The hormone insulin has a central role in human physiology, yet the answer to a fundamental biochemical question — how it binds to its cell-surface receptor — has remained elusive, until now. See Letter p.241
Insulin holds a storied place in the fields of physiology and biochemistry1. First isolated in 1921, it was soon used to save the lives of individuals with diabetes. This medical breakthrough and subsequent biochemical research on insulin have garnered three Nobel prizes. The three-dimensional structure of insulin was determined by X-ray crystallography in 1969. Yet, despite intense interest, an atomic view of insulin bound to its receptor has been lacking — a situation rectified by the report of Menting et al.2 in this issue.
Over the years, researchers have used protein crystallography to determine the structure of numerous hormones, growth factors and cytokines bound to their receptors3,4. In general, structures of these ligand–receptor systems were tractable because the ligand-binding domains of the receptors were of moderate size and many could be produced in bacteria. Moreover, the binding modes of these ligands to their receptors are relatively simple, involving one or two receptor subdomains that are contiguous in amino-acid sequence. However, the production of insulin-receptor protein for structural studies is fraught with challenges.
The insulin receptor belongs to the receptor tyrosine-kinase family of cell-surface receptors5. These are single-pass transmembrane proteins with an extracellular region that binds a protein ligand (typically a growth factor) and a cytoplasmic region that contains a tyrosine-kinase domain, which phosphorylates specific tyrosine amino-acid residues on other signalling proteins, as well as on the receptor itself. On ligand binding, many receptor tyrosine kinases form a dimer, which facilitates tyrosine phosphorylation of one cytoplasmic domain by the other, thereby activating the receptor6. By contrast, the insulin receptor is a preformed, disulphide-bridged dimer, and insulin binding is thought to induce a conformational change in the receptor that initiates cross-phosphorylation of the two cytoplasmic domains.
The mature insulin receptor consists of two copies of two polypeptide chains, α and β. The α-chain (723 amino-acid residues) is entirely extracellular and heavily glycosylated. The β-chain (620 residues) starts on the extracellular side and spans the membrane (by way of an α-helix) to the cytoplasmic side, which harbours the tyrosine-kinase domain (Fig. 1a). Each α-chain is linked to a β-chain through a disulphide bridge to form an αβ half-receptor, and the two half-receptors are linked by at least two (and possibly four) disulphide bridges. The extracellular region of each half-receptor contains a series of folded domains: L1, C, L2, F1, F2 and F3 (Fig. 1a).
Mature insulin consists of two polypeptide chains: an A chain of 21 residues and a B chain of 30 residues. It contains one intrachain disulphide bridge (in the A chain) and two interchain disulphide bridges.
Previous biochemical studies1,7 established that a single insulin molecule binds, with sub-nanomolar affinity, to the dimeric receptor (1:2 stoichiometry) and activates it. The two equivalent insulin-binding sites on the receptor are composed of two distinct interfaces. The primary interface consists of the L1 domain of one half-receptor and the carboxy-terminal region of the α-chain (αCT) of the second half-receptor. The secondary interface comprises the loop regions near the junction of F1 and F2 in the second half-receptor.
In 2006, some of the authors of the current study determined the atomic structure of the dimeric extracellular portion of the insulin receptor in the apo conformation (devoid of insulin). This landmark structure8 revealed an inverted V shape, with the two half-receptors arranged in an antiparallel fashion (Fig. 1b,c). Although an insulin-mimetic peptide was present in the crystallization trials, it could not be located in the electron-density map and was presumed to be unbound.
Menting et al. now use some of the same strategies to crystallize truncated versions of the α-chain, containing only the primary binding interface (involving L1 and αCT), in the presence of insulin. Because the αCT region is a considerable distance in sequence from L1, they either added it as a synthetic peptide or spliced it onto the C-terminal end of the truncated α-chain construct. This time, the catch was made: insulin was visible in the electron-density maps along with the receptor.
In the four crystal structures the authors determined, insulin is bound to the L1 domain, but just barely (Fig. 1d). Almost all of the hydrophobic residues extending from the flat β-sheet surface of L1, which were previously mapped1 as insulin-binding 'hotspots', contact the αCT helix rather than insulin, and it is the αCT helix that is in intimate contact with insulin.
Two structural transitions occur on insulin binding, one in insulin and one in αCT (L1 is not appreciably altered). In structures of insulin alone, residues at the C-terminal end of the B chain pack against the rest of the molecule, but are displaced on binding to αCT. This observation confirms a long-standing prediction that such a structural transition occurs in insulin on binding to its receptor1. What was not predicted, however, was the behaviour of αCT. In the apo structure, the αCT helix is bound to L1, but insulin binding causes a repositioning of the helix on the L1 surface such that it interacts with insulin, as well as with L1.
Menting and colleagues' structures also provide insights into the receptor-activation mechanism. If the L1 domain from the insulin-bound structure is superimposed on L1 from the apo structure, it is apparent that a conformational change must take place in the other half-receptor (near F1 and F2, the secondary binding interface) to accommodate insulin binding to αCT–L1 (Fig. 1e). Evidently, this change underlies the structural mechanism whereby insulin binding triggers cross-phosphorylation of the cytoplasmic kinase domains of its receptor. The precise nature of this conformational change and how it repositions the kinase domains remain unresolved.
Fast-acting insulin analogues have been used to treat patients with diabetes for many years, and the three-dimensional structure of insulin was instrumental in their design9. The structure of insulin bound to its receptor is not only the latest milestone in insulin research, but should also afford opportunities for the design of insulin analogues with enhanced receptor binding, as well as favourable pharmacokinetics.
Ward, C. W. & Lawrence, M. C. Front. Endocrinol. 2, 76 (2011).
Menting, J. G. et al. Nature 493, 241–245 (2013).
Stroud, R. M. & Wells, J. A. Sci. STKE 2004, re7 (2004).
Wang, X., Lupardus, P., Laporte, S. L. & Garcia, K. C. Annu. Rev. Immunol. 27, 29–60 (2009).
Lemmon, M. A. & Schlessinger, J. Cell 141, 1117–1134 (2010).
Hubbard, S. R. & Miller, W. T. Curr. Opin. Cell Biol. 19, 117–123 (2007).
De Meyts, P. Trends Biochem. Sci. 33, 376–384 (2008).
McKern, N. M. et al. Nature 443, 218–221 (2006).
Pandyarajan, V. & Weiss, M. A. Curr. Diab. Rep. 12, 697–704 (2012).
About this article
Could insulin receptor H1085H C > T single nucleotide polymorphism predict insulin resistance in type 2 diabetic and chronic hepatitis C virus patients in Egypt?
Egyptian Liver Journal (2021)
Insulin-like plant proteins as potential innovative drugs to treat diabetes—The Moringa oleifera case study
New Biotechnology (2017)