I remember being at a chemistry meeting in New Jersey back in around 1990 or 1991, where a speaker mentioned in passing that most of the people in the room would probably soon be offered a chance to move to California and work for some small company trying to develop antisense drugs. There was a wave of laughter, especially from people who’d already had such headhunter calls, and it really was a bit like that for a while. There was a great wave of enthusiasm for the idea of controlling gene expression this way – antisense oligonucleotides seemed to open up a whole new mode for therapy, promising near-magical powers. The exact same enthusiasm drove investment in the various RNA-interference companies years later, and is currently behind the enthusiasm for mRNA approaches and for CRISPR.
Although some of these trends can get irrational, the motivations behind them aren’t. You can see the common themes: control of gene expression, and thus control of protein expression. The obvious application is for the many diseases, many of them rare, that can be traced back to a defective gene sequence that in turn cranks out defective proteins that can’t function well enough, aggregate in cells over time, and so on. Start with sickle cell anemia, which thanks to Linus Pauling and co-workers is the original molecular genetic disease, and work your way on up through cystic fibrosis, Huntington’s, and many others. But there are plenty of normal proteins whose expression and cellular abundance we’d like to be able to alter at will, too. And for that, see the enthusiasm over targeted protein degradation, in oncology and other therapeutic areas.
But getting access to those control panels is not easy. Regulation of gene expression and of protein levels is accomplished in the cell by a completely bewildering mass of overlapping, interlocking, multilayered mechanisms, as well they might be. The work of the Department of Redundancy Department is very much in evidence as you look closer, too, which makes it hard to tweak just one thing without something else making up the difference. As it is with higher-level stuff like feeding behavior and reproduction, all the living creatures in which these pathways could be easily hijacked or broken are long gone. Our ancestors were the other guys.
Another big challenge, if you choose to operate at the DNA or RNA level, is that there are very, very few good small-molecule openings for therapy. That’s what led, by the way, to another burst of enthusiasm in the late 1990s and early 2000s for nuclear receptors, and a bit later on for epigenetic enzyme pathways. Both of those promised a small-molecule route into messing around with gene expression, but both of them, despite a few successes, proved too wooly and multifaceted for easy manipulation. So if you really want to step in and gum up the expression of this particular gene right over here and not much else, you’re forced into playing as the away team on the oligonucleotides’ home field.
That was the idea behind antisense. Come in with a complementary oligo strip to the mRNA that corresponds to a gene of interest, and it would stick to it and block up the works. What could be easier? But plain old oligonucleotides make terrible drug substances. The body does not look with favor on big stretches of unfamiliar DNA and RNA wandering around loose. There are lots of enzymes whose job is to rip such things to shreds on contact, and lots of innate-immune tripwires that are constantly looking for signs of viral infection (which is a main reason a living organism would suddenly experience such oligonucleotide events). What’s more, you obviously have to get into cells for these things to work their magic, and cell membranes are not geared to let them in very easily, either.
Thus the huge amount of work that’s gone into packaging of RNA and DNA molecules over the years – work that was, very fortunately, at a mature enough stage by now to allow the quick deployment of the mRNA vaccines. The famous lipid nanoparticles are a culmination of attempts to get the mRNA constructs to survive long enough in vivo and to actually be able to enter cells in decent numbers. The few siRNA therapies that are out there use very similar packaging, and even then are often targeted at diseases involving the liver. That’s because when you inject these things into the bloodstream (rather than giving them as an intramuscular dose like the vaccines, where you just want to hit the local tissue and a few lymph nodes), the lipid nanoparticles and their RNA payloads will invariably have to go through the liver (since our entire circulatory system loops through there) and there they will come to a screeching halt, never to emerge on the other side. The RNA and antisense ideas that aren’t targeted at the liver are generally targeted to the eye (retinal degeneration and the like), because the eye is a protected compartment where such exotica can linger without being degraded so quickly. Pharmacokinetics constrain us, big-time. Did you have some therapeutic oligonucleotide ideas that don’t involve the liver, the eye, or the lymphatic system? Better raise some more money, because you’ll need it. If you would like (for example) to target an antisense drug at neurons in the brain, then to re-use a favorite line of mine, you’re going to have to break out the Black and Decker, because for now that’s the only way you’re getting that stuff in there.
That is more or less how a potential antisense therapy for Huntington’s has been dosed, via intrathecal injection (right into the spinal column and thus into the cerebrospinal fluid space). Ionis, basically the sole survivor of the early 1990s antisense days, has been working on this for some years now as HTTRx, now known as Tominersen after Genentech stepped in with a deal to develop it. The drug is a complementary stretch of modified oligos that binds to a 20-residue stretch of mRNA produced by transcription of the HTT gene, and the protein produced by it, huntingtin, is absolutely the problem with Huntington’s disease. Wild-type huntingtin has a variable number of glutamines repeating down at the N-terminal region. If you have up to 26 of them, you’re completely normal. Up to 35 of them? Less common, but you’re still going to be OK. 36 to 40 glutamines means that you may or may not develop symptoms – that’s the grey zone. But more than 40 (the repeats can get up to over 200 in some cases) means that you will develop Huntington’s disease, and I am aware of no exceptions that have been found. Now, we’re not completely sure of all of the huntingtin protein’s functions, but losing it completely is embryonic lethal, and it has a wide profile of interacting proteins that mark it down as a big player in neuronal development and function. There’s a whole list of such polyglutamine repeat disorders, and they are all very bad news. The jury is still out on which is causing more damage – the repeated oligonucleotide CAG codons or the repeated glutamine residues in the protein, but the repeats in general are where you want to be if you want to fix Huntington’s. And it would thus seem to be a perfect candidate for an antisense therapy.
In 2018 and 2019, early stage human trials from the 2015-17 period were reported. The drug was administered in four injections, four weeks apart, and then patients were monitored for several months afterwards. CSF samples were taken to both quantify levels of the antisense oligo, and to measure reductions in mutant HTT. The therapy appeared to be well tolerated, and dose-dependent reductions in mutant HTT were observed, all of which sounded promising.
Well, that makes the news this week even harder to take. The data monitoring committee for the ongoing Phase III trial reviewed the results so far, and decided to stop things based on risk-to-benefit ratio. The press release goes on to say that no new safety signals were picked up, so that makes you think that there was little or no clinical benefit showing up, or at least not enough to justify giving people periodic injections into their spines. Intrathecal dosing as a therapeutic mode has always been problematic – you see it used as a one-shot in (for example) anesthesiology, but doing it all the time, even every few weeks, is another matter entirely. The company will continue to follow the patients who have already been dosed, but the drug itself is now abandoned.
I really am curious to hear more about what happened. You would have thought that reducing mutant HTT would have helped, but if this trial followed the Phase I/II results, then they reduced it (at least to some degree) without seeing any real benefit. Do you need to knock it down even more than Tominersen could achieve? Or are we learning something about Huntington’s that will complicate its etiology even more? This might end up adding more evidence to the line of argument that I blogged about before (“jury still out” link above), which was this paper’s focus. It presented evidence that it’s the snarled DNA with all those CAG repeats that is the fundamental problem, and in that case everything is downstream, including using antisense on the mRNA. If that’s the case, is it going to be DNA editing or bust? That’s going to be a significant challenge to do in the CNS. We shall see!