Development of a 3D-3 co-culture microbead consisting of cancer-associated fibroblasts and human umbilical vein endothelial cells for the anti-tumor drug assessment of lung cancer

Introduction

Lung cancer is one of the most prevalent cancers worldwide. According to the 2020 Global Cancer Observatory (GLOBOCAN) statistics, lung cancer contributes approximately 2.2 million new cases and 1.8 million deaths each year (1). Recent research has indicated that the occurrence and progression of lung cancer, as well as the development of drug resistance, are influenced not only by intrinsic regulatory mechanisms of the tumor cells themselves but also by the components of the tumor microenvironment (TME) (2). Due to the heterogeneity of the lung cancer microenvironment, some patients ultimately experience disease progression and exhibit resistance to initial therapeutic agents after prolonged treatment (3). Furthermore, the mechanism of resistance of lung cancer drugs varies due to different classes of drugs. Resistance to chemotherapy can come from changes in the DNA repair system, drug efflux, prosurvival signaling, cell cycle arrest, epigenetic regulation, upregulation of microRNA, upregulation of epithelial-mesenchymal transition (EMT) phenotype, and changes in cancer metabolism. Specifically with the TME, chemotherapy resistance can be triggered by upregulation of cancer-associated fibroblasts (CAFs), hypoxia, and PD-L1. Resistance to immunotherapy can come from an immunosuppressive TME in which there are high levels of regulatory T cells (Tregs), myeloid-derived suppressor cells (MDSCs), tumor-associated macrophages (TAMs), and CAFs, an oxidative environment that triggers hypoxia and affects effector T cells, neo-angiogenesis which inhibits the infiltration of effector immune cells and recruits Tregs, TAMs, and MDSCs, upregulation of alternative immune checkpoints such as LAG-3, TIGIT, and dysregulation of cellular and molecular pathways leading to resistance to anti-CTLA-4 and anti-PD-1 (4). Resistance to tyrosine kinase inhibitors (TKIs) can happen through on-target resistance such as through acquired T790M mutation in the ATP binding pocket of EGFR gene or off-target resistance that can bypass signaling such as MET amplification (5). In order to better study the impact of the TME on the sensitivity to anticancer drugs, various three-dimensional (3D) tumor culture models have been extensively developed. Current clinically utilized 3D lung cancer culture models can be broadly categorized into scaffold-free 3D tumor models, scaffold-based 3D tumor models (6), bioreactors, microfluidic chips, patient-derived tumor xenografts (PDTX) (7), and organoids (8). Each type of 3D lung cancer model has its own advantages. For example, scaffold-based 3D culture systems provide structural support for cell adhesion, and by selecting different biomaterials (proteins, polysaccharides, and decellularized matrices) based on various extracellular matrix (ECM) components, it is possible to somewhat mimic the tumor’s ECM (9). Meanwhile, organoids can be cultured in vitro from normal or malignant human biopsy tissues through enzymatic or mechanical digestion, preserving the in vivo characteristics of the original tissue, including the heterogeneity of cancer cells (10,11). Although these various 3D lung cancer tumor models have distinct advantages, they still fall short in replicating the complexity and heterogeneity of the in vivo TME. This limitation arises from a lack of crucial ECM and cellular components within the TME, such as CAFs, tumor endothelial cells (TECs), and infiltrating immune cells. CAFs play a major role in growth, invasion, and metastasis through ECM remodeling, angiogenesis, interaction with immune cells, intracellular communication, and resistance from chemotherapy (12). TECs are involved in promoting tumor angiogenesis, trafficking of immune cells, T cell priming, and formation of tertiary lymphoid structures (13). Consequently, constructed lung cancer 3D tumor models struggle to accurately recreate cell-ECM and cell-cell interactions, which may negatively impact the development of novel drugs for lung cancer and the personalization of cancer treatment (14,15).

To address this issue, recent studies have begun to focus on 3D co-culture models that aim to replicate the interactions among various cellular components within the TME and their contributions to the mechanisms of tumor drug resistance. Zhu et al. observed that CAFs isolated from tissues resistant to TKIs, when co-cultured with parental lung cancer cells, promoted osimertinib resistance (16). In the case of hepatocellular carcinoma (HCC). Lim et al. utilized a PDTX model combined with human umbilical vein endothelial cells (HUVECs) cultured in hyaluronic acid (HA) hydrogel. They discovered that the co-culture model significantly activated tumor necrosis factor (TNF) signaling pathways and complement and coagulation cascades, revealing the complex vascular secretory cross-talk between HCC and endothelial cells (ECs) and its impact on the tumor immune microenvironment (17). However, the co-culture of only two cell types in 3D models struggles to adequately reproduce the interactions among all cell types present in the TME. Therefore, based on these findings, further development of tri-cell co-culture systems has been undertaken. A lung cancer study showed that when lung cancer cells were co-cultured with fibroblasts (MRC-5), their growth rate significantly increased; furthermore, if the interaction between MRC-5 and HUVECs was introduced into this co-culture system, it would further promote the proliferation and invasive growth of lung cancer cells (18). Lazzari et al. established a 3D co-culture model for pancreatic cancer that included MRC-5 and HUVECs, further demonstrating the influence of fibroblasts and EC components within the TME on the chemotherapeutic sensitivity of cancer cells (19). Thus, incorporating various stromal cells from the TME into co-culture systems is a promising strategy. Specifically, the introduction of HUVECs and CAFs into 3D cell cultures can facilitate cell-ECM interactions and mimic vasculature, thereby achieving a more accurate replication of the TME (20,21). However, most of the studies mentioned above are confined to 3D models built with tumor cell lines, leaving research on patient-derived primary lung cancer cells, CAFs, and HUVEC-containing 3D co-culture systems still unexplored.

Compared to traditional 3D tumor models, encapsulation technology represents a cutting-edge method for 3D cell culture, particularly the use of sodium alginate (Alg)-HA hydrogel microbeads. This approach signifies a promising microscale 3D cell culture technology that can provide an environment more physiologically similar to the ECM of tumors, thereby more effectively modulating cellular behaviors, including morphology, proliferation, differentiation, and functional expression (22,23). Our group previously reported a 3D model of sodium Alg-gelatin microbeads for lung cancer in vitro (24). Building upon this foundation, we employed conditional reprogramming (CR) techniques to isolate conditionally reprogrammed lung cancer cells (CRLCs) and CAFs from lung cancer patients, integrating HUVECs into the co-culture system to construct a 3D-3 co-culture model using Alg-HA microbeads, thus preserving the heterogeneity of the TME from lung cancer patients.

Using 3D hydrogel microbeads, we investigated differences in the chemotherapy sensitivity of primary cells from different lung cancer patients under various conditions: two-dimensional (2D) solo culture, 3D solo culture, and co-culture conditions. The first-line chemotherapy drugs analyzed included cisplatin, carboplatin, paclitaxel, vinorelbine, and gemcitabine, along with EGFR-TKIs such as gefitinib, osimertinib, and almonertinib. We also explored the transcriptional changes in lung cancer cells induced by tumor-stromal interactions. Furthermore, we revealed a close relationship between the increased expression of lung cancer stemness markers (such as ALDH1A1, NANOG, and SOX9) by CAFs and HUVECs and the modulation of lung cancer cell resistance to therapy. We present this article in accordance with the MDAR reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-525/rc).

Methods

Materials

The high-glucose variant of Dulbecco’s modified Eagle medium (DMEM) and fetal bovine serum (FBS) were sourced from Thermo Fisher Scientific, based in Waltham, MA, USA. Calcium chloride (CaCl₂) was procured from Shanghai Aladdin Biochemical Technology Co., Ltd., located in Shanghai, China. Additionally, collagenase IV was acquired from MP Biomedicals in Santa Ana, CA, USA. The 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) cell proliferation assay kit was obtained from Promega (Madison, WI, USA). Alg and HA were purchased from Sigma-Aldrich (St. Louis, MO, USA). Calcein-AM/propidium iodide (PI) double-staining kit, radioimmunoprecipitation assay (RIPA) lysis buffer, and phenylmethylsulfonyl fluoride (PMSF) were purchased from Beyotime Biotech Inc. (Beijing, China). The ROCK inhibitor Y27632 (carboplatin, cisplatin, paclitaxel, vinorelbine, gemcitabine, osimertinib, gefitinib, and almonertinib) was purchased from MedChemExpress (MCE, Monmouth Junction, NJ, USA). Primary antibodies against glyceraldehyde-3-phosphate dehydrogenase (GAPDH), NANOG, ALDH1A1, OCT-4, C-Myc, CD24, CD44, TP53, HIF-α, and SOX-9 were used. GAPDH (CAT: 2118), SOX-9 (CAT: 82630), OCT-4 (CAT: 2750), C-Myc (CAT: 5605), CD44 (CAT: 3570), TP53 (CAT: 2527), HIF-α (CAT: 36169), and SOX-9 (CAT: 82630) were purchased from Cell Signaling Technology (CST, Danvers, MA, USA). ALDH1A1 (CAT: ab134188) was purchased from Abcam (Cambridge, MA, USA). NANOG (CAT: 14295-1-AP) and CD24 (CAT: 10600-1-AP) were purchased from the Proteintech Group (Rosemont, IL, USA). Goat anti-mouse antibody (CAT: 7076) and goat anti-rabbit antibody (CAT: 7074) antibodies were purchased from CST. A 5× loading buffer (Bio-Rad, Hercules, CA, USA) was added to the protein samples at a 1:4 dilution and mixed. Swiss-3T3-J2 cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA, CRL-1658). The HUVECs cell line (CAT: iCell-h110) was purchased from iCell Inc. (Shanghai, China), and maintained in an endothelial cell medium (ECM, CAT: 1001, ScienCell, Carlsbad, CA, USA).

Preparation of Alg-HA hydrogel precursor

A 4% (w/v) Alg solution (CAT: A1112, Sigma-Aldrich) was autoclaved, and a 0.1% (w/v) HA aqueous solution (CAT: 53747, Sigma-Aldrich) was sterilized using a 0.22 µm filtration membrane (Merck Millipore, Darmstadt, Germany). Both solutions were prepared in ultrapure water and subsequently stored at 4 ℃ in a refrigerator for future use.

Patient-derived lung tumors dissociation and cell culturing

Newly harvested lung tumor specimens were gathered from individuals who had undergone surgical excision at the First Affiliated Hospital of Sun Yat-sen University in Guangzhou, China. The research adhered to the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Board of the First Affiliated Hospital of Sun Yat-sen University (No. [2018]-153), and all patients gave their informed consent. The surgical extracts were kept in sanitized centrifuge containers for subsequent examination. Details of the patients are detailed in Table 1. In this study, tumor tissue samples were surgically resected from 29 lung cancer patients. Through CR culture techniques, tumor cells from 24 cases were successfully cultured in vitro, yielding a success rate of 82.75%.

The tumor tissue samples, obtained from patients, were first removed from their preservation solution and briefly sterilized by dipping them in 95% ethanol for no more than three seconds. Afterward, they were rinsed three times with phosphate-buffered saline (PBS), sectioned into small 2–3 mm fragments, and spun at 300 ×g for 5 minutes. For digestion, each sample was treated with a solution of advanced DMEM/F12 medium, enhanced with 2.5 mg/mL collagenase IV and 0.1 mg/mL DNase I, then placed in a 37 ℃ water bath for 6 hours. Following digestion, the mixture was strained through a 100-µm cell mesh sieve (Shandong Biologix Biotech Co., Ltd., Jinan, China), centrifuged again at 300 ×g for 5 minutes, and treated with erythrocyte lysate for an additional 5 minutes. The isolated cells were then pelleted via centrifugation at 300 ×g for 5 minutes and suspended in freshly prepared F+Y medium. The F+Y medium was made by enriching complete medium with 10% FBS, 100 µg/mL penicillin-streptomycin (Gibco, Waltham, MA, USA), GlutaMAX (100×, Gibco), and a 3:1 ratio of conditioned medium. This mixture was further supplemented with 25 ng/mL hydrocortisone (Solarbio Technology Co., Ltd., Beijing, China), 8.4 ng/mL cholera toxin (Sigma-Aldrich, CAT: C8052), 10 ng/mL human recombinant epidermal growth factor (EGF; MCE), 5 µM Y-27632 (MCE), 10 µg/mL gentamicin (MCE), and 5 µg/mL amphotericin B (Gibco).

Cell culturing

CAFs were digested following the aforementioned method for isolating primary cells from lung cancer patient tissues (patient ID: LC-17), and the resulting cell suspension was cultured in Petri dishes using CAFs medium to obtain CAFs. For example, using 225 mL of DMEM, 225 mL of F12 medium, 50 mL of FBS, 5 mL of penicillin-streptomycin (100×), and 0.5 mL insulin (5 mg/mL), the CAFs medium had a total volume of 505 mL.

Mouse embryonic fibroblast cell line 3T3-J2 was cultured in high-glucose DMEM supplemented with 10% (v/v) newborn calf serum (NBCS; Gibco), 100 IU/mL penicillin, and 100 mg/mL streptomycin. In the CR system, 3T3-J2 cells were mitotically inactivated by irradiation (30 Gy). The HUVECs cell line was purchased from iCell Inc., China, and maintained in an ECM (ScienCell).

All cells were maintained in a humidified incubator at 37 ℃ with 5% CO₂.

3D micro-beads production

This strategy is based on the Alg microbeads technology previously described by our team and has been summarized graphically (25). The Alg solution (4%) and HA aqueous solution (0.1%) were mixed in a ratio of 4:1 for later use. The cell pellet was resuspended in a sodium Alg-HA mixture, adjusting the cell density to 5×10⁵–7×10⁵/mL. Next, the Alg-HA hydrogel mixed with the cells was extruded through 1-mL syringes and slowly dropped into a CaCl₂ solution (2%, autoclaved). The solution was allowed to stand for 5 minutes to ensure adequate cross-linking. After washing three times with aseptic ultrapure water, the 3D microbeads were cultured in F+Y media. CRLCs, CAFs, and HUVECs were co-cultured in a 2:2:1 ratio. The cells were collected by adding 55 mM sodium citrate solution to dissolve 3D microbeads.

Rheological determination

Rheological measurements were made using a rotary rheometer (ARES-G2; TA Instruments, New Castle, DE, USA). At 25 ℃, the Alg-HA hydrogel microbeads were measured in the mode of (strain: 0.1–40%; angular frequency: 6.28 rad/s; clearance: 0.84 mm) shear modulus.

Cytocompatibility of the Alg-HA hydrogel

The CRLCs were centrifuged and then suspended in an Alg-HA hydrogel precursor to achieve a final cell density of 5×10⁵ cells/mL, forming a uniform cell-hydrogel blend. Cell proliferation was assessed via MTS assay on days 1, 3, 5, and 7. For the assay, individual cell microbeads were transferred into 96-well plates containing 10 µL of MTS reagent and 90 µL of fresh medium. Following a 4-hour incubation period, absorbance readings were taken at 490 nm using a Thermo Fisher Scientific microplate reader. In parallel, a 2D culture was prepared by seeding 3,000 cells per well in 96-well plates. After allowing the cells to adhere, the medium was replaced with a mixture of 10 µL MTS solution and 90 µL fresh medium. The plates were then incubated at 37 ℃ for 2 hours before measuring the optical density (OD). Cell viability was determined by comparing the OD values of the experimental group to those of the control group.

Viability staining

To assess cell distribution and viability, a calcein-AM/PI dual-staining assay was conducted. Following a 2-hour incubation at 37 ℃, the microbeads were examined under a Leica confocal microscope (Wetzlar, Germany) to visualize live and dead cell staining patterns.

Histology and immunofluorescence

3D-3 microbeads were lysed using 55 mM sodium citrate solution, followed by centrifugation at 300 ×g for 5 minutes to collect the lung cancer cells, which were washed three times with PBS, and agarose gel was added to encapsulate the cellular precipitates. The 3D-3 microbeads and their corresponding parental tumors tissues were fixed in 4% paraformaldehyde, followed by paraffin embedding, sectioning, deparaffinization, dehydration, and hematoxylin and eosin (H&E) staining. Lung cancer tissues, 2D-LC (2D cultured lung cancer cells), CAFs, and HUVECs were fixed with 4% paraformaldehyde. The samples were then washed thrice with PBS and permeabilized with 0.5% Triton X-100 for 20 minutes. After permeabilization, the samples were washed thrice with PBS. Subsequently, immunostaining blocking solution (Beyotime, P0102) was applied for 1 hour. The samples were then incubated overnight with diluted primary antibodies including anti-NANOG (1:500, CST, 14295-1-AP), anti-cytokeratin 7 (CK7; 1:200, CST, 4465T), anti-alpha-smooth muscle actin (α-SMA; 1:500, CST, 19245S), and anti-CD31 (1:500, Invitrogen, Carlsbad, CA, USA, MA3100). Following this, the samples were incubated for 1 hour with goat anti-rabbit secondary antibody (1:500, Invitrogen, A11008) in immunofluorescence staining secondary antibody dilution buffer (Beyotime, P0108), followed by three washes of 5 minutes each with PBS. 4',6-diamidino-2-phenylindole (DAPI; Solarbio, S2110) was used for the nuclear staining. 3D-3 microbeads and their corresponding parental tumors were analyzed using antibodies targeting thyroid transcription factor-1 (TTF-1) and CK7 for adenocarcinoma (ADC). Immunofluorescence imaging was performed using a LSM780 confocal microscope (ZEISS, Oberkocken, Germany).

Western blot

The protein expression was evaluated through a western blot technique. To summarize, the cells were subjected to two washes with PBS, followed by lysis in RIPA buffer with PMSF and protease inhibitors chilled at 4 ℃ for half an hour. The mixture was then shaken and spun at 12,000 rpm for 15 minutes. The resulting supernatant was diluted with 5× loading buffer and boiled at 100 ℃ for 5 minutes. The samples were promptly added to the wells at a concentration of 20 µg protein each. The proteins were separated using a 10% gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes via electro-blotting. These membranes were then blocked with 5% bovine serum in tris-buffered saline with Tween-20 (TBST) for an hour at room temperature. Post-blocking, the membranes were washed with TBST three times, each for 10 minutes, and incubated overnight at 4 ℃ with a cocktail of primary antibodies, including GAPDH, NANOG, ALDH1A1, OCT-4, c-Myc, and SOX-9. The membranes were washed again with TBST three times before being exposed to secondary antibodies for an hour. The protein expression was detected via enhanced chemiluminescence using Thermo Fisher Scientific’s enhanced chemiluminescence (ECL) reagents on a Bio-Rad ChemiDoc MP imaging system.

RNA sequencing (RNA-seq)

In this part of the transcriptome sequencing experiment, CRLCs were indirectly co-cultured using Transwell membranes (0.1 µm pore size, Biofil, Guangzhou, China) (Figure S1A). Mesenchymal cell (CAFs, HUVECs, or CAFs-HUVECs) microbeads were placed in the upper chamber and Mono-LC (lung cancer cells cultured in 3D conditions) microbeads were placed in the lower chamber to allow intercellular interactions across the membrane. After 7 days of culture, the microbeads were dissociated and the CRLCs in the lower chamber were collected for RNA extraction and transcriptome sequencing.

In summary, messenger RNA (mRNA) was isolated from total RNA using poly-T oligo-conjugated magnetic beads. The RNA was fragmented using divalent cations at elevated temperatures in NEBNext first strand synthesis reaction buffer (5×) [New England Biolabs (NEB), Ipswich, MA, USA]. First-strand complementary DNA (cDNA) was generated with random hexamer primers and M-MuLV reverse transcriptase (RNase H⁻, NEB), followed by second-strand cDNA synthesis employing DNA polymerase I and RNase H. Any remaining overhangs were blunted through exonuclease/polymerase activity. Following 3' end adenylation of DNA fragments, a hairpin loop-structured NEBNext Adaptor was ligated for hybridization preparation. cDNA fragments of 250–300 bp were selectively purified using the AMPure XP system (Beckman Coulter, Brea, CA, USA). Size-selected, adaptor-ligated cDNA was then treated with 3 µL USER enzyme (NEB) at 37 ℃ for 15 minutes, followed by 95 ℃ for 5 minutes prior to polymerase chain reaction (PCR) amplification. PCR was conducted with Phusion high-fidelity DNA polymerase (NEB), universal PCR primers, and index (X) primer. The resulting PCR products were purified (AMPure XP system), and library quality was evaluated using the Agilent Bioanalyzer 2100 system (Agilent, Santa Clara, CA, USA).

Index-coded samples were clustered on a cBot Cluster Generation System with the TruSeq PE Cluster Kit v3-cBot-HS (Illumina, San Diego, CA, USA) following manufacturer protocols. Post-clustering, libraries were sequenced on an Illumina NovaSeq platform to generate 150 bp paired-end reads.

Bioinformatics analysis

Raw data (raw reads) in fastq format were first processed using in-house Perl scripts. The index of the reference genome was built using STAR v2.7.8a and paired-end clean reads were aligned to the reference genome using STAR v2.7.8a (26).

FeatureCounts v2.0.4 was used to count the read numbers mapped to each gene (27). Subsequently, the fragments per kilobase of transcript per million mapped reads (FPKM) for each gene were computed according to the gene’s length and its corresponding read count. Comparative expression analysis between the two experimental groups/conditions (each with two biological replicates) was conducted using the DESeq2 R package (version 1.16.1). DESeq2 implements statistical methods for identifying differentially expressed genes (DEGs) from digital expression data through a negative binomial distribution model. The obtained P values underwent adjustment via the Benjamini-Hochberg procedure to regulate the false discovery rate. Genes meeting the criteria of an adjusted P value <0.05 and an absolute log₂fold change >1, as identified by DESeq2, were designated as differentially expressed.

Differential expression analysis under the two conditions was performed using the edgeR R package (3.18.1) (28). The P values were adjusted via the Benjamini-Hochberg method. A corrected P value threshold of 0.05 and an absolute fold change of 2 were established as the criteria for statistically significant differential expression. Gene Ontology (GO) enrichment analysis of DEGs was performed using the clusterProfiler R package, with correction for gene length bias. GO terms exhibiting a corrected P value below 0.05 were deemed significantly enriched by DEGs. Additionally, the clusterProfiler R package was employed to evaluate the statistical enrichment of DEGs in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways.

Drug evaluation

Drug sensitivity testing (DST) was conducted on primary tumor cells from 10 patients who underwent surgical resection, comparing the efficacy of various drugs such as cisplatin, carboplatin, paclitaxel, vinorelbine, gemcitabine, gefitinib, osimertinib, and almonertinib across different experimental groups, as shown in Table 2. CRLCs, CAFs, and HUVECs were directly co-cultured to form 3D microcapsules, which were subsequently utilized for drug sensitivity assessment. After 5 days of culture, the 3D microcapsules reached the logarithmic growth phase, following which they were transferred to a 96-well plate for 3 days of drug incubation, using DMEM with 1% dimethyl sulfoxide (DMSO) as a control. For the 2D cells, approximately 30,00 cells per well were cultured in a 96-well plate for 24 hours before being subjected to DST. Drug selection criteria: patients diagnosed with EGFR-sensitive mutations during postoperative stages IB to II should receive osimertinib as adjuvant targeted therapy. For those in stages IIA or IIB, a platinum-based adjuvant chemotherapy regimen is the recommended treatment approach.

Statistical analysis

All the data were expressed as the mean ± standard deviation of at least three individual biological experiments. Comparisons between two groups were performed using Student’s t-test, whereas comparisons among multiple groups were conducted using one-way analysis of variance (ANOVA). Significance levels were set at *, P<0.05; **, P<0.005; ***, P<0.001; and ns, not significant (P>0.05).

Results

Alg-HA 3D micro-beads support CRLCs and stromal cells co-culture

We established a 3D co-culture system for non-small cell lung cancer (NSCLC) that incorporates tumor stromal components. The average diameter of the Alg-HA 3D microbeads was 1.5±0.4 mm, containing three types of cellular components: CRLCs, CAFs, and HUVECs (Figure 1A). After 7 days of culture, multiple spheroids and cell clusters were observed within the 3D co-culture microbeads. To identify the various cell types, we employed CK7 to label CRLCs, α-SMA to label CAFs, and CD31 to label HUVECs (Figure 1B), thereby confirming the accuracy of identifying the isolated cells from tumor tissues as CRLCs, CAFs, and HUVECs.

Figure 1 Characterization of the biological properties of sodium Alg-HA 3D micro-beads. (A) Bright-field microscopy images showing the morphology of CRLCs, CAFs, and HUVECs, as well as the spheroid formation of cancer cells within the micro-beads. The red arrow indicates 3T3-J2 mouse fibroblasts, and the black arrows indicate CRLCs. Scale bar: 250 µm. (B) Expression of CK7, α-SMA, and CD31 in CRLCs, CAFs, and HUVECs. The cells were stained with DAPI (blue) to indicate the nuclei, anti-CK7 antibody (green) for CRLCs, anti-α-SMA antibody (violet) for CAFs, and anti-CD31 antibody (red) for HUVECs. Scale bar: 50 µm (upper) and 20 µm (lower). (C,D) The relationship among oscillatory strain, deformation percentage, and energy storage modulus of sodium Alg micro-beads at varying concentrations. (E) The effect of the F+Y to ECM medium ratio on the proliferation of CRLC, CAF, and HUVEC co-cultures in 2D environments. (F) The impact of irradiated cancer fibroblasts on the proliferation of lung cancer progenitor cells. (G) Proliferation levels of CRLCs, CAFs, and HUVECs within micro-beads measured from culture days 1 to 7. (H) Visualization of live cells (green) and dead cells (red) using calcein-AM and PI staining. Scale bar: 250 µm. 3D-2-LC_1: CRLCs-CAFs co-culture; 3D-2-LC_2: CRLCs-HUVECs co-culture; 3D-3-LC: CRLCs-CAFs-HUVECs co-culture. 2D, two-dimensional; 3D, three-dimensional; α-SMA, alpha-smooth muscle actin; Alg, alginate; CAFs, cancer-associated fibroblasts; CK7, cytokeratin 7; CRLCs, conditionally reprogrammed lung cancer cells; DAPI, 4',6-diamidino-2-phenylindole; ECM, extracellular matrix; HA, hyaluronic acid; HUVECs, human umbilical vein endothelial cells; LC, lung cancer; PI, propidium iodide.

To further replicate the stiffness of lung cancer tissue, we prepared 3D microbeads using sodium Alg solutions at concentrations of 1%, 2%, and 4%. The storage modulus of the Alg-HA microbeads increased with the concentration of sodium Alg; microbeads made from 1% and 2% sodium Alg had storage module lower than 4 kPa, whereas those made from 4% sodium Alg approached 12 kPa, closely resembling the viscoelastic characteristics of lung tumor tissue (Figure 1C). Furthermore, the microbeads produced with 4% sodium Alg exhibited higher elasticity and demonstrated greater deformational elasticity (Figure 1D).

To optimize the culture medium for 3D-3 co-culture, we tested different ratios of F+Y media. Under these conditions, there were no significant differences in the proliferation rates of CRLCs, CAFs, and HUVECs (Figure 1E). In this study, two-dimensionally cultured 3T3-J2 cells were irradiated with a dose of 30 Gy. The treatment modulated the cytokine secretion profile of 3T3-J2 cells, subsequently influencing tumor cell growth. To further investigate, CAFs were irradiated with an equivalent dose. The experimental results indicated that this irradiation protocol similarly failed to significantly alter cancer cell proliferation rates (Figure 1F). To evaluate the biocompatibility of Alg-HA hydrogels, we performed live/dead staining and MTS assays on CRLCs, CAFs, and HUVECs cultured within 4% Alg-HA microbeads. After 7 days, Calcein-AM/PI staining confirmed high cell viability, while MTS assay revealed good cell proliferation in hydrogels, especially in3D-2-LC_1 (CRLCs-CAFs co-culture) and 3D-2-LC_2 (CRLCs-HUVECs co-culture) groups (Figure 1G,1H).

These findings suggest that hydrogel microbeads constructed with 4% sodium Alg-10 mg/mL HA at a ratio of 4:1 can effectively support cell growth and exhibit excellent cytocompatibility and suitability for the development of in vitro tumor models without toxic effects.

3D-3 micro-beads maintain the histological features and genomic characterization of the parental tumors

To evaluate the capability of 3D co-culture microbeads in maintaining the histological and genomic characteristics of tumor tissues, we conducted a detailed morphological characterization of the tumor cells within hydrogels microbeads and performed RNA-seq analysis. H&E staining and morphological analysis indicated that lung cancer cells in the 3D-3 co-culture microbeads retained the cellular and nuclear atypia characteristic of the parental tumor tissue, and the microbeads in this model formed spherical structures resembling in vivo tumor tissue (Figure 2A). Additionally, we assessed the expression of key biomarkers of lung ADC, such as CK7 and TTF-1, using immunofluorescence staining. These markers were well preserved within the microbeads compared to lung cancer tissue, further supporting the similarity of this 3D model in replicating the original tumor features (Figure 2B).

Figure 2 Transcriptomic correlation analysis of CRLCs in Alg-HA micro-beads across different groups. (A) Bright-field microscopy images showing the structure of 3D-3 micro-beads. Histological stain showing the structure and status in 3D-3 micro-beads and primary tumors. The arrows point to spherical cell clusters. Scale bar: 100 µm. (B) Immunofluorescence images showing the ADC related markers (CK7 and TTF-1) in primary tumors and 3D-3 micro-beads. Scale bar: 50 µm. (C) PCA of RNA-seq data comparing tumor tissue, 2D-LC, Mono-LC, 3D-2-LC_1, 3D-2-LC_2, and 3D-3-LC conditions. (D) Volcano plots illustrating upregulated (red) and downregulated (blue) genes in pooled 3D-3-LC relative to 3D-2-LC_1 and 3D-2-LC_2. (E) Volcano plots and intersectional clustering were used to analyze the DEGs related to tumor stem cells in primary lung cancer cells under both 3D-3 co-culture and 3D mono-culture microbeads. (F) Dot plot shows the up-regulated Reactome pathways (FDR <0.1) enriched for different contrasts. The size of the dot is based on gene count enriched in the pathway, and the color of the dot shows the pathway enrichment significance. (G) Differences in gene expression between lung cancer tissues and the 3D-3-LC model. The heatmap shows the differences in gene expression levels related to the extracellular matrix-receptor interaction pathway between lung cancer tissues and the 3D-3-LC model. Tissue: lung cancer tissue; 2D-LC: 2D cultured lung cancer cells; Mono-LC: lung cancer cells cultured in 3D conditions; 3D-2-LC_1: CRLCs-CAFs co-culture; 3D-2-LC_2: CRLCs-HUVECs co-culture; 3D-3-LC: CRLCs-CAFs-HUVECs co-culture. 2D, two-dimensional; 3D, three-dimensional; ADC, adenocarcinoma; Alg, alginate; BF, bright-field microscopy; CAFs, cancer-associated fibroblasts; CK7, cytokeratin 7; CRLCs, conditionally reprogrammed lung cancer cells; DEGs, differentially expressed genes; ECM, extracellular matrix; FDR, false discovery rate; H&E, hematoxylin and eosin; HA, hyaluronic acid; HUVECs, human umbilical vein endothelial cells; LC, lung cancer; PC, principal component; PCA, principal component analysis; RNA-seq, RNA sequencing; TTF-1, thyroid transcription factor-1.

Principal component analysis (PCA) of the transcriptomic features of CRLCs from different co-culture groups revealed distinct clustering of lung cancer tissues from different patients, highlighting significant transcriptomic variation among lung cancer samples. In contrast, 2D lung cancer cells derived from different patients clustered closely together, indicating a high degree of transcriptomic similarity among lung cancer cells from different patients under 2D conditions (Figure 2C). The volcano plot illustrated subtle differences in gene expression among the 3D-2-LC_1, 3D-2-LC_2, and 3D-3-LC (CRLCs-CAFs-HUVECs co-culture) groups. This suggests that both 3D-2 co-culture and 3D-3 co-culture displayed similarities in their influence on the transcriptomic characteristics of tumor cells (Figure 2D). Next, we analyzed the expression of 12 genes previously identified as specific markers for lung cancer stem cells (CSCs) in different 3D models. Volcano plot analysis demonstrated that the four genes—NANOG, EpCAM, SOX2, and ALDH1A1—exhibited significantly higher expression levels in the 3D-3-LC and Mono-LC culture systems (with corresponding data points clustered in the orange right-hand section of the heatmap). In contrast, their expression was notably lower in the 2D-LC culture system (with data points concentrated in the blue left-hand region of the heatmap) (Figure 2E). Conversely, CD44 and ABCG2 showed downregulation in both 3D-3-LC and Mono-LC (Figure 2E). Subsequently, we conducted gene set enrichment analysis (GSEA) to explore the differential enrichment of tumor-related signaling pathways across the different groups. Compared to the 2D-LC model, the KEGG enrichment pathways in the 3D microbeads model primarily involved cell adhesion molecules, ECM-receptor interactions, cancer pathways, the PI3K-Akt signaling pathway, AGE-RAGE signaling pathway in diabetic complications, and regulation of actin cytoskeleton. Notably, the leukocyte transendothelial migration signaling pathway was enriched in the 3D-2-LC_2 and 3D-3-LC groups (Figure 2F).

Finally, we compared the differences in genes associated with ECM reconstruction pathways between lung cancer tissues from four patients (LC-18, LC-19, LC-21, and LC-22) and their corresponding 3D-3 microbeads. Finally, we performed a comparative analysis of ECM remodeling pathway-related gene expression patterns between tumor tissues from four lung cancer patients (LC-18, LC-19, LC-21, and LC-22) and their matched 3D-3 microbead culture systems. The results revealed distinct expression profiles of integrin family genes and collagen-related genes among patient tumors: LC-18 (KRAS-mutated) and LC-21 (ERBB2-mutated) shared nearly identical gene expression signatures, clustering together, whereas LC-19 (ALK-mutated) demonstrated expression patterns resembling those of EGFR-mutated cases, forming a separate cluster. Importantly, lung cancer cells cultured in 3D-3 microbeads consistently showed downregulated expression of both integrin genes (ITGA/ITGB) and collagen-related genes (Figure 2G). These observations demonstrate substantial heterogeneity in ECM remodeling regulation across lung cancers harboring different driver mutations.

In summary, the 3D-3 co-culture microbeads we constructed preserved the histological and genomic characteristics of the original cancer tissue. All RNA-seq data will be made publicly available on the National Center for Biotechnology Information (NCBI) database, with BioProject ID PRJNA1182023.

Personalized drug evaluation

To evaluate the consistency of drug sensitivity in the 3D-3 hydrogel microbeads with treatment responses in tumor patients in vivo, we conducted DST on multiple patients and nine drugs, determining their half-maximal inhibitory concentration (IC₅₀). The results indicated that there were no statistically significant differences in the IC₅₀ values across the experimental groups when examining the impact of stromal cell types (CAFs and HUVECs) on the sensitivity to chemotherapeutic agents such as cisplatin, carboplatin, and paclitaxel. This lack of difference suggests that the regulatory effects of stromal cells on the sensitivity to these chemotherapeutic drugs are limited. This phenomenon may be attributed to patient heterogeneity, as individual responses to drugs can vary significantly.

Further analysis showed that while the addition of CAFs and HUVECs increased the IC₅₀ values of vinorelbine and gemcitabine, in the 3D-3-LC model, the IC₅₀ values of these drugs were lower than those in the co-culture groups with CAFs or HUVECs, and were closer to the levels observed in the Mono-LC group. For targeted therapies like gefitinib, osimertinib, and almonertinib, the IC₅₀ values were similar between 2D-LC and Mono-LC, suggesting that the effectiveness of these drugs was minimally impacted by the dimensionality of cell culture. In 3D co-culture settings, gefitinib, osimertinib, and almonertinib generally showed higher IC₅₀ values, and the presence of CAFs and HUVECs markedly decreased the drug sensitivity of tumor cells. Importantly, in the 3D-3-LC group, the IC₅₀ values for these targeted therapies peaked, clearly demonstrating that CAFs and HUVECs contributed to increased drug resistance (Figure 3A).

Figure 3 Drug sensitivity evaluation results in different lung cancer patients. (A) Sensitivity differences to various drugs among different lung cancer patients across distinct culture conditions (cisplatin, carboplatin, paclitaxel, vinorelbine, gemcitabine, gefitinib: n=10; osimertinib and almonertinib: n=9). (B) Dose-response curves for anticancer drugs in the 3D-3-LC models. Data are presented as mean ± standard deviation (n=4). *, P<0.05; **, P<0.005; ***, P<0.001. 2D-LC: 2D cultured lung cancer cells; Mono-LC: lung cancer cells cultured in 3D conditions; 3D-2-LC_1: CRLCs-CAFs co-culture; 3D-2-LC_2: CRLCs-HUVECs co-culture; 3D-3-LC: CRLCs-CAFs-HUVECs co-culture. 2D, two-dimensional; 3D, three-dimensional; CAFs, cancer-associated fibroblasts; CRLCs, conditionally reprogrammed lung cancer cells; HUVECs, human umbilical vein endothelial cells; IC₅₀, half maximal inhibitory concentration; LC, lung cancer.

It is important to note that compared to traditional 2D-LC models, the IC₅₀ values in the 3D culture system were generally higher, revealing an increased resistance of tumor cells to anti-tumor drugs in the 3D environment. This may more accurately reflect the complexity and drug resistance of tumors in vivo.

In particular, for targeted therapies such as gefitinib, osimertinib, and almonertinib, as well as certain chemotherapeutic agents, the IC₅₀ values between 2D-LC and Mono-LC were found to be similar, indicating that the effects of these specific drugs are less influenced by variations in cell culture dimensions. In the 3D-3 co-culture models from patients LC-19, LC-21, LC-23, and LC-24, cisplatin exhibited a very high sensitivity, with IC₅₀ values significantly below 10 µM. Conversely, most patient’ 3D-3 microbeads displayed resistance to carboplatin. Further analysis revealed that within the same patient samples, the IC₅₀ values for chemotherapeutic agents were generally higher than those for targeted therapies. Notably, the sensitivities of paclitaxel, vinorelbine, osimertinibb, and almonertinib showed large fluctuations in the 3D-3 co-culture system, whereas the sensitivities of cisplatin and carboplatin showed relatively small changes, suggesting that the co-culture of CAFs and HUVECs had a limited effect on the sensitivities of these two drugs (Figure 3B). The IC₅₀ values for different subgroups, drugs, and patients are presented in Table S1.

For specific lung cancer patient populations, particularly cases LC-18 (with KRAS gene mutation) and LC-23 (with MET gene mutation), we conducted an in-depth analysis of drug sensitivity, yielding results highly consistent with theoretical expectations. Both of these patients demonstrated significant drug resistance to TKIs (gefitinib, icotinib, and osimertinib) and conventional chemotherapy agents, further validating the strong correlation between in vitro drug sensitivity measured by the 3D-3-LC model and in vivo treatment responses.

Next, we investigated the relationship between DST results and clinical outcomes, with a detailed analysis of the treatment of patient LC-22. During LC-22’s treatment, gefitinib achieved a progression-free survival (PFS) of 11.1 months, whereas osimertinib extended the PFS to 17.7 months. DSTs performed using the 3D-3 co-culture model revealed that LC-22 had sensitivities of 13.62 µM to gefitinib, 5.93 µM to osimertinib, and 15.72 µM to aumolertinib. These findings suggest that LC-22 was more sensitive to osimertinib than to gefitinib, potentially explaining the longer PFS observed with osimertinib. Thus, the DST results from the 3D-3 co-culture model offer valuable insights for predicting clinical efficacy.

Overall, our study underscores the potential and advantages of the 3D-3 co-culture microbeads model in accurately predicting clinical responses to anti-tumor drugs, providing a robust tool for assessing drug sensitivity and screening for potentially effective therapies.

The impact of stromal cells on tumor cell drug resistance through enhancing lung cancer cell stemness

An increasing body of evidence suggests that CSCs play a critical role in tumor progression, metastasis, and drug resistance. Therefore, we further explored the mechanisms by which CAFs and HUVECs influence stemness in tumor cells. Referring to previously reported lung cancer stemness markers, we analyzed the expression of 17 lung CSC-specific marker genes. Based on the levels of expression, all groups could be roughly categorized into lung cancer tissues, 2D-LC, Mono-LC, and 3D co-culture groups. Most tumor stem cell markers in patient-derived lung cancer cells isolated by CR techniques were well-preserved under 2D conditions (Figure 4A). Some key stemness markers exhibited an increasing trend under 3D conditions (Figure 4A).

Figure 4 The influence of co-culturing with CAFs and HUVECs on lung cancer cell stemness. (A) A heatmap displaying the fold changes in expression levels of lung CSC-specific markers across co-cultured, mono-cultured, and lung tissue samples. The categories include: tissue, 2D-LC, Mono-LC, 3D-2-LC1, 3D-2-LC2, and 3D-3-LC. (B) A comparison of expression levels (FPKM) of lung CSC-specific marker genes among tissue, 2D-LC, Mono-LC, 3D-2-LC1, 3D-2-LC2, and 3D-3-LC samples (n=7). (C) Immunofluorescence analysis of lung cancer tissue and 3D-3 microbeads. Cells were stained with DAPI (blue) to indicate cell nuclei and target proteins in CRLCs were labeled with antibodies against SOX2 (red), SOX9 (red), ALDH1A1 (red), NANOG (red), CD44 (red) and CD24 (red). All scale bars represent 100 µm. (D) Quantitative analysis of western blot results depicting the expression of stemness markers in lung cancer cells under different co-culture conditions. *, P<0.05; **, P<0.005; ***, P<0.001; ****, P<0.0001. Tissue: lung cancer tissue; 2D-LC: 2D cultured lung cancer cells; Mono-LC: lung cancer cells cultured in 3D conditions; 3D-2-LC_1: CRLCs-CAFs co-culture; 3D-2-LC_2: CRLCs-HUVECs co-culture; 3D-3-LC: CRLCs-CAFs-HUVECs co-culture. 2D, two-dimensional; 3D, three-dimensional; CAFs, cancer-associated fibroblasts; CRLCs, conditionally reprogrammed lung cancer cells; CSC, cancer stem cell; DAPI, 4',6-diamidino-2-phenylindole; FPKM, fragments per kilobase of transcript per million mapped reads; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HUVECs, human umbilical vein endothelial cells.

Further analysis of stemness gene expression across each group revealed that, under 3D conditions, the expression of genes such as EpCAM, ALDH1A1, SOX2-OT, CD24, and CDKN1 significantly increased, approaching levels found in lung cancer tissues. Notably, in the 3D-3-LC microbeads, the expression levels of ALDH1A1, SOX2-OT, EpCAM, and SOX9 closely matched those in lung cancer tissues, whereas the expression of CD24 was highest in the 3D-3 co-culture (Figure 4B). In the Mono-LC group, ALDH1A1 and SOX2-OT had higher expression levels. Immunofluorescence staining results indicated that stemness markers ALDH1A1 and NANOG were well-preserved in tumor tissue (LC-21) and the corresponding 3D-3 microbeads, with expression levels similar to those in the original tissue (Figure 4C). Transcriptomic results corroborated the immunofluorescence findings (Figure 4C).

Finally, we utilized western blot analysis to assess the expression of stem cell markers at the protein level in CRLCs after co-culture with CAFs and HUVECs, including NANOG, SOX9, OCT-4, c-Myc, and ALDH1A1 (Figure 4D). Compared to Mono-LC, SOX9 expression was significantly elevated in CRLCs. In the 3D-3-LC co-culture group, NANOG expression markedly increased, whereas c-Myc remained stable. For the 3D-2-LC_1 and 3D-2-LC_2 models, c-Myc expression was significantly reduced. Additionally, co-culture with CAFs and HUVECs resulted in increased expression of OCT-4 and ALDH1A1 in CRLCs, both of which were significantly absent in the Mono-CRLC group.

In summary, the incorporation of stromal cells such as CAFs and HUVECs in 3D co-culture microbeads increases the expression of stemness markers such as NANOG, SOX9, OCT-4, and ALDH1A1, potentially revealing mechanisms by which the TME promotes drug resistance in tumor cells.

Discussion

There is an increasing body of evidence indicating that the cellular components within the TME play a crucial role in drug resistance, particularly through the interactions between different constituents, such as CAFs and ECs. Activated CAFs secrete a variety of growth factors, cytokines, and chemokines that promote cancer progression (29,30). A key limitation of current 3D lung cancer models is the lack of a cancer microenvironment that includes stromal and immune cells. Most existing 3D models are derived solely from epithelial cell lines, resulting in a loss of the heterogeneity found in tumor tissues from clinical lung cancer patients (31). Therefore, when modeling 3D cell culture systems for physiological or drug screening research, it is essential to incorporate cellular components from the TME (such as CAFs, ECs, and immune cells) along with patient-derived tumor cells, thus reintroducing the missing cell-ECM interactions present in traditional 3D models to accurately replicate natural tissue models.

Here, we describe a novel 3D-3 hydrogel co-culture model. This approach utilizes CRLCs to accurately represent the pathological characteristics of patient lung cancer tissues, incorporating HUVECs and patient-derived CAFs into the co-culture model, providing an individualized platform for predicting lung cancer drug sensitivity. We constructed 3D microbeads containing CAFs and HUVECs using Alg-HA scaffolds. The 4% hydrogel microbeads demonstrated excellent cell compatibility and mechanical strength, closely resembling lung cancer tissue, thereby allowing for a precise simulation of natural lung tumor architecture (Figure 1C,1D). To establish our culture system, we performed an initial evaluation of the medium’s composition, creating an in vitro 3D co-culture system. Notably, under 3D culture conditions, the growth rate of cells in the 3D-3 group accelerated compared to those in the Mono-LC group (Figure 1E-1G). Histological analysis indicated that the 3D microbeads retained the histological features of the parental tumor tissue, including cellular and nuclear atypia.

Furthermore, to explore the effects of different co-culture conditions on gene expression in tumor cells, we conducted transcriptomic analysis. PCA revealed that 2D lung cancer cells from different patients clustered tightly together, suggesting that these cells exhibited a high degree of transcriptomic similarity under 2D conditions, likely due to the absence of a TME (Figure 2C). Volcano plots among different groups highlighted similarities in gene expression among the 3D-2-LC_1, 3D-2-LC_2, and 3D-3-LC groups. This indicates that both 3D-2 co-culture and 3D-3 co-culture exhibited similar influences on the transcriptomic characteristics of tumor cells. In contrast, the lung cancer transcriptomes from tissue samples, 2D-LC, Mono-LC, and 3D co-cultures displayed significant differences, further confirming the impact of CAFs and ECs on cancer cell transcriptomes under 3D conditions (Figure S1B).

In the 2D-LC, Mono-LC, and 3D-3-LC experimental models, CAFs and HUVECs exhibit distinct regulatory influences on lung cancer cell transcriptomes. This phenomenon likely stems from their ability to reprogram the stem-like properties of tumor cells. An implication of this viewpoint is that stem cell characteristics could be acquired through genetic modifications and/or interactions with the TME (32). Therefore, we analyzed the expression of 12 genes previously identified as lung CSC-specific markers across different 3D models. Several stem cell markers, including NANOG, EpCAM, SOX2, and ALDH1A1, were upregulated in 3D-3-LC and Mono-LC, yet downregulated in 2D-LC. Further GSEA indicated significant enrichment of key pathways such as cell adhesion molecules, ECM-receptor interaction, cancer pathways, the PI3K-Akt signaling pathway, the AGE-RAGE signaling pathway in diabetic complications, and regulation of actin cytoskeleton under 3D conditions, with pathways such as ECM-receptor interaction, leukocyte transendothelial migration, and cell adhesion molecules being upregulated in 3D-3-LC. These findings underscore the significant impact of 3D co-culture models incorporating CAFs and HUVECs on ECM remodeling and intercellular communication in tumor cells.

Existing research highlights the intricate interactions between tumor cells and TECs within the microenvironment, which contribute to increased resistance of tumor cells to TKIs (33). Our study is the first to validate the connection between TKI resistance and the TME in patient-derived lung cancer models in vitro. In the context of targeted therapies, particularly for gefitinib, osimertinib, and almonertinib, the presence of CAFs and HUVECs in the 3D co-culture significantly enhanced the drug resistance of tumor cells, as evidenced by a general increase in IC₅₀ values across experimental groups. Although targeted therapies represent a more refined approach to cancer treatment, their efficacy can be severely undermined by the TME. The heterogeneity induced by the TME, along with its ability to activate alternative signaling pathways and modulate cancer cell responses through complex intercellular and matrix interactions, poses significant challenges to targeted therapies. Understanding these interactions and developing strategies to overcome TME-induced resistance are crucial for enhancing the effectiveness of targeted cancer treatments and finding new strategies to overcome resistance to these targeted therapies (34). Moreover, our research revealed that for specific lung cancer patient cohorts, particularly cases LC-18 (with KRAS mutations) and LC-23 (with MET mutations), the in-depth drug sensitivity analysis results were highly aligned with theoretical expectations. Both patients exhibited significant resistance to TKIs and conventional chemotherapy drugs, further corroborating the close relationship between in vitro drug sensitivity observed in the 3D-3-LC model and in vivo drug response. Critically, our 3D co-culture model addresses a major translational gap: compared to resource-intensive patient-derived organoids (PDOs) (6–8 weeks) and patient-derived xenograft (PDXs) (4–8 months) models, this platform generates drug sensitivity profiles within 3 weeks post-surgery (35). This rapid approach enables scalable, personalized therapeutic guidance, particularly for refractory cases where traditional models fail to inform timely clinical decisions. This could help further personalize treatments available to patients as to whether patients should be receiving platinum-based chemotherapy. It also may help with risk stratification of patients with EGFR mutations for example and help determine whether the patient would benefit from a combination strategy such as osimertinib and chemotherapy or amivantamab and lazertinib versus osimertinib monotherapy (36-38). Further directions with this model can include evaluating immune checkpoint inhibitors along with new antibody-drug conjugates targeting Trop-2 and HER-3 to see if patients from this 3D model would be candidates for these drugs (39-41).

An increasing body of evidence suggests that CSCs play a pivotal role in tumor progression, metastasis, and drug resistance. CAFs in the TME secrete a variety of growth factors, including hepatocyte growth factor (HGF), insulin-like growth factor II (IGF-II), stromal cell-derived factor 1 (SDF-1), basic fibroblast growth factor (bFGF), Wnt, and oncostatin M (OSM). These factors regulate the “stemness” of NSCLC stem cells through paracrine signaling, activating key pathways such as EMT, TGFβ1, Wnt, Notch, and Hedgehog (HH) signaling pathways, as well as upregulating stemness-related factors such as Oct3/4, Sox2, and Nanog (42). In primary lung cancer tissues, transcriptional regulators critical for stemness, including ALDH1A1, CD44, CD24, and NANOG, have been closely associated with cancer progression and chemoresistance (43). Our results indicate that co-culture of CRLCs with CAFs and HUVECs significantly upregulated the expression of stem cell markers (such as SOX9, NANOG, OCT-4, and ALDH1A1) in CRLCs. Various stemness genes (EpCAM, ALDH1A1, SOX2-OT, CD24, and CDKN1) exhibited significantly elevated expression levels under 3D conditions, aligning more closely with levels found in lung cancer tissues. Notably, ALDH1A1 has been identified as a CSC marker that promotes self-renewal, regulates proliferation, and mediates drug resistance, confirming its direct transcriptional targeting by SOX9. Overexpression of SOX9 enhances the stemness characteristics of NSCLC cells and is a key mechanism driving chemoresistance (44,45). In lung CSCs, elevated ALDH1A1 expression demonstrates a significant correlation with resistance to EGFR-TKIs (46). Furthermore, among advanced-stage lung cancer patients undergoing platinum-based chemotherapy, NANOG overexpression is closely associated with adverse therapeutic outcomes and reduced overall survival (OS) (47). In our 3D-3 co-culture model, the expression levels of ALDH1A1, SOX2-OT, EpCAM, and SOX9 approached those in lung cancer tissues, whereas the expression of the stemness marker CD24 was significantly higher than in lung cancer tissues. Similarly, the expression levels of TP53 and HIF-α were also significantly elevated under 3D-3 conditions (Figure S1C). These results further elucidate the mechanisms by which CAFs and ECs in the TME reduce drug sensitivity in lung cancer cells and enhance the stemness of tumor cells. This is also particularly important in the realm of coming up with improved strategies for immunotherapy efficacy. CAFs have been shown to promote Treg recruitment leading towards CD8⁺ T cell dysfunction, secreting Wnt2 to inhibit the anti-tumor response of dendritic cell mediated CD8⁺ T cells, remodeling of the ECM to prevent infiltration of the immune cells, tumor associated neutrophil differentiation, and M2 macrophage differentiation from TAMs leading to immune checkpoint inhibitor resistance (48). However, the 3D co-culture model in this study was only evaluated for chemotherapeutic drugs and targeted therapeutic agents, and has not yet included research on immune checkpoint inhibitors. Given the central role of immune checkpoint inhibitors in the clinical treatment of solid tumors such as lung cancer, we plan to further conduct in-depth research on the TME and immune checkpoint blockers (ICBs) based on the 3D co-culture model system constructed by our team, thereby providing predictive evidence for the formulation of personalized immunotherapy regimens for lung cancer patients. Furthermore, current models are predominantly developed using surgically excised tumor specimens, which limits their ability to accurately replicate the TME in advanced-stage patients ineligible for surgery. In late-stage cases diagnosed through needle biopsies, the minimal tissue sample size (often as little as 10–50 mg) could compromise the effectiveness of 3D microbeads culture system establishment. Future research should incorporate circulating tumor cell (CTC) enrichment techniques to expand this model’s applicability to metastatic cancer cohorts.

Conclusions

Altogether, our study demonstrates that the sodium Alg-HA 3D-3 co-culture hydrogel microbeads model, which incorporates patient-derived lung cancer cells, CAFs, and HUVECs, can more effectively simulate the intercellular interactions within the lung cancer TME and their impact on the sensitivity to chemotherapy and targeted therapy compared to 2D cultures or monocultures. Furthermore, the mechanisms by which CAFs and ECs reduce drug sensitivity in lung cancer cells are associated with the enhancement of tumor cell stemness. Additionally, considering the short time frame from establishment to drug testing, our developed 3D-3 hydrogel microbeads can be utilized to predict patient-specific drug responses in lung cancer, offering important insights for personalized treatment strategies. Future research should aim to expand the application of the 3D-3 hydrogel microbeads to other drug classes such as immune checkpoint inhibitors, antibody drug conjugates, and adoptive cell therapies, with the goal of providing more forward-looking solutions to address treatment resistance in cancer patients.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the MDAR reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-525/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-525/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-525/prf

Funding: This study was supported by the Natural Science Foundation of Guangdong Province (Nos. 2023A1515030091, 2020B1515120094, and 2021B1515120053).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-525/coif). R.H. is a consultant for MJH Life Sciences, EMD Serono, and Takeda. He has received honoraria from DAVA Oncology and The Dedham Group. He has participated in the advisory board meeting for OncoHost. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was approved by the Ethics Board of the First Affiliated Hospital of Sun Yat-sen University (No. [2018]-153) and written informed consent was provided by all the patients. The study was conducted according to the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
2. Gao Q, Zhan Y, Sun L, et al. Cancer Stem Cells and the Tumor Microenvironment in Tumor Drug Resistance. Stem Cell Rev Rep 2023;19:2141-54. [Crossref] [PubMed]
3. Li Y, Yan B, He S. Advances and challenges in the treatment of lung cancer. Biomed Pharmacother 2023;169:115891. [Crossref] [PubMed]
4. Ashrafi A, Akter Z, Modareszadeh P, et al. Current Landscape of Therapeutic Resistance in Lung Cancer and Promising Strategies to Overcome Resistance. Cancers (Basel) 2022;14:4562. [Crossref] [PubMed]
5. Chhouri H, Alexandre D, Grumolato L. Mechanisms of Acquired Resistance and Tolerance to EGFR Targeted Therapy in Non-Small Cell Lung Cancer. Cancers (Basel) 2023;15:504. [Crossref] [PubMed]
6. Atat OE, Farzaneh Z, Pourhamzeh M, et al. 3D modeling in cancer studies. Hum Cell 2022;35:23-36. [Crossref] [PubMed]
7. Karim NA, Zaza M, Subramanian J. Breaking barriers: patient-derived xenograft (PDX) models in lung cancer drug development-are we close to the finish line? Transl Lung Cancer Res 2024;13:2098-102. [Crossref] [PubMed]
8. Lee E, Lee SY, Seong YJ, et al. Lung cancer organoid-based drug evaluation models and new drug development application trends. Transl Lung Cancer Res 2024;13:3741-63. [Crossref] [PubMed]
9. Rao SS, Dejesus J, Short AR, et al. Glioblastoma behaviors in three-dimensional collagen-hyaluronan composite hydrogels. ACS Appl Mater Interfaces 2013;5:9276-84. [Crossref] [PubMed]
10. Zahmatkesh E, Khoshdel-Rad N, Mirzaei H, et al. Evolution of organoid technology: Lessons learnt in Co-Culture systems from developmental biology. Dev Biol 2021;475:37-53. [Crossref] [PubMed]
11. Heydari Z, Moeinvaziri F, Agarwal T, et al. Organoids: a novel modality in disease modeling. Biodes Manuf 2021;4:689-716. [Crossref] [PubMed]
12. Glabman RA, Choyke PL, Sato N. Cancer-Associated Fibroblasts: Tumorigenicity and Targeting for Cancer Therapy. Cancers (Basel) 2022;14:3906. [Crossref] [PubMed]
13. Nagl L, Horvath L, Pircher A, et al. Tumor Endothelial Cells (TECs) as Potential Immune Directors of the Tumor Microenvironment - New Findings and Future Perspectives. Front Cell Dev Biol 2020;8:766. [Crossref] [PubMed]
14. Zhang H, Yue X, Chen Z, et al. Define cancer-associated fibroblasts (CAFs) in the tumor microenvironment: new opportunities in cancer immunotherapy and advances in clinical trials. Mol Cancer 2023;22:159. [Crossref] [PubMed]
15. Holloway EM, Capeling MM, Spence JR. Biologically inspired approaches to enhance human organoid complexity. Development 2019;146:dev166173. [Crossref] [PubMed]
16. Zhu K, Lv Z, Xiong J, et al. MET inhibitor, capmatinib overcomes osimertinib resistance via suppression of MET/Akt/snail signaling in non-small cell lung cancer and decreased generation of cancer-associated fibroblasts. Aging (Albany NY) 2021;13:6890-903. [Crossref] [PubMed]
17. Lim JTC, Kwang LG, Ho NCW, et al. Hepatocellular carcinoma organoid co-cultures mimic angiocrine crosstalk to generate inflammatory tumor microenvironment. Biomaterials 2022;284:121527. [Crossref] [PubMed]
18. Chen S, Giannakou A, Golas J, et al. Multidimensional Coculture System to Model Lung Squamous Carcinoma Progression. J Vis Exp 2020;
19. Lazzari G, Nicolas V, Matsusaki M, et al. Multicellular spheroid based on a triple co-culture: A novel 3D model to mimic pancreatic tumor complexity. Acta Biomater 2018;78:296-307. [Crossref] [PubMed]
20. Satpathy A, Datta P, Wu Y, et al. Developments with 3D bioprinting for novel drug discovery. Expert Opin Drug Discov 2018;13:1115-29. [Crossref] [PubMed]
21. Lee SW, Kwak HS, Kang MH, et al. Fibroblast-associated tumour microenvironment induces vascular structure-networked tumouroid. Sci Rep 2018;8:2365. [Crossref] [PubMed]
22. Kang SM, Lee JH, Huh YS, et al. Alginate Microencapsulation for Three-Dimensional In Vitro Cell Culture. ACS Biomater Sci Eng 2021;7:2864-79. [Crossref] [PubMed]
23. Shin YM, Kim TG, Park JS, et al. Engineered ECM-like microenvironment with fibrous particles for guiding 3D-encapsulated hMSC behaviours. J Mater Chem B 2015;3:2732-41. [Crossref] [PubMed]
24. Zhao Z, Feng X, Wu H, et al. Construction of a lung cancer 3D culture model based on alginate/gelatin micro-beads for drug evaluation. Transl Lung Cancer Res 2024;13:2698-712. [Crossref] [PubMed]
25. Liu XH, Tang KJ, Lei YY, et al. Establishment of a 3D Culture Model Based on Patient-derived Lung Cancer Cells and Anticancer Drug Sensitivity Study. Pharmacy Today 2023;33:335-41, 351.
26. Dobin A, Davis CA, Schlesinger F, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 2013;29:15-21. [Crossref] [PubMed]
27. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 2014;30:923-30. [Crossref] [PubMed]
28. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550. [Crossref] [PubMed]
29. Luis G, Godfroid A, Nishiumi S, et al. Tumor resistance to ferroptosis driven by Stearoyl-CoA Desaturase-1 (SCD1) in cancer cells and Fatty Acid Biding Protein-4 (FABP4) in tumor microenvironment promote tumor recurrence. Redox Biol 2021;43:102006. [Crossref] [PubMed]
30. Wang M, Herbst RS, Boshoff C. Toward personalized treatment approaches for non-small-cell lung cancer. Nat Med 2021;27:1345-56. [Crossref] [PubMed]
31. Anthon SG, Valente KP. Vascularization Strategies in 3D Cell Culture Models: From Scaffold-Free Models to 3D Bioprinting. Int J Mol Sci 2022;23:14582. [Crossref] [PubMed]
32. Prasetyanti PR, Medema JP. Intra-tumor heterogeneity from a cancer stem cell perspective. Mol Cancer 2017;16:41. [Crossref] [PubMed]
33. Bani M, Decio A, Giavazzi R, et al. Contribution of tumor endothelial cells to drug resistance: anti-angiogenic tyrosine kinase inhibitors act as p-glycoprotein antagonists. Angiogenesis 2017;20:233-41. [Crossref] [PubMed]
34. Hirata E, Sahai E. Tumor Microenvironment and Differential Responses to Therapy. Cold Spring Harb Perspect Med 2017;7:a026781. [Crossref] [PubMed]
35. Gao J, Lan J, Liao H, et al. Promising preclinical patient-derived organoid (PDO) and xenograft (PDX) models in upper gastrointestinal cancers: progress and challenges. BMC Cancer 2023;23:1205. [Crossref] [PubMed]
36. Cho BC, Felip E, Hayashi H, et al. MARIPOSA: phase 3 study of first-line amivantamab + lazertinib versus osimertinib in EGFR-mutant non-small-cell lung cancer. Future Oncol 2022;18:639-47. [Crossref] [PubMed]
37. Passaro A, Wang J, Wang Y, et al. Amivantamab plus chemotherapy with and without lazertinib in EGFR-mutant advanced NSCLC after disease progression on osimertinib: primary results from the phase III MARIPOSA-2 study. Ann Oncol 2024;35:77-90. [Crossref] [PubMed]
38. Planchard D, Jänne PA, Cheng Y, et al. Osimertinib with or without Chemotherapy in EGFR-Mutated Advanced NSCLC. N Engl J Med 2023;389:1935-48. [Crossref] [PubMed]
39. Yu HA, Goto Y, Hayashi H, et al. HERTHENA-Lung01, a Phase II Trial of Patritumab Deruxtecan (HER3-DXd) in Epidermal Growth Factor Receptor-Mutated Non-Small-Cell Lung Cancer After Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitor Therapy and Platinum-Based Chemotherapy. J Clin Oncol 2023;41:5363-75. [Crossref] [PubMed]
40. Ahn MJ, Tanaka K, Paz-Ares L, et al. Datopotamab Deruxtecan Versus Docetaxel for Previously Treated Advanced or Metastatic Non-Small Cell Lung Cancer: The Randomized, Open-Label Phase III TROPION-Lung01 Study. J Clin Oncol 2025;43:260-72. [Crossref] [PubMed]
41. Gandhi L, Rodríguez-Abreu D, Gadgeel S, et al. Pembrolizumab plus Chemotherapy in Metastatic Non-Small-Cell Lung Cancer. N Engl J Med 2018;378:2078-92. [Crossref] [PubMed]
42. Chen WJ, Ho CC, Chang YL, et al. Cancer-associated fibroblasts regulate the plasticity of lung cancer stemness via paracrine signalling. Nat Commun 2014;5:3472. [Crossref] [PubMed]
43. Fatma H, Siddique HR, Maurya SK. The multiple faces of NANOG in cancer: a therapeutic target to chemosensitize therapy-resistant cancers. Epigenomics 2021;13:1885-900. [Crossref] [PubMed]
44. Voronkova MA, Rojanasakul LW, Kiratipaiboon C, et al. The SOX9-Aldehyde Dehydrogenase Axis Determines Resistance to Chemotherapy in Non-Small-Cell Lung Cancer. Mol Cell Biol 2020;40:e00307-19. [Crossref] [PubMed]
45. Wei Y, Wu S, Xu W, et al. Depleted aldehyde dehydrogenase 1A1 (ALDH1A1) reverses cisplatin resistance of human lung adenocarcinoma cell A549/DDP. Thorac Cancer 2017;8:26-32. [Crossref] [PubMed]
46. Huang CP, Tsai MF, Chang TH, et al. ALDH-positive lung cancer stem cells confer resistance to epidermal growth factor receptor tyrosine kinase inhibitors. Cancer Lett 2013;328:144-51. [Crossref] [PubMed]
47. Chang B, Park MJ, Choi SI, et al. NANOG as an adverse predictive marker in advanced non-small cell lung cancer treated with platinum-based chemotherapy. Onco Targets Ther 2017;10:4625-33. [Crossref] [PubMed]
48. Pei L, Liu Y, Liu L, et al. Roles of cancer-associated fibroblasts (CAFs) in anti- PD-1/PD-L1 immunotherapy for solid cancers. Mol Cancer 2023;22:29. [Crossref] [PubMed]

(English Language Editor: J. Jones)