> SPHEROIDS

Explore all the potential of these 3D cell culture assays 

> ABSTRACT

The usage of 3D cell culture is increasing in oncology, cell biology and tissue engineering. It provides the accessibility and flexibility of in vitro culture while providing a platform that more closely resembles the 3D structure of in vivo tissues. This review focuses on 3D aggregates of cells, also called spheroids or organoids. Here we discuss their application, their biology, the interest of using such a model, the different methods currently used to generate these and the areas ripe for improvement in this field. 

> INTRODUCTION 

A spheroid is a 3D aggregate of cells which forms when the intercellular interactions are favored over the attachment to a substrate. Such conditions are met when cells cannot create contacts with a fixed support, the plastic or glass surface of a culture plate for instance, and they start to self assemble in loose aggregates. It relies on the same process observed in embryogenesis, morphogenesis and organogenesis (Ryu et al. 2019). Integrins and E-cadherins have been shown to be of key importance in this process (Lee et al. 2012). The integrins are transmembrane proteins which can bind to extracellular matrix (ECM) components and initiate the agglomeration of the cells as a network. Then, homologous E-cadherin interactions tightly pack the cells together and make the structure to evolve into a compact spheroid.
If the cells used are able to generate ECM components, they can consolidate the structure while the cells autonomously arrange themselves due to contact interactions (Takezawa et al. 1993). This creates an environment adapted to the cells similar to in vivo conditions: the signaling pathways relying on ECM and cell-cell interaction are properly activated, the composition and density of the cell’s microenvironment can be autoregulated within the spheroid and cells of different origins can segregate with time to reproduce an in-vivo like organisation.


When compacted a spheroid can display up to three different layers which modulate the phenotype of the cells (Nath et al. 2016). The diffusion phenomena are limited inside a dense aggregate, the oxygen reaches its core only in small quantities, the biological waste remains in the innermost regions and only the outside layer of a spheroid receives enough nutrients to proliferate (Fig.1). It can lead to a three layer structure where the center is composed of necrotic and hypoxic cells, the core is surrounded by a layer of quiescent cells and the cells in the periphery are highly proliferative and produce metabolites. The size of these different domains depends on the cells used, if the structure is very compact or not and on the dimension of the spheroid. For most of the cell culture considered in this review, the presence of a necrotic core has been observed for spheroids with a diameter larger than a few hundreds micrometers (Kim et al. 2020, Qiu et al. 2015, Nath et al. 2016). Below such dimensions diffusion of nutrients and waste across the spheroid is sufficient to maintain the whole structure healthy.


Many cell types have been demonstrated to be able to form aggregates such as differentiated epithelial cells (Xia et al. 2012 (Hepatocytes), Takezawa et al. 1993 (Fibroblasts)), mesenchymal stem cells (MSCs) (Qiu et al. 2015 (Adipose derived MSCs), Wang et al. 2009 (Human MSCs)), primary stem cells (Kim et al. 2014 (Murine embryonic stem cells)) and cancer cell lines (Froehlich et al. 2016 (MCF-7, SK-BR-3, MDA-MB-231, breast cancer), Sirenko et al. 2015 (HCT116, colon cancer, DU145, prostate cancer, HepG2, liver cancer), Yamazaki et al. 2014 (HepG2, liver cancer)). A coculture of different cell types is also possible and can be of interest to create a more realistic model (Otsuka et al. 2013). For instance, Fibroblasts can be cultured alongside cancer cells to form a more robust structure (Lee, J.M. et al. 2018). Different cell types originating from the same tissue can form a spheroid, which better mimics the organisation and function of the organ studied. In this case, endothelial cells and the different cells forming the islet of Langherans might be used (Jun et al. 2019).

figure 1

Figure 1. Interior organisation of a spheroid (A) three layer organisation composed of necrotic core, a layer of quiescent cells and an outer proliferating enveloppe. (B) The oxygen concentration decreases inside the spheroid because of the limit of diffusion and its consumption by the cells. (C) Biological waste accumulates in the core of the spheroid and makes the core more acidic. (D) Several cell types can be cocultured in a spheroid and auto arrange themselves.

Moshksayan et al. 2018

> WHY STUDY SPHEROIDS?

Conventional 2D monolayer cell culture assays are known to be easy to use but often do not recreate physiologically relevant behaviours. However, they are extensively used for drug screening since they are affordable and automatisable. More precisely, cells growing on a flat and hard surface lack the tissue-specific architecture, the mechanical and biochemical cues and the cell-cell communication necessary for the maintenance of their phenotype (Pampaloni et al. 2007). What is more, it is the organisation of a tissue, its stiffness and the cross talk between its different cell types and functional units which assure its function. ECM signaling is also a key parameter, which determines if an organ grows normally or develops malignancies (Bissel et al. 2003).  For instance, the increased pressure applied by a tumor can activate oncogenes in surrounding cells (Fernandez-Sanchez et a. 2015), so a model with an inadapted stiffness can significantly modify the gene expression of the cells. The use of such assays may lead to imprecise observations and conclusions which are not representative of human biology. Regarding the drug screening pipeline, only 10% of the drugs entering clinical trials receive FDA approval, and most are discarded in phase 2 (Hay et al. 2014). This small success rate and finding 70% of the drugs to be toxic or uneffective in phase 2 may stem from the lack of relevance of the assays used in the preclinical step. Using more physiologically relevant assays such as 3D cell culture would improve the predictive power of preclinical trials and increase the chances of successful drug commercialization (Pampaloni et al. 2007). 

More specifically, many experiments involving spheroids or organoids showed promising results for a wide range of applications. The better maintenance of the phenotype in such assays enables to go further in the comprehension of a tissue in vitro. 

Stem cells grown in spheroids show better survival and maintain their stemness and their multipotency (Liu et al. 2021). Such features are assured in the body by the cellular niche granting the needed signals to the stem cells to maintain and proliferate. As the cell-cell and cell-ECM interaction are re-established within a spheroid, it is possible to include the components of the niche and then avoid the loss of stemness observed in 2D models. In addition, the stiffness and the geometry of a spheroid is adapted to the cells, so they keep a round nucleus and maintain a physiological gene expression, while the nucleus can be flattened in 2D assays (Cesarz et al. 2016). This inhibits the replicative senescence and maintains the differentiation potential of stem cells longer. 

It has also been demonstrated that somatic cells cultured as 3D aggregates maintain their in vivo phenotype and are able to secrete metabolites. Pancreatic cells extracted from islets of Langerhans cultured in a microfluidic chip form spheroids under medium perfusion and achieve insulin secretion (Jun et al. 2019). A liver organoid can metabolize a drug into a molecule effective on a connected tumor spheroid (Skardal et al. 2016). 

Furthermore, the three layer organisation of a spheroid mimics the in vivo structure of an avascular tumor (Nath et al. 2016). This similarity in behaviour makes spheroids great models to assess the effect of a molecule on a specific tumor. While 2D cell culture can indicate the effect a treatment has on isolated cells, 3D assays take into account the complexity of the cell arrangement, the diffusion phenomenon, the heterogeneity of cell phenotypes within a tumor and the specificities of the microenvironment created in its vicinity (Kucinska et al. 2021). Using spheroids in drug testing pipelines provides better insight in the effect of a molecule on a whole tumor.

More than a model, spheroids are also of interest for regenerative medicine and transplantation. In fact, MSC-derived spheroids proliferate, foster angiogenesis and display low levels of apoptosis, which make them interesting for autograft or allograft. Such biological materials can be derived from the patient’s own cells to decrease the immune reaction which usually rejects a tissue coming from a donor. What is more, the cells of the patient can be modified, by gene therapy for instance, before being reimplanted to replace diseased cells.  For these reasons spheroids are a promising solution to restore tissue function and are a great tool for personalized medicine and cell therapy.  However, the low scalability of spheroid generation techniques and the absence of vascularization in such structures still limit their use in clinical applications (Ong et al. 2018).

> HOW TO PRODUCE SPHEROIDS 

> Ultra low adherence plates

Seeding cells on a surface to which they cannot attach may be the most straightforward way to generate spheroids (Fig.2A). To obtain low-adherence surfaces, the surface of a petri dish, a well or a tissue culture plate is coated with a protein-repellent material such as agarose, poly-Hema, poly-ethylene glycol (PEG), galactose, polyvinyl alcohol (PVA) or poly(N-isopropylacrylamide) (PNIPAAm) (Liu et al. 2021). When seeded on such surfaces, anchor-dependent cells will aggregate and form spheroids. This method is low cost and easy to carry out, but generates a heterogeneous population of spheroids. Compared to the bioreactor and the hanging drop methods (see below), ultra low adherence plates induce aggregates that are less compact and more sensitive to chemical treatments (Raghavan et al. 2016). This simple method struggles to properly foster cell aggregation and intercellular interactions seem to be less strong than with other techniques. The coating can also be finely tuned to generate a semi-adhesive surface (Xia et al. 2012). By tuning the relative amount of a anti-adhesive (galactose) and an adhesive (RGD peptide) molecules on a surface, it is possible to switch from a fully adhesive surface (cell monolayer) to a semi-adhesive surface (tethered spheroids) or a non-adhesive surface (detached spheroids).

> Bioreactor

Cells cultured in a bioreactor are under constant agitation, which prevents stable attachment to a surface. In these conditions, cells tend to aggregate and compact into spheroids (Fig.2B). Maintained agitation during the aggregation phenomenon keeps the spheroid alive for a long period of time with active tissue-like functions and proper cellular organisation (Tostões et al. 2012). The viability and compaction of such aggregates are comparable with the one of spheroids obtained with the hanging drop methods and they are more resistant than those derived on low attachment plates (Raghavan et al. 2016). However, a study working on three breast cancer cell lines showed that a spheroid formation protocol has to be optimized for each different cell type, and the bioreactor method appeared to be less effective than the hanging drop or pellet culture approaches to generate cell aggregates (Froehlich et al. 2016). The bioreactor method is suited for long term cultures and can be scaled up to produce more aggregates, but it generates heterogeneous spheroids and prevents any individual following or real time monitoring. The spheroids have to be transferred into another culture environment to be studied. 

> Cell sheet engineering

Cell sheet engineering relies on the ability of cells to contract into a dense structure when they are not attached to a substrate (Fig.2C). A monolayer of cells is cultured on a material, the adhesive properties of which can be altered at will, by modifying the temperature for instance. Then the monolayer is detached by altering the properties of the substrate and it shrinks into a floating spheroid (Takezawa et al. 1993). Thermoresponsive hydrogels are most commonly used for this method such as PNIPAAm (Lee,Y.B. et al. 2018) or PDEGMA (Jiang et al. 2018). These hydrogels are hydrophobic above their lower critical solution temperature (LCST), allowing cell adhesion at typical cell culture temperatures, but become hydrophilic at lower temperatures, leading to their swelling and subsequent cell detachment. By patterning the thermoresponsive hydrogel with another cell repellent material, such as PEG, the size of the generated monolayers and thus the size of the spheroids can be tuned (Kim et al. 2020). This method allows control of the size of the spheroids and their harvesting time. The resulting aggregates show good production of ECM proteins. However, this method requires precise control over the temperature to avoid premature release of cells and appropriate design and manipulation to prevent fusion of adjacent monolayers together, which would generate bigger spheroids. 

> Pellet culture 

A pellet culture, or liquid overlay method, aims to aggregate cells by sedimentation in non-adherent cavities (Fig.2D). The cell culture can be done on a multiwell plate with U or V shaped wells coated with a cell repellent material (Froehlich et al. 2016, Sirenko et al. 2015). Scaling down the size of such wells allows to have a better control over the size of the spheroids and to produce more spheroids in the same assay. We then speak of microwells which are usually made out of a cell repellent material such as hydrogels (Qiu et al. 2015). These microwells can be a few hundreds of micrometers wide, thus many can be placed in a single well of a multiwell plate for instance. This leads to one spheroid in each microwell, all uniformly-sized, and their size can be tuned by controlling  the number of cells seeded.  As the limit of diffusion inside a spheroid is a few hundred micrometers, microwell dimensions are usually designed in the same size range to avoid large aggregates containing a necrotic core. The cells sediment inside the microwells, and as they cannot bond to a substrate they develop cell-cell and cell-ECM interactions, which eventually generate a compact spheroid within a few days (Lee, J.M. et al. 2018). Spheroids generated in hydrogel microwells are uniform in size and show good proliferation properties when replated on other ECM-based substrates. The phenotype maintenance of such a culture as be demonstrated for embryonic stem cells by targeting the expression of endothelial genes  (Kim et al. 2014). Increasing the area of collection of the cells to sediment into a well and preventing cells from staying in between the microwells increases the monodispersity of the generated spheroids (Cha et al. 2017). This method offers great reproducibility, scalability and is adapted to automatic liquid handlers, allowing automated drug screening assays (Sirenko et al. 2015, Liu et al. 2021). However, the uniformity and number of the spheroids can be reduced during medium exchanges, as fluid convection can cause these unattached cell aggregates to transfer into a nearby microwell and fuse with other aggregates, or to be lost during medium aspiration. To circumvent this issue it is possible to add a small adhesive area at the bottom of each well to anchor the spheroid (Mori et al. 2008).

> Hanging drop

The hanging drop assay relies on cell sedimentation to form aggregates, but rather than confining the cells in a well, they are confined within the bottom of a droplet (Fig.2E). These microliter droplets contain a set amount of cells allowing generation of spheroids of a given size. These hanging drops can be created by depositing liquid on the lid of a petri dish, which is then turned upside down (Froehlich et al. 2016), or they can be created in more sophisticated designs (Wu et al. 2016). This technique shows similar performances as pellet culture regarding the reproducibility of the spheroid generation and can work at a high throughput rate. However, due to the small volumes used, handling is challenging and spheroids can easily be lost. Furthermore, it is not straightforward to renew the medium in such an assay and evaporation of it can be an issue. 

> Microfluidics

As microfluidic technologies enable precise control over flows at the micron scale, it is possible to elaborate strategies to aggregate a given amount of cells inside of a chip (Moshksayan et al. 2018). Droplet microfluidics (Fig.2F) entrap cells in aqueous drops of a few microliters separated by a hydrophobic medium (such as oil) (Wang et al. 2015). Without an attachment substrate, the cells evolve into a spheroid and reproducibility in cell number is easily achieved using current technologies. It is also possible to entrap different cues in each droplet and introduce a barcode system to run many independent experiments in parallel. Another strategy consists in designing traps in the path of the microchannels, which accumulate cells in specific locations and force them to aggregate (Liu et al. 2015). Traps can also be used to capture already formed spheroids and facilitate their imaging and analysis (Ruppen et al. 2014). To better mimic in vivo conditions several spheroids of same (Jun et al. 2019) or different (Skardal, Devarasetty et al. 2016) nature can be cultured under perfusion in connected traps or microfluidic chambers. This widens the range of applications for spheroids using them as constitutive pieces to create a more complex model. Spheroid-on-chip can give insights into the interactions between different organs to create an advanced body-on-chip model (Skardal, Shupe & Atala. 2016). 

figure.2v3

Figure 2. Main approaches used to generate spheroids (A) Low adherence plate. A substrate, in black, is coated with a cell repellent material, in pink, so that the cells cannot grow a monolayer and self assemble into 3D aggregates, shown in green. (B) Bioreactor. The medium of culture, in blue, is maintained under agitation, preventing cell adhesion to the surface. (C) Cell sheet engineering. Cells grow monolayer on an adhesive substrate, in blue, which can be delimited by a non adherent region, in pink. The adhesive material can be turned into a cell repellent one, by decreasing the temperature for instance, releasing the cell monolayer which then contracts into a spheroid. (D) Pellet culture. Cells sediment into a non adhesive well or microwell and aggregate into a spheroid. (E) Hanging drop. Cells sediment inside a drop hanging from a substrate and aggregate into a spheroid. (F) Droplet microfluidics. Droplets of medium are dispersed in an hydrophobic phase, in yellow. The cells contained in the droplets then form a spheroid.

> Summary of main technologies

> DISCUSSION

The main challenges of the current methods used to generate spheroids are the control of size and scalability for high throughput assays. Techniques such as low adherence plates or bioreactors usually produce a wide variability of spheroid sizes within the same batch. Uniformity in spheroid size within a population is of paramount importance to carry out reproducible experiments and obtain trustworthy observations. Being able to precisely tune the size of the generated cell aggregates and assure their monodispersity would be the major features of an ideal process.  Furthermore, drug screening platforms require a large amount of tests to obtain significant results but most current techniques struggle to provide such an amount of spheroids. To enable the use of 3D cell aggregate models at large scales it is necessary to develop high throughput generation techniques. 

Another limitation of the use of spheroids as a standard model is the difficulties of handling cell aggregates (often in suspension) compared to a classical 2D cell culture. Some techniques use very small volumes to form the aggregates, leading to a high probability of losing and fusing spheroids when exchanging media or transferring them to another environment. Removing these issues would enhance the use of this kind of cell culture by facilitating both its manual and automatic handling.  

Acquiring data from  3D structures is also non-trivial and conventional optical imaging techniques are usually limited. More advanced techniques such as fluorescent microscopy and confocal microscopy are needed to fully grasp the organization of a spheroid. However,  the cost of these equipments and the time needed to perform the acquisition can prevent some researchers from using these models. In addition, spheroids are highly scattering structures, making optical observation challenging, even with sophisticated approaches (Pampaloni et al. 2007). New methods to finely analyse such 3D structures  would be a plus to foster their use. 

> CONCLUSION

Spheroids can be an extremely useful biological model which recapitulates many essential physiological features. This 3D model bridges the gap between the 2D cell culture and in vivo assays since it includes cell-cell and cell-ECM interactions, depicts tissue-like stiffness and cell organisation and exposes the cells to appropriate mechanical and biochemical cues. Such cell cultures maintain in-vivo phenotype and show tissue specific functions that cannot be achieved with 2D cultures. The use of 3D cell aggregates significantly improves the predictive power of drug screening assays, the observations made in vitro are more relevant than with conventional assays, the understanding of some diseases can go further and together  this paves the way for personalized medicine. 

> REFERENCES

Bissell, M. J., Rizki, A., & Mian, I. S. (2003). Tissue architecture: the ultimate regulator of breast epithelial function. Current opinion in cell biology, 15(6), 753. 

Cesarz, Z., & Tamama, K. (2016). Spheroid culture of mesenchymal stem cells. Stem cells international, 2016.

Cha, J. M., Park, H., Shin, E. K., Sung, J. H., Kim, O., Jung, W., … & Kim, J. (2017). A novel cylindrical microwell featuring inverted-pyramidal opening for efficient cell spheroid formation without cell loss. Biofabrication, 9(3), 035006. 

Fernandez-Sanchez, M. E., Barbier, S., Whitehead, J., Béalle, G., Michel, A., Latorre-Ossa, H., … & Farge, E. (2015). Mechanical induction of the tumorigenic β-catenin pathway by tumour growth pressure. Nature, 523(7558), 92-95.

Froehlich, K., Haeger, J. D., Heger, J., Pastuschek, J., Photini, S. M., Yan, Y., … & Schmidt, A. (2016). Generation of multicellular breast cancer tumor spheroids: comparison of different protocols. Journal of mammary gland biology and neoplasia, 21(3), 89-98. 

Hay, M., Thomas, D. W., Craighead, J. L., Economides, C., & Rosenthal, J. (2014). Clinical development success rates for investigational drugs. Nature biotechnology, 32(1), 40-51.

Jiang, S., Lyu, B., Müller, M., Wesner, D., & Schönherr, H. (2018). Thickness-encoded micropatterns in one-component thermoresponsive polymer brushes for culture and triggered release of pancreatic tumor cell monolayers and spheroids. Langmuir, 34(48), 14670-14677. 

Jun, Y., Lee, J., Choi, S., Yang, J. H., Sander, M., Chung, S., & Lee, S. H. (2019). In vivo–mimicking microfluidic perfusion culture of pancreatic islet spheroids. Science advances, 5(11), eaax4520. 

Kim, G., Jung, Y., Cho, K., Lee, H. J., & Koh, W. G. (2020). Thermoresponsive poly (N-isopropylacrylamide) hydrogel substrates micropatterned with poly (ethylene glycol) hydrogel for adipose mesenchymal stem cell spheroid formation and retrieval. Materials Science and Engineering: C, 115, 111128. 

Kim, J. E., Lee, J. M., & Chung, B. G. (2014). Microwell arrays for uniform-sized embryoid body-mediated endothelial cell differentiation. Biomedical microdevices, 16(4), 559-566. 

Kucinska, M., Plewinski, A., Szczolko, W., Kaczmarek, M., Goslinski, T., & Murias, M. (2021). Modeling the photodynamic effect in 2D versus 3D cell culture under normoxic and hypoxic conditions. Free Radical Biology and Medicine, 162, 309-326.

Lee, E. J., Park, S. J., Kang, S. K., Kim, G. H., Kang, H. J., Lee, S. W., … & Kim, H. S. (2012). Spherical bullet formation via E-cadherin promotes therapeutic potency of mesenchymal stem cells derived from human umbilical cord blood for myocardial infarction. Molecular Therapy, 20(7), 1424-1433.

Lee, Y. B., Kim, E. M., Byun, H., Chang, H. K., Jeong, K., Aman, Z. M., … & Shin, H. (2018). Engineering spheroids potentiating cell-cell and cell-ECM interactions by self-assembly of stem cell microlayer. Biomaterials, 165, 105-120. 

Lee, J. M., Yang, L., Kim, E. J., Ahrberg, C. D., Lee, K. B., & Chung, B. G. (2018). Generation of uniform-sized multicellular tumor spheroids using hydrogel microwells for advanced drug screening. Scientific reports, 8(1), 1-10. 

Liu, D., Chen, S., & Win Naing, M. (2021). A review of manufacturing capabilities of cell spheroid generation technologies and future development. Biotechnology and Bioengineering, 118(2), 542-554. 

Liu, W., Wang, J. C., & Wang, J. (2015). Controllable organization and high throughput production of recoverable 3D tumors using pneumatic microfluidics. Lab on a Chip, 15(4), 1195-1204.

Mori, R., Sakai, Y., & Nakazawa, K. (2008). Micropatterned organoid culture of rat hepatocytes and HepG2 cells. Journal of bioscience and bioengineering, 106(3), 237-242.

Moshksayan, K., Kashaninejad, N., Warkiani, M. E., Lock, J. G., Moghadas, H., Firoozabadi, B., … & Nguyen, N. T. (2018). Spheroids-on-a-chip: Recent advances and design considerations in microfluidic platforms for spheroid formation and culture. Sensors and Actuators B: Chemical, 263, 151-176. 

Nath, S., & Devi, G. R. (2016). Three-dimensional culture systems in cancer research: Focus on tumor spheroid model. Pharmacology & therapeutics, 163, 94-108. 

Ong, C. S., Zhou, X., Han, J., Huang, C. Y., Nashed, A., Khatri, S., … & Hibino, N. (2018). In vivo therapeutic applications of cell spheroids. Biotechnology advances, 36(2), 494-505. 

Otsuka, H., Sasaki, K., Okimura, S., Nagamura, M., & Nakasone, Y. (2013). Micropatterned co-culture of hepatocyte spheroids layered on non-parenchymal cells to understand heterotypic cellular interactions. Science and technology of advanced materials. 

Pampaloni, F., Reynaud, E. G., & Stelzer, E. H. (2007). The third dimension bridges the gap between cell culture and live tissue. Nature reviews Molecular cell biology, 8(10), 839-845. 

Qiu, X., Zhang, Y., Zhao, X., Zhang, S., Wu, J., Guo, H., & Hu, Y. (2015). Enhancement of endothelial differentiation of adipose derived mesenchymal stem cells by a three-dimensional culture system of microwell. Biomaterials, 53, 600-608.

Raghavan, S., Mehta, P., Horst, E. N., Ward, M. R., Rowley, K. R., & Mehta, G. (2016). Comparative analysis of tumor spheroid generation techniques for differential in vitro drug toxicity. Oncotarget, 7(13), 16948. 

Ruppen, J., Cortes-Dericks, L., Marconi, E., Karoubi, G., Schmid, R. A., Peng, R., … & Guenat, O. T. (2014). A microfluidic platform for chemoresistive testing of multicellular pleural cancer spheroids. Lab on a Chip, 14(6), 1198-1205.

Ryu, N. E., Lee, S. H., & Park, H. (2019). Spheroid culture system methods and applications for mesenchymal stem cells. Cells, 8(12), 1620. 

Sirenko, O., Mitlo, T., Hesley, J., Luke, S., Owens, W., & Cromwell, E. F. (2015). High-content assays for characterizing the viability and morphology of 3D cancer spheroid cultures. Assay and drug development technologies, 13(7), 402-414. 

Skardal, A., Shupe, T., & Atala, A. (2016). Organoid-on-a-chip and body-on-a-chip systems for drug screening and disease modeling. Drug discovery today, 21(9), 1399-1411.

Skardal, A., Devarasetty, M., Forsythe, S., Atala, A., & Soker, S. (2016). A reductionist metastasis‐on‐a‐chip platform for in vitro tumor progression modeling and drug screening. Biotechnology and bioengineering, 113(9), 2020-2032. 

Takezawa, T., Mori, Y., Yonaha, T., & Yoshizato, K. (1993). Characterization of morphology and cellular metabolism during the spheroid formation by fibroblasts. Experimental cell research, 208(2), 430-441. 

Tostões, R. M., Leite, S. B., Serra, M., Jensen, J., Björquist, P., Carrondo, M. J., … & Alves, P. M. (2012). Human liver cell spheroids in extended perfusion bioreactor culture for repeated‐dose drug testing. Hepatology, 55(4), 1227-1236.

Wang, W., Itaka, K., Ohba, S., Nishiyama, N., Chung, U. I., Yamasaki, Y., & Kataoka, K. (2009). 3D spheroid culture system on micropatterned substrates for improved differentiation efficiency of multipotent mesenchymal stem cells. Biomaterials, 30(14), 2705-2715. 

Wang, Y., Zhao, L., Tian, C., Ma, C., & Wang, J. (2015). Geometrically controlled preparation of various cell aggregates by droplet-based microfluidics. Analytical Methods, 7(23), 10040-10051.

Wu, H. W., Hsiao, Y. H., Chen, C. C., Yet, S. F., & Hsu, C. H. (2016). A PDMS-based microfluidic hanging drop chip for embryoid body formation. Molecules, 21(7), 882.

Xia, L., Sakban, R. B., Qu, Y., Hong, X., Zhang, W., Nugraha, B., … & Yu, H. (2012). Tethered spheroids as an in vitro hepatocyte model for drug safety screening. Biomaterials, 33(7), 2165-2176. 

Yamazaki, H., Gotou, S., Ito, K., Kohashi, S., Goto, Y., Yoshiura, Y., … & Nakazawa, K. (2014). Micropatterned culture of HepG2 spheroids using microwell chip with honeycomb-patterned polymer film. Journal of bioscience and bioengineering, 118(4), 455-460.