Who I'm Reading Today

https://www.science.org/content/blog-post/condensate-dissolver

I’ve written several times over the last few years about biomolecular condensates, also known as “membraneless organelles” and (confusingly) by several other names. Whatever you call them, these are phase-separated droplets that form reversibly under various cellular conditions, and they often consist of mixtures of proteins and RNA species. There are particular protein sequences and charge distributions that are more common among condensate-formers, and disordered proteins are over-represented as well, but there’s a lot of variety in composition, lifetime, and localization.

I was invited to a meeting on these topics back in 2018 and found the field very interesting. It seemed obvious that these things were performing important cellular functions and that they had been largely overlooked. Well, except by microscopists, who had given them names like “speckles”, “granules”, “bodies” and so on. It was quite a while before their common features were recognized. Not long after that meeting several condensate-focused startups launched, not all of which are still with us. I spent a good amount of time digesting the research in the field, as much as it was actually digestible, and trying to come up with a good actionable application to drug discovery.

But I was unable to. No matter how I came at it, the whole field seemed to still be in too early a state for these opportunities. Indeed, my impression of some of the startups was that they set about trying to bring some order to things to see what they had and what might present the best opportunities, and I suspect that process is still continuing. A few things did stand out: everyone was agreed that proteins like TDP43 and FUS formed stress-granule condensates, and that these seemed to somehow go haywire in some neurodegenerative disorders. The simplest way to look at it was that the condensates didn’t seem to dissolve on cue once the cellular stress was removed, but rather hung around, aging from liquid droplets to something more like gelatinous blobs and eventually to outright solid deposits that were associated with pathological states in neurons. That might be too simple a way to look at it - nothing in neurodegeneration is ever as simple as it looks, from what I’ve seen - but it’s not a bad place to start. And a lot of people have indeed been starting right there.

This recent paper makes the case that interfering with that “condensate aging” process might be beneficial, as hoped, and the authors searched for small molecules that could induce this phenotypic outcome. There had been some condensate-dissolving molecules identified over the years, but they tended to be like the prototype of that group, 1,6-hexanediol. That one really does seem to dissolve condensates, but probably in the same way that it might make a useful additive to a spray for cleaning shower doors: in some ways it’s just a solvent. And as you’d figure, it has a lot of other effects on cells, making it a pretty poor tool for assays and outright hopeless as a potential mode of treatment.

This new paper settled on lipoamide. That makes some chemical sense as a hit - the formation and dissolution of stress granules seems to involve redox changes in some key proteins, and the disulfide group in lipoamide might be expected to get into the middle of that chemistry. The paper shows that various cell lines expressing fluorescently tagged versions of stress granule proteins (such as FUS and TDP43) are sensitive to lipoamide treatment, acutely or with pretreatment before the induced formation of the granules as a preventative. Interestingly, and array of other known condensates in both the cytoplasm and nucleus were unaffected. Even the source of stress granule formation mattered: granule formation on exposure to arsenate, oxidative stress, or osmotic shock was inhibited, while formation after heat treatment or inhibition of glycolosis went on apparently as before. There are stress granules and there are stress granules, it seems.

Nitrogen-isotope-labeled versions of lipoamide or a derivative of it with a photoreactive diazirine group both showed that it partitions well into cells, and that it also gets concentrated in the stress granules (as well as other organelles). SAR exploration around the structure showed that the lipoamide stereochemistry didn’t seem to matter, indicating that it’s probably not a specific protein-binding event that’s in play. And the activity around the amide group was rather flat - some a bit better, some a bit worse. In fact, the entire carboxamide could be deleted without completely losing activity. But the disulfide (the dithiolane ring) was crucial. Six- and seven-membered disulfides were actually inactive (the size of such rings is known to affect the redox potential).

Proteomic work showed that disordered protein regions with lots of arginine and tyrosine residues interacted with lipoamide. FUS itself does not, but two other proteins that are known to localize to stress granules do, SFPQ and SRSF1. Their own sequence oddities are found in their names, which are acronyms for “splicing factor proline- and glutamine-rich” and “serine-arginine rich splicing factor 1”. Knocking out the former impaired the ability of lipoamide to keep stress granules from forming, and knocking out the latter stops it from working at all.

After this came the real test in cells. Lipoamide does actually seem to rescue the functions of FUS and TDP43, presumably by allowing them to be either released from stress granules or by preventing their formation outright, and it actually seems to alleviate the ALS phenotypes in relevant models with both human- and animal-derived cells expressing mutant forms FUS and TDP43. On top of that, it also showed beneficial effects in small-animal model systems in C. elegans and Drosophila. That’s really interesting, but it also raises a lot of questions about the formation of those stress granules and their role in normal cells. The granules in those mutant cells aren’t really associated with oxidative stress, for example, and you also have to wonder about the consequences for normal cells if they are indeed unable to form stress granules after treatment with lipoamide or something like it. They’re presumably there for a reason! And more work will be needed to show a solid link between stress granule dissolution and these phenotypic improvements. But now it looks like there are some good tools with which to get started on that, which is a real improvement.

https://www.science.org/content/blog-post/condensate-dissolver