Also with @m-albert.bsky.social doing a great job on the image analysis and registration part, everything at @pasteur.fr

Take-home message?

Macrophage metabolism is shaped by two opposing forces: bacterial uptake activates it, while Legionella effectors attenuate and redirect that response, creating distinct metabolic states.

Pre-print: www.biorxiv.org/content/10.6...

And T4SS effectors do more than shift the average macrophage.

WT infection generated additional metabolic substates: 5 metabolic clusters versus 2 in every other condition.

Infection is metabolically heterogeneous, cell by cell.

But virulent Legionella does something very different.

Its Type IV secretion system effectors broadly counteract the metabolic response triggered by bacterial uptake, while promoting a glycolytic shift.

The first surprise: simply encountering and taking up bacteria triggers a broad metabolic activation in macrophages.

Live avirulent and heat-killed Legionella produced remarkably similar responses.

Infected macrophages are heterogeneous and bulk metabolomics hides this diversity.

Thus, we developed SpatialMetProfiler:

MALDI mass-spectrometry imaging + microscopy + image registration → linking the metabolic spectrum of each macrophage to its infection status.

We profiled >11,000 single cells

Thus, a key job of T3SS effectors is overcoming host-imposed nutrient restriction. When host metabolism supplies preferred nutrients, ST can bypass that need. Host-pathogen metabolic interactions are central to intracellular survival of Salmonella!
📄 doi.org/10.1371/jour... 9/9

The benefit needs bacterial catabolism: ΔgalK, ΔfruK & ΔglpK mutants each lost the advantage of their matching sugar. So supplemented carbon sources reach ST in the SCV and are used by the bacterium, not just taken up by the host. 8/9

Where does the T3SS mutant grow? Using Galectin-3 & SINA reporters we show it stays inside the Salmonella-containing vacuole (SCV). The ΔprgH/ΔssaV double mutant does NOT escape to the cytoplasm. Nutrients fuel its growth within the vacuole. 7/9

This was amplified in highly glycolytic RAW 264.7 cells. But it's not glycolysis itself: blocking glycolysis with 2-DG didn't stop growth if glycerol was available, and fructose lowered glycolysis yet failed to support ST. 6/9

The surprise: a T3SS-deficient ΔprgH/ΔssaV double mutant replicated in a subset of macrophages when glycerol was present. Access to preferred nutrients alone can partially substitute for T3SS-mediated virulence. 5/9

Supplementing macrophages with glycerol strongly boosted replication: 58% of cells harbored replicative ST vs 18% in control. Glycerol, more than glucose, is a preferred carbon source for ST inside human macrophages. 4/9

Using human primary macrophages + live single-cell imaging (our BATLI pipeline), we tracked thousands of infected cells over 16h and scored replication cell by cell. Key point: ST does NOT replicate in every infected macrophage. 3/9

Salmonella survival in macrophages is classically pinned on its two T3SSs (SPI-1 & SPI-2) and their effectors. But grabbing nutrients inside the host cell is a core challenge. How does macrophage metabolic heterogeneity shape ST replication? 2/9

Two pathogens, two tactics to preserve macrophage mitochondrial Δψm: Legionella reverses FOF1-ATPase with ANT forward, Salmonella does not. But the involvement of mito Complex I is common during infection of both pathogens if they can inject bacterial effectors.

journals.asm.org/doi/10.1128/...

Great work! HCS❤️AI

Thus, single-cell metabolic dynamics can forecast infection outcomes. In the spirit of @stephenwolfram.bsky.social's ideas, bacterial infection exhibits pockets of computational reducibility at the single-cell scale. Might this point to new ways to predict infection progression?

So we trained an explainable Machine Learning model using Δψm + mROS dynamics, and the AI model predicted which infected cell will support replication with 83% accuracy before bacterial replication begins