Arylsulfonamide 64B Inhibits Hypoxia/HIF-Induced Expression of c-Met and CXCR4 and Reduces Primary Tumor Growth and Metastasis of Uveal Melanoma

Increased expression of c-Met and CXCR4 correlates with poor prognosis and liver metastasis in patients with UM

To identify hypoxia-induced genes whose expression might underlie metastasis and UM patient survival, we performed comparative gene expression analysis on primary melanoma samples of patients who had or not developed metastasis using 2 publicly available databases [GSE27831 (ref. 39) and GSE22138 (ref. 40)]. 168 genes were found upregulated in common in the metastatic groups, and pathway analysis revealed that 32 of them were highly interconnected. Outcome analysis using an independent dataset (TCGA UM; ref. 4) evidenced that gene expression of 22 of these was correlated with poor outcome, 2 of which (CXCR4 and MET) are hypoxia-inducible (Supplementary Fig. S2A and S2B and Supplementary Table S1). Kaplan–Meier analyses on 2 datasets (GSE22138 and TCGA) showed that elevated mRNA expression of CXCR4 and MET is strongly correlated with patient survival (Fig. 1C; Supplementary Fig. S2C). In contrast, expression levels of genes encoding HGF and CXCL12, the respective ligands for c-Met and CXCR4 and other main hypoxia response genes showed no difference between groups (Fig. 1A; Supplementary Fig. S2D). Unbiased hierarchical clustering further showed that most tumors with low expression of both MET and CXCR4 mRNAs did not metastasize, whereas those with elevated MET metastasize (Supplementary Fig. S2E). Immunohistochemistry on a UM tissue array confirmed that expression levels of CXCR4, c-Met, and HIF-1α were significantly higher in samples collected from metastatic (n = 17) as compared with non-metastatic UM patients (n = 14; Fig. 1D and E; Supplementary Fig. S3 and Supplementary Table S2). Survival analysis of matching patients evidenced that those with metastasis had worse outcome (Fig. 1F; Supplementary Table S1). Altogether, these results show that c-Met and CXCR4 expression are useful markers for prediction of UM metastatic progression and prognostic for patient outcome. They also suggest that both c-Met and CXCR4 may play an important role in the metastatic process in UM.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Elevated expression of CXCR4 and MET correlates with poor prognosis and metastasis in patients with UM. A, Heatmap comparison of CXCR4, CXCL12, and MET, HGF mRNA expression in human patients with UM with or without metastasis (n = 46/group). B, Quantification of (A). C, Kaplan–Meier survival curves for patients with UM (GSE22138) with high (n = 35) or low (n = 28) MET (Left) and CXCR4 (Right) mRNA levels. D, Representative immunofluorescence staining of CXCR4, c-Met and HIF-1α expression in ocular tumor tissue array from UM patients with or without metastasis. DIC, differential interference contrast. E, Quantification of positive tumor cells in (D). F, Survival time for metastatic patients and follow-up time for non-metastatic patients used for tissue array; **, P < 0.01; ***, P < 0.001.

c-Met/CXCR4 are hypoxia/HIF-1 target genes in UM cells and their expression is inhibited by arylsulfonamide 64B, a novel HIF inhibitor

To examine whether tumor hypoxia could contribute to upregulated c-Met and CXCR4 expression in UM, we exposed cultured UM cells to 1% hypoxia. Hypoxia led to HIF-1α stabilization and increased expression of both c-Met and CXCR4 mRNA and protein, responses that were inhibited by HIF pathway inhibitor 64B (Fig. 2A–C; Supplementary Fig. S4A). 64B dose-dependently inhibited hypoxia-induced transcription of a luciferase reporter gene under the control of a promoter driven by hypoxia-response elements in UM cells (Fig. 2D). This effect was not due to cytotoxicity or direct inhibition of luciferase activity as 64B had no effect on cell viability of Mel290 cells under the culture conditions used in this assay, and did not affect luciferase expression driven by a constitutive CMV promoter (data not shown). 64B was not cytotoxic to normal cells either, as tested in 3 human fibroblast cell cultures (Supplementary Fig. S4B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

64B blocks hypoxia-induced expression of c-Met and CXCR4 and UM cell migration. A, Quantification of mRNA expression level of CXCR4 and MET by QRT-PCR in Mel290-V6R UM cells under normoxia (21% O₂) or hypoxia (1% O₂) with or without treatment with 64B (2 μmol/L) for 24 hours. B and C, Western blot analysis for indicated proteins from Mel290-V6R cells cultured under normoxia or hypoxia with or without 64B treatment (at indicated doses) for 24 hours. D, Dose–response inhibition of hypoxia-inducible gene expression with 64B in Mel290-V6R cells stably expressing an HRE-driven luciferase reporter gene. n = 18 independent experiments carried out in triplicate. E, Tumor cell scratch wound assay. Percentage of wound closure was measured in 64B-treated versus control from 92.1 (28 hours) and Mel290 (16 hours) cells grown under hypoxia (top). Representative pictures of wound closure in 92.1 UM cells after 28 hours. of hypoxia (bottom). Cells were pre-treated for 48 hours with control (Ctrl.), cMet or CXCR4 siRNAs for 48 hours. with or without treatment with 64B (2 μmol/L; 1 hour pre-treatment), and silencing of cMet and CXCR4 cell surface expression was verified by FACS (right). Representative data of 3 independent experiments; *, P < 0.05; **, P < 0.01; ***, P < 0.001.

To further test whether 64B-mediated inhibition of c-Met/CXCR4 expression had any effect on UM cell motility, a prerequisite for metastasis, we performed scratch wound closure assays. 64B caused a significant decrease in hypoxia-induced wound closure in the assay (Fig. 2E), although had little effect under normoxic conditions (Supplementary Fig. S4C). Treatment of the cells with siRNA for CXCR4 and c-Met also inhibited cell migration under hypoxia, supporting their role in this process (Fig. 2E). Combination with 64B slightly increased inhibition, likely because the siRNA effect was incomplete (Fig. 2E), or due to additional 64B-inhibited factors. These results show that hypoxia enhances UM cell migration at least in part through c-Met and CXCR4 expression and that this response can be potently antagonized by HIF transcription inhibitor 64B.

64B blocks HIF transcriptional activity by disrupting the interaction between HIF-1α and p300, independently of Factor inhibiting HIF-1 (FIH-1)

We next aimed to define the mechanism underlying 64B inhibition of HIF transcriptional activity in UM cells. We first tested if it alters hypoxia-activated HIF-1α protein stability. A 24 hours dose–response experiment showed that 64B at concentrations up to 16 μmol/L did not disrupt hypoxia-mediated stabilization of HIF-1α (Fig. 3A). A time course experiment showed that HIF-1α is stabilized by hypoxia as early as 4h, remains stable up to 48 hours, and is not affected by 64B (Fig. 3B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

64B prevents the binding of p300 and CBP to HIF-1α. A, Western blot of indicated protein expression after dose-response with 64B in Mel290-V6R cells after 24 hours of hypoxia. B, Western blot of indicated protein expression after time-course of hypoxia ± 64B (2 μmol/L) treatment. C–E, Co-immunoprecipitation (co-IP) of cell extracts from Mel290-V6R and 92.1 UM cells grown for 24 hours. under normoxia or hypoxia ± 64B (2 μmol/L) treatment. Note reduction in binding between HIF-1α and p300/CBP cofactors, but not between HIF-1α and HIF-1β. Western blot analysis of an aliquot (4%) of the cell extracts was used as a control for input protein (input). F, Western blot analysis of indicated protein expression in 92.1 cells under normoxia and hypoxia ± FIH silencing. G, Effect of siRNA-mediated silencing of FIH, on 64B's ability to inhibit p300 -HIF-1α interaction in hypoxic 92.1 cells as measured by co-IP.

Under hypoxia, HIF-1α heterodimerizes with HIF-1β and forms a transcriptional complex with p300/CBP cofactors in the nucleus. Co-immunoprecipitation (co-IP) experiments showed 64B could disrupt the interaction between HIF-1α and both p300 and CBP paralogs, but not with HIF-1β (Fig. 3C–E). Levels of hypoxia-induced HIF-1α, and constitutive expression of p300, and CBP were unaffected by 64B (Fig. 3 C–E; see input fraction).

The interaction of HIF with p300/CBP co-factors is regulated by the oxygen-dependent hydroxylation of asparagine 803 by Factor Inhibiting HIF-1 (FIH) on the C-terminus of HIF-1α (42). To probe for FIH role in 64B action, we silenced it by siRNA. FIH knockdown did not abrogate 64B disruption of HIF-1α/p300 interaction, suggesting that 64B does not act through FIH (Fig. 3F and G). Altogether, these results show that 64B can disrupt the recruitment of p300/CBP co-factors to the HIF transcription complex in a FIH-independent manner.

64B can bind to p300 and directly interferes with the interaction between HIF-1α and p300

To further examine whether 64B disrupted HIF-1α/p300 interaction through direct binding of p300, we used the cellular thermal shift assay (38). As a negative control, we used BW-HIF13, a structural analog of 64B in which the cyclobutane is replaced with a cycloheptyl ring (Fig. 4A) that does not inhibit hypoxia-activated luciferase gene expression (Fig. 4B). Cell extracts of UM cells were prepared and then treated with 64B (2 μmol/L) or BW-HIF13 in vitro for 30 minutes. Thereafter, aliquots (20 μg protein each) were prepared and each tube was heated to a different temperature, then centrifuged to eliminate any heat-denatured precipitated proteins, and the supernatants examined by Western blot for the protein of interest. In this assay, protein binding to a small molecule of interest will thermo-stabilize the protein and shift its denaturation temperature. This assay revealed that 64B could increase the thermal stability of p300 by 8°C (Fig. 4C), supporting a direct interaction between 64B and p300. In contrast, BW-HIF-13 did not elicit any shift (Fig. 4C), indicating 64B structural specificity to bind p300.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

64B directly binds to p300 and interferes with p300 recruitment to HIF-1 activated MET and CXCR4 genes. A, Chemical formula of 64B and inactive analog BW-HIF13. B, Dose-response of 64B and BW-HIF13 on hypoxia-induced luciferase expression in Mel290-V6R cells with a stably integrated HRE-luciferase reporter; (n = 3). C, Cellular thermal shift assays on 92.1 cell extracts treated with 64B or BW-HIF-13 (2 μmol/L for 2 hours). Western blot (Left) of soluble fraction of individually heated cell extract aliquots (44°C–72°C) show 64B (but not BW-HIF-13) extends the thermal stability of p300. Quantification of p300 bands (Right; n = 3). D, ChIP analysis shows hypoxia-induced binding of p300 on the CXCR4 and MET promoter regions is inhibited by 64B (2 μmol/L) in 92.1 cells. P300 recruitment to the control gene DAPK1 was not affected. Data are representative of 3 independent experiments.

These findings suggested that 64B might prevent CXCR4 and MET gene activation under hypoxia by preventing p300 co-factor recruitment to the HIF transcriptional complex on the gene promoters in the nucleus. To test this, we used ChIP and found that 64B reduced p300 binding to both genes under hypoxia (Fig. 4D). In contrast, 64B did not prevent recruitment of p300 to the DAPK1 gene, which is activated by transcription factor C/EBPβ, independently of hypoxia/HIF. P300 binds to HIF-1α through its CH1 domain (43), whereas it binds to C/EBPβ via its CH3 domain (44). These findings suggest that 64B may exhibit CH1 domain specificity in its ability to interfere with p300/CBP transcription factor interactions.

64B inhibits primary tumor growth, liver metastasis formation, and prolongs survival in orthotopic mice models of UM

Next, we examined the in vivo effects of 64B on primary and metastatic UM growth. To establish a treatment dose, we carried out a pilot acute toxicity study with a single intraperitoneal injection of 64B at 100, 200, and 400 mg/kg. This did not result in any mice death, significant loss of mice weight or other abnormal behavioral changes (data not shown). On the basis of this, we selected a dose of 120 mg/kg for testing in 2 complementary orthotopic UM models in mice. Both models involve the injection of melanoma cells in the sub-uveal area of the mouse eye where they form melanomas (Fig. 5A), which can spontaneously metastasize to the liver, mimicking human UM disease process. The first is an immunocompetent model where mouse skin melanoma cells (B16LS9) are injected in syngeneic C57BL/6 mice. The second uses human UM cells (92.1) injected in immunocompromised athymic nude mice. After eye inoculation, mice are treated intraperitoneally with 64B or vehicle control 5 d/wk. After primary tumor growth, tumor-bearing eyes are removed at 7 to 9 days, fixed, and stained with hematoxylin and eosin (H&E) to evaluate tumor burden. Mice are continually treated with 64B till they are sacrificed to observe metastasis development in the liver. In the B16LS9 model, 64B treatment strongly suppressed eye tumor formation with an approximately 60% decrease in tumor size 7 days post-inoculation (Fig. 5B). At day 28, mice were terminated and the formation of liver micro-metastases quantified in liver sections. Mice treated with 64B showed an approximately 50% reduction in metastatic foci (Fig. 5C). To further examine whether 64B could prolong mice survival, we conducted a separate experiment where after enucleation on day 7, mice were kept until they reached IACUC criteria for termination. 64B significantly extended mice lifespan by approximately 30% (Fig. 5D). Moreover, treatment of 64B (60 mg/kg/d) in mice was accompanied by slight body weight gain (Fig. 5E). Similarly, in the 92.1 model we observed approximately 90% decrease of eye tumor size 7 days post-inoculation (Fig. 5F), and 64B significantly extended mice lifespan (Fig. 5G). Altogether, these findings demonstrate that 64B reduces primary and metastatic growth of melanoma in the eye and prolongs mice life span.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

64B inhibits primary eye tumor growth, liver metastasis, and prolongs survival in UM mouse models. A, Schematic showing timeline and procedure for the animal experiments. B–E, B16LS9 mouse skin melanoma cells (2 × 10⁶/eye) were injected in the sub-uveal area of C57BL/6 mice eyes where they form melanomas in the uvea. One day after eye inoculation (day 1), mice (8/group) were treated with either vehicle control or 60 mg/kg/i.p. of 64B 5 days per week. On day 7, tumor-bearing eyes were enucleated, fixed, and stained to evaluate tumor burden (B). Representative H&E sections are shown (Left) and quantification of eye tumor (*) size was performed in 4 randomly selected mice/group (right). Mice were sacrificed on day 28 and number of liver metastases counted (8 mice/group; C). In a repeat experiment with identical design, mice were kept under treatment after eye enucleation and terminated when reaching IACUC criteria to measure survival by Kaplan–Meier analysis (15 mice/group; P = 0.0012, log-rank test, D). E. Weight of mice used in (D). F and G, 92.1 human UM cells (2 × 10⁶/mouse) were injected in the suprachoroidal space of 3 ½-month-old female nu/nu mice (Envigo Corp; Strain: Hsd: Athymic Nude-Foxn1^nu; 15 mice/group). On day 1, mice were treated with either vehicle control or 120 mg/kg/i.p. of 64B 5 days per week. On day 7, eyes were removed and evaluated for tumor burden. Representative H&E-stained eyes are shown (left) and tumor size (*) quantified in 6 mice per group (F). Survival rate of mice (15 mice/group, P < 0.0001, log-rank test, G). **, P < 0.01; ***, P < 0.001.

64B reduces circulating UM cells, antagonizes HIF-1–induced activation of c-Met and CXCR4 expression, and inhibits the metastasis of UM cells to the liver

To test the efficacy of 64B in the inhibition of circulating tumor cells and liver metastasis formation, we conducted another independent experiment with the 92.1 model (Supplementary Fig. S5A–S5D). GFP-labeled 92.1 cells were used for cell tracking, and mice were injected with pimonidazole, a hypoxia marker before enucleation or sacrifice. Eyes were enucleated on day 9, and sections showed reduced tumor size in 64B treatment group (Fig. 6A and B). Immunofluorescence analysis of eye tumor sections showed CXCR4 and c-Met expression was strongly reduced in 64B-treated mice compared with vehicle controls (Fig. 6A and C; Supplementary Fig. S5E). Staining for pimonidazole revealed that vehicle control primary tumors were hypoxic, whereas control tumors had little evidence for hypoxia, likely due to their smaller size (Fig. 6D). Flow cytometry of blood collected on day 9 showed that mice treated with 64B exhibited reduced amounts of circulating tumor cells (Fig. 6E). Mice were sacrificed on day 56, and livers from mice treated with 64B had a reduced metastatic burden as evidenced by fewer micrometastases (Supplementary Fig. S5F). Immunofluorescence staining of Pimo and HIF-1α in liver sections evidenced that micro-metastases were hypoxic in both control and 64B (Fig. 6F). Prolonged systemic treatment with 64B was well tolerated as mice did not lose weight (Supplementary Fig. S5A), and examination of blood cell populations at day 56 did not evidence any alteration (Supplementary Fig. S5D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

64B reduced UM tumor growth, circulating tumor cells and metastasis and inhibited c-Met and CXCR4 expression in hypoxic/HIF-1α–expressing tumors in eye and liver in a UM mouse model. 92.1-GFP UM cells (2 × 10⁶/eye) were injected in 6-week-old female nu/nu mice [JAX Laboratory; Strains: Outbred athymic nude; J:NU (007850; 10 mice/group)]. On day 1, mice received either vehicle control or 120 mg/kg/i.p. of 64B 5 days per week. Eyes were enucleated on day 9. A, Representative H&E pictures (left) and immunofluorescence analysis for CXCR4 and c-Met (right) in tumor-bearing eyes. B, Quantification of tumor size on 6 representative mice/group). C, Quantification of CXCR4- and cMet-positive cells from (A). D, HIF-1α and Pimo immunofluorescence staining in representative control and 64B-treated tumor-bearing eyes (5 mice/group). E, Representative flow cytometry analysis (left) and quantification (right) of circulating tumor cells (GFP-labeled) in mice blood collected on day 9 (6 mice/group). F, Representative pictures of HIF-1α and Pimo immunofluorescence staining in livers from vehicle and 64B-treated mice (7 mice/group); *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Finally, we sought to determine whether the reduced metastatic load in 64B-treated mice was primarily due to a reduction in primary eye tumor burden, or was also related to tumor cell survival in blood circulation and colonization of liver. We directly injected tumor cells (92.1 and Mel290) in the tail vein and counted liver metastasis formation 5 weeks postinjection. We observed approximately 10-fold reduction in liver metastases in 64B-treated mice versus vehicle-treated controls (Supplementary Fig. S6A–S6B). The treatment was well tolerated, with no weight loss (Supplementary Fig. S6C).