AbstractProgrammable RNA sensors such as toehold switches are used to detect specific RNA sequences. However, their reliance on protein-based or RNA-based outputs limits their use in multiplexed and sequencing-based applications. Here we introduce Detectrons, modular biosensors that couple programmable toehold switches with retron-mediated reverse transcription to transduce RNA inputs into unique DNA barcodes, converting dynamic RNA signals into durable DNA records within living cells. The framework enables alternative modes of transcript-based sensing with applications including viral infection detection. Through the construction of a synthetic toehold retron library and application of machine learning, we uncover key design principles that improve signal strength and specificity. We apply Detectrons to the multiplexed live-cell detection of specific phage infections, enabling transcript-triggered barcode synthesis and quantitative host susceptibility profiling in pooled bacterial populations. Detectrons provide a scalable and generalizable strategy for phage screening and for recording transcriptional events in complex bacterial communities.This is a preview of subscription content, access via your institutionAccess options Access through your institutionAccess Nature and 54 other Nature Portfolio journalsGet Nature+, our best-value online-access subscription27,99 € / 30 dayscancel any timeLearn moreSubscribe to this journalReceive 12 print issues and online access269,00 € per yearonly 22,42 € per issueLearn moreBuy this articlePurchase on SpringerLinkInstant access to the full article PDF.39,95 €Prices may be subject to local taxes which are calculated during checkoutFig. 1: Detectrons specifically report on target RNA by DNA barcode synthesis.Fig. 2: Optimization of Detectrons using a high-throughput variant library.Fig. 3: Machine learning on a Detectron variant library identifies key design principles for optimal on/off ratios.Fig. 4: Programmable Detectron-V2 achieves enhanced on/off ratios with high specificity across phage transcripts.Fig. 5: Detectrons enable discrimination of phage infections in a single-pot culture.Fig. 6: Ratiometric Detectrons enable multiplexed assessment of host susceptibility to phage infection.Data availabilitySequencing data associated with this study are available from the National Center for Biotechnology Information Sequence Read Archive under BioProject (PRJNA1366125). Statistical source data underlying the figures are provided in Source Data. Complete variant, plasmid, primer and statistical summary information are provided in Supplementary Tables. Source data are provided with this paper.Code availabilityCustom code used to process and analyze data from this study is available through the Shipman Lab GitHub repository (https://github.com/Shipman-Lab/Detectron).ReferencesBetzer, O. et al. In vivo neuroimaging of exosomes using gold nanoparticles. ACS Nano 11, 10883–10893 (2017).Article CAS PubMed Google Scholar Sheth, R. U., Cabral, V., Chen, S. P. & Wang, H. H. Manipulating bacterial communities by in situ microbiome engineering. Trends Genet. 32, 189–200 (2016).Article CAS PubMed PubMed Central Google Scholar Ganeshan, S. D. & Hosseinidoust, Z. Phage therapy with a focus on the human microbiota. Antibiotics 8, 131 (2019).Article CAS Google Scholar Zhan, Y., Buchan, A. & Chen, F. Novel N4 bacteriophages prevail in the cold biosphere. Appl. Environ. Microbiol. 81, 5196–5202 (2015).Article CAS PubMed PubMed Central Google Scholar Green, A. A., Silver, P. A., Collins, J. J. & Yin, P. Toehold switches: de novo-designed regulators of gene expression. Cell 159, 925–939 (2014).Article CAS PubMed PubMed Central Google Scholar Green, A. A. et al. Complex cellular logic computation using ribocomputing devices. Nature 548, 117–121 (2017).Article CAS PubMed PubMed Central Google Scholar Pardee, K. et al. Rapid, low-cost detection of Zika virus using programmable biomolecular components. Cell 165, 1255–1266 (2016).Article CAS PubMed Google Scholar Takahashi, M. K. et al. A low-cost paper-based synthetic biology platform for analyzing gut microbiota and host biomarkers. Nat. Commun. 9, 3347 (2018).Article PubMed PubMed Central Google Scholar Yan, Z. et al. Programmable fluorescent aptamer-based RNA switches for rapid identification of point mutations. Nat. Chem. 17, 1826–1838 (2025).Article CAS PubMed Google Scholar Yan, Z. et al. Rapid, multiplexed, and enzyme-free nucleic acid detection using programmable aptamer-based RNA switches. Chem 10, 2220–2244 (2024).Article CAS PubMed PubMed Central Google Scholar Zhou, Y. et al. Conditional RNA interference in mammalian cells via RNA transactivation. Nat. Commun. 15, 6855 (2024).Article CAS PubMed PubMed Central Google Scholar Millman, A. et al. Bacterial retrons function in anti-phage defense. Cell 183, 1551–1561 (2020).Article CAS PubMed Google Scholar Yee, T., Furuichi, T., Inouye, S. & Inouye, M. Multicopy single-stranded DNA isolated from Myxococcus xanthus. Cell 38, 203–209 (1984).Article CAS PubMed Google Scholar González-Delgado, A. et al. Simultaneous multi-site editing of individual genomes using retron arrays. Nat. Chem. Biol. 20, 1482–1492 (2024).Article PubMed PubMed Central Google Scholar Crawford, K. D., Khan, A. G., Lopez, S. C., Goodarzi, H. & Shipman, S. L. High-throughput variant libraries and machine learning yield design rules for retron gene editors. Nucleic Acids Res. 53, gkae123 (2025).Article Google Scholar Khan, A. G. et al. An experimental census of retrons for DNA production and genome editing. Nat. Biotechnol. 43, 914–922 (2024).Article PubMed PubMed Central Google Scholar Lopez, S. C., Crawford, K. D., Lear, S. K., Bhattarai-Kline, S. & Shipman, S. L. Precise genome editing across kingdoms of life using retron-derived DNA. Nat. Chem. Biol. 18, 199–206 (2022).Article CAS PubMed Google Scholar Kong, X. et al. Precise genome editing without exogenous donor DNA via retron editing system in human cells. Protein Cell 12, 899–902 (2021).Article CAS PubMed PubMed Central Google Scholar Zhao, B., Chen, S. A. A., Lee, J. & Fraser, H. B. Bacterial retrons enable precise gene editing in human cells. CRISPR J. 5, 31–39 (2022).Article CAS PubMed PubMed Central Google Scholar Lim, D. & Maas, W. K. Reverse-transcriptase-dependent synthesis of a covalently linked, branched DNA–RNA compound in E. coli B. Cell 56, 891–904 (1989).Article CAS PubMed Google Scholar Palka, C., Fishman, C. B., Bhattarai-Kline, S., Myers, S. A. & Shipman, S. L. Retron reverse-transcriptase termination and phage defense are dependent on host RNase H1. Nucleic Acids Res. 50, 3490–3504 (2022).Article CAS PubMed PubMed Central Google Scholar Chen, T. & Guestrin, C. XGBoost: a scalable tree boosting system. In Proc. 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (eds Krishnapuram, B. & Shah, M.) 785–794 (ACM, 2016).Gootenberg, J. S. et al. Nucleic acid detection with CRISPR–Cas13a/C2c2. Science 356, 438–442 (2017).Article CAS PubMed PubMed Central Google Scholar Kellner, M. J., Koob, J. G., Gootenberg, J. S., Abudayyeh, O. O. & Zhang, F. SHERLOCK: nucleic acid detection with CRISPR nucleases. Nat. Protoc. 14, 2986–3012 (2019).Article CAS PubMed PubMed Central Google Scholar Gootenberg, J. S. et al. Multiplexed and portable nucleic acid detection platform with Cas13, Cas12a, and Csm6. Science 360, 439–444 (2018).Article CAS PubMed PubMed Central Google Scholar Jiao, C. et al. Noncanonical crRNAs derived from host transcripts enable multiplexable RNA detection by Cas9. Science 370, 1087–1094 (2020).Google Scholar McKenna, A. et al. Whole-organism lineage tracing by combinatorial and cumulative genome editing. Science 353, aaf7907 (2016).Article PubMed PubMed Central Google Scholar Frieda, K. L. et al. Synthetic recording and in situ readout of lineage information in single cells. Nature 541, 107–111 (2017).Article CAS PubMed Google Scholar Perli, S. D., Cui, C. H. & Lu, T. K. Continuous genetic recording with self-targeting CRISPR–Cas in human cells. Science 353, aag0511 (2016).Article PubMed Google Scholar Choi, J. et al. A time-resolved, multi-symbol molecular recorder via sequential genome editing. Nature 608, 98–107 (2022).Article CAS PubMed PubMed Central Google Scholar Tang, W. & Liu, D. R. Rewritable multi-event analog recording in bacterial and mammalian cells. Science 360, eaap9751 (2018).Article Google Scholar Farzadfard, F. et al. Single-nucleotide-resolution computing and memory in living cells. Mol. Cell 75, 769–780.e4 (2019).Article CAS PubMed PubMed Central Google Scholar Shipman, S. L., Nivala, J., Macklis, J. D. & Church, G. M. Molecular recordings by directed CRISPR spacer acquisition. Science 353, aaf1175 (2016).Article PubMed PubMed Central Google Scholar Bhattarai-Kline, S. et al. Recording gene expression order in DNA by CRISPR addition of retron barcodes. Nature 608, 217–225 (2022).Article CAS PubMed PubMed Central Google Scholar Sheth, R. U., Yim, S. S., Wu, F. L. & Wang, H. H. Multiplex recording of cellular events over time on CRISPR biological tape. Science 374, eabi5919 (2021).Google Scholar Farzadfard, F. & Lu, T. K. Genomically encoded analog memory with precise in vivo DNA writing in living cell populations. Science 346, 1256272 (2014).Article PubMed PubMed Central Google Scholar Roquet, N., Soleimany, A. P., Ferris, A. C., Aaronson, S. & Lu, T. K. Synthetic recombinase-based state machines in living cells. Science 353, aad6350 (2016).Article Google Scholar Mosberg, J. A., Gregg, C. J., Lajoie, M. J., Wang, H. H. & Church, G. M. Improving lambda red genome engineering in Escherichia coli via rational removal of endogenous nucleases. PLoS ONE 7, e44638 (2012).Article CAS PubMed PubMed Central Google Scholar Download referencesAcknowledgementsWe thank K. S. Pollard for discussions regarding machine learning, K. D. Crawford for discussions regarding Detectron library construction and A. G. Delgado for assistance in assessing Detectron portability.FundingWork was supported by funding from the Bachrach Family Foundation and the Robert J. Kleberg, Jr. and Helen C. Kleberg Foundation. S.L.S. is a San Francisco Biohub Investigator.Author informationAuthors and AffiliationsGladstone Institute of Data Science and Biotechnology, San Francisco, CA, USAJihoon Han & Seth L. ShipmanDepartment of Bioengineering and Therapeutic Sciences, University of California, San Francisco, San Francisco, CA, USASeth L. ShipmanChan Zuckerberg Biohub San Francisco, San Francisco, CA, USASeth L. ShipmanAuthorsJihoon HanView author publicationsSearch author on:PubMed Google ScholarSeth L. ShipmanView author publicationsSearch author on:PubMed Google ScholarContributionsJ.H., conceptualization, methodology, investigation, writing, visualization and supervision. S.L.S., conceptualization, methodology, supervision, project administration and funding acquisition.Corresponding authorCorrespondence to Seth L. Shipman.Ethics declarationsCompeting interestsJ.H. and S.L.S. are named inventors on US provisional patent application no. 63/900,194, related to the technologies described in this work.Peer reviewPeer review informationNature Biotechnology thanks Alexander Green, Baojun Wang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.Additional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Extended dataExtended Data Fig. 1 Detectrons specifically respond to target RNA expressed under an inducible promoter.a, Schematic of the experimental promoters used to express No Toehold control, Detectron, and trigger RNA, along with representations of the expressed transcripts. msr, multicopy single-stranded RNA. msd, multicopy single-stranded DNA. trigRNA, trigger RNA. b, Representative urea PAGE analysis of RT-DNA from the No Toehold control (lane 1, excluding ladder) and from Detectron in the absence (lane 2) or presence of trigger RNA expressed under the pBAD promoter (lane 3), or in the absence (lane 4) or presence of scrambled RNA expressed under the pBAD promoter (lane 5). The experiment was repeated independently three times with similar results.Extended Data Fig. 2 Construction of the Detectron variant library.a, Schematic of retron Eco1 ncRNA variants tested. b, Representative urea PAGE analysis of RT-DNA from the indicated ncRNA variants (lanes 1-4, excluding ladder) and wild-type (WT) ncRNA (lane 5). c, Representative urea PAGE analysis of RT-DNA from the No Toehold control (lane 1, excluding ladder) and Detectron-V1 in the absence (lane 2) or presence of trigger RNA expressed under the J23105 promoter (lane 3) or J23100 promoter (lane 4). The experiments shown in b and c were repeated independently three times with similar results. d, Design of Detectron variant oligonucleotides. The invariable region containing the full msr and partial msd sequences was omitted from synthesized oligos. e, Pooled oligos were cloned into either on (constitutively expressing trigger RNA) or off (no trigger RNA expression) backbones via SapI digestion. The invariable region was subsequently reinserted into both on and off libraries via BsaI digestion.Source dataExtended Data Fig. 3 Effect of shorter loop lengths on Detectron performance.a, Schematic of the focused loop-length analysis. To isolate the effect of loop length from changes in a2 length, a2 was fixed at 12 nt by excluding the loop from the a2 sequence, while loop length was varied from 12 nt to shorter lengths of 11, 9, 7, 5, 3, and 1 nt. b, Enrichment of RT-DNA + plasmid over plasmid alone for Detectrons with the indicated loop lengths, as measured by qPCR; One-way ANOVA followed by two-sided Tukey’s multiple comparisons test with adjustment for multiple comparisons for on-state signal: 11 nt versus 12 nt: P = 0.0970; 9 nt versus 12 nt: P = 0.995; 7 nt versus 12 nt: P = 4.82 × 10−7; 5 nt versus 12 nt: P = 3.35 × 10−10; 3 nt versus 12 nt: P = 3.35 × 10−8; 1 nt versus 12 nt: P = 0.543. Closed circles indicate three biological replicates. Bars represent the mean (±s.d.) of three biological replicates. c, on/off ratios of Detectrons with the indicated loop lengths; One-way ANOVA followed by two-sided Tukey’s multiple comparisons test with adjustment for multiple comparisons: 11 nt versus 12 nt: P = 0.230; 9 nt versus 12 nt: P = 0.374; 7 nt versus 12 nt: P = 5.62 × 10−8; 5 nt versus 12 nt: P = 1.44 × 10−10; 3 nt versus 12 nt: P = 4.13 × 10−6; 1 nt versus 12 nt: P = 0.430. Bars represent the mean (±s.d.) of three biological replicates.Source dataExtended Data Fig. 4 Machine learning-based model training identifies optimal structural feature set for Detectron performance prediction.a, Schematic of the machine learning workflow using XGBoost regression trained on the indicated feature sets. b, Model performance for each fold was evaluated using coefficient of determination (R2), root mean squared error (RMSE), mean absolute error (MAE), and Pearson’s correlation coefficient (r). c, Average of performance metrics across the five outer cross-validation folds. d, Paired t-tests comparing performance metrics between the indicated feature-set models.Extended Data Fig. 5 Average on/off ratios and SHAP feature effects across Detectron structural parameters.a-b, SHAP analysis of on/off ratios across (a) stem2 and (b) stem1 lengths. Each circle represents an individual Detectron variant. Bars represent the mean SHAP value of the variants. c-f, Average on/off ratios across different feature distributions: (c) combined stem lengths; (d) loop lengths; (e) bulge lengths; and (f) distance from stem to priming G. Closed circles indicate three biological replicates. Bars represent the mean (±s.d.) of three biological replicates. g, Average on/off ratios plotted against free energy of the hairpin (ΔGhairpin). Each circle represents an individual Detectron variant.Source dataExtended Data Fig. 6 Effect of extended trigger-complementary stem pairing on Detectron performance.a, Normalized on/off ratios of Detectron variants in which stem2 was present (stem2 ≥ 1 bp) and stem1 length was varied. b, Normalized on/off ratios of Detectron variants in which stem2 was absent (stem2 = 0) and stem1 length was varied, such that the full stem consisted of the trigger-complementary stem1 region. In both panels, bulge, loop and d lengths were held constant at 0 nt, 12 nt and 1 nt, respectively.Source dataExtended Data Fig. 7 Target-site selection affects Detectron-V2 performance within gp23 mRNA.a, Schematic of four 33-nt target sites selected within the full-length T4 gp23 mRNA. The 25-nt binding region within each target site represents the sequence expected to mediate initial hybridization with the Detectron-V2 switch RNA. Predicted target-site opening energies (ΔGopen) for the 25-nt binding regions were calculated using RNAplfold; lower ΔGopen values indicate greater predicted local accessibility. b, on/off ratios of Detectron-V2 devices targeting the indicated sites, plotted against the predicted opening energy of the corresponding 25-nt binding region. Closed circles indicate three biological replicates. Bars represent the mean of three biological replicates.Source dataExtended Data Fig. 8 MOI-dependent response of T7-targeting Detectron.Barcode enrichment from the T7-targeting Detectron strain following infection with T7 phage at the indicated multiplicities of infection (MOIs). For the MOI titration, the initial bacterial density was kept constant and the amount of T7 phage was varied to achieve the indicated MOIs. Barcode abundance was normalized to the WT barcode and plotted as fold change relative to the no-infection control. Bars represent the mean (±s.d.) of three biological replicates.Source dataSupplementary informationSupplementary Information (download PDF )Supplementary Text, Figs. 1–5 and legend.Reporting Summary (download PDF )Peer Review file (download PDF )Supplementary Tables 1–5 (download XLSX )Complete Detectron variant and barcode sequence table, including skpp15 primer assignments, variable switch-region sequences, barcode sequences, msd/a2 sequences, padding sequences and full synthesized oligonucleotide sequences with and without padding. Statistical summary table, including statistical tests, sample sizes, exact P values, test statistics, degrees of freedom, confidence intervals, effect sizes and multiple-comparison adjustments where applicable. Plasmid information table, including plasmid names, descriptions and relevant construct information. Detectron and trigger RNA sequences information table, including sequences for Detectron construct and trigger RNA used in this study. Primer sequence table, including primer names and oligonucleotide sequences used in this study.Source dataRights and permissionsSpringer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.Reprints and permissionsAbout this article