Figure S3. Arf6Q67L-HA expression in tumors from BrafCA;Cdkn2af/f;Ptenf/f mice. qRT-PCR detection of Arf6 Q67L-HA from fresh, frozen primary tumor fragments. Bars represent the absolute difference in crossing threshold (Ct) between Arf6 Q67L-HA and Gapdh. The Ct threshold that defines detection was established above the background signal for controls in both the BrafCA;Cdkn2af/f and the BrafCA;Cdkn2af/f ;Ptenf/f cohorts (see also Figure S1). All mouse tumor negative by qRT-PCR showed detectable HA-tagged ARF6Q67L by immunohistochemistry (not shown, see Figure S1 for example).

Galpha and MEK1/2 inhibition in combination with Baf A1 and CQ in MP41 cells

Figure S4. Primary tumor incidence, growth, survival and metastasis in Ptenf/f vs. Arf6Q67L mice. (a) Increased tumor incidence (% = # of mice that developed a tumor / # TVA positive, injected mice) in Ptenf/f vs. Arf6Q67L cohorts. Fisher's exact test, two-tailed, α <0.05. (b) Primary tumor onset, unpaired t test with Welch's correction. (c) Primary tumor growth in Ptenf/f vs. Arf6Q67L mice upon tumor onset. Bars = SD. Two-way ANOVA, a=0.05. (d) Overall survival upon tumor onset, log-rank (Mantel-Cox) test. (e) Comparison of spontaneous metastatic volume in each mouse cohort (see also schematic, Figure 1g). ANOVA (p < 0.0001) with Kruskal-Wallis multiple comparisons test (p values shown). ns=not significant. Identical data from the Control and Arf6Q67L cohorts are also shown in Figure 1c with a slightly different Y axis spacing.

Figure S6. ARF6-GDP inhibits PI3K-AKT activation and is necessary for AKT activation. (a-b) Quantification of western blots shown in Figure 5c-d. Reproducible reduction in pAKT by Myc-ARF6T27N over expression in A375 cells (n=4), A2058 cells (n=4), and HEY-T30 cells (n=5). Reduced pAKT with siRNA knockdown of ARF6 in SK-MEL-147 (n=5), A2058 (n=4), and CACL (n=4) and HEY-T30 (n=6). Graphs show individual data points normalized to control along with geometric means, 95% CI, ratio paired t test. (c) Phosphokinase array images (left) and quantification of signal by densitometry (right, histogram), n=1. Expression of ARF6T27N in A2058 cells preferentially reduced pAKT and the AKT substrate PRAS40 over most other proteins in the array.

Validation of hepatic tumor model

YAP requirement in MP41 cells

Supplementary Materials and Methods

Figure S5: ARF6-GTP induces morphologic changes in mouse melanoma cells. Bright field images (100x magnification) of primary mouse melanoma cell lines derived from control BrafCA;Cdkn2af/f (5588) or BrafCA;Cdkn2af/f + Arf6Q67L (6431, 6455) mice. ARF6-GTP induces elongation/spindling of tumor cells relative to the more polygonal-shaped controls. Scale bars = 400um.

Figure S2. Primary tumor incidence and growth in BrafCA;Cdkn2af/f mice is not altered by ARF6-GTP. (a) Primary tumor incidence (% = # of mice that developed a tumor / # TVA positive, injected mice) in control and overlapping Arf6Q67L cohorts. Two Arf6Q67L cohorts, classified according to detectable or undetectable HA-tagged Arf6Q67L expression by qRT-PCR and/or by anti-HA immunohistochemistry (see Figure S1). Cohort A includes all mice injected with RCAS Cre + Arf6Q67L. Cohort B is limited to only those mice from cohort A with detectable Arf6Q67L in tumors. Fisher's exact test, two tailed,a<0.05. b) Primary tumor growth upon tumor onset, bars = SD. (c) Overall survival upon tumor onset, log-rank (Mantel-Cox) test. (d) Metastatic incidence (# mice with metastasis / # mice with primary tumors). (e) Proliferation, BrafCA;Cdkn2af/f mouse melanoma cells. (f). Apoptosis, BrafCA;Cdkn2af/f mouse melanoma cells. (g) Representative images of Ki67 immunostains, 200x magnification, scale bars = 50um.

Response of uveal melanoma cell lines to MEK1/2 inhibition

Figure S1. Detection of HA-tagged ARF6Q67L in BrafCA;Cdkn2af/f melanoma. a) Anti-HA immunostain of FFPE sections of primary tumors with variable levels of detection of ARF6Q67L -HA. The strongest staining was observed in tumors with the highest level of mRNA detected by qRT-PCR (b). Absent staining of FFPE tumors was observed in tumors with low (6456) and undetectable (6457) levels of mRNA by qRT-PCR of frozen tumors. Scale bars = 20mm, 400x magnification. Control = BrafCA;Cdkn2af/f (Cre-only injection). b) qRT-PCR of Arf6 Q67L-HA from fresh, frozen primary tumor fragments. Bars represent the absolute difference in crossing threshold (Ct) between Arf6Q67L-HA and Gapdh. The Ct threshold that defines detection was established above the background signal for controls in both the BrafCA;Cdkn2af/f and the BrafCA;Cdkn2af/f ;Ptenf/f cohorts (see Figures 1g and S3). With rare exception, positive tumors demonstrated Arf6 Q67L-HA expression levels lower than Gapdh.

Immunofluorescence analysis of autophagy in OMM2.5 cells with or without trametinib

Table S1: ARF6 pathways and related genes.

Trametinib in combination with macropinocytosis or mitophagy inhibition

Figure S7. Low expression of ARF GAP genes correlates with reduced overall survival. (a) Individual gene plots of four ARF6 GAPs (ACAP1, ADAP1, ARAP2, SMAP2), an ARF1/ARF5 GAP (AGAP2) and one predicted ARF GAP (AGFG2). n = 439 patients. (b) Individual gene plots in Stage III specific cohort, n=142 patients. Log-rank (Mantel-Cox) test.

YAP activation in OMM1 cells

Bioluminescent images of hepatic tumors

Preventing or effectively treating metastatic uveal melanoma (UM) is critical because it occurs in about half of patients and confers a very poor prognosis. There is emerging evidence that hepatocyte growth factor (HGF) and insulin-like growth factor 1 (IGF-1) promote metastasis and contribute to the striking metastatic hepatotropism observed in UM metastasis. However, the molecular mechanisms by which HGF and IGF-1 promote UM liver metastasis have not been elucidated. ASAP1 , which acts as an effector for the small GTPase ARF6, is highly expressed in the subset of uveal melanomas most likely to metastasize. Here, we found that HGF and IGF-1 hyperactivate ARF6, leading to its interaction with ASAP1, which then acts as an effector to induce nuclear localization and transcriptional activity of NFAT1. Inhibition of any component of this pathway impairs cellular invasiveness. Additionally, knocking down ASAP1 or inhibiting NFAT signaling reduces metastasis in a xenograft mouse model of UM. The discovery of this signaling pathway represents not only an advancement in our understanding of the biology of uveal melanoma metastasis but also identifies a novel pathway that could be targeted to treat or prevent metastatic uveal melanoma.

PURPOSE. Acute retinal necrosis ( ARN) is a herpesvirus infection of the retina with blinding complications. In this study, we sought to create a reproducible mouse model of ARN that mimics human disease to better understand innate immunity within the retina during virus infection. METHODS. C57Bl/6J wild type (WT) and type I interferon receptor-deficient (IFNAR-/-) mice were infected with varying amounts of herpes simplex virus type 1 (HSV-1) via subretinal injection. Viral titers, optical coherence tomography (OCT) and fundus photography, the development of encephalitis, and ocular histopathology were scored and compared between groups of WT and IFNAR-/- mice. RESULTS. The retina of WT mice could be readily infected with HSV-1 via subretinal injection resulting in retinal whitening and full-thickness necrosis as determined by in vivo imaging and histopathology. In IFNAR-/- mice, HSV-1-induced retinal pathology was significantly worse when compared with WT mice, and viral titers were significantly elevated within two days after infection and persisted to day 5 after infection within the retina. These results were also observed in the brain where there were significantly higher viral titers and frequency of encephalitis in IFNAR-/- when compared to WT mice. CONCLUSIONS. Collectively, these findings show that our new mouse model of ARN mimics human disease and can be used to study innate immunity within the retina. We conclude that type I interferons are critical in containing HSV-1 locally within retinal tissues and prohibiting spread into the brain.