Allele‐Specific GNAQ Q209 Mutations Reveal Context‐Dependent Oncogenic Programs in Uveal Melanoma

ABSTRACT

Uveal melanoma (UM) is driven by activating GNAQ/GNA11 mutations, predominantly at the hotspot Q209 residue. However, the functional impact of distinct GNAQ Q209 substitutions on downstream signaling and malignant progression remains unclear. Using immune‐competent, multistep UM mouse models combining eye‐restricted mutant GNAQ with BAP1 loss and MYC activation, we directly compared GNAQ^Q209L and GNAQ^Q209R in an otherwise matched genetic context. Tumors arising from either allele shared similar histology and cell‐state composition. However, GNAQ^Q209L tumors progressed faster and were associated with reduced survival. In contrast, GNAQ^Q209R tumor‐derived cell lines exhibited greater ex vivo fitness (proliferation, clonogenicity, and migration), whereas GNAQ^Q209L‐derived lines were relatively fragile in 2D culture. At the molecular level, GNAQ^Q209R tumors preferentially engaged the FAK‐YAP axis in primary lesions and derived cell lines. Together, these results support an allele‐dependent, context‐specific trade‐off in which higher output GNAQ^Q209L favors rapid in vivo expansion, while GNAQ^Q209R promotes adaptability, potentially via FAK‐YAP signaling, nominating allele‐stratified therapeutic opportunities.

Highlights

• Matched UM models reveal distinct oncogenic properties of GNAQ Q209L and Q209R mutations.

• Q209L accelerates in vivo tumor progression, whereas Q209R promotes greater in vitro fitness.

• Preferential FAK‐YAP activation in Q209R reveals an allele‐associated signaling distinction.

Using matched immune‐competent, multistep uveal melanoma models, we show that GNAQ Q209L and Q209R mutations are not functionally equivalent. Q209L drives faster in vivo progression but reduces in vitro proliferation and migration, whereas Q209R promotes greater in vitro fitness and preferential FAK‐YAP activation, revealing an allele‐dependent, context‐specific trade‐off between tumor expansion and adaptability.

Uveal melanoma (UM) is the most common primary intraocular malignancy in adults and remains a major clinical challenge because metastatic relapse is frequent and deadly. Up to ~50% of patients ultimately develop distant metastases (most commonly to the liver), and survival after metastatic diagnosis is typically ~1 year, with limited effective systemic options (Jager et al. 2020). Genetically, UM is distinct from cutaneous melanoma: it is initiated by mutually exclusive activating mutations in GNAQ or GNA11 (often affecting Q209), while progression and metastatic risk stratification are strongly associated with secondary, largely mutually exclusive alterations in BAP1, SF3B1, or EIF1AX, alongside recurrent copy‐number changes (e.g., monosomy 3; gain of 8q/6p; loss of 1p/6q/8p) (Smit et al. 2020).

We have developed a multistep, immune‐competent genetic mouse model that recapitulates key genetic and phenotypic features of UM by enabling eye‐restricted, physiologic expression of mutant GNAQ, followed by stepwise BAP1 loss and MYC activation. This platform produces intraocular tumors with the potential to disseminate to the liver, captures malignant‐state plasticity (melanocytic and neural crest‐like programs), and preserves a tumor‐microenvironment resembling human UM, including immunosuppressive macrophages and exhausted T cells (Xu et al. 2026).

A major unresolved question is whether distinct codon 209 substitutions in oncogenic GNAQ are functionally interchangeable. In human UM, Q209L and Q209P predominate, whereas Q209R is present in a smaller subset of cases (~2.5%) (Carvajal et al. 2014). This raises the question of whether the Q209L and Q209P mutations are specifically selected for due to their stronger oncogenic potential, and if so, whether the Q209R mutant GNAQ protein differentially activates downstream signaling pathways compared to GNAQ harboring Q209L or Q209P mutations. To address this question, we generated a genetically engineered GNAQ^G209R mouse model akin to our previously published GNAQ^G209L model. This enabled the direct comparison of the Q209L versus Q209R mutations in an otherwise matched genetic context that includes BAP1 loss and MYC activation. Using these mouse models and cell lines derived from them, we performed a controlled test of allele‐specific signaling and phenotypic consequences across in vivo and ex vivo settings.

We generated Gnaq^CA; Myc^CKI; Bap1^CKO mice expressing either GNAQ^Q209L or GNAQ^Q209R. In vivo, Q209L mice exhibited significantly accelerated disease progression and reduced overall survival compared with Q209R mice (Figure 1A), suggesting stronger oncogenic activity for GNAQ^Q209L. Histopathologic evaluation showed that tumors arising in response to activation of either Gnaq mutant allele were broadly similar in architecture, expanding within the posterior segment/subretinal compartment as predominantly poorly pigmented epithelioid sheets with round nuclei (Figure 1B,C). PAS staining showed minimal positivity in tumor cells but complex extracellular matrix patterns in both genotypes, consistent with the extravascular matrix loops characteristic of a uveal melanoma phenotype (Figure 1D,E). Multiplex immunofluorescence (mIF) demonstrated intratumoral heterogeneity with intermingled GLI2⁺ (neural crest‐like) and TYRP1⁺/Melan‐A⁺ (melanocytic) populations. Quantification of individual tumors (n = 2 per genotype) showed that Melan‐A positivity was higher in Q209L tumors than in Q209R tumors (74% vs. 41%), whereas TYRP1 positivity was similar between genotypes (14% vs. 12%). GLI2 positivity showed substantial inter‐tumor variability, particularly among Q209L tumors, and no clear allele‐associated pattern was evident (Figure 1F,G). Thus, while Q209L and Q209R tumors share similar histologic features, the Q209L allele drives a more aggressive phenotype in vivo.

FIGURE 1.

Q209L drives faster UM progression in vivo. (A) Kaplan–Meier overall survival curve of Gnaq^CA; Bap1^CKO; Myc^CKI mice after lentiviral administration (Xu et al. 2026) (n = 37 for GNAQ^Q209L cohort and n = 25 for GNAQ^Q209R cohort; log‐rank test). (B, C) Representative images of H&E staining of intraocular tumors from Gnaq^CA; Bap1^CKO; Myc^CKI mice harboring GNAQ^Q209R (B) or GNAQ^Q209L (C). (D, E) Representative images of Periodic acid–Schiff (PAS) staining of tumors from Gnaq^CA; Bap1^CKO; Myc^CKI mice harboring GNAQ^Q209R (D) or GNAQ^Q209L (E). Tumor cells displayed minimal PAS positivity, while extravascular matrix loops were highlighted. (F, G) Representative images (F) and quantification (G) of multiplex IF (mIHC) showing Gli2, Melan‐A, and Tyrp1 expression in unpigmented tumors from Gnaq^CA; Bap1^CKO; Myc^CKI mice. Quantification was performed separately for two independent tumors from each genotype. Individual points represent independent tumors. Data are presented descriptively without formal statistical comparison. (H) Schematic diagram of orthotopic transplant via suprachoroidal injection (left), metastatic transplant via tail vein injection (middle), and liver metastatic transplant via intrasplenic injection followed by splenectomy (right) of Gnaq^CA; Bap1^CKO; Myc^CKI primary tumor cells in NSG mice. Resulting transplant tumors were dissociated to generate the indicated mouse UM cell lines. (I) PCR amplification and Sanger sequencing of Gnaq exon 5 from cDNA derived from the indicated mouse UM cell lines confirmed expression of the Q209L and Q209R mutations.

To establish experimental models for mechanistic and functional studies, primary ocular tumors were dissociated to single‐cell suspensions and inoculated into NSG mice by suprachoroidal (orthotopic), tail vein, or intrasplenic injection (Figure 1H). We successfully established orthotopic tumor‐derived cell lines (QBM12.1 [Q209R], QBM12.3 [Q209R], QBM24 [Q209R], QBM45 [Q209L], QBM47 [Q209L]) from suprachoroidal transplants, as well as metastatic lines (QBM12L [Q209R], QBM12L2 [Q209R], QBM51L [Q209L], QBM53L [Q209L]) from liver metastases by tail vein and/or intrasplenic inoculation (Figure 1H). Sanger sequencing of Gnaq exon 5 confirmed the retention of the expected c.626A > T (Q209L) or c.626A > G (Q209R) mutation in all derived lines (Figure 1I).

To define intracellular signaling differences associated with these allele‐specific phenotypes, we profiled downstream pathway activity in tumor‐derived lines. GNAQ expression and BAP1 loss were confirmed by immunoblot (Figure 2A). Melan‐A expression generally tended to be higher in orthotopic tumor‐derived lines than metastasis‐derived lines, although heterogeneity was observed across individual lines (Figure 2A). Melan‐A expression in derived cell lines did not fully mirror primary tumors, likely reflecting lineage‐state shifts during transplantation or ex vivo culture. Although total ERK levels were stable, p‐ERK levels varied across lines without a uniform shift based on the Gnaq mutation (Figure 2A). Intriguingly, in Q209L lines, p‐ERK level was significantly higher in lines derived from liver metastasis compared to orthotopic transplants. Consistently, FRA1, a downstream effector of the MAPK pathway and driver of cutaneous melanoma metastasis (Xu et al. 2024, 2025), was notably increased in metastasis‐derived Q209L lines (Figure 2A). Quantification of downstream effectors revealed divergent signaling programs. p‐AKT (S473) showed a nonsignificant decrease in Q209L lines relative to Q209R (p = 0.35; Figure 2B), while p‐4EBP1 was significantly reduced in Q209L (p = 0.009; Figure 2C), suggesting attenuated mTOR pathway output in this context. Moreover, we observed significantly higher p‐FAK (Y397) expression in Q209R cells (p = 0.021; Figure 2D,E). Q209R lines also showed lower levels of inhibitory p‐YAP (S127) compared to the expression observed in Q209L lines (p = 0.071; Figure 2D,F). Intriguingly, in Q209L lines, the inhibitory p‐YAP (S127) level was significantly higher in lines derived from liver metastasis compared to orthotopic transplants. Immunohistochemistry demonstrated significantly greater nuclear YAP localization in Q209R primary tumors compared with Q209L (Figure 2G,H). This difference persisted in metastatic settings, where Q209R liver lesions displayed higher overall YAP staining and increased nuclear localization relative to Q209L lesions (Figure 2I,J). Collectively, these data indicate that Q209R tumors preferentially engage mTOR signaling and the FAK‐YAP axis.

FIGURE 2.

Q209R preferentially engages FAK‐YAP signaling. (A) Western blot analysis of GNAQ, BAP1, Melan‐A, p‐ERK1/2 (Thr202/Tyr204), ERK1/2, FRA1, p‐AKT (Ser473), AKT, p‐4EBP1 (Ser65), and 4EBP1 expression in mouse UM cell lines harboring GNAQ^Q209R or GNAQ^Q209L. HSP90 was used as an internal loading control. (B) Quantification of the relative ratio between p‐AKT (Ser473) and AKT in Q209R and Q209L UM cell lines from (A). (C) Quantification of the relative ratio between p‐4EBP1 (Ser65) and 4EBP1 in Q209R and Q209L UM cell lines from (A). (D) Western blot analysis of p‐FAK (Tyr397), FAK, p‐YAP (Ser127), and YAP in mouse UM cell lines harboring GNAQ^Q209R or GNAQ^Q209L. HSP90 was used as an internal loading control. (E) Quantification of the relative ratio between p‐FAK (Tyr397) and FAK in Q209R and Q209L UM cell lines from (D). (F) Quantification of the relative ratio between p‐YAP (Ser127) and YAP in Q209R and Q209L UM cell lines from (D). Orthotopic tumor‐derived cell lines (QBM12.1 [Q209R], QBM12.3 [Q209R], QBM24 [Q209R], QBM45 [Q209L], QBM47 [Q209L]) and liver metastatic lines (QBM12L [Q209R], QBM12L2 [Q209R], QBM51L [Q209L], QBM53L [Q209L]) were used in A–F. (G) Representative IHC images showing YAP expression in primary tumors from Gnaq^CA; Bap1^CKO; Myc^CKI mice harboring either GNAQ^Q209R or GNAQ^Q209L. (H) Quantification of the nuclear‐to‐cytoplasmic YAP intensity ratio (H‐score) among primary tumors from Gnaq^CA; Bap1^CKO; Myc^CKI mice harboring either GNAQ^Q209R or GNAQ^Q209L. (I) Representative IHC images showing YAP expression in liver metastasis derived from intrasplenic transplantation of mouse UM cell lines harboring GNAQ^Q209R or GNAQ^Q209L. (J) Quantification of nuclear YAP intensity (H‐score) among liver metastasis derived from intrasplenic transplantation of mouse UM cell lines harboring GNAQ^Q209R or GNAQ^Q209L.

We next asked whether these signaling differences translated into divergent in vitro behaviors. In growth assays, Q209R‐derived lines exhibited significantly faster proliferation compared to Q209L cells (Figure 3A). Within the Q209R group, no significant difference in proliferation was observed between orthotopic tumor‐derived and metastatic cells; however, within the Q209L group, the metastasis‐derived QBM51L line proliferated markedly faster than its orthotopic tumor‐derived counterparts (Figure 3A). In clonogenic assays, Q209R lines demonstrated robust colony‐forming capacity, while Q209L lines formed few to no colonies under identical conditions (Figure 3B). Transwell assays further revealed that metastasis‐derived lines generally possessed higher migration capacity than orthotopic tumor‐derived lines, with overall migration being higher in the Q209R background (Figure 3C). These findings suggest an inverse relationship between in vivo aggressiveness and in vitro fitness for the Q209L allele, potentially reflecting specific microenvironmental requirements for Q209L tumor growth that are absent in standard 2D culture.

FIGURE 3.

Q209R confers greater ex vivo fitness. (A) Proliferation as measured by relative confluence of the indicated mouse UM cell lines harboring GNAQ^Q209R or GNAQ^Q209L (n = 5 biological replicates). (B) Quantification of percentage occupied area of low‐density colony formation assays of the indicated mouse UM cell lines harboring GNAQ^Q209R or GNAQ^Q209L (n = 4 biological replicates). Representative images are shown on the right. (C) Quantification of cell numbers of transwell migration assays of the indicated mouse UM cell lines harboring GNAQ^Q209R or GNAQ^Q209L (n = 5 biological replicates). Representative images are shown on the right. Orthotopic tumor‐derived cell lines (QBM12.1 [Q209R], QBM12.3 [Q209R], QBM24 [Q209R], QBM45 [Q209L], QBM47 [Q209L]) and liver metastatic lines (QBM12L [Q209R], QBM12L2 [Q209R], QBM51L [Q209L]) were used. Statistical analyses were performed using unpaired two‐tailed t‐tests. (D–I), Endpoint IVIS images of NSG mice after intrasplenic injection of QBM51L (D), QBM47 (F), and QBM12L (H) cells. Representative H&E staining of liver metastasis generated by QBM51L (E), QBM47 (G), and QBM12L (I) cells at indicated endpoint days. For intrasplenic transplantation, 500,000 cells were injected into NSG mice, with two mice used for each indicated cell line, and liver metastatic progression was monitored by IVIS imaging.

To determine whether UM‐derived cell lines differ in liver metastatic outgrowth in vivo, we performed intrasplenic transplantation into NSG mice. A total of 500,000 cells of QBM12L, QBM47, or QBM51L cells were inoculated into the spleen, followed by splenectomy, and metastatic progression was monitored by IVIS imaging. Among the Q209L lines, the liver metastasis‐derived QBM51L cells reached end point rapidly, with strong liver bioluminescence and histologically confirmed liver metastases at Day 16 (Figure 3D,E). In contrast, the orthotopic tumor‐derived Q209L line QBM47 reached endpoint later, at Day 32, indicating that metastatic derivation is associated with enhanced liver colonization capacity within the Q209L background (Figure 3F,G). Notably, QBM51L also reached end point earlier than the liver metastasis‐derived Q209R line QBM12L, which showed liver metastatic burden at endpoint Day 42 (Figure 3H,I). H&E staining of endpoint livers confirmed metastatic lesions composed of poorly pigmented epithelioid tumor cells growing within the liver parenchyma, broadly resembling the histology of the parental intraocular UM tumors. Together, these data suggest that metastatic outgrowth in this model is influenced by both derivation history and GNAQ Q209 allele context, with QBM51L displaying the most rapid liver metastatic progression among the tested lines.

In summary, we directly compared the two codon 209 mutant Gnaq alleles Q209L and Q209R in the context of BAP1 loss and MYC activation. Consistent with allele‐specific biology, we observed an intriguing context dependence: Q209L drives more rapid tumor progression and liver metastatic outgrowth in vivo but yields cells with reduced robustness in standard 2D culture, whereas Q209R tumors progress more slowly yet generate lines with greater proliferative/adaptive capacity ex vivo (Table 1).

TABLE 1.

Summary of mouse UM cell lines.

Cell line	12.1	12.3	24	12L	12L2	45	47	51L
Genotype	Q209R	Q209R	Q209R	Q209R	Q209R	Q209L	Q209L	Q209L
Derivation site	Orthotopic	Orthotopic	Orthotopic	Liver Met.	Liver Met.	Orthotopic	Orthotopic	Liver Met.
In vitro growth	Moderate	Fast	Fast	Fast	Fast	Slow	Slow	Moderate
In vitro migration	Moderate	Low	Moderate	High	High	Low	Low	Moderate
Liver metastasis	ND	ND	ND	Yes	ND	ND	Yes	Yes

Note: The table summarizes GNAQ Q209 genotype, derivation site, in vitro growth, migration capacity, and liver‐metastasis capacity of the indicated mouse UM cell lines.

Abbreviation: ND, not determined in the current analysis.

A plausible mechanistic interpretation is that the two alleles impose graded differences in signaling amplitude and downstream pathway engagement, which in turn shape an evolutionary life‐history trade‐off between rapid proliferation and environmental adaptability (Aktipis et al. 2013). Interestingly, a similar phenomenon has been noted in cutaneous melanoma where the Braf^V600E; Pten^−/− driver genotype combination promotes rapid tumor growth, yet primary cells derived from this model enter a senescence state when grown in cell culture ex vivo (Bok et al. 2022, 2020). In endothelial models, Q209R is a gain‐of‐function allele but weaker than Q209L, suggesting the two mutations engage largely overlapping programs that differ mainly in activation magnitude (Galeffi et al. 2022). In our UM‐derived system, we propose that this higher‐output Q209L state drives a “fast” program that maximizes immediate expansion within permissive in vivo niches, as further illustrated by the rapid liver outgrowth of the metastasis‐derived cell line, but may require dependence on microenvironmental support, rendering cells fragile upon ex vivo culture. Conversely, the weaker Q209R driver appears to favor a “slow” trajectory in vivo, potentially allowing time to accumulate adaptive stress‐response circuitry or mutations. Consistent with this, Q209R lines exhibit stronger FAK‐YAP axis output (higher p‐FAK and nuclear YAP), an ECM‐sensing survival program associated with anchorage independence, stress response, and drug‐tolerant states (Franklin et al. 2023). We speculate that this intrinsic FAK‐YAP activity functions as a survival buffer during the transition from tissue to plastic, enabling Q209R cells to trade speed for robustness and broader environmental resilience (Zhao et al. 2012). This framework also suggests allele‐specific therapeutic opportunities: Q209R tumors may be preferentially sensitive to FAK inhibition and/or YAP‐TEAD blockade, whereas Q209L disease may be better targeted by strategies that disrupt tumor‐microenvironment interactions rather than relying solely on tumor‐intrinsic pathway inhibition.

1. Methods

1.1. Animals

The conditional GNAQ^Q209L mouse allele and multistep uveal melanoma (UM) model were described previously (Xu et al. 2026). The GNAQ^Q209R allele was generated using an analogous targeting strategy. Gnaq^CA; Cas9^CKI; Myc^CKI mice received suprachoroidal delivery of lentivirus expressing Cre recombinase, firefly luciferase, and sgRNA targeting Bap1 to induce ocular tumors, as previously described (Xu et al. 2026). NOD/SCID gamma (NSG) mice were used for transplantation studies. All animal experiments were performed under an approved Institutional Animal Care and Use Committee protocol and in accordance with institutional guidelines.

1.2. Histology, IHC, and Multiplex IF

Tumor tissues were formalin‐fixed and paraffin‐embedded, and H&E, periodic acid–Schiff (PAS), immunohistochemistry (IHC), and multiplex immunofluorescence (mIF) were performed as previously described (Xu et al. 2026). Multiplex staining for GLI2 (1:100, Invitrogen, PA1‐28838, RRID:AB_2111904), Melan‐A (1:100, Abcam, ab210546, RRID:AB_2889292), and TYRP1 (1:100, Abcam, ab83774, RRID:AB_2211142) was performed using the automated Opal platform and quantified by digital image analysis. Tissues were segmented into individual cells based on DAPI staining, and marker‐specific positivity thresholds were applied to determine the percentage of positive cells and cells co‐expressing multiple markers. YAP IHC (1:100, Cell Signaling Technology, 4912, RRID:AB_2218911) was quantified by digital image analysis based on staining intensity and subcellular localization. Nuclear‐to‐cytoplasmic YAP intensity ratios were calculated for primary tumors, whereas nuclear YAP staining intensity was quantified in liver metastases.

1.3. Mouse Uveal Melanoma Cell Isolation and Culture

Mouse UM cell lines were established from orthotopic transplant tumors or experimental liver metastases as previously described (Xu et al. 2026). Briefly, tumor tissues were dissected and enzymatically dissociated using the Tumor Dissociation Kit (Miltenyi Biotec, RRID:SCR_020285) at 37°C for 30–45 min. Cell suspensions were washed with PBS containing 2% FBS, centrifuged at 300 × g for 5 min, and used for transplantation or establishment of tumor‐derived cell lines. Gnaq exon 5 was analyzed by Sanger sequencing to confirm retention of the expected Q209L or Q209R mutation. For initial establishment from freshly dissociated tumors or recovery from cryopreservation, T25 flasks were precoated with 1 mL DMEM containing 10% FBS, 2.5% Collagen I (Corning, cat. #354236), and 2.5% Matrigel (Corning, cat. #356237) for 1 h at 37°C. The coating medium was removed before cells were plated in DMEM supplemented with 10% FBS. Cells typically attached within 1–2 days but could initially retain a rounded or spheroid morphology; approximately 5–7 days were generally required for cells to fully spread and acquire their characteristic spindle morphology. After proliferation was established, cells were routinely passaged in 60‐ or 100‐mm culture dishes. Culture dishes were initially precoated as above, with Collagen I and Matrigel concentrations gradually reduced to 1.5% each as the cells adapted to ex vivo growth. Fully adapted cell lines could subsequently be maintained without extracellular matrix coating. Cells were cultured at 37°C in a humidified atmosphere containing 5% CO₂.

1.4. Western Blot Analysis

Western blotting was performed as previously described (Xu et al. 2025, 2026). Briefly, 15 μg of protein was separated on NuPAGE 4%–12% gels, transferred to nitrocellulose membranes, blocked in 5% nonfat milk in TBST, and incubated with primary antibodies overnight at 4°C. HRP‐conjugated secondary antibodies and chemiluminescent detection were used for visualization. Antibodies against GNAQ (CST, #14373, RRID:AB_2665457), BAP1 (Abcam, ab255611), Melan‐A (Abcam, ab210546, RRID:AB_2889292), ERK1/2 (CST, #4695, RRID:AB_390779), p‐ERK1/2 (Thr202/Tyr204) (CST, #4370, RRID:AB_2315112), FRA1 (Abcam, ab252421, RRID:AB_2904513), AKT (CST, #4691, RRID:AB_915783), p‐AKT (Ser473) (CST, #4060, RRID:AB_2315049), p‐FAK (Tyr397) (CST, #8556, RRID:AB_10891442), FAK (CST, #3285, RRID:AB_2269034), 4E‐BP1 (CST, #9452, RRID:AB_331692), p‐4E‐BP1 (Ser65; CST, #9451, RRID:AB_330947), YAP (CST, #4912, RRID:AB_2218911), and p‐YAP (Ser127) (CST, #4911, RRID:AB_2218913) were used at 1:1000 as indicated. HSP90 (1:1000, CST, #4874, RRID:AB_2121214) served as a loading control.

1.5. Proliferation Assay

Cell proliferation was measured by live‐cell imaging as previously described (Xu et al. 2024, 2025). Briefly, 3000 cells were seeded in 96‐well plates and, after attachment, imaged daily using a Cellcyte‐X live‐cell analyzer (ECHO). Cell confluence was quantified using the instrument software.

1.6. Clonal Formation Assay

Low‐density colony formation assays were performed as previously described (Xu et al. 2024). Briefly, 1500 cells were seeded at low density in 12‐well plates and cultured for 8 days. Colonies were fixed with 4% paraformaldehyde, stained with 0.5% crystal violet, and the occupied area was quantified using ImageJ.

1.7. Transwell Migration Assay

Transwell migration assays were performed using previously established methods (Xu et al. 2024, 2025). Briefly, 30,000 cells were seeded into the upper chamber of the transwell (Costar, 3422) in serum‐free DMEM, with 15% FBS‐containing DMEM in the lower chamber as a chemoattractant. After 48 h, cells that had migrated to the lower surface of the membrane were fixed, stained with crystal violet, imaged, and quantified.

1.8. Intrasplenic Injection and Splenectomy

Experimental liver metastasis was generated by intrasplenic transplantation as previously described (Xu et al. 2026). Briefly, 500,000 UM cells in 100 μL sterile PBS were slowly injected into the splenic parenchyma of anesthetized NSG mice, followed by immediate splenectomy. Liver metastatic progression was monitored longitudinally by IVIS bioluminescence imaging, and livers were harvested at endpoint for histologic analysis.

1.9. Statistical Analysis

Statistical analyses were performed using GraphPad Prism. Survival was compared using the log‐rank test, and comparisons between two groups were performed using unpaired two‐tailed t‐tests. p < 0.05 was considered statistically significant.

Author Contributions

Vinesh Jarajapu: investigation, validation, writing – original draft, methodology, formal analysis, visualization, data curation. Sathya Neelature Sriramareddy: investigation, validation. Nicol Mecozzi: investigation, validation. Kaizhen Wang: investigation, validation. Florian A. Karreth: conceptualization, supervision, funding acquisition, project administration, resources, writing – original draft, writing – review and editing, software. Xiaonan Xu: conceptualization, methodology, data curation, supervision, validation, investigation, funding acquisition, visualization, writing – original draft, writing – review and editing, project administration, formal analysis, resources.

Funding

N.M. was supported by an F99 award from the NCI/NIH (1F99CA305524–01). F.A.K. was supported by grants from the Melanoma Research Alliance (935329) and the CDMRP Melanoma Research Program (ME230182). X.X. was supported by a Melanoma Research Foundation Career Development Award (1068914) and a Miles for Moffitt Postdoctoral Milestone Award. This work was also supported by the Gene Targeting Core, Small Animal Imaging Lab Core, Comparative Medicine Core, Advanced Analytical and Digital Laboratory Core, and Analytical Microscopy Core, which are funded in part by Moffitt Cancer Center Support Grant (P30CA076292).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.