Sp1 is involved in Akt-mediated induction of VEGF expression through an HIF-1-independent mechanism

Abstract

Increased expression of vascular endothelial growth factor (VEGF) contributes to the growth of many tumors by increasing angiogenesis. Although hypoxia is a potent inducer of VEGF, we previously showed that epidermal growth factor receptor amplification and loss of PTEN, both of which can increase phosphatidylinositol-3-kinase (PI3K) activity, increase VEGF expression. Using both adenoviral vectors and a cell line permanently expressing constitutively active myristoylated Akt (myrAkt), we show that activation of Akt, which is downstream of PI3K, increases VEGF expression in vitro and increases angiogenesis in a Matrigel plug assay. Transient transfection experiments using reporter constructs containing the VEGF promoter showed that up-regulation of VEGF by Akt is mediated through Sp1 binding sites located in the proximal promoter. Small interfering RNA directed against Sp1 prevented the induction of VEGF mRNA in response to myrAkt but not to hypoxia. Expression of myrAkt is associated with increased phosphorylation of Sp1 and its increased binding to a probe corresponding to the -88/-66 promoter region. In conclusion, our results indicate that Sp1 is required for transactivation of the VEGF by Akt. Others have proposed that the PI3K/Akt pathway can increase VEGF expression via the hypoxia-inducible factor 1 (HIF-1); however, our results suggest an alternative mechanism can also operate.

Figure 1.

Akt modulates expression of VEGF in different cell lines. (A–C) Northern blots; 10 μg of total RNA were harvested from cells and run on gel. Membranes were probed for VEGF and ribosomal 18S (loading control). (A) VEGF mRNA levels from various cell lines grown under normal tissue culture conditions. (B) U87MG cells were infected with adenovirus expressing either dominant negative Akt or GFP at a multiplicity of infection (MOI) of 10. Cells were harvested 48 h later for RNA isolation. (C) SF188, LN229, or DU145 cells were infected with adenovirus expressing either myrAkt or GFP (MOI of 10). Cells were harvested 24 or 48 h later for RNA isolation. Numbers at bottom of gels in A–C represent relative VEGF levels (the ratio of VEGF band intensity to 18S band intensity). (D–G) VEGF protein levels. VEGF protein levels were determined by ELISA and normalized to the number of cells in each dish. (D and E) Media were sampled 24 h after infecting SF188 or DU145 cells with virus expressing either myrAkt or GFP or no virus. (F) Medium was sampled 24 h after infecting U87MG cells with virus expressing either dominant negative Akt or GFP or no virus. (G) Medium was taken from NHA and NHAAkt cells 24 h after plating.

Figure 2.

Akt increases angiogenesis in vivo. Matrigel mixture containing U87MG, NHA, or NHA-Akt cells was injected subcutaneously. into nude mice at sites lateral to the abdominal midline. As a negative control, Matrigel containing 100 μl PBS was injected in similar way. Animals were sacrificed 5 d after injection. The mouse skin was detached along the abdominal midline, and the Matrigel plug was recovered and photographed immediately (A). (B) The relative level of hemoglobin present in each plug was determined using commercially available kit (Sigma). The amount of hemoglobin was calculated from standard hemoglobin curve. The amount of hemoglobin normalized to the weight of each Matrigel plug is plotted on y-axis.

Figure 3.

Role of Akt and Sp1 in VEGF promoter activation. In all panels, pSV-β-galactosidase was cotransfected as a control for transfection efficiency along with the indicated plasmids. Cells were harvested 48 h after transfection, and lysates were harvested for determination of luciferase and β-galactosidase activity. The ratio of luciferase to β-galactosidase is plotted on the y-axis. (A) SF188 cells were cotransfected with the 1.5-kb VEGF promoter luciferase reporter (2 μg) and 0.2 μg of either pCMV6/myr-Akt or filler control DNA (pCMV6). (B) Sequences of -88/-66 human VEGF promoter region. In the top sequence labeled wt, the AP-2 binding site is overlined, and Sp1 binding sites are underlined. The nucleotides in the Sp1 or AP-2 binding sites that were mutated in the three mutant promoter constructs (Sp1mut, AP-2mut, Sp1/AP-2mut) are boldfaced. (C) The four different VEGF promoter constructs (wt and three mutant constructs) were cotransfected separately into U87MG cells. (D) pCMV6/myrAkt (0.2 μg) and one of four VEGF promoter constructs (2 μg) was cotransfected into SF188 cells. (E) The 1.5-kb VEGF promoter luciferase reporter was cotransfected along with pSG5 vectors expressing one of the AP-2 isoforms (α, β, or γ; 0.2 μg each) into SF188 cells. (F) Vectors expressing Sp1 or Sp3 (0.2 μg each) were cotransfected with one of four VEGF promoter reporter constructs (2 μg) into SF188 cells. The results shown in this figure are representative of at least two independent experiments.

Figure 4.

Mithramycin interferes with Akt-mediated induction of VEGF expression and VEGF promoter transactivation. (A) U87 MG, SF188, DU145, and PC3 cells were treated with 100 nM mithramycin. After 24 h RNA was harvested from cells. Northern blotting was performed, and membranes were probed for VEGF and ribosomal 18S (loading control). (B) SF188 cells were infected with adenovirus expressing myrAkt or GFP or no virus 24 h after infection cells were treated with 100 nM mithramycin for another 24 h. Aliquots were taken, and VEGF protein was measured by ELISA. (C) Sp1 or Sp3 expression vectors (0.2 μg) were cotransfected along with pSV-β-galactosidase (0.1 μg) and the wt -88/+54 VEGF promoter reporter (2 μg) into SF188 cells. Panel D is identical to C except that the luciferase reporter used was the AP-2mut promoter rather than the wt -88/+54 reporter. (E) Sp1 or Sp3 expression vectors were cotransfected with pSV-β-galactosidase and the wt -88/+54 VEGF promoter reporter and in some cases with pCMV/myrAkt (0.2 μg) into SF188 cells. In some cases cells were exposed to mithramycin (100 nM). Panel F is identical to E except that the luciferase reporter used was the AP-2mut promoter rather than the wt -88/+54 reporter. The results shown in this figure are representative of at least two independent experiments.

Figure 5.

Sp1 RNAi abolishes Akt-mediated induction of VEGF mRNA expression. (A) Cells were transfected without RNAi, with Sp1 RNAi (600 pmol/well), or with control GFP RNAi. Forty-eight hours later, cells were harvested. (B) Cells were transfected without RNAi, with Sp3 RNAi (600 pmol/well) or with control GFP RNAi. Forty-eight hours later, cells were harvested. For both A and B, protein lysates were separated on SDS-PAGE gel and transferred to nitrocellulose membrane. Membrane was probed for Sp1 (A) and Sp3 (B). In both A and B, membranes were subsequently reprobed for β-actin. (C) RNA was extracted from SF188 cells. RT-PCR was performed using a range of starting quantities of RNA (0.06, 0.12, 0.24, 0.36, 0.48 μg). DNA was run on a gel that was stained with ethidium bromide. (D) DNA quantitation from RT-PCR was plotted vs. quantity of starting RNA. (E) SF188 cells were transfected with RNAi targeting Sp1, Sp3, or GFP. After 24 h, cells were transduced with adenovirus expressing myristoylated Akt or GFP (control). Forty hours after viral transduction, cells were harvested for RNA. (F) SF188 cells were transfected with RNAi targeting Sp1 or GFP. After 40 h, cells were exposed to 0.2% oxygen and then harvested for RNA after 8 h. (E and F) Results of RT-PCR performed for VEGF and 18S using 0.12 μg of total RNA.

Figure 6.

Gel shift assays performed with nuclear extracts from cells transduced with PTEN or Akt expressing adenovirus. (A–C) Oligonucleotides corresponding to -88 to -66 base pairs of the human VEGF promoter were labeled with [γ-³²P]ATP. Gel shift assays were performed using nuclear extract from cells transduced with adenovirus expressing dead PTEN, wild-type PTEN, dominant negative Akt, myrAkt, or GFP as indicated in the headings above the figures. The DNA-binding reaction was performed with 100-fold molar excess of unlabeled competitor, Sp1, AP-2 consensus oligonucleotide as indicated. Cell line (SF188 or U87 MG) is indicated at top of gel. (A–C) Arrows point to shifted band. (D) Supershift gel shifts were performed using nuclear extract (5 μg) and 0.2 μg of AP-2 or HIF-1α (control) antibody. Arrow indicates position of supershifted complex. (E) Supershift gel shifts were performed using nuclear extract (5 μg) and 0.2 μg of Sp1, Sp3, or AP2 antibody. The position of supershifted complexes is indicated by arrow. (F) Lysates from U87MG cells were incubated with alkaline phosphatase for 15 min and then treated with phosphatase inhibitors in order to inactivate the enzyme. Then the radioactive probe was added, and gel shift assay was performed. Arrow points to shifted band.

Figure 7.

myrAkt is associated with phosphorylation of Sp1 and Sp3 in SF188 cells. SF188 cells were infected with adenovirus expressing myrAkt or GFP. Thirty-six hours after infection, cells were in vivo–labeled with orthophosphate. In the top part of A labeled IP, in vivo–labeled proteins were immunoprecipitated with either Sp1 or Sp3 antibody as indicated. Immunoprecipitated complexes were separated on 10% SDS-PAGE gel and transferred to nitrocellulose membrane and autoradiographed. In the bottom part of A labeled IB (immunoblot), these same lysates were separated on 10% SDS-PAGE gel, transferred to nitrocellulose membrane, and probed with either Sp1 or Sp3 antibody to serve as a loading control. (B) The same procedure used in A was followed, but after immunoprecipitation with anti-Sp1 antibody, each protein extract was incubated with or without 10 mU of calf intestinal phosphatase (CIP) for 30 min at room temperature before running on an SDS-PAGE gel. The identical procedure was followed for C as in B, except lysates were immunoprecipitated with an anti-Sp3 antibody. (D) U87MG cells were treated with LY294002 (20 μM) or DMSO (control) for 16 h. At this time, cells were lysed and the procedure described for A was followed.

Figure 8.

Mutation of HIF-1 binding site in VEGF promoter does not prevent its regulation by Akt. (A) Schematic of VEGF promoter constructs. The mut1 construct contains a mutation in the HIF-1 binding site of CGT to AAA as indicated (first box in solid black). The mut2 construct contains mutations in Sp1/AP-2 binding sites in the proximal promoter (second box in solid black). The mut3 construct contains mutations in both the HIF-1 binding site and Sp1/AP-2 region (fist and second boxes in solid black). (B) SF188 cells were cotransfected with (a) 0.2 μg of pCMV/myrAkt or pCMV6 (empty vector), (b) 2 μg of 1.5-kB wt VEGF, mut1, mut2, or mut3 vector, and (c) pSV-β-galactosidase (0.1 μg). Forty-eight hours later, dishes were harvested for both luciferase and β-galactosidase determination. (C) U87MG cells were cotransfected with (i) 2 μg of mut2 or mut3 vector and (ii) pSV-β-galactosidase (0.1 μg). Twenty-four hours later they were subjected to 1% O₂. Samples were harvested after 8 h in hypoxia. (D) U87MG cells were cotransfected with (a) pCMV6/Akt K179M (0.2 μg), which is a plasmid expressing dominant negative Akt, (b) 2 μg of either a 1.4-kB wt VEGF reporter or mut1, and (c) pSV-β-galactosidase (0.1 μg). Forty-eight hours later, dishes were harvested.

Figure 9.

Effect of myrAkt in DU145 prostate carcinoma cells. (A) Oligonucleotides corresponding to -88 to -66 base pairs of the human VEGF promoter were labeled with [γ-³²P]ATP. Gel shift assay was performed using nuclear extract from cells transduced with adenovirus expressing myAkt or GFP (control). The DNA-binding reaction was performed with 100-fold molar excess of unlabeled competitor, Sp1, AP-2 consensus oligonucleotide as indicated. (B) DU145 cells were infected with adenovirus expressing myrAkt or GFP. Thirty-six hours after infection, cells were in vivo–labeled with orthophosphate. In the top part of B labeled IP, in vivo–labeled proteins were immunoprecipitated with either Sp1 or Sp3 antibody as indicated. Immunoprecipitated complexes were separated on 10% SDS-PAGE gel and transferred to nitrocellulose membrane and autoradiographed. In the bottom part of B labeled IB (immunoblot), these same lysates were separated on 10% SDS-PAGE gel, transferred to nitrocellulose membrane, and probed with either Sp1 or Sp3 antibody to serve as a loading control.