Radically different
A cup of coffee can mean many things: a daily part of our routine, a moment of calm, a midday boost. But zoom in past the steam, past the roasted aromatics, down to the caffeine molecule itself, and a different story emerges. At the Texas A&M Health Institute of Biosciences and Technology, researchers have turned this everyday stimulant into something far more unexpected: a molecular “pause button” for engineered cells.
In a study published in the Journal of the American Chemical Society (JACS), the team unveiled CODS, a caffeine‑operated dissociation system built using AI‑guided de novo protein design. The paper, “AI‑Guided De Novo Design of a Caffeine‑Induced Protein Dissociation System,” describes how the group reprogrammed “an existing caffeine-responsive chemically induced proximity (CIP) module into a ligand-dependent dissociation system.”
“AI is changing how we design biology,” said senior author Yubin Zhou, MD, PhD. “Instead of relying only on protein parts that already exist in nature, we can now design new mini proteins with specific behaviors. Here, we used AI to help turn caffeine into a precise trigger for controlling engineered cells.”
The CODS system pairs a caffeine‑binding protein with a synthetic mini‑binder designed using the BindCraft platform. In the absence of caffeine, the two components stay locked together. Add caffeine, and the complex snaps apart, releasing the binder and shutting down the attached cellular function. As Tianlu Wang, PhD, a postdoctoral fellow in the Zhou lab, put it, “Many genetically-encoded molecular tools act like accelerators. CODS gives us something closer to a brake or pause button.”
The team demonstrated CODS across several biological contexts. In engineered gene circuits, caffeine addition sharply reduced transcriptional activity. In a rewired pyroptosis pathway, caffeine triggered inflammatory cell death by freeing the active domain of gasdermin D. And in perhaps the most translational example, CODS served as a conditional deactivator for CAR T cells, temporarily dampening their activity without destroying the therapeutic cells.
“Powerful therapies need powerful control,” Zhou said. “By combining AI‑designed proteins, high‑performance computing, and familiar small molecules, we are building a new language for communicating with engineered cells.”
The design process itself leaned heavily on computation. Graduate student Brendan McKee led the AI‑guided binder design and molecular modeling, while Tatsuki Nonomura spearheaded the molecular engineering and live‑cell validation. The Texas A&M High Performance Research Computing service provided the infrastructure needed to run large‑scale simulations. “High‑performance computing was essential for this project,” Zhou noted. “It helped us move from a conceptual idea to a functional molecular switch much faster.”
Although caffeine is not a therapeutic molecule, its safety and familiarity make it an appealing control signal. As Zhou emphasized, “Coffee will not replace medicine. But caffeine can help us imagine medicines that are more controllable, more responsive, and safer for patients.” The researchers’ next steps include further testing in therapeutic cells, animal models, and disease-relevant settings before moving toward clinical use.
CODS now joins a growing toolkit of AI‑designed molecular switches, offering a blueprint for future systems responsive to other safe, accessible molecules. As programmable cell therapies advance, the ability to modulate them with something as simple as caffeine may prove unexpectedly powerful.
The authors report that a patent application covering the CODS platform has been filed by Texas A&M University, with Y.Z., T.N., B.M., and T.W. listed as inventors (U.S. Provisional Patent Application No. 64/022,078).
The post AI Reimagines Caffeine as a Molecular Off‑Switch for Engineered Cells appeared first on GEN – Genetic Engineering and Biotechnology News.
Legal tech startup BrentWorks reports that it has launched CiteSentinel, a dedicated platform built specifically to detect and prevent AI hallucinations in legal citations, including those related to biotechnology. The tool scans legal documents and flags case law, statutes, and legal authorities that may be fabricated, misstated, or otherwise erroneous, before they reach a judge, according to the company.
Courts across the country are increasingly sanctioning attorneys who submit briefs containing invented case citations, a well-documented byproduct of generative AI drafting tools that produce authoritative-sounding, but entirely fictional, legal authority, says BrentWorks co-founder Brent Britton, a technology attorney and MIT-trained engineer. CiteSentinel was designed to close that verification gap, giving attorneys a fast and easy way to confirm that every citation in a filing corresponds to a real case, a real statute, and a real legal authority, he adds.
“The legal profession is learning, in very public ways, that AI doesn’t just make mistakes, it confidently lies to your face,” continues Britton. “CiteSentinel is about restoring trust. It lets lawyers move fast with the irresistible efficiencies of generative AI while still filing documents reciting authorities they can stand behind. It also enables them to scan opposing counsel’s documents, giving them a competitive edge in the courtroom.”
Many attorneys who do not personally use AI to draft documents are discovering they have a problem anyway, Britton points out. Opposing counsel may have used AI. Co-counsel may have. Contract attorneys and paralegals almost certainly have access to it and may be using it without disclosing that fact. When a brief containing fabricated citations reaches the court, the question of who drafted it quickly becomes secondary to the question of whose name is on it, he explains.
CiteSentinel lets attorneys scan any document, their own, a colleague’s, or an adversary’s, for citation errors before those errors become their problem, notes Britton. Attorneys who review opposing counsel’s filings with CiteSentinel gain an additional advantage: the ability to identify and challenge citations to authorities that simply do not exist, he says.
Unlike traditional research platforms that focus on finding more information, states Britton, CiteSentinel was created to confirm that the law cited in a document is real. Attorneys can scan:
BrentWorks’ other co-founder is Brent Hunter, a technologist who applied neural networks to finance in 1993. He cites CiteSentinel as the first in a series of products the company will be releasing for the practice of law in the age of AI.
Both BrentWorks’ co-founders agree that AI hallucinations pose particular risks in biotechnology-related legal matters because cases often depend on highly technical evidence, including patent claims, prior art, clinical trial data, FDA regulatory history, scientific publications, expert witness testimony, freedom-to-operate analyses, and licensing agreements. In this context, an AI system could invent scientific references that do not exist, mischaracterize FDA guidance documents, fabricate patent precedents, incorrectly summarize clinical trial results, or generate inaccurate prior-art searches. Such errors can undermine legal arguments, regulatory submissions, and intellectual property strategies.
“Biotech litigation is where AI hallucinations turn genuinely dangerous. You have a system trained to sound authoritative now injecting phantom patent precedents and counterfeit clinical data into documents that determine whether a drug reaches patients or a patent survives challenge,” explains Brent Britton. “In this domain, where the technical record is everything, a ghost FDA guidance document or a fabricated prior art reference can unravel an entire legal strategy and years of work along with it. The law has always been a high-stakes information game, and right now the machines are playing it with synthetic cards.”
As CiteSentinel expands beyond just case citation verification, “it will be the truth layer that keeps all players honest,” he predicts.
The post CiteSentinel Launched to Detect and Prevent AI Hallucinations in Legal Citations appeared first on GEN – Genetic Engineering and Biotechnology News.
When Jingkun Zeng, PhD, joined the lab of Nobel laureate, Jennifer Doudna, PhD, as a postdoctoral researcher in 2024, he was not interested in applying CRISPR for gene editing.
The molecular scissors had demonstrated extraordinary clinical promise in correcting single-point mutations, most strikingly in Baby KJ’s case, where a rare metabolic disorder once presented a 50% mortality rate in infancy.
Yet, Zeng had his ambitious sights on stopping cancer progression, where the biology “became messy.” Cancer can be driven by hundreds of thousands of mutations, making it nearly impossible to correct each mutation one-by-one to restore healthy function.
Zeng, who completed his PhD training in cancer evolution at The Francis Crick Institute, aimed to develop new CRISPR-based technology that could therapeutically access the undruggable tumor suppressor protein, p53. Mutations in this “guardian of the genome” are found in nearly half of all cancers, and up to 70–90% of cases of the most deadly tumors, including ovarian, pancreatic, and non-small cell lung cancer.
In a new study published in Nature titled, “Targeting Cancer-Specific Mutations with RNA-Triggered Chromatin Shredding,” Zeng and colleagues from Innovative Genomics Institute (IGI), University of California (UC) Berkeley, UC San Francisco (UCSF), and Gladstone Institutes, have now engineered a CRISPR system to selectively trigger cancer cell death by chromatin shredding.
The approach recognizes cancer cells using the RNA-guided nuclease, CRISPR-Cas12a2, to recognize mutant p53 mRNA transcripts. Therapeutic effectiveness was demonstrated in mouse models of lung and liver tumors.
Mutations in p53 are early drivers in the cancer-causing cascade, making the tumor suppressor one of the most sought-after targets in cancer therapy. Yet despite decades of effort, no approved p53 drugs exist on the market.
Unlike many druggable proteins, p53 lacks a well-defined binding pocket traditionally required by established modalities, such as small molecules or antibodies. Additionally, most cancer therapeutics are designed to inhibit disease-driving proteins, whereas restoring p53 function demands precise, controlled activation of a tumor suppressor.
“It’s the first time we managed to target p53 with such precision,” Zeng told GEN, emphasizing that CRISPR-Cas12a2 can distinguish healthy and disease cells that differed by just one nucleotide.
The novel drug modality takes advantage of CRISPR’s bacterial roots as a defense system that protects against infection by cutting the genetic material of invading viruses, preventing replication and spread.
Zeng also emphasizes that the guide RNA is easily programmable for additional therapeutic areas, such as destroying viral infected cells or abnormal cells due to aging. The technology can also be multiplexed to recognize multiple cancer mutations simultaneously.
The work joins a growing industry effort to develop scalable and generalizable genetic medicines.
Looking ahead, the authors aim to improve the delivery efficiency to cancer cells, a longstanding challenge across CRISPR therapies. The team is also undergoing collaborations to apply the technology across diverse cancer types, including brain, prostate, and ovarian cancer.
The post CRISPR Shreds Undruggable Cancer Cells with Precision appeared first on GEN – Genetic Engineering and Biotechnology News.
Want to stay on top of the science and politics driving biotech today? Sign up to get our biotech newsletter in your inbox.
Gene therapies can now cure disease, but America still hasn’t figured out how to pay for them. Is there a better way? Also, Novo executives are downplaying a once-coveted obesity drug, Incyte is making a $2 billion blood disease bet, and new data suggest Boehringer’s obesity contender may struggle to stand out in an increasingly crowded market.
Revolution Medicines’ experimental pancreatic cancer drug has been the star of the oncology field in recent weeks, with new data showing the medicine produced unprecedented outcomes for patients.
Nature Biotechnology, Published online: 08 June 2026; doi:10.1038/s41587-026-03171-8
We have developed single-cell spatial pharmacobiology (SSP), which combines in situ imaging of a systemically infused fluorescent therapeutic antibody with high-plex spatial proteomics. Applied to head and neck and pancreatic tumors from patients treated in phase 1 trials, SSP revealed marked spatial heterogeneity in antibody delivery and target engagement, which was shaped by conserved stromal barriers.
Detailed data from a mid-stage study offered further evidence that the obesity drug Pfizer acquired from the biotech Metsera could be dosed monthly. But it’s not clear how competitive the treatment would be against weekly injectables on the market and in development that may lead to greater weight loss.
In the study, called VESPER-3, patients with obesity took weekly doses of the drug, called berobenatide, for 12 weeks and then transitioned to higher monthly doses out to 28 weeks. By then, patients lost up to 12.1% of their weight, when analyzing just those who stayed on treatment, as Pfizer previously reported.
New data presented at the annual meeting of the American Diabetes Association on Saturday show that when patients transitioned from weekly to monthly dosing, the rate of weight loss continued at a similar pace; they had not yet hit a plateau by 28 weeks. That’s a promising sign, but the rate of weight loss at 28 weeks was still less than what was seen at a similar time point in the pivotal trial of Eli Lilly’s Zepbound.
NEWS RELEASE: Cadence earns second consecutive EcoVadis silver medal for sustainability STAUNTON, Va — Cadence, Inc., a leading provider of vertically integrated contract manufacturing solutions to the MedTech and Pharma markets, proudly announces it has earned a Silver Medal from EcoVadis for the second consecutive year. EcoVadis is the world’s most trusted provider of business sustainability…
The post Cadence earns second consecutive EcoVadis silver medal appeared first on Medical Design and Outsourcing.
A study tracking thousands of B cells across more than 100 germinal centers (GCs) in mice has revealed how the system consistently produces highly effective antibodies. The findings overturn longstanding ideas about how germinal centers function, revealing that they are far more selective than once thought, and challenge the idea that antibody improvement is driven mainly by rare growth “bursts” among the most successful B cells. The discovery could have implications for immune cell evolution, and ultimately guide the design of vaccines against rapidly mutating pathogens like influenza. It could also lead to new ways of studying evolution itself.
“The traditional, mechanistic view of germinal centers is to think of them as selection machines sorting out the best antibodies,” said research lead Gabriel D. Victora, PhD, head of the Laboratory of Lymphocyte Dynamics at The Rockefeller University. “But when you look very, very closely, you see a process that’s almost essentially random—a little bit better than a coin toss—which repeats many times until the immune system arrives at the right answer consistently. That’s much more akin to how evolution operates than the way a machine does.”
Victora and colleagues reported on their findings in Cell, in a paper titled “Replaying germinal center evolution on a quantified affinity landscape.”
Inside germinal centers, B cells rapidly mutate and compete to produce antibodies that bind successively better to pathogens. “Darwinian evolution of immunoglobulin genes within germinal centers (GCs) underlies the progressive increase in antibody affinity following antigen exposure,” the authors wrote. That puts B cells under intense pressure to optimize a single trait: binding affinity, or how well an antibody recognizes its target.
But how they accomplish that feat has very much remained an open question, the team noted. “Whereas the cellular mechanics of how competition between B cells increases affinity are well established, the evolutionary dynamics of this process are less clear.” Because weak and strong B cells often coexist side by side in the germinal center, scientists have long wondered whether the immune system temporarily preserves weaker cells in case they later acquire useful mutations. The phenomenon of clonal bursts, in which the descendants of a single B cell rapidly take over an entire germinal center, are also poorly understood.
The authors explained that GC B cells evolve by rapidly mutating only two Ig genes, which are the heavy chain (Igh) and light chain (either Igk or Igl). Victora’s team engineered mice in which all competing B cells began with the same antibody sequences, allowing them to replay a single evolutionary process across more than 100 germinal centers at once. “… we established a system in which GCs are composed entirely of B cells carrying the same pre-rearranged Igh and Igk genes, ensuring identical starting specificity and affinity,” they explained. Victora added, “We simplified it to the bare bones, and asked how repeatable is the exact sequence of mutations that leads to stronger antibodies.”
Once each of the B cells was primed with the exact same unmutated antibody sequence, the team triggered germinal center formation through immunization. They then tracked the resulting sprint toward immune efficiency with multiphoton microscopy and laser-based photoactivation, and sequenced thousands of individual B cells across 119 germinal centers.
With this data, the team managed to construct a detailed family tree that mapped how different lineages of B cells had developed. They also built a mutational dictionary, using deep mutational scanning (DMS), a technique that links almost every possible amino-acid change to antibody performance. This advance allowed the team to determine how mutations affected binding strength and structural stability simply by reading a cell’s DNA sequence.
“DMS was the big technical advance here,” says first author Ashni Vora, PhD, a graduate fellow in the lab. “With it we could determine the affinities of thousands of cells just by looking at their sequence, without having to produce an antibody.”
The researchers compare the resulting picture to a casino game. Watching a single B cell evolve inside a germinal center looked almost random, with some cells rapidly expanding, others disappearing, and even promising mutations failing as if random chance ruled the day. Some germinal centers were overtaken by clonal bursts while others contained many competing lineages with no clear winner. The differences had little to do with affinity or merit. “We find that, even in this simplified setting, GC selection yields widely divergent tree topologies, ranging from clonal-burst-type structures to multi-pronged GCs where multiple line ages evolve in parallel,” they noted.
But the team discovered that the germinal center game is rigged. In a casino, the house always wins not because of the odds on any individual game, but because a slight statistical bias is built into the system and repeated thousands of times. Germinal centers appear to operate similarly. Each round of cellular competition is only slightly biased toward cells carrying beneficial mutations, and random chance means that there is often little correlation between affinity and success. But by repeating that same noisy, almost random process over and over across many germinal centers, the immune system ultimately produces stronger antibodies.
“If you see someone get a jackpot, you might wonder how the casino makes money,” Victora says. “The answer is that the casino puts in a little bit of bias, so that you win and you lose, but on average, you lose more than you win. If there are just one or two people playing, the casino might lose money due to random chance. But if there are a thousand people playing, it’s going to average out and the house wins. That’s essentially how germinal centers work.”
The researchers also found that the immune system favors mutations that are easiest for its cellular machinery to generate, rather than the mutations that would produce the strongest antibodies. And by tracking B cell lineages over time, they also showed that germinal centers are far more selective than previously thought, rapidly eliminating inferior B cells. “By combining phylogenetic reconstructions with a fitness landscape inferred from populations sampled over time, we show that both the apparent permissiveness of GCs to low-affinity lineages and the apparent early plateau in affinity maturation are best explained by survivorship biases that distort the histories of lineages present at sampling,” the investigators wrote in summary.
Taken together, the findings overturn several longstanding ideas about how germinal centers function and may provide new tools for vaccine developers hoping to steer antibody evolution against influenza and HIV. “What was once theoretical speculation about what must happen in the germinal center, we are now showing in action—the real thing,” Victora says.
At the same time, this work also illustrates how germinal centers could become a powerful model for studying evolution more broadly. Scientists have long relied on bacteria grown in the lab over many generations to plumb the depths of evolutionary biology and determine how much of evolution is driven by random chance. In clarifying the rules governing germinal centers, the researchers revealed why the immune system could offer a potentially more tractable experimental avenue: Unlike bacterial evolution, which centers around adapting to many possible survival strategies, B cells are all aiming for the same target. “I see this as an opening salvo in a longer effort to understand evolution by using the immune system as a model,” Victora added.
The post How Germinal Centers Generate Antibodies Through Noisy Rounds of Mutation and Selection appeared first on GEN – Genetic Engineering and Biotechnology News.
Researchers working to advance genome engineering in crops face many challenges, including simultaneously introducing diverse genome edits. Although a major goal of modern crop breeding is to efficiently combine multiple desirable traits by “stacking” the favorable alleles that contribute to those traits in a single crop variety, current strategies are time-consuming and inefficient.
Now, a team led by Caixia Gao, PhD, professor at the Institute of Genetics and Developmental Biology of the Chinese Academy of Sciences, has developed a genome engineering platform that allows multiple trait stacking in crops by combining gene knockout, precise sequence editing, and chromosome engineering within a single framework. The advance is “a twin prime editing-based knockout (TKO) system that installs stop codon clusters (SCCs) for precise translational termination with minimal in-frame mutations.” TKO achieved knockout efficiencies of up to 70.5%, 58.6% and 75.1% in rice, maize, and wheat protoplasts, respectively.
This work was published in Nature Biotechnology in the article, “Multiplexed, precise genome engineering in monocots with twin prime editing systems.”
The researchers first developed a precise and efficient gene knockout tool called twin prime editing (twinPE)-mediated gene knockout (TKO), which precisely inserts a small fragment containing a stop codon cluster at the target site. TKO achieves predictable gene disruption through precise installation of stop codons, avoiding in-frame indels caused by insertions or deletions in multiples of three nucleotides, which are often seen in Cas9 systems.
In protoplasts, TKO demonstrated efficient knockout capabilities in monocot crops such as rice, wheat, and maize. In regenerated T0 rice plants, the average efficiency for single gene knockout reached 96.8%.
To eliminate cross-editing between different loci and to achieve precise, safe multiplex gene knockout, the researchers developed 10 orthogonal TKO systems, enabling efficient simultaneous knockout of up to 10 genes. Unlike Cas9-mediated multiplex editing, which can lose effectiveness because in-frame mutations accumulate across multiple targets, the orthogonal TKO systems maintain high knockout efficiency even when multiple genes or homologous gene copies are edited simultaneously.
Building on TKO, the researchers then developed two integrated genome engineering platforms, TRIM1 and TRIM2—forming a unified platform known as TRIM.
TRIM1 combines TKO with prime editing-based sequence modification, enabling simultaneous gene knockout, base substitution, insertion, deletion, duplication, and inversion within a single editing framework. In regenerated T0 rice plants, TRIM1 achieved simultaneous knockout of one gene together with homozygous precise editing of three additional targets with an efficiency of 22.8%.
TRIM2 incorporates a prime editor–Cre recombinase fusion protein and enables kilobase-scale DNA insertion, replacement, deletion, inversion, and chromosomal translocation through recombinase-assisted genome engineering.
Unlike existing genome editing tools that typically perform only a limited number of sequence modifications, TRIM integrates gene knockout, small-scale precise sequence editing, and large-scale chromosome engineering into a single platform. This “all-in-one” platform provides a powerful way to rapidly stack multiple favorable alleles, thus enhancing precision breeding of complex traits in monocot crops.
The post Twin Prime Editing Enables Rapid Trait Stacking in Crops appeared first on GEN – Genetic Engineering and Biotechnology News.