The cellular response to low oxygen levels is regulated by a process in which one protein is ousted from a binding site by another. It emerges that protein disorder allows the displacement to occur remarkably efficiently. See Letter p.447
Imagine two different ligand molecules, either of which can individually bind to the same site on a protein. Now imagine those ligands competing with each other to bind to a population of the protein molecules: a mixture of two complexes will form, in which either one or the other ligand is bound to the protein. It is generally assumed that the proportions of the complexes in the mixture will be defined by the binding affinities of the ligands for the protein. But on page 447, Berlow et al.1 report an unexpected result that challenges this assumption — a finding that could alter our understanding of the behaviour of protein-interaction networks.
The authors studied hypoxia-inducible factor 1α (HIF-1α), a protein whose production increases when cellular oxygen levels are low. HIF-1α binds the protein CBP, and together they initiate a response to the low oxygen level2. This response is tempered by a feedback mechanism3 in which HIF-1α promotes production of the protein CITED2. CITED2 in turn displaces HIF-1α from the HIF-1α–CBP complex by occupying a binding pocket on CBP that largely overlaps with the binding pocket for HIF-1α, thus inhibiting the hypoxic response. HIF-1α and CITED2 bind to CBP with equal affinity, which means that, if HIF-1α and CITED2 were present in equal amounts, then equal populations of CBP molecules should be bound by HIF-1α and CITED2. But instead, Berlow et al. report that an equivalent concentration of CITED2 can rapidly displace HIF-1α from CBP.
The regions of HIF-1α and CITED2 that bind to CBP are both intrinsically disordered when unbound, meaning that they do not assume a single, low-energy folded conformation, but remain flexible so that they can adopt multiple conformations. Many disordered proteins settle into a predominantly folded conformation on binding to a folded binding partner, as is the case for the region of CITED2 that binds to CBP. Berlow et al. use nuclear magnetic resonance (NMR) spectroscopy to show that the αB and αC helices of HIF-1α are also ordered when bound to CBP, but the αA helix remains relatively mobile. The retained flexibility of the HIF-1α αA helix might allow the αA helix of CITED2 — which becomes less mobile when bound — to 'get a foot in the door' of HIF-1α-bound CBP.
Berlow et al. propose a displacement mechanism in which the conformational dynamics of the CBP-bound HIF-1α αA helix allow the CITED2 αA helix to squeeze into the binding pocket occupied by HIF-1α (Fig. 1). Binding by the CITED2 αA helix in turn leads to a structural shift in CBP that decreases its affinity for the αB and αC helices of HIF-1α, and increases its affinity for the full interaction with CITED2. The authors used fluorescence-based assays to show that the rate of HIF-1α displacement by CITED2 is proportional to the concentration of the latter, indicating that the process probably involves a ternary CBP–HIF-1α–CITED2 complex. The researchers also used NMR to observe molecular interactions during the process, confirming that HIF-1α and a protein fragment corresponding to the CITED2 αA helix can bind simultaneously to CBP. These experiments again show that binding of the CITED2 αA helix pushes the CBP structure towards a conformation that favours CITED2 binding.
The proposed mechanism is, at its core, an allosteric one — that is, a process in which an input or interaction at one site on a protein leads to a modified output or interaction at a different site. The discovery of allostery fundamentally changed our understanding of enzymatic functions4, and Berlow and colleagues' findings suggest that it will also radically change our interpretation of how binding affinities describe protein-interaction networks. In the present case, binding of the CITED2 αA helix at one site leads to a structural shift in CBP that decreases the latter's affinity for the αB and αC helices of HIF-1α at a second site.
Although the molecular details for this particular example are coming into focus, it is unclear where the energy required to shift the conformation of CBP comes from. The answer might lie in the allosteric response of the intrinsically disordered regions. The allosteric mechanisms of intrinsically disordered proteins have surprised researchers before. For example, interactions between CBP and the disordered protein E1A can be either enhanced or impaired by formation of a ternary complex with another protein, pRb, depending on the available binding sites on E1A (ref. 5).
In another example, the disordered 'tails' of the DNA-binding domain of the dimeric Phd protein inhibit binding of a second Phd dimer to a DNA molecule6. There are no inhibitory interactions between the ordered regions of the Phd dimers, so it was not initially clear how binding of the second Phd dimer was inhibited. It was shown that the tails provide a 'fuzzy' steric hindrance (an obstacle associated with the physical volume of parts of a molecule) to binding of a second Phd dimer that acts like an entropy barrier, with the size of the barrier proportional to the degree of disorder. The measured affinity of a single Phd dimer for DNA, and the absence of inhibitory interactions between the ordered regions, did not reveal the whole story in this case.
Just as interactions between the disordered tails of the Phd dimers inhibit binding of a second Phd dimer, one might speculate that disordered portions of CITED2 and HIF-1α interact in a manner that modulates their affinity for CBP. The implication is that affinities determined for protein pairs may not be sufficient to determine the outcome of competitive protein interactions, especially when allostery and intrinsically disordered proteins are involved.
Protein–protein interactions were previously understood to involve relatively static interfaces, but examples of substantially dynamic interactions7, including fuzzy complexes8, are becoming more common. These interactions can involve an equilibrium state in which multiple binding motifs on a ligand dynamically bind and unbind a single binding pocket (or multiple pockets)9. Dynamic protein-interaction networks have also recently been discovered10, in which thousands to millions of protein molecules undergo phase separation into protein-dense regions or clusters, observable as micrometre-sized, visually distinct droplets in cells. Many of the proteins involved have large, intrinsically disordered regions, and the droplets have some of the characteristics of liquids. Additional complexity in these systems is introduced through protein modifications, such as phosphorylation.
Returning to CBP, it will be interesting to see if cellular pH changes associated with respiratory distress modulate the competition between CITED2 and HIF-1α for their pH-sensitive binding pockets on CBP. More broadly, the modulation of intrinsically disordered proteins through allosteric interactions in ternary complexes may well form the basis of a range of possibilities for fine-tuning regulatory protein interactions that control biological responses to stimuli.Footnote 1
Berlow, R. B., Dyson, H. J. & Wright, P. E. Nature 543, 447–451 (2017).
Arany, Z. et al. Proc. Natl Acad. Sci. USA 93, 12969–12973 (1996).
Yin, Z. et al. Proc. Natl Acad. Sci. USA 99, 10488–10493 (2002).
Swain, J. F. & Gierasch, L. M. Curr. Opin. Struct. Biol. 16, 102–108 (2006).
Ferreon, A. C. M., Ferreon, J. C., Wright, P. E. & Deniz, A. A. Nature 498, 390–394 (2013).
Garcia-Pino, A. et al. Nature Chem. Biol. 12, 490–496 (2016).
Wright, P. E. & Dyson, H. J. Nature Rev. Mol. Cell Biol. 16, 18–29 (2015).
Sharma, R., Raduly, Z., Miskei, M. & Fuxreiter, M. FEBS Lett. 589, 2533–2542 (2015).
Csizmok, V. et al. Nature Commun. 8, 13943 (2017).
Mitrea, D. M. & Kriwacki, R. W. Cell Commun. Signal. 14, 1–20 (2016).
Related links in Nature Research
About this article
Cite this article
Chong, P., Forman-Kay, J. A hidden competitive advantage of disorder. Nature 543, 325–326 (2017). https://doi.org/10.1038/nature21893
Archives of Biochemistry and Biophysics (2020)
Biogenic amines: a key freshness parameter of animal protein products in the coming circular economy
Current Opinion in Food Science (2018)
Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics (2017)
Science Signaling (2017)