Thread by @connor_rosen

cGAS-STING work continues to boggle the mind - flurry of new papers in the last two weeks branching the field in new directions. I’ll try for quick (not really) takes… #cGAS #microbiology #Immunology #Inflammation @nature @biorxivpreprint @NatureMicrobiol @CellReports

Paper 1: @aaronwhiteley et al writing in @nature - nature.com/articles/s4158…. cGAS-like nucleotidyltransferases in bacteria (CD-NTases) synthesize a wide range of cyclic dinucleotides (CDNs) - and more, including trinucleotides (CTNs?)!

Different CDNs and CTNs activate different host sensors - for example, many that don’t activate STING instead activate RECON. Dozens of CD-NTases are shown to synthesize a wide range of products, including CTNs - which can also activate RECON.

The crystal structures, bioinformatics, and biochemistry in this paper are amazing - great examples, fantastic detail into the reactions and products! Will be very interesting to see the normal role of these enzymes / products in bacterial physiology.

Speculation: These CDN and CTN products were all formed in vitro from simple NTPs. I would be shocked if there isn’t, at some point, production of CDN/CTN with modified nucleotides - whether from modified starting products or by enzymatic modification of the final product.

Given the arms race with host sensing of cyclic nucleotides, it would almost be shocking at this point if there weren’t enzymes devoted to altering CDN/CTNs to have methylated nucleotides, pseudouridine, deoxyuridine, etc… And that these show different binding to host sensors!

Paper 2: Another #preprint from @lingyinli1 on STING @biorxivpreprint, Ergun et al - this time showing a polymer structure induced by ligand binding in human STING. #Crystallography biorxiv.org/content/early/…

For structure, nothing beats a picture (showing the polymer structure with cartoon TMs and membrane, zooming in on a salt bridge at the polymer interface - which is not required for polymerization, however):

So, STING activation depends on polymer formation and disulfide bonds formed (in the cytosol!), rather than just changes in a dimer structure, and the conformational changes involving the C-terminal tail are different than expected.

Human mutations at this polymer interface may lead to inflammatory diseases. And, different ligands (e.g. c-di-GMP vs cGAMP) engage different steps of this assembly process (dimer closing vs polymerization), explaining differences in pharmacology.

These structures and biochemical experiments, together with some solid pharmacology at the end (look at CDG acting as a competitive antagonist of cGAMP at low, but not high concentrations in Fig 5!), will really help our understanding of and ability to target STING.

Interlude before the next papers: What do all these new CDN/CTNs discovered by Whiteley et al, and their varying activity to STING, do when we combine that with new pharmacological insights from Ergun et al, about ways that different CDNs agonize / antagonize STING?

It’s interesting to note that many bacteria can express multiple CD-NTases (check out Supplemental Figure 2 from Whiteley et al) - so the same bacterium could be making multiple ligands of STING or other nucleic acid sensors.

Look at Extended Figure 6 - CD-NTases 1 and 56, both from E coli, are pretty active - and give different products (!). 5 and 26 are both from Desulfotomaculum, and 5 is definitely active and 26 may be at high pH?

So, given what we know about CDG-cGAMP competitive activity from the Ergun preprint, it’s entirely within reason that a bacterium could be making a CDN specifically to block STING sensing another CDN it’s making, by carefully controlling ratios!

Let’s just take a second to appreciate how complicated things could get if this is true. And, since bacteria seem to find a way to break every possible immune response, I’m going to go ahead and expect that this scenario probably happens. But that's just a hypothesis!

Paper 3: Bystander activation of cGAS/STING in uninfected cells, by transfer of pathogen DNA in extracellular vesicles (rather than cGAMP transfer), from the Paludan group in @NatureMicrobiol. nature.com/articles/s4156…

Conditioned supernatant from cells infected by Listeria (or other intracellular pathogens) stimulated IFN production in naive cells in a DNA and STING-dependent, but TLR-independent, manner. Blocking exosome biogenesis blocked this activation.

They work out the cellular pathway leading to packaging of pathogen DNA into exosomes - through STING-TBK1-MVB12b in the infected cell, leading to DNA transfer. Interestingly, these exosomes / EVs stimulate apoptosis in T cells.

Now, they don’t actually show that pathogens do this “deliberately” - there’s no clear indication that an evolved Listeria effector (for example) modulates this pathway - if anything, the opposite, since DNA alone is sufficient to stimulate MVB12b phosphorylation.

So this is a clearly an endogenous response that’s part of the cGAS-STING pathway. The role in inducing T cell apoptosis is interesting, and it certainly is possible that bacteria could hijack this process to enhance that effect.

I *personally* expect that yes, some pathogens do manipulate this process - but in some cases it may even go the other way, where they shut it down. Seems like balance of innate vs T cell control of particular infection will determine that…

I ran out of tweets on this thread, but I continued looking at the localization of cGAS in the nucleus -

https://twitter.com/connor_rosen/status/1100989925808394240

.
Although now I'm going to have to revisit all of that with a super interesting looking paper from @jkagan1 in @CellCellPress on membrane-bound cGAS!