Using Detectron to read which phage was active in which bacterial strain

Finding a phage in a bacterial culture tells us little about which strain it became active in. Detectron, developed by researchers at Gladstone Institutes, makes the host cell produce a DNA barcode when phage RNA appears inside it. A short-lived RNA signal thus leaves a trace that researchers can read later by sequencing. Original study in Nature Biotechnology ↗

The sensor combines a switch that changes shape when it binds its target RNA with a retron, a bacterial reverse-transcription mechanism. When the switch opens, the system produces DNA with a sequence assigned during design. In the study, the researchers mixed E. coli carrying sensors for T4, T5 and T7 and added one phage at a time. For each phage, its corresponding barcode gave a stronger signal than the other two. Experiment distinguishing three phages ↗

The readout can also identify the host strain. The researchers pooled known strains and distinguished which ones responded. In a separate experiment with T7, they combined the wild-type strain and five mutants with differences in cell surface structure. The barcode produced in response to infection encoded both the phage and the bacterial strain. By taking the ratio to a reference barcode continuously produced by the same strain, the researchers could compare responses in mixed culture. They assigned the strain identifiers before the experiment. Strain-by-strain comparison in mixed culture ↗

How accurately and quickly can it measure?

For T7, the researchers held the amount of bacteria constant and varied the amount of phage to measure the sensor response. They also found that changing where the switch recognized the same T4 RNA altered the signal. Signal strength depends on the fraction of infected cells, the expression level of the target RNA and how readily the switch binds the sequence. Converting barcode abundance to the proportion of infected cells would therefore require separate calibration. Original study, Extended Data Figures 7 and 8 ↗

The experiments demonstrated discrimination among three phages and a parallel comparison of six host strains exposed to T7. These were separate experiments. Testing all 18 combinations together remains a further step. The researchers also delayed lysis to let barcodes accumulate inside cells before recovering and sequencing the DNA. The published figures leave open how few infected cells the assay can detect in real samples, how often it confuses phages with similar sequences, and how many hours pass between culturing the sample and obtaining a result. Practical throughput depends on both the number of combinations the assay can distinguish and the time to result. Phage identification experiment ↗

Screening therapeutic phages and another possible use

In the search for therapeutic phages, a screen of known candidates could narrow the list to combinations that show phage transcription inside the target strain. Separate assays would then establish whether the phage replicates and reduces the bacterial population sufficiently. Direct application to a patient isolate would first require introducing the sensor into the isolate and verifying that it responds as intended. Gladstone’s explanation ↗

A promising fit at this stage is research into where bacteria stop a phage infection. Bacteria with altered surface receptors that make phage entry harder and bacteria that block replication after entry and transcription can both appear resistant in growth or survival assays. In a panel of strains with known genetic differences, recording early and late phage transcripts separately and comparing them with measures of adsorption and production of progeny phages could help pinpoint where an infection stalled. That is a specific application to test beyond the study. Existing E. coli resistance screen ↗ · Example of bacterial defense that blocks phage assembly ↗

What happens to cells that have experienced infection?

Could Detectron work in human cells? Its components offer a starting point. Researchers have used bacterial retrons to produce DNA inside cultured human cells. In separate work, switches that recognize a specific RNA sequence and then produce a protein have worked in mammalian cells. Connecting the two steps so that Detectron writes a DNA barcode only when its target RNA appears remains a research goal. Retrons in human cells ↗ · RNA sensing in mammalian cells ↗ · Gladstone’s development update ↗

One possible application is to study how airway epithelium recovers after infection. In a study of respiratory syncytial virus in human airway epithelium, infected ciliated cells showed lower expression of genes involved in cilia formation and innate immunity. Nearby uninfected cells, meanwhile, strongly expressed genes that respond to interferon. Cells in the same tissue had different experiences. Original study of respiratory syncytial virus in airway epithelium ↗

The experiment would record the detection of viral RNA and the host interferon response with separate barcodes. Days later, once viral RNA has declined, researchers could compare the two records with gene expression and the state of the cilia in individual cells. Do cells that survived infection recover their cilia? Or do neighboring cells replace them? Establishing replacement would also require a marker that follows the original cells and their descendants.

A separate method has already recovered a record of an earlier immune response in cultured human cells. Recordanelle, reported in October 2026, retained RNA associated with a transient interferon response inside cells and recovered it at single-cell resolution after the stimulus was removed. The opportunity to test with Detectron is whether selected events in the virus and host can leave distinct short DNA sequences. That would require measuring how many days those records last, checking that the readout can assign them to their cells of origin, and verifying that the sensor itself leaves the immune response intact. Recordanelle original study ↗