The growth factor TGF-ß1 is located inside a protein cage, and is thought to be released by force applied through integrin proteins. A structure of TGF-ß1 in complex with integrin αVß6 sheds light on the uncaging process. See Article p.55
Much like humans, cells often exchange information. Proteins of the transforming growth factor-β (TGF-β) family act as key chemical signals for such exchange. These proteins are unusual in that they are packaged inside a protein cage, and can be accessed by cells only when the cage is opened. Mounting evidence1 suggests that mechanical forces applied by the cell through membrane proteins called integrins are responsible for rattling the TGF-β proteins out of their cages. On page 55, Dong et al.2 report crystal structures of integrin bound to a TGF-β cage, and show with unprecedented clarity how this rattling is achieved.
Integrins are heterodimers composed of α and β integrin subunits. They connect the extracellular matrix, which surrounds the cell, to the cytoskeleton, which consists of filaments of actin protein and provides intracellular structure. There are 24 different integrin heterodimers3, of which αVβ6 activates the protein TGF-β1.
Caged TGF-β1 is linked to the extracellular matrix through an adaptor protein, and consists of a latent version of the TGF-β1 protein surrounded by an extended amino-terminal 'pro-domain' that has been cleaved away from the protein proper — this region stably wraps around the protein to form the cage, preventing receptor binding. Integrin αVβ6 uncages TGF-β1 through interactions with a sequence of three amino-acid residues (arginine, glycine, aspartate; RGD) on the pro-domain4. This integrin–RGD binding is essential for TGF-β1 signalling: mutation of a single RGD residue in mice produces traits similar to those caused by complete deletion of TGF-β1, including multi-organ inflammation and defects in formation of the blood-vessel network5.
An attractive model has emerged in which a cell that needs TGF-β1 yanks on the pro-domain through integrins, which are presumably powered by actin-filament movements (Fig. 1). This action loosens the pro-domain and releases the active growth factor for local use (or to make it available to nearby cells). In a sense, TGF-β1 comes in a box, gift wrapped with ribbons — opening the box involves pulling on the ribbons in the correct direction.
To understand this process better, Dong et al. used X-ray crystallography to determine the structure of caged TGF-β1 bound to αVβ6. The structure of the pro-domain, although not its contents, is considerably different from a previous structure of caged TGF-β1 alone6. Interestingly, biophysical measurements performed by the authors suggested that the binding affinity between TGF-β1 and the pro-domain is not affected by integrin binding. What, then, is the role of this structural change?
Dong and colleagues noted that the binding interface between the pro-domain and αVβ6 is highly interdigitated, resembling interlocking fingers — different from the flat contacts found at most protein–protein interfaces. Such an interface would not have been predicted either by computational models of docking between pre-existing structures of caged TGF-β1 and integrins (which assume that the two proteins maintain a stable structure during the interaction) or analyses of previous structures of integrins bound to their ligands (which revealed relatively small contact areas). However, this interdigitation and the associated large binding interface are probably responsible for the stable orientation between integrin and caged TGF-β1, which the authors observed using a second technique for structure determination, small-angle X-ray scattering. This low-resolution technique does not suffer from the potential for artefacts caused by crystal packing, which can force proteins into an abnormal orientation.
The importance of a stable binding orientation was highlighted by Dong and co-workers' computer simulations of the uncaging process. When progressively increasing force was applied to the pro-domain through β6 integrin, the cage opened, breaking most of its contact with the growth factor. By contrast, applying the force through αV caused this subunit to unfold before any major structural changes to the pro-domain occurred. These results suggest that force application in the correct orientation — that is, through the region of the pro-domain that contacts β6 — is essential for opening the cage.
And it seems that nature has specifically engineered integrins this way. As the authors noted, all subdomains in β integrins are connected to adjacent subdomains by two covalent bonds, whereas in α integrins they are linked through a single bond. This ensures that force is securely transmitted through β integrins, and that α integrins are prone to failure. Because TGF-βs and integrins arose during the same evolutionary period and in the same phylogenetic branch, the mechanism of force-induced TGF-β1 activation is probably shared across many species. However, unlike TGF-β1 and TGF-β3, the pro-domain of TGF-β2 lacks a recognizable integrin binding motif such as RGD (ref. 1). The mechanism by which TGF-β2 is released from its cage is unknown.
Much remains to be understood about TGF-β uncaging. Because Dong et al. used very high forces in their simulations, they could not determine the minimum force required to release TGF-β1. Certain more-subtle changes to the pro-domain structure would probably be sufficient for uncaging. Would αVβ6 remain bound to the pro-domain at these forces, as it did in the simulations, or is the binding seen in the simulations an artefact of the extreme forces used? Answering this question will provide a key test of the current model, because the unbinding force between the integrin and the pro-domain should be as high or higher than the force required to open the cage. To address this issue, single-molecule methods could be used to directly measure forces across integrins during unbinding and uncaging7,8.
The method used by the authors to detect TGF-β1 involves measuring the activity of the released growth factor, and as such is indirect, with low time resolution and sensitivity. By contrast, the role of integrins in another process (remodelling extracellular matrices) can be directly examined at single-cell resolution using fluorescence-based methods9. If a similar resolution could be achieved for TGF-β activation, it would allow researchers to determine whether the released growth factors are used by the cells that do the uncaging or by the surrounding cells.
More broadly, Dong and colleagues' structure is a major leap forward in our understanding of how integrins interface with the extracellular environment. This should and will inspire researchers to ask deeper and more difficult questions about the links between mechanical and chemical communication between cells.Footnote 1
Robertson, I. B. & Rifkin, D. B. Cytokine Growth Factor Rev. 24, 355–372 (2013).
Dong, X. et al. Nature 542, 55–59 (2017).
Mas-Moruno, C. et al. Angew. Chem. Int. Edn 55, 7048–7067 (2016).
Munger, J. S. et al. Cell 96, 319–328 (1999).
Yang, Z. et al. J. Cell Biol. 176, 787–793 (2007).
Shi, M. et al. Nature 474, 343–349 (2011).
Kong, F., Garcia, A. J., Mould, A. P., Humphries, M. J. & Zhu, C. J. Cell Biol. 185, 1275–1284 (2009).
Wang, X. & Ha, T. Science 340, 991–994 (2013).
Pankov, R. et al. J. Cell Biol. 148, 1075–1090 (2000).
About this article
Journal of Medicinal Chemistry (2019)
Der MKG-Chirurg (2018)
Comprehensive transcriptome analysis of fluid shear stress altered gene expression in renal epithelial cells
Journal of Cellular Physiology (2018)
The CRISPR Journal (2018)
Growth Factors (2017)