The CSC theory postulates that cancer cells are organized in a hierarchical structure, with only a small portion of cells exhibiting stem cell characteristics such as self-renewal and multipotentiality, with a propensity to metastasize, display treatment resistance, and induce tumor recurrence. Furthermore, CSCs are distinguished by their ability to initiate tumors when transplanted into an immunocompromised host[30]. In our previous studies, we used flow cytometry to identify CSCs in primary Rb tumors based on their size and surface marker expression[5]. This study aimed to evaluate two other characteristics of CSCs (CD44+/CD133), i.e., drug resistance in in vitro assays and tumor-initiating properties in the CE-CAM model, and to document the increased drug resistance, enhanced expression of stem cell-related genes and expression, and ability to generate tumors in vivo with metastatic potential to the embryo.
Based on our previous work[5], CD44+/CD133- cells were identified as CSCs in Rb primary tumors, and these cells were enriched by phenotypic sorting using FACS followed by MACS. Typically, CSCs are present in a small population, whereas in Rb, the frequency of CSCs varies among tumors. This is due to the involvement of several factors, such as intratumor heterogeneity and its microenvironment[31, 32]. Winter et al. demonstrated that bilateral Rb exhibits intrapatient and intratumoral heterogeneity and is responsible for the differential expression of genes and proteins[33]. In most cases, the surface markers CD44 and CD133 have been used either alone or in combination for the isolation and enrichment of CSCs. However, in Rb, CD133 is involved in photoreceptor differentiation. Our prior research on Y79 cells similarly confirmed CD133- cells as a CSC marker utilizing functional gene expression and in vivo tumorigenicity and metastatic potential[6]. On the other hand, Y79 cells do not express CD44[6, 29]. These observations are corroborated by the findings of Ma et al., who found that CD44 expression was elevated in long-term serum-free cultures of neurospheres derived from primary Rb in comparison to that in CD133-differentiated cells in vitro[34]. In contrast, using in vitro propagation and stem cell-like properties, Zhixin Tang et al. reported that CD133 is a CSC marker in Rb primary tumors and cell lines with highly expressed stem cell markers and self-regenerative growth in culture. However, this study did not demonstrate that CSCs were enriched in the CD133 surface marker[10].
Furthermore, isolated CSCs have shown greater intrinsic multidrug resistance to CPT, ETP, VCR, and MLP than non-CSCs of Rb, a hallmark characteristic of CSCs[35]. The inherent resistance of CSCs is due to the increased expression of multidrug resistance proteins such as ABCB1, ABCC1, and ABCG2[36, 37]. These isolated Rb CSCs exhibited enhanced multidrug resistance gene expression and exhibited stemness, EMT, invasion, and metastasis; these genes were shown to regulate many pathways essential for the survival and metastasis of CSCs[38]. The property of drug resistance was also documented by the CSCs in Rb Y79 cell lines, which could be reversed with the use of nano-formulations. Hence, understanding drug resistance properties and pathways is important to pave the way for future targeted therapies.
Another important characteristic of CSCs is their tumor-initiating properties in vivo. Several established transgenic and xenograft animal models have been developed for Rb, and the majority of them are mammalian models[39, 40]. However, Rb is a neonatal tumor, and developmental animal models are difficult to recapitulate in this system due to several technical difficulties. In order to create strategies for efficient therapeutic targeting, it is crucial to model Rb carcinogenesis and metastasis within a developing microenvironment. The CE-CAM model provides the innate tumor microenvironment that resembles Rb and is also a reliable model for drug testing[41, 42]. Our study employed CM-Dil dye to label cells to evaluate the effects of CAM on tumor nodules and metastatic dissemination to the embryo. Chen et al. effectively used CM-Dil dye in an earlier Rb zebrafish orthotopic xenograft model to track tumor cell invasion locally as well as through the optic nerve[43]. However, it should be noted that the intensity of the CM-Dil dye may decrease as a result of cell proliferation. Busch et al. investigated the potential of the CE-CAM model in Rb studies and showed the tumor-forming capacity of various cell lines (normal and chemoresistant) in soft agar as well as CAM and assessed the tumor suppressor effect of the trefoil factor family (TFF) peptides in Rb cells using the CE CAM model[25–28]. They do not provide any evidence for metastasis within the embryo, although the intention of their investigation was to characterize cell lines and determine chemo-resistant mechanisms. The authors may have found no evidence of invasion because they evaluated the metastasis on E17/18, an embryonic stage at which the immune system is fully developed and activated. We agree with their findings that Y79 cells can create nodules on CAM and have the potential to spontaneously invade CAM tissue. This finding is consistent with research on glioblastoma, ovarian cancer, prostate cancer, and uveal melanoma, among other solid tumors[20, 23, 42]. A novel aspect of our study is the use of this model to study the tumorigenic and metastatic potential of Rb CSCs. Since mouse xenograft models are thought to be the gold standard way to assess CSCs in vivo, they are invariably expensive and are linked to stringent ethical constraints. Pinto, M. et al. demonstrated breast CSC activity using a limit dilution assay with a CE-CAM model and found that BCSCs (CD44 + CD24−) were isolated from MDA-MB-231 breast cancer cells[44]. Similarly, Muenzner, J. K et al. generated hepatocellular carcinoma stem cells via in vitro 2D and 3D cell culture systems as well as an in vivo CE-CAM model by IVIS spectrum imaging and histology[45]. The presence of feeder blood vessels around the tumor nodules generated by Rb CSCs is a possible indicator of tumor neovascularization. In a neuroblastoma CAM xenograft model, Ribatti et al. showed that there were numerous host arteries around and inside the tumor nodule, indicating that the tumor cells were able to induce angiogenesis by exploiting the chorioallantoic vasculature[20].
The IVIS spectral imaging and histopathological reports confirmed the evidence of metastasis of Rb CSCs in the cephalic region of chick embryos. Further quantitative studies of alu-qPCR analysis confirmed the presence of metastasis in the brain. Clinical evidence suggests that Rb can metastasize to the brain by invasion of the optic nerve, as well as to the bone and liver via hematogenous dissemination[46, 47]; hence, the involvement of the optic nerve, uvea, and sclera is regarded as an important prognostic marker for predicting metastasis. Our findings support the findings of Palmer and colleagues, who investigated the metastasis of the human epidermis carcinoma cell line HEp3 to the liver and lungs of chicks[48]. Another study from the Liverpool Ocular Oncology Group (LOORG) used a chick embryo tumor model and reported the metastasis of uveal melanoma cells to their metastatic sites, which are the liver and eyes[49, 50].
The clinical relevance of this study relies on utilizing a simple, low-cost in vivo model comprising patient-derived xenografts and cell lines to evaluate and develop new therapeutics from the bench to the clinic. There are certain limitations with the CE model, and more investigations are required to determine how cells adjust to the avian microenvironment appropriately. The incubation period of approximately one week may not be optimal for slow-growing tumor cells, and in such cases, it may be challenging to detect evident metastases. We believe that the CE-CAM Rb xenograft model will be advantageous for exploring CSC signaling mechanisms and developing new therapeutic targets for testing against them.