What happens when you photocopy a photocopy 58 times? If you're a mouse, apparently, you stop existing. This week, we've got the world's longest cloning experiment finally hitting a wall, a single tobacco plant producing psilocybin and DMT (because why pick just one), a massive clinical trial that gave Mars Inc. exactly the marketing ammo it wanted, and Chinese scientists trying to turn your garden into a nightlight. Let's go.

RESEARCH

Picture this: it's January 2005, and Teruhiko Wakayama (the guy who cloned the first mouse from an adult somatic cell back in 1997) decides to answer the dumbest question in biology - what happens if you just keep cloning the clone of a clone of a clone? Twenty years, over 30,000 nuclear transfer attempts, and more than 1,200 cloned mice later, he has an answer. And it's not great.

The experiment, published in Nature Communications, started with a single female mouse and used somatic cell nuclear transfer to produce generation after generation. For a while, things actually improved - success rates peaked at 15.5% around generation 26. Then the wheels started coming off. By generation 57, the success rate had crashed to 0.6%. At generation 58, every single pup died within a day of birth. No visible abnormalities. They just... stopped working.

Here's what they actually figured out: the problem wasn't epigenetic drift, which is what everyone assumed. Telomeres didn't shorten significantly. Instead, the clones were accumulating genomic mutations at roughly three times the normal rate - about 69 new single-nucleotide variants per generation versus 22 in sexually reproducing mice. By generation 57, that's roughly 3,700 mutations, 80 indels, and up to 84 structural variants piling up with no way to shake them off. It's the first direct evidence of Muller's Ratchet (the idea that asexual lineages inevitably accumulate lethal mutations) in any mammal.

The silver lining? When late-generation clones were allowed to mate naturally, their grandchildren partially recovered normal litter sizes - from 2.2 pups back up to 7.0. Sexual reproduction, turns out, is nature's error-correction protocol. Though Wakayama put it more bluntly: "mammals cannot sustain their species through cloning." Bad news for anyone who was planning an infinite army of themselves.

RESEARCH

Researchers at the Weizmann Institute of Science just turned a tobacco relative into arguably one of the most interesting plants on Earth. They engineered Nicotiana benthamiana to simultaneously produce five psychedelic tryptamines - psilocybin, psilocin, DMT, bufotenin, and 5-MeO-DMT - sourced from mushrooms, plants, and toad venom. That's three biological kingdoms in one leaf. As senior author Asaph Aharoni told journalists: "In one leaf, we get five different psychedelics from three different kingdoms."

The team, led by co-first authors Paula Berman and Janka Höfer, introduced genes from across plants, fungi, animals, and bacteria via Agrobacterium-mediated transient expression (basically a temporary gene delivery system that means the modifications can't be passed through seeds or pollen). They chose N. benthamiana because it naturally overproduces tryptophan, the amino acid precursor all five compounds share.

The coolest technical detail? They used AlphaFold to predict the 3D structure of a key enzyme, identified a structural clash in its active site, and designed a single amino acid swap that boosted 5-MeO-DMT production 40-fold. AI-assisted drug farming. We live in the future.

To be fair, yields vary - DMT reached 89 micrograms per gram of fresh leaf, which is respectable but still below some natural producers, and all five pathways compete for the same tryptophan pool when running simultaneously. Independent experts like Miami University bioengineer Andrew Jones called it "exciting work," but noted industrial fermentation will probably scale better. Still, the paper also produced halogenated psychedelic analogs not found in nature - compounds in the same class have shown antidepressant-like activity in prior mouse studies. That's the real play here: not growing drugs in your garden, but building a platform to discover entirely new ones.

NEWS & RESEARCH

Every few months, someone shares an article claiming chocolate is a health food, and you want to believe it so badly you don't read past the headline. The COSMOS trial - the largest randomized trial ever conducted on cocoa flavanols - is exactly the kind of study that generates those headlines. And the headlines are, predictably, wrong.

Here's what actually happened. 21,442 older Americans (women 65+, men 60+) took either 500 mg of cocoa flavanols daily or a placebo for a median of 3.6 years. The trial was run by Howard Sesso and JoAnn Manson at Brigham and Women's Hospital. The primary endpoint - total cardiovascular events - came back statistically non-significant (HR 0.90, 95% CI 0.78-1.02, P = 0.11). In plain English, cocoa flavanols did not significantly reduce heart attacks, strokes, or other cardiovascular events.

But here's the number you'll see in every "chocolate is healthy" article: a 27% reduction in cardiovascular death. Sounds impressive, right? Except that was a secondary outcome, not what the trial was designed to test. The confidence interval was wide (HR 0.73, CI 0.54-0.98), and the cocoa extract's cognitive function results were even clearer - no significant effect on global cognition, memory, or executive function over two years. The one exception? People with poor baseline diets saw some memory benefit, which tells you more about eating vegetables than about eating chocolate.

Now for the part that should make you pause. COSMOS was co-funded by Mars Inc., the company behind M&Ms, Snickers, and Milky Way. Mars provided the cocoa extract supplements. Mars's Chief Science Officer Hagen Schroeter co-authored the trial's design paper and publicly stated he was "leading Mars's scientific contributions." The main results paper claims Mars had no role in data analysis or manuscript preparation, but that's a hard sell when your employee helped design the study. Meanwhile, Mars's supplement brand CocoaVia now leads with an even bolder 39% reduction figure from per-protocol analysis on its product pages, while the failed primary endpoint gets buried. As nutrition researcher Marion Nestle wrote: Mars got its money's worth from what was a very expensive study. This is fine.

NEWS

A Chinese biotech startup called Magicpen Bio made its high-profile international debut at the Zhongguancun Forum 2026 in Beijing, showing off genetically engineered glowing plants and headlines that wrote themselves: "Avatar-like plants could light cities without electricity." The plants glow continuously without UV charging or electricity, needing only water and fertilizer. Founder Li Renhan, a graduate of China Agricultural University, has modified over 20 plant species, including orchids, sunflowers, and chrysanthemums, using genes from both fireflies and luminous fungi.

Commercially, they're already moving. Their "Star Firefly Night Flower" brand of glowing plants launched in early 2025 and sold over 10,000 units within a week at about $10-13 each. The company claims brightness an order of magnitude greater than previous bioluminescent plant efforts.

Now for the cold water. Magicpen Bio has published no peer-reviewed papers and released zero quantitative brightness data - no lumen, lux, or candela measurements. Street lighting delivers tens to hundreds of lux at the footpath level. Bioluminescent plants produce a soft ambient glow that's pretty, but not functional. Cambridge plant scientist John Carr doesn't see them replacing streetlights anytime soon. The fundamental problem is metabolic: producing photons costs the plant ATP and chemical substrates that compete with growth and flowering. Making enough light to read by would probably kill the plant.

For comparison, the US company Light Bio already sells USDA-approved glowing petunias for $29, using fungal bioluminescence genes from Neonothopanus nambi mushrooms (a more metabolically efficient system than firefly genes). Light Bio offered 50,000 plants in its initial 2024 batch and was named one of TIME's Best Inventions of 2024. So glowing houseplants? Absolutely real. Glowing city lighting? Maybe don't cancel the power grid just yet.

This week, we learned that biology has hard limits (sorry, cloners), creative workarounds (thank you, trippy tobacco), uncomfortable conflicts of interest (looking at you, Mars), and a whole lot of unfounded optimism (glowing streetlights, really?). The through-line? The gap between what a study actually shows and what the headline claims is roughly the same distance as between a glowing houseplant and a functioning streetlamp.

If you enjoyed this edition, forward it to someone who still thinks eating a Snickers counts as preventive cardiology. And if you've got opinions on which kingdom should contribute the sixth psychedelic, hit reply. We're placing bets on archaea.

Keep questioning everything (especially secondary endpoints funded by candy companies),

Prateek & Jere

P.S. Somewhere in Hefei, a scientist is genuinely pitching "what if we just grew more plants" to a city planning committee, and that's the most optimistic thing we've heard all month.