New Pipeline Reveals RNA Fusions Formed During Inflammation, Opens New Frontier in Biology

By Amy Reyes 

October 20, 2026 | Scientists have discovered that during inflammation, mammalian immune cells can create entirely new messenger RNAs by fusing segments from two different genes into functional chimeric mRNA. Such fused RNAs have been previously associated with DNA damage, but a new study published Sept. 2 in Nature (DOI: 10.1038/s41586-026-10982-x) shows that some protein‑coding chimeric mRNAs are not accidental events caused by disease-causing DNA rearrangements, but in fact have a functional role within the immune system to fight infection.  

Led by Ruaidhrí Jackson, PhD, assistant professor of immunology at Harvard Medical School, researchers describe how inflammation physically reshapes the DNA inside immune cells, pulling distant genes closer together. That structural shift allows distinct RNA transcripts to fuse into chimeric mRNAs, which for years were largely viewed as a byproduct of damaged DNA or malfunctioning transcription.  

Jackson’s findings challenge long‑held assumptions about how genes operate and suggest the mammalian proteome is far larger than previously understood. 

“The protein‑coding genome is actually vastly larger than we previously thought,” said Jackson. “We've found that genes, even in different chromosomes, can donate mRNA together to create hybrid RNAs and hybrid proteins. There are tons of new proteins out there that people couldn’t even look for because they didn’t know they existed, but now we can … We've shown you can actually identify these fairly well, and they're important,” he said. 

Demonstrating that healthy immune cells generate these fused RNAs during infection is a paradigm shift, revealing that a mechanism once considered an accident of disease actually plays an active, functional role in normal immune defense. By bringing distant genes closer together, an inflamed cell can create the conditions needed for transcript fusion, allowing the immune system to use chimeric mRNAs as part of its normal functional toolkit. 

The mainstream central dogma is that mammalian mRNAs and proteins originate from single genes, with variations arising from cis-splicing, which occurs when exons within the same gene are rearranged. Until now, it wasn’t clear whether two different genes, located far apart in the genome could have their RNAs stitched together through trans-splicing to create chimeric mRNA that can produce a functional protein in mammals. 

It was known that chimeric RNAs can form when chromosomes break and reattach out of order, creating abnormal gene fusions that produce hybrid mRNAs and, ultimately, fusion proteins that can act as potent drivers of oncogenic elements. For example, the BCR–ABL1 fusion in chronic myeloid leukemia produces a constantly active tyrosine kinase that promotes uncontrolled cell proliferation. Additional gene fusions TMPRSS2–ERG in prostate cancer and EML4–ALK in lung cancer generate chimeric proteins that promote tumor development.  

A Hidden Layer of Transcript Diversity 

As a postdoctoral researcher at Yale University, Jackson used ribosome profiling to show that several RNAs labeled as “non‑coding” were in fact being translated into proteins, revealing that long‑standing rules for identifying protein‑coding genes were incomplete. That finding, he said, prompted him to question whether the definition of a gene itself needed revisiting—a line of thinking that ultimately led to his search for functional chimeric mRNAs. 

Jackson said the Harvard study might never have moved beyond an idea “sitting in a folder” if not for graduate student Olivia Venezia, a Ph.D. candidate at Harvard Medical School, who agreed to take on the unproven concept when she joined his lab. Venezia, the paper’s first author, “drove the whole thing,” Jackson said. Harry Kane, Ph.D., a research fellow in immunology at Harvard Medical School and the study’s co–first author, was instrumental in building and refining the computational pipeline. 

The multi‑technology workflow the Harvard team built to detect, verify, and functionally test chimeric mRNAs, was a crucial part of the work because chimeric RNAs are notoriously difficult to identify by short- or long-read sequencing alone. The Jackson team built a multilayered pipeline that combines long‑read direct RNA sequencing, short‑read fusion detection, NanoString probes that physically bind across RNA junctions, Hi‑C genome mapping to show inflammation‑driven gene proximity, and CRISPR technology for validation, essentially demonstrating how the right combination of tools can uncover an entire layer of gene regulation that has been hiding in plain sight. The pipeline took four to five years to develop [and] now runs in about eight hours. 

With RNA sequencing, artifacts are errors that can be introduced during reverse‑transcription (RT), such as stitching together pieces of RNA that were never actually fused inside the cell. This can create fake chimeric RNAs that look real in sequencing data but do not exist biologically. These artifacts were a major reason why early studies reported many chimeric RNAs that later turned out to be false positives. That’s why Jackson’s team relied on the multilayered pipeline, which filtered out the noise and confirmed which fused RNAs are real and functional. 

Without the pipeline, most chimeric RNAs would remain invisible or be dismissed as sequencing errors. The workflow provides the first reliable method for finding endogenous, functional chimeric mRNAs in mammalian cells—opening the door for other labs to identify additional chimeras across immunity, development, cancer, and disease. 

“This is a natural phenomenon that occurs across many cell types,” Jackson said. “We didn’t appreciate it until now.” 

A Fusion Protein That Influences Infection and Sepsis 

The Harvard team focused on a chimeric mRNA called Gsdmd–Tmem106a, found in mouse immune cells, that encodes a fusion protein. The protein sits on the cell’s outer membrane and attaches to the activated, pore‑forming fragment of GSDMD. Because the mRNAs from Gsdmd and Tmem106a are spliced together, the resulting hybrid protein changes how immune cells respond to infection and inflammation. Mice that couldn’t make this fusion protein had a harder time clearing Salmonella infections, showing weaker antibacterial protection. However, those same mice were far more resistant to severe sepsis, staying warmer and surviving doses of inflammatory toxins that normally prove fatal. 

“It showed physiologically that the chimeric protein had a non‑redundant, important function in life‑and‑death outcomes,” Jackson said. 

A New Valuable Resource in Research 

The work caught the attention of outside experts, including Hui Li, Ph.D., chief of experimental pathology at the University of Virginia School of Medicine, who was not involved in the study but reviewed the findings. 

Li said the results highlight several important points: chimeric RNAs are not transcriptional noise; their formation may be mediated by CTCF, a protein that regulates genome organization, and 3D proximity; and they represent “a hidden pool of transcriptome.” 

“The big picture is that humans share a similar number of genes with mouse, drosophila, or even worm. We already know alternative splicing, RNA editing, protein modifications, etc. can increase the final functional gene products. However, chimeric RNAs can be a hidden pool of transcriptome, sort of like dark matter in the universe,” Li said. “My lab has found trans-spliced chimeric RNAs before, but so far only a few functional trans-spliced chimeric RNAs have been reported in the literature. In the past, we relied on short-reads RNA-seq, but the technology suffers from RT artifacts, and many can’t be validated. This paper found a large number of interchromosomal and long-distance intrachromosomal chimeras, using direct long-reads RNA-seq and hybridization-based NanoString technology. Both eliminate the concern for template switching artifacts during the RT step,” he said. 

Li added that the study contributes to scholarship by identifying a large number of interchromosomal and long‑distance intrachromosomal chimeras. His lab held a journal club on the paper and immediately began discussing next steps. “This paper provides a very valuable resources as well as a pipeline for the community. At the same time, as [with] all significant work, it raises additional questions. For instance, the authors found some interesting discrepancy between different platforms. It is not clear to me that why a subset of long-read candidates cannot be validated by matching junction short-reads. The devil is in the details,” he said. 

Jackson said computational biologists have already requested access to the raw code.  

Conserved Patterns Suggest Broader Biological Relevance 

Although the exact GSDMD‑TMEM106A fusion was not found in human macrophages, the team identified 900 chimeric RNAs (chRNAs) in humans, and 33 of these were built from the same genes as chRNAs identified in mice. The specific junctions differ, but the underlying pattern is conserved across species. 

“In very few cases, we’ve actually found ones that are conserved between mice and humans,” Jackson said. “That’s wild; how that has co‑evolved is crazy.”  

Li noted that only a few functional trans‑spliced chimeric RNAs have been reported in mammals.  

Despite the progress, Jackson said the biggest unanswered question is how specific exons consistently fuse together. In the Gsdmd‑Tmem106a chimera, for example, two exons from one gene always splice into a specific exon of the second gene. 

“We have some ideas [why that happens], but we don’t have data yet,” he said. “That’s ongoing work.” 

The findings open potential avenues for drug development and diagnostics. Chimeric proteins could become targets for modulating inflammation in sepsis or other immune disorders. Chimeric mRNAs might serve as biomarkers for infection or inflammatory risk. And synthetic mRNA therapeutics could one day deliver beneficial chimeric proteins. 

Jackson’s lab is now profiling chimeric RNAs in chronic viral infections, neurodegenerative diseases, and inflammatory disorders. “We think there are absolutely disease‑causing ones out there,” he said. “And they could be targeted.” 

Jackson said the field is only beginning to understand the implications. “Now that we know how to find them, anyone can look,” he said. “We need a broader effort to characterize them.”