Defects of the heart, eye, and megakaryocytes in peroxisome proliferator activator receptor-binding protein (PBP) null embryos implicate GATA family of transcription factors.

Peroxisome proliferator activator receptor (PPAR)-binding protein (PBP) is an important coactivator for PPARgamma and other nuclear receptors. It has been identified as an integral component of a multiprotein thyroid hormone receptor-associated protein/vitamin D(3) receptor-interacting protein/activator-recruited cofactor complexes required for transcriptional activity. Here, we show that PBP is critical for the development of placenta and for the normal embryonic development of the heart, eye, vascular, and hematopoietic systems. The primary functional cause of embryonic lethality at embryonic day11.5 observed with PBP null mutation was cardiac failure because of noncompaction of the ventricular myocardium and resultant ventricular dilatation. There was a paucity of retinal pigment, defective lens formation, excessive systemic angiogenesis, a deficiency in the number of megakaryocytes, and an arrest in erythrocytic differentiation. Some of these defects involve PPARgamma and retinoid-sensitive sites, whereas others have not been recognized in the PPAR-signaling pathway. Phenotypic changes in four organ systems observed in PBP null mice overlapped with those in mice deficient in members of GATA, a family of transcription factors known to regulate differentiation of megakaryocytes, erythrocytes, and adipocytes. We demonstrate that PBP interacts with all five GATA factors analyzed, GATA-1, GATA-2, GATA-3, GATA-4, and GATA-6, and show that the binding of GATA-1, GATA-4, and GATA-6 to PBP is not dependent on the nuclear receptor recognition sequence motif LXXLL (where L is leucine and X is any amino acid) in PBP. Coexpression of PBP with GATA-3 markedly enhanced transcriptional activity of GATA-3 in nonhematopoietic cells. These observations identify the GATA family of transcription factors as a new interacting partner of PBP and demonstrate that PBP is essential for normal development of vital organ systems.

Peroxisome proliferator activator receptor (PPAR) 1 -binding protein (PBP) was originally identified as a coactivator for PPAR␥ (1) and later shown to participate in the transactivation of other nuclear receptors (1)(2)(3)(4)(5). Some of the nuclear receptor cofactors such as CBP/p300 (6), SRC-1 (7), and ACTR (8), possess intrinsic histone acetyltransferase activity and modify the chromatin organization of the target gene promoter regions, but PBP lacks histone acetyltransferase activity (9,10). PBP has been found to be a key component of the thyroid hormone receptor-associated protein (TRAP) (3), vitamin D 3 receptor-interacting protein (DRIP) (4), and activator-recruited cofactor (5) complexes. In these multiprotein complexes, PBP designated as TRAP220/DRIP205 serves as the major anchor for bridging various components with activated nuclear receptors (3-5, 9, 10). Recent evidence suggests that PBP is involved not only in the activation of nuclear receptors but also in the activation of other groups of transcription factors. One such example is a PBP-containing cofactor complex cofactor required for Sp1 that was purified from HeLa cells and found to be required for transcriptional activation by Sp1 in association with the RNA polymerase II apparatus (11). We and others (12,13) have reported that the ablation of PBP/TRAP220 gene resulted in embryonic lethality around gestational day 11.5 (E11.5). The PBP null placenta failed to become adequately vascularized, similar to the placenta of PPAR␥-deficient embryos (12,14), and PBP null-derived fibroblasts had a diminished ability for ligand-dependent transcriptional activation of PPAR␥ (12), supporting the requirement of PBP for PPAR␥ functions in vivo.
We now report that PBP null embryos exhibit a diverse group of developmental defects involving the heart, eye, megakaryocytes, erythrocytes, and vasculature. Some of these findings involve PPAR␥ or retinoid-sensitive tissues; however, several cell-specific abnormalities have not been attributed to the PPAR-signaling pathway, suggesting PBP may interact with additional transcription factors. We now report that PBP null embryos exhibit phenotypic overlap in four organ systems with embryos deficient in various members of the GATA family of zinc finger transcription factors. Here, we further demonstrate that PBP interacts with GATA-1, GATA-2, GATA-3, GATA-4, and GATA-6, the members of the zinc finger-containing GATA family of multifunctional transcription factors. The GATA family of transcription factors controls the development and differentiation of a wide spectrum of cell lineages including cardiac muscle, megakaryocytes, placental trophoblasts, adipocytes, and T-lymphocytes (15)(16)(17)(18)(19)(20). GATA-1 is essential for differentiation and maturation of megakaryocytes and erythrocytes (21), although the recruitment of cofactor, FOG (friend of GATA), is needed to regulate cell type-specific gene transcription (22)(23)(24). GATA-1 can be modified by acetylation (25) and interacts with the transcriptional coactivator CBP (26). GATA-2 is abundantly expressed in hematopoietic and nonhematopoietic tissues (27), whereas GATA-3 is more restricted and is expressed in T-lymphocytes, specific neurons, the endocardial cushion, and lens of the eye (15,19,28). The identification of FOG-1 (22) and FOG-2 (24) and the interaction of GATA-1 with the general coactivator CBP (26) and of GATA-4 with p300 (29) have prompted speculation that GATA factors may participate in multiprotein-DNA complexes. The similarities in the phenotype of certain organs of PBP null embryos with embryos deficient in various members of the GATA family along with the dependence of hematopoietic differentiation on GATA-1 (21) and the role of PPAR␥ in a GATA-2-and GATA-3-dependent transition of preadipocytes to adipocytes (19) led us to test members of the GATA family as activation targets for PBP. We found that PBP is capable of binding all five GATA factors analyzed, GATA-1, GATA-2, GATA-3, GATA-4, and GATA-6, and that the binding of mouse and chicken GATA-1 and mouse GATA-4 and GATA-6 was independent of the conserved nuclear receptor-interacting sequence motif LXXLL (where L is leucine and X is any amino acid) (9,10,30,31). When PBP was coexpressed with GATA-3, an augmentation of GATA-dependent transcriptional activity was observed. These observations, taken together with the phenotypic overlap of PBP null embryos with mice deficient in PPAR␥, retinoid X receptor, and the GATA family strongly suggest that PBP acts as a pivotal coactivator in orchestrating the complex array of transcriptional signals required for normal embryonic development.

MATERIALS AND METHODS
Mutant Mice-A generation of mutant mice carrying disrupted PBP gene was performed as described previously (12). Embryos from heterozygous intercross were collected at gestational day 11.5, and the DNA from the amnion was used for genotyping by PCR using primers 5Ј-CCTTCTTTCTCCGCAGTCAC-3Ј and 5ЈAGTGATGAGTTCATA-CAGGGG-3Ј to detect the wild type allele and primers 5Ј-CCACAGTC-GATGAATCCAGAA-3Ј and 5Ј-TGAATGAACTGCAGGACGAGG-3Ј to detect the targeted allele. Twenty-two embryos (wild type, n ϭ 11; null embryos, n ϭ 11) were used for gross and microscopic examination.
Histological Examination and Immunohistochemistry-Agematched (E11.5) wild type (n ϭ 11) and PBP null embryos (n ϭ 11) were fixed in paraformaldehyde or 10% buffered formalin, embedded in paraffin, serially sectioned at 5 M thickness in sagittal or transverse planes, and stained with hematoxylin-phloxine-saffarin. Sections of normally developed wild type murine eyes (n ϭ 4) and bone marrow (n ϭ 3) were used to compare localization of PBP at different stages of development. Immunohistochemical stains were performed using a standard avidin-biotin-peroxidase complex protocol as described previously (32) and an affinity-purified rabbit polyclonal antibody directed against PBP (12). The slides were counterstained with hematoxylin and reviewed in a blinded fashion. Positivity was denoted by a brown color and its intensity graded as 1ϩ (low) to 3ϩ (strong). The specificity of the antibody was confirmed previously with in vitro translated full-length PBP protein and Western blotting with the nuclear extract prepared from HeLa cells (12).
GST Pull-down Assays-GST-cGATA-1, GST-cGATA-2, and GST-cGATA-3 are chicken GATA proteins that were expressed as GST fusion proteins by the bacteria using the expression vector pGEX-2T (Amersham Biosciences, Inc.). GST-mGATA-1, GST-mGATA-3, GST-mGATA-1 N-terminal zinc finger, and GST-mGATA C-terminal zinc finger) are mouse GATA proteins expressed as GST fusion proteins. The GST and GST-GATA fusion proteins were produced in Escherichia coli BL21 and bound to glutathione-Sepharose beads according to the in-structions by the manufacturer (Amersham Biosciences, Inc.). The 35 Slabeled full-length PBP was produced by in vitro translation using rabbit reticulocyte lysate (Promega). pCMV-PBP used in the in vitro translation was as described previously (1,2). To determine the PBP region(s) interacting with cGATA-1 or mGATA-1, truncated PBP was generated from T 7 promoter-linked PBP cDNA fragments by in vitro translation. In the GST pull-down assay, GST-fusion protein from 10 ml of bacteria culture bound to glutathione-Sepharose beads was incubated with 10 l of [ 35 S]methionine-labeled PBP protein for 2 h in 500 l of NETN (20 mM Tris-HCl, pH 7.5, 100 mM KCl, 0.7 mM EDTA, 0.05% Nonidet P-40, 1 mM phenylmethylsulfonylfluoride). Beads were washed three times with binding buffer, and bound protein eluted by boiling for 2 min in 20 l of SDS sample buffer, analyzed by SDS-PAGE, and subjected to autoradiography. To assess the interaction of mGATA-4 and mGATA-6 with PBP, 35 S-labeled in vitro translated full-length mGATA-4 and mGATA-6 were used in GST pull-down assays with GST-PBP-truncated fusion proteins.
Cell Culture and Transfection-Mammalian expression vectors expressing cGATA proteins were constructed with TFAneo backbone (27), and the transcription was directed by Rous sarcoma virus-long terminal repeat promoter and enhancer. C3␤-hGH reporter construct contains the rabbit ␤-globin TATA box and three copies of the C␤E oligonucleotide in plasmid p0GH (27). QT6 (immortal quail fibroblasts) cells were grown in Dulbecco's modified Eagle's medium containing 10% newborn calf serum, 2% chicken serum, and 2% tryptose phosphate broth, and NIH 3T3 cells were cultured in Dulbecco's modified Eagle's medium with 10% calf serum. Cells plated in six-well plates were transfected 24 h later with hGH reporter (0.25 g), the appropriate expression plasmid DNAs (0.5 g), GATA and PBP, and 0.5 g of ␤-galactosidase expression vector pCMV␤ (CLONTECH) DNA using CaPO 4 transfection methods (CLONTECH) according to manufacturer instructions. Transfection without pCMV-PBP was compensated by adding the same amount of pCDNA3.1. After 24 h, the medium containing secreted hGH was collected and assayed by radioimmunoassay using the Allegro hGH system from Nichols Institute Diagnostics. The cell extracts were used to assay ␤-galactosidase activity (Tropix).

RESULTS
PBP Immunolocalized to the Eye, Heart, Hematopoietic Elements, and Nervous System-To determine the expression pattern of PBP, immunohistochemical studies were performed on adult murine tissues using an anti-PBP antibody (12). Overall, the cellular distribution of PBP closely paralleled the sites of morphological disturbances in the null embryos. Sections of the eye showed strong (3ϩ) staining of the retinal pigment epithelial cells (Fig. 1a, arrows), the corneal epithelial and endothelial cell layers (see Fig. 1b), lens epithelium, vascular endothelial cells, and the retinal ganglion cell layer (Fig. 1c, arrow). A well defined band of positivity was observed within the interphotoreceptor region of the retina (Fig. 1c, arrowhead). To assess whether the neural retina was only nerve-related tissue expressing PBP, sections from the murine brain and spinal cord were stained. Intense staining (3ϩ) was noted in cerebellar granule cells, vascular endothelial cells, and selective cells of the spinal ganglia ( Fig.  1d, arrowhead) and its adjacent nerve (Fig. 1d, arrow).
Loss of PBP affected two other major organ systems, the heart and the hematopoietic system. Sections of the heart showed strong (3ϩ) immunolocalization to the myocardial cells (see myocardium adjacent to ventricle in Fig. 1e), endothelial cell nuclei within the endocardium, and the endocardial cushion near the ventricular outflow tract (see OT, Fig. 1e). When sections of bone marrow were examined, the population of megakaryocytes consistently demonstrated cytoplasmic and nuclear positivity for PBP (see arrows in Fig. 1f), whereas focal staining was evident in erythroid elements. These observations reveal that PBP immunolocalizes to a wide spectrum of organ systems, many of which are targets for pathological complications in PPAR␥-dependent disease processes (14).
Noncompaction of the Ventricular Myocardium Contributes to Cardiac Failure in PBP Null Embryos-Early embryonic demise at E11.5 in PBP null embryos is attributed to severe cardiac failure because of underlying maldevelopment of the heart (12,13). In all embryos, there was a large pericardial effusion associated with a dilated, blood-filled ventricle (Fig.  2a, arrow). Histologically, the heart of the null embryos demonstrated noncompaction of the ventricular myocardium and trabecular hypoplasia resulting in a thin ventricular wall. Unlike the thickened ventricle of the wild type embryos, the ventricular myocardium of PBPϪ/Ϫ embryos failed to stratify into the multilayer structure that is required to sustain adequate cardiac function (Fig. 2, compare c with d). The interface between the endocardial endothelial cells and myocardium appeared to be the region where striking differences are evident between controls and null embryos. The heart endothelium (endocardium) is a key regulator of myocardial growth. It provides signals to facilitate the endothelial-mesenchymal transformation that is a prerequisite for normal ventricular development (33). In the control hearts, the endocardium made direct contact with the stratified myocardial cells, whereas in all of the PBPϪ/Ϫ embryos, the endothelial cells separated from the underlying pyknotic-appearing myocardial cells (Fig.  2f, arrows). Similar contact gaps exist at the interface of the endocardium and myocardium in other lines of mice with hypoplastic ventricles including those deficient in PPAR␥, retinoid X receptor ␣, or GATA-3 (14,15,34,35). Taken together, these data suggest that impaired signaling at the endocardial-myocardial junction may account for the loss of myocytes contributing to ventricular hypoplasia.
Defective Angiogenesis in Multiple Organ Systems in PBP Null Embryos-Maintaining a delicate balance of angiogenic mediators is important in the placenta during the early phases of development (36,37). Cells of trophoblast lineage are especially important, because they secrete several potent angiogenic factors and must interact with both mesenchymal and vascular components to permit the placenta to function as a rich blood source for the growing fetus (37,38). GATA-2 and GATA-3 are highly expressed by placental trophoblasts (16), and differentiating trophoblasts are influenced by PPAR␥ (39) and acquire several endothelial cell surface markers (40). We previously noted that the E11.5 PBP null placentas were histologically different from the wild type controls (12). Unlike the controls, the PBPϪ/Ϫ placentas lacked a complex capillary network and exhibited marked erythrophagocytic activity within the trophoblasts (Fig. 3, compare a with b). Given the critical role of the trophoblast in secreting pro-angiogenic mediators, it is plausible that dysfunctional trophoblasts create an imbalance in placental angiogenesis.
The placental disturbances in angiogenesis appeared quite distinct from those that occurred within the PBP null embryo itself in which vessels were in excess. Grossly, all null embryos could be recognized by a generalized hyperemic appearance (Fig.  3d). This feature was in contrast to the more uniformly tanned appearance of the wild type controls (Fig. 3c). Microscopically, the hyperemia corresponded to a marked increase in vascularization (not illustrated), especially notable along the paraspinal region (Fig. 3d, arrows) within the developing brain and the retina. In some instances, vascular permeability appeared impaired with extravazation of red blood cells into adjacent tissue. These data suggest that PBP may function as a negative regulator of endothelial cell proliferation possibly through a loss of PPAR␥ regulation of angioinhibitory activity (41).
Ablation  (43). When the eyes of E11.5 PBP null mice were examined, significant eye abnormalities had already emerged. On the gross level, the eyes of all null embryos were difficult to visualize because of the absence of a pigmented ring when compared with the well defined structure in the controls (Fig. 4, compare b with a). These findings were confirmed at the histological level at which the wild type eyes had a linear rim of pigment (Fig. 4, c and e, arrows) and the null eyes showed a paucity of retinal pigment (Fig. 4, d and f,  arrows), a feature also described in GATA-3-deficient mice (15). The lens of the PBP null embryo eyes was poorly developed, and it failed to form the lining epithelium that was clearly present in the wild type controls (Fig. 4, compare d with f). In addition, small vessels surrounding the rudimentary lens of all null embryos were consistently more numerous than control eyes. These observations support a role for PBP in normal eye development and suggest that it might regulate the neural retina as well as influence the functions of the retinal pigment epithelial cells and its associated vasculature.
Loss of PBP Altered Megakaryocytic and Erythrocytic Cell Lineages-Unlike most organ systems, hematopoiesis is a dynamic process that must shift from the placenta to the liver with definitive hematopoiesis residing within the bone marrow by maturity (44). Disturbances in one or more hematopoietic cell populations can trigger systemic complications ranging from anemia to massive hemorrhage (15). PBP null embryos showed abnormalities in both megakaryocytes and erythrocytes, two cell lineages that are known to arise from a common progenitor and are regulated by similar families of transcrip-tion factors (45). In the wild type controls, multinucleated megakaryocytes were easily visualized (Fig. 5, a and b, arrows), whereas the liver of null animals lacked mature megakaryocytes (Fig. 5, c and d). Instead, many of the cells showed a high nuclear to cytoplasmic ratio with a significant number of cells showing fragmented nuclei suggestive of apoptosis.
Similar to the GATA-1-or GATA-3-deficient mice (15,22), the PBP null livers showed an excessive number of nucleated erythrocytes within markedly dilated sinusoids of the liver (Fig. 5, c and d, arrowheads). In the GATA-deficient mice, the abnormally high percentage of nucleated erythrocytes was attributed to a failure in cellular differentiation (15,22). The current observations suggest that proper development of two essential hematopoietic cell lineages require PBP, and that definitive hematopoiesis may be dependent on the interaction of PBP with the members of GATA.
PBP Interacts with Five Members of the GATA Family-GST pull-down assays were performed to determine whether PBP physically interacts with GATA transcription factors. Three members of the chicken or mouse GATA family, GATA-1, GATA-2, and GATA-3, were expressed as GST fusion proteins, bound to glutathione-beads, and incubated with in vitro translated full-length mouse PBP. In contrast to GST alone, GST-GATA-1, GST-GATA-2, and GST-GATA-3 were able to retain the radiolabeled PBP (Fig. 6A). The interactions between PBP and GATA factors were not species-specific, because both chicken and mouse GATA-1, GATA-2, and GATA-3 interacted with PBP. We found that GATA-4 and GATA-6 also interacted with PBP as described below. GST pull-down assay was used to determine the interaction of 35  200 -254) and C-terminal zinc finger (aa 249 -318) (Fig. 6B). The N-terminal zinc finger of GATA-1 fragment demonstrated a marked increase in its ability to interact with PBP, whereas the C-terminal zinc finger of GATA-1 bound only a reduced amount of PBP (Fig. 6B). These results suggest that N-terminal zinc finger of GATA-1 is important for PBP binding.

LXXLL Motif Is Not Required for the Interaction between
PBP and GATA-1-The sequence motif LXXLL is found in several coactivators and has been shown to be necessary and sufficient for mediating the binding of cofactors to nuclear receptor (9,10,29,30). Using GST-cGATA-1 or GST-mGATA-1 immobilized by glutathione beads and in vitro translated truncated PBP peptides, we investigated the region of PBP that is able to interact with GATA-1. As shown in Fig. 7A, the two N-terminal fragments of PBP (aa 1-380 and 230 -626) bound to GATA-1, whereas two C-terminal fragments failed to bind GATA-1. Because the PBP 1-380 amino acid fragment lacks the LXXLL motif, the presence of LXXLL motif is not considered a requirement for the binding between GATA-1 and PBP. Because there is a reverse LXXLL (LLXXL, aa 4 -8) in the N-terminal PBP, we used aa 1-380 and a regular LXXLL in the aa 230 -626 of PBP, it was necessary to use additional PBP truncations. Fragments aa 1-230 and 230 -440 of PBP exhibited GATA-1 binding, suggesting that there are two GATA-1 binding sites in the N-terminal aa 1-440 of PBP (Fig. 7A). Furthermore, aa 440 -887 of PBP contains two LXXLL motifs at aa 589 -593 and 630 -634 but failed to bind GATA-1, clearly indicating that LXXLL in PBP is not important for GATA-1 binding (Fig. 7A). To define the GATA-4 and GATA-6-interacting domain of PBP, we used in vitro translated [ 35 S]methionine-labeled mGATA-4 or mGATA-6 and studied the binding to GST-PBP fragments (Fig. 7B). We found that mGATA-4 binds to PBP in the aa 900 -980 and 1370 -1560 region, and mGATA-6 binds PBP in the aa 335-400 region, and because these regions are devoid of LXXLL, it appears that LXXLL of PBP is not required for GATA binding.
PBP Enhances GATA-3-dependent Transcription-The ability of PBP to influence GATA transcriptional activity was examined using transient transfection assay. In these experiments, cGATA-dependent transcriptional activity was detected by transcription of reporter hGH directed by the rabbit ␤-globin TATA box and three copies of the C␤E oligonucleotide (25). In NIH 3T3 cells, the coexpression of cGATA-3 with PBP resulted in a 5-fold increase in the production of hGH over that expressed by cells transfected with cGATA-3 alone (Fig. 8). The rather modest increase in the transcription of the reporter gene by the addition of cGATA-3 may be attributed to the endogenous level of GATA factors present in NIH 3T3 cells, because transfections of QT6 increased the transcription of the reporter more than 10-fold (data not shown). The functional interaction of PBP and cGATA-3 may indicate a requirement for PBP in selected biological actions attributed to GATA-3.

DISCUSSION
PBP is an important coactivator for PPAR␥ as well as for other nuclear receptors. Here, we introduce a new interacting partner of PBP, the GATA family of zinc finger transcription factors. The GATA family of transcriptional activators contains six members (GATA-1, GATA-2, GATA-3, GATA-4, GATA-5, and GATA-6) divided into two subfamilies, GATA-1, GATA-2, and GATA-3 and GATA-4, GATA-5, and GATA-6. GATA-1, GATA-2, and GATA-3 genes appear to play critical roles in the development and differentiation of the hematopoietic stem cells toward T-lymphocytes, erythroid cells, and megakaryocytes, whereas GATA-4, GATA-5, and GATA-6 are expressed in mesoderm-and endoderm-derived tissues including heart, liver, lung, gut, and gonads (20 -23, 46 -48). GST pull-down assays revealed that PBP interacts with GATA-1, GATA-2, GATA-3, GATA-4, and GATA-6, (GATA-5 was not tested), and that the binding of GATA-1, GATA-4, and GATA-6 to PBP is independent of the sequence motif LXXLL or the nuclear receptor box (10,30,31). Moreover, the coexpression of PBP with GATA-3 markedly enhanced transcriptional activity of GATA-3 in nonhematopoietic NIH 3T3 cells. PBP null embryos died at the same gestational age between E11 and E12 (12, 13) as did GATA-3-deficient mice (15,28). When the phenotypes of the PBP null embryos were compared systematically to mice deficient in various members of the GATA family, there was a striking overlap in systems involving ventricular development, formation of retinal pigment, and the development of two hematopoietic cell lineages, megakaryocytes and erythrocytes. These observations demonstrate that PBP is critical for normal development of vital organ systems and suggest that the recruitment of PBP is important in the regulation of gene transcription by the members of the two GATA subfamilies.
Embryonic lethality of PBP null embryos is attributed to defects in placental development, ventricular maldevelopment, and cardiac failure. Ventricular function appears to be hampered by the failure of the myocardial cells to stratify normally, thus resulting in a thin ventricle. The cardiac phenotype of ventricular hypoplasia has been observed in several transgenic animals including those deficient in PPAR␥, retinoid X receptor ␣, GATA-3, and GATA-4 (14, 28, 35, 36, 46 -49) suggesting convergence into a common signaling pathway. Mice harboring a single amino acid replacement in GATA-4 that impairs its ability to interact with its presumptive cardiac cofactor FOG-2 die just after E12.5, establishing the importance of GATA-4 in heart development (49). It is conceivable that PBP interaction with GATA-4 might influence its role in heart development and lack of PBP, just as lack of FOG-2 can affect GATA-4 function (20,49). A "placenta-heart axis" hypothesis was postulated after the successful reversal of cardiac phenotype by supplementing PPAR␥ null embryos with wild type embryos through aggregation with tetraploid embryos (14). An alternative explanation involves the shared use of soluble factors such as angiogenic mediators in both placental and cardiac development. When secretion of one or more of the angiogenic factors is altered, it interferes with the endothelial-related transformation processes that are required for normal development in both organ systems. The cardiac phenotype observed in the PBP null embryos is remarkably similar to an isolated and rare human cardiac entity called noncompaction of the ventricular myocardium (48). This disease has been linked to an arrest of endocardial-myocardial development (33). The sensitivity of this endocardial-myocardial junction to perturbations in local factors has been highlighted by a study that showed that targeted overexpression of angiogenic inducer, vascular endothelial growth factor, in the heart caused poor compaction of the myocardium in the mouse (50) and a report that elevated cardiac lipids in obese Zucker diabetic rats induced an increase in myocardial apoptosis (51). Our data support a role for PBP and/or the cardiac-sensitive factors such as GATA-4 with which it interacts in the development of normal ventricular myocardium.
PBP null embryos demonstrated an angiogenic phenotype characterized by excessive systemic vascularization when compared with age-matched controls (Figs. 3 and 5). The abnormalities in the vasculature of the PBP-deficient embryos may be attributed, at least in part, to a loss of PPAR␥ function, because it is known to play a role in vasculopathies (52), and its ligands have been shown to act as potent inhibitors of angiogenesis in vitro and in vivo (53). Similar to thrombospondin-1, another angiogenic inhibitor, caspase-dependent endothelial cell apoptosis appears to be one mechanism underlying the inhibition (54,55). The angiogenic phenotype could also be related to the interaction of PBP with at least two members of GATA, namely GATA-3 and GATA-4. GATA-3 is expressed in endothelial cells, 2 and GATA-3-deficient mice had markedly reduced levels of catacholamines, a pathway recently impli-2 S. E. Crawford and J. D. Engel, unpublished observations. FIG. 8. PBP stimulates GATA-3-mediated transactivation. NIH 3T3 cells were cotransfected with 1.5 g of reporter construct C3␤GH, 0.25 g of TFAneo-cGATA3, 0.5 g of pCMV-PBP, and 0.5 g of pCMV␤. Transfection without pCMV-PBP was compensated with the same amount of pCDNA3.1. Transfection efficiency was normalized to the internal control of ␤-galactosidase expression. The levels of transfection were calculated as described previously (27). cated in blocking angiogenesis through modulation of vascular endothelial growth factor (56). As a negative regulator of endothelial cell proliferation, PBP may have therapeutic applications in angiogenic-dependent diseases.
A myriad of transcription factors are responsible for both erythroid and megakaryocytic development; however, one of the most important is GATA-1 (57,58). Zinc finger protein GATA-1 is highly expressed in erythroid and megakaryocytic cells and is critical in the maturation of megakaryocytes, as precursor cells lacking GATA-1 exhibit a developmental arrest (21,59). Although megakaryocytes have not been implicated in the PPAR-signaling pathway, platelets are known to play an important role in the vascular complications associated with diabetes and atherosclerosis, two PPAR␥-related diseases (60). It seems likely that some of the hematopoietic abnormalities in PBP null embryos involve GATA factors; however, whether these abnormalities arise from an arrest in differentiation remains to be determined. The failure of megakaryocyte and erythroid differentiation in PBP null embryos strongly suggest that this coactivator is necessary for GATA-1-mediated effects.
The eyes are the targets of pathological complications in several diseases, the most notable disease being diabetes. The retinal pigment epithelium is a vital multifunctional component of the retina. It is intimately involved in the vision through its role in regeneration of the visual pigment, opsin, via the vitamin A cycle (61). It is strategically located between the capillary bed of the choroidal layer and photoreceptors allowing it to facilitate the maintenance of the blood retinal barrier (62) by inhibiting pathological neovascularization through the secretion of potent inhibitors of angiogenesis (32). Of particular interest is that pigment epithelium-derived factor, a member of serpin superfamily, which inhibits angiogenesis, immunolocalizes to the same sites in the retina like PBP, suggesting that PBP may be involved in regulating homeostasis of the eye vasculature (32). Pigment epithelium-derived factor is the most potent inhibitor of angiogenesis in the mammalian ocular compartment, and in view of the similar cellular localization, it is possible that PBP also acts as an inhibitor of angiogenesis. PPAR␥ is selectively activated by the phagocytosis-related photoreceptor renewal process (63), a function that may require PBP given the proximity of this coactivator to the photoreceptor region. Normal eye development is highly dependent on members of another group of transcription factors, the Pax gene family (64), whereas lens development is driven predominantly by Pax-6 (65). Given the severity of the null eye phenotype, it is possible that the interaction of PBP with GATA and Pax gene families may prove to be important in controlling gene transcription in the eye.
The results presented here demonstrate that the loss of PBP results in a wide spectrum of developmental defects such as defects in hematopoiesis and the maldevelopment of the heart and eye. These observations implicate the GATA family of transcription factors as functional interacting proteins with PBP and highlights the overlapping phenotypes of PBP null embryos to mice deficient in various members of GATA. The physical interaction of PBP with all five GATA family members tested and the functional cooperation between PBP and GATA-3 reported here point to the possible tissue-specific enhancement of the function of various members of GATA factors by PBP. PBP did not appear to interact with FOG-1 and FOG-2, 3 suggesting that PBP and FOG may function independently as coactivators to enhance GATA transcriptional activity in cell type-specific genes. CBP has been shown to interact with GATA-1 and play a role in erythroid differentiation (26). Like-wise, p300 appears to function as a coactivator for GATA-4 (29). Our data suggest that PBP has several important functions including ones involving vascular maintenance, cellular differentiation in the hematopoietic system, and the development of the heart and eye. These effects may be mediated by the ability of PBP to interact with various members of the GATA family of transcription factors. In this regard, PBP appears to be a key player in development and differentiation and appears to be as important as CBP/p300 in its ability to interact with and influence the functions of various transcription factors.