One macrophage line changes the organoid infection readout

What the work claims

This is a methods-plus-mechanism primary study, still a preprint. Goertz, Cipelli, Buettner and Grassl at Hannover Medical School built a two-compartment model: a confluent monolayer derived from human colon organoids on a 1-micrometer-pore Transwell membrane, with THP-1 monocyte-derived macrophage-like cells attached to the underside, so the epithelium and immune cells face each other across the porous support. Apical infection with Listeria monocytogenes or Salmonella Typhimurium then proceeds through the natural route, from the luminal side.¹

The claims: first, the co-culture is stable, with transepithelial electrical resistance developing normally and tight-junction ZO-1 staining intact. Second, under resting conditions the macrophage-like cells reduce intracellular Listeria burden but not Salmonella. Third, after 48 hours of activation with 20 ng/mL interferon-gamma and 100 ng/mL LPS, the macrophages produce a pronounced, significant reduction of both pathogens at 24 hours post-infection (multiplicity of infection 10, gentamicin protection assay, colony-forming units), alongside significantly lower infection-associated cytotoxicity measured by LDH release. Fourth, co-culture under those inflammatory conditions shifts the epithelial transcriptional program itself.¹

How it works

The engineering solves a real problem in organoid infection work. Three-dimensional intestinal organoids enclose their apical surface, so pathogens that naturally attack from the lumen must be microinjected or the organoids must be mechanically disrupted. Organoid-derived monolayers on permeable supports restore independent apical and basolateral access while keeping barrier properties: here, organoids were dissociated to single cells, seeded at 150,000 cells per insert on Matrigel-coated 1-micrometer Transwells, expanded six days to confluence, then differentiated for 48 hours, which dropped the stem marker LGR5 and raised the enterocyte marker ALPI and goblet marker MUC2. TEER climbed from day 2 to day 8.¹

The immune cell placement is the clever step. Inserts with confluent epithelium were inverted, and 70,000 PMA-differentiated THP-1 cells (500 ng/mL PMA for 48 to 72 hours) were applied directly to the basolateral membrane; after 4 hours of attachment the inserts were flipped back and differentiation continued for 48 hours. The macrophages stayed attached throughout. For activation, 20 ng/mL IFN-gamma and 100 ng/mL LPS were added basolaterally during the co-culture period, with epithelial-only controls receiving the same inflammatory cocktail, an important control because the cytokine stimulus alone perturbs the epithelium.¹

Readouts were conventional and quantitative. Infection at MOI 10 was followed by gentamicin protection and CFU enumeration at 24 hours, and LDH release quantified cytotoxicity. PMA-differentiated THP-1 cells alone significantly cut Listeria burden (representative of at least three independent experiments, three replicates per group) but left Salmonella unchanged in a single-experiment comparison; after IFN-gamma and LPS activation, both pathogens dropped significantly with at least three independent experiments each. A murine parallel using RAW264.7 macrophages under C57BL/6J colon organoid monolayers reproduced the Salmonella restriction. Bulk RNA sequencing then showed that under activated, infected conditions, co-cultures lost HLA-D and MHC class II-associated gene expression, altered S100A8 and S100A9, and shifted epithelial lineage markers: ALPI fell, MUC2 rose, LGR5 rose and OLFM4 fell, a pattern the authors validated by RT-qPCR and interpret as a move away from an enterocyte-associated state, with a non-canonical stem-cell signature.¹

Where a skeptic should push

The most load-bearing assumption is that a THP-1 cell line, chemically differentiated and then strongly activated, stands in for the lamina propria's tissue-resident macrophage compartment. The authors themselves reject this in their limitations: THP-1 macrophage-like cells are reproducible but do not reproduce the phenotype or functional diversity of primary intestinal macrophages, which in the gut are continuously renewed from blood monocytes and conditioned by the epithelium to be hyporesponsive. An IFN-gamma-plus-LPS-primed monocytic line is arguably closer to an inflammatory infiltrate than to the resident, tolerogenic macrophage that normally sits under the epithelium. The protection observed may therefore be a property of activated recruited macrophages, not of the steady-state mucosa.¹

Second, the generalization base is one donor. The human colon organoids came from biopsies of a single healthy donor (ethics permit 3082-2016). Every epithelial result, including the lineage shift that is the paper's most novel observation, is one genotype, one immune history, one microbiota exposure. The authors list extension to additional donors and patient-matched macrophages as future work. Third, there is a structural confound: the 1-micrometer pores let bacteria pass through to the macrophage compartment, and imaging indeed showed Salmonella in both layers. So the reduced CFU counts partly reflect killing in macrophages beneath the membrane, not protection of the epithelium per se; the model measures a two-cell system and cannot cleanly attribute the effect.¹

Fourth, the transcriptional data are bulk and therefore ambiguous by construction; epithelial and macrophage transcripts are pooled, so the lineage-marker changes cannot be assigned to altered gene expression versus altered cell composition, as the authors note. Single-cell or spatial methods would resolve it. Finally, the basal-state result is the honest one: without inflammatory priming, macrophages restricted Listeria but not Salmonella, meaning the robust, generalizable claim is narrower than the abstract's framing. Weight this as a well-built model with a context-dependent effect, not as a demonstration that mucosal macrophages universally protect.¹

What immune co-culture asks of organoid screens

The non-obvious implication cuts both ways, and it should unsettle any program that screens anti-infectives or mucosal modulators on epithelium alone. In one experiment this paper shows that adding a single immune cell type changes the primary efficacy readout, intracellular bacterial burden, and the toxicity readout, LDH release, in opposite directions that an epithelial-only assay would misestimate in both cases: it would overstate bacterial persistence and either over- or understate epithelial damage depending on the pathogen and the immune state. If one macrophage line does that, the true mucosal compartment, with dendritic cells, T cells, innate lymphoid cells and a stromal scaffold, will do more. Epithelial-only organoid infection screens are not simplified versions of the mucosa; they are a different, less predictive system.¹

The opportunity is a tractable upgrade path. The inverted-insert technique is simple, uses standard Transwell hardware, and the readouts (CFU, LDH, TEER, flow cytometry) are already in every screening lab. The authors also show cross-species transferability with the RAW264.7 murine system, which matters for preclinical bridging. For drug discovery, the near-term uses are concrete: testing whether candidate antimicrobials or immunomodulators work when the immune compartment is present, screening for epithelial-protective effects that only appear under inflammatory co-culture, and building pathogen restriction assays for mucosal vaccines. The GEO-deposited transcriptomic data (accession GSE345254) give the field a reference signature to compare against.¹

The threat is co-culture theater. The moment immune co-culture becomes a checkbox on grant applications and platform brochures, the field risks standardizing on exactly the cheapest surrogate, a monocytic cell line, and calling it a human immune system. This paper is unusually honest about that failure mode, but the incentives run the other way. There is also a supply-chain problem: primary tissue-resident intestinal macrophages are hard to obtain and even harder to keep in their resident phenotype ex vivo, so the lazy default will win unless funders and journals demand patient-matched validation. The other threat is subtler: the lineage shift the authors document means the epithelial state is a moving target set by the immune compartment. Drug-response baselines recorded in epithelial-only organoids may not even describe the same cell state the drug will meet in a patient.¹

The bottom line

Established: basolateral macrophage-like cells can be stably co-cultured with colon organoid monolayers without breaking the barrier; IFN-gamma and LPS-activated THP-1 cells significantly reduce intracellular Listeria and Salmonella and infection-associated cytotoxicity at 24 hours; and inflammatory co-culture measurably shifts epithelial lineage-marker programs. Established in mice only for the cross-species parallel. Not established: that resident, non-activated human mucosal macrophages do the same, that the effect generalizes beyond one organoid donor, or that the lineage shift reflects epithelial reprogramming rather than composition change. What would confirm it: patient-matched primary macrophage co-cultures across multiple donors with single-cell or spatial attribution. What would break it: failure of the restriction effect to survive replacement of THP-1 cells with unprimed tissue-resident macrophages.