CREB-binding protein and p300: molecular integrators of hematopoietic transcription

CBP was originally discovered based on its ability to interact with the cAMP response element-binding protein (CREB),10 whereas p300 was isolated as a cellular target of the adenoviral oncoprotein E1A.11 Although E1A binds to various cellular proteins, including the Rb family of tumor suppressor proteins, its ability to block cell differentiation and to induce cell cycle progression in many cell types depends, at least in part, on its interaction with CBP and p300. The functions of CBP and p300 appear interchangeable in many published reports, yet both molecules also fulfill unique roles as revealed by gene inactivation studies.5,12 13 

During the last 5 years, numerous transcriptional regulators have been found to interact with CBP and p300 (see Figure1 for examples; for review see Shikama et al8). CBP and p300 are widely expressed and are believed to regulate gene expression in most cell types. Consistent with a function in a wide range of tissues, CBP-1, a C elegans factor closely related to CBP and p300, acts at an early stage in development and is essential for all non-neuronal differentiation pathways.14In mammals, the situation is more complex because of the existence of at least 2 such molecules, CBP and p300.

The complexity of protein-protein interactions surrounding CBP and p300 has led to their description as molecular integrators. Their ability to integrate multiple transcriptional signals is illustrated by the observation that many nuclear factors that interact with CBP and p300 can synergize with each other when bound to the same promoter incis. On the other hand, inhibition between these factors might occur if they are bound to different promoters. Inhibition has been proposed to result at least in some cases from competition between these factors for limiting amounts of CBP and p300 in the nucleus.15,16 Genetic evidence for the idea that CBP and p300 are limiting stems from the discovery that patients who lack one allele of CBP suffer from RTS. Finally, normal development of Drosophila embryos is highly dependent on CBP gene dosage.17 18 

Many of the protein interactions surrounding CBP and p300 are regulated by cellular signals. For example, phosphorylation of the transcription factor CREB regulates its interaction with CBP and p300, and hormones such as estrogens, glucocorticoids, and retinoic acid stimulate CBP and p300 binding to nuclear hormone receptors.

To add to the complexity, CBP and p300 can stimulate both the activating and repressive functions of certain nuclear factors. For example, although CBP and p300 increase p53 activity on certain p53-dependent promoters,19-21 they can also augment p53-mediated transcriptional repression on others.22Moreover, CBP and p300 support cellular differentiation, but can also cooperate with gene products that interfere with it. Thus, promoter and cellular context are critical determinants of CBP and p300 function.

A breakthrough in the understanding of CBP and p300 function was the discovery that they act not only in a stoichiometric fashion, as is the case for most transcriptional cofactors, but that they also possess enzymatic activity. The laboratories of Bannister23 and Ogryzko Nakatani24 found that CBP and p300 possess intrinsic histone acetyltransferase (HAT) activity. Acetylation of histones is associated with a relaxed chromatin configuration, which is thought to facilitate transcription factor access to DNA. For example, work by Hebbes and colleagues25 demonstrated a strong correlation between the presence of acetylated core histones and DNase I sensitivity at the chicken β-globin locus. DNase I sensitivity occurs before transcription is initiated and might reflect a state poised for transcriptional activation. The importance of a balance between the acetylated and nonacetylated state of histones in transcriptional regulation is supported by the discovery that certain transcriptional repressors are associated with histone deacetylases (for review see Pazin and Kadonaga26).

More recently, CBP and p300 have been shown to acetylate nonhistone nuclear proteins, including the tumor suppressor protein p53,27-29 dTCF,30 EKLF,31GATA-1,32,33 NF-Y,34 the basal transcription factors TFIIE and TFIIF,35,36 and the architectural transcription factor HMG I(Y).37 In the case of p53, acetylation strongly increases DNA binding in vitro, providing a potential mechanisms for CBP and p300-mediated transcriptional control.27-29 Given the large number of factors that interact by CBP and p300, it is likely that some of these are also regulated by acetylation. Additional mechanisms by which CBP and p300 might operate are discussed later.

The viral oncoprotein E1A has been an invaluable tool for examining the requirements of CBP and p300 in gene expression and differentiation in various cell types. The N-terminus of E1A binds to dedicated domains within CBP and p300 and blocks their function.38 39 Indeed, in numerous studies, the first clues suggesting a requirement for CBP and p300 during gene regulation derived from experiments showing that forced expression of E1A, but not mutant forms of E1A defective for CBP and p300 binding, interfered with expression of certain myeloid, erythroid, and B-lymphocytic genes (Figure 3).

The following recurring themes are found in many of the studies summarized here. First, the activities of most transcription factors that interact with CBP and p300 are sensitive to coexpressed E1A. Inhibition by E1A is possible even if transcription factor binding occurs outside the E1A-binding domain of CBP and p300, suggesting that simple competition for CBP and p300 binding cannot account for all the effects of E1A. Second, stimulation of transcription factor activity by CBP or p300 usually ranges between 2-fold and 10-fold in transient transfection assays, indicating that CBP and p300 are limiting under these conditions. Third, various combinations of nuclear factors regulated by CBP and p300 synergize with each other when bound to the same promoter.

The following section is divided according to classes of CBP and p300-regulated hematopoietic transcription factors (summarized in Figure 2) rather than according to hematopoietic cell lineages, because most transcription factors are expressed in multiple cell types. Moreover, the biological functions of CBP and p300 in hematopoiesis are linked to the functions of the transcription factors with which they interact.

c-Myb is among the first hematopoietic transcription factors found to be regulated by CBP. c-Myb is the cellular counterpart of the v-Myb oncoprotein identified in the avian myeloblastosis virus (AMV). In the E26 virus, which causes mixed leukemia in chickens, v-Myb is part of a Gag-Myb-Ets fusion protein. Interestingly, Ets itself is regulated by CBP and p300 (see below). c-Myb expression is highest in progenitor cells of the myeloid, erythroid and lymphoid lineages and is downregulated during maturation/differentiation of these cells (for review see Weston40). Forced expression of c-Myb blocks differentiation of erythroid and myeloid cell lines.41-45Expression of a dominant interfering form of c-Myb results in enhanced erythroid differentiation,46 whereas treatment with antisense oligonucleotides directed against c-Myb reduces proliferation of immature cells of the erythroid, myeloid and T-lymphoid lineages.47-49 Disruption of the c-Myb gene in mice leads to lethal anemia during fetal liver hematopoiesis.50 Along with the leukemogenic potential of c-Myb, the previously mentioned studies suggest that c-Myb functions in maintaining hematopoietic precursor cells in a proliferative state.

CBP was found to stimulate both c-Myb and v-Myb transcriptional activity in transient transfection experiments.51,52 c-Myb binds CBP in vivo and in vitro in a phosphorylation-independent manner at a site that overlaps with the CREB-binding domain of CBP. Expression of E1A, of antisense CBP RNA, or of dominant-negative CBP interferes with c-Myb-dependent transactivation.51,52 Although CBP moderately enhances c-Myb activity (approximately 3-fold), the presence of another CBP-regulated DNA binding protein such as NF-M, strongly increases the effects of CBP in a synergistic fashion.52 

Given the requirement for CBP and p300 during differentiation of various cell types, it seems paradoxical that CBP would cooperate with gene products such as c-Myb or v-Myb that block differentiation. A possible explanation is that factors inducing differentiation and those stimulating proliferation compete for the action of CBP, depending on their expression levels during cellular maturation, or depending on cellular signals that regulate their interaction with CBP and p300.

Work from more than a decade ago demonstrated that E1A can repress the activity of the gamma 2b heavy chain (IgH) and the kappa light chain genes in lymphoid cells.53,54 However, at that time, the identity of transcription factors inhibited by E1A was unknown. Recent studies suggest that the basic helix-loop-helix (bHLH) proteins E47 and E12 might present critical targets for inhibition by E1A. E12 and E47, which are both encoded by the E2A gene (not to be confused with E1A), are essential regulators of B-cell gene expression. In most cell types, E12 and E47 proteins bind to DNA and regulate transcription as heterodimers with tissue-specific bHLH proteins, such as the hematopoietic transcription factor tal-1/SCL or the muscle-determining factors of the MyoD family. Remarkably, despite its broad distribution, only in B-cells can E47 bind DNA and activate gene expression as a homodimer.55 Targeted disruption of the E2A gene in mice leads to perinatal death and a selective ablation of mature B-cells.56,57 The cause of death is uncertain, but surprisingly, there are no obvious abnormalities present in other hematopoietic and nonhematopoietic tissues.56-58 

Work by Eckner and colleagues59 demonstrated that p300 forms a stable complex with E47 on DNA. In addition, p300 stimulates E47 activity in transient transfection experiments by using a reporter gene driven by an intact IgH enhancer or by isolated E47-binding sites. p300 also interacts with bHLH proteins involved in myogenesis,59 suggesting that it has the capacity to target various members of bHLH protein superfamily that might include those involved in hematopoiesis. Along with the findings outlined later, this suggests a role for CBP and p300 in B-lymphoid gene expression.

GATA-1, one of the best studied hematopoietic transcription factors, is a zinc finger protein involved in the regulation of virtually all erythroid and megakaryocytic genes. GATA-1 is required for survival and maturation of primitive and definitive erythroid precursor cells.60-64 In addition, GATA-1 plays a critical role during megakaryocytic proliferation and differentiation.61,65 GATA-1 can trigger terminal differentiation and cell cycle arrest when reintroduced into a GATA-1–deficient immortalized proerythroblastic cell line.66 

Among the genes regulated by GATA-1 are the globin genes, which, in turn, are under the influence of the locus control regions (LCRs). The LCRs, which contain multiple functionally important GATA-binding sites, are thought to act in part by regulating the chromatin structure at the globin gene loci.67 Given that CBP has histone acetyltransferase activity, it is noteworthy that GATA-1 interacts with CBP in vivo and in vitro.68 This interaction involves the zinc finger region of GATA-1 and the E1A-binding domain of CBP. CBP strongly augments GATA-1 activity in transient expression assays.68 Expression of E1A in the erythroid cell line MEL leads to a complete block in differentiation and to reduced expression of GATA-1–dependent genes, including the α- and β-globin genes (Figure 2).68 These findings are consistent with a mechanism by which CBP and p300 mediate at least some functions of GATA-1 in intact erythroid cells.

Other GATA factors, including GATA-2 and GATA-3, which have distinct expression patterns in hematopoietic cells, are also stimulated by CBP (G. A. Blobel, unpublished). GATA-2 levels are high in progenitor cells and decline during erythroid maturation.69,70 In contrast, GATA-1 levels increase as cells mature.69,70 Thus, it is possible that as its levels rise, GATA-1 recruits CBP away from factors required for proliferation of precursor cells such as GATA-271 and c-Myb,50 using them for the activation of differentiation-specific genes.

One mechanism by which CBP regulates GATA-1 activity appears to involve direct acetylation of GATA-1 itself. Two reports showed that CBP and p300 acetylate GATA-1 at 2 highly conserved lysine rich motifs near the zinc fingers.32,33 In addition, CBP stimulates acetylation of GATA-1 in vivo at the same sites acetylated by CBP in vitro.33 In vivo acetylation of GATA-1 by CBP is inhibited by E1A but not by mutant E1A defective for CBP and p300 binding,33 establishing a correlation between acetylation of GATA-1 and its transcriptional activity. Although Boyes et al32 reported that acetylation by p300 stimulates DNA binding of chicken GATA-1 in vitro, no change in DNA binding upon acetylation was observed by Hung et al.33 This discrepancy may be the result of using chicken GATA-1/p300 versus murine GATA-1/CBP, respectively. However, several lines of evidence suggest that changes in DNA binding might not be the mechanism by which acetylation regulates GATA-1 activity in vivo. First, mutations in the acetylation sites do not affect DNA binding of mammalian expressed GATA-1 molecules but do affect the transcriptional response to CBP and p300.32,33 Second, although CBP and p300 stimulate GATA-1 activity in transient transfection assays, no evidence exists showing that this stimulation is associated with an increase in DNA binding of GATA-1. Third, when assayed in the context of differentiating erythroid cells, mutations in either of the 2 acetylation motifs impair the ability of murine GATA-1 to trigger erythroid differentiation without affecting its ability to bind DNA.33 This indicates that the biological activity of the acetylation sites can be uncoupled from their putative role in DNA binding.

Although acetylation of GATA-1 is likely to be important for GATA-1 function in vivo, the underlying molecular mechanism remains to be determined. Acetylation of GATA-1 does not affect its interaction with Fog, CBP, or GATA-1 itself.33 However, it is possible that acetylation leads to changes in the conformation of GATA-1 or affects interaction with other as yet unidentified cofactors. The acetylation motifs of GATA-1 might serve as docking sites for interaction with such cofactors.

Given the large number of CBP-interacting proteins, it is likely that the strong inhibitory effects of E1A on MEL cell differentiation and globin gene expression might involve multiple CBP-interacting factors. Indeed, a very recent report showed that NF-E2 binding sites in the LCR are important in mediating E1A sensitivity of the β-globin LCR.72 Moreover, both NF-E2 and EKLF (see below), have been reported to physically interact with CBP. The basic zipper (bZip) transcription factor NF-E2 is composed of a hematopoietic-restricted p45 subunit and a widely expressed p18 subunit, which is a member of the maf family of proteins73-75 (for review see Blank and Andrews76). Other p45-related molecules capable of interacting with maf family members include Nrf1, Nrf2, Nrf3, Bach 1, and Bach 2 (for references see Kobayashi et al77). Multiple functionally important NF-E2-binding sites are present in the α- and β-globin LCRs. Loss of a functional p45 gene leads to a pronounced defect in platelet formation,78 whereas globin gene expression and erythroid development are only mildly affected.79 This suggests that other members of the p45 family might substitute for p45 function in erythroid cells.

In vitro binding experiments showed that the p45 subunit of NF-E2 binds directly to CBP.80 This study further suggests that CBP might participate in mediating the ligand-dependent stimulation of the thyroid hormone receptor by p45. This is of biologic interest given the role of thyroid hormone during erythropoiesis.81 Although the functional and molecular consequences of the p45-CBP interaction have not been studied in detail, it is conceivable that NF-E2 cooperates with GATA-1 and EKLF in the formation at the LCR of a high molecular weight transcription factor complex (enhanceosome) surrounding CBP and p300.

It is important to point out that NF-E2 activity on chromatinized templates cannot be attributed solely to the recruitment of histone acetyltransferases. A report by Armstrong and Emerson82demonstrated that NF-E2 can disrupt chromatin structure on templates containing regulatory regions of the β-globin locus, and that the NF-E2-associated chromatin modifying activity is ATP-dependent.

Another transcription factor regulated by CBP is the zinc finger-containing erythroid Krüppel-like factor EKLF.83 EKLF is specifically required for the expression of adult β-globin but not α-globin genes, and loss of EKLF function leads to lethal β-thalassemia in mice.84,85 Moreover, EKLF −/− mice carrying a human globin gene locus display a delayed γ- to β-globin switch that normally occurs at the onset of adult bone marrow erythropoiesis.86,87 Interestingly, absence of EKLF also results in a loss of DNase 1 hypersensitive site formation at both the transgenic and endogenous β-globin promoters,87 consistent with a role of EKLF in remodeling chromatin at these promoters.

EKLF can interact with both CBP and p300, and the CBP- and p300-associated acetyltransferase p/CAF in transfected cells. However, CBP and p300, but not p/CAF, acetylate EKLF in vitro.31Acetylation most likely occurs at 2 residues that are part of an inhibitory domain adjacent to the zinc finger region. Metabolic labeling experiments that used [³H]acetate further suggest that EKLF is acetylated in vivo.31 CBP and p300, but not p/CAF, stimulate EKLF activity in transient transfection experiments that used the erythroleukemia cell line K562.31It will be interesting to determine whether acetyltransferase activity of CBP and p300 is required for stimulation of EKLF activity. Acetylation did not affect DNA binding of EKLF, and the molecular consequences of acetylation are not yet known.31 

Together, the above reports suggest that erythroid transcription factors controlling globin gene expression might cooperate in the formation of a high molecular weight complex in which GATA-1, NF-E2, and EKLF are linked through CBP and p300 (Figure4). Consistent with such a model is the observed synergy between GATA-1 and EKLF in transactivation experiments.88 

CCAAT-box/enhancer binding proteins (C/EBPs) belong to the basic region/leucine zipper class of transcription factors and play a role in the differentiation of a broad range of tissues. In the hematopoietic system, C/EBP family members are expressed mostly in the myelomonocyctic lineage and participate in the regulation of macrophage and granulocyte-restricted genes, such as the M-CSF receptor, G-CSF receptor, and GM-CSF receptor genes (for review see Lekstrom-Himes and Xanthopoulos, and Yamanaka et al89,90). Targeted disruption of the C/EBPd, C/EBPβ, or C/EBPε genes resulted in defects predominantly affecting the granulocytic lineage,91-94whereas other hematopoietic lineages remained intact. C/EBP transcription factors are also critical mediators of inflammatory and native immune functions (for review see Poli95).

Studies by Mink et al96 showed that C/EBPβ-dependent transcription is E1A-sensitive and that overexpressed p300 stimulates C/EBPβ activity on the macrophage/granulocyte-specific mim-1 promoter and, importantly, also on an endogenous C/EBP-regulated gene, called 126. Moreover, p300 increases the synergy between c-Myb and C/EBPβ. C/EBPβ binds to the E1A-binding region of p300 through its N-terminus. Overexpression of the minimal C/EBPβ-binding domain of p300 reduced the activity of C/EBPβ presumably by interfering with the C/EBPβ-p300 interaction.96 The N-terminus of C/EBPβ contains stretches of amino acids conserved among C/EBP family members suggesting that other C/EBP molecules might also be regulated by CBP and p300.96 Together, these results implicate CBP and p300 as important cofactors during granulocytic gene expression.

The Ets family of transcription factors is a diverse group of approximately 30 proteins that share a conserved DNA binding domain.97 The c-ets-1 proto-oncogene is transduced by the E26 avian acute leukemia virus to form part of the Gag-Myb-Ets gene fusion. This virus induces both erythroid and myelomonocytic leukemias. Full transforming activity of E26 requires the presence of both the Myb and Ets portions of the fusion protein.98,99Ets-1 is expressed predominantly in lymphoid cells and regulates a number of lymphocyte-specific genes. Gene knockout studies demonstrated a role for Ets-1 in T-cell proliferation and survival.100,101 Effects on B-cell differentiation were also observed.100,101 Ets-1 and some of its relatives synergize with a number of transcriptional regulators known to interact with CBP and p300, such as AP-1,102 and Myb.103-106 Especially striking is the frequently observed cooperativity between Ets-like factors and GATA-1 during the expression of several megakaryocyte-restricted genes, including the αIIb,107 GPIX,108 GP1bα,109 the thrombopoietin receptor (c-mpl),110 and PF4 genes.111 The synergy of Ets proteins with CBP and p300-regulated factors led to the hypothesis that they too are regulated by CBP. Indeed, Yang et al112 showed that the Myb- and Ets-dependent promoter of the myeloid-expressed gene CD13/APN is sensitive to the expression of E1A but not mutant E1A defective for CBP and p300 binding. Ets-1 activity is stimulated by coexpressed CBP, and Ets-1 associates with CBP in nuclear extracts. In vitro, the N-terminus of Ets-1 can form 2 contacts with CBP involving the CH1 and CH3 domains of CBP. In support of the functional importance of the physical interaction between Ets-1 and CBP, the authors demonstrated a good correlation between binding of Ets-1 to the CH1 region and its ability to transactivate. In addition, Ets-1 coprecipitates with histone acetyltransferase activity, consistent with its association with CBP and p300 and/or other acetyltransferases in vivo.112 

Of note, another Ets family transcription factor, PU.1, was recently found to interact with CBP through the activation domain of PU.1 in a yeast 2-hybrid assay.113 CBP stimulates PU.1 transcriptional activity in transient transfection assays. PU.1 is specifically expressed in hematopoietic organs with the highest levels detected in myeloid and lymphoid cells.114 Thus, CBP and very likely p300 target a broad range of myeloid and lymphoid expressed transcription factors.

Another leukemogenic transcription factor controlled by p300 is AML1.115 The AML1 gene is rearranged in several distinct chromosomal translocations associated with acute myeloid leukemia (AML; t[8;21]), acute lymphatic leukemia (ALL; t[12;21]), and myelodysplastic syndrome (t[3;21]) (for review see Look116). The AML1 gene is the most frequent target for chromosomal translocations in human leukemias. AML1 constitutes a family of at least 3 factors derived from the same gene by alternative splicing. The AML1 gene products bind to DNA as heterodimeric complexes with CBFβ. Of note, the CBFβ gene itself is involved in chromosomal rearrangements found in cases of AML.116 Consistent with its broad expression pattern and the presence of functionally important AML1 binding sites in the promoters and enhancers of myeloid and lymphoid expressed genes, knock-out studies revealed that both AML1 and CBFβ genes are essential for the formation of all definitive blood lineages.117-121 

AML1b, one of the AML1 isoforms containing an activation domain, and p300 associate in vivo and in Far Western blots, and p300 stimulates AML1b activity on the myeloperoxidase promoter in transient transfection experiments.115 Overexpression of the t(8;21) translocation product AML1-ETO in the IL-3-dependent myeloid cell line L-G interferes with G-CSF–induced differentiation along the neutrophilic lineage. Forced expression of wild-type AML1b can overcome the effects of AML1-ETO and restore differentiation.115 In contrast, AML1a, which lacks an activation domain, is inactive in this assay. The potential of various AML1b constructs to induce differentiation is further enhanced by coexpression of p300 and correlates well with their ability to interact with p300.115 This indicates that p300 plays a role in myeloid cell differentiation and suggests that the rearranged AML genes found in chromosomal translocations act as dominant negative alleles. The latter notion is consistent with the recent finding that AML-ETO associates with a transcriptional repressor complex containing histone deacetylases and that this deactylase complex is required for blocking differentiation of myeloid cells.122-124 This raises the interesting possibility that the intrinsic (or associated) acetyltransferase activity of p300 might be required to overcome the repressive effects of AML-ETO. Indeed, a truncated form of p300 lacking the acetyltransferase domain was impaired in its ability to synergize with AML-1b. However, a more detailed mutagenesis of p300 will be required to establish a correlation between its HAT activity and its ability to cooperate with AML1b.

Finally, AML-1 synergizes with c-Myb and with C/EBP on myeloid and lymphoid promoters.125-127 This synergy is apparently not the result of cooperative DNA-binding,127 128 suggesting that it is instead mediated through recruitment of a common cofactor such as CBP and p300, similar to what has been proposed for other CBP and p300 regulated factors.

Both CBP and p300 bind the viral oncoproteins E1A and SV40 T. This raised the possibility that alterations in the functions of CBP and p300 might play a role in the development of malignancies in humans. This suspicion was supported by the finding that 1 copy of the CBP gene is inactivated in the rare disease Rubinstein-Taybi syndrome,3 which is manifested by an increased propensity for tumors (mostly of the nervous system), craniofacial malformations, and mental retardation.129 130 

The involvement of CBP and p300 in hematologic malignancies was realized through the discovery of leukemia-associated chromosomal translocations involving the CBP and p300 genes. These translocations generally result in fusion products that preserve most of the CBP and p300 molecules, suggesting that the disease mechanism does not simply involve loss of function of CBP, as is the case in Rubinstein-Taybi syndrome. Instead, they suggest altered function (dominant positive or dominant negative) through fusion to another molecule. For example, AML-derived leukemic blast cells containing the t(8;16)(p11;p13) translocation, which is often associated with acute myelogenous leukemia subtype M4/M5, have the CBP gene fused to the MOZ (monocytic leukemia zinc finger) gene.131 This fusion results in a small deletion of the N-terminal 266 amino acids of CBP leaving the rest of the molecule intact.131 Interestingly, the MOZ gene also has a putative acetyltransferase domain that is retained in the MOZ-CBP fusion.

In principle, any translocation event could lead to gain or loss of function of either fusion partner, to the formation of dominant interfering alleles, or to entirely new activities. Fusion of CBP to a given transcription factor might result in aberrant recruitment of CBP to certain promoters, leaving less free CBP available for other transcription factors involved in balancing proliferation and differentiation. In addition, it is possible that misdirected or deregulated acetyltransferase activity by CBP and p300 fusion products causes changes in gene expression profiles that contribute to the transformed state. One likely mechanism by which the MOZ-CBP fusion contributes to malignant transformation involves constitutive recruitment of CBP to MOZ-regulated genes. The MOZ gene contains 2 C4HC3 zinc finger regions, also found in CBP and p300, and a C2HC zinc finger. These regions might serve as protein-protein interaction domains and might target MOZ to chromatin-associated proteins and DNA.131 The MOZ-CBP fusion protein contains the CBP-derived and the putative MOZ acetyltransferase domain that together could be powerful regulators of chromatin structure and transcriptional activity at MOZ-regulated genes.

Since their initial discovery, additional cases of AML with t(8;16) translocations resulting in CBP and MOZ gene arrangements have been reported.132 However, in these cases no MOZ-CBP fusion transcripts were detected, raising the possibility that CBP or MOZ gene rearrangements might contribute to leukemogenesis by alternative mechanisms.

Another clinically relevant example of the importance of balancing histone acetylation and deacetylation comes from studies of acute promyelocytic leukemia (APL)-associated translocations that fuse the retinoid acid receptor alpha (RARα) to the PLZF or PML genes. PML-RARα and PLZF-RARα fusion proteins have a high affinity for a transcriptional repressor complex containing histone deacetylases. Although normal RAR responds to retinoic acid (RA) by shedding the deacetylase complex, followed by association with an acetyltransferase complex (which contains CBP), PML-RARα responds only to very high concentrations of RA, and PLZF-RARα is RA resistant.133-135 The ability of leukemic cells to differentiate upon RA treatment correlates with the ability of their translocation fusion proteins to displace the repressor complex in response to RA. In fact, patients with PML-RARα APL typically achieve remission upon treatment with high doses of RA, whereas PLZF-RARα APL patients do not.

The chromosomal translocation, t(11;16), which is associated with therapy-induced acute myeloid leukemia, therapy-induced chronic myelomonocytic leukemia, and myelodysplastic syndrome, fuses the MLL and CBP genes such that most of the CBP molecule stays intact.136-139 The MLL gene was also found to be fused to the p300 gene in an AML patient carrying a t(11;22) translocation.140 The MLL gene encodes a large multidomain protein containing zinc fingers and AT-hook motifs,141-143and is involved in translocations with at least 40 different fusion partners (for references see Sobulo et al137). This raises the question whether the structural alterations of MLL itself or of its fusion partners are critical for leukemogenesis. Together, these findings underscore the importance of CBP and p300 function in balancing growth and differentiation of hematopoietic cells.

Some unexpected insights into the function of CBP and p300 have come from gene knock out studies. The CBP and p300 null mice display similar phenotypes.13 The p300−/− embryos die between days 9 and 11.5. Their main defects are severe developmental retardation, reduced size, failed neural tube closure, and altered cardiac ventricular trabeculation. A fraction of the p300 +/− mice die early, displaying neural tube closure defects similar to the p300−/− mice, indicating a requirement for full p300 gene dosage during neural development. Mice heterozygous for CBP deficiency suffer from skeletal abnormalities and growth retardation, a phenotype resembling RTS in humans.4 CBP and p300 compound heterozygous mice die early and display a phenotype very similar to the individual homozygous knock outs.13 

More extensive analysis of mice heterozygous for CBP deficiency revealed defects in the hematopoietic system that only became apparent in newborn pups beginning at 3 months of age.5 The CBP +/− animals have extramedullary myelopoiesis and erythropoiesis, and display enlarged, hypercellular spleens. In the peripheral blood, the most striking defect is a decrease in the number of B-lymphocytes, whereas in the bone marrow, cells of the erythroid, myeloid, and B-lymphocytic lineage were significantly reduced. No overt malignancies were observed in the CBP +/− mice until the mice reached at least 1 year of age. Then, 4 of the 18 mice analyzed had overt tumors, 2 had histiocytic sarcomas, 1 had myelomonocytic leukemia, and 1 had lymphocytic leukemia. In light of the small number of cases studied, it is conceivable that other types of hematologic neoplasms might occur at an increased rate in CBP +/− mice. When splenocytes or bone marrow cells from apparently tumor-free CBP+/− donors were engrafted into sublethally irradiated wild-type mice, the recipients developed histiocytic sarcomas at a high rate with latency periods of 3 to 5 months. Grafts derived from 1 CBP +/− donor resulted in the formation of plasmacytomas with monoclonal gammopathy and renal amyloid deposition. DNA analysis of 1 plasmacytoma and 1 histiocytic sarcoma from bone marrow–transplanted mice revealed the specific loss of the wild-type CBP allele with retention of the targeted one. Loss of heterozygosity in these cases suggests that CBP is a tumor supressor gene, similar to the RB family of proteins that are also targeted by the E1A oncoprotein. Surprisingly, no hematologic defect or cancer predisposition was observed in age- and strain-matched p300 targeted mice.5 This suggests that, despite their similarity, CBP and p300 might play distinct roles in certain cell types.

The tumors observed in CBP +/− in mice appear to be restricted to the hematopoietic system, although additional types of neoplasms might be found as more mice are analyzed. In contrast, patients with RTS have an increased risk for tumors of various origins, the most common tumors being neurally derived. Hematologic malignancies observed in RTS patients occur less frequently and include acute lymphocytic leukemia, acute myelogenous leukemia, and non-Hodgkin lymphoma.130 

A very recent report describes the hematologic consequences of homozygous CBP-deficiency in mice.6 In this study, disruption of the CBP gene resulted in the formation of a truncated form of CBP that retains the N-terminal 1084 amino acids (of 2441) but lacks the HAT domain. Mice homozygous for this defect die between day 9.5 and 10.5 of embryogenesis similar to the CBP knock-out mice. Before their deaths, embryos are anemic, and their yolk sacs contain fewer erythroid cells and display a defective vascular network. Although the number of yolk sac–derived erythroid colony forming units is reduced, a few mature erythroid cells are found, suggesting that CBP is not absolutely required for erythroid maturation and that p300 might be able to partially compensate for the CBP defect. To examine definitive hematopoiesis in the CBP −/− mice, organ culture was performed from E9.5 embryos with tissue from the aorta-gonad-mesonephros (AGM) region, followed by colony forming assays. These experiments revealed dramatically reduced numbers of definitive erythroid and granulocyte/macrophage progenitor cells. Organ cultures from these embryos also revealed a strong reduction in vasculo-angiogenesis.

The mechanisms by which CBP deficiency cause RTS in humans and the severe hematologic and nonhematologic defects in mice are entirely unknown. The answer to this question is complicated by the enormous complexity of protein interactions surrounding CBP and the multitude of mechanisms by which CBP regulates gene expression. Analysis of gene expression profiles in tissues from CBP-deficient mice, as well as gene complementation experiments with mutant CBP gene constructs, could be used to tackle this question. Progress in the understanding of the phenotypic defects that result from CBP deficiency requires a reductionistic approach involving the study of individual CBP- and p300-binding transcription factors and the genes that they control. For example, it is conceivable that the reduced number of B lymphocytes in CBP +/− mice results from reduced activity of the E47 transcription factor that interacts with CBP and p300, and that is required for B-cell development.59 

CBP and p300 interact with numerous transcription factors. Many of these interactions might take place simultaneously because they are mediated by distinct domains. This could account for the observed synergy between factors regulated by CBP. Thus, CBP might provide a platform for the assembly of high molecular weight complexes (enhanceosomes; for review see Carey144) containing multiple DNA-binding proteins that position the complex in a sterically correct fashion at promoters and enhancers. Because this complex is likely to include non-DNA-binding proteins such as p/CAF, ACTR, or SRC-1, which also possess acetyltransferase activity, it would constitute a powerful regulator of chromatin structure.145-147 For example, a high molecular weight complex centered on CBP and p300 could form at the LCR, which participates in regulating chromatin structure at the β-globin locus (Figure 4). The LCR contains binding sites for GATA-1, EKLF, and NF-E2 all of which bind to CBP and p300.31,33,68 80 Thus, CBP and p300 might integrate signals from multiple transcriptional regulators and perhaps even present targets for global regulators of gene expression, such as signaling cascades used by growth/differentiation factors. The latter notion is supported by the observation that CBP and p300 are acetylated and phosphorylated.

CBP and p300 are also thought to mediate negative cross-talk between transcription factors. Competition for limiting amounts of CBP and p300 has been invoked to account for mutual inhibition of CBP- and p300-regulated transcription factors when bound to separate DNA templates.15 This might explain the inhibition of GATA factors by ligand-activated nuclear hormone receptors (NR).148-150 The observation that overexpression of CBP alleviates NR-mediated repression of GATA-1, and that ligand-bound NR reduce the stimulation of GATA-1 activity by CBP (G. A. Blobel, unpublished) are consistent with such a model. Together, these findings support a role of CBP and p300 as molecular integrators of positive and negative transcriptional signals that govern hematopoietic gene expression.

The large number and diversity of genes and transcription factors regulated by CBP and p300 could be explained if CBP and p300 were components of the basal transcription apparatus. In support of such a model, CBP and p300 have been found to interact with TFIIB,151 TBP,152-155 and RNA polymerase II.156-160 Thus, recruitment of CBP by a DNA-bound transcription factor could facilitate the formation of a preinitiation complex at relevant promoters (Figure 5). Such a mechanism would imply that CBP and p300 act in a stoichiometric fashion. Although this might be true on some promoters, additional evidence suggest that CBP and p300 also act catalytically (see next paragraph).

The observation that CBP, p300, and some of its associated factors possess acetyltransferase activity suggests an enzymatic mechanism of gene regulation. Targeting of CBP and p300 to the appropriate sites could lead to local increases in histone acetylation, followed by rearrangement of chromatin structure (Figure 4). This in turn could favor access of other transcriptional regulators. Again, the LCR provides an example where such a mechanism might be operating. As previously mentioned, histone acetylation and open chromatin correlate well at the chicken β-globin gene locus.25 However, depending on transcription factor/promoter context, CBP and p300 can also act in a HAT-independent fashion.161 

If some nuclear factors act by recruiting a histone-modifying enzyme to trigger chromatin opening, how do they find access to DNA in the first place? One possibility is that other transcription factors might pave their way by opening chromatin structure in an acetylation-independent fashion. An example for such a scenario is provided by the observation that NF-E2 disrupts chromatin structure in a ATP-dependent manner on a chromatinized template containing DNase1 hypersensitive site 2 of the β-globin LCR.82 This leads to increased access of GATA-1 to adjacent GATA sites.

Alternatively, GATA-1 might find access to chromatin without the assistance of other factors. A recent report162demonstrated that chicken GATA-1, or a peptide comprising just its DNA-binding domain, can bind to DNA packaged into a nucleosome. This leads to a reversible breakage of histone/DNA contacts, thus perturbing nucleosome structure.162 Once bound to DNA, the GATA-1-associated acetyltransferase complex might modify adjacent histones, thus facilitating access of other transcription factors to DNA.

It is important to keep in mind that modification of chromatin is not restricted to acetylation, and that numerous regulated chromatin modifying complexes have been identified (for review see Kadonaga163). For example, an elegant study by Armstrong et al164 demonstrated that EKLF interacts with a complex, called E-RC1, which contains components of the mammalian SWI/SNF complex, an ATP-dependent chromatin remodeling machine.163However, E-RC1 does not appear to contain histone acetyltransferases (Beverly Emerson, personal communication).

Acetylation of nonhistone proteins, including transcription factors, might turn out to be of equal importance for CBP and p300 function. For example, acetylation of p53 leads to an increase in DNA binding activity.27-29 It is likely that acetylation regulates transcription factor activity by a variety of mechanisms. In the case of the drosophila transcription factor dTCF, acetylation by CBP decreased its affinity for its cofactor β-catenin/Armadillo, leading to transcriptional inhibition.30 An interesting variation of this theme is the finding that acetylation of the architectural transcription factor HMG)-I(Y) by CBP leads to destabilization of an enhanceosome complex at the interferon γ gene promoter, resulting in termination of transcription.37 

It is conceivable that acetylation might be a widely used mechanism to trigger allosteric changes in proteins, thereby regulating protein-protein and protein-DNA interactions, similar to what has been observed upon protein phosphorylation. In both cases, the modification results in a change of charge, addition of a negative charge in the case of phosphorylation, and neutralization of a positive charge in the case of acetylation. Moreover, acetylation changes the size of the lysine side chain, which could be important in protein folding.

CBP and p300 are large, multifunctional molecules that can exert both positive and negative effects on transcription and cell differentiation. It is likely that additional factors will be discovered to interact with CBP and p300, and that a subset of these might be regulated by acetylation. The challenge that lies ahead will be to determine the significance of such interactions in physiologically relevant settings. Given that CBP and p300 share many functions this will not be an easy task, especially because it has not been possible so far to generate CBP and p300 double knock-out cell lines. The mechanisms by which CBP and p300 act likely depend on promoter and cellular context as well as the chromatin configuration in which a given target gene is embedded. One approach that would allow dissection of CBP and p300 functions in a physiologic context would be to knock in mutant CBP and p300 alleles bearing mutations in domains associated with specific functions such as the HAT domain or important protein docking sites. Such experiments might also yield insights into the mechanism by which loss of CBP leads to RTS.

Given the broad variety of CBP and p300 regulated factors, an important and challenging task will be the identification of the relevant downstream target genes that mediate their function in vivo. Subtractive hybridization and microarray technologies might be useful approaches to identify genes most sensitive to changes in CBP and p300 levels.

Although CBP and p300 are expressed in most tissues, their importance in regulating gene expression and differentiation in hematopoietic cells is illustrated by their involvement in leukemia-associated chromosomal translocations. It remains to be determined why these chromosomal translocations result in leukemias mostly of the myeloid/monocytic lineage.

Because CBP and p300 have intrinsic and associated acetylase activity, they might present targets for pharmacological intervention. It can be envisioned that novel drugs might be developed that alter their specific activity or substrate specificity, thereby allowing for modulation of gene expression and cell differentiation. For example, in cases in which CBP acetyltransferase activity might be activated as a result of chromosomal translocations or point mutations, interference with this activity might reverse the cellular phenotype whether it is hypo- or hyperplastic. Alternatively, in cases in which cellular CBP activity is reduced as a result of haploinsufficiency or inactivating mutations, a compound that stimulates acetyltransferase activity of the remaining allele or that of p300 might allow for compensation of the defect. Treatment of any disorder caused by defects involving the CBP and p300 genes or CBP- and p300-regulated transcription factors rests entirely on a thorough understanding of the molecular and cellular environment in which CBP and p300 function.

An example for the successful manipulation of the acetylation balance in the cell comes from studies that use the drug trichostatin A. Trichostatin A, which is a deacetylase inhibitor, has been successfully used to activate silenced transgenes in the context of gene delivery vectors designed for use in gene therapy.165 Furthermore, drugs targeting histone deacetylases have been used against malaria and toxoplasmosis.166 Thus, a detailed understanding of the role of protein acetylation might reveal new approaches to controlling gene expression and treating human diseases.

I want to thank Margaret Chou, Merlin Crossley, Richard Eckner, Stuart Orkin, Morty Poncz, and Mitchell Weiss for helpful suggestions and critical reading of the manuscript. Naturally, the survey of a burgeoning field such as this might not do justice to all contributions. Therefore, I apologize to those whose work is not represented here.