To overcome the treatment failure due to the cancer recurrence and metastasis related to CSCs biological features and their interference with intrinsic drug resistance mechanisms, design of appropriate preclinical models is one of the most pressing issue to test targeted therapies [35, 36]. Identification of the ideal tools to clarify CSC biology is still proceeding, and great efforts have been paid to improve isolation and treatment modalities in CSCs research . Spheres culture in which CSCs are trapped and enriched, has been recommended as an extremely effectual CSC isolation method for cancer cell lines and solid tumors [37-39]. Such an anchorage independent sphere forming system under non-adherent, nutritional deficient and serum free conditions prevents the differentiation of stem-like cells and leads to eliminate non-CSCs cells through apoptosis [40-42]. The superiority of this method is based entirely on the intrinsic properties of CSCs and leads to enriching the CSC subpopulations regardless of the expression patterns of cell surface markers [43, 44]. Moreover, spheroid cultures offer an ideal platform for routine toxicity and drug efficacy testing to determine safe exposure doses in designed cellular models [45-48]. In addition, previous studies have shown that different cancer cell lines form spheres with distinctive morphological and functional features based on sphere-formation method [49-51]. Hence, the status of morphological characteristics, CSC properties and genes expression profile of EMT and drug resistance is important in selecting appropriate model.
In this regard, in the current study, the forced ﬂoating using low attachment plates and hanging drop methods, as more applicable and feasible methods were used to generate CSC-enriched spheroids from Caco-2 and HT-29 cell lines, respectively. We chose these two cell lines because of their widespread use in research and drug discovery. Caco-2 cell line is widely used in pre-clinical investigations, drug permeability and solubility studies and nanoparticle translocation. The cell line is accepted by the U.S. Food and Drug Administration (FDA) to support new drug applications. In addition both Caco-2 and HT-29 cells are widely applied in the study of intestinal transport and predict bioavailability [52-54].
To generate spheroids serum-free condition was applied to prevent the differentiation of stem-like cells and elimination of non-CSCs population using suspension culturing in low attachment plates [55-58]. Although, both cell lines showed capacity to form spheroids, the HT-29 cells generated loose and disintegrating spheres in poly-HEMA plates, whereas, they formed compact spheroids when were cultured as hanging drops. The comparison of morphological and molecular characteristics of Caco-2 and HT-29 derived spheroids has been summarized in Table 3.
Generation of dense and practical spheroids strongly depend on cellular interactions through the gap junctions between single cells [59, 60]. Hence, hanging drop method was selected to generate HT-29 spheroids with compact structure due to its advantages in improvement of cells accumulation and adhesion potential with least damage to the spheres through gravitational force. Our observation was in line with the previous study in rat pancreatic beta cells which hanging drop method enhanced connexin protein accumulation and generation of tight spheres . Moreover, murine and human brain tumor cell lines showed more consistent structures in brain tumor spheroids with applying of hanging drop method compared to spinner culture . Therefore, it seems that different cell lines can be compatible with different spheroid formation methods, as in our study as well; hanging drop and low attachment poly-HEMA coated plates were suitable, rapid and time efficient for spheroid formation from HT-29 and Caco-2 cells, respectively. Some limitations of hanging drop including possible shattering of spheroids due to mechanical shock in transferring of spheroids to the conventional culture plates and high spheroid-diameter variation compared to low attachment culture can lead to changes in initiate characteristics in HT-29 spheroids. Whereas, forced ﬂoating method using low attachment culture with more reproducibility than hanging drop could be the most suitable method for high-through put screening. Therefore, appropriate cells and spheroid formation methods must be accurately chosen based on aim of study and tumor type.
Although both spheroids derived from HT-29 and Caco-2 cells exhibited similar roundness and smooth surfaces, the light and electron microscopic examination revealed some differences. The Caco-2 spheroids presented hollow core structure as distinct morphological feature which has been reported before by others [62, 63]. Samy et al, also found that after 5 days culturing in Matrigel, the Caco-2 cells self-organized into intestinal epithelial like cells as spheroids with a confluent monolayer surrounding a hollow lumen . This hollow structure in Caco-2 cells is account to the functional polarization through E-cadherin-dependent cell-cell adhesion [65, 66]. Since the hollow core is the common feature of the Caco-2 spheres formed by different methods [62, 63], it can inference that this feature is cell line dependent and may not related to the sphere formation methods.
The Caco-2 spheroids features in expression of drug transporters similar to human intestinal and also their ability to recreate the spatial organization similar to intestinal epithelial cells have been made them as an improved platforms in drug screening [64, 67, 68]. However, Caco-2 spheroids with round shape and hollow lumenized morphology have widely employed as a reproducible in vitro model for studying intestinal features and functions as well as intestinal drug metabolism and uptake [64, 69], they are not compatible with anti-cancer drug researches which are based on evaluation of the spheroid size reduction and drug penetration into inner layers. The roundness and smooth surface of spheroids which hide individual cells, is ascribed to high ECM secretion and strong cell-cell adhesion and could help analysis of drugs efficacy more accurately [49, 70]. Accordingly HT-29 spheroids with compact round shape may be more suitable platforms for anti-cancer drug testing in CRC than Caco-2 spheres. In general, in order to create more practical and functional CRC spheroid models, it is essential to characterize the gene expression alterations, invasion, and drug transporters in spheroids to standardize them based on research requirements [61, 64, 71-73]. Hence, CRC spheroid models were further assessed to determine their potential in enrichment of CSCs-related characteristics, including the cell surface marker patterns, serial sphere formation capability, and gene expression profiles of multipotency, EMT and drug resistance transporters compared to the parental cells. Our findings showed that the expression levels of stemness genes could be affected by CRC spheroid culture, and both spheroids displayed similar high expression levels of pluripotent stem cell genes (KLF4, OCT4 and C-MYC) when compared to their parental counterparts, while drastic higher expression level of SOX-2 and NANOG was found in Caco-2 and HT-29 spheres, respectively. It has been also documented that the overexpression of NANOG as an oncogene along with OCT4 is a prominent characteristic of CSCs and is associated with EMT transition of CSCs and drives tumor progression and poor prognosis in patients with breast and colorectal cancer [74-79]. Furthermore, many studies have proved that SOX-2 overexpression is correlated with self-renewal capacity, a poorly differentiated-aggressive phenotype and clinicopathological characteristics of CRC patients [80-82]. Therefore, high expression levels of stemness genes, specially, SOX2 and NANOG that were observed in our CRC spheroids could be served as an indicator of self-renewal potential.
We extended our study by assessing the expression of CSC surface markers in CRC spheroids compared to the parental cells. Our findings independently corroborated the expression profiles of stem cell-related genes which evidenced by enriching the CD166, CD44, CD133 positive populations of CSCs in CRC spheroids. This was in line with other reports in various cancers, where non-adherent spheroid cultures possessed more characteristics of CSCs [44, 83-86]. Moreover, serial sphere formation which has been applied as an appropriate platform for long-term expansion of cells with self-renewal capacity and imitates the tumor heterogeneity , was maintained in CRC spheroids cells during the long-term cultures. Thus, these results further advocated the efficiency of the serum free and non-adherent condition in enrichment of CSCs.
The complexity of cross-talks between stem cell–related genes and EMT are still unclear and need to be more clarified. To determine the association of the EMT genes expression level with the enriched CSC nature in spheroids, we compared the expression of Twist1, Snail1, ZEB1, Vimentin, E-cadherin and N-cadherin in spheroids than parental cells. Our results showed that in spite of up-regulation of Twist1, Snail1, ZEB1 and Vimentin, unexpectedly E-cadherin was up-regulated and N-cadherin was down-regulated in spheroids which are in contrast to the results of EMT process from other similar studies [87-90]. Several reports suggested the strong correlation of E-cadherin down-regulation and N-cadherin up-regulation as the main hallmark of EMT [77, 87-91]. It can postulate that the expression level of EMT genes, might be varied widely due to the absent of definite correlation between the gene transcripts levels and their corresponding proteins and do not necessarily reflect their protein levels in CSCs [92-96]. In addition, Jolly, Jia et al. displayed that the partial EMT is associated with stemness , deciphering the coupling of EMT and stemness needs further investigations.
The increased expression of ATP-binding cassette (ABC) transporter genes such as ABCB1, ABCC1, and ABCG2, as other CSC-related characteristic is involved in regulation of self-renewal and multidrug resistance (MDR) in ovarian and colon cancer cell lines [98-103]. In agreement with Collura et al. study , our results also demonstrated significant increased expression of ABC B1 and G2 genes in HT-29 spheres compared to monolayers and Caco-2 spheres which further verified the CSCs enrichment.