Biomolecular condensates have been a big topic in the cell biology literature for the last few years, and I’ve blogged about them many times. One of the things I’m fond of saying (as my colleagues can confirm) is “We probably already have small-molecule condensate modulating compounds – we call them kinase inhibitors”. That’s because kinase inhibitors, of course, keep various proteins sites from being phosphorylated, and it’s been clear for some time that phosphorylation state is one of the (several) important factors in whether a given protein forms condensates on its own, joins in on ones that have formed elsewhere, or alternatively is excluded from them. Some of the condensate formation/dissolution behavior we see in living cells may well be driven by activation (or inhibition) of kinases and phosphatases (their opposite number, enzymes that cleave phosphate groups that have already been attached).
One reason to think this is that many of the condensates we’ve been able to profile have a lot of RNA species in them – yet another role for RNA molecules that no one would have even thought about years ago, and that’s turning into an ever-longer list. These RNA species are decorated with phosphate backbones, every one, so the interior of the RNA-rich condensates must have a rather “phosphatey” feel to it already, if I can coin another adjective to go with some of the others I’ve tried to add to the language of science over the years. By “phosphatey” I mean that there are a lot of negatively charged oxygens in there, things that have to be balanced out by positive charges somewhere. Those could just be ions like sodium, potassium, calcium and so on, but they also could be positively charged residues on the proteins involved as well – and indeed, some of the the proteins that are known to be in these systems are heavy in such amino acids (such as arginine). That would give you what the polymer chemists call a “complex coacervate”, a mixture of polyvalent species where one type has excess negative charge and the other type has excess positive charge. There’s a lot of work going on around that idea – see this paper for an example (open access), but be braced for pictures of condensate droplets inside other condensate droplets, among other stuff. In an RNA-rich condensate droplet, you could picture the process of adding more negatively charged phosphates to the outside of an associated protein might be enough to make the whole thing start flying apart.
This recent paper shows that arginine-rich proteins have their solubilities regulated by phosphorylation in general, which is a clue towards condensate behavior (and likely towards what may be its pathological sequel, insoluble aggregate formation). And here’s a new paper that brings some more data to the party. The authors are using a phosphoproteomics approach combined with an assay where you spin down cell lysates to separate out the condenate droplets that are present. They use these approaches (sites of phosphorylation and general likelihood of being present in condensates) to get some general mapping going. A limitation of this approach is that you’re probably only going to get the most sturdy condensate-dwelling proteins – ones with weaker interactions may well be stripped out by the cellular lysis workup. Varying that workup to digest out RNA showed that about a third of the proteins had to have RNA present in order to remain in the condensate at all, though. Looking at these condensate-enriched phosphorylation sites, it turns out that many of them are in intrinsically disordered sections of the proteins involved, which is another clue. Disordered regions are far more common in condensate-associated proteins (and indeed, more common in nuclear proteins in general), and phosphorylating them is likely to change their conformations, solubilities, and ability to interact with a variety of positively-charged species. These sites are also enhanced in nearby aromatic residues, and that’s also suggestive. Aromatic interactions (pi-pi and cation-pi) have been implicated in protein condensation and aggregation, and these would likely also be affected by nearby phosphate groups.
The authors zoomed in on two specific proteins, HNRNPA1 (which is important in RNA splicing) and NPM1 (which is essential in the nucleolus and forms the “granular” structure seen there). HNRNPA1 had several phosphorylation sites on its C-terminal disordered region, which otherwise has a positive charge overall when unphosphorylated. This phosphorylation seems to reduce its ability to be present in nuclear condensates. Meanwhile, NPM1 has several phosphorylation sites, but five in particular seemed important, and they were also largely located in a positively charged region that is predicted to be disordered. Two of these sites, when phosphorylated, are enough to completely exclude NPM1 from the nucleolus, interestingly, and phosphorylation there also reduced its interactions with RNA.
So we’re getting a bit of focus on this process – positively charged disordered regions getting phosphorylated to change condensate behavior – but there’s sure a lot more to come. Just unpacking this behavior is going to take many years of study. You also have to wonder about other post-translational modifications (should anyone run out of stuff to do, the devil finds work for idle hand, y’know). Phosphates are the kings for adding localized charge, so they’re definitely the place to start looking, but what about the less-common sulfates? What about acylation? For affecting hydrophobicity, you have glycosylation (either glucose or N-acetylglucosamine, from from the hydrophilic direction) and things like farnesylation from the other. Ah, cellular biology. . .
