An Interferon-Braked Monocyte State and Compartment-Specific Myeloid Programming Across Severe Respiratory Infections

Severe COVID-19 is accompanied by marked remodeling of airway myeloid compartments, yet how infiltrating monocytes are routed toward inflammatory versus pro-fibrotic fates, and why resident alveolar macrophages fail to recover, remains incompletely understood. We re-analyzed BALF single-cell data (GSE145926; 3 healthy, 3 mild/moderate, 6 severe; 27,807 cells), re-clustering myeloid cells (n = 19,485) into 37 fine states (13 families) with functional module scoring, diffusion pseudotime, PAGA, a curated ligand–receptor screen, and per-donor decomposition; validation used the independent fatal-COVID-19 BALF cohort GSE157344 and the healthy-BALF cohort GSE151928 as a specificity control, a human endotoxin bone-marrow comparator (GSE212092), two sepsis blood cohorts (GSE175453; GSE167363, paired 0 h/6 h), three cross-pathogen blood cohorts (influenza A GSE243629, infant RSV GSE283746, bacterial ARDS GSE157789), and cross-disease comparison with IPF lung (GSE128033). Healthy BALF comprised resident alveolar macrophages, depleted in severe disease and replaced by interferon-activated, CXCL8-high, and SPP1+ populations. A CD14+SPP1−FCN1+ population co-expressing interferon-stimulated genes and IL1RN—an interferon-braked state distinct from LPS-type trained immunity—was enriched in severe disease, driven predominantly by one patient. PAGA delineated an FCN1+→intermediate→SPP1+ continuum; severe-enriched circuits included MIF–CD74 and an autocrine CXCL8–CXCR2 loop. The interferon-braked signature and SPP1+ compartments were reproduced in an independent fatal-COVID-19 BALF cohort and were at baseline in a healthy-BALF reference cohort (0.343 ± 0.091); LPS-challenged marrow showed the reciprocal pattern (TLR4-high, interferon-silent); and IPF lung shared SPP1/TREM2 induction with resident-AM collapse (SPP1 p = 0.0005). In sepsis blood, monocytes retained the interferon-braked signature with the LPS/TLR4 module near baseline; the signature was likewise elevated in bacterial ARDS, influenza A, and infant RSV monocytes, extending the compartment model beyond COVID-19. The data support a two-signal model: interferon restrains recruited monocytes at entry, while TLR4/HIF-1α- and SPP1-associated programs act downstream. The brake state is a measurable, stage-specific marker; MIF–CD74, CXCL8–CXCR2, and SPP1–CD44 are candidate nodes for stage-specific intervention. These mechanistic interpretations are grounded in the data and are proposed as testable hypotheses.