Medicinal chemists spend the vast majority of their time targeting proteins. Enzyme active sites, receptors, allosteric sites, interfacial sites – it’s one protein after another, to the point that you can mentally assume that your compounds are going to be hitting the familiar landscape of backbone amide bonds, pi-interacting tryptophan side chains, hydrogen-bonding aspartates and the rest of the crew. But (as I and others mention every so often) there are a lot of other classes of targets out there and one of them, RNA in all its forms, has been moving into the spotlight recently.
Here’s an example of that, from Matt Disney’s lab at Scripps-Florida. They’ve been dealing (as have the others in this field) with some challenges. For one thing, there are an awful lot of different RNAs out there, and when you’ve gotten through the ones that code for proteins you’ve only started. MicroRNAs (miRNAs), long noncoding RNAs (lncRNAs), circular RNAs (circRNAs), double-stranded RNAs (dsRNAs) such as silencing RNAs (siRNAs), and a whole list of others. . .one loses track after a while of all the different varieties of RNA swimming around in a living cell, and don’t think for a minute that we understand what all those things are doing, either. (If I had to pick something about cell biology that is totally different from the way we imagined it to be when I started working in the drug industry, RNA in general would be a solid choice).
Another difficulty is that some (many?) of these RNAs are not very structured, and we already know about the joys of targeting disordered proteins. Add to that the fundamental problem that there are only four nucleic acids making up all of these species, and you start to worry not only about finding small molecules that bind, but whether any of those small molecules are just going to bind one thing at a time. There have been numerous reports of small molecules that bind to or affect levels of miRNA species, but building on those results has not been so easy. Disney’s group, though, has been able to target sites in RNA sequences that are recognized by the Drosha enzyme, a ribonuclease III enzyme that’s involved in processing microRNA precursors (it’s just upstream of the Dicer enzyme in the sequence). They have identified a molecule that hits two of those sites with pretty much the same affinity – the ones for miR-515 and miR-885. A key discovery in this new paper is that there’s an adjacent binding site in the 515 species, and that a dimeric form of the ligand binds selectively to the miR-515 over the 885 ones. It took searching a library of different spacer groups, as you’d imagine, but there’s one with four propylamino groups that works well – the original compounds binds to both miR species with micromolar affinity, but the dimer binds just to the 515 one at about 60 nanomolar.
And that lets you into some more selective biology. None of the compounds in this series bind to DNA, which is good, and selectivity against other RNA species looked good as well. Cellular activity is the real test, though – as the paper notes there are demonstrated artifacts when working with RNA species in vitro, due both to folding problems of the RNA species and their interactions with various proteins in lysate, etc. that they might not see at all under physiological conditions. In MCF-7 cells, the dimer compound was indeed much more selective for lowering the amount of miR-515, and at least tenfold more potent besides. An RNAseq experiment to look at general effects showed that miR-515 was definitely the most strongly affected species, as it should be. Looking at a number of other RNA species that might be predicted to also show binding showed most of them not to be a problem at all – in fact, there are a number of other (putatively unrelated) RNAs that show equal or greater effects, which at least to my untrained eye suggests that the algorithms used to pick RNA structural similarities may still need some work.
The key thing, though, is what miR-515 itself does. It’s been shown to lower expression of sphingosine kinase 1 (SK-1) and thus also lower the amount of sphingosine-1-phosphate (S1P), which that enzyme would be producing. That species is well-known to be involved in important inflammatory signaling cascades, in cell proliferation and migration, and other processes. The dimer RNA-targeting molecule induces just the effects that you’d expect: increased levels of SK-1 and S1P. The cells show increased migration, and downstream proteins (notably HER2) whose expression is known to be affected by S1P levels were increased. All of these effects could be shut down by adding an SK-1 inhibitor compound, by siRNA targeting SK-1 itself, or by increased expression of miR-515 to cancel things out. MCF-7 cells don’t normally have much HER2 protein, but treatment with the RNA-targeting ligand rendered them sensitive to the Herceptin antibody, in a dose-dependent manner.
So signifigant cellular effects can indeed be brought on by selectively targeting microRNA species – this paper is pretty convincing in that regard. What we don’t know yet is how general the idea is – will that second-binding-site effect be necessary to get the required selectivity and potency, or are there other tricks that can be tried? In other words, how many miR species are targetable in general? The good news and the bad news is that nobody knows yet: these are blanks spaces waiting to be filled in. And there are, as we speak, people moving in to do just that.