Repression of Vascular Endothelial Growth Factor Expression by the Runt-Related Transcription Factor 1 in Acute Myeloid Leukemia

VEGFA is considered one of the most important regulators of tumor-associated angiogenesis in cancer. In acute myeloid leukemia (AML) VEGFA is an independent prognostic factor for reduced overall and relapse-free survival. Transcriptional activation of the VEGFA promoter, a core mechanism for VEGFA regulation, has not been fully elucidated. We found a significant (P < 0.0001) inverse correlation between expression of VEGFA and AML1/RUNX1 in a large set of gene expression array data. Strikingly, highest VEGFA levels were demonstrated in AML blasts containing a t(8;21) translocation, which involves the AML1/ RUNX1 protein (AML1/ETO). Overexpression of AML1/RUNX1 led to downregulation of VEGFA expression, whereas blocking of AML1/RUNX1 with siRNAs resulted in increased VEGFA expression. Cotransfection of AML1/RUNX1 and VEGFA promoter luciferase promoter constructs resulted in a decrease in VEGFA promoter activity. ChIP analysis shows a direct binding of AML1/RUNX1 to the promoter of VEGFA on three AML1/RUNX1 binding sites. Silencing of AML1/ETO caused a decrease in VEGFAmRNA expression and a decrease in secreted VEGFA protein levels in AML1/ETO-positive Kasumi-1 cells. Taken together, these data pinpoint to a model whereby in normal cells AML1/RUNX1 acts as a repressor for VEGFA, while in AML cells VEGFA expression is upregulated due to AML1/RUNX1 aberrations, for example, AML1/ETO. In conclusion, these observations give insight in the regulation of VEGFA at the mRNA level in AML. Cancer Res; 71(7); 1–11. 2011 AACR.


Introduction
In cancer, including hematological malignancies, VEGFA is considered as one of the most important regulators of tumorassociated angiogenesis (1)(2)(3).Besides its role in angiogenesis, VEGFA has also been shown to enhance leukemic cell survival and/or leukemic cell growth through an autocrine loop, and a paracrine pathway in which VEGFA stimulates the production of hematopoietic growth factors by endothelial cells and/or stromal cells (4)(5)(6)(7).Moreover, AML-derived VEGFA is an independent adverse prognostic factor related to worse treat-ment outcome in AML and is significantly upregulated in leukemic blasts compared with normal controls (1)(2)(3).
Previous studies have shown that VEGFA gene expression can be influenced by extracellular growth factors including platelet-derived growth factor, transforming growth factor beta, hepatocyte growth factor, and/or by cytokines like interleukin 1 alpha, interleukin 3, and interleukin 6 (8)(9)(10)(11)(12).Several oncogenes have also been implicated in increased VEGFA expression including Ras and Src proteins (13,14).Transcriptional regulation of VEGFA is known to be mediated by many transcription factors including SP1, HIF-1a, and STAT3 (15)(16)(17).Although all these studies clearly indicate an important role for VEGFA in the pathogenesis of AML, surprisingly little is known regarding AML specific transcription factors involved in VEGFA mRNA regulation.
AML1/RUNX1 encodes the alpha subunit of the Runt domain transcription factors and binds a non-DNA-binding b subunit (CBF b) to form a heterodimeric complex, corebinding factor (CBFl; refs.24,25).Its function depends on the availability of cofactors such as HDAC (26).AML1/ RUNX1 has important roles in hematopoiesis; AML1/RUNX1 is required for the generation of definitive hematopoietic stem cells during embryogenesis.Furthermore, haploinsufficiency of AML1/RUNX1 has been shown to impair numbers of long-term repopulating HSCs (27), although conditional knockout models have suggested that AML1/RUNX1 is not strictly required for steady-state HSC maintenance during adult hematopoiesis (28).AML1/RUNX1 is essential for differentiation of myeloid progenitor cells into granulocytes (24,25,29,30).Interestingly, VEGFA mRNA expression was found to be higher in AML blasts characterized by t(8;21), AML1/ETO compared with other cytogenetic AML subtypes (31).
In this study, we identified wild-type AML1/RUNX1 as an important transcriptional repressor of VEGFA mRNA expression in AML cells, whereas repression of VEGFA mRNA expression is attenuated by AML1/RUNX1 mutations, such as in AML1/ETO-positive AML blasts.

Statistical analysis
Statistical analyses were performed with SPSS software, release 14.0.Correlations between VEGFA and RUNX1 were calculated with the Spearman rank correlation coefficient.The probes for VEGFA and AML1/RUNX1 were averaged.All tests were 2-tailed, and a P value of less than 0.05 was considered statistically significant.To investigate the relation between VEGFA and AML1/RUNX1 mRNA expression in primary AML blasts, statistical analysis has been performed on gene expression data from 525 previously published adult AML patients (32,33).Detailed clinical, cytogenetic, and molecular cytogenetic information is available at the Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo,accession number GSE6891).

Cell culture and reagent
Human acute myeloid leukemia cell lines HL-60, and TF-1 were obtained from the ATCC in 2006.Cells have last been tested and authenticated by flow-cytometry analysis in 2009.The cells were maintained in RPMI 1640 (Lonza) and 293T cells in DMEM (Lonza), supplemented with 10% fetal calf serum (FCS; Bodinco B.V.) and 1% penicillin/streptomycin (Invitrogen).TF-1 cell culture was supplemented with an additional 10 ng/mL GM-CSF (Sandoz AG).Kasumi-1 were obtained from the German Resource Centre for Biological Material (DSMZ) in 2007 and were directly passaged and used in experiments (method of characterization of DSMZ is Fingerprint: multiplex PCR of minisatellite markers).Kasumi-1 cells were cultured in RPMI 1640 containing 20% FCS.Trichostatin A (TSA; Sigma-Aldrich), an antifungal antibiotic that is a reversible, potent, and specific inhibitor of mammalian histone deacetylase was dissolved in ethanol.The functionality of secreted VEGFA from transfected cells was assessed by adding its supernatant to HUVEC and to quantify the expression level of the VEGFA responsive gene EGR-3 in HUVEC using realtime PCR as described in detail by Liu and colleagues (34).

RUNX1b overexpression construct
Full length human RUNX1b was subcloned from a pCMV5RUNX1b construct into the EcoRI sites of pMSCViGFP.Overexpression of RUNX1b was established on mRNA and protein level.

Reporter constructs
A 2.4-kb VEGFA promoter fragment from À2,274 to þ50 relative to transcription start site in pGL-2 basic vector (kind gift from Dr. Georg Breier, Institute of Pathology, Medical Faculty, University of Technology, Dresden, Germany) was subcloned into the KpnI and NheI sites of pGL3-basic vector (Promega).This reporter construct is designated as pGL3-VEGFA.
Due to low transfection efficiency of AML cells by electroporation or chemical reagents, the deletion promoter constructs were introduced into cells with a Moloney murine leukemia virus (Mo MLV) based retrovector pCBsinDelta (Supplementary Fig. 1A) which is developed from the pBabe vector (35).It is designed as a self-inactivating (SIN) vector by mutagenesis and deletion of retroviral promoter and enhancer machinery in the 3 0 LTR of pBabe vector.As the U3 region in the 3 0 LTR of a provirus serves as a template for the formation of both U3 regions in the progeny provirus, the mutation and deletion in the U3 of the 3 0 LTR will be transferred to both LTRs in the progeny virus.As a consequence of mutagenesis and removal of the enhancer and promoter from both LTRs, the viral transcriptional unit is eliminated and the luciferase reporter gene is under the regulatory control of the VEGFA promoter (Supplementary Fig. 1A).The VEGFA promoterluciferase fragment from the KpnI/XbaI-digested pGL3-VEGFA was further inserted into retrovector by replacing the NGFR in the pCBsinDelta-NGFR construct which was cut with EcoRI and XbaI (Fermentas); KpnI and EcoRI sites were blunted with T4 DNA polymerase (Fermentas).The retrovector incorporating full-length VEGFA promoter was designated pCBsinD-VEGFA1.5 0 -deletion promoter reporter constructs were generated by digestion of different restriction sites within the promoter region and a XbaI site behind the luciferase: BamHI (À1,287) for pCBsinD-VEGFA2, PstI (À790) for pCBsinD-VEGFA3, BmgBI (À507) for pCBsinD-VEGFA4, and SacII (À268) for pCBsinD-VEGFA5 (Fermentas).Deletion promoter/luciferase fragments were inserted into retrovector through the same procedure for generation of pCBsinD-VEGFA1.pCBsinD-VEGFA6 and pCBsinD-VEGFA7 were made by PCR using forward primers corresponding to the positions À86 to À65, and À52 to À31, which contained an additional EcoRI site (Primer for pCBsinD-VEGFA6: 5 0 -ACGAATTCGGGGCGGGCCGGGGGCGGGG-3 0 ; Primer for pCBsinD-VEGFA7: 5 0 -ACGAATTCAGCCATGCGCCCCCCCC-TTT-3 0 ), a reverse primer to the region behind VEGFA promoter (5 0 -GTGGCTTTACCAACAGTACC-3 0 ).PCR products were digested with EcoRI and HindIII and inserted to retrovector containing luciferase which was generated by digestion of pCBsinD-VEGFA1 with EcoRI and HindIII (Fermentas).To generate the promoter-less control construct pCBsinD-Luc, the luciferase gene was cut from pGL3 basic with NheI and XbaI (Fermentas) and inserted into the retrovector.All constructs were verified by restriction and sequence analysis (ServiceXS).
QuickChange site-directed mutagenesis kit (Stratagene) was used to generate reporter constructs with mutant elements according to the manufacturer's instructions.All mutagenic primers were designed using Stratagene's web-based QuickChange Primer Design Program http://www.stratagene.com/qcprimerdesign).All mutant constructs were confirmed by sequence analysis (ServiceXS).

Retroviral transduction of leukemic cell lines
Retroviral supernatants were generated by cotransfection of 5 mg RUNX1b construct or reporter constructs and 5 mg packaging plasmid pCLampho into 293T cells by using FuGENE HD transfection reagent (Roche).Before transduction of leukemic cells, 293T cells were incubated with RPMI containing 10% FCS for 24 hours.Then, 5 Â 10 4 cells were incubated with retroviral supernatants which were filtered through 0.45-mm pore size syringe-mounted filters.The incubation was supplemented with 8 mg/mL polybrene.This procedure was repeated for 2 consecutive days after which stably transduced cells were expanded.
HL-60 cells transduced with pCBsinD-Luc were sorted into single-cell clones.From 3 randomly chosen single cell-derived clones luciferase activity was measured and correlated with vector copy number (determined with luciferase specific real time PCR), following earlier studies showing that real-time PCR represents an effective method to correct for varying transduction efficiency (36).We demonstrate that luciferase activity is positively correlated with transduction efficiency (Supplementary Fig. 1B).In further experiments no clonal selection was performed, mixed populations of transduced cells were used for subsequent experiments and normalized accordingly.

VEGFA stimulation
A total of 4 Â 10 6 HL-60 empty vector and HL-60 RUNX1b cells were starved on 0.1% FCS for 3 hours.100 ng/mL recombinant VEGFA was added (Sigma-Aldrich) and after 1 hour cells were washed and RNA was isolated.

Luciferase assay
Transduced leukemic cells were lyzed with luciferase cell culture lysis reagent (Promega).Then, a 20 mL aliquot of each lysate supernatant was mixed with 100 mL luciferase assay reagent (Promega) and light intensity was measured based on light emission using Wallac 1420 multilabel counter (Perkin Elmer).The protein concentrations of the lysate supernatants were determined by BCA Protein Assay Kit (Pierce).The luciferase activity was calculated as light intensity/mg protein.

VEGF protein measurement
The VEGF protein level in the culture supernatants were determined using the Quantikine VEGF ELISA kit (R&D Systems), which is a quantitative immunometric sandwich enzyme immunoassay.A curve of the absorbance of VEGF versus its concentration in the standard wells was plotted.By comparing the absorbance of the samples with the standard curve, we determined the VEGF concentration in the unknown samples.

Cell survival assays
For quantification of HL60 cell viability after serum deprivation and VEGFA stimulation, cell survival assays were performed.A WST-1 colorimetric viability assay was used following the procedures recommended by the manufacturer (Roche).HL60 cells were seeded at a density of 500 cells per well in medium supplemented with 10% FCS (serum condition) or 0% FCS (serum free condition).The cells were subjected to serum-free conditions with or without addition of 10 or 25 ng/ mL VEGFA for 24 hours of incubation (6 replicates for each concentration).After addition of the WST-1 cell survival reagent the absorbance was measured at 450 nm in a microplate reader (Benchmark; Bio-Rad).The data are presented as the cell survival percentage relative to the serum condition cells.

Tube formation assay
The surface of a 4-well culture slide (BD biosciences) was coated with 150 mL of Matrigel basement membrane matrix (BD biosciences), which was allowed to polymerize at 37 C for 30 minutes.Conditioned medium of HL60 cells treated with either scrambled siRNA or AML1/RUNX1 siRNA was added to each well together with 5 Â 10 5 HUVECs.After 16 hours, endothelial tubes were examined using light microscopy.The web based analysis program S.CORE -Image Analysis (http:// www.sco-lifescience.de)was used to calculate the total tube formation skeleton length.

Inverse correlation between VEGFA and AML1/RUNX1 expression in AML patients
To investigate the relationship between VEGFA expression and RUNX1 expression A highly significant inverse correlation was found between VEGFA and AML1/RUNX1 mRNA expression (n ¼ 525, rho ¼ À0.173, P ¼ < 0.0001; Fig. 1).In AML, aberrations of AML1/RUNX1 are frequent and one of the most common chromosomal translocation in AML is t(8;21), resulting in the fusion gene AML1/ETO.Interestingly, the highest VEGFA levels were apparent in AML samples with an AML1/ ETO fusion protein compared with other cytogenetic subtypes (P < 0.0001; Fig. 1) The inverse relationship between VEGFA and AML1/ RUNX1 suggested that AML1/RUNX1 may be involved in negative regulation of VEGFA expression.To determine the functional relationship between VEGFA and AML1/RUNX1, HL-60 cells that overexpress AML1/RUNX1b were established.These cells and empty vector control HL-60 cells were stimulated with recombinant VEGFA.After stimulation VEGFA mRNA levels were reduced for 20% in HL-60 RUNX1b cells compared with HL-60 empty vector control cells (Fig. 2A).

Decreased VEGFA promoter activity in response to overexpression of AML1/RUNX1
AML1/RUNX1 encodes the alpha subunit of the Runt domain transcription factors.Therefore, the ability of this transcription factor to repress transcription of VEGFA was investigated.VEGFA promoter luciferase reporter promoter constructs were used to analyze the effect of overexpression of RUNX1 on VEGFA promoter activity.HL60 cells were cotransduced with the complete VEGFA1 promoter construct, the deletion construct VEGFA4, or the mutated construct VEGFA1-mut, and the RUNX1 expression construct.Overexpression of RUNX1 resulted in decreased VEGFA1 promoter activity, as expected, overexpression of RUNX1 had no effect on the promoter activity of the deletion construct VEGFA4 (promoter construct without AML1/ RUNX1 binding sites) or the mutated construct VEGFA1mut (Fig. 4A).This demonstrates RUNX1 specific repression of the VEGFA promoter.

ChIP analysis reveals in vivo binding of AML1/RUNX1 to the promoter of VEGFA
To investigate whether AML1/RUNX1 directly binds to the VEGFA promoter in vivo we used a ChIP assay.Three sets of PCR primers flanking the putative AML1/RUNX1-binding sites and one set of control primers were constructed (Fig. 5A).Chromatin fragments from HL60 and TF1 cells were immuneprecipitated overnight with specific antiRUNX1 antibody or control IgG antibody.All 3 sites were found to bind with RUNX1, with the highest affinity in site 1 and lowest in site 2 (Fig. 5A and 5B).Therefore, AML1/RUNX1 negatively regulates VEGFA transcription by direct binding to its promoter, suggesting that loss of AML1/RUNX1 expression may lead to VEGFA overexpression.

AML1/RUNX1-mediated repression of VEGFA is important for autocrine and paracrine AML cell survival mechanisms
To determine the effect of AML1/RUNX1 on cell survival, HL60 cells were subjected to stress conditions.After 24 hours of serum starvation the cell survival of AML1/RUNX1b transduced HL60 cells was found to be 55% decreased compared with serum conditions.In contrast, the survival of HL60 control cells was not affected by serum starvation (Fig. 6A).These data underscore the importance of Runx-mediated regulation of VEGFA expression for the survival of AML cells.
To investigate whether AML1/RUNX1-mediated inhibition of cell survival could be overcome by VEGFA, recombinant VEGFA was added to the serum free cultures.As can be appreciated in Figure 6B addition of recombinant VEGFA results in a dose-dependent increase in cell survival.This indicates that the cell survival effect of transcriptional inhibition of VEGFA by AML1/RUNX1 can be overcome by addition of VEGFA protein.
VEGFA is a strong angiogenic factor, therefore we hypothesized that modulation of VEGFA through AML1/RUNX1 will result in a different angiogenic response of endothelial cells.To test this hypothesis, HUVECs were treated with conditioned medium of AML1/RUNX1 siRNA transduced HL60 cells or scrambled siRNA transduced HL60 cells.Tube formation, used as a measure for in vitro angiogenesis, was increased after incubation with conditioned medium of AML1/RUNX1 siRNA transduced HL60 cells compared to scrambled siRNA control transduced HL60 cells (Fig. 6C).In addition, tube formation of HUVECS treated with conditioned medium of AML1/RUNX1 Interestingly, we found VEGFA mRNA expression to be increased in AML samples characterized with t(8;21).This translocation generates the fusion protein AML1/ETO consisting of the amino-terminal portion of AML1/RUNX1 including the Runt domain but lacking the carboxyl terminal transactivation domain.AML1/ETO retains the ability to bind to the consensus sequence but loses its transactivation ability on AML1/RUNX1 target genes (26).Two models have been proposed for the interference of AML1/ETO with normal AML1/RUNX1 function (Supplementary Fig. 1A; ref. 37).In the first model, AML1/ETO is a constitutive repressor of all AML1/RUNX1 target genes through binding via its runt domain (Supplementary Fig. 1B).In the second model, AML1/ETO acts as a dominant-negative molecule by competing with AML1/RUNX1 for common cofactors (Supplementary Fig. 1C; ref. 37).
We therefore investigated whether inhibition of AML1/ETO expression would have an effect on VEGFA mRNA expression using the AML cell line Kasumi-1, containing the t(8;21) translocation.When AML1/ETO is acting as a constitutive repressor factor, inhibition of AML1/ETO expression would result in elevated levels of VEGFA expression.When AML1/ ETO is, however, acting as a dominant-negative factor, blocking of AML1/ETO would result in downregulation of VEGFA expression, as AML1/RUNX1 is able to repress VEGFA expression with cofactors that are no longer titrated away by AML1/ ETO.
Silencing of AML1/ETO using siRNAs caused a 40% decrease in VEGFA mRNA expression (Fig. 7).As a control for transfection efficiency the AML1/ETO mRNA level was reduced by 90% in Kasumi-1 cells compared with treatment with the nonsilencing control siRNA.This indicates that blocking of AML1/ETO results in repression of VEGFA promoter activity and AML1/ETO acts as a dominant-negative factor in VEGFA expression.

Knockdown of AML1/ETO results in reduced functional VEGFA
To investigate the functionality of AML1/ETO-mediated upregulation of VEGFA protein expression of VEGFA was measured and the effect on a downstream target was assessed.
A significant suppression of secreted VEGFA protein was evident in the medium of Kasumi-1 cells upon transfection with a siRNA against AML1/ETO compared with medium of Kasumi-1 cells transfected with a scrambled control siRNA as determined using ELISA (Fig. 8A).Because the secreted VEGFA protein is decreased upon inhibition of AML1/ETO we tested the paracrine route of VEGFA.HUVEC cells were treated with conditioned medium of Kasumi-1 cells transfected with either a siRNA against AML1/ETO or a scrambled control RNA, and EGR3 mRNA expression was measured.A 2fold decrease in EGR3 mRNA expression was found in AML1/ ETO siRNA transfected Kasumi-1 treated HUVECs, compared with scrambled control siRNA transfected Kasumi-1 treated HUVECs (Fig. 8B).These data suggest that inhibition of AML1/ETO results in a reduction of functional VEGFA protein.

Discussion
In this study we identify 3 AML1/RUNX1 binding sites in the VEGFA promoter, we show direct binding of AML1/RUNX1 to the promoter of VEGFA and show that wild-type AML1/ RUNX1 acts as a transcriptional repressor of VEGFA.
The RUNX family of transcription factors proteins consists of 3 known members RUNX1-3.RUNX2, a bone-specific transcriptional regulator has been shown to activate the transcription of VEGFA (38)(39)(40).In contrast, screening of RUNX3 transcriptional binding to the promoter of VEGFA has revealed that RUNX3 is involved in the transcriptional repression of VEGFA (41).Our study shows for the first time that AML1/RUNX1, a gene required for hematopoiesis, is involved in the transcriptional regulation of VEGFA.In addition, AML1/RUNX1-mediated repression of VEGFA has functional implications in AML cells.Alterations of the AML1/RUNX1 gene occur at high frequency in AML.To test the effect of AML1/RUNX1 restoration on VEGFA expression we used an AML1/ETO model.We found a "restored" repression of VEGFA transcription after blocking of AML1/ETO using a AML1/ETO-specific siRNA.AML1/ETO most probably titrates away cofactors, for example, HDAC, which are necessary for AML1/RUNX1-mediated transcriptional repression.Although, in a Drosophila eye model, Wildonger and colleagues have found evidence to support a constitutive repressor model of AML1/ETO function (37), several other studies support a dominant-negative model, which is consistent with our observations in AML.For example, AML1/ETO binds CBFb more efficiently than wild-type AML1/RUNX1 (42); and AML1/ETO knock-in mice show a similar phenotype of AML1/RUNX1-deficieny mice (43,44).Specific mutations in AML1/RUNX1 have also been found to exert a dominant-negative effect on transcriptional regulation.These include in frame mutations in the N-terminal runt domain of AML1/RUNX1 and short truncating mutations in the C-terminal region of AML1/RUNX1 (22).
Recently, it was shown that RUNX1 mutations are associated with poor prognosis in patients with de novo AML (45).The association with poor prognosis may be explained by increased VEGFA levels in RUNX1 mutated blasts.
We found a significant inverse correlation between VEGFA and AML1/RUNX1 mRNA expression in AML.Recently, Silva and colleagues demonstrated a distinct gene expression profile for AML-M0 cases with a RUNX1 mutation compared with AML-M0 cases without RUNX1 mutations (46).Remarkably, VEGFA was one of the genes higher expressed in patients containing a RUNX1 mutation compared with wild-type RUNX1.
We identified the highest VEGFA expression among AMLs harboring a t(8;21) translocation compared with other cytogenetic subtypes, which is consistent with previous observations in a relatively small cohort (31).
We found that inhibition of AML1/ETO in Kasumi-1 cells resulted in a decreased VEGFA mRNA expression and, a decreased VEGFA protein secretion.In corroboration, functionally the effect of AML1/ETO inhibition was detected in the downregulation of the VEGFA responsive gene, EGR3 (34,47).In t(8;21) AML cells, a possible autocrine VEGFA/VEGFR pathway was identified.In addition, Imai and colleagues have shown VEGFA/VEGFR dependency of the t(8;21) Kasumi-1 cell line for growth and survival in contrast to the HL60 and NB4 AML cell lines which were less sensitive to VEGFR2 inhibition (48,49).This might indicate the existence of an AML subtype specific VEGFA/VEGFR dependency, initiated by alterations in RUNX1.Patients with an RUNX1 alteration might therefore benefit the most from anti-VEGFA/VEGFR therapy.
We show for the first time the involvement of AML1/RUNX1 in the transcriptional repression of VEGFA.In addition, we show that alterations in AML1/RUNX1 can result in a dominant-negative effect on the repression of VEGFA expression, a major mediator of angiogenesis, proliferation, migration and survival in AML.

Figure 2 .
Figure 2. Overexpression and silencing of RUNX1.A, HL-60 pMSCViGFP RUNX1b and empty vector control cells were stimulated with recombinant VEGFA.VEGFA mRNA levels were reduced 20% in stimulated HL-60 pMSCViGFP RUNX1b cells compared with unstimulated cells and empty vector control cells.Data are the mean AE SD of 4 replicates.The amount of VEGFA mRNA was determined 1 hour after stimulation.mRNA levels were normalized for b-actin mRNA.B, HL-60 were transfected with 500 nmol/L nonsilencing control siRNA (siCTL) and siRNA against AML1/RUNX1 (siAML1/RUNX1) by electroporation.A, decrease of AML1/RUNX1 mRNA (in grey), and increase of VEGFA mRNA (in black) by AML1/RUNX1 siRNA treatment of HL-60 and TF-1 cells.The amount of AML1/RUNX1 mRNA and VEGFA mRNA at 6 hours after transfection was determined by real-time RT-PCR.mRNA levels were normalized for b-actin mRNA.

Figure 3 .
Figure 3. RUNX1 repressor elements in the VEGFA promoter.A, location of AML1/RUNX1 binding sites relative to the 7 different promoter constructs and location of the different primer sets used in ChIP analysis.HL-60 (B) and TF-1 (C) transduced with retroviruses incorporating reporter constructs were starved in RPMI containing 0.1% FCS for 16 hours and then stimulated with 100 ng/mL GM-CSF or left untreated for 24 hours.Luciferase activities normalized by vector copy levels are shown and represent the mean AE SD of 3 replicates.Data are representative of 2 independent experiments.D, HL-60 cells were transduced with control pCBsinD-VEGFA1 construct and with 4 mutated pCBsinD-VEGFA1 constructs.Luciferase activities normalized by vector copy levels are shown and represent the mean AE SD of 3 replicates.

Figure 5 .
Figure 5.In vivo binding of AML1/RUNX1 to 3 AML1/RUNX1 binding sites in the VEGFA promoter.A and B, detection of the VEGFA promoter region containing the 3 putative AML1/RUNX1 binding sites after AML1/RUNX1 immunoprecipitation.The VEGFA promoter sequence present in IP samples and input controls was amplified by real-time PCR after ChIP using AML1/RUNX1 antibody or control IgG antibody, on cross-linked lysates of HL60 or TF1 cells.Bars represent the percentage of total input DNA for each ChIP sample.*, statistical significance P < 0.05 compared with respective control groups.

Figure 4 .
Figure 4. Promoter activity VEGFA changes upon overexpression and silencing of AML1/RUNX1.A, decreased promoter activity in HL60 cotransduced with pCBsinD-VEGFA1 and pMSCViGFP RUNX1b.No decrease in promoter activity in HL60 cotransduced with pCBsinD-VEGFA4 and pCBsinD-VEGFA1-mut.Data are the mean AE SD of 4 replicates.B, increased promoter activity in HL-60 transduced with pCBsinD-VEGFA1, after AML1/RUNX1 siRNA treatment.No increase in promoter activity in HL-60 transduced with pCBsinD-VEGFA4 after AML1/ RUNX1 siRNA treatment.Luciferase activity was measured 72 hours after transfection of siRNA.Data are the mean AE SD of 3 replicates.*, statistical significance P < 0.01 compared with respective control groups.

Figure 7 .
Figure 7. SiRNA-mediated depletion of AML1/ETO causes a decrease in VEGFA mRNA expression.Kasumi-1 cells were transfected with control siRNA or siRNA against AML1/ETO.A decrease of VEGFA mRNA in Kasumi-1 cells treated with siRNA against AML1/ETO (in black) is shown compared with control siRNA (in grey).The amounts of VEGFA and AML1/ ETO mRNA at 6 hours after transfection were determined by real-time RT-PCR.Mean AE SD of 3 replicates of VEGFA and AML1/ETO mRNA levels normalized for b-actin mRNA are shown.*, statistical significance P < 0.01 compared with respective control groups.

Figure 6 .
Figure 6.The effect of AML1/RUNX1 on cell survival and in vitro angiogenesis.A, cell survival assay.HL-60 pMSCViGFP RUNX1b and empty vector control cells were cultured under serum conditions and serum-free conditions.Bars represent the mean % of cell survival relative to the control (serum condition).B, cell survival assay.HL-60 pMSCViGFP RUNX1b cells were cultured under serum conditions, serum-free conditions, or serum-free conditions supplemented with 10 ng/mL or 25 ng/mL of recombinant VEGFA.Bars represent the mean % of cell survival relative to the control (serum condition).C, tube formation assay.Appearance of endothelial cell organization into capillary-like structures on basement membrane proteins (Matrigel) in the presence of conditioned medium of HL60 cell treated with scrambled siRNA or medium of HL60 treated with siRNA against RUNX1.D, tube formation assay.Bars represent the total skeleton length.Inhibition of RUNX1 resulted in an increase of 9% total skeleton length.

Figure 8 .
Figure 8. Inhibtion of AML1/ETO in Kasumi-1 cells results in reduction of secreted VEGFA protein level and reduction of a paracrine downstream target.Kasumi-1 cells were transfected with siRNA against AML1/ETO or scrambled control siRNA.A, secreted VEGFA protein levels were measured using ELISA.ELISA values are expressed as concentration (pg/mL) per milligram.*, statistical significance P < 0.05 compared with respective control groups.B, real-time PCR analysis of EGR3 mRNA expression in recombinant VEGFA-treated HUVECs [standard curve 0, 1, 5, 10, 20 ng/ mL of added recombinant VEGFA (white bars)] and HUVECs treated with conditioned medium of Kasumi-1 cells transfected with a siRNA against AML/ETO or a scrambled control siRNA (dark grey bars).Results are expressed as mean fold-increases above control level normalized to GAPDH expression.*, statistical significance P < 0.05 compared with respective control groups.