A 6,500-Interaction Map Exposes New Weak Points in Tuberculosis

Researchers have built the first comprehensive map of how genes essential to Mycobacterium tuberculosis affect one another, turning a difficult pathogen into a network of measurable dependencies. The Nature study tested 762 essential genes and identified nearly 6,500 genetic interactions, including relationships that point toward drug combinations worth investigating.

Tuberculosis treatment already relies on multiple antibiotics, often taken for months. Yet choosing better combinations is difficult because many genes needed for the bacterium's survival remain poorly characterized, and the number of possible drug pairings is enormous. Studying essential genes is especially awkward: deleting one can kill the bacterium before researchers learn how it works with other genes.

Turning essential genes down instead of deleting them

The Rockefeller University-led team adapted CRISPR interference, or CRISPRi, which blocks gene expression without cutting DNA. Its system carries two guide RNAs, allowing researchers to reduce two genes simultaneously. They first refined the technique in the related bacterium Mycobacterium smegmatis, then applied it to the disease-causing M. tuberculosis strain H37Rv.

The tuberculosis library contained 1,483 guide RNAs covering 762 genes and about 2.2 million guide-pair permutations. Each engineered bacterial strain carried a barcode, so its abundance could be tracked during competitive growth. The researchers compared the observed effect of suppressing a pair with the effect expected from suppressing each gene alone.

Key Number: nearly 6,500 interactions. That is how many likely genetic relationships the first tuberculosis map detected among the 762 targeted genes.

A stronger-than-expected growth defect indicates a negative genetic interaction. Such a relationship can reveal parallel survival routes or connected processes that become vulnerable when both are disturbed. A milder-than-expected effect is a positive interaction and can expose compensation between genes. Similar interaction profiles also helped cluster genes into functional groups, including cell-wall synthesis, energy production and nucleotide metabolism.

From a network to drug hypotheses

The team combined the genetic map with AlphaFold 3 predictions across roughly 600,000 gene pairs to identify relationships that might also reflect direct protein interactions. Follow-up experiments linked several poorly understood genes to the electron transport chain, a core part of bacterial energy generation.

The map also generated testable treatment ideas. Researchers validated four synergistic drug combinations and one antagonistic combination. One experimental compound increased the killing activity of ethambutol, an established tuberculosis antibiotic. The useful result is not that the map has found a ready-made regimen, but that it can reduce a vast search space to a smaller set of combinations that deserve laboratory and animal testing.

Why this matters for drug-resistant tuberculosis

Combination therapy helps suppress resistance because the bacterium must withstand several attacks at once. However, combinations can also add toxicity, interfere with each other or work only at particular concentrations. A systematic interaction map offers a biological basis for selecting pairs, including genes that sensitize the bacterium to disruptions across several pathways.

The researchers have released the interaction data through an interactive website and published their analysis code. Other groups can now examine individual relationships, test predicted targets and compare the map with chemical, structural and clinical data.

What happens next

The immediate task is to validate the strongest interactions in infection-like conditions and animal models, then test whether drugs can reproduce the effects of genetic suppression. The team also plans to extend mapping beyond essential genes. A full all-by-all tuberculosis map would require tens of millions of guide combinations, but even this first network gives drug researchers a far more structured place to start.