Regulation of Vascular Endothelial Growth Factor D by Orphan Receptors Hepatocyte Nuclear Factor-4α and Chicken Ovalbumin Upstream Promoter Transcription Factors 1 and 2 Free

Angiogenesis, the formation of new blood vessels from endothelial precursors, is a prerequisite for growth and dissemination of solid malignancies (1), and the vascular endothelial growth factor (VEGF) superfamily of endothelial growth factors has been identified to critically influence tumor-related angiogenesis (1). Recently, it was shown that basic mechanisms of hemangiogenesis also apply to the lymphatic system and that VEGF-C and its close homologue VEGF-D are intimately involved in the regulation of lymphangiogenesis and lymphatic tumor spread (2). The lymphangiogenic effects of VEGF-C and VEGF-D are essentially mediated via activation of VEGF receptor (VEGFR)-3, which is primarily expressed by lymphatic vessels but can also be found on angiogenic endothelial cells of tumor blood vessels (2). In line with their proposed role in cancer pathobiology, the expression of VEGF-C, VEGF-D, and/or VEGFR-3 has been shown to correlate with lymphatic spread, tissue invasion, and/or poor prognosis in solid malignancies (3) including gastric cancer (4). In other studies, however, conflicting observations were made (5, 6) indicating that the role of the VEGF-C/VEGF-D/VEGFR-3 system in cancer is complex and may differ between malignancies.

Despite the accumulating evidence supporting the importance of VEGF-C and VEGF-D for lymphangiogenesis and cancer spread, the current knowledge on molecular determinants and regulating pathways is very limited. In vitro studies showed that proinflammatory cytokines, including interleukin-1β and tumor necrosis factor (TNFα), as well as phorbol esters and growth factors can stimulate vegf-C expression (7–10), whereas these stimulators had no effect on vegf-D (7–10). Although resistant to various extracellular stimulators, vegf-D mRNA levels were found to be stimulated by a cell-cell contact–triggered mechanism involving cadherin-11 (8). In another report, it was shown that a β-catenin–dependent mechanism can down-regulate vegf-D expression by influencing mRNA stability (11). Because the human vegf-D cDNA was initially isolated using an activator protein (AP-1) overexpression strategy (12), it was suggested that AP-1 may be directly involved in regulation of vegf-D gene expression. Because the transcription factors and regulatory promoter elements controlling vegf-D gene expression have not yet been identified, the role of AP-1 in regulation of VEGF-D, however, remains hypothetical.

Orphan receptor hepatocyte nuclear factor 4α (HNF-4α) represents a key regulator of liver specific gene expression and influences important metabolic pathways regulating glucose and lipid homeostasis (13, 14). Additionally, hepatocyte-specific deletion of hnf-4α in mice leads to severely altered hepatocyte differentiation as well as distorted organization of sinusoidal endothelium (15). Recent studies showed that at some gene promoters, HNF-4α functionally cooperates with orphan receptors chicken ovalbumin upstream promoter transcription factor (COUP-TF)-1 and COUP-TF2 (16, 17). Both COUP-TFs display high (96–98%) homology on the protein level and influence key biological processes (18, 19): homozygous disruption of the coup-tf 1 allele in mice leads to severe defects in the central and peripheral nervous system, whereas loss of coup-tf2 results in disturbed angiogenesis and defective development of the murine heart leading to embryonic death (18, 19). Similar to HNF-4α, target genes executing the effects of COUP-TFs in context with endothelial morphogenesis and angiogenesis have not yet been described.

To characterize the molecular determinants of vegf-D gene transcription, we cloned and sequenced the human vegf-D gene promoter and found that a novel, atypical direct repeat (DR) element located in the proximal human vegf-D gene promoter is essential for vegf-D gene expression. This element is bound and activated by a complex comprising the orphan receptors HNF-4α and COUP-TF1/COUP-TF2 as well as transcriptional coactivators glucocorticoid receptor interacting protein 1 (GRIP-1) and cyclic AMP–responsive element binding protein–binding protein (CBP) to drive vegf-D gene transcription. Moreover, our results strongly suggest that the vegf-D gene is under control of epigenetic mechanisms. These results for the first time uncover the molecular components controlling vegf-D transcription and thereby add to our current understanding of mechanisms regulating the process of lymphangiogenesis.

Cell culture and transfections. Cell lines were grown in DMEM (Invitrogen) supplemented with 2 mmol/L glutamine, 100 units/mL penicillin, 100 μg/mL streptomycin, and 10% bovine calf serum in a humified atmosphere (Sanyo). Transient transfections were carried out as previously described (20) and luciferase activities were detected using the Dual-Luciferase-Reporter Assay System (Promega). In a set of experiments, cells transfected with expression constructs encoding peroxisome proliferator–activated receptor (PPAR)-γ, retinoic acid receptor (RAR)-γ, and/or retinoid X receptor (RXR)-α, together with reporter gene construct vegf-D(−135/−81), were treated with prostaglandin J2, 9-cis retinoic acid, or all-trans retinoic acid at indicated concentrations for 24 or 48 h. The same treatment was applied to transfectants receiving empty expression constructs along with vegf-D(−135/−81) (negative controls). The functionality of nuclear hormone receptor agonists and expression constructs was confirmed in a series of transfections using the reporter construct 5× HRE-Luc as a readout (data not shown). Incubations were done in triplicates and results were normalized for transfection efficiency and calculated as mean ± SE. Values were expressed as fold increases of control value. Statistical significances were calculated using Student's t test (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

Immunofluorescence in cell lines. Immunofluorescence in cell lines was done as previously described (4). Briefly, cells were grown on chamber slides, fixed, and permeabilized with saponin solution (Sigma). Following incubation with polyclonal VEGF-D antibody (C-18, Santa Cruz Biotechnology), stainings were visualized with an antigoat secondary antibody labeled with Texas red (Vector Laboratories).

Quantitative real-time reverse transcription-PCR. Quantitative reverse transcription-PCR (RT-PCR) analysis was done as previously described (4). Briefly, total RNA was isolated from cell lines or tissues using the TRIzol reagent (Invitrogen) and reversely transcribed using oligo-dT primers and SuperScript II polymerase (Invitrogen). cDNAs generated from 50 ng of total RNA were used for quantification using the TaqMan-Universal or SYBR Green PCR Master Mix (Applied Biosystems) on an ABI PRISM 7700 Sequence Detection System. Primers and probes were applied as listed in Supplementary Table S1.

DNA constructs and reporter plasmids. Using the Genome Walker Kit (Clontech) and two gene-specific primers (VEGF-D/GSP1, 5′-gcagttgacatcaggcagctagag-3′, and VEGF-D/GSP2, 5′-caaaaggcattctccagaaggaaag-3′) located within the first exon of the vegf-D gene (NM_004469), a genomic fragment comprising 2,011 nucleotides of the 5′ flanking region of the human vegf-D gene as well as 110 nucleotides of noncoding exon 1 was isolated (GenBank accession no. AY874421). Subcloning of this DNA fragment into the promoterless luciferase reporter gene vector pGL3 (Promega) yielded the promoter-reporter gene construct vegf-D(−2011/+110). Using vegf-D(−2011/+110) as a template, 5′-deletion constructs were generated using a common 3′ primer and different 5′ primers (Supplementary Table S2) as previously described (20). To study vegf-D promoter elements in a heterologous promoter system, vector pT81-Luc, in which the firefly luciferase gene is under control of the enhancerless herpes simplex thymidine kinase viral promoter, was used as described earlier (20). Site-directed mutagenesis was done with the QuikChange kit (Stratagene). To increase efficiency and reliability of mutagenesis, construct vegf-D(−264/+110) was used as a template instead of a plasmid containing a longer fragment of vegf-D 5′ sequence. Primers for mutagenesis were applied as listed in Supplementary Table S2. Constructs encoding COUP-TF1, COUP-TF2, HNF-4α, its mutant HNF-4α-DBD (DNA binding domain deletion Δ49–115), or RAR isoforms α, β, and γ, as well as RXRα, have been described before (21–23). Similarly, the construct encoding wild-type PPARγ was previously used (24).

Electrophoretic mobility shift assays. Electrophoretic mobility shift assays (EMSA) were carried out as previously described (20). In brief, nuclear protein extracts (5 μg) were incubated with [γ-³²P]ATP–radiolabeled double-stranded oligonucleotides (Supplementary Table S3). For competition experiments, nuclear extracts were incubated with 100-fold molar excess of double-stranded, unlabeled competitor oligonucleotides (Supplementary Table S3). For supershift experiments, nuclear extracts were incubated with antibodies recognizing PPARγ, RARγ, RXRα (Santa Cruz Biotechnology), HNF-4α (25), or COUP-TF (26).

Chromatin immunoprecipitation. Chromatin immunoprecipitation assays were carried out as previously described (27) with 5 μg of anti–acetylated H3 or H4 antibodies (Upstate Biotechnologies) and rabbit pre-immunosera. Quantitative amplification using the qPCR SYBR Green Core Kit (Eurogentec) was done according to the manufacturer's conditions and results were normalized to input controls.

Trichostatin A and 5-aza-2-deoxycytidine treatment of cells. Exposure of cells to indicated concentrations of trichostatin A (TSA) and/or 5 μmol/L 5-aza-2-deoxycytidine (Sigma-Aldrich) was done for 24 and/or 72 h. Control incubations received the appropriate solvent for identical time periods. In experiments examining the influence of TSA on vegf-D promoter-reporter gene constructs, transfected cells were exposed to 100 ng/mL TSA or vehicle for 24 h. To investigate the influence of hypoxia on TSA-induced effects, cells were treated with various concentrations of TSA or solvent and were immediately afterwards transferred to normoxic or hypoxic conditions for 24 h.

vegf-D gene expression and promoter activity in cancer cell lines. Using a specific anti–VEGF-D antibody, we detected a paranuclear granular staining pattern in all cell lines investigated (Fig. 1A). In line with staining results, real-time PCR analysis revealed low levels of vegf-D mRNA in AGS, KATO-III, and MKN-28 cells, whereas MKN-45 cells showed 5- to 8-fold higher amounts of vegf-D mRNA (Fig. 1B). Compared with the empty pGL3 construct, transfection of vegf-D(−2011/+110) into MKN-45 cells produced a 6-fold increase in luciferase activity (Fig. 1C). In contrast, the activity of vegf-D(−2011/+110) in other gastric cancer cell lines was close to pGL3 control.

A DR element between −138 and −79 is indispensable for vegf-D promoter activity. To identify the basal enhancer element(s) controlling vegf-D gene promoter activity, functional 5′ deletion analysis was done in MKN-45 cells. Whereas progressive deletion from −2011 to −141 had no influence on vegf-D promoter activity, loss of additional 20 nucleotides [vegf-D(−122/+110)] reduced promoter activity by 50% (Fig. 2A). Shortening of the vegf-D 5′ sequence by additional 40 bp [construct vegf-D(−83/+110)] abolished vegf-D transcription (Fig. 2A). Because these results indicated that vegf-D −141 to −83 contains essential regulatory elements, this sequence was used as a radiolabeled probe (probe A) in EMSAs. In parallel, a single copy of the fragment was subcloned into the heterologous, enhancerless luciferase reporter construct pT81 and analyzed in functional transfection studies. For further analysis, the vegf-D −138/−79 fragment was split into a 5′ [−138 to −109 (probe B)] and a 3′ fragment [−108 to −79 (probe C)]. Computer-assisted sequence analysis (28) revealed the presence of two consensus binding sites for GATA transcription factors and a consensus DR half-site at −119 to −125 within the −138/−79 sequence (Fig. 2B). In EMSAs, incubation of probe A with nuclear extracts yielded two typical nuclear complexes (Fig. 2C, lane 1). The lower complex (complex II) bound to probe A was also formed at probe B (−138/−109; Fig. 2C, lane 2), whereas complex I was not found with probe B but was clearly detectable with probe C (Fig. 2C, lane 3). In functional transfection assays, vegf-D −138/−79 showed a 10- to 15-fold higher transcriptional activity when compared with empty pT81, showing that this sequence confers true enhancer characteristics (Fig. 2D). In contrast, although showing unaffected binding of complex II or complex I, the activity conferred by vegf-D −135/−109 and vegf-D −109/−81 in pT81 was drastically reduced (Fig. 2D), suggesting that enhancer elements exist within the −138/−79 region that interact in a cooperative fashion.

After confirming the enhancer characteristics of vegf-D(−135/−79), this sequence was investigated by site-directed mutagenesis in context of a larger vegf-D fragment [vegf-D(-264/+110)]. Whereas individual or simultaneous disruption of GATA sites had no significant influence, mutation of the 5′ DR half-site at −125/−119 (mutant ΔDR5′) reduced vegf-D promoter activity by 50% (Fig. 3A). Moreover, mutation of the sequence vegf-D −99/−94, which resembles characteristics of an degenerated DR half-site (mutant ΔdDR3′), blunted vegf-D promoter activity, whereas simultaneous mutation of −125/−119 and −99/−94 abolished vegf-D transcription (Fig. 3A). EMSA competition experiments confirmed that the mutations used in transfection assays were sufficient to abrogate formation of complex I at the 3′ dDR element (competitor ΔdDR3′) as well as formation of complex II at the 5′ DR half-site (competitor ΔDR5; Fig. 3B).

HNF-4α and COUP-TF transcription factors bind to the vegf-D DR element. DR elements are typically bound by nuclear receptors and/or orphan receptors (29). To evaluate whether such proteins interact with the vegf-D DR element, EMSA competition experiments using consensus binding sequences for PPAR, RAR, and RXR as unlabeled competitors and vegf-D −138/−79 as radiolabeled probe were conducted. Application of PPAR and RXR competitors abolished formation of complex II at the vegf-D −138/−79 probe (Fig. 3C, lanes 2 and 6), whereas addition RAR oligonucleotides disrupted binding of both complex I and complex II (Fig. 3C, lane 4). The lack of effectiveness observed for mutated competitors confirmed the specificity of results obtained with wild-type oligonucleotides (Fig. 3C, lanes 1, 3, 5, and 7). To further confirm findings from competition studies, EMSA supershifts with antibodies recognizing PPARγ, RARγ, RXRα, HNF-4α, and/or COUP-TF1/COUP-TF2 were done. At the vegf-D 138/−79 probe, α-PPARγ and/or α-RXRα antibodies reduced formation of complex II, whereas complex I was not altered (Fig. 3D, lanes 2, 4, and 5). In contrast, α-RARγ antibody did not influence complex formation (Fig. 3D, lane 3) unless it was applied together with α-PPARγ or α-RXRα antibodies, which disrupted complex II (Fig. 3D, lanes 5 and 7). In contrast, application of HNF-4α antibody diminished formation of complexes I and II at the vegf-D −138/−79 probe and produced a supershift with probes B and C (Fig. 3D, lane 8). Similarly, anti–COUP-TF1/COUP-TF2 antibody reduced complex formation at the vegf-D −138/−79 probe (Fig. 3D, lane 9, top) and produced supershifts from probes B and C (Fig. 3D, lane 9, middle and bottom). Previous studies suggested that transcription factor AP-1 may regulate vegf-D gene transcription via an AP-1 consensus site located in the human vegf-D 5′-flanking region (12). Our studies revealed that the loss of a putative AP-1 site at −192 had no influence on basal vegf-D promoter activity (Fig. 2A). Similarly, an additional AP-1 core motif at −54 was not able to compensate for the loss of −136/−79 (Fig. 2A), confirming that these AP-1 binding sites are not relevant for control of basal vegf-D gene transcription.

Ectopic expression of HNF-4α and COUP-TF1/COUP-TF2 stimulates vegf-D promoter activity. In contrast to PPARγ, RARγ, or RXRα, which had no influence on vegf-D transcription regardless whether experiments were carried out with or without stimulation, ectopic expression of HNF-4α as well as COUP-TF1 and COUP-TF2 significantly stimulated vegf-D promoter activity (Fig. 4A, data for COUP-TF2 not shown). In accordance with results obtained in gastric cancer cells, enforced expression of wild-type HNF-4α concentration-dependently stimulated vegf-D promoter activity in HNF-4α–deficient HeLa cells, confirming the direct effects of HNF-4α on the vegf-D gene promoter. Expression of mutant HNF4-DBD, which lacks a functional DNA binding domain (30), had no effect (Fig. 4B). Similarly, ectopic expression of COUP-TF1 or COUP-TF2 stimulated vegf-D promoter activity 4- to 5-fold in HNF-4α–negative HeLa cells (Fig. 4C), whereas coexpression of COUP-TF1 or COUP-TF2 along with HNF-4α was not more effective. To confirm that HNF-4α activates the vegf-D promoter via the vegf-D DR element, reporter constructs carrying mutated or wild-type vegf-D DR sequences were used (Fig. 4D). Mutation of the 5′ DR half-site abrogated HNF-4α–triggered promoter activity, whereas the mutant targeting the 3′ DR half-site reduced vegf-D promoter activity by ∼50%. Loss of both DR half-sites (mutant ΔDR5′ + ΔDR3′) abrogated the stimulatory effect of HNF-4α.

Transcriptional coactivators GRIP-1 and CBP bind and activate the vegf-D gene promoter. Previous reports showed that HNF-4α and COUP-TFs cooperate with p160 transcriptional coactivator GRIP-1 (31, 32) and/or CBP (33) to drive target promoters. To evaluate whether these two cofactors may be involved in vegf-D regulation, chromatin immunoprecipitations were carried out. In accordance with EMSA studies, HNF-4α was found at the proximal vegf-D promoter; however, we were not able to find COUP-TFs although several commercial antibodies and experimental conditions were tested. In addition to HNF-4α, coactivators CBP and GRIP-1 were found to be associated with the vegf-D promoter (Fig. 5A). Interestingly, in HNF-4α–negative HeLa cells, CBP, but not GRIP-1, was detected in the vegf-D binding nuclear complex, suggesting that interaction of CBP may require the presence of HNF-4α (Fig. 5A). To investigate the functional effect of GRIP-1 and CBP on vegf-D regulation, ectopic expression experiments were carried out. In HNF-4α– and COUP-TF–competent AGS cancer cells, expression of CBP or GRIP-1 significantly elevated vegf-D promoter activity (Fig. 5B), showing a more robust transcriptional response when both coactivators were coexpressed. In HNF-4α–deficient HeLa cells, neither GRIP-1 nor CBP significantly stimulated vegf-D promoter activity (Fig. 5C). Expression of GRIP-1 and/or CBP together with HNF-4α resulted in a 3- to 4-fold elevation of vegf-D promoter activity (Fig. 5C), strongly suggesting that the presence of HNF-4α is required for GRIP-1 and CBP to exert their transcriptional effects on the vegf-D gene.

Regulation of vegf-D gene transcription by histone H3 and H4 acetylation. Because previous reports showed that application of histone deacetylase inhibitors can influence expression of angiogenesis genes including vegf-A (34, 35), we investigated the effect of TSA on vegf-D mRNA levels in various cancer cell lines. We found that TSA treatment potently stimulated vegf-D mRNA levels (results for AGS cells shown in Fig. 6A and B). Similarly, vegf-C mRNA levels were potently up-regulated by TSA (Fig. 6B). Application of TSA potently stimulated the activity of a transfected vegf-D reporter gene construct in HeLa and AGS cells (Fig. 6C). Additionally, chromatin immunoprecipitation assays revealed that TSA treatment resulted in a 10- and 20-fold increase of acetylated histones H3 and H4 at the vegf-D promoter (Fig. 6D), strongly supporting the concept that histone deacetylase inhibition stimulates vegf-D transcription through histone-dependent chromatin remodeling.

In this study, we identify a novel, atypical DR element located in the proximal vegf-D gene promoter as indispensable for vegf-D gene transcription. DR elements were initially identified as binding sites for ligand-activated hormone receptors such as RAR, RXR, and PPARs as well as orphan receptors like HNF-4α and COUP-TFs (36). Typical DR elements consist of two consensus 5′-AGGTCA-3′ half-sites spaced by one to five nucleotides (36, 37). In addition, atypical DR sites comprising degenerated variants of the consensus motif, palindromic configuration, and/or spacing sequences of up to 150 bp were described (36, 37). Importantly, the nucleotide sequence of DR half-sites as well as their spacing sequence determines the selectivity of DR elements toward binding nuclear factors (29). Based on its primary structure comprising a degenerated 3′ DR half-site sequence separated from a consensus half-site by a 20-bp spacer, the vegf-D DR element has to be classified as an atypical DR site. To our knowledge, structure and sequence of the vegf-D DR element are unique and DR elements with this configuration have not yet been described. In accordance with the concept that the half-sites of a given DR element cooperate functionally, we found that individual vegf-D DR half-sites conferred only very weak transcriptional activity, whereas the complete vegf-D DR element potently elevated (10–15-fold) transcription in a heterologous promoter system (Fig. 2). Accordingly, site-directed mutagenesis experiments showed that the integrity of both DR half-sites is indispensable for full vegf-D gene promoter activity (Fig. 3). Previous studies showed that DR elements can bind nuclear receptors without showing responsiveness toward corresponding receptor agonists (38). In line with this observation, we detected binding of PPARγ, RARγ, and RXRα to the vegf-D DR element, whereas application of appropriate agonists failed to stimulate vegf-D gene transcription. This may be explained by DNA-dependent allosteric effects influencing the activating capacity and/or ligand-binding abilities of hormone receptors involved (38). Alternatively, such effects are due to the lack of transcriptional cofactors required for the inducibility of a given complex forming at a DR element (38).

In contrast to PPARγ, RARγ, and RXRα, we found that orphan receptors HNF-4α and COUP-TF1/COUP-TF2 are potent activators of the human vegf-D gene promoter. Because mutations abrogating the binding of HNF-4α and COUP-TF1/COUP-TF2 to the vegf-D DR element also blunted vegf-D transcription, the presence of HNF-4α and COUP-TF1/COUP-TF2 at the proximal DR element seems to be indispensable for vegf-D transcription. Transcriptional effects of HNF-4α are typically mediated via DR1-like response elements (13, 14), but interaction of HNF-4α with DR2-like elements or single atypical DR half-sites has also been shown (39, 40). Similar to our findings with the vegf-D DR element, HNF-4α–activated DR sites were previously shown to be regulated by orphan receptors COUP-TF1/COUP-TF2 (16, 17). COUP-TFs preferentially interact with DR1 elements, but interaction with DR0, DR2, DR6, DR7, and DR9 sites as well as everted DRs has been reported (for review, see ref. 41). Interestingly, COUP-TFs were initially considered as transcriptional repressors of nuclear receptor action (41), but more recently, it became clear that COUP-TFs can also act as transcriptional activators via activation of DR elements (42, 43). In the light of in vivo studies showing that targeted disruption of HNF-4 and/or COUP-TF2 in mice influences endothelial morphogenesis and angiogenesis, our results make it tempting to speculate that at least some of the vascular effects of HNF-4 and/or COUP-TF2 are mediated by VEGF-D.

Activation of gene transcription by DNA-binding transcription factors is mediated by coactivators recruiting RNA polymerase II and its transcription initiation complex to target promoters (44). Complexes formed on stimulation of nuclear receptors frequently comprise p160 transcriptional coactivators such as GRIP, SRC-1, and/or pCIP (44–46). Moreover, coactivator CBP and/or its functional homologue p300, both of which possess histone acetylase activity, were detected in such complexes (45). p160 proteins typically link other coactivators to the transcription initiation complex, whereas CBP/p300 can directly acetylate histones and/or other components of the transcription initiation complex (47). Similarly, orphan receptor HNF-4α was shown to interact with CBP/p300 and GRIP-1 (48, 49). In line with these studies, we found that CBP and GRIP-1 were associated with HNF-4α at the proximal vegf-D promoter (Fig. 5A). Functional expression studies confirmed that GRIP-1 and CBP contribute to the regulation of the human vegf-D promoter and strongly suggested that the presence of HNF-4α is required for GRIP-1 and CBP to exert their transcriptional effects. To our knowledge, our study for the first time provides an analysis of coactivator proteins regulating the transcription of a gene of the VEGF ligand/receptor family.

The acetylation/deacetylation of histones plays an important role in the regulation of gene expression (50). The acetylation status of histones is controlled by the opposing activity of two types of enzymes: histone acetylases and histone deacetylases (50). Most of the identified histone acetylases act as transcriptional coactivators, whereas histone deacetylases function in opposition to histone acetylases by deacetylating lysine residues on histone tails. Inappropriate recruitment of histone deacetylases in malignant cells has been identified to result in activation of transcriptional programs supporting the onset and progression of cancer. Conversely, histone deacetylase inhibitors such as TSA were shown to display potential as anticancer agents (50), an effect that could be partially mediated through inhibition of proangiogenic factors such as VEGF-A (34). As a potential mechanism underlying TSA-dependent suppression of the vegf-A gene, interruption of hypoxia-inducible factor 1α–dependent vegf-A activation was proposed (34, 35). Interestingly, in our study, application of TSA stimulated accumulation of acetylated histones H3 and H4 at the vegf-D promoter and potently induced vegf-D gene transcription (Fig. 6). How reversible epigenetic silencing of the vegf-D gene contributes to control of vegf-D expression in the setting of vascular differentiation and/or cancer, however, remains to be elucidated in future investigations.

We thank Anja Beyer and Feng Gao for excellent technical assistance, Dr. S. Tarivas (Division of Molecular Biology of the Cell I, German Cancer Research Center, Heidelberg, Germany) for providing the HNFα expression plasmids, Dr. M.M. Moasser (Laboratory of Molecular Medicine, Sloan-Kettering Institute, Memorial Sloan-Kettering Cancer Center, New York, NY) for providing the RARγ and RXRα plasmids, Dr. M. Gurnell (Department of Medicine, University of Cambridge, Addenbrooke's Hospital, Cambridge, United Kingdom) for providing the PPARγ plasmid, and Dr. M.J. Tsai (Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, TX) for the expression constructs encoding COUP-TF1 and COUP-TF2 and for the antibody against COUP-TFs.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.