PHD2 Levels Decrease in Human Cancers
(A) Variation of PHD2 levels in the NCI-60 panel of cell lines immunoblotted for PHD1, PHD2, PHD3, HIF-1α, HIF-2α, and α-tubulin.
(B) Immunohistochemical staining for PHD2 in colon adenocarcinoma (bottom panels) compared to corresponding normal adjacent colon tissue (top panels) from several different patients. Scale bar = 500 μm.
(C) Oncomine database analysis to examine PHD2 mRNA expression levels in adjacent, non-tumoral colon versus colon carcinoma (p<0.0005).
(D) HCT116 cells with shRNA designed to specifically target PHD2. Cells were treated with DFO (100 μM; 4 hours) or proteasome inhibitor (MG132: 10 μM; 4 hours) and harvested to examine protein levels of PHD2, HIF-1α, and α-tubulin.

HIF-Independent Tumor Suppressor Function of PHD2
(A) Tumor growth of HCT116 cells expressing shRNA to HIF-1α alone or in combination with shRNA to PHD2 (*p<0.05).
(B) PHD2 and HIF-1α silencing is maintained in vivo. Protein extracts from tumors in Figure 3A were immunoblotted to examine levels of HIF-1α, PHD2, and α-tubulin.
(C) Tumor sections of the given genotype were stained for CA-IX expression. Scale bar = 200 μm.
(D) Tumor growth of SU.86.86 cells with shRNA to PHD2 or shRNA to PHD2 and HIF-1β (*p<0.05).
(E) In vivo growth of RKO cells deficient in HIF-1α or HIF-1α-deficient cells expressing PHD2 shRNA (*p<0.05).
(F) In vivo growth of HIF-1α-deficient HCT116 cells or those expressing PHD2 shRNA (p<0.05). All error bars represent ±SEM.

Hydroxylase function of PHD2 is not necessary for regulating angiogenesis
(A) The ability of conditioned media from wild-type or PHD2 shRNA expressing HCT116 cells with or without hypoxia condition (2% O₂, 16 hours) in inducing in vitro tube formation (*p<0.001).
(B) The ability of conditioned media from HIF1^-/- HCT116 cells or HIF1^-/- HCT116 cells expressing PHD2 shRNA under normoxia or hypoxia (2% O₂, 16 hours) in inducing in vitro tube formation (*p<0.01).
(C) Conditioned media from HCT116 HIF1^-/- cells or cells expressing shRNA to PHD2 treated with DMOG (1 mM, 16 hours) was subjected to in vitro tube formation (*p<0.001).
(D) In vitro tube formation of conditioned media from RKO HIF1^-/- cells or cells expressing shRNA to PHD2 treated with hypoxia (2% O₂) or DMOG (1 mM) for 16 hours (*p<0.01).
(E) HCT116 cells expressing PHD2 shRNA were transfected with either wild-type PHD2 or hydroxylase mutant PHD2 (H313A H315D). Conditioned media from these cells were collected and subjected to a tube formation assay (*p<0.001). All error bars represent ±SEM.
(F) Representative photographs of primary endothelial cells seeded onto Matrigel in conditioned media from HCT116 cells stably expressing shRNA to PHD2 and transfected with either wild-type PHD2 or a hydroxylase mutant PHD2. Scale bar = 2 mm.

PHD2 Regulation of ANG and IL-8 Mediates Tumor Vasculature
(A) Conditioned media from HCT116 cells expressing PHD2 shRNA and shRNA targeting either ANG or IL-8 were subjected to a tube formation assay (*p<0.05).
(B) Tube formation with conditioned media from the given genotype. Scale bar = 2 mm.
(C) Tumor staining for CD11b myelomonocytic precursor cells (red) and pericytes with vascular smooth muscle actin (green). Scale bar = 100 microns. Bottom panels are higher magnification of PHD2-silenced tumors. Scale bar = 500 microns. Quantification of CD11b-positive cells (*p<0.001).
(D) Tumor staining for CD45 (red), hypoxia with pimonidazole (green) and nuclei with DAPI (blue). Scale bar = 100 microns. Quantification of CD45-positive cells (*p<0.001).
(E) Higher magnification images of wild-type HCT116 tumors stained for CD31 (red) and pimonidazole (green). Scale bar = 100 microns.
(F) Tumor growth of cells expressing the given shRNA in SCID mice (*p<0.005). All error bars represent ±SEM.

Loss of PHD2 Increases Tumor Growth
(A) In vivo growth of wild-type or PHD2 knockdown (Construct A) HCT116 xenograft tumors (*p<0.00001).
(B) In vitro growth curve of HCT116 cells with or without stable knockdown to PHD2 under either normoxic (N, 21% O₂) or hypoxic (H, 2% O₂) conditions.
(C) Growth of wild-type or PHD2 knockdown (construct C or E) HCT116 tumors in mice (*p<0.001).
(D) Growth of wild-type SU.86.86 or cells expressing PHD2 shRNA in SCID mice (*p<0.000005).
(E) In vivo growth of HT29 cells expressing shRNA to PHD2 (Construct A) or control (construct E)(*p<0.000005).
(F) Growth of wild-type RKO cells or cells expressing shRNA to PHD2 in SCID mice (*p<0.05). All error bars represent ±SEM.

PHD2 Regulates Angiogenesis and Recruitment of Bone Marrow-Derived Cells
(A) CD31 staining in sections from HCT116 tumors expressing PHD2 shRNA compared to wild-type tumors (*p<0.01). Scale bar = 500 μm.
(B) CD31 staining in sections from HCT116 tumors expressing shRNA to both PHD2 and HIF-1α compared to those expressing HIF-1α shRNA alone (*p<0.0005). Scale bar = 500 μm.
(C) In vitro tube formation of endothelial cells in conditioned media from wild-type HCT116 cells or cells expressing PHD2 (*p<0.000005). Scale bar = 1 mm.
(D) In vitro tube formation of endothelial cells in conditioned media from HIF-1α-deficient HCT116 or HIF-1α-deficient HCT116 cells expressing PHD2 shRNA (*p<0.00005). Scale bar = 1 mm.
(E) Staining of HCT116 tumors for bone marrow-derived myelomonocytic cells with CD11b (red) and nuclei with DAPI (blue) (*p<0.005). Scale bar = 100 μm.
(F) HCT116 tumor staining for CD45 (green) and nuclei with DAPI (blue) (*p<0.005). Scale bar = 100 μm. All error bars represent ±SEM.

NF-κB regulates PHD2-mediated angiogenesis
(A) NF-κB reporter assay of wild-type or HCT116 cells expressing PHD2 shRNA (*p<0.05).
(B) NF-κB is hypoxia-responsive. HCT116 cells were transfected with a NF-κB luciferase reporter and treated with hypoxia (2% O₂, 16 hours)(*p<0.005).
(C) Wild-type or hydroxylase mutant PHD2 were overexpressed along with a NF-κB luciferase reporter construct in HCT116 HIF-1^-/- cells stably expressing shRNA to PHD2. Following transfection, these cells were not treated or treated with TNF-α (20 ng/ml; 6 hours), hypoxia (2% O₂; 16 hours), or DMOG (1 mM; 16 hours) (*p<0.05).
(D) Quantitative real-time PCR to demonstrate knockdown of RELA in HCT116 cells with HIF-1α knocked out.
(E) Conditioned media from HCT116 HIF-1^-/- cells expressing PHD2 shRNA transfected with pool of non-targeting siRNA or siRNA targeting RELA were subjected to a tube formation assay (*p<0.0005).
(F) Primary endothelial cells were seeded onto Matrigel in conditioned media from HIF-1α knockout cells or those expressing shRNA to PHD2 and assayed for tube formation. Scale bar = 1 mm.
(G) Angiogenesis antibody arrays were carried out in duplicate with conditioned media from either wild-type HCT116 or cells stably expressing PHD2 shRNA.
(H) Reporter assays with ANG or IL-8 promoters along with the promoters with mutant NF-κB sites. Two days after transfection, cells were treated with TNF-α (20 ng/ml; 6-8 hours) and luciferase activity was determined. (*p<0.005).
(I, J) RELA is recruited to the ANG promoter. HCT116 cells expressing shRNA to PHD2 were treated with TNF-α (20 ng/ml) for the given amount of time. ChIP with α-RELA antibodies was performed, and RELA occupancy is shown. MnSOD promoter is used as a positive control for recruitment of RELA to a NF-κB target promoter. IgG control ChIPs were also performed as a negative control. All error bars represent ±SEM.

Decreased PHD2 Expression Correlates with Increases in NF-κB Activity and Tumor Angiogenesis in Human Cancers
(A, B) Left panels: Significant negative correlation between the expression level of PHD2 (x-axis) and CD31 (y-axis) in the breast cancer samples profiled in two expression studies of breast cancers. Middle panels: Significant positive correlation between NF-κB pathway activity (x-axis) and expression of CD31 (y-axis). Right panels: Significant negative correlation between the expression level of PHD2 (x-axis) and NF-κB activity (y-axis).
(C) Immunohistochemical staining for PHD2 in invasive breast carcinomas (bottom panels) compared to matched, uninvolved, adjacent breast tissue (top panels) from several different patients. Scale bar = 100 μm.