Three-Dimensional Spheroids as In Vitro Preclinical Models for Cancer Research

Most cancer biologists still rely on conventional two-dimensional (2D) monolayer culture techniques to test in vitro anti-tumor drugs prior to in vivo testing. However, the vast majority of promising preclinical drugs have no or weak efficacy in real patients with tumors, thereby delaying the discovery of successful therapeutics. This is because 2D culture lacks cell–cell contacts and natural tumor microenvironment, important in tumor signaling and drug response, thereby resulting in a reduced malignant phenotype compared to the real tumor. In this sense, three-dimensional (3D) cultures of cancer cells that better recapitulate in vivo cell environments emerged as scientifically accurate and low cost cancer models for preclinical screening and testing of new drug candidates before moving to expensive and time-consuming animal models. Here, we provide a comprehensive overview of 3D tumor systems and highlight the strategies for spheroid construction and evaluation tools of targeted therapies, focusing on their applicability in cancer research. Examples of the applicability of 3D culture for the evaluation of the therapeutic efficacy of nanomedicines are discussed.


Introduction
Significant investments are made in cancer research for drug discovery and development. Yet, the approval rate (≤5%) of drugs that reach the clinic remains very low [1,2]. Typically, anticancer compounds are tested in two dimensional (2D) cell culture models, that involve a panel of cancer cell lines, such as those used by the US National Cancer Institute [3]. Drugs that show promising cytotoxicity in 2D in vitro system progress to animal models of human cancers (mainly mice) for anti-tumor efficacy testing [4]. Unfortunately, most of the promising preclinical drugs have no or weak efficacy in real patients with tumors, resulting in a significant delay of anticancer drug development [5]. One of the main factors underlying this poor success is the inadequacy of the preclinical 2D cultures and animal models to recapitulate the human tumor microenvironment (TME). TME is a complex and heterogeneous structure made of cellular (e.g., transformed epithelial cells, fibroblasts, infiltrating lymphocytes, mesenchymal stem cells, endothelial cells) and non-cellular (e.g., extracellular matrix-ECM, growth factors, cytokines and chemokines) components, with a critical role in cancer development and progression [6,7]. The 2D culture systems lack the structural architecture and the microenvironment of the tumor, and display altered gene expression and activation of cell signaling pathways, compared to the in vivo tumor tissues (Table 1) [8][9][10]. Besides the associated higher cost and ethical issues, animal models also display significant limitations and poorly reflect the  The ECM provides structural support for cells in the extracellular space, and is composed of structural fibrous proteins (e.g., collagens and elastin), multiadhesive proteins (e.g., fibronectin and laminin), glycosaminoglycans (e.g., heparan sulfate, hyaluronan), proteoglycans (e.g., perlecan, syndecan), and sequestered growth factors, as well as secreted proteins [23][24][25]. The cross-talk between the different TME cells and components plays an essential role in tumor growth, progression and metastasis [26].
Many factors present in the TME, including transforming growth factor beta (TGF-β), cytokines (IL-10 and IL-1β), members of the VEGF, plateled-derived growth factor (PDGF), FGF, angiopoietin families, Bv8/PROK2, and hypoxia-inducible factor (HIF)-1α, provide molecular support to tumor growth and progression [27][28][29][30]. Additionally, cancer cells are experts in modifying their surrounding environments. For instance, cancer cells can co-opt fibroblasts to obtain growth factors, such as basic fibroblast growth factor (bFGF), necessary to sustain their growth and proliferation. additionally, tumor cells can interact with the surrounding endothelial cells, promoting the release of soluble factors, like vascular endothelial growth factor (VEGF), to trigger the angiogenic process. Tumor cells can also evade the immune-mediated cellular destruction through different strategies. For instance, loss of tumor antigen expression precludes their recognition by the immune system, production of immunosuppressive cytokines protects them from the cytotoxic lysis by immune cells, and development of immunosuppressive forces leads to local immunosuppression in the TME that shifts the phenotype and function of normal immune cells from an anti-tumor state to a pro-tumor state [31][32][33][34][35][36].
Currently, treatment options against cancer include surgery, chemotherapy, radiation therapy, hormonal therapy, and targeted therapy [37]. Basically, anticancer therapies aim to target tumor cells either directly, through DNA damage by cytotoxic drugs or local radiation causing apoptosis, or indirectly, through the destruction of TME so as to deprive cancer cells of the machinery that fuel The ECM provides structural support for cells in the extracellular space, and is composed of structural fibrous proteins (e.g., collagens and elastin), multiadhesive proteins (e.g., fibronectin and laminin), glycosaminoglycans (e.g., heparan sulfate, hyaluronan), proteoglycans (e.g., perlecan, syndecan), and sequestered growth factors, as well as secreted proteins [23][24][25]. The cross-talk between the different TME cells and components plays an essential role in tumor growth, progression and metastasis [26].
Many factors present in the TME, including transforming growth factor beta (TGF-β), cytokines (IL-10 and IL-1β), members of the VEGF, plateled-derived growth factor (PDGF), FGF, angiopoietin families, Bv8/PROK2, and hypoxia-inducible factor (HIF)-1α, provide molecular support to tumor growth and progression [27][28][29][30]. Additionally, cancer cells are experts in modifying their surrounding environments. For instance, cancer cells can co-opt fibroblasts to obtain growth factors, such as basic fibroblast growth factor (bFGF), necessary to sustain their growth and proliferation. additionally, tumor cells can interact with the surrounding endothelial cells, promoting the release of soluble factors, like vascular endothelial growth factor (VEGF), to trigger the angiogenic process. Tumor cells can also evade the immune-mediated cellular destruction through different strategies. For instance, loss of tumor antigen expression precludes their recognition by the immune system, production of immunosuppressive cytokines protects them from the cytotoxic lysis by immune cells, and development of immunosuppressive forces leads to local immunosuppression in the TME that shifts the phenotype and function of normal immune cells from an anti-tumor state to a pro-tumor state [31][32][33][34][35][36].
Currently, treatment options against cancer include surgery, chemotherapy, radiation therapy, hormonal therapy, and targeted therapy [37]. Basically, anticancer therapies aim to target tumor cells either directly, through DNA damage by cytotoxic drugs or local radiation causing apoptosis, or indirectly, through the destruction of TME so as to deprive cancer cells of the machinery that fuel their growth and progression. However, these therapies induce new biological tumor responses, mainly through immunological and angiogenic modulation, contributing to drug resistance, which remains a serious consequence of most anticancer treatments, impacting the patient's prognosis and quality of life [31,38].
Various cancer cells can spontaneously assemble into spheroids in culture environment that privileges cell-cell and cell-ECM interactions over cell-substrate interactions [14]. These predominant cell-cell and cell-ECM interactions result in the formation of a 3D structure that closely reproduces mimics the native spatial organization and environment of avascular tumors, where cells can proliferate, aggregate and differentiate ( Figure 2) [56]. Common methods for spheroid generation are described in the next section. Spheroids have a diameter of 200 micrometers or more, generally with a spherical shape, and display three concentric zones of heterogeneous cell populations: an external zone of highly proliferating and migrating cells; a middle zone of quiescent cells, and an internal zone of necrotic cells [57,58]. These cell layers are so defined due to the nutrients and oxygen gradients that are established, as a result of limited diffusion, from the outside to the center of the spheroids. Thus, cells of the peripheral layer of the spheroids are exposed to sufficient oxygen and Pharmaceutics 2020, 12, 1186 5 of 38 growth factors from the medium, which stimulate their proliferation. At the middle layer, limited diffusion of growth factors forces cell entry into quiescent state of the cell cycle. In large spheroids (>500 micrometers), oxygen deficiency (hypoxia) in the innermost zone induces altered gene expression, through stabilization of the transcription factor HIF-1α, and, consequently, triggers the Warburg effect, promoting aerobic glycolysis and lactic acid production, thereby lowering pH of the inner layer of spheroids [59]. Nutrient and oxygen deprivation, together with the accumulation of metabolic waste, triggers the necrotic death of cells at the innermost layer.
Pharmaceutics 2020, 12, x 5 of 40 internal zone of necrotic cells [57,58]. These cell layers are so defined due to the nutrients and oxygen gradients that are established, as a result of limited diffusion, from the outside to the center of the spheroids. Thus, cells of the peripheral layer of the spheroids are exposed to sufficient oxygen and growth factors from the medium, which stimulate their proliferation. At the middle layer, limited diffusion of growth factors forces cell entry into quiescent state of the cell cycle. In large spheroids (>500 micrometers), oxygen deficiency (hypoxia) in the innermost zone induces altered gene expression, through stabilization of the transcription factor HIF-1α, and, consequently, triggers the Warburg effect, promoting aerobic glycolysis and lactic acid production, thereby lowering pH of the inner layer of spheroids [59]. Nutrient and oxygen deprivation, together with the accumulation of metabolic waste, triggers the necrotic death of cells at the innermost layer. Figure 2. Typical structure of a multicellular tumor spheroid. The geometric rearrangement of the cells within the spheroid forms three concentric zones of heterogeneous cell populations: an external proliferating zone (Proliferative zone); a middle zone of quiescent cells (Quiescent zone), and an internal zone of necrotic cells (Necrotic zone). These cell layers are caused by the gradients of nutrients, oxygen, and pH (yellow), from the outside to the center of the spheroid, and by the gradients of CO2, waste, and lactate, from the center to the outside. Created with BioRender.com.
Therefore, 3D culture systems recapitulate many characteristics of in vivo tumors, such as cellcell and cell-ECM interactions, nutrient and oxygen gradients, and distinct layers of cell populations. Besides, the morphology and polarity of the cells, as well as gene expression and activation of cell signaling pathways, are also close to those of real tumors [8,9,60,61]. These features make spheroids a promising model for the study of cancer biology, cancer initiation, invasion and metastatic processes, as well as drug testing.

Methods for Spheroid Generation
Cells grown in culture environment of low binding or absence of adhesive surface can assemble into 3D spheroids, as these conditions favor cell-cell and cell-ECM interactions over cell-substrate Figure 2. Typical structure of a multicellular tumor spheroid. The geometric rearrangement of the cells within the spheroid forms three concentric zones of heterogeneous cell populations: an external proliferating zone (Proliferative zone); a middle zone of quiescent cells (Quiescent zone), and an internal zone of necrotic cells (Necrotic zone). These cell layers are caused by the gradients of nutrients, oxygen, and pH (yellow), from the outside to the center of the spheroid, and by the gradients of CO 2 , waste, and lactate, from the center to the outside. Created with BioRender.com. Therefore, 3D culture systems recapitulate many characteristics of in vivo tumors, such as cell-cell and cell-ECM interactions, nutrient and oxygen gradients, and distinct layers of cell populations. Besides, the morphology and polarity of the cells, as well as gene expression and activation of cell signaling pathways, are also close to those of real tumors [8,9,60,61]. These features make spheroids a promising model for the study of cancer biology, cancer initiation, invasion and metastatic processes, as well as drug testing.

Methods for Spheroid Generation
Cells grown in culture environment of low binding or absence of adhesive surface can assemble into 3D spheroids, as these conditions favor cell-cell and cell-ECM interactions over cell-substrate interactions. Spheroids can be obtained after 1 to 7 days of culture, with various morphologies, depending on the cell line and the approach used. Examples of studies that performed spheroid  Table 2, providing information on cell lines, density and the time required to obtain the spheroids. According to the literature, spheroids of 300-500 µm of size are those that best mimic in vivo tumors in terms of hypoxia and proliferation gradients. Typically, spheroids are constructed from tumor cells, using scaffold-free or scaffold-based techniques [62].

Scaffold-Free Techniques
In scaffold-free techniques, different factors (e.g., low-adhesion substrates, gravity force and magnetic action) contribute to cellular aggregate formation and spheroid generation. During this process, the ECM is originated through continuous deposit of proteins produced by spheroid cells [63]. The most common scaffold-free techniques currently used are ultra-low attachment plates, hanging drop, magnetic levitation and magnetic 3D printing. The advantages and disadvantages of each method are summarized in Table 3.

Ultra-Low Attachment Plates
The surfaces of the plates are coated with a substrate to prevent cell adhesion, promoting cell aggregation and spheroid formation. Besides presenting low adhesion, the wells of these plates have a defined shape (round bottom, V-shaped or conical), allowing the positioning of a single spheroid [64,65]. Generally, the main substrates used to coat the plate are agar/agarose or poly(2-hydroxyethyl methacrylate), by adding 50 µL of solution in each well of 96-well plates, at concentrations of 15 mg/mL and 5 mg/mL, respectively [66][67][68]. With this technique, a large number of spheroids can be generated simultaneously on the same plate, facilitating experimental reproducibility, in addition to enable the monitoring of spheroid formation and growth. As disadvantages, some tumor cell lines do not form tight spheroids in ultra-low attachment plates [69].

Hanging Drop
In this method, approximately 25 µL of cell suspension is positioned inside of a petri dish lid, which contains phosphate-buffered saline (PBS) to avoid dehydration of the cellular solution. Then, the lid is inverted and due to surface tension, the droplets remain suspended. Owing to the force of gravity, the cells within the droplets spontaneously form cellular aggregates, giving rise to a single spheroid [70,71]. This method allows large production of spheroids and an easy control of their size. However, it is labor intensive due to its multistep process, and there is a risk of cell damage in case of media evaporation, requiring constant monitoring of the culture medium [71][72][73][74]. Moreover, the hanging drop method can originate spheroids with heterogeneous sizes and morphologies, impacting the spheroid standardization, which is essential for new drug screening. Recently, studies have developed different tools to minimize these limitations and facilitated the realization of this method [75][76][77]. For instance, the pressure-assisted network for droplet accumulation (PANDA) system consists of a pressure chip capable to create homogeneous and compact hanging drop array, enabling the fast and economical production of spheroids [75]. Another way to circumvent these barriers is through the use of 3D printed hanging-drop dripper array that allows in situ analysis of drug screening, tumor metastasis and tumor transendothelial migration, besides promoting heterotypic spheroid interaction [77].

Magnetic Levitation and Magnetic 3D Printing
Through a mixture of magnetic particles/nanoparticles, the cells are magnetized and incubated under magnetic forces to overcome the gravitational force, allowing their levitation and, consequently, formation of cellular aggregate [78]. In this method, after the cells absorb the magnetic particles, a magnet is positioned above (magnetic levitation) or below (magnetic 3D printing) the plate, promoting cells aggregation and spheroid generation [70]. Spheroids are usually formed in less than 16 h, being considered a fast-acting technique [72]. However, prior preparation of magnetic nanoparticles is necessary, and limited number of spheroids are generated [78].

Scaffold-Based Techniques
In scaffold-based techniques, external cell anchoring systems are used to mimic the ECM structure, allowing greater cell-cell and cell-matrix interaction. These systems can consist of porous microcarriers of natural, synthetic, and semisynthetic hydrogels made of cross-linked polymers. Porous scaffolds are widely used in the bioengineering sector, and have gained prominence in 3D cell culture implementation [63,72,79]. Due to their interconnected pores, these structures allow greater diffusion of nutrients, oxygen and metabolic debris, in addition to mimicking the ECM architecture, providing cellular support, attachment and proliferation [80][81][82]. 3D porous scaffolds also promote the formation of bigger spheroids, compared to non-porous scaffolds, and enhance tumor cell invasion and therapeutic resistance [83,84]. Although they are synthesized mainly by polymers such as poly(ε-caprolactone), porous microparticles can consist of different substances, including natural (e.g., chitosan, hyaluronic acid, alginate, collagen, gelatin, silk fibroin), and synthetic (e.g., poly(lactide-co-glycolide)) materials [85][86][87][88][89][90]. The porosity and pore size of the scaffold are essential for the establishment of effective 3D models, as they can affect the transport of oxygen, metabolites and nutrients, as well as cell adhesion and cell growth [91]. While porous 3D scaffold methods are useful to control the spheroid size, effective collecting and separation of spheroids from 3D scaffolds may be difficult [92]. Common scaffold-based methodologies include spinner flasks, micropatterned plates, matrix encapsulation, matrix on top, matrix embedded, microcarriers beads, and microfluidic devices.

Spinner Flasks
Continuous rotating agitation inhibits cell adhesion to the surface, leading to spheroid formation. The main means of rotation used are through spinner flasks and rotating flasks. In spinner flasks, a magnetic stirrer is positioned inside the flask, allowing homogeneous distribution of oxygen and nutrients. However, the cells are subjected to direct shearing force, which increases the risk of their damage. In the rotating flasks, the flask itself is rotated, allowing the dispersion of oxygen and nutrients, and the reduction of the shear forces on the cells [70]. As an advantage, this method allows large scale generation of spheroids. However, the continuous rotation prevents the visualization of the aggregates, can damage the cells, and is hard to monitor [71]. Yet, this method is considered as one of the most efficient systems for obtaining large amounts of spheroids under controlled nutritional conditions [72].

Micropatterned Plates
The plates are modified to create micrometer sized compartments with a low adhesion surface within each microspace, providing a micropattern or microwells which induce cells to grow as clusters. First, a layer of 3-trimethoxysilyl polymethacrylate is added to the glass plate, to ensure fixation of the hydrogel microwells to the plate, followed by a uniform layer of hydrogel. Soon after, using photolithography techniques, polydimethylsiloxane is added to the hydrogel for microwell formation [70]. The cell suspension is then seeded into hydrogel microwells, which can vary in size from 150-600 µm [93]. This method allows large scale production of spheroids. However, bubbles often form during the culture, and pipetting can damage micropatterned surfaces due to pipetting [64].

Matrix Encapsulation
Suspended cells are surrounded by hydrogel and placed in calcium free solution, forming cellular microcapsules. In these microcapsules, cells aggregate to form matrix encapsulated spheroids [70]. Generally, microcapsules have a size between 100 and 500 µm, are capable of generating monotypic or heterotypic spheroids, and allow cell-cell and cell-ECM interaction [94]. In these systems, the transport of nutrients and metabolic residues occurs by simple diffusion and, as the microcapsule increases, the nutrient transport becomes limited, which can cause cellular necrosis. Due to their viscoelastic capabilities, alginate hydrogels has been widely used to generate microcapsules [95]. An important advantage is that this method yields homogeneous sized spheroids.

Matrix on Top and Matrix Embedded
The matrix on top and matrix embedded methods are quite similar. In the matrix-on-top method, the cells are seeded and trapped on the top of the solid matrix, and spheroids are formed through cellular aggregation. In the matrix embedded method, cells suspended in the liquefied matrix are only incorporated into the matrix after the gelation process [70]. Several compounds have been used as a matrix, including agarose, matrigel, collagen, and synthetic polymers [96]. Matrix-on-top method facilitates post-culture processing and imaging of the generated spheroids.

Microcarrier Beads
This system has been used for more than 25 years for to generate 3D cell culture [97]. In this method, cells adhere to natural (e.g., collagen, cellulose) or synthetic (e.g., dextran, poly(d,l-lactide-co-glycolide)) matrix-coated beads, forming spheroidal structures [98,99]. The microcarrier beads provide a cell attachment surface, allowing the aggregation, especially of cells unable to aggregate spontaneously. This method is considered a fast, easy, and reproducible spheroid generation system, and allows the adhesion of different cell types to form heterogeneous spheroids. However, the presence of microcarrier beads in spheroids does not mimic the tumor physiological conditions in vivo [100].

Microfluidic Devices
The cells are placed in microchannels with a free perfusion system, allowing the distribution of oxygen and nutrients, and the elimination of metabolic waste. As an advantage, this system can mimic tumor microvasculature in vivo. However, this method requires specialized laboratories and equipment [101][102][103]. Due to its ability to guarantee gases permeability, polydimethylsiloxane (PDMS) is the most used material for making microfluidic devices [104]. In addition, PDMS are biocompatible, easy to make, and are low cost. However, under high pressure, PDMS microchannels can be deformed, causing changes in fluid speeds. Depending on the type of sealing, reversible or irreversible, the PDMS microfluidic devices can withstand pressure up to 0.3 or 2 bar, respectively. Moreover, when exposed to some fluids, PDMS microfluidic can swell, which impacts in device function [105][106][107]. Other microfluidic device polymers, such as thermoset polyester, polyurethane methacrylate and Norland Adhesive 81, also undergo structural changes when exposed to pressures above 10, 8 and 5 bar, respectively [108].  [115] • Human pancreatic β-cell line (EndoCbH3) and human umbilical vein endothelial cells (HUVECs).
• 5000 cells/50 mL in corresponding cell culture media per well; • Cell ratio: 5000 cells to 5000 HUVECs.
• The exact beginning of spheroid formation was not described; spheroid formation was observed from day 5; • β-cells and HUVECs were previously treated with NanoShuttle™-PL at a concentration of 40 µL/mL in media culture. [117] • Mesenchymal stem cells (MSCs).

•
The exact beginning of spheroid formation was not described; spheroid formation was observed from day 3; • Cell were seeded in single inlet multi-size spheroid synthesis (SIMSS) chips. [128] • Human colorectal cancer cell line: HT-29, and human normal fibroblast cell line: CCD-18Co.

Tools to Evaluate Targeting Effect
Several techniques are available to characterize spheroids, either in viable, fixed, or dissociated state, before and after anticancer drug treatment. These techniques were described in details in a number of excellent review papers [63,140,141], and allow the characterization of the organization, size, shape, gene and protein expression, metabolic status, migration and invasion of anticancer drug-treated spheroids. In general, standard biological assays used for 2D culture can be applied to spheroids, with some drawbacks as outlined below (Table 4).

Optical Microscopy
Morphologic changes such as size and shape can be monitored over time by optical microscopy and analyzed by appropriate software [142,143]. For instance, with a standard phase-contrast microscope, the difference in the size or volume between treated and untreated spheroids at a defined endpoint, or even during treatment, can be used to evaluate the efficacy of an anticancer drug.
Fluorescence microscopy can provide information on ECM deposition in spheroids immunostained with antibodies against fibronectin, laminin, and collagen IV [144], while relevant information such as cytoskeletal arrangement, proliferation, and apoptosis in the spheroids can be obtained by Hoechst or DAPI, phalloidin, Ki-67, caspases, Annexin V, Propidium iodide, and TUNEL staining [63,145]. Confocal laser microscopy is required to obtain higher spatial resolution, needed to analyze spheroid architecture. However, this analysis is restricted to small spheroids due to limited light penetration and to light scattering in thick tissues [143].
To overcome these issues, spheroids can be processed for histological sectioning. Then, staining methods such as hematoxylin and eosin staining allow distinction of pyknotic nuclei and eosinophilic cytoplasm in spheroid sections. For proliferating and quiescent cell populations, the use of specific antibodies in immunohistological staining is required. However, spheroid fixation used in the histological procedure precludes the study of dynamic alterations in the spheroids over time. Additionally, sample fracture and morphology deformation can occur during spheroid sectioning. Due to the delicate nature and small size of spheroids, the fixation time may need to be reduced, comparatively to biopsies or organ fragments. Further spheroid processing has also presented some challenges. For example, the inclusion of several spheroids in a unique paraffin block may involve a more arduous and costly sectioning process, since the spheroids will localize in different section planes. The development of microwell-containing apparatuses facilitated this process, allowing the simultaneous analysis of multiple spheroids in a more organized and cheaper manner [146].
To overcome these drawbacks of spheroid fixation and sectioning, faster and noninvasive microscopy approaches have been developed in the last years to image the innermost layer of live and fixed spheroids, such as light sheet fluorescence microscopy (LSFM), single or selective plane illumination microscopy (SPIM), and multi-photon microscopy (MPM) [147][148][149]. These new microscopic approaches allow deep tissue imaging study without the need of physical sectioning, while allowing dynamic processes to be studied in live 3D cultures at high resolution, under reduced light exposure and phototoxicity.

Electron Microscopy
Electron microscopy techniques are widely used to characterize spheroids because they provide high resolution, at nanoscale levels. High-resolution images of the internal structures can be generated by transmission electron microscopy (TEM) while high-resolution images of the surface of spheroids can be achieved by scanning electron microscopy (SEM) [56].
The TEM technique provides information on cell-cell interaction in the spheroids, such as cell junctions, and ECM deposition, as well as information on treatment outcomes such as apoptosis, cell shrinkage and organelle swelling [150]. Importantly, TEM is mostly used to analyze the distribution of drugs or nanoparticles in the spheroid [151].
The SEM technique provides high-resolution images and is used to analyze, for instance, cellular protrusions, integrity of cell-cell interactions, integrity of cellular membrane after anticancer drug treatment [56,152,153].
Both TEM and SEM are very informative although specimen collapse and morphological alterations can be associated with the steps involved in the procedures [154].

Flow Cytometry
Quantitative measurements such as cell viability, proliferation kinetics, cell cycle, apoptosis, and uptake of anticancer drugs and nanomedicines in spheroids can be performed using flow cytometry. Mechanical or enzymatic disaggregation of spheroids by trypsin or less toxic enzyme cocktail (Accutase ® ) is needed to obtain single cell suspension that can be stained and manipulated similarly to 2D cultures, and analyzed by flow cytometry [155]. For instance, single cells can be stained with calcein and ethidium to evaluate live cells and dead cells, respectively [56]. Other fluorescent dyes are used to analyze proliferating or quiescent cells (e.g., Propidium iodide), entry intro S phase of the cell cycle (e.g., 5-bromo-2 -deoxyuridine (BrdU) detected by a fluorescently labeled secondary antibody), or the expression of specific cellular proteins with fluorescently labelled antibodies. Flow cytometry analysis does not enable evaluation of penetration of anticancer drugs due to spheroid disaggregation and cell dissociation. However, it was reported that Hoechst 33342 (a fluorescent DNA dye) forms a marked diffusion gradient into the inner space of spheroids, therefore enabling cells of the different layers to be sorted on the basis of Hoechst staining intensity [156,157]. One major limitation of flow cytometry analysis is the need of a large amount of spheroids due to loss of cells during the process of cell dissociation [140].

Colorimetric Methods
Cell viability in the spheroids can be evaluated without the need of cell dissociation. For this purpose, are used colorimetric, fluorometric and luminescent methods that include acid phosphatase assay, Alamar blue, MTT assay, and lactate dehydrogenase quantification [140,158,159]. Nowadays, specialized kits for cytotoxicity assessment in spheroids are made available from many manufacturers. For instance, commercially available cell viability assays such as CellTiter-Glo 3D with better penetration of the reagents into the spheroids are easy to implement, and enable more accurate cytotoxicity determination [142,160]

Molecular Biology Tools
Standard molecular biology assays such as Western blot and qRT-PCR are useful to evaluate differential protein and gene expression, respectively, between 2D and 3D systems and/or before and after drug treatment. These techniques involve the use of cell lysis during the procedures of cellular protein and RNA extraction from the spheroids [59,161]. Table 4. Methods currently used to characterize spheroids and to evaluate drug effect.

Method Description Staining Methods/Markers Feature Evaluated Advantages (↑) and Limitations (↓) References
Phase contrast microscopy Monitorization of morphology and general state of spheroids.
-Size/volume and shape.
↑ Low cost and easy method to observe the general data on spheroids size and shape. ↑ Noninvasive. ↓ Does not provide enough quality in focus to obtain detailed data from complex 3D spheroid structures. [140,142,162] Fluorescence microscopy Uses fluorescent dyes to analyze specific structures in the sample; Monitorization of stained/immunostained spheroids or spheroid sections.
DNA staining by Hoechst or DAPI. DNA, nucleus.
↑ Allows easy monitoring of a wide range of features. ↓ For larger spheroids, processing for histological sectioning is required-spheroid fixation used in the histological procedure precludes the study of dynamic alterations in the spheroids over time. [144,145,158,[163][164][165][166][167] Fibronectin, laminin, and collagen IV staining. ECM deposition.
Live/cell death assays.

Bright field microscopy
Light is transmitted through the sample, and denser areas attenuate light transmission, originating contrast.
Distinction of nuclei and cytoplasmic structures.
↑ Low-cost method that offers a general overview of the sample structure (of a section). ↓ Requires spheroid processing for histological sectioning. [143,[168][169][170] Confocal laser microscopy The use of a focused laser spot with the removal of the out-of-focus light allows to acquire higher spatial resolution images.
Same markers described for fluorescence microscopy.
Spheroid architecture. ↑ High resolution data. ↑ 3D reconstruction. ↓ Restricted to small spheroids due to limited light penetration and to light scattering in thick tissues.
[143,171,172] The features described for fluorescence microscopy can also be evaluated.

Method Description Staining Methods/Markers Feature Evaluated Advantages (↑) and Limitations (↓) References
Light sheet fluorescence microscopy (LSFM) and single or selective plane illumination microscopy (SPIM) High resolution data from thick experiments through the use of planar illumination incident orthogonally to the direction of observation.
Same markers described for fluorescence microscopy.
The innermost layer of live and fixed spheroids.
↑ High spatial resolution. ↑ 3D reconstruction. ↑ Noninvasive. ↑ Does not require physical sectioning. ↑ Reduced light exposure and phototoxicity. ↓ LSFM may imply high processing time and memory in order to produce high-resolution 3D images; scattering and absorption of light may limit the penetration into specimens, although some efforts have been recently made to improve those issues. ↓ The upgrading of conventional microscopes to LSFM and/or SPIM technology may be complex and, in some cases, the optical sectioning capability may be limited. ↓ Some MPM limitations have been reported, such as weak endogenous signal strength, limited imaging materials, insufficient imaging depth.  Cell cycle analysis. [56,192,193] 5-bromo-2 -deoxyuridine (BrdU) + PI (or analog).
Live/dead cell analysis, detection of quiescent cells. [56] Hoechst 33342 DNA staining intensity dependent on the depth of cells in the spheroid. [156,157,196] Fluorescent staining of specific cellular proteins. [197,198]

Application of 3D Cultures in Anti-Cancer Drug Discovery and Delivery
The capacity to reproduce the in vivo 3D tumor environment such as cellular heterogeneity, gene expression patterns, cell differentiation, generation of hypoxia, activation of cell signaling pathways, and cell-cell and cell-ECM adhesions, are amongst the many advantages that prompted the use of spheroids for in vitro evaluation of chemoresistance, migration and invasion, and other aspects of tumor biology (e.g., cancer stem cells/tumorigenicity, hypoxia and tumor metabolism). We will focus on chemoresistance and migration/invasion, and provide a brief overview on the use of spheroids to study drug delivery. Details of the other aspects were reviewed elsewhere [64,70,212,213].

Chemoresistance
Drug resistance is a major concern responsible for the failure of the current chemotherapeutics and their ability to fight cancer, especially in aggressive and highly metastatic tumors. It is now well established that cancer cells, grown in vitro as 3D spheroids, more accurately mimic the drug behavior in terms of sensibility and resistance than cells grown as 2D monolayers [214]. This difference is probably due to the TME and the spatial organization of the spheroids [215]. Increased cell-cell and cell-matrix adhesions may lead to changes in gene expression. Upregulation of cell-adhesion molecules, such as lumican, SNED1, DARP32, and miR-146a, was reported to increase chemotherapeutic resistance in pancreatic tumor spheroids as compared to 2D monolayers [59]. Fibronectin protected DU145 prostate cancer cell spheroids against ceramide and docetaxel-induced apoptosis through interaction with Insulin like growth factor-1 receptor [216]. A variety of apoptotic stimuli, including combinations of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), ribotoxic stressors, histone deacetylase, and proteasome inhibitors, were reported to be highly effective against mesothelioma cells when grown as monolayers than when grown as multicellular spheroids [214].
Increased resistance to chemotherapeutic drugs in spheroids is attributed to many factors associated with their constitution and organization, such as hypoxia, altered cellular energy metabolism, the acidic microenvironment, the cellular heterogeneity including the presence of cancer stem cells, and cell-cell and cell-ECM interactions [215,[217][218][219][220][221][222]. The mechanisms by which these factors confer chemoresistance to spheroids were nicely reviewed in [223]. While most studies showed that cells in spheroids are more chemoresistant than cells in 2D monolayers, some studies reported that cells in MCTS are equally or even more sensitive to anticancer agents than their 2D monolayer counterparts. For example, the proteasome inhibitor PS-341 was shown to be equally effective in killing ovarian and prostate tumor cells grown in the form of multicellular spheroids, and tumor cells grown in monolayer cell culture [224].
A number of studies reported that spheroids are also more radioresistant than 2D monolayers. For instance, increased cell compaction increased the resistance of human colon adenocarcinoma spheroids to ionizing radiation [225]. Besides the aforementioned factors, radioresistance may be due to decreased radiation-induced DNA damage as a consequence of lack of oxygen in the spheroids, given that oxygen seems to be required to stabilize DNA damage upon radiation [226][227][228].

Migration and Invasion
The acquisition of motility and migratory ability is an important hallmark of malignant tumors. Common characteristics of solid tumors, such as hypoxia and soluble mediators-mediated interactions with stromal cells, drive tumor cell migration and invasion, through essential steps that involve, amongst others, actin cytoskeleton remodeling, changes in cell-cell and cell-ECM adhesion, and protein degradation of the surrounding ECM [229,230]. Therefore, the success of studying the multistep process of metastasis relies on a 3D microenvironment through which tumor cells can move and disseminate. In this sense, tumor spheroids are viewed as relevant in vitro models for studying invasion and migration processes [70,166,231,232]. For instance, 3D spheroids display adhesion and ECM molecule expression pattern similar to that of the tumor in vivo, and can also induce expression of proteins associated with metastasis [70,231,233]. Importantly, non-tumor cells are also present in the TME and continuously interact, through paracrine signaling, with cancer cells. For instance, fibroblasts were shown to promote contact-dependent cancer cell motility and invasion of 3D spheroids in co-culture with colorectal cancer cells, a finding validated in vivo [234]. Therefore, ideal migration/invasion assays should be performed in 3D co-cultures that also include non-tumor cells, such as macrophages, dendritic cells, endothelial cells, CAFs and immune cells, in order to better simulate the migration and invasion process found in tumor tissues. CAFs, through the release of cytokines and growth factors, together with the other stromal cells, promote the epithelial-mesenchymal transition in heterotypic 3D cell cultures, resulting in tumor development and metastasis [111,[234][235][236]. At the same time, endothelial cells in 3D co-cultures tend to accumulate in the peripheral layer, facilitating the adhesion and infiltration of immune cells [28]. In fact, immune cells can secrete interleukin 6 and MMP-9, which cause inflammation, angiogenesis and ECM degradation, thereby promoting tumor invasion and metastasis [237].
Several assays are available to determine the invasion and migration potential of cells in spheroids [70,232]. In the transwell-based or Boyden chamber assays, the spheroids are seeded on the top of a filter coated with a thick layer of ECM-derived components, usually collagen I, and invasion, in response to a chemo-attractant such as growth factors, can be measured by determining the number of cells that move from the top chamber to the lower chamber [70,232,238]. Additionally, the ability of the cells to invade cellular barriers can be determined by adding a layer of fibroblasts or endothelial cells on top of the matrix [70]. This latter is particularly relevant to mimic the ability of cancer cells to cross the blood vessel barrier and to invade deeply the tissues. Alternatively, spheroids can be completely embedded into different matrices, usually between two layers of ECM gel, where cells leave the spheroids and invade the surrounding matrix [96,239]. Sophisticated techniques combined with computerized quantification are now available to reproducibly perform optimized experimental conditions and to calculate the invasive index of cells [70,[239][240][241]. For instance, the extent and rate of tumor spheroid invasion, using the 3D spheroid invasion assay, was rapidly and reproducibly measured using imaging cytometer [238]. Spheroid invasion assays can also be used as a metric to measure drug efficacy [96]. For example, lower concentrations of the adjuvant gamma-linolenic acid caused an increase in glioma spheroid invasion, but increased the apoptotic index at higher concentrations [242]. In sum, spheroids have been widely utilized to study the role of mechanisms involved in cellular invasion, and represent a valuable tool for preclinical evaluation of therapeutic agents targeting invasion [96,166,232].

Spheroids and Nanomedicines
Systemic drug toxicity and poor efficacy remain a major concern in cancer therapy due to the lack of selective drug delivery to tumor tissues, stressing the need to improve tumor targeting [243]. Nanomedicines have thus emerged as promising approach to (actively) target tumor and improve drug delivery. These nanostructures are biocompatible, biodegradable, non-toxic, can be prepared on a large scale, can provide controlled drug release, and enhance tissue/cell-specific targeting, in addition to reducing side effects [244][245][246][247][248]. However, despite the promising preclinical outcome that was reported for a significant number of nanotherapeutics, only few nanodrugs reached the clinic and achieved the expected results in patients [243]. Many barriers influence the efficiency of nanomedicine delivery to the target tumor, that are not recapitulated by the 2D monolayer cultures.
Tissue penetration of nanoparticles (NPs) relies on their diffusion capacity through the ECM, which varies in density and size, and is also influenced by cell-cell interactions, necrotic core, hypoxia, and by the intravascular pressure irregularities due to vessel compressions applied by growing tumors [249][250][251]. In this sense, as outlined above, spheroids have gained in popularity over traditional 2D culture systems because their pathophysiological features are close to those of the native tumors, being an excellent model to evaluate nanodrugs and to better predict their clinical outcomes [101,197,212,252]. Consequently, spheroids have been used as valuable tool to study different physico-chemical proprieties of nanocarriers such as chemical composition, size, shape and surface properties, which are crucial for their penetration and antitumor efficacy [197,253,254].
A general observation from studies that used spheroids is that nanoparticles (NPs) penetration is inversely correlated to the particle size [159,[254][255][256]. NPs with small size (<100 nm) penetrate deeply and faster in the spheroids and distribute homogeneously, as compared to larger NPs (>100 nm) which remain confined to the superficial layers [159,[257][258][259]. However, NPs <50 nm were reported to interact with liver cells, and to be poorly retained in the tumor [260].
The surface charge of NPs also influences their penetration in the spheroids: negatively charged NPs penetrate deeply while their positive counterparts remain at the outer layers [56,199]. Yet, more effective drug delivery is warranted by NPs with positive surface charge due to electrostatic interactions with negatively charged cell membranes. To overcome this issue, it has been proposed the use of pH-responsive negatively charged NPs that can turn to positively charged ones once in contact with acidic conditions (e.g., tumor microenvironment), so that negative surface charge ensures deep penetration in the spheroids, while positive surface charge enables more effective drug delivery [199,261].
Although little information exists on the influence of NP shape on penetration and accumulation in the spheroid, the existing literature indicates that nanorods seem to diffuse more rapidly in spheroids compared to nanospheres, and that short nanorods (400 nm in length) accumulate more rapidly and are better internalized than long nanorods (<2000 nm in length) [262][263][264].
Interestingly, NP penetration into spheroids has been enhanced by modification of the surface coating. For instance, ECM-degrading enzymes such as collagenases have been used to coat NPs of up to 100 nm in size, which demonstrated superior (4-fold increase) penetration over control NPs [258]. Drug efficacy is the most important endpoint of any formulation, and it depends greatly on the penetration and accumulation into the spheroids [254]. In general, nanocarrier formulations with high penetration and accumulation in the spheroids exhibited better antitumor activity [159].
Comparison between NP delivery and efficacy between 3D tumor spheroids and animal models revealed key similarities between the two systems. For instance, the photosensitizer verteporfin encapsulated into lipid nanocarriers strongly reduced tumor cell viability of ovarian spheroid cancer cells, and also inhibited tumor growth in an orthotopic murine ovarian cancer model, when compared to free drug [265]. Similar to in vivo tissues, HepG2 cells in 3D hydrogels were more resistant to biotin-conjugated pullulan acetate nanoparticles (Bio-PA NPs) treatments compared to the 2D system [266]. Moreover, Bio-PA NPs exhibited similar anti-tumor activity in 3D culture cells and in in vivo xenografted hepatic tumor model [266]. Studies also observed that iRGD-conjugated nanoparticles with doxorubicin were able to accumulate with more efficacy and penetrate deeply into tumor in both SH-SY5Y spheroids and H22 tumor-bearing mice, restraining tumor growth in both systems [267]. Overall, this highlights the predictive power of spheroids for in vivo therapeutic efficacy, and their potential as promising alternative to animal models for cancer study, hopefully resolving high cost and ethical issues associated with animal use.

Concluding Remarks and Perspectives
It is consensual that 3D tumor models enable evaluation of anticancer drugs and nanomedicines in a condition closer to the real tumor, owing to their key features such as spatial organization, cell-cell and cell-ECM, diffusive gradients, complex cell signaling, drug resistance and metabolic adaptation. As reviewed here, these features are missing in 2D culture systems and, consequently, 3D culture models in preclinical evaluation are expected to provide more accurate results of the therapeutic potential of anticancer drug candidates, thereby increasing the predictability of the in vivo efficacy. Identifying and eliminating those therapeutics that did not show any interesting efficacy in 3D cultures will reduce animal use and speed up the number of therapeutics that reach the clinic.
It is noteworthy that most of the published works used spheroids made of only cancer cells, and, thus, do not represent the complexity associated with the diversity of the cellular and non-cellular components present in the real tumor. Spheroids that incorporate cell types recapitulating the vasculature (e.g., endothelial cells), the immune system (e.g., leukocytes) and ECM production (e.g., fibroblasts) are, thus, highly recommended. This is important as it would make the geometry of drug penetration in the spheroids closely similar to that in vivo, therefore providing better prediction of drug effects and delivery mechanisms and, at the same time, reducing costly investments associated with the ultimate step of clinical investigations.
Standardized methodologies for generation and characterization of spheroids are urgently needed, this would avoid variability in size and homogeneity, as well as in biological effect evaluation. Although considerable progress has been made to adapt existing 2D culture analysis assays to the spheroid model, many challenges remain to be addressed. Enabling acquisition of high-resolution images from intact spheroids remains a major challenge, due to the size of spheroids and poor light scattering. On the other hand, histological procedures for spheroid sectioning require special care in handling, as specimen tend to collapse or fracture easily. Mass production, together with developing easy to handle spheroids that are time and cost effective, with reduced workflows of culture and analysis, is crucial in order to encourage their routine use in drug discovery research. We are, yet, still far from giving up using animal models for safety and efficacy studies of drugs. Meanwhile, and ideally, the use of spheroids in preclinical testing could reduce the number of compounds progressing to in vivo testing, thereby reducing the numbers of animals used.
In conclusion, the use of 3D models to assess tumor penetration, accumulation and antitumor activity of drug and nanomedicine candidates is becoming a reality, and should turn out a mandatory step between 2D and in vivo models in the near future, with a great impact on the transferability of new anticancer drugs from bench to bedside. Hopefully, the generation of tumor spheroids from the patient's own cells may enable personalized approaches to screening and selecting the appropriate drugs for the patients.