As early as 1911, Rous and Murphy were the first to transplant primary tumor tissue on the highly vascularized chicken embryo CAM, demonstrating the rapid growth of Rous sarcomas early after engrafting. Starting from those findings, the in ovo model gained emerging interest as an alternative for costly, time-consuming mammalian in vivo models in pre-clinical oncological research (1). For centuries a drawback in rodent models, tumor cells and tissues can be engrafted at high efficiencies without species-specific restrictions due to the embryo’s natural immunodeficiency (2, 3). By that oncogenesis can be studied in a humanized system with widespread adoption. Developing tumors perform neo-vascularization and matrix deposition, mimicking the complex tumor microenvironment largely limited in conventional in vitro models (4). The extraembryonic membranes are connected to the embryo through a continuous circulatory system, readily accessible for manipulation and versatile visualization techniques to study tumor growth, metastasis and angiogenesis in and ex ovo, deep mechanistic insights at cellular and molecular levels can be achieved after tumor excision. Ethical approval is omitted if experiments are terminated at embryonic day 14 in most countries, facilitating screenings of pharmacological or physics-based therapies with high reproducibility at large scales supporting the 3Rs principle (5).
In spite of a valuable history in pre-clinical oncology, the applicability of the tumor chorioallantoic membrane (TUM-CAM) model in immunoncological research is largely underexplored. Only few studies have taken advantage of a naturally immunodeficient host to study tumor-immune interactions without species-specific restrictions in ovo. Rodent models have yielded fundamental insights into key aspects of the human immune system, including its dual role in elimination and guarding of malignant cells. As such, the discovery that culture supernatants of activated T cells can boost the reactivity of previously generated cytotoxic T lymphocytes gave rise to the first, IL-2 based, immunotherapy already 30 years ago (6–8) and new waves of excitement are dedicated to recent advances in antibody and cell based therapy approaches. On the other hand, it is undisputable that cancer cells are excellent maskers, hijacking and coopting immune and stromal cells residing in their microenvironment to aid in tumor progression and metastasis. Tumor-immune cell interactions evolve in a highly dynamic, heterogenic environment, and mammalian models remain state-of-the-art reflecting the patient’s situation. However, establishment of clinically relevant, cost and time efficient 3D models for basic and translational immunooncological research at larger scales is urgently needed. The TUM-CAM model is ideal to bridge this gap.
The present review aims to summarize, highlight and emphasize opportunities and limitations of the chicken embryo model for its use in pre-clinical (immuno-)oncological research. Moreover, its potential for tailored treatments based on patient-derived xenografts (PDX) in the context of personalized medicine is discussed.
The in ovo model in cancer research
Table 1 Advantages and limitations of methods that have been established in ovo.
General experimental procedure and ethical guidelines
Several techniques have been established to engraft tumor cells and tissues on the vascularized CAM, with main differences between in ovo and shell-less ex ovo approaches. The latter offers better accessibility of the CAM but increases drop-out rates due to frequent rupture of the yolk membrane and contamination, requiring high levels of experience (36, 37). With respect to its prevalent use in cancer research, the present review provides a step-by-step protocol of the general experimental procedure in ovo.
Table 2 List of cell lines that have been engrafted in ovo.
Figure 1 General experimental procedure of the tumor chorioallantoic membrane model.
Monitoring of tumor growth, invasion, and metastasis
Figure 2 Intravital monitoring of angiogenesis, tumor growth, and assessment of tumor weight (11, 13).
Tumor angiogenesis and vascular remodeling
Expanding tumors hijack physiological angiogenesis to deliver adequate oxygen and nutrient supply, enable waste disposal, and facilitate dissemination of cancer cells to distant sides. During the avascular phase early in tumor progression, the tumor size is largely limited to 1-2 mm. Nutrient deprivation can trigger an ‘‘angiogenic switch’’, resulting in vascular branching and proliferation of endothelial cells, enabling the tumor to grow beyond a restricted size by ensuring sustained energy supply (68). Deregulation of angiogenesis as a hallmark of cancer (69) plays a major role in disease progression, and inhibition of tumor angiogenesis was introduced as a therapeutic strategy more than 50 years ago (70). In 2003, a clinical trial demonstrated prolonged survival of colorectal cancer patients in combination regimes with humanized neutralizing antibodies (bevacizumab, approved by FDA 2005) targeting vascular endothelial growth factor (VEGF), providing proof-of-concept of the successful use of anti-angiogenic therapies in oncology. Likewise, several tyrosine kinase inhibitors designed to target pro-angiogenetic signaling (e.g., sunitinib, approved by FDA 2006) are applied in the treatment of gastrointestinal neoplasms, renal cell carcinoma, and glioblastoma at the front line setting. As a major drawback, hypoxia in the context of aberrant perfusion has been linked to increased resistance to conventional chemotherapeutics, radiotherapy, and immunotherapy, raising the need to unravel further signaling pathways and molecules involved in vascular remodeling and characteristics of vascular structure in the tumor microenvironment.
Investigating the hallmarks of cancer at cellular and molecular levels
Evaluation of tumor growth, metastasis, and vascular remodeling in ovo can give important implications for the therapeutic efficacy and safety of novel agents or the influence of future targets based on knock-in/knock-out models. Notwithstanding, once the tumor is excised, an unlimited field of feasible downstream assays opens up (Figure 3). By that, detailed characterization of classical hallmarks of cancer can be achieved to provide mechanistic insights underlying oncogenesis or therapeutic effects (Figure 4).
Figure 3 Downstream analysis of excised in ovo-grown tumors at macroscopic, cellular, and molecular levels.
Figure 4 Addressing the hallmarks of cancer in ovo.
Excised tumors can be embedded for further histological, immunohistochemical, or –fluorescent staining. By that, the architecture of the tumor microenvironment can be evaluated using conventional HE stainings (27), endothelial surface markers for evaluation of tumor perfusion (view chapter 2.3), or selective staining of chick and xenograft cells to evaluate their spatial distribution within the tumor (9). Staining of cleaved caspase 3 or TdT-mediated dUTP-biotin nick end labeling (TUNEL) (29) has been used to evaluate cell death in tissues combined with Ki-67 as a biomarker reflecting the proliferative state (15, 47). Various surface markers of interest, including integrins related to EMT (view chapter 2.2) (26), the immunogenicity of cell death, myeloid (28), and classical checkpoints (75) (view chapter 3.3) as well as diagnostic (9, 15, 35, 45) and predictive biomarkers (14, 31, 73) can be assessed. Tumor dissociation using adequate dissociation kits can be used to validate previous results at single cell levels using flow cytometry after intra- and extracellular staining with high throughput, including intracellular ROS levels, translocation, and phosphorylation of transcription factors, and an unlimited range of surface markers (13). Characterization of the TME can moreover be achieved using ELISA or bead-based multiplex assays for quantification of cytokine, chemokine, and growth factor release (26). As a consequence of aggressive growth, tumor cells exhibit a deregulated metabolism with shifts in carbon consumption characterized by high glycolytic activity, referred to as the Warburg effect (76). In the context of metabolic reprogramming during embryonal development, adaptions characteristic for aerobic glycolysis in cancer have been investigated by uptake measurements of fluorescent glucose analogs, such as 2-NBDG, basal oxygen consumption rates, and extracellular acidification rates using Seahorse technology, or conventional lactate and glucose uptake assay kits in ovo recently (77). Several studies have further focused on metabolic profiling via mass spectrometry (33, 34) or positron emission tomography (PET) imaging for evaluation of glucose metabolism and protein synthesis (53). At the simplest, the metabolic activity can be assessed by conventional colorimetric or fluorescence-based methods, including resazurin, MTT, and WST assays before or after tumor dissociation (10). While non-malignant cells exhibit a limited number of divisions, cancer cells bypass this barrier by hijacking telomerases to extend the length of their telomeres. Replicative immortality can be evaluated in ovo by sub-culturing of excised tumors after dissociation in the context of microtissues, spheroids, or conventional colony formation assays (78). Molecular insights into the mechanistic action of novel therapeutic agents, oncogenesis, and tumor progression have been achieved using western blot, PCR, and transcriptomics (79). Antibody panels with high specificity for chicken tissues based on the complete characterization of the chick genome enable to investigate interactions between xenograft and chicken tissues and help to distinguish between both.
Aspects in cancer immunology
Experimental studies in animal models, rodents in particular, have yielded fundamental insights into key aspects of the development and regulation of the human hematopoietic and immune systems. However, as 65 million years of evolution may suggest, many aspects of mammalian biological systems, particularly their immune systems, display distinct differences (80). As therapeutic approaches become ever more sophisticated and specifically targeted, it becomes increasingly important to address the limitations of extrapolating pre-clinical discoveries to humans using animal models that more closely recapitulate human biological systems. Since the early 2000s, key advances have been made in the development of immunodeficient mice for generating humanized mice based on the mutant IL2rγ gene introduced in non-obese diabetic (NOD)/severe combined immunodeficiency (SCID) (81) and RAG1/2null mice (82). Xenografting of human primary hematopoietic cells and tissues generating a functional human immune system became feasible, opening up novel avenues in basic and translational preclinical research, permitting insights into cause and cure of human diseases, including cancer immunity. As a major drawback, rodent models, and humanized mice in particular, are accompanied by ethical, cost, and efficiency constraints, limiting widespread adoption and pre-clinical investigation at larger scales. As a naturally immunodeficient host, the chick embryo model has served as an ideal preclinical alternative to rodent tumor models for almost a century. Engrafting of human immortalized cell lines and patient-derived xenografts already featuring a complex and unique microenvironment has successfully been applied. However, few studies have taken advantage of a naturally immunodeficient host to study tumor-immune responses without species-specific restrictions in ovo.
Development of the avian immune system
Primary lymphoid organs include the thymus and bursa of Fabricius, which are colonized by hematopoietic stem cells to become immunologically competent T and B cells before re-entering the circulation and colonizing peripheral lymphoid organs, broadly similar to mammalian immune systems. Early lymphoid cells deriving from the yolk sac and spleen are present in the thymus starting from ED8 and the bursa of Fabricius on ED11. Like mammals, avian T cells recognize MHC presented antigens via the heterodimeric T cell receptor (TCR), subdivided in αβ and γδ TCRs. Precursor hematopoietic cells enter the thymus in three waves, until TCR-γδ + and TCR-αβ1+ mature T cells migrate to the spleen by ED15 and ED19, respectively (96). Mature B cells leave the bursa of Fabricius only post-hatch (97). Generation of the avian antibody repertoire relies on somatic gene conversion, a process taking place during bursal development, as chickens only have a single copy of functional variable (V) and joining (J) segments for both chains of immunoglobulins (95, 98, 99), representing a major difference compared to mammals.
Although the avian immune system can respond to tumor cells by infiltration of monocytes and inflammatory-like cells such as avian heterophils, it is incapable of mounting an immune response before ED18. A non-specific inflammatory reaction has been reported if the experiment extends beyond ED15 but is dampened if the xenografts are implanted early during development when the avian immune system is still immature (100). Notwithstanding, due to ethical restrictions, in ovo experiments are widely terminated before cell-mediated immunity occurs.
Addressing cancer immunity in ovo
The history of immune-oncological research in ovo is surprisingly short. Until now, two comprehensive studies present in the literature addressed tumor-immune cell interactions related to immune contexture in pancreatic ductal carcinoma (43) and reactive oxygen species-based therapy approaches in ovo (101) (Table 3). Other studies focused on the role of immune cells in inflammation-induced angiogenesis, partially translating results to the angiogenic switch in cancer (view chapter 2.3).
Table 3 Studies that focused on tumor-immune cell interactions in ovo.
In the course of studying plasticity of tumor-associated macrophages, Khabipov and colleagues were the first to engraft immortalized RAW264.7 murine macrophages in a coculture model of pancreatic cancer in ovo. In a hydrogel onplant, murine PDA6606 pancreatic cancer cells were inoculated either with non-stimulated (naïve, M0) or pre-stimulated (polarized; M2) RAW264.7 macrophages at a 1:1 effector-target ratio, with 1 mio. cells, respectively. Pre-stimulated macrophages were generated by exposing naïve RAW264.7 cells to pre-conditioned PDA6606 supernatants 1:2 in fresh medium for 72 h. Interestingly, MRI measurements revealed pre-stimulated macrophages to increase PDA6606 tumor growth compared to monoculture tumors, while naïve did not. The authors hypothesized that the short time span (72 h) limited M2 polarization in naïve macrophages during tumor development (view chapter 3.3). Flow-cytometric analysis of dissociated PDA6606:RAW264.7(pre-stimulated) co-culture tumors confirmed the M2-phenotype of the latter, while tissue sections emphasized their role in promoting angiogenesis and matrix remodeling in pancreatic cancer (43). The second study focused on redox-based effects of oxidant-enriched carrier solutions for adjuvant peritoneal lavage in the context of peritoneal carcinomatosis. In vitro, the approach increased immunogenicity and uptake of three human carcinoma cell lines by monocyte-derived dendritic cells. The authors used the in ovo model as a screening platform to validate the therapeutic efficacy observed in vitro in a physiologically more complex model. Here, human monocyte-derived dendritic cells isolated from healthy donors were engrafted 1:1 with the immortalized tumor cell lines and grown on the CAM for 7 days. Reduction in tumor burden correlated well with results obtained in a model of peritoneal carcinomatosis in mice, emphasizing the translational relevance of the chicken embryo model for pre-clinical studies (101). A widely different approach was employed by Wang and colleagues, who took advantage of homologies in the PD1/PDL1 axis between chickens and humans and proposed the TUM-CAM model as an alternative for immunooncological drug development. Clinically approved checkpoint inhibitors mitigated tumor growth in ovo and partially restored T cell-mediated tumor toxicity of chicken lymphocytes in vitro (75). However, appropriate controls and quantification of crucial mechanistic insights, including checkpoint antibody binding assays, were largely missing, questioning the scientific relevance of such findings. Furthermore, tumors were excised at ED18, only shortly after mature T cells could be detected in the avian circulatory system (view chapter 3.2).
Obstacles or chance?
Figure 5 Applicability and limitations of the in ovo model in immuno-oncological research.
Overall, extensive research is needed to clarify if the in ovo model represents a faithful avatar for evaluating immunotherapies in preclinical oncological research. Nonetheless, considering previous research in humanized in ovo models of inflammation-induced angiogenesis and latest reports studying tumor-immune cell interactions in ovo, it is conceivable that fertilized chicken embryos could represent a comprehensive research model in immunooncological research in the future.
A major advantage relies in the CAM’s easy accessibility and low rejection rate to facilitate increasing the model’s complexity by engrafting 3D spheroids (118, 119) or multicellular organoids. For instance, mature organoids derived from human-induced pluripotent stem cells (PSCs) rapidly connected to the vascular network of the chick embryo after transferring them on the CAM (120), and PSC–derived inner ear and kidney organoids demonstrated the model’s potential to optimize developmental maturity and functionality of organoids based on vascularization (121, 122). Along similar lines, grafting of tumor organoids, also combined with immune cells and/or fibroblasts, reflecting tissue heterogeneity in cancer to a greater extent, can be employed (123).
Figure 6 Cells and tissues that can be engrafted in ovo.
Needless to say, the chicken embryo model can clearly not be suggested as a complete replacement for conventional pre-clinical animal models. Despite other advantages, humanized mice models aim, e.g., for genetic modifications of HLA expression to ensure appropriate antigen presentation in peripheral tissues or transgenic expression of human cytokines to improve development and function of transplanted cells at large time scales, which widely outperforms the limits of studying oncogenesis and cancer immunity in ovo. On the contrary, the well-vascularized CAM provides high efficiency of tissue grafting, bridging the gap between in vitro and complex but costly mammalian in vivo models while supporting the 3Rs guidelines. With some major limitations and caveats to keep in mind, the humanized chicken chorioallantoic tumor model could serve as an alternative for pre-clinical immunooncological drug screenings and basic immunological research. It provides a quick, reproducible and effective evaluation of different therapeutic options and has, based on PDX, the potential for development of tailored treatments, including personalized immunotherapy (131).
SB designed the review; LM and JB designed the figures; LM and JB performed experiments; all authors wrote the manuscript draft and reviewed the draft. All authors contributed to the article and approved the submitted version.
Funding was received by the German Federal Ministry of Education and Research (BMBF, grant numbers to SB: 03Z22DN11 and 03Z22Di1).
Figure design was supported by the biorender.com platform.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.