The Changing Face of in vitro Culture Models for Thyroid Cancer Research: A Systematic Literature Review

Background: Thyroid cancer is the most common endocrine malignancy worldwide. Primary treatment with surgery and radioactive iodine is usually successful, however, there remains a small proportion of thyroid cancers that are resistant to these treatments, and often represent aggressive forms of the disease. Since the 1950s, in vitro thyroid culture systems have been used in thyroid cancer research. In vitro culture models have evolved from 2-dimensional thyrocyte monolayers into physiologically functional 3-dimensional organoids. Recently, research groups have utilized in vitro thyroid cancer models to identify numerous genetic and epigenetic factors that are involved with tumorigenesis as well as test the efficacy of cytotoxic drugs on thyroid cancer cells and identify cancer stem cells within thyroid tumors. Objective of Review: The objective of this literature review is to summarize how thyroid in vitro culture models have evolved and highlight how in vitro models have been fundamental to thyroid cancer research. Type of Review: Systematic literature review. Search Strategy: The National Institute for Health and Care Excellence (NICE) Healthcare and Databases Advanced Search (HDAS) tool was used to search EMBASE, Medline and PubMed databases. The following terms were included in the search: “in vitro” AND “thyroid cancer”. The search period was confined from January 2008 until June 2019. A manual search of the references of review articles and other key articles was also performed using Google Scholar. Evaluation Method: All experimental studies and review articles that explicitly mentioned the use of in vitro models for thyroid cancer research in the title and/or abstract were considered. Full-text versions of all selected articles were evaluated. Experimental studies were reviewed and grouped according to topic: genetics/epigenetics, drug testing/cancer treatment, and side populations (SP)/tumor microenvironment (TME). Results: Three thousand three hundred and seventy three articles were identified through database and manual searches. One thousand two hundred and sixteen articles remained after duplicates were removed. Five hundred and eighty nine articles were excluded based on title and/or abstract. Of the remaining 627 full-text articles: 24 were review articles, 332 related to genetic/epigenetics, 240 related to drug testing/treatments, and 31 related to SP/TME. Conclusion: In vitro cell culture models have been fundamental in thyroid cancer research. There have been many advances in culture techniques- developing complex cellular architecture that more closely resemble tumors in vivo. Genetic and epigenetic factors that have been identified using in vitro culture models can be used as targets for novel drug therapies. In the future, in vitro systems will facilitate personalized medicine, offering bespoke treatments to patients.


INTRODUCTION
Thyroid cancers are the most common endocrine malignancies worldwide (1). In most developed countries the incidence of thyroid cancer has been steadily rising, partially attributed to an increased diagnosis of subclinical papillary micro-carcinomas (2). Despite its prevalence, the overall mortality rate of thyroid cancer has remained low (0.5 per 100,000 patients) (3). Surgery followed by radioactive-iodine (RAI) therapy continue to be the first line treatment modalities for thyroid cancer. Overall survival rates following primary treatment are high (>98% 5-year survival), however, for the 1-2% of patients with aggressive forms of the disease or the 5-10% of patients with distant metastases, the prognosis is far worse (4).
Thyroid cancer research is focused on improving our understanding of the biological mechanisms that initiate and propagate the disease in the hope of refining diagnoses and formulating bespoke treatments to improve patient outcomes. An essential foundation of this research is the use of in vitro experimental models. In the simplest terms, an in vitro culture model is comprised of a vessel (e.g., dish, plate, or well) containing a culture medium to support and maintain cells outside of the body for experimental purposes. Culture models have evolved from growing homogenous cell populations in a 2-dimensional (2D) monolayer into complex 3-dimensional (3D) heterogeneous multicellular structures that resemble tissues in vivo. Cells used in these models can be derived from immortalized cell lines, pluripotent stem cells or ex-vivo human tissue. Individual patients' explanted thyroid tissue can be maintained in vitro for several days using microfluidic technology, an advancement which will open the gateway for personalized cancer medicine (5). This review summarizes how in vitro culture models have evolved and how they have been applied to thyroid cancer research.

Search Strategy
The National Institute for Health and Care Excellence (NICE) Healthcare and Databases Advanced Search (HDAS) tool was used to search EMBASE, Medline and PubMed databases. The following terms were included in the search: "in vitro" and "thyroid cancer". The search period was confined from January 2008 until June 2019. A manual search of the references of review articles and other key articles was also performed using Google Scholar.

Article Selection
All experimental studies and review articles that explicitly mentioned the use of in vitro models for thyroid cancer research in the title and/or abstract were considered (Figure 1). Full-text versions of all selected articles were evaluated. Experimental studies were reviewed and grouped according to topic: genetics/epigenetics, drug testing/cancer treatment and side populations (SP)/tumor microenvironment (TME; Figure 2).

THE EVOLUTION OF CELL CULTURE MODELS 2D vs. 3D
There are two basic systems for growing cells in culture, as a single layer of cells on an artificial substrate (adherent culture) or free-floating in the culture medium (suspension culture). Thyrocyte 2D monolayer culture systems have been used since the late 1950s (6). Their main limitation is that thyrocytes are unable to arrange themselves into their normal physiological follicular structures when cultured on adherent plates in standard culture medium (7). Instead, thyrocytes are arranged into a continuous epithelial sheet, with the apical aspect of the cell facing the culture medium above and the basal aspect facing the surface of the dish (Figure 3).
When thyrocytes are suspended in non-adherent vessels containing culture medium, they arrange themselves into a follicular structure ( Figure 4A). However, the orientation of the cells is such that the apical aspect with microvilli are facing outwards in contact with the culture medium (7) (Figure 4B), thus creating an "inside-out" follicle. If these insideout follicles are then embedded into a 3D substrate emulating thyroid extracellular matrix (ECM; e.g., type 1 collagen gel) the cellular polarity inverts with the microvilli facing inwards toward the follicular lumen, creating a true physiological thyroid follicle ( Figure 4C). This leads to the conclusion that thyroid folliculogenesis is dependent on the presence of ECM.
Such collagen gel cultures were developed in the 1970s to enable reconstruction of thyroid follicles in vitro (8). These 3D culture systems have not only allowed study of thyroid folliculogenesis but also thyroid function under stimulation by factors such as thyroid stimulating hormone (TSH) and iodine, as well as interactions between thyrocytes and the ECM (9).
By replicating the in vivo thyroid cellular structure, 3D cell culture systems allow researchers to study the complex spatial morphology that facilitates cell-cell and cell-matrix interactions and signaling-a huge advantage over 2D monolayer culture systems (10). Advances in cell biology, microfabrication and tissue engineering have facilitated development of a wide range of 3D cell culture techniques including spheroids, organoids and microfluidic systems (11).

Spheroids
One of the most common 3D culture models used in thyroid cancer research today is the multicellular spheroid. Spheroids are cellular aggregates consisting of several thousand phenotypically distinct cells. A technique for developing spheroids was pioneered by Sutherland et al. in the early 1970s for testing the response of radiation exposure on tumor cell lines (12). The resulting dose response curves were similar to those produced when irradiating ex-vivo solid animal tumors. Since then, several techniques to create spheroids have been established including hanging drop plates, low-adhesion surface methods, and suspension culture in bioreactors which drive cells to selfaggregate under dynamic conditions ( Figure 5). Spheroids can be grown to a variety of sizes, depending on the needs of the study     (13). As well as thyrocytes, these models have been developed to include co-culture with immune cells such as macrophages (14) and neutrophils (15).
The aggregate structure of 3D spheroids supersedes 2D monolayers in terms of their ability to reproduce the cellular heterogeneity of tumors in vivo. Depending on the size of the spheroid, the structure usually consists of an outer layer of proliferating cells, a middle layer of quiescent cells and a central core of necrotic cells caused by a nutrient and oxygen diffusion gradient (Figure 6). This structural heterogeneity is important to consider when spheroids are used to test drug sensitivity.
Despite their many advantages, spheroids do present some practical challenges. Firstly, it can be difficult establishing Frontiers in Surgery | www.frontiersin.org spheroids from a small seed number of cells. Also, controlling proliferation, specific ratios of various co-cultured cell types and maintaining spheroids of a uniform size is not always achievable (16). This leads to issues with standardizing culture and assay protocols as well as evaluating output data (17). Currently there is no reliable, standardized highthroughput assay that allows spheroid use for drug screening. Furthermore, despite having the ability to co-culture thyroid cancer cells with select immune cells, spheroids do not entirely mimic the TME in terms of representing all the cell types present in vivo.

Organoids
Organoids are 3D in vitro cellular structures derived from either embryonic stem cells (ESC), induced pluripotent stem cells (iPSC), organ-specific adult stem cells (ACS), or primary cancer cells (18) (Figure 7). Organoids are defined by three characteristics: self-organization, multicellularity, and functionality (19). The constituent cells of an organoid are arranged into a 3D structure characteristic of the organ in vivo. They generally contain all the cell types found within that organ and execute the same functions they would normally carry out.
Novel culture systems containing laminin-rich Matrigel as a substitute for ECM and growth factors including EGF, Noggin, Wnt, and R-spondin allow the development of organoids from stem cells (20). The in vitro process utilizes the defining characteristics of stem cells: namely, the clonal expansion capacity and production of daughter cells that can differentiate into multiple cell types (21). It then relies on cell-cell and cellmatrix interaction and signaling to form the organoid structure.
Organoids have been applied to understand stem cell biology, organogenesis and pathogenesis of various diseases (10,18,(20)(21)(22)(23)(24)(25)(26). Organoids have huge potential for modeling cancer and many organoid systems such as breast (27), colorectal (28), and prostate (29) have already been established in experimental studies. In 2018, Saito et al., established a thyroid organoid culture system from murine stem cells (25). These organoids successfully functioned as thyroid tissue, producing thyroglobulin and thyroid hormone (T3) when exposed to thyroid stimulating hormone (TSH). After p53 knockout, these organoids were xenografted into recipient mice which subsequently developed poorly differentiated thyroid cancer. Presently this is the only published study that has established a thyroid organoid as a novel experimental model.
Although organoids closely represent the cellular structure and function of in vivo tissues, there are still limitations-they often only demonstrate the initial stages of organogenesis/tumorigenesis, they lack the full range of cells that exist in the TME, and they do not develop tissue support structures such as a vascular or neuronal network (11).

Microfluidic Systems
Microfluidic systems (MFS) are devices that maintain and analyze small ex-vivo tissue samples or 3D cultured cells in a pseudo-in vivo state (30). The basic design is made up of a "chip" which houses the tissue sample connected to inlet and outlet tubing for circulating fluids. MFS mimic the human body's vasculature and lymphatics through continuous perfusion of nutrients via micro-volumes of fluid while simultaneously removing waste products. A significant advantage of maintaining tissue in these devices is that the cells remain viable and maintain tissue architecture for longer periods (3-7 days) than conventional in vitro culture systems (31).
MFS have been used to interrogate numerous types of cancer such as breast (32), lung (33), head, and neck squamous cell carcinoma (31), and only very recently thyroid cancer (5). These devices offer the ability to test drugs and RAI sensitivity of individual thyroid tumors to potentially customize/personalize therapeutic regimes.

Genetics and Epigenetics
In the past ten years there have been over 300 published papers utilizing in vitro models to study the molecular biology of thyroid cancer. Mutations of genes such as RET, BRAF and RAS are widely recognized as contributing to thyroid carcinogenesis (34). These mutations lead to uncontrolled cellular proliferation, dedifferentiation and metastasis through signaling pathways such as mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase (PI3K)/Akt, and Wnt/β-catenin (35). Epigenetic factors, including messenger RNA (mRNA), micro RNA (miRNA) and long non-coding RNA (lncRNA), control gene expression through mechanisms such as DNA methylation and histone modification. Epigenetic factors can be over-expressed or underexpressed in thyroid cancers and activate the same signaling pathways mentioned above ( Table 1).
An important feature of these epigenetic studies is the ability to reverse the effects on gene expression and observe phenotypic characteristics of the cells. The general experimental methodology described in the studies begins with collection of human thyroid cancer specimens taken at the time of surgery and extraction of the messenger RNA. Real-time polymerase chain reaction (RT-PCR) is then performed on both the malignant and adjacent healthy thyroid tissue to compare the epigenetic profiles. Following this, established thyroid cancer cell lines are manipulated to replicate the particular expression profile identified in the human tissue samples. These cell lines are then cultured in vitro as monolayers and/or spheroids to examine phenotypic characteristics such as proliferation (e.g., Ki67 staining), cell viability and apoptosis (e.g., TUNEL assay), migration and invasion (e.g., wound-closure and Transwell chamber assays).
The study of cell migration in thyroid cancer research is essential as metastatic dissemination is a significant prognostic factor. For 2D monolayer culture systems, the wound-closure and the transwell migration/invasion assays are widely used. These assays can be applied to study the migratory ability of a whole cell mass but also an individual cell's morphology (122)(123)(124)(125). Thyroid cancers that are known to have an aggressive phenotype can be studied for morphological features such asb invadopodia (126).
Alternatively, thyroid cancer cells can be completely immersed into a 3D matrix-either as a single cell suspension or more commonly as a spheroid (122,125). This allows cells to migrate away from the tumor mass in any direction. The extent of migration/invasion is monitored at set intervals over the course of several days. This technique offers the benefit of performing   the assay without having to re-plate the cells, as well as more closely simulating cell migration from a tumor mass in vivo. The oxygen and nutrient diffusion gradient present within a spheroid structure can promote migration and invasion through changes in gene expression-not present in 2D culture models. The effect of co-cultured cells such as macrophages on migration and invasion of thyroid cancer cells has also been explored as these immune cells have roles in epithelial-mesenchymal transition and subsequent tumor progression (14,15). These assays are also applied in research studies testing the cytotoxic effects of chemotherapy agents. Although no single genetic/epigenetic change has been reported, in all cases of thyroid cancer there are common signaling pathways which are affected. The epigenetic factors examined in the studies listed in Table 1  The conclusion is that epigenetic molecules have the potential to be used as biomarkers as well as targets for drug therapy. The observation that thyroid cancer progression is associated with an accumulation of epigenetic changes has led to the development of drug treatments targeting these pathways such as multitargeted tyrosine kinase inhibitors (TKIs), demethylating agents and histone deacetylase inhibitors (HDACi).

Drug Testing
Drug attrition rates for cancer are much higher than in other therapeutic areas. Only 5% of agents that have anticancer activity in preclinical development demonstrate a sufficient efficacy in phase III testing (127). Although surgery and RAI therapy are primary therapeutic modalities for all subtypes of thyroid cancer, targeted drug therapies such as the tyrosine kinase inhibitors (TKIs) vemurafenib, sunitinib and lenvatinib are available for those patients with either rapidly progressing, recurrent or RAI-resistant thyroid cancers (128). Recent studies using in vitro models have focused on testing drug combinations to enhance tumor sensitivity to established chemotherapy agents, as well as testing novel anticancer agents and drug delivery systems ( Table 2).
Nanontechnology, and more specifically manufacture of nanoparticles for drug formulation and delivery has been a promising area of research. Drug products that contain proteins or nucleic acids are susceptible to pharmacokinetic degradation. Nanoparticles can be customized for targeted delivery of drugs to improve bioavailability and provide controlled release of medication (167). For thyroid cancers, nanoparticles have been used to deliver sorafenib (149), anti-hTERT siRNA (150), capsacin for use in photothermal therapy (156), and I 131 labeled anti-VEGFR2 antibodies for targeted drug delivery and have shown promising results in both in vitro and in vivo cancer models.
As with the epigenetic studies, most drug studies utilize a combination of primary human thyroid cancer cells and established cell lines. Cells are cultured in monolayers and spheroids and exposed to incremental drug levels to observe cell survival and apoptosis. Thyroid cancer cells are also injected into animal models (patient derived xenografts-PDX) and treated with drugs to validate in vitro findings.
3D culture models are emerging as improved experimental models for preclinical target identification. Although spheroids more closely resemble the tumor in vivo, there is currently no commercially available standardized high-throughput assay for drug screening. Earlier studies by Li et al. (168), Guiffrida et al. They all observed that drug resistance was much higher in cells that formed spheroids than in monolayers. These findings have been attributed to the diffusion dynamics seen in spheroids as well as the discovery of side populations of cells within tumors that demonstrate stem cell-like properties.

Cancer Stem Cells
Cancer stem cell (CSC) theory challenges the classical model of carcinogenesis (where any cell in an organ has the potential to transform through gene mutations) following the discovery of distinct populations of pluripotent tumor stem-like cells within solid tumors (171). In thyroid cancer, small populations of cells within a tumor have displayed distinct CD surface antigens and gene expression (e.g., Oct 3 and 4, Nanog) that are known to be associated with stem cells identified in other forms of cancer (172) ( Table 3).
Generally, the studies have used in vitro sphere formation assays and PDX (using immunedeficient murine models) to confirm the existence of CSC. In 2007, Mitsutake et al., were the first group to identify and characterize a very small side population of putative thyroid cancer stem cells (CSC; 0.02-0.25% of total number of cells) from thyroid cancer cell lines (187). The highest percentage of CSC was seen in anaplastic thyroid cancer cell lines (ATC). The CSC in this study demonstrated stem-like properties of self-renewal and differentiation potential as well as altered gene expression profiles compared with non-CSC cells.
Since then more researchers have used in vitro models to identify specific tumor markers in thyroid CSC previously validated in other types of cancer. High levels of Oct-4, SOX-2, NANOG, and CD44 have been associated with thyroid CSC (53,168,170,177). Conversely, the cells isolated in these studies expressed low or completely absent levels of thyroid-specific differentiation markers such as TTF1, PAX8, and TSH-R.
When proliferation, migration, and cell survival assays have been applied to CSC in vitro they have demonstrated increased metastatic potential and reduced apoptosis (170,172,182,188). Additionally, they are largely quiescent which allows them to escape chemotherapy agents that normally target rapidly dividing cells (169).

CONCLUDING STATEMENT AND FUTURE PERSPECTIVE
In this review we have established that in vitro cell culture models have been the workhorse in thyroid cancer research for decades. There have been many advances in culture techniquesdeveloping complex cellular architecture that more closely resemble tumors in vivo.
In vitro culture models have provided researchers with a reliable platform to study the molecular and cellular biology of thyroid cancer, as well as for testing drugs prior to human trials. In the future, the promising field of personalized cancer medicine will establish effective treatment strategies based on an individual tumor's genetic profile and predicted drug response through in vitro culture techniques.

AUTHOR CONTRIBUTIONS
DC was the primary author of this article. VG and JG also reviewed articles that were included in the literature review. AR, RE, VG, and JG contributed to proof reading and editing of the final version. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
The authors would like to acknowledge the reproduction of Figure 7 from the Open Access article by Xu et al. (22). Institutions: Hull University Teaching Hospitals NHS Trust, Faculty of Health Sciences-University of Hull.