Co-isolation of human donor eye cells and development of oncogene-mutated melanocytes to study uveal melanoma

Abstract

Background

Uveal melanoma (UM) is the most common intraocular tumor in adults, arises either de novo from normal choroidal melanocytes (NCMs) or from pre-existing nevi that stem from NCMs and are thought to harbor UM-initiating mutations, most commonly in GNAQ or GNA11. However, there are no commercially available NCM cell lines, nor is there a detailed protocol for developing an oncogene-mutated CM line (MutCM) to study UM development. This study aimed to establish and characterize premalignant CM models from human donor eyes to recapitulate the cell populations at the origin of UM.

Results

Given the precious value of human donor eyes for studying multiple ocular cell types, we validated a co-isolation protocol of both human NCMs and retinal pigment epithelial (RPE) cells from a single eye. To this end, NCMs and RPE cells were sequentially isolated from 20 donors, with success rates of 95% and 75%, respectively. MutCMs were generated from 10 donors using GNAQ^Q209L-carried lentivirus with high mutant copies (up to 98.8% of total GNAQ copies being mutant). NCM growth and behavior were characterized under different culture conditions (i.e., supplementation with serum and 12-O-tetradecanoylphorbol-13-acetate) to determine optimized protocols. Particularly, Matrigel™ coating induced spheroid growth under certain coating thickness and cell seeding density but did not improve NCM metabolic activity. Current methodologies in NCM isolation, culture, and research applications were summarized. Proteomic profiling of 4 NCMs, 1 MutCM, and 3 UMs allowed to discover significant differences in UMs including a downregulation of proteins linked to melanocyte differentiation and an upregulation of proteins involved in RNA metabolism. RNA sequencing revealed enriched pathways related to cancer, notably PI3K-Akt and MAPK signaling pathways, in MutCMs and UM cells compared to NCMs, providing insights into molecular changes in GNAQ^Q209L-mutated pre-cancer cell models and UM cells.

Conclusions

We successfully isolated and established NCM, RPE, and MutCM cell lines. We describe efficient methods for the isolation and growth of NCMs and report their phenotypic, proteomic, and transcriptomic characteristics, which will facilitate the investigation of UM development and progression. The co-isolated RPE cells could benefit research on other ocular pathologies, such as age-related macular degeneration.

Graphical Abstract

Created with https://www.biorender.com/.

Background

While UM is a rare cancer (5 cases per million and 5% of all melanomas), it represents the most common primary intraocular malignancy in adults [1,2,3]. Despite the success in controlling intraocular disease, over 50% of patients develop fatal metastases (median survival less than 1 year) [4,5,6]. UM develops following malignant transformation of uveal melanocytes either de novo or through the transformation of premalignant nevi [1, 2, 4, 5] and occurs most commonly in the choroid. A driving event in UM development is a mutually exclusive mutation in GNAQ or GNA11 that arises in over 80% of primary UMs and induces the constitutive activation of oncogenic signaling pathways [7,8,9]. Notably, choroidal nevi harboring the GNAQ/GNA11 mutations progress into melanoma in 2%, 9%, and 13% of eyes at 1, 5, and 10 years, respectively [10].

The understanding of the etiology of ocular diseases and the implementation of new therapies requires the development and validation of in vitro cell culture models. Research on UM is based on a panel of cell lines that were derived from primary or metastatic lesions [11,12,13,14]. Although they play an important role in helping researchers understand the biological aspects of the disease, these cell lines do not permit the study of UM development from a normal melanocyte. Indeed, while previous studies revealed that GNAQ/GNA11 mutations drive the initiation of UM, no human uveal melanocytes with this oncogenic mutation were developed and used to study tumorigenesis [15, 16].

Developing methods for isolating and culturing ocular cells from human donor eyes is of tremendous value for studying the pathogenesis of ocular disorders. Methods have been previously described for the isolation of individual cell populations [17,18,19,20], but infrequently for multiple cell type co-isolations [21,22,23,24], meaning that the potential uses of precious donor eyes are not maximized. For example, retinal pigment epithelial (RPE) cells could be used to study age-related macular degeneration (AMD), affecting up to 30% of the population over 75 years of age and causing 5% of all cases of blindness [25,26,27,28]. A few human-immortalized RPE cell lines have been successfully established from newborn or adult donor eyes and could be accessed commercially [29,30,31,32]. However, many of them stem from donors that do not represent the corresponding age groups for AMD onset. Thus, primary RPE cells derived from aged human donors may be more relevant for disease-designed studies. A sub-aim of this paper is to report the success rate of co-isolation of NCMs and RPE cells.

Moreover, while methods for RPE cell isolation and culture have been reviewed [20], choroidal melanocyte isolation and subsequent characterization have not been summarized. Indeed, diverged culture methods were rarely compared and discussed [18, 19, 33]. There is a need to compare individual protocols on isolation and culture, growth medium supplements and culture substrates that influence choroidal melanocyte survival, morphology, proliferation, and melanin synthesis [19, 33,34,35,36]. This could lead to a map of optimal protocols that can be suitable for specific downstream applications.

In this paper, we describe successful co-isolation and phenotypic and immunofluorescence characterization of NCMs and RPE cells from single eyes obtained from 20 human donors, with success rates of 95% and 75%, respectively. We derived GNAQ^Q209L MutCM lines from over 10 donors, validating an efficient method to develop an important cell model to study UM development. We report the proteomic profiles of a pair of NCMs and MutCMs derived from a same donor. Moreover, we compared NCM growth using different media and substrates and provided an overview of current methods for melanocyte culture and downstream applications, aiming to facilitate future studies on UM. Taken together, this is the first comprehensive characterization of human NCM isolation and co-isolation with other cell types, NCM growth under different conditions, and methodologies for developing MutCM, thereby maximizing the use of precious donor material to establish pre-clinical models for UM and other ocular diseases.

Results

NCMs and RPE cells were successfully co-isolated from human donor eyes and expanded in culture

In this study, we validated a protocol for the sequential co-isolation of these cell types from single human donor eyes (Fig. 1a, b and Table 1) [19]. NCMs expressed MLANA, vimentin, HMB45, TYRP1, and S100 and were negative for the RPE markers (i.e., cytokeratin 18 (KRT18)) indicating a pure NCM culture (Fig. 1c). In parallel, isolated RPE cells expressed specific phenotypic markers (i.e., KRT18, RPE65) and were negative for the melanocyte marker HMB45, indicating a pure RPE cell culture (Fig. 1c).

Fig. 1

NCMs and RPE cells were successfully co-isolated and cultured. a An illustration of human eye anatomy and the location of RPE cells and NCMs. b Images of eye dissection and sequential RPE cell and NCM isolation. c Immunofluorescence analyses validated cell identity and purity. Scale bars = 30 μm

Table 1 Summary of donor information. Clinical information (age, sex, cause of death), days between death to cell isolation, and cell cumulative population doubling (cpd) times of NCMs and RPE cells. f: female, m: male. n.d.: not determined. NCM cpd with †: cpd of NCM culture estimated based on the splitting ratio during passages and the last passage in culture. The cultures were used up for experiments before proliferation ceased (indicated by “ > ”) or deliberately terminated due to ceased proliferation before reaching confluence (indicated by “ < ”). RPE cell cpd with †: cells analyzed as previously published [37]

In total, we obtained eyeballs from 20 deceased human donors with an age range of 64 to 85 years (72.8 ± 6.2 years, 11 men and 9 women; Table 1). The success rate of viable cultures was 19/20 (95%) and 15/20 (75%) for NCMs and RPE cells, respectively. On average, over 1 to 3 months of cell culture, NCMs doubled 7.3 ± 3.6 times (n = 18 recorded cultures), while RPE cells doubled 15.2 ± 7.9 times (n = 9 recorded cultures) (Fig. 2a–d and Table 1). Days from death to cell isolation ranged from 2 to 6 days (4 ± 1.1 days). This, as well as donor sex, did not affect cell recovery and the success of cell culture (Additional file 1a-d). The donor age did not affect the cumulative population doubling (cpd) times of RPE cell cultures (P = 0.4670) but was significantly inversely correlated to the cpd of NCM cultures (P = 0.0303) (Fig. 2e, f).

Fig. 2

Characterization of the growth potential of NCMs and RPE cells. Cpd versus cultured days of a NCMs from 7 donors and b RPE cells from 4 donors, from day of isolation to end passage where proliferation ceased, observed by significant cell detachment. Cpd of all characterized c NCMs (n = 18) and d RPE cells (n = 9) are shown as the mean ± SD. e Cpd of NCMs significantly correlated to donor age (P = 0.0303). f Cpd of RPE cells did not significantly correlate to donor age (P > 0.05). g Pigment and morphology change of NCMs and RPE cells over 2 months. Cells eventually stained positive (blue) for β-galactosidase. Scale bars = 100 μm. d, days

Within 10 days of culture, both NCMs and RPE cells grew as a monolayer of adherent, brown-pigmented cells, displaying characteristic shapes (Fig. 2g). Morphologically, NCMs were pigmented and had various morphologies (bipolar, tripolar, or dendritic). RPE cells presented an epithelioid shape and were bigger in size with darker pigmentation. Both cell types gradually lost their pigmentation, reached senescence (stained positive for β-galactosidase), and ceased proliferating in around 2 months (Fig. 2g).

Since the recovery rate of NCMs from freeze–thaw was rarely reported, we analyzed the recovery rate and cpd of 14 batches of NCMs from 9 donors stored in liquid nitrogen (Additional file 1e-g). Total frozen days ranged from 9 to 618 days. Over half of the batches showed poor attachment after thawing (9/14, 64.29%) and a negative increment of cpd (8/14, 57.14%). The maximum cpd reached by freeze-thawed cells was significantly reduced compared to non-cryopreserved cells (P = 0.0028). This suggests that NCMs are highly susceptible to damage by the freeze–thaw cycle.

Together, these data confirm that NCMs and RPE cells can be efficiently co-isolated from a single eye and maintained in culture long enough to perform subsequent studies. Since we previously studied the behavior of primary RPE cells under blue light exposure [37], we focused in the present study on deciphering the best conditions for NCM maintenance and subsequent mutagenesis for the study of UM.

Serum and TPA were not necessary for NCM culture

NCMs exhibited lower growth potential when compared to RPE cells in general. Therefore, we aimed to optimize the NCM culture by comparing the morphology and confluence change in different media and the metabolic activity on different substrates.

Both serum and 12-O-tetradecanoylphorbol-13-acetate (TPA) are known to stimulate NCM growth [38, 39], but how they compare to serum-free and phorbol ester-free M2 medium has not been reported to our knowledge. We tested NCM growth of 4 different donors in 3 types of media: (1) DMEM/F12 + 10% FBS, (2) DMEM/F12 + 10% FBS + 100 nM TPA, and (3) a serum-free M2 medium. After 7 days of exposure to different media, NCMs exhibited apparent differences in cell morphology (Fig. 3a). While we observed donor-to-donor differences in confluence change, NCMs cultured in M2 medium exhibited the best growth rate consistently (Fig. 3b). Surprisingly, supplementation of TPA to DMEM/F12 + 10% FBS only improved the growth of 1/4 (25%) cultures (i.e., NCM713). In summary, serum and TPA were not necessary or insufficient to stimulate NCM growth, and NCMs responded to these two factors variably.

Fig. 3

Different cell culture media were assessed to optimize NCM culture. a Morphology of NCMs in different media on day 7. Scale bars = 100 μm. b Confluence change (%) of NCMs cultured compared to day 0. a, b n = 4 biological replicates (NCMs from 4 different donors), with each repeated in n = 4 wells. Data are shown as mean ± SD. Two-way ANOVA with Dunnett’s multiple comparisons tests (baseline: DMEM/F12 + 10% FBS) at each time point. ns: not significant. *, **, ***, and **** indicate P < 0.05, 0.01, 0.001, or 0.0001, respectively

The addition of a basement membrane matrix altered NCM growth patterns

A basement membrane matrix is a specialized extracellular matrix (ECM) containing collagen IV, laminins, heparan sulfate proteoglycans, and a variety of growth factors [40]. It provides physiological culture conditions and has been used to enhance cell differentiation and survival of primary cells. It was reported that Matrigel™, a commercial basement matrix, promotes neural, melanocytic, and chondrogenic differentiation of trunk neural crest cells [41]. In addition, the thicknesses of the Matrigel™ layer and cell seeding density have been shown to affect the cell morphology of breast metastatic cells [40]. The effect of Matrigel™ on melanocyte morphology and survival has not been previously reported. To understand whether Matrigel™ could stimulate the NCM growth, we first tried culturing a late passage of NCM114 in a 24-well plate coated with 80 μL Matrigel™ per well or on uncoated wells (Fig. 4a). While the NCMs constantly detached on non-coated wells over the 6 days, they did not detach on Matrigel™-coated wells, and they migrated to form spheroids. Based on this observation, we next sought to study the effect of Matrigel™ on NCM cultures more comprehensively by coating wells with different matrix thicknesses and using different cell seeding densities in 96-well plates.

Fig. 4

Different cell culture substrates were assessed for optimizing NCM culture. a Morphology of NCM114 (late passage, P4) seeded on a 24-well plate coated without or with 80 μL of Matrigel™ per well at 18,333 cells per well. Scale bars = 100 μm. b Morphology of NCMs on different substrates. In a 96-well plate, 2500 cells were seeded per well and were imaged after 3 days. Scale bars = 100 μm. c The metabolic activity of NCMs was assessed after 3 days of culture on different substrates. Data are shown as mean ± SD. One-way ANOVA with multiple comparisons test (baseline: no coating), ns: not significant. **P < 0.01, ****P < 0.0001. d Morphology of NCMs on Matrigel™ at various cell seeding densities. In a 96-well plate coated with 45 μL of Matrigel™ per well, 1000, 3000, 5000, or 7000 cells were seeded per well and were imaged after 7 days. b–d n = 4 biological replicates (NCMs from 4 different donors), with each repeated in n = 3 wells. Scale bars = 400 μm

We tested NCM growth on different amounts of Matrigel™ (15 μL, 30 μL, and 45 μL per well of a 96-well plate) and a cell attachment enhancer, poly-l-lysine. The poly-l-lysine-coated surface is positively charged and thus enhances the electrostatic interaction between the negatively charged cell membrane and the surface. Morphologically, NCMs cultured without coating presented typical melanocyte shapes (bi/tripolar and dendritic), while they were more flattened and rounded on poly-l-lysine (Fig. 4b). With the increasing thickness of Matrigel™, the dendrites of NCMs became more retracted. On 45 μL of Matrigel™, NCMs exhibited mostly rounded morphology. After 3 days of culture, a CCK-8 assay was performed to analyze cell metabolic activity, an index for cell vitality (Fig. 4c). Surprisingly, the metabolic activity was not enhanced by the addition of poly-l-lysine or Matrigel™ coating but was inhibited in 3/4 (75%) cultures (P < 0.01).

Lastly, we tested different cell seeding densities on a thick layer of Matrigel™ (45 μL per well) to determine if the confluency would impact cell growth characteristics. Interestingly, over 7 days of culture, NCMs remained mostly as single cells when seeded at low density (1000 or 3000 cells per well), while spheroids were seen when seeded at higher density (5000 or 7000 cells per well) (Fig. 4d). Based on daily observation, the spheroids appeared to result from cell migration, instead of growth in foci. Further study on this phenomenon is warranted. In summary, although donor-to-donor variations were seen, Matrigel™ did not enhance NCM metabolic activity but altered cell morphology and growth behavior.

GNAQ ^Q209L MutCMs were generated successfully with high mutant GNAQ copies

GNAQ/11^Q209L mutation is believed to be a major initial event for UM oncogenesis. However, the generation of mutant premalignant choroidal melanocytes of human origin has not been previously reported. The efficient isolation of NCMs has prompted us to generate a GNAQ^Q209L MutCMs from CMs that would serve as a powerful model to study UM oncogenesis and development.

As NCMs are precious cells that are limited in quantity, we practiced and validated the lentiviral transduction method on multiple helper cell types including BJ fibroblasts and HEK293 (Additional file 2a-b). These cell types are easy to grow and were used for testing the virus preparation infectivity before its use on NCMs. A viral dose-dependent mutant copy fraction was found, and the freshly prepared virus was more infectious than the freeze-thawed virus. In addition, polybrene, a cationic polymer, was a critical factor in enhancing the transduction efficiency (Additional file 2c).

After validating the virus infectivity on other cell types, we sought to test it on NCMs. Four NCM donors were transduced at the same time under the same condition (50,000 cells were mixed with 50 μl of fresh virus in 1 ml of M2 medium supplemented with 4 μg/ml polybrene). Consistently, all NCMs were efficiently transduced with high mutant copies (0.9783 ± 0.0081 fraction of all GNAQ copies being mutant, Fig. 5a, b). Successful introduction of the point mutation (GNAQ (c.626A > T)) was further validated by Sanger sequencing (Additional file 3). MutCMs with high mutant copies (mutant fraction > 0.8) were heterogenous in morphology, with a dendritic shape and long processes (Fig. 5a, b, Additional file 2d-e). MutCMs expressed MLANA, vimentin, HMB45, TYRP1, and S100 and were negative for KRT18 as expected (Fig. 5c).

Fig. 5

The insertion of the mutant gene (GNAQ (c.626A > T)) by lentiviral transduction was highly efficient. a Phase contrast images and ddPCR 2D plots for MutCMs. b The transduction efficiency, indicated by wildtype (WT) and mutant (Mut) fraction. c MutCMs (MutCM713 was shown) expressed MLANA, Vimentin, HMB45, TYRP1, and S100 and were negative for KRT18. Scale bar = 30 μm. d The fraction of WT and Mut GNAQ/11 gene in 4 UM cell lines (92.1, MP46, MEL270, and MP41) were profiled. e Representative images and digital droplet PCR (ddPCR) of NCMs and MutCMs and f their WT and Mut copy fraction. Scale bars = 100 μm. ddPCR plots: droplets that are positive for WT, Mut, and both gene copies are shown in green, blue, and orange, respectively. Negative droplets are shown in black

To compare to the mutant GNAQ/11 mutant fraction in UM cell lines, we analyzed the wildtype (WT) and GNAQ/11 mutant (Mut) copies in 4 human UM cell lines: 92.1 and MP46 (GNAQ (c.626A > T)), MEL270 (GNAQ (c.626A > C)) and MP41 (GNA11 (c.626A > T)). Mutant fractions ranged from 0.3225 to 0.5789 (Fig. 5d). Therefore, we believed the mutant copy fraction similar to this range may be more relevant to the context of UM. By titrating various virus dosages on NCMs (early passage, P < 3) or helper cell lines in order to tune the virus dosage, 6 MutCMs were successfully transduced with mutant fractions ranging from 0.2095 to 0.6296 (Fig. 5e, f, Additional file 2d-e). Interestingly, compared to their parental NCMs, these MutCMs tended to grow in clusters instead of a monolayer (Fig. 5g, Additional file 2e). The level of clustering and spheroid formation varied between donors.

In addition to the lentiviral transduction method for the genetic engineering of NCMs, we also explored electroporation for future genome editing applications such as creating relevant melanocyte mutants using CRISPR-Cas9. We tested the electroporation efficiency of an EGFP plasmid in NCMs and UM cells (Additional file 4). Optimal electroporation conditions differed between NCMs (1400 V, 20 ms, 2 pulses) and UM cells (1300 V, 20 ms, 1 pulse) (Additional file 4a) as well as the percentage of obtained fluorescent cells (48.9% and 83.5%, respectively) (Additional file 4b). Electroporating NCMs could thus be a valuable technique in various research fields related to vision science research.

In summary, the described methods were highly efficient for establishing MutCMs. Further characterization of MutCMs is warranted to understand behavioral and molecular changes that may recapitulate the formation of nevi.

Proteomic profiling of NCMs, MutCMs and UMs

To further our understanding of premalignant CM models, we sought to compare proteomic profiles of NCMs (NCM602, 713, 809, 1009) and UMs (92.1, MP46, MP41). Secondly, we compared a pair derived from the same donor (NCM713 and MutCM713). Tandem mass tag (TMT)-mass spectrometry identified 3268 proteins after quality control of the data. All cells expressed melanocyte markers (e.g., MLANA, VIM, PMEL, TYRP1, and S100B) with higher levels in NCMs and MutCMs compared to UM cell lines (Fig. 6a). NCMs (n = 4) and UMs (n = 3) showed robust difference in the volcano plot (Fig. 6b), with 549 and 191 proteins being significantly downregulated and upregulated, respectively (P < 0.05, fold change < 0.5 or > 2). Gene ontology (GO) term analysis revealed distinct enrichment of biological processes, molecular functions, and cellular components in NCMs and UMs. Notably, ATP metabolic process, oxidation–reduction, cell–matrix adhesion, and cell morphogenesis involved in differentiation were upregulated in NCMs, while mRNA metabolic process, RNA splicing, and gene expression were upregulated in UMs (Fig. 6c). The Reactome pathway enrichment analysis is shown in Table 2. In concordance with GO terms, NCM proteome profiles were enriched in pathways such as tricarboxylic citric acid (TCA) cycle and respiratory electron transport, cell-extracellular matrix interactions, cell junction organization, and biological oxidation. In addition, innate immune system and apoptosis pathways are elevated in NCMs. UM proteome profiles were enriched in pathways such as metabolism of RNA, translation, mRNA splicing, and regulation of TP53 activity, which highly aligns with the findings of a recently published study [42].

Fig. 6

Proteomic comparisons of NCMs and UMs. a Heatmap of relative expression of melanocyte markers shown in z-score. b Volcano plot of proteins in NCMs vs. UMs. Significantly upregulated and downregulated proteins in NCMs and UMs appear in blue and red, respectively (P < 0.05, Log₂(fold change) < − 1 or > 1). c Gene ontology analysis of the significantly different proteins. Selected terms are with P < 0.05, Benjamini < 0.05. A full list can be found in Additional file 6

Table 2 Selected upregulated pathways in NCMs and UM cells, respectively (P < 0.05). A full list can be found in Additional file 6

In contrast, differences between NCM713 and MutCM713 were subtle (Table 3), with only 36 proteins being differentially regulated (P < 0.05, fold change < 0.67 or > 1.5) (Table 3). Interestingly, 5 out of 8 downregulated proteins in MutCM713 are involved in mitochondrion, and 2 downregulated proteins (i.e., STK10, COMMD5) are thought to have tumor suppressor properties [43, 44]. Upregulated proteins in MutCM713 include ALDH1A1, which is a cancer stem cell marker [45] and has been found overexpressed in melanoma cells [46]. CASP3 and UBE2E3 are involved in regulating apoptosis and were elevated. Importantly, numerous upregulated proteins in MutCM713 have been implicated in cancer progression, including PAK4 [47], MKI67 [48], PSMD10 [49], ELP3 [50, 51], and STAT6 [52]. In summary, the proteomic changes in MutCM713 may provide novel insights and understanding of UM oncogenesis.

Table 3 Differentially expressed proteins in NCM713 and MutCM713 (P < 0.05, fold change < 0.67 or > 1.5)

Transcriptomic profiling of NCMs, MutCMs, and UMs

To validate the proteomic data, we performed RNA sequencing on NCMs and MutCMs derived from the same donor (NCM713, MutCM713) and a UM cell line (92.1). Over 26,000 identities of RNA with unique official gene symbols were identified after quality control. Consistent with the proteomic results, NCM713 and MutCM713 have generally higher expression levels of melanocyte markers (i.e., MLANA, VIM, PMEL, TYRP1, S100B, MITF, DCT, SOX10, EDNRB, GPNMB) compared to 92.1 (Fig. 7a). Differential gene expression was analyzed and a P value < 0.05 and Log₂(fold change) < − 1 or > 1 was considered significant. Specifically, for NCM713 vs. 92.1, 3536 and 2918 genes are downregulated and upregulated in 92.1, respectively (Fig. 7b). NCM713 and MutCM713 have fewer differences, as expected, with 1541 and 1957 genes downregulated and upregulated in MutCM713, respectively (Fig. 7c). Five hundred eighteen and 700 genes are commonly downregulated and upregulated in 92.1 and MutCM713, respectively (Fig. 7d).

Fig. 7

Transcriptomic comparisons of NCM713, MutCM713, and 92.1. a Heatmap of relative expression of melanocyte markers shown in z-score. Volcano plot of b NCM713 vs. 92.1 and c NCM713 vs. MutCM713. Significantly upregulated and downregulated genes appear in blue and red, respectively (P < 0.05, Log₂(fold change) < − 1 or > 1). d Venn diagram of differential gene expression analysis of NCM713 vs. 92.1 and NCM713 vs. MutCM713 (P < 0.05, Log₂(fold change) < − 1 or > 1). e Selected regulated pathways identified by KEGG Pathway Enrichment Analysis of commonly regulated genes in 92.1 and MutCM713. A full list can be found in Additional file 6. f Heatmap of selected gene expression of interest

To understand whether MutCM713 shares cancer-associated malignant properties with 92.1, the commonly significant genes were queried against KEGG Databases (Fig. 7e, f). Top enriched pathways of downregulated genes include cell adhesion molecules, motor proteins, cell cycle, gap junctions, and p53 signaling pathway, with reduced expression of genes such as CDH1, CDH4, ITGA6, TP53I3, AIFM2, and CHECK2. The top 3 enriched pathways of upregulated genes are pathways in cancer, PI3K-Akt signaling pathway, and MAPK signaling pathway, followed by other prominent signaling associated with cell survival, migration, and aggressiveness, including Ras, WNT, Rap1, Hippo, Notch, and TGF-beta signaling pathways. Notably, upregulated gene expression include PLCB4, ERBB4, RASGRP3, TRAF6, SOS1, PIK3CA, AKT3, MYC, WNT4, WNT11, WNT88, VEGFA, VEGFC, JAG2, and NOTCH3. Complete lists of differentially expressed genes and significant KEGG pathways have been provided in the Additional file 6. Collectively, transcriptomic profiling data provides further insights into the potential molecular changes in GNAQ^Q209L-mutated pre-cancer cell models (MutC713) and UM cells (92.1).

Overview of current culture methods and research applications using human NCMs

There is a lack of literature review on melanocyte culture methodologies, especially within the context of ocular research. Considering this gap, we summarized the current methodologies for establishing ocular cell cultures and research applications, focusing particularly on human NCMs, that are not commercially available (Fig. 8). This can serve as a reference guide of validated protocols and techniques according to downstream analyses.

Fig. 8

Overview of current methods in generating NCMs and RPE cells, and NCM culture methods and research applications. Created in https://BioRender.com. MSH, melanocyte-stimulating hormone; bFGF, basic fibroblast growth factor; cAMP, cyclic adenosine monophosphate

The isolation procedures include eyeball dissection and sequential enzymatic dissociation aimed at isolating NCMs and RPE cells. Notably, Hu et al. compared the use of trypsin, collagenase, or both enzymes for the isolation of melanocytes from the choroidal stroma [18]. Among these methods, a trypsin treatment followed by an incubation in collagenase produced the highest number of viable NCMs. Contamination by RPE cells and fibroblasts was eliminated by Geneticin. Another significant development in this arena was by Valtink and Engelmann, who described a protocol for the co-isolation of RPE cells and NCMs [19, 53]. The authors observed that collagenase treatment for one hour was sufficient in dissociating RPE cells but not for NCMs, allowing for sequential isolation of RPE cells followed by NCMs using dispase. Apart from the conventional ex vivo extraction techniques, multiple studies have created RPE cells from induced pluripotent stem cells [20]. Lastly, oncogene-mutated melanocytes, which mimic melanoma precursor cells, have been created from human and murine skin melanocytes [8, 54], but to the best of our knowledge, nothing similar has been documented for NCMs. We therefore propose the methodologies for creating MutCMs by lentiviral transduction or electroporation for genome editing.

While RPE cells can be easily grown in an FBS-supplemented medium [20], the culture of NCMs requires specific growth and melanin-stimulating agents [18, 39, 55]. Hu et al. suggested that a melanocyte culture medium requires three classes of supplements, such as serum, mitogens (e.g., phorbol esters or bFGF), and cAMP enhancers (e.g., cholera toxin, isobutyl methylxanthine) [39, 55]. Phorbol esters were found to be mitogenic and melanin-stimulating [39, 55]. Basic FGF, insulin, and cAMP enhancers were shown to be important for melanocyte proliferation [39, 55]. Some neurotransmitters could influence melanocyte growth and melanogenesis [56]. Stem cell factor was mitogenic in the presence of certain essential factors [57]. Although melanocyte-stimulating hormone was proven to induce melanocyte growth and melanogenesis for skin melanocytes, its effect on NCMs was controversial [33, 58].

As crucial as the culture medium, the substrate could influence melanocyte survival and morphology. Thereby, the choices of culture conditions rely on the research objectives and applications. Research on NCMs mainly focused on melanin biogenesis [24, 55, 56, 59,60,61,62], oncogenesis (primarily as a control for studying UM) [63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83], and immune regulation [84,85,86,87,88,89,90,91,92,93]. Most studies cultured NCMs on uncoated plastic surfaces. The attachment of melanocytes to specific ECM proteins was assessed by several studies. Melanocytes preferentially attached to fibronectin compared to plastic or other ECM proteins including collagen types I, III, or IV, and laminin [35, 36]. ECM proteins influence melanocyte morphology, tyrosinase activity, and proliferation, and intracellular calcium may be a key signaling pathway involved in the response to ECM proteins [35, 36]. Melanocytes could grow in a monolayer on uncoated plastic surfaces or on surfaces coated with individual ECM proteins. However, NCMs cultured on Matrigel™ (basement membrane) could grow either in monolayer or as spheroids depending on the thickness of the Matrigel™ layer and the cell seeding density, as reported in the present study. Moreover, dissociated NCMs have been used in tissue engineering, by embedding them into choroidal fibroblast-derived ECM sheets [21]. The development of this technology sheds light on innovative in vitro ocular models for disease comprehension, and on ocular tissue transplantation for future therapies.

Discussion

Research on many human diseases has benefited from the implementation of cell culture models. Our understanding of ocular diseases and pre-clinical discoveries for novel therapies depend on the development of efficient in vitro eye tissue-derived models. Attempts have been made in this optic using animal and human eyes [17,18,19,20, 33, 59, 94]. In most studies, individual cell types are isolated to be used for a specific study, while the remaining tissues are discarded. Yet, fresh human donor eyes are precious research materials that are hard to obtain. While methods for consecutive isolation of NCMs and RPE cells have been described [19, 24, 53], the efficiency of co-isolation was not reported donor-by-donor. In the current study, we collected eyes from 20 human donors and validated the high efficiency of obtaining viable cultures of NCMs and RPE cells from the same donor eyes (95% and 75%, respectively), two cell types that are involved in the development of important ocular diseases, such as UM or AMD. We also provide a detailed summary of best practices for their co-isolation, followed by a focus on the characterization of NCMs under different culture conditions.

The major challenge for studying NCMs lies in the high variability of cell viability and their short-lived culture. A serum-based medium supplemented with TPA and other growth and melanin stimulators has been widely used for NCM culture [18, 39], while a serum-free and TPA-free medium (M2) was also reported for efficient NCM culture [19]. We found NCMs consistently had the highest proliferation rate and maintained typical melanocyte morphology in M2 medium, but variable proliferation rates and reduced dendritic morphology in the serum-based medium supplemented with or without TPA. These data suggest that serum and TPA alone may not be necessary or sufficient for NCM culture. In addition, serum contains undefined factors and TPA is a tumor inducer [95], which can confound the downstream analysis of NCMs when studying ocular physiology or pathologies. Thus, the M2 medium may be more suitable for NCM culture. It is essential to note that serum-free formulations for melanocyte culture have been comprehensively studied in cutaneous melanocytes [34, 96, 97]. Considering that both cutaneous and choroidal melanocytes originate from neural crest cells, a formulation effective for one might be similarly suitable for the other. However, direct analyses to compare the optimal growth conditions for these two types of melanocytes are yet to be explored. In addition, melanocytes could be generated from induced pluripotent stem cells or their progenitor cells (e.g., neural crest cells for melanocytes) [41, 98, 99]. These technologies hold the promise to mitigate the existing challenges surrounding limited tissue availability and cellular quantity. However, the identity and behavior of cells derived under in vitro conditions require careful validation.

Enhancing cell attachment by growing cells on biological or chemical substrates is another avenue to improve cell growth. However, both poly-l-lysine and Matrigel™ at all tested thicknesses did not improve the metabolic activity but generally inhibited it, when compared to the plastic surface without substrate. Interestingly, on thick Matrigel™, NCMs seeded at high density migrated and formed spheroids while they remained as single cells when seeded at low density. Matrigel™ previously supported the differentiation of neural crest cells into neurons, melanocytes, and chondrocytes [41]. Further studies on the effect of Matrigel™ on melanocyte differentiation and behavior would be of interest to investigate the interaction of NCMs with ECM proteins. Similarly, seeding cutaneous melanocytes at high density on chitosan-coated surfaces has been shown to induce melanocyte aggregation into spheroids [100]. Moreover, coating with agents that prevent cutaneous melanocyte attachment led to spheroidal growth in suspension [101]. The same study showed that melanocytic spheroids can return to a monolayer state by subculturing them on uncoated or specific ECM-coated surfaces, meaning that melanocyte spheroids hold a high potential to be used for tissue transplantation for diseases such as vitiligo [101]. Cutaneous melanocyte spheroids have also been used for drug screening, as they showed elevated melanin content and were more sensitive to skin-lightening drugs [102]. Collectively, the choices of substrates map to corresponding melanocyte behavior and morphology, thus downstream applications.

Besides uncovering the best practices of NCM culture, we propose using modified NCMs to advance our understanding of UM development. In this view, we developed human MutCM cell lines with the most common UM driver mutation GNAQ^Q209L. Based on the two-hit hypothesis for full malignant transformation, this predisposing hit (i.e., GNAQ^Q209L mutation) may represent the prerequisite and priming event for these cells to acquire a nevus-like state [103]. While GNAQ^Q209L-mutated melanocytes have been previously developed using immortal human and murine skin melanocytes [8, 54], such a model of primary human NCMs that represents the cell of origin for UM has not been previously reported. With the described method, we obtained MutCMs with mutant GNAQ copies accounting for up to 0.988 (98.8%) fraction of total copies, indicating a highly efficient transduction. To estimate the GNAQ mutant fraction in the UM context, we analyzed 4 human UM cell lines and found mutant fractions between 0.3225 to 0.5789. In addition, a study reported that the median fraction of GNAQ mutant cells in choroidal nevi was 0.13 (13%) [104]. Thus, we created MutCMs with mutant copies (0.2095 to 0.6296) similar to the range found in UM cells. Interestingly, many of these MutCM cultures tended to aggregate or grow in foci, a phenomenon that may be relevant to the oncogenic growth of nevi. Furthermore, genome editing using CRISPR-Cas9 could be a newer option to generate mutant cell models for punctual mutations such as GNAQ [105]. Electroporation is recommended to deliver the Cas9 protein and sgRNAs in hard-to-transfect cell lines or primary cells [106], and we demonstrated that this method can be used successfully with primary NCMs and UM cells. To the best of our knowledge, no studies have been published with human NCMs edited with CRISPR-Cas9. A limitation that remains to be addressed in the future is that one or two mutant allele copies per cell (heterozygous or sometimes homozygous) are present in natural UM cases [107]. The number of mutant copies per cell of the MutCM models could be better controlled in the lentiviral transduction method by defining the multiplicity of infection (MOI). Using a low MOI and performing a clonal selection of transduced cells by antibiotic-resistant gene or fluorescence-activated cell sorting could further optimize the models to be relevant to the UM context. While we have attempted extensively to select the transduced cells by puromycin resistance with the psd44-GqQL plasmid, MutCM viability was consistently and significantly compromised by this method. Therefore, future development of plasmid design, such as by incorporating a fluorescence gene, could facilitate the sorting of MutCM.

Both GNAQ and GNA11 mutations are crucial initial drivers of UM development. In this study, we chose to focus on one of the two mutations (GNAQ^Q209L) as an example of a UM-initiating mutation that could represent a pre-malignant cell state. Given that GNA11^Q209L is associated with more metastatic lesions [7], developing GNA11^Q209L MutCM models will likely garner significant research interest in the future. Future studies could focus on inducing secondary mutations in BAP1, SF3B1, and EIF1AX, known for their significant impact on metastatic potential [108]. Conducting in vivo studies would provide valuable insights into the tumorigenic and metastatic capabilities of these models. Moreover, immortalizing cell lines may be another future direction to overcome the short in vitro lifespan limitation. In summary, the methodologies we established for NCM isolation, culture, and genetic engineering through both lentiviral transduction and electroporation pave the way for further in-depth investigation into the mechanisms underlying UM oncogenesis, as well as the development of other MutCM models.

Both proteomic and transcriptomic results revealed enriched pathways in NCMs in cell adhesion and cell junction. Moreover, RNA sequencing identified numerous differentially expressed genes involved in cancer pathways, notably MAPK and PI3K-Akt pathways, which have been extensively associated with UM development [109, 110]. These changes were not evident in the proteomic results, which could be attributed to the regulation of RNA metabolism and translation, which are highly regulated pathways as identified in the proteomic comparison of NCMs vs. UMs in Table 2. Another reason could be attributed to the sensitivity of techniques. Transcriptomic analyses identified over 26,000 unique genes, offering a larger dataset compared to proteomic analyses (3268 proteins), allowing for a more in-depth characterization. However, validating transcriptomic changes at protein levels is important for ensuring functionality.

A challenge of the study is the limited number of biological replicates, with 4 NCMs, 1 MutCM, and 3 UMs for mass spectrometry and one cell line of NCMs, MutCMs, and UMs for RNA sequencing. Specifically, MutCM713 was chosen for the comparison to NCMs (NCM713) because of its sufficient cell number for protein and RNA extraction. 92.1 was chosen as a representative UM cell line for transcriptomic comparison because it contains GNAQ^Q209L mutation and presents a proliferation rate close to the early passage of NCM713 (data not shown, supporting data in Additional file 6-Sheet 25 and Fig. 3). Given the challenge in increasing the sample size, it is important to consider that the greater differences observed between NCMs and UMs (from different donors) compared to NCMs and MutCMs (from the same donor) in both omics studies may partially be due to variations in the genetic backgrounds of different donors. An ideal comparison would involve NCM, MutCM, and UM cells derived from the same individuals, but to our knowledge, this does not exist. Moreover, increasing the sample size of the experimental model could facilitate identifying common molecular mechanisms underlying the malignant changes among these cells. Our datasets contribute to this collective effort in UM research. To the best of our knowledge, this is the first proteomic and transcriptomic profiling of a MutCM developed from NCMs. These datasets could be a useful resource for understanding UM oncogenesis.

Conclusions

In conclusion, we validated the co-isolation of NCMs and RPE cells from the same donors, aiming to optimize the use of human donor eyes in research. Different mediums and substrates were compared to identify the best practices for NCM culture. Furthermore, we developed a protocol for the generation of a premalignant UM model (GNAQ^Q209L MutCMs) and characterized the proteomic and transcriptomic profiles of NCMs, MutCMs, and UMs, which could be beneficial for future studies on the mechanisms for UM initiation.

Methods

Eye collection

Human eyeballs were provided by the Centre Universitaire d’Ophtalmologie Eye Bank (CHU de Québec-Université Laval Research Centre), following informed consent from donor’s next-of-kin. Eyeballs were acquired from May 2021 to March 2023 and used in accordance with protocols approved by the Research Institute of the McGill University Health Center (RI-MUHC, IRB#2019–5314), the CHU de Québec-Université Laval Research Centre (IRB#2021–5273), and the Code of Ethics of the World Medical Association. The age, sex, cause of death of donors, and days from death to cell isolation were recorded (Table 1).

RPE cells and NCM isolation and culture, UM cell culture

Cell isolation protocols were established essentially as reported previously [19]. Upon receiving, the eyeballs were dissected in PBS containing 1% penicillin/streptomycin (pen/strep) antibiotics, by removing the cornea, lens, iris, ciliary body, vitreous humor, and neural retina (Fig. 1a, b). Four incisions were made towards the optic nerve to obtain a petal-like structure for the removal of the neural retina. The tissues laying on the sclera (RPE + choroid) were immersed in a collagenase solution (collagenases IA and IV (Sigma), each at 0.5 mg/mL in RPE medium (DMEM/F12 + 10% FBS + 1% pen/strep) at 37 °C for 1 h. The tissue was carefully rinsed to release RPE cells, and the homogenate was pelleted at 1500 rpm for 5 min. Afterwards, the choroid was detached from the sclera and put in 0.6 U/mL Dispase II solution (Sigma) in M2 medium (Melanocyte Growth Medium M2 (PromoCell) supplemented with 1% pen/strep at 37 °C overnight. The homogenate was collected, shaken to obtain a single-cell suspension, and pelleted at 1500 rpm for 5 min. Both cell types were washed with PBS containing 1% pen/strep, filtered through a 40-μm cell strainer, then pelleted at 1500 rpm for 5 min and resuspended in RPE or M2 media on uncoated culture dishes. For NCM cultures, 50–100 μg/mL geneticin was supplemented until the contaminating RPE cells or fibroblasts were removed (usually within 10 days). Cells were grown at 37 °C in a humidified atmosphere with 5% CO₂, and the medium was changed every 3 days. When confluent, RPE cells were passaged using 0.05% trypsin and neutralized by RPE medium, while NCMs were passaged using 0.025% trypsin and neutralized by Trypsin Neutralizing Solution (PromoCell). Cell stocks were made by freezing RPE in 90% FBS + 10% DMSO or NCMs in Cryo-SFM medium (PromoCell).

MP41 (#CRL-3297) and MP46 (#CRL-3298) cell lines were purchased from American Type Culture Collection (ATCC) [12]. 92.1 cells were kindly gifted from Dr. Martine Jager (Leiden University, Netherlands) [111]. MEL270 were gifted by Dr. Vanessa Morales (University of Tennessee) [112]. All UM cell lines (92.1, MEL270, MP41, MP46) were cultured in RPMI-1640 (Corning™ or Wisent Inc.) supplemented with 10% fetal bovine serum (FBS), 10 mM HEPES, 2 mM glutagro, 1 mM NaPyruvate, 1% penicillin and streptomycin (pen/strep), and 10 μg/ml insulin (Roche).

Lentiviral transduction for GNAQ ^Q209LMutCM generation

The lentiviral transduction method was developed for NCMs with knowledge built on the guidance for lentiviral transduction for mammalian epithelial cells by Iggo et al. [113] and lentivirus production protocol on https://www.addgene.org. The GNAQ^Q209L-carried lentiviral vector (psd44-GqQL; Addgene plasmid #46,826, gift from Dr. Agnese Mariotti) was co-transfected with the lentiviral packaging and envelope expressing plasmids (psPAX2 and pMD2.G; Addgene plasmids #12,260 and #12,259, gifts from Dr. Didier Trono) into HEK293T cells (ATCC, #CRL-3216) using the Xtreme-GENE 9 transfection reagent (Sigma). Specifically, 5.6 μg of psPAX2, 2.4 μg of pMD2.G, and 8 to 16 μg of psd44-GqQL were mixed with 27 μL of XtremeGene transfection reagent (Roche) and 750 μL of Opti-Mem™ (Gibco) at room temperature for 30 min. The mixture was then added to 8 mL fresh antibiotic-free medium to transfect HEK293T cells in a T75 flask. After 16 h, the medium was refreshed. After 48–72 h, the virus-containing medium was collected, spun at 500 × g for 10 min to remove cell debris, and the supernatant was filtered through a 0.45-μm filter and concentrated using the Lenti-X™ Concentrator (Takara Bio) following the manufacturer’s protocol. The virus pellet was resuspended in the antibiotic-free medium at 1:10 of the original volume, aliquoted and used freshly or stored at – 80 °C for future use.

For viral transduction, 50,000 NCMs of early passages (P1 to P3) were mixed with virus aliquots (1:30 to 1:5 virus to medium volume ratios, multiple virus dosages could be tested for generating MutCMs with desired mutant gene fraction) and polybrene (4–8 μg/mL) in 1 mL of M2 medium and seeded per well of 12-well plate and cultured for 48 h prior to two washes with PBS and reincubation with M2 medium. The transduced cells were not selected by puromycin due to the high toxicity of puromycin on the NCM culture (data not shown).

To determine the efficiency of viral transduction, ddPCR analyses were used as we previously reported [114]. Genomic DNA was isolated from NCMs, MutCMs, and UM cell lines (92.1, MEL270, MP41, MP46 [12, 111, 112]) using the DNeasy Blood & Tissue Kit (Qiagen). 92.1 and MP46 are GNAQ (c.626A > T) mutated, MEL270 is GNAQ (c.626A > C)) mutated, and MP41 is GNA11 (c.626A > T) mutated. The PCR reaction was run at 95 °C for 10 min, 50 cycles (95 °C for 30 s, 56 °C for 1 min, 72 °C for 30 s), 98 °C for 10 min, and held at 12 °C. The ddPCR data were read and analyzed by the QX200 Droplet Reader and QuantaLife software (Bio-Rad).

Electroporation of cells for genome editing purposes

Electroporation is a method used to introduce foreign DNA, RNA, or other substances into cells by applying an electrical field, which temporarily permeabilizes the cell membrane, allowing these substances to enter the cell [115]. Using the Neon® Transfection System 10 μL Kit (Thermo Fisher Scientific), 1 to 5 × 10⁵ NCMs or UM cells (Mel185 cell line (WT for GNAQ/GNA11) [116]; gift from Dr. Bruce R. Ksander, Harvard Medical School) were used per electroporation. The Neon® 24 optimization protocol was executed as per the manufacturer’s guidelines to ensure maximal transfection efficiency. Two to 12 × 10⁶ cells were collected and centrifuged at 500 × g for 5 min. The supernatant was carefully removed, and the cell pellet was subsequently washed in 1 mL of PBS without Ca²⁺/Mg²⁺. Cells were then mixed with the Resuspension SK Buffer containing an EGFP plasmid (pRRLSIN.cPPT.PGK-GFP.WPRE (12 μg in 240 μL of resuspension buffer); Addgene plasmid #12,252, a gift from Dr. Didier Trono). For the 24 unique optimization conditions, each varying in terms of pulse voltage/width/count, a 10 μL aliquot of the cell suspension was utilized. Once the electroporation was completed, cells were transferred to 24-well plates and incubated in growth medium for 48 h before analysis. The phase contrast and fluorescence images were taken with a Lumascope 620 microscope (Etaluma), and both cell types were analyzed by flow cytometry to determine the percentage of EGFP-positive cells (BD Accuri C6 Plus flow cytometer).

Cumulative population doubling (cpd)

Cpd was estimated starting at the end of passage 1 (the seeding of freshly isolated cells corresponds to passage 0) using the displayed formula. The number of cells at the start (Nseed) and end (Nend) of passage N was recorded. $cpd\left(N\right)=\frac{\log(\frac{N_{end}}{N_{seed}})}{\log(2)}+cpd\left(N-1\right),cpd\left(0\right)=0.$

Immunofluorescence characterization of isolated cells

Cells were cultured in 4-well chamber slides (Ibidi) for 2 days, then fixed in 4% paraformaldehyde for 10 min, and permeabilized by 0.1% Triton X-100 for 10 min. Cells were incubated with PBS 1% BSA/0.1% Tween 20 (B-PBST) for 45 min to block unspecific binding of antibodies and incubated overnight at 4 °C with antibodies against MLANA (#TA801623, 1:100; Thermo Fisher Scientific), HMB45 (#59,305, 1:200; Santa Cruz Biotechnology), vimentin (#ab92547, 1:250), S100 (#ab4066, 1:50), and TYRP1 (#ab235447, 1:500) as well as RPE65 (#ab231782, 1:300) and KRT18 (#ab32118, 1:300) (all from Abcam). Cells were then incubated with secondary antibodies in B-PBST for 1 h at room temperature (Alexa Fluor™ 488-conjugated goat anti-mouse (#A11001, 1:1000), Alexa Fluor™ 594-conjugated donkey anti-rabbit (#A21207, 1:1000), both from Invitrogen), followed by a 20-min incubation with Hoechst 33,342 (Invitrogen) to stain nuclei. Cells were visualized using an LSM 780 confocal microscope (Zeiss).

Cell confluence assay

Cell confluence was analyzed by live cell imaging (IncuCyte® S3, Essen BioScience). Cells were plated in a 96-well plate at 2500 cells per well in 100 μL medium in quadruplicates. After 2 days, the medium was refreshed, and the plate was scanned by the IncuCyte phase contrast microscope at 10X. Cell confluence was analyzed with the IncuCyte® S3 software.

Cell metabolic activity assay

Cells were plated in a 96-well plate at 2500 cells per well in 100 μL medium in quadruplicates. Before the assay, 10 μL of CCK-8 reagent (Dojindo Molecular Technologies) was added to the culture medium and the plate was incubated at 37 °C for 2 h. The absorbance was measured at 450 nm by an Infinite M200Pro microplate reader (Tecan).

β-galactosidase enzyme assay

The detection of cellular senescence was performed with the Beta Galactosidase Staining Kit (#ab102534, Abcam) according to the manufacturer’s protocol. Bright-field images were acquired by a phase contrast microscope (EVOS) after overnight incubation at 37 °C.

Mass spectrometry analysis

Total cell lysates from NCMs, MutCMs, and UM cell lines [12, 111] were extracted in RIPA buffer (Thermo Fisher Scientific) containing protease and phosphatase inhibitor cocktails (Roche). Protein preparations were quantified with the BCA Protein Assay Kit (Thermo Fisher Scientific) and submitted to the Proteomics Platform at the RI-MUHC. Samples were treated with TMT-16plex reagents (Thermo Fisher Scientific) according to the manufacturer’s instructions and run on an Orbitrap Fusion instrument (Thermo Fisher Scientific) which was operated in DDA-MS3 mode. Detailed procedures given in Additional file 5. LC–MS/MS data were searched with Proteome Discoverer 2.5 against a Uniprot human database (20,359 entries). False discovery rate was set to 1% for peptide spectrum matches. Two-sample t-test were performed with Perseus 2.0.5.0 after imputation. GO term enrichment and Reactome pathway analyses were performed with DAVID [117] (https://david.ncifcrf.gov/).

Whole transcriptome analysis

Total RNA was extracted using Qiagen RNeasy Plus Micro kit following the manufacturer’s protocol. RNA quality and concentration were evaluated by Nanodrop. PolyA RNASeq was conducted with Illumina NovaSeq PE100-25 M reads. The raw Fastq files were preprocessed using fastp [118]. The trimmed FASTQ files were aligned to human reference GRCh37 using STAR [119]. The aligned reads were processed with HTSEQ [120] to obtain the read counts. Quality control on the reads was performed using FASTQC [121] and RNASEQC [122]. Differential expression analysis was conducted using the Deseq2 [123] R package. Differentially expressed genes (DEGs) were selected with a Log₂(fold change) threshold of 1 and a P value lower than 0.05. KEGG pathway analyses were performed with DAVID (https://david.ncifcrf.gov/).

Statistical analyses

Graphs and statistical analyses were performed using Excel, Prism GraphPad, Python, and R programming. All in vitro experiments were performed for at least 4 donors. Data were analyzed using Student’s t-test or one-way ANOVA with multiple comparisons. Spearman correlation analyses were performed to measure the degree of association between eye donor age or the time length from death to cell isolation and cell cpd. P < 0.05 was considered statistically significant. Mass spectrometry for proteomic profiling was performed on NCMs derived from n = 4 donors, MutCMs derived from n = 1 MutCM, and n = 3 UM cell lines. Transcriptome analysis was performed on NCMs and MutCMs derived from the same donor and 1 UM cell line (92.1), each with n = 3 technical replicates. Statistical analyses for omics data were described in their corresponding method sections.