background option selected. The mean fluorescence intensity of each spheroid was then calculated as integrated density divided by area.

Gene expression evaluation by quantitative real-time PCR (qPCR)

Total RNA was isolated directly from tumor-core, stroma-core, and randomly mixed heterospheroids within the microwell array chip without prior collection using TRIzol reagent (Thermo Fisher Scientific). Each condition comprised approximately 470 spheroids per sample. Complementary DNA (cDNA) was synthesized with the PrimeScript FAST RT Reagent Kit with gDNA Eraser (TAKARA, RR092A) following the manufacturer’s instructions. qRT-PCR was performed using TB Green Premix Ex Taq II (TAKARA, RR820A) on a CFX96 Real-Time PCR Detection System (Bio-Rad, USA). The relative mRNA expression of α-SMA, MMP-2, FAP, E-cadherin, and Vimentin was detected according to the primers in table S1, and normalized to GAPDH as the internal control and expressed as fold change. Each experiment was performed in triplicate.

Cultivation of PDAC heterospheroids post drug treatment

Following drug treatment, the tumor-core/stroma shell heterospheroids within the microarray chip were placed into an 11-mm diameter ring chamber. They were subsequently encapsulated with 120 µL of an 8% (w/v) GelMA solution prepared with 0.5% (w/v) lithium phenyl-2, 4, 6-trimethylbenzoylphosphinate (LAP). Finally, the constructs were photo-crosslinked for 30 seconds under 405 nm blue light (200 mW cm⁻²) prior to subsequent culture.

Statistical analysis

Data are presented as the mean ± SD. All experiments were performed with at least three independent biological replicates. Graphs were generated using Origin software and further refined with Inkscape. Statistical significance was assessed by two-way ANOVA with Tukey’s post-hoc test, and a p-value < 0.05 was considered statistically significant.

Supplementary information

Supplemental Material (367.2KB, docx)

Acknowledgements

This work was supported by the National Key Research and Development Program of China (2022YFA1104600), the National Natural Science Foundation of China (82303978), the Key Research and Development Program of Zhejiang Province (2024C03068), the Key Research and Development Program of Hangzhou City (2024SZD1B07) and the Central Government Guiding Funds for Local Science and Technology Development (2025ZY01048).

Conflict of interest

The authors declare no competing interests.

Contributor Information

Xiaoyun Wei, Email: wxyun@hdu.edu.cn.

Keke Chen, Email: kkchen@hdu.edu.cn.

Mingen Xu, Email: xumingen@hdu.edu.cn.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41378-026-01437-4.

References

  • 1.Wang, S. et al. The molecular biology of pancreatic adenocarcinoma: translational challenges and clinical perspectives. Signal Transduct. Target. Ther.6, 249 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Amaral, M. J. et al. Tumor stroma area and other prognostic factors in pancreatic ductal adenocarcinoma patients submitted to surgery. Diagnostics13, 655 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Apte, M. V., Wilson, J. S., Lugea, A. & Pandol, S. J. A starring role for stellate cells in the pancreatic cancer microenvironment. Gastroenterology144, 1210–1219 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Zhang, T., Ren, Y., Yang, P., Wang, J. & Zhou, H. Cancer-associated fibroblasts in pancreatic ductal adenocarcinoma. Cell Death.Dis.13, 897 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Rimal, R. et al. Cancer-associated fibroblasts: Origin, function, imaging, and therapeutic targeting. Adv. Drug. Deliv. Rev.189, 114504 (2022). [DOI] [PubMed] [Google Scholar]
  • 6.Schnittert, J., Bansal, R. & Prakash, J. Targeting pancreatic stellate cells in cancer. Trends Cancer5, 128–142 (2019). [DOI] [PubMed] [Google Scholar]
  • 7.Ho, W. J., Jaffee, E. M. & Zheng, L. The tumour microenvironment in pancreatic cancer-clinical challenges and opportunities. Nat. Rev. Clin. Oncol.17, 527–540 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Kota, J., Hancock, J., Kwon, J. & Korc, M. Pancreatic cancer: stroma and its current and emerging targeted therapies. Cancer Lett391, 38–49 (2017). [DOI] [PubMed] [Google Scholar]
  • 9.Chhabra, Y. & Weeraratna, A. T. Fibroblasts in cancer: unity in heterogeneity. Cell186, 1580–1609 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Boyd, L. N. C., Andini, K. D., Peters, G. J., Kazemier, G. & Giovannetti, E. Heterogeneity and plasticity of cancer-associated fibroblasts in the pancreatic tumor microenvironment. Semin. Cancer. Biol.82, 184–196 (2022). [DOI] [PubMed] [Google Scholar]
  • 11.Khaliq, A. M. et al. Spatial transcriptomic analysis of primary and metastatic pancreatic cancers highlights tumor microenvironmental heterogeneity. Nat. Genet.56, 2455–2465 (2024). [DOI] [PubMed] [Google Scholar]
  • 12.Liu, Y. et al. Conserved spatial subtypes and cellular neighborhoods of cancer-associated fibroblasts revealed by single-cell spatial multi-omics. Cancer. Cell.43, 905–924.e906 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Öhlund, D. et al. Distinct populations of inflammatory fibroblasts and myofibroblasts in pancreatic cancer. J. Exp. Med.214, 579–596 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Biffi, G. et al. IL1-Induced JAK/STAT Signaling Is Antagonized by TGFβ to Shape CAF Heterogeneity in Pancreatic Ductal Adenocarcinoma. Cancer. Discov.9, 282–301 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Vaskovich-Koubi, D. et al. Patient-derived 3D-bioprinted models of pancreatic cancer: toward personalized therapy and overcoming tumor microenvironment challenges. Adv. Drug. Deliv. Rev.225, 115670 (2025). [DOI] [PubMed] [Google Scholar]
  • 16.Velasco, V., Shariati, S. A. & Esfandyarpour, R. Microtechnology-based methods for organoid models. Microsyst. Nanoeng.6, 76 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Lee, J. H. et al. Establishment of patient-derived pancreatic cancer organoids from endoscopic ultrasound-guided fine-needle aspiration biopsies. Gut. Liver.16, 625–636 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Baker, L. A., Tiriac, H., Clevers, H. & Tuveson, D. A. Modeling pancreatic cancer with organoids. Trends Cancer2, 176–190 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Lazzari, G. et al. Multicellular spheroid based on a triple co-culture: a novel 3D model to mimic pancreatic tumor complexity. Acta.Biomater.78, 296–307 (2018). [DOI] [PubMed] [Google Scholar]
  • 20.Monteiro, M. V., Rocha, M., Gaspar, V. M. & Mano, J. F. Programmable Living Units for Emulating Pancreatic Tumor-Stroma Interplay. Adv. Healthc.Mater.11, e2102574 (2022). [DOI] [PubMed] [Google Scholar]
  • 21.Monteiro, M. V., Gaspar, V. M., Mendes, L., Duarte, I. F. & Mano, J. F. Stratified 3D microtumors as organotypic testing platforms for screening pancreatic cancer therapies. Small Methods5, e2001207 (2021). [DOI] [PubMed] [Google Scholar]
  • 22.Haque, M. R. et al. Patient-derived pancreatic cancer-on-a-chip recapitulates the tumor microenvironment. Microsyst. Nanoeng.8, 36 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Choi, D. et al. Microfluidic organoid cultures derived from pancreatic cancer biopsies for personalized testing of chemotherapy and immunotherapy. Adv. Sci (Weinh).11, e2303088 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Monteiro, M. V. et al. Embedded Bioprinting of Tumor-Scale Pancreatic Cancer-Stroma 3D Models for Preclinical Drug Screening. ACS. Appl. Mater. Interfaces.16, 56718–56729 (2024). [DOI] [PubMed] [Google Scholar]
  • 25.Huang, B., Wei, X., Chen, K., Wang, L. & Xu, M. Bioprinting of hydrogel beads to engineer pancreatic tumor-stroma microtissues for drug screening. Int. J. Bioprint.9, 676 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Wei, X., Wu, Y., Chen, K., Wang, L. & Xu, M. Embedded bioprinted multicellular spheroids modeling pancreatic cancer bioarchitecture towards advanced drug therapy. J. Mater. Chem. B. 12, 1788–1797 (2024). [DOI] [PubMed]
  • 27.Kang, S. M., Kim, D., Lee, J. H., Takayama, S. & Park, J. Y. Engineered Microsystems for Spheroid and Organoid Studies. Adv. Healthc. Mater.10, e2001284 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Kim, S. J., Kim, E. M., Yamamoto, M., Park, H. & Shin, H. Engineering Multi-Cellular Spheroids for Tissue Engineering and Regenerative Medicine. Adv. Healthc. Mater.9, e2000608 (2020). [DOI] [PubMed] [Google Scholar]
  • 29.Chen, X. et al. Integrated 3D microstructured digital microfluidic platform for advanced 3D cell culture. Microsyst. Nanoeng.11, 239 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Glia, A. et al. Spheromatrix: a paper-based platform for scalable 3D tumor model generation, cryopreservation, and high-throughput drug assessment. Microsyst. Nanoeng.11, 219 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Qu, F. et al. Double emulsion-pretreated microwell culture for the in vitro production of multicellular spheroids and their in situ analysis. Microsyst. Nanoeng.7, 38 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Tian, D. et al. Rocking- and diffusion-based culture of tumor spheroids-on-a-chip. Lab. Chip.24, 2561–2574 (2024). [DOI] [PubMed] [Google Scholar]
  • 33.Liu, H., Shi, Y. & Qian, F. Opportunities and delusions regarding drug delivery targeting pancreatic cancer-associated fibroblasts. Adv. Drug Deliv. Rev.172, 37–51 (2021). [DOI] [PubMed] [Google Scholar]
  • 34.Tanaka, H. Y. et al. Heterotypic 3D pancreatic cancer model with tunable proportion of fibrotic elements. Biomaterials251, 120077 (2020). [DOI] [PubMed] [Google Scholar]
  • 35.Brancato, V. et al. Bioengineered tumoral microtissues recapitulate desmoplastic reaction of pancreatic cancer. Acta Biomater49, 152–166 (2017). [DOI] [PubMed] [Google Scholar]
  • 36.Struth, E. et al. Drug resistant pancreatic cancer cells exhibit altered biophysical interactions with stromal fibroblasts in imaging studies of 3D co-culture models. Sci. Rep.14, 20698 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Ermis, M. et al. Tunable hybrid hydrogels with multicellular spheroids for modeling desmoplastic pancreatic cancer. Bioact. Mater.25, 360–373 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Priwitaningrum, D. L. et al. Tumor stroma-containing 3D spheroid arrays: A tool to study nanoparticle penetration. J. Control Release.244, 257–268 (2016). [DOI] [PubMed] [Google Scholar]
  • 39.Ito, A. et al. Co-culture of human breast adenocarcinoma MCF-7 cells and human dermal fibroblasts enhances the production of matrix metalloproteinases 1, 2 and 3 in fibroblasts. Br J. Cancer.71, 1039–1045 (1995). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Francou, A. & Anderson, K. V. The Epithelial-to-Mesenchymal Transition (EMT) in Development and Cancer. Annu. Rev. Cancer. Biol.4, 197–220 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Zhao, L., Liu, Y., Liu, Y., Zhang, M. & Zhang, X. Microfluidic control of tumor and stromal cell spheroids pairing and merging for three-dimensional metastasis study. Anal. Chem.92, 7638–7645 (2020). [DOI] [PubMed] [Google Scholar]
  • 42.Berlin, J. & Benson, A. B. 3rd Chemotherapy: Gemcitabine remains the standard of care for pancreatic cancer. Nat. Rev. Clin. Oncol.7, 135–137 (2010). [DOI] [PubMed] [Google Scholar]
  • 43.Tanaka, H. Y. & Kano, M. R. Stromal barriers to nanomedicine penetration in the pancreatic tumor microenvironment. Cancer Sci109, 2085–2092 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Tanaka, H. Y. et al. Therapeutic strategies to overcome fibrotic barriers to nanomedicine in the pancreatic tumor microenvironment. Cancers (Basel)15, 724 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Ishii, G., Ochiai, A. & Neri, S. Phenotypic and functional heterogeneity of cancer-associated fibroblast within the tumor microenvironment. Adv. Drug Deliv. Rev.99, 186–196 (2016). [DOI] [PubMed] [Google Scholar]
  • 46.Schuth, S. et al. Patient-specific modeling of stroma-mediated chemoresistance of pancreatic cancer using a three-dimensional organoid-fibroblast co-culture system. J. Exp. Clin. Cancer Res.41, 312 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Kozono, S. et al. Pirfenidone inhibits pancreatic cancer desmoplasia by regulating stellate cells. Cancer. Res.73, 2345–2356 (2013). [DOI] [PubMed] [Google Scholar]
  • 48.Lei, Y. et al. Pirfenidone alleviates fibrosis by acting on tumour-stroma interplay in pancreatic cancer. Br. J. Cancer.130, 1505–1516 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Microsystems & Nanoengineering are provided here courtesy of Nature Publishing Group

RESOURCES