The branch of the immune system called innate immunity has a pivotal role in host defence by recognizing general hallmarks of disease-causing agents. The intracellular protein STING, a transmembrane protein usually located on an organelle called the endoplasmic reticulum, is a key regulator of this type of immune response1. Writing in Nature, Shang et al.2 and Zhang et al.3 report full-length structures of STING, including STING in complex with the kinase protein TBK1, which initiates the downstream signalling pathway that is triggered on STING activation.
The abnormal presence of double-stranded DNA in the cytoplasm is a potent danger signal that activates STING. If the enzyme cGAS senses such DNA4,5, it makes the molecule cGAMP; when this binds and activates STING, a signalling cascade begins that eventually alters gene expression to generate proinflammatory molecules. Abnormalities in this defence mechanism can underpin a spectrum of conditions, including cancer, autoinflammatory syndromes or neurodegenerative diseases6–9. STING is thus a highly promising drug target10, but for such efforts to succeed, it is essential to understand how STING activity is regulated.
Previous studies11–13 have illuminated how the cytoplasmic domains of STING interact with cGAMP and how the downstream signalling events that follow STING activation are triggered. For example, cGAMP binds to a V-shaped pocket that is formed by the cytoplasmic domains of two STING proteins that make a dimer11 and, after undergoing a conformational change, STING is transported to an organelle called the Golgi complex, where it recruits TBK112. Despite such progress, how cGAMP binding causes STING activation and facilitates the subsequent downstream signalling events was unknown.
Shang and colleagues used the technique of cryo-electron microscopy (cryo-EM) to generate near atomic-resolution structures of full-length human STING without cGAMP bound to it, and full-length chicken STING with and without cGAMP bound. Without cGAMP bound, the STING dimer is stabilized by interactions between different domains in each STING protein and between domains of the two different STING proteins.
A transmembrane helix is linked to the cGAMP-binding domain through a connector element — made up of a connector helix and a connector loop — that forms a crossover point between the two STING proteins. In the cGAMP-bound state, the connector elements and the cGAMP-binding region of each subunit unwind, causing a 180° rotation (see Supplementary Video 1 of ref. 2) of the cGAMP-binding domain (Fig. 1). This movement is probably initiated by cGAMP, which might push apart the junction of the connector element and the cGAMP-binding domain. Some disease-causing mutations of STING are in this junctional region, suggesting that these mutations might cause this rotation to occur even in the absence of cGAMP.
The cryo-EM data for cGAMP-bound STING generated by Shang et al. offers some intriguing clues about how this unwinding and 180° rotation trigger STING activation. In an activated, cGAMP-bound state, STING dimers are tightly packed and arranged side by side in the lipid membrane. The dimers can make connections with adjacent dimers, and these connections are stabilized by a loop that connects the dimers at their interface. The authors’ modelling suggests that, in the absence of cGAMP binding, this interface loop would be in an orientation that would block tight binding between adjacent STING dimers. The connector element probably stabilizes this inhibitory orientation of the interface loop when cGAMP isn’t bound, suggesting that the rearrangement of the connector element on cGAMP binding could promote tetramerization and is associated with STING activation. The authors observed the formation of activated STING tetramers, and it is probable that oligomers of activated STING form that are larger than this.
Zhang et al. investigated the interaction of STING with TBK1. Their cryo-EM data reveal that a dimer of TBK1 proteins is located on top of the cGAMP-binding domain of the STING dimer. This interaction between STING and TBK1 is mediated by an evolutionarily conserved stretch of eight amino-acid residues in the carboxy-terminal ‘tail’ of STING — a part of the protein that was not visible in earlier STING structures. This C-terminal part of STING is tethered to the cGAMP-binding domain by a flexible linker region, allowing STING and TBK1 to adopt different orientations relative to each other and to interact independently of whether cGAMP has bound STING. This suggests that the role of cGAMP binding in promoting the interaction between STING and TBK1 is probably indirect; it might enforce an oligomeric state of STING or initiate STING movement to the Golgi complex.
The structure of STING in complex with TBK1 suggests that the autophosphorylation of TBK1 (the addition of a phosphate group to one TBK1 by another TBK1) that is necessary for TBK1 activation cannot be carried out by TBK1’s dimer partner. Moreover, although activated TBK1 phosphorylates STING, the structural information indicates that the phosphorylation site on STING is probably located beyond the reach of the catalytic domains of a TBK1 dimer bound to a STING dimer. Together, these features suggest that a complex of one STING dimer and one TBK1 dimer would fail to phosphorylate the constituent proteins, and supports a model in which oligomerization of activated STING leads to the phosphorylation of neighbouring TBK1 dimers, which, in turn, phosphorylate neighbouring STING molecules. The authors speculate that the flexibility in the possible orientations for interaction between TBK1 and STING could aid this activation process.
Shang, Zhang and their respective colleagues have pushed the limits for high-resolution cryo-EM of transmembrane proteins. Solving the structure for this type and size of protein complex is a major challenge14, but their success will probably motivate others to try to solve the cryo-EM structures of similarly sized (about 80-kilodalton) transmembrane protein complexes.
The structural models emerging from the authors’ studies might aid investigations that seek to answer other questions about how STING functions. For example, the mechanism that regulates STING movement from the endoplasmic reticulum to the Golgi complex remains to be determined. STING also has functions that do not require TBK1 activity, including the initiation of a degradation process called autophagy, and how such processes are controlled is not fully understood. Moreover, the structures will certainly be useful for drug-discovery programmes by pinpointing protein regions that might offer a targeting opportunity to precisely manipulate STING activity. Such efforts might result in STING-targeting therapies for the treatment of human disease.
Nature 567, 321-322 (2019)