Systemic inflammation impairs myelopoiesis and interferon type I responses in humans

nature immunology Article https://doi.org/10.1038/s41590-025-02136-4 Systemic inflammation impairs myelopoiesis and interferon type I responses in humans Received: 15 March 2023 Accepted: 17 March 2025 Published online: 18 April 2025 Check for updates Farid Keramati 1,2,3,14, Guus P. Leijte4,14, Niklas Bruse4,14, Inge Grondman4,5, Ehsan Habibi 6,7,8, Cristian Ruiz-Moreno 1,2, Wout Megchelenbrink2,9, Annemieke M. Peters van Ton4, Hidde Heesakkers4, Manita E. Bremmers10, Erinke van Grinsven 11, Kiki Tesselaar11, Selma van Staveren11,12, Walter J. van der Velden10, Frank W. Preijers10, Brigit te Pas 2, Raoul van de Loop4, Jelle Gerretsen 4, Mihai G. Netea 5,13, Hendrik G. Stunnenberg 1,2,15 , Peter Pickkers 4,15 & Matthijs Kox 4,15 Systemic inflammatory conditions are classically characterized by an acute hyperinflammatory phase, followed by a late immunosuppressive phase that elevates the susceptibility to secondary infections. Comprehensive mechanistic understanding of these phases is largely lacking. To address this gap, we leveraged a controlled, human in vivo model of lipopolysaccharide (LPS)-induced systemic inflammation encompassing both phases. Single-cell RNA sequencing during the acute hyperinflammatory phase identified an inflammatory CD163+SLC39A8+CALR+ monocyte-like subset (infMono) at 4 h post-LPS administration. The late immunosuppressive phase was characterized by diminished expression of type I interferon (IFN)-responsive genes in monocytes, impaired myelopoiesis and a pronounced attenuation of the immune response on a secondary LPS challenge 1 week after the first. The infMono gene program and impaired myelopoiesis were also detected in patient cohorts with bacterial sepsis and coronavirus disease. IFNβ treatment restored type-I IFN responses and proinflammatory cytokine production and induced monocyte maturation, suggesting a potential treatment option for immunosuppression. Systemic inflammation plays a pivotal role in the pathophysiology of several diseases, such as sepsis and viral infections, including COVID-19, and contributes to 20% of global mortality1–4. Systemic inflammation is classically characterized by an acute hyperinflammatory phase, often termed as ‘cytokine storm’, followed by a late immunosuppressive phase5 that renders patients susceptible to secondary infections6. This sustained refractory state of the immune system is associated with high late-onset mortality7–10. Despite the substantial burden on healthcare systems, unraveling the molecular mechanisms underpinning systemic inflammation-related pathogenicity is extremely difficult, as a result of extensive heterogeneity in terms of inflammation A full list of affiliations appears at the end of the paper. Nature Immunology | Volume 26 | May 2025 | 737–747 onset time, etiology, site of infection and underlying comorbidities. Mouse models of systemic inflammation are ubiquitously used and valuable, but suffer from important interspecies differences and poorly replicate the human inflammatory response11, thereby limiting translatability12. The human in vivo model of lipopolysaccharide (LPS)-induced systemic inflammation13, also known as experimental endotoxemia, is a tightly controlled systemic inflammation model in which healthy volunteers are injected with bacterial LPS intravenously to elicit transient but profound acute hyperinflammation, followed by a late immunosuppressive phase14. Previous studies utilizing the experimental e-mail: 737 Article endotoxemia model primarily conducted transcriptomic analyses at the whole-blood level and focused on the hyperinflammatory phase, thereby limiting the ability to dissect unbiased cell-type-specific effects of systemic inflammation and examine the mechanistic underpinnings of the immunosuppressive phase15. In the present study, we combined massively parallel single-cell RNA sequencing (scRNA-seq) and cellular functional assays to characterize the hyperinflammatory and immunosuppressive phases after LPS administration in peripheral blood and bone marrow leukocytes. Leveraging of publicly available large cohorts of patients with sepsis and COVID-19 indicated the clinical relevance of our comprehensive longitudinal dataset. Our results identified conserved gene expression programs, mainly in monocytes and T cells, during the acute LPS-induced hyperinflammatory phase, early sepsis and COVID-19. In the immunosuppressive phase, we observed a significant impairment of monocyte functionality and maturation accompanied by decreased interferon type I (IFN-I) signaling that could be restored by IFNβ treatment. Results LPS induces hyperinflammation followed by immunosuppression To study systemic inflammation in humans in vivo, healthy male volunteers (Supplementary Table 1) were intravenously injected with 2 ng kg−1 of LPS (n = 7, age 24 (19–30) years) or placebo (n = 4, age 19 (18–28) years; Fig. 1a). The appearance of clinical systemic inflammation symptoms, such as fever and tachycardia (Extended Data Fig. 1a), and a transient profound increase in circulating cytokines, including tumor necrosis factor (TNF), interleukin (IL)-10, CCL4, CCL3, IL-6, CXCL8, CCL2, IL-1RN (IL-1ra gene) (Extended Data Fig. 1b) up to 8 h after LPS administration, confirmed the induction of hyperinflammation. Severe monocytopenia was observed at 1 h after LPS administration (Extended Data Fig. 1c), with CD14+CD16− classic monocytes (cMonos) starting to repopulate the blood ~3 h post-LPS and returning to baseline levels (defined as immediately before the LPS challenge) at ~6 h post-LPS16 (Extended Data Fig. 1c). Between 6 h and day 7 (d7) after the LPS challenge, cMonos gradually differentiated into CD14+CD16+ intermediate monocytes (iMonos) and CD14−CD16+ nonclassic monocytes (ncMonos) (Extended Data Fig. 1d)16; however, the abundance of these subsets in blood remained significantly decreased at d7 after the LPS challenge (Fig. 1b), suggesting impaired monocyte maturation. Bulk RNA-seq of blood CD14+ monocytes obtained from LPSchallenged volunteers (n = 3) showed significant perturbation at 4 h and 8 h post-LPS compared with baseline (1,459 upregulated and 1,222 downregulated genes), which substantially normalized at 24 h post-LPS (Fig. 1c), with no significant differentially expressed genes (DEGs) at d7 compared with baseline (Fig. 1d and Extended Data Fig. 2a). Gene ontology (GO) of upregulated DEGs at 4–8 h post-LPS was mainly attributed to inflammatory response (CXCL8 and IL-6) and IFN type I (IFN-I) signaling pathways (MX1 and IRF7) (Fig. 1e), whereas antigen presentation through major histocompatibility complex (MHC) class II was the most significant GO term of downregulated genes at 4 h (HLA-DRA and HLA-DRB5) (Fig. 1e and Extended Data Fig. 2b). As decreased expression of the MHC-I receptor human leukocyte antigen isotype DR (HLA-DR) in CD14+ monocytes indicates the induction of T cell inhibitory, myeloid-derived suppressor cells (MDSCs), a correlate o (...truncated)