Physiology meets biophysics: Visualizing the interaction of hypoxia-inducible factor 1α with p300 and CBP

Complex circulatory and respiratory systems are established during embryonic development and used in fetal and postnatal life to ensure oxygen delivery to every cell in the human body. Within each cell, the utilization of O₂ as a substrate for biochemical reactions, most notably as an electron acceptor in the mitochondria, is also highly regulated. As a result of systemic and cellular physiological mechanisms that control O₂ delivery and consumption, O₂ concentrations are precisely maintained within a narrow range that represents a balance between cellular metabolic requirements and the risk of oxidative damage. Since its identification ten years ago (1), an exponentially growing body of experimental data indicates that the transcription factor hypoxia-inducible factor 1 (HIF-1) functions as a global regulator of O₂ homeostasis, as it is required for the establishment of the circulatory and respiratory systems as well as for physiological responses to hypoxia in prenatal and postnatal life (2–7).

Altered O₂ homeostasis plays a critical role in the pathophysiology of ischemic and neoplastic disorders, which are the most common causes of mortality in the U.S. population. In ischemic disorders, HIF-1 activates transcription of genes encoding vascular endothelial growth factor (VEGF), which promotes angiogenesis; glucose transporters and glycolytic enzymes, which allow for the production of ATP in the absence of O₂; and survival factors such as VEGF, erythropoietin (EPO), and insulin-like growth factor 2, which inhibit ischemia-induced apoptosis (reviewed in ref. 8). Thus, therapeutic strategies designed to increase HIF-1 activity may prevent ischemia or infarction in patients with advanced atherosclerosis (9). In cancer, these same pathways are used to promote tumor cell survival, proliferation, invasion, and metastasis (reviewed in refs. 10–12). Increased HIF-1 expression is associated with increased patient mortality in several types of cancer, and therapeutic strategies designed to inhibit HIF-1 activity may improve the survival of such patients (12).

Over the same decade during which the involvement of HIF-1 in development, physiology, and pathophysiology has been established, novel molecular mechanisms that regulate HIF-1 activity have also been delineated. Protein purification revealed that HIF-1 was a heterodimer composed of HIF-1α and HIF-1β subunits (13), which dimerize and bind to DNA by means of basic helix–loop–helix–PAS (bHLH-PAS) domains (14). Expression of the 826-aa HIF-1α subunit is tightly regulated by the cellular O₂ concentration (15) by means of ubiquitination and proteasomal degradation (16, 17) that is mediated by binding of the von Hippel–Lindau tumor suppressor protein (VHL) to HIF-1α (18). This interaction occurs only when HIF-1α has been hydroxylated at residues 402 and 564 (19–23). O₂ is a rate-limiting substrate for the prolyl hydroxylases that modify HIF-1α (20), thus providing a molecular mechanism for hypoxia-inducible expression of HIF-1α (Fig. 1).

Figure 1.

Oxygen-dependent hydroxylation events that regulate HIF-1α protein stability and transcriptional activity. Under normoxic conditions, HIF-1α is hydroxylated (OH) on proline residues 402 and 564 (P402 and P564) and asparagine residue 803 (N803), which are located within the carboxyl- (TAD-C) and amino- (TAD-N) terminal transactivation domains, respectively. bHLH, basic helix–loop–helix domain. Hydroxylation of P402 and P564 by the prolyl hydroxylases PHD-1, -2, or -3 (also known as HIF-1α prolyl hydroxylases HPH-1, -2, or -3) is required for the binding of the von Hippel–Lindau tumor suppressor protein (VHL), which in turn recruits an E3 ubiquitin–protein ligase complex (containing elongin B, elongin C, CUL2, and RBX1) that targets HIF-1α for proteasomal degradation. VHL also recruits histone deacetylases that repress transactivation. Hydroxylation of N803 by FIH-1 prevents the interaction of TAD-N with the CH1/TAZ1 domain of the coactivators p300 and CBP (CREB-binding protein) which is necessary for transactivation of target genes by HIF-1.

HIF-1α transactivation domain function is also subject to O₂-dependent negative regulation (24, 25). The amino-terminal transactivation domain is negatively regulated by the recruitment of histone deacetylases by VHL and by factor inhibiting HIF-1 (FIH-1), which binds to both VHL and HIF-1α (26). The binding of the coactivators CBP (CREB-binding protein) and p300 to the carboxyl-terminal transactivation domain (TAD-C; also known as CAD and CTAD) is critical for HIF-1 transcriptional activity (27, 28). HIF-1α TAD-C binds to a domain in p300 and CBP that is known as the cysteine/histidine-rich 1 (CH1) or transcriptional adapter zinc-binding 1 (TAZ1) domain (29). CBP and p300 perform both structural and enzymatic functions in mediating the effects of dozens of DNA-binding transcriptional activators (reviewed in refs. 30 and 31). First, they function to physically link these sequence-specific DNA-binding proteins to the transcription initiation complex that contains RNA polymerase II and more than 50 associated proteins. Second, they function as histone acetyltransferases which perform chromatin remodeling that is required for gene transcription. Binding of CBP and p300 to HIF-1α is negatively regulated by the O₂-dependent hydroxylation of asparagine-803 in TAD-C by FIH-1 (32, 33).

Thus, both the stability and transcriptional activity of HIF-1α are negatively regulated by hydroxylation (Fig. 1). Hydroxylation plays a role similar to that of other posttranslational modifications (e.g., phosphorylation) in regulating protein–protein interactions, but it is unique in its dependence upon the cellular O₂ concentration. These exciting discoveries have provided an elegant molecular basis for the O₂-dependent regulation of gene transcription by HIF-1. This understanding has now been advanced to atomic resolution with the publication of two recent papers in PNAS describing NMR spectroscopic analyses of the structural interaction of HIF-1α TAD-C with the CH1/TAZ1 domain of either p300 (34) or CBP (35).

In one study (34), the structure of HIF-1α amino acid residues 786–826 in a complex with p300 residues 323–423 was determined. The p300 CH1 domain consists of four α-helices that are folded into a triangular structure that is stabilized by the coordination of three Zn²⁺ atoms by means of a histidine and three cysteine residues and also by a single hydrophobic core that is formed by all four helices. The complex of p300 CH1 and HIF-1α TAD-C has three remarkable characteristics. First, the interactions between the two domains are extensive: the entire TAD-C is embedded in CH1 with almost half of its surface buried in the interface. The buried surface area is more than twice the average for protein complexes in general and almost three times that of the complex consisting of the CREB transactivation domain bound to the KIX domain of CBP. Second, the tertiary structure of TAD-C in the complex is determined exclusively by its interactions with CH1 and, in the absence of CH1, TAD-C is essentially unstructured. The proposed induced-fit model provides a basis for the interaction of CH1 with a variety of transcriptional activators (ETS1, HNF4, p53, PIT1, STAT2, and others) that share no common sequence features, although this will require experimental verification. Third, Asn-803 is the single most buried HIF-1α residue in the complex. Although the chemistry of the asparagine hydroxylation reaction has not been determined, the structural data indicate that hydroxylation in either the pro-R or pro-S position would be highly disruptive, providing a structural basis for the regulation of complex formation by O₂-dependent hydroxylation.

In the other study published in PNAS (35), the structure of HIF-1α amino acid residues 776–826 in a complex with the TAZ1 domain of CBP (residues 323–423) was determined. Although the HIF-1α fragment analyzed in this study contained an additional 10 amino acid residues, it was also intrinsically disordered in the unbound state and became structured upon binding to CBP TAZ1. When isothermal titration calorimetry was used, the dissociation constant of the complex between TAD-C and TAZ1 was determined to be 7 nM, indicating a remarkably high affinity as predicted by the extensive interactions between these domains. Whereas binding of HIF-1α(786–826) to p300 CH1 induced the formation of two α-helices involving residues 797–803 and 816–822, the binding of HIF-1α(776–826) to CBP TAZ1 induced the formation of three α-helices involving residues 784–788, 796–804, and 815–824. In the complex with CBP, Asn-803 of TAD-C is located on an α-helix and is buried deep in the protein–protein interface (as was observed for the p300 complex), where it participates in multiple hydrogen-bonding interactions that are predicted to stabilize both the α helix and the complex.

The absence or presence of hydroxylation of Asn-803 can be viewed as a switch that turns TAD-C function on or off by determining the ability of the domain to interact with p300 or CBP. Mutation of Asn-803 to alanine completely eliminates the O₂-dependent negative regulation of TAD-C function and interaction with p300 (33). Taken together with the structural studies, these data suggest that a small molecule targeted to the α-helix containing Asn-803 might inhibit the interaction of TAD-C with p300 or CBP. Such an inhibitor would be potentially useful in the treatment of patients with cancers in which HIF-1α overexpression is associated with increased mortality (reviewed in ref. 12).

The pharmaceutical industry has in general avoided transcription factors as molecular targets for drug development and inhibition of protein–protein interactions as a mechanism of drug action. The rationale has been that a nuclear target is more difficult to hit because the drug must pass through an additional membrane bilayer. It is also considered less likely that a small molecule can specifically disrupt extensive interactions between two protein domains as compared with targeting an enzyme for inhibition by occupying its active site. However, the recent identification of small molecules that inhibit the dimerization of the transcription factors MYC and MAX in vitro and MYC-induced transformation in cell culture (36) suggests that the aforementioned historical biases against targeting transcription factors and protein–protein interactions may not have as compelling a scientific basis in the present era of high-throughput molecular screening.

Expression of TAD-C peptides that block the interaction between HIF-1α and p300 inhibit tumor xenograft growth in nude mice (29). However, such peptides also inhibit the interaction of p300 CH1 with other transcription factors (see above) and thus the extent to which the observed effects are attributable to reduced interaction of HIF-1α and p300 per se is unknown. Herein lies a potential problem with this pharmacologic approach: One would like to inhibit the interaction of p300 and CBP with HIF-1α specifically, and thus the latter protein is the preferred target for binding of a small molecule inhibitor. However, the disordered structure of TAD-C before interaction with p300 or CBP binding may reduce the likelihood of identifying a small molecule capable of specific high-affinity binding. Adding to the challenge is the high affinity of the competing interaction of p300 or CBP with HIF-1α. Nevertheless, the demonstration that hydroxylation of Asn-803 can disrupt these interactions suggests that this critical molecular switch may represent an Achilles' heel that can be targeted therapeutically.

Clinical applications as yet reside in that fantastic realm where every basic science discovery is immediately translated into a powerful disease-conquering therapy, and such dreams are not disturbed by the harsh light of empiricism. For now, it is exciting to appreciate the molecular portraits that are presented in the papers by Freedman et al. (34) and Dames et al. (35). These data, along with recently reported structures of the HIF-1α–VHL interaction (37, 38), provide us with a remarkable opportunity to view physiology through the eyes of the biophysicist.