Loss of Phd2 cooperates with BRAFV600E to drive melanomagenesis

Abstract

Prolyl hydroxylase domain protein 2 (PHD2) is a well-known master oxygen sensor. However, the role of PHD2 in tumor initiation remains controversial. We find that during the transition of human nevi to melanoma, the expression of PHD2 protein is significantly decreased and lower expression PHD2 in melanoma is associated with worse clinical outcome. Knockdown of PHD2 leads to elevated Akt phosphorylation in human melanocytes. Mice with conditional melanocyte-specific expression of Phd2^lox/lox (Tyr::CreER;Phd2^lox/lox) fail to develop pigmented lesions. However, deletion of Phd2 in combination with expression of BRaf^V600E in melanocytes (Tyr::CreER;Phd2^lox/lox;BRaf^CA) leads to the development of melanoma with 100% penetrance and frequent lymph node metastasis. Analysis of tumor tissues using reverse phase protein arrays demonstrates that Phd2 deletion activates the AKT-mTOR-S6 signaling axis in the recovered tumors. These data indicate that PHD2 is capable of suppressing tumor initiation largely mediated through inhibiting of the Akt-mTOR signaling pathway in the melanocyte lineage.

Fig. 1

PHD2 expression and function in human melanocytes and melanomas. a IHC staining of PHD2 in 126 nevus and 266 melanoma specimens. A melanoma progression tissue microarray (TMA) was stained with the anti-PHD2 antibody and PHD2 staining positive and negative cases were calculated. Representative PHD2 positive (upper panel) and negative cases (lower panel) are shown. Bars indicate 100 µm. b Percentage of cases with PHD2 expression. There was significantly less percentage of melanoma cases with PHD2 expression than that of nevi; *p < 0.01. c. TCGA SKCM patients were sorted into PHD2-high and PHD2-low groups according to PHD2 mRNA expression levels (top 5% versus bottom 5%). d PHD2 expression and survival curve. The Kaplan–Meier survival curves showed that the PHD2-high group have a significantly improved patient survival than PHD2-low group (log rank p value = 0.00643). e PHD2 knockdown induces Akt phosphorylation. Immortalized human melanocytes (hTERT/p53DD/Cdc24R24C-BRAFV600E) were transfected with two independent PHD2 shRNAs. Cells transfected with shGFP were included as a negative control. Western blots were performed using the indicated antibodies. f Reintroduction of PHD2 in PHD2^−/− MEFs reduces Akt phosphorylation. PHD2^−/− MEFs (left lane) were transfected with wild-type PHD2 (lane adjacent to the left lane). PHD2-positive MEFs which express endogenous PHD2 were used as a control (right lane). g DFO or DMOG treatment. Immortalized human melanocytes were transfected with control vectors or EGLN knockdown vectors. These cells were than treated with DFO or DMOG. h Overexpressed HIF-2α in immortalized human melanocytes. i The melanoma-derived PHD2-P317 mutation inhibits PHD2 binding to HIF-1α. Biotin-labeled HIF1α-ODD peptides were incubated with WCLs derived from HEK293T cells transfected with the indicated PHD2 constructs (beads as negative control). Binding of PHD2 to HIF1α-ODD is abolished with the P317 mutation. j The PHD2-P317 mutation inhibits PHD2 and HIF-1α interaction. Unlike WT-PHD2, P317S-PHD2 is largely impaired in inhibiting HRE reporter activities detected by the Dual-Luciferase Reporter Assay System. The error bars indicate s.d.

Fig. 2

Phd2 loss cooperates with BRaf^V600E to induce malignant melanoma in mice. a Treatment scheme. Mice carrying various conditional alleles of BRaf (BRaf^CA) and/or Phd2 (Phd2^lox/lox) were crossed to the Tyr::CreER mice with melanocyte-specific expression of a hormone-dependent form of Cre recombinase (CreERT2). 4-OHT-dependent activation of the CreER recombinase leads to a melanocyte-specific conversion of BRaf^CA→BRaf^V600E and the conversion of the Phd2^lox alleles to Phd2^−/− alleles. b–g Melanoma progression in mice. Tyr::CreER;BRaf^CA;Phd2^lox/lox mice were treated topically with 4-OHT on the ear (b–d) and flank (e–g). 4-OHT-treated mice were killed at 40, 80 and 120 days after the induction. The pigmented lesions at flank and ear was photographed at days 40 (b, e), 80 (c, f), and 120 (d, g) following the 4-OHT induction. h–m Histology of melanoma in mice. Mouse tissues were processed and histology was examined using hematoxylin and eosin stain. Increased pigmented cells in the dermis were present at day 40 (h) and pigmented cells formed small tumor nodules in the dermis at day 80 (i). At day 120, melanoma cells formed bigger tumor nodules (j, k). Cytologically, these tumor cells were focally pigmented and had enlarged nuclei, a morphology that is similar to human melanoma (l). Tumor cells were stained positive for the S100 antibody (m). Bars in b–g indicate 5 mm. Bars in h–j indicate 200 μm. Bar in k indicates 400 μm. Bars in l, m indicate 50 μm

Fig. 3

Survival and lymph node metastasis in Tyr::CreER;BRaf^V600E;Phd2^−/− mice. a Kaplan–Meier survival analysis of Tyr::CreER;BRaf^V600E (n = 20), Tyr::CreER;Phd2^−/− (n = 20), Tyr::CreER;BRaf^V600E;Phd2^−/− postnatal mice (n = 20) and Tyr::CreER;BRaf^V600E;Phd2^−/− adult mice (6 weeks, n = 20) after 4-OHT induction. Log rank tests of survival plots of the data indicated a statistically significant difference between the following survival curves: Tyr::CreER; BRaf^V600E; versus Tyr::CreER;BRaf^V600E;Phd2^−/− mice (p < 0.0001), and Tyr::CreER;BRaf^V600E;Phd2^−/− adult versus Tyr::CreER;BRaf^V600E;Phd2^−/− postnatal mice (p < 0.0001). b–d Lymph node metastasis in Tyr::CreER;BRaf^V600E;Phd2^−/− mice. A positive lymph node was present near the area with melanoma induction (arrow points to the lymph node, b). The lymph node was enlarged with pigmented cells (c) and high-power view showed melanoma cells in the lymph node (arrows point to the tumor cells, d). e Percentage of mice that developed lymph node metastasis. Bar in c indicates 200 μm. Bar in d indicates 50 μm. Bar in b indicates 3 mm. *p < 0.05 compared to Tyr::CreER;Braf^V600E;Phd2^−/− group using one-way ANOVA with pairwise comparisons

Fig. 4

Haplo-deficient Phd2 is sufficient to cooperate with BRaf^V600E to induce melanoma in mice. a–h Haplo-deficient Phd2 is sufficient to cooperate with BRAF^V600E to induce melanoma in mice. The 6-week old adult Tyr::CreER; BRaf^CA; Phd2^lox/+ mice were treated topically on the ear, flank and tail with 4-OHT. The presence of pigmented lesions was assessed at days 45 (a), 90 (b), 135 (c) and 180 (d) following 4-OHT administration. 4-OHT-treated mice were killed at 45, 90 and 180 days after the induction. The pigmented lesions were photographed at day 45 (e), 90 (f) and 180 (g, h) following the 4-OHT induction. i Kaplan–Meier survival analysis of Tyr::CreER; BRaf^V600E; Phd2^-/+ neonatal mice (n = 20) and Tyr::CreER; BRaf^V600E; Phd2^-/+ (n = 20) adult mice (6 weeks to 1 year old). j Comparison of tumor volume in mice with homozygous deletion of Phd2 or haplo-deficient Phd2 mice. Tumor volume in Tyr::CreER; BRaf^V600E; Phd2^-/+ (n = 20) mice was significantly smaller than that of Tyr::CreER; BRaf^V600E; Phd2^−/− mice (p = 0.00171). k Melanoma shown on uninduced part of body (%) in homozygous deletion or haplo-deficient Phd2 mice. Melanoma present at uninduced body part was quantified (n = 20 in each group). l Average number of positive lymph nodes in homozygous deletion or haplo-deficient Phd2 mice (n = 20 in each group). m, n CD31 immunohistochemical stain in in homozygous deletion or haplo-deficient Phd2 mice (n = 5 in each group). Bars in a–d indicate 1 mm. Bars in e–h, m, n indicate 100 μm. Arrows in m, n point to CD31+ blood vessels. *p < 0.05 (t-test) compared to Brafv600E;Phd2^−/− group in j and l; error bars indicate s.d.

Fig. 5

Phd2 deletion induced HIF target and glucose uptake changes. a Phd2 regulated gene expression in mouse melanomas. Quantitative RT-PCR assay for Phd2, CAIX, VEGFA, GLUT1, PGK, PGM and LDHA mRNA expression in Tyr::CreER; BRaf^V600E; Phd2^−/− mice melanoma tissues (n = 3 replicate experiments; *p < 0.01 compared with Tyr::CreER; BRaf^V600E; Phd2^−/+ or Tyr::CreER; BRaf^V600E; Pten^−/− mouse melanoma tissues). β-Actin is used as loading control. b Phd2 regulated protein expression in mouse melanomas. Expression of Phd2, HIF-1α, VEGFA, pAkt and Akt protein was determined by western blot analysis in Tyr::CreER; BRaf^V600E; Phd2^−/−, Tyr::CreER; BRaf^V600E; Phd2^-/+ and Tyr::CreER; BRaf^V600E; Pten^−/− mouse melanoma tissues (n = 3 replicate experiments). β-Actin was used as a loading control. c Uptake of a fluorescent deoxyglucose analog (2-NBDG) is increased in Phd2^−/− mouse melanoma cells. The 2-NBDG uptake assay was performed using BRaf^V600E; Phd2^−/− or BRaf^V600E; Pten^−/− mouse melanoma cells (left panel) (n = 3 replicate experiments). Quantity of 2-NBDG uptake is assessed in these mice (right panel) (n = 3 replicate experiments, *p < 0.05). d HIF inhibition reverses Phd2 deletion-induced glucose uptake increase. 2-NBDG uptake is assessed in FM19G11-treated BRaf^V600E; Phd2^−/− melanoma cells (n = 3 replicate experiments, *p < 0.05 compared with control group). e Knockdown of HIF-1α reverses Phd2 deletion-induced glucose uptake increase. 2-NBDG uptake is assessed in BRaf^V600E; Phd2^−/− melanoma with HIF-1α knockdown (n = 3 replicate experiments, *p < 0.05 compared with control group). One way ANOVA or t-test was used for statistical analysis and error bars indicate s.d.

Fig. 6

Phd2 deletion leads to activation of the Akt–mTOR pathway. a RPPA analysis of tumors with homozygous deletion of Phd2. Tumor tissues from Tyr::CreER; BRaf^V600E; Phd2^−/− or Tyr::CreER; BRaf^V600E mice were processed and analyzed by RPPA assays. The analyses identified proteins that were significantly changed in mouse melanomas compared to nevi. b Activation of Akt–mTOR pathway after phd2 deletion. Tumor tissues were processed and western blots showed stabilization of HIF-1α and HIF-2α proteins after Phd2 depletion. Increased phosphorylation of Akt, 4EBP1 and S6K was observed in tumors from Tyr::CreER; BRaf^V600E; Phd2^−/− compared with those of Tyr::CreER; BRaf^V600E mice. c Re-expression of Phd2 inhibits the Akt–mTOR pathway. A BRaf^V600E; Phd2^−/− mouse melanoma cell line was established from melanomas in Tyr::CreER; BRaf^CA; Phd2^lox/lox mice. Phd2 was ectopically reintroduced in these tumor cells. Western blot analysis showed that degradation of HIF-1α and HIF-2α proteins with decreased expression of VEGFR2 decreased phosphorylation of Akt, 4EBP1 and S6K. d Pharmacological inhibition (FM19G11) of HIF pathway in BRaf^V600E; Phd2^−/− melanoma cells. A similar but more pronounced inhibition of the Akt–mTOR pathway was observed using the HIF inhibitor. β-Actin was used as a loading control. Results are representative of three independent experiments

Fig. 7

Inhibition of mTOR pathway suppresses the growth of melanoma in vivo. a Scheme of rapamycin treatment. b Representative images of Tyr::CreER; BRaf^V600E; Phd2^−/− mouse treated with the vehicle control (left panel) or rapamycin (right panel). Rapamycin treatment lasted for 8 weeks. Tumor masses are highlighted by red circles. c Kaplan–Meier survival curves of mice treated with vehicle control (n = 10) or rapamycin (n = 10) and p < 0.05. d–i Representative images of skin and lymph nodes in mice treated with the vehicle control or rapamycin. After killing the mice, underside of skin was exposed and photographed (d, g). The entire dermis was occupied by tumor cells (e) and lymph node was positive for melanoma in the vehicle-treated mice (f). On the contrary, there were significantly fewer tumor cells in the dermis (h) with negative lymph node (i) in the rapamycin-treated mice. j Inhibition of the Akt–mTOR pathway resulted in tumor growth suppression. Tumor tissues with or without rapamycin treatment were processed and analyzed using RPPA assays. Proteins with most significant changes were used to generate the heatmap. Notably, phospho-Rb^S807/811, phospho-AKT^S473, Cyclin B1, TFRC, phospho-S6^S235/236, phospho-EGFR^Y1068, MEK1 and phospho-AKT^T308 proteins were significantly reduced in mice treated with rapamycin. k Torin1 inhibits Akt–mTOR pathway. BRaf^V600E; Phd2^−/− melanoma cells were treated with torin1 for 48 h. Western blot analysis was performed. Results are representative of three independent experiments. l Torin1 inhibits BRaf^V600E; Phd2^−/− melanoma cell growth. The melanoma cells were treated with different concentrations of torin1 (Control, 0.01, 0.1, 1, 10 and 100 μM) and different lengths of time (24 or 48 h). Results are summary of three independent experiments. Bars in b indicate 6 mm. Bars in d, g indicate 3 mm. Bars in e, f, h, i indicate 100 µm. *p < 0.05 compared to control (t-test); error bars indicate s.d.