How can we turn a disease-causing gene off without changing its DNA sequence?
Some people have a mutation in the PCSK9 gene, which can be the best thing possible for their hearts. This helps maintain a low level of cholesterol and lowers the risk of heart disease for life.
So, could we give everyone this gene mutation? Technically, CRISPR can edit the PCSK9 gene, but it can also cause unpredictable mutations. But what if we could “turn off” the gene and leave the DNA exactly as it is?
1.0 What does “turning off without changing the gene” mean?
Almost every cell in your body has identical DNA, so why is a liver cell completely different from a neuron? The answer lies in how your genes are read. When a gene is “on,” it gets transcribed and produces a protein. A gene that is “off” is still included in the DNA, but isn’t transcribed.
This is where epigenetics comes in. It is an additional layer of chemical tags on top of the DNA that tells the cell which genes to use and modifies the readability of the gene. [1]
So how do we turn a specific gene off?
2.0 How does a cell switch a gene off naturally?
A gene is off when the cell never starts reading it. One of the mechanisms your cells use to do this is by making the gene inaccessible to the machinery of reading. Your DNA is around 2 meters long, but it has to fit inside a nucleus a few micrometers wide, so it is tightly coiled and compressed. However, the cell can tighten or loosen certain genes. To do that, it has three main ways: chemical tags stuck directly onto DNA, chemical tags stuck onto the proteins DNA is packed around, and RNA molecules that steer the first two.
2.1 DNA methylation: tags on the DNA itself
The simplest switch is a tiny chemical group called a methyl group. This is added to the DNA when cytosine is followed by guanine. Such pairs are known as CpG.
Many genes have a dense cluster of CpGs around their promoter. The promoter is the part of DNA near the beginning of a gene that tells the cell where it should start reading. Proteins called transcription factors land there and recruit RNA polymerase, the enzyme that copies the gene into RNA. This particular CpG is called a CpG island. When the gene is active, the island is usually free of methyl groups. When it gets methylated, the gene usually goes quiet. Methylation can either physically prevent transcription factors from binding to the promoter, and thus the RNA polymerase never gets recruited. Or it can also attract proteins that recognize methylated DNA, and those proteins bring in enzymes that compact the chromatin around the gene. [2]
But the most important part is how the cell keeps the methyl group. Enzymes called DNMT3A and DNMT3B put new methyl groups on DNA. Then, when the cell copies its DNA before dividing, the new strand comes out unmethylated, while the old strand still has its tags. Another enzyme, DNMT1, finds this mismatch and adds the corresponding methyl groups to the new strand. So the pattern gets copied along with the DNA, and that’s why a methylated gene in one cell stays methylated in its daughter cells. [3]
It is also possible to reverse it. A group of enzymes called TET proteins removes the methyl groups, which is how a cell can switch a gene back on. [4]
2.2 Histone Modification
In order to fit into the nucleus, DNA is wound around clusters of proteins called histones, a bit like thread around spools. Spools along with the thread wrapped around it is called a nucleosome (about 147 DNA letters wrapped around eight histone proteins), and the entire structure is called chromatin.
When nucleosomes are packed loosely, the DNA is open, and transcription factors and RNA polymerase can get in so the gene can be on. When it's packed tighter, the gene is switched off.
Each histone has a loose “tail” sticking out from the nucleosome, and those tails can have chemical tags that either loosen or tighten the packaging.
- Acetylation: This is adding an acetyl group. The DNA molecule is negatively charged, and histone tails are positively charged, so they stick together. Acetylation cancels out some of that positive charge, so the DNA loosens its grip, and the gene becomes easier to read. Enzymes called HATs add acetyl groups, and enzymes called HDACs remove them. [5]
- Methylation: This one is adding a methyl group. It is more complicated since the effect depends on which histone and which position gets tagged. Scientists name these marks with a short code. For example, H3K4me3 means “histone 3, lysine at position 4, three methyl groups.” That mark sits at active promoters. H3K9me3 and H3K27me3 are the opposite: they’re linked to tightly packed, silenced regions. The H3K27me3 mark is put down by a protein complex called PRC2. [6]
These tags work like signals that other proteins read. For example, a protein called HP1 binds specifically to H3K9me3 and helps compact the chromatin around it.
DNA methylation and histone marks also reinforce each other. H3K9me3 tends to recruit DNA methylation, and DNA methylation tends to bring in enzymes that maintain the histone marks. So once a region is silenced, the two layers lock it in together.
3.0 Why PCSK9?
At the start of this article, I mentioned people whose broken PCSK9 gene keeps their cholesterol low for life. To see why that matters, you need to know what PCSK9 actually does inside the body.
3.1 What does PCSK9 actually do?
Your liver is your main cholesterol-clearing organ. Liver cells have receptors on their surface called LDL receptors, which are able to remove LDL cholesterol particles from the blood. When the receptor grabs the particle, both the particle and the receptor itself are taken into an organelle called an endosome. Cholesterol is processed there, while the receptor comes to the membrane to repeat the whole cycle.
PCSK9 breaks that loop. It is a protein the liver releases into the blood, and it binds to the LDL receptors. Receptors that are bound to PCSK9 do not recycle but get internalized and destroyed.
So the logic is simple. More PCSK9 means fewer receptors, so LDL piles up in the blood. Less PCSK9 means more receptors survive, so LDL goes down. [7]
3.2 What happens when PCSK9 is too active or broken?
You can see this play out in real people. A study conducted in 2003 has revealed that some specific mutations in the PCSK9 found that these mutations were behind a form of very high cholesterol that runs in families. These mutations make PCSK9 more active, so more receptors get destroyed, and LDL climbs, often starting from an early age in life.
Then researchers found the opposite. Some people carry mutations that stop PCSK9 from working properly. With fewer receptors destroyed, they end up with lower LDL for life, and much lower rates of heart disease.
What makes the PCSK9 gene a pure target is that there is evidence beyond cells or mice. Human beings provide evidence regarding what can happen when one switches off the gene throughout one’s lifetime, and it seems nothing wrong is happening. Nobody can guarantee that a drug will behave like that, but this is definitely much better than what most drugs start with.[8]
4.0 How do we turn PCSK9 down today, and what’s the catch?
Doctors can lower PCSK9 using drugs, but how long do they actually last?
4.1 Blocking the protein: Antibodies
Evolocumab and alirocumab were the first PCSK9 drugs, approved in 2015. They’re antibodies, which are proteins built to stick to one specific target. Since the antibody binds to the PCSK9 protein, it is not able to attach to the LDL receptor. [9]
They are effective; however, it is an injection that has to be given every couple of weeks. Once it is stopped, the PCSK9 activity will increase again.
4.2 Blocking the message: inclisiran
A newer drug, inclisiran, goes one step further. Instead of grabbing the PCSK9 protein, this drug blocks the creation in the first place. It’s a small piece of RNA (an siRNA) that finds the message the cell copied from the PCSK9 gene and destroys it, so the protein never gets built.
It lasts longer with injections about twice a year after the first couple. But the PCSK9 gene remains the same, so the liver continues to copy it.[10]
4.3 Why not just use epigenetic drugs?
Epigenetic drug therapies are already available. Drugs that block DNMT, preventing DNA methylation, and drugs that prevent histone deacetylation are available against some types of blood cancers.
However, the drugs do not target any specific gene. The drugs act by altering the tags throughout the genome, affecting the expression of thousands of genes simultaneously, including PCSK9. This is alright for cancer treatments but not for something people would be taking for decades to lower their cholesterol. [11]
5.0 Can we keep PCSK9 silent?
Regular CRISPR editors, including the base editors being tested on PCSK9, work by changing the DNA. Epigenetic editors borrow their aiming system and leave the DNA alone. Here’s how that works.
5.1 Step one: a protein that finds one gene
Get Ashimabassi’s stories in your inbox
Join Medium for free to get updates from this writer.
Normal CRISPR has two parts. A short guide RNA matches the exact stretch of DNA you want, and a protein called Cas9 follows it there and cuts. Scientists figured out how to disable the cutting part with a few mutations, creating “dead” Cas9 (dCas9). It still goes to the right spot, but it doesn’t cut anything.
That leaves a tool that can find one gene out of about 20,000 and sit on it. On its own, it isn’t very useful. When researchers aimed a dCas9 that only blocks a gene (a method called CRISPRi) at PCSK9, the drop in PCSK9 was fast but didn’t last. Sitting on the gene turns it down, but once the editor is gone, the gene wakes back up. [12]
5.2 Step two: attach something that writes the marks
So, the next idea is to bolt a second protein onto the finder, one that adds the same marks from section 2. Two are used most:
- KRAB is a repressor that helps install the histone marks that tighten chromatin
- DNMT3A (often paired with a helper called DNMT3L) adds methyl groups to the DNA itself
Putting these together matters because, as we saw, the two kinds of marks reinforce each other. Researchers have used combinations of DNMTs and KRAB for lasting gene silencing, including on PCSK9 in mice. A version called CRISPRoff also kept a gene silenced through cell divisions after the editor was only made temporarily.
The editor shows up, writes the marks, and leaves. The cell’s own copying machinery is supposed to do the rest. [13]
5.3 Getting it into the liver
Since PCSK9 is made in the liver, that’s where the editor needs to go. The standard method is lipid nanoparticles (LNPs), tiny fat bubbles that tend to end up in the liver. They carry the editor as mRNA instructions, so the liver cells build the editor themselves, use it, and then break it down. That’s why this is called “hit-and-run” editing: the tool is only there briefly, but the change is meant to stay.
5.4 The study that tested it
The key study is a 2024 paper in Nature by Cappelluti and colleagues, from the lab of Angelo Lombardo. They built several editors, packaged their mRNAs in lipid nanoparticles, and gave mice a single dose, which almost halved the amount of PCSK9 in the blood for nearly a year. [14]
Two details make it more convincing. First, it was tested against the strongest comparison: the design reduced PCSK9 about as well as conventional gene editing, but without breaking the DNA. Second, they tested whether the effect survives cell division, which is the big question from section 2. They forced the liver to regenerate, so the cells had to divide, and the silencing and its marks were still there afterward. [15]
It has also been followed up. A later study in Nature Medicine tested an optimized editor in mice carrying the human PCSK9 gene and reported a reduction of over 98% that lasted at least a year. The researchers also switched PCSK9 back on in some mice using an editor that removes methylation, which shows the change is reversible. And a study in macaque monkeys, using a different type of finder protein, reported more than 90% silencing of PCSK9 in the liver lasting 343 days after a single LNP dose, with minimal off-target effects.
6.0 Where is this in humans?
So far, everything in section 5 was done in mice and monkeys. The human question has two halves: what the DNA-changing versions have done, and where the epigenetic version is.
6.1 The comparison point: editing the DNA
The PCSK9 base editor is the version people have actually tested. A phase I trial started dosing in July 2022 in New Zealand and the UK, and participants showed lower PCSK9 protein and dose-dependent drops in LDL within a month. A Chinese company, AccurEdit, started a similar trial in 2024 and saw LDL fall by about 50% in most cases, with no severe adverse reactions. [16]
6.2 The epigenetic version
It’s much earlier. In May 2026, Scribe Therapeutics got clearance from Australia’s regulator to start a phase I trial of STX-1150 in adults with high LDL and increased heart disease risk. It uses a version of Cas that can’t cut DNA, fused to proteins that add histone and DNA methylation marks at the PCSK9 promoter in liver cells. In animal tests, a single dose in mice with the human PCSK9 gene cut circulating PCSK9 by more than 95%, and the drop lasted over 310 days. [17]
However, the trial’s main goal is safety and tolerability, and LDL lowering is a secondary measure to see if the monkey results carry over to people. So the first results won’t tell us if it lasts. And Scribe isn’t the only company. Chroma is also going after PCSK9, and it began dosing people in January 2026 with an epigenetic silencer for hepatitis B, so this type of treatment is now reaching people for other diseases. I couldn’t find any published human data on epigenetic editing of PCSK9 itself. [18]
7.0 What’s still unsolved?
It was shown to be effective in animals, but in humans, it is not known yet.
Will it last in people? The longest results are about a year in mice, and slightly less in monkeys. A patient suffering from high cholesterol would require treatment lasting for decades. These marks are copied during cell division, which is why the silencing survived liver regeneration in mice, but nobody knows if that holds for 30 years in a human liver.
Does it stay on target? Although the target is PCSK9, the human genome is much larger, and there might be proteins that are able to add marks at incorrect locations in the genome, leading to silencing of the incorrect gene. The editor has proven minimal off-target effects in one monkey experiment, but there is only one small-scale experiment so far. The first clinical trials are aimed at testing the safety of the technology.
Does it work outside the liver? PCSK9 happens to be an easy target partly because lipid nanoparticles are mostly delivered to the liver, where PCSK9 is made. Other organs don’t have an efficient mechanism for delivery, so cholesterol may be one of the easiest cases for epigenetic editing.
8.0 So, can we turn off a disease gene without changing its DNA?
In animals, yes. But in people, we don’t know yet.
Nobody knows if the silencing will last for decades, stay on target, or work outside the liver. First clinical trials using the PCSK9 approach have just been approved to begin.
For myself, I approached the problem being sure that the difficult thing would be to discover how to switch a gene off. But turned out that cells do that all the time, and the difficult part is not to let them switch the gene back on.
References
[1] https://www.cdc.gov/genomics-and-health/about/index.html
[2]https://pmc.ncbi.nlm.nih.gov/articles/PMC3093116/
[3]https://pmc.ncbi.nlm.nih.gov/articles/PMC4063825/
[4]https://www.nature.com/articles/s41392-023-01537-x
[5]https://www.news-medical.net/health/What-is-Acetylation.aspx
[6]https://www.healthline.com/health/methylation
[7]https://www.webmd.com/cholesterol-management/pcsk9-inhibitors-treatment
[8]https://australianprescriber.tg.org.au/articles/pcsk9-inhibitors-mechanisms-of-action.html
[9]https://pubmed.ncbi.nlm.nih.gov/40511660/
[11] https://www.sciencedirect.com/topics/medicine-and-dentistry/dna-methyltransferase-inhibitor
[12] https://www.nature.com/articles/s41591-025-03508-x
[13] https://pmc.ncbi.nlm.nih.gov/articles/PMC12052198/
[14] https://pmc.ncbi.nlm.nih.gov/articles/PMC10937395/
[15]https://pmc.ncbi.nlm.nih.gov/articles/PMC10937395/
[16] https://innovativegenomics.org/news/crispr-clinical-trials-2026/
[18]https://crisprmedicinenews.com/news/first-patient-dosed-with-crispr-hbv-silencer/
