RNA Interference, the Long Way Round

There is a thing in biology called the central dogma of molecular biology. DNA becomes RNA, and RNA becomes protein. That is the whole statement. And that one sentence is basically the operating instructions for every living thing, including you. It is the reason you have the skin color you have, the height you have, and it is the reason every single cell in your body can do whatever it is that specific cell does, whether that is a muscle cell contracting or a red blood cell carrying oxygen around your body.

That only makes sense once you back up one step further, to something more basic underneath it, which is cell theory: there is no life apart from the life of cells. And a cell is not a simple thing either. It is a complex biological construct, built out of complex macromolecules, running complex chemical reactions inside it more or less constantly, what biologists call metabolism. But however complicated a given cell is, and whatever job it happens to be doing for your body, the way it actually does that job comes back to that same pipeline: DNA to RNA to protein. DNA holds the instructions, RNA is a working copy of one specific instruction, and protein is the thing that goes and actually does something, inside a cell, inside you.

I am starting here on purpose, because none of what I am about to describe, RNA interference, sickle cell disease, any of it, makes sense without this sitting underneath it first. So, here is how I actually got into thinking about any of this.

How I actually got here

I have been trying to learn some bioinformatics for the past couple of months, working through problems in R, the kind of thing where you are measuring gene expression, looking at how much of a given RNA is present in a sample and how that changes between conditions. At some point in the course I was following, there was a problem where you measure something called Ct, the cycle threshold, from a technique called qPCR, or RT-qPCR really, since what you are actually measuring is RNA, and RNA gets turned into a matching bit of DNA first before it can be amplified and counted. Either way, the basic idea is measuring how much of a specific RNA was present in a sample to begin with, by counting how many amplification cycles it takes before you can detect it. Fewer cycles needed, more of it was there at the start. More cycles needed, less of it was there.

I could follow the analysis part of it fine. Work out the delta Ct, compare a treated sample against a control, get your fold change out the other end, that part is really just maths once you have the formula. But something stood out to me while I was going through it, which is that I did not actually understand what was happening at the bench to produce those numbers in the first place. We were measuring a change in Ct, but I could not have told you what was actually happening biologically to cause that change. And that started to bother me, because at that point this was not really a coding problem or a statistics problem anymore, it had become an analytical problem, and you cannot properly analyze something you do not understand.

The specific gap was RNA interference. The exercise was built around an experiment that used it: silence a particular gene, not the gene itself, but the message it produces, meaning stop that gene's RNA from actually becoming protein, and then see whether you could measure a downstream effect because of it. In the version I was working through, we were looking at whether silencing one specific gene changed the concentration of fetal hemoglobin. So before those Ct numbers were going to mean anything to me at all, I needed to understand what RNA interference actually was.

I do not have much hands on wet lab experience, and qPCR was not something I had actually run into before, so this was not something I already had sitting in my head somewhere. I went and did some digging. Read around online, asked a couple of language models to walk me through it with examples, and one example in particular is the one that actually made it click, and it has stuck with me since. It was a hypothetical: a made up scenario where you use RNA interference to treat sickle cell disorder.

The hypothetical

Sickle cell disorder is a disease of red blood cells. Specifically, the red blood cells end up shaped like a sickle, like a crescent, instead of their normal round disc shape, and that happens because of a particular protein. Which, going back to that first bit about cells, tracks: there is no life apart from the life of cells, and almost everything a cell does, it does through proteins. I like to think of proteins as verbs, the things that actually go and perform an action. And the specific protein in a red blood cell whose job is to carry oxygen around your body is hemoglobin.

Hemoglobin is not just one fixed thing, though. There are different variants of it. The one you are running on right now, from around six months of age onward, is called hemoglobin A, or HbA. But before that, before you were even born, you were running on a different variant entirely, called hemoglobin F, or HbF, fetal hemoglobin.

Both of them share the same basic shape. Hemoglobin is a tetramer, meaning it is built out of four separate protein chains stuck together. In HbA, that is two chains called alpha and two called beta. In HbF, it is the same two alpha chains, but instead of beta you have two chains called gamma. Each of those chains comes from its own gene, and remember, DNA to RNA to protein, so somewhere in your genome there is a gene for the alpha chain, a separate one for the beta chain, and a separate one again for the gamma chain, and depending on which combination of those is actively being read at a given point in your life, you end up with a different version of hemoglobin.

Here is the part that matters for sickle cell disorder specifically. The mutation behind it lives entirely inside the gene for the beta chain. Something like a single letter changed, in the part of your DNA that codes for that chain, is enough to swap one amino acid for another right on the surface of the protein, and that is enough to make copies of it start sticking to each other and clumping into long stiff fibers once the molecule gives up its oxygen. That clumping is what physically bends the red blood cell into that sickle shape.

Notice what that means, though. The mutation only touches the beta chain gene. It says nothing at all about the gamma chain. Which means hemoglobin F, the fetal version, built from alpha and gamma chains instead, is completely unaffected by it.

Where RNA interference actually comes in

So here is the idea behind the hypothetical. If the sickle mutation only affects the beta chain, and hemoglobin F does not use a beta chain at all, then one way to treat sickle cell disorder, at least in theory, is to get the body making more hemoglobin F again, instead of switching almost entirely over to hemoglobin A the way it normally does after about six months. More HbF in the mix means less of the sickle version of HbA is doing the clumping, and HbF itself does not take part in that clumping at all, so it actively gets in the way of it happening, rather than just watering it down. The red blood cells stay closer to their normal shape.

But your body already knows how to make hemoglobin F, it did it for the first six months of your life. The reason it stops is that a protein called BCL11A becomes active as part of that whole switch from making fetal blood cells to making adult ones, and it represses the gamma chain gene, meaning it works to keep that gene from being read, by binding near it and around the wider stretch of DNA that controls it. It is less a light switch and more a gradual handover, but BCL11A is the main thing driving it.

So if you could get rid of that repressor, or even just reduce how much of it there is, the gamma chain gene should come back on, and you would start making hemoglobin F again.

This is where RNA interference actually fits into the whole thing. Remember, DNA to RNA to protein. If you want less of a protein, in this case less of that repressor, BCL11A, you do not necessarily have to touch the gene itself. You can intervene one step downstream, at the RNA stage, and get rid of the message before it ever becomes protein. That is what RNA interference does. You introduce a short piece of RNA whose sequence matches the RNA you are trying to get rid of, the cell's own machinery grabs onto it, finds the matching message, and destroys it before it can be translated into protein. Less message, less protein. In this case, less BCL11A.

Put the whole hypothetical together and it reads like this: silence BCL11A using RNA interference, the gamma chain gene comes back on, you start producing hemoglobin F again, and because the sickle mutation was never in the gamma chain to begin with, that hemoglobin F is not affected by it at all.

It is not actually a permanent fix, though

Here is the part I had to sit with a little, once I understood it properly. RNA interference works on the RNA, on the message. It never actually touches the gene itself, the DNA. Which means if you stopped supplying that interfering RNA, the cell would just go back to reading the BCL11A gene normally, start making the repressor again, and the gamma chain gene would get shut back off. So on its own, this is not something you do once and are done with. It behaves more like ongoing maintenance than an actual repair.

If you wanted something more lasting, a real one time fix instead of ongoing maintenance, you would have to go one level further back and edit the DNA itself, so the change is baked into the gene rather than something you have to keep reapplying at the RNA level. That is closer to what something like CRISPR-Cas does, a one time edit meant to hold, rather than a course of treatment you keep repeating. And I do remember reading, not too long ago, about an actual approved treatment for sickle cell disorder that works more or less along those lines, editing the DNA around that same BCL11A gene directly, rather than just silencing its RNA. So the hypothetical I was working through was not purely made up, it was just using the simpler of the two tools, RNA interference, to explain the underlying idea, because it is the easier one to actually picture happening. Correct me if any of that is off, I am still fairly new to this side of it.

Back to the Ct values

Once I actually understood all of that, the exercise finally made sense. We were measuring, by qPCR, a drop in the BCL11A message after the RNA interference step, a higher Ct for BCL11A, since there was less of it around to detect, meaning it took more cycles to show up, alongside a rise in the gamma chain gene's message, a lower Ct there, more of it to detect, fewer cycles needed. Two numbers moving in opposite directions, and once you actually understand what RNA interference is, and what those two genes are doing, it stops being two numbers and turns into one story: less repressor, more fetal hemoglobin. That is what I was actually looking at the whole time, I just could not see it yet.