The first study concerns one of the oldest puzzles in genome biology: how genes are switched on by enhancers, DNA sequences that often sit tens of thousands of base pairs away from the genes they control. For more than four decades, researchers have struggled to explain how a regulatory element so distant can communicate with its target promoter, the region that acts as the gene’s on switch. A new study led by first author Lingling Cheng, PhD, a postdoctoral researcher in the lab of senior author Alexandros Pertsinidis, PhD, a structural biologist at MSK’s Sloan Kettering Institute, offers a detailed answer. Using cutting-edge live-cell imaging at nanometer resolution, the team watched gene activity unfold in real time in mouse embryonic stem cells. The study also offers key new insights into the puzzle of transcriptional bursting, the tendency of genes to fire in short, sporadic bursts rather than steady streams.

The study showed that cohesin, a protein complex, acts as a molecular motor, physically reeling in DNA between an enhancer and its promoter. It does this by extruding a loop of DNA along the chromosome until the two regions are brought face-to-face, within just 100 nanometers of each other. “Our data suggests that using this mechanism, cohesin significantly contributes to how frequently enhancers and promoters come together,” Dr. Cheng says. “And the process is far more efficient than waiting for the two regions to bump into each other as they randomly drift through the cell nucleus — though that does also happen.” The finding also provides a direct explanation for transcriptional bursting: each time cohesin brings an enhancer and promoter together, a burst is triggered; when the regions drift apart, the burst ends. The team went further, determining how the activation signal is transmitted once the two regions meet. Clusters of proteins called General Transcription Factors, including Taf1 and Med19, sit near the enhancer, and when the loop delivers the promoter into these clusters, the gene’s on switch is boosted. Removing either Taf1 or Med19 alone had only a modest effect on gene firing, but removing both caused a dramatic drop, revealing a previously unknown redundancy that protects the system against disruption. The work was published in Cell.

The second study, published in Science, ventures into the mitochondria and names a new form of programmed cell death: baciptosis. A team overseen by senior author Emily Cheng, MD, PhD, of MSK’s Human Oncology and Pathogenesis Program, with co-first authors Shaoyuan Yang, PhD, a postdoctoral fellow at MSK, and Hyungjin Kim, PhD, a former graduate student of Dr. Cheng at Washington University who is now a faculty member at Stony Brook University, found that two proteins called BAK and BAX can come together in mitochondria in two distinct ways, each triggering a fundamentally different fate. In the familiar arrangement, the proteins release cytochrome c, which activates enzymes called caspases and triggers apoptosis, a form of cell death that typically does not provoke an inflammatory immune response. In the newly described arrangement, BAK and BAX assemble with other proteins spanning the outer and inner mitochondrial membranes, allowing mitochondrial DNA to escape into the cytosol, the fluid inside the cell.

When caspases are not activated, that mitochondrial DNA release triggers inflammatory responses and contributes to catastrophic energy failure, culminating in baciptosis and a strong response from the immune system. The researchers demonstrated that baciptosis can be induced in human cancer cells with impaired caspase activation, suggesting potential relevance to cancers carrying similar defects. The discovery expands the known roles of BAK and BAX beyond apoptosis and reveals how these proteins can regulate immunogenic cell death. Unlike apoptosis, baciptosis can stimulate immune responses. Finding ways to trigger baciptosis instead of apoptosis in cancer cells could help the immune system recognize and attack tumors and improve cancer immunotherapy, the researchers suggest.

The third study addresses a stubborn gap in human genetics: why the genetic basis of autism spectrum disorder has long puzzled scientists. Although hundreds of genomic variants have been linked to autism, they account for a genetic diagnosis in only about 15 percent of autistic individuals. A study co-led by MSK computational biologist Quaid Morris, PhD, with collaborators at the University of Toronto, King’s College London, and the Francis Crick Institute, focused on microexons, very short DNA segments that can be selectively included or excluded when cells build proteins, which prior work had implicated in the disorder. The team developed a machine learning model that decodes the regulatory rules governing microexon splicing, the process by which these tiny segments are stitched into or left out of messenger RNA. When the researchers ran their model on genetic data from thousands of autistic individuals and their neurotypical siblings, predicted splicing disruptions appeared almost entirely in the 85 percent of autistic individuals who had never received a genetic explanation for their diagnosis. That pattern suggests microexons may account for much of that missing piece. The researchers believe the model can also be applied to other neurological conditions, including schizophrenia and epilepsy, where microexons are thought to play a role. The work was published in Nature.

The fourth study harnesses artificial intelligence to solve a fundamental limitation of modern biology: single-cell sequencing can read the genetic activity of thousands of individual cells at once, but it captures only a single moment in time. Researchers from MSK and Weill Cornell Medicine built a tool called DynaVelo that uses AI to analyze these snapshots and learn the underlying rules governing cell state changes, allowing it to infer the full course of a cell’s development. The effort was overseen by senior authors Christina Leslie, PhD, a computational biologist at MSK, and Darko Barisic, PhD, of Weill Cornell Medicine, and led by first author Alireza Karbalayghareh, PhD. The model was tested on B cells, immune cells from which many lymphomas arise, tracking how gene activity and the transcription factors that control it change as cells mature. DynaVelo’s most powerful feature is its capacity for virtual experiments. The team used it to study how losing Arid1a, a protein involved in organizing chromatin and regulating gene activity, disrupts normal B cell maturation and pushes cells toward an immature fate associated with B cell lymphoma. When the model was asked which genetic changes might counteract the loss of Arid1a, it nominated three candidates: Ctcf, Stat3, and Bcl6. The scientists then built new mouse models to test these predictions, and in each case the changes restored normal B cell behavior, validating the algorithm. In silico experiments can screen thousands of possibilities that would be prohibitively slow and expensive to test in animal models. “Our findings suggest DynaVelo could help accelerate the search for new treatments by identifying promising therapeutic targets virtually,” Dr. Leslie says. The study appeared in Nature Biotechnology.

The fifth study, published in Science Immunology, reveals that a protein called STAT3 acts as a molecular switch inside natural killer cells, changing its behavior based on the severity of a viral infection. NK cells are a key part of the innate immune system’s defense against infection, and the research team, led by Simon Grassmann, MD, a postdoctoral fellow in the lab of senior author Joseph Sun, PhD, at MSK’s Sloan Kettering Institute, found that STAT3 activates entirely different sets of genes in response to a mild infection than during a severe one. During low-grade infection, STAT3 promotes expression of a protein called MYB that drives NK cell proliferation and builds long-term immune protection. During severe infection, when levels of the inflammatory molecule type I interferon are high, STAT3 instead activates a protein called BLIMP-1 that pushes NK cells toward a powerful but short-lived state to quickly eliminate the pathogen. STAT3 accomplishes this switch by repositioning itself on the genome depending on infection severity, a process governed by epigenetic changes in how DNA is packaged, which determine the accessibility of specific genomic regions, as well as by STAT3’s preference for binding to particular DNA sequences depending on context. The findings raise an important question for cancer immunotherapy: STAT3 inhibitors are currently being tested against cancer, and their effectiveness may depend on the level of inflammation within a tumor. That suggests pairing STAT3 inhibitors with other treatments that modulate tumor inflammation could make them more effective. Taken together, the five studies show that the decisive events — whether in a looping chromosome, a leaking mitochondrion, a skipped microexon, a maturing B cell, or a vigilant NK cell — are dynamic, context-dependent, and now increasingly visible to science.

Subject of Research: Five Memorial Sloan Kettering studies on gene activation, cell death, autism genetics, AI cell-fate prediction, and natural killer cell immunity

Article Title: MSK research highlights, October 22, 2026

Article References: MSK research highlights, October 22, 2026. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: cohesin, enhancers, transcriptional bursting, baciptosis, BAK, BAX, microexons, autism spectrum disorder, DynaVelo, single-cell sequencing, STAT3, natural killer cells