Mark ở vũ trụ khác · Another Universe Genes and Rare Diseases #9 · Gene editing · Bacteria

CRISPR: scissors borrowed from bacteria

Jinek, Doudna, Charpentier and colleagues · Science · 2012

The storyfrom odd repeats in bacteria to a tool for fixing genes
Key peopleMojica, Barrangou, Charpentier, Doudna, Šikšnys, Zhang
Todayfirst CRISPR medicine (2023), baby KJ (2025)
LabelSolid as science, Caution as a cure

Source: Jinek et al. (2012), Science 337(6096): 816–821; Nobel Prize in Chemistry 2020 press release; Frangoul et al. (2024), NEJM 390: 1649–1662; Musunuru et al. (2025), NEJM.

For reading and understanding, not a substitute for medical advice.

Recap in 1 minute

Essence

Bacteria keep a "virus logbook"A bacterium copies a bit of DNA from a virus that attacked it into its own genome, so next time it can recognise and cut it.
2012: programmable scissorsJinek, Doudna and Charpentier showed the Cas9 protein cuts DNA exactly where a guide RNA points it.
Nobel Prize in Chemistry 2020Awarded to Charpentier and Doudna. Many others had supplied pieces of the puzzle before.
First medicine, 2023The UK, then the US, approved a CRISPR therapy for sickle cell disease: 29 of 30 people had no pain crises for at least 12 months.
Baby KJ, 2025A drug that changes one letter in a gene, made for one baby. Good results, but one case and still very early.

Source: Barrangou et al. (2007), Science 315: 1709–1712; Jinek et al. (2012); Nobel Prize in Chemistry press release 7 Oct 2020; Frangoul et al. (2024), NEJM; Musunuru et al. (2025), NEJM.

Part 1

The question

How did a way bacteria defend themselves against viruses, found by people studying microbes in salt marshes and yogurt cultures, become a tool for editing human genes? And what can it cure today, and what not yet?

The question and wording are Mark's, based on the 2020 Nobel press release and the papers in Part 9.

1987–2007Odd repeatsEveryone saw them, nobody understood. Then the answer: an anti-virus memory.
→
2012–2013The scissorsCas9 cuts where it is told. Then it works in human cells.
→
2023–2025To patientsThe first medicine is approved, and one baby gets a gene fix of his own.
Part 2

The story

Who, where, when: a discovery pieced together by many people in many countries.

1987 · Japan: odd repeats in E. coliYoshizumi Ishino's team sequenced a gene of the gut bacterium and found, right next to it, a short sequence repeated again and again with different stretches in between. Nobody knew what it was for.
Early 1990s · Spain: the Santa Pola salt marshesFrancisco Mojica, a PhD student in Alicante, found similar repeats in a salt-loving archaeon from the salt ponds. He chased the question "what are they for" for more than ten years.
2002 · A nameRuud Jansen's team (Netherlands) coined CRISPR, for "clustered regularly interspaced short palindromic repeats", and found the cas genes that always sit nearby.
2005 · The answer: bits of virus DNAMojica compared the in-between stretches with sequence databases and found they matched DNA from viruses and plasmids. His guess: this is the immune memory of bacteria.
2007 · Yogurt proves itAt the dairy company Danisco, Rodolphe Barrangou, Philippe Horvath and colleagues showed that yogurt bacteria add a new piece of virus to their CRISPR after an attack, and from then on resist that virus.
2011 · The missing pieceEmmanuelle Charpentier (then in Sweden) found tracrRNA, an RNA molecule the scissors need. The same year she began working with Jennifer Doudna (Berkeley, USA).
28 June 2012 · The Science paperJinek, Doudna, Charpentier and colleagues showed that the Cas9 protein, guided by two RNAs, cuts both strands of DNA at the right place. Join the two RNAs into one and you have "programmable" scissors: change the RNA, change the cut. The same year, Virginijus Šikšnys's team (Lithuania) independently reported Cas9 cutting DNA guided by RNA.
2013 · Into human cellsFeng Zhang's team (Broad Institute) and George Church's team (Harvard) showed Cas9 can edit genes in human and mouse cells. From then on, thousands of labs could use CRISPR.
7 Oct 2020 · Nobel Prize in ChemistryAwarded to Charpentier and Doudna "for the development of a method for genome editing".

Source: Ishino et al. (1987), J Bacteriol 169: 5429–5433; Nakata et al. (1989), J Bacteriol 171: 3553–3556; Mojica, Juez and Rodríguez-Valera (1993), Mol Microbiol 9(3): 613–621; Jansen et al. (2002), Mol Microbiol 43: 1565–1575; Mojica et al. (2005), J Mol Evol 60: 174–182; Barrangou et al. (2007), Science 315: 1709–1712; Jinek et al. (2012), Science 337: 816–821 (online 28 June 2012); Gasiunas et al. (2012), PNAS 109(39): E2579–E2586; Cong et al. (2013), Science 339: 819–823; Mali et al. (2013), Science 339: 823; Nobel press release 7 Oct 2020. Santa Pola and Danisco details from The Scientist and the ASU Embryo Project (secondary sources).

Why do many people think CRISPR was "invented by two people"?

A Nobel Prize goes to at most three people a year, and in 2020 it went to the two scientists who turned a bacterial mechanism into a tool. But the story spans at least four countries: Japan (the first repeats), Spain and the Netherlands (understanding and naming), France and Denmark (Danisco's yogurt experiments), Lithuania and the US (the scissors in a test tube and in human cells).

Two US institutions (the University of California, where Doudna works, and the Broad Institute, where Zhang works) are still fighting over patents. In May 2025 a US federal appeals court sent the case back to the Patent Office. Who "thought of it first" is still legally unsettled.

Source: Nobel Prize in Chemistry 2020 press release; US Court of Appeals for the Federal Circuit decision of 12 May 2025 in Regents of the University of California v. Broad Institute, via summaries by MoFo and White & Case (secondary sources).

Part 3

How the gene works

1. A guide RNA carries about 20 letters that match the DNA stretch to be found. It walks the Cas9 protein along the genome until it hits the match.

2. At the match, Cas9 cuts both DNA strands. The cell rushes to repair the cut; in doing so it may break the gene (switch it off) or copy a repair template the scientists supplied.

3. Newer versions such as "base editing" do not cut through the DNA but change exactly one letter, for example A to G. This is the approach used for baby KJ.

Three-step diagram. Step 1: the guide RNA (pink) leads the Cas9 protein (green) along the DNA. Step 2: at the matching stretch, Cas9 cuts both strands. Step 3: the cell repairs the cut, either switching the gene off or copying a repair template; base editing changes one letter without cutting through. 1 · search 2 · cut both strands 3 · the cell repairs Cas9 guide RNA (~20 letters) scans along until it matches both strands cut exactly where the RNA points quick patch → gene offa few letters added/lost copy a repair templaterarer, harder base editing: A → Gno cut through (2016–2017) illustrative diagram, not to scale
Scissors that follow a map. Change the guide RNA and you change the cut, which is why CRISPR is far cheaper and faster than earlier gene-editing methods that needed a redesigned protein for every site.Source: Jinek et al. (2012) (Cas9, guide RNA, double-strand cut); Cong et al. (2013) (editing in human cells); Komor et al. (2016), Nature 533: 420–424 and Gaudelli et al. (2017), Nature 551: 464–471 (base editing without a cut through). About 20 letters: guide length per Jinek (2012) and Cong (2013).
Two ways to bring the scissors to a patient. Left, outside the body (Casgevy, 2023): take blood stem cells, edit them in the lab, give high-dose chemotherapy, infuse them back. Right, inside the body (baby KJ, 2025): wrap the drug in lipid nanoparticles, infuse into a vein, the drug reaches the liver and changes one letter. outside the body · Casgevy (2023) 1. collect the patient's own blood stem cells2. edit the gene in the lab3. high-dose chemo to "clear" the marrow4. infuse the edited cells back switches fetal haemoglobin back on inside the body · baby KJ (2025) 1. read the gene, find the wrong letter2. build a custom drug in about 6 months3. wrap in lipid nanoparticles, infuse4. drug reaches the liver, A → G no marrow-clearing chemo needed
Same CRISPR roots, two different routes. Both edit body cells, not eggs or sperm, so the changes are not passed on to children.Source: Frangoul et al. (2024), NEJM 390: 1649–1662 (exa-cel process, busulfan conditioning); Vertex release and ASGCT on the FDA decision of 8 Dec 2023 (BCL11A gene, fetal haemoglobin); Musunuru et al. (2025), NEJM; Penn Medicine and CHOP releases 15 May 2025 (about 6 months, lipid nanoparticles); MedlinePlus Genetics (body cells and germline cells).
Part 4

Fact check: what is true and what is rumour

14 things often said about CRISPR, checked against the original papers and regulators' decisions. Tap a row for the reasoning.

SOLIDCRISPR started as a bacterial immune system against viruses

The in-between stretches match virus DNA (Mojica 2005); yogurt bacteria copy a new piece of virus and become resistant to it (Barrangou 2007).

SOLIDThe odd repeats were first seen in 1987, when nobody understood them

Ishino et al. (1987) saw repeats next to the iap gene in E. coli; Nakata et al. (1989) described a 29-base-pair sequence repeated 14 times.

SOLIDThe name CRISPR dates from the early 2000s

Jansen et al. (2002) used it in Molecular Microbiology and identified the accompanying cas genes.

SOLIDIn 2012, Cas9 was "programmed" with RNA to cut a chosen site

Jinek et al. (2012), Science 337: 816–821; the same year Šikšnys's team reported a similar result in PNAS.

SOLIDThe 2020 Nobel Prize in Chemistry went to Charpentier and Doudna

Press release of the Royal Swedish Academy of Sciences, 7 October 2020.

SOLIDThe first CRISPR therapy was approved in 2023

The UK granted conditional authorisation on 16 Nov 2023; the US FDA approved it on 8 Dec 2023 for sickle cell disease from age 12 (MHRA, Vertex; ASGCT).

SOLIDIn the trial, 29 of 30 people had no pain crises for at least 12 months

30 people with enough follow-up out of 44 infused; 29/30 (97%) free of severe vaso-occlusive crises for at least 12 consecutive months; 30/30 free of hospital stays for them (Frangoul et al. 2024, NEJM).

SOLIDEditing body cells is not passed on to children

Only changes to eggs, sperm or embryos pass to the next generation (MedlinePlus Genetics). The story of embryo editing is piece #10.

SOLIDBaby KJ was the first person to receive a custom gene-editing drug infused straight into the body

Musunuru et al. (2025), NEJM, published 15 May 2025; Penn Medicine and CHOP releases the same day.

CAUTION"CRISPR has cured sickle cell disease for everyone"

Trial results are very good, but patients need high-dose chemotherapy (which can affect fertility), median follow-up was only about 19 months, and the US price is 2.2 million USD (Reuters, AAFP; secondary sources). Very few hospitals can do it.

CAUTION"KJ is completely cured"

His doctors say it is "still really early days" and do not call it a cure; he still needs care. No liver biopsy was done, so how many cells were edited is unknown (Penn, CHOP; press, secondary).

CAUTION"Anyone with a rare disease can now get a custom gene-editing drug"

Only one case has been published. KJ's team plans a trial for babies with similar disorders; the FDA has only a draft guidance (23 Feb 2026) for this kind of individualised drug.

FALSE"Doudna and Charpentier invented CRISPR on their own"

They turned the mechanism into a tool. Discovering the mechanism took many teams: Ishino, Mojica, Jansen, Barrangou and Horvath, Šikšnys; bringing it into human cells involved Zhang's and Church's teams.

FALSE"CRISPR is perfectly precise and never cuts the wrong place"

It can cut off-target, and right at the cut site large deletions or scrambled DNA can occur (Kosicki, Tomberg and Bradley 2018, Nature Biotechnology 36: 765–771). The Casgevy label warns of off-target editing.

Source: as given in each row; full list in Part 9.

Part 5

Where we are today

From the first medicine to baby KJ, and what is still pending.

16 Nov 2023 and 8 Dec 2023 · The first CRISPR medicineThe UK, then the US, approved exa-cel (brand name casgevy) for sickle cell disease; the UK also approved it for transfusion-dependent beta thalassaemia. It edits blood stem cells to switch fetal haemoglobin back on.
2024 · Results published in NEJM29/30 people had no crises for at least 12 months. US price 2.2 million USD per treatment (secondary source).
Aug 2024 · KJ is bornA baby in Philadelphia (USA) with severe CPS1 deficiency: his liver cannot clear ammonia, which builds up in the blood and harms the brain. Very rare, estimated at about 1 in 200,000 to 1 in 1,300,000 babies.
Feb–Apr 2025 · Three infusionsThe team of Kiran Musunuru and Rebecca Ahrens-Nicklas built his drug in about 6 months; the FDA allowed its use one week after the application. First dose at 6–7 months of age.
15 May 2025 · NEJM paperThe paper describes the treatment and early follow-up. According to the press, he eats more normally and needs less medicine; in June 2025 he went home after 307 days in hospital (secondary source).
Dec 2025 · First stepsHe took his first steps (MIT Technology Review, Jan 2026, secondary). His doctors still say: very early days.
23 Feb 2026 · FDA draft guidanceA "plausible mechanism framework" for custom genetic drugs for ultra-rare diseases. It is a draft, not a new legal approval pathway.

Source: MHRA and Vertex releases 16 Nov 2023; FDA 8 Dec 2023 (via ASGCT); Frangoul et al. (2024), NEJM 390(18): 1649–1662; Musunuru et al. (2025), NEJM, doi 10.1056/NEJMoa2504747; Penn Medicine and CHOP releases 15 May 2025; ClinGen report AC1046 (CPS1 frequency); FDA draft guidance 23 Feb 2026 via AJMC and Hogan Lovells (secondary); price per Reuters and AAFP; 307 days per NBC/AP; first steps per MIT Technology Review 12 Jan 2026 (secondary). Checked 16 Oct 2026.

Part 6

Verdict

Scores are Mark's assessment based on the sources in Part 9, not data from a study.

Strength of the science
5/5
Impact on research
5/5
Already helping patients
3/5
Reaching most people
1/5
Strengths
  • Mechanism confirmed by many independent teams
  • An approved medicine with trial data
  • Opens a path to custom drugs for ultra-rare diseases
Weaknesses
  • Off-target cuts and unwanted deletions possible
  • Very high price, heavy procedure
  • Follow-up still short

Solid as science, Caution as a cure: the scissors have proven they cut true; getting them to every patient is a long road.

More: why does a case like KJ matter if it is only one person?

One case is not enough to say a drug works. But KJ's case shows a new process is feasible: read the gene, design the drug, test safety in cells and animals, get permission, all in about half a year.

For thousands of rare diseases, each affecting only a few people, no company can run a large trial. If the "made to measure" process can be repeated for many babies, that would be the real change. Those data do not exist yet.

The "More" section is Mark's commentary, based on the sourced facts in Parts 2 to 5.

Part 7

Lessons: how to read a "cured by gene editing" headline

There will be many more CRISPR headlines. Four questions tell you how far a story really goes.

For reading and understanding, not a substitute for medical advice.

1 · How many people?One case, a few dozen in a trial, or thousands? KJ is one case; Casgevy had 30 people with enough follow-up.
2 · Followed for how long?A few months or many years? Gene-editing drugs need long follow-up to catch late side effects.
3 · Which cells were edited?Body cells (not passed to children) or embryos, eggs, sperm (passed to the next generation, banned in many places)?
4 · Who approved it, where?A peer-reviewed paper? Which regulator allowed it? "In trials" is not "approved".
Beware "CRISPR" advertsReal gene-editing drugs are given only in hospitals, in trials or under licence. There is no "gene-fixing" pill or cosmetic.
Ask your doctor the right questionIf someone in your family has a genetic disease: "Are there gene therapy trials for this? Where? Who is eligible?"
Self-check: read one headline in 5 minutes

1. Find a "cured by gene editing" story in the news or on social media.

2. Answer the four questions above. Which ones does the story fail to answer?

3. Does it name the paper or the approving regulator? If not, treat it as unverified.

4. Rewrite the headline at the right strength, e.g. "one baby treated experimentally, early results good".

The four questions and the self-check are Mark's suggestions, based on the facts in Parts 4 and 5; body vs germline cells per MedlinePlus Genetics.

Part 8

Read next

Genes and Rare Diseases #6: Sickle cellThe disease the first CRISPR medicine targets, and why it was once a shield against malaria. Read
Genes and Rare Diseases #5: ThalassaemiaAn inherited blood disease very common in Vietnam, also covered by the UK licence. Read
Genes and Rare Diseases #10: The gene-edited babiesWhen the scissors were used on human embryos in 2018, and why the whole scientific world spoke up. Read
Jennifer Doudna and Samuel Sternberg (2017), A Crack in CreationA book by one of the insiders. Read it alongside Walter Isaacson's The Code Breaker (2021) for more sides.
Part 9

Sources

Primary sources are the original papers, the Nobel press release and regulators' decisions. The working environment blocks PubMed, the FDA, the Nobel site and hospital pages, so figures were checked against those pages' own excerpts in search results, without opening the full text. Secondary sources are marked.

  1. Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E (2012), "A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity", Science 337(6096): 816–821. PubMed
  2. Nobel Prize in Chemistry 2020, press release 7 Oct 2020. nobelprize.org
  3. Ishino Y et al. (1987), J Bacteriol 169: 5429–5433. PubMed · Nakata A et al. (1989), J Bacteriol 171: 3553–3556.
  4. Mojica FJM, Juez G, Rodríguez-Valera F (1993), Mol Microbiol 9(3): 613–621 · Mojica FJM et al. (2005), "Intervening sequences of regularly spaced prokaryotic repeats derive from foreign genetic elements", J Mol Evol 60: 174–182.
  5. Jansen R, van Embden JDA, Gaastra W, Schouls LM (2002), "Identification of genes that are associated with DNA repeats in prokaryotes", Mol Microbiol 43: 1565–1575.
  6. Barrangou R et al. (2007), "CRISPR provides acquired resistance against viruses in prokaryotes", Science 315: 1709–1712.
  7. Gasiunas G, Barrangou R, Horvath P, Siksnys V (2012), PNAS 109(39): E2579–E2586. PMC
  8. Cong L et al. (2013), "Multiplex genome engineering using CRISPR/Cas systems", Science 339: 819–823 · Mali P et al. (2013), Science 339: 823.
  9. Komor AC et al. (2016), Nature 533: 420–424 · Gaudelli NM et al. (2017), Nature 551: 464–471.
  10. Kosicki M, Tomberg K, Bradley A (2018), Nature Biotechnology 36: 765–771. PMC
  11. Frangoul H et al. (2024), "Exagamglogene autotemcel for severe sickle cell disease", NEJM 390(18): 1649–1662. PubMed
  12. FDA, "FDA approves first gene therapies to treat patients with sickle cell disease", 8 Dec 2023. fda.gov · MHRA, Casgevy authorisation 16 Nov 2023.
  13. Musunuru K, Ahrens-Nicklas RC et al. (2025), "Patient-specific in vivo gene editing to treat a rare genetic disease", NEJM. nejm.org
  14. MedlinePlus Genetics, "What are the ethical issues surrounding gene therapy?". medlineplus.gov · ClinGen, actionability report AC1046 (CPS1 deficiency).
  15. Secondary sources: Penn Medicine and CHOP releases 15 May 2025; NBC/AP and Philadelphia Inquirer 2025 (307 days); MIT Technology Review 12 Jan 2026; Reuters and AAFP (2.2 million USD price); AJMC, Hogan Lovells (FDA draft 23 Feb 2026); MoFo, White & Case (ruling of 12 May 2025); The Scientist, ASU Embryo Project (Danisco, yogurt).