CM is the second most common ocular surface malignancy representing melanocytic lesions and occur specifically in the conjunctiva (1-3). CM arises from melanocytes. While CM arises from melanocytes, CM can form from three different precursors: primary acquired melanosis (PAM) in 74% of cases, nevus in 7%, and 19% arise spontaneously de novo (4). The prevalence is 0.5–1.0 cases/million in the US and Europe, with an increasing trend in recent years (5-7). The current treatment of choice is surgical excision in combination with adjuvant radio-, cryo-, chemo- and immunotherapies, with recurrence occurring in 30–60% of patients, leading to metastasis in approximately 12–15% of cases and not infrequently fatal (8-10). Metastases in general befall predominantly the adjacent lymph nodes, which suggests transmission of metastases via lymphangiogenic route (11-14). Treatment options for metastatic CM are very limited; therefore, a better understanding of the mechanisms is essential to develop novel successful therapeutic approaches. All melanomas arise originally from melanocytes, sharing the same origin but often have different disease progression (15,16). At genetic and clinical level, many CMs share similarities with cutaneous melanomas that include genetic aberrations, metastatic behavior, and clinical course (2,17-19). Mutations shared between cutaneous and CM include BRAF (V600E), NRAS, and NF-1 mutations (19-24). Often, the mutations involve cytosine-to-thymine transitions, which may indicate damage from UV exposure, once again showing parallels to cutaneous melanoma (25). Although there are new promising therapies in development, e.g., the application of systemic BRAF/MEK inhibitors (26,27), there is still a tremendous lack of knowledge and reliable preclinical testing possibilities.
UM is the most common melanoma entity after cutaneous melanoma and account for approximately 5% of all melanoma (28-31). The incidence is approximately 5.1/1,000,000 cases in the Caucasian population, making them the most common intraocular tumors in adults (32-34). UMs most commonly manifest in the choroid, but could also affect the iris and/or the ciliary body (35-38). Risk factors for the development of UMs are light skin, blue iris color, nevi in the choroid/iris or also skin (39,40). In addition to conventional pathological features such as tumor size and location, the clinical prognosis is influenced by genetic factors, such as the presence of monosomy 3 (41) or gain on chromosome 8q (42), which generally correlate with a poorer prognosis. Also unfavorable are mutations in genes such as GNA11/GNAQ (43-46), BAP1 (41,47,48), SF3B1 (45,49), and EIF1AX (45). Although uveal and CMs both originate in ocular melanocytes, the clinical pathology and genetic features are very different (33,50). Therefore, the molecular understanding and therapeutic options of CMs cannot be easily transferred to uveal melanomas.
Therapeutic options range from transscleral resection, enucleation and radiotherapy, with distinctions between plaque brachytherapy with plaques loaded with iodine-125, ruthenium-106, palladium-103, cobalt-60, and radiotherapy with proton beam therapy, helium ion therapy or a stereotactic radiosurgery such as Cyber Knife, Gamma Knife or linear accelerator (51-56). None of the listed therapies, however, provides an evidence-based treatment option for metastatic uveal melanoma. Approximately 25% of all UMs develop metastases at 5 years and 34% at 10 years (53) after local treatment, which predominantly settle to the liver via the hematogenous route (57-59), which is one of the main distinction between CM and UM, as CM predominantly metastasizes via lymphatic routes. Long-term survival in metastatic melanoma has remained very low and unaffected by therapy.
Retinoblastomas are the most common tumors of the eye in childhood and occur in about 1 in 15,000–18,000 births in European countries (60). Retinoblastomas are almost in all cases caused by biallelic loss of Rb1 and develop after additional genetic alterations (61,62).
Retinoblastomas have a very good chance of cure if treated in time. In cases with late discovery of the tumor, which still occurs in developmental countries an appropriate treatment cannot be initiated. Subsequently the vision may be enormously affected and distant metastases may occur, leading to a fatal diagnosis. Therapy options range from classic chemotherapy alternative treatment options like cryo- and thermotherapy, plaque radiation, external beam radiation.
Even if retinoblastoma is one of the most well studied tumor entities, there are still a lot unanswered questions about the origin of the tumor making it difficult to find new development approaches in the design of novel drugs. Late detection at advanced stages often makes enucleation unavoidable, decreasing the life quality of young patients making it a top priority to be able to reproduce the disease using different models and be able to develop more efficient and non-invasive therapy options.
Preclinical models are a top priority to gain a better understanding of tumorigenesis and key mechanisms in tumor development, as well as for testing therapeutic approaches (63). In vitro testing with cell lines and primary tumor cells are popular in this respect. These are easy to perform, inexpensive and provide rapid results. In the case of cell lines, the resources are self-replicating and are well suited for a variety of tests, such as new drugs. On the other hand, pure two-dimensional cell cultures cannot reflect the heterogeneity of tumors and do not map the tumor microenvironment and its effects. Furthermore, it is not always easy to establish new cell lines, especially for rare tumors that are not frequently diagnosed. In this case, an improvement could be achieved by using organoids and 3D cell cultures, where the tumor tissue could be better remodeled (64-71). Clear advantage is the use of co-cultivation, as well as higher heterogeneity and cell-environment interaction. At this point, however, these models are still in their infancy, so further optimization is also needed and the tumor cannot be completely replicated and is not yet possible for every tumor type. In addition, optimization and establishment is more expensive and time-consuming compared to two-dimensional cell culture.
Still, there is no way to avoid in vivo testing in preclinical research. In this case, a distinction is made between different forms of applications, as well as between different host animals. There are many studies with non-mammalian animals, such as zebrafish embryos and chicken embryos, which are relatively easy to use and inexpensive and allow a high-test throughput, making them ideal for testing agents (72-81). Of course, biologically the differences to mammals are immense here, so that these systems are not always ideally suited. Xenografts using mammals, such as mice, remain quite popular as they are easy to establish and can provide rapid results using human cells (74,82-91). However, a disadvantage is the use of immunodeficient animals and the associated loss of microenvironment and tissue specificity in the tumor, as well as lack of heterogeneity when cell lines are used. Instead of cell lines, however, so-called patient-derived xenografts (PDX) can be used, whereby a tissue section can be utilized in order to map heterogeneity and epigenetic factors.
Another method is the use of syngeneic models, in which case species-derived cell lines are used to induce a tumor in the host animal. In this case, effects of the immune system and tumor microenvironment can be considered. However, there are not enough suitable murine or other syngeneic cell lines for every type of cancer available at date (92).
Through biotechnological progress in the field of gene engineering, genetically engineered models (GEM) are becoming increasingly common. Through the targeted regulation of oncogenes it is possible to study tumorigenesis and to investigate the effects with a fully functional immune system and microenvironment. However, these models are still relatively complex to handle, especially in multigenic diseases. Targeted spontaneous tumors are not available for every species and there are also discrepancies to actual disease in humans (63,93-96).
It is important to evaluate in advance which type of model is best suited for one’s research, depending on what is to be studied. In this narrative review, we discuss the most current and promising models for three types of tumors in ophthalmic oncology: UM, CM, and retinoblastoma. We present the article in accordance with the Narrative Review reporting checklist (available at https://aes.amegroups.com/article/view/10.21037/aes-21-39/rc).
Using electronic bibliographic databases, PubMed, Embase, Web of Science and Google Scholar were searched for the following keywords with different combinations: “ocular melanoma”, “retinoblastoma”, “uveal melanoma”, “conjunctival melanoma”, “animal models”, “preclinical testing”, “syngeneic”, “xenografts”, “transgenic mice”, “in vitro”, “in vivo”, “modeling” and “preclinical research”. Searches were limited to English and German studies until May 30th, 2021.
Syngeneic models are very useful when investigating immune responses in addition to tumorigenesis in experiments designed for this purpose. A common model in CM involves the use of cutaneous melanoma cell lines, most commonly using the B16 murine melanoma cell line derived from a spontaneously occurring melanoma in a C57Bl/6J mouse (97-99). In this process, to induce intraocular tumors, the cells previously cultured in vitro are introduced in murine conjunctiva by microinjection, in most cases resulting in solid tumors after a few weeks. In order to develop a more invasive model, some alterations were made by serial passaging, resulting in a more invasive B16LS9 subline (100), which actually showed occurrence of liver metastases after injections. Another approach for a syngeneic model with focus on metastasis used C57BL/6N-derived HGF-Cdk4R24C melanoma cells (92), which also lead to solid tumors and metastases, due to the impairment of the cell cycle by Cdk4R24C mutant and allowing maintenance of melanocytes in interfollicular epidermis through HGF overexpression.
Certainly, there were also numerous studies apart from mice, using other animals, such as the application of Greene melanoma cells (hamster origin) in rabbits (101,102). Rabbit models in eye diseases have the advantage that compared to mice, rabbits have larger eyes, making application and monitoring much easier. Recently, this approach is hardly used in basic research, but much more for testing treatment options, since rapid tumor growth and the absence of metastases, compromised the model (103).
However, syngeneic models have some advantages, providing the perfect basis for studying angiogenesis and metastasis, as well as immune responses and, consequently, a reasonably reliable assessment of treatment strategies. Unfortunately, there are no syngeneic CM cell lines, but only murine cutaneous melanoma cell lines available at date (92).
Besides syngeneic model, the use of xenografts in preclinical studies is one of the tools of choice. Human tumor cell lines are cultured in vitro and subsequently injected in conjunctiva of immune suppressed hosts, including mice, rabbits and zebrafish. These models are mainly used for drug screening, different therapeutic options and tumor growth in general (92,97,104-106). Permanent human cell lines have the advantage that they are already characterized immunohistologically and genetically, so that biological and pharmaceutical effects can thus be better viewed in context. Unfortunately, relatively few established cell lines are available (Table 1) from CM cells (107-110), so that the heterogeneity of a tumor population can only be mapped to a limited extent, due to testing variety limitations.
| Experimental model | Advantages | Limitations |
|---|---|---|
| Syngeneic mice models | ? Immune-competent hosts | ? In CM and UM no murine cell lines available at date |
| ? Ideal conditions for investigations on TME and immune system | ? Many differences in tumor biology between mice and humans | |
| Xenograft model | ? Ideal for high throughput testing of novel chemotherapeutic agents | ? Purchase and maintenance of immunodeficient animals is often very cost-intensive |
| ? Metastasis models possible | ? Effects of immune system and TME interaction are disregarded | |
| ? Many different hosts available—mice, chick embryos, zebrafish | ? In the case of non-mammals very high differences in biology | |
| Genetically engineered models | ? Optimal for studies on cancerogenesis and fundamental research | ? Generation very time and cost-intensive |
| ? Immunocompetent hosts | ? One or several genes mostly not sufficient for exact representation of the disease | |
| ? Metastatic models possible | ? Manipulation of genes often leads to multiple tumors | |
| 3D culturing | ? Time-saving alternative - relatively fast establishment | ? Long-term studies in terms of tumorigenesis and relapses not possible |
| In vitro testing | ? No living organisms needed | ? Not all cell types spontaneously form spheroids |
| Stem cell culture | ? Scientific progress enables simulation of real tumors in cell culture-making testing of compounds and therapy options easier | ? Despite diverse cell cultures with different cell types no complete representation of a TME so far |
| In silico | ? Fast and inexpensive method for pre-selection of chemotherapeutic candidates | ? Calculation of binding affinities alone not indicative for actual effect—efficacy must always be additionally evaluated in vitro or in vivo |
| ? Prediction of molecular docking, protein-protein interactions and pathway analyses | ? For a better prediction, an expansion of biobanks would be necessary to complement pathways and to be able to include interactions and side effects |
CM, conjunctival melanoma; UM, uveal melanoma; TME, tumor microenvironment.
Another possibility for extensive investigation of the efficacy of new therapeutics is the use of so-called PDX. In this case, instead of cells from a cell line, biopsies from patients are transplanted into a model animal and subsequently investigated in further approaches. This offers the advantage that more consideration can be given to heterogeneity of a tumor population and also provides an opportunity for personalized medicine (89). To date, there are no current studies in CM with PDX but some with cutaneous melanoma that have been able to provide not only opportunities in drug discovery but also basic insights into metabolism and metastatic behavior of melanoma cells (84,111).
In addition to animal testing, there are also in vitro alternatives in preclinical testing. For example, there is an interesting study by Fiorentzis and colleagues from 2020 that uses 3D cell cultures with CRMM1 and CRMM2 cell lines to test an approach with electrochemotherapy (69,70). In contrast to standard cell cultures with cells grown in a two-dimensional environment, 3D cultures are much better able to reproduce the spatial complexity of tumor tissue and mimic the tumor microenvironment (68). Thus, clearly representative results on the efficacy of e.g., new compounds can be obtained. 3D spheroids consists of aggregates of tumor cells that provide more natural conditions in terms of metabolism and oxygen distribution than 2D cell cultures (112). They can be evaluated using different assays on tumor growth while testing therapeutic application and furthermore bear the possibility to be transplanted as a PDX in xenograft animal studies. Apart from spheroids, there are also carrier substrates such as Matrigel (113) widely used in in vitro preclinical testing investigating the migration and invasion potential of melanoma cells.
Certainly, any of these models have to be viewed with caution, as they only partially represent the systems, cells and tissues of human organisms. Nevertheless, they are an important tool for understanding basic biochemical processes of tumor biology, such as proliferation, expression of angiogenic factors, intra- and extravasation and migration.
Another increasingly important method is the use of in silico predictions. In silico modeling can be used to accurately test binding affinities and efficacies of chemical compounds, as well as predict protein-protein interactions. Thereby, costly testing can be reduced to truly promising therapeutic candidates and is an interesting possibility for high-throughput screening of drug libraries in the future (114). Currently, there are several studies in cutaneous melanoma, e.g., on the efficacies of BRAF inhibitors but also on proteomic profiles in tumors (115-117), implying that this techniques could also be applicable for CM.
Similar as in studies with CM, cutaneous melanoma cells have been used for syngeneic models of UMs for decades. For this purpose, cell lines derived from different animal species are used, such as Greene melanoma cell lines in rabbits (101,103,118) and B16 melanoma cells in mice (100). Even though the cutaneous cell lines are not derived from the choroid and accordingly have different properties, these models are still suitable for studying the intraocular growth of melanoma cells, and many of these models, actually lead to metastasis in the liver. Thus, the metastatic process including intra- and extravasation and growth in other organs can be monitored. Besides studying metastatic behavior, it is obviously very advantageous that the experiments take place in immunocompetent animals, to date remaining the greatest strength of these models.
As with CM, there are many murine models of UM most commonly involve inoculation of C57BL/6 mice with the B16LS9 cell line, a derivative of the B16 skin melanoma line. Serial passaging induced the metastatic potential of this cell line to form hepatic metastases, which has led to valuable insights into the biology of metastatic melanoma (43,46,100).
Although the syngeneic models are very useful, they are not suitable for UM. Since the used melanoma cells are from cutaneous origin, the molecular drivers differ from drivers in UM. There were efforts made to compensate this by introducing canonical UM mutations in some studies. For this purpose, e.g., immortalized melanocytes were transduced with specific mutations like GNAQQ209L, actually leading to solid tumors and metastasis as well (119).
The desirable outcome would be the use of murine UM cell lines, so eventually in the future it will be possible to establish a stable murine cell line from transgenic animals, allowing the investigation of interactions of an actual UM in an immunocompetent host.
Just as in cutaneous and CM, xenografts are also being used in UM. Commonly, permanent UM cell lines are used, whose genetic and histological profile is already known (88,120-124). The great advantage of human xenografts is that cells derived directly from patients display characteristics of UMs at molecular level. The models are therefore ideal for testing new drugs and for screening intraocular tumor growth. Hence, the biological and pharmacological aspects can be studied in vivo with a view to interaction with new compounds identifying potential candidates for clinical studies. The selection of cell lines is of utmost importance as very few UM cell lines are available, as some cell lines turned out to be of cutaneous origin, by exhibiting e.g., BRAF mutations (125), which usually are not present in UM. Furthermore, some could be identified as identical cell lines by short tandem repeat (STR) analysis, which further limits the choice of reliable cell lines (126). Orthotopic mouse models of UM basically result in inoculation to the iris, ciliary body, or choroid. Suprachoroidal injection models have been described (98) and rapidly demonstrated tumors in the ciliary body and choroid. There are also methods to perform the injection intravitreally. Although in humans UMs do not arise in the vitreous body, animal models showed similar invasive behavior to human melanoma, making it well suited for preclinical testing (127,128)
A major disadvantage of these models is that even if the characteristics of the cell lines are well described to date, cell lines yet evolve through frequent passaging and become increasingly distant from the tumor of origin. In this way, results obtained in animals are not readily translationally applicable with regard to original tumors (71,120,129).
PDX models are a suitable solution and are becoming increasingly popular in cancer research. This often involves implanting tumor samples into mice, with resulting in solid tumors in nearly one-third of cases. Often the studies are using severe-combined immunodeficient (SCID) (63,87,90,91,120,122,130) and next-generation sequencing (NGS) (87,131) mice.
This method enables producing heterogeneous tumors that share the same molecular and genetic abnormalities as tumors in patients, making them particularly good models for testing combination therapies.
Xenotransplantation is primarily performed in immunodeficient mice and less frequently in rabbits (118,132). Nevertheless, in both cases, these are very costly and time-consuming variants. For this reason, approaches that allow high-throughput testing of entire compound libraries in UMs are becoming increasingly common.
Another suitable model for preclinical screening is zebrafish, due to its low maintenance cost and ease of manipulation of zebrafish embryos, as the adaptive immune system of the animals is not formed until 4 weeks after fertilization. In addition, there are similarities between zebrafish and human tumors at the histopathological level, as well as there is the presence of tissue-specific transgenic zebrafish lines that facilitate imaging. In general, melanoma cells are implanted by injection into the embryos and then growth and migration are identified using various imaging methods (63,74,75,80,133). The studies showed that the zebrafish xenograft model is useful for preclinical testing of a variety of compounds and has the advantage over mouse models in terms of cost-efficiency and time-saving.
Another method for preclinical testing is the use of chicken embryos in so called chorioallantoic membrane (CAM) assays. UM cells are applied to the CAM on tenth day after fertilization and subsequently tumor growth, angiogenesis and metastasis can be observed and analyzed in this model. The immune advantage of inducing tumor cells without rejection is due to the lack of an immune system in chicken embryos to this point of embryonic development (72). Considering the accessibility and application diversity of this method, it could be as well a very helpful tool in the future.
Like all preclinical models, xenografts also have deficiencies. In addition to the aforementioned use of cell lines, some of which are not translatable to the tumors in patients, the use of immunodeficient hosts is also an extreme disadvantage. With the increasing importance of immunotherapies, it does not seem reasonable to work with models that cannot represent the immune interaction with tumor cells. Although response rates to PD-1 and CTLA4 inhibition have been low in UM (134), there are other aspects of the immune system that play important roles and fundamentally affect angiogenesis, metastasis, and ultimately prognosis. In addition, xenografts, even PDX, are often very expensive and have a low transplantation rate or, as with CAM assays or zebrafish, are not performed in mammals, which also challenges the transferability of these models.
Besides xenografts and syngeneic models, there is another subclass of animal models—GEMs. In this case, GEMs allow the study of autochthonous tumorigenesis coupled with the influence of the immune system (85,93,95)—given that spontaneous tumors arise.
This method, in combination with the use of immunocompetent hosts, has the advantage that signal transduction of tumorigenesis can be studied at genetic level and fundamental understandings of the disease can be gained.
Older models of UM primarily used pigment-specific promoters such as tyrosinases and HRAS, which resulted in not only uveal tumors but also retinal and cutaneous malignancies. Apart from that, these were gene alterations that cannot be observed in UM patients, which limits the clinical applicability immensely (135).
The discovery of the oncogenic drivers GNAQ/GNA11 has been one of the most important contributions in recent years, allowing the development and study of several mouse models using GNAQQ209L transgenic mice. These mice also did not develop UMs initially, but still showed some molecular similarity in the cutaneous lesions, such as activation of the YAP protein (133,136-138).
Another GNAQQ209L model resulted in the formation of UMs within a few months, although intravasation and metastasis were also observed here. However, there were also dermal neoplasms derived from melanocytes in addition to lesions on the choroid. Other models combined, for example, BAP1 deletion and expression of GNAQQ209L, in which, unexpectedly, choroidal melanomas turned out smaller but with overall increased dermal tumor burden (119).
These models allow us to have a more detailed look on tumorigenesis than before, for example, the role of GNAQ and GNA11 as oncogenes could be verified by this kind of modelling (40,119,136,137).
By all means, like other models, GEMs are not without drawbacks. First, the introduction of GNAQQ209L leads to melanocytic neoplasms in other organs leading to undesirable side effects and to a premature termination of the studies without the possibility to sufficiently observe the development of UMs. In addition to this, the time factor also plays a role, since the occurrence of spontaneous lesions certainly is more time-dependent than inoculation in xenografts and syngeneic models. Another factor is the lack of distinction between primary tumors and metastases, as the tumor burden of the transgenic mice is generally very high.
Finally, yet importantly, despite all the advances, the immense differences in the biology of mice and humans are not to be overlooked and always have to be considered in preclinical testing.
In addition to basic preclinical metastasis assays such as migration assays, ring assays, and chemotaxis assays (139), there are a number of approaches to overcome two-dimensional cell cultivation and to establish more realistic methods for the evaluation of preclinical tests in UM. Based on the success of PDX, it has been shown that three-dimensional cultures from tumor samples can grow in mice. Thus bears the possibility to represent the molecular phenotypes and possibly also include the role of tumor microenvironment with fibroblasts and lymphocytes (71,129). There are already successful approaches to cultivate cell culture lines with e.g., added macrophages as well as to cultivate patient cells (66), which result in 3D spheroids and could also be used successfully in first tests. If these 3D cell models become established, this will be a superior tool for testing cell-matrix interactions and tumor microenvironment and probably could be used for in vitro testing as well as act as a xenograft model.
Although the extraction of live cells from retinoblastomas was initially very difficult, some cell lines are now available that allow the application of xenografts. Therefore, retinoblastoma cell lines could be transplanted into immunodeficient animals by microinjection, which also lead to formation of tumors (83,132,140,141).
Among the animals used were rabbits, in which immunohistochemistry was used to demonstrate that the tumors had indeed grown (132,142). Both stronger vascularization and persistent tumor growth could be detected, as well as the presence of necrotic areas and hypoxic conditions, which can also be observed in human retinoblastomas. However, these models were inaccurate since the subretinal space was affected rather than the retina.
Another approach used athymic nude mice including the injection of a primary cell line and freshly derived patient samples after surgery (143). The fresh cells were transplanted into the anterior chamber and grew in the eye, but failed to grow if injected subcutaneously. On the other hand, cells from the Y-79 cell line invaded orbit, brain and optic nerve and showed severe tumor growth.
There are also several models using zebrafish in an orthotopic approach, showing promising results in terms of high throughput screening of drug libraries (76-79,81,144). Like the most xenograft models they are used to test new chemotherapeutic agents and photodynamic therapies and allows a high number of agents to be tested in short time.
Besides the preclinical testing of novel drugs, xenograft models delivered some diagnostic advances in non-invasive imaging, such as micro CT, MRI and fluorescence and bioluminescence imaging (145-147). By labeling cell lines with fluorescent proteins like green fluorescent protein (GFP), it is possible to study the metastatic behavior of retinoblastomas.
However, like all xenograft models, these models have the disadvantage that they only reflect the disease to a limited extent and the tumor microenvironment of tumors cannot be monitored due to immunodeficiency. In addition, there are also differences in the various biological species: for instance, the optimal body temperature for zebrafishes is 28 ℃, while tumor cells show the best growth performance at 37 ℃ degrees, so naturally, the conditions are not limitless comparable (77). This limitation has been addressed by many groups and the Langenau group actually succeeded in performing preclinical experiments with zebrafish at 37 ℃. For this purpose, adult zebrafish of the Casper strain prkdc?/?, il2rgα?/? with weakened immune systems were used, in which human cancer cells could actually be implanted over a period of more than 28 days (73).
In addition to xenografts, there are a number of approaches to knockout models of retinoblastoma. Even though the Rb1 gene is considered an oncopromoter in human retinoblastomas, it was initially insufficient to generate a retinoblastoma in the mouse eye. Only with the identification of p107 in the signal pathways and the knockout combination of both genes was it possible to generate successful transgenic models.
One of the first models is the LH-β T-Ag mouse model, in which the oncogenic unit of the SV40 protein is expressed and can be induced by an LH-β promoter, and is combined with T-Ag-expressing mice (148-151). On histological level, the resulting retinoblastomas were also very similar to human retinoblastoma samples and therefore became a first basis for better understanding of retinoblastoma. The model was primarily used to test local therapies, yet had the disadvantage of using viral oncoproteins, whose impact on tumor and experimental animal is not completely evaluated. Consequently, it was inevitable to develop other transgenic models that could mimic retinoblastoma tumorigenesis in humans.
Although the original model provided some important results on therapeutic testing, the advent of gene knock-out technology (152) first provided important tools for developing new translational retinoblastoma models. The first attempts resulted in the knockout of the Rb1 gene in the retinal hat of mice (153). However, this alone didn’t bring the desirable retinoblastoma. It was not until the discovery of another protein, p107 (154-157) and its role in inhibiting retinoblastoma formation in Rb-deficient mice until multiple knockout mouse lines were developed from it actually getting retinoblastoma tumors. Unfortunately most of the animals died very early in development, since Rb, as well as p107, plays a major role in embryonic development (158). Hence, a Cre-Lox model was generated to facilitate viable mutants and still knock out the genes. This way, it was indeed possible to obtain a group of animals in which retinoblastoma developed (156,157). This model also showed high apoptotic sensitivity and cell death resistance, which is very similar to human retinoblastomas. Unfortunately, the delayed tumorigenesis was unfavorable; in addition, little penetrance and invasiveness could be observed in this model.
Inclusion of an additional mutation of p130 in these models resulted in enhancement of several developmental phenotypes seen with loss of Rb in the retina (157,159,160). This indicated that there is a functional synergy between these family members. The α-Cre Rb/p130 DKO mouse turned out as a suitable model to study advanced retinoblastoma with tumors rapidly progressing and also producing metastases. Rb/p130-DKO retinoblastomas appear similar to Rb/p107-DKO retinoblastomas on histologic examination, and both resemble human retinoblastomas with neuroblastic differentiation. Numerous variations of this murine model exist, also e.g., including p53 mutations as well, while a lot of them are considered genetically and histologically translatable to humans. Obviously, there is always the fact that tumor biology of mice and humans is distinctive and the findings on tumorigenesis can only be transferred to a limited degree.
While many mouse models rely on the deactivation of Rb1 and Rb1l, there are other models that make use of this principle. For example, there is a promising study in Xenopus tropicalis where means CRISPR/Cas9 techniques using a combination of double mutation of Rb1 and Rbl1 resulted in tumors, while knockout of Rb1 or Rbl1 alone did not result in tumor formation in the tadpoles (161).
Many approaches have tested the use of immortalized Rb cell lines for 3D cell culture for biological and preclinical testing, but failed due to self-assembled structures and formation of cell-cell matrices in three-dimensional space.
A successfully tested study instead uses human pluripotent stem cells (hPSCs) that could mimic retinogenesis in vitro. Consequently, human embryonic stem cells with biallelic Rb1 mutation were generated and did grow stepwise into Rb organoids. Through this model, it was indeed possible to identify genetic signatures and successfully test potential therapeutics (162). The use of pluripotent stem cells could therefore become an interesting innovative way to screen retinoblastoma therapies, without the time and cost effort of mice models.
In summary, it is certain that the one perfect preclinical model does not exist for any of the three cancer entities described here. All animal models as well as in vitro testing have their advantages and disadvantages. Crucially, to realize the experimental potential, it is definitely necessary to create more molecular datasets for ocular melanomas in order to create successful therapies and treatment options despite the rarity of the respective diseases. Besides the optimization of the respective preclinical models, it is also a top priority to support the expansion of biological databases in order to consolidate the state of knowledge and to combine all known findings and use them well in future scientific approaches.
