Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Cancer Cell Int . 2026 Mar 6;26:166. doi: 10.1186/s12935-026-04250-1 Search in PMC Search in PubMed View in NLM Catalog Add to search A new inject-embed 3D culture method enables the spheroid and aggregate formation from single or dual liver cancer cell types Hanyuan Zhang Hanyuan Zhang 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China 2 Center for Life Sciences, Shaoxing Institute, Zhejiang University, Shaoxing, 321000 Zhejiang China 3 Cancer Center, Zhejiang University, Hangzhou, 310058 Zhejiang China 4 Department of Gastroenterology, the First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, 310003 China Find articles by Hanyuan Zhang 1, 2, 3, 4, # , Shuting Xue Shuting Xue 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China 2 Center for Life Sciences, Shaoxing Institute, Zhejiang University, Shaoxing, 321000 Zhejiang China 3 Cancer Center, Zhejiang University, Hangzhou, 310058 Zhejiang China Find articles by Shuting Xue 1, 2, 3, # , Jinchang Pan Jinchang Pan 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China 2 Center for Life Sciences, Shaoxing Institute, Zhejiang University, Shaoxing, 321000 Zhejiang China 3 Cancer Center, Zhejiang University, Hangzhou, 310058 Zhejiang China Find articles by Jinchang Pan 1, 2, 3 , Jianjuan Zhang Jianjuan Zhang 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China 2 Center for Life Sciences, Shaoxing Institute, Zhejiang University, Shaoxing, 321000 Zhejiang China 3 Cancer Center, Zhejiang University, Hangzhou, 310058 Zhejiang China Find articles by Jianjuan Zhang 1, 2, 3 , Danduo Wei Danduo Wei 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China 2 Center for Life Sciences, Shaoxing Institute, Zhejiang University, Shaoxing, 321000 Zhejiang China 3 Cancer Center, Zhejiang University, Hangzhou, 310058 Zhejiang China Find articles by Danduo Wei 1, 2, 3 , Peng Xia Peng Xia 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China Find articles by Peng Xia 1 , Guangming Li Guangming Li 5 Department of Liver Disease, Henan Provincial Infectious Disease Hospital, the Affiliated Infectious Disease Hospital of Zhengzhou University, Zhengzhou Sixth People’s Hospital, Zhengzhou, 450000 China Find articles by Guangming Li 5 , Stephanie Roessler Stephanie Roessler 6 Institute of Pathology, Heidelberg University, University Hospital Heidelberg, Heidelberg, Germany Find articles by Stephanie Roessler 6 , Chaohui Yu Chaohui Yu 4 Department of Gastroenterology, the First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, 310003 China Find articles by Chaohui Yu 4, ✉ , Junfang Ji Junfang Ji 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China 2 Center for Life Sciences, Shaoxing Institute, Zhejiang University, Shaoxing, 321000 Zhejiang China 3 Cancer Center, Zhejiang University, Hangzhou, 310058 Zhejiang China Find articles by Junfang Ji 1, 2, 3, ✉ Author information Article notes Copyright and License information 1 The MOE Key Laboratory of Biosystems Homeostasis & Protection, Zhejiang Provincial Key Laboratory for Cancer Molecular Cell Biology, Life Sciences Institute, Zhejiang University, Hangzhou, 310058 Zhejiang China 2 Center for Life Sciences, Shaoxing Institute, Zhejiang University, Shaoxing, 321000 Zhejiang China 3 Cancer Center, Zhejiang University, Hangzhou, 310058 Zhejiang China 4 Department of Gastroenterology, the First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, 310003 China 5 Department of Liver Disease, Henan Provincial Infectious Disease Hospital, the Affiliated Infectious Disease Hospital of Zhengzhou University, Zhengzhou Sixth People’s Hospital, Zhengzhou, 450000 China 6 Institute of Pathology, Heidelberg University, University Hospital Heidelberg, Heidelberg, Germany ✉ Corresponding author. # Contributed equally. Received 2025 Nov 4; Accepted 2026 Feb 23; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13077891 PMID: 41792756 Background Three-dimensional (3D) cell culture techniques have emerged as a bridge between traditional two-dimensional (2D) cell culture and the complex 3D architecture of living organisms. To overcome the limitations of existing 3D culture methods, we aimed to develop a new 3D culture platform that provides a liquid-matrix interface for cells, enabling both cell-cell and cell-matrix interactions. Methods We generated an inject-embed 3D culture platform with a standardized procedure for operation and quantification. This platform was evaluated using seven cell lines: six liver cancer cell lines including four hepatocellular carcinoma (HCC) lines and two cholangiocarcinoma (CCA) lines, and LX2 hepatic stellate cells representing the liver cancer microenvironment. Cells were cultured in both mono-culture and co-culture setups to assess spheroid and aggregate formation, proliferation, assembly, and their 3D architecture. We also evaluated its utilization in assays of stemness-related gene expression, chemo-resistance, and the tumor malignancy phenotypes. Results In mono-culture, spheroids from all seven cell lines exhibited varying sizes and shapes. In co-culture setup, HCC cells with LX2 predominantly formed mixed LX2-HCC hybrid aggregates, while CCA cells with LX2 formed well-organized CCA-centered/LX2-surrounded aggregates. In both mono-culture and co-culture systems, this inject-embed method supported significant cell proliferation, spheroid/aggregate aggregation, and cell-cell communication. Compared with conventional 2D culture, cells in this 3D system altered gene expression profiles, enhanced stemness-associated gene expression and increased cell resistance to chemotherapeutic agents. Moreover, a known HCC malignancy regulator altered spheroid size and number in this method, demonstrating its suitability for functional studies in cancer field. Conclusions This newly established inject-embed 3D culture method is highly reproducible, effectively promotes spheroid/aggregate growth and assembly, and enables the investigation of cell malignancy features under 3D microenvironment conditions. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1186/s12935-026-04250-1. Keywords: Three-dimensional cell culture, Spheroid, Aggregate, Liver cancer, Hepatic stellate cell Highlights Inject-embed 3D culture provides liquid-matrix interface and high reproducibility. Inject-embed method shows notable cell growth, spheroid and aggregate formation. Spheroids from various cell lines with this method differ in size and shape. Co-culture with LX2 of various cell lines yields distinct 3D aggregate structures. Inject-embed method detects cell malignancy features in a 3D microenvironment. Supplementary Information The online version contains supplementary material available at 10.1186/s12935-026-04250-1. Introduction Currently, the primary methods for functional research in oncology studies are in vitro cell culture and in vivo animal experiments. In the traditional two-dimensional (2D) cell culture system, cells grow as monolayers on flat, solid surfaces with unrestricted access to oxygen and nutrients. Concurrently, cells in 2D cultures also experience mechanical stretching, which triggers cytoskeletal rearrangement and induces of artificial polarity [ 1 ]. In contrast, animal experiments, conducted entirely in vivo, pose challenges in controlling individual or intermediate processes and come with drawbacks such as high costs, time demands, and inherent species variability [ 2 ]. Cancer is the leading cause of death globally. To meet the growing need for more accurate in vitro tumor models, significant progress has been made in three-dimensional (3D) cell culture technology in recent years. 3D cell culture models effectively replicate key features of in vivo tumors, including cell-cell and cell-matrix interactions [ 3 , 4 ]. Moreover, the 3D structure creates osmotic gradients of oxygen, nutrients, metabolites, and soluble factors, resulting in distinct proliferation rates between the interior and exterior of the 3D cell mass. This closely mimics the metabolic characteristics of living tumor environments [ 5 , 6 ]. Thus, 3D cell culture models have become essential in vitro tools, enabling researchers to gain deeper insights into the complex biological behavior of tumors. Current 3D cell culture technologies for cancer cell lines are broadly classified into two categories based on the presence or absence of scaffolds. Scaffold-free methods, including ultra-low attachment culture plates, the hanging drop method, rotary vessel/spinner flask, aqueous two-phase system, and magnetic levitation etc. These approaches share a common principle: they utilize non-adhesive surfaces to prevent cell attachment and encourage cell aggregation into spheroids [ 7 – 11 ]. On the other hand, scaffold-based 3D cell culture typically uses a classical embedding method, where cells are mixed with scaffold material and cultured within it [ 12 ]. The scaffold provides structural support and serves as a source of external signals, influencing cellular interactions and functionality [ 13 ]. Commonly used scaffold materials include natural and synthetic hydrogels, each with distinct properties that allow for customization based on experimental needs [ 14 – 16 ]. Among these, Matrigel, a soluble basement membrane extract derived from the Engelbreth-Holm-Swarm (EHS) mouse sarcoma, is the most widely utilized natural scaffold material, rich in various extracellular matrix (ECM) proteins [ 17 ]. In general, scaffold-free 3D culture allowed better cell-cell interactions at a liquid interface with minimal ECM, whereas scaffold-based methods provided ECM but restricted cell-cell interactions. Meanwhile, it is important to incorporate the tumor microenvironment (TME) including cellular components like cancer-associated fibroblasts (CAFs) and non-cellular components, alongside tumor cells in 3D culture to more accurately recapitulate real tumors. Therefore, we sought to develop a more physiologically relevant 3D culture approach that supports both cell-ECM and homotypic/heterotypic cell-cell interactions, with both liquid and ECM interfaces. Based on the traditional “mixing and then embedding” 3D culture technique (the mix-embed method), we developed an innovative approach called the inject-embed 3D culture method. In this method, a single cell line or multiple cell lines in culture medium are injected into a Matrigel mixture. According to the literature, spheroid generated in our study describes spherical masses formed by a single cell type and aggregate describes 3D structures formed by proliferation or aggregation of multiple cell types [ 18 ]. We have tested this approach with seven cell lines, i.e., six primary liver cancer cell lines with both hepatocellular carcinoma (HCC) pathological type ( n = 4) and cholangiocarcinoma (CCA) pathological type ( n = 2) [ 19 , 20 ], and one activated hepatic stellate cell line (LX2) representing a primary source of CAFs in liver cancer [ 21 , 22 ]. With the inject-embed method, 3D spheroids of all cell lines successfully formed, displaying various sizes and shapes. When liver cancer cells were co-cultured with LX2, aggregates with unique 3D architectures emerged, predominantly in two types. These cells also exhibited proliferation and aggregation. We have also described other phenomena observed with the inject-embed method and explored its applications, including transcriptomic alterations. Overall, the inject-embed method represents a novel and practical 3D cell culture approach, offering a practical and versatile 3D culture strategy with broad applicability in cancer research. It may also allow researchers to better capture the complexity of 3D architectures in physiological or pathological organisms, enhancing our understanding of biological systems. Methods Cell lines Human HCC cell lines (Huh7, HLF, Huh1, HLE); human CCA cell lines (HUCCT1, RBE); human hepatic stellate cell line LX2; and human embryonic kidney HEK 293T cells were routinely cultured in our lab as we described before [ 19 , 20 , 23 ]. Huh7, HLF, Huh1, HLE were originally from Japanese Collection of Research Biosources Cell Bank (JCRB). RBE and LX2 were from Chinese Academy of Sciences (Shanghai, China). HEK 293T was from American Type Culture Collection (ATCC). They were authenticated via short tandem repeat profile done by GTB Corporation. Cell lines were confirmed to be negative for Mycoplasma by a TransDetect PCR Mycoplasma Detection Kit (FM311-01, Transgen Biotech). CCA cell lines were cultured in RPMI 1640 medium and other cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM), which were supplemented with 10% fetal bovine serum, 100 U/ml penicillin–streptomycin and 1% L-glutamine. All cells were cultured in a humidified atmosphere of 5% CO 2 at 37 °C. Plasmids and siRNAs Lentiviral construct pCDH-CMV-MCS-EF1α-copGFP was obtained from SBI Biosciences and stored in our lab. pCDH-CMV-MCS-EF1-copRFP was constructed by replacing copGFP with Ds-Red between the XbaI and NotI sites. pCDH-CMV-CYP39A1-3×flag-EF1α-copGFP were generated via amplifying whole length of CYP39A1 through RT-PCR and inserting them to the multiple cloning site (ClaI/NotI). Lentiviruses were packaged with plasmids psPAX2 and pMD2.G (Addgene) in HEK 293T cells. For infection, 5 MOI of each lentivirus was used for all our studies. In the co-culture experiments, lentiviruses carrying GFP and RFP were utilized to specifically infect liver cancer cells and LX2 cells, respectively. CYP39A1 siRNAs and scramble negative control siRNAs were purchased (Ribo, Guangzhou). All information of constructs and sequences are listed in Table. S1. Lipofectamine 2000 Reagent (Cat# 11668019, Invitrogen) was used for transfections of plasmids and Rfect siRNA Transfection Reagent (Cat# 11011, BIOTRAN) was used for transfections of siRNAs. The expression efficiency of the plasmids and siRNAs has been validated in previously published studies [ 20 ]. Inject-embed 3D culture method Matrigel (Cat# 354234, Corning), after being initially thawed overnight at 4 °C, was gently aspirated and then mixed with culture medium at a 1:1 ratio to achieve a homogeneous mixture with a final concentration of 4–6 mg/ml. All reagents and materials coming into contact with Matrigel were adequately pre-chilled. 100 µl of diluted Matrigel was added to each well of the 96-well plate, followed by a 30-minute incubation at 37 °C. 10 µl of the prepared cell suspension was injected into each well using a 10 µl pipette. The tip was then vertically inserted into the gel’s center to a suitable depth, verifying that it was properly embedded in the gel without reaching the well’s bottom. With the tip in position, the cell suspension was injected into the gel at a steady rate, ensuring the process was smooth and not too rapid. After a 30-minute incubation at 37 °C, 100 µl of complete medium was added to each well, and the medium was changed every other day. For the spheroid proliferation and aggregation examination, initial cell numbers of mono-culture were established at 125, 250, 500, 1,000, 2,000, and 4,000 per well. For co-culture of liver cancer cells and LX2 cells, the initial cell numbers were established at 500/500, 500/1,500, 1,000/1,000, 1,500/500, and 2,000/2,000 for liver cancer cell/LX2 cell per well. For the comparison of this method with the other two 3D culture methods, cell numbers were set for the inject-embed method at 1,000 cells per well for mono-culture and 1,000/1,000 for liver cancer cell/LX2 cells per well for co-culture. In experiments using the inject-embed method to evaluate the expression of stemness genes, Huh7 and HLF were initiated with 2,000 and 1,000 cells, respectively, and spheroids were harvested on the eighth day of cultivation. In experiments validating the CYP39A1 phenotype, Huh7 and HLF were initiated with 4,000 and 2,000 cells, respectively, with images captured on the eighth day. Mix-embed 3D culture method Matrigel was diluted to a concentration of 50% as described above. Subsequently, the mixture was evenly distributed into 96-well plates at 50 µl per well. The plates were then incubated at 37 °C for 15–30 min to facilitate gel solidification. A cell suspension of suitable density was combined with an equal volume of undiluted Matrigel and dispensed into a 96-well plate at 50 µl per well. After a 30-minute incubation at 37 °C, each well received 100 µl of complete growth medium, which was replaced every other day. For the comparison of with the other two 3D culture methods during mono-culture, two cell numbers were mainly used in this method. 1,000 cells per well per cell line were used for the comparison at the level of the same initial cell number. 5,000 cells per well per cell line were used for the comparison at the level of the identical cell density (100 cells/µl). For co-culture, liver cancer cell lines and LX2 cells were both harvested and generally mixed at a 1:1 ratio, and a total of 2,000 mixed cells were used under the condition of the same initial cell number and a total of 10,000 mixed cells were used under the condition of the identical cell density. For specific occasion, other cell number might also be used as indicated in the manuscript. Hanging drop 3D culture method Droplets (20 µl) of cell suspension were placed on the inner surface of the Petri dish lid, which was then gently inverted back onto the dish surface. To prevent evaporation of the droplets, an adequate volume of phosphate-buffer solution (PBS) was pre-added to the Petri dish. For the comparison with the other two 3D culture methods, two cell numbers were mainly used in this method. 1,000 cells per well per cell line were used for the comparison at the level of the same initial cell number. 2,000 cells per well per cell line were used for the comparison at the level of the identical cell density (100 cells/µl). For co-culture, liver cancer cell lines and LX2 cells were harvested and mixed at a 1:1 ratio, and a total of 2,000 mixed cells were used under the condition of the same initial cell number and a total of 4,000 mixed cells were used under the condition of the identical cell density. Spheroid imaging and quantification Images were captured using an inverted fluorescence microscope (Leica). In observing specific spheroid or aggregate clusters, a “cross” on the 96-well plate bottom divided it into quadrants for target positioning relative to the origin, which also horizontally labeled the X-axis and Y-axis for the targeted objects. Marking imminently fusing spheroids or aggregates with interconnected dashed circles in the images increased tracking reliability. They were further confirmed across different days by comparing their morphology and relative positions to each other and to nearby structures. For all three culture methods, the number and size of the spheroids and aggregates were quantified in ImageJ software (NIH) using standardized criteria for regions of interest (ROIs) annotation and measurement. Only clearly in-focus objects were included automatically or manually as appropriate. Area-based parameters were obtained using the “Analyze Particles” function in ImageJ, with identical calibration and measurement settings applied across methods to ensure consistency. For the inject-embed method, spheroids predominantly positioned within the specific focal plane along the vertical axis of the Matrigel based on the observation from the 40 sequenced images along the Z-axis of the culture plate (Supplementary movie AVI file 1). Thus, images in this method could be captured with clear focus plane and then processed in bulk using the Batch Macro function of ImageJ software, which allows the automatic quantification of spheroids’ numbers and sizes. For the mix-embed and hanging drop methods, target spheroids were manually outlined using the ImageJ software. In the mix-embed method, the spheroids were distributed throughout the different focal planes of the gel based on the observation from the 40 sequenced images along the plate Z-axis (Supplementary movie AVI file 2). In this case, for spheroid quantification, one representative focal plane was selected. Given the approximately uniform distribution of spheroids within the gel, the plane with relative more focused spheroids was chosen, and the clearly in-focus spheroids with sharp boundaries were then manually outlined in ImageJ. For the hanging drop method, all spheroids settled at the droplet bottom and could be clearly visualized within a single focal plane under the microscope. In this case, the representative image was taken but manual annotation for the spheroids was required due to edge artifacts of the image, which was caused by the natural shape of the droplet. The un-focused spheroids were manually excluded. After outlining, parameter determination was performed using the “Analyze Particles” tools in ImageJ, which was the same across all methods. For size comparisons, statistical analyses were performed based on spheroid/aggregate size parameters. The diameter ( d ) of each spheroid or aggregate was calculated from the area measured by ImageJ, employing the formula d = 2 × √(area/π) . Only spheroids and aggregates with a surface area exceeding 800 μm² were considered for statistical inclusion. For total volume of all spheroids, the formula V = (4/3) × π × (d/2)³ was used. For spheroid-number analysis, spheroid number was defined as the count of clearly in-focus spheroids in the quantified representative focal plane per image. Given the distinct physical organization of cells in each culture method, direct comparisons of absolute spheroid numbers across methods might not biologically equivalent and were therefore not emphasized. Harvest spheroids from the gel After carefully aspirating the culture medium from the gel surface, each well was rinsed twice with 100 µl of pre-chilled PBS. 200 µl of 5 mM PBS-EDTA was added to each well and the mixture was allowed to stand at 4 °C for 1 h. Post this interval, the macroscopic observation of a hue transition from pale pink to light yellow within the gel, coupled with the microscopic detection of spheroids settling at the well bottoms, signified the complete dissolution of the Matrigel. The liquid in the wells was gently pipetted up and down using a 100 µl pipette to ensure that the spheroids could be aspirated. The spheroid-containing liquid was aliquoted into Eppendorf tubes and centrifuged at 200 ×g for 1–2 min, resulting in a loose pellet of spheroids at the tube base. The harvested spheroids were then ready for downstream experimental processes. Sphere formation with ultra-low attachment plate For sphere formation assay, 2,000 Huh7 cells and 1,000 HLF cells were plated in Ultra Low Attachment 24-well plates (Cat# 3473, Corning), respectively. The cells were cultured in 1 ml of DMEM medium at 37 °C for a duration of 8 to 10 days. RNA extraction and quantitative real-time PCR Total RNA was extracted using TRIzol RNA isolation Reagents (Invitrogen) following the manufacturer’s instructions. cDNA was reverse transcribed with 1 µg of total RNA using PrimeScriptTM RT reagent Kit (Cat# RR047, TaKaRa). Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed with the TB Green Premix Ex Taq II (Cat# RR420, TaKaRa). 18 S was used as the reference gene. All primer sequences are listed in Table. S1. Drug treatment and cell viability detection Cells in 2D and 3D cultures were treated with different chemotherapeutic drugs including Doxorubicin (Dox, Cat# S1208, Selleck), Cisplatin (Cat# HY-17394, MCE), and 5-Fluorouracil (5-Fu, Cat# HY-90006, MCE) at indicated concentrations with 4–6 replicates per condition. In 2D culture, 4,000 Huh7 and 2,000 HLF cells per well were seeded in 96-well plates (BeyoGold™ black 96-well plates, Cat# FCP966-80pcs, Beyotime). Post a 24-hour interval, 100 µl of pre-prepared drug solution was added to each well. After a 48-hour treatment, the plates were left at room temperature for 10 min. Then, 100 µl of CellTiter-Lumi™ reagent (Cat# C0061M, Beyotime) was added to each well. In 3D cultures, 4,000 Huh7 and 2,000 HLF cells per well were also seeded in 96-well plates. After a 6-day culture, 100 µl of drug solution was added to each well and incubated for 48 h. At the end of drug treatment, the solution was carefully removed, and 50 µl of 10 mM EDTA-PBS solution was added to each well, followed by a 1-hour incubation at 4 °C to dissolve the gel. Then, 150 µl of CellTiter-Lumi™ reagent was added per well. The plate was subsequently placed on a shaker at 150 rpm for 25 min at room temperature. Luminescence signals were then measured using a multifunctional plate reader (SynergyNEO2, BioTek). RNA sequencing and data analysis RNA sequencing was performed with RNA extracted from three culture conditions (2D, inject-embed, and mix-embed) and two cell lines (Huh7 and HLF). Two biological replicates were used for each condition. For 2D cultures, cells were harvested after 2 days when the confluency reached approximately 90%. For the inject-embed and mix-embed 3D cultures, cells were harvested at 8 days after seeding. RNA samples were sent to Novogene Co., Ltd. (Tianjin, China) for library construction and sequencing. RNA integrity was assessed using the Bioanalyzer 2100 system (Agilent Technologies, CA, USA). Libraries that passed quality control yield to the paired-end Illumina sequencing. The resulting RNA-seq data have been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession number GSE298810. Gene expression was analyzed with their FPKM (Fragments Per Kilobase of exon model per Million mapped fragments) value. The sample correlation analysis was performed with R package (version 4.2.2) using the ‘cor()’ function based on the Pearson correlation. The resulting correlation matrix was visualized with the ‘pheatmap’ function. The principal component analysis (PCA) was done using the ‘prcomp()’ function in R package. PCA visualization was generated using the R packages ‘ggplot2’ and ‘ggrepel’. Genes in any group (2D culture, inject-embed method, mix-embed method) with a median FPKM > 3 were used for class comparison analysis, and 3,698 genes satisfied this criterion. Class composition between each 3D culture group and 2D culture group was then performed to obtain the differential expressed genes based on Student’s t-test (P-value < 0.05, |log 2 of fold changes| > 0.2). These significantly differentially expressed genes were used for Gene Set Enrichment Analysis (GSEA), Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analysis. GSEA analysis was performed using GSEA 4.0.0. KEGG and GO analysis were performed using the ShinyGO online platform ( https://bioinformatics.sdstate.edu/go/ ). Statistical analysis All the experiments in the study were repeated at least three times. The statistical graphs for this study were created by Graphpad Prism 9 software. Class comparison was performed to obtain the differential gene expression between the 3D culture group and 2D culture group. Two-way ANOVA and Student’s t-test were used for statistical analysis of comparative data between groups. All P-values were 2-sided, and P-values less than 0.05 were considered significant. Results A newly generated inject-embed 3D culture method enabled spheroid formation of seven tested cell lines Two main traditional 3D cell culture methods often used in cancer research are the mix-embed method (featuring an ECM interface) and the hanging drop method (featuring a liquid interface). Here we developed a more physiological relevant 3D culture approach, termed the inject-embed method, which integrates both liquid and ECM interfaces to enable interactions between cells and the ECM, as well as among cells themselves. It consists of three key steps, i.e., (1) preparing a low concentration Matrigel layer in a culture plate, (2) preparing a cell suspension in culture medium, and (3) injecting 10 µl of resuspended cells into the pre-solidified Matrigel using a pipette (Fig. 1 A). In this system, the culture medium in the gel provides a liquid interface while Matrigel serves as the ECM interface for the cultured cells. Fig. 1. Open in a new tab The newly established inject-embed 3D culture method allowed the spheroid formation of all tested cells. ( A ) Schematic diagram of the inject-embed 3D culture method. The process includes three steps, i.e., prepare 100 µl of Matrigel diluted 1:1 with culture medium in each well of a 96-well plate; prepare a cell suspension; inject 10 µl of the cell suspension into the solidified gel with a pipette. Then 100 µl of culture medium was added to each well, and replace the medium as required. ( B ) 3D culture of HUCCT1, HLF, and Huh7 cell lines using the inject-embed method, traditional mix-embed method, and hanging drop method. An initial cell number of 1,000 was used for each cell line. ( C ) The sizes and morphologies of spheroids from seven cell lines under the inject-embed 3D culture. HCC, hepatocellular carcinoma. CCA, cholangiocarcinoma. TME, tumor microenvironment. An initial cell number of 4,000 per well in a 96-well plate was used for each cell line. The images were captured 4 days after injection, and scale bars are shown in the images We first applied this method to three liver cancer cell lines (HUCCT1, HLF, and Huh7) and compared its performance with the mix-embed and hanging drop methods (Fig. 1 B). Using an identical starting cell number ( n = 1,000), the inject-embed spheroids were generally larger across all three cell lines compared to those in the mix-embed method. Notably, the inject-embed spheroids predominantly positioned within a specific focal plane along the vertical axis of the Matrigel matrix under the microscope, unlike those in the mix-embed method (Fig. S1A, Supplementary movie AVI files). This facilitated a standardized bulk imaging processing and quantification of spheroids with ImageJ software for the inject-method approach. In contrast, the hanging drop method yielded varied results. HUCCT1 failed to form spheroids, but exhibited only loose aggregation, accompanied by partial cell death and abundant cellular debris around the clusters (Fig. 1 B). However, HLF and Huh7 cells formed one single, large, compact spheroid in the hanging drop method after 6 days of culture. A limitation of the hanging drop method is its inability to accommodate medium changes or supplementation, resulting in stagnant spheroid sizes for HLF and Huh7 cells beyond 4 or 6 days of culture. Similar results were observed when assays were conducted at the identical cell densities (100 cells/µl) among three methods (Fig. S1B). Collectively, the new inject-embed method allowed spheroid formation of all three tested cells with noticeable advantages, i.e., reproducible procedure, a liquid interface allowing better cell-cell interactions, an ECM interface supporting a cell-ECM interactions, tolerable for extended culture times, and allowing clear image capture and standardized data quantification (Table. S2). We further extended the inject-embed method to all six liver cancer cell lines and LX2 cell line. As shown in Fig. 1 C and Fig. S2, all cell lines formed spheroids exhibiting a variety of morphologies and sizes. Morphologically, spheroids from HCC cell lines often exhibited cellular protrusions at their sphere edges, whereas Huh1 spheroids presented a compact structure with relatively smooth edges. Meanwhile, spheroids from CCA cell lines typically featured smooth edges and a dense composition. In contrast, LX2 spheroids displayed stellate protrusions. Regarding size, all tested cell lines demonstrated a progressive increase in spheroid diameter over the culture period (Fig. S2). A size comparison conducted at 4 days of culture presented noticeable differences (Fig. 1 C). Among HCC cell lines, HLE and HLF spheroids were similar in size and larger than those of Huh7 and Huh1. Within CCA cell lines, RBE spheroids were significantly smaller than those of HUCCT1. Meanwhile, LX2 spheroids were generally smaller than those formed by liver cancer cell lines. These results confirm that the inject-embed method reliably supports spheroid formation across all tested cell lines in 3D mono-culture conditions. The inject-embed 3D culture method allowed the aggregate formation of liver cancer cells co-cultured with LX2 To determine the utilization of inject-embed method for co-culturing cancer cells with their microenvironment cells, we labeled liver cancer cell lines with GFP lentivirus and LX2 cells with RFP lentivirus (Fig. S3A). Both CCA HUCCT1-GFP cells and LX2-RFP cells were harvested and mixed at a 1:1 ratio. A total of 2,000 mixed cells were injected into Matrigel following the inject-embed method procedure. This resulted in the formation of yellow aggregates containing both GFP- and RFP-labeled cells (Fig. 2 A). Similar results were observed for the co-culture of HCC Huh7-GFP and LX2-RFP cells (Fig. 2 B), as well as for co-cultures of LX2-RFP with the remaining four liver cancer cell lines (Fig. S3B-E). Over the culture period, the size of these aggregates with mixed population increased, while their structural architecture remained stable. Fig. 2. Open in a new tab The inject-embed 3D culture allowed the aggregate formation of liver cancer cells co-cultured with LX2. ( A ) Images of an aggregate at 2, 4, 6 days after the injection of HUCCT1-GFP with LX2-RFP. An initial cell number of 1,000/1,000 (HUCCT1/LX2) were used. ( B ) Images of an aggregate at 2, 4, 6 days after the injection of Huh7-GFP with LX2-RFP. An initial cell number of 1,000/1,000 (Huh7/LX2) were used. ( C ) Images of HUCCT1-GFP and LX2-RFP co-cultured in 3D using the inject-embed method, mix-embed method, and hanging drop method. All images were captured on the 6th day of culture. Scale bars are shown in the images We have also compared the aggregate formation across three 3D culture methods, i.e., the inject-embed method, the mix-embed method, and the hanging drop method, using co-cultures of GFP-labeled liver cancer cells and RFP-labeled LX2 cells. Two conditions were considered, both at a 1:1 GFP: RFP cell ratio, i.e., with identical initial cell number and identical cell densities (Fig. S4). Under both conditions, yellow aggregates were more prevalent in the inject-embed method than in the other two methods. In the inject-embed method, yellow aggregates constituted 19.1–55.0% of the structures formed by liver cancer cells and LX2 cells, with their size increasing over prolonged culture periods (Fig. S5A). In contrast, the mix-embed method produced very few yellow aggregates (< 1%) under both conditions. In all three cells, both the number and size of yellow aggregates appeared to be lower than those in the inject-embed method even when their cell densities were the same (Fig. S5B-C). In the hanging drop method, cells tended to quickly merge into one single, large yellow aggregate as culture duration extended (Figs. S4-5). In addition, the spatial organization of aggregates formed by each liver cancer cell line with LX2 was consistent across all three different methods (Fig. S4). Comparably in Fig. 2 C, all three methods with the same cell density presented a clear consistent architecture of well-organized aggregates with HUCCT1 at the core surrounded by LX2. Together, the inject-embed method stably supports co-culture of liver cancer cells and LX2 cells and facilitates the formation of aggregates with defined spatial organization. Spheroids of 3D mono-culture exhibited both proliferation and aggregation capabilities To further characterize mono-culture spheroids from the inject-embed method, we evaluated liver cancer cell lines with different initial cell numbers over an 8-day culture period. For CCA HUCCT1 cells, the higher the initial cell number was, the faster the spheroids formed and the larger they were (Fig. 3 A). At 2 days post-injection, spheroids had formed across all groups. Quantitatively, spheroid diameter increased with higher initial cell numbers and continued to grow over the culture period (Fig. 3 B). Meanwhile, the number of spheroids initially increased as the initial cell count rose, but subsequently decreased with this reduction being more pronounced in groups with higher initial cell counts (Fig. 5 C). Within each group, spheroid numbers also decreased over time (Fig. 3 C). Similar patterns were observed for HCC Huh7 cells (Fig. 3 D-F), LX2 cells (Fig. S6), and the other four liver cancer cell lines (Fig. S7A-D). Since that the reduction in spheroid numbers over time coupled with the observation that higher initial cell densities yielded larger but fewer spheroids, it was likely that the increase in spheroid diameter resulted from both the growth of individual spheroids and their aggregation. The proliferation of spheroids was then confirmed by calculating of the total volume of all spheroids. As shown in Fig. 3 G-H, S6D, and S7E, the total spheroid volume increased over the culture time in each group across different initial cell numbers and in six out of seven tested cell lines. The exception was the RBE cell line, which showed no increase in total spheroid volume, likely due to its characteristically small spheroid size (Fig. 1 C, S2, and S7E). Spheroid aggregation was assessed by continuous image acquisition over the culture period. We observed that spheroids gradually moved closer together and progressively merged to form larger spheroids. This phenomenon was consistently noted in HUCCT1 (Fig. 4 A), Huh7 (Fig. 4 B), LX2 (Fig. 4 C), and other liver cancer cell lines (Fig. S8). Additional confirmation of this aggregation behavior was also provided through time-lapse photography over 64 h using HLE cell line (Fig. 4 D). Thus, these data highlight the occurrence of spheroid aggregation in the inject-embed method. Fig. 3. Open in a new tab Spheroids from 3D mono-culture of cells showed proliferation. (A) Images of 3D mono-culture of HUCCT1 cell line with the inject-embed method at the indicated days. Different initial cell numbers were used. (B-C) Quantitative diameters (B) and numbers (C) of the HUCCT1 spheroids at 2, 4, 6 days with different initial cell numbers. (D) Images of 3D mono-culture of Huh7 cell line with the inject-embed method at the indicated days. Different initial cell numbers were used. (E-F) Quantitative diameters (E) and numbers (F) of the Huh7 spheroids at 2, 4, and 6 days with different initial cell numbers. (G-H) Total spheroid volume of HUCCT1 (G) and Huh7 (H) at 2, 4, and 6 days. Scale bars are shown in the images Fig. 5. Open in a new tab The features of aggregates formed by liver cancer cell lines and co-cultured LX2. (A) Representative images of spheroids and aggregates of the co-cultured GFP-labeled liver cancer cell lines (Huh7, HLF, Huh1, HLE, HUCCT1, RBE) with RFP-labeled LX2 cells at 4 days, with an initial cell number of 1,000 for each cell type. (B) The quantitative analysis of the spheroids and aggregates from panel (A) . Green, liver cancer cell spheroids. Red, LX2 spheroids. Yellow, aggregates of liver cancer cells and LX2. (C) Representative images when different ratios of GFP-labeled liver cancer cells and RFP-labeled LX2 were co-cultured with a total number of 2,000. Images were taken at 6 days of co-culture. (D-E) Quantitative data of spheroids and aggregates at their diameters (D) and numbers (E) for panel C. (F) Representative images when different numbers of GFP-labeled liver cancer cells and RFP-labeled LX2 were co-cultured with a 1:1 ratio. Images were taken at 6 days of co-culture. (G-H) Quantitative data of spheroids and aggregates at their diameters (G) and numbers (H) for panel F Fig. 4. Open in a new tab Spheroids from 3D mono-culture of cells showed aggregation. (A-C) Images of spheroid aggregation within the same field for HUCCT1 at 2, 4, 6 days (A) , Huh7 at 6, 8, 10 days (B) , and LX2 at 2, 4, 6 days (C) . Initial cell numbers were 4,000 for HUCCT1, 1,000 for Huh7, and 2,000 for LX2. (D) The 64 h time-lapse spheroid images for HLE cells. One photo was taken at every 8 h. (E) Representative images of cell protrusions at the surface of spheroids. Protrusions were categorized based on morphology into “ball hands”, “needle hands”, “leaf hands” and “no hands”. (F) Images of HLE cell spheroids at 8, 10, 12 days, with an initial cell number of 1,000. Scale bars are shown in the images We also compared the proliferation and aggregation features of inject-embed method with the other two 3D culture method under two conditions, i.e., identical initial cell numbers and identical initial cell densities (Figs. S9-11). In general, consistent data were observed among three methods that the spheroid diameter increased with culture duration, accompanied by a reduction in spheroid number. However, spheroids in the mix-embed method were smaller than those in inject-embed method under each condition (Fig. S10). Spheroid aggregation and fusion in the mix-embed method occurred at a much slower rate compared to the inject-embed method (< 4 days vs. >6 days), suggesting its limited spheroid mobility (Fig. S11). In contrast, in the hanging drop method, spheroids fused rapidly, with nearly all cells aggregating into a single spheroid by day 4 of culture (Fig. S9). Taken together, spheroid enlargement in the inject-embed method was not solely attributable to growth of individual spheroids but also significantly influenced by aggregation effects. This method enables reliable quantification of the spheroid proliferation and aggregation. Spheroids of 3D mono-cell culture exhibited size limitations and unique edge morphology Given the observed variations in spheroid numbers, sizes, and total volumes across liver cancer cells and LX2, we further assessed any potential constraints on their spheroid size and culture duration. An initial cell number 4,000 was used for each cell line. The maximum spheroid diameter was defined as the average diameter of spheroids when spheroid growth ceased. The longest culture duration was defined as the number of days required to reach this maximum diameter. Notable limitations in the spheroid size and culture duration were evident for different cell lines (Table. S3). Maximum spheroid diameter varied widely, ranging from 48.9 μm (RBE cells) to 445.7 μm (HLF cells), with corresponding culture durations spanning 6 to 12 days. Beyond the longest culture duration, spheroids either showed enhanced sphericity without further diameter increase or underwent cell death. Moreover, during the culture process, cellular protrusions were frequently observed along the spheroid edges, displaying diverse morphologies that resembled tentacles. Four distinct types were observed, i.e., “ball hands”, “needle hands”, “leaf hands” and “no hands”. Huh7 spheroids commonly exhibited “ball hands”. “Needle hands” were typical for HLE, HLF, and LX2. “Leaf hands” were also noted in HLE and HLF spheroids (Fig. 4 E). In contrast, spheroids from HUCCT1, RBE, and Huh1 typically featured smooth edges without protrusions (“no hands”) (Fig. 4 E). Interestingly, some spheroids were interconnected by cellular protrusions prior to aggregation, which then gradually shortened over the culture period, bringing the spheroids closer until they ultimately aggregated (Fig. 4 F, S8C). These observations suggest that cellular protrusions likely play a role in facilitating spheroid approximation. Aggregates of liver cancer cells with LX2 actively formed in the inject-embed 3D co-culture system Six GFP-labeled liver cancer cell lines were individually co-cultured with RFP-labeled LX2 cells using the inject-embed 3D co-culture method. As shown in Fig. 5A and S12A, each co-culture produced three types of spheroids, i.e., green spheroids (liver cancer cell spheroids), red spheroids (LX2 spheroids), and yellow aggregates (mixture of green and red cells) indicating aggregates of liver cancer cells and LX2. At 4 days post-injection, with an initial cell ratio of 1:1 (1,000 cell each), yellow aggregates comprised approximately one-third of the total spheroids formed across all co-cultures (Fig. 5 B, S12B). To further assess the aggregate formation efficiency of liver cancer cells with LX2 in the inject-embed method, we tested different initial cell ratios while maintaining a total of 2,000 cells with Huh7-GFP and LX2-RFP cells. Three groups were examined with Huh7:LX2 ratios of 1:3 (500:1,500), 1:1 (1,000:1,000), and 3:1 (1,500:500) (Fig. 5 C). The sizes of green or red spheroids showed minimum increase over time, whereas yellow aggregate sizes increased significantly (Fig. 5 D). Concurrently, the number of all three spheroid types generally decreased over culture period (Fig. 5 E). These findings indicated active aggregation between Huh7 and LX2 spheroids during co-culture, rather than aggregation within Huh7 or LX2 populations alone. Meanwhile, among the three ratios, the 1:1 group exhibited the highest number of yellow aggregates (Fig. 5 E), implying a 1:1 initial cell ratio as an optimal condition for inject-embed co-culture of Huh7 and LX2. Next, three groups with a 1:1 Huh7:LX2 ratio and varying total cell numbers (1,000, 2,000, and 4,000) were tested (Fig. 5 F). Consistent with prior observations, yellow aggregate sizes increased markedly over time, while green and red spheroid sizes remained largely unchanged (Fig. 5 G). The number of all three spheroid types was generally reduced over time (Fig. 5 H). Additionally, lower initial cell numbers correlated with fewer initial mixed aggregates (Fig. 5 H), likely due to the lower initial cell density and greater intercellular distances. These data suggest that higher initial cell numbers enhance effective aggregate formation in the co-culture system. To further investigate spheroid aggregation dynamics, Huh7-GFP and LX2-RFP co-culture spheroids or aggregates were monitored over time via photography. The data revealed that the nearby yellow aggregates approached each other, merged morphologically, and formed larger aggregates while maintaining a consistent 3D architecture before and after fusion (Fig. 6 A). This behavior was also observed in HUCCT1 (Fig. 6 B) and the other four liver cancer cells (Fig. S13). Additionally, smaller green or red spheroids gradually integrated into yellow aggregates with larger diameters over the culture period. Similar to the inject-embed 3D mono-culture system, cellular protrusions were also observed during the merging process of aggregates in the co-culture system. As shown in Fig. 6 C, hand-like projections extended from either the LX2-RFP cells or HLF-GFP cells, indicating that both cell types were capable of generating protrusions during aggregation in the co-culture system. Comparable results were also noticed in the other three HCC cell lines, i.e., Huh7, Huh1 and HLE (Fig. S14A-C). Intriguingly, in co-cultures of CCA cell lines with LX2, all protrusions originated exclusively from LX2-RFP cells (Fig. 6 D, S14D). Quantitative data confirmed these findings (Fig. 6 E). Together, these results demonstrate that aggregates in the co-culture system actively undergo further aggregation, in which cell protrusions were involved. Fig. 6. Open in a new tab Aggregates formed with liver cancer cells and LX2 further actively approached and merged together. ( A ) Representative images of the aggregates in inject-embed 3D co-culture system of 1,000 HUCCT1 and 1,000 LX2 over culture time. The representative approached aggregates were circled out. ( B ) Representative images of the aggregates in inject-embed 3D co-culture system of 1,000 Huh7 and 1,000 LX2 over culture time. ( C ) Representative images of the HLF/LX2 aggregates with the hand-like projections between the aggregations. Both HLF-GFP and LX2-RFP hand-like projections were noticed. ( D ) Representative images of the HUCCT1/LX2 aggregates with the hand-like projections between the aggregations. Only LX2-RFP hand-like projection was noticed. ( E ) The quantified “hand-like” projections from either liver cancer cells labeled with GFP or LX2-RFP for each cell lines Aggregates formed by liver cancer cells with LX2 exhibited two distinct 3D architecture types In inject-embed 3D co-culture method, two distinct 3D architectures were observed, a well-organized structure with a core of green cells surrounded by red cells, and a hybrid structure with intermingled green and red cells (Figs. 2 and 6 , S13-14). Theoretically, co-culturing liver cancer cell lines with LX2 cells could result in three assembly patterns, i.e., type 1, a disordered hybrid where both cell types intermingled; type 2, liver cancer cells encircled by the LX2 cells; and type 3, LX2 cells encircled by the liver cancer cells (Fig. 7 A). Fig. 7. Open in a new tab HCC and CCA cells exhibited distinct 3D architectures during co-culture with LX2. ( A ) Schematic illustration of the assembly patterns of liver cancer and LX2 cells within their aggregates. ( B ) Representative aggregate images and their assembly patterns of four HCC cell lines (Huh7, HLF, Huh1, HLE) co-cultured with LX2. ( C ) Representative aggregate images and their assembly patterns of two CCA cell lines (HUCCT1 and RBE) co-cultured with LX2. ( B - C ) Initial cell numbers had been set to 1,000 for both liver cancer cells and LX2 cells. The frequency of the different assembly pattern was also quantified. Images of aggregates and quantitative data were from day 6. Student’s t-test was used for statistical analysis We then quantified the 3D architecture types formed by the co-culture of different liver cancer cells with LX2. When the four HCC cell lines (Huh7, HLF, Huh1, HLE) were co-cultured with LX2, they predominantly formed the type 1 aggregates, characterized by a hybrid intermingling of both cell types (Fig. 7 B, > 90%). In contrast, two CCA cell lines (HUCCT1, RBE) with LX2 predominantly formed type 2 aggregates, exhibiting an architecture where CCA cells formed a central core surrounded by LX2 cells (Fig. 7 C, > 90%). Type 3 aggregates were rare in co-culture of LX2 with either HCC or CCA pathological subtype of liver cancer cells. Thus, the inject-embed 3D co-culture enabled the formation of specific 3D architectures, with distinct aggregate types emerging based on the pathological subtype of liver cancer cells co-cultured with LX2. Such a co-culture method likely provided a new tool for exploring cell-cell interactions under certain spatial organization and investigating the related mechanisms. Functional enrichment analysis based on RNA sequencing date of 3D spheroids from the inject-embed method 3D culture models can exhibit distinct gene expression and protein expression profiles compared to 2D cultures, which has been documented in liver cancer [ 24 ]. We thus conducted the RNA sequencing on two commonly used HCC cell lines (Huh7 and HLF) under three different culture conditions, i.e., 2D, inject-embed, and mix-embed 3D cultures. Two biological duplicates were used per condition (Fig. 8 A). A total of 27,516 genes were detected in these 12 samples. An unsupervised Pearson correlation analysis showed that samples from the same cell line grouped together regardless of the culture methods, indicating that heterogeneity among cancer cell lines outweighed the variability introduced by different culture conditions (Fig. 8 B). The principal component analysis (PCA) was then performed in each cell line, which showed clear separation among three culture conditions in both cell lines (Fig. S15A), indicating that transcriptomic profiles are associated with culture conditions. We then conducted class comparison analysis of transcriptome data between each 3D culture method and 2D culture. There were 561 genes with the significantly altered expression in the inject-embed 3D culture condition compared to 2D culture ( P < 0.05, |Log2 of fold| > 0.2), while 1,014 genes were identified in the mix-embed culture condition compared to 2D culture (Fig. 8 C). Although 213 genes were commonly altered in two 3D culture conditions, a larger number of genes presented the unique alterations in each 3D culture method, suggesting the potential similarities and differences of the transcriptome alteration between two 3D culture methods. Fig. 8. Open in a new tab The biological features of spheroids from the inject-embed 3D culture method in cancer related studies. ( A ) Schematic illustration of the cells cultured under different condition used for RNA sequencing. ( B ) Pearson correlation analysis of all sequenced samples based on all the detected genes. ( C ) Venn diagram analysis with differentially expressed genes between two different 3D culture methods and 2D culture. ( D ) GSEA results with the differentially expressed genes between two 3D culture methods and 2D culture. The enriched signatures with P < 0.05 and Normalized Enrichment Score (NES) > 1.5 were shown. The X-axis represents NES and the Y-axis lists the gene set names. The enriched pathways are color-coded by functional category as indicated. ( E ) Drug resistance to cisplatin, doxorubicin, and 5-Fu was compared between Huh7 and HLF cells under 2D culture and their spheroids from the inject-embed 3D culture method. Two-way ANOVA analysis was used. ( F - G ) The spheroid formation with inject-embed method for Huh7 cells ( F ) and HLF cells ( G ) transfected with Si-Ctrl and Si-CYP39A1. ( H - I ) The spheroid formation with inject-embed method for Huh7 cells ( H ) and HLF cells ( I ) transfected with Ctrl plasmid and CYP39A1 plasmid. ( F - I ) Student’s t-test was used for spheroid number comparison and non-parametric t-test was used for spheroid diameter comparison In this case, Gene Set Enrichment Analysis (GSEA) (Fig. 8 D and Table. S4), Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) analysis (Fig. S15B-E) were then performed with these altered genes to understand the biological implications of these genes. The results demonstrated significant enrichment of gene sets related to stemness, development and tumor malignancy in both 3D culture methods, indicating these culture methods enhanced cancer stemness properties compared to 2D culture (Fig. 8 D, S15B-C). In addition to shared pathways, each 3D culture methods exhibited distinct enrichment patterns. In the inject-embed 3D method, pathways related to mitochondrial electron transport and oxidative phosphorylation activities were noticeably enriched, suggesting enrichment of gene expression signatures related to mitochondrial respiration and oxidative phosphorylation (Fig. 8 D, S15B, S15D). In contrast, the mix-embed method showed enrichment of HIF-1α signaling activation, immune-related gene signatures, and glucose metabolism (Fig. 8 D, S15C, S15E). These results indicate enrichment of hypoxia-associated transcriptional signatures together with glucose metabolism–related pathways in the mix-embed culture condition. Meanwhile, the enriched immune-related signatures might be a consequence of the possible cell death and/or hypoxic environment in this method. Together, the RNA sequencing results suggest a cancer stemness transcriptomic alteration for cells cultured in both inject-embed 3D culture method and mix-embed 3D culture method, and a possible different favorable energy metabolism process with mitochondrial respiration in the inject-embed 3D culture method and glucose metabolism in mix-embed 3D culture method. Spheroids of the inject-embed method expressed stemness genes and exhibited chemoresistance We next evaluated some of the biological features of the spheroids from the inject-embed 3D culture method with Huh7 and HLF. The spheroid formation assay with ultra-low attachment plates was known to evaluate self-renewal ability and enrich cancer stem cell populations [ 25 , 26 ]. Here, we collected the spheroids from the ultra-low attachment plates and the inject-embed 3D culture method, with cells from 2D culture serving as the control. The HCC stemness related genes were examined, including CD24, CD44, CD133, EpCAM, OCT4, NANOG, and SOX2. As shown in Fig. S16, overall, the stemness gene expression pattern in spheroids from both 3D methods were similar and generally higher than those in 2D-cultured cells. These results indicated that the inject-embed method also induces gene expression patterns associated with tumor stemness compared to 2D culture, as does the ultra-low attachment approach. Meanwhile, while HLE and RBE cell lines were not able to form spheroid with the ultra-low attachment plate method [ 23 , 27 ], both successfully formed spheroids with the inject-embed method (Fig. 1 C). Collectively, these data indicate that the inject-embed method can be applied to the analysis of tumor stemness-associated phenotypes. As previously shown, 3D cultured cells were more chemo-resistant than 2D cultured cells [ 2 , 28 – 30 ]. In this case, we compared the chemotherapeutic responses of the inject-embed 3D spheroids and the conventional 2D cultured cells. Huh7 and HLF cells, under 2D culture and inject-embed 3D culture, were treated with three different chemotherapeutic reagents, i.e., cisplatin, doxorubicin and 5-fluorouracil (5-Fu). The relative viability of 3D spheroids was significantly higher than that of 2D cells in response to three different chemo-drugs, suggesting greater chemo-resistance of 3D spheroids ( P < 0.001 for each comparison, Fig. 8 E). This increased resistance may be partially attributed to diffusion gradients within spheroids, which more closely mimicked the in vivo conditions. Thus, the results from the inject-embed method appeared to be more representative in drug experimentation than the 2D culturing method. Spheroid formation of the inject-embed method could be regulated by functional genes in HCC development Previous research findings from our lab demonstrate that CYP39A1, a member of the cytochrome P450 family with female-biased expression, strongly inhibits HCC development [ 20 ]. With the inject-embed method, we examined the spheroid formation in cells with the altered CYP39A1 gene expression. As shown in Fig. 8 F, silencing CYP39A1 led to larger spheroids and an increased spheroid count compared to the control siRNA group. Comparable data were obtained in both Huh7 and HLF cells (Fig. 8 G). Consistently, overexpressing CYP39A1 significantly reduced spheroid size and number in both cell lines (Fig. 8 H-I). These data not only showed that the spheroids from inject-embed method could be responsive to regulatory changes, but also consistently demonstrated the tumor-suppressive attributes of CYP39A1. Similar data were also obtained with other regulatory factors (data not shown). Therefore, the inject-embed method could serve as a complementary approach in cellular phenotyping experiments. Taken together, the inject-embed method, as a novel 3D cell culture technique, exhibits broad applications in stemness evaluation, drug testing, and cellular phenotyping. Discussion 3D cell culture techniques effectively address the limitations of conventional 2D cell culture and animal experiments, capturing the growing interest in the oncology research community. These technologies are predominantly utilized for constructing patient-derived organoids. For in vitro cell lines, common 3D cell culture methods include the scaffold-free method (such as the hanging drop method) and scaffold-based method (such as the mix-embed method). While the traditional scaffold-based method offers ease of operation and uniform cell distribution, it limits cell motility and cell-cell interaction. Meanwhile, spheroids formed in this method were located on different planes of scaffold, making observation and quantification challenging. In contrast, the scaffold-free method allowed greater cell mobility but lacked a matrix environment. In this study, we developed a new 3D cell culture method, the inject-embed method, which not only provides both liquid interface and ECM interface for cultured cells, but also can accommodate both single cell type cultures and co-cultures of two cell types. This method mainly involved injecting a cell suspension in culture medium into the internal layer of the Matrigel scaffold using a pipette. In this case, cells in this 3D culture system were then both within a matrix environment and possessed a certain degree of mobility allowing active cell-cell interactions. With this method, spheroids and aggregates formed effectively, exhibiting distinct size and morphology representing their biological features. Notably, this method enabled spheroid formation even for cell lines that could not generate spheroids using ultra-low attachment plates, and also for cell lines that showed poor aggregation in the hanging drop method. As spheroids and aggregates generated by the inject-embed method are distributed within a relatively confined Z-plane, this approach offers improved consistency for imaging and quantitative analysis. We applied the inject-embed method to perform 3D mono-cultures of seven cell lines and all cell lines formed cell spheroids with 3D structures. Over the culture period, the spheroid diameters gradually increased and eventually approached a plateau, a phenomenon that is partly consistent with the presence of gradients of oxygen, nutrients and metabolic waste observed in tumor tissues. Once a certain size is reached, cells within the core may experience hypoxia, nutrient deprivation, and difficulties in waste excretion, leading to halted growth or even necrosis [ 31 ]. Additionally, spheroids formed by different cell lines varied in shape. HCC cell line-derived spheroids often exhibited cellular protrusions at their periphery. CCA spheroids had smoother edges. LX2 spheroids featured stellar protrusions. Meanwhile, different initial cell number could be adjusted based on experimental needs, for which lower counts facilitated observation of individual spheroid formation and higher counts enhanced intercellular aggregation effects. With the inject-embed 3D co-culture of liver cancer cells with LX2 cells, aggregates formed and progressed toward further aggregation. These aggregates featured two distinct 3D architecture. A disordered hybrid architecture, where both cell types intermingled, was formed by HCC cells and LX2. In contrast, aggregates of CCA cells with LX2 exhibited a well-organized architecture, with CCA cells at the core surrounded by LX2 cells, suggesting the potential cellular polarity at the CCA cell-LX2 interface. Notably, the in vivo growth characteristics of HCC and CCA differ, with CCA tumors exhibiting more pronounced fibrotic features compared to HCC [ 32 , 33 ]. Thus, it is interesting to investigate the biological implication of such a 3D structure and the biological mechanisms behind these unique architectures. In future studies, it will be great to further incorporate additional cell types, such as Kupffer cells and vascular endothelial cells, into the co-culture system to better recapitulate the in vivo microenvironment. Moreover, the biological features of these spheroids from the inject-embed 3D culture were briefly examined, including their altered transcriptomic profiles, enrichment of gene sets associated with oxidative phosphorylation, elevated expression of stemness-related genes and chemo resistant phenotypes compared to 2D-cultured cells. The spheroid formation was also regulated by the tumor related genes, indicating its potential broad utilization in oncology research. It would be valuable to deeply investigate how the transcriptomic alterations contribute to phenotypes such as spheroid aggregation. In addition to using established cell lines for 3D culture, future studies may extend this method to novel applications, such as culturing primary cells, cells derived from murine tumors, and even patient-derived cells. Transplanting spheroids or aggregates into mice to establish xenograft models could further expand its utility. We also observed a ubiquitous tendency for spheroids and aggregates to migrate towards each other and ultimately merge into a larger one. Cellular protrusions extending from the periphery of the spheroids and aggregates may initiate this process, and further research is also needed to elucidate the underlying mechanisms. Although the primary aim of this study was to develop and characterize the inject-embed method, a direct comparison with the other two 3D methods (mix-embed and hanging drop) would be valuable. Here we only performed a relative comparison and mainly used them as references. The current analysis was based on the representative images, which potentially limited the direct comparison. The inject-embed method was naturally suitable for imaging capture and current analysis as its spheroids were largely confined to a single focal plane, unlike the spheroids generated by the mix-embed method (spatially dispersed). Thus, it would be good to collect all the spheroids for a much direct comparison, such as a 3D reconstruction of the entire unit with all spheroids through imaging. Unfortunately, we faced difficulties in optimizing the imaging condition. We are currently working on applying advanced microscopy and specialized culture plates to achieve this goal, and hope to share it in future studies. In summary, we have established an inject-embed 3D culture method as a novel technique that enabled the formation of spheroids and aggregates from both single cell type cultures and co-cultures of two cell types. This approach provides a reproducible and practical platform for analyzing cancer cell–associated phenotypes in a 3D context with significant potential for advancing oncology research. Conclusion The newly generated inject-embed 3D culture method provides a novel, highly reproducible and scalable platform to study cancer spheroid/aggregate behavior and interactions between cancer cells and their surrounding microenvironment. With these, this method offers a 3D in vitro research tool with potential applications in cancer-related studies such as functional studies and drug screening for personalized medicine. Supplementary Information Below is the link to the electronic supplementary material. 12935_2026_4250_MOESM1_ESM.zip (13.2MB, zip) Supplementary Material 1: Movie AVI file 1. The AVI file for the HLF spheroids from the inject-embed method. Z-stack images were sequentially acquired with 33 μm spacing between adjacent frames. A total of 40 consecutive frames were captured and the corresponding AVI file were generated for visualization. Movie AVI file 2. The AVI file for the HLF spheroids from the mix-embed method. Z-stack images were sequentially acquired with 33 μm spacing between adjacent frames. A total of 40 consecutive frames were captured and the corresponding AVI file were generated for visualization. Supplementary Material 2 (8.2MB, docx) Acknowledgements The authors thank Dr. Shengda Lin for his critical reading and comments, and Ms. Julia Zheng for her language editing. Abbreviations 2D Two-dimensional 3D Three-dimensional HCC Hepatocellular carcinoma CCA Cholangiocarcinoma CAFs Cancer-associated fibroblasts ECM Extracellular matrix PBS Phosphate-buffered saline GO Gene ontology GSEA Gene set enrichment analysis KEGG Kyoto encyclopedia of genes and genomes NES Normalized enrichment score PCA Principal component analysis TME Tumor microenvironment 5-Fu 5-fluorouracil Author contributions Conception and design: JJ, HZ, SX; Development of methodology: HZ, SX, JP, JZ, DW, PX; Acquisition of data: HZ, SX, DW; Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): HZ, PX, JJ; Writing, review, and/or revision of the manuscript: HZ, SX, GL, SR, CY, JJ; Administrative, technical, or material support: PX, GL, CY; Study supervision: JJ. Funding This work was supported by the ‘Leading Goose’ R&D Program of Zhejiang Province (2023C03052) (J.Ji), National Natural Science Foundation of China (82361148723; 82073055) (J. Ji), the Fundamental Research Funds for the Central Universities in China (J. Ji), the MOST-NSFC grants of China (2022YFA0806800, 32170768) (P. Xia), and the National Natural Science Foundation of China (82030019) (C. Yu). Data availability The RNA-seq data of this manuscript have been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession number GSE298810. The authors also confirm that all other data supporting the findings of this study are available within the article and/or its supplementary information file. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Hanyuan Zhang and Shuting Xue wish it to be known that, in their opinion, the first two authors should be regarded as joint first authors. Contributor Information Chaohui Yu, Email: [email protected]. Junfang Ji, Email: [email protected]. References 1. Nath S, Devi GR. Three-dimensional culture systems in cancer research: Focus on tumor spheroid model. Pharmacol Ther. 2016;163:94-108. [ DOI ] [ PMC free article ] [ PubMed ] 2. Jubelin C, Muñoz-Garcia J, Griscom L, Cochonneau D, Ollivier E, Heymann M-F, et al. Three-dimensional in vitro culture models in oncology research. Cell bioscience. 2022;12(1):155. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Kim JB. Three-dimensional tissue culture models in cancer biology. Sem Cancer Biol. 2005;15(5):365–77. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Costard LS, Hosn RR, Ramanayake H, O’Brien FJ, Curtin CM. Influences of the 3D microenvironment on cancer cell behaviour and treatment responsiveness: A recent update on lung, breast and prostate cancer models. Acta Biomater. 2021;132:360–78. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Laschke MW, Menger MD. Life is 3D: Boosting Spheroid Function for Tissue Engineering. Trends Biotechnol. 2017;35(2):133–44. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Kang S-M, Kim D, Lee J-H, Takayama S, Park JY. Engineered Microsystems for Spheroid and Organoid Studies. Adv Healthc Mater. 2021;10(2):e2001284. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Landry J, Bernier D, Ouellet C, Goyette R, Marceau N. Spheroidal aggregate culture of rat liver cells: histotypic reorganization, biomatrix deposition, and maintenance of functional activities. J Cell Biol. 1985;101(3):914–23. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Bartosh TJ, Ylöstalo JH, Mohammadipoor A, Bazhanov N, Coble K, Claypool K, et al. Aggregation of human mesenchymal stromal cells (MSCs) into 3D spheroids enhances their antiinflammatory properties. Proc Natl Acad Sci USA. 2010;107(31):13724–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Bhang SH, Cho S-W, La W-G, Lee T-J, Yang HS, Sun A-Y, et al. Angiogenesis in ischemic tissue produced by spheroid grafting of human adipose-derived stromal cells. Biomaterials. 2011;32(11):2734–47. [ DOI ] [ PubMed ] [ Google Scholar ] 10. Wang A, Madden LA, Paunov VN. Advanced biomedical applications based on emerging 3D cell culturing platforms. J Mater Chem B. 2020;8(46):10487–501. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Haisler WL, Timm DM, Gage JA, Tseng H, Killian TC, Souza GR. Three-dimensional cell culturing by magnetic levitation. Nat Protoc. 2013;8(10):1940–9. [ DOI ] [ PubMed ] [ Google Scholar ] 12. Lee GY, Kenny PA, Lee EH, Bissell MJ. Three-dimensional culture models of normal and malignant breast epithelial cells. Nat Methods. 2007;4(4):359–65. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Abuwatfa WH, Pitt WG, Husseini GA. Scaffold-based 3D cell culture models in cancer research. J Biomed Sci. 2024;31(1):7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Nikolova MP, Chavali MS. Recent advances in biomaterials for 3D scaffolds: A review. Bioact Mater. 2019;4:271–92. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Cao H, Duan L, Zhang Y, Cao J, Zhang K. Current hydrogel advances in physicochemical and biological response-driven biomedical application diversity. Signal Transduct Target Ther. 2021;6(1):426. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Caliari SR, Burdick JA. A practical guide to hydrogels for cell culture. Nat Methods. 2016;13(5):405–14. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Kleinman HK, Martin GR. Matrigel: basement membrane matrix with biological activity. Sem Cancer Biol. 2005;15(5):378–86. [ DOI ] [ PubMed ] [ Google Scholar ] 18. Sumbalova Koledova Z. 3D Cell Culture: Techniques For and Beyond Organoid Applications. Methods Mol Biol. 2024;2764:1-12. [ DOI ] [ PubMed ] 19. Zhang JJ, Ji FB, Tan YQ, Zhao L, Zhao YZ, Liu JX, et al. Oncogenic Roles of Laminin Subunit Gamma-2 in Intrahepatic Cholangiocarcinoma via Promoting EGFR Translation. Adv Sci. 2024;11:21. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Ji F, Zhang J, Liu N, Gu Y, Zhang Y, Huang P, et al. Blocking hepatocarcinogenesis by a cytochrome P450 family member with female-preferential expression. Gut. 2022;71(11):2313–24. [ DOI ] [ PubMed ] [ Google Scholar ] 21. Carloni V, Luong TV, Rombouts K. Hepatic stellate cells and extracellular matrix in hepatocellular carcinoma: more complicated than ever. Liver international: official J Int Association Study Liver. 2014;34(6):834–43. [ DOI ] [ PubMed ] [ Google Scholar ] 22. Affo S, Filliol A, Gores GJ, Schwabe RF. Fibroblasts in liver cancer: functions and therapeutic translation. Lancet Gastroenterol Hepatol. 2023;8(8):748–59. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Zhao YZ, Xue ST, Wei DD, Zhang JJ, Zhang NC, Mao LP et al. Membrane RRM2-positive cells represent a malignant population with cancer stem cell features in intrahepatic cholangiocarcinoma. J Exp Clin Canc Res. 2024;43(1):255. [ DOI ] [ PMC free article ] [ PubMed ] 24. Aa AH, Cervantes-Gracia K, Chahwan R, Amin A. Modelling liver cancer microenvironment using a novel 3D culture system. Sci Rep. 2022;12(1):8003. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Lee TK-W, Guan X-Y, Ma S. Cancer stem cells in hepatocellular carcinoma - from origin to clinical implications. Nat Rev Gastroenterol Hepatol. 2022;19(1):26–44. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Yang L, Shi P, Zhao G, Xu J, Peng W, Zhang J, et al. Targeting cancer stem cell pathways for cancer therapy. Signal Transduct Target Ther. 2020;5(1):8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Gu Y, Wei X, Sun Y, Gao H, Zheng X, Wong LL, et al. miR-192-5p Silencing by Genetic Aberrations Is a Key Event in Hepatocellular Carcinomas with Cancer Stem Cell Features. Cancer Res. 2019;79(5):941–53. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Zanoni M, Cortesi M, Zamagni A, Arienti C, Pignatta S, Tesei A. Modeling neoplastic disease with spheroids and organoids. J Hematol Oncol. 2020;13(1):97. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Decarli MC, Amaral R, Santos DPD, Tofani LB, Katayama E, Rezende RA et al. Cell spheroids as a versatile research platform: formation mechanisms, high throughput production, characterization and applications. Biofabrication. 2021;13(3):1-37. [ DOI ] [ PubMed ] 30. Hwang HJ, Oh M-S, Lee DW, Kuh H-J. Multiplex quantitative analysis of stroma-mediated cancer cell invasion, matrix remodeling, and drug response in a 3D co-culture model of pancreatic tumor spheroids and stellate cells. J Experimental Clin Cancer Research: CR. 2019;38(1):258. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Thoma CR, Zimmermann M, Agarkova I, Kelm JM, Krek W. 3D cell culture systems modeling tumor growth determinants in cancer target discovery. Adv Drug Deliv Rev. 2014;69–70:29–41. [ DOI ] [ PubMed ] [ Google Scholar ] 32. Feng M, Pan Y, Kong R, Shu S. Therapy of Primary Liver Cancer. Innov (Camb). 2020;1(2):100032. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Banales JM, Marin JJG, Lamarca A, Rodrigues PM, Khan SA, Roberts LR, et al. Cholangiocarcinoma 2020: the next horizon in mechanisms and management. Nat Rev Gastroenterol Hepatol. 2020;17(9):557–88. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials 12935_2026_4250_MOESM1_ESM.zip (13.2MB, zip) Supplementary Material 1: Movie AVI file 1. The AVI file for the HLF spheroids from the inject-embed method. Z-stack images were sequentially acquired with 33 μm spacing between adjacent frames. A total of 40 consecutive frames were captured and the corresponding AVI file were generated for visualization. Movie AVI file 2. The AVI file for the HLF spheroids from the mix-embed method. Z-stack images were sequentially acquired with 33 μm spacing between adjacent frames. A total of 40 consecutive frames were captured and the corresponding AVI file were generated for visualization. Supplementary Material 2 (8.2MB, docx) Data Availability Statement The RNA-seq data of this manuscript have been deposited in the NCBI Gene Expression Omnibus (GEO) database under accession number GSE298810. The authors also confirm that all other data supporting the findings of this study are available within the article and/or its supplementary information file. Articles from Cancer Cell International are provided here courtesy of BMC ACTIONS View on publisher site PDF (3.8 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top