miR-1908, miR-199a-3p, and miR-199a-5p Promote Metastatic Invasion, Endothelial Recruitment, and Angiogenesis

We next sought to determine the cellular mechanisms underlying miRNA regulation of metastasis. Importantly, over-expression of each miRNA reduced cell proliferation (Figure S2A) and did not increase tumor growth (Figure 2A), indicating that the pro-metastatic effects of miR-1908 and miR-199a are not secondary to tumor growth promotion or enhanced proliferation.

Open in a separate window

Figure 2

miR-1908, miR-199a-3p, and miR-199a-5p Display Cell-Autonomous/Non-Cell-Autonomous Pro-Metastatic Roles in Melanoma

(A) Primary tumor growth by 1×10⁶ MeWo cells over-expressing the miR-199a or miR-1908 precursor hairpins or expressing an shCTRL following subcutaneous injection into mice. n=4–6.(B) Matrigel invasion by 1×10⁵ MeWo cells over-expressing miR-199a or miR-1908 or expressing an shCTRL was quantified by counting the number of cells that invaded into the basal side of matrigel-coated trans-well inserts after 24 hours. n=7. (C–D) Matrigel invasion by MeWo-LM2 (C) and A375-LM3 (D) cells with miR-199a-3p KD, miR-199a-5p KD, miR-1908 KD, or a control KD. n=6–8. (E) Trans-well recruitment of 1×10⁵ human umbilical vein endothelial cells (HUVEC) by 5×10⁴ MeWo cells over-expressing miR-199a or miR-1908 or control MeWo cells. Endothelial recruitment capacity was measured by counting the number of HUVECs that migrated to the basal side of each trans-well insert after 16 hours. n=7. (F–G) Endothelial recruitment by MeWo-LM2 (F) and A375-LM3 (G) cells inhibited for miR-199a-3p, miR-199a-5p, miR-1908, or a control sequence. n=6–10. (H–I) Cumulative fraction plots depicting the endothelial cell (H) or dextran perfusion (I) density distributions for metastatic nodules formed by 2×10⁵ MeWo-LM2 cells with miR-199a-3p KD, miR-199a-5p KD, miR-1908 KD, or a control KD. Lung sections were double-stained for vimentin and MECA-32 (H) or vimentin and intravenously injected dextran (I). n=211, 30, 138, and 39 nodules for control KD, m199a3p KD, m199a5p KD, and m1908 KD, respectively (H); n=30, 18, 38, and 40 nodules for control KD, m199a3p KD, m199a5p KD, and m1908 KD, respectively (I). All data are represented as mean ± SEM. Scale bar, 100 µm. See also Figure S2

We next asked whether these miRNAs regulate cell invasion, a key metastatic phenotype. Over-expression of either miR-199a or miR-1908 enhanced the ability of parental MeWo cells to invade through matrigel (Figure 2B), while metastatic LM2 cells, which express higher levels of these miRNAs, displayed significantly greater matrigel invasion capacity (Figure S2B). Conversely, individual inhibition of miR-199a-3p, miR-199a-5p, or miR-1908 decreased the invasive capacity of MeWo-LM2 (Figure 2C) and A375-LM3 (Figure 2D) metastatic melanoma cell derivatives.

Given the robust effects of these miRNAs on metastatic progression, we examined whether they regulate additional pro-metastatic phenotypes. While over-expression of miR-199a or miR-1908 did not modulate melanoma cell adhesion to endothelial cells (Figure S2C), resistance to anoikis (Figure S2D), survival during serum starvation (Figure S2E), or colony formation (Figure S2F), over-expression of each miRNA dramatically enhanced the ability of parental MeWo cells to recruit endothelial cells in trans-well endothelial recruitment assays (Figure 2E). Consistent with this, metastatic MeWo-LM2 cells were more efficient at recruiting endothelial cells relative to their parental line (Figure S2G). Furthermore, inhibition of miR-199a-3p, miR-199a-5p, or miR-1908 in the metastatic MeWo-LM2 (Figure 2F) as well as A375-LM3 (Figure 2G) cells suppressed endothelial recruitment.

To determine whether endogenous miR-199a-3p, miR-199a-5p, and miR-1908 mediate endothelial recruitment by metastatic cells in vivo, we examined metastatic endothelial density in lung nodules formed by melanoma cells over-expressing or knocked-down for each miRNA. Consistent with the robust miRNA-dependent endothelial recruitment phenotype, each of the three miRNAs was both required and sufficient for enhanced metastatic nodule endothelial content (Figures 2H and S2H–I). Remarkably, individual inhibition of each miRNA robustly suppressed metastatic nodule perfusion by roughly 5-fold following dextran injection (Figures 2I and S2J), indicating that each of these miRNAs also promotes functional metastatic angiogenesis. Our findings reveal miR-199a-3p, miR-199a-5p, and miR-1908 as necessary and sufficient for enhanced invasion, metastatic endothelial recruitment (MER), and angiogenesis during melanoma progression.

miR-1908, miR-199a-3p, and miR-199a-5p Convergently and Cooperatively Target ApoE and DNAJA4

We next employed a systematic and global approach to identify the direct molecular targets of these miRNAs. Since miR-1908, miR-199a-3p, and miR-199a-5p mediate the same sets of in vitro and in vivo phenotypes, and since miR-199a-5p and miR-199a-3p arise from the same precursor hairpin, we hypothesized that the pro-metastatic phenotypes of these miRNAs may emerge through silencing of common target genes. We thus performed transcriptomic profiling of melanoma cells in the context of both loss- and gain-of-function for each miRNA (Figure S3A). This analysis, followed by qRT-PCR validation, revealed two genes (ApoE and DNAJA4) to be repressed by both exogenous and endogenous miR-199a and miR-1908 and also present at lower levels in the metastatic LM2 derivatives—which display elevated levels of these miRNAs (Figures 3A and S3B–D).

Open in a separate window

Figure 3

Identification of ApoE and DNAJA4 as Direct Target Genes of miR-199a and miR-1908

(A) Heat-map depicting the mRNA expression levels of ApoE and DNAJA4, measured by qRT-PCR, in MeWo parental cells expressing an shCTRL or over-expressing miR-199a or miR-1908 and in MeWo-LM2 cells. Color-map illustrates standard deviations change from the mean for each gene’s expression. (B–C) Heterologous luciferase reporter assays measuring the expression of wild type or miRNA target site mutant ApoE and DNAJA4 CDS’s and 3’UTR’s in MeWo cells over-expressing miR-199a or miR-1908 or in control MeWo cells (B) and in MeWo-LM2 cells with miR-199a-3p KD, miR-199a-5p KD, miR-1908 KD, or a control KD (C). n=3–4. (D) Schematic of an experimentally derived model depicting the convergent targeting of the CDS’s and 3’UTR’s of ApoE and DNAJA4 by miR-199a-3p, miR-199a-5p, and miR-1908. (E) Luciferase activity of wild type and miRNA target site mutant ApoE and DNAJA4 3’UTR/CDS luciferase constructs transfected into MeWo parental and MeWo-LM2 cells. n=4. (F) Linear regression analyses of the expression levels of each gene (ApoE and DNAJA4) and the aggregate expression levels of the three miRNAs (average of the individual miRNA expression values) in multiple MeWo and A375 metastatic derivatives and their parental lines. n=12. (G–H) Trans-well matrigel invasion (G) and endothelial recruitment (H) by parental MeWo cells expressing short hairpin RNAs (shRNAs) targeting ApoE (shAPOE), DNAJA4 (shDNAJA4), or a control sequence (shCTRL). n=6–8. All data are represented as mean ± SEM. Scale bar, 100 µm. See also Figure S3.

To determine whether ApoE and DNAJA4 are directly targeted by miR-1908, miR-199a-3p, and miR-199a-5p, we examined the effects of each miRNA on the expression of its putative targets through heterologous luciferase reporter assays. Interestingly, both miR-199a and miR-1908 individually repressed the expression levels of the 3’ untranslated regions (UTR) and coding sequences (CDS) of both ApoE and DNAJA4. Consistent with direct targeting, mutating the miRNA complementary sequences on each target abrogated miRNA-mediated regulation (Figure 3B). In a direct test of endogenous targeting, individual miRNA inhibition in metastatic LM2 cells led to an increased target-driven luciferase activity (Figure 3C) that was abrogated upon mutating the miRNA target sites (Figure S3E), revealing ApoE to be directly targeted by miR-1908 and miR-199a-5p and DNAJA4 to be directly targeted by all three miRNAs (Figure 3D). Importantly, the CDS’s and 3’UTR’s of both genes were expressed at lower levels in highly metastatic MeWo-LM2 cells, which exhibit physiologically higher endogenous levels of the three regulatory miRNAs (Figure 3E). These findings establish ApoE and DNAJA4 as direct targets of these miRNAs in melanoma. In agreement with this, we found a significant anti-correlation between the endogenous levels of both ApoE and DNAJA4 and the aggregate expression levels of the three miRNAs across our collection of parental and metastatic lines (Figure 3F). Consistent with the phenotypes of their miRNA regulators, both ApoE and DNAJA4 were required and sufficient for suppression of cell invasion and endothelial recruitment–phenotypes that drive metastatic progression (Figures 3G–H and S3F–I).

ApoE and DNAJA4 Mediate miR-199a- and miR-1908-Dependent Metastatic Invasion, Endothelial Recruitment, and Colonization

To determine whether ApoE and DNAJA4 are the direct biological effectors downstream of miR-199a and miR-1908, we examined whether these two target genes are epistatic to each miRNA. Consistent with direct and epistatic miRNA/target gene interactions, knock-down of ApoE or DNAJA4 in the setting of miRNA inhibition (Figures S4A–D) significantly occluded the suppressed invasion (Figures 4A and 4C) and endothelial recruitment phenotypes (Figures 4B and 4D) seen upon silencing of each miRNA and fully rescued the dramatically suppressed metastatic colonization phenotype resulting from miRNA inhibition (Figures 4E–F, and S4E). Conversely, over-expression of the coding regions of ApoE or DNAJA4 in cells over-expressing miR-1908 (Figures 4G–H and S4F) or miR-199a (Figures S4G–I) was sufficient to suppress cell invasion and endothelial recruitment, as well as miRNA-dependent metastatic colonization (Figure S4J). Importantly, downregulation of ApoE and DNAJA4 was also required for miRNA-induced enhancement of cell invasion and endothelial recruitment by the highly metastatic A375-LM3 line (Figures 4I–J and S4K). Notably, endogenous ApoE and DNAJA4 were also found to suppress metastatic endothelial recruitment in vivo (Figure 4K). These effects of ApoE and DNAJA4 were both epistatic to and dependent on the miRNAs. Our findings implicate ApoE and DNAJA4 as direct downstream effectors of miRNA-dependent metastatic invasion, colonization, and endothelial recruitment phenotypes in melanoma.

Open in a separate window

Figure 4

ApoE and DNAJA4 Epistatically Interact with miR-199a and miR-1908 in Regulating Metastatic Invasion, Endothelial Recruitment, and Colonization

(A–D) MeWo-LM2 cells expressing a control shRNA or shRNAs targeting ApoE or DNAJA4 in the context of miR-1908 KD (A–B) or miR-199a-5p KD (C–D) were subjected to the trans-well invasion (A and C) and endothelial recruitment (B and D) assays. n=6–8. (E–F) Bioluminescence imaging plots and H&E-stained lungs corresponding to lung metastasis by 1×10⁵ MeWo-LM2 cells expressing a control shRNA or shRNAs targeting ApoE or DNAJA4 in the setting of miR-1908 KD (E) or miR-199a-5p KD (F). n=5. (G–H) MeWo cells over-expressing the coding regions of ApoE or DNAJA4 or expressing a control vector in the context of miR-1908 over-expression were analyzed for the invasion (G) and endothelial recruitment (H) phenotypes. (I–J) A375-LM3 cells expressing a control shRNA or shRNAs targeting ApoE and DNAJA4 were transfected with a cocktail of LNAs targeting miR-199a-3p, miR-199a-5p, and miR-1908 or a control LNA and subjected to the invasion (I) and endothelial recruitment (J) assays. n=4. (K) Cumulative fraction plots of metastatic nodule endothelial content distributions corresponding to nodules formed by 1×10⁵ MeWo-LM2 cells knocked down for ApoE, DNAJA4, or a control sequence in the setting of miR-1908 KD. Thresholded images depicting MECA-32 signal are shown in lower panels. n=39 nodules (shCTRL); n=97 (shAPOE¹); n=38 (shAPOE²); n=200 (shDNAJA4¹); n=19 (shDNAJA4²). All data are represented as mean ± SEM. Scale bar, 100 µm. See also Figure S4.

Melanoma Cell-Secreted ApoE Suppresses Invasion and Endothelial Recruitment

ApoE is a secreted factor. As such, we wondered whether melanoma cell-secreted ApoE could suppress invasion and endothelial recruitment. Consistent with this, poorly metastatic parental cells displayed higher levels of secreted ApoE than highly metastatic LM2 cells (Figure 5A). Secreted ApoE levels were also significantly suppressed by endogenous and exogenous miR-199a and miR-1908 (Figures 5B and S5A). Next, inhibiting extracellular ApoE through the use of a neutralizing antibody (1D7) that recognizes its receptor-binding domain enhanced both cell invasion (Figure 5C) and endothelial recruitment (Figure 5D) by parental MeWo cells. Conversely, addition of recombinant human ApoE protein significantly suppressed invasion and endothelial recruitment by metastatic LM2 cells (Figure 5E). Importantly, antibody-mediated neutralization of extracellular ApoE significantly abrogated the suppressed invasion and endothelial recruitment phenotypes seen with inhibition of each miRNA (Figures S5B–C)— consistent with secreted ApoE being downstream of miR-199a and miR-1908. Notably, recombinant ApoE addition did not affect melanoma cell or endothelial cell proliferation (Figures S5D–E) or survival in serum starvation conditions (Figures S5F–G), indicating that suppression of invasion or endothelial recruitment by ApoE is not secondary to decreased proliferation or impaired survival. Our findings reveal melanoma cell-secreted ApoE as a necessary and sufficient suppressor of miRNA-dependent invasion and endothelial recruitment phenotypes in melanoma.

Open in a separate window

Figure 5

Extracellular ApoE Inhibits Invasion, Endothelial Recruitment, and Metastasis, while Genetic Deletion of ApoE Enhances Metastasis

(A–B) Extracellular ApoE levels, quantified by ELISA, in conditioned media from MeWo parental and LM2 cells (A) and in MeWo-LM2 cells silenced for miR-199a-5p, miR-1908, or a control sequence (B). n=3. (C) Matrigel invasion by MeWo cells in response to the ApoE-neutralizing antibody 1D7 (5, 10, 20, or 40 µg/mL) or IgG control (40 µg/mL). n=4–6. (D) Endothelial recruitment by MeWo cells in the presence of 1D7 or control IgG antibody at 40 µg/mL. n=4. (E) The invasion and endothelial recruitment phenotypes were assessed in MeWo-LM2 cells in response to bovine serum albumin (BSA) or recombinant ApoE3 added to the cell media at 100 µM. n=7–10. (F) Extracellular ApoE levels, quantified by ELISA, in conditioned media from MeWo cells expressing shRNAs targeting DNAJA4 or an shCTRL. n=3. (G–H) MeWo cells with shRNA-induced silencing of DNAJA4 were analyzed for the matrigel invasion (G) and endothelial recruitment (H) phenotypes in the presence of either BSA or recombinant ApoE3 at 100 µM. n=4. (I) Array-based ApoE transcript expression levels in nevi (n=9), primary melanomas (n=6), and distal melanoma metastases (n=19). (J) ApoE protein expression determined by blinded immunohistochemical analysis in a tissue microarray (TMA) melanoma progression set (NIH) comprised of primary melanoma lesions (n=66), nodal (n=36), and distal (n=66) melanoma metastases. (K) Bioluminescence quantification and representative ex vivo lung images corresponding to lung metastasis 19 days post-injection of 5×10⁴ B16F10 mouse melanoma cells pre-treated with recombinant ApoE3 (100 µg/mL) or control for 24 hours and intravenously injected into ApoE genetically null or wild type mice. n=12–22 (control) and n=5–7 (ApoE pre-treatment). (L–N) Bioluminescence imaging plots of lung colonization by 4×10⁴ MeWo-LM2 cells (L), 2×10⁵ A375-LM3 cells (M), and 1.5×10⁵ WM-266-4 cells (N) pre-treated with ApoE3 or BSA at 100 µg/mL for 24 hours prior to injection. n=5–8. All data are represented as mean ± SEM. Scale bar, 100 µm. See also Figure S5.

We next wished to investigate the mechanism by which DNAJA4, a poorly characterized heat-shock protein, mediates endothelial recruitment and invasion. Given the phenotypic commonalities displayed by ApoE and DNAJA4, we hypothesized that DNAJA4 may play a regulatory role and enhance ApoE levels. Indeed, knock-down of DNAJA4 reduced both ApoE transcript levels (Figure S5H) as well as secreted ApoE levels (Figure 5F), while DNAJA4 over-expression substantially elevated ApoE transcript expression (Figure S5I). Consistent with DNAJA4 acting upstream of ApoE, addition of recombinant ApoE abrogated the enhanced cell invasion and endothelial recruitment phenotypes resulting from DNAJA4 knock-down (Figures 5G–H), while antibody-mediated neutralization of ApoE significantly occluded the suppression of invasion and endothelial recruitment resulting from DNAJA4 over-expression (Figures S5J–K). These findings reveal DNAJA4 to suppress melanoma invasion and endothelial recruitment through positive regulation of ApoE expression, leading to its increased extracellular levels.

The regulatory convergence of three metastasis-promoting miRNAs and the DNAJA4 gene onto ApoE motivated us to determine whether ApoE expression correlates with human melanoma progression. To this end, we analyzed published array-based expression data for ApoE (Haqq et al., 2005) in nevi, primary, and metastatic lesions. Consistent with a metastasis-suppressive role, ApoE transcript levels were significantly lower in distal organ metastases relative to primary melanoma and nevi lesions (Figure 5I). We also examined ApoE protein expression in a large (N = 168) human tissue-microarray melanoma progression panel. Blinded immunohistochemical analysis revealed that ApoE protein levels were significantly lower in lymph node metastases (P = 0.035) and even further downregulated in distal melanoma metastases (P < 0.0001) relative to primary melanomas (Figure 5J).

Genetic inactivation of ApoE promotes metastasis, while ApoE pre-treatment blocks metastatic initiation

Given the robust suppression of metastatic phenotypes exerted by ApoE and its clinical association with melanoma progression, we next investigated the impact of genetic deletion of systemic ApoE on melanoma progression in an immunocompetent mouse model of melanoma metastasis. Consistent with a major suppressive role for extracellular ApoE in metastasis, B16F10 mouse melanoma cells injected into the circulation exhibited a 10-fold increase in metastatic colonization in ApoE genetically null mice compared to their wild type littermates. Importantly, the metastatic capacity of B16F10 cells was dramatically abrogated by ApoE cell pre-treatment in both wild type (60-fold inhibition) and ApoE-null (460-fold inhibition) genetic contexts (Figure 5K). These findings establish both systemic and cancer-secreted ApoE as a robust suppressor of metastasis by human and mouse melanoma cells.

We next wondered whether enhanced ApoE signaling in melanoma cells could have therapeutic efficacy in preventing melanoma metastasis. We thus pre-incubated melanoma cells with recombinant ApoE or BSA for 24 hours prior to injection into mice. Notably, ApoE pre-treatment robustly suppressed metastatic colonization by the highly aggressive MeWo-LM2 cells by over 300-fold (Figure 5L). Remarkably, ApoE pre-incubation abrogated the metastatic capacity of four additional human melanoma cell lines: A375-LM3 (145-fold, Figure 5M), WM-266-4 (80-fold, Figure 5N), HT-144 (119-fold, Figure S5L), and A2058 (85-fold, Figure S5M).

Extracellular ApoE Divergently Targets Melanoma Cell LRP1 and Endothelial Cell LRP8 Receptors

We next investigated the molecular mechanisms by which ApoE suppresses metastasis. In order to identify the ApoE receptor(s) that mediate(s) invasion, we knocked down all four known ApoE receptors (VLDLR, LRP1, LRP8, and LDLR) in melanoma cells. Interestingly, knock-down of LRP1, but not the other ApoE receptors, abolished ApoE-induced suppression of invasion (Figures 6A and S6A), while LRP1 knock-down in the setting of miRNA silencing rescued the suppressed invasion phenotype resulting from miRNA inhibition (Figures 6B and S6B) and significantly enhanced in vivo metastatic colonization by MeWo-LM2 cells depleted for miR-1908 (Figures 6C and S6C). These findings reveal LRP1 to be a downstream mediator of miRNA/ApoE-dependent melanoma invasion and metastatic colonization phenotypes.

Open in a separate window

Figure 6

Distinct Melanoma and Endothelial Cell Receptors Mediate the Effects of ApoE on Invasion and Endothelial Recruitment

(A) Trans-well invasion by MeWo-LM2 cells transfected with short interfering RNAs (siRNAs) targeting LDLR, VLDLR, LRP8, LRP1, or a control siRNA (siCTRL) in the presence of either BSA (100 µM) or recombinant ApoE3 (100 µM). n=4–7. (B) MeWo-LM2 cells with miR-1908 KD or a control KD were transfected with siRNAs targeting LRP1 or siCTRL and subjected to the matrigel invasion assay. n=4. (C) Lung colonization by 1×10⁵ MeWo-LM2 cells transfected with siRNAs targeting LRP1 or siCTRL in the setting of miR-1908 KD. n=5. (D) Trans-well recruitment of endothelial cells pre-incubated with BSA or recombinant ApoE3 at 100 µM for 24 hours by MeWo-LM2 cells. n=3–4. (E) Recruitment of HUVECs transfected with siRNAs targeting LDLR, VLDLR, LRP1, LRP8, or siCTRL by MeWo-LM2 cells with miR-1908 KD or control KD. n=4–12. (F) Trans-well migration by HUVECs in the presence of IgG or 1D7 antibody at 40 µg/mL. n=6–8. (G) Trans-well migration by endothelial cells transfected with siRNAs targeting LRP8 or siCTRL in the presence of BSA or recombinant ApoE3 at 100 µM. n=6–7. (H) Trans-well migration along an ApoE matrigel gradient by endothelial cells transfected with siRNAs targeting LRP8 or siCTRL. n=6–8. (I) Endothelial recruitment into subcutaneously implanted matrigel plugs containing BSA (10 µg/mL), VEGF (400 ng/mL) + BSA (10 µg/mL), or VEGF (400 ng/mL) + ApoE3 (10 µg/mL). The number of MECA-32 positive endothelial cells was quantified after three days by immunohistochemical detection. n=3–6. (J) Endothelial cell content, determined by staining for MECA-32, within lung metastatic nodules formed 19 days following injection of 5×10⁴ B16F10 cells into wild type or ApoE genetically null mice. n=17–20. All data are represented as mean ± SEM. Scale bar, 100 µm. See also Figure S6.

While ApoE’s invasion phenotype reflects its cell-autonomous effect on melanoma cells, the endothelial recruitment phenotype suggests a non-cell-autonomous role of cancer-expressed ApoE on endothelial cells. Consistent with this, pre-treatment of endothelial cells with ApoE significantly reduced their ability to migrate towards highly metastatic cancer cells (Figure 6D). We next wished to identify the ApoE receptor(s) on endothelial cells that mediate(s) the endothelial recruitment phenotype. Interestingly, unlike for cancer cell invasion, knock-down of endothelial LRP8 receptors selectively and significantly abrogated the suppressed endothelial recruitment phenotype induced by miRNA silencing (Figures 6E and S6D–F). These findings are consistent with the LRP8 receptor being the endothelial mediator of miRNA/ApoE-dependent effects on endothelial recruitment. We next investigated whether ApoE might also regulate general endothelial migration in a cancer cell-free system. Indeed, antibody neutralization of ApoE, which is present in endothelial cell media, significantly enhanced endothelial migration (Figure 6F), while recombinant ApoE was sufficient to inhibit endothelial migration in a trans-well assay (Figure 6G) and a gradient-based migration assay (Figure 6H) in an endothelial cell LRP8 receptor-dependent manner. Importantly, ApoE dramatically suppressed by more than 40-fold VEGF-induced in vivo endothelial recruitment into subcutaneous matrigel plugs (Figure 6I). Consistent with the requirement and sufficiency of ApoE for suppressing endothelial migration, genetically null ApoE mice displayed higher blood vessel densities within their lung metastatic nodules formed by B16F10 mouse melanoma cells compared to their wild type littermates (Figure 6J), implicating systemic ApoE as a suppressor of metastatic angiogenesis.

Cell Culture

Various cell lines were obtained from ATCC and maintained in standard conditions. Cell line information, miRNA and gene knock-down/over-expression studies in cell lines, and in vitro functional assays are described in Extended Experimental Procedures.

Microarray Hybridization

In order to identify miRNAs deregulated across highly metastatic derivatives, small RNAs were enriched from total RNA derived from MeWo and A375 cell lines and profiled by LC Sciences. For identification of miRNA gene targets, total RNA from MeWo cell lines was labeled and hybridized onto Illumina HT-12 v3 Expression BeadChip arrays by The Rockefeller University genomics core facility. Microarray-based expression findings were validated by qRT-PCR using TaqMan miRNA expression assays (Applied Biosystems) or SYBR® Green-based detection of gene expression (Invitrogen). See Extended Experimental Procedures.

Analysis of miRNA Expression in Human Melanoma Skin Lesions

All human clinical samples were obtained, processed, and analyzed in accordance with IRB guidelines. Total RNA was extracted from paraffin-embedded tissue sections of primary melanoma skin lesions previously resected from patients at MSKCC, and specific miRNA expression levels were analyzed in a blinded manner using TaqMan miRNA Assays. See Extended Experimental Procedures.

ApoE ELISA

Conditioned cancer cell media was prepared by incubating 70%-confluent cells in 0.2% FBS serum starvation DMEM-based media for 24 hours. ApoE levels in conditioned media were quantified using the APOE ELISA kit (Innovative Research).

Histochemistry

For gross macroscopic metastatic nodule visualization, 5-µm-thick lung tissue sections were H&E stained. For metastatic endothelial content and perfusion analyses, lung sections were double-stained with antibodies against MECA-32 (Developmental Studies Hybridoma Bank, University of Iowa) and human vimentin (Vector Laboratories) or biotinylated dextran (Vector Laboratories) and vimentin, respectively. ApoE protein expression in the melanoma tissue microarray progression set (NIH) was detected by staining with D6E10 anti-ApoE antibody (Abcam). See Extended Experimental Procedures.

We thank the members of the Tavazoie Laboratory for thoughtful comments on previous versions of this manuscript and Masoud Tavazoie for insightful discussions regarding melanoma. ApoE null mice were generously provided by Jan Breslow. We thank Jordana Ray-Kirton for tremendous help with clinical sample procurement and Yves Marcel for kindly providing ApoE neutralization antibody. S.F.T. was supported by the Rita Allen and the Elizabeth and Vincent Mayer foundations and a DOD Era of Hope Scholar award. This work was also generously supported by the Melanoma Research Alliance and the Melanoma Research Foundation.