This weekend brought some really significant news in the long-running effort to use gene editing to treat human disease. As most readers will have heard, Boston Children’s Hospital and a Vertex/CRISPR effort both published papers in the NEJM addressing sickle-cell anemia and beta-thalassemia. (Update: edit to fix attribution).
These diseases have long been linked when it comes to gene therapy ideas, because both of them have defects in the hemoglobin protein as their cause. And it’s long been thought that both could be treated by getting adults to re-express the fetal hemoglobin protein – it’s on a different gene entirely, and thus does not have any of the genetic problems that affect the adult hemoglobin gene. The normal course of events is for babies to stop expressing the fetal form and switch over to “regular” hemoglobin, and it’s been worked out that a particular transcription factor called BCL11a is a key player in that transcriptional repression of the fetal hemoglobin gene. That plays right into the usual way that we tend to think about therapeutic possibilities: whether it’s enzymes, receptors, or expression of whole proteins, we have a lot more tools to mess things up and interrupt processes than we have to make them run faster or better. So the possibility of interrupting BCL11a’s function has been a tempting one for many years.
It’s hard to do by traditional means, though. (Full disclosure: I have, at different times in my career, been involved with such efforts, but none have ever come near the clinic.) Transcription factors are notoriously hard to get a handle on with small molecule therapeutics, and many unsuccessful runs have been taken at BCL11a ligands to try to interrupt its functions in one way or another. My general impression is that the protein doesn’t much care about recognizing small-molecule ligands (and it’s far from the only one in that category, for sure). You’d think that if you ran a few hundred thousand (or a few million) various molecules past any given protein that you’d find a few of them that bind to it, but that assumption is too optimistic for most transcription factors. You’re also going to have a hard row to hoe (to use an old Arkansas expression) if you try to break up their interactions with their DNA binding sites: a significant amount of capital has gone down the chute trying to get that to work, with (as far as I can tell) not much to show for it.
There’s another complication: BCL11a has a lot of other functions. Every protein has a lot of other functions, but for transcription factors, the issue can be especially fraught. If you had a small molecule that really did interfere with its activity, what would happen if you just took a stiff dose of it? Probably a number of things, including some interesting (and not necessarily welcome) surprises. There have been a number of ideas about how to get around this problem, but a problem it is.
So it’s on to biological mechanisms. The BCH team reports on using RNA interference to do the job – they get cells to express a short hairpin RNA that shuts down production of BCL11a protein, with some microRNA work to target this to the right cell lines. And the Vertex/CRISPR team, naturally, uses CRISPR itself to go in and inactivate the BCL11a gene directly. Both approaches take (and have to take) a similar pathway, which is difficult and expensive, but still the best shot at such therapies that we have. You want the fetal hemoglobin expressed in red blood cells, naturally, and red blood cells come from CD34+ stem cells in the bone marrow. Even if you haven’t thought about this, you might see where it’s going: you take a bone marrow sample, isolate these cells, and then do your genetic manipulation to them ex vivo. Once you’ve got a population of appropriately re-engineered cells ready to go, you go kill off the bone marrow in the patient and put the reworked cells back in, so they’re the only source there for red blood cells at all. A bone marrow transplant, in other words – a pretty grueling process, but definitely not as much as having some sort of blood-cell-driven cancer (where the therapy uses compatible donor cells from someone else without such a problem), or as much as having full-on sickle cell disease or tranfusion-dependent thalassemia.
You can also see how this is a perfect setup for gene therapy: there’s a defined population of cells that you need to treat, which are available in a specific tissue via a well-worked-out procedure. The problem you’re trying to correct is extremely well understood – in fact, it was the first disease ever characterized (by Linus Pauling in 1949) as purely due to a genetic defect . And the patient’s own tissue is vulnerable to chemotherapy agents that will wipe out the existing cell population, in another well-worked-out protocol, giving the newly reworked cells an open landscape to expand in. You have the chance for a clean swap on a defined target, which is quite rare. In too many other cases the problem turns out to involve a fuzzy mass of genetic factors and environmental ones, none of which by themselves account for the disease symptoms, or the tissue doesn’t allow you to isolate the defective cells easily or doesn’t allow you to clear them out for any new ones you might generate, and so on.
Both the Vertex/CRISPR and BCH techniques seem to work – and in fact, to work very well. There are now people walking around, many months after these treatments, who were severely ill but now appear to be cured. That’s not a word we get to use very often. They are producing enough fetal hemoglobin, more than enough to make their symptoms completely disappear – no attacks, no transfusions, just normal life. And so far there have been no side effects due to the altered stem cells. An earlier strategy from Bluebird (involving addition of a gene for a modified adult hemoglobin) also seems to be holding up.
These are revolutionary proofs of concept, but at the same time, they are not going to change the course of these diseases in the world – not right now, anyway. Bone marrow transfusion is of course a complex process that costs a great deal and can only be done in places with advanced medical facilities. But what we’ve established is that anything that can cause fetal hemoglobin to be expressed should indeed cure these diseases – that idea has been de-risked. As has the general idea of doing such genetic alteration in defined adult tissues (either RNA interference or CRISPR). From here, we try to make these things easier, cheaper and more general, to come up with new ways of realizing these same goals now that we know that they do what we hoped that they would. This work is already underway – new ways to target the affected cell populations rather than flat-out chemotherapy assault, new ways to deliver the genetically altered cells (or to produce them “on site” in the patients), ways to make the switchover between the two more gradual, and so on. There are lot of possible ways, and we now know where we’re going.