An Orthotopic Model of Uveal Melanoma in Zebrafish Embryo: A Novel Platform for Drug Evaluation

Uveal melanoma is a highly metastatic tumor, representing the most common primary intraocular malignancy in adults. Tumor cell xenografts in zebrafish embryos may provide the opportunity to study in vivo different aspects of the neoplastic disease and its response to therapy. Here, we established an orthotopic model of uveal melanoma in zebrafish by injecting highly metastatic murine B16-BL6 and B16-LS9 melanoma cells, human A375M melanoma cells, and human 92.1 uveal melanoma cells into the eye of zebrafish embryos in the proximity of the developing choroidal vasculature. Immunohistochemical and immunofluorescence analyses showed that melanoma cells proliferate during the first four days after injection and move towards the eye surface. Moreover, bioluminescence analysis of luciferase-expressing human 92.1 uveal melanoma cells allowed the quantitative assessment of the antitumor activity exerted by the canonical chemotherapeutic drugs paclitaxel, panobinostat, and everolimus after their injection into the grafted eye. Altogether, our data demonstrate that the zebrafish embryo eye is a permissive environment for the growth of invasive cutaneous and uveal melanoma cells. In addition, we have established a new luciferase-based in vivo orthotopic model that allows the quantification of human uveal melanoma cells engrafted in the zebrafish embryo eye, and which may represent a suitable tool for the screening of novel drug candidates for uveal melanoma therapy.


Introduction
The zebrafish embryo has been successfully employed as a platform for modeling human diseases and for large-scale screening of new drugs [1][2][3][4]. Ease of manipulation, relatively low costs of maintenance, and optical transparency, combined with the opportunity to perform high-quality imaging, led to an extensive use of this model in cancer research. In this regard, mammalian tumor cell grafting in zebrafish embryos can be achieved in different anatomical sites, giving opportunity to study various aspects of the disease, such as tumor progression, angiogenesis, cancer cell spreading, and metastasis formation. Tumor cells have been successfully implanted in the perivitelline space, yolk ball, blood stream, pericardial cavity, eye, and brain (see [5][6][7][8][9] and references therein).
One of the major drawbacks of the use of the zebrafish embryo as a model in oncology is the quantification of tumor xenograft growth in the different anatomical sites, generally performed by measuring the fluorescence signal generated by engrafted fluorescent tumor cells [10,11]. This approach has also been used for the study of cancer growth following ocular transplantation of fluorescent tumor cells in zebrafish embryos [7][8][9]. However, the presence of the lens and the cup-like structure of the eye make difficult the acquisition of high-quality fluorescence images, which may lead to misleading results. This with increasing concentrations of the different anticancer drugs. After a further 48 h or 72 h incubation, cells were trypsinized and viable cell counting was performed with the MACSQuant ® Analyzer (Miltenyi Biotec, Bergisch Gladabach, Germany), as reported [21].

Apoptosis Assay
92.1-RFP + /luc + cells were seeded on 6-well plates at 1.0 × 10 4 cells/cm 2 . After 24 h, cells were treated with 140 nM paclitaxel, 20 nM panobinostat, and 60 nM everolimus. After 72 h of treatment, apoptotic cell death was assessed by Annexin-V/propidium iodide double staining according to the manufacturer's instructions, and cytofluorimetric analysis was performed using the MACSQuant ® Analyzer.

Zebrafish Maintenance and Cell Transplantation
The transgenic tg(kdrl:EGFP) zebrafish line was maintained in the facility of the University of Brescia at 28 • C under standard conditions [22], and embryos were staged by h post-fertilization (hpf), as described [23]. To prevent pigmentation, embryo fish water was added with 0.2 mM 1-phenyl-2-thiourea (PTU) starting from 24 hpf. For cell injection and in vivo observation, embryos were anesthetized using 0.16 mg/mL tricaine. For cell engrafting, 48 hpf embryos were microinjected in the eye with tumor cells using a borosilicate needle and an Eppendorf FemtoJet microinjector equipped with an InjectMan NI2 manipulator. A single eye was injected with tumor cells in each zebrafish embryo. When indicated, 2.0 nL of a solution containing the anticancer drug under testing was injected in the same eye. After tumor cell injection, zebrafish embryos were selected under a fluorescence microscope to ensure that tumor cells were located only within the eyeballs and then grown at 33 • C.

Fluorescence and Light Sheet Microscopy
Live embryos were photographed at 1 h (t 0 , 48 hpf), 1 day (t 1 ) and 4 days (t 4 ) post implantation on agarose-coated dishes using an AxioZoom V16 fluorescence stereomicroscope (Zeiss, Oberkochen, Germany, EU) equipped with a digital Axiocam 506 color camera (Zeiss). The mean area of the tumor was manually measured using FIJI software [24]. Light sheet microscopy experiments were performed using a Light Sheet Z.1 microscope (Zeiss). For this purpose, t 0 , t 1 , and t 4 embryos were embedded in a low melting agarose cylinder (1% low melting agarose:fish water, 1:1) and immersed in the observation chamber filled with fish water and anesthetic. Maximum intensity projections were obtained using the Zen software (Zeiss) and 3D reconstructions were made after z-stack processing with Arivis software (Zeiss).
To detect apoptotic cells, 48 hpf embryos were microinjected in the eye with 2.0 nL of a solution containing the anticancer drug under testing. After injection, zebrafish embryos were grown at 33 • C for 4 days. At t 4 , live embryos were soaked in fish water containing 2 µg/mL acridine orange and incubated at 28 • C for 20 min. After 8 washes for 5 min each with fish water, embryos were anesthetized and analyzed immediately with a fluorescence stereomicroscope (Zeiss).

Luciferase-Based Quantification Method
At different time points after intraocular grafting of luc + cells, enucleated eyes or anesthetized embryos were singularly placed in a well of a white polystyrene 96-well plate (Sigma-Aldrich). Embryo medium was removed and replaced with 50 µL of lysis buffer (80 mM Na 2 HPO 4 , 9.3 mM NaH 2 PO 4 , 2% TritonX100, 1.0 mM DTT in MilliQ water) and 50 µL of ONE-Glo™ Reagent. The luminescence was measured using an EnSight ® Multimode Plate Reader (PerkinElmer, Milan, Italy) and expressed as relative luminescence units (RLUs).
To generate the calibration curve, a fixed number of B16-LS9-luc + cells (ranging from 0 to 1000 cells) was added to non-injected embryos and then the bioluminescence signal quantified as described above.

Statistical Analysis
Statistical analysis was performed with GraphPad Prism 8 (San Diego, CA, USA) using a Student's t-test for 2 groups of samples or one-way analysis of variance followed by Tukey's multiple comparison post hoc test for more than 2 groups. Differences were considered significant when p-values < 0.05.

Zebrafish Embryo Eye Is a Permissive Environment for the Growth of Engrafted Melanoma Cells
To evaluate whether the zebrafish embryo eye represents a microenvironment suitable for the grafting of melanoma cells, we first assessed the behavior of the well-characterized model of invasive murine melanoma represented by B16-BL6-DsRed + cells [2], which were injected into the eye of zebrafish embryos at 48 hpf. At this stage, the embryo eye consists of the retina (mainly composed of neuronal cells that will progressively organize in stratified retinal layers [26]), the hyaloid, and the ciliary vascular systems [27]. On this basis, B16-BL6-DsRed + cells were orthotopically injected in the posterior side of the developing eye of tg(kdrl:EGFP) embryos (100 cells/embryo) and monitored for the following 4 days by light sheet fluorescence microscopy. One hour after implantation (t 0 ), maximum intensity projection of the z-stacks and 3D reconstructions confirmed that DsRed + cells were present at the bottom of the eye in the proximity of the developing choroidal vasculature ( Figure 1A,A'). One day post implantation (t 1 ), cells relocate towards the eye surface, interacting with the surrounding vasculature ( Figure 1B,B'). At 4 days post implantation (t 4 ), DsRed + cells invaded the lens surface and grew without exerting a significant impact on the anatomical architecture of the eye ( Figure 1C,C'). To confirm these observations, paraffin sections of tumor cell-grafted eyes were analyzed at t 0 and t 4 by H&E staining and by Ki-67 and vimentin immunostaining. As shown in Figure 2, implanted B16-BL6-DsRed + cells were able to proliferate, as demonstrated by the presence of Ki-67 + cells, without affecting the physiological development of the retina. Moreover, cells moved towards the eye surface and invaded the lens ( Figure 2B). Notably, preliminary observations suggest that the displacement of melanoma cells observed at t 4 is in part a consequence of the invasive properties of cancer cells and in part due to the remodeling of the eye that occurs during embryo development, which plays a not negligible role in tumor cell localization within the eye (data not shown).
liminary observations suggest that the displacement of melanoma cells ob part a consequence of the invasive properties of cancer cells and in part du eling of the eye that occurs during embryo development, which plays a no in tumor cell localization within the eye (data not shown).

Quantification of Melanoma Xenograft Growth in the Zebrafish Embryo E
To obtain a reliable and reproducible quantification of melanom zebrafish embryo eyes, we performed a first set of experiments exploiting signal of B16-BL6-DsRed + cells. For this purpose, we attempted to mea tumor areas in engrafted embryos at t0, t1, and t4 after injection. As a though the analysis of digitalized images demonstrated an increase of Ds at t4 when compared to the other time points (Supplementary Material results suffered significant drawbacks. Indeed, although extended depth stacks provided a good quality lateral view of the xenografts at t0 and t1, t

Quantification of Melanoma Xenograft Growth in the Zebrafish Embryo Eye
To obtain a reliable and reproducible quantification of melanoma cell growth in zebrafish embryo eyes, we performed a first set of experiments exploiting the fluorescence signal of B16-BL6-DsRed + cells. For this purpose, we attempted to measure fluorescent tumor areas in engrafted embryos at t 0 , t 1 , and t 4 after injection. As anticipated, even though the analysis of digitalized images demonstrated an increase of DsRed + tumor areas at t 4 when compared to the other time points (Supplementary Materials, Figure S1), the results suffered significant drawbacks. Indeed, although extended depth of focus of the z-stacks provided a good quality lateral view of the xenografts at t 0 and t 1 , the acquisition of images required to cover the entire thickness of the tumor was difficult at t 4 and was affected by the position of tumor cells that were close to the lens or deeply immersed in the eye.
In addition, the three-dimensional structure of the embryo eye and the presence of the lens, which may generate distorted images, made problematic the choice of the best angle for image acquisition. In this context, the optical accessibility of the zebrafish eye is further limited by the presence of pigmented cells, including neural crest-derived chromatophores (i.e., melanophores, iridophores, and xanthophores) and the retinal pigment epithelium [28]. Moreover, the blockade of zebrafish pigmentation by the addition of PTU in the fish water [22] has no effect on iridophores and on their nonspecific fluorescence signal [29], which impairs the reproducibility of the quantification technique.
On the other hand, it has been shown that the use of transparent crystal zebrafish mutants does not completely avoid refraction of the light due to the presence of the lens and of residual xanthophores present in the mutant eyes [29]. Finally, even though high-quality images may be obtained by confocal microscopy [11], acquisition and analysis procedures are time consuming and not suitable for high-throughput analysis.
To overcome these limitations, we developed an alternative quantification method exploiting the bioluminescence signal generated by tumor cells transduced with firefly luciferase. To this end, we took advantage of a red fluorescent and luciferase expressing human melanoma cell line (A375M-RFP + /luc + ) available in our laboratory. A375M-RFP + /luc + cells were grafted in the eye of 48 hpf-old zebrafish embryos at 50, 100, and 200 cells/injection. Then, injected and not injected contralateral eyes were enucleated 1 h after grafting. As shown in the Supplementary Materials, Figure S2, analysis of grafted eyes indicates that the bioluminescence signal increases in a cell dose-dependent manner, being distinct from the basal levels measured in the contralateral control eyes. Similar results were obtained by measuring the bioluminescence signal generated by the lysates of the whole embryos engrafted with A375M-RFP + /luc + cells (data not shown), thus avoiding the technically difficulty and the time-consuming eye enucleation procedure and confirming the reliability of this quantification method.
To assess whether this procedure allowed a quantitative evaluation of the growth of grafted tumors, A375M-RFP + /luc + cells (100 cells/embryo) were injected in the eye of tg(kdrl:EGFP) embryos at 48 hpf. Then, injected embryos were analyzed at t 0 and t 4 by light sheet fluorescence microscopy or by evaluation of the bioluminescence of the lysates of the whole animals. As shown in Figure 3A, A375M-RFP + /luc + cells were clearly visible 1 h after grafting in the embryo eye. At 4 days post implantation, grafted cells were alive and had moved from the injection site toward the lens surface, as already observed for B16-BL6-DsRed + cells. In parallel, a significant increase of the A375M-RFP + /luc + cell-related bioluminescence signal was measured at t 4 when compared to t 0 , thus confirming the capacity of this protocol to monitor the relative growth of tumor grafts ( Figure 3B).  Figure S2, analysis of grafted e indicates that the bioluminescence signal increases in a cell dose-dependent manner, ing distinct from the basal levels measured in the contralateral control eyes. Similar resu were obtained by measuring the bioluminescence signal generated by the lysates of whole embryos engrafted with A375M-RFP + /luc + cells (data not shown), thus avoiding technically difficulty and the time-consuming eye enucleation procedure and confirm the reliability of this quantification method.
To assess whether this procedure allowed a quantitative evaluation of the growth grafted tumors, A375M-RFP + /luc + cells (100 cells/embryo) were injected in the eye tg(kdrl:EGFP) embryos at 48 hpf. Then, injected embryos were analyzed at t0 and t4 by li sheet fluorescence microscopy or by evaluation of the bioluminescence of the lysates the whole animals. As shown in Figure 3A, A375M-RFP + /luc + cells were clearly visible after grafting in the embryo eye. At 4 days post implantation, grafted cells were alive a had moved from the injection site toward the lens surface, as already observed for B BL6-DsRed + cells. In parallel, a significant increase of the A375M-RFP + /luc + cell-related oluminescence signal was measured at t4 when compared to t0, thus confirming the cap ity of this protocol to monitor the relative growth of tumor grafts ( Figure 3B).

Orthotopic Ocular Grafting in the Zebrafish Embryo as a Model for Uveal Melanoma Treatment
Given the promising capacity of luciferase-expressing melanoma cells to grow and to be quantified after grafting in zebrafish eyes, we decided to extend this assay to a wellestablished murine melanoma model suitable for investigating the mechanisms responsible for uveal melanoma liver tropism [30][31][32], immunologic and angiogenic aspects [33], and drug response [34][35][36][37]. On this basis, B16-LS9-luc + cells were injected in the zebrafish embryo eye, grafts were analyzed at t 0 and t 4 , and immunohistochemical analysis of cell grafts showed that tumor cells proliferate, as already observed for B16-BL6 tumors ( Figure 4A). In addition, bioluminescence quantification performed at different time points after injection (t 0 , t 1 , t 2 , t 3 , and t 4 ) showed that, after a slight decrease in cell growth at t 1 , B16-LS9-luc + cells proliferate rapidly, their cell number increasing up to four times at t 3 /t 4 when compared to t 0 ( Figure 4B and Supplementary Materials, Figure S3).  In order to assess the response of tumor cells grafted in the embryo eye to anticancer drugs, preliminary experiments were carried out in which B16-LS9-luc + cells were treated in vitro for 72 h with increasing concentrations of the microtubule-disrupting agent paclitaxel [38]. As shown in Figure 4C, the compound inhibits the growth of B16-LS9-luc + In order to assess the response of tumor cells grafted in the embryo eye to anticancer drugs, preliminary experiments were carried out in which B16-LS9-luc + cells were treated in vitro for 72 h with increasing concentrations of the microtubule-disrupting agent paclitaxel [38]. As shown in Figure 4C, the compound inhibits the growth of B16-LS9-luc + cells with an ID 50 equal to 50 nM. On this basis, three different routes of in vivo administration of the drug were attempted in engrafted zebrafish embryos: (i) 24 h in vitro pretreatment of B16-LS9-luc + cells with 0.5 µM paclitaxel, followed by their injection in the zebrafish eye; (ii) injection of B16-LS9-luc + cells in the embryo eye, followed by incubation of engrafted embryos with 10 µM paclitaxel dissolved in fish water-an experimental procedure frequently used to test compounds in zebrafish [39]; (iii) engraftment of cells in the zebrafish embryo eye, followed by injection of the drug at 0.4 pmoles/embryo in the same eye. At the end of each protocol, the growth of B16-LS9-luc + grafts was assessed by bioluminescence-based quantification of luc + tumor cells performed at t 4 .
As anticipated, pretreatment with paclitaxel resulted in a significant inhibition of the growth of the tumor grafts ( Figure 4D). No inhibition of the growth of B16-LS9-luc + grafts was observed when engrafted embryos were treated with paclitaxel dissolved in the fish water, possibly as a consequence of the limited entry of the drug in the eye compartment ( Figure 4E). Interestingly, a significant inhibition of B16-LS9-luc + tumor growth occurred when paclitaxel was directly injected in the embryo eye after cell grafting ( Figure 4F).
Based on these observations, we decided to extend this experimental model by setting up an orthotopic experimental protocol in which human 92.1-RFP + /luc + uveal melanoma cells were grafted (100 cells/embryo) in zebrafish embryo eyes at 48 hpf, followed by injection in the same eyes with 0.4 pmoles of different canonical chemotherapeutic drugs (i.e., paclitaxel [38], the histone deacetylase inhibitor panobinostat [40], the mTOR inhibitor everolimus [41], or vehicle). As shown in Figure 5A, all drugs inhibited the growth of uveal melanoma 92.1-RFP + /luc + cells in vitro, with ID 50 values ranging between 10 nM and 67 nM. Accordingly, treatment of engrafted uveal melanoma cells by eye injection of paclitaxel, panobinostat, or everolimus caused a significant inhibition of tumor growth when assessed by measurement of bioluminescence ( Figure 5B). Notably, panobinostat exerts a pro-apoptotic effect on cancer cells, both in vitro and in vivo (Supplementary Materials, Figure S4A,B). Moreover, no significant toxic or pro-apoptotic effect was observed in the zebrafish embryo eye tissue after the injection of the three drugs (Supplementary Materials, Figure S4C).
In addition, light sheet fluorescence microscopy confirmed the efficacy of drug treatment, uveal melanoma cells remaining confined at the bottom of the eye in the proximity of the choroidal vasculature ( Figure 5C). These results are in line with previous observations about the efficacy of these drugs on uveal melanoma growth in in vitro and in vivo experimental models [42,43]. Relevant to this point, it must be pointed out that phase 2 clinical trials designed to evaluate the clinical benefits of paclitaxel or everolimus administration showed only a limited efficacy in uveal melanoma metastatic patients [44,45], whereas no data are available about the effect of panobinostat. Further studies will be required to assess the efficacy of histone deacetylase inhibitors in uveal melanoma.
In this paper, we describe the first orthotopic model of uveal melanoma in zebrafish, previous models of uveal melanoma being limited to the injection of cancer cells into the yolk sac of embryos [46][47][48][49]. Even though orthotopic models are usually less rapid and more technically challenging with respect to the heterotopic implants, these approaches are more tissue-specific and allow a more realistic recapitulation of the natural microenvironment in which the tumor originated. Altogether, our data extend previous observations about the possibility of engrafting tumor cells, including retinoblastoma and conjunctival melanoma cells, in zebrafish embryo eyes, thus generating orthotopic models of different ocular neoplasms [7][8][9]. In addition, it should be considered that the eye represents a metastatic site for various tumor types, including cutaneous melanoma, breast, and lung cancer, with choroidal metastases occurring in approximately 8% of human malignancies [50]. Thus, tumor cell grafting in the zebrafish embryo eye may be exploited as a useful orthotopic model to investigate novel therapeutic approaches not only for primary tumors but for eye metastases as well. Relevant to this point, our work focuses on providing a simple and reliable strategy for the accurate quantification of engrafted tumor cells by exploiting the bioluminescent signal of firefly luciferase-expressing cells. Indeed, the presence of the lens and the cup-like structure of the eye make difficult the acquisition of high-quality fluorescent images and may lead to misleading results. Moreover, the autofluorescent properties of zebrafish embryos and mammalian cells increase the nonspecific background and decrease the sensitivity of signal detection. On the other hand, bioluminescence displays a higher detection capacity and allows for greater sensitivity because of the enzymatic nature of the bioluminescent reporter and the absence of the endogenous bioluminescence of cellular components.

Conclusions
In this paper, we described an orthotopic model of uveal melanoma in which cells are grafted in the eye of zebrafish embryos in the proximity of the developin roidal vasculature. In the following 3-4 days, grafted cells proliferate and move to the eye surface, thus demonstrating that the zebrafish embryo eye is a permissiv ronment for the growth of UM cells. In addition, the use of firefly luciferase biolu cent murine and human tumor cells allowed the assessment of the antitumor act candidate drugs when injected into the grafted eyes. In conclusion, we have establ The luminescence-based method herein described allows for a precise quantification without relying on any image analysis software and it provides a simple and quick in vivo evaluation of the efficacy of anticancer drugs after intraocular delivery. In this context, this approach may be exploited for high-throughput analysis and may have relevant implications for the evaluation of new low molecular weight compounds for the treatment of uveal melanoma and other primary ocular neoplasms and metastatic tumors endowed with ocular tropism.

Conclusions
In this paper, we described an orthotopic model of uveal melanoma in which tumor cells are grafted in the eye of zebrafish embryos in the proximity of the developing choroidal vasculature. In the following 3-4 days, grafted cells proliferate and move towards the eye surface, thus demonstrating that the zebrafish embryo eye is a permissive environment for the growth of UM cells. In addition, the use of firefly luciferase bioluminescent murine and human tumor cells allowed the assessment of the antitumor activity of candidate drugs when injected into the grafted eyes. In conclusion, we have established a new quantification method based on the ocular implantation of bioluminescent uveal melanoma cells in zebrafish embryos that may represent a useful in vivo orthotopic model suitable for the screening of novel drug candidates for uveal melanoma therapy.