One Word in Three Billion
Bacteria are under attack from viruses all the time, and many bacteria fight back with a memory. A bacterium keeps short snippets of virus DNA, copied from past attacks, inside its own DNA, like a wall of wanted posters. [1]
When the same virus attacks again, the bacterium copies the matching poster and hands the copy to a pair of molecular scissors. The scissors search the virus's DNA, find the stretch that matches the poster, and cut the virus's DNA in two. [1]
In 2012, Emmanuelle Charpentier and Jennifer Doudna showed that you can hand those scissors a poster you wrote yourself. [2] The right 20 letters send the scissors to one spot among the 3 billion letters of human DNA. [3] The tool is called CRISPR, and in 2023 the first medicine built on CRISPR was approved, for sickle cell disease. [4, 5]
Make your guess. How long must a poster be to pick out one spot in 3 billion letters?
- About 5 letters
- About 20 letters
- About 1,000 letters
Show answer
B. About 20 letters
About 20 letters. Each extra letter makes a match by pure chance 4 times rarer, so 20 letters can be written about a trillion different ways: far more than the spots in your DNA. You'll check this yourself below.
Before any cutting, Cas9 needs a poster that fits one spot and nowhere else. Can you write one?
Write a guide
DNA is a long string written in four letters: A, C, G and T. [3] The scissors are a protein called Cas9, and the poster Cas9 carries is called the guide: a short string of the same four letters. Cas9 cuts the DNA wherever the DNA matches the guide, letter for letter. [1]
Your goal: tap a letter near the blue dot to start your guide there, and use + and − to change the guide's length. Find a guide that makes exactly one cut, on the dot.
A 3-letter guide matches all over the strand. A longer guide helps, but only if the guide also avoids the stretch of letters that repeats. Every extra match is an off-target cut: a cut in a place you never meant to touch.
Your strand had 96 letters. Human DNA has about 3 billion. [3] How long must a guide be for a strand that long?
Why twenty letters
Each spot in a guide holds one of 4 letters. So every extra letter makes a match by pure chance 4 times rarer.
Your goal: press + until the bar passes the black line, which marks the number of spots in your DNA. Then keep going to 20.
At about 16 letters, matches by pure chance drop to about one. Real guides use 20 letters, [2] which leaves a safety margin.
Real DNA needs the margin. Real DNA is full of repeats, like the strand in the game. And Cas9 is a sloppy reader: Cas9 sometimes cuts a spot that differs from the guide by a few letters. [6, 7] So scientists use software to check each guide against the whole genome before using the guide. [7]
A bacterium never had to search 3 billion letters. So how did a weapon from bacteria end up in labs searching human DNA?
Borrowed from bacteria
In 1987, a team in Japan reading a gene of E. coli bacteria noticed something odd beside the gene: one short stretch of letters, repeated again and again, with different letters in the gaps. [8] Nobody knew what the repeats were for.
In Spain, Francisco Mojica kept finding the same pattern in microbes from the salt ponds of Santa Pola, near Alicante. [9] In 2005 he reported something strange: the letters in the gaps between the repeats match the DNA of viruses. [10]
Why would a microbe keep pieces of virus DNA in its own DNA?
- The pieces are junk left behind by old infections
- The pieces are a memory that helps the microbe fight that virus
- The pieces help the microbe build viruses of its own
Show answer
B. The pieces are a memory that helps the microbe fight that virus
The pieces are a memory. Mojica noticed that a virus could not infect a microbe that carried a matching piece. [10] But a pattern is not proof. Someone still had to show the memory at work.
The proof came from yogurt. Scientists at Danisco, a company that sells the bacteria used to make yogurt and cheese, attacked those bacteria with viruses. The survivors had added new snippets of the virus's DNA to their repeats, and the new snippets made the bacteria resistant. [11]
Then, in 2012, Charpentier and Doudna rebuilt the scissors in a test tube, simplified the parts and reprogrammed the scissors with a guide of their own. [2] In 2020 the two shared the Nobel Prize in Chemistry. [12]
So a guide sends Cas9 to one spot. What exactly happens when Cas9 gets there?
Inside the scissors
Cas9 holds the guide and pries the two strands of DNA apart. If the guide's letters pair with one strand, Cas9 cuts both strands.
An animation of Cas9 at work. Cas9, drawn as a dashed oval that carries the guide, slides along the DNA, stops at a short tag called the PAM, unzips the DNA beside the PAM, lets the guide pair with one strand, and cuts both strands. The animation plays by itself in a loop and pauses for a moment after the cut. Three step names under the drawing light up in turn: find the PAM, unzip and pair, and cut both strands.
Text version
A stretch of DNA runs across the drawing as two strands joined by rungs. Cas9 slides in from the left with the guide tucked inside, below the DNA, and stops where the upper strand carries a short tag called the PAM. Starting next to the PAM and moving away from it, Cas9 pulls the two strands apart: the rungs between the strands break, the upper strand bulges up and the lower strand bulges down. As the strands part, the guide rises and pairs with the lower strand, letter by letter, beside the PAM. Then Cas9 cuts both strands at the same spot, a few letters from the PAM, and a small gap opens in each strand. The drawing holds on the cut DNA, fades, and starts again.
How it really works (the molecules)
- The guide is RNA. The guide is made of RNA, DNA's chemical cousin, and the guide's 20 letters pair with one strand of DNA: C with G, and A with T (RNA uses U in place of T). In bacteria the guide comes in two RNA pieces. In 2012 Jinek and colleagues joined the two pieces into one single guide RNA, the form most labs now use. [2]
- Cas9 checks a tag first. Cas9 doesn't read DNA one letter at a time. Cas9 looks for a short tag, NGG (any letter, then two Gs), called the PAM. Only beside a PAM does Cas9 unzip the DNA and compare the guide with the 20 letters there. [2]
- Two blades. Cas9 has two cutting parts. One part cuts the strand that pairs with the guide, and the other part cuts the opposite strand, about 3 letters from the PAM. [2]
- Near-misses. How often Cas9 cuts a near-miss depends on how many letters differ and where the wrong letters sit along the guide. [7] One study found off-target cuts at spots with up to five wrong letters. [6]
- Two repairs. Gluing the ends back together is called non-homologous end joining. Copying a patch is called homology-directed repair: the patch must match the DNA on both sides of the cut. [13]
- Into human cells. In early 2013, teams led by Feng Zhang and George Church showed that Cas9 edits DNA inside human cells. [13, 14]
An animation from MIT's McGovern Institute shows the same steps in motion:
Cas9 has cut the DNA in two, and a cut strand is an emergency for a cell. What does the cell do next?
Cut, then repair
The cell repairs the break, and the repair is where the editing happens. [1] Cells read a gene three letters at a time, so a gene works like a sentence of three-letter words. This sentence has one typo.
Your goal: cut the DNA, let the cell glue the ends, and read the words after the cut. Then start over and hand the cell a patch.
Gluing is quick but messy. A letter gained or lost shifts every word after the cut, and the gene stops working. A gene broken on purpose is called a knockout. A patch lets the cell copy the right letters in, and the typo is fixed.
Fixing a typo sounds like the obvious way to treat a disease, and breaking a gene sounds useless. Is that how the first CRISPR medicine works?
A cure for sickle cell
Sickle cell disease starts with one changed building block in hemoglobin, the protein that carries oxygen in red blood cells. [15] The faulty hemoglobin bends red blood cells into a crescent, or sickle, shape. Sickled cells block blood vessels and cause attacks of severe pain and organ damage. About 100,000 people in the US have the disease. [5]
The first CRISPR medicine, called Casgevy, treats sickle cell disease. What do you think Casgevy edits?
- Casgevy fixes the typo in the hemoglobin gene
- Casgevy breaks a different stretch of DNA on purpose
- Casgevy adds a new copy of the hemoglobin gene
Show answer
B. Casgevy breaks a different stretch of DNA on purpose
Casgevy breaks something. Before birth, your blood makes a different hemoglobin, called fetal hemoglobin, and a gene called BCL11A turns fetal hemoglobin off after birth. Casgevy cuts the stretch of DNA that switches BCL11A on in blood-making cells, and the cell's messy glue breaks the switch. [16] The cells go back to making fetal hemoglobin, which keeps red blood cells from sickling. [5]
Casgevy is not a pill. Doctors collect the patient's own blood stem cells, and a lab edits the cells. The patient then needs high-dose chemotherapy to clear out the bone marrow, and the edited cells go back in. [5]
The UK approved Casgevy on 16 November 2023, the first approval anywhere, and the US followed on 8 December 2023. [4, 5] In the trial, 29 of 31 patients followed long enough had no severe pain attacks for at least a year. [5]
Sickle cell is one use. Where else does CRISPR show up?
Where it shows up
- In labs everywhere. To learn what a gene does, scientists knock the gene out in cells or animals and watch what changes. Cas9 edits human and mouse cells alike. [13]
- On your plate. In 2021, a tomato edited to hold more of a natural compound called GABA went on sale in Japan, reported as the first CRISPR-edited food sold to the public. [17]
- In the clinic. Casgevy is approved for sickle cell disease and for a related blood disorder, beta thalassemia. [4]
Every one of these edits stays in the cells that were edited. What if an edit could be passed on to someone's children?
A line crossed
In November 2018, the Chinese scientist He Jiankui announced that he had used CRISPR on human embryos, and that twin girls had been born from the embryos. He Jiankui said he had edited a gene called CCR5, which HIV uses to infect immune cells, to protect the girls from HIV. [18]
Casgevy edits one patient's blood stem cells. He Jiankui edited embryos. Which edit can be passed on to future children?
- The edit in Casgevy patients
- The edit in the embryos
- Both edits
Show answer
B. The edit in the embryos
Only the embryo edit. An edit in an embryo can reach the cells that later make eggs or sperm, so the edit can pass to future generations. An edit in blood stem cells stays with one patient. [1]
Many scientists were alarmed: nobody knew the side effects, and any mistake could pass down to future generations. Later research suggested that the CCR5 change might leave people more vulnerable to other viruses, such as West Nile and flu. [18]
In December 2019, a court in China sentenced He Jiankui to three years in prison for illegal medical practice. The court also confirmed that a third gene-edited baby had been born. [18] Editing embryos to start a pregnancy is barred in the United States and many other countries. [1]
CRISPR, explained
Now you can explain how CRISPR works, from the bacterium to the clinic:
- The poster. Bacteria keep snippets of virus DNA, and Cas9 cuts any DNA that matches a snippet. Write your own snippet, the guide, and Cas9 goes where you send it.
- Twenty letters. Each letter makes a chance match 4 times rarer, so 20 letters pick out one spot in 3 billion. Repeats and near-misses still cause off-target cuts, so guides are checked by software.
- The repair. After the cut, the cell glues the ends and breaks the gene, or copies a patch and fixes the gene.
- The medicine. Casgevy uses the messy glue to break a switch, so blood cells make fetal hemoglobin again.
- The line. An edit in an embryo can pass to children. In 2018 He Jiankui crossed that line, which the United States and many other countries bar.
Where to go next
CRISPR opens a lot of doors. Here are the questions this page leads to:
- Can scientists change a single letter of DNA without cutting both strands?
- How do you deliver CRISPR to the right cells inside a living body?
- How does a bacterium stop Cas9 from cutting the wanted posters in its own DNA?
- Could CRISPR wipe out the mosquitoes that spread malaria, and should it?
- Should anyone ever edit a human embryo, and who gets to decide?
Ask Phy
- Walk me through how Cas9 finds its target, from the PAM to the cut.
- How does Casgevy turn fetal hemoglobin back on? Explain it like a light switch.
- Give me the strongest arguments for and against editing human embryos.
- What are genome editing and CRISPR-Cas9? (MedlinePlus Genetics, US National Library of Medicine) — How bacteria use CRISPR, how labs use it, and germline editing.
- Jinek, Chylinski, Fonfara, Hauer, Doudna and Charpentier: A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity (Science, 2012) — The paper that turned Cas9 into a programmable tool.
- What is DNA? (MedlinePlus Genetics) — Four bases; about 3 billion bases in human DNA.
- MHRA authorises world-first gene therapy that aims to cure sickle-cell disease and transfusion-dependent β-thalassemia (UK government, 16 November 2023)
- FDA Approves First Gene Therapies to Treat Patients with Sickle Cell Disease (US FDA, 8 December 2023) — How Casgevy works, the trial result, and the chemotherapy it needs.
- Fu et al.: High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells (Nature Biotechnology, 2013)
- Hsu et al.: DNA targeting specificity of RNA-guided Cas9 nucleases (Nature Biotechnology, 2013)
- Ishino et al.: Nucleotide sequence of the iap gene ... in Escherichia coli (Journal of Bacteriology, 1987) — The first report of the strange repeats.
- Francisco Mojica: "I'm incredibly proud of having been part of the CRISPR revolution" (El·lipse, PRBB) — Interview: the salt ponds of Santa Pola.
- Mojica et al.: Intervening sequences of regularly spaced prokaryotic repeats derive from foreign genetic elements (Journal of Molecular Evolution, 2005)
- Barrangou et al.: CRISPR provides acquired resistance against viruses in prokaryotes (Science, 2007) — The yogurt bacteria experiment.
- Press release: The Nobel Prize in Chemistry 2020 (NobelPrize.org)
- Cong et al.: Multiplex genome engineering using CRISPR/Cas systems (Science, 2013)
- Mali et al.: RNA-guided human genome engineering via Cas9 (Science, 2013)
- HBB gene (MedlinePlus Genetics) — The hemoglobin change behind sickle cell disease.
- Frangoul et al.: CRISPR-Cas9 gene editing for sickle cell disease and β-thalassemia (New England Journal of Medicine, 2021) — The first patients treated with the edit that became Casgevy.
- GABA-enriched tomato is first CRISPR-edited food to enter market (Nature Biotechnology, 2021)
- Chinese researcher who created gene-edited babies sentenced to 3 years in prison (NPR, 30 December 2019)