Picking apart TB-infected cells one at a time reveals metabolic changes bulk methods miss

Research conducted at King's College London and the University of Surrey. Published in Analytical Chemistry, 2026.

A single human macrophage is, on average, 10 micrometers across. Its volume is less than a picolitre - roughly 100 million times smaller than a raindrop. Measuring the chemical contents of that speck has, until now, been essentially impossible when you also need to know whether the cell was infected by a bacterium, and where it sat in relation to its neighbors.

A team from King's College London and the University of Surrey has developed a technique that does exactly this. Published in Analytical Chemistry, the method selects individual macrophages under a microscope, extracts them, and runs each through liquid chromatography-mass spectrometry (LC-MS) to generate a metabolic fingerprint - a chemical snapshot of what that single cell was doing at the moment it was captured.

Why single cells matter for tuberculosis

Tuberculosis remains the world's deadliest single-agent infectious disease, caused by Mycobacterium tuberculosis. The bacterium has an unsettling survival strategy: it takes up residence inside macrophages, the very immune cells whose job is to destroy pathogens. But not every macrophage in a population becomes infected. Some are invaded; others, sitting right beside infected cells, remain uninfected. Understanding what distinguishes these two groups could point toward new therapeutic strategies.

The problem is that conventional analytical methods crush thousands or millions of cells together and analyze the mixture. This bulk approach averages out the very differences researchers need to see. If 30% of macrophages in a dish are infected, a bulk analysis gives you the mean chemistry of a 70-30 mixture - not the distinct profiles of infected and uninfected cells.

Some existing single-cell methods sort cells into groups before analysis using techniques like flow cytometry. But sorting disrupts the cells' spatial context - you lose track of which cells were neighbors, which were isolated, which were clustered near an infection hotspot. The new approach avoids this by using a microscope to visually identify and individually select cells from their original positions, preserving information about their spatial relationships.

Reading chemistry at the picolitre scale

The technical achievement here is sensitivity. Metabolites - the small molecules that are the byproducts and intermediates of cellular metabolism - exist at vanishingly low concentrations in a single cell. The researchers had to push LC-MS detection limits to extract meaningful signals from these tiny samples.

Using bacteria that model TB infection (rather than live M. tuberculosis, which requires biosafety level 3 containment), the team infected populations of human macrophages in culture dishes. Under the microscope, they could see which cells contained bacteria and which did not. They then selected individual cells from both categories, captured them, and ran their chemical contents through the mass spectrometer.

The results showed distinct metabolic profiles between infected and uninfected macrophages. The bacteria alter the host cell's metabolism in measurable ways - changes that disappear into noise when cells are analyzed in bulk. The specific metabolic shifts were not detailed in the available information, but the demonstration that such shifts are detectable at the single-cell level is itself the primary contribution.

Mapping infection neighborhoods

Because the technique preserves each cell's location, it opens the possibility of spatial metabolic mapping. Researchers can ask not just whether a cell is infected, but whether uninfected cells near an infected cell differ from uninfected cells farther away. Do infected macrophages send chemical warning signals to their neighbors? Does proximity to infection alter an uninfected cell's metabolism in ways that make it more resistant - or more vulnerable - to subsequent infection?

These spatial questions have been largely inaccessible with existing methods. Flow cytometry-sorted populations lose their positional information. Mass spectrometry imaging can map chemistry spatially but lacks the molecular specificity and sensitivity of LC-MS for metabolites. The new technique bridges these limitations, though it comes with its own constraints: manually selecting cells under a microscope is slow, and the number of cells that can be analyzed in a single experiment is small compared to high-throughput methods.

Beyond tuberculosis

The team at King's College London has already begun extending the approach beyond TB. Professor Melanie Bailey, the study's senior author, noted that the method is being applied to questions about other bacterial, viral, and fungal infections, as well as cancer biology and fundamental cell communication research. The SEISMIC Facility at King's College London, which specializes in single-cell studies, will house the ongoing work.

Dr. Dany Beste of the University of Surrey, a co-author, emphasized that the project required genuine collaboration between biologists and chemists - neither discipline alone had the tools to pose and answer these questions.

Practical constraints and open questions

The technique has clear limitations that should temper expectations. Manual cell selection under a microscope is labor-intensive and low-throughput. Each cell must be individually identified, picked, and processed. This constrains sample sizes and makes the approach impractical for large-scale screening studies.

The current work used model organisms rather than virulent M. tuberculosis. While model bacteria can recapitulate some aspects of TB infection, they do not possess the full arsenal of virulence factors that the real pathogen deploys to subvert macrophage defenses. The metabolic changes seen with model organisms may not perfectly mirror those caused by actual TB infection.

The sensitivity advances, while impressive, likely cannot capture every metabolite present in a single cell. Less abundant signaling molecules, short-lived intermediates, and compounds that ionize poorly in mass spectrometry may remain below detection limits. The metabolic fingerprint is real but incomplete.

And the leap from observing metabolic differences between infected and uninfected cells to understanding the causal mechanisms of infection susceptibility is substantial. Correlation between infection status and metabolic profile does not reveal which came first: a cell might have a distinct metabolism because it was infected, or it might have been infected because its pre-existing metabolism made it vulnerable. Longitudinal studies - tracking cells before, during, and after infection - would be needed to disentangle cause and effect, and those present additional technical challenges at the single-cell level.

Still, having a way to chemically interrogate individual infected cells while preserving their spatial context is a genuine methodological advance. For a disease that kills more than a million people annually and whose intracellular biology remains incompletely understood, any new window into the host-pathogen interaction is worth opening.