Role of Intrinsic Protein Disorder in the Function and Interactions of the Transcriptional Coactivators CREB-binding Protein (CBP) and p300*

The transcriptional coactivators CREB-binding protein (CBP) and p300 undergo a particularly rich set of interactions with disordered and partly ordered partners, as a part of their ubiquitous role in facilitating transcription of genes. CBP and p300 contain a number of small structured domains that provide scaffolds for the interaction of disordered transactivation domains from a wide variety of partners, including p53, hypoxia-inducible factor 1α (HIF-1α), NF-κB, and STAT proteins, and are the targets for the interactions of disordered viral proteins that compete with cellular factors to disrupt signaling and subvert the cell cycle. The functional diversity of the CBP/p300 interactome provides an excellent example of the power of intrinsic disorder to facilitate the complexity of living systems.

The transcriptional coactivators CREB-binding protein (CBP) and p300 undergo a particularly rich set of interactions with disordered and partly ordered partners, as a part of their ubiquitous role in facilitating transcription of genes. CBP and p300 contain a number of small structured domains that provide scaffolds for the interaction of disordered transactivation domains from a wide variety of partners, including p53, hypoxiainducible factor 1␣ (HIF-1␣), NF-B, and STAT proteins, and are the targets for the interactions of disordered viral proteins that compete with cellular factors to disrupt signaling and subvert the cell cycle. The functional diversity of the CBP/p300 interactome provides an excellent example of the power of intrinsic disorder to facilitate the complexity of living systems.
Intrinsically disordered proteins (IDPs) 3 and protein regions (IDRs) are highly abundant in eukaryotic transcription factors (1,2) and perform critical functions in regulation of the transcriptional machinery (3)(4)(5). The intrinsic lack of structure in IDPs and IDRs provides functional advantages that make them ideally suited to mediate transcriptional regulatory processes. These include (i) the presence of small recognition motifs that fold upon binding, (ii) the flexibility to interact with multiple targets, (iii) accessible sites for post-translational modification (enables IDPs to function as molecular switches and rheostats), (iv) efficient utilization of a small number of residues to mediate binding interactions, and (v) the ability to bind with high specificity but modest affinity (an important attribute that could facilitate spontaneous dissociation or displacement after signaling is complete) (5)(6)(7)(8)(9)(10)(11).
The present review focuses on the interactions of the transcriptional coactivator CREB-binding protein (CBP), and its paralog p300, to illustrate the diverse functions of protein disorder in the regulation of gene expression. CBP and p300 are central nodes in eukaryotic transcriptional regulatory networks (12). They interact with more than 400 transcription factors and other regulatory proteins (13), which must compete for binding to the limiting amounts of CBP/p300 present in the cell (14,15). CBP and p300 regulate crosstalk and interference between numerous cellular signaling pathways, and are targeted by tumor viruses to hijack the cellular regulatory machinery.
CBP and p300 are large proteins that contain seven folded domains (Fig. 1). The regions outside these globular domains, containing more than 1400 amino acids (nearly 60% of the sequence), are predicted to be intrinsically disordered. In addition, the histone acetyltransferase domain (HAT) domain contains an ϳ60-residue, disordered, autoinhibitory loop that regulates acetyltransferase activity (16). The nuclear coactivator binding domain (NCBD, also called the IRF-3 binding domain, IBiD (17)) has the properties of a molten globule in its unbound state, but folds upon binding to its protein targets (18). The TAZ1, TAZ2, KIX, and NCBD domains form the interaction sites for intrinsically disordered activation domains of cellular transcription factors and other regulatory proteins and are also targeted by viral oncoproteins. Structures have been determined for all of the globular domains of CBP or p300, either free or bound to protein ligands. A graphic structural model for full-length CBP/p300 is shown in Fig. 1B.

Role of Disordered, Flexible Linkers in Promoter Recognition
CBP and p300 have been found at the promoters of more than 16,000 human genes (19). Smith et al. (20) suggested that CBP and p300 are "molecular interpreters that can parse and/or conjugate the regulatory words, phrases, and sentences of the genome." Their ability to perform this function derives from two attributes: promiscuous interactions with the disordered transactivation domains of hundreds of cellular transcription factors, and the presence of long, intrinsically disordered regions between the various CBP/p300 interaction domains. These attributes impart to CBP and p300 the flexibility to bind diverse arrays of transcription factors at promoters of variable architecture and with different spacing between binding sites, thereby "reading" the language encoded in these gene regulatory sequences (Fig. 2).
Although the overall "structure" of CBP/p300 is unknown, intrinsic flexibility is suggested by a recent cryo-EM analysis of a complex formed between DNA-bound estrogen receptor, the steroid receptor coactivator SRC-3, and full-length p300 (21). Free p300 is somewhat compacted, and there are changes in the overall conformation upon binding to estrogen receptor and SRC-3.

TAZ1 Interactions
The TAZ1 and TAZ2 (transcriptional adapter zinc finger (22)) domains share the same fold, a bundle of four helices (designated ␣ 1 -␣ 4 ) stabilized by three zinc atoms (23,24). The TAZ domains are selective in their interactions and bind different subsets of intrinsically disordered regulatory motifs.
The TAZ1 domain mediates transcription of genes involved in regulation of the hypoxic response, the immune and inflammatory response, and cellular proliferation and survival. IDRs that bind with high affinity to TAZ1 tend to be relatively long, with multiple amphipathic regions that make extensive hydrophobic contacts in a deep binding groove (25). These contacts are complemented by electrostatic interactions between acidic residues in the ligand and positively charged side chains on the surface of TAZ1. Structures have been determined for the complexes of TAZ1 with the intrinsically disordered transactivation domains of hypoxia-inducible factor 1␣ (HIF-1␣) (26,27), CITED2 (28,29), STAT2 (25), and NF-B (30). Apart from a conserved LP(Q/E)L motif in HIF-1␣ and CITED2, there is no sequence conservation between these activation domains, all of which bind in the hydrophobic groove and wrap almost entirely around TAZ1. Although all four activation domains fold upon binding to form local elements of helical structure, there is little similarity in their locations on the TAZ1 surface (Fig. 3, A and  B). The structure of the bound state is determined by the amino acid sequences of the amphipathic regions, the distribution of these regions in each IDR, and the physicochemical character-istics of their interactions with the TAZ1 hydrophobic groove and surrounding electrostatic charge. The lack of directionality in the binding pose of the different activation domains, as well as the fact that any given region of TAZ1 may bind different secondary structures, demonstrates the promiscuous nature of TAZ1 recognition.
Competition between transcription factors for binding to TAZ1 plays an important role in cellular regulation. For example, HIF-1␣ and CITED2 compete for binding to TAZ1 in a negative feedback circuit that attenuates the hypoxic response (28,29,31). Crosstalk between the HIF-1␣ and p53 pathways also occurs through competition for binding to TAZ1 (32).

TAZ2 Interactions
The TAZ2 domain binds promiscuously to IDRs from numerous cellular regulatory proteins. Unlike TAZ1, interactions with the TAZ2 domain are mostly localized to a hydrophobic binding surface at the interface of the ␣ 1 , ␣ 2 , and ␣ 3 helices. This surface functions as a docking site for intrinsically disordered activation domains, which adopt helical structure upon binding to TAZ2 (Fig. 3C). The promiscuity of this binding site has resulted in some potentially misleading interactions in x-ray structures. In the crystal structure of free TAZ2, a C-terminal extension of the ␣ 4 helix extends beyond the globular core and docks to the hydrophobic surface on a neighboring molecule in the crystal lattice, mimicking the interactions of TAZ2 with the amphipathic helices of regulatory IDRs (33). The long ␣ 4 helix observed in the x-ray structure appears to be stabilized by lattice contacts. The helices formed by the isolated AD1 motif of the intrinsically disordered N-terminal activation domain of p53 and by the activation domain of transcription factor E2␣ bind to the same region of TAZ2, but in opposite orientations (34). The isolated AD2 motif of p53 binds in a partly helical conformation to a different region of the TAZ2 hydrophobic surface (35). The STAT1 activation domain utilizes a similar binding site but forms supplementary interactions on the surface formed by helices ␣ 3 and ␣ 4 of TAZ2 (25) (Fig. 3C). It is of note that STAT1 and the p53 AD2 share a common sequence motif (25) yet differ significantly in their interactions with TAZ2. The bipartite activation domain of C/EBP binds in a similar mode to STAT1, with helices docked to the primary ␣ 1 , ␣ 2 , ␣ 3 surface and also to the secondary ␣ 3 , ␣ 4 site (36). However, the longer of the two C/EBP helices is displaced relative to that of STAT1 (Fig. 3C); this displacement may well be an artifact of crystallization, because the C/EBP IDR was engineered into the ␣ 4 helix of TAZ2 to promote crystallization and makes numerous contacts with neighboring molecules in the crystal lattice (36).
Numerous promoter-bound transcription factors activate transcription by recruiting CBP/p300 through interactions with TAZ2 (12). Binding to TAZ2 also brings the transcription factor close to the HAT domain, providing an additional level of regulation through acetylation-mediated mechanisms (37).

KIX Interactions
The CBP/p300 KIX domain is the primary interaction site for numerous cellular transcription factors that function in hematopoietic differentiation: CREB, c-Myb, MLL (mixed  (18), as well as the x-ray crystal structure of the combined BRD, CH2, and HAT domains of p300 (62). Gray spheres in the TAZ1, CH2, ZZ, and TAZ2 structures represent Zn 2ϩ . SUMO, small ubiquitin-like modifier; pCREB, phosphorylated CREB.
lineage leukemia protein), c-Jun, E2A, and FOXO3, among many others. Some hematopoietic transcription factors compete for binding to the same surface of KIX (e.g. CREB and Myb (38, 39)) or bind synergistically to different surfaces of the KIX domain (e.g. MLL and Myb, MLL, and CREB (40,41)). Because KIX is a central hub for interactions with hematopoietic transcription factors, it is likely to perform a key function in integrating and regulating crosstalk between different signaling pathways.
KIX is a small domain, comprising three ␣and two 3 10 -helices (4), that utilizes two distinct binding surfaces (Fig. 4). The intrinsically disordered activation domains of CREB and c-Myb fold into helical conformations and bind in a shallow hydropho-bic groove in the ␣1-␣3 surface of KIX (4, 42) (Fig. 4A). The kinase-inducible activation domain of CREB (KID), phosphorylated at Ser 133 (pKID), forms a pair of orthogonal helices upon binding to KIX, with the C-terminal helix (␣ B ) dominating the binding interaction. c-Myb forms a slightly bent helix that binds in the same site as the pKID␣ B helix. The MLL activation domain binds in a helical conformation on the opposite face of KIX, in a hydrophobic groove formed by the ␣ 2 , ␣ 3 , and 3 10 helices (43) (Fig. 4B). MLL and Myb, or MLL and pKID, can bind simultaneously and cooperatively to KIX to form a ternary complex, resulting in allosteric enhancement of the binding affinity (40,41). Binding cooperativity has been attributed to stabilization and decrease in dynamics in local regions of the  KIX structure upon binding of MLL (43)(44)(45), resulting in a decrease in the dissociation rates of both activation domains (46).
The complexes described in the preceding paragraph each involve interactions between KIX and an amphipathic binding motif (⌽XX⌽⌽, where ⌽ is a bulky hydrophobic residue) located within an intrinsically disordered region of the transcriptional activation domain. The ⌽XX⌽⌽ motif interacts promiscuously with both the c-Myb/pKID and MLL binding surfaces of KIX; specificity is imparted by flanking residues (47). For example, although the pKID and c-Myb activation domains bind preferentially to their cognate high affinity site on the ␣1-␣3 surface of KIX, both domains also bind with lower affinity to the MLL site (48); conversely, MLL binds tightly in its cognate site and weakly in the c-Myb site (49). Activation domains with bipartite interaction motifs can potentially bind simultaneously to both KIX sites. The disordered activation domain of FOXO3a, a transcription factor that regulates cell differentiation, survival, and apoptosis genes, contains two amphipathic interaction motifs (CR2 and CR3) that bind synergistically to KIX (50). The bipartite FOXO3a activation domain makes promiscuous multivalent interactions with KIX. The complex is heterogeneous, adopting two equally populated conformational states with helical structure in the CR2 and CR3 motifs. In one conformer, the CR2 helix binds in the c-Myb site and CR3 occupies the MLL site, whereas in the second conformer, CR2 is in the MLL site and CR3 occupies the c-Myb site. The bipartite N-terminal transactivation domain of p53 also forms a heterogeneous complex on binding to KIX (51). The AD1 and AD2 regions of p53, which contain ⌽XX⌽⌽ motifs that fold into helices on the surface of KIX, bind simultaneously and synergistically to the MLL and c-Myb sites, thereby enhancing the binding affinity. Similar to FOXO3a, the p53 AD1 and AD2 helices each interact with both the MLL and c-Myb binding surfaces. However, in contrast to FOXO3a, each of the p53 helices binds in each site in alternate conformations, which differ by an ϳ180°rotation of the helix axis (51). Some level of disorder is frequently retained in the interface between IDPs and their target proteins, a property that is referred to as "fuzziness" (52, 53).

NCBD Interactions
In the absence of binding partners, the NCBD, also known as the IBiD, has the properties of a molten globule, with substantial helical structure (18). Upon binding to its physiological partners, the NCBD folds to form a bundle of three helices that differ in their topological arrangements in complexes with intrinsically disordered targets, such as the p160 nuclear coactivators ACTR (coactivator for thyroid and retinoid receptors) and SRC-1 and the p53 activation domain (18,54,55), and in the complex with the stably folded interferon regulatory factor IRF-3 (56). Detailed NMR and computational studies suggest that in the unbound state, the NCBD is flexible and fluctuates over an ensemble of conformations that includes both the ACTR/SRC-1-bound conformation and the IRF-3-bound conformation (57)(58)(59)(60)(61). The inherent flexibility of the NCBD and its ability to sample a number of conformational states in the unbound state are likely of functional importance in allowing it to fold into different structures upon binding to different physiological targets.

The Bromodomain
CBP and p300 contain a bromodomain fused to a novel zinc binding domain that contains both RING and plant homeodomain (PHD) fingers (62,63). In common with other bromodomains, the CBP/p300 bromodomain recognizes acetyl-lysine residues in histone tails and in transcription factor IDRs including those of p53 and CREB (64).

Mechanisms of Coupled Folding and Binding
CBP/p300 has been a paradigm for detailed studies, using NMR, stopped flow kinetics, and molecular simulations, of the mechanism by which intrinsically disordered proteins fold upon binding to their targets. NMR provides unique atomic resolution insights into the mechanisms by which intrinsically disordered proteins bind to a folded protein target. The addition of an IDP to an isotopically labeled ordered protein target allows rapid and accurate identification of the binding site on the target, using chemical shift mapping (65). Analysis of NMR relaxation dispersion profiles for the isotopically labeled IDP in the presence of unlabeled binding partner provides a comprehensive picture of the kinetics and mechanism of binding and folding (66,67). Two extreme mechanisms can be envisaged: conformational selection and induced fit. In the conformational selection mechanism, the target protein "selects" a conformation that closely approximates that of the bound form from the ensemble of conformations populated by the unbound IDP; i.e. folding of the IDP occurs before binding. In the induced-fit mechanism, the IDP associates with its binding partner in a fully disordered or incorrectly folded state and folds in association with the target protein; i.e. folding after binding to the target. The binding of IDPs to the KIX domain provides examples of both of these types of interaction. However, it should be noted that IDP binding processes rarely occur by pure conformational selection or induced-fit mechanisms, and most interactions involve some combination of the two processes; it is to be expected that some structural rearrangement will be necessary even when a preformed structural motif in an IDP binds to its target.

Induced-fit Binding of pKID to KIX
In the unbound state, the ␣ B helix of pKID is fully unstructured, whereas the ␣ A helix spontaneously forms ϳ50% population of helix (68). Using NMR and 15 N relaxation dispersion measurements with 15 N-labeled pKID, Sugase et al. (66) showed that binding of pKID to KIX occurs by an induced-fit mechanism, forming an intermediate in which pKID is docked to the surface of KIX in a partly folded state. The ␣ A helix of pKID is fully folded in the intermediate state, whereas the ␣ B region is only partially structured in the intermediate and folds to form stable helical structure only after pKID has docked to the surface of KIX. The weak propensity of the ␣ B region of pKID to spontaneously fold into helix may impose a barrier to folding, such that stabilization of the ␣ B helix requires formation of favorable intermolecular interactions after docking to KIX (48). Molecular simulations were able to capture the principal features of the experimental binding and folding pathway, providing molecular level insights into the ensemble of intermediate states formed by the partially folded ␣ B helix (69,70).

Elements of Conformational Selection: The c-Myb-KIX Complex
The c-Myb activation domain occupies the same binding site as the ␣ B helix of pKID, forming a long amphipathic helix with a distinct bend that positions the side chain of Leu 302 in a deep hydrophobic pocket in the surface of KIX and effectively breaks the helix into N-and C-terminal halves (42,43). The free c-Myb activation domain is dynamically disordered, fluctuating between helical and more extended conformational states. The N-terminal half spontaneously forms helical structure (ϳ70% population) in aqueous solution, whereas the C-terminal region has a much weaker propensity to form helix in the absence of binding (48). Stopped-flow kinetics and NMR relaxation dispersion measurements show that the c-Myb activation domain binds to KIX in an extremely fast two-state process, without formation of observable intermediates (48,71,72). The N-terminal region, which has strong propensity for spontaneous helix formation, binds by the conformational selection mechanism, as evidenced by the correlation observed (48) between the population of helix in a series of mutants and the association rates reported by Giri et al. (73). In contrast, folding of the C-terminal region of the c-Myb peptide occurs after binding to KIX, by an induced-fit mechanism. Thus, the binding mechanism of c-Myb is complex and involves elements of both conformational selection and induced fit. A similar mechanism has been observed for binding of the intrinsically disordered ACTR to the NCBD of CBP; the N-terminal helix of ACTR folds rapidly and binding involves conformational selection (74), whereas other regions of ACTR fold by a slower, induced-fit process following the initial binding event (75).
The differences in the propensities of the disordered pKID and c-Myb activation domains to fold spontaneously into helical structure appear to be related to their functions and may be, at least in part, determinants of the folding rate and binding mechanism (48). CREB is an inducible transcriptional activator that must be phosphorylated at Ser 133 in the KID to bind with high affinity to the KIX domain of CBP/p300 (76,77). The high degree of disorder of the ␣ B region in the unbound state and its limited propensity to form helical structure are expected to facilitate binding, in an extended conformation, to the active site of protein kinase A (39). In turn, this imposes a barrier to folding of the ␣ B helix on association with KIX, such that development and stabilization of helical structure in the ␣ B region occur slowly, after the initial binding event (66). In contrast, c-Myb is a constitutive transcriptional activator. No post-translational modifications of the c-Myb activation domain are needed for interaction with KIX, and the strong propensity for spontaneous helix formation may be beneficial for promoting rapid, high affinity binding to CBP/p300 and other target proteins.
The KIX association mechanisms for the pKID and c-Myb peptides show significant differences, despite binding to an identical site on KIX. Clearly, the characteristics of the IDP itself, its sequence, charge, and population of transiently structured states in the conformational ensemble, play a role in determining the binding kinetics and mechanism. It seems likely that the coupled binding and folding processes of all but the simplest IDPs will occur by complex multi-step mechanisms, certainly for those IDPs that interact through more than one binding motif (78). In the case of pKID and c-Myb, it appears that the intrinsic secondary structure propensities of the IDP play an important part in determining the binding mechanism (48), but much experimental and theoretical work will be required before the principles governing folding and binding are sufficiently well defined that mechanisms can be predicted.

The Tumor Suppressor p53 as a Promiscuous Multivalent Regulator
The intrinsically disordered N-terminal transcriptional activation of p53 binds with varying affinities to four domains of CBP/p300 (TAZ1, TAZ2, KIX, and NCBD), raising the possibility of multivalent binding where each activation domain of the p53 tetramer interacts with a different domain of CBP/ p300, thereby increasing the binding avidity (79,80). The p53 activation domain is bipartite, with two amphipathic interaction motifs (termed AD1 and AD2) that can bind synergistically to a single target protein or bind simultaneously to two different targets to form a ternary complex. Binding to the CBP/p300 domains is dominated by interactions with AD2, whereas interactions with the p53 binding domain of the ubiquitin ligase Mdm2 are dominated by AD1. In unstressed cells, p53 can bind simultaneously to both Mdm2 and CBP/p300 to form a ternary complex that promotes polyubiquitination and degradation of p53 (80). Following genotoxic stress, multisite phosphorylation of the disordered p53 activation domain weakens binding to Mdm2 and enhances binding to CBP/p300, resulting in stabilization of p53, acetylation of the disordered C-terminal regulatory region, and transcriptional activation (80,81). Phosphorylation at Thr 18 functions as a simple switch to control binding to Mdm2. Binding to CBP/p300 is determined by the extent of phosphorylation, with successive phosphorylation events functioning as a rheostat to progressively enhance the binding affinity for the TAZ1, TAZ2, and KIX domains (82). The complex interplay of interactions between CBP/p300, Mdm2, and p53 is made possible by the characteristics of the intrinsically disordered, bipartite activation domain of p53, which facilitates promiscuous interactions with binding partners and dynamically exposes serine and threonine residues for phosphorylation by stress response kinases.

Viral IDPs Compete with Cellular Proteins for CBP/p300 Binding
Intrinsically disordered proteins are highly abundant in viral proteomes (83) and are utilized by viruses to mimic cellular IDR motifs and as hubs or scaffolds to rewire cellular signaling networks (11,84,85). Given the role that CBP and p300 play in the regulation of critical cellular signaling and transcriptional networks, it is perhaps not surprising that they are targeted by numerous viral proteins (12,86). These include oncoproteins encoded by DNA tumor viruses (adenovirus E1A, human papillomavirus E7, and simian virus SV40 large T antigen), Tax and HBZ (HTLV-1 b-zip factor) encoded by the human T-cell leukemia virus HTLV-1, and the Tat protein of HIV-1.
Adenovirus E1A mediates cell transformation by activating viral gene expression and deregulating the host cell cycle and forcing S-phase entry (87). E1A functions as an intrinsically disordered molecular hub that can bind to diverse cellular proteins and subvert the host cell regulatory machinery (88). E1A binds tightly to the TAZ2 domain of CBP/p300, with conserved region 1 (CR1) occupying the ␣ 1 , ␣ 2 , ␣ 3 surface binding surface that is utilized by cellular IDRs (89). The N-terminal region of E1A makes additional, disordered ("fuzzy") contacts with the ␣ 3 , ␣ 4 surface that increase the binding affinity (89,90). E1A binds much more tightly to TAZ2 than do cellular transcription factors (e.g. E1A binds with K d ϳ2 nM (90), whereas p53 activation domain binds with K d ϭ 26 nM (80) and STAT1 binds with K d ϳ50 nM (25)), allowing it to compete efficiently for CBP/ p300, hijack the transcriptional machinery, and inhibit p53mediated cell cycle arrest.
The intrinsically disordered N-terminal half (residues 1-140) of E1A contains four binding motifs, which function synergistically in their interactions with cellular proteins (90). Interactions at these sites are allosteric and display either positive or negative cooperativity (90). The disorder in this region provides E1A with the flexibility to target multiple cellular proteins and organize them into higher order complexes (89,90). In particular, E1A is able to recruit CBP/p300 and the retinoblastoma protein pRb into a ternary complex that promotes acetylation and degradation of pRb and loss of cell cycle control (89).
The human papillomavirus (HPV) E7 protein encoded by strains with a high risk of promoting cancerous transformation employs a mechanism similar to that used by E1A to deregulate the host cell cycle. The CR1 and CR2 regions of E7 are intrinsically disordered (91,92), whereas CR3 is a zinc binding domain that mediates formation of a homodimer (93,94). The disordered CR1 and CR2 regions of E7 from high risk HPV16 bind with high affinity to the TAZ2 domain of CBP (95). Although the TAZ2 binding site overlaps the binding site for pRb, the full-length E7 dimer functions as a scaffold to recruit TAZ2 and the pRb pocket domains into a ternary complex. By bringing the CBP/p300 HAT domain into proximity to pRb, the E7 dimer stimulates acetylation and subsequent degradation of pRb, and drives S phase entry and deregulation of the host cell cycle (95). In contrast, E7 from the low risk strain HPV6b binds TAZ2 with much lower affinity than high risk HPV16 E7. This would impair the ability of HPV6b to suppress p53-mediated transcriptional programs and disfavor formation of a ternary complex with CBP/p300 and pRb (95).
The HTLV-1 Tax and HBZ proteins, as well as the HIV-1 Tat protein, recruit CBP/p300 to viral promoters to activate transcription of viral genes. All three proteins utilize intrinsically disordered interaction motifs to bind to the MLL site on the KIX domain (96 -98). Tax and HBZ compete with cellular transcription factors for binding to KIX, thereby interfering with CBP/p300-mediated transcriptional processes that regulate critical hematopoietic signaling pathways (98,99). Infection by these viruses is frequently a precursor to diseases such as leukemia.

Conclusion
The central role of the transcriptional coactivators CBP and p300 in regulation and integration of eukaryotic transcriptional and signaling pathways is entirely dependent upon the existence and utilization of intrinsically disordered regions, both within the coactivators themselves and in the interaction domains of their partners. Disorder mediates the formation of both highly specific and promiscuous complexes, with dissociation constants that range from nanomolar to micromolar; these complexes can be dissociated either spontaneously or by competition with other partners. This facility is particularly important for signaling, where the activation of genes in response to extracellular signals must not only be turned on with high specificity, but must also be turned off when these gene products are no longer required. Disorder also allows sequences to interact with a number of different partners, promoting redundancy and efficiency in the utilization of cellular resources, as well as facilitating crosstalk between signaling pathways. The coactivators CBP and p300 provide a paradigmatic example of a system that employs both order and disorder to fulfill complex and multifarious functions.