Episode Summary
Executive Summary: Peter Attia interviews MIT neuroscientist Feng Zhang about the origins and mechanics of CRISPR, how it transformed gene editing from cumbersome protein engineering into programmable RNA-guided editing, and where the field stands clinically. They cover optogenetics, bacterial immunity, delivery bottlenecks, therapeutic uses in blood, liver, and eye diseases, AI’s role in protein engineering, and the ethics of germline editing.
Main Topics: Feng Zhang’s scientific path and optogenetics (Priority: 5/5): Zhang describes his training with Karl Deisseroth at Stanford, where optogenetics used algal light-sensitive proteins delivered by viral vectors to control neurons with high temporal precision. This work exposed a key limitation: precise genomic targeting was needed for broader and more therapeutic applications. CRISPR’s bacterial origins and mechanism (Priority: 5/5): The conversation traces CRISPR from repetitive bacterial DNA sequences to the discovery that spacer segments came from viruses, and then to Cas proteins and guide RNA forming an adaptive immune system. Cas9 uses a PAM-dependent recognition system to cut invading DNA while avoiding self-targeting. Why CRISPR beat zinc fingers and TALENs (Priority: 5/5): Zhang explains that earlier gene-editing systems were powerful but difficult to engineer, slow to build, and cumbersome because each target required custom protein design. CRISPR’s RNA-guided programmability made it far easier, faster, and more scalable. Therapeutic gene editing and delivery (Priority: 5/5): They discuss how CRISPR can inactivate harmful genes or, with more advanced methods, precisely rewrite DNA. The main bottleneck is now delivery: getting the right payload into the right cells in vivo, especially outside the liver and eye. Clinical applications and limitations (Priority: 4/5): Examples include sickle cell disease, liver disorders, Huntington’s disease, and eye diseases such as LCA2/LCA10. The discussion emphasizes that some conditions are already treatable or in trials, but full restoration is limited by delivery efficiency, tissue biology, and the difficulty of precise insertion or regeneration. AI, smaller Cas proteins, and next-generation tools (Priority: 4/5): Zhang highlights AI advances like AlphaFold2 for protein structure prediction and the search for smaller Cas systems that fit better into delivery vehicles. He also discusses Cas13 for RNA targeting and diagnostics, especially during COVID. Ethics of germline editing and human enhancement (Priority: 5/5): The transcript revisits the controversial CCR5 embryo-editing case in China and the broader debate over germline modification. Zhang supports treating serious disease when safe and necessary, but warns that enhancement applications raise major scientific, ethical, and societal concerns.
Key Arguments: CRISPR became transformative because it replaced custom protein engineering with a programmable RNA guide, making gene editing much easier to design and deploy. The original bacterial CRISPR system is an adaptive immune defense: bacteria capture viral DNA snippets, store them as spacers, and use Cas proteins plus guide RNA to cut matching viral genomes on reinfection. Earlier tools like zinc finger nucleases and TALENs worked, but were too cumbersome, slow, and unreliable for broad use compared with CRISPR. The biggest current barrier is not the editing chemistry itself but delivery to the correct tissue and cell type in the body. Many monogenic diseases are now plausible targets for gene editing, especially in the liver, blood, and eye, where delivery is more tractable. For some diseases, knockout strategies are easier than precise correction; for example, sickle cell therapy can work by turning on fetal hemoglobin rather than directly fixing the causal mutation. Base editing and prime editing expand the toolbox beyond double-strand breaks, enabling more precise nucleotide changes. Cas13 extends CRISPR to RNA targeting and has been useful for diagnostics because it can detect viral RNA in simple assay formats. AI is accelerating protein engineering and structure prediction, which may help design smaller, more effective CRISPR systems and other therapeutic proteins. Germline editing remains ethically fraught: even if technically possible, the field lacks sufficient safety, specificity, and societal consensus for broad use, especially for enhancement rather than disease treatment.
Data Points: PhD completion year: 2009 - Zhang finished his PhD at Stanford before moving into Harvard/MIT work. MIT/Broad start year: 2011 - He began at MIT and the Broad Institute in 2011. CRISPR spacer length: ~30 letters - During first infection, bacteria can capture a viral DNA snippet of about 30 nucleotides. Human genome size: ~3 billion letters - Used as context for why 18-base recognition can be unique in the genome. Number of known genetic mutations causing disease: more than 5,000 - Zhang cites this as the set of mutations with direct causative roles in disease. Cas9 protein length: 1,300 amino acids - Discussed as a delivery challenge because it is large for viral vectors. Cas13 protein length: ~1,000 amino acids - Also large, though smaller than Cas9, and used for RNA targeting. ISCB protein length: ~450 amino acids - An ancestral Cas-like protein Zhang mentions as a smaller scaffold with a larger guide RNA. Mouse gestation period: 21 days - Used to illustrate how quickly CRISPR can generate transgenic mice once embryos are edited. Liver editing efficiency: 80-90% - Zhang says lipid nanoparticle delivery to liver can be quite robust. Laboratory editing efficiency in some cell settings: approaching 100% - He notes ex vivo editing of harvested cells can be highly efficient. Eye disease gene therapy outcome: some light sensitivity restored - Laxterna for LCA2 can help patients move around in a room with large obstacles. Sickle cell therapy cost estimate: tens of thousands of dollars initially - Zhang speculates on future cost for one-time gene-editing treatment. Current PCSK9 drug cost comparison: about $6,000/year - Attia notes current PCSK9 inhibitors are now around this annual price, down from about $15,000/year. Original PCSK9 drug cost: about $15,000/year - Used as a benchmark for comparing one-time gene-editing economics.
Pivotal Quotes: "CRISPR is really a brilliant acronym. And so C-R-I-S-P-R stands for exactly how these repeats look." — Feng Zhang: Explaining the naming and structure of clustered, regularly interspaced, short palindromic repeats. "The promise is that CRISPR is like the smartphone. You can load software onto it to recognize different genes." — Feng Zhang: Comparing CRISPR’s programmability to earlier protein-based gene-editing systems. "The bottleneck is how do we put these really powerful payloads into the right cells, in the right tissue in the body?" — Feng Zhang: Summarizing the main remaining challenge in therapeutic gene editing.
Implications: CRISPR is already reshaping biology, but the next leap depends on better delivery, smaller editors, and safer precision. Near-term gains are strongest for blood, liver, and eye diseases; germline enhancement remains scientifically immature and ethically contentious.
About Peter Attia Drive
Expert insight on health, performance, longevity, critical thinking, and pursuing excellence. Dr. Peter Attia (Stanford/Hopkins/NIH-trained MD) talks with leaders in their fields.