1. Introduction
Cell therapy has emerged as a promising therapeutic strategy for cancer, with chimeric antigen receptor T (CAR-T) cell therapy leading current advances in the field [
1]. However, although studies indicate that the rapid expansion of CAR-T research pipelines outpaces clinical translation in oncology, one critical reason for this disparity is the lack of reliable preclinical models for efficacy prediction [
1]. Lung cancer remains the leading cause of cancer-related mortality worldwide [
2], necessitating the urgent development of novel therapies to improve patient outcomes. Lung adenocarcinoma, the most prevalent pathological subtype, harbors epidermal growth factor receptor (
EGFR) mutations in approximately 50% of Asian cases [
3]. Research indicates that tyrosine kinase inhibitor (TKI) resistance in these patients primarily involves downstream pathway alterations and bypass activation (e.g., mesenchymal epithelial transition (MET) amplification), with loss of the original
EGFR mutation being rare [
4]. Critically, these mutations are consistently associated with persistent EGFR protein expression [
5], rendering them viable targets for CAR-T therapy. This molecular rationale supports EGFR-directed CAR-T cells as a promising strategy for treating EGFR-expressing lung adenocarcinomas, particularly in TKI-resistant settings.
Substantial efforts have been devoted to developing reliable preclinical models for predicting therapeutic efficacy in oncology. Current model systems include cell lines, genetically engineered mouse models, murine-derived organoids, patient-derived xenografts (PDXs), and patient-derived organoids (PDOs) [
6]. Among these, PDOs have emerged as a particularly promising tool for personalized drug-response assessment due to their unique advantages: rapid generation timeline, faithful recapitulation of tumor heterogeneity, high genetic/phenotypic fidelity to primary tumors, and clinically predictive drug-response profiles [
7]. These characteristics have established PDOs as a widely adopted platform for evaluating targeted therapies and chemotherapeutic agents [
8], [
9]. Nevertheless, a critical limitation exists in conventional Matrigel-embedded PDO cultures—the surrounding extracellular matrices (ECMs) form a physical barrier that prevents direct cell-cell contact (
Fig. 1(a)). This architectural constraint significantly impedes their utility for evaluating cell-based therapies such as CAR-T. Therefore, innovative matrix-free culture systems are urgently needed.
Currently, only a limited number of studies have investigated matrix-free organoid culture methodologies. Notably, Jacob et al. [
10] established patient-derived glioblastoma organoids through tissue-block suspension culture using surgically resected tumor specimens. Recent advances in three-dimensional (3D) printing technology have further enabled innovative microdevice-based approaches. Chen et al. [
11] developed a microwell platform that generated mature human cerebral organoids without Matrigel dependency. Wiedenmann et al. [
12] similarly cultured pancreatic duct-like organoids from human induced pluripotent stem cells (iPSCs) using a microwell chip system, eliminating the need for Matrigel. To overcome the limitations of matrix-dependent cultures, we created a Matrigel-free microdevice that directly addresses the need for reliable organoid-based CAR-T drug screening, serving as a new translational tool for predicting response in personalized cellular immunotherapy (
Fig. 1(b)).
In this study, we developed a microwell platform through 3D printing and precision molding to enable Matrigel-free organoid culture. The hierarchically nested microwell structure allowed the successful generation of lung adenocarcinoma-derived LCOs with minimal cell input. Immunohistochemistry (IHC) confirmed EGFR expression in the LCOs, consistent with their matched primary tumors. These Matrigel-free LCOs demonstrated high susceptibility to EGFR-CAR-T cells, exhibiting efficient target recognition and lysis in the microwell platform. Importantly, the antitumor efficacy observed in vitro was recapitulated in vivo, with EGFR-CAR-T cells significantly inhibiting the growth of PDO xenograft (PDOX) tumors. This concordance between in vitro and in vivo models validates our Matrigel-free LCO platform as a reliable tool for preclinical evaluation of CAR-T cell therapies. Collectively, we present a novel Matrigel-free organoid culture platform that effectively predicts EGFR-CAR-T cells efficacy, providing a promising approach for precision screening of cellular immunotherapies.
2. Methods
2.1. Microwell fabrication
The microwells were fabricated using a combined 3D printing and molding approach. A mold was first produced using a digital light processing (DLP) 3D printer with photopolymer resin (Nova Robotics Co., Ltd., China). The mold design consisted of a 22-mm-diameter disc with a central 18-mm-diameter cylindrical protrusion, the surface of which was patterned with a truncated hexagonal pyramid structure and a micro-cylinder array. After printing, the mold was cleaned with anhydrous ethanol, ultrasonicated, and baked at 80 °C for further processing. Polydimethylsiloxane (PDMS) was prepared by mixing the base and curing agent at a 10:0.8 ratio. The PDMS solution was poured into the 3D-printed mold, degassed in a vacuum oven (BLUEPARD, China) to remove bubbles, and thermally cured at 80 °C for 2 h. The cured PDMS structure was sterilized by immersion in 75% ethanol, followed by 30-min UV exposure, and air-dried prior to use.
2.2. Patient specimens
This study protocol was approved by the Research Ethics Committee of West China Hospital, Sichuan University, China (approval No. SKMBT_28322022515102). Written informed consent was obtained from all participating patients or their legally authorized representatives. Malignant pleural effusion (MPE) samples were collected from patients with advanced lung adenocarcinoma at the Cancer Center, West China Hospital. Samples were aseptically drained into sterile collection bags, and all diagnoses were pathologically confirmed by the Department of Pathology, West China Hospital.
2.3. Processing of MPE collection and organoid culture
To standardize tumor cell isolation, 100-400 mL of MPE was collected per patient. Samples were immediately placed on ice and processed within 6 h of collection. Then, they were centrifuged at 400
g for 5 min to separate supernatant and cell pellets. Red blood cells were lysed using ACK Lysis Buffer (Lanjie Ke Technology Co., Ltd., China). Isolated cells were resuspended in ice-cold Matrigel (Corning Incorporated, USA) and seeded as 30 μL drops in 48-well plates (Corning Incorporated) at a density of 2 × 10
4-3 × 10
4 cells per well. Matrigel was polymerized at 37 °C for 15 min, followed by overlay with 150 μL of lung cancer organoid (LCO) medium (LCOM; Dulbecco’s modified Eagle’s medium (DMEM)/F12 supplemented with growth factors; see Na et al. [
13] for detailed formulations). The culture medium was refreshed every 48 h. Organoids were mechanically dissociated with TrypLE™ Select (12605-028; Gibco, USA) every 7-10 d.
2.4. Hematoxylin and eosin (H&E) staining and IHC of parental tumors and organoids
Organoids and matched parental tumor cells were fixed overnight in 4% paraformaldehyde, processed through paraffin embedding, sectioned at 4 μm, and subjected to standard H&E staining after deparaffinization and dehydration. For IHC, the following primary antibodies were used: EGFR (RMA-0804), cytokeratin 7 (CK7; Z0071), thyroid transcription factor-1 (TTF-1; ZM-0270), Napsin A (MAB-0704), and Ki67 (ab16667). Antibodies were diluted 1:200 in 2% goat serum and incubated overnight at 4 °C. After phosphate-buffered saline (PBS) washes, slides were incubated with species-matched secondary antibodies for 60 min at room temperature. Signal detection was performed using a DAB horseradish peroxidase color development kit (P0202; Beyotime, China), followed by hematoxylin counterstaining of nuclei.
2.5. Detection of CAR expression by flow cytometry
EGFR-CAR-T cells (1 × 106) were resuspended in 100 μL PBS and stained with 1 μL anti-Myc-AF488 antibody (BioLegend, USA) and 1 μL allophycocyanin (APC)-Cy7 live/dead viability dye (BioLegend). The mixture was incubated on ice for 30 min protected from light, centrifuged at 300g for 5 min, discarded supernatant, and resuspended in 200 μL PBS. EGFR expression was analyzed using fluorescein isothiocyanate (FITC)-conjugated recombinant human EGFR protein (BioLegend). Dead cells were excluded by gating with Zombie NIR™ (BioLegend). Data were acquired using an ACEA NovoCyte flow cytometer and analyzed with NovoExpress software (v1.2; ACEA Biosciences, China).
2.6. Organoid dissociation and microwell seeding
Organoids were incubated with TrypLE™ Select at 37 °C for 15 min with gentle agitation every 5 min. The dissociated cells were washed twice with PBS and centrifuged at 400g for 5 min. Cells were then resuspended in LCOM at a final concentration of 2 × 105 cells·mL−1. The suspension was seeded into each microwell and allowed to settle at 37 °C for 60 min. Subsequently, 150 µL of LCOM was gently added along the well walls. The plates were transfer to a 37 °C, 5% CO2 incubator for continuous culture.
2.7. EGFR-CAR-T cells mediated organoid cytoxicity assay
Organoids were seeded in microwells as previously described and cultured for 48 h to establish mature structures prior to co-culture. LCOM was aspirated and replaced with X-VIVO 15 medium (Lonza Bioscience, Swiss). EGFR-CAR-T cells or control-T cells were added at 1:1 effector:target (E:T) ratio. Co-culture was maintained for 48 h under standard conditions (37 °C, 5% CO2). For visualization, organoids were labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) (red; C1991S; Beyotime), and T cells were labeled with 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO) (green; C1993S; Beyotime). The dyes were incubated at 37 °C for 10 min according to the manufacturer’s instructions. Image data were acquired using a DeltaVision Ultra imaging system (GE HealthCare, USA). Fluorescence co-localization and cell migration trajectories were analyzed with ImageJ software.
2.8. Cytokine release assay
The levels of cytokines (interferon-γ (IFN-γ), interleukin-2 (IL-2), and tumor necrosis factor-α (TNF-α)) in the co-culture supernatants were measured using human enzyme-linked immunosorbent assay (ELISA) kits (BioLegend) in accordance with the manufacturer’s instructions.
2.9. Live-cell imaging for cytotoxicity evaluation
To enable long-term dynamic monitoring, the culture medium was supplemented with propidium iodide (PI). Time-lapse imaging was performed using a Cytation5 system (Agilent BioTek, USA) to track interactions between organoids and T cells within the microwells.
2.10. Animals
All animal procedures were conducted in compliance with the ethical guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC International) and were approved by the Institutional Animal Care and Use Committee (IACUC) of the Clinical Medical College, Sichuan University (protocol No. 20231117010). Female non-obese diabetic (NOD)-scid IL-2 receptor gamma chain (IL-2Rγ) null (NSG) mice (6-8 weeks old) were obtained from Beijing Huafukang Biotechnology Co., Ltd. (China). Mice were housed under specific pathogen-free (SPF) conditions in the State Key Laboratory of Biotherapy’s Animal Center at Sichuan University, with autoclaved food and water and maintained on a 12 h light/dark cycle.
2.11. Establishment of the PDOX model
First-generation LCOs were cultured for 7 d, dissociated with TrypLE (37 °C, 5 min), and centrifuged (400g, 5 min). Cell pellets were resuspended in ice-cold 50% Matrigel (diluted in PBS). After homogenization, 100 μL of the suspension was subcutaneously injected into the right flank of NSG mice. Tumor growth was monitored every 3 d by caliper measurements (volume = length × width2 × 0.5). When tumor volumes reached 500-1000 mm3, mice were euthanized, and tumors were aseptically excised. For serial transplantation, tumors were minced into 2-3 mm3 fragments and re-implanted into secondary recipient mice under identical conditions.
2.12. EGFR-CAR-T efficacy assessment in PDOX model
After the second transplantation, tumor growth was monitored every 3 d via caliper measurements. Mice bearing tumors (40-70 mm3) were randomized into groups (n = 6 per group) and administered 5 × 106 control-T cells or EGFR-CAR-T cells via tail vein injection. Tumor volumes (calculated as length × width2 × 0.5) and body weights were recorded every 3 d as indicators of therapeutic efficacy and systemic toxicity, respectively. On day 10 post-infusion, mice were euthanized, and tumors were harvested for immunohistochemical analysis of T cell infiltration (CD3ε+/CD8+ cells), tumor cell proliferation (Ki67+), and target antigen retention (EGFR+). Major organs (heart, liver, spleen, lungs, kidneys, and gastrointestinal tract) were collected and subjected to H&E staining for histopathological evaluation of off-target toxicity.
2.13. Statistical analysis
All statistical analyses were performed using GraphPad Prism 8.0.2 (GraphPad Software, USA). Continuous data are presented as mean ± standard error of the mean (SEM). Comparisons between two groups were performed using unpaired Student’s t-test for normally distributed data or Mann-Whitney U test for non-parametric data. Comparisons among multiple groups were conducted using one-way analysis of variance (ANOVA) with Tukey’s post hoc test for parametric data or Kruskal-Wallis test with Dunn’s correction for non-parametric data. A two-tailed p-value < 0.05 was considered statistically significant.
3. Results
3.1. Fabrication of nested microwells
In this study, we designed a microdevice with a multi-layer nested structure for Matrigel-free organoid culture, specifically tailored to evaluate CAR-T cell cytotoxicity. The microdevice is a 22 mm diameter dish containing 9 nested wells and 63 microwells. Each microwell consists of two layers: an inverted truncated hexagonal pyramid structure in the upper layer and cylindrical microwells (600 µm in depth) in the bottom layer. Seven microwells are arranged into one array, forming a hierarchical structure termed a “nested well” (
Fig. 2(a)). The hexagonal close-packed arrangement of nested wells enhances spatial efficiency by minimizing dead space, ensures uniform cell distribution through consistent settling, and reduces cell loss via the confined geometry.
The microdevices were fabricated using an approach combining DLP-based 3D printing and molding. First, a photopolymerizable resin mold was produced via DLP-based 3D printing. The digital model was sliced and projected onto a digital micromirror device (DMD) chip, followed by rapid photopolymerization under 405 nm light exposure. PDMS was then cast onto the mold, thermally cured, and demolded to obtain the multi-layer nested-well microdevice (
Fig. 2(b)). Structural characterization identified: ① a close-packed honeycomb-like inlet at the top, ② a convergent inverted triangular intermediate layer that facilitates cell aggregation, and ③ cylindrical culture microwells at the bottom (
Fig. 2(c)). Compared with conventional cylindrical microwell arrays, this nested design significantly increased the number of cells in each microwell (
Fig. 2(d)). A key contributing factor is the inverted triangular design of the hexagonal well array (
Fig. 2(e)). The organoid formation process involved uniform sedimentation within 1 h, cell aggregation within 8 h, and organoid formation within 24 h (
Fig. 2(f)).
3.2. Matrigel-free tumor organoid culture using microwell arrays
We collected MPE samples from patients with advanced lung adenocarcinoma and processed them by centrifugation to obtain a cancer cell-enriched sediment. The cells were then seeded in Matrigel and cultured in optimized LCOM for 7 d. H&E staining and IHC demonstrated that the LCOs maintained morphological and molecular characteristics consistent with the original tumors. The LCOs exhibited key malignant features, including prominent nucleoli, significant nuclear heterogeneity, and acinar-like architectures formed by clustered tumor cells (
Fig. 3(a)). IHC analysis confirmed the expression of canonical lung adenocarcinoma markers (CK7, TTF-1, and Napsin A) in both primary tumor cells and their corresponding LCOs (
Fig. 3(b)). Collectively, these results demonstrate that our LCOs successfully preserves the histopathological and phenotypic characteristics of the original tumors.
Previous studies have shown that organoid formation in microwell cultures is influenced by factors such as well diameter and seeding cell density. To determine the optimal conditions, we first evaluated LCO viability in microwells of varying diameters (400, 600, and 800 µm). Organoid morphology was assessed by staining F-actin with Alexa Fluor™ 488 phalloidin (Fig. S1(a) in Appendix A). The results demonstrated that the 600 µm microwells achieved a significantly higher organoid formation rate than the other sizes on days 2, 4, and 6 of culture (Fig. S1(b) in Appendix A). Further analysis revealed that larger microwells accommodated more cells (Fig. S1(c) in Appendix A). Notably, the 600 µm microwells exhibited the highest cell density with a more uniform distribution, likely contributing to their superior organoid formation efficiency (Fig. S1(d) in Appendix A).
Seeding density also critically impacts organoid culture efficiency. Insufficient cell numbers hinder self-assembly, whereas excessive cell numbers lead to oversized organoids (> 250 µm), impairing nutrient diffusion [
14]. To evaluate this effect, we tested seeding densities of 1000, 2000, 4000, and 8000 cells per nested well. Within 8 h of seeding, cells aggregated and formed intercellular connections, progressing to complete organoid structures by 24 h (Fig. S2(a) in Appendix A). Higher seeding densities increased cell occupancy per microwell (Fig. S2(b) in Appendix A). After 24 h, organoid diameter correlated with cell concentration. The 1000-cell group yielded the smallest organoids, while the 8000-cell group produced significantly larger structures (Fig. S2(c) in Appendix A). However, by day 2, the 8000-cell group showed high organoid mortality [
12]. Thus, we selected 2000-cell per nested well as the optimal seeding density for subsequent LCO cultures.
Next, we cultured LCOs both in microwells and in Matrigel. Organoid structures formed within 24 h post-seeding, with progressive increases in diameter over the subsequent 5 d culture period (
Fig. 3(c)). No significant size differences were observed between the two groups during the first 3 d (
Fig. 3(d)), demonstrating the potential of the microdevice to enable rapid and standardized assessment of CAR-T cell efficacy. To further characterize microwell-cultured LCOs, we performed F-actin immunofluorescence staining. The results revealed interconnected cell clusters with defined 3D organoid morphology (
Fig. 3(e)).
3.3. In vitro evaluation of EGFR-CAR-T cells efficacy using Matrigel-free LCOs
Based on the EGFR protein expression observed in both original tumor specimens and their corresponding LCOs by IHC (
Fig. 4(a)), we constructed EGFR-CAR-T cells. These cells were generated using a retroviral vector encoding an EGFR-specific CAR, designed according to the established strategy previously reported by Ma et al. [
15] for human epidermal growth factor receptor 2 (HER2)-CAR-T cells. A detailed schematic of the construct is shown in
Fig. 4(b). Flow cytometric analysis confirmed efficient CAR expression, with 53.92% of T cells derived from healthy donors testing positive (
Fig. 4(c)). Control group were performed using unmodified T cells. After 48 h of LCO culture in either Matrigel or microwells, EGFR-CAR-T cells or control-T cells were introduced into the respective groups. Brightfield imaging analysis revealed that Matrigel-free LCOs cultured in microwells efficiently recruited EGFR-CAR-T cells, with significant apoptotic body formation observed around lysed LCOs within 24-48 h of co-culture (
Fig. 4(d)). In contrast, control-T cells showed no detectable cytotoxic activity in either culture system (
Fig. 4(d)). Importantly, in Matrigel cultures, LCOs became fully embedded within the matrix, resulting in complete physical segregation from both control-T and EGFR-CAR-T cells (
Fig. 4(d)). To determine the optimal E:T ratio for the microwell co-culture system, we conducted dose-response experiments. As shown in
Fig. 4(e) and Fig. S3 in Appendix A, cytotoxic efficiency exhibited a positive correlation with increasing E:T ratios. Notably, EGFR-CAR-T cells demonstrated significantly enhanced LCO elimination compared with control-T cells (
p < 0.05), and this cytotoxic superiority was maintained throughout the 48 h observation period (
Fig. 4(e)). Based on these results, we established a 1:1 E:T ratio for all subsequent functional assays. Using this optimized ratio, we evaluated the cytotoxic efficacy of EGFR-CAR-T cells against three independent patient-derived LCO lines on the microwell platform. Following 48 h of co-culture with either EGFR-CAR-T or control-T cells, residual LCOs were quantitatively assessed using automated image analysis. Time-lapse imaging at 24 h intervals was used to track absolute organoid counts, normalized values to baseline (0 h), and calculated condition-specific averages across triplicate LCO lines. Comparative analysis demonstrated that EGFR-CAR-T cells maintained superior cytotoxic activity compared with control-T cells (
Fig. 4(f)). Supernatants were collected from patient 2 (P2) and P3 patient-derived LCOs after 24 h of co-culture with either control-T or EGFR-CAR-T cells, and the secretion levels of IFN-γ, TNF-α, and IL-2 were analyzed by ELISA. The EGFR-CAR-T cells secreted significantly higher levels of these cytokines than the control-T cells (Fig. S4 in Appendix A). For real-time monitoring of cytolytic activity, we employed a dual-fluorescence system in which LCOs were labeled in red and T cells (EGFR-CAR-T and control-T) were labeled in green. Time-lapse confocal microscopy showed that EGFR-CAR-T cells specifically bound to LCOs and formed cytotoxic clusters around them (
Fig. 4(g)). This coordinated activity was further supported by fluorescence co-localization and migration trajectory analyses, which confirmed efficient tumor targeting and killing by the CAR-T cells (Fig. S5 in Appendix A). Complementing the live-imaging data, a PI-based cytotoxicity assay further verified the killing capacity of CAR-T cells, as evidenced by a substantial increase in PI influx into target LCOs following CAR-T cell attack (Video S1 in Appendix A).
3.4. In vivo antitumor efficacy of EGFR-CAR-T cells
To assess the therapeutic potential of EGFR-CAR-T cells, we established PDOX models by subcutaneously implanting LCOs into immunodeficient mice. Once tumor volumes reached 40-70 mm
3, mice received intravenous administration of either 5 × 10
6 EGFR-CAR-T or control-T cells (
n = 6 per group). Subsequent monitoring revealed significant growth inhibition in the EGFR-CAR-T group compared with controls (
Fig. 5(a)). No significant body weight fluctuations were observed in either group (
Fig. 5(b)). Histopathological evaluation using H&E staining showed no evidence of treatment-associated tissue damage in any of the examined organs, including the heart, liver, kidneys, lungs, intestines, and spleen (
Fig. 5(c)). To further evaluate treatment-induced immune infiltration and target antigen modulation, IHC staining was performed on harvested tumor tissues. Comparative analysis revealed that the EGFR-CAR-T group exhibited higher CD3
+ lymphocyte infiltration compared to controls (
Figs. 5(d) and
(e)). Furthermore, tumors from EGFR-CAR-T cell-treated mice showed significantly reduced EGFR expression relative to the control group (
Figs. 5(d) and
(f)).
4. Discussion
CAR-T cell therapy represents an innovative approach for lung cancer treatment; however, the lack of efficient preclinical predictive models remains a critical barrier to clinical translation. Although conventional models (e.g., cell lines [
16] and PDXs [
17]) have been widely used for CAR-T efficacy evaluation, their limitations in recapitulating human tumor biology are increasingly recognized. PDOs have therefore emerged as promising tools for therapy prediction. Previous studies have employed diverse strategies. Dijkstra et al. [
18] dissociated PDOs into single cells for T-cell interaction assays in Matrigel-free 2D co-culture systems. Schnalzger et al. [
19] assessed CAR natural killer (CAR-NK) cell cytotoxicity against colorectal cancer organoids cultured in Matrigel, even though the ECM may impede immune cell migration. Jacob et al. [
10] utilized glioblastoma organoids from surgical specimens in suspension tissue culture systems, which require large amounts of tumor tissue. In this study, we established a Matrigel-free microwell system to culture liquid biopsy-derived 3D LCOs from patients with advanced lung adenocarcinoma. These LCOs retained the morphological and molecular characteristics of the original tumors. Using this platform, we successfully predicted EGFR-CAR-T cell efficacy by co-culturing Matrigel-free LCOs with EGFR-CAR-T cells
in vitro.
Conventional PDOs generation methods rely heavily on Matrigel or other exogenous ECMs [
20]. These matrices create physical barriers that impede direct contact between cell-based therapeutics and organoids, severely limiting PDOs’ utility in cell-based drug testing. To address this limitation, Matrigel-free organoid culture strategies have been explored, including tissue slice suspension culture [
21], 2D monolayer culture [
18], and bioengineered microdevices (e.g., microwells [
11], [
14], [
22], microfluidics [
23], [
24], and microarrays [
25], [
26], [
27]). Among these approaches, microdevice-based platforms have emerged as particularly promising solutions for supporting scaffold-free organoid growth while enabling precise cell-drug interaction studies [
28], [
29]. Currently, commercially available products such as AggreWell™ and Elplasia® have been developed based on this matrix-free, microstructure-based culture approach. Designed in multi-well plate formats, they offer user-friendly protocols and reliable performance, holding significant promise for the large-scale production and application of organoids or spheroids. These cellular aggregates can also serve as “organ building blocks” for large-scale tissue construction via 3D bioprinting [
30], [
31]. However, the high-density layouts of their plates present practical challenges, including difficult cell seeding and cumbersome microscopic localization. To overcome these limitations, we introduced a subdivided nested-well architecture, in which each primary unit contains multiple microwells. This design enables simultaneous multi-well culture from a single seeding step, dramatically improving throughput. In addition, the clearly partitioned layout greatly facilitates sample localization and tracking using standard microscopy. In clinical drug screening scenarios where sample availability is extremely limited, our design directly addresses this bottleneck. The nested hexagonal honeycomb structure with inverted pyramidal features enables single-step seeding to generate sufficient material for seven parallel tests within a single unit. This geometry also enhances the initial cell seeding density within each microwell, demonstrating high potential for cultivating precious clinical samples.
Traditional microdevice fabrication techniques face significant challenges. Molding-based approaches require ultra-high-precision mold fabrication, leading to substantial manufacturing costs and technical complexity [
32]. Although soft lithography enables high-resolution microstructure production, its widespread adoption is hindered by time-consuming protocols, labor-intensive processes, and specialized technical requirements [
33], [
34]. In contrast, 3D printing technology offers a transformative solution for microstructure mold fabrication, enabling rapid production of complex, high-precision architectures [
35], [
36], [
37], [
38]. Its adoption in microfluidics and chip manufacturing has revolutionized these fields [
39]. In this study, we employed a hybrid strategy that integrates 3D-printed molds with conventional PDMS molding. This approach enables cost-effective production of high-precision microdevices using PDMS while overcoming the material limitations associated with direct 3D printing. Leveraging this microfabrication method, our approach enables rapid prototyping and multi-material integration, providing a reproducible and scalable strategy for developing microdevices suitable for clinical applications. Specifically, we exploited the inherent hydrophobicity of PDMS in conjunction with the self-assembly properties of tumor cells to fabricate a simple sheet-like microdevice. This device can be directly placed into standard cell culture plates for subsequent culture and drug screening applications, including the evaluation of CAR-T cell efficacy. The microwells successfully supported the growth of LCOs in a completely Matrigel-free system and enabled
in vitro prediction of EGFR-CAR-T cell therapeutic efficacy, highlighting their potential for future clinical translation. Moreover, the interconnected hexagonal inverted-cone structure of the microwell significantly reduced PDO loss during handling while maintaining optimal culture conditions for limited organoid samples. These microwells supported the growth of LCOs in a completely Matrigel-free system and allowed for
in vitro prediction of EGFR-CAR-T cell therapeutic efficacy.
5. Conclusions
In this study, we developed a microwell platform fabricated via rapid 3D printing and precision molding. This system enabled Matrigel-free culture of LCOs while preserving their original tumor characteristics. The efficient targeting and lysis of LCOs by EGFR-CAR-T cells demonstrated the antigen-specific cytotoxicity, and the EGFR-CAR-T cell efficacy was predicted in vitro through standardized co-culture assays. Our platform is a scalable and physiologically relevant approach for screening cell-based therapeutics, addressing a critical bottleneck in precision immunotherapy development. By eliminating Matrigel dependence and enabling high-throughput testing, this platform may provide a promising translational tool for evaluating and personalizing CAR-T therapies.
PDF
(3532KB)
