Detailed critical analysis (evidence-first)

1) Novelty and conceptual advance

The study reframes granulomas as heterocellular ecological niches in which non-immune stromal cells act as active immune modulators: (a) stromal PDPN+ cells uptake and harbor mycobacteria in vivo and in vitro; (b) infected stromal cells adopt IFN/xenophagy/autophagy and lipid metabolic programs; (c) stromal cells secrete CCL2/CCL7 and CSF1, recruiting monocytes and promoting Mφ maturation; (d) SC-to-Mφ signaling (CCL axis, CSF1, TNF feedback) shapes granuloma architecture and is conserved in macaque/human reanalyses — this conceptual shift (non-immune cell as host + immune-organizer) is an important advance and well-supported by multimodal data (Core findings (Lösslein et al.)).

2) Experimental strengths

• Model that generates mature necrotic granulomas in C57BL/6 mice (omentum) — enables fate mapping and spatial analysis.
• Multiple orthogonal methods: fate-mapping (Cxcr4 creER:R26 tdTomato), adoptive transfers, TEM, in vitro survival assays (gentamicin protection), bulk Smart3 sequencing, 10x scRNA-seq, CellChat/NicheNet signaling inference, and functional perturbation (Pdgfrα CreER:R26 DTA/+ stromal reduction, antibiotics).
• Cross-species reanalysis (Gideon et al. macaque; Wang et al. human) to test conservation of stromal–myeloid ligand signatures.

Citation for methodological breadth and translational linking: Methods & multimodal data (Lösslein et al.).

3) Limitations, sources of bias, and alternative explanations

1. BCG vs virulent M.tb and anatomical site: intraperitoneal BCG in the omentum is an experimentally tractable granuloma model but differs from pulmonary M.tb in virulence, immune cues, and tissue architecture; results may not fully translate to lung TB granulomas (Model limitations (Lösslein et al.)).
2. Reproducibility / sample size: scRNA-seq used 3 infected and 5 PBS controls (pooled) with post-QC cell counts reported; modest biological n reduces power to detect rare states and increases dependency on concordant orthogonal validation (which authors partly provide) — raw sequencing deposit is not clearly stated for independent reanalysis (paper notes protocols and reanalyses but explicit raw access must be requested) (scRNA sample sizes / data availability (Lösslein et al.)).
3. Tamoxifen/DTA systemic effects and partial depletion: fate-mapping and Pdgfrα-driven DTA ablation can have off-target/systemic effects and do not achieve complete localized stromal depletion; observed systemic changes (e.g., reduced spleen weight) may partly reflect global effects; authors note these caveats (DTA depletion caveats (Lösslein et al.)).
4. Cause vs correlation for bacterial containment: SC depletion reduced macrophage numbers and serum M-CSF but did not change CFU at 4w — interpreting whether SC promote containment or simply shape architecture requires longer timelines and direct functional perturbation of SC-derived CSF1 or CCLs in vivo (blockade experiments) to prove causation.
5. Cell–cell transfer mechanism: SC-to-Mφ transfer of bacteria shown by live imaging and co-culture; authors suggest non-lytic expulsion (vomocytosis-like) but mechanism and molecular mediators (e.g., ERK5, actin remodeling) remain untested here and should be validated pharmacologically or genetically.

4) How convincing is the evidence supporting the central claims?

Overall: quite convincing for the existence of stromal infection, stromal transcriptional reprogramming (IFN/autophagy/lipid metabolism), and stromal chemokine/CSF1-driven monocyte recruitment that shapes macrophage composition. The multi-level orthogonal evidence (EM + CFU survival in SC cultures + bulk Smart3 + scRNA + CellChat/NicheNet + fate-mapping + depletion) gives strong internal consistency. However, translation to human pulmonary TB and the effect of SC manipulation on sterilizing immunity remain unresolved and require targeted perturbations (e.g., in vivo CCL2/CSF1 neutralization, conditional SC-specific CSF1 KO, and M.tb pulmonary models) (Overall evidence evaluation (Lösslein et al.)).

5) Key blindspots and what would change the conclusions

• If in virulent M.tb pulmonary granulomas stromal cells do not become infected, do not upregulate CCL/CSF1, or their ablation worsens rather than improves containment, that would overturn the implied generality.
• If targeted blockade of SC-derived CCL2/Ccl7 or CSF1 in vivo does not alter monocyte recruitment or macrophage maturation (or paradoxically improves CFU clearance), the functional role of SC-derived signals would be questioned.
• Independent scRNA-seq cohorts (larger n) that fail to replicate S2 IFN-rich stromal subsets or their xenophagy/autophagy signatures would reduce confidence in SC immune-transition claims.

6) Concrete next experiments (testable, prioritized)

1. In vivo neutralization of SC-derived CSF1 and CCL2/7 during i.p. and aerosol M.tb: use stromal-targeted (Pdgfrα-Cre-driven) conditional Csf1 or Ccl2/7 flox-knockouts to test impact on monocyte recruitment, macrophage maturation (marker panels), granuloma architecture and CFU at multiple timepoints.
2. Mechanistic dissection of SC→Mφ bacterial transfer: live-cell imaging with ERK5 inhibitors and actin-remodeling blockers + mutant BCG strains to test vomocytosis-like pathways; quantify frequency and molecular dependencies of non-lytic expulsion.
3. Pulmonary translation: isolate PDPN+/Pdgfrα+ stromal cells from human TB granuloma resections (when available) or high-burden human biopsies and perform spatial transcriptomics and ex vivo infection/transfer assays to test conservation of signatures (IFN, xenophagy, CCL, CSF1).
4. Test whether SC metabolic reprogramming (cholesterol lipid droplet accumulation) contributes to bacterial survival: pharmacological manipulation of cholesterol metabolism in SC (e.g., DHCR24 inhibitors) and measure intracellular BCG/M.tb survival and subsequent transfer to macrophages.

7) Short practical takeaways for researchers

• Include stromal compartments (PDPN+/Pdgfrα+ cells) when profiling granulomas; they can be infection reservoirs and active immune modulators.
• When interpreting granuloma scRNA-seq, explicitly test for CCL and CSF1 stromal modules and IFN/xenophagy signatures.
• Functional tests of host-directed therapy that target microenvironmental signals should assay architecture and cell composition (not just CFU), because stromal effects may change immunopathology independent of bacterial killing.

8) Reproducibility & data access

The paper details analytical pipelines (DESeq2, Seurat v5, CellChat, NicheNet) and references protocols (Smart3Seq protocol DOI provided). However, the primary raw 10x and Smart3 fastq/cellranger outputs are not explicitly given with an accession number in the main text — authors note that data may be requested. For full reproducibility the authors should deposit raw scRNA and Smart3 fastqs + processed matrices in GEO/ENA with clear sample metadata and provide code repository with analysis notebooks (Data & pipelines (Lösslein et al.)).

Quantitative Summary Scores (critical)

Short justifications (one-liners)

• Novelty 9: Demonstrates stromal infection + active immune programming inside granulomas — an underexplored cell type-function in TB granuloma biology.
• Quality 9: Multiple orthogonal assays and computational pipelines; robust mechanistic framing though some causal steps need further perturbation tests.
• Generality 8: Strong within omental/serous models and supported by cross-species ligand analyses, but pulmonary/virulent M.tb generalization requires more data.
• Usefulness 8: Suggests new host-directed targets (SC CSF1/CCL axes) and compels inclusion of stromal compartments in granuloma studies.
• Reproducibility 6: Detailed methods but absent explicit public raw 10x/Smart3 fastq accession in main text; scRNA sample n modest.
• Depth 9: Mechanistic depth (autophagy/xenophagy, CCL/CSF1, metabolic reprogramming) is high; gaps exist in the molecular vesicle/expulsion mechanism.

Novel hypotheses generated (testable)

1. SC-derived CSF1 sustains local macrophage maturation and niche stability that limits granuloma necrosis but also permits mycobacterial persistence: predict that conditional SC-specific Csf1 deletion will reduce mature iNOS+ macrophages and increase necrotic dissemination (falsifiable).
2. Non-lytic expulsion (vomocytosis-like) from stromal cells depends on ERK5-mediated signaling: predict ERK5 inhibition will reduce SC→Mφ transfer frequency in live imaging, trapping bacteria within SC and altering granuloma outcome (falsifiable).

Two high-value experiments to decide mechanism vs consequence

1. Stromal-specific Csf1 flox x Pdgfrα-CreER conditional knockout + i.p. BCG and aerosol M.tb: measure macrophage maturation markers, granuloma architecture, CFU kinetics, and survival — decisive to prove SC-origin CSF1 function.
2. Live imaging of SC-infected cultures with fluorescent reporters of phagosome integrity, coupled to ERK5 and actin inhibitors, quantifying non-lytic expulsion and subsequent macrophage uptake — identifies molecular drivers of bacterial handover.

Confidence & recommendation

Confidence in internal validity (mouse i.p. BCG model & reported mechanisms): high. Confidence in direct translation to human pulmonary TB and sterilizing immunity: moderate to low until validated with virulent M.tb lung models and human granuloma samples. I recommend: 1) authors deposit raw scRNA/Smart3 fastqs and analysis notebooks; 2) perform targeted in vivo perturbations (SC-CSF1/CCL axis) and pulmonary M.tb validation; 3) pursue human granuloma spatial transcriptomics for translational relevance (Translational caution (Lösslein et al.)).

Useful actionable items (what I can do next for you)

• Run a focused reanalysis of the authors' scRNA marker lists (Ccl2/Ccl7/Gbp2/Gbp3 modules) and cross-compare to Gideon et al. and Wang et al. datasets (I can run this if you want raw matrices) — would quantify conservation across species.
• Generate a prioritized experimental plan and sample-size calc for conditional SC-Csf1 KO and pulmonary M.tb experiments.

Primary source (this review is entirely grounded on the full paper text and figures): Main paper (Lösslein et al. 2026)