DRT10 belongs to a growing catalog of bacterial defense systems that rival CRISPR in their ingenuity, but it departs from the familiar playbook in a fundamental way. CRISPR-based defenses and the genome editors derived from them work primarily by cutting DNA at specified sites. DRT10 instead deploys a reverse transcriptase, an enzyme capable of synthesizing DNA from an RNA template, to produce a long strand of DNA bearing a repetitive sequence. That capacity to write new DNA sequences rather than merely sever existing ones is precisely what makes the system so intriguing to biotechnologists. Samuel Sternberg, professor of biochemistry and molecular biophysics at Columbia University Vagelos College of Physicians and Surgeons and a Howard Hughes Medical Institute investigator, who led the study, describes it as an exotic departure from the typical way scientists think about genome engineering. Rather than cutting DNA like CRISPR or swapping single letters of the genetic code like base editors, the system uses RNA as a template to write new sequences of DNA, offering a different starting point for building tools to engineer cells.

The road to the discovery began with a longstanding mystery in bacterial genomics. In 2008, researchers grouped together a collection of bacterial DNA-writing enzymes under the informal label of the Unknown Group, a family of dozens of reverse transcriptases whose functions had never been determined. Because reverse transcription creates new DNA from strands of RNA, these enzymes have long been considered promising raw material for developing genetic tools that could write therapeutic sequences into genomes. Sternberg’s laboratory, which has built its reputation by mining the genomes of bacteria and archaea for inventive ways these organisms control and alter their own DNA, found the family irresistible. As Sternberg notes, the team had a large family of enzymes that could write DNA but almost no idea what bacteria were using them for, a biological mystery with the potential to yield entirely new biotechnology.

It was Stephen Tang, a graduate student in the Sternberg lab, who cracked the case. Searching within the Unknown Group, Tang identified DRT10 and demonstrated that the system fights viral infection by using its reverse transcriptase to generate a long piece of DNA with a repetitive sequence read from a template RNA. Tang worked alongside co-first authors Josephine L. Ramirez and Qixiang He and an international team of collaborators. Ramirez led the biochemical studies that characterized the system’s activity, while He and Israel S. Fernández directed structural work that revealed how DRT10’s protein and RNA components fit together at the molecular level. Together, these complementary approaches provided a first mechanistic portrait of a defense strategy that no one had previously described.

The pivotal realization came as Tang probed how the reverse transcriptase operates. The enzyme, he recognized, works much like telomerase, the specialized reverse transcriptase found in eukaryotes, an evolutionarily distant group of organisms that includes yeast, plants, fungi, and animals. In humans and most other eukaryotes, telomerase creates DNA caps that protect the ends of linear chromosomes. Every time a cell divides, these caps, called telomeres, erode, and without telomerase to replenish them, chromosomes gradually lose protective sequence until cells stop dividing and enter a senescent state. That a bacterial antiviral enzyme and the guardian of human chromosome ends should operate by such similar chemistry was far too striking to dismiss as coincidence.

The functional similarity might have remained an interesting footnote had it not been for a visit to the Sternberg lab by Mario Rodríguez Mestre, a colleague at the University of Copenhagen who uses computational methods to identify and classify bacterial defense systems. During a talk, Mestre presented a preliminary evolutionary tree of all known reverse transcriptases, with branches color-coded by domain of life. One side of the tree was almost entirely brown, representing bacterial enzymes, while the other side was turquoise, representing eukaryotic enzymes. Sternberg interrupted the presentation, intrigued by a narrow turquoise sliver of eukaryotic reverse transcriptases that clustered tightly with the bacterial branch. That sliver turned out to be telomerase. For Sternberg, it was a eureka moment: the functional parallels between DRT10 and telomerase might reflect a shared evolutionary ancestry rather than convergence, unifying what had seemed like two separate fields into a single large family of enzymes.

Additional evolutionary analyses bolstered that interpretation. The data support a common origin for bacterial defense-associated reverse transcriptases and telomerase, indicating that the capacity to synthesize repetitive DNA arose in a shared ancestor and was later adapted by eukaryotes to maintain linear chromosomes. This raises a provocative possibility about the history of life itself. As Sternberg suggests, ancient enzymatic capabilities of a DRT10-like system may have been part of the molecular machinery that allowed the eukaryotic branch of life to emerge with a different, linear chromosomal topology. Bacteria carry their genomes on circular chromosomes that need no end protection, whereas eukaryotes adopted linear chromosomes, which create the telomere problem that telomerase solves. An ancestral DNA-writing enzyme capable of generating repetitive telomeric sequence could therefore have been a prerequisite for the transition, helping set the evolutionary stage for complex organisms.

Beyond its evolutionary significance, the discovery carries immediate translational promise. Telomere shortening drives cells into senescence, and the accumulation of senescent cells is linked to a range of age-related diseases, raising the possibility that restoring telomeres could help preserve healthy cell function. Defects in telomere maintenance also underlie a class of disorders known as telomeropathies, and Sternberg’s team hopes to explore whether DRT10 could eventually be adapted to address them. The lab has already used the enzyme to synthesize human telomeric DNA repeats in human cells and is now working to determine whether lengthening chromosomes can prevent cells from senescing and extend their healthy lifespans. Although the work remains at an early stage, the prospect of turning a bacterial defense enzyme into a tool for rebuilding protective DNA at chromosome ends represents a striking repurposing of microbial immunity for human medicine.

Cancer immunotherapy offers another potential application. CAR-T cell therapy, which engineers a patient’s own T cells to recognize and attack cancer cells, is often limited by the exhaustion and finite lifespan of the therapeutic cells themselves. Because telomere length governs how long a cell can continue dividing, maintaining or lengthening telomeres could help therapeutic T cells remain active for longer periods in patients. DRT10, with its ability to write repetitive DNA from an RNA template, gives researchers a brand-new way to explore that question, and potentially a means of engineering longer-lived cell therapies.

The study also stands as a testament to the value of basic research pursued without a predetermined application in mind. Curiosity about what bacteria do with a family of uncharacterized enzymes has illuminated deep questions about aging, cancer, and the evolution of life, revealing an unexpected connection between microbial antiviral defense and one of the most consequential enzymes in human biology. Columbia University has filed a patent application related to the work, with Sternberg, Tang, and Ramirez listed as inventors, and Sternberg holds advisory roles with several genome-editing companies, signaling commercial interest in the technology. Whether DRT10 matures into a clinical tool for telomeropathies, an enhancer of CAR-T therapy, or a novel platform for writing DNA into genomes, its discovery has already accomplished something profound: it has traced a single molecular thread from the viral wars of bacteria to the chromosome caps that make complex life possible, and offered an awe-inspiring look back at how life evolved.

Subject of Research: A bacterial antiviral reverse transcriptase system, DRT10, and its shared evolutionary origin with eukaryotic telomerase

Article Title: Scientists uncover an ancient link between viral defense and chromosome protection

Article References: Scientists uncover an ancient link between viral defense and chromosome protection. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: DRT10, reverse transcriptase, telomerase, bacterial antiviral defense, telomeres, genome engineering, evolution, senescence, CAR-T cell therapy, telomeropathies, CRISPR, Columbia University

Defense-associated reverse transcriptases, abbreviated DRTs, form a diverse family of enzymes that bacteria deploy to repel phage infection. Unlike the reverse transcriptases encoded by retroviruses such as HIV, which convert viral RNA genomes into DNA for integration into the host, DRTs appear to have been repurposed entirely for defense. In recent years, researchers have characterized several members of this family, including DRT2, DRT3, and DRT9, each of which uses cDNA synthesis in a distinctive way to detect or neutralize invading viruses. Yet the majority of DRT systems remain unexplored, their mechanisms hidden within the vast sequence space of bacterial genomes. The new study set out to illuminate one such dark corner of the family and, in doing so, uncovered a mechanism with unexpected parallels to one of molecular biology’s most famous enzymes.

The DRT10 system consists of two components: a reverse transcriptase protein and an associated noncoding RNA, an RNA molecule that does not encode a protein but instead serves a structural or functional role. When the researchers reconstituted the system in the laboratory, they found that the enzyme catalyzes a processive, protein-primed synthesis of tandem-repeat DNA. In other words, once initiated on a protein primer, the reverse transcriptase repeatedly copies the same segment of the noncoding RNA template over and over, churning out a long DNA product composed of head-to-tail repeats. This iterative mode of polymerization is the defining feature that links DRT10 to telomerase, which likewise adds repetitive DNA sequences, the telomeric repeats, again and again to chromosome ends using its own built-in RNA template.

A central question in understanding such systems is which part of the RNA serves as the template for repeat addition. The team found that reverse transcription in the DRT10 system is templated by a defined region of the noncoding RNA characterized by a distinctive sequence pattern, and that the template is flanked by template boundary elements. Both the sequence pattern and the boundary elements are conserved across related systems, and both proved essential for DNA repeat addition in biochemical assays. This architecture solves a fundamental problem for any enzyme that copies the same template repeatedly: how does the polymerase know where the template begins and ends so that it resets after each round rather than drifting off the RNA? The boundary elements act as molecular fences, delimiting the stretch of RNA that is copied and enabling the enzyme to cycle back and synthesize the next repeat in the tandem array.

To see how the protein and RNA components are physically organized, the researchers turned to cryo-electron microscopy, a technique that flash-freezes molecular complexes and images thousands of copies to reconstruct their three-dimensional structure. The reconstructions revealed an unanticipated 2:1 architecture: two reverse transcriptase monomers bind to opposite sides of a single pseudo-symmetric noncoding RNA molecule. The RNA, in effect, acts as a scaffold that dimerizes the enzyme, presenting two template regions, one to each monomer. This two-fold symmetric arrangement was not predicted from sequence analysis alone and illustrates how noncoding RNAs can choreograph the assembly of multi-enzyme complexes in ways that protein-only systems do not.

Biochemical experiments then dissected what each of the two bound monomers actually does. The results showed that each reverse transcriptase monomer reverse transcribes the template encoded on its respective side of the noncoding RNA, but only one of the two generates the long repetitive product. The identity of that productive template is defined by the distance between two flanking stem-loop anchors, structured elements in the RNA that bracket the template region. By varying this distance, the researchers could alter the properties of the synthesized DNA, demonstrating that the RNA scaffold does more than merely position the enzymes: it actively specifies which template is used for processive repeat synthesis. The second monomer, meanwhile, copies its own template without producing the long repetitive cDNA, a division of labor whose functional consequences for antiviral defense remain an open question.

The mechanistic echoes of telomerase are hard to miss. Telomerase carries its own RNA subunit with a short template segment, uses boundary elements to keep the enzyme copying only that segment, and repeatedly adds telomeric repeats to chromosome termini through iterative cycles of template positioning and DNA synthesis. The mode of iterative, boundary-delimited cDNA synthesis observed for DRT10 exhibits direct parallels to this process, despite the two systems operating in organisms separated by billions of years of evolution and serving entirely different biological purposes, one defending bacteria against phages and the other preserving chromosome integrity in eukaryotes. The convergence on the same polymerization logic suggests that iterative templated repeat addition is a particularly effective solution to a recurring biochemical problem.

That suggestion gains weight from the study’s phylogenetic analyses. When the researchers built evolutionary trees of reverse transcriptases across the domains of life, telomerase did not appear as an isolated lineage. Instead, it fell within a monophyletic clade together with the DRT family, meaning that telomerase and these bacterial defense enzymes appear to share a common ancestor. The implication is provocative: the DNA polymerization activity of telomerase, long studied as a eukaryotic specialization, may have roots in the ancient bacterial machinery of antiviral defense. If so, the telomere-maintaining enzymes that influence aging and cancer in humans would represent evolutionary descendants of, or at least relatives of, molecular weapons that bacteria once wielded against their viruses.

Together with earlier work on DRT2, DRT3, and DRT9, the new findings point to a conserved mechanistic logic underlying RNA-templated DNA repeat addition across diverse DRT systems, even as those systems arrive at vastly different architectural solutions. Some use single enzymes, others dimeric assemblies; some employ protein priming, others rely on different initiation strategies; yet all converge on the core idea of using a noncoding RNA template, delimited by structural boundaries, to generate repetitive cDNA. This recurring design principle now stands as a unifying theme in a rapidly expanding branch of bacterial immunity research, one in which reverse transcription, once considered a viral specialty, has become a central player.

The work also broadens the known repertoire of reverse-transcription-based antiviral strategies in bacteria, a catalogue that biotechnologists are watching closely. Bacterial defense systems have repeatedly proven fertile ground for molecular tools, with CRISPR being the most famous example, and enzymes that synthesize DNA from RNA templates with programmable specificity hold obvious appeal for genome engineering, molecular recording, and biosensing applications. While the immediate significance of the study lies in basic biology, revealing how a bacterial immune system works at atomic and mechanistic resolution, the deep evolutionary link it draws between phage defense and telomerase ensures that its implications will reach well beyond microbiology, touching questions about the origins of the very machinery that protects the chromosomes of complex life.

Subject of Research: A bacterial antiviral defense system that uses telomerase-like reverse transcription to synthesize tandem-repeat DNA

Article Title: Antiviral immunity via telomerase-like reverse transcription in bacteria

Article References: Tang, S., Ramirez, J. L., He, Q., Rodríguez Mestre, M., Wiegand, T., Zhang, D. J., Wang, M., Lampe, G. D., Wang, J., Ma, Y., Jovanovic, M., Pinilla-Redondo, R., Fernández, I. S., & Sternberg, S. H. (2026). Antiviral immunity via telomerase-like reverse transcription in bacteria. Nature. https://doi.org/10.1038/s41586-026-11250-8

Image Credits: AI Generated

DOI: 10.1038/s41586-026-11250-8

Keywords: antiviral immunity, reverse transcriptase, telomerase, bacterial defense, noncoding RNA, cryo-electron microscopy, phage biology, cDNA synthesis, molecular evolution, DRT10, tandem-repeat DNA, CRISPR

HIV-1 currently comprises four lineages: group M, for major; group N, for non-M non-O; group O, for outlier; and group P, named to continue in alphabetical order. Group M is by far the most prevalent lineage and the only pandemic strain, believed to have originated in Kinshasa, the capital of the DRC, with the emergence of HIV-1 estimated to have occurred around the 1920s in the DRC. Within group M, ten subtypes have been delineated, labeled A through D, F through H, and K through L, along with nine sub-subtypes such as A1 through A8 and F1 and F2. These subtypes are thought to have diverged from one or a few ancestral HIV-1 strains after a single chimpanzee-to-human transmission, some of them around the 1950s. The most recently recognized subtype, L, was formally established in 2020 on the basis of complete genome sequences identified in the DRC between 1983 and 2001.

The designation of a new HIV-1 subtype is governed by strict criteria. Researchers must identify at least three viral strains or complete genomes from distinct, epidemiologically unlinked individuals. Phylogenetically, each subtype should form a distinct monophyletic clade, with approximately equal genetic distances separating it from the other subtypes, producing a characteristic star-like phylogeny. Established group M subtypes differ by a mean nucleotide divergence of approximately 15 percent at the genome level, with considerable variation between genes. In addition, genomes classified within a given subtype must be non-mosaic, distinguishing them from recombinant genomes that arise when viruses of different subtypes exchange genetic material inside coinfected cells.

The new report describes an HIV-1 genome recovered from residues of peripheral blood mononuclear cells from a single blood sample collected during the summer of 2023. The team extracted viral DNA using a commercial kit and amplified the genome with conventional polymerase chain reaction primers as well as a whole-genome HIV-1 genotyping assay. Sequencing was performed using three complementary platforms: Sanger sequencing, Oxford Nanopore technology on a GridION instrument, and Illumina technology on a MiSeq instrument. The resulting near-full-length consensus genome, deposited in GenBank under accession number PZ307881, was 8,950 nucleotides long and covered 92 percent of the subtype B reference genome, with the shortfall mostly due to sequencing defects at the 5-prime and 3-prime ends.

Comparative genomic analysis revealed that the ten best matches in the GenBank database exhibited nucleotide similarities of only 83.29 to 83.60 percent to the new sequence. These closest genomes had been collected from patients in the DRC, the United States, the Netherlands, and Gabon between 1987 and 2018. When aligned against representative genomes of the ten established subtypes, the new genome showed a mean nucleotide divergence of 12.69 percent, ranging between 11.87 and 13.28 percent, while the ten established subtypes diverged from one another by a mean of 12.64 percent, ranging between 10.01 and 13.73 percent. In other words, the candidate subtype is as distant from every known subtype as those subtypes are from each other, and pairwise comparisons showed no statistically significant differences in divergence. Similar results held when the genomes were cut into eight fragments corresponding to different genomic regions, with regional divergences for the candidate genome ranging from 8.55 to 20.23 percent, closely mirroring the 8.49 to 20.25 percent range observed among established subtypes.

Phylogenetic analyses reinforced this picture. Using IQ-TREE and Nextclade software, with a simian immunodeficiency virus genome from Gabon as an outgroup, the researchers found that the candidate genome formed a distinct branch within an overall star-like tree, the hallmark topology of genuine subtypes. The closest established subtype was J, supported by a bootstrap value of 99 percent. Independent trees built from each of the eight genome fragments likewise showed star-like topologies with the candidate sequence on its own branch. Bootscan analysis, which uses a sliding window of 200 nucleotides to detect recombination, found no recombination events, satisfying another key criterion for subtype status. Nextclade classified the genome as an other clade, and a tree incorporating 1,020 genomes from the Los Alamos National Laboratory HIV Sequence Database placed it on a new node. Taken together, the researchers concluded that the genome belongs to a candidate new subtype, which they named M following subtype L in alphabetical order, while acknowledging the potential nomenclatural ambiguity of reusing the letter that already designates the pandemic group.

The clinical course of the patient offers an intriguing backdrop to the genomic findings. At the start of antiretroviral treatment, her CD4-positive T lymphocyte count was 351 cells per cubic millimeter with a CD4-to-CD8 ratio of 0.23, and plasma HIV-1 RNA stood at 5.03 log10 copies per milliliter. Subsequently, her CD4 count fluctuated between 406 cells per cubic millimeter in 1999 and 2,394 cells per cubic millimeter in 2008, and has remained above 1,000 since 2007. Plasma viral load has been undetectable during regular monitoring from 2008 onward, except for a reading of 2.2 log10 copies per milliliter in 2024. Notably, a Western blot performed in 2023 showed antibody reactivities only to envelope and capsid proteins, with no response to Pol proteins. The authors suggest this unusual profile may partly reflect reduced antibody affinity for antigenically divergent proteins, or an overall weak antibody response to HIV-1 antigens, which particularly affects Pol-encoded proteins such as reverse transcriptase, protease, and integrase, especially after more than three decades of infection and long-term antiretroviral therapy.

Genotypic resistance analysis using algorithms from the French National Agency for AIDS Research and the Stanford University HIV drug resistance database found no naturally present amino acid mutations conferring resistance to any currently available antiretroviral drug, apart from two substitutions acquired during treatment failures: Y181C in reverse transcriptase, detected transiently in 2003 during a virological failure episode on nevirapine, and I84V in protease, detected in 2007 during failure on atazanavir and ritonavir. Viral tropism was predicted to be R5, indicating use of the CCR5 co-receptor. The researchers cautioned that the absence of natural resistance mutations should be interpreted carefully given the markedly different genetic background of this genome, though the finding is congruent with observations for both rare and prevalent subtypes.

The team also turned to molecular modeling to explore how the roughly 40 amino acid differences in the candidate subtype’s reverse transcriptase, the central enzyme of HIV replication and a prime drug target, might alter its structure. Mapping the mutations onto the reference enzyme structure revealed changes predicted to reshape several surface loops, including a closure of the F130 to P150 buckle and a flattening of the K281 to P294 loop. Alterations in the N57 to D76 region, corresponding to the functionally important beta3-beta4 loop within the fingers subdomain, were associated with a greater space between the fingers and palm subdomains. The surface electrostatic potential was modified but remained globally electropositive, particularly in the ligand-interacting area. The modeled enzyme-ligand interaction energy decreased slightly, from minus 646 to minus 512 kilojoules per mole, because the number of amino acids involved in ligand interaction fell from 15 to 7. The authors speculate that such changes could modulate enzyme processivity and, by promoting dissociation and re-association of the enzyme-substrate complex, potentially increase template switching and genetic recombination, though they stress these hypotheses derive entirely from in silico modeling and require in vitro biochemical validation.

The discovery underscores that HIV-1 diversity remains far from fully charted and reinforces the status of the DRC as the epicenter of the group M pandemic and a region where rare subtypes continue to circulate. Rare established subtypes, including F, H, J, K, and L, each account for less than 1 percent of global HIV-1 infections and circulate primarily in the DRC, Angola, and Cameroon, with limited data on their transmissibility, virulence, and pathogenicity. The researchers also observed notable intra-sample diversity, with mean intra-host nucleotide divergence of 1.61 percent and a maximum of 5.52 percent, suggesting intra-patient viral evolution or, less likely given the candidate subtype’s apparent rarity, superinfection with multiple variants during the patient’s decades of stays in the DRC. Formal recognition of the new subtype will require detecting it in at least two more epidemiologically unlinked individuals, and the authors call for further studies, primarily in the DRC, to establish its prevalence and genotypic and phenotypic characteristics.

Subject of Research: Discovery of a candidate new HIV-1 group M subtype in a patient from the Democratic Republic of the Congo

Article Title: A case of infection with a candidate new subtype, tentatively termed M, of human immunodeficiency virus type 1 of group M

Article References: COLSON, P., BOSCHI, C., BEDOTTO, M., MOTTE, A., SOLAS, C., LEFEBVRE, M., MILLION, M., TOMEI, C., DELERCE, J., & FANTINI, J. (2026). A case of infection with a candidate new subtype, tentatively termed M, of human immunodeficiency virus type 1 of group M. New Microbes and New Infections, Article 101863. https://doi.org/10.1016/j.nmni.2026.101863

Image Credits: AI Generated

DOI: 10.1016/j.nmni.2026.101863

Keywords: HIV-1, subtype M, Democratic Republic of the Congo, viral phylogenetics, whole-genome sequencing, reverse transcriptase, group M, subtype L, molecular modeling, antiretroviral resistance, HIV diversity, New Microbes and New Infections

The core of the Detectron design marries two molecular technologies that synthetic biologists have refined over the past decade. The first is the toehold switch, a de novo-designed RNA regulator first reported by Green and colleagues in 2014. A toehold switch is a carefully folded strand of RNA that hides a protein-coding or functional sequence behind a hairpin structure. When a complementary trigger RNA binds to an exposed single-stranded ‘toehold’ region, the hairpin opens and the previously sequestered sequence becomes active. Toehold switches are prized because they can be designed almost arbitrarily, letting researchers program sensors for nearly any RNA sequence of interest. Their weakness, however, is that their outputs are typically proteins or RNA molecules that are themselves transient, difficult to multiplex, and poorly suited to sequencing-based readouts.

That is where the second technology enters. Retrons are bacterial genetic elements that naturally produce large amounts of single-stranded DNA inside living cells. A retron consists of a reverse transcriptase enzyme together with a noncoding RNA scaffold, traditionally divided into the msr and msd regions, that the enzyme partially copies into DNA. The result is a covalently linked RNA-DNA hybrid molecule, a peculiar product first characterized in the 1980s and later recognized as part of bacterial anti-phage defense systems. Recent years have seen retrons repurposed as tools for genome editing and DNA recording, and the Shipman laboratory has been at the forefront of that effort. Detectrons fuse the two systems: the toehold switch controls when the retron’s reverse transcriptase activity produces its DNA product, and the sequence of that DNA product serves as a barcode unique to the RNA that triggered it.

In practice, a Detectron is a single engineered RNA molecule. In its default state, the switch region folds into a hairpin that blocks productive priming of reverse transcription. When the target RNA appears and hybridizes to the toehold, the structure rearranges, the retron reverse transcriptase gains access to its template, and a short single-stranded DNA barcode is synthesized. Cells that never encounter the target produce little or no barcode; cells that do encounter it accumulate a measurable DNA record. Because the barcode is DNA, it is stable, it survives conditions that would destroy RNA, and it can be pooled, amplified, and sequenced alongside thousands of other samples. This converts what was once a per-sample fluorescence measurement into a massively parallel counting problem, one that modern sequencing platforms are exceptionally good at solving.

Getting the architecture to work well was not trivial, and much of the new paper is devoted to systematic optimization. The authors constructed a large synthetic library of toehold-retron variants, systematically varying structural parameters such as the lengths of the two stems in the hairpin, the loop that connects them, the bulges that interrupt base pairing, and the distance between the stem and the priming guanine where reverse transcription begins. By measuring barcode production for each variant in both the presence and absence of trigger RNA, they generated a quantitative map of how each design choice affects the on/off ratio, the critical figure of merit that determines how cleanly a sensor distinguishes true signal from background.

Machine learning then extracted design principles from that map. Using XGBoost, a scalable tree-boosting algorithm, the team trained regression models on the variant library and evaluated them with cross-validated metrics including the coefficient of determination, root mean squared error, and Pearson correlation. SHAP analysis, a technique that attributes model predictions to individual features, revealed which structural parameters mattered most. The findings were sometimes counterintuitive: shortening the loop below roughly seven nucleotides sharply degraded both signal strength and on/off ratios, while specific combinations of stem lengths and hairpin folding free energy proved far more predictive of performance than any single parameter alone. Armed with these rules, the researchers built Detectron-V2, a redesigned second-generation device that achieved substantially enhanced on/off ratios and high specificity across a panel of phage transcripts.

To demonstrate the platform’s real-world utility, Han and Shipman turned to bacteriophages, the viruses that infect bacteria. Detectrons were programmed to recognize transcripts from specific phage genes, including the major capsid gene gp23 of phage T4 and sequences from phage T7. When a bacterial cell carrying the appropriate Detectron was infected by the matching phage, the viral RNA triggered barcode synthesis; unrelated phages did not. The team even showed that the response was dose-dependent, with barcode enrichment scaling with the multiplicity of infection, meaning the devices do not merely register that an infection occurred but carry quantitative information about how intense it was. Target-site selection within a viral transcript also mattered, with sites predicted to be more locally accessible by RNA folding algorithms yielding better sensor performance.

The most striking demonstration came in pooled cultures. The researchers mixed together bacterial strains, each carrying a Detectron tuned to a different phage, and infected the pooled population. After sequencing the barcode pool, they could determine which phages had infected which strains within a single pot, without any physical separation of the samples. They extended this to ratiometric susceptibility profiling, in which the relative abundance of barcodes reports quantitatively on how susceptible different host strains are to different phages. For the phage-therapy field, where matching the right phage to the right bacterial infection is the central clinical challenge and current methods require growing each phage-host pair separately, a pooled sequencing-based assay of this kind could dramatically accelerate screening.

Beyond virology, Detectrons join a growing family of molecular recording technologies that include CRISPR-based recorders and recombinase state machines, but they occupy a distinctive niche. Where CRISPR recorders typically write into the genome through DNA cleavage events, Detectrons produce their records through reverse transcription of an engineered RNA template, a process that does not require the recorder to target the cell’s own DNA. That makes the approach potentially safer and more modular, and the authors note that the framework enables alternative modes of transcript-based sensing with applications including viral infection detection more broadly. The ability to record transcriptional events in complex bacterial communities, such as microbiomes, suggests uses in environmental monitoring and diagnostics where cells must be observed in place rather than disrupted.

The work also carries the familiar caveats of early synthetic biology. The demonstrations were performed in engineered bacterial strains with inducible promoters and defined phage infections, and porting the technology to other organisms, to natural microbial communities, or to mammalian cells will require re-optimization of the switch architecture for each new context. Han and Shipman have filed a provisional patent application on the technology, signaling commercial interest, and the complete variant libraries, sequences, code, and statistical data have been released to support replication and extension by other groups. Still, the conceptual advance is clear: a programmable RNA sensor whose output is not a fading fluorescent glow but a stable, sequenceable, multiplexable DNA barcode. As the design rules mature, devices of this kind could turn living cells into their own record-keepers, quietly logging which genes spoke and when, for anyone with a sequencer to read the archive.

Subject of Research: Synthetic RNA biosensors that record transcript detection as DNA barcodes via retron reverse transcription

Article Title: Detectrons convert transient RNA sequences into stable DNA barcodes for high-throughput analysis of RNA-dependent processes

Article References: Han, J., & Shipman, S. L. (2026). Detectrons convert transient RNA sequences into stable DNA barcodes for high-throughput analysis of RNA-dependent processes. Nature Biotechnology. https://doi.org/10.1038/s41587-026-03334-7

Image Credits: AI Generated

DOI: 10.1038/s41587-026-03334-7

Keywords: Detectrons, toehold switches, retrons, reverse transcriptase, DNA barcodes, RNA sensing, synthetic biology, bacteriophages, phage therapy, machine learning, molecular recording, Nature Biotechnology

Prime editing, introduced by David Liu’s group at the Broad Institute in 2019, combined a catalytically impaired Cas9 nickase with an engineered reverse transcriptase and a guide RNA that carries both the targeting information and the template for the desired edit. The system writes new sequence directly into the genome without making a full double-strand break, avoiding the chaotic repair outcomes that plague conventional CRISPR-Cas9 cutting. Yet prime editing has always faced a ceiling: the efficiency and precision of the approach decline sharply as the requested edit grows larger. Insertions of more than a few dozen base pairs become unreliable, and integrating entire genes or rearranging chromosome segments has remained largely out of reach.

Prime assembly tackles that ceiling by reconceiving the edit as a stepwise construction process rather than a single copying event. Instead of forcing the reverse transcriptase to polymerize a long, unwieldy DNA tract in one continuous reaction, the system orchestrates a series of coordinated sub-edits, each installing a defined fragment at the target locus. Because each fragment is short enough to be written with high fidelity, the assembled product can extend to kilobase scale while maintaining the sequence accuracy that defines prime editing. The design encodes overlapping junctions within the delivered template RNAs so that successive fragments anneal to one another and to the genomic target, stitching the pieces into a contiguous, correctly ordered insert.

The molecular choreography relies on a redesigned editing complex. The researchers report engineering the prime editor apparatus and its guide RNA architecture so that multiple template modules can be loaded and processed in a defined sequence at the same target site. Timing and polarity matter enormously in genome writing: a fragment installed out of order, or in the wrong orientation, would leave behind scars or truncate the intended product. The prime assembly system addresses this by controlling the order in which template segments are reverse-transcribed and by using nicking patterns that favor progressive extension of the nascent strand. In effect, the genome itself becomes the scaffold on which the new sequence is assembled, and the cell’s own repair machinery seals the final junctions.

What distinguishes the work most sharply from earlier large-insert methods is the degree of control over rearrangement. Transposon-based delivery systems can move large cargoes but do so at their own preferred genomic sites, and double-strand-break-dependent methods such as CRISPR paired with homologous recombination or non-homologous end joining frequently generate a messy mixture of deletions, inversions, and random integrations. Prime assembly, by contrast, specifies both the landing site and the structure of the rearrangement. The paper demonstrates targeted integration of sizable genetic payloads, precise excision of unwanted segments, and programmable reorganization of DNA segments within a locus, all without introducing a double-strand break at any point in the procedure.

The technical benchmarks reported in the study emphasize both efficiency and purity. Across the demonstrated edit classes, the authors describe insertion products that are predominantly exact, with detectable byproducts reduced relative to break-dependent alternatives. Sequence-level analysis of the assembled products shows the expected junction architecture, an important indicator that the overlapping-fragment design behaves as intended rather than relying on stochastic recombination. Editing outcomes were profiled at the DNA level with deep sequencing and, for the larger rearrangements, with long-read sequencing capable of confirming the structure of multi-kilobase changes end to end, a level of verification that smaller-scale editing studies rarely require.

The implications for disease research are considerable. A large fraction of pathogenic mutations are not simple point substitutions. Thousands of known genetic disorders arise from deletions, duplications, insertions of mobile elements, or larger structural changes that conventional single-nucleotide editors cannot address. Gene-addition therapies using viral vectors can deliver a working copy of a gene, but they insert it at a semi-random safe-harbor location rather than restoring the native locus, losing native regulation and occasionally provoking insertional complications. Prime assembly offers a route to writing a functional gene back into its endogenous position, complete with its own regulatory context, or to rebuilding a damaged locus from the bottom up.

Synthetic genome engineering stands to benefit even more directly. Building entire chromosomes and synthetic genomes has so far depended on laborious cycles of homologous recombination in yeast or on assembly in vitro followed by transplantation, approaches that are slow and species-limited. A method that can integrate large designed sequences at user-chosen genomic addresses in mammalian or other difficult cells would compress that workflow dramatically. The authors’ demonstration of programmable rearrangement suggests a longer-term vision in which genome architecture itself, not merely gene sequence, becomes an editable design parameter, allowing researchers to test in weeks what once took years of strain construction.

As with every genome-editing advance, the distance between a demonstration in cultured cells and a therapeutic reality is substantial. Delivery remains the field’s perennial bottleneck: prime editors are large multi-component systems, and a prime assembly platform that carries multi-part template cargoes is larger still, making efficient in vivo delivery a formidable engineering problem in its own right. Off-target activity, immunogenicity of the bacterial-derived editor proteins, and the long-term stability of large engineered loci will all need rigorous assessment. The authors and the field more broadly will also need to establish how the method performs in primary cells, tissues, and whole organisms, where chromatin context and cell-cycle state strongly influence repair outcomes.

Even with those caveats, prime assembly represents a conceptual milestone in the transition of genome editing from correction to construction. The first decade of CRISPR made cutting routine; the second made precise rewriting of individual letters increasingly reliable. Prime assembly points toward a third phase in which geneticists can move, add, and reorganize whole stretches of the genome with the same programmability that made base editing famous. For patients with structural variants that no current therapy can touch, and for engineers attempting to build genomes to specification, the ability to assemble DNA in place, fragment by fragment, at the exact location of choice, may prove to be one of the more consequential ideas to emerge from the genome-writing field in years.

Subject of Research: Prime assembly genome editing for targeted DNA integration and rearrangement

Article Title: Targeted genomic integration and rearrangement using prime assembly

Article References: Levesque, S., Kawashima, N., Hwang, G.-H., Zeng, J., Toskov, V., Barry, T., Mannherz, W., Homfeldt, L., Becerra, B., Schoonenberg, V. A. C., Pinello, L., Agarwal, S., & Bauer, D. E. (2026). Targeted genomic integration and rearrangement using prime assembly. Nature. https://doi.org/10.1038/s41586-026-11024-2

Image Credits: AI Generated

DOI: 10.1038/s41586-026-11024-2

Keywords: prime assembly, prime editing, genome editing, gene writing, CRISPR, reverse transcriptase, DNA integration, structural variants, synthetic genomics, gene therapy, molecular biology, biotechnology

Endogenous retroviruses, or ERVs, are the fossilized remnants of retroviruses that once infected the germ-line cells of ancestral hosts. When a retrovirus inserts its provirus into the DNA of a sperm or egg cell that goes on to contribute to the next generation, the insertion can be inherited vertically, copying itself into every cell of every descendant. Over millions of years these elements accumulate mutations, deletions and recombination events, but their diagnostic architecture, including long terminal repeats flanking the core retroviral genes gag, pol and env, often remains recognizable. Because integration events are effectively permanent and their ages can be estimated, ERVs function as paleontological records of the viruses that circulated in host populations deep in evolutionary time.

To recover these viral fossils from the macaque genome, the researchers employed a homology-based, genome-wide screen designed to detect relatively intact proviruses rather than the far more numerous degraded fragments and solo long terminal repeats that litter mammalian genomes. The survey exploited the current highest-quality chromosome-level assembly available for M. fascicularis, an important practical consideration because fragmented assemblies can obscure proviral structure and misplace insertions. Candidate elements were then subjected to structural annotation and phylogenetic classification, anchoring the new macaque sequences within the broader landscape of vertebrate retroviral diversity.

Classification relied on one of the most conserved regions of the retroviral toolkit: the reverse transcriptase domain of the pol gene. Because this enzyme performs the essential step of copying viral RNA into DNA, its sequence is under strong functional constraint and retains enough phylogenetic signal to place divergent elements into their correct lineages. Building multiple alignments of the recovered reverse transcriptase domains and reconstructing their evolutionary relationships, the team found that the macaque proviruses fall into betaretroviral, gammaretroviral and unclassified lineages, with the beta and gamma groups clearly predominating. This pattern mirrors what has been observed in many other mammalian genomes, where betaretroviruses and gammaretroviruses have historically been prolific colonizers of germ lines.

The compositional result carries evolutionary weight. Betaretroviruses, relatives of modern mouse mammary tumor virus and the simian retrovirus SERV, and gammaretroviruses, relatives of murine leukemia virus, differ in their envelope biology, host receptor usage and genomic preferences. A predominance of both classes suggests that the macaque lineage experienced repeated, independent waves of retroviral germline invasion rather than a single dominant radiation. Consistent with that interpretation, dating based on the divergence between the paired long terminal repeats of each provirus, which are identical at the moment of integration and accumulate mutations thereafter, indicated that the 106 loci span a broad range of integration ages, recording multiple distinct epochs of retroviral activity in the species’ ancestry.

The macaque proviruses are also distributed across all chromosomes, a pattern consistent with the idea that integration sites reflect a combination of retroviral target-site preferences and post-insertion forces such as selection, recombination and random genetic drift. Studies in human and mouse have shown that different retroviral genera favor different genomic neighborhoods, with some elements preferentially accumulating in gene-poor regions while others insert near transcription units. The chromosome-wide coverage reported for M. fascicularis means that the new dataset can support analyses of integration preference in a primate genome at a resolution previously unavailable for this species.

The biomedical significance of the work extends beyond evolutionary curiosity. The crab-eating macaque, also known as the cynomolgus macaque, is one of the most widely used primate models in drug development, infectious disease research and transplantation studies, and its endangered status in the wild makes the genomic record it carries all the more valuable to document. ERVs in macaques are directly relevant to laboratory safety as well: previous work has identified and characterized simian endogenous retroviruses in captive macaque populations in Indonesia, raising questions about how these elements behave in breeding colonies. A curated, high-confidence catalog of intact proviral loci gives researchers a framework for tracking ERV expression, recombination and potential co-option in an animal model whose transcriptomes are routinely interrogated.

The broader context of ERV biology reinforces that relevance. In humans, endogenous retroviral sequences have been co-opted for essential functions, most famously the syncytin genes derived from retroviral envelopes that mediate placental cell fusion, and ERV-derived regulatory sequences now contribute to innate immune gene networks. Deregulated ERV expression has also been implicated in a growing list of pathologies, including systemic lupus erythematosus, Parkinson’s disease, neuronal aging and cancer immunotherapy responses, where translated retroviral products can provoke interferon signaling or serve as tumor-associated antigens. Establishing which proviruses are intact in the macaque genome is therefore a prerequisite for asking whether analogous host-virus co-option and immune interactions operate in this model species.

The authors are careful to delineate the scope of their catalog. Because the screen targeted relatively intact proviruses, the dataset deliberately excludes the vastly more abundant degraded ERV fragments and solo long terminal repeats, which form when recombination between the two LTRs of an integrated provirus excises the internal sequence. Those eroded elements represent older and numerically dominant chapters of the retroviral record, and their omission means the 106 loci described here constitute a conservative, high-confidence floor rather than a complete inventory of retroviral genetic material in the macaque genome. Even so, the team notes that the curated set provides a reference resource for investigating the evolutionary history and genomic impact of preserved proviruses in an endangered primate.

The study also joins a rapidly expanding comparative effort. Recent surveys have uncovered endogenous retroviruses in red pandas, ducks and primitive ruminants, and genome-mining approaches have shown that retroviruses have pervasively invaded vertebrate genomes across the tree of life. By adding a systematic account for a close human relative, the macaque analysis helps triangulate which retroviral lineages circulated in the common ancestors of Old World monkeys and apes, and which invasions were lineage-specific. The genome assembly data underlying the work are publicly available through the National Center for Biotechnology Information, and the researchers suggest their dataset will serve as a durable foundation for exploring how viral inheritance has shaped, and continues to shape, the genome of one of biomedical science’s most important animal models.

Subject of Research: Discovery and evolutionary characterization of endogenous retrovirus proviruses in the genome of the crab-eating macaque (Macaca fascicularis)

Article Title: Discovery and evolution of endogenous retroviruses in the genome of crab-eating macaque (Macaca fascicularis)

Article References: Shi, W., Mao, L., Chen, Y., Alfred, N., Fu, Y., Bao, Y., Wang, X., Liu, Y., Yang, S., Ji, L., Zhou, C., Xu, J., Li, W., Shan, T., Wang, J., Zhang, W., & Shen, Q. (2026). Discovery and evolution of endogenous retroviruses in the genome of crab-eating macaque (Macaca fascicularis). Molecular Genetics and Genomics, 301(1), Article 189. https://doi.org/10.1007/s00438-026-02518-9

Image Credits: AI Generated

DOI: 10.1007/s00438-026-02518-9

Keywords: endogenous retroviruses, crab-eating macaque, Macaca fascicularis, retrovirus, genome evolution, provirus, phylogenetic analysis, long terminal repeats, reverse transcriptase, genomics, non-human primate model, viral fossils

The work addresses a gap that has shaped the trajectory of genome engineering for more than a decade. Since the advent of programmable nucleases, researchers have been able to direct double-strand breaks to almost any chosen sequence, and cellular repair machinery can sometimes stitch in a new DNA cassette at the break site. But that approach leans on the cell’s own repair pathways, which are error-prone, unpredictable and often disabling to the very sequences scientists want to insert. Broken DNA is dangerous DNA, and cells treat integration events as injuries to be patched rather than as opportunities for precise reconstruction.

Prime editing, first described in 2019, took a fundamentally different route. Rather than cutting both strands of the DNA double helix, a prime editor pairs a Cas9 nickase — an engineered enzyme that cuts only one strand — with an engineered reverse transcriptase. The editing instructions are carried on a prime editing guide RNA, or pegRNA, which both locates the target site and encodes the new genetic information the reverse transcriptase should write into the nicked strand. Because the process avoids double-strand breaks and does not require an additional donor DNA template supplied in bulk, prime editing has proven remarkably clean for small substitutions, insertions and deletions.

Where prime editing has historically faltered, however, is scale. The reverse transcriptase copies a sequence encoded within the pegRNA itself, and practical constraints on RNA length, delivery and synthesis efficiency have limited the size of the DNA payload that a single prime-editing event can install. For applications in which a functional gene, a large regulatory element or a multi-kilobase cassette must be placed at a defined genomic address, the technology’s ceiling has been a persistent frustration. Complementary systems — including CRISPR-associated transposases and integrase-based platforms — can move larger cargoes, but they typically bring their own constraints on target-site selection, orientation and cargo compatibility.

The new study, led by researchers working at the interface of protein engineering and genome technology, tackles the size problem by rethinking how the donor DNA participates in the reaction. In the reported strategy, termed donor-directed annealing, the genetic cargo is carried on a separate donor molecule rather than being encoded within the pegRNA. The prime editor still performs its characteristic task of opening the target site and synthesizing an exposed stretch of new DNA on the nicked strand, but that newly synthesized sequence is designed to serve as a molecular landing pad. Once exposed, it is complementary to sequences at the end of the donor fragment, and the two single-stranded regions find each other and anneal, drawing the donor cargo into the editing site.

The elegance of the design lies in what happens next. Cellular DNA repair enzymes process the annealed intermediate, ligating the donor fragment into the genome through the natural resolution of the flap-like structure that the prime editor has created. Because the specificity of the event is dictated by sequence complementarity between the editor-generated overhang and the donor terminus, the cell is never asked to recognize a double-strand break or to improvise an end-joining reaction. The authors report that this mechanism allows fragments substantially larger than the payloads accessible to conventional prime editing to be incorporated at defined loci, with precision determined largely by the programmed overlap rather than by stochastic cellular repair.

From a biochemical standpoint, donor-directed annealing converts what has been an intramolecular copying reaction into a hybridization-guided assembly step. Conventional prime editing is, in essence, a controlled form of DNA synthesis: the pegRNA templates every base that the reverse transcriptase installs. The new method retains that templated synthesis for a short anchoring sequence but delegates the bulk of the payload to a separate donor, which can be produced synthetically or by standard cloning at lengths far beyond what a pegRNA can encode. The trade-off is that the donor and the pegRNA must be co-delivered and their sequences coordinated, but the payoff is a system in which cargo size is no longer bound to the physical limits of the guide RNA.

The practical implications extend across both research and therapeutic arenas. In basic biology, the ability to drop large regulatory modules, reporter constructs or engineered gene circuits into precise genomic contexts would simplify experiments that currently require laborious screening of random integration events. In medicine, many inherited disorders are caused by mutations in genes that are too large, too structurally complex or too mutationally diverse to be addressed base by base. Delivering a corrected copy of a gene, or a functional cDNA, into its native locus under the control of endogenous regulatory elements — rather than scattering it randomly through the genome as viral vector gene therapy does — remains the aspirational gold standard, and integration strategies of this kind are among the most credible paths toward it.

The reported system also speaks to a recurring theme in the genome-editing literature: the value of avoiding double-strand breaks altogether. Studies across multiple cell types have associated double-strand-break-based editing with p53 activation, chromosomal rearrangements and large unintended deletions, concerns that are particularly acute for ex vivo cell therapies and in vivo applications alike. By building integration on a nicking enzyme and sequence-programmed annealing rather than on blunt-ended break repair, the approach aligns with the field’s broader movement toward editing chemistries that leave the genome’s integrity machinery largely undisturbed.

As with any genome-engineering advance, several questions will shape how the technique matures. The efficiency of integration across different genomic loci, cell types and species will need systematic mapping; cargo lengths will have practical ceilings set by delivery vehicles rather than by chemistry; and off-target activity — a concern for any nuclease-fusion system — will require careful characterization at both the sequence and chromosomal level. The study’s authors report encouraging precision at the sites they examined, and the strategy’s dependence on designed sequence complementarity offers a built-in specificity checkpoint that many integration methods lack. Independent replication and optimization in therapeutically relevant primary cells will be the next milestones.

What the work illustrates most clearly is how quickly the conceptual boundaries of genome editing continue to move. In barely a decade, the field has progressed from cutting DNA at chosen addresses, to rewriting individual letters of the genetic code, to contemplating the programmed installation of whole functional modules at will. Donor-directed annealing extends prime editing’s core strengths — precision, minimized DNA damage and programmability — into a size regime that those strengths had not previously reached. If the method’s efficiency and reliability hold up as it is tested more broadly, large-fragment insertion could shift from a heroic, low-yield exercise to a routine operation in the genome engineer’s toolkit, with consequences for drug discovery, synthetic biology and, ultimately, the treatment of diseases that small edits alone cannot fix.

Subject of Research: Precise integration of large DNA fragments into genomic target sites using prime editing with donor-directed annealing

Article Title: Precise genomic integration of large DNA fragments by donor-directed annealing using prime editing

Article References: Jung, H., Jeong, B., Kim, Y.-W., Jung, C., Lee, S., Uhm, H., Kim, H., Oh, Y. E., Park, Y., Lee, Y., Kang, M., Im, H. W., Kim, D., Lee, S., Kim, Y., Choi, K., & Bae, S. (2026). Precise genomic integration of large DNA fragments by donor-directed annealing using prime editing. Nature Biotechnology. https://doi.org/10.1038/s41587-026-03301-2

Image Credits: AI Generated

DOI: 10.1038/s41587-026-03301-2

Keywords: prime editing, genome editing, CRISPR, gene therapy, DNA integration, pegRNA, reverse transcriptase, Cas9 nickase, large DNA fragments, donor-directed annealing, biotechnology, genetic engineering