CRISPR and genome editing — how we rewrite the code

What CRISPR actually is

CRISPR stands for Clustered Regularly Interspaced Short Palindromic Repeats — a mouthful that refers to a pattern of DNA sequences found in bacterial genomes. Before the 2010s, it was an obscure feature of prokaryotic genetics studied by a handful of researchers.

Bacteria have immune systems. When a virus infects a bacterium, the bacterium can capture a short piece of the viral DNA and store it in its genome between these CRISPR repeat sequences. If the same virus attacks again, the bacterium transcribes those stored sequences into RNA, which guides a protein called Cas (CRISPR-associated) to cut any matching DNA it encounters — destroying the viral genome.

In 2012, Jennifer Doudna and Emmanuelle Charpentier published a landmark paper in Science showing that the CRISPR-Cas9 system from Streptococcus pyogenes could be programmed to cut any DNA sequence by simply changing the guide RNA. The bacterium's immune memory could be repurposed as a programmable molecular scissors.

This was the insight that launched a revolution. Previously, genome editing required engineering custom proteins (zinc finger nucleases, TALENs) for every target — a process that took months and cost hundreds of thousands of dollars. CRISPR reduced it to designing a short RNA sequence, a task that takes hours and costs almost nothing.

Doudna and Charpentier received the 2020 Nobel Prize in Chemistry for this work.

---

Mechanism — how Cas9 finds and cuts DNA

The CRISPR-Cas9 editing system has three core components:

1. The guide RNA (gRNA): A short synthetic RNA molecule (~20 nucleotides) designed to match the target DNA sequence. The gRNA base-pairs with the target strand of DNA using the same Watson-Crick complementarity that governs DNA replication and transcription.

2. The Cas9 protein: A large endonuclease (DNA-cutting enzyme) that associates with the gRNA and carries out the actual cut. Cas9 has two nuclease domains — one cuts each strand of the double helix — producing a double-strand break (DSB).

3. The PAM sequence: Cas9 also requires a short sequence adjacent to the target called a protospacer adjacent motif (PAM). For SpCas9, the PAM is NGG (any nucleotide followed by two guanines). The PAM must be present immediately downstream of the target sequence in the genome. This is a constraint on what sites can be targeted — not every sequence in the genome has an accessible PAM — but in practice, a PAM exists approximately every 8 base pairs in the human genome, so most targets can be reached.

The cutting process:

1. Cas9-gRNA complex scans the genome, briefly interacting with many sequences
2. When it finds a sequence matching the gRNA with an adjacent PAM, it unwinds the DNA and checks for full complementarity
3. On successful matching, Cas9 undergoes a conformational change that activates both nuclease domains
4. Both strands of DNA are cut, creating a blunt-ended double-strand break

The DSB is the initiating event. What happens next depends on which cellular repair pathway takes over.

---

What happens after the cut — NHEJ vs. HDR

A double-strand break is a genomic emergency. The cell has two primary mechanisms to repair it, and they produce very different outcomes.

Non-Homologous End Joining (NHEJ)

NHEJ is the cell's fast, sloppy repair pathway. It simply rejoins the two broken ends — but in doing so, it frequently introduces small insertions or deletions (indels) at the cut site. These indels are typically 1–20 base pairs and are essentially random.

If the cut is in a coding exon, indels will usually cause a frameshift — and as you learned in Module 6, frameshifts almost always destroy protein function (the protein is truncated by nonsense-mediated decay or produces a garbled sequence). This makes NHEJ useful for gene knockout: you can use CRISPR to disrupt a gene's function by introducing indels that destroy the reading frame.

NHEJ is the default repair pathway and occurs in virtually all cell types and cell cycle stages. This makes CRISPR-mediated knockout relatively efficient and straightforward.

Homology-Directed Repair (HDR)

HDR uses a template with sequences homologous to the region flanking the break to repair the DNA accurately. In cells with an intact sister chromatid (primarily during S and G2 phases of the cell cycle), HDR can repair DSBs without errors.

If you provide an exogenous DNA template alongside the Cas9-gRNA system, the cell can use that template during HDR — allowing you to insert a specific sequence, correct a point mutation, or make any precise edit you design.

This is CRISPR knock-in or correction: instead of destroying a gene, you fix it or insert new information.

The catch: HDR is much less efficient than NHEJ, and it's essentially absent in non-dividing cells (neurons, cardiomyocytes, most differentiated adult cells). This is a major limitation for therapeutic genome editing — most disease-relevant cell types divide infrequently or not at all, making precise correction extremely difficult.

The core tradeoff:

• NHEJ: efficient, available in all cell types, but imprecise (knockout only)
• HDR: precise, allows any edit, but inefficient and restricted to dividing cells

---

Beyond scissors — base editors and prime editors

The limitations of HDR drove researchers to develop editing tools that don't rely on a double-strand break at all.

Base editors (2016 — David Liu lab, Broad Institute)

Base editors are fusion proteins combining a catalytically impaired Cas9 (which nicks one strand but doesn't cut both) with a chemical enzyme that directly converts one base to another — without cutting the DNA.

Two main classes:

• Cytosine base editors (CBE): Convert C to T (via deamination of cytosine to uracil, which the cell reads as thymine)
• Adenine base editors (ABE): Convert A to G (the trickier direction — required engineering an enzyme that doesn't exist in nature)

Base editors can correct roughly 30% of known pathogenic point mutations — the subset caused by single base changes amenable to C→T or A→G conversion. They're more efficient than HDR-based correction and don't produce double-strand breaks, reducing the risk of unintended large deletions or chromosomal rearrangements.

Prime editors (2019 — David Liu lab)

Prime editors are a further evolution: a Cas9 nickase fused to a reverse transcriptase, guided by a specially designed "prime editing guide RNA" (pegRNA) that encodes both the target sequence and the desired edit sequence.

Instead of cutting and using a template, prime editing essentially writes new sequence directly into the genome by reverse-transcribing a short RNA template at the nick site. Prime editors can in principle make any substitution, small insertion, or small deletion — without a DSB and without requiring an exogenous DNA template.

Prime editing is more flexible than base editing (any base change, not just C→T or A→G) but is currently less efficient in many contexts. Both technologies are being actively developed for clinical use.

CRISPRi and CRISPRa (no editing at all)

A further variation uses a catalytically dead Cas9 (dCas9) — it binds the target sequence via the gRNA but doesn't cut. Fused to repressor or activator domains, dCas9 can:

• CRISPRi: Repress gene expression without editing the DNA
• CRISPRa: Activate gene expression without editing the DNA

These tools are powerful for research — you can dial gene expression up or down without permanently altering sequence — and are being explored therapeutically for diseases where you want to modulate expression rather than fix a mutation.

---

Delivery — the unsolved problem

Having a precise editing tool is only half the challenge. Getting it into the right cells in a living person is the other half, and it is arguably harder.

Ex vivo editing: Remove cells from the patient, edit them outside the body, then reinfuse. This is the most clinically advanced approach — it's what the first CRISPR medicines use. It works best for blood cells (hematopoietic stem cells) and T cells, which can be extracted, manipulated, and reinfused.

In vivo editing: Deliver the editing machinery directly into the body to reach target tissues. Requires a delivery vehicle that can reach the target, enter cells, and release the editing cargo without being destroyed by the immune system.

Current delivery vehicles:

• Adeno-associated virus (AAV): The workhorse of gene therapy. Small, non-integrating, tissue-tropic (different serotypes target different tissues). Limitation: small cargo capacity (~4.7 kb), which barely fits a Cas9 gene. Cannot be re-dosed because of immune responses.
• Lipid nanoparticles (LNPs): Synthetic fat-based particles that encapsulate mRNA or ribonucleoprotein (Cas9 protein + gRNA). Highly efficient for liver delivery — used in the first approved in vivo CRISPR therapy. Less efficient for other tissues.
• Base editor/prime editor mRNA: Delivered via LNP as mRNA, which is translated into protein transiently and then degraded — reducing off-target risk because the editor isn't permanently present.

Getting CRISPR to the brain, muscle, lung, or eye in sufficient quantities remains a major unsolved problem. Most current therapies target the liver or blood because those are where delivery is most tractable.

---

Off-target effects — why precision matters

Cas9 is precise but not perfect. It can cut at unintended sites in the genome where the sequence partially matches the gRNA — called off-target editing. Off-target cuts can cause:

• Indels in unintended genes (potentially disrupting function)
• Chromosomal rearrangements if two off-target cuts occur on different chromosomes
• In dividing cells, off-target edits in tumor suppressor genes could contribute to cancer

The frequency of off-target editing depends on gRNA design, Cas9 variant, and the target sequence. Multiple approaches reduce off-target risk:

• High-fidelity Cas9 variants (eSpCas9, HiFi Cas9): engineered to require tighter guide-target matching
• Paired nickases: Two Cas9 nickases targeting opposite strands near each other; each individually creates only a nick, requiring both to cut at the correct location for a DSB — greatly reducing off-target activity
• Careful gRNA design: Bioinformatic screening for potential off-target sites before clinical use
• Transient delivery: Using mRNA rather than DNA encoding Cas9 limits the duration of exposure

Off-target assessment is now a required part of preclinical development for CRISPR therapeutics. The FDA requires comprehensive off-target profiling before clinical trials.

---

Approved therapies and clinical pipeline

CRISPR moved from a 2012 research tool to an approved medicine in 12 years — one of the fastest translations in biomedical history.

Casgevy (exa-cel) — approved December 2023 (FDA)

The first approved CRISPR therapy, developed by Vertex Pharmaceuticals and CRISPR Therapeutics. It treats sickle cell disease and transfusion-dependent beta-thalassemia — both caused by mutations in the beta-globin gene (HBB) that produce dysfunctional hemoglobin.

The approach is indirect and elegant: instead of fixing the HBB mutation directly, Casgevy uses CRISPR to disrupt the BCL11A gene in the patient's hematopoietic stem cells. BCL11A normally represses fetal hemoglobin (HbF) — a form of hemoglobin that is functional but is normally switched off after birth. By disrupting BCL11A, the therapy reactivates fetal hemoglobin production, compensating for the defective adult hemoglobin.

Process:

1. Extract hematopoietic stem cells from the patient
2. Edit with CRISPR ex vivo to disrupt BCL11A
3. Give the patient chemotherapy to eliminate the remaining unedited bone marrow
4. Reinfuse the edited cells, which engraft and produce HbF-expressing red blood cells

Clinical trial results: in trials, 29 of 29 patients with sickle cell disease had no severe vaso-occlusive crises for at least 12 months post-treatment. This is transformative for a disease that previously required regular hospitalizations.

Cost: approximately $2.2 million per patient — a number that immediately raised access and equity questions.

Nexiguran ziclumeran — under FDA review (2024–2025)

An in vivo CRISPR therapy from Intellia Therapeutics for transthyretin amyloidosis (ATTR), a disease caused by misfolding of the TTR protein in the liver. Delivered via lipid nanoparticles directly to the liver; knocks out TTR expression. Phase 3 data showed 90%+ reduction in TTR protein levels — competitive with approved RNA-targeting drugs for the same disease.

Oncology: CRISPR-engineered T cells

Multiple clinical trials are using CRISPR to engineer T cells for cancer therapy — disrupting genes that normally limit T cell activity (PD-1, CTLA-4) or that cause T cells to attack normal tissue (TCR), while engineering them to target tumor antigens. Early results are promising.

---

Germline editing — the line that was crossed

Everything described so far is somatic editing: editing cells in a patient's body. Somatic edits affect only that person, are not inherited, and are regulated as standard medical therapies.

Germline editing means editing sperm, eggs, or early embryos — edits that would be present in every cell of any resulting person and would be heritable by their children. This is a categorically different intervention.

The scientific community had reached a broad consensus, articulated at a 2015 international summit, that clinical germline editing was premature and should not proceed, for several reasons:

• Off-target effects cannot be fully screened in an embryo before implantation
• Mosaicism: not all cells in an edited embryo may carry the edit, making outcomes unpredictable
• The heritable nature of changes means errors would be passed to future generations
• The medical justification is unclear — preimplantation genetic testing (PGT) already allows selection of unaffected embryos without editing

In November 2018, Chinese biophysicist He Jiankui announced he had implanted CRISPR-edited embryos into women, resulting in the birth of twin girls (and later a third child) with edits to the CCR5 gene — a co-receptor used by HIV to infect cells. He claimed this would make them resistant to HIV.

The response from the scientific community was swift and near-universal condemnation:

• Medical justification was weak: The children's father was HIV-positive but his virus was undetectable on treatment. Standard IVF already prevents transmission. No medical need existed.
• Off-target edits were present and uncharacterized: The sequencing data released showed evidence of mosaicism and uncharacterized edits.
• CCR5 deletion has known risks: CCR5 loss increases susceptibility to West Nile virus and may increase severity of influenza — trading one risk for others in children who never consented.
• Consent was inadequate: The parents were not adequately informed of the risks, and the procedure was conducted outside any regulatory oversight.
• The children cannot un-be edited: They carry permanent changes they never chose.

He Jiankui was sentenced to three years in prison by a Chinese court in 2019. The International Commission on the Clinical Use of Human Germline Genome Editing (convened by national academies of science) concluded in 2020 that germline editing should not proceed to clinical use until safety can be established and broad societal consensus exists — criteria not yet met.

The He Jiankui case established germline editing as the defining ethical flashpoint of the CRISPR era. It is also why genomics researchers working on therapeutic applications of CRISPR are intensely attentive to regulatory and ethical frameworks.

---

Check yourself

1. A researcher uses CRISPR-Cas9 to introduce a frameshift in exon 3 of a gene in HEK293 cells (a dividing cell line). The experiment works — 70% of alleles show indels. She then tries the same approach in primary neurons (post-mitotic, non-dividing). She gets only 5% editing. What molecular explanation accounts for the difference? What tool might she use instead to achieve higher editing efficiency in neurons?

2. A patient with sickle cell disease asks why Casgevy doesn't directly fix their HBB mutation. Instead it disrupts BCL11A. What is the biological logic of this indirect strategy? What would be required to directly correct HBB — and what makes that harder?

3. A pharmaceutical company develops a new CRISPR therapy using a gRNA with high predicted off-target scores at three sites in the genome, but the actual off-target editing frequency at those sites in cell culture is below detection. An FDA reviewer says this is insufficient evidence of safety. Why might the reviewer be right, and what additional experiments would address the concern?

4. He Jiankui's stated goal was to protect the twins from HIV. List four specific scientific or ethical objections to his experiment — distinct from "he broke the rules."

---

Key facts to remember

• CRISPR-Cas9: bacterial immune system repurposed as programmable DNA scissors; guide RNA provides specificity, Cas9 makes the cut; PAM (NGG for SpCas9) required adjacent to target
• DSB repair: NHEJ = fast, imprecise, indels → gene knockout; HDR = precise, requires template, restricted to dividing cells
• Base editors: C→T (CBE) or A→G (ABE) without DSB; covers ~30% of pathogenic point mutations
• Prime editors: any base change/small indel via reverse transcription; no DSB, no exogenous template
• CRISPRi/CRISPRa: dCas9 fused to repressors/activators modulates expression without editing
• Delivery: ex vivo (blood cells) is most clinically advanced; LNPs efficient for liver; other tissues remain challenging
• Off-target edits: real risk; mitigated by high-fidelity Cas9, paired nickases, transient delivery, careful gRNA design
• Casgevy (December 2023): first approved CRISPR therapy; sickle cell disease / beta-thalassemia; $2.2M/patient
• Somatic editing ≠ germline editing: germline edits are heritable and permanently alter the human germline
• He Jiankui (2018): first human germline editing births; condemned by scientific community; sentenced to 3 years prison

---

Primary sources & references

• Jinek, M. et al. (2012). "A programmable dual-RNA–guided DNA endonuclease in adaptive bacterial immunity." Science, 337, 816–821.
• Komor, A. C. et al. (2016). "Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage." Nature, 533, 420–424.
• Anzalone, A. V. et al. (2019). "Search-and-replace genome editing without double-strand breaks or donor DNA." Nature, 576, 149–157.
• Frangoul, H. et al. (2021). "CRISPR-Cas9 gene editing for sickle cell disease and β-thalassemia." NEJM, 384, 252–260.
• Lander, E. S. et al. (2019). "Adopt a moratorium on heritable genome editing." Nature, 567, 165–168.
• International Commission on the Clinical Use of Human Germline Genome Editing (2020). Heritable Human Genome Editing. National Academies Press.